The Technology World Quarterly Journal · GPS, Satellite clock bias, GAGAN. 1. INTRODUCTION GPS is an all weather, line-of-sight radio navigation and positioning system. GPS was developed

The Technology WorldQuarterly Journal

Gps Position Error Analysis Comparison Using Point Solution Approach And Least Squares 3 - 8Approximation Algorithms For Surveying And Gagan ApplicationsV. VENKATA RAO (1), DR. G. SASIBHUSHANA RAO(2) AND DR. M. MADHAVI LATHA(3)

Usage Of Extreme Programming Process Methodology When Implemented On 9 - 12The Global ProjectsN.GANESH1, S.THANGASAMY2

Geo position based routing for mobile ad - hoc networks (manet’s) implementation through 13- 18 Embedded systems1. RAJARAM, 2. MUTHULAKSHMI, 3. SHANTHI, 4. Dr. V.SUMATHY

Design Of Smith-Predictors For Time-Delay Processes With Auto-Tuning Techniques 19 - 28 DR.G.SARAVANA KUMAR1DR.R.S.D.WAHIDABANU2 ANDMR.K.G.ARUN RAAJESH3

Textual Analysis Of Stock Market Prediction Using Financial News Articles 29 - 38 M.SURESH BABU, DR. N.GEETHANJALI, V.RATNA KUMAR I

Cryptography Using Neural Network 39- 44 T.GODHAVARI, DR.N.R.ALAMELU

Optimisation and enhancing cathode life of space helix twt 45-50 SIVANANDAM.M1, RAVI.S2

Experimental Investigation Of Forced Convective Heat Transfer In Rectangular Micro- 51 - 56 Channels R.KALAIVANAN1 and R.RATHNASAMY2

Domestic Electrical Energy Modeling For India 57- 70N V RAO1 AND VVS KESAVA RAO2

Effect Of Temperature On Microwave Conductivity In Polyvinyl Chloride 71 - 74 R.MURUGESH*

a novel approach for secure data storage in desktop data grid 75 - 80

DR. G.VASANTH1 , DR B.S.VISHWANATH 2, MR. T. SUDALAI MUTHU 3 MRS. SAVITA HARKUDE4

statistical modeling of gtaw for weld strength and hardness in welding zone w. r. t. 304 stainless steel 81 - 86MR. A.R. DESHPANDE, PROF. DR. M.L. KULKARNI, PROF. DR. P.J.AWASARE, PROFESSOR

Numerical Simulation And Experimental Validation On Diabatic Helical Capillary Tube With R-22 87 - 96

DR HEMACHANDRA REDDY.K1 , KRISHNA REDDY.V 2

Land Use Change Detections In Tirumullaivasal Village Using Remote Sensing And Geographic 97 - 100Information System

THE TECHNOLOGY WORLD QUARTERLY JOURNAL 3 DECEMBER 2010 VOLUME II ISSUE 4 ISSN : 2180 - 1614

Contents Page No

V.RAJESH KUMAR1 and G.VICTORRAJAMANICKAM2

Pitch Rate And Reynolds Number Effects On Hysteresis Behavior Of Flow Past A Pitching Airfoil 101 - 110 D.M.SHARMA1, KAMAL PODDAR2, A. MUTHUKUMAR3, AND K.S.V.REDDY4

Robust Image Watermarking Using Dwt 111 - 116S.DHANDAPANI1 AND DR.A.KRISHNAN2

Multiobjective approach for feeder overloading And service restoration through 117 - 122Reconfiguration

P. RAVI BABU1, K. PRAPOORNA2, N.V. PRASHANTH3, A.SHRUTI4, D. PRABHUVARDHAN5, V.P. SREE DIVYA6

An Innovative Design And Energy Output Estimation Of Wind-Solar Hybrid Renewable Energy 123 - 132 Generation System With Rain Water Collection FeatureCHONG WEN TONG*, POH SIN CHEW,HAKIM S. SULTAN, PAN KOK CHEN AND DEEP CHEAH HOW

An investigation into validation of deep Drawing with abs 133 - 140M. MORADIA , A. R. FATHIB

Characteristics Of Convective Mass Transfer Under Influence Of Turbulence Control With 141 - 146Wall-Recess In A Parallel Plate Electrochemical Flow Cell HARINALDI, DAMORA RHAKASYWI AND HANIFA AKROM

Power Quality Disturbance Classification Using Wavelet And Fuzzy Logic 147 - 154PRAMILA P1 , S.PURUSHOTHAMAN2 AND PUTTAMADAPPAC3

Malisous Node Detection System For Mobile Ad Hoc Networks 155 - 164A.RAJARAM, DR. S. PALANISWAMI

Numerical Prediction Of Cavitating Flows In A Ball Valve 165 - 172Y. GAWAS**, DR. V. R. KALAMKAR**, V. MALI***

Optimal Power Flow In Power System Network By Using Facts Device 173 - 180J.SUNILKUMAR*M.TECH, CH. CHENGAIAHA** M.TECH (P.HD)

An Experimental Analysis Of Two Sided Assembly Line Balancing Using Ga 181 - 188

MUGALE UMESH S. 1, DR. NANDEDKAR VILAS M. 2, MUGALE RAMESH S. 3Performance And Emission Characteristics Of A Diesel Engine Fueled With Palm Oil Methyl Ester 189 - 192And Its Blends With Diesel

B. DEEPANRAJ1, P. LAWRENCE2

Durability Properties Of Self Curing Concrete By Addition Of Vegetative Material As Admixtures 193 - 1961GEETHA M AND 2Dr. R.MALATHY

Investigations On Effect Of Natural Shading For Thermal Comfort In Hot Climatic Anantapur, India 197 - 202P. SANJEEVA RAYUDU A*, DR. K. HEMACHANDRA REDDY A


GPS POSITION ERROR ANALYSIS COMPARISON USING POINT SOLUTION APPROACH AND LEAST SQUARES APPROXIMATION

ALGORITHMS FOR SURVEYING AND GAGAN APPLICATIONSV. VENKATA RAO (1), DR. G. SASIBHUSHANA RAO(2) AND DR. M. MADHAVI

LATHA(3)

(1) Tirumala Engg College, Narasaraopet, Guntur (Dt), A.P.,E-mail: [email protected](2) Department of ECE, AU College of Engg, Andhra University, Visakhapatnam-530003

(3) Department of ECE, JNTU College of Engg, JNTU, Hyderabad

ABSTRACTThe Global Positioning System (GPS) is a worldwide satellite-based navigation system. GPS provides highly accurate positions of objects, their velocity and time data, and its accuracy relies in the precise knowledge of the satellite orbits and the time. Different approaches have been developed to estimate the receiver position. In this paper a non recursive point solution approach algorithm is proposed for the receiver position estimation and it is compared with the recursive least squares approximation method. The receiver position is estimated without considering the various error sources namely satellite clock bias, atmospheric delay, error in the broadcast ephemeris data, multipath error and receiver tracking error. The results are analyzed by computing the mean position ( ), standard deviation (σ) and variance (σ2) over a period of the day for both the methods. The error analysis is carried out by collecting the several days of dual frequency (L 1 and L2) GPS receiver data from the Andhra University Engineering College, Visakhapatnam (Latitude /Longitude 17.73o/83.32o). The position error analysis shows that the proposed method provides a better position estimation when compared to recursive least squares approximation method. The error analysis made in this paper will be helpful to the surveyors and the GPS Aided Geo Augmented Navigation (GAGAN) users over the Indian subcontinent in determining the obtained position accuracy over the 24 Hour period particularly for the aircraft purposes.

Key WordsGPS, Satellite clock bias, GAGAN.

1. INTRODUCTION

GPS is an all weather, line-of-sight radio navigation and positioning system. GPS was developed primarily for the military purpose. The multipurpose usage of Navigation Signal Timing And Ranging (NAVSTAR) GPS has developed enormously within the last three decades. With the elimination of Selective Availability (SA) on May 2nd, 2000, the usefulness of the system for civilian users was even more pronounced. The system became fully operational in 1994 [1]. Currently there are 32 operational satellites in the constellation. The GPS consists of three major segments: space, control and user [2]. The space segment consists of a nominal 24 operational satellites, which are constantly orbiting the surface at an altitude of approximately 3 Earth radii, and broadcast signals which travel at approximately the speed of light. Each satellite has a unique identification number. The control segment monitors the health and status of the space segment.

The user segment consists of antennas and receiver processors which receive the signals broadcasted by the satellites, and decode them to provide precise information about the receiver’s position and velocity. There are two fundamental GPS observables, namely the pseudorange and the carrier phase, which can be used to estimate the receiver position.

2. NONRECURSIVEPOINTSOLUTION APPROACH ALGORITHMThe GPS navigation and position determination is based on measuring the distance from the user position to the precise locations of the GPS satellites as they orbit [3]. The standard GPS positioning system of pseudorange nonlinear equations are

+ c dt (1)Where (Xi, Yi, Zi) is the satellite position of the ith

satellite ( i =1 to 4), (x, y, z) is receiver location and c dt is the range bias. C is speed of light and dt is receiver clock offset.


Different approaches have been developed to solve the system of nonlinear Eq. (1), some involving closed-form solution, and some through linearization [4]. Since these approaches involve calculation of the receiver position from a single measurement of pseudoranges, they are called point solution approaches. An algorithm is proposed which is a closed form point solution method and is a non-iterative algorithm requiring less computational time compared to the conventional iterative algorithms such as recursive least squares and Kalman filter etc [6] and is described as follows.

The pseudorange equation corresponding to the ith

satellite can be written as

(2)

where b=c dt. Squaring on both sides of Eq. (2) and regrouping terms, we get

(3)

If the Lorentz inner product is defined as

Kf

where K=diagonal matrix [1,1,1,-1] (4)

Equation (3) can be written as

(5)

Where Li=[ ]T and =[x, y, z]T

Considering that each pseudorange generates an equation of the form (5), four equations are sufficient to solve for the unknowns. If we define matrix A with all known quantities as

The four equations may be written in compact form as

(6)

Where e=[1 1 1 1]T

The jth entry of β is given by

and

Then from equation (6), we get

= KA – 1( ) (7)

By substituting (7) into (6), we will get the quadratic equation in

(8)

Equation (8) gives two possible values for . Now the two solutions are obtained by inserting the values of into equation (7) and one among them is correct one.3. RESULTS AND DISCUSSION

The GPS data required for the paper was collected from a newly installed dual frequency GPS receiver (NovaTel make DLV3) at Andhra University College of Engineering, Visakhapatnam. The two data files (Navigation data and Observation data) were collected for 19th February, 2010. The navigation data file consists of 38 parameters. But for the calculation of satellite position, elevation angle etc., only 23 parameters are used. The satellite positions estimated at a particular epoch 10:00:00 hours along with the pseudorange observed on L1 due to C/A code are presented in Table 1.The actual user position coordinates are X=706970.909 m, Y=6035941.022 m and Z=1930009.582 m. Using the proposed algorithm the user position is estimated considering pseudorange observed on L1 due to C/A code and the corresponding X position versus local time is plotted in Figure 1. The error in user position is plotted in Figure 2. The X position of the user after smoothing over an hour is plotted in Figure 3. The error in user positions after smoothing over an hour is also plotted in Figure 4. The maximum error in user position is found at 16:00:00 hours and the reason for that is observed to be visibility of less number of satellites.The user position is estimated using the proposed and least squares approximation methods after smoothening over an hour and the error in user position are presented in Table 2 and Table 3 respectively. The mean ( ), Standard deviation (σ) and Variance (σ2) are also calculated and presented. The comparison of estimated user position error analysis is presented in Table 4. From Table 4 it is clear that the proposed non recursive point solution


approach algorithm provides better user position estimation.

Figure 1 Local Time vs. X position

Figure 2 Local Time vs. Error in User Position

5 1 0 1 5 2 0 2 57 . 0 6 9 4

7 . 0 6 9 4

7 . 0 6 9 5

7 . 0 6 9 5

7 . 0 6 9 6

7 . 0 6 9 7

7 . 0 6 9 7

7 . 0 6 9 8x 1 0

5

L o c a l T i m e i n H o u r s

X p

os

itio

n o

f th

e u

se

r in

Me

ters

Figure 3 Local Time vs. X positionsmoothed over an hour

Figure 4 Local Time vs. Error in User position smoothed over an hour

4. CONCLUSIONS

A new point solution algorithm based on non recursive approach is proposed for accurate GPS receiver position estimation. The results of the proposed method are compared with the recursive least squares method computation. The pseudo range errors due to the atmosphere, multipath and satellite clock are not taken into consideration in the receiver position computation. The maximum and minimum position error observed in the horizontal (X) position due to the proposed method are 28.405 m and 6.356 m and for the least squares approximation method are 62.135 m and 39.081 m, which are more. Using the proposed algorithm, better user position accuracy (± 1meter) can be obtained, if all the errors are taken into account. Therefore, the proposed algorithm will be helpful to the precise accuracy requirement users over the Indian subcontinent like carrier phase based surveying and GAGAN users.


REFERENCES

[1] G. S. Rao, “Global Navigation Satellite Systems”, Mc Graw Hill, New Delhi, 2010.

[2] Bradford W. Parkinson and J. J. Spilker Jr, “Global Positioning System: Theory and Applications – vol. I and II”, American Institute of Aeronautics and Astronautics, Washington, 1996.

[3] Elliot D. Kaplan, “Understanding GPS”, Artech House, Boston, 1996.

[4] Farrell J, Barth M, “Global positioning system & inertial navigation”, New York: McGraw-Hill, 1999.

[5] Seiji Yamaguchi, Toshiyuki Tanaka, “GPS Standard Positioning using Kalman filter”, SICE-ICASEInternational Joint Conference, Busan, Korea, Oct. 18-21, 2006.

Table 1 Satellite positions along with pseudorange observed on L1 (1575.42MHz) due to C/A code

S. No. SV No.

Satellite Position in Meters Pseudorange observed on

L1 due to C/A code (Meters)

X position Y position Z position

1 3 24709979.959 9671417.231 3006665.228 24146014.0002 6 23289532.856 12889496.515 806924.914 23551751.6003 14 11111868.900 19587667.382 14315889.112 21099178.4004 19 22374584.724 5027682.195 13682277.361 24675000.3005 21 -4312562.315 25585351.285 -3475509.263 20907167.9006 22 1063540.881 16523095.777 20913552.598 21637020.8007 24 -10255557.017 21981750.045 -11050774.253 23222544.8008 26 199982.656 16540119.640 20109939.533 21008229.4009 27 -14925544.822 4751752.406 22131066.547 25537658.80010 30 -17705836.449 17602711.884 -9368558.533 24438821.20011 31 7772645.116 21482156.832 -13186935.093 22753809.500

Table 2 User position after smoothening over an hour and the error in user position due to the proposed method

S.No.Local

Time in Hours

User Position in Meters Error in Meters

X position Y position Z position X position Y position Z position1 7 706958.479 6035827.593 1929972.835 12.429 113.429 36.7462 8 706958.475 6035838.521 1929972.923 12.434 102.501 36.6583 9 706957.148 6035834.393 1929975.518 13.760 106.629 34.0644 10 706947.980 6035833.428 1929978.263 22.928 107.594 31.3185 11 706948.796 6035832.762 1929975.676 22.113 108.260 33.9056 12 706946.136 6035838.410 1929978.089 24.772 102.612 31.4927 13 706951.370 6035833.205 1929975.697 19.539 107.817 33.8858 14 706942.503 6035815.279 1929980.697 28.405 125.743 28.8849 15 706947.860 6035809.015 1929974.421 23.048 132.007 35.16010 16 706960.715 6035797.768 1929961.697 10.194 143.254 47.88411 17 706964.552 6035832.374 1929971.361 6.356 108.647 38.22012 18 706960.901 6035832.167 1929981.154 10.007 108.854 28.42713 19 706952.209 6035817.442 1929967.814 18.700 123.579 41.76714 20 706953.929 6035820.802 1929969.905 16.979 120.219 39.676


15 21 706957.695 6035826.741 1929972.482 13.213 114.281 37.09916 22 706961.518 6035815.382 1929969.681 9.390 125.639 39.90017 23 706956.004 6035832.084 1929972.803 14.905 108.938 36.77818 24 706953.618 6035842.313 1929979.925 17.291 98.709 29.657

Mean ( ) 706954.438 6035826.649 1929973.941 16.470 114.373 35.640Standard

deviation (σ) 6.088 11.694 4.925 6.088 11.694 4.925

Variance (σ2) 37.065 136.753 24.258 37.065 136.753 24.258

Table 3 User position after smoothening over an hour and the error in user position due to the least squares approximation method

S.No.Local

Time in Hours

User Position in Meters Error in Meters

X position Y position Z position X position Y position Z position1 7 706925.834 6035832.032 1929972.834 45.075 108.990 36.748 2 8 706926.058 6035841.170 1929972.617 44.850 99.851 36.9643 9 706924.536 6035837.777 1929975.731 46.372 103.245 33.8504 10 706915.170 6035835.274 1929978.323 55.738 105.748 31.2595 11 706915.538 6035837.034 1929975.505 55.371 103.988 34.0766 12 706912.539 6035841.487 1929978.229 58.370 99.535 31.3527 13 706917.332 6035834.639 1929975.082 53.576 106.382 34.4998 14 706908.773 6035818.085 1929980.064 62.135 122.936 29.5179 15 706914.113 6035811.379 1929973.922 56.795 129.643 35.65910 16 706927.557 6035798.580 1929960.929 43.352 142.442 48.65211 17 706931.827 6035836.293 1929971.693 39.081 104.728 37.88812 18 706927.653 6035834.611 1929981.231 43.255 106.410 28.35113 19 706919.128 6035820.589 1929967.530 51.780 120.433 42.05114 20 706921.220 6035823.080 1929969.481 49.688 117.942 40.10015 21 706925.670 6035829.216 1929972.299 45.239 111.806 37.28216 22 706929.759 6035818.125 1929969.346 41.150 122.897 40.23517 23 706922.574 6035835.922 1929973.147 48.334 105.099 36.43418 24 706921.657 6035846.673 1929979.893 49.251 94.349 29.688

Mean ( ) 706921.497 6035829.554 1929973.770 49.412 111.468 35.811Standard deviation

(σ) 6.472 12.234 5.045 6.472 12.234 5.045

Variance (σ2) 41.898 149.678 25.461 41.898 149.67 25.461


Table 4 Comparison of the performance of recursive least squares approximation method and the proposed point solution approach method in terms of position error

S.No. Parameter Recursive Least Squares Approach Proposed

X Position Y Position Z Position X Position Y Position Z Position1 Mean ( ) (m) 49.412 111.468 35.811 16.470 114.373 35.640

2Standard

deviation (σ) (m)

6.472 12.234 5.045 6.088 11.694 4.925

3 Variance (σ2) 41.898 149.678 25.461 37.065 136.753 24.258

AUTHORS PROFILESMr. V. Venkata RaoHe is an Associate Professor in the Department of Electronics and Communication Engineering, Tirumala Engineering College, Jonnalagadda, Andhra Pradesh, India. He has over 16 years of experience in varied fields like Industry, R & D and academics. He has published and presented 15 research papers in various reputed international/national journals and conferences respectively. He is a Life Member of ISTE. His research areas include GPS and Image processing.

Dr. G. Sasibhushana RaoHe is a Professor in the Department of Electronics and Communication Engineering, Andhra University College of Engineering, Visakhapatnam, India. He has over 25 years of experience in varied fields like R & D, industry and academics. He has published and presented more than 185 research papers in various reputed international/national Journals and conferences respectively. His awards include the Best Researcher Award and the Dr Survepalli Radhakrishnan Award for Best Academician.

Dr Rao is a Senior Member of IEEE, Fellow of IETE, Member of IEEE Communication Society, IGU and International Global Navigation Satellite System (IGNSS), Australia. He was also the Indian member in the International Civil Aviation organization (ICAO), Canada Working Group for developing the SARPPS. He has also been involved in the development of the GAGAN system which has been jointly developed by the ISRO and AAI. His current research areas include GPS/INS signal processing, Mobile Communication, SONAR and Acoustic Signal Modelling.

Dr. M. Madhavi LathaShe is a Professor in the Department of Electronics and Communication Engineering, Jawaharlal Nehru Technological University, Hyderabad, India. She has over 20 years of teaching experience. She has published and presented more than 35 research papers in various reputed international/national Journals and conferences respectively. She is a Fellow of IETE, Life Member of ISTE. Her research areas include Image processing, Signal processing and VLSI.


USAGE OF EXTREME PROGRAMMING PROCESS METHODOLOGY WHEN IMPLEMENTED ON THE GLOBAL PROJECTS

N.GANESH1, S.THANGASAMY2

1 Research Scholar, Department of Computer Science and Engineering,Anna University, Coimbatore – 641 006, India

2 Dean, Research and Development,Kumaraguru College of Technology, Coimbatore – 641 006, India

[email protected]

ABSTRACT

Extreme Programming (XP) is a software development process as well as a methodology. XP is also a process framework because it can be tailored to the specific needs of teams, projects, companies etc. XP has been applied to business problems only, e.g. projects with a external customer that wants a specific product. The projects usually ranged from 6 to 15 months. XP was used by small teams ranging from two to twelve members. This article presents the artifacts faced in using the extreme programming on the global projects.

KeywordsExtreme Programming, XP, global software projects, pair programming, XP practices

1. INTRODUCTIONDelivering a system that satisfies customer requirements and which is on time and within budget with few defects is the ultimate goal of any software development activity. The principle behind the Agile Methods include specific strategies for satisfying the customer through involving the customer regularly, relying on face-to-face communication, responding to evolving requirements and providing early and regular feedback. When the customers are not satisfied there is a gap between customer expectation and experiences. In XP, there is an explicitly-defined role for the customer in development team so customer can work with developer closely thereby the customer is made to get satisfied with the activities performed. The essence of XP truly is simple. Be together with your customer and fellow programmers, and talk to each other. Use simple design and programming practices, and simple methods of planning, tracking, and reporting. Test your program and your practices, using feedback to steer the project. Working together this way gives the team courage [6].” [11] discusses the relationship of customers with the Organizations in a global software projects.XP uses rapid iterative planning and development cycles in order to force trade-offs and deliver the highest value features as early as possible.The systemic testing that is part of XP ensures high quality via early defect detection and resolution.

XP is one of best method for the Global software projects wherein we have clients in the onsite.

2. USAGE OF XP IN REAL WORLDAgile methodology advocates often find it difficult to obtain management support for implementing changes in application development. These methodologies require developers, managers and users alike to change the way they work and think. For example, the XP practices of pair programming, test-first design, and the continuous integration. A few studies have considered the social factors of the development team, especially with regard to pair programming [4,5] Furthermore, these methodologies tend to be developer-centric and seem to dismiss the role of management in ensuring success. The project managers also need a set of simple guiding practices that provide a framework within which to manage, rather than a set of rigid instructions. XP teams create and monitor their own iteration plans in collaboration with the customers. The customer creates stories (features) and prioritizes them based on business value. The developers divide up the tasks themselves as they work and measures progress for each iteration (time-boxed development cycle), adjusting plans with the customer as necessary.XP teams have no experts – all developers work on all aspects. In reality, sometimes experts are needed when the team is learning some new tools or a specific component


requires technology with which the organization has no experience. [9] discusses the rational of practicing XP using an acknowledged scientific framework


designed to explain how knowledge is created when several communities are present.3.PROMOTING OPEN INFORMATIONIn order to promote open information a variety of techniques can be followed:

• Place team members within close proximity of each other.

• Make use of whiteboards, charts, etc to disseminate information

• Bring the project sponsor / customer to the project room for public status reports and hands-on demos.

• Use a team to upload the contents on a web page and share the information.

• Establish daily status meetings to promote the flow and exchange of information.

• Sustain open information exchange between business domain experts and the development team.

Communication and customer participation are two parts of the methodology, hence poor participation and bad communication can decrease customer satisfaction.XP also focus important issues such as feedback, simplicity and courage. Since XP embrace late changes, the documentation is minimal. It does not place great emphasis on producing documents. It rather prefers to deliver tested software, but it still needs detailed requirements [12]. Short iterations and release makes it possible to deliver quickly and the customer can see a return on the investments they have done. The iterations can be two to three weeks long and leads to a release every three or four month.4. USER STORIESThe user stories (features) are written by the customers as things that they think should be included in the system. Each one of the stories must be business-oriented, testable and estimable [2]. The user stories build up a foundation for the release planning and acceptance tests. User stories serve the same purpose as the use cases. The stories are formulated and usually written on index cards.5. XP PRACTICESThe main characteristics of XP can be summarized as communication and coordination, customer participation, continues integration and testing, limited documentation, pair programming, collective owner ship of the code. Beck [2] introduces the XP practices as follows:Planning game: Customers decide the scope and timing of releases based on estimates provided by programmers. Programmers implement only the

functionality demanded by the stories in this iteration.Small releases: The system is put into production in a few months, before solving the whole problem. New releases are made often, anywhere from daily to monthly.Metaphor: The shape of the system is defined by a metaphor or set of metaphors shared between the customer and programmers. Simple design: At every moment, the design runs all the tests, communicates everything the programmers want to communicate, contains no duplicate code, and has the fewest possible classes and methods. Tests: Programmers write unit tests minute by minute. These tests are collected and they must all run correctly. Customers write functional tests for the stories in iteration. These tests should also all run, although practically speaking, sometimes a business decision must be made comparing the cost of shipping a known defect and the cost of delay.Re-factoring: The design of the system is evolved through transformations of the existing design that keep all the tests running. Pair programming: All production code is written by two people at one screen / keyboard / mouse.Continuous integration: New code is integrated with the current system after no more than a few hours. When integrating, the system is built from scratch and all tests must pass or the changes are discarded.Collective ownership: Every programmer improves any code anywhere in the system at any time if they see the opportunity.On-site customer: A customer sits with the team full-time.40-hour weeks: No one can work a second consecutive week of overtime. Even isolated overtime used too frequently is a sign of deeper problems that must be addressed. During the 8 hours team development on one particular day [1], the entire team is located in one room, implementing customer supported stories while practicing the techniques.

Open workspace: The team works in a large room with small cubicles around the periphery. Pair programmers work on computers set up in the center.Standup meeting: Every day all team members (developers, customers etc.) meet at a specified time for a couple of minutes. Everybody briefly describes what he or she is working on, how it is going, interesting stuff that he or she found out, and any problems.


6. XP ROLESXP defines a separation between the roles for the different tasks. The roles are divided into

programmer, customer, tester, tracker, coach, consultant and the manager.Programmers: Programmers write test and keep the program code as simple and definite as possible. The first issue making XP successful is to communicate and coordinate with other programmers and team members.Customer: The customer writes the stories and functional tests, and decides when each requirement is satisfied. The customer sets the implementation priority for the requirements.Tester: Testers help the customer writes functional. The run functional tests regularly, broadcast test results and maintain testing tool.Tracker: Tracker gives feedback in XP. He traces the estimates made by the team (e.g. effort estimate) and gives feedback on how accurate they are in order to improve future estimations. He also traces the progress of each iteration and evaluates whether the goal is reachable within the given resources and time constraints or if any changes are needed in the process.Coach: Coach is the person responsible for the process as a whole. A sound understanding of XP is important in this role enabling the coach to guide the other team members in following the process.Consultant: Consultant is an external member possessing the specific technical knowledge needed. The consultant guides the team in solving their specific problems.Manager: Manager makes the decisions. In order to be able to do this, he communicates with the project team to determine the current situation, and to distinguish any difficulties or deficiencies in the process.The roles clearly define the tasks and responsibilities within them and through this way the task misunderstandings can increase. The necessary documentation (except the user stories) is done at the last phase of the development cycle.

7. BENEFITS OF COMMUNICATION AND ITS USAGE IN THE GLOBAL SOFTWARE PROJECTSCommunication is important throughout the entire software development lifecycle. There are several kinds of communication we needed:

• Communication between project manager and customers;

• Communication between developers and customers;

• Communication between developers and project manager;

• Communication between developers;

• Communication between customers.

Communication is also one of the core values of XP discipline. When it comes to the globally distributed software development [12], some of the benefits become hard to achieve due to the kinds of reasons stated earlier. Key elements in the project facilitate the deployment of XP. Those key elements include project information, project site information, project team member information, user story information, project release plan information, project iteration information, and project events information. The goal of managing this information is to provide every stakeholder a clear vision of the project, in such a way to decrease the communication difficulty.

8.BARRIERSINPAIR PROGRAMMINGProgrammers tend to think that the first pair programming session has to be perfect in terms of productivity [8]. The programmers need to accept instruction and suggestion from their partners and consider them as a constructive feedback and not as a personal criticism. Both partners have to waive part of their freedom because they have to synchronize their work hours.

9. POSITIVE AND NEGATIVE ASPECTS OF XPThe XP practices are exclusively intended for use with small, co-located teams. The XP’s practices have both positive and negative aspects. They sound like rules – “Do this, Do not do that”. Creating and maintaining highly functional collaborative environment challenges management far beyond in making up task lists and checking off their completion [2].

10. THE RISK OF CHANGEXP involves the customer in the development cycle more than many other structured processes. Other structured processes typically involve the customer primarily during the early and late phases of the development process, specifically requirements elicitation and analysis, budget and contract negotiation, and acceptance testing [3].


XP provides techniques for monitoring how changes are affecting the code at large, but XP does not offer specific methods for estimating the up-front cost of making a change. Without sufficient risk analysis, making changes to the requirements is another instance of the customer being forced to make decisions in the absence of valuable information.Moreover, there is business value in accountability, auditing, and assessment of risky decisions. Combined with the perception that requirement changes necessarily negatively affect cost and quality, customers may not be maximizing the business value of a process that embraces changes.

11. CONCLUSIONXP practices of on-site customers, planning game, small release, simple design, testing and collective ownership as the ones that can be customized to alleviate communication problems between the onsite and off-shore developers. The major problem found on XP is that it does not provide support for project management, management approaches are needed when practicing the same. If the communication gap and the timings offered due to the geographically located customers and developers are resolved with the proper management support, then the usage of XP on the global projects will become a grand success.

12. REFERENCES[1]. Görel Hedin, Lars Bendix and Boris Magnusson, “Introducing Software Engineering by means of Extreme Programming”, in proceedings of the 25th

International Conference on Software Engineering, Portland, Oregon, May 3-10, 2003

[2]. Beck, kent “Extreme Programming explained: Embrace change”, pp: 70 – 77, Addison Wesley, 1999

[3]. Brown, S. W. and Swartz, T. A. “A Gap Analysis of Professional Service Quality.” Journal of Marketing, Pg: 92-98, April 1989

[4]. Muller, M. M. and Padberg, F. “An Empirical Study about the Feelgood Factor in Pair Programming”. In the 10th International Symposium on Software Metrics, 151-158, September 11-17, 2004, IEEE 2004

[5]. Cao, L. and Xu, P. “Activity Patterns of Pair Programming”. In Proceedings of the 38th Annual

Hawaii International Conference on System Sciences, Jan. 3-6, 2005, IEEE, 2005.

[6]. Jeffries, R. et al., “Extreme Programming Installed”, Addison Wesley Longman, Pg: 172, 2001

[7]. Kent Beck, “Embracing change with extreme programming”, Computer Journal, Vol. 32, Iss. 10, pp: 70-77, IEEE, dated: 1999.

[8]. “www.pairprogramming.com”, last accessed on 25-02-2010

[9]. Kahkonen, T.; Abrahamsson, P., “Digging into the fundamentals of extreme programming building the theoretical base for agile methods”, Proceedings of 29th Euromicro Conference, pg: 273 – 280, 1-6 Sept. 2003

[10]. Lowell Lindstrom, Ron Jeffries, “Extreme Programming and Agile Software Development Methodologies”, Vol. 21, Issue. 3, pg: 41-52, Information Systems Management, 2004

[11]. Damian, D., “Stakeholders in Global Requirements Engineering: Lessons learned from practice”. IEEE Software, vol.24, pg: 21-27, March /April 2007

[12].http://www.it.uu.se/edu/course/homepage/acsd/ht03/Fowler.pdf, last accessed on 24–02-201013. ABOUT THE AUTHORSN.GANESH1 is a research scholar in the faculty of Computer Science and Engineering, Anna University, Coimbatore, India. He has obtained his B.E.[CSE] from Bangalore University, M.E.[CSE] from Anna University Chennai, M.B.A.[HR] from IGNOU New Delhi. He has to his credit a rich experience of Teaching and Training from both Educational institutions and from the Corporate World.

S.THANGASAMY2 is currently working as Professor and Dean, in the Department of Computer Science and Engineering, Kumaraguru College of Technology, Coimbatore, India. He holds his Engineering Degree in the field of Electrical Engineering from A.C. College of Technology, Karaikudi, and then holds the Ph.D Degree in Control and Computers from Indian Institute of Technology, Bombay. He has a rich professional experience on varied fields for more than 30 years.



GEO POSITION BASED ROUTING FOR MOBILE AD - HOC NETWORKS (MANET’S) IMPLEMENTATION THROUGH

EMBEDDED SYSTEMS

1. RAJARAMFaculty of Electronics and Communication Engineering Park College

Of Engineering and [email protected]. MUTHULAKSHMI

Faculty of Electronics and Communication Engineering Park CollegeOf Engineering and Technology [email protected]

3. SHANTHIFaculty of Electronics and Communication Engineering Park College of Engineering

and [email protected]. Dr. V.SUMATHY

Assistant Professor, Dept of Electronics and Communication,Government of Technology Coimbatore

ABSTRACT Using location information to help routing is often proposed as a means to achieve scalability in large mobile ad-hoc networks. However, location based routing is difficult when there are holes in the network topology and nodes are mobile. Terminode routing, presented here, addresses these issues. It uses a combination of location based routing(Terminode Remote Routing, TRR), used when the destination is far, and link state routing (Terminode Local Routing, TLR), used when the destination is close. TRR uses anchored paths, a list of geographic points (not nodes) used as loose source routing information. Anchored paths are discovered and managed by sources, using one of Geographical Map-based Path Discovery. Our simulation results show that terminode routing performs well in networks of various sizes. In smaller networks, the performance is comparable to MANET routing protocols In this system implementatiom through real time systems with Speclized CAN (Control Area Network.).

key words: Ad hoc network, scalable routing, location-based routing method, RTOS, CAN

I. INTRODUCTION

A mobile ad-hoc network (MANET) consists of mobile computing entities such as laptop and palmtop computers which communicate with each other through wireless links and without relying on a static infrastructure such as a base station or access point. Without centralized administration, a MANET is highly unpredictable due to its unstable links and resource-poor as most of the nodes have limited battery power. Due to these physical limitations, nodes require the cooperation of other nodes to successfully send a message to a destination. Figure 1.1 shows the mobile Ad hoc network.

Figure 1 Mobile Ad hoc Networks

There are numerous scenarios that do not have an available network infrastructure and could benefit from the creation of an ad hoc network: a) Rescue/Emergency operations: Rapid installation of a communication infrastructure during a natural/environmental disaster (or a disaster due to terrorism) that demolished the previous communication infrastructure. Law enforcement activities: Rapid installation of a communication infrastructure during special operations. Tactical missions: Rapid installation of a communication infrastructure in a hostile and/or unknown territory.


1.2.CHARACTERISTICS OF MOBILE AD HOC NETWORKSBecause of their very nature, mobile ad hoc networks have a certain number of peculiarities like the following: They can be independent of any provider. Dynamic topologies: Due to the presence of mobile nodes, the network topologies varies from time to time. Bandwidthconstrained,variablecapacitylinks:Wireless links have significantly lower capacity than their hardwired counterparts. Moreover, the realized throughput of wireless communications is often much less than a radio’s maximumtransmissionrate. 1.3CLASSIFICATIONOFAD-HOCPROTOCOLSRouting protocols can be classified into different categories depending on their properties:i. Centralized vs. Distributed

ii. Static vs. Adaptive

iii.Reactive vs. Proactive

In centralized algorithms, all route choices are made at central node, while in distributed algorithms, the computation of routers is shared among the network nodes.

1.4 MANET ROUTING PROTOCOLS

The routing protocols in ad-hoc networks are categorized in to two groups: Proactive (Table-driven) and Reactive (On-demand) routing.

1.4.1PROACTIVE(TABLE-DRIVEN)ROUTING PROTOCOLS

These routing protocols are similar to and some as a natural extension of those for the wired networks. In proactive routing , each node has a one or more tables that contain the latest information of the routes to any node in the network. Each rows ha the next hop for reaching a node/subnet and the cost of this route. Various table–driven protocols differ in the way the information about a change in topology is propagated through all nodes in the network.

1.4.2 REACTIVE (ON-DEMAND) ROUTING PROTOCOLS

These protocols take lazy approach to routing. They do not maintain or constantly update their route tables with the latest route topology. Reactive routing is also known as on-demand routing. These Examples of reactive routing protocols are the dynamic Source

Routing (DSR), Ad hoc on-demand distance vector routing (AODV) and temporally ordered routing algorithm (TORA).

1.5 CHALLENGES IN AD-HOC NETWORKS

The following are the current challenges in the ad-hoc wireless networks i.Multicasting, ii.Qos support iii.Power aware routing iv.Location aided routing.

2. GEOGRAPHICAL MAP PATH DISCOVERY

The basic idea behind GMPD is that, mapping information of the network density is known to all nodes in the network. Areas with high nodedensityarecalled “towns”. The Source node determines the town area in which it is actually situated; also determines the town area of the destination node. The Source node refers the network map in order to find anchor path from source to destination. The shortest path between source and destination is chosen for routing the packets. For

GMPD [7] each town, a map gives the location of its center and the size of the square area. A map of the network can be presented as a graph with nodes corresponding to towns and edges corresponding to highways. Macroscopically, the graph of towns does not change frequently.

FIGURE 2 GMPD with a given map of towns works as followsFor Figure 2 Source S determines from its own location LDAS the town area ST (Source town) in which S is situated (or, the nearest town to LDAS if it


is not in the town area). In addition, since S knows the location of destination D (LDAD), it can determine from the LDAD the town area DT (Destination town), where D is situated (or, the nearest town to LDAD if it is not in the town area). Then, S accesses the network map in order to find the anchored path from S to D. An anchored path is the list of the geographical points: The points correspond to centers of the towns that the packet has to visit from ST in order to reach DT. One possible realization of the map lookup operation.


1. LOCATION BASED-ROUTING

In ad hoc network location information to help routing is often routing (Terminode Remote Routing, TRR), used when the destination is far, and link state routing (Terminode Local Routing (TLR), proposed as a means to achieve scalability in large mobile ad hoc networks. In order to enhance the life time of the network a location based routing protocol which uses anchored paths a list of geographic points is used as loose source routing information .It is a high probability that there are nodes to ensure connectivity from one town to another source and destination of the location have sufficient of its centre and size of the square area.

3.1 TERMINODE ROUTING First, it combines a location-based routing method [6] with a link state-based mechanism. Second, it uses a special form of restricted search mode (Restricted Local Flooding, RLF). These first two ingredients solve problems due to the inaccuracy of location information, in particular for control packets. Third, it introduces the concept of anchors, which are geographical points imagined bysources for routing to specific destinations. This helps efficiently route around connectivity holes. In order for the comparison to be fair to MANET protocols, we implemented an ad hoc location management scheme. In smaller ad hoc networks we compared Terminode routing to some existing MANET like routing protocols (AODV and LAR1) and found similar performance. In larger mobile ad hoc networks of 500 nodes, MANET-like routing protocols do not perform well (except when mobility is small), while our routing protocol still performs well. In networks that are regularly populated with nodes, Terminode routing performs comparable to GPSR when the location management accuracy is high; however, terminode routing performs better when the location information accuracy is low. We also consider irregular networks with holes in node distribution. Here, too, we find that terminode routing outperforms GPSR. An existing MANET and location-based routing protocols we compared it to. Non uniform topologies are likely to appear in metropolitan areas with mountains or lakes, like the Lake of Geneva area.

4.PERFORMANCE EVALUATION OF GEOGRAPHICALTERMINODEROUTINGIn the NS 2 settings, The IEEE 802.11 Medium Access Control (MAC) protocol is used. And the radio range is 250 meters. The channel capacity is 2Mb/s. The propagation model is two-ray. It uses free space path loss for near sight and plane earth path loss for far sight.4.1 PROTOCOL CONSTANTS

We used the following configuration for the Helloing protocol. The HELLO timer is 1 second. Each entry in the routing table expires after two seconds, if it is not updated. All nodes promiscuously listen to all HELLO messages within their radio range. Nodes that have data or control packets to send should defer sending HELLO messages (up to the timer value) and piggyback the HELLO message to the data or control packet. 4.2. EMBEDDED APPROACH AND METHODS SOFTWARE:In order for the hardware to function, the firmware code for the system has to be written. Theoretically, software that resides in the non-volatile memory and handles the operationas well as function of a system is known as firmware. The firmware holds the information that the Microcontroller needs to operate or run. Thus, it needs to be free of bugs and errors for a successful application or product There are various types of software that could be used to program a PIC micro. Program can be written in a variety of languages such as C shows the standard 11 bit identifier format Basic, Pascal or even Assembler. In this paper, the program would be written in C language to generate the required firmware for the system. The program code is written in C language using CCS C compiler (by Custom Computer Services) to generate the hex file This hex file is then downloaded into the microcontroller for it to function as programmed. There are several advantages of using CCS as the compiler because functions related to CAN system are already available and ready to be used. This reduces the time to write the program code as well as ease the development process. The CAN protocol supports CAN base frame” supports a length of 11 bits for the identifier o message frame formats.


Fig. 4 Security system block diagram

Figure 4 shows the block diagram of several subsystems security system. The same pattern of block diagram discussed earlier in figure 1 is implemented here in figure 3 except that this one only shows more general overview of the hardware The overall security system is based on the integration. However, a 7AH battery is recommended to power the overall security system in case of power failure. A fully charged battery can last up to approximately 13 hours at minimum power consumption and Smart card systems are used to detect which is node is reset and send beacon message to Concern system.

5.NETWORK SIMULATOR 2

The simulator used and the experimental results are explained in this chapter.

5.1 NS-2TheNSnetwork simulator[1] (www.isi.edu/nsnam/ns), from U.C. Berkeley/LBNL, is a object oriented discrete event simulator targeted at networking research and available as public domain. Its first version (NS-1) began in 1989 as a variant of the REALnetworksimulator(www.cs.cornell.edu/skeshav/real/overview.html), and was developed by the Network Research Group at the Lawrence Berkeley National Laboratory (LBNL), USA.

ItsdevelopmentwasthenpartoftheVINTproject(www.isi.edu/nsnam/vint/index.html), supported by DARPA, at LBNL, Xerox PARC, and UCB, under which NS version 2.0 (NS-2) was released, evolving substantially from the first version. NS-2 is widely used in the networking research community and has found large acceptance as a tool to experiment new ideas, protocols and distributed algorithms. Currently NS-2 development is still supported through DARPA. NS has always included substantial contributions from other researchers, including wireless code for both mobile ad hoc networks and wireless LANs from the UCB Daedelus and CMU Monarch projects and Sun Microsystems. At the time being, NS-2 is well-suited for packets switched networks and wireless networks (ad hoc, local and satellite), and is used mostly for small scale simulations of queuing and routing algorithms, transport protocols, congestion control, and some multicast related work.It provides substantial support for simulation of TCP, routing, and multicast protocols over wired and wireless networks.NS-2 is suitable not only for simulation but also for emulation, that is, it is possible to introduce the simulator into a live network. Special objects within the simulator are capable of introducing live traffic into the simulator and injecting traffic from the simulator into the live network. NS-2 plays an important role in the research community of mobile ad hoc networks, being a sort of reference simulator. NS-2 is the most used simulator for studies on mobile ad hoc networks, and it comes with a rich suite of algorithms and models. In this perspective, NS-2 would be the natural candidate to be used in BISON too. Unfortunately, its software architecture is such that adding new components and/or modifying existing ones is not a straightforward process. That is, in terms of ease to implement/test new algorithms or scenarios, NS-2 scores poorly with respect to other candidates. Moreover, NS-2 does not scale well in terms of number of nodes and it is reported to be in general quite slow from a computational point of view. Implementation and simulation under NS-2 consists of 4 steps

i.Implementing the protocol by adding a combination of c++ and Otcl code to NS-2,’ source base;ii.Describing the simulation in an Otcl script;iii.Running the simulationiv. Analyzing the generated trace files


6.EXPERIMENTAL RESULTS6.1 SIMULATION ENVIRONMENTAn event driven simulator ns-2 was used for simulations. The simulation setup 10 nodes and Radio range of each node is assumed to be 250m. Reliable connections are established at random n the network. The connections with constant bit rate . Two ray ground propagation model was used. The size of the data payload was 512 bytes.

6.1.1 CUMULATIVE SUM OF NUMBERS OF RECEIVED PACKETS In Figure 6.1 shows the Cumulative sum of Numbers of Received Packets vs received event time (s) of Geographical location aided routing

Figure 6.1 Cumulative sum of Numbers of Received Packets vs receive event time(s)

6.1.2 PACKET DELIVERY FRACTIONIt is the ratio between the number of data packets delivered to that of the number of packets supposed to be received by it. This metric gives a measure about the packet loss. A high packet delivery ratio implies an efficient protocol. The figure 6a shows the packet delivery fraction vs simulation time of AODV and Geographical location aided routing.Packet Delivery Ratio (in%)=Actual packets received X % Packets supposed to be receivedOn an average, the packet delivery ratio improved by 5.08% than AODV.

PACKET DELIVERY FRACTION

78

80

82

84

86

88

90

92

94

96

0 100 200 300 400 500 600 700 800 900

SIMULATION TIME(S)

PA

CK

ET

DE

LIV

ER

Y F

RA

CT

ION

A0DV

GEOGRAPHICAL LOCATIONAIDED ROUTING

Figure6 Packet Delivery fraction vs Simulation time(s)

6.1.3 THROUGHPUT

Throughput are mainly of network parameters in the throughput receiving packets vs simulation time in that manner slightly increase in simulation of throughput receiving packets in simulation time of 300 sec. It is observed that throughput is better than AODV by 37.5%

6.1.4 IMAGE OF NUMBER OF RECEIVED BYTESIn figure 6.8 shows the Number of received bytes at all the nodes x-receive node y- send node and proposes low overhead methods for computing anchors.

Figure shows the Number of received bytes at all the nodes x-receive node y- send node

7.CONCLUSION

Terminode routing aims to support Geographical map path discovery (GMPD) location-based routing on irregular topologies with mobile nodes. It achieves its goal by combining a location-based routing method with a link state-based mechanism. Further, it introduces the concept of anchors, which are geographical points imagined by sources for routing to specific destinations, and proposes low overhead methods for computing anchors.


Last, a special form of restricted search mode (Restricted Local Flooding, RLF), solves problems due to the inaccuracy of location information, in particular for control packets The performance analysis shows that, in large mobile ad hoc networks, terminode routing performs better than MANET-like, or existing location-based routing protocols.It does so by maintaining its routing overhead low and by efficiently solving location inaccuracies GMPD(Geographical Map Path Discovery) It is a high probability and packet received information’s are good that there are nodes to ensure connectivity from one town to another source and destination of the location have sufficient of its centre and size of the square area. In this system implementatiom through real time systems with Speclized CAN (Control Area Network.).

7.1 FUTURE WORK

In this implementation ,the communication is established between single sender and receiver in future versions ,this can be extended to multiple senders and receivers and the performance levels and using 4g & Global positioning system still can be improved using real time systems with Speclized CAN (Control Area Network.).

8. BIBLIOGRAPHY[1] http://ica1www.epfl.ch/TNRouting, 2004, Simulation Source Code of Terminode Routing (NS-2 ).

[2] S. Basagni, I. Chlamtac, V. Syrotiuk, and B. Woodward, 1998, “A Distance Routing Effect Algorithm for Mobility (DREAM),” Proc. Fourth Ann. ACM/IEEE Int’l Conf. Mobile Computing and Networking (MobiCom ’98),

[3] L. Blazevic, S. Giordano, and J.-Y. Le Boudec, 2002, “Anchored Path Discovery in Terminode Routing,” Proc. Second IFIP-TC6 Networking Conf. (Networking 2005).

[4] L. Blazevic, S. Giordano, and J.-Y. Le Boudec. Anchored Path Discovery in Terminode Routing. In Proceedings of The Second IFIP-TC6 Networking Conference (Networking 2002), May 2002.

[5] L. Blazevic, S. Giordano, and J.-Y. Le Boudec. Self Organized Terminode Routing. Cluster Computing Journal, April 2002.

[6] L. Blazevic and J.-Y. Le Boudec. Terminode Routing: Scalable Routing for Large Mobile Ad-Hoc Networks. In Proceedings of Communication networks and distributed systems modeling and simulation conference (CNDS), January 2002.

[7] Ljubica Blazevic. Scalable Routing Protocols with Applications to Mobility, PhD Thesis. EPFL, PhD Thesis no. 2517, 2002.

[8] P. Bose, P. Morin, I. Stojmenovic, and J. Urrutia. Routing with guaranteed delivery in ad hoc wireless n/w.newtorks. Descrete Algorithms and methods for mobile computing communications (DIAL M), August 1999.

[9] J. Broch, D.A. Maltz, D.B. Johnson, Y.C Hu, and J. Jetcheva. A performance comparison of multi-hop wirelessad hoc network routing protocols. Proceedings of the Fourth Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom’98),Dallas, Texas, USA, August 1998.

[10] Srdjan Capkun, Maher Hamdi, and Jean-Pierre Hubaux. GPS-free positioning in mobile Ad-Hoc networks. Proceedings of the 34thHICSS, January 2001.

[11] M. Gerla G. Pei and X. Hong. Lanmar: Landmark routing for large scale wireless ad hoc networks with group mobility.


DESIGN OF SMITH-PREDICTORS FOR TIME-DELAY PROCESSES WITH AUTO-TUNING TECHNIQUES

DR.G.SARAVANA KUMAR1DR.R.S.D.WAHIDABANU2 ANDMR.K.G.ARUN RAAJESH3

1Dept. Of Electronics & Instrumentation Engineering, Birla Institute of Technology and Science, Pilani – Dubai, P.O.Box 345055, Dubai International Academic City, Dubai, UAE [email protected]

2Dept. Of Computer science and Engineering, Government College of Engineering, Salem, India [email protected]

3B.E.(HONS) Electronics and Instrumentation Engineering, Birla Institute of Technology and Science, Pilani – Dubai, P.O.Box 345055, Dubai International Academic City, Dubai, UAE [email protected]

ABSTRACT:

All the dynamic components of the process control loop may exhibit significant time delays in their response. A Modification of the classical feedback control system proposed by O.J.M Smith for the compensation of Dead-Time effects is studied. In this paper the system’s stability criteria is analyzed with relation to system gain and Dead-time for Two-Degree-Freedom Self-Adaptive controller.

Keywords: Modified Smith-Predictor, Dead-Time, Self-Adaptive Controller, Two-Degree-Freedom

I. INTRODUCTION

Smith-Predictor (SP) proposed by Smith [2] is a very effective Dead-Time Compensator for a stable system and with definite delay. Matausek [3] added another controller to the feedback loop which proved to show bad results for uncertain system with integral action. Wang and Liu [1] proposed a delay feedback automatic tuning controller for stable and integrating process. They are only few papers describing about the robustness of the controller for uncertain system with variable delay for set-point response. Hence in this paper the robustness of the new modified Smith-Predictor proposed by Wand and Liu is verified for uncertain, unstable system with variable time-delay.

Whenever an input variable of a system changes, there is a time interval (short or long) during which no effect is observed on the outputs of the system. This time interval is called dead time or transportation lag or pure delay or distance-velocity lag..

A time lag (time delay or dead time) limiting the permissible process gain (PG) reduces the ability to control the process. So, a controller mechanism is necessary to reduce this limitation. This mechanism is called the ‘Dead-time compensator’. Smith Predictor is extensively used to reduce Dead-time. Smith Predictor gives a new controlled variable that is the reaction of the process variable to its controller output without Dead-time. It needs three parameters such as process gain, Dead- time and the time constant Smith Predictor uses these parameters to construct models of the process from the controller output. The maximum permissible controller gain is inversely proportional to Dead-time. The controller gain can be hypothetically increased without any limit provided ignoring justifying circumstances such as loop interaction, measurement noise, resolution and the final element dead band.

II. SMITH PREDICTORSmith Predictor proposed by Smith is shown in Fig.1


mailto:[email protected]



Gc Go e-τo Y

e-τmGm

+-

+-

+-

r

Inner-loop

External -loop

Y1

Fig.1 Smith PredictorSmith-Predictor structure which has the transfer

function as,

( ) ( )MSM

OSOCMC

OSOC

R eGeGGGGeGGsY ττ

τ

−−

−

−++=

1

The system’s characteristic equation can be

expressed as follows,

0)(1)( =−++= −− SM

SOCMC

MO eGeGGGGsL ττ

We assume MG = OG( )

( )( )SS

MC

SS

MC

MC

SS

SSMC

OM

OM

MO

MO

eeGG

eeGG

GGee

eeGG

ττ

ττ

ττ

ττ

−−−

−−

−−

−−

−=+

−=

+

−=−+

=+−+

1)(1

11

11

011

mnasasasabsbsbsbG n

nn

n

mm

mm

O >=++++++++= −

−

−− 0

......

01

11

1

01

11

1

So we can design )(sG c and obtain,

0))()...((

...11

)()(1

1,...,1121

01

11

1

1

⟩−−−

+++∗+=

+

++

++

−

nn

mm

P

MC

pppspsps

bsbsbK

sGG

Thesystem characteristic equation can be divided into two parts:

)()(1)(

0)()()(1

1

21

sGGsLsLsLsL

MC−+=

=−=

=)(2 sLTherefore,

)()( 21 sLsL =Then the dual-locus diagram is obtained when s traverses the Nyquist contour. The argument of

)()( 21 sLsL − is the angle between the vector

joining the corresponding points on loci )(1 sL and

)(1 sL and the real axis. Therefore the following is derived.

))()((11)(

)()))()...((

...(11)(

1

121

01

11

11

ωωω

ωω

YjXK

jL

jpspsps

bsbsbK

jL

P

n

mm

P

∗+∗+=

−−−+++

∗+=+

++

=)(2 ωjL

Due to 2)(lim,0)(lim,)(lim,1)(lim 220110

≤=∞==∞→→∞→→

ωωωωωωωω

jLjLjLjL

and )(2 ωjL


being a periodic function of ω and symmetric to real axis, it is obvious that the Nyquist curves have numerous intersection between and . Let ω1 and ω2 be the frequency value, respectively, at the intersection point.

The system will be unstable, if the point on curve crosses the Nyquist line before the point on the Nyquist line reaches the intersection. Then the

system will be stable if ω1 < ω2. When for ω1, ω2 ϵ R, a proposition is achieved.So


III.NEW MODIFIED SMITH-PREDCITOR

A New Modified Smith-Predictor proposed by Wang and Liu is shown in Fig.2

Gc Go e-to Y

e-tmGm +-

r

Y1

+-

+-

Gc1

E1

Fig.2 New Modified Smith-Predictor

The symbol r denotes the set-point input. GC is employed to take care of the set-point servo-tracking. GC1 makes the system output error. If GO = GM together with τO = τM is satisfied, and GC1 works nothing to be zero and the standard Smith-Predictor structure is obtained. The Output error feedback is

))(1())(1(

1

1S

MC

SOC

M

M

eGGeGG

τ

τ

−

−

++

The

closed-loop response to set-point input is given as,

( )

))(1())(1(1

))(1())(1())(1(

1

1

11

1

SMC

SOCMC

SOC

SOCMC

SMC

SMC

SOC

r

M

M

O

MM

MO

eGGeGGGG

eGGeGGGGeGG

eGGeGGsY

τ

τ

τ

ττ

ττ

−

−

−

−−

−−

+++

=

++++

=

If SM

SO

MO eGeG ττ −− = , then (11) becomes,


( )

( )

( )3

22

cos2

sin2)(1

12

sin2

sin2)(11

21

221

221

ωω

ωττωττω

ωττωττω

⟨

+−=∗

+−=∗+

MOMO

P

MOMO

P

YK

XK

( ) ( )OC

SOC

r GGeGGsY

O

+=−

1τ

IV.CONTROL SCHEME FOR NEW MODIFIED SMITH-PREDCITOR

There are two controllers GC and GC1 in our MSP structure to be designed. The controller GC1 has the self-adaptive mechanism, which can make the simulated model output change adaptive to real system output. Assume E1 and Y1 denote the main controller output and simulated output. Therefore

( )( )

SMS

MC

SM

SO

CO

M

MO

eGeGG

eGeGGsEsY τ

τ

ττ−

−

−−

+

−∗+=

11

1

1

11

Then apply steady-state final value theorem,If S

MS

OMO eGeG ττ −− = then,

( )( )

SOSS

OeGsEsY τ−

→→=

01

1

0limlim

Then the characteristic equation is stable. Now we can achieve the design of the controller GC1. It is designed, for simplicity, to be P type for the plant assumed as an integrator,tracking this form ( ) 11 PC KsG = and to be a PI type for the plant assumed as a form

( )

+= sTKsGC

PC1

1111

V.CONTROLLER TUNING FOR NEW MODIFIED SMITH-PREDCITOR

A. For the plant assumed as an integrator with dead-time:The model assumed as sKG MM = and

11, CCCC KGKG == respectively. The delay free part of the above equation becomes,

( )1

1

111

+∗=

+∗

=

ssKK

sY

MC

λ

where λ

=MC KK

1λ is the closed-loop

parameter. So the parameter MC KK λ1= can be obtained by a suitable choice of λ. Based on the analysis above, we obtain,

)(1

011

1

11

ωωτ

ττ

jwejKK

esKKeGG

M

MM

MC

MCSMC

=

=−

=

+=+

−∗

−∗

−

Where MM KK =∗ . The thre threshold value of KC1

can be obtained easily by dual-diagram Nyquist curve as follow:

*1 2 MMC KK τ

Π⟨

Introducing the parameter φPM, the phase margin of the closed-loop system with the loop transfer function, we obtain:

( )*12 MCMPM KKarctg τϕ −Π=

The choice of φPM = 600 gives the system satisfactory performance, correspondingly

)(577.0

*1MM

C KK τ=

B. For the plant assumed as first-order stable plant with dead-time:

Consider )1( += sTKG

M

MM with

sTsTKG

C

CCC

)1( += and

sTsTKG

C

CCC

1

11

)1( += .Correspondingly

MMC KTK λ= and choice of φPM = 600 gives

)(577.0

*1MM

C KK τ=

VI.I-PD CONTROLLER

I-PD Controller proposed by Saravanakumar &

Wahidhabanu

Consider the plant with transfer function

(6.1)

Where represents the time delay of L seconds.

Assume that the time delay L is known and process

is

Here


Lse−

1( ) LspG s e

s−=

0

( )( )

1pG s

G ssL

=+

1s

and L (process) = L (model). The modification in

Figure 3.1 consists of the additional feedback path (

0k ≠0) from the difference of the plant output Y and

the model output my to the control input u. When 0k

= 0 the Smith predictor is used. When 0k ≠0, then

modified Smith predictor is obtained. The controller

used in this Smith predictor is an I-PD controller

where the integrator is in the forward path and the

proportional and derivative control are in the

feedback acting on the feedback signal. The

controller was tuned by Ziegler –Nichols method and

by relay method, since relay method gives the

immediate value of ultimate gain and Ultimate

period, when compared to ZN method it is also

preferred for auto-tuning.

( ) )()()()()1(1

)()(1

)(1)(

)(sG

sGsGsGesGsGKsG

sG

sH

popdi

Lspdi

op

p

d

+−++

+=

−

Fig.3 Proposed modified smith predictor

The set point and the load disturbance

response is given by

( ) ( ) ( ) ( ) ( )r dY s H s R s H s D s= +

Where

( )

( ) )()()()1(1)()(

1

)()()()1(1)()(

)( 0

sGsGsGesGsG

sGsGsGesGsG

sH

opdiLs

pdi

pdiLs

ip

r

+−++

+

−+=

−

−

(6.

3)

(6.4)

where,

(6.5)

and represents the estimated transfer

function of the plant. From (6.3) and (6.4)

we have

( ) )()()(1)()(

)(0

0

sGsGsGesGsGsH

pdi

Lsi

r ++=

−

(6.6)


Y

-

-

ski

+

R(s)

+++

D(s)

+

++

-

-

+

+

-

d

)(sGopkdsk

Lse −

K0

Lses

−1

ym

dppd sKKsG +=)(

sKsG i

i =)()(0 sG

( )( ) )()()(1

)()()()1(1)(1)(

)(sGsGsG

sGsGsGeKsG

sGsH

opdi

opdiLs

op

pd ++

+−++

=−

Here tuning of the P, I and D values are performed by the relay feedback test. There are two parameters that result from this auto tune test. One is the period (time between successive peaks), P, and the other is the


sKsG i

i =)(

amplitude of the process output, a.

Theperiod has units of time; the ultimate

frequency can be found from

2u p

πω = (6.7)

And the ultimate gain can be found from the

amplitude

4cu

hkaπ

= (6.8)

A.VALIDATION OF TIME CONSTANT

& DEAD TIME RELATIONSHIP

Model of the process transfer function is

sLsG

sG p

+=

1)(

)(0 (6.9)

and L (process) = L (model). That is, process dead

time is equal to time constant of the model. Some

method should be there to validate this assumption.

Table 1 Validation through time domain

specifications

Thus simulation studies shown in the Table 1 are

undergone to validate this assumption. In this

validation the time constant was assumed to be 3 sec

as that of the dead-time. Simulation studies are

carried out for time constant varying from 0.75 to 6

seconds and various responses are obtained in time

domain specifications such as overshoot, rise time,

peak time, delay time and settling time etc.

From the detailed study of these specifications it is

validated that time constant of the model should be

equal to dead-time of the process. For other time

constant a small steady state error is obtained for

constant load disturbance. The table above shows the

various time constant and dead-time relationship with

values of rise time, peak time, delay time and settling

time.

A - Represents the time domain

specifications of the set point response.

B - Represents the time domain


Dead time

Time constant

Rise time A(sec)

Peak overshoot magnitude A

Settling time A(sec)

Peak Undershoot magnitude B

Settling time B (sec)

3 sec 0.75 16 1.1 60 0.5 150

1.5 15.5 1.09 58 0.45 148

2.25 15 1.08 55 0.4 145

3 14 1.05 45 0.3 140

3.75 14 1.05 58 0.44 145

4.5 15 1.055 65 0.47 150

5.25 15 1.06 70 0.52 155

6 16 1.07 80 0.55 190

specifications of the load response.

i) Dead-time is considered to be constant for all

the simulative study.

ii) The time constant is varied from one fourth

of the dead-time to double the dead time.

iii) Rise time is the time taken by the

process to reach the 90% of the final value for the

very first time. Here the observation shows that the

rise time are slow in the initial stages and fast at time

constant of 3 sec and again slows down at double the

time constant.

iv) Peak magnitude represents the overshoot of

the process and is high at the initial and final values

of time constant, but found to be low when time

constant matches the dead time.

v) Settling time is the time taken by the process

to settle with its set point value very soon, which is

also fast when time constant matches dead time ,and

is slow in all other cases.

So this study shows that when time constant matches the dead time, it is possible to achieve a critically damped system.2. SIMULATION RESULTS FOR

VALIDATION OF TIME CONSTANT

AND DEAD-TIME RELATIONSHIP

In simulation shown in Figure 4 for a dead time of 3

sec and time constant of 3 sec, it is noticed that the

error is of only one polarity (i.e., it never oscillates

about the set point). Here the measure of quality are

the duration (starting point of disturbance to return of

the process variable) of the excursion, for a load

change and a maximum error for transient change.

Fig.4 Time constant = 1/4 OF dead time

Transient change --- settling time = 60 sec

Load change --- settling time = 150 sec

The duration is the time for exceeding the allowable

error and to regain the allowable error. The overshoot

magnitude is 1.1 and for the transient change the

process returns to set point at 60 sec. For load change

of .1 unit at 100 sec the process variable returns at

150 sec which is comparatively higher than the 45

and 140 sec of the, time constant equal to dead time


ratio.

Fig.6 Time constant=3/4 of dead time

In simulation shown in Figure 6 when the time

constant is 3/4th of the dead time, the settling time of

the set point response and the load response is 55 sec

and 145 sec respectively, which is comparatively

higher than the 45 and 140 sec of the, time constant

equal to dead time ratio

Fig.7 Time constant=2 times dead time

In simulation shown in Figure 7 for a dead time of 3

sec and a time constant of 6sec (double the dead

time), the settling time for transient change is about

80 sec. For a load change of 1 unit at 100sec, the

settling time is around 190sec, which is

comparatively higher than the 45 and 140 sec of the,

time constant equal to dead time ratio.

Thus the ratio of the time constant and dead time relation has been validated using simulative study.

VII.RESULTS OF VARIATION IN PARAMETERS OF WANG AND LIU SMITH-PREDICTOR

1. Decreasing value of phase margin with constant time-constant and λ

Fig.3 shows the graph for decreasing value of phase margin with constant time-constant and λ

Fig.3 Decreasing value of phase margin with constant time-constant and λ

Inference made from the graph:

For different values the rise-time, settling-time, delay-time are all the same. There is no over-shoot or under-shoot. There is no change in the graph.

2. Increasing value of phase margin with constant time-constant and λ

Fig.4 shows the graph for increasing value of phase margin with constant time-constant and λ


Fig.4 Increasing value of phase margin with constant time-constant and λ Inference made from the graph:


For different values the rise-time, settling-time, delay-time are all the same. There is no over-shoot or under-shoot. There is no change in the graph.

3. Constant value of phase margin with constant time-constant and λ decreasing

Fig.5 shows the graph for constant value of phase margin with constant time-constant and λ decreasing

Fig.5 Constant value of phase margin with constant time-constant and λ decreasing


For different values the settling-time increases when λ decreases. There is no over-shoot in the graph.

4. Constant value of phase margin with constant time-constant and λ increasing

Fig.6 shows the graph for constant value of phase margin with constant time-constant and λ increasing

Fig.6 Constant value of phase margin with constant time-constant and λ increasing


For different values the settling-time decreases when λ increases. There is no under-shoot in the graph.

5. Maximum value of process gain and time-constant and time-delay

Fig.7 shows the graph for maximum value of process gain and time-constant and time-delay

Fig.7 Maximum value of process gain and time-constant and time-delay

The Maximum value of process gain and time-constant and time-delay are 1, 5, and 5 simultaneously.


For different values the settling-time increases for maximum value of process gain and time-constant and time-delay.

VIII. CONCLUSIONDead-time compensator proposed by Smith, Modified Smith-Predictor proposed by Wang and Liu, Modified smith predictor proposed by Saravanakumar and Wahidha banu were discussed and compared in the stability point of view by parameter variations such as dead-time, Time constant and gain of the system for both Integrator with time delay process and first order system with time delay process. Design and controller tuning of Saravanakumar et.al is discussed in the reference paper. Only the effect of parameter variations of time constant and dead time has been shown through simulative study. It is clear that in Wang Smith-Predictor the change in time constant with constant


phase margin produces changes in the Settling time of the process which indicates the over damped phase margin produces changes in the Settling time of the process which indicates the over damped response. Whereas in Saravanakumar Smith-Predictor the change in the time constant w.r.t. Dead-time produces slight change in overshoot and under shoot response by keeping the settling time as the same. By the stability consideration is almost the same in connection with the phase margin. At the same time the tuning methods are different in both the Smith-Predictors. Both the Smith- Predictors are verified for Integrating and time delay processes and also with first order systems with time delay processes. Auto-tuning is performed by Least-Square in Saravanakumar Smith-Predictor and auto-tuning is performed by self –adaptive controller in Wang and Liu process.

IX.ACHNOWLEDGEMENTS

I take this opportunity to thank Prof. Dr. M. Ramachandran, Director BPD who has given us this opportunity to apply our core engineering concepts in a theoretical atmosphere. We offer our regards and blessings to all of those who supported us in any respect during the completion of this work.

X.REFERENCES

• Shihua Wang, Bugong Xu and and

Qingyang Wang and Yun-Hui Liu, Modified

Smith Predictor and Controller for Time-

Delay Process with Uncertainty,

Proceedings of the 6th World Congress on

Intelligent Control and Automation, June

21-23, 2006, Dalian, China

• O.J.Smith, “A controller to overcome dead-

time,” ISA, J., vol.6, no.2, pp 28-33, 1959.

• M.R.Matausek and A.D.Micic, “On the

modified Smith Predictor for controlling a

process with an integrator and long dead-

time,” IEEE Transactions on Automatic

Control, vol.44, no.8, August 1999.

• Paul W.Murril, Process control theory,

Instrument of society of America, North

Carolina, 1991.

• Kevin Warwick, “An Introduction to

Control Systems,” World Scientific

Publishing Company, 2nd Edition, January

1996.

• K.J.Astrom, C.Chang and B.C.Lim, “A New

Smith-Predictor for controlling a process

with an integrator and long dead-time,”

IEEE Transactions on Automatic Control,

vol.39, no.2, February 1996.

• S.Majhi and Derek P.Atherton, “Automatic

tuning of the modified Smith Predictor

controllers,” Proceedings of the 39th IEEE

Conference on Decision and control,

Sydney, Australia, December 2000.

• George Stephanopoulos,

“Chemical Process Control, An

Introduction to Theory and Practice,”

Pearson Education Book Company, 2nd

Edition, 2007.


• Donald R.Coughanowr, Process

Systems Analysis and control, McGraw-

Hill, Chemical Engineering Series, U.S.A.


TEXTUAL ANALYSIS OF STOCK MARKET PREDICTION USING FINANCIAL NEWS ARTICLES

M.SURESH BABU,MCA,M.Phil,PGDBA,,MISTE,MISCA., Principal, Intel Institute of Science, Anantapur, Andhra Pradesh , India. Email :

[email protected]. N.GEETHANJALI

M.Sc,MISTE, Ph.D. Associate Professor, Department of Computer Science, S.K. University, Anantapur. India, Email : [email protected].

V.RATNA KUMARIB.Tech,M.Tech, Associate Professor, Department of CSE, Nigama Engineering College, Karimnagar. Email :

[email protected]

ABSTRACTThis paper examines the role of financial news articles on three different textual representations; Bag of Words, Noun Phrases, and Named Entities and their ability to predict discrete number stock prices twenty minutes after an article release. Using a Support Vector Machine (SVM) derivative, we show that our model had a statistically significant impact on predicting future stock prices compared to linear regression. We further demonstrate that using a Named Entities representation scheme performs better than the de facto standard of Bag of Words.

1 INTRODUCTION

Stock Market prediction has always had a certain appeal for researchers. While numerous scientific attempts have been made, no method has been discovered to accurately predict stock price movement. The difficulty of prediction lies in the complexities of modeling human behavior. Even with a lack of consistent prediction methods, there have been some mild successes.Stock Market research encapsulates two elemental trading philosophies; Fundamental and Technical approaches (Technical-Analysis 2005). In Fundamental analysis, Stock Market price movements are believed to derive from a security’s relative data. Fundamentalists use numeric information such as earnings, ratios, and management effectiveness to determine future forecasts. In Technical analysis, it is believed that market timing is key. Technicians utilize charts and modeling techniques to identify trends in price and volume.These later individuals rely on historical data in order to predict future outcomes.

One area of limited success in Stock Market prediction comes from textual data. Information from quarterly reports or breaking news stories can dramatically affect the share price of a security. Most existing literature on financial text mining relies on identifying a predefined set of keywords and machine learning techniques. These methods typically assign weights to keywords in proportion to the movement of a share price. These types of analysis have shown a definite, but weak ability to forecast the direction of share prices. In this paper we experiment using several linguistic textual representations, including Bag of Words, Noun Phrases, and Named Entities approaches. We believe that using textual representations other than the de facto standard Bag of Words will yield improved predictability results.This paper is arranged as follows. Section 2 provides an overview of literature concerning Stock Market prediction, textual representations, and machine learning techniques. Section 3 describes our research questions. Section 4 outlines our system design.





Section 5 provides an overview of our experimental design. Section 6 expresses our experimental findings and discusses their implications. Finally, Section 7 delivers our experimental conclusions with a brief oratory on future directions for this stream of research.

2 LITERATURE REVIEW

When predicting the future prices of Stock Market securities, there are several theories available. The first is Efficient Market Hypothesis (EMH) (Fama 1964). In EMH, it is assumed that the price of a security reflects all of the information available and that everyone has some degree of access to the information. Fama’s theory further breaks EMH into three forms: Weak, Semi-Strong, and Strong. In Weak EMH, only historical information is embedded in the current price. The Semi-Strong form goes a step further by incorporating all historical and currently public information into the price. The Strong form includes historical, public, and private information, such as insider information, in the share price. From the tenets of EMH, it is believed that the market reacts instantaneously to any given news and that it is impossible to consistently outperform the market.A different perspective on prediction comes from Random Walk Theory (Malkiel 1973).

In this theory, Stock Market prediction is believed to be impossible where prices are determined randomly and outperforming the market is infeasible. Random Walk Theory has similar theoretical underpinnings to Semi-Strong EMH where all public information is assumed to be available to everyone. However, Random Walk Theory declares that even with such information, future prediction is ineffective.It is from these theories that two distinct trading philosophies emerged; the fundamentalists and the technicians. In a fundamentalist trading philosophy, the price of a security can be determined through the nuts and bolts of financial numbers. These numbers are derived from the overall economy, the particular industry’s sector, or most typically, from the company itself. Figures such as inflation, joblessness, industry return on equity (ROE), debt levels, and individual Price to Earnings (PE) ratios can all play a part in determining the price of a stock.

In contrast, technical analysis depends on historical and time-series data. These strategists believe that market timing is critical and opportunities can be found through the careful averaging of historical price and volume movements and comparing them against current prices. Technicians also believe that there are certain high/low psychological price barriers such as support and resistance levels where opportunities may exist. They further reason that price movements are not totally random, however, technical analysis is considered to be more of an art form rather than a science and is subject to interpretation.Both fundamentalists and technicians have developed certain techniques to predict prices from financial news articles. In one model that tested the trading philosophies; LeBaron et. al. posited that much can be learned from a simulated stock market with simulated traders (Lebrun, Arthur et al. 1999). In their work, simulated traders mimicked human trading activity. Because of their artificial nature, the decisions made by these simulated traders can be dissected to identify key nuggets of information that would otherwise be difficult to obtain. The simulated traders were programmed to follow a rule hierarchy when responding to changes in the market; in this case it was the introduction of relevant news articles and/or numeric data updates. Each simulated trader was then varied on the timing between the point of receiving the information and reacting to it. The results were startling and found that the length of reaction time dictated a preference of trading philosophy. Simulated traders that acted quickly formed technical strategies, while traders that possessed a longer waiting period formed fundamental strategies (Learn, Arthur et al. 1999). It is believed that the technicians capitalized on the time lag by acting on information before the rest of the traders, which lent this research to support a weak ability to forecast the market for a brief period of time.In similar research on real stock data and financial news articles, Gidofalvi gathered over 5,000 financial news articles concerning 12 stocks, and identified this brief duration of time to be a period of twenty minutes before and twenty minutes after a financial news article was released (Gidofalvi 2001). Within this period of time, Gidofalvi demonstrated that there exists a weak ability to predict the direction of a security before the market corrects itself to equilibrium. One reason for the weak ability to forecast is because financial news articles are typically reprinted throughout the various news wire


services. Gidofalvi posits that a stronger predictive ability may exist in isolating the first release of an article. Using this twenty minute window of opportunity and an automated textual news parsing system, the possibility exists to capitalize on stock price movements before human traders can act.

2.1 Textual RepresentationThere are a variety of methods available to analyze financial news articles. One of the simplest methods is to tokenize and use each word in the document. While this human friendly approach may help users to understand the syntactic structure of the document, machine learning techniques do not require such structural markings. This technique also assigns

importance to determiners and prepositions which may not contribute much to the gist of the article. One method of circumventing these problems is a Bag of Words approach. In this approach, a list of semantically empty stop-words are removed from the article (e.g.; the, a, and for). The remaining terms are then used as the textual representation. The Bag of Words approach has been used as the de facto standard of financial article research primarily because of its simple nature and ease of use.Building upon the Bag of Words approach, another tactic is to use certain parts of speech as features. This method addresses issues related to article scaling and can still encompass the important concepts of an article (Tolle and Chen 2000). One such method using this approach is Noun Phrasing. Noun Phrasing is accomplished through the use of a syntax where parts of speech (i.e., nouns) are identified through the aid of a lexicon and aggregated using syntactic rules on the surrounding parts of speech, forming noun phrases.A third method of article representation is Named Entities. This technique builds upon Noun Phrases by using a semantic lexical hierarchy where nouns and noun phrases can be classified as a person, organization, or location (Sekine and Nobata 2003).

This hierarchy operates by analyzing the synonyms of each noun and generalizing their lexical profile across the rest of the noun phrase (McDonald, Chen et al. 2005). Named Entities in effect provide for a more abstract representation than Bag of Words or Noun Phrases.

2.2 Machine Learning Algorithms

Like textual representation, there are also a variety of machine learning algorithms available. Almost all techniques start off with a technical analysis of historical security data by selecting a recent period of time and performing linear regression analysis to determine the price trend of the security. From there, a Bag of Words analysis is used to determine the textual keywords. Some keywords such as ‘beat earnings’ or ‘unexpected loss’ can lead to predictable outcomes. These outcomes are then classified into stock movement prediction classes such as up, down, and unchanged. Much research has been done to investigate the various techniques that can lead to stock price classification. Table 1 illustrates a Stock Market prediction taxonomy of the various machine learning techniques.

Table 1 :Taxonomy of Prior Algorithmic research.


From Table 1, several items become readily noticeable. The first of which is that a variety of techniques have been used. The second is that almost all instances commonly classify predicted stock movements into a set of classification categories, not a discrete price prediction.Lastly, not all of the studies were conducted on financial news articles, although a majority was.The first technique of interest is the Genetic Algorithm. In this study, discussion boards

were used as a source of independently generated financial news (Thomas and Sycara 2002). In their approach, Thomas and Sycara attempted to classify stock prices using the number of postings and number of words posted about an article on a daily basis. It was found that positive share price movement was correlated to stocks with more than 10,000 posts. However, discussion board postings are quite susceptible to bias and noise. Another machine learning technique, Naïve Bayesian, represents each article as a weighted vector of keywords (Seo, Giampapa et al. 2002). Phrase co-occurrence and price directionality is learned from the articles which leads to a trained classification system. One such problem with this style of machine learning is from a company mentioned in passing. An article may focus its attention on some other event and superficially reference a particular security. These types of problems can cloud the results of training by unintentionally attaching weight to a casually-mentioned security. One of the more interesting machine learners is Support Vector Machines (SVM). In the

work of Fung et. al., regression analysis of technical data is used to identify price trends while SVM analysis of textual news articles is used to perform a binary classification in two predefined categories; stock price rise and drop (Fung, Yu et al. 2002). In cases where conflicting SVM classification ensues, such as both rise and drop classifiers are determined to be positive, the system returns a ‘no recommendation’ category. From their research using 350,000 financial news articles and a simulated Buy-Hold strategy based upon their SVM classifications, they showed that their technique of SVM classification was mildly profitable.Mittermayer also used SVM is his research to find an optimal profit trading engine (Mittermayer 2004). While relying on a three tier classification system, this research focused on empirically establishing trading limits. It was found that profits can be maximized by buying or shorting stocks and taking profit on them at 1% up movement or 3% down movement. This method slightly beat random trading by yielding a 0.2% average return.2.3 Financial News Article SourcesIn real-world trading applications, the amount of textual data available to stock market traders is staggering. This data can come in the form of required shareholder reports, government-mandated forms, or news articles concerning a company’s outlook. Reports of an unexpected nature can lead to wildly significant changes in the price of a security. Table 2 illustrates taxonomy of textual financial data.

Table 2 : Taxonomy of textual financial data.

Textual data itself can arise from two sources; company generated and independently generated sources. Company generated sources such as

quarterly and annual reports can provide a rich linguistic structure that if properly read can indicate how the company will perform in the future


(Kloptchenko, Eklund et al. 2004). This textual wealth of information may not be explicitly shown in the financial ratios but encapsulated in forward-looking statements or other textual locations. Independent sources such as analyst recommendations, news outlets, and wire services can provide a more balanced look at the company and have a lesser potential to bias news reports. Discussion boards can also provide independently generated financial news; however, they can be suspect sources.News outlets can be differentiated from wire services in several different ways. One of the main differences is that news outlets are centers that publish available

financial information at specific time intervals. Examples include Bloomberg, Business Wire, CNN Financial News, Dow Jones, Financial Times, Forbes, Reuters, and the Wall Street Journal (Cho 1999; Seo, Giampapa et al. 2002). In contrast, news wire services publish available financial information as soon as it is publicly released or discovered. News wire examples include PRNewsWire, which has free and subscription levels for real-time financial news access, and Yahoo Finance, which is a compilation of 45 news wire services including the Associated Press and PRNewsWire.

Besides their relevant and timely release of financial news articles, news wire articles are also easy to

automatically gather and are an excellent source for computer-based algorithms.While previous studies have mainly focused on the classification of stock price trends, none has been discovered to harness machine learning to determine a discrete stock priceprediction based on breaking news articles. Prior techniques have relied on a Bag of Words approach and not the more abstract textual representations. From these gaps in the research we form the crux of our research with the following questions.

3 RESEARCH QUESTIONS

Given that prior research in textual financial prediction has focused solely on the classification of stock price direction, we ask whether the prediction of discrete values is possible. This leads to our first research question.

• How effective is the prediction of discrete stock price values using textual financial news articles?We expect to find that discrete prediction from textual financial news article is possible. Since prior research has indicated that certain keywords can have

a direct impact on the movement of stock prices, we believe that predicting the magnitude of these movements is likely.

Prior research into stock price classification has almost exclusively relied on a Bag of Words approach. While this de facto standard has led to promising results, we feel that other textual representation schemes may provide better predictive ability, leading us to our second research question.

• What textual representation can best predict future stock prices?

Since prior research has not examined this question before, we are cautious in answering such an exploratory issue. However, we feel that other textual representation schemes may serve to better distill the article into its essential components.

4 SYSTEM DESIGN

From these questions we developed the AZFinText system illustrated in Figure 1.


Table 3: Representative Terms.

As the reader will note, the Named Entities representation used the least number of termsof 3 unique terms, while Bag of Words used the most at 6 unique terms. Applying these terms to the weights obtained by the SVM model, returns the following +20 minute prediction; Bag of Words $15.645, Noun Phrases $15.629, and Named Entities $15.642. For this particular article the stock price at the time of article release was $15.65 and dropped to $15.59 in +20 minutes. While all three representations correctly predicted the downward

direction of price movement, Noun Phrases was closer to the +20 minute price. As for the Simulated Trading Engine, none of these three representations predicted a value that would have triggered a trade.

6 EXPERIMENTAL FINDINGS AND DISCUSSION


In order to answer our research questions on the effectiveness of discrete stock prediction


and the best textual representation; we tested our model against a regression-based predictor using the three dimensions of analysis; measures of Closeness,

Directional Accuracy, and a Simulated Trading Engine. Table 4 shows the results of the Closeness measures, Table 5 illustrates Directional Accuracy, and Table 6 displays the results of the Simulated Trading Engine.

Table 4. MSE analysis of the data models

Table 5. Direction Accuracy of the data models

Table 6. Simulated trading engine on the data models

6.1 Our model predicts stock prices significantly better than regression

For our first research question, how effective is the prediction of discrete stock price values using textual financial news articles, we compare the regression estimate against our model using the three textual representations to yield a Closeness measure of MSE which is shown in Table 4. From this table, our model had statistically lower MSEs than the linear regression counterparts for each textual representation (p-value < 0.01); Bag of Words was 0.04713 to 0.07523, Noun Phrases was 0.05826 to 0.07257, and Named Entities was 0.03346 to 0.07244 respectively. This significantly lower MSE score across all three textual representations means that our trained system was significantly closer to the actual +20 minute stock price than the regression estimate.In looking at the Directional Accuracy metric of Table 5, we can further see that our predictive model was again consistently better in all three textual representations (p-value < 0.01); Bag of Words was 49.3% to 47.6%, Noun Phrases was 50.7% to 47.6%,

and Named Entities was 49.2% to 47.4% respectively, where the percentage refers to the number of times the predicted value was in the correct direction as the +20 minute stock price. This result would further imply that our model was better able to use the textual financial news articles in a predictive capacity.The third test was a Simulated Trading Engine. Table 6 shows that our predictive modelagain performed better than regression. Bag of Words gained $5,111 using our predictive model versus losing $1,809 for regression. Noun Phrases had a similar performance gaining $6,353 versus losing $1,809. Named Entities also gained $3,893 in our model versus losing $1,879 using regression.From these three metrics it is quite clear that our machine learning method using article terms and the stock price at the time of article release, performed much better at predicting the +20 minute stock price than linear regression. While Random Walk theory may explain the failures of regression, it would appear that our model was better able to predict future stock prices.


6.2 Named Entities was the better textual representation

To answer our second research question, what textual representation can best predict future stock prices, we again look at Tables 4, 5, and 6, while scrutinizing the differences between various textual representations. For the Closeness measures of Table 4, Named Entities had the lowest MSE score of 0.03346. Bag of Words was next lowest at 0.04713 and Noun Phrases had the highest MSE of the textual representations, at 0.05826 (p-values < 0.01).In the Directional Accuracy measures of Table 5, quite the opposite effect was observed.

Named Entities had the lowest directional accuracy at 49.2%, followed by Bag of Words at 49.3%, and Noun Phrases performed the best by predicting the stock price direction 50.7% of the time (p-values < 0.01).As for the Simulated Trading Engine results from Table 6, Noun Phrases had the highest net profit of $6,353 followed by Bag of Words at $5,111 and Named Entities at $3,893. These seemingly confusing results, where some textual representations perform better than others on particular metrics do not fully explain the observed effects. To answer this problem, we look deeper into the outlay of cash and percentage returns of the Simulated Trading Engine yielding Table 7.

Table 7. Simulated trading engine showing outlay and percentage return

From this table, Noun Phrases invested the most money, $295,000, and consequently also had the lowest percentage return of 2.15% of the three representations. Whereas Named Entities had the lowest investment amount of $108,000 and the highest return of the three at 3.60%. Putting all of this information together; Noun Phrases had the best directional accuracy at 50.7%, but its investment strategy yielded the lowest return through poor stock picks. We believe that the level of abstraction used by this textual representation was insufficient to capture the essence of the article for prediction purposes. This assertion can be backed up by the worst Closeness score of the three representations at 0.05826. While this representation was conservative enough to predict the direction of future stock prices, its wild investment strategy led to too many bad choices and thus reduced its appeal as a representation scheme for financial news articles. By contrast, Named Entities had the lowest directional accuracy at 49.2%, but its more

conservative investment approach coupled with its superior closeness score of 0.03346, led it to the best investment return (3.60%) of the three representations. We believe that the success of Named Entities is directly attributable to its ability to abstract away many of the semantically lesser important terms in an article, and generate a minimally representative essence of the article, suitable for near-term prediction.

7 CONCLUSIONS AND FUTURE DIRECTIONS

From our research we found that our machine learning model using article terms and the stock price at the time of article release performed much better predicting the +20 minute stock price than linear regression. These results were consistent throughout the three evaluation metrics of Closeness, Directional


Accuracy, and Simulated Trading across all three textual representations.We further found that Named Entities performed best of the three representations tested. Named Entities conservative investment style steered it to the best investment return (3.60%) as well as the best Closeness score of 0.03346. While it did not have the best Directional Accuracy, we feel that there is some room for further optimization. We believe that this representations success arises from its ability to abstract the article in a minimally representative way.

Future research includes using other machine learning techniques such as Relevance Vector Regression, which promises to have better accuracy and fewer vectors in classification. It would also be worthwhile to pursue expanding the selection of stocks outside of the S&P500. While the S&P500 is a fairly stable set of companies, perhaps more volatile and less tracked companies may provide interesting results. Lastly, while we trained our system on the entire S&P500, it would be a good idea to try more selective article training such as industry groups or company peer group training and examine those results in terms of prediction accuracy. Finally, there are some caveats to impart to readers. While the findings presented here are certainly interesting, we acknowledge that they rely on a small dataset. Using a larger dataset would help offset any market biases that are associated with using a compressed period of time, such as the effects of cyclic stocks, earnings reports, and other unexpected surprises. However, datasets of several years duration could be imaginably unwieldy for this type of research.

REFERENCES

1. Cho, V. (1999). Knowledge Discovery from Distributed and Textual Data. ComputerScience. Hong Kong, The Hong Kong University of Science and Technology.

2. Fama, E. (1964). The Behavior of Stock Market Prices. Graduate School of Business,University of Chicago.

3. Fung, G. P. C., J. X. Yu, et al. (2002). News Sensitive Stock Trend Prediction. Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD), Taipei, Taiwan.

4. Gidofalvi, G. (2001). Using News Articles to Predict Stock Price Movements. Department of Computer Science and Engineering, University of California, San Diego.

5. Joachims, T. (1998). Text Categorization with Suport Vector Machines: Learning with Many Relevant Features. Proceedings of the 10th European Conference on Machine Learning, Springer-Verlag: 137-142.

6. Kloptchenko, A., T. Eklund, et al. (2004). "Combining Data and Text Mining Techniques for Analysing Financial Reports." Intelligent Systems in Accounting, Finance & Management 12(1): 29-41.

7. LeBaron, B., W. B. Arthur, et al. (1999). "Time Series Properties of an Artificial StockMarket." Journal of Economic Dynamics and Control 23(9-10): 1487-1516.8. Malkiel, B. G. (1973). A Random Walk Down Wall Street. New York, W.W. Norton & Company Ltd.

9. McDonald, D. M., H. Chen, et al. (2005). Transforming Open-Source Documents to Terror Networks: The Arizona TerrorNet. American Association for Artificial Intelligence Conference, Stanford, CA.

10. Mittermayer, M.-A. (2004). Forecasting Intraday Stock Price Trends with Text Mining Techniques. Proceedings of the 37th Hawaii International Conference on Social Systems, Hawaii.

11. Platt, J. C. (1999). Fast training of support vector machines using sequential minimaloptimization. Advances in kernel methods: support vector learning, MIT Press: 185-208.

12. Sekine, S. and C. Nobata (2003). Definition, dictionaries and tagger for Extended Named Entity Hierarchy. Proceedings of the LREC.

13. Seo, Y.-W., J. Giampapa, et al. (2002). Text Classification for Intelligent Portfolio Management. Robotics Institute, Canegie Mellon University.


14. Technical-Analysis (2005). The Trader's Glossary of Technical Terms and Topics,http://www.traders.com. 2005.

15. Thomas, J. D. and K. Sycara (2002). Integrating Genetic Algorithms and Text Learning for Financial Prediction. Genetic and Evolutionary Computation Conference (GECCO), Las Vegas, NV.

16. Tolle, K. M. and H. Chen (2000). "Comparing Noun Phrasing Techniques for Use with Medical Digital Library Tools." Journal of the American Society for Information Science 51(4): 352-370.

About Authors

1. M.Suresh Babu, obtained his MCA from Osmania University and M.Phil Degree in Computer Science from Bharathiar Univeristy. At Present he is doing research in Data Mining. He has organized several workshops and training Programmes in the field of Computer Science and attended a number of Workshops and seminars. He is a life member of ISTE and Science & Society. He was elected as Convener, State Education Sub Committee, A.P.Jana Vignana Vedika. He has written two books 1. Crisis

in Higher Education – Ailments & remedies, 2. Database Management System and contributed more than ten papers in international journals and thirty papers in national journals.

2. Dr.N.Geethanjali, has obtained PhD in 2004 from S.K.University. She is working as Associate Professor in Department of Computer Science & Applications . She has more than 18 years of teaching experience for both UG and PG Courses.Her area of interest are Data mining, Data Communications, Artificial Intelligence, Cryptography, Network Security, Programming Languages.


CRYPTOGRAPHY USING NEURAL NETWORK

[email protected]

[email protected] University

ABSTRACT:

The goal of any cryptographic system is the exchange of information among the intended users without any leakage of information to others who may have unauthorized access to it. In 1976, Diffie & Hellmann found that a common secret key could be created over a public channel accessible to any opponent. Since then many public key cryptography have been presented which are based on number theory and they demand large computational power. More over the process involved in generating public key is very complex and time consuming. To overcome these disadvantages, the neural networks can be used to generate common secret key. This is the motivation for this present work on interacting neural networks and cryptography[1].In the case of neural cryptography, both the communicating networks receive an identical input vector, generate an output bit and are trained based on the output bit. The dynamics of the two networks and their weight vectors is found to exhibit a novel phenomenon, where the networks synchronize to a state with identical time –dependent weights. This concept of synchronization by mutual learning can be applied to a secret key exchange protocol over a public channel. The generation of secret key over a public channel has been studied and the generated key is used for encrypting and decrypting the given message using DES algorithm which is simulated and synthesized using VHDL.

Keywords: neural cryptography, mutual learning, time-dependent weights.

1.INTRODUCTION

Artificial neural networks are massively parallel adaptive networks of simple nonlinear computing elements called neurons which are intended to abstract and model some of the functionality of the human nervous system in an attempt to partially capture some of its computational strengths.[2] A novel phenomenon of dynamic neural network is applied in cryptography systems. The limitation in the general cryptographic process led to the development of cryptographic systems with shorter keys called the secret key systems. The security of such cryptographic system depends on the secrecy of the key. Public key cryptosystems solve the distribution problem by using different secret keys for the internal users. Thus key exchange between users is not required and the security of such system depends on the computational complexity. But in large systems the distribution of key between the users and key bank is still a bottleneck. Moreover, the

problem of passive eavesdropping is still not addressed. So the special characteristic of neural

networks can be used in generating secret key over public channel.

2.NEURAL CRYPTOGRAPHY

2.1 Interacting neural network and cryptography

Two identical dynamical systems, starting from different initial conditions, can be synchronized by a common external-signal which is coupled to the two systems.Synchronization has recently been observed in artificial neural networks as well. Two networks which are trained on their mutual output can synchronize to a time dependent state of identical




synaptic weights [3].This phenomenon has been applied to cryptography as well [4]. In this case, the two partners A and B do not have to share a common secret but use their identical weights as a secret key needed for encryption. The secret key is generatedover a public channel. An attacker E who knows all the details of the algorithm and records any communication transmitted through this channel finds it difficult to synchronize with the parties, and hence to calculate the common secret key. Synchronization by mutual learning (A and B) is much faster than learning by listening(E).Neural cryptography is much simpler than the commonly used algorithms which are mainly based on number theory [5] or on quantum mechanics [6].

2.2 Basic Algorithm

Fig 1.Tree Parity Machine

The mathematical model used for representing a party is a Tree Party machine (TPM) as shown in figure 1. It consists of k different hidden units; each of them is a perceptron with an n-dimensional weight vector wk, when a hidden unit k receives an n-dimensional input vector xk it provides input bit

σ k = sign (wk . xk) ………..(1)

Output bit τ of the total network by k τ = П σi …………………….(2) i=1 The input vector is generally a binary input such that xk Є{ -1,+1 } and both discrete or continuous weights wk can be considered. However, generally discrete weights are considered for ease in explanation.

Each of the two communication parties A and B has their own network with an identical TPM

architecture. Each party selects a random initial weight vector wk (A) and wk (B) at t = 0.Both the networks are trained by their output bits τ (A) and τ (B). At each training step, the two networks receive the common input vector xk and the corresponding output bit τ of its partner. The following is the learning rule

1. If the output bits are different, τ(A) ≠ τ(B) , nothing is changed

2. If τ (A) = τ (B) = τ only the hidden units are trained which have an output bit identical to the common output σk (A/B) = τ (A/B)

3. Three different rules can be considered for training4. Anti- Hebbian learning, Hebbian

learning ,Random walk. If any component wk moves out of the interval [- L, L] it is replaced by sign (wk)L. This can treated as a random walk with reflecting boundaries. Using this algorithm the two neural networks synchronize to a common secret key.

3. DESIGN CONSIDERATIONS

Method of applying Hebbian rule

The two output bits are exchanged and used for the mutual learning and synchronization. At each training step, the machines A, B etc. receive identical input vectors x1, x2 etc. The training algorithm is as follows. Only if all output bits are equal i.e τ(A) = τ(B) =τ the weights can be changed. In this case, only the midden unit ‘i’ whose output bit is σi is identical to the actual output bit τ changed its weights using the following rule

wi(A/B) (t+1) = wi(A/B) (t)+xi …….(3)

If the learning step pushes any component weight out of the interval –L....L, then the component is replaced by ± L. The benefit of this scheme is that the partner as well as any opponent does not know which one of the weight vectors is being updated.

4.SECRET KEY GENERATION

4.1 Key Generation


5 0 6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 00 . 2

0 . 3

0 . 4

0 . 5

0 . 6

0 . 7

0 . 8

0 . 9

1

1 . 1

1 . 2

N o o f i n p u t u n i t s

Syn

chro

nisa

tion

time

The different stages in the secret key generation procedure which is based on neural networks, can be stated as follows [10]

Determination of neural network parameters:k, the number of hidden layer units,n, the input

1. layer units for each hidden layer unit,l, the range of synaptic weight values

2. The network weights to be initialized randomly

3. The following steps are repeated until synchronization occurs.

4. Inputs are generated by a third party (say the server) or one of the communicating parties

5. The inputs of the hidden units (σ) are calculated

6. The output bit τ is generated and exchanged between the networks

7. 7. If the output values agree for each of σi which agree with the output value τ, the corresponding weights are modified using the Hebbian learning rule

8. 8.When synchronization is finally occurred, the synaptic weights are same for both the networks.

4.2 Use of continuous weight

The range of weight values that the perceptrons take can be continuous, instead of being discrete. By using continuous weights, the protocol is enhanced because of the precision involved. [7]

5.CRYPTANALYSIS

Any attacker who knows all the details of the protocol and all the information exchanged between A and B should not have the computational power to calculate the secret key.

Assume that the attacker E knows the algorithm, the sequence of input vectors and the sequence of output bits. In principle, E could start from all of the (2L+ 1)3N initial weight vectors and calculate the ones which are consistent with the input/output sequence. It has been shown, that all of these initial states move towards the same final

weight vector, the key is unique [8]. However, this task is computationally infeasible. This is not true for the simple perceptron [9,10]. The most successful cryptanalysis has two additional ingredients: First, an ensemble of attackers is used. Second, E makes additional training steps when A and B are quiet .Therefore, in the limit of sufficiently large values of L neural cryptography is secure.

6. RESULTS AND DISCUSSION

6.1 Simulation Results

The data set obtained for synchronization time by varying number of input units (n) is shown in fig 2. The number of iterations required for synchronization by varying number of input units(n) is shown in fig 3. The two figures show that as the value of n increases, the synchronization time and number of iterations also increases. A peculiar but favorable behavior is observed at n = 101. In this case both synchronization time and number of iterations required are very less.Figure 2 &3 show the behavior before and after n = 101. Though this behavior is in explicable at present, the value of input units (n) = 101 is chosen to be optimal.

5 0 6 0 7 0 8 0 9 0 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0

2

4

6

8

1 0

1 2

1 4

1 6

N o o f i n p u t u n i t s

No

of it

erat

ions

Fig.2Simulation ResultNo of Input Units Vs No of Iteration


Fig. 3 Simulation ResultsNo of Input Units Vs Sync. Time

Fig 4&5 show the simulation results for two parity machine.Now the synchoronization time is minimum for N=50.

Fig 4.For multiple TPM(3)

Fig 5.For multiple TPM(3)

6.2 VHDL Implementation:DES algorithm

The DES (Data Encryption Standard) algorithm is the most widely used encryption algorithm.DES works by encrypting groups of 64 message bits, which is the same as 16 hexadecimal numbers. In this present work 56 bit key is generated using neural cryptography and used in VHDL implementation of DES algorithm.Figure4 shows the VHDL timing diagram for encryption.


6 10 25 50 80 90 101110

synchronization time(sec)

No of input nodes

No of input nodes vs …

Fig.4 VHDL timing diagram for DES Encryption

7. CONCLUSION

Interacting neural networks have been calculated analytically. At each training step two networks receive a common random input vector and learning. The two partners can agree on a common secret key over a public channel.

An opponent who is recording the public exchange of training examples cannot obtain full information about the secrete key used for encryption .This works if the two partners use multilayer networks, parity machines.

The opponent has all the information (except the initial weight vectors) of the two partners and uses the same algorithms. Nevertheless he does not synchronize. This phenomenon may be used as a key exchange protocol. The two partners select secret initial weight vectors, agree on a public sequence of input vectors and exchange public bits. After a few steps they have identical weight vectors which are used for a secret encryption key. For each communication they agree on a new secret key, without having stored any secret information before. In contrast to number theoretical methods the networks are very fast; essentially they are linear filters, the complexity to generate a key of length N scales with N (for sequential update of the weights)..

In fact, ensembles of opponents have a better chance to synchronize. These may be good news for a possible attacker.

In the new network based approach is faster than conventiona ( number theory) methods.In this paper neural network based mutual learning approach for secret key generation has been attempted as hardware implementation which would be still faster.learn their mutual output bits. A new phenomenon has been observed: Synchronization by mutual

8. FUTURE WORK

The new concepts of public key cryptography, which are not based on number theory method can be used for multi party key exchange protocols.Future research can make use of the secret key generated using neural networks in advanced algorithms like triple DES to improve the security of neural cryptography.

REFERENCES

[1]. Wolfgang Kinzel and Ido Kanter. Interacting neural networks and cryptography Advances in Solid State Physics, edited by B. Kramer (Springer, Berlin, 2002), Vol. 42, p. 383 arXiv: cond-mat/0203011

[2]. Jacek M. Zurada: Introduction to Artificial Neural systems

[3] R. Metzler and W. Kinzel and I. Kanter, Phys. Rev. E 62, 2555 (2000)

[4] I. Kanter, W. Kinzel and E. Kanter, Europhys. Lett. 57,141-147 (2002)

[5] D. R. Stinson, Cryptography: Theory and Practice (CRC Press 2002)

[6] C.P. Williams and S.H. Clearwater, Explorations in Quantum Computing (Springer Verlag, 1998)


http://www.arxiv.org/cond-mat/0203011

[7] G K Patra,Tahir Ali,Anil Kumar and R P Thangavelu: MultipartySecure Key Exchange Algorithm using Neural Cryptologyy.National Workshop On Cryptology:2004

[8]. R. Urbanczik, private communication

[9]. D. R. Stinson, Cryptography: Theory and Practice (CRC Press 1995)

[10] M. Rosen-Zvi, E. Klein, I. Kanter and W. Kinzel, Mutual learning in a treeparity machine and its application to cryptography, Phys. Rev. E (2002).


OPTIMISATION AND ENHANCING CATHODE LIFE OF SPACE HELIX TWT

SIVANANDAM.M1, RAVI.S2

1Research scholar, 2Professor & Head1Sathyabama University, Chennai- 600 119, India

2Dr.M.G.R. Engineering college, Chennai- 600 095, India1E-mail: [email protected]

ABSTRACT:

The life characteristics of Oxide cathodes, M cathodes and Mixed Metal cathodes studied and optimized for space helix TWT. The life of cathode is estimated to be better than 15 years. Instrumental techniques used by NASA cathode life facility, operating for more than 20 years is presented in the Paper. Data analysis by Miram curve, Emission current plot, Knee point migration plot and work function distribution are presented. The cathode life is enhanced by controlling the heater voltage. A heater circuit is designed using Pspice Orcad and simulated with bus variation. The output monitored by Virtual Instrumentation for constant heater load of 1.6Amperes. The heater voltage is controlled very closely resulting in cathode life variation of less than 1000 hrs in 15 years life.

1. INTRODUCTION

In space helix travelling-wave tube (TWT), high energy electron beam transfers its energy to microwave signal and results in microwave amplification. To ensure, a long life for communication satellite the cathode supplying electron beam function satisfactorily at least for 15 years. In low power TWT, the cathode which is made of oxide coating such as barium, calcium or strontium on a nickel base operate from 650°C. In high power TWT dispenser cathode with barium oxide, calcium and aluminum are impregnated on porous tungsten and operate from 1000°C to 1200°C. The following factors control the cathode life:

• The Initial thickness of the impregnated materials on the cathode surface

• The Heater voltage • The Cathode operating temperature• The Cathode or beam current• Vacuum Integrity of TWT

2. CHARACTERISTICS OF CATHODE

A Cathode emits electrons when operated in the vicinity of 650°C to 1200°C depending upon the material. The current density at the emitting surface

of the cathode, called current loading is expressed in Amperes per Square centimeter. The cathode current and beam current are the same. The cathode life is considered to be end, when beam current falls by 5% from the initial beam current.

3.0 STUDY OF CATHODE PARAMETERS AND LIFE EXPEPANCY

3.1 Classification of Cathodes

Cathodes used in space TWT is broadly classified as Oxide cathodes and Dispenser cathodes. Dispenser cathodes are further classified as M- type cathode and MM-type cathodes. The parameters of the cathodes and its life characteristics are analyzed. To enhance the life, cathode loading of less than 2A/cm2 is considered.

3.2 Oxide Cathodes


Oxide cathodes use Nickel substance coated with oxides of Barium, Calcium and Aluminum. The cathode loading is 2A/cm2 and the operating temperature is 670°C to 750°C. The typical specifications are:Coating Substance : Ni sponge.

Fig.1 Life prediction for oxide cathodes.

3.3 m-types cathodes

The M-Types cathode has a sintered tungsten matrix impregnated with barium oxide, calcium oxide and aluminum oxide. The emitting surface has osmium/ruthenium film.

Typical specifications are:Matrix composition : 100%wImpregnate molar Ratio : 5:3:2, Bao: Cao:Al2O3.

Percent Impregnations: 75%Particle size : 4.5 micronFilm Type : OS/RSFilm Thickness : 4000-6000ÅOperating Temperature: 900°C-1100°CCathode Loading : 2A/cm2

Average Life : 45,000hours.

The life curve M-Cathode life is shown inFig.2. Since the end of Cathode Life is short, it is not considered for communication satellite.

Fig.2 Life Prediction for M-Cathode

3.4 Mixed Metal Matrix Cathode

Mixed Metal Matrix (MMM) Cathode has Matrix compose of 80% Tungsten and 20% iridium. The matrix is impregnated with M-Type cathode impregnated compounds. TM refers to another mixed metal matrix of 95% tungsten and 5% Iridium. The emitting surface has Tungsten/iridium film.

Typical specifications of MMM-type cathode are:

Matrix Composition :80%W/ 20% IreImpregnate molar ratio : 6:1:2, Bao: Cao:

Al2O3

Percent Impregnation : 79.11%Matrix Density : 71.52%Particle Size : 15.16/ 325meshCathode Load : 2A/Cm2

Average Life : 1, 00,000hours.

The life Curve for MMM Cathode is shown in Fig.3 [2].The Life of Cathode is good and preferred for space communication.

Fig.3 Life prediction for MMM-type cathode.

4. CATHODE LIFE PREDICTION MODEL

Cathode life Prediction model provides a method of calculating cathode current changes with time and temperature. Tungsten dispersion B-type and osmium-ruthenium dispenser M-type cathode used in space TWT are modeled. The results for M-cathode is compiled and compared with accelerated life test data run in a TWT electron gun. The life curve is shown in Fig.4 [3].


Fig.4 Life prediction for M-type cathode.

5. CATHODE LIFE TEST FACILITY AND INSTRUMENTATION

The Cathode life Test facility is an Independent Laboratory under US Navy and offers an environment for life testing cathode system. During 20years, this facility has life tested various cathodes and accumulated 3,959,889hours of life testing. Each test system consists of a vacuum envelope with integral ion pump, cathode mounted on assembly containing a heater filament, anode and collector. The cathode power supply is 0 to - 6 KV dc, Collector supply fixed at 30% of cathode voltage and heater voltage 0 to 10V ac [2]

6. DATA ANALYSIS METHOD

Cathode performance data is presented in four different forms:

• Miram Curve• Plot of Emission current versus time• Knee point migration plot• Work function Distribution

In Miram curve, X-axis is cathode operating temperature and Y-axis is cathode emission current at the first data point. The long term performance of each cathode system is presented as a degradation plot. In each plot, the current at the operating temperature is plotted as a function of life in hours. The resultant plot shows the behavior of the cathode system over time and provides a convenient method of comparing the potential life of different cathode systems. The analysis also contains Knee point migration plots that shows how quickly the Knee temperature of a cathode approaches its operating temperature. Miram curve, Knee point curve and work function curve are shown in Fig.5 [2].

.

Fig.5(a) Miram Curve

Fig.5(b) Knee point Curve

Fig.5(c) Work function curve.

7. ENHANCING CATHODE LIFE BY HEATER VOLTAGE

A heater is used to raise the cathode to its operating temperature. The heater consists of a coil of tungsten and potted with cathode assembly. The heater operates at 6.2V or 6.3V DC at a load current of 1.6A and dissipates 10 watts. The heater voltage variations affect the cathode temperature, beam current and life of cathode. A logarithmic relation exists between the heater voltage and life of cathode. Life curve for an impregnated tungsten cathode is shown in Fig.6 [4].


Fig.6 Variation of cathode ife vs. Heater voltage.

LA heater voltage variation of ±5% cause a variation of ±3.8% in cathode temperature, ±7.5% in beam current and tenfold change in cathode life (90,000 hours/ 10 years) [4]

Hence it is essential to control the heater voltage variations to close tolerance. To study the variations of heater voltage, a regulated circuit was designed using Pspice– Orcad with IC regulator LM405 and diodes IN4376 & IN4099. The potential of common terminal is elevated to obtain a regulated output of 6.2V. Capacitors are provided at input and output for ripple reduction and RF noise suppression. The regulator supplies a load of 1.6A at output power of 10W, Fig.7

Fig.7 Regulated Heater circuit.

The heater is powered by satellite bus which varies from 18V to 30V DC, based on the solar power, To simulate the condition, supply voltage was varied from 18V to 30V and output monitored by virtual oscilloscope. The output voltage for different supply voltage is shown in Fig. 8.

Fig.8(a) Load current maintained at 1.6A.

Fig.8(b) Output for 26V, Load 1.6A

Fig.8(c) Output for 24V,1.6A

Fig.8(d) Output for 22V, 1.6A


Fig.8(e) Output for 20V, 1.6A

Fig.8(f) Output for 18V Load, 1.6A.

The observations are given in Table.1. It is observed that a bus variation of 12V to 26V is acceptable and heater voltage is within close tolerance.

Tabel 1 Measurement of Heater VoltageLoad: 1.6A

Bus voltage, V

Heater Output, V Remarks

18.0 6.206 Acceptable20.0 6.206 ”22.0 6.206 ”22.5 6.206 ”23.0 6.206 ”23.5 6.205 ”

24.0 6.201 Designed point

24.5 6.205 Acceptable25.0 6.204 ”25.5 6.204 ”26.0 6.203 ”

30.0 5.255 Beyond Limit

8. CONCLUSIONS

• For Space applications, Mixed Metal cathode gives more life, stable and degradation is linear.

• For the proposed circuit, bus voltage of 18V- 26V is acceptable. Beyond 26V, regulation is poor.

• For the bus varied, heater variation is within +0.1%, Cathode temperature is -0.07%, beam current variation is +0.25% and cathode life variation is less than 1000hours, in a life time of 15 years.


REFERENCE

• [1] TWT/TWTA Handbook, 13ed, PP42, L3 Communications Electrons Technology, INC,USA,2007

• [2] Lawrence Dressman; “The Tri service/ NASA Cathode Life Test facility 2000 Annual Report”, PP 6-1, 6-2, A-1, A-2, C-1, C-2, C-3, Indiana, USA, 2002.

• [3] Longo, R.T; Adler, E.A; and Falce, L.R;” Dispenser Cathode Life prediction Model”, Hughes Aircraft company, USA, 1984.

• [4] Sivan, L;”Microwave Tube Transmitters”, 1ed, PP113-114, Chapman & Hall, London, 1994.


EXPERIMENTAL INVESTIGATION OF FORCED CONVECTIVE HEAT TRANSFER IN RECTANGULAR MICRO-CHANNELS

R.KALAIVANAN1 and R.RATHNASAMY2

1 Lecturer (Selection Grade), 2Professor,Mechanical Engineering, Annamalai University, Annamalainagar, Tamilnadu, 608002,

India 1Corresponding author,email: [email protected]

ABSTRACT:This paper investigates the experimental program on the study of heat transfer characteristics in micro-channels. The two test sections used are of 47 and 50 micro-channels in rectangular cross-section of equivalent diameters 387 and 327 µm respectively. Each channel of length 192 mm is fabricated on a 304 stainless steel substrate (230 mm x 160 mm x 1.6 mm) by photo chemical etching process. Covering the top with another plate of 0.5 mm thickness forms the channels by vacuum brazing. Experiments cover laminar region using the fluids ethanol, methanol and an ethanol-methanol mixture. The heat transfer coefficient is evaluated based on the heat carried away by the coolant and an average wall to mean fluid temperature difference. The Nusselt number is correlated through empirical correlations involving Reynolds number and Prandtl number with length parameter, the hydraulic diameter.

Keywords: Micro-channels, alcohols, alcohol mixture, forced convection, laminar.

NOMENCLATURE Symbolsa indexcp specific heat (J/kg K)C constantd diameter (m)Deq equivalent diameter (m)H heat transfer coefficient (W/m2 K)H channel height (m) L channel length (m)M mass flow rate (kg/s)n Prandtl indexq heat transfer rate (W) qepi electrical power input (W)q” heat flux (W/m2)Re Reynolds number ( =ρvd/µ)t temperature (oC)v velocity (m/s)

W channel width (m)x mass fractiony mole fraction

Greek Letters“ differentialλ thermal conductivity ( W/m K)µ dynamic viscosity, (N s/m2 or Pa s)ρ fluid density ( kg/m3)

Subscriptseq equivalent

f fluid

i inletm mean mix mixture o outletw wall conditions1 component 12 component 2

1. INTRODUCTION

The systematic research into micro-scale flow and heat transfer studies were in vogue for the past three decades, started by initiation of Tuckermann & Pease [1,2] on the thermal management of high heat dissipating electronic components both covering chip and board levels. As far as micro-scale flow passages are concerned, the literature sources are not more recent. The emergence of micro-electro-mechanical systems (MEMS) has generated significant interest in the area of micro scale heat transfer. Several investigations ensued those dealt with flow of gases [3] and liquids [4] through micro-geometries. Fluid flow through micro-scale flow geometries is encountered in numerous engineering systems such



as cooling of electronic devices and compact heat exchangers.Micro-channel flows have been used for liquid dosing and flow measurement [5]. Non-circular geometries are often adopted because of their relative simplicity in fabrication as compared to circular channels. Issues pertaining to micro-channel heat exchangers were dealt with by Peterson [6] and Ravigururajan et al [7]. Theoretical approaches to fluid flow and heat transfer as well were reported [8]. The substrates used were silicon, glass, copper and stainless steel. The test section geometries varied from a fraction of a µm to a few 100s of µm. There had been limited study with mixtures of fluids [9] and Rathnasamy et al [10] investigated flow of alcohols and its mixtures in long serpentine micro-channels. Steinke and Kandlikar [11] reviewed previous studies on the topic of fluid flow and heat transfer and tried to explain the deviation in data by conducting experiments. It is concluded that deviations came from the measurement uncertainty of micro-channel dimension in conjunction, with inlet and exit losses. Fluid flow and heat transfer experiments were conducted in rectangular micro-channels by Jung and Kwak [12]. There is general agreement in the literature that the high relative roughness of micro-channels reduces the critical Reynolds number for transition to turbulent flow. It is well known that roughness influences the transition to turbulent in Schlichting [13].

These studies provide substantial evidence to prove that flow and heat transfer in micro-channels need to be addressed differently compared to conventional channels. Wang et al [14] have attempted to provide some possible explanation to peculiarities of micro-channel heat transfer. It appears that the results quite good agreement with experimental data. Moreover Jung-Yeul Jung et al [15] have reported that the Nusselt number increases with increasing Reynolds number in laminar regime by doing experiments with nano fluids. There is a need for more experimental data on a variety of fluids and flow geometries (i.e; channel dimensions and relative roughness) so that some generalized conclusions can be evolved. 2. TEST SECTION

Micro-components are mostly fabricated using etching, deposition and photo-lithographic techniques.

The EDM is considered to be a non-conventional machining technique. It was reported that with a precision EDM dimensional tolerances up to 0.5 µm could be obtained. All dimensions are in mm In the present case, two test sections herein after designated as MCP1 and MCP2 having common features; each channel of length 192 mm were fabricated on a 304 stainless steel substrate (230 mm x 160 mm x 1.6 mm) by photo chemical etching process as shown in Figure 1. MCP1 and MCP2 have 47 and 50 micro-channels of rectangular cross-section 1000 by 240 µm and 900 by 200 µm in width and depth, respectively. The channel header portions were deepened after etching by EDM in order to have negligible pressure loss. The surface roughness (ε) of MCP1 and MCP2 was done to check the uniformity of the channel using a surface profilometer (Rank Taylor Hobson). The surface roughness measured have shown rms values of the order 1.11-6.90 µm for machined surface and 0.19-0.34 µm for cover plate. The flow passage is formed by vacuum brazing the machined plate, covering with another stainless steel plate of thickness 0.5 mm on top. Inlet and out conduits were attached together with the two plates and brazed in a vaccum furnace at 10-5 torr and about 1000oC. The mating surfaces were coated with some plating technique to assist brazing.

Fig. 1. Details of micro-channel test section.3. EXPERIMENTAL SETUP AND PROCEDURE

The experimental set-up, shown in Figure 2, consists of a liquid reservoir (capacity ~ 20 lit) to supply fluid to the test section. A diaphragm operated pump is used to pump fluid to the test section through a micro-filter (~ 50 micron) built-in in the main line to avoid any dirt that may enter into the test section. In the absence of micro-filter poor measurement data


could be resulted due to blockage of the channels to the flowing fluid.

Fig. 2. Schematic of experimental setup.Legend: 1 Sump 2 Pump 3 By-pass control valve 4 Flow control valve 5 Micro-filter 6 Flow meter 7 Test section 8 Differential pressure transducer 9 Temperature bath Experiments were conducted with liquids ethanol, methanol and a mixture 50%E-50%M by volume (i.e; 0.53E-0.47M by mole fraction). Further the test setup is provided with by-pass line and control valves to establish the required flow rate in the test section and as well the pressure drop, measured with the aid of differential pressure transducer (make: KELLER Druckmesstechnik; piezoresistive pressure transmitter, PD-23/ 5 bar/8666.1). Flow rate through the test section was measured by using the flow meter (make: DIGMESA; magnetic turbine flow meter). However, the flow meter was calibrated by measuring known volume of methanol manually with a maximum absolute deviation 4% over the measuring range 0.03- 8.00 lpm. Nichrome ribbon heaters wound on a mica sheet and encapsulated between a pair of mica sheets, were used as heat sources. Two identical heaters were mounted on either side of the test section. This enabled application of equal heat inputs from both sides of the test module. The entire test section was adequately insulated with 2-mm thick asbestos sheet followed by 75-mm thick glass wool and 20-mm thick polystyrene foam on each side and at the edges. The fluid coming out of the channel was passed through a cooling coil and then in to the sump. The cooling coil was submerged in a temperature bat The flow rates were varied with a corresponding variation of heat inputs. Thermocouples (T-type) and few Platinum resistance thermometers were affixed to the surface at 15 designated locations on the each side of the test section. A few thermocouples were used for measuring the inlet and outlet temperatures of conduits of the micro-channel test section. A

digital display unit is used with a resolution of 0.1oC. in the measurement. The flow medium used to test is filled in the reservoir. To start the experiment all the required precautions are taken into consideration. The pump is switched on while keeping the by-pass valve in open position and control valves to test section closed. The required discharge flow rate of the pump is obtained through stroke adjustment in the pump system. Control valve is somewhat opened allowing the flow to take place in the test section. Later, the control valve is tuned to set the required flow rate indicated by the display unit connected. Pressure drop at the test section for the corresponding flow rate indicated by the display unit is noted. The steady value of flow rate is taken to reduce the measurement uncertainty. Several trials are carried out and the average value is taken to evaluate flow rate in terms of cc/min basis. The primary data were heat flux, flow rate, pressure drop, and temperatures at various locations on the test section. Each experiment is continued till the steady state conditions had stabilised, which took about one hour after the commencement of a test. The experiment was continued for a further period of 10-15 minutes before switching to the next run. In the present case of heat transfer experiments the Reynolds number covered the laminar region (30 d” Re d” 500). All tests were conducted with the test section in a horizontal plane.

4. DATA REDUCTIONThe Reynolds number is defined in the conventional way, Re = ρvd/µ. The velocity v (average) is calculated from flow rate based on the cross-sectional area of the channel. The velocity is evaluated using the mass flow arte and the equivalent diameter (Deq= 2 WH/(W+H)). The mass flow arte was evaluated based on the density at inlet condition. 4.1 Property evaluation

Thermophysical properties require d for the heat transfer calculations were evaluated at the mean fluid temperature, defined as

2)t(tt fofi

fm+=

(1)


The linear mixing rules for specific volume were used for evaluating the density, specific heat and thermal conductivity of the mixtures. The logarithmic mixing rule was used for viscosity of liquid mixed mixtures [16]. Temperature-dependent pure property data used were taken from reference [17]. 4.2 Heat transfer coefficientThere is an uncertainty over the amount of heat transferred to the fluid. Because of the inevitable losses from the test section, all the electrical power given to the heaters is not transferred to the fluid. Thus, the actual heat transferred to the fluid was calculated from the enthalpy change. The heat flux was calculated based on energy balance of the fluid.

)t(tq fifo −= pcm(2)

)(2qm HWL

q+

=′′(3)

The average heat transfer coefficient is defined as

mt∆′′

= mm

qh(4)

where

)(m fmwm ttt −=∆

(5)The mean wall temperature (twm) was based on the measurements at 15 locations. The Nusselt number is correspondingly defined as

fm

eqmDhNu

λ=

(6)This is procedure is similar to the one used by other researchers also [18,19].Table 1 shows the electrical heat input, velocity, fluid inlet and outlet temperatures, mean wall temperature and heat transfer rate for ethanol flow in MCP1. The last column indicates the percent of heat transfer to the fluid with respect to electrical power supplied to the heater. Similar data for experiments with methanol and one liquid mixture were generated for both channels. It is seen that about 10% of the input power is lost to the ambient in experiments.5. RESULTS AND DISCUSSIONThe Nusselt number data obtained from previous section are shown in Figure 3 as a plot of Nusselt number versus Reynolds number. These include all the experiments with liquids. These data show a perceptible fluid species dependence.

Several researchers [8,19,20,21] have brought out that the transition to turbulent regime occurs at lower Reynolds numbers in micro-channels than in normal channels. These transitions have been identified in the present fluid flow work also communicated [22]. The experimental data from the present work had been fitted to relation as in figure 4 with the following relation:

PrReC=Nu na(7)

The values in eq. (7) for correlation coefficient ‘C’ and indices a and n are obtained.

Fig. 3. Nusslet number versus Reynolds plots for all liquids in micro-channels

Legend: ΅% MCP1 ethanol, % MCP2 ethanol Λ% MCP1 methanol Ο% MCP2 methanol Δ MCP1 0.53E-0.47M mixture ²% MCP2 0.53E-0.47M mixture.

Fig. 4. Nusselt number versus Reynolds number plot for all liquids in micro-channels with fit to equation

(7) as solid line.6. CONCLUSIONSAn experimental setup was devised to study the characteristics of single phase forced convective heat transfer in micro-channels. The two test sections MCP1 and MCP2 employed has equivalent diameter of 387 and 327 µm on 230 x 160 mm2 steel plate. The flow media studied were alcohols and their mixture. The measured flow rate and temperatures data were used to evaluate the Nusselt number. An empirical correlation is obtained for Nusselt number in terms of


Reynolds number and Prandtl number for laminar region as below:

PrRe949.0Nu 0.406.0= (8)More experiments on different channel dimensions, surface roughness and fluids are needed to evolve a generalised correlation for heat transfer at micro-scale length. These experimental data could set off

the ongoing efforts on micro-channel studies. The dependence of Nusselt number on Reynolds number and Prandtl number in micro-channel is quite different from that of normal channels.ACKNOWLEDGEMENTS The test modules used in this paper are financially supported by Department of mechanical engineering, Indian Institute of Science, Bangalore. The authours are grateful to Liquid Propulsion Systems Centre, Indian Space Research Organisation, Bangalore for deploying their expertise in the fabrication of test sections.

27 12.50 1.420 25.1 42.5 51.6 11.47 91.728 14.20 1.831 25.2 41.0 54.0 13.47 94.829 14.50 1.754 25.5 42.0 53.4 13.76 94.930 14.60 1.670 25.2 43.0 52.8 13.91 95.3


Experiment qepi(W) v (m/s) tfi (oC) tfo (oC) twm(oC) q (W) q/qepi (%)

1 2.00 0.136 24.1 52.0 41.7 1.83 91.42 2.50 0.165 24.5 53.0 42.0 2.28 91.13 2.70 0.185 24.3 52.0 42.4 2.47 91.44 3.20 0.208 24.1 54.2 44.2 3.03 94.85 3.80 0.232 24.3 57.0 46.9 3.69 97.06 4.00 0.248 24.3 54.7 45.0 3.66 91.57 4.50 0.278 24.3 54.6 46.3 4.07 90.68 5.25 0.324 24.8 56.2 48.8 4.93 93.99 5.50 0.306 24.8 58.5 49.1 5.03 91.410 5.80 0.377 25.2 54.4 49.7 5.33 91.911 6.00 0.428 25.0 52.0 47.0 5.58 93.112 6.60 0.467 25.0 53.3 48.3 6.39 96.813 7.00 0.443 24.2 54.5 49.2 6.49 92.814 7.50 0.502 24.0 54.0 51.2 7.29 97.115 8.00 0.829 24.6 43.5 48.1 7.30 91.316 8.50 0.916 25.1 43.5 47.1 8.07 94.917 9.00 0.798 24.8 46.4 50.1 8.17 90.818 10.00 0.572 25.0 57.9 55.2 9.18 91.819 10.00 0.689 25.2 53.4 54.0 9.39 93.920 10.00 0.730 24.2 51.2 51.9 9.49 94.921 10.20 0.764 24.2 49.1 51.1 9.13 89.522 11.00 0.660 25.2 58.0 55.5 10.56 96.023 11.00 1.086 25.3 44.8 49.8 10.01 91.024 12.00 1.210 25.1 44.8 50.4 11.15 92.925 12.00 1.507 25.3 42.0 52.2 11.82 98.526 12.40 1.285 25.2 43.7 51.0 11.16 90.0

Table 1 Raw data for ethanol flow in micro-channelReferences

[1] Tuckermann, D.B., and RFW. Pease, 1982. Optimized convective cooling using micro machined structures. J.Electro Chem. Soc. 129, (3) C98.

[2] Tuckermann, D.B., 1984. Ph. D. Thesis, Department of Electrical Engineering, Stanford University, Alto, CA.

[3] Wu, P., and W.A. Little, 1983. “Measurement of friction factors for the flow of gases in very fine channels used for micro-miniature Joule-Thomson refrigerators”, Cryogenics 23, (5): pp.273-277.

[4] Adams, T.M., S.L. Abdel-Khalik, S.M. Jeter, and Z.H. Qureshi, 1997. “An experimental investigation of single-phase forced convection in micro- channels”, AIChE Symposium series., Heat transfer-Baltimore. pp: 87-94.

[5] Richter, M., P. Woias, and D. Weiß, 1997. Micro-channels for applications in liquid dosing and flow-rate measurement. Sensors and Actuators A Physical 62, pp: 480-483.

[6] Peterson, G.P.,(1992). Overview of micro heat pipe research and development. Appl. Mech. Rev. 45, (5) pp :175-189.

[7] Ravigururajan, T.S., J. Cuta, C.E. McDonald, and M.K. Drost, 1996. Single- phase flow thermal performance characteristics of a parallel micro-channel heat exchanger. HTD-329, National Heat Transfer Conference, 7 ASME, pp: 157-166.

[8] Mala, G. M., D. Li, and J.D. Dale, 1997. Heat transfer and fluid flow in micro- channels. Int. J. Heat Mass Transfer 40, (13) pp : 3079-3088.

[9] Peng, X.F., and G.P. Peterson, 1996. Convective heat transfer and flow friction for water flow in micro channel structures, Int. J. Heat Mass Transfer 39, pp: 2599-2608.

[10] Rathnasamy, R., J.H. Arakeri, and K. Srinivasan, 2005. Experimental investigation of fluid flow in long rectangular micro-channels. Proc. IMechE Part E: J. Process Mechanical Engineering, 219, pp: 227-235

[11] Steinke, M.E. and S.G. Kandlikar, 2006. Single- phase liquid friction factors in micro-channels. Int. J. Heat Mass Transfer 45, pp: 1073-1083.

[12] Jung, J.Y., and H.Y. Kwak, 2008. Fluid flow and heat transfer in micro- channels with rectangular cross section. Int. J. Heat Mass Transfer 44, pp: 1041-1049.

[13] Schlichting, H., 1979. Boundary Layer Theory, McGraw-Hill Book Company, New York.

[14] Wang, B. X., and X. F. Peng, 1994. Experimental investigation on liquid forced convection heat transfer through micro-channels. Int. J. Heat mass Transfer 37, pp: 73-82.

[15] Jung-Yeul Jung., Hoo-Suk Oh, and Ho-Young Kwak, 2009. Forced convective heat transfer of nanofluids in micro-channels. Int. J. Heat Mass Transfer 52, pp: 466-472.

[16] Heide, R., 1980. An instrument for measuring viscosity of refrigerants and refrigerant mixtures. Proc. Int. Inst. Of Refrig., Comm. B1,B2,E1,E2, Mons. Belgium.

[17] Beaton, C. F., and G.F. Hewitt, 1989. Physical Property Data for the Design Engineers. Hemisphere publishers, New York.

[18] Peng, X. F., and Wang, B. X, 1994. Liquid flow and heat transfer in micro-channels with / without phase change heat transfer. In proceedings of the 10th International Heat Transfer Conference, Brighton, UK, 1, pp: 159-177.

[19] Wu, P., and W.A. Little, 1984. Measurement of heat transfer characteristics of gas flow in fine channel heat exchangers used for micro-miniature refrigerators. Cryogenics, 24, pp: 415- 420.

[20] Pfahler, J. N., J. Harley, H.H. Bau, and J. Zemel, 1990. Liquid transport in micron and submicron channels. Sensors and Actuators., 38 pp: 431- 434.

[21] Harely, J., Y. Huang, H.H. Bau, and J. N. Zemel, 1995. Gas flow in micro- channels. J. Fluid Mech., 284, pp: 257 -274.

[22] Kalaivanan, R., and R. Rathnasamy, 2009. Investigation of low density fluid flow through parallel micro-channels, Paper communicated.


DOMESTIC ELECTRICAL ENERGY MODELING FOR INDIA

N V RAO1 AND VVS KESAVA RAO2

Mechanical Engineering Department Andhra University, Visakhapatnam,Andhra Pradesh, India E-mail1&2: [email protected]

Mailing address1: Anil Neerukonda Institute of Technology & Sciences, Sangivalasa, Visakhapatnam-531163, IndiaCell No:91- 98492198381(1) 91-9989087001(1) 1Corresponding author

ABSTRACT:This paper examines the residential demand for electricity in India using two methods. In first method residential electricity consumption is used to ARIMA (P,D,Q) (p, d, q) and in the second method Transfer functions model is applied with residential electricity consumption as dependent variable and Urbanization index and literacy rate as explanatory variables. It is found that Transfer function model between residential electricity consumptions and literacy rate is found to be more suitable for predicating future residential consumption than ARIMA (P, D, Q) (p, d, q) model. Using that model demand predications are made for next 10 years. In the literature survey transfer function model is not used for residential energy modeling purpose earlier and definitely this model will be helpful in predicating energy consumption in any where.Keywords: Arima, Prewhitening, white noise, transfer function.

1.0 LITERATURE SURVEY:

R.E.ABDEL-AAL, A.Z.AL-GARNI and Y.N.AL-NASSAR(1) (1997) in their study have used Abductive neural networks for modeling and forecasting electricity energy consumption in Eastern Saudi Arabia. Raymond Y.C.Tse, John Raftry(2) (1999) in their paper titled “Income elasticity of housing consumption in Hong Kong: a co integration approach”. Journal of Property research, 1999, have found that Income elasticity of housing expenditure is of considerable interest particularly to applied researchers in housing economics. However, there is a wide variation in the estimation of income elasticity. Such a variation is attributed to different specifications of estimation. This paper applies techniques developed in the literature of co integration analysis to re – examine the income elasticity of housing consumption based on a long-run equilibrium model fitted with time-series data. Tests for unit roots are conducted to avoid superiors result in the estimation of the model. The use of the time series data generates some attractive properties. These results can be used to estimate the value of income elasticity of public and private housing residents, and of owners and renters separately. Soteris A. Kalogirous(3) [2000] has used Artificial Neural- Networks for modeling Energy. Pernilalli Holtedahl and Frederick L.Joutz(4) (2000) examined the residential demand for the electricity in Taiwan as a

function of household disposable income, population growth, the price of electricity and the degree of urbanization. Short run and long term effects are separated through error correction model. They used a proxy variable, urbanization, to capture economic development characteristics and electricity using capital stock not examined by income. Eric Girardin, Lucio Sarno (5) (2001) in their paper titled “Private consumption behavior, liquidity constraints and financial deregulation in France: a nonlinear analysis”. Empirical Economics examined the effect of financial deregulation on consumption expenditure in France during the period 1970-1993. A nonlinear model for consumption, which allows for liquidity constraints through a time varying parameter dependent on proxy for financial deregulation, is estimated using nonlinear instrumental variables. It is concluded that in France financial deregulation has significantly reduced liquidity constraints faced by consumers, allowing a higher percentage of the population to smooth consumption over time. Evidence is also provide that the inter temporal electricity of substitution is not significantly different from zero at conventional normal levels of significance. Herath M. Gunatilake, Inoka Chandrasena(6) (2001) estimated the demand for domestic water in a fast-growing city of a developing country. Monthly data for 40 randomly selected households for a six-year period were used for the estimation. A log-log model was selected as a proposed


specification for the demand function. Marginal price, difference price, income, and household size were used as the independent variables. Resulted shown that water is neither price-nor income-elastic. However, very low price responsiveness can be used to increase water revenues of the municipality. Planning commission (7)

(2002) Electricity demand in Indian households. In this paper, price and income elasticity’s of electricity demand in the residential sector of all urban areas of India are estimated for the first time using disaggregate level survey data. Three electricity demand function have been estimated using monthly data for winter, monsoon & summer. Geoffrey K.F.Tso, Kelvin K.W.Yaw(8) (2003) have applied multiple regressive model for modeling domestic energy usage in Hon Kong. Melody Stokes, Mark Rylatt, Kevin Lomas(9) (2004) used a cross – spectral model for modeling domestic lighting demand in UK. Merih aydinalp, V.Ismet Ugursal, Alan S.Fung(10) (2004) have applied neural network method for modeling purpose. Dorte Grinderslev, Kennet Karlsson, Erik Bjorsted (11)

(2004) presented a model a for household demand for electricity estimation on Danish macro economic date. They set up a theoretical model, where the consumer maximizes utility subject to a budget constraint given by the total private consumption. A variable for the aggregated stock of household electrical appliances is constructed using National Account time series for the private purchase of different categories of appliance and average depreciation rates for each category (an imputed capital stock). Zaid Mohammad, Pat Bodger(12) (2005) in their paper titled Forecasting electricity consumption in New Zealand using economic and demographic variables have used multiple regression technique and compared their results with the results of Logistic model. Devood manzoor(13) (2005) presented an approach to the modeling of residential demand in Iran. The key determinants of residential energy demand are the total house hold expenditure as a proxy for disposable income and the real price energy carriers (electricity, natural gas and oil products). He focused on price and income elasticity’s for forecasting purpose. Taking in to account the energy carrier substitution, it was modeled in two interrelated levels and aggregate level and disaggregates level. As a general modeling technique in econometrics lagged variables are introduced to account for dynamic effects over time. V. Paatero and Peter D Lund(14) (2005) in their paper titled, A model for generating house hold electricity load profiles demand side management measures have been adopted Runming Yao, Koen Steemers(15) (2005).have applied cluster analysis Method for modeling domestic energy demand in UK. S.Jebaraj and S.Iniyan (16)(2006) have discussed different energy planning models, forecasting models etc. F.Gerard Adams and Yochanan Shachmurove(17) (2008) have discussed different growth models for energy planning of China.

Holger Fabig in his IMF working paper titled Modelling Macro –Critical Energy sectors in low income countries(18) (2008) have used linear models to project the energy demand. Mohammad Reza Faraji Zonooz, .Z.M. Nopiah, Ahmad Mohd. Yusof and Kamaruzzaman Sopian(19) (2009) have used a MARKAL software in their study titled A review of MARKAL energy modeling. Aneta Strrzalka, Jurgen Bogdahm and Ursula Eichker(20) (2010) in their paper titled 3D City modeling for Urban scale Heating energy demand forecasting have used a simulation model for computation of energy demand . Aneta Strrzalka, Ursula Eichker, Volker Coors and Jurgen Schumacher(21)

(2010) in their paper titled Modeling energy demand for heating at city scale have also used a simulation software for arriving the results. Hence various authors have adopted different engineering techniques to compute the energy requirements. But ARIMA(P,D,Q)(p,d,q) model and transfer function method so far not adopted for domestic electrical energy modeling by any researchers either in India or in any countries and hence in this work those two techniques have adopted to determine domestic electrical energy demand for India

2.0 INTRODUCTION:

The energy modeling has been very crucial issue for people in energy planning and many people attempted it in different directions. Since the energy consumption is dependent on many factors and in each sector the energy consumption is again affected by certain key determinants which determine the demand of energy. Even though energy is basically divided in to two basic types like conventional and non conventional but these days the dependence on conventional energy cannot replaced by the other one. Hence it has become very important to explore the different technologies that provide availability of energy at an affordable price.In this paper Domestic electrical energy demand has been attempted in two different environments. In the first environment the energy consumption in the sector is alone is taken and modeling is done. Here the assumption is energy consumption in the previous periods will have a key role in determining the future consumption. In reality the society will demand more energy than that has been consumed in the previous period. Over period of years in India the accessibility of electrical energy has reached the remote areas of country. Hence the people are able to consume whatever that has been supplied and in turn the society will become more energy intensive year after year. To address this phenomena an ARIMA (P, D, Q)(p, d, q) model is used. In the second environment the urbanization index (which is defined as ratio of number of people living in urban areas to total population.) and literacy rate (which is defined as


ratio of number of people who are literate to total population) are considered in isolation along with electrical energy consumption. Since both variable will affect domestic electrical energy consumption. Urbanization will provide 100% accessibility to electrical energy and literacy rate will also affect domestic electrical consumption since person who is literate will use electrical appliances more. To address the impact of these two variables along with assumption of in the first method two transfer function models have been designed. In first model a transfer function model with electrical energy consumption and urbanization index is attempted and in the second model a transfer function model with electrical energy consumption and literacy rate is attempted 3.0 METHODOLOGY

The procedure adopted for ARIMA (P, D, Q) (p, d, q)

Step-I:

The collected data (Domestic electrical consumption per annum in Giga watt hours for last fifty years are plotted against time. It is to check whether the data is stationary or non stationary. If the data is found to be non stationary then suitable differencing is to be done for the data (first differencing or second differencing) so that stationarity is achieved.

Step-II:Once the data is made stationary then calculate SPAC (Sample partial auto correlation coefficients) and SAC(Sample auto correlation coefficients)for the above to know the auto regressive order(P) and moving average order(Q).

Step-III:Then apply ARIMA (P,D,Q) to the data in any standard computer package and calculate the Nt s (The Nt s will be one of out put of the package results.)

Step-IV:Check for the Nt s whether they are exhibiting

any structure by verifying their SPAC and their SAC.Step-V: If the Nt s SPAC and SAC do not exhibit any structure then stop the process here it self then go to step-7 or if they exhibit any structure move to step-6

Step-VI:If the Nts SPAC and SAC exhibit any structure then convert the Nt s into Ats by applying Arima (p,d,q) process on the N t s and the value of “p” will be known from the Nt s SPAC and the value of “q” will be known from the Nt s SAC. Then move to step-7 Formulate the final equation of Arima (P,D,Q)(p,d,q)

Step-VII:Then using values of coefficients calculate final expected Domestic electrical consumption 2nd differenced value

Step-VIII:From the expected domestic electrical consumption 2nd

differenced value it can be converted in to actual expected domestic electrical consumption value

Step-IX:Now calculate the error of the above model by taking the differences between the expected values and actual values

Step-X:For the errors of model the mean squared value and also the test statistic given by the equation S= mΣr2

aa

Now for these errors calculate in step-9, verify whether they are white noise or not. For this for errors the SPAC (Sample partial auto correlation coefficients) and SAC (Sample auto correlation coefficients ) are calculated from lag1 to lag10. If they are white noise then the model is adequate and correct.

Step-XI:Now using the above model forecast the future values of Expected Domestic Electrical consumption. For this purpose the Ats future values are required and these values are also calculated by using either Arima process or any other extrapolation method

3.2 Second method:The procedure adopted for transfer function model as follows

Step-I The procedure adopted for the first method is repeated for both dependant variable(domestic electrical consumption) and for independent variable(that is for urbanization index or for literacy rate) up to step-7in the first method .After


completing above process it yields the pre whitened errors series for both dependent variable and as well as for independent variable.

Step II:

After the process of pre – whitening as stated in step-1 above It yields two time series βt, αt, (βt, αt, are pre whitened series related to Xt (domestic electrical consumption), Zt (that is for urbanization index or for literacy rate)) .

Step III:

Calculate the cross-correlation coefficient between βt and αt

Step IV:

Compute Variances of βt, αt, and impulse transfer function weights between βt and αt from lag0 to lag10, Impulse transfer coefficient weight= Cross correlation coefficient ×Variance ratio Where variance ratio= Variances of βt/ = Variances of αt

Step V:

Plot impulse transfer weights on graph with (X-axis: Lags and Y-axis: values of impulse transfer weight).

Step VI:

Using the rules that define the values of b, r, s. Determine from the graph the values of b, r, s.

Step VII:

Formulate the transfer function and calculate the expected initial steel production first differences Xt value for all t

values using the transfer function and the form of initial transfer function is

Xt zt-b

= (B)

Step VIII:

Determine the value of Nt = (Xt - t)

Step IX:

For Nts determine suitable ARIMA model so that from Nts we can determine At(Ats are the noise of Nts)Step X:

Check whether At follows white-noise.

Step XI:

Combined the initial transfer function and noise model to formulate the final transfer function model.

Step XII: If Ats are white noise then using ARIMA model of statistical calculate auto correlation of A ts, which are raa(k).,And also calculate the cross correlations between A ts and pre whitened input series

Step XIII: Calculate value of Q by using the following formula.

Q = m ( m + 2 ) ∑=

k

k 1

(m - k)-1rαa2 (k)

Where m = length of series k = Number of lags taken p = Partial auto- correlation order of noise model. q = auto - correlation order of noise model. rαa(k)= auto-correlation value of at at lag k.

Step XIV: Check the calculated Q value with table Q (in Chi-square distributed table at (k-p-q)degree of freedom). If calculated Q< table Q , then model is good fit.

Calculated Q value should be lower then table value for chi-square distribution at (k-p-q) degree of freedom at all probabilities. Here the value of mean squared error is also taken another measure for goodness of fit. If the forecasted values by the models satisfy the above two tests then the models can be utilized future projection for a short term period Then the fit is good fit.

Step-XV:

Now using the final transfer function model calculate expected 2nd differenced domestic electrical consumption values and convert them into expected domestic electrical consumption values.

Step-XVI:

Now calculate the error of the above transfer function model by taking the differences between the expected values and actual values

Step-XVII:


Now for these errors calculate in step-16 verify whether they are white noise or not. For this for error the SPAC (Sample partial auto correlation coefficients) and SAC (Sample auto correlation coefficients) are calculated from lag1 to lag10. If they are white noise then the model is adequate and correct

4.0 Data sources: The data for the present work has been collected from the records of Department of Economics Andhra University Visakhapatnam India. The data dates from 1950 to 2009 covering nearly 59 observations

5.0 ANALYSIS OF DATA:

ARIMA (P, D,Q)(p, d, q) method is applied for Domestic electrical consumption of India . The data (Domestic electrical consumption of India) is plotted against time as shown below in figure-1

X- Axis: 1950-1960- 1970- 1980 - 1990- 2 000- 2009.Y-axis: values of domestic electric consumption in

G.W.Hr.

It is found that the data is not stationary and hence stationarity is achieved by taking 2nd differences and this 2nd

difference is plotted against time in the figure -2.

Figure-2 X- Axis: 1950 – 1960 – 1970 – 1980 - 1990 - 2000 - 2

009. Y-axis : 2nd difference of Domestic electric consumption

in G.W.hr.

Since near stationary is achieved for this 2nd differences data as per the methodology the SPAC (Sample partial auto correlation coefficients) and SAC (Sample auto correlation coefficients) are calculated from lag1 to lag10. They are plotted in figure-3 and figure-4.

From these plots it is clear that ARIMA (5, 0, 3) model is suitable for 2nd differences of domestic electrical consumption. And accordingly the ARIMA (5, 0, 3) is applied to 2nd differences of domestic electrical consumption and then from the results the following equation is arrived EquationNo-1

t= Nt

The values of Nts are automatic output from the computer package used for Arima modeling and now it is necessary to verify whether the Nts are exhibiting any structure or else they are white noise. For this purpose again for Nts the SPAC ( Sample partial auto correlation coefficients ) and SAC( Sample auto correlation coefficients ) are calculated from lag1 to lag10 and these are shown in figure-5and figure-6.

And from these it clear that Nts are exhibiting the structure and hence the Nts are again converted into Ats by applying Arima (4,03)model on Nts.Equation-2


Nt = At

The Ats are also automatic output from the computer package and it is important to verify whether A ts are white noise or not. For this purpose Ats the SPAC (Sample partial auto correlation coefficients) and SAC (Sample auto correlation coefficients) are calculated from lag1 to lag10 and they are shown in figure-7 and figue-8.

By observing the figure-7 and figue-8 it is clear that A ts are white noise and the final equation for Arima (P, D, Q) (p, d, q) is displayed belowEquation-3

t =

t = (-0.164xt-1+0.192xt-2+1.06xt-3+0.216xt-4+0.366xt-5-0.08xt-6-0.056xt-7- 0.32xt-8 + at-0.6256at-1-0.00786at-2-1.37at-

3+1.019at-4-0.2at-5+0.55at-6 - 0.355at-7)

Using the above equation the forecasted value of 2nd differences of Domestic electrical consumption is calculated. These 2nd differences of Domestic electrical consumptions are converted to back to first differences and from first differences again original expected domestic electrical consumption values are calculated. After calculating the forecasted values, now the error of this process is also evaluated and checked whether the errors are white or not. For that purpose again for these errors the SPAC (Sample partial auto correlation coefficients) and SAC (Sample auto correlation coefficients) are calculated from lag1 to lag10 and they are shown in figure-7A and figue-8A.

Now original data values and values arrived from the above method are plotted below in figure-9.

Figure-9

X axis: No of years from 1958 to 2009 Y axis : Series-1 is existing domestic electrical consumption

Series-2 is forecasted domestic electrical consumptionNow as stated in methodology procedure the mean squared error and portmanteaus test are conducted using the respective equations and it is found that the calculated value of from portmanteaus test is less compared to table value at all probabilities. It is presented in table-1.The results of portmanteaus test are presented in Table-1A.

TABLE -1

e=(Xt- t) e2

-0.10327 0.010664-0.06876 0.0047280.105305 0.0110890.726341 0.527571-1.43973 2.0728190.497176 0.247184


0.055331 0.0030610.33356 0.111262-0.49261 0.2426650.065549 0.004297-0.04259 0.0018140.361482 0.130669-0.28591 0.081746-0.1935 0.037443-0.13192 0.0174030.19366 0.037504

0.025468 0.000649-0.36334 0.132013-0.12552 0.0157550.049151 0.002416-0.27922 0.077963-0.49579 0.2458080.456218 0.208135-2.83162 8.0180561.904274 3.62626-2.76354 7.637166-0.53096 0.281914-0.71538 0.511764-1.58649 2.5169532.129343 4.534102-0.81813 0.669334-0.45781 0.209586-0.31579 0.099724-0.32168 0.1034760.327874 0.1075010.660602 0.436394-2.0242 4.097404

0.398517 0.158816-0.42729 0.182577-0.05126 0.0026270.776593 0.6030960.931138 0.867017-3.28388 10.78384

SUM OF e2 = 49.67226

(∑e2)/n=49.67226

43 =0.000248

TABLE-1A

raa r2aa

0.010894 0.0001190.010068 0.000101-0.087512 0.007658-0.054809 0.0030040.034312 0.0011770.049477 0.002448-0.006003 3.6E-05-0.013398 0.000179-0.133659 0.017865-0.060910 0.00371

-0.100985 0.010198-0.079941 0.006391-0.053013 0.00281


-0.105479 0.011126-0.118802 0.0141140.123532 0.01526-0.066556 0.00443-0.168030 0.0282340.157359 0.0247620.018862 0.0003560.167589 0.0280860.010502 0.000110.110885 0.0122950.010824 0.000117-0.052292 0.0027340.077028 0.005933-0.015896 0.000253-0.045620 0.0020810.037055 0.0013730.003756 1.41E-05-0.016558 0.0002740.041704 0.001739-0.058991 0.00348-0.035624 0.0012690.023517 0.0005530.001238 1.53E-06-0.066751 0.0044560.071102 0.0050550.026528 0.0007040.029011 0.000842-0.078896 0.006225-0.001299 1.69E-060.001572 2.47E-06-0.008303 6.89E-05-0.004948 2.45E-05

Σ r2aa = 0.231672

S=m Σ r2aa = 10.42522

Now using the above Arima equation Expected Domestic

electrical consumption is calculated for next 10 years. And

values of Expected Domestic electrical consumption are

displayed in table-4

Transfer function method:

In this method as per methodology as a part of

step-1 it is required to do process of pre-whitening for both

dependent variable(Domestic electrical consumption )and

also for independent variable( which may be urbanization

index or literacy rate). In first method the valuation of A ts

will do the job for pre whitening for domestic electrical

consumption and those values will be utilized. Now for pre

whitening the independent variables (which may be

urbanization index or literacy rate) the procedure is

repeated (what was done in first method up to calculation

of Nts or if necessary up to Ats will be carried.

In case of urbanization index the series is plotted

against time as shown in figure -10 and when the data is not

stationary it is made stationary after making 2nd

differencing. And it is plotted against time in figure-11.

Figure-10

X- Axis: 1950 -1960 - 1970 – 1980 – 1990 – 2000 - 2

009.

Y-axis: values of Urbanization index.


Figure-11

X- Axis: 1950 -1960 - 1970 – 1980 – 1990 - 2000 - 2 009.

Y-axis: values of 2nd difference of Urbanization index.

Same procedure is repeated for literacy rate also

and the corresponding figures are plotted in figure 12 and

figure-13.

Figure-12 X- Axis: 1950 – 1960 – 1970 – 1980 –

1990 – 2000 - 2 009.

Y-axis: values of Literacy rate.

Figure-13

X- Axis: 1950 – 1960 - 1970 - 1980 – 1990 - 2000 - 2 009.

Y-axis: values of 2nd differences of Literacy rate.

Then for this second difference urbanization

index the SPAC (Sample partial auto correlation

coefficients) and SAC (Sample auto correlation

coefficients) are calculated from lag1 to lag10 and these are

shown in figure-12A and figure-13A.

Same procedure is repeated for 2nd difference

literacy rate and the corresponding figure is shown in

figure-14 and figure-15.

Now for 2nd difference urbanization index ARIMA (4,0,3)

model is applied to get its pre whitened series. In case of

2nd difference literacy rate Arima(4,0,3) model is applied to

get its pre whitened series. Now for the pre whitened series

of 2nd difference urbanization index its SPAC ( Sample

partial auto correlation coefficients ) and SAC( Sample

auto correlation coefficients ) are calculated from lag1 to

lag10 and these are shown in figure-16and figure-17.

And both the plots are clearly indicating that the pre

whitened series of 2nd difference urbanization index is

white noise. Now for the pre whitened series of 2nd

difference literacy rate its SPAC ( Sample partial auto


correlation coefficients ) and SAC( Sample auto correlation

coefficients ) are calculated from lag1 to lag10 and these

are shown in figure-18and figure-19.

And both the plots are clearly indicating that the pre

whitened series of 2nd difference literacy rate is white noise.

After the process of pre whitening now it is required to

calculate the cross correlation coefficients between the pre

whitened out put series(2nd difference Domestic electrical

consumption) and pre whitened input series(it may be 2nd

difference urbanization index or 2nd difference literacy rate)

are calculated from lag0 to lag10 and these are shown in

figure-20 and figure-21.

From these cross correlations the impulse transfer weights

are calculated (which are obtained by multiplying the

variance ratio (pre whitened output series variance/ pre

whitened input series variance) with values of cross

correlation coefficients. Now these impulse transfer

weights are plotted against lags (lag 0 to lag 10) as shown

in figure-22 and and in figure 23.From these cross

correlation coefficients are converted into Impulse transfer

function weights by using the formula mentioned in the

methodology.

Figure-22

X-axis: Lag 0 to Lag 10.

Y-axis: Value of Impulse transfer weights

Figure-23

X- axis : Lag0 to lag 10

Y- axis: values of impulse transfer function weights

These two graphs are used to calculate the values of b,r,s

which decide the structure of initial transfer function.

From the graph as shown in figure-22 the values of

b=1,r=2,s=2 and the first initial transfer function between

2nd difference domestic electrical consumption (Xt) and 2nd

difference urbanization index(Zt) is as follows

First Initial transfer function-1

t = Zt-1

t=1.4878xt-1 1.8xt-2 65.68zt-1 163.664zt-2+

127.33zt-3 From the graph as shown in figure-23 the values

of b=1,r=1,s=2 and the second initial transfer function

between 2nd difference domestic electrical

consumption(Xt) and 2nd difference literacy rate (Zt) is as

followsSecond Initial transfer function-2

t = zt-1

t =1.456xt-1+32zt-1+18.3zt-2 44.84zt-3


From these initial transfer functions the values Nts are

calculated using the following formula. For both transfer

function models Nts are calculated using the formula given

below

Nt = (Xt- t)

Now Nts are again converted into Ats and final

transfer function is shown for both models. For this purpose

Nts the SPAC ( Sample partial auto correlation

coefficients ) and SAC( Sample auto correlation

coefficients ) are calculated from lag1 to lag10 and they are

shown in figure-24 and figure-25 (these two figures are

relating Domestic electrical consumption versus

urbanization index)

And also figure-26 and figure-27 (these two figures are

relating Domestic electrical consumption versus literacy

rate)

It is also important to know that the Ats are white

noise or not . For this purpose Ats the SPAC ( Sample

partial auto correlation coefficients ) and SAC( Sample

auto correlation coefficients ) are calculated from lag1 to

lag10 and they are shown in figure-28 and figure-29 (these

two figures are relating Domestic electrical consumption

versus urbanization index)

And also figure-30 and figure-31 (these two figures are

relating Domestic electrical consumption versus literacy

rate)

All these four figures indicate that the Ats are white noise.

Now final first transfer function between 2nd difference

domestic electrical consumption and 2nd difference

urbanization index is as follows

First final transfer function ( 2nd difference of with domestic

electrical consumption with 2nd difference of urbanization

index)

t=

t = (0.313xt-1-0.3xt-2-4.466xt-3-1.54xt-4-0.6768xt-5-65.68zt-1-

281.88zt-2-244.07zt-3+13zt-4+87.363zt-5+ 47.876zt-6+at-

1.319at-1+1.15at-2+0.502at-3-0.12at-4-0.72at-5)

Now second final transfer function is formulated between

2nd difference domestic electrical consumption and 2nd

difference literacy rate

t=

t=(-0.444xt-1+0.966xt-2+2.0xt-3+0.917xt-4+32zt-1+ 79.1zt-

2+47.53zt-3-32.326zt-4-69.17zt-5-28.25zt-6+at-0.426at-1-0.76at-

2-1.077at-3)

From these final transfer functions the values of expected

2nd difference of domestic electrical consumption are

calculated. These expected 2nd differences of domestic

electrical consumption are used to arrive first the first


difference values and from these first difference values the

actual expected domestic electrical consumption values are

calculated. In this way two series of actual expected

domestic electrical consumption values are calculated (first

using urbanization index and later literacy rate) and these

expected values are plotted with exisisting values in figure

-32 and in figure-33.

For both the models the errors are calculated and checked

whether are they are white noise or not. For this for errors

the SPAC (Sample partial auto correlation coefficients) and

SAC (Sample auto correlation coefficients) are calculated

from lag1 to lag10. They are shown in figure-34, figure-35,

figure-36, and figure-37

If they are white noise then the model is adequate and

correct.

For both models portmanteaus tests are conducted and they

are tables in table-2 and in table-3. Both test indicate that

the models are fit enough to give future values.

TABLE-2

Domestic Electrical Consumption Versus Urbanization

Index Portmanea’s Test

rαa r2αa (m-k) r2

αa/(m-k)

0.181 0.032761 41 0.000799

0.148 0.021904 40 0.000548

-0.069 0.004761 39 0.000122

0.037 0.001369 38 3.6E-05

0.015 0.000225 37 6.08E-06

0.181 0.032761 36 0.00091

-0.04 0.0016 35 4.57E-05

-0.107 0.011449 34 0.000337

-0.154 0.023716 33 0.000719

-0.155 0.024025 32 0.000751

-0.029 0.000841 31 2.71E-05

-0.059 0.003481 30 0.000116

0.157 0.024649 29 0.00085

-0.117 0.013689 28 0.000489

-0.122 0.014884 27 0.000551

0.197 0.038809 26 0.001493

0.05 0.0025 25 0.0001

0.285 0.081225 24 0.003384

-0.042 0.001764 23 7.67E-05

-0.025 0.000625 22 2.84E-05


-0.092 0.008464 21 0.000403

-0.112 0.012544 20 0.000627

-0.06 0.0036 19 0.000189

-0.025 0.000625 18 3.47E-05

-0.031 0.000961 17 5.65E-05

-0.045 0.002025 16 0.000127

-0.047 0.002209 15 0.000147

Q=m (m+2) ∑=

k

k 1

(m-k)-1 r2αa (k) = 23.974

TABLE-3Domestic Electrical Consumption

Versus Literacy rate Portmanea’s Test

rαa r2αa (m-k) r2

αa/(m-k)

0.23 0.0529 41 0.00129

0.041 0.001681 40 4.2E-05

0.1153 0.013294 39 0.000341

0.0808 0.006529 38 0.000172

0.193 0.037249 37 0.001007

0.15 0.0225 36 0.000625

0.0928 0.008612 35 0.000246

0.1293 0.016718 34 0.000492

0.0475 0.002256 33 6.84E-05

0.199 0.039601 32 0.001238

0.1217 0.014811 31 0.000478

0.053 0.002809 30 9.36E-05

0.0265 0.000702 29 2.42E-05

0.0419 0.001756 28 6.27E-05

0.081 0.006561 27 0.000243

0.019 0.000361 26 1.39E-05

0.0057 3.25E-05 25 1.3E-06

0.0531 0.00282 24 0.000117

0.0529 0.002798 23 0.000122

0.001 0.000001 22 4.55E-08

0.018 0.000324 21 1.54E-05

0.0344 0.001183 20 5.92E-05

0.0659 0.004343 19 0.000229

0.092 0.008464 18 0.00047

0.046 0.002116 17 0.000124

Q=m (m+2) ∑=

k

k 1

(m-k)-1 r2αa (k) = 13.35281

But looking at figure 32 and figure-33 it is clear that figure-33 is better. And hence the transfer function model with literacy rate is taken for using to evaluate the future values.The existing literacy are growth is taken as a consideration for fixing the future literacy rate values and also the values of Ats are extrapolated for next ten years to provide values for using in transfer function modelThe future Domestic electrical consumption values are from Arima model and transfer function model are presented in table-4.

TABLE-4 Expected Domestic Electrical Consumption

YEAR

ARIMA

(P,D,Q,)(p, d, q)

TRANSFER FUNCTION MODEL-2 Average of

both models

2011 89.29094 104.5598 96.925372012 100.0764 114.3538 107.21512013 102.8055 129.0045 115.9052014 96.70938 143.87 120.28972015 110.9866 153.3796 132.18312016 105.889 173.3114 139.60022017 119.6866 201.3425 160.51462018 109.56 205.6495 157.60482019 105.2229 221.2419 163.23242020 116.9114 283.405 200.1582

In Table -4 in column-4 the expected average domestic electrical consumption from both models is also presented . 6.0 Future scope of study: Domestic electrical consumption can be used in a transfer function model along with literacy rate and urbanization index. REFERENCES:1. R.E.Abdel-aal,A.Z.AL-Garni and Y.N.AL-Nassar

[1997].Modeling And Forecasting Monthly Electric


Energy Consumption in Eastern Saudi Arabia Using Abductive Networks. Energy Vol.22 No.9 pp.911-921.

2. Raymond Y.C.Tse, John Raftrey, Department of building & real Estate, The Hong Kong polytechnic University, Hong Kong, “Income elasticity of housing consumption in Hong Kong: a co integration approach”. Journal of Property research, 1999, 16(2)123-138.

3. Soteris A. Kalogirous [2000].Applications Of Artificial Neural- Networks For Energy Systems. Applied Energy 67(2000)17-35.

4. Pernilalli Holtedahl and Frederick L.Joutz, Department of economics, The George Washing ton university,"Residential Demand in Taiwan” ,January 2000.

5. Eric Girardin, Lucio and Mark P. Taylor, Warwick

Business School, University of Warwick, UK, “Private consumption behavior, liquidity constraints and financial deregulation in France: a nonlinear analysis”. Empirical Economics (2001)25:351-368.

6. Herath M. Gunatilake, Inoka Chandrasena, Chennat Gopalakrishna, University of Peradeniya, Sri Lanka, “The economics of household demand for water: The case of Kandy Municipality, Sri Lanka”. Water Resource Development, vol.17, no.3, 277-277, 2001.

7. Planning commission ;Electricity demand in Indian households published in CMIE papers in2002

8. Geoffrey K.F.Tso, Kelvin K.W.Yaw [2003]. A Study of Domestic Energy Usage Patterns in Hon Kong. Energy 28(2003)1671-1682.

9. Melody Stokes, Mark Rylatt, Kevin Lomas [2004]. A Simple Model of Domestic Lighting Demand. Energy and Buildings 36(2004) 103-116.10. Merih aydinalp, V.Ismet Ugursal, Alan S.Fung [2004]. Modeling Of the Space And Domestic Hot Water Heating Energy – Consumption In The Residential Sector Using Neural Networks. Applied Energy 79(2004)159-17811. Dorte Grinderslev, Kenneth Karsson,Erik Bjosted House hold Demand for electricity and electrical appliances July 200412. Zaid Mohamed, Pat Bodger [2005]. Forecasting electricity consumption in New Zealand using economic and demographic variables. Energy 30 (2005)1833-1843.13. Davood Manzoor (2005) Estimating Residential Energy Demand in Iran14. Jukka V.Paatero and Peter D. Lund [2006]. A model for generating household electricity load profiles. International Journal of Energy Research (2006) 30: 273-290.

15. Runming Yao, Koen Steemers [2005].A Method of Formulating Energy Load Profile for Domestic Buildings in the UK. Energy and Buildings 37(2005) 663-671.16. S.Jebaraj, S.Iniyan [2006]. A review of energy models. Renewable and sustainable Energy Reviews 10(2006)281-311.17. F.Gerard Adams and Yochanan Shachmurove (2008) energy planning of China18. Holger Fabig [2008] Modeling Macro-Critical Energy Sector in Low Income Countries: A General Framework and an Application to Cote d Ivoire. International Monitory Fund .WP/08/156.19. Mohammad Reza Faraji Zonooz , .Z.M. Nopiah, Ahmad Mohd. Yusof and Kamaruzzaman Sopian(2009) (19) A review of MARKAL energy modeling. European Journal of Scientific Research ISSN 1450-216 X Vol.26 No-3(2009) pp 352-36120. Aneta Strzalka, Jurgen Bogdahn and Ursula Eicker [2010]. 3D City Modeling for Urban Scale Heating Energy Demand Forecasting. Stuttgart University of Applied Sciences, Germany.21. Aneta Strzalka, Ursula Eicker, Volker Coors and Jurgen Schumacher [2010] Modeling Energy Demand for heating at City Scale. Stuttgart University of Applied Sciences, Schellingstrasse 24, D-70174 Stuttgart, Germany.


EFFECT OF TEMPERATURE ON MICROWAVE CONDUCTIVITY IN POLYVINYL CHLORIDE

R.MURUGESH*

Lecturer, [email protected],[email protected] Teresa College of Engineering, Thirunavalur, Cell no: 9786521390, 7667080768

ABSTRACT:The dielectric constant, loss tangent and the a.c. conductivity of Polyvinyl Chloride (PVC) at X-band and Ku-band micro frequencies at temperatures ranging from 300C-1000 C have been measured by Von-Hippel method. Our preliminary studies show a peak value of the dielectric constant as 4.73 at 800 C at a Ku-band frequency of 15.9 GHz, while the peak value of the conductivity at the same conditions is 9.89x10-3 s/cm. The results of the temperature variation study of the above mentioned electrical properties at 15.9 GHz are presented. These results are compared with the values obtained for the same sample at X-band frequency of 9.4GHz.Keywords: Microwave, dielectric constant, conductivity, waveguide

1. INTRODUCTION

Vinyl chloride and its polymers are

among the very oldest pure organic substance.

The monomer was first obtained by LieBig in

1835 and characterized by his student, Regnault

[1]. Properties of the monomer, techniques of

polymerization and polymer properties and

applications have been treated in several

monographs [2-4]. In the recent years electrically

conducting polymers represent the important

research area with diverse scientific problems of

fundamental importance with the potential for

commercial applications [5]. Polymers, which

show high electrical conductivity, have been

synthesized successfully in the last two decades

[6]. The theory of doping led to dramatic

increase in the conductivity of conjugated

polymers. The doping of conjugated polymers

generates high conductivities primarily by

increasing the carrier concentration. This is

accomplished by oxidation or reduction with

electron acceptors or donors respectively.

The polymer is oxidized by the

acceptors removal of an electron, thereby

producing a radical caution (or hole) on the chain

of the hole can overcome the coulombic binding

energy to the acceptor onion with thermal energy

or, at high dopant cones using screening of the

coulombic change of the onions, it moves

through the polymer and contribute to

conductivity. The process of doping is

somewhat misleading since dopant cones are

exceptionally high compared to those commonly

encountered in semiconductors. Thus a doped

system would be more appropriately described as

a conducting charge transfer complex rather than

a doped polymer. This paper contains discussion

of dielectric properties of polyvinyl chloride at

microwave frequencies. The dielectric

properties are of interest fortwo main reasons. It

is because of materials used as dielectrics and


insulators and the interpretation of dielectric

behaviour in terms of molecular structure is a

long-term objective. Properties cannot usually

state as constants but must be expressed as

functions of temperature and frequency.

This was well illustrated by solid

models constructed by Reddish [7]. Polymers by

virtue of their light weight and ease of

fabrication and low cost have replaced metals in

several areas of applications. Nowadays

electrically conducting polymers, which are

stable even in doped form, have been prepared

[8]. Measurements of dielectric parameters of

pure and iodine doped PVC films at X-band and

Ku-band microwave frequencies have already

been reported [9]. In the present work we report,

the effect of temperature on the electrical

properties of pure and iodine doped PVC at X-

band and Ku-band microwave frequencies.

2. EXPERIMENT DETAILSPoly Vinyl Chloride in the form of fine powder (Molecular Weight 48,000) obtained from Chemika was used in the experiment. Rectangular blocks of the PVC powder with wave guide dimensions were made using a specially made dye and press arrangement. The packing density of the doped and undoped PVC was measured as 0.979 gm/cc. Iodine doping was done by Chemical vapour Deposition (CVD) method [10]. For temperature variation the solid dielectric cell in the microwave bench was covered with a mica sheet and over which heating element was wound. White cement was applied over the heating element with their free ends protruding out. Provision was made for inserting the probe of temperature indicator. A digital temperature indicator with mineinsulated

flexible thermocouple as the probe with an accuracy of ± 1oC was used for temperature measurements. Heating was done by a variac system. When the temperature of dielectric cell attains the stable state, temperature measurements were done. As a sample is in good contact with the metallic wave-guide, it is assumed that the temperature of the waveguide and the sample is the same. To avoid the heat conduction to other parts of the waveguide, a thin layer of thermal insulator was introduced between the dielectric cell and other parts of the bench. Adjusting the a.c. voltage by the variac to the required level can be maintained the temperature. The dielectric constants were calculated using the relations,

Dielectric strength,

ε′=(λ0/λc)2+(λ0/λd)2(1+ (αdλd)/2π)

Dielectric loss,

ε′′=1/π(λ0/λd)2αdλd Loss tangent,

Tanδ=ε′′/ε′Dielectric Constant,

εr =ε′(1+tanδ)Where, λ0 -Free space wave

lengthλd -Wavelength in the dielectric αd

-Attenuation in the dielectric, and

Conductivity was calculated using,

σ = ωε0ε’’ s/cm

Where, ω - Angular frequency; ε0 –

Permittivity of free space (80854x10-14F/cm)

3. RESULTS AND DISCUSSIONDielectric constant and the conductivity of undoped PVC at the X-band frequency of 9.4 GHz at various temperatures ranging form 300C to 1000C were measured, and the results are shown in Fig.1. The dielectric constant shows a peak value of 2.12 at 300C and a minimum value

of 2.00 at 1000C, while the conductivity shows a peak value of 3.88x10-3s/cm at 300C and a minimum of 3.54x10-3s/cm at 1000C. The result of the dielectric constants and conductivity at the Ku-band microwave frequency of 15.9GHz for


the undoped sample at different temperature are shown in Fig. 2. The result show a peak value of dielectric constant as 4.73 at 800C and a


minimum value of 4.16 at 700C, while the conductivity shows a maximum value of 9.89x10-3s/cm at 800C and a minimum value of 9.09x10-3s/cm at 400C. The result of the peak values iodine doped PVC at 15.9GHz is given ion table1 along with the result of undoped sample.

The d.c conductivity studies of PVC by [11-12] etal show a value of the order of 10-10

to 10-12

s/cm. Dielectric and loss tangent study of pure and doped PVC at 9.4GHz microwave frequency [13] gives the value of 3.4 for the dielectric constant by a technique developed by [14] to study thin films

4. CONCLUSIONIn the present study at 9.4 GHz, 4.26 at room temperature. This small variation may be due to the reason that the Von- Hipple [10] method gives less

accurate measurements as claimed by due [13] and difference in the packing of the molecules in thin films and the method used in our studies. The a.c


conductivity in the present study has been estimated at 9.4 GHz as 3.4x10-3s/cm and at the 15.9GHz as 9.25x10-3s/cm at room temperature. The value of the conductivity measured a9.4GHz more or less agrees with the value obtained by Dube, the peak value of dielectric constant (4.73) and conductivity (9.89x10-

3s/cm). It has been observed that at 800C at Ku-band frequency of PVC [12] gives a change of slope in the logσ versus 1000/T grape at 840C, predicting a phase transformation at 840C. The sudden jump value of dielectric constant and conductivity observed at 800C is in agreement with their observation of phase transition at 840C.

5. REFERENCE

1."The science and Technology of polymer film" edited by Origle’J.Sweeting.Vol IIWiley interscience (1971).

2. Franz Kainer,"Polyvinylecholorid and Vinylchlorid Mischpolymerisate",springer,Berlin,(1951)

3. C.E.Schildknecht,"Vinyl and Related Polymers",Wiley,New York,(1952).

4.RudolfPummerer(Ed.),ChenischeTexilfasern,Film,andFolien,FerdinandEnke,Stuttgart,(1953).

5. Jane E.Former & Ronald R.Chance, "Electrical and Electronic Properties of polymers", (Kroschwitz Ed) Jone Wiley $ Sons (1985).

6. Ramakrishnan S.Resonance,Nov(1997).

7.Reddish,W.,Trans.Far.Socb,459(1950).

8. Bakshi A K,Indian J. Chem.31A (1992)291

9. Natarajan R, Ph.D Thesis entitled Study of dielectric properties of film using Microwave techniques,IIT, Delhi(1975).

10. Von-Hippel A.R, "Dielectric materials and applications" new York Wiley) (1961).

11. Indar Mohan Talwar & Bhawalkar D.R, Indian J.Pure $ Appl. Phy,7(1969)681. 12. Meenakshi Maruthamuthu,Selvaraj M & Annadurai S, Bull.Mater.Sci,16(4)(19930273)13. Dube D.C, Bull.Mater.Sci.6(6)(1984)1075. 14. Dube D.C & Natarajan R, J.Appl. Phys. 44 (197304927.


A NOVEL APPROACH FOR SECURE DATA STORAGEIN DESKTOP DATA GRID

DR. G.VASANTH1 , DR B.S.VISHWANATH 2, MR. T. SUDALAI MUTHU 3 MRS. SAVITA HARKUDE4

1professor And Hod, Department Of Information Science And Engineering2professor, Electronics And Communication Engineering

3lecturer, Department Of Information Science And EngineeringBtl Institute Of Technology,Bangalore, Karnataka, India.

4sir M V Institute Of Technology Bangalore

ABSTRACTVolunteer Computing is becoming a new paradigm because of the enormous storage potential that may be achieved at a low cost. However, the Data Grid” depends on a set of widely distributed and untrusted storage nodes, therefore offering no guarantees about neither availability nor protection to the stored data. These security challenges must be carefully managed before fully deploying Data Grids in sensitive environments. In this paper we propose a cryptographic protocol able to fulfil the storage security requirements related with a generic Desktop Data Grid scenario. The proposed protocol uses three basic mechanisms to accomplish its goal: (a) symmetric cryptography and hashing, (b) an Information Dispersal Algorithm and the novel (c) “Quality of Security” (QoSec) quantitative metric.

KeywordsVolunteer Computing, Data Grid, Cryptography, Hashing, Information Dispersal Algorithm, Petaflops, Exabyte.

1. INTRODUCTION

The Desktop Data Grid is a specific type of distributed system, where shared resources (processor or storage) are provided in a volunteer fashion by the participants. These environments potentially provide commodity resources not only for CPU-intensive tasks, but also for applications that require significant amounts of memory, disk space and network throughput. However the potential computing power of volunteer systems is even greater: The number of Internet connected PCs is projected to reach 1 billion by 2015 [5], which means several PetaFLOPS of computing power and a storage capacity (around one Exabyte) able to exceed the one provided by any centralized system. Therefore, it is not surprising that nowadays a lot of ongoing effort is targeting Volunteer Computing as a new paradigm for both the Computational and the Data Grid.Tapping unused PC capacity promises efficiency and reliability comparable to traditional storage attached networks (SANs) minus the associated costs and management headaches that come with large-scale SAN deployments. This emerging approach is quite attractive for small to midsize businesses that would like to store large amounts

of data in their premises, or even to backup critical data [4].

However, for the successful deployment of the Volunteer Computing paradigm into any data storage-related scenario (this particular application field will be referred to as the Desktop Data Grid throughout this paper), the security aspect arises as a critical parameter.Figure-1 shows the basic elements and interactions that usually take place in a Desktop Data Grid. Grid Users interact with the Desktop Grid Service, either generating or using information (data producers and/or consumers). Also there is a set of security-related services grouped under the generic term Door node, which are the servers performing all the authentication and authorization processes for involved entities (users and resources). On the other hand, the Project Server (e.g a BOINC server) manages the metadata and the data itself prior to its stage to the pool of Volunteer Storage Clients (VSCs) which provide storage for the Desktop Data Grid. In most volunteer installations, all communication is initiated by the VSCs, in order to facilitate firewall and NAT traversal.


Figure 1: A Typical Desktop Data Grid.As a preliminary step in designing the security mechanisms for a generic Desktop Data Grid, the first part of this paper extends the analysis framework to identify its particular security issues and challenges. The second part of this work relies on the requirements obtained via the security analysis to propose a cryptographic protocol able to protect the data both on the wire and at rest (in VSCs),applying three basic mechanisms(i)cryptography, (ii) data fragmentation, and (iii) a novel quantitative security evaluation metric.The rest of this paper is organized as follows. Section 2 analyzes the security issues of the typical Desktop Data Grid by applying an extended framework. Section 3 introduces the proposed security protocol and analyzes the security guarantees it provides to Desktop Data Grids, while section 4 shows how data assurance can be improved by using the proposed mechanisms. Finally, section 5 summarizes our conclusions and outlines future work.

2. SECURITY ANALYSISFrom the point of view of the Desktop Data Grid being studied, its subsystems may become attacked in several ways; for the purposes of our research, the framework proposed by [14] and extended in [9] will be used to find out the main concerns related to the security of its data and metadata. The first step in our analysis is to identify the elements that play a security-related role in a generic Desktop Data Grid: 1. Players:Three data readers/writers are involved (i) theVolunteer Storage Client; (ii) the Project Server, which also implements a metadata service that gathers information about stored files; and (iii) the WAN links (referenced also as the “wire”) conveying information between VSCs and Project Servers. For this framework the Grid Service and the Grid user are not considered as players.2. Attacks:The generic attacks that may be executed over the Desktop Data Grid are related with (i) Adversaries on the wire; (ii) Adversaries on the infrastructure servers (Project Server or door node); (iii) Revoked users on the door node; and (iv) Adversaries with full control of the site services (Volunteer Storage Clients). Each one of these attacks may result in data being leaked, changed or even destroyed.

3. Security Primitives:Two security operations take place into a typical Desktop Data Grid: (i) when Grid users are authenticated by door nodes before accessing a Grid Service; and (ii) when the door node also authorizes Grid users.4.Trust Assumptions: We have considered that in general (i) VSCs have full control over the data stored on them (which means that a malicious VSC may leak, change or destroy it); (ii) data are encrypted on the wire using a secure channel (i.e. SSL); (iii) data are unprotected on the VSCs’ disks, but protected at the Project Server; (iv) infrastructure servers are trusted; and finally (v)the VSC’s client software is trusted onlywhen coming from a Project Server (i.e. digitally signed code distribution).5.User inconvenience:Desktop Data Grids only require the installation of a client-side software at the volunteer nodes, and these nodes are only expected to provide storage space, therefore keeping processor overhead to a minimum.

2.1 Results of the security analysis

Using the elements identified in section 2.1 and after applying the analysis framework from [9] we have obtained a set of security issues which have been summarized in table1. Results are categorized by possible attacks (main columns) and types of damage – the Leak, Change, Destroy sub-columns. Cells marked with a “Y” mean that the system (row) is vulnerable to the type of damage caused by this particular attack. Cells marked with a “N” mean that the attacks are not feasible or may not cause a critical damage.

Table 1Summary of security issues related with Desktop Data Grids

Our conclusion from the previous security analysis is that there are two main security concerns in a Desktop Data Grid. The first issue is related to the Infrastructure Servers, in particular when revoked users have colluded with door nodes. We believe, however, that it is feasible to extrapolate the proposed mechanisms on validation for Computational Grid’s users [10]; therefore these attacks do not generally represent a major security risk. On the other hand, we also identified a second group of critical vulnerabilities related with the VSCs, which are considered untrusted and heterogeneous


(different hardware and software configurations), thus offering a non-uniform level of protection to the stored information. Consequently, the rest of this paper is focused on a security protocol designed to protect a Desktop Data Grid environment from adversaries colluded with untrusted Volunteer Nodes.

3. SECURITY PROTOCOLA security mechanism conveying sufficient data protection for the Desktop Data Grid should provide a fair balance between performance overheads and data confidentiality, integrity and availability. With these quirements, we designed a security protocol that achieves its goals based on three techniques. The first is an application of symmetric cryptography, whereby, before staging data into the VSCs, the Project Server encrypts the data file with a symmetric master key composed of the file’s hash appended to a nonce (i.e. a timestamp). With this technique, we provide confidentiality (as the VSCs are unaware of the encryption key) and integrity (via the hash function). As an optional authorization feature, when a Grid Client is retrieving a file, we propose the use of a public-key cryptosystem for encrypting the symmetric master key, thus ensuring that only the legitimate owner of the corresponding private key is able to decrypt the symmetric master key and therefore the requested file.

A second technique, the Information Dispersal Algorithm, provides high availability and assurance for the Desktop Data Grid by means of data fragmentation. In a fragmentation scheme [13], a file f is split into n fragments; all of these are signed and distributed to n remote servers, one fragment per server. The user then can reconstruct f by accessing m arbitrarily chosen fragments (m d” n). The fragmentation mechanism can also be used for storing long-lived data with high assurance: f is fragmented in just a few pieces (small m) and dispersed to a large number of nodes (large n) (a replication-like approach). When fragmentation is used along with encryption, data security is enhanced: An adversary colluded with the VSCs has to compromise and retrieve m fragments of file f, and then has to break the encryption system K. For the proposed protocol, the dispersal algorithm is based on the Reed-Solomon erasure code [12]. This means that when a device fails, shuts down or is compromised, then the system recognizes this event, as opposed to an error, where a device failure is manifested by storing and retrieving incorrect values, that can only be detected using codes that are embedded within the data We selected the Reed-Solomon erasure code for our protocol taking into consideration its successful applications in several fault-tolerant systems.Finally, a third security technique takes into account the heterogeneity of the Desktop Data Grid’s nodes.

Our premise is that if subsets of VSCs with analogous features are clearly identified and their security level quantified, then any user of the Desktop Data Grid is able to request storage space only at nodes fulfilling some minimum guarantees or Quality of Security (QoSec) level. To this end, we propose a policy-based approach, where (i) each VSC is associated with a Policy modeling its securitybehaviorand,(ii)anevaluation methodology (presented in section 3.1) for computing a quantitative QoSec metric associated with such policy. we believe that an analogous approach for the Volunteer nodes is quite feasible. This approach does not add considerable overhead to the protocol, mainly because the evaluation can be performed by the Project Server at any time (not necessarily in real-time) as soon as the VSC’s policy has been retrieved. Once evaluated, a numeric QoSec factor can be associated with every VSC; moreover, a proprietary set of reference levels can be established to group these discrete values into a major set, for example (L0, L1, L2, L3) which represents four different QoSec. Any Grid Client storing data will request a minimum QoSec to be fulfilled by the VSCs. An important challenge with this proposal is the definition of VSC’s security policy itself.

Ongoing research in the context of the CoreGRID NoE [3] is moving towards defining the security policy associated with a generic storage element, and as members of this group we plan to implement such a policy on the Desktop Data Grid. A second challenge is the VSC security policy’s audit process, because the proposed evaluation would be pointless if the storage node is not honoring its policy. However, there are some “controlled” scenarios where this audit can be performed with confidence; for example, a Desktop Data Grid overlapping different institutions that are members of the same Grid Policy Management Authority (PMA), that is, clients using in their installations digital certificates issued by a Certification Authority that has passed a quite strict accreditation process, enabling it to inter-operate in Grid projects only with institutions members of the same PMA.The following subsection presents important details of the proposed evaluation mechanism, called the Reference Evaluation Methodology (REM). Subsections 3.2 and 3.3 then describe the interactions taking place when a Grid client writes and reads a file using the proposed protocol. When presenting the proposed protocol it must be taken into account that (i) all communication taking place among theinvolved partiesis conveyed throughencrypted channels (i.e. via SSL tunnels) and, (ii) communication is initiated by the VSCs. Finally, file deletion is unaffected by our protocol and follows the same principles from [2], where the Project Server creates an specific message for this operation (after authenticating and authorizing the requestor).


3.1 The Reference Evaluation MethodologyThe core of our proposal to quantify the Quality of Security associated with a Volunteer node is the REM methodology introduced in [7], which has been successfully applied in other security-related environments, i.e. Public Key Infrastructures (PKI). Basically the REM takes a policy to evaluate, formalize it to ease the further evaluation step over an homogeneous metric space, uses a set of reference levels to apply a distance criteria and finally obtains a number that corresponds to the policy’s security level. The rest of this section presents an illustrative example that shows basic use of the REM methodology to evaluate a single security provision for a VSC. To evaluate the full security policy, this same procedure should be applied to each one of its provisions. Interested readers that would like to take a closer lookat the details and formalisms behind REM should refer to [6].Step 1: Policy formalization. In REM, a security policy is formed via a set of individual security provisions, so let us suppose that a VSC’s security policy contains the provision called Px =”User log-on mechanism”. To formalize a provision it is necessary to state the possible values that are allowed for it to take into the VO. For our example let us suppose the following values for Px (from least to most secure) “None, Password, Smartcard, Biometric”, this means that Px has a cardinality of 4.Step 2: The evaluation technique. If for a particular VSC the user has to login via Smartcard, then the provision Px takes as value the vector (1, 1, 1, 0), which means that it is more secure than the Password-only mechanism (vector (1, 1, 0, 0)), but less than Biometry (1, 1, 1, 1). Using REM’s nomenclature, this implies a Local Security Level or LSLx = 3. It is easy to see that after evaluating a whole formalized security policy for a particular VSC, we will have a set of vectors or in other words, a matrix Mx.To obtain the Global Security Level (QoSec) for this particular VSC, it is necessary to have a new zero-matrix called Mφ which contains only 0’s as its elements. Notice that Mφ represents the least secure VSC that can be found into a Desktop Data Grid installation. The QoSec in our example is defined as the Euclidean distance between Mx and Mφ, that is:

d(Mx,Mφ) = σ(Mx “Mφ,Mx “Mφ)....... (1)Where: σ(Mx”Mφ,Mx”Mφ) = Tr (Mx “Mφ)(Mx “Mφ)T ............(2)Obviously the REM technique considers more complex scenarios,where for example different provisions have different security weights (some

provisions are more important than others) and even different cardinalities [6]. 3.2 Grid Client writing a file

Any authenticated and authorized Grid Client must perform the following interactions when writing a file to the Desktop Data Grid with the proposed protocol: 1. The Producer P, generates the clear-text data f and sends it to the Project Server.

2. On the Project Server, the cryptographic hash H(f) is computed, concatenated with the nonce T and used as the symmetric key to encrypt the file f. That is: EH(f)|T (f). .............................(3)This master encryption key is stored as part of the associated metadata into the Project Server. Also a numeric QoSecf, representing the Security Level requested by the Producer, is associated with the data f.

3. The encrypted data from (3) is fragmented with the IDA and the Project Server stages it for downloading only to those Volunteer Storage Clients fulfilling the minimum requested QoSecf . Notice that the Project Server has previously computed offline the QoSec associated with each participating VSC.

4. The VSCs queries the Project Server and if required to do so will download and store the encrypted fragments.

3.3 Grid Client reading a fileWhen an authenticated and authorized Grid Client

requests a file from the Desktop Data Grid, the following protocol is performed:

1. The Project Server requests from the VSCs to upload the fragments required to rebuild the file solicited by the user.

2. The VSCs query the Project Server and if required to do so, will upload the encrypted fragments.

3. The Project Server defragments and decrypts with H(f)|T the uploaded data from (3) to rebuild the file f.

4. Optional: If required, instead of directly decrypting (3), the Project Server may re-enforce authorization by encrypting the stored master key H(f)|T with the public key of the Consumer (PubC), that is:

EPubC (H(f)|T).................... (4)5. The data consumer retrieves the clear-text file f

or, if still encrypted (previous step), then (i) with its private key obtains H(f)|T from (4) and, (ii) uses this master key to obtain the clear-text file (f ) from (3), also verifying f ’s integrity and freshness.

The following section presents via an analytical model the strong relationship between the security of the data at rest, the QoSec of the Volunteer Storage


Client and the number of fragments required to rebuild the original file.

4. QOSEC AND DATA ASSURANCE FOR PRODUCTION GRIDS

We use the model presented in [11] to evaluate the assurance of a file stored in a Desktop Data Grid according to the proposed protocol. Assurance is defined as the probability that the cited file has not been compromised under the assumption that the system was the target of a successful attack. Formally speaking, we assume that a Desktop Data Grid with N-Volunteer Storage Clients is built in such a way that the same configuration (and therefore the same security level) is present in at most λN volunteer

nodes, where » “ R and 0 < » < 1. A potential

attacker could then compromise at most »N servers, because the same vulnerabilities are to be found in all of them.

Therefore, it makes sense to state that a Desktop Data Grid can be considered secure if it is composed of heterogeneous volunteer computers, where of

course » H” 0. According to the authors of [11], this latter fact can be represented by the conditional

probability of the event “at most m-1 VSCs, each storing a fragment of f, are down or compromised” given the event “the Desktop Data Grid has been attacked”; this probability is known as the distribution assurance Aφ(μ) and for a dispersal algorithm μ applied over a file f will be denoted by (5).

————————(5)According to the analytical model (5), it is more

likely to find subsets of λN Volunteer Nodes sharing the same configuration. Thus, a better data assurance

is obtained with highly secure (QoSec ’! “) or

heterogeneous (λ ’! 0) volunteer nodes. Therefore:

QoSec ”” 1 Ò! QoSec = k ë ë

————————(6)We can rewrite (5) in terms of the proposed QoSec with the factor k = 1:

————————(7)The relationship between the assurance for the data at rest and the computed QoSec can be obtained by plotting expression (7), just as seen in figure 2. For the evaluation, we consider a Grid with 100 VSCs

(N=100), implementing file fragmentation into 15 pieces (n=15).

The figure shows that, despite the small difference

between the computed QoSecs, CERN’s nodes

achieve the greatest data assurance (H” 1) with the smallest number of fragments.

Figure 2. Relationship between data assurance and QoSec. The horizontal axis represents the number of fragments required to retrieve from the VSCs to rebuild the original file.

On the other hand, a Grid installation fulfilling only the minimum security requirements (QoSec=4.24) requires the retrieval of more fragments from the volunteer nodes in order to fully defragment the stored file with the same data assurance. The results presented in this section are a first approach to compute the Storage Element’s QoSec using a subset of provisions from a Certification Policy; however, in a realworld installation we should evaluate a comprehensive Security Policy specifically built for this resource.

5.CONCLUSIONSProviding security to the Desktop Data Grid is critical for its deployment in production environments, where its full potential can be developed in terms of storage (i.e. for backup and caching) and computing power. However, it is mandatory to establish guarantees over the security provided to the data and metadata being managed in these systems in order to avoid unnecessary risks. Towards this goal, we presented a proposal for enhancing the overall security of the Desktop Data Grid, with a protocol that makes use of three basic


techniques (cryptography, data fragmentation and a quantitative security metric) to cope with untrusted Volunteer nodes participating in these environments.

Even though the use of cryptography and Information Dispersal Algorithms is not new for securing data and metadata in distributed systems, the main contribution of our research consists of enhancing these mechanisms with the adoption of a numeric Quality of Security -QoSec- level representing the guarantees offered by the Volunteer Storage Client to the Grid User’s stored data. Applying an analytical model we have shown the strong relationship between the data assurance, the number of fragments required to retrieve a file, and the volunteer node’s QoSec. On the other hand, we have to consider an important limitation inherent to Desktop Data Grids: the VSC’sbandwidth. To improve computational performance over the stored data, future research will consider executing code at the VSC directly. An important security challenge here are the fragmented and encrypted data, because for the computation to take place it might be necessary to retrieve fragments from other VSCs and, of course, decrypt them afterwards (which means disclosing the corresponding encryption key to the VSC). A first approach could be to do this only on those VSCs with a “high QoSec” (e.g. those using cryptographic hardware).Despite the usefulness of the QoSec concept, it is important to avoid abuses from Grid Clients (such as a client requesting always a high QoSec even for non-critical data). To this end, future research should provide the Project Server with a QoSec validation mechanism, so that QoSecs are also assigned to Grid Clients or to the data objects themselves, to decide the maximum security level that should be requested from VSCs. This process can be seen as “QoSec based authorization”.

REFERENCES[1] Classic ap profile version 4.03.

http://www.eugridpma.org/igtf/IGTF-AP-classic- 20050905-4-03.pdf, 2005.

[2] Berkeley open infrastructure for network computing. http://boinc.berkeley.edu/, 2007.

[3] Coregrid network of excellence. http://www.coregrid.net, 2007.

[4] Tapping pc resources for storage needs. http://www.internetnews.com/storage/article.php/3720931, 2007.

[5] D. P. Anderson. Boinc: A system for public-resource computing and storage. In R. Buyya, editor, GRID, pages 4–10. IEEE Computer Society, 2004.[6] V. Casola, A. Mazzeo, N. Mazzocca, and V.

Vittorini. A policy-based methodology for security evaluation: A security metric for public key infrastructures. Journal of Computer Security, 15(2):197–229, 2007.

[7] V. Casola, R. Preziosi, M. Rak, and L. Troiano. A reference model for security level evaluation: Policy and fuzzy techniques. J. UCS, 11(1):150–174, 2005.

[8] R. Housley, W. Polk, W. Ford, and D. Solo. Internet X.509 Public Key Infrastructure - Certificate and Certificate Revocation List (CRL) Profile. RFC 3280 (Informational), 2002.

[9] J. Luna et al. An analysis of security services in grid storage systems. In CoreGRIDWorkshop on Grid Middleware 2007, June 2007.

[10] J. Luna, M. Medina, and O. Manso. Using ogro and certiver to improve ocsp validation for grids. In Y.-C. Chung and J. E. Moreira, editors, GPC, volume 3947 of Lecture Notes in Computer Science, pages 12–21. Springer, 2006.

[11] A. Mei, L. V. Mancini, and S. Jajodia. Secure dynamic fragment and replica allocation in large-scale distributed file systems. IEEE Trans. Parallel Distrib. Syst., 14(9):885–896, 2003.

[12] J. S. Plank. A tutorial on reed-solomon coding for faulttolerance in raid-like systems. Technical Report CS-96-332, University of Tennessee, Department of Computer Science, 1997.

[13] M. O. Rabin. Efficient dispersal of information for security, load balancing, and fault tolerance. J. ACM, 36(2):335–348,1989.

[14] E. Riedel, M. Kallahalla, and R. Swaminathan. A frameworkfor evaluating storage system security. In D. D. E. Long, editor, FAST, pages 15–30. USENIX, 2002.

[15] M.W. Storer, K.M. Greenan, E. L. Miller, and K. Voruganti. Secure, archival storage with potshards. In FAST’07: Proceedings of the 5th conference on USENIX Conference on File and Storage Technologies, pages 11–11, Berkeley, CA, USA, 2007. USENIX Association.


STATISTICAL MODELING OF GTAW FOR WELD STRENGTH AND HARDNESS IN WELDING ZONE W. R. T. 304 STAINLESS STEEL

MR. A.R. DESHPANDE, LECTURERVishwakarma Institute of Information Technology, Pune

[email protected]. DR. M.L. KULKARNI,

Professor, SVERI’s College of Engineering, [email protected]

PROF. DR. P.J.AWASARE, PROFESSORHead, Dyanaganga College of Engineering, Pune

[email protected]

ABSTRACTIndustrially gas tungsten arc welding (GTAW), is a widely used for welding stainless steel, as it have good weld quality and relatively lower capital investment. However the numbers of parameters involved in the process are large, thus it is difficult to set process parameters so as to obtain the desired weld quality. In the present paper attempts are made to investigate the parametric effect on the weld strength and hardness near the weld zone for 304 stainless steel. Parameters under consideration were welding current shielding gas flow and work piece thickness. Multiple regression analysis is carried out to establish relation between stated parameters and dependent parameters such as weld strength and hardness near the weld zone. It is observed that weld strength increases with increase in welding current upto certain limit, also with increase in work piece thickness weld current required is high. The effect of shielding gas flow on the weld strength is negligible. The variation in the hardness is observed near the weld zone with increase in welding current and shielding gas flow. The validation of the developed empirical model show good agreement with the experimental results.

Key Words: Weld Strength, Hardness, Welding Current, Shielding Gas Flow, Work piece Thickness

1.0 INTRODUCTION TIG is an arc welding process, wherein coalescence is produced by heating the work piece with an electrical arc struck between a tungsten electrode and the work piece. TIG welding involves a number of process variables. Each variable has its effect on the weld and there is an interrelationship among variables that affects the weld characteristics. Welding parameters have significant effect on mechanical properties of material. Heating and subsequent cooling of the work piece during welding, [1,2] results in the change in the microstructure of the base metal. The change in microstructure results changes in mechanical properties of the welded joint. Therefore effect of welding parameter needs to be systematically investigated. In the present study, major factors affecting the weld characteristics are investigated and their effect on weld characteristics is studied. By using the result obtained from experimentation, an empirical relations for breaking load and hardness of work piece ( 3mm from the center of weld pool), based on the selected welding parameters are established using multiple regression analysis. The established empirical relations can be used for predicting the probable values of breaking load of welded joint and hardness near the welded zone.

2.0 EXPERIMENTAL PROCEDURE It was observed that, welding current, shielding gas flow and work piece thickness are the parameters which affect the hardness and strength of the welded joint. Initial experiments were carried out to determine the working range limits [3] of selected welding parameters. An AXT -400 TIG welding machine with DC output and 440V input supply voltage was used for experimentation. The welding machine was having a forced air cooled welding torch; it was working in the current range of 10 – 400A. Welding current is manually regulated with the help of current regulator. A gas regulating attachment was also provided with the welding machine for controlling shielding gas flow. Argon was used as shielding gas for welding the work pieces. As Argon is slightly heavier than air, provides more efficient gas shielding at lower flow rate. The density of argon (1.656 kg/m3) is higher than that of the atmosphere; therefore better protection of molten weld pool is possible. Argon [4,5] is better for arc starting and operates at a lower arc voltage.1.6 mm diameter 304 stainless steel rod was used as consumable filler metal. For study SS304 was selected as work piece material, as it has a very large scale application in the process industry. The



material is selected in three thickness ranges i.e. 1.2mm, 2mm and 3mm. The specimen size selected was 25mm X 100mm as per ASTM standards.The SS304 sheet was converted in the desired work piece size by using shearing operation. After shearing the work pieces were straighten by holding them in a manual press. The burr from the cut edges of the work piece was removed by manual filing. The welding on SS304 work piece was performed in two steps. Initially the work pieces were tagged for alignment and then the final welding run was performed. For holding the work piece in both of these operations, a fixture was fabricated in MS angle, an asbestos sheet (8mm Thick) was used as support for work piece. The asbestos guide ways were glued with the base asbestos sheet, this help in proper alignment of the work piece during welding.

2.1WORK PIECE CHARACTERISATION

2.1.1 Tensile Load MeasurementThe work pieces welded by TIG welding process were tested for breaking load on a FSA make universal testing machine with 60T capacity. 2.1.2 Hardness Measurement

A PRIME make Rockwell cum Brinell hardness testing machine with maximum 5 kN capacity was used to test the work piece for hardness. For present study ‘A’ scale (HRA) with diamond indenter with 60 kg load was used.Hardness measurement by using Rockwell cum Brinell hardness testing machine is dependent on the supporting anvil size. It was observed that hardness measurement of work piece get affected with the supporting base of the anvil. If a larger diameter anvil is used for thin plate, larger warped area of thin sheet comes in contact with the anvil, which deviates hardness value from actual value. To avoid variation in the hardness value small supporting area of the anvil is to be used, which supports smaller area of the warped sheet exposing small amount of irregularities of the surface while taking hardness reading. Therefore the anvil used for the hardness measurement.2.2 Process Parameters The process parameters[4,5] governing the TIG welding process are welding current, arc voltage, shielding gas, travel speed, pulsing in case of programmed welding, rod feed speed, torch angle, pool geometry, work piece thickness, temperature distribution, bead geometry, clamping, fixtures, heat sink, heat buildup, distortion, high frequency of arc starting.It was not possible to consider all above mentioned parameters for study, therefore few parameters were selected for studying their effect on breaking load and hardness of the material. It was found that welding current, shielding gas flow and work piece thickness has greater effect on mechanical properties [6] of material. The increase in welding current results in increase in heat input given to the work piece, which affects the mechanical properties of the material. The increase in shielding gas flow result in increases the rate of heat

transfer [7], it also affect the penetration, which plays important role in improving the strength of the welded joint. As thickness of work piece increases, welding current and shielding gas flow used for welding were also to be increased significantly. With increases in thickness of work piece, the increase in welding current and shielding gas flow results into heating and subsequent cooling [1] of the work piece that result in the change in the microstructure of the base metal. The change in microstructure results in change in the hardness of the base metal, which affects strength of welded joint. It, also, observed that hardness of the base metal near the weld was lowest and it increases in the HAZ. So, welding current, shielding gas flow and work piece thickness were considered as variables for present study. Breaking load of joint and hardness of the material were considered as dependent variables.2.3 Pilot Experiments

Working limits of experimentation were decided through pilot experiments. Initially experiments were performed to identify the working range of the process parameters [2] for all work piece thicknesses. The welded work pieces were observed visually and the range of parameter for the pilot experiments was decided. The selected working range is given in table 1 and 2.The pilot experiments were performed within above mentioned working range. One parameter varied by keeping other two parameters constant. From results, it was observed that, variation in hardness due to change in welding current and shielding gas flow is observed in the region 3mm from the center of weld pool along the length of work piece. Therefore the variation in the hardness is plotted against the welding current and shielding gas flow in the region of 3mm from the center of weld pool along the length of work piece. Similarly, variation in the breaking load is plotted against varying current and shielding gas flow as shown in Fig 1–4. These plots are instrumental in deciding the working range for the final experiments.

Table 1 Working Limits for Shielding Gas Flow for Pilot Experiments

Table 4 Working Limits for Shielding Gas Flow for Final Experiments


Shielding Gas Flow vs Breaking Load

152025303540455055

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Shielding Gas Flow (LPM)

Bre

akin

g L

oad

(K

N)

Current 25A,Thickness1.2mmCurrent 65A, Thickness2mmCurrent 110A, Thickness3mm

Fig. 1 Shielding Gas Flow vs. Hardness at 3mm

Shielding Gas Flow vs Hardness at 3mm From Center of Weld Pool

30

35

40

45

50

55

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Shielding Gas Flow (LPM)

Har

dne

ss (H

RA

)

Current 25A,Thickness1.2mmCurrent 65A, Thickness2mmCurrent 110A, Thickness3mm

Fig. 2 Shielding Gas Flow vs. Breaking Load


Welding Current vs Hardness at 3mm From Center of Weld Pool

30

35

40

45

50

15 30 45 60 75 90 105 120 135 150 165

Welding Current (A)

Har

dn

ess

(HR

A)

Gas Flow 1LPM, Thickness1.2mmGas Flow 2LPM, Thickness2mmGas Flow 3LPM, Thickness3mm

Fig. 3 Welding Current vs. Hardness at 3mm

Welding Current vs Hardness at 3mm From Center of Weld Pool

30

35

40

45

50

15 30 45 60 75 90 105 120 135 150 165

Welding Current (A)

Har

dnes

s (H

RA

)

Gas Flow 1LPM, Thickness1.2mmGas Flow 2LPM, Thickness2mmGas Flow 3LPM, Thickness3mm

Fig. 3 Welding Current vs. Hardness at 3mm

Welding Current vs Breaking Load

17

22

27

32

37

42

47

52

15 30 45 60 75 90 105 120 135 150 165

Welding Current (A)

Brea

king

Loa

d (K

N) Gas Flow 1LPM, Thickness1.2mmGas Flow 2LPM, Thickness2mmGas Flow 3LPM, Thickness3mm

Fig. 4 Welding Current vs. Breaking Load2.4 Final ExperimentationAfter performing initial experiments, suitable working range of welding parameters for further experimentation was selected. Three working values of welding current and shielding gas flow were selected for each work piece thickness, which is given in Table 1 and 2. Parameter maintained constant was gap between the plates. As work piece thickness increases the gap between the work pieces

needs to be increased accordingly. The gap between the work pieces for higher thickness of work piece is instrumental in proper penetration, which gives better joint quality. The maintained gap was 1.2 mm for 1.2 mm thickness of work piece whereas it was 2 mm for 2 and 3 mm thickness of work piece, as it affects the tensile breaking load, that can be achieved. For designing final


experiments, L9 orthogonal array was used. Using L9 orthogonal array total 9 experiments for each thickness of work piece were designed. Using orthogonal array sets of welding parameters such as welding current, shielding gas flow for all thickness of work piece were selected for further experimentation. These set of parameters are useful for selecting the larger combinations of parameter range with optimum number of experiments. Working limit for parameters for final experiments is shown in Table 3and 4.

Table 3 Working Limits for Welding Current for Final Experiments

Table 4 Working Limits for Shielding Gas Flow for Final Experiments

3.0 Regression AnalysisIn the present study breaking load and hardness were dependent on welding current, shielding gas flow and thickness of work piece. Breaking load and hardness were dependent variables and welding current, shielding gas flow and thickness were influencing variables. The data generated from experiments was collected and tabulated using Microsoft Excel software. Multiple regression analysis was applied to the data by using Microsoft Excel worksheet. Relation between breaking load and influencing variables, welding current, shielding gas flow and work piece thickness was derived.The equation obtained by multiple regression analysis for breaking load is given below.

ó = b*( m3Ò * m2

³ * m1q ) ……….4.1

Where,ó = Breaking Load Ò = Work Piece Thickness ³ = Welding Currentq = Shielding Gas Flow

m1 , m2 , m3 and b are constants generated using multiple regression analysis.The constants m1, m2, m3 and b are determined from the multiple regression analysis. The values of influencing variables can be selected form obtained working range for these parameters. The value for breaking load can be obtained from empirical relation for values of influencing variables. Thus, predicted value of the breaking load is helpful in selecting suitable welding process parameters.From the experiments it was observed that the major variation in the hardness value of the material was in the region of 3mm from the center of weld pool along length of the work piece. Therefore, equation obtained by multiple regression analysis for hardness at 3mm from the center of weld pool along length of the work piece. The equation obtained by multiple regression analysis for breaking load

H = bh*( mh3Ò * mh2

³ * mh1q ) …...4.2

Where,H = Hardness at 3mm from the center of weld pool along length of the work PieceÒ = Work piece thickness ³ = Welding current

q = Shielding gas flowmh1, mh2 , mh3 and bh are constants generated using multiple regression analysis. The constants mh1, mh2, mh3 and bh are determined from the multiple regression analysis. The values of influencing variables can be selected form obtained working range for these parameters. Using multiple regression analysis the equation generated for breaking load and hardness at 3mm from the center of weld pool along length of the work piece. The values for breaking load and hardness generated by the regression analysis are compared by the values generated by experiments. The comparison between the values generated by multiple regression analysis and experiments are shown in Table 5 and 6.

Fig. 5 represents comparison of breaking load calculated by using multiple regression analysis and actual experiments. Form the Fig. 5, it is observed that, results obtained by multiple regression analysis are close to actual experimental values. The error observed is less than 10%. Maximum error observed 9.54% for 3mm work piece thickness with 160 A welding current and 4 LPM shielding gas flow. Minimum error observed is 4.24% for 2mm work piece thickness with 55 A welding current and 2 LPM shielding gas flow. As the value of the error is less than10% it can be stated that the obtained equation is close to the actual equation. Therefore, values generated by the multiple regression analysis can be accepted for breaking load.Table 5 Result of Validation Experiment for Breaking Load

Table 6 Result of Validation Experiment for Hardness

Fig. 6 represents comparison of hardness calculated by using multiple regression analysis and actual


experiments. Form the Fig. 6; it is observed that, results obtained by multiple regression analysis are close to actual experimental values. The error observed is less than 10%. Maximum error observed 4.94% for 3mm work piece thickness with 160 A welding current and 4 LPM shielding gas flow. Minimum error observed is 1.93% for 3mm work piece thickness with 160 A welding current and 3 LPM shielding gas flow. Therefore, values generated by the multiple regression analysis can be accepted for hardness of the material. Within defined range of process parameters, any of the combination of the process parameter leads to same results as found from experimentation and model validation. Thus, the developed model is validated and is found to be correct.

Fig. 5 Comparison of Breaking Load

CONCLUSIONSIt is found that the parameter considered i.e. welding current, shielding gas flow and work piece thickness have great influence on the considered weld characteristics i.e.

\ breaking load and hardness. With increase in welding current and shielding gas flow, breaking load increases and hardness decreases. However beyond

certain limit the breaking load and hardness fluctuates. In the present case maximum strength of 55 kN and 46HRA hardness is obtained for 160 A welding current and 3 LPM shielding gas flow.Through regression analysis empirical relations are developed for evaluating the breaking load of welded joint obtained by TIG welding, wherein use of filler material is made. It is found that the developed model gives the result in accordance with experimental validation and hence it is concluded that within the defined process parameter limit developed model can be used effectively.

REFERENCES[1] B.Y.Kang, Yarlagadda K.D.V. Prasad, M.J. Kang, H.J.Kim, I.S.Kim, The effect of alternate supply of shielding gases in austenite stainless steel GTA welding, Journal of materials processing technology,(2008), p.1109-1120

[2] M.A. Wahab, M.S. Alam, M.J.Painter and P.E.Stafford, Experimental and numerical simulation of restraining forces in gas metal arc welded joints, welding journal, 85(2006), p. 35-43

[3] Shanping Lu, Hidetoshi Fujii and Kiyoshi Nogi, Arc ignitability, bead protection and weld shape variation for He-Ar-O2 shielded GTA welding on SUS304 stainless steel, Journal of materials processing technology, 209(2009), p. 1231-1239[4] Cary HB, Modern welding technology, Printice-Hall, Eaglewood Cliff, NJ, 1989.

[5]Sacks R.J. Welding : Principles and Practices,Glencoe,Peoria, IL,1981.

[6] S.C. Juang and Y.S.Tarng, Process parameter selection for optimizing the weld pool geometry in the tungsten inert gas welding of stainless steel, Journal of materials processing technology, 122(2002), p. 33-37.

[7] S.A.A. Akbari Mousavi and R. Miresmaeili, Experimental and numerical analysis of residual stress distribution in TIG welding process for 304L stainless steel, Journal of material processing technology, 208(2008), p. 383-394


NUMERICAL SIMULATION AND EXPERIMENTAL VALIDATION ON DIABATIC HELICAL CAPILLARY TUBE WITH R-22

DR HEMACHANDRA REDDY.K1 , KRISHNA REDDY.V 2

1. Professor and Director, academic planning J.N.T.UniversityAnanthapur, Ananthapur Dt, A.P. India.

2. Research Scholar, J.N.T.U. Ananthapur, A.P. India. E.Mail: [email protected].

Mobile No: 9441681902

ABSTRACTCapillary tubes are often used as an expansion and controlling devices in small refrigeration and air conditioning systems. In the present work a separated flow model has been developed for the non adiabatic flow of refrigerant through helical coiled capillary tubes, which simulates a situation nearer to that existing in real systems. The homogeneous two phase and adiabatic flow model has been used exclusively in capillary tube flow studies; however many of previous research results show that two phase flow in small diameter tubes may be not entirely homogeneous due to phase interaction. Much of the work done on straight adiabatic capillary tubes in modeling studies of capillary tube. However the capillaries used in most air-conditioners are coiled to be installed in a confined space and are not insulated. In this paper a separated flow model of refrigerants through coiled capillary tubes is derived. Mass flow rate through the coiled capillary is measured with different tube geometries and under various operating conditions. This paper attempts to compare the flow characteristics of R 22 in non adiabatic coiled tubes and also to study the effect of the coiled diameter and on the capillary tube length. Comparison of the present model has been performed by comparing with the previous models and experimental data presented in the technical literature. The results of Zhou and Zhang [9] model, experiments, Kim et al [6] results are making use of to validate the developed model. . The predicted results show that the separated flow model using the Chisholm’s slip ratio and appropriate friction factors at different regions gives a better estimation with the available experimental

Keywords:Capillary Tube, Helical Tube, Two Phase Flow, Separated Flow, Flow Characteristics, Slip Ratio data.;.

1. INTRODUCTIONCapillary tube is one of the simplest kinds of expansion device employed in small refrigeration systems and air conditioners due to their simplicity, low cost and reliability. These tubes tender many advantages over the other expansion devices such as thermostatic valves like inexpensive and cause the compressor to start at a low torque as the pressure across the tube equalize during the off cycle. The study of flow behavior of refrigerant through capillary tube is decisive to the performance of refrigeration system. Regardless of of its simplicity the flow inside the capillary tube is a complex phenomenon. In the past decades, number of numerical and experimental studies has been made to understand its flow characteristics. Due to the concern of the depletion of the earth’s ozone layer and global warming, many conventional refrigerants are being phased out and hence to meet the demand

for saving the environment and also improving the performance redesign and assessment of the capillary tube has received considerable attention in the recent years. In practical consideration of the use of a capillary tube the main concern is to determine the mass flow rate of the refrigerant flowing through the tube and hence to a given refrigeration capacity the appropriate length, diameter of the tube and the helical diameter and inlet and outlet conditions. As liquid refrigerant flows through a capillary tube for the first part of its flow it remains in the liquid state and it can be considered as single-phase flow region. After passing through some length due to flashing refrigerant changes from liquid to a two-phase mixture. The flow within the capillary tube is separated into a single-phase region and two-phase


flow region. In this two phase region the phases are artificially segregated into two streams: one is liquid and the other one is vapor.In a two-phase flow due to the differences in thephysical properties of the phases, vapour tends to flow at a higher velocity than the liquid phase; hence slip exists between the two phases.2. LITERATUREBolstad and Jordon [1] were probably the first to have conducted the experimental investigation on adiabatic capillary tubes on R-12 refrigerant and observed that very little effect of varying evaporator pressure on the mass flow rate. Lin et al [2] developed correlations to calculate the single and two phase flow friction factors. These correlations were used by Li et al [3] in their numeric model to calculate the adiabatic capillary tube length. Choi et al [4] investigated, experimentally the flow through adiabatic capillary tubes for different geometries. With the obtained results from experiments, a generalized correlation was developed for the calculation of mass flow rate and he compares both studies and found a reasonable agreement. Later Choi et al [5] again investigates experimentally and developed empirical correlation and rating charts for different refrigerants. S.G.Kim et al [6] investigate the performance of different refrigerants through the experiments and developed a dimensionless correlation on the basis of the experimental results. They compared the results of straight capillary tubes with those of helical capillary tubes. By taking several lengths and inner diameter combinations of capillaries and the performances of adiabatic tubes are estimated. P.K Bansal et al [7] presented a homogeneous two phase model, CAPIL, which is designed to study the performance of adiabatic capillary tubes where the tube is assumed as straight and showed little improvement over their simpler homogeneous two-phase flow model.

Valladares et al [8] performed a theoretical

study on adiabatic capillary tubes with different

refrigerants like R-407C, R-22, and R-410A and

developed a numerical simulation model and

proposed a theoretical relation for mass flow rate.

Zhang [9] analyzed and modeled an adiabatic

capillary tube and he has used the artificial neural

network technique to predict the mass flow rate of the

adiabatic capillary tubes.

Neeraj agarwal and Bhattacharyya [10]

made a relative study of flow characteristics of an

adiabatic capillary tube by employing separated and

homogeneous two phase flow models. And from his

studies he concluded that the discrepancy between

the separated and homogeneous flow models has a

maximum about 8-11%.

3. RESEARCH SIGNIFICANCE

Substantial efforts have been made for

straight and adiabatic capillary tube modeling.

Available technical literature indicate that many

studies are available on modeling of the adiabatic

capillary tube sizing, study the effect of geometric

parameters on mass flow rate such as tube diameter,

coiling and sub cooling of the refrigerant. And also

modeling of adiabatic tubes for homogeneous flow

and in some cases straight adiabatic tubes with

separated flow. Modeling efforts or experimental data

for the coiled capillary tubes with separated flow are

fairly limited and very little information available in

the current literature on helical coiled adiabatic tubes

with separated flow considering metastable region.

Considering the scarcity of literature on the above

said situation, the present investigation has been

conducted to model the capillary tube and to compare

with the experimental data as well as numerical

model results available in the literature.

4. Model description and Solution

Methodology

This section discusses the numerical model that has

been developed to design and study the performance

of diabatic capillary tubes.


It includes the effect of various design

parameters namely the tube diameter, tube relative

roughness, tube length, level of sub cooling and mass

flow rate of refrigerant. This method is based on the

fundamental equations of conservation of mass,

energy and momentumThe coiling effect of a tube

can be embodied in the calculation of friction factors.

So, a straight tube model can also be applied to coiled

capillary tubes by changing the corresponding

friction factor equations. For convenience, the

developed four-ûow-region model is brieûy reviewed

and then friction factors for coiled capillary tubes are

discussed.

Also, the ûow inside it is considered to be one-

dimensional, steady, diabatic, no external work and

the separated two-phase ûow.After the observation of

the flow mechanism inside the capillary tube, the

coiled capillary tube can be divided into different

regions. They are single phase sub cooled region,

metastable liquid region, metastable two phase region

and equilibrium two phase region.inspite of the above

regions, to understand clearly the flow behavior, flow

restrictions at inlet and outlet sections are also to be

considered.The pressure loss in a sudden contraction

at inlet and pressure drop due to enlargement can be

evaluated by the weisbach classical equations

Pcon-Pin K (1) Pin-Pevap K ñ (í

1 – í

2)2 / 2 (2)

4.1 Single-phase sub cooled region

The region from the inlet of the capillary to

the position where the saturation pressure

corresponds to inlet temperature is known as sub

cooled liquid region. In this region the liquid is

considered as incompressible, therefore

v1=v2=vsc (3)

h1=h2=hsc.

(4)

For straight capillaries, friction factor is evaluated

from the expressions proposed by Churchill

( )

11212

32

8 18Re

fA B

= + +

Where

(5)

16

0.9

12.457 ln7 0.27

Re

A

dε

= +

For coiled tubes the friction factor in this region is

evaluated from the equations proposed by Mori and

Nakayama with different coefficients considering the

tube wall roughness.

1637,530Re

B =

0.5

12

1.6 1/62.5 2.51

Re Re

c

c c

DC D CfD D

D D

= +

(6) Where


4.2 Metastable liquid region

In this region the refrigerant changesfrom

sub-cooled liquid to liquid-vapor two phase mixture.

In this process the inception of vaporization does not

take place at the corresponding saturated state, but at

a point other than at downstream. This process of

delay in vaporization is known as metastability or

under pressure of vaporization.Chen et al proposed a

correlation to predict the under pressure for R 12 and

it has been extended by Bittle and pate [12 ] from R

12 to R 134a and R 22.

(7)

Where D1 is the reference length ,

410satkTσ

×

k is Boltz-mann constant (1.38066210-23 J K-1).Friction factor is evaluated in the same way as for the single-phase region.

Metastable two-phase region

There exists a metastable two-phase region after

vapor appears and before the fluid reaches

thermodynamic equilibrium. Due to the lack of

specific information, the friction factor is evaluated

in the same as for two-phase equilibrium phase.

Equilibrium two-phase region

In this region the fluid enthalpy and volume are

determined by the following equations.

h= (1-xg) hls+xghg (8)

v= (1-xg) vls+xgvg (9) Considering the flow as separated, a void fraction is estimated from the semi empirical correlation of Premoli.

For coiled tubes, three methods are considered for

calculating friction factor .1) M&N + GIRI method 2)

M&N method 3) C-M&N method. Among the above

three, C-M &N equation is used as proposed by Zhou

and zhang [ ] where the tube wall roughness is also

taken into account.

Figure 2

( )( ) ( )

0.5

1, 1 12.5 2.56 6

1

Re Retp

dDf

d dD D

ω ψ

= × + (10)

Table 1

4.3 Diabatic sub cooled single-phase region

Applying energy balance equation for the

element in the single phase region we get

m Cp [ Ti+1 - Ti ] = UA [Ta - (Ti + Ti+1)/2](11)


Where

( )

12 3 410.728 lw v lw

wlw w c

g h KhT T d

ρµ

∆= − Figure 1

Where the mass flow rate of the refrigerant, Cp is is the specific heat of the refrigerant, Ta is the ambient temperature and U is the overall heat transfer coefficient.

Since the incremental length is very small errors caused by this approximation are very minimal. The heat transfer process from the capillary tube side to the surroundings includes convective heat transfer from the refrigerant inside the capillary tube and the natural convection over the tube. Therefore the overall thermal conductance is given as 1/ UA = 1/hcd ð dz + 1/had ð dz (12)The heat transfer coefficients are calculated from hc = NucKc/d (13) Where Nuc is the Nusselt number of the refrigerant flowing in the tube and Kc is the thermal conductivity of the refrigerant inside the tube.The Nusselt number for the flow inside the coil is given as

Nuc = 0.023[Re]0.85 [d/D]0.1 [Pr]0.4 (14)4.4 Diabatic two phase region

For the refrigerant flow in a capillary tube with no externally applied work the energy equation is obtained

d/dz [xhg + (1 – x) hl ] + d/dz [x Vg2 / 2 + (1-x)

Vl2 /2] + dQ/m = 0 (15)

Where dQ is the natural convection heat transfer

from the surroundings or the latent heat of the water

condensed on the capillary tube.The natural

convection heat transfer from the capillary tube to the

surroundings occur till the temperature of the

refrigerant inside the capillary tube drops below the dew point temperature (Tw = 200C) .Water vapour

condensation starts from that point till the

temperature drops to 00C. The natural convection

heat transfer is given as

dQ= haðd[Tc-Ta] (16)

Tc is the corresponding saturation temperature at the

quality xand Tw is the dew point temperature at atmospheric

conditions (taken as 20 0C)

The latent heat of vaporization is given as dQ = M Ähv = hw ( Tw - Tc ) ðd

(17)The heat transfer coefficient ha is calculated in the similar manner as explained in the single phase region.

Substituting the value of in equation (5) and after subsequent rearranging the terms we obtain an expression for total pressure gradient.

ha is calculated as explained earlier in the single phase region and Ta is the ambient temperature and is taken as 280C. The momentum equation is written as an explicit equation for the pressure gradient. In a coiled tube, the centrifugal force of a flowing fluid produces a pressure gradient in a cross section.

This pressure gradient yields secondary flows. The secondary flows cause a larger amount of pressure drop or heat transfer rate than that for a straight tube.The total pressure gradient is expressed as the sum of two different components


Ø Frictional pressure dropØ Acceleration pressure drop

dP/dZ = (dP/dZ) f + (dP/dZ) a (18)After rearranging the terms, to calculate the total pressure gradient, void fraction, slip ratio and the frictional pressure gradient (dp/dg) f are required.

Slip ratio expresses the ratio of the gas phase velocity

to the liquid phase velocity. Chisholm correlation is

employed to calculate slip ratio between the two

phases (refrigerant vapor & liquid)S= [1-x (1-(vg/vl))

]½ (19)

The void fraction is calculated by employing the premoli

correlation

á = x vg / (1-x) vl S

+X vg (20)

5. METHDOLOGYThe refrigerant examined for simulation in the

present study was R 22. Refrigerant thermodynamic

and transport properties like dynamic viscosity,

specicific volume, specific enthalpy, thermal

conductivity are taken from REFPROP7.0 [11] and

are developed as a functionof pressure. A capillary

tube is characterized by the following parameters (i)

geometry: inside diameter and roughness, length and

diameter of the connecting tubes; (ii) Refrigerant;

(iii) temporary distribution of both inlet pressure

( mass ûux) and discharge pressure; (iv) temporary

distribution of the inlet temperature or vapor quality.

The total length of the capillary tube is the sum of the four region lengths, i.e. Lcap= Lsc+ Llm+ Ltm+ Ltp.

For the case of the mass ûow rate simulation, the

mass ûow rate cannot be solved explicitly from the

total length of the capillary tube, due to the

interdependence between the refrigerant mass ûow

rate and the friction factors. So, an iteration process is

required for solution. Initially premised mass ûow

rate is used for a capillary tube length calculation

process.The governing equations for the flow include

conservation equations for mass continuity,

momentum and energy that are solved

simultaneously through iterative procedure. The

computational domain extends from tube inlet to

outlet along the curvature of the tube.In adiabatic

single-phase flow region the changes in specific

volume can be ignored and the pressure varies

linearly along the capillary tube due to friction. In

this region after calculating the Reynolds number,

single-phase friction factor and single-phase length

were calculated.In the diabatic region due to the heat

exchange with the ambient, the temperature of the

refrigerant does not remain constant, resulting in

variation of viscosity, thermal conductivity and

specific heat and therefore change in Reynolds

number and Prandtl number in this region. This

explains the reason for change in local friction factor

and convective heat transfer coefficient throughout

the sub-cooled region. The mathematical model

presented in the previous section for the single-phase

flow region is analyzed by dividing the length into

multiple elements. In this approach the outlet analysis

of one element becomes the inlet analysis of adjacent

element. The above equations are solved to calculate

the pressure drop and temperature drop at each

element. The calculation procedure is repeated until

the saturated conditions are reached.


In the two-phase flow region, the equations developed are solved by means of fourth order Runge Kutta method. The output parameters calculated from the single-phase region are fed as input parameters for two phase region.6. RESULTS AND DISCUSSION6.1 Case of straight capillary tube

Fig: 1 shows the comparison of present

numerical mass ûow rates for different capillary tube

diameters with Zhou and zang [9] simulation model

results for straight capillary tubes. It appears that the

Zhou and zang straight tube model slightly

underestimates the present developed model and

shows a reasonable agreement with the present one.

It is observed that with the increase in capillary tube

diameter and with the increase of degree of sub

cooling the mass flow rate increases for R 22. For

small capillary diameters, it is observed that the

deviation from Zhou and zang model is very little,

where as it increases for larger tube diameters. For

example for a capillary diameter of 1 mm the mean

deviation from Zhou and zang model is about

2.45% and for a capillary diameter of 2.2 mm the

deviation is 4.043%. At higher degree of sub cooling,

the deviation of mass flow rates with the Zhou model

is small compared to lower sub cooling temperatures.

6.1 Case of straight capillary tube

Fig 2: shows the comparison of modeled mass

ûow rates with the Zhou experimental data. It

indicates that the straight model agrees quite well

with the measured mass ûow rates except the slightly

underestimation for d = 2.2 mm tube.

6.3 Case of coiled capillary tube

Fig. 3 compares the numerical coiled capillary mass ûow rates with the Zhou simulation model in which the friction factor is calculated with the C–M&N equation.

Zhou simulation model estimates the mass flow rates

with three friction factor equations and concluded

that C–M&N equation give almost the same

predicting results and underestimate the mass ûow

rate with the a small discrepancy about 5.64% with

Zhou experimental results.

For coiled tubes also same tendency continues as in straight tubes i.e. with the increase in sub cooling temperatures mass flow rate also increases. the discrepancy between the present developed model with the Zhou model is about 9%.

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12 14 16Degree of subcooling (ºC)

Mas

s flo

w r

ate

(kg/

h)

Zhou (Capillary dia 2 mm)Zhou (Capillary dia 1.4 mm)Zhou (Capillary dia 1 mm)Present (Capillary dia 2 mm)Present (Capillary dia 1.5 mm)Present (Capillary dia 1 mm)

6.4 Comparison with experimental data by Kim et al.

The Experimental Results Presented By Kim Et Al. [6] For Adiabatic Capillary Tubes Using R-22 As A Working Ûuid For Coiled Tubes Has Been Used Fig. 4.it Compares The Mass Ûow Rates Obtained By The Experimental Cases And The Developed Numerical Model Results. In This, The Numerical Model Under Predicted By A Small Amount With The Experimental Results; Although The Tendency And The Relative Discrepancy Of The Model Against The Experimental Results Are In General Very Good.


Figure 4

30

35

40

45

50

55

60

0 2 4 6 8 10 12Degree of subcooling (ºC)

Mas

s flo

w r

ate

(kg/

h)

C-M&N line Zhou simulationC-M&N line Present model

6.5 Comparison with Zhou simulation model

Comparison of Zhou model (C-M&N equation) with

the results predicted from the model developed was

illustrated in Fig: 5. From this we can observe that,

similar to a straight tube, the mass flow rate of the

coiled capillary decreases rapidly as tube length

increases, and it increases as condensing temperature

increases. The deviation from the Zhou model is

almost same for different capillary tube lengths and it

is in the range of 2.78-4.2%.

Figure 5

0

10

20

30

40

50

60

70

80

90

0 0.5 1 1.5 2 2.5Capillary tube length (m)

Mas

s flo

w r

ate

(kg/

h)

Zhou model

Present model

6.6 Comparison with experimental data by Zhou et al.

As the length of the capillary tube influences a lot in

the mass flow rate through the tube, it has been

investigated through the model developed and the

results are validated with the experimental results by

Zhou Figure 6

0

20

40

60

80

100

120

140

160

0 2 4 6 8 10 12 14 16Degree of subcooling (ºC)

Mas

s flo

w r

ate

(kg/

h)

Zhou Exp Capillary dia 2 mmZhou Exp Capillary dia 1.5 mmZhou Exp Capillary dia 1 mmpresent model Capillary dia 2 mmpresent model Capillary dia 1.5 mmpresent model Capillary dia 1 mm

Figure 7

0

10

20

30

40

50

60

70

0 2 4 6 8 10 12Degree of subcooling (ºC)

Mas

s flo

w r

ate

(kg/

h)

Kim et al experimental resultsPresent model

Figure 8

0

10

20

30

40

50

60

70

80

90

0 0.5 1 1.5 2 2.5Capillary tube length (m)

Mas

s flo

w r

ate

(kg/

h)

Zhou experimental results

Present model


7. CONCLUDING REMARKSThe performance of a refrigeration system with R22 in straight and adiabatic capillary tubes with separated flow is studied numerically and compared with previous numerical models and experimental results available in the technical literature. This flow model has been developed by means of a one dimensional analysis of the basic equations of conservation of mass, energy and momentum.

The mass flow rate along the capillary at

different regions is calculated with the appropriate

friction factor equations. In this study, Churchill’s

equation is used for straight tube and Mori &

Nakayama method for coiled tubes at early regions of

capillary tube and C-M&N equation is used for the

later regions in the capillary.

With the usage of the above equations in

conjunction with Premoli’s void fraction for

separated flow the mass flow rate is calculated to

estimate the performance of the system.

The predicted results show that the mass

flow rate increases with the increase in degree of sub

cooling and also with the coil diameter. For a coil

diameter of 40mm the decrease in mass flow rate is

about 8-10% and with the increase in coil diameter ,

flow rate also increases to certain limit and thereafter

there is no considerable change in flow rate. Between

the separated flow and homogeneous flow there is a

discrepancy about 7-9% is observed. In general a

good degree of correlation between numerical and

experimental results has been observed for the cases

using different tube geometries and flow patterns.

REFERENCES

1. M.M. Bolstad, R.C. Jordan, Theory and use of the capillary tube expansion device: part II – non-adiabatic, Refrigerating Eng. 57 (1949) 519–523.

2. Lin, S., R.Y. Li and Z.Y. Chen Local frictional pressure drop during vaporization for R12 through capillary tubes, International Journal of Multiphase Flow, 17 (1991) 995-1020.

3. R.Y.Li, S.Lin, Z.Y.Chen and Z.H.Chen, Metastable flow of R12 through capillary tubes, International Journal of Refrigeration.13 (1990) 181-186.

4. J. Choi, Y. Kim, H.Y. Kim, Generalized correlation for refrigerant mass flowrate through adiabatic capillary tubes, Int. J. Refrigeration 26 (2003) 881–888.

5. J. Choi, Y. Kim, J.T. Chung, An empirical correlation and rating charts for the performance of adiabatic capillary tubes with alternative refrigerants, Appl.Thermal Eng. 24 (2004) 29–41.

6. Kim, S.G., M.S. Kim and S.T. Ro (2002) Experimental Investigation of the performance of R22, R407C and R410A in several capillary tubes for air-conditioners, International Journal of Refrigeration, 25, 521-531.

7. P.K.Bansal and A.S.Rupasinghe, an homogeneous model for adiabatic capillary tubes. Applied Thermal Engineering 18 (1998) 207-219.

8. P.K.Bansal and A.S.Rupasinghe, an homogeneous model for adiabatic capillary tubes. Applied Thermal Engineering 18 (1998) 207-219.

9. O.G. Valladares, Review of numerical simulation of capillary tube using refrigerant mixtures, Appl. Thermal Eng. 24 (2004) 949–966.

10. G. Zhou, Y. Zhang, Numerical and experimental investigations on the performance of coiled adiabatic capillary tube, Appl. Thermal Eng. 26 (2006) 1106–1111.

11. Neeraj Agrawal, Souvik Bhattacharyya Homogeneous versus separated two phase flow models:


1. Adiabatic capillary tube flow in a transcritical CO2 heat pump, International Journal of Thermal Sciences 47 (2008) 1555–1562.

2. REFPROP (2000) Thermodynamic properties of refrigerants and refrigerant mixtures, version 7.0, National Institute of Standards and Technology

3. R.R. Bittle, M.B. Pate, A theoretical model for predicting adiabatic capillary tube performance with alternative refrigerants, ASHRAE Trans. 102 (2) (1996) 52e64.

Bibliography of Authors Dr.Hemachandra Reddy Principal and Professor

Graduated from S.V University in Mechanical Engg, did his masters from J.N.T.University,

Hyderabad and did his PhD from J.N.T.University.

Areas of interest are Refrigeration and Air Conditioning, Heat Power Engineering. He actively involved in setting up various laboratories in JNTU Ananthapur.

Presently working as Director,Academic planning JNTU Ananthapur. He is presently guiding 12 Ph.D candidates and nearly guided 75 M.Tech projects. He has published nearly 50 publications.

Krishna Reddy

Research ScholarGraduated from V.R. Siddhartha Engineering College in Mechanical Engineering in 1995. Post graduated from J.N.T.University, Hyderabad. Specialized in Thermal Engineering. Doing research in the field of Refrigeration and Air conditioning. Areas of interest: refrigeration & conditioning, I.C.Engines Member of I.S.T.E.


LAND USE CHANGE DETECTIONS IN TIRUMULLAIVASAL VILLAGE USING REMOTE SENSING AND GEOGRAPHIC INFORMATION

SYSTEMV.RAJESH KUMAR1 and G.VICTORRAJAMANICKAM2

1) 1Assistant Professor III, School of Civil Engineering, SASTRA University,2) Thanjavur

3) 2Dean, CARISM, SASTRA University,Thanjavure-mail :[email protected]

ABSTRACTLand cover and its change is a key to many diverse applications such as Environment, Forestry, Hydrology, Agriculture, Geology and Ecology. Various Natural Resource Management, Planning and Monitoring programs depend on accurate information about the land cover in a region. Many coastal areas are undergoing dramatic changes and experiencing the impact of human activities dealing with economic, land use/abuse and resource development. Coastal zones are constantly undergoing wide ranging changes in shape and environment due to natural as well as human development activities. Natural processes such as waves, erosion, changes in river courses etc., cause long time effect at slower rate; but manmade activities, such as settlement, industrial activities, recreational activities, waste disposal etc., affect the coastal environment at comparatively much faster rate. This change influences more on sustainable development. Hence it is necessary to predict the land use changes accurately and fastly The recently developed spatial technologies, remote sensing and GIS have paramount roles in the understanding of our environment and management of resources. The present study focuses on the evaluation of the changes in land uses of the coastal village Thirumullaivasal of Sirkazhi taluk, Nagapattinam by considering four different years such as 1992, 1997, 2004 and 2006.The satellite imageries of the four years were acquired from different sources and were visually interpreted for land use categories supplemented with ground truth. The spatial database were created using GIS and intersected to study the changes occurred in the area through years. The reasons for changes in different categories were analysed through visiting the area in person and surveys and are presented in the paper.Key Words: Coastal land use, Remote Sensing, GIS

INTRODUCTIONThe surface of the earth has always been influenced and transformed under the effect of natural forces through the material flows and energy transfer that take place at different scales-global, regional and local scales. In all these changes, one area stands out very markedly that being the changes in the land use and land cover. Land use/cover is continuously changing, both under the influence of humans and nature resulting in various kinds of impacts on the ecosystem (Rajan, et. al., 1997). The most important factor in the modification of the land cover and its conversion is the human use component rather than the natural changes (Turner, et.al., 1993). Land cover and its change is a key to many diverse applications such as Environment, Forestry, Hydrology, Agriculture, Geology and Ecology. Various Natural Resource Management, Planning and Monitoring programs depend on accurate information about the land cover in a region. Many coastal areas are undergoing dramatic changes and experiencing the impact of human activities dealing with economic, land use/abuse and resource development. Conventional maps are quite useful; however, they do not provide up-to-date information. Since coastal zone is very dynamic, periodic mapping is vital for planning effective strategies.It is of primary importance to undertake a synthetic land use and land cover investigation and develop a monitoring system in order to generate fundamental

land use data and some useful references for the local governments in their sustainable land use planning and decision making. Land-use/land-cover change is a widespread, accelerating and significant process. Information about change is necessary for updating land cover maps and the management of natural resources. Many researches have been undertaken to develop methods of obtaining change information. For large areas, however, to the ground surveys and aerial photo-interpretation are impractical due to large amounts of data to be collected (in space and time). Research has shown that mapping of land cover often is significant by improved using of combination of different techniques (Manier et al., 1984). In this paper, a method involving remote sensing and geographic information system for the prediction of the land use changes in Thirumullaivasal village has been carried out overcoming the conventional methods. The present study focuses on the evaluation of the changes of land uses of Thirumullaivasal village of Nagapattinam district of Tamilnadu. Thirumullaivasal is a coastal village along Bay of Bengal, covering an area of 13975322.87 sq m and extends between 11 14’ and 11 17’N latitudes and 79 48’30” and 79 51’E longitudes.It falls in the survey of India topo sheet 58M/16 prepared in the scale of 1:50,000.


The maximum and minimum atmospheric temperatures are 32.0ºC and 25.3ºC. The primary source of irrigation is the ancient network of canals of river Cauvery and its tributaries. Nearly 97 % of the net area irrigated is connected by canal irrigation system. The normal rainfall is 1168 - 1500 mm. About 80 % of the population depends on agricultural farming for livelihood.

MethodologySurvey of India topographical maps on 1: 50,000 scale were used in the preparation of base maps onto which the interpreted thematic details are transferred. The land use maps of the study area for the years 1992, 1997, 2004 and 2006 are prepared using the satellite imageries through the process of visual interpretation (vink, 1964) of satellite imageries of respective years, as listed in table 1. The methodology is based on monoscopic visual interpretation of IRS imagery using National Remote Sensing Agency classification system (Anon, 1989). The interpreted details are to be ground checked to verify the doubtful areas. Based on the ground verification, the boundaries of the different land use/land cover unit are to be finalized. The corrected details are to be transferred to the single base map on 1:50,000 scale. The land use maps interpreted in hard copy form for different years followed by ground truth information of the year 2004 and 2006 were scanned in tiff format and inputted in the ArcMap module of ArcGIS software. Six tic points were selected for all the maps and were geo-referenced using ArcMap. The accuracy was checked with more tic points. A different shape file was created for each map and the maps were digitized as shape files with polygon mode available in the ArcMap so as area changes can easily be estimated. The shape files were then converted into coverages using ArcTool box. The tables of the coverages were updated with the categories of the interpreted land use. Since the projection system was not defined in the ArcMap software, the area of the polygons was got in decimal units and this cannot be converted into square units. So, Geomatica software was used to get the area of the themes in square meters using update geometry option available in the software. Then layout was prepared for the land use of different years of the study area using ArcMap. The spatial data base land use created for the study area for years 1992, 1997,2004 and 2006 were subjected to intersection analysis available in Geoprocessing Wizard of ArcMap. Then the resulting table was saved in the form of Excel table so as the table may be supported in Microsoft Excel. The land use change matrix for the three different years was derived using Microsoft Excel.Results and DiscussionThe interpreted land use categories include built-up land, cropland, plantations, fallow/harvested land, waterlogged area, scrub land, sandy area, aquaculture, tank and river. The details of the land use categories of the study area are shown in the figures 1, 2, 3 and 4 respectively for the years 1992, 1997, 2004 and 2006. The area covered by the different land use categories are shown in table 2. The dominated land use, cropland during 1992 with 42.15 % of the total area increases to 45.71 % of the total area during

1997 and reduces to 14.36 % by mostly converted to fallow/harvested land during 2004. The fallow/harvested land with 19.03 % during 1992 got reduced due its conversions to cropland during 1997 to 11.16 % and reclaims 48.80 % in 2004 from the cropland. The built up land found mostly along the banks of the river in the southern and eastern parts of village during 1992 with 13.36 % of total area extends its coverage towards northern and western parts with a toll of 18.92 % and emergence of new hamlets on north-eastern boundary rises the total coverage to 28.28 % of the total area. The salt affected area reduces to none during 2004. The plantations and waterlogged area reduces through the years. The 26th December 2004 Tsunami brought many changes to the land use and is discussed as below, with the supplementation of land use for the year 2006: A reasonable decrease of 3.29 %, of the total area in the built-up area category was found after tsunami. The cropland is changed to salt pan, tank, scrub land, fallow/harvested land and agricultural plantations and also shows a considerable decrease with 14.04 %, of the total area. The key decrease is observed with 48.80 % of the total area. The fallow / harvested land has been transformed to sandy area, tank, Scrubland, cropland, built up area, Agricultural plantations and river. Thirumullaivasal also show a vast increase of 111.08 % of the total area. Here the agricultural plantations are transformed to sandy area, built up land, saltpan, tank, scrubland, cropland, agricultural plantations, fallow/harvested land and river. The harbor wave affected the agricultural and horticultural croplands by means of seawater intrusion. The invasion and receding action of harbor waves lead to the removal of soil by erosion and deposition of large amounts of sand, salt and other debris. Paddy fields are affected by erosion, salt deposition, water logging and other deposited sediment. The widening of canals, tanks and newly formed brackish water channels occupied some area in paddy fields. The village shows increase in salt affected land category with 44.43% of total area. Here, it has been transformed to sandy area, built up land and agricultural plantations. Thirumullaivasal show considerable decrease in waterlogged area with 0.15 % of the total area and has been converted to sandy area, built up area, saltpan, cropland, tank, agricultural plantations and river. Thirumullaivasal also show a vast increase in sandy area with 9.64% of total area. Here the transformations of the sandy area are to saltpan, built-up area, cropland, tank, scrubland, agricultural plantations, salt affected land and river. The shift of sand towards land due to receding action of tsunami waves was the main reason for increase in sandy area.Due to tsunami the collapsed built-up area is occupied by sandy area. There is a minimal increase of 1.14% of total area in river. The intrusion and receding action of the waves caused widening of canals in some parts of the study area. The encroachment of tank by human being decreases the area of the tank. In some places, new tanks have been drilled to put up more area in tanks. But the net result is water from the tanks is not found suitable for drinking (According to VAO’s statement).Even though the entire saltpan / aquaculture area was lost during tsunami, the area occupied from them goes on increasing. After Tsunami, the owners of aqua farms have bought most of the agricultural


land and other lands, for very lesser cost since people sell their land as they move out for survival crediting

Table 2 Land use details of Thirumullaivasal village

to the increase in aquaculture. But most of the lands under aquaculture are found uncultivated due to the coastal

regulations imposed on them.Table 1 Details of the satellite data used for the studyConclusions

Since the remote sensing imageries are used, the interpretation of the same will always give more efficient and accurate solutions. The time consumed for the interpretation was very less compared to the conventional methods. The GIS software used handles spatial data and hence accurate overlaying of maps has been got. The GIS makes us to feel where our study area is in the earth. The land use changes detected from the procedure used can be used for sustainable development.Most of the agricultural lands have been taken up by the settlements. If this change scenario continues, the

availability of agricultural lands will be a question mark during 2030’s. The study region being the tail end of Cauvery River, inadequate water supply to the agricultural lands has made them to raise only single crop or double crops per year. This forms the datum for decrease in agricultural lands. Water harvesting structures are the most essential things that need to be given fore most importance in this area since the area depends only on Cauvery water so as at least, the recharged water may give hands to the farmers, if water released from dam is late as per the general schedule. The salinization of aquaculture land and the threat of salinity intrusion into the agricultural lands force agriculturists to sell their land. This made the land unfit for agriculture which supports increase in scrubland in the study area.


The land use changes due to tsunami conclude that an implication of coastal regulations in the study area is mandatory. The built up lands are mostly found on the coastal stretch that too even within 100 m from the coast. So the losses due to any of the hazards may be

high, which necessitates strict adherence of coastal zone regulations in the study region.

References

1. Anand, P.H., and Rajesh Kumar, V., (2003) “Principles of Remote Sensing and GIS”, Sri Venkateswara Publications.

2. Anon, 1989.Manual of Nationwide Land Use / Land cover mapping using satellite imagery, part I, National Remote Sensing Agency, Hyderabad.

3. ESRI, “Understanding GIS The ARC/INFO

Method”

4. Kang-fsung Chang, (2002) “Introduction to Geographic Information System” Tata McGraw Hill Edition.

5. Maniere, R., Poisson, M., and Lecharpentier, M., (1984),”Land use mapping by remote sensing in theMediterranean region: Development of a software package for processing Landsat data”. Proceeding of the Eighteenth International Symposium on Remote Sensing on Environment, Paris, 1-5 October (Ann Arbor, MI: ERIM), pp. 1937-1944.

6. Peter A.Burrough and Rachael A. Mc.Donnell, (1998) “Principles of Geographic Information System” Oxford University Press.

7. Rajan, K.S. and Shibasaki, R., (1997). “Estimation of Agricultural Productivity and Its Application to Modelling the Expansion of Agricultural Land in Thailand”. Journal of Agricultural Meteorology, 52(5) pp. 815-818.

8. Turner II, B.I., moss R.H. and Skole, D.L. (1993) “Relating land use and global land-cover change: a proposal for an IGBP-HDP core Project”. IGBP report No. 5, 65pp.

9. Vink A. P. A, 1964. Some thoughts of Photo Interpretation. Publication of the International Institute for Aerial Survey and Earth Sciences, The Netherlands.


PITCH RATE AND REYNOLDS NUMBER EFFECTS ON HYSTERESIS BEHAVIOR OF FLOW PAST A PITCHING AIRFOIL

D.M.SHARMA1, KAMAL PODDAR2, A. MUTHUKUMAR3, AND K.S.V.REDDY4

Corresponding author: Research Scholar, Department of Aerospace Engineering, IndianInstitute ofTechnology, Kanpur-208016, Email: [email protected]

Professor, Department of Aerospace Engineering, Indian Institute of Technology, Kanpur-208016, IndiaTIMES Scientist B, NAL, Bangalore, Karnataka 560037, India.

ABSTRACTWind tunnel experiments were conducted on NLF-0416 airfoil model to investigate the effect of the Pitch Rate & Reynolds Number on the aerodynamic characteristics and hysteresis behavior associated with the pitching motion of the airfoil. Pressure measurements were conducted on the mid span of the airfoil and in the wake for quantitative results. The hysteresis behavior was observed in aerodynamic characteristics as function of pitch rates and Reynolds number and loops gets enlarged when pitch rates was increased. The present investigation of ramp motion of airfoil showed that, under the operating conditions studied, the airfoil undergoes light dynamic stall. This stall process is characterized by the formation, growth, and upstream movement of a vortical layer in the aft region of the airfoil, leading to large-scale flow separation. No LEV (large eddy vortex) formation or the scenario associated with deep stall was observed during the ramp motions. The experiments also indicated strong hysteresis effects in the flow evolution. The initiation of the stall process from the trailing edge is believed to be characteristic of the NLF-0416 airfoil section tested. As the reduced pitch rate (k) increases, hysteresis loop gets enlarged and á-stall also increases due to increment in lag associated with the movement of separation towards leading edge during pitching motion, because boundary layer is unable to keep up with the change in angle of attack.Keywords: Hysteresis effect, Reynolds number effect, reduced pitch rate, light dynamic stall.

INTRODUCTIONIt has been recognized for many years that the maximum lifting force generated by the wing can be substantially enhanced, if it is subjected to rapid pitching motions. Aerodynamics generated by pitching airfoils is generally classified under the heading of “Dynamic Stall.” Dynamic stall affects helicopter rotor blades in forward flight, maneuvering, and descent, wind turbine rotors. Dynamic stall phenomena are the result of airfoil and wing undergoes ramp or oscillatory motion and having a maximum angle of attack greater than the static stall angle. These unsteady flows are characterized by massive separation and formation of large-scale vortical structures. Numerous experimental and computational investigations [Johnson& Ham, 1972, McCroskey et al. (1974, 1975, 1981), Jumper et al.(1987), Carr, 1988, Ericsson & Reding et al. (1985, 1988), Walker et al (1985), Biber & Zumwalt, 1993, Green & Galbraith 1994, Krothapalli et al. (1995), Rhee, 2002, Mittal & Saxena, 2002, Ferrecchia et al. 2003, Soltani & Davari, 2004] have shown that the unsteady flow can be separating or reattaching over a large portion of the top surface of the airfoil.

The predominant feature of dynamic stall is the formation, shedding, and convection over the upper surface of the airfoil of an energetic vortex-like disturbance from the leading edge of the airfoil. It induces a nonlinearity fluctuating pressure field and produces transient variations in forces and moments that are fundamentally different from their steady state counterparts. As a result, the maximum values of lift, drag, and pitching moment can highly exceed their static counterparts. After the leading edge vortex passes the airfoil trailing edge and goes in to the wake, the flow progresses to a state of full separation over the upper surface. This is accompanied by a sudden loss of lift and a decrease in pitching moment. Furthermore, when the angle of attack becomes low enough, the flow will finally reattach again from the leading edge [McCroskey et al. (1982)]. The causes of this delay of stall, often accompanied with undesirable hysteresis in lift and moment coefficients, have challenged aerodynamicists for many years. Delay in the reattachment of the flow during the pitch down mode is mainly responsible for hysterisis in the sectional properties near the stall angles.



An experimental investigation on the two element airfoil was carried out by Reddy & Poddar (2005) to study the hysterisis effect and the effect of flap deflection & overhang on the aerodynamic characteristics. Among the considerable dynamic-stall prediction efforts, Ericsson & Reding et al. (1985) reported that for a constant pitching NACA0012 airfoil, the unsteady airfoil stall is characterized by two distinctly different flow phenomena: the delay of stall due to time-lag and boundary-layer-improvement effects, which is quasi-steady in nature, and the transient behavior of the forward movement of the separation point and the following ‘spillage’ of a leading-edge or dynamic vortex (LEV). Dynamic stall studies can be performed by oscillatory as well as ramp motions. The current investigation focused on two-dimensional light dynamic stall (lower reduced pitch rate) by constant pitch-rate ramp motion, which is a simpler flow regime in which flow separation and reattachment plays a fundamental role.

EXPERIMENTAL METHODOLOGYExperiments were carried out on NLF - 0416 airfoil model, having a chord length (c) of 0.75m, at the National Wind Tunnel Facility (NWTF), Indian Institute of Technology Kanpur. It is a low speed, closed-return, continuous, atmospheric wind tunnel having test section 2.25m height, 3m width, and 8.75m length with maximum free-stream velocity 80m/s. The free-stream turbulence level in the test section is within 0.2%. Airfoil model for the present investigation was fabricated out of solid teak wood blocks. The spanwise length of the model is 1.28 m. Two square spars spanning the tunnel height were used as main load bearing members. The airfoil model was mounted vertically between the bottom and upper turntable in such a manner that the mid chord of the airfoil coincided with the centre of the turntable Fig. 3. The model was aligned to set the angle formed between the tunnel centre line and the model chord line to zero degree (within ±0.020

accuracy). End plates were installed at both ends of the model to eliminate the tip effect and to ensure 2D flow at the centre of the model.Surface pressure measurements were made with 60 pressure ports on the model located chordwise at mid span both on

upper and lower surfaces of airfoil model. Pressure port locations are shown in Fig. 2. Pressure ports were mounted on the surface of the airfoil by inserting steel tubes, which were flush to the airfoil surface and projecting inside the model. These pressure ports were connected to pressure scanners (ESP-32HD from PSI) through flexible tubes. For the wake survey, a multi-port rake covering the entire width of the tunnel was mounted perpendicular to the airfoil at a distance of twice the chord length and at mid span. Wake survey rake shown in Fig. 1 consists of 96 pressure ports (91 total pressure ports and 5 static pressure ports). The total pressure ports at the centre of the rake are more closely spaced than the ports at the outer part of the rake. The gaps between the ports are linearly varying (10mm to 50mm) from centre to outer part of the rake. This is done to capture the wake accurately where the velocity gradient is large. Static pressure ports are located at the center, and two each on both sides of the rake. The total pressure sensing was done using steel tubes of sufficient stiffness and the length to ensure that the upstream influence of the rake body does not alter the measurement. Pressure measurement was carried out using ESP pressure scanners from PSI, USA. Data acquisition and analysis was carried out using in-house developed LabVIEW® based application software and PXI–based data acquisition system from National Instruments, USA. For surface pressure measurement, at each measurement point, 5sec data were acquired at a rate of 500 samples per sec. The solid blockage for the airfoil model was about 6% for the maximum angle of incidence of 25° and data are not corrected for blockage and presented as it acquired. Airfoil surface pressure data was used to estimate lift, while the wake survey data was used for drag calculation. Drag coefficient was computed form the wake pressure using momentum deficit method and lift on the airfoil was computed by integrating the static pressures over the model surface. Uncertainty in the calculation of surface pressure coefficients (Cp) is about 0.15% and for drag coefficient (Cd) & lift coefficient (Cl) is about 0.2%.

To determine the aerodynamic characteristics of the hysteresis associated with the pitching motion of the airfoil, its effects were analyzed at varied pitch rate (1 deg/s to 10 deg/s) & Reynolds number (1.5 to 2.5 million) for pitching angle (α) ranges from -5° to +30° deg at c

Uακ

∞= ˙

Reduced pitch rates () ranging


from 0.0004 to 0.004, whereis the pitch rate in radians per second and U

” is the free stream velocity of flow over

airfoil.RESULTS & DISCUSSIONStudy of hysteresis effect in aerodynamic performance of airfoil is important since it is not desirable. Hysteresis has been considered as a low Reynolds number phenomenon [Biber et al. (1993), Marchman et al. (1985)], occurring in post stall recovery of single element airfoils. However, the present study shows the hysteresis effect on the aerodynamic force coefficients at a relatively high Reynolds numbers also [Lorber et al. (1988)].This also can be described as lag of symmetry between the pitch-up and pitch-down parts of the cycle. The difference between pitch-up and pitch-down lift coefficient values at the same angle of attack defines a measure of the hysteresis loop.

Effect of Pitch Rate and Reynolds Number on

To study the effects of pitch rate and Reynolds number on the Hysteresis for a given NLF-0416 airfoil, surface pressure measurements were carried out by varying the pitch rates of the ramping motion at different Reynolds numbers based on the airfoil chord. The range of the ramping motion in a sweep mode of experiments has been selected from -50 to 300 assuring the airfoil to stalled within this range while Reynolds number is varied in the steps viz. 1.5 x 106, 2 x 106 & 2.5 x 106 corresponding to free stream velocity of 30m/s, 40m/s & 50m/s respectively.

a. Hysteresis effects on surface pressure distribution.

Hysteresis effect on the surface pressure distribution at varied pitch rates viz. 1 deg/sec, 4 deg/sec & 8 deg/s for all Re range during pitch up & pitch down ramping motion were analyzed. Results for Re=2.5 x 106 are shown in 3D plots Fig. 4. Stall angles for all the range of pitch rates are identified for both pitch up and pitch down ramp motions. Distinct variation is observed in Cp distribution between the pitch-up and pitch-down motion. A range of predefined similar angle of attack for both the pitch-up and pitch-down are selected to understand Hysterisis effect at varied pitch rates. The surface pressure distribution has been analyzed at unstalled region (50 & 150) and near stall regions viz. pre-stall region (200) & post-stall region (250). The results(Fig. 5) indicates that there is no significant change in the Cp distribution during pitch up and pitch down motion at unstalled & post-stall region (Fig.5a, 5b & 5d), but indicates significant change in trend for pre stall region (Fig.5c). This is because the flow in the pre-stall and in the close vicinity of stall angle, do not cope up with the adverse pressure gradient thus looses momentum and gets separated during pitch up butadversely

causes delay in the reattachment of flow while pitch down. Further at high post-stall angles this time lag in minimal similar to unstalled Cp distribution.different Reynolds numbers.For pitch up and pitch down at different Reynolds numbers and pitch rates, the stalling angle of attack and reattachment angle of attack variations are indicated in Fig. 6. It shows that the stall angle at pitch-up and reattachment angle at pitch-down motion significantly varies with the increase in pitch rates and the effect is almost unaltered with varying Reynolds number for a given pitch rate. So for a given pitch rate at particular position no significant change in observed in stall & reattachment angles by varying Reynolds number. Based on this analysis for a given pitch rate, at each Reynolds number, there is a significant shift in loop without much change in size of the loops. Moreover, increase in pitch rate shifts the separation point and reattachment point to higher angles of attack.

(lift and drag coefficients)

To study the pitch rate effect on lift and drag coefficients, for Re 2.5 million, as can be seen from the corresponding Fig.7, the effect of pitching motion is greatly reduced for low pitch rate (1deg/s) in comparison to high pitch rate (8 deg/s) case. The pitch-up and pitch-down lift for low pitch rate is almost same from 00 to180 angle of attack, in comparison the lift was equal only up to 160 for the high pitch rate. The hysteresis loop that occurs close to stall (@ 200) is relatively small with the pitch-up lift co-efficient of 1.32 and the pitch-down lift of 1.10. The lift coefficient change was only 0.22, this represents a 17 % drop in comparison to the pitch-up. Whereas for the high pitch rate, the change was 0.4 indicating 29% drop To measure the hysteresis loop, two lift coefficient values at the same angle of attack were selected, one on the pitch-up and other on the pitch-down. For example (Fig. 7c), for 8 deg/s pitch rate & Re 2.5 million, at 200 angle of attack, the pitch-up lift coefficient was 1.35 and the pitch-down was0.95. The maximum lift coefficient of 1.37 occurred at 220 and the corresponding hysteresis lift was 0.97. The change in lift coefficient at 8 deg/s pitch rate at 220 was 0.4, a 29 % drop in lift in comparison to the pitch-up. These changes between the pitch-up and the pitch-down are significant. The loads in the actual airfoil change drastically and these have to be incorporated into the structural design, since otherwise structural failure may occur.


with only a rise of approximately 4% in the lift coefficient. The hysteresis loop size was approximately thrice with the pitch rate increase from 1 deg/s to 8 deg/s. The hysteresis loop increases in size for high pitch rate and then decreases for low pitch rate, since at that high angle of attack for the low pitch rate, airfoil surface is largely stalled. While for the 8 deg/s pitch rate case, Most of the curves do not depart from its linear slope until 150 angle of attack. When the pitch rate increases, the size of the hysteresis loop gets enlarged due to the lag of boundary layer response behind the angle of attack variation. The reattachment angle of attack is also being delayed for each increase in pitch rate. This due to the flow is unable to keep up with the angle of attack change. However, does not rule out the possible existence of hysteresis behavior at low angles of attack and that can be seen from the plots. At low angle of attack, during pitch-down the lift coefficient slightly increases when compared to pitch-up case with the parallel shift in lift curve. The drag curve shows that the loop gets enlarged for each increase in pitch rates. The pitching moment generally for pitch-up is more negative than the pitch-down. For the high pitch rate case, the moment stall is more sharp when compared to low pitch rate case. To see the Reynolds number effect on hysteresis behavior, the lift coefficient curves of same pitch rate with different Reynolds number were chosen. For example at 8 deg/s pitch rate case at 220 angle of attack, the pitch-up lift coefficient was 1.37 and the pitch-down lift coefficient was 0.97, the change in lift coefficient was 0.4, a 29 % drop in lift in comparison to the pitch-up for the Reynolds number of 2.5 million. At the same angle of attack, the pitch-up lift coefficient was 1.31 and the pitch-down lift coefficient was 0.90, the change in lift coefficient was 0.41, a 31 % drop in lift comparison to the pitch-up for the Reynolds number of 1.5 million and a 30 % drop in lift for the Reynolds number of 2.0 million.

d. Hysteresis effects in unstalled region

Experiments were also conducted to study the hysteresis behaviour in the unstalled regions for various pitch rates. Fig. 8, 9, & 10 shows the hysteresis behavior in unstalled region (low angle of attack region covers the aerodynamic coefficients less than 200 angle of attack) at Reynolds number 2.5 million. At low pitch rate say 2 deg/s, the hysteresis is almost negligible. But for high pitch rate say 10 deg/s in Fig. 8 shows the well formed hysteresis loops.

In unstalled region, the hysteresis loops are counter clockwise direction, where in closure to stall region, loops formed in clockwise direction. That is closure to stall region, pitch-down lift is lower than the pitch-up but in unstalled region, pitch-down lift is higher than pitch-up. However, during pitch-down the values of the aerodynamic coefficients are high when compared to pitch-up motion in unstalled region. There may be reason that moving wall effect, when the airfoil pitches down, that is equivalent to plate moving upstream and hence once the flow is reattached, the relative instantaneous velocity may be increased during pitch-down and hence increase in negative pressure in turn it increases the lift coefficient.Increasing the value of aerodynamic coefficients in each increase in pitch rate reflects the delay in movement of separation point, while the increasing incidence is the result of the pitch rate suppressing the stall. Pitch rate was the dominant influence; however, motion time history was also important. Nevertheless, unfortunately in this experiments do not have time series data for analysis. The boundary layer may become thinner on the pitch-up of the pitching motion and stays attached until higher angles of attack. After separation, the boundary layer characteristics do not return to those of the pitch-up, thus creating the hysteresis loop, in which a different boundary layer thickness exists. These changes produce stress on the structure of the wing and have to be incorporated into the design.

CONCLUSIONS

From the above analysis, following are the general conclusions:· From the examination of the instantaneous pressure distributions, it is concluded that the NLF 0416 airfoil exhibits trailing edge separation in which separation moves from trailing edge to leading edge pitch up motion and vice versa for all the conditions studied.

· As the reduced pitch rate (k) increases, dynamic increments to the aerodynamic coefficients and the stall incidence also increases. Moreover, hysteresis loop is enlarged for each increase in pitch rate due to increment in lag associated with the movement of separation and reattachment point. The reason is the lag of movement of separation/reattachment during pitch-up/pitch-down motions, as boundary layer is unable to keep up with change in angle of attack.


For the given pitch rate, increasing the Reynolds number delays the stall angle of attack and it shifts the hysteresis loops to higher angle of attack with slight reduction in size of hysteresis loops. Since Reynolds number fixes the separation point and for high Reynolds number, the flow is able to sustain more friction and adverse pressure gradient.

· During pitching motion, once the flow is fully separated at high angle of attack, there may be chance of development of mild vortices at the leading edge and it convects along the surface and at the same time formation of trailing edge vortices. That is why there is an increase in suction pressure in stalled region in the aft chord on the upper surface of airfoil.

· During the course of experiments, in almost all cases, the imprint of the dynamic stall vortex is not evident in all of the instantaneous pressure distributions. The aerodynamic characteristics showed in the results shares the many of the features of the static stall but in addition, results in large hysteresis loops on aerodynamic loading, under which conditions the flow is expected to be characterized by light dynamic stall, in the sense already defined.

ACKNOWLEDGMENTSTechnical support and assistance provided by the research engineers and staff of National Wind Tunnel Facility at IIT-Kanpur is gratefully acknowledged.

REFERENCES

Biber, K., and G. W. Zumwalt, (1993), “Hysteresis Effects on Wind Tunnel Measurements of a Two-Element Airfoil”, AIAA Journal, 31 (2), 326-330.

Carr, L. W., (1988) “Progress in Analysis and Prediction of Dynamic Stall”, Journal of Aircraft, 25 (1), 6-17.

Ericsson, L. E., and J. P.Reding, (1985), “Dynamic Overshoot of static Stall Angle”, Journal of Aircraft, 22 (7), 637-638.

Ericsson, L. E., (1988), “Moving Wall Effects in Unsteady Flow” Journal of Aircraft, 25 (11), 977-990.Francis, M. S., and J. E. Keesee (1985), “Airfoil Dynamic stall Performance with Large-Amplitude Motions”, AIAA Journal, 23 (11), 1653-1658.

Ferrecchia, A., F. N. Coton, R. A. McD. Galbraith, and R. B. Green (2003), “An Investigation of dynamicstall onset on the pitching wing”, Aeronautical Journal,

107(2803), 487-494.

observed during aerofoil ramp-down motions from the fully separated state”, Aeronautical Journal, 98 (979), 349-356.Jumper, E. J., S. J.Schreck, and R. L.Dimmick (1987),

“Lift-Curve Characteristics for an Airfoil Pitching at Constant Rate”, Journal of Aircraft, 24 (10), 680-687.

Johnson, W. & Ham, N. D. 1972 “On the mechanism of dynamic stall”, J. Am. Helicopter Soc. 36–45.

Lorber, P. F., F. O. Carta, , and A. F. Cavino (1988), “Airfoil Dynamic Stall at Constant Pitch Rate and High Reynolds Number”, Journal of Aircraft, 25 (3), 548-556.

McCroskey, W. J., and J. J. Philippe (1975), “Unsteady Viscous Flow on Oscillating Airfoils”, AIAA Journal, 13 (1), 71-79.

McCroskey, W. J., K. W. McAlister, L. W.Carr, S. L. Pucci, O. Lambert, R. F. Indergrand, (1981), “Dynamic Stall on Advanced Airfoil Sections”, Journal of the American Helicopter Society, 26(3),40-50.

McAlister, K. W., L. W. Carr (1979), “Water Tunnel Visualizations of Dynamic Stall”, ASME Journal of

Fluids Engineering, 101, 376-380.Martin, J. M., R. W. Empey, W. J.McCroskey, and F. X.

Caradonna (1974), “An Experimental Analysis ofDynamic Stall on an Oscillating Airfoil”, Journal

of American Helicopter Society, 19, 26–32.McCroskey, W. J. 1982 “Unsteady airfoils”. Annu. Rev.

Fluid Mech. 14, 285–311.Mittal, S., and P.Saxena (2002), “Hysteresis in Flow Past a

NACA 0012 Airfoil”, Computer methods inapplied mechanics and engineering, 191 (19-20),

2207-2217. Reddy, K.S.V., and K.Poddar (2005), “Experimental study

of a two-element airfoil”, Journal of Institutionof Engineers (India), Aerospace Engineering

division, 86, 58-68.Rhee, M. J., (2002), “A Study of Dynamic Stall Vortex

Development using Two-Dimensional data fromthe AFDD Oscillating Wing Experiment”,

NASA/TM-2002-211857, AFDD/TR-02-A-009. Shih, C., L. M.Lourenco, and A. Krothapalli (1995),

“Investigation of Flow at Leading and Trailing Edgesof Pitching-Up Airfoil”, AIAA Journal, 33 (8),

1369-1376.Soltani, M. R., and A. R.Davari (2004), “On self-similar

behavior of the hysteresis loops in pitching motions”, Aeronautical Journal, 108 (2869), 427-434.

Walker, J. M., H. E. Helin, and J. H. Strickland (1985), “An Experimental Investigation of an AirfoilUndergoing Large-Amplitude Pitching Motions”, AIAA Journal, 23(8),


Fig. 1. Airfoil model and wake survey rake mounting positions in the wind tunnel.

Fig. 2. Pressure port locations on the airfoil.

Fig. 3. Model mounting in the tunnel




Fig.6. Stalling angle for pitch up and pitch down at different Reynolds numbers.


Fig. 7a Hysterisis effect on Cl and Cd at Pitch rate: 1 deg/sec at Re = 2.5 x 106

Fig. 7b Hysterisis effect on Cl and Cd at Pitch rate: 4 deg/sec at Re = 2.5 x 106

Fig. 7c Hysterisis effect on Cl and Cd Pitch rate: 8 deg/sec at Re = 2.5 x 106


Fig. 8 Hysteresis loop in the Cl and Cd during the airfoil ramped up to 20° at pitch rate of 10 deg/s for Re = 2.0 ×106.




ROBUST IMAGE WATERMARKING USING DWT

S.DHANDAPANI1 AND DR.A.KRISHNAN2

1Asst. Professor, ECE Department,Sree Sastha Institute of Engineering and Technology,

Chennai – 602 103, IndiaEmail: [email protected]

2Dean,K.S.Rangasamy College of Technology,

Tiruchengode – 637 215, India

ABSTRACT

Digital Watermarking have gained popularity in recent years as a means of protecting multimedia data from illegal copying and unlawful reproduction. In this paper an image watermarking algorithm which is blind (do not require the presence of host image for detection), and robust is proposed. The proposed watermarking scheme embeds the watermark in the Discrete Wavelet Transform (DWT) domain. The watermark is a binary logo and is embedded in the subband of the wavelet decomposition. This watermarking scheme is robust to several attacks like JPEG compression, rotation, Gaussian filtering, median filtering, cropping, scaling, sharpening, Gaussian noise and salt & pepper noise.

Keywords – Digital Watermarking, Robustness, DWT

I. INTRODUCTION

Over the past few years digital watermarking has become popular due to its significance in content authentication and legal ownership for digital multimedia data. A digital watermark is a sequence of information containing the owner’s copyright for the multimedia data. It is inserted invisibly in host image so that it can be extracted at later times for the evidence of rightful ownership. An ideal ownership-identifying watermarking should include Imperceptibility, Robustness and Unambiguity.

Imperceptible: The image must not be visibly degraded by the

presence of the watermark. Robustness: The watermark must be robust to attack and must be tolerant to reasonable lossy compression. Standard image processing operations such as low pass filtering, cropping, translation and rescaling should not remove the mark.Unambiguity: Retrieval of the watermark should unambiguously identify the owner.Available digital watermarking techniques can be categorized into one of the two domains, viz., spatial and transform, according to the embedding domain of the host image. The simplest technique in the spatial domain algorithms is to insert the

watermark image pixels in the least significant bits (LSB) of the host image pixels. The data hiding capacity in these algorithms is high. However, these algorithms are hardly robust for various attacks and prone to tamper by unauthorized users.Watermarking in transform domain is more secure and robust to various attacks. Among the transform domain watermarking techniques Discrete Wavelet transform (DWT) based watermarking techniques are gaining more popularity since DWT has a number of advantages over other transforms including space frequency localization, multi resolution representation, superior Human Visual System (HVS) modeling, linear complexity and adaptivity. These are the key reasons for the success of wavelets in many signal processing applications including watermarking. The wavelet transform decomposes the image into four subbands called LL, LH, HL and HH. Wavelet transform is computationally efficient and it reflects the anisotropic properties of HVS. Magnitudes of DWT coefficients are larger for LL band compared to other bands. The larger the coefficient the more significant it is. Using DWT an image can be shown at different levels of resolution. The blocking artifact problem is less severe in DWT than DCT, since there is no block processing in DWT. ‘Cox et al. [1] proposed a watermarking method to embed a watermark by the spread-spectrum technique. The watermark is inserted in the perceptually significant portion of an image wherein a predetermined range of low-frequency components



excludes the dc component. The watermark is spread over many frequency coefficients, so that the number of coefficients which are modified is very small and difficult to detect. Choi et al [2] proposed a multi-purpose logo embedding method based on vector projection on Gaussian random subspace from DCT domain. Their approach could obtain better visual quality for watermarked image, but cannot being robust enough to Gaussian noise. An additive watermarking technique using image fusion technique is proposed by Kundur and Hatzinakos [3]. Zhang et al. [4] divided the original image into n x n blocks and transformed them into a DWT domain. The watermark is embedded by using the mean and the variance of a subband to modify the wavelet coefficient of a block.

II. PROPOSED SCHEME

A Discrete wavelet transform when applied to an image transforms the image to images of various resolutions. A one level DWT decomposition gives four sub-bands, namely LL, HL, LH and HH. These sub bands are shown in Fig 3. Most of the energy is contained in the LL band. Human Visual System is known to be less sensitive to, such as the high resolution detail bands LH, HL, and HH. The HH band will be easily removed by JPEG compression. So, the proposed watermarking scheme embeds the watermark in the LH and HL band using Haar wavelet. A Haar wavelet transform is conceptually simple and fast. It is exactly reversible without any edge effects. The LH and HL values are modified according to a Pseudo random Number (PN) sequence for the watermark bit 0. Then inverse DWT is used to get the watermarked image. During watermark extraction process, if the correlation of LH and HL values of watermarked image and the PN sequence is greater than the mean correlation then the watermark bit is set to 0.

A. Watermark Embedding Algorithm

Host Image – it is a gray-scale image to be watermarked.Watermark Image – it is a binary image act as watermark.Key –numeric key used for watermark embeddingAlgorithm1. Perform DWT decomposition of the Host Image at level one. Four bands LL, HL, LH and HH are the host images of various resolutions. And store Approximation, horizontal, vertical and diagonal

coefficients in LL1, LH1, HL1 and HH1 respectively. 2. Find the size of HL1 matrix and store it in S.

3. Initialize the state of Random number generator to Key.

4. For each bit of watermark perform the following steps:

(i) Create a random matrix of size S with random number generator and store it in RHL.

(ii) Calculate

RHL1 = round (2*(RHL – 0.5))

(iii) Create a random matrix of size S with random number generator and store it in RLH.

(iv) Calculate

RLH1 = round (2*(RLH – 0.5))

(v) If bit at current position in watermark has value Zero, then

Set HL1 = HL1+ k * RHL1.

Set LH1 = LH1 + k * RLH1.

5. Perform Inverse Discrete Wavelet Transform (IDWT), to create the watermarked image.

Output

Watermarked Image – it is a gray-scale image watermarked with binary image.

B. Watermark Extraction Algorithm

Input

SW – size of original binary watermark

Key – key for watermark extraction

Algorithm

1. Perform DWT decomposition of the Watermarked Binary Image at level one. And store Approximation, horizontal, vertical and diagonal coefficients in LL1, LH1, HL1, and HH1 respectively.

2. Find the size of HL1 matrix and store it in S.

3. Initialize the state of Random number generator to Key.

4. Find number of bits in the watermark and store in N.

5. Create a matrix with one row and N columns with all ones and store in variable Watermark.


6. Repeat the following for i = 1 to N

(i) Create a random matrix of size S with random number generator and store it in RLH.

(ii) Calculate

RLH1 = round (2*(RLH – 0.5)).

(iii) Create a random matrix of size S with random number generator and store it in RHL.

(iv) Calculate

RHL1 = round (2*(RHL – 0.5)).

(v) Find the correlation between LH1 and RLH1 and store it in corr_h(i).

(vi) Find the correlation between HL1 and RHL1 and store it in corr_v (i).

(vii) Calculate

corr (i) = (corr_h + corr_v) / 2

7. Find mean corr and store it in mean.

8. Repeat the following for i = 1 to N

If (corr (i) > mean)

Set Watermark (i) = 0

9. Reshape the Watermark in size SW.

Output

Watermark – it is the recovered binary watermark.III. EXPERIMENTAL RESULTS AND OBSERVATIONS

Experiments were conducted using Lena as host image. The two images host image and watermark image are shown in Figure 1 and 2 respectively. The size of the host image is 512x512 pixels. The size of the watermark image is 12x9 pixels. A Haar Wavelet filter is used for wavelet decomposition. The Host image is decomposed into four subbands LL, LH, HL

and HH. This is shown in Figure 3. The watermark image is embedded in the LH and HL sub-bands.

The visual appearance of the watermarked image is good showing no significant artefacts or distortions because of the process of watermarking. The Peak Signal to Noise Ratio (PSNR) of the watermarked image is 34.3 dB and is shown in Figure 4.

Figure 1. Host Image Lena (512x512)

Figure 2. Watermark Image (12x9)

LL HL

LH HH


Figure 3. 1-Level Haar Wavelet Decomposition

Figure 4. Watermarked imageVarious attacks [7] used to test the robustness

of the watermark are JPEG compression, Rotation, Resizing, Gaussian filtering, Median filtering, cropping, sharpening, Gaussian noise, salt & pepper noise. The extracted watermarks after applying various attacks are shown in Figure 5.

A. JPEG Compression

The watermarked image is compressed using lossy JPEG compression using MATLAB. The index of the JPEG compression ranges from 0 to 100, where 0 is best compression and 100 is best quality. The reconstructed watermarks for various indices are shown in Figure 5. The proposed scheme works well even for extreme compression.

B. RotationAlthough an image is rotated in a small

degree, the positions of pixels will be shifted. It is then difficult to recover the original image from a rotational image. The result of grouping coefficients into blocks for a rotation image will be quite different from that of the original image. Though the rotation attack does not cause the image serious visual distortion, the watermark extraction will result in error. The results are shown in Table 1.

C. Scaling

Resizing operation first reduces or increases the size of the image and then generates the original image by using an interpolation technique. This operation is a lossy operation and hence the watermarked image also looses some watermark information. In this experiment first the watermarked image is reduced from 512x512 size to 256x256. By

using bilinear interpolation its dimensions are increased to 512x512. The extracted watermark as shown in Figure 5 is good.

D. Gaussian and Median filter

For Gaussian low pass filtering attack with a standard deviation of 0.5 and 1.0 is used. The recovered image is perfect showing its resistance to Gaussian low pass filtering attack. Median filter is a non linear spatial filter which is usually used to remove noise spikes from an image. The watermarked image is attacked by median filtering with a 3x3 mask, 5x5 mask and 7x7mask. The median filtered watermarked image is more blurred than low pass filtered image. The extracted watermark is still recognizable.

E. Cropping

Cropping operation deletes some portion of the image. This is a lossy operation. In this experiment half of the watermarked image is cropped and then watermark is extracted. The extracted watermark is very good even after 25% cropping.

F. Noise

The watermarked image is attacked by Sharpening, Gaussian noise with a variance of 2 and salt & pepper noise of 0.05 noise intensity. The extracted watermark is good after all these attacks.For all kind of attacks, the extracted watermark is good compared to the watermark extracted by other methods [5] and [6]. Finally the proposed algorithm has shown resistance to all attacks as shown in Fig 5.The Normalized Correlation is used as a metric to compare the robustness and summarized in Table 1. After extracting the watermark, the normalized correlation coefficient (NC) is computed using the original watermark and extracted watermark to judge the existence of watermark. It is defined in Eq.(1).

NC = / (h.w) ———— (1)where h and w are the height and width of the watermark, respectively; and w(i,j) and v(i,j) are the values located at coordinate (x,y) of the original watermark and the extracted watermark. Here w(i,j) is set to 1 if it is a watermark bit 1; otherwise, it is set to -1. v(i,j) is set in the same way. So the value of w(i,j) . v(i,j) is either -1 or 1.

Table 1. NC values for Various Attacks


JPEG-100% JPEG-90%

JPEG-80% JPEG-70%

JPEG-60% JPEG-50%

JPEG-40% JPEG-30%

JPEG-20% JPEG-10%

Rotation-0.250 Rotation-0.300

Rotation- 10 Resize(512-256-512)

Median Filtering(3x3) Median Filtering(5x5)

Sharpening Gaussian Filtering

Gaussian noise Salt and Pepper noise

Figure 5. Extracted Watermarks for various attacks


IV. CONCLUSION

In this paper a blind image watermarking algorithm based on the wavelet domain is presented. The watermark image is embedded in the LH and HL sub-bands. The proposed algorithm is shown to be robust to various attacks, JPEG compression, rotation, scaling, cropping, median filtering, Gaussian low pass filtering, sharpening, Gaussian noise and salt &pepper noise. The robustness of the proposed algorithm is very good to all types of attacks as compared with other methods [5] and [6]. This means that an embedded watermark is still recoverable even after common image processing operations on the watermarked image. Hence the proposed method is suitable for copyright protection.

REFERENCES

[1] Cox, I. J.; Kilian, J.; Leighton, F. T. and T. Shamoon, T. 1997. “Secure spread spectrum watermarking for multimedia,” IEEE Transactions on Image Processing.,vol. 6, no. 12, pp. 1673-1687.

[2] Choi, Y.H. and Choi, T.S. 2006. “Multipurpose logo embedding method for copyright protection and

authentication” IEEE International Conference on Consumer Electronics, pp. 241-242.

[3] Kundur, D. and Hatzinakos, D. 2004. “Towards Robust Logo Watermarking using Multiresolution Image Fusion,” IEEE Transactions on Multimedia, vol. 6, pp. 185-198.

[4] Zhang, G.; Wang, S. and Wen, Q. 2004. “An adaptive block-based blind watermarking algorithm,” Proc. IEEE ICSP, pp. 2294-2297.

[5] Wang,S. and Lin,Y. 2004. “Wavelet Tree Quantization for Copyright Protection Watermarking”, IEEE Trans. Image Processing, vol.13, no.2, pp. 154-165.

[6] Wei-Hung Lin et al, 2008. “An Efficient Watermarking Method Based on Significant Difference of Wavelet Coefficient Quantization” IEEE Transactions On Multimedia, vol. 10, no. 5. pp. 746-757.

[7] Ping Dong, Brankov, G.; Galatsanos, P.; Yongyi Yang, and Davoine,F. 2005. “Digital Watermarking Robust to Geometric Distortions” IEEE Transactions on image processing, vol.14, no.12. pp. 2140-2150.


MULTIOBJECTIVE APPROACH FOR FEEDER OVERLOADING AND SERVICE RESTORATION THROUGH

RECONFIGURATIONP. RAVI BABU1, K. PRAPOORNA2, N.V. PRASHANTH3, A.SHRUTI4, D.

PRABHUVARDHAN5, V.P. SREE DIVYA6

1. Associate Professor, EEE Department, CVR College of Engineering, Hyd, [email protected]. Student, EEE Department, CVR College of Engineering, Hyd, [email protected]. Student, EEE Department, CVR College of Engineering, Hyd, [email protected]

4. Student, EEE Department, CVR College of Engineering, Hyd, [email protected]. Student, EEE Department, CVR College of Engineering, [email protected]

6. Student, EEE Department, SNIST, Hyd, [email protected]

ABSTRACTThis paper presents the method to solve the feeder over loading problem and service restoration strategies for affected zones due to fault in a three feeder distribution network which deals with the estimation of load changes through feeder reconfiguration using the heuristic search technique This method is rooted in the compulsion that, the number of switchings and total I2R losses are minimum Software has been developed using C for the above work. Keywords: Reconfiguration, loss reduction, service restoration, distribution network, heuristic search.

I. INTRODUCTIONDistribution feeders are usually radial to simplify the over current protection. The functions carried out through a distribution system control centre depend on the state, the system is in. If the load is being supplied and all operating constraints are being satisfied, then the objective may be say minimization of system losses. But, if some load has been disconnected due to a major fault, then the objective will be to restore service as soon as possible. Such apparently distinct functions share a common feature solving a combinatorial problem whose solution is a set of swithching operations. To help restore power to customers following a fault, most feeders have several interconnecting tie switches to neighboring feeders which are used for fault isolation and service restoration. There are numerous number of switches in the distribution system which results in tremendous number of possible switching operations. Feeder reconfiguration, thus becomes a complex decisionmakingprocess.[1] Distribution feeder reconfiguration can be used as a planning tool as well as a real-time control tool. It is a very important and useable operation to reduce distribution feeder losses and improve the system security and also enables load transfer from heavily loaded regions to the lightly loaded, which are performed by changing the status of network switches in such a way that radiality is ensured, after the operations are completed.Service restoration problem can be solved with the application of automation techniques to electric utility distribution systems which would allow better investment utilization as well as improved service quality. Most specifically, automation

may reduce the time for fault isolation and service restoration to unfaulted areas. The conventional approach to power system computer applications rely on the use of numerical algorithms coded in procedural languages like C and C++. While the conventional approach is suited for quantitative applications, the Artificial Intelligence approach is more oriented to qualitative resolving. Heuristic programming technique used in this paper for service restoration and overloading falls somewhere between the conventional and the Artificial Intelligence approach. This programming technique is used to solve combinatorial problems i.e problems in which finding a solution normally involves analysis of a large number of alternatives or possible combinations.[2]Feeder reconfiguration problem is solved by using the solution method whose features include:

Ø The capability to estimate with minimal computational efforts and the change in losses resulting from feeder reconfiguration.

Ø The criteria that may be used to eliminate undesirable switching options inorder to alleviate the dimensionality problem.The formula used in estimating the change in losses requires little additional information over the base case and it also suggests a filtering mechanism for eliminating those switching options which would not yield loss reduction. Optimal solutions have been obtained both in the case of system overloading as well as restoration by using the methods proposed.

II. PROBLEM STATEMENT



Fault location and isolation of a faulty section in a distribution network in a real-time is made possible with the help of monitoring and control functions in an automated distribution system. But, service restoration to the non-faulted out of service area in real-time poses a real challenge. Fast restoration strategies are necessary to reduce the inconvenience to the user during such interruptions. [3]

Overloading is also one such real-time problem that can be solved by feeder reconfiguration which is performed by opening or closing two types of switches i.e. sectionalizing and tie switches respectively. A whole feeder or part of a feeder may be served from another feeder by using a tie switch linking the two, while an appropriate sectionalizing switch must be opened to maintain radial structures.

Even for a distribution system of moderate size, the number of switching options is so great that, conducting many load flow studies for all the possible options becomes not only extremely inefficient from a computational stand point but also impractical as a real-time feeder reconfiguration. [4]

The problem addressed in this paper is to identify the tie and sectionalizing switches that should be opened and closed respectively to achieve a maximum reduction in losses.III. OUTLINE OF THE APPROACH

Transition problems such as service restoration network configuration, feeder load balancing and loss minimization of a combination of two or more than two individual problems involve the determination of control strategies that are implemented either automatically or with the help of a system operator. Though different problems require different solution procedures most of them share the common base i.e., we are dealing with discrete decision problems. Pruning is important to avoid the uncontrolled growth of the number of alternatives to be investigated. It is possible to design algorithms and flowcharts that generalize practical procedures utilized by system operators. The advantage of the proposed approach is the capability to incorporate practical operating procedures as part of the overall solution process.

The fault problem in the distribution system is solved by reconfiguration considering the minimum switching operations as shown in the flow chart (Fig 4).The problem of overloading is solved by the expression for power loss reduction through load

transfer. The objective of this expression is to determine

Ø Whether a specified switching option would result in a loss increase or decrease.

Ø Among the candidate switching options, which option would yield the greatest reduction in losses.

In other words, relative rather than absolute accuracy is sought here. IV. Heuristic search

The decision problem we face involves deciding on switch/breaker status (open/closed). We proposed to solve this problem by heuristic search on a binary decision tree. By heuristic search, we mean a general search method armed with domain-specific knowledge to guide the search. The search methods allow us to traverse the space of possible system states, whereas domain-specific knowledge is essential in limiting the size of the decision tree.

IVA. BINARY DECISION VARIABLES The decision whether the switch is

open/closed is denoted by the binary decision variable. xk = 1 if the k-th switch is closed. 0 if the k-th switch is open.IVB. DECISION VECTOR

There are 2m possible combinations of switch positions in the system, each one corresponding to a decision vector.xi = { x1, x2, x3, ……xm}Where xk = 0 or 1.IVC. DECISION TREE

The decision process is best illustrated by a binary tree as shown. At the beginning of the process (root node), all the decision variables are undeclared i.e., we have the original problem with m unknowns. As we move down on the tree, the decision variables are declared either as 0 or 1: at level i of the tree, we have (m-i) undeclared variables left. At the bottom of the tree (level m), all the decision variables are declared: each of the 2m nodes (leaves of the tree) is associated with a possible solution (not necessarily a feasible one). Though the tree that must be searched could in principle be built explicitly, this is neither practical nor necessary. An implicit representation of the tree, given by the rules used to generate new nodes from the existing ones, is adopted instead. Thus, as the search process is carried out, only part of the tree is constructed explicitly and most of the tree never needs to be traversed. Domain-specific is used to avoid unnecessary search.


At each node of each tree level, we can choose any undeclared variable to be declared next. When a variable is declared, two new nodes (called successors) are created. The process of tree expansion (called

branching) is illustrated. The choice of next variable to be declared will depend on:

Ø Previous decisions taken on a particular pathØ Practical rules, based on operator

experience, used to guide the search.Thus, it is possible to simulate what an

experienced operator would do under the same circumstances. In addition to guiding the search, practical rules can be used to prune the tree, which may be important in limiting the growth of the number of alternatives to be considered.

IVD. SEARCH STRATEGYThe effectiveness of the method depends on the strategy used to control the search. Strategies such as breadth-first, depth-first can be adopted. In the present implementation of the method, we are using a depth-first search, which allows the direct stimulation of operator procedures as part of the search process like the search shown in the Fig 1a, 1b. Used without heuristics, the depth-first search is simply an exhaustive search on the decision tree. Even though exhaustive search is not suited for practical implementation, it is very useful in testing the effect of heuristics on the optimality of these solutions. As for practical implementations, the appropriate use of heuristics (the ones whose effect on the solution process is known) may be crucial regarding the efficiency of the method. [2]

V. Flow chartIn this section, we describe an implementation of the proposed approach by presenting a simple flowchart as shown in Fig 4. For simplicity, neither heuristics nor pruning rules are considered at this stage. This knowledge includes practical rules and procedures normally applied by system operators to carry out switching operations aimed at restoring service or improving system operating conditions.

Let us consider there is a permanent fault occurring at the load point-(6) which results in affecting the load point-(7) in the given three feeder distribution system shown in Fig 2. The optimal solution obtained in this case by applying the flow chart shown in Fig 4 is “to open the section switch 13 and close the tie switch 26”. The switch positions for this optimal solution are given below in Table 1:

Table-1 Switch positions


VI. Estimation of the loss changes

The amount of loss change resulting from transferring a group of loads from Feeder-II to Feeder-I can be estimated from the following simple equation: DP=Re { 2(SIi? (Em-En)¢ } + RloopS |Ii|

2 ......(1)

iÎDWhere :

D: Set of buses which are disconnected from feeder-II and connected to Feeder-I.

m: The bus of Feeder-I to which loads from Feeder –II will be connected.

n: Tie bus of Feeder-II that will be connected to bus m via a tie switch.Ii: complex bus current at bus iRloop: series resistance of the path connecting the two substation buses of Feeder-I and Feeder-II via closure of the specified tie switch.Em: component of E= RBUS I BUS corresponding to bus m. RBUS is the “bus resistance matrix” of Feeder-I before the load transfer which is found using the substation bus as reference.I BUS is the vector of bus currents for Feeder-I.En; Similar to Em but defined for bus n of Feeder-II.Re: real part, complex conjugate and magnitude operators respectively.It is to be noted that Em and En are computed using base-case bus currents Ii before the load transfer. It is suggested to incorporate the effects of capacitors into bus currents to facilitate computational efficiency. DP represents a kW loss reduction (increase) when it is negative (positive). The second term on the right-hand side of Eq.(1) is always positive. Therefore, a reduction in losses cannot be achieved unless the first term becomes significantly negative. Since complex values are dealt within the first term, it may not be simple to draw any definite conclusions. However, we note that voltage phase-angle differences are small on most distribution systems, and those complex bus currents Ii may be mostly in phase with voltage phasors due to capacitor VAR compensation on well designed systems. Under these circumstances, loosely speaking, the first term becomes negative. |Em| <. |En| It follows the above observation that loss reduction can be attained only if there is a significant voltage difference across the normal open tie switch and if the loads on the higher voltage drop side of the tie switch are transferred to the other side. It will be seen that the above observation can be used as a most attractive criterion to eliminate undesirable switching

options during the elimination process. It is also note worthy that in Eq.(1), information regarding E is required only at the terminal buses where the tie switch is located, and that the configuration of the group of loads to be

transferred or the geographic extent of the overall distribution system does not matter to the result. [4]

Let us consider a three feeder distribution system shown in Fig 3 having the data as shown in Table 2 whose capacities are:20 MW, 14.5 MW & 7 MW respectively.

Table-2 Data of the three feeder system


The loadings of each feeder being 8.5 MW, 15.1 MW & 5.1 MW respectively. The loading & capacity values show that the second feeder is overloaded and the cases of reconfiguration are presented: Case-I: The load at bus 11 is transferred from Feeder-II to Feeder-I by closing the tie line switch 15 and opening the sectionalizing switch 19. In this case, D={11},m=5,n=11 and DP= Re [2I11 ( E5 – E11 )

¢ +Rloop |I11|2

Where Rloop is the total resistance of the path along the branches 11, 12, 15, 19, 18 and 16.Case-II: The load at bus 10 is transferred from feeder-II to feeder-III by closing the tie switch 21 and opening the sectionalizing switch 17.In this case, D={10}, m=14,n=10 and DP= Re [2I10 (E14 – E10 ) ¢ ] +Rloop |I10|

2

Where Rloop is the total resistance of the path along the branches 16, 17, 21, 24 and 22.Case-III: The loads at the buses 9, 11 and 12 are transferred from Feeder-II to Feeder-I by closing tie switch 15 and opening Sectionalizing switch 18. In this case, D={9,11,12},m=5,n=11 andDP= Re [2(I 9+I12 + I11) ( E5 – E11 ) ¢ ]+ Rloop |I 9+I12 +I11|2

Where Rloop is the total resistance of the path along the branches 11, 12, 15, 19, 18 and 16.

Table-3 Percentage of Real Power Losses (DP)

Losses are calculated as percentage of the total load of all the three feeders and are tabulated in Table 3. The calculation of the system losses in the above three cases show that, the CASE-II has the minimum increase in losses (DP) which can be considered as an optimal solution for the overloading problem considered.VII. ConclusionA heuristic search approach to distribution system restoration has been proposed. The method uses a knowledge guided search strategy to solve problems such as service restoration and is based on operating procedures which investigate alternatives that normally would not be considered by system operators which may be very helpful under certain critical operating conditions. It also facilitates to investigate the effect of practical rules on the optimality of the final solution. Computational complexity arising in reconfiguration due to

overloading is identified and a criterion is developed for reducing the number of candidate options which eliminates the study of numerous load flow studies.The algorithm and expression presented in this paper shows promising flexibility that will allow their ready incorporation into the overall feeder reconfiguration and restoration strategies.VIII. AcknowledgementThe authors would like to express their sincere gratitude to the Principal and Authorities of CVR College of Engineering, SNIST, Hyderabad for their encouragement and co-operation.IX. References[1]. Whei-Min Lin et al, “A New approach for Distribution Feeder Reconfiguration for Loss Reduction and Service Restoration”, IEEE Transactions on Power Delivery, Vol.13,No.3,July 1998, pp. 870-875.

[2]. A.B.Morelalo et al, “Heuristic search Approach to Distribution system Restoration”, IEEE Transactions on Power Delivery, Vol.4, October 1989, pp. 2235-2241.

[3]. N.D.Sarma et al, “Real Time Service Restoration in Distribution Systems”, IEEE Transactions on Power Delivery, Vol.9, October 1978, pp. 2064-2070.

[4]. S.Civanlar et al, “Distribution Feeder Reconfiguration for Loss Reduction”, IEEE Transactions on Power Delivery, Vol.3, July 1988, pp. 1217-1223.

[5]. Flavio Vanderson Gomes et al, “A new Heuristic Reconfiguration algorithm for Large Distribution Systems”, IEEE Transactions on Power Delivery, Vol.20, No.3, August 2005, pp. 1373-1378.

[6]. D.W.Ross et al, “New Method for Evaluating distribution automation and Control (DAC) System Benifits”, IEEE Transactions on Power Apparatus & Systems, June 1981, pp. 2978-2986.

[7]. C.H.Castro et al, “generalized Algorithms for Distribution Feeder Deployment and Sectionalizing”, IEEE Transactions on Power Apparatus & Systems, March 1980, pp. 549-557.[8]. C.A.Podbuory et al, “An Intelligent Knowledge Based System for Maintenance Scheduling in a Power System”, Proceedings of Ninth Power Systems Computation Conference, pp. 708-714, Portugal, 1987. [9]. E.Rich, Artificial Intelligence, Mc Graw-Hill, New York, 1983.



AN INNOVATIVE DESIGN AND ENERGY OUTPUT ESTIMATION OF WIND-SOLAR HYBRID RENEWABLE ENERGY GENERATION

SYSTEM WITH RAIN WATER COLLECTION FEATURE

CHONG WEN TONG*, POH SIN CHEW,HAKIM S. SULTAN, PAN KOK CHEN AND DEEP CHEAH HOW

Department of Mechanical Engineering, Faculty of Engineering, University of Malaya,50603 Kuala Lumpur, Malaysia.*Corresponding Author. Tel: +6012-7235038,

Email: [email protected]

ABSTRACTThe proposed wind-solar hybrid renewable energy generation system is a feasible design and new concept of utilization, integration and optimization of existing renewable energy and rain water harvesting technologies. It is compact and can be built on the top of high rise buildings in order to provide on-site green power to that building or feed into the grid line. This system fully utilizes the advantages of Malaysian climate, i.e. high solar radiation and high rainfall over a year for green energy generation and free water supply. It also overcomes the inferior aspect on the low wind speed in Malaysia by guiding and increasing the speed of the high altitude free-stream wind from all directions radially through fixed or yaw-able power-augmentation-guide-vane (PAGV) before entering the wind turbine at center portion. The PAGV is an innovative design used to guide and create venturi effect for increasing the wind speed before the wind-stream enters wind turbine. The system must also be designed to provide optimum surface area for solar panels installation. In addition, rain water collection feature must be built-in to the system design. The system is recommended to be sited on the top of high rise buildings or structures with its appearance can be blended into the building architecture without negative visual impact. With this system, the size of wind turbine can be reduced for a given power output and the noise is also reduced since it is contained in the PAGV. The rain water can be treated, stored, and used for general purposes. The integration of photovoltaic (PV) and solar thermal collectors into the upper wall improves the efficiency of PV panel and supplies hot water for building needs. The design is also safer and suitable to be used in populated area. Mesh can be mounted at the entrance of the PAGV to prevent the bird-strike problem faced by current wind turbine design. This system can eliminate or minimise the current problems concerning public acceptance of wind energy, i.e. poor starting behaviour in low wind speed, wind turbine noise, visual impact, electromagnetic interference, safety and environmental factors. With the 30-metre diameter and 12-metre height of system, the estimated annual energy saving = 160 MWh. With these benefits, the renewable energy can be promoted in populated urban area.

Keywords: on-site renewable energy; power-augmentation-guide-vane; urban wind turbine; solar panel; rain water collector

INTRODUCTIONWind and solar energy has long been recognized as a potential source of free, clean and inexhaustible energy; and it could play an important role in contributing to the energy requirements. Generation of both solar and wind energy requires no usage of water and thus it would not bring out water crisis problem compared to the energy generation from other sources such as biomass and nuclear. Malaysia is situated at equatorial zone, and experiences low speed wind belt consisting of Southwest and Northeast Monsoons a year.

Most of the areas in mainland experience low (free-stream, V∞< 4m/s for more than 90% of total wind hours) and unsteady wind speeds.As a result, most of the existing wind generators (rate wind speed, Vrated = 9-15 m/s) are not suitable for Malaysian applications since they are designed for high wind speeds which prevail in the USA and Europe. From the study conducted by Chong [1], wind energy could only be an economically viable generation of electricity for isolated areas far away from national grid system.In Europe, due to the decreasing number of economic sites, organizations involved in planning are urged to place wind turbine closer



to the populated areas [2]. In order to design a wind energy generation system that can be used in rural and urban areas (lower wind speeds), good starting characteristic of wind turbine or system design that can speed-up the on-coming wind speed at high altitude is required.

1. There are several possibilities that wind energy collection systems can be integrated into urban environments and they can be categorized into three types [3]: Sitting stand alone wind turbine in an urban locations

2. Retro-fitting wind turbines onto existing buildings

3. Full integration wind turbines together with architectural form.

For first option, it may encounter constrains of limitation of land in urban area and low wind speed due to other existed high rise buildings. It may also arise of public acceptance over safety, issues of noise and vibration, as well as visual impact in urban areas.

Meanwhile, for third strategy, it might seem to be fascinating in architectural side as well as quite effective in aerodynamics point of view. World Trade Centre in Bahrain was an example of high-rise building that integrating building augmented design incorporated with three horizontal axis wind turbines of 29 m diameter [4]. This building consists of twin towers facing the prevailing wind direction in order to concentrate the airflow on the three turbines between the towers. However, in realistic, the architecturally fascinating building augmented wind turbine cannot always feasible since it involves huge capital, well-defined construction plan, thus the growth of installed generating system could be very slow. For second approach, to-date, horizontal axis as well as vertical axis Darrieus and H-rotor type’s small-scale micro-wind turbines are commercially available for building integration [5] hence small-scale turbines are easily possible as a viable retrofit solution.

However, such small-scale wind turbines for building integration may not always be visually appropriate and may be hazardous due to turbine blades failures would be occurred.

Apart from wind collection systems; from the demonstrative study for the wind and solar power hybrid system conducted at Ashikaga Institute of Technology in Japan, it was confirmed that there is a complementary relationship between the solar and wind energy after one year operation of the system [6]. Perhaps with development of solar panel technologies, the wind-solar harvester could be widespread in urban areas.

Therefore, an innovative design and compact system that integrates several green and renewable energies harvesting technology (wind-solar hybrid energy generation system and rain water collector) is aimed for the use on high rise buildings or structures. The system can be retrofitted as a part of building architecture without negative visual impact and overcomes the conventional wind turbine weakness such as acoustic or noise problem, bird strike, electromagnetic interference, sustainability and safety concern caused by the blades failure.

GENERAL ARRANGEMENT AND WORKING PRINCIPLES This patented design integrates and optimizes several green elements; including urban wind turbine, solar array and rain water collector [7]. The general arrangement of this system is shown in Figure 1. The system can be of cylindrical shape or any shape of design, depending on the building architecture profile, such as in ellipse shape, etc. The vertical axis wind turbine (VAWT), [A] is located in the middle of the system, surrounded by the power-augmentation guide vane (PAGV). The PAGV consists of upper wall duct [D], lower wall duct [E] and guide vanes [B]. The PAGV is designed to be fixed or yaw-able with the help of rudder, [C] or pressure sensors and servo-motor, to face the on-coming wind stream. The PAGV collects wind stream radially from a larger area and create a venturi effect to guide and increase wind speed before entering the wind turbine. The wind turbine can take shape of any existing vertical axis or horizontal axis wind turbine design or their combinations. A preferred Darrieus straight-bladed vertical axis wind turbine is chosen and is powered by phenomenon of aerodynamic lift force. The wind turbine has a common rotating axis with PAGV.

The center drive shaft of VAWT is coupled with the generator, [I] through the power transmission shaft and mechanical


drive system, [H] such as gear system. The radius of PAGV is greater than the radius of VAWT. The PAGV consists of vanes in variable sizes or shapes

having constant or varying thickness, in which they are positioned surrounding the turbine. The vanes form multiple flow channels which are utilized to speed up the wind stream by creating venturi effect and to guide wind stream to the optimized angle of attack on the VAWT blades.

Figure 1: General arrangement of wind-solar hybrid renewable energy generation system with rain water collection feature – (a) side sectional view; (b) perspective view.

The upper wall duct and lower wall duct are inclined at an angle, θ from the vertical plane. The exterior surface of the upper wall of PAGV provides the base for placement of solar panels (solar photovoltaic and/or solar thermal panel), [F]. The solar panel can be fixed or pivoted with solar tracking system. The multi-sector arrangement of solar panels, on top of upper wall duct, is inclined to harness solar energy from multiple angles of the sun. At the same time, the multi-sector arrangement of these inclined solar panels also forms the flow path for guiding the rain water towards the center of the system. The raiwater

then flows through rain water passage, [G] in the middle of the system and stores in the water storage compartment, [K] at the bottom. A rain water mesh or filter, [N] prevents the flow of foreign objects into the passage which can cause the blocking of passage. The mesh, [M] is mounted at the entrance side of PAGV to avoid foreign objects from striking VAWT. The power generated from wind turbine and solar panel is stored in a battery bank, [J] or fed into electricity grid line. A layer of thermal insulation, [L] is embedded at the bottom of the system to prevent the heat transfer into the interior part of the building.


(a) Side sectional view (b) Perspective view

FEATURES OF SYSTEM (BENEFIT AND APPLICATION)The features in this patented system are designed and integrated to harvest wind energy, solar energy and rain water. The PAGV not only speeds up and guides the oncoming wind stream, it also reduces the turbulent. The VAWT, [A] and PAGV can also be built in multiple layers, by stacking up vertically, on top or between the upper layers of the buildings or structures. Figure 2 shows a tandem design of two layers wind turbine and guide vane. The VAWT in different layers can be connected to common power-transmission shaft, gear system and generator, or separate power-transmission shaft, gear system, and generator. The upper PAGV channels wind to the upper wind turbine while the lower PAGV channels wind to the lower wind turbine. The wind turbines on adjacent layers are designed to rotate in a contra rotating motion (create a balanced-force between these two VAWTs). The wind turbines can be mounted on two co-axial output shafts which rotate in opposite direction as shown in Figure 2. One of the output shafts is coupled to rotor of generator and the other one to the stator of generator to improve the relative angular velocity. A multi-stage generator can be matched to the system. When the wind speed is below a speed threshold, only first stage of generator is engaged. When the wind speed increases and is above a speed threshold the second stage generators are engaged, and so on. A high speed wind cut-off system is activated to cut off generator to prevent strong wind damage.

Figure 2: Tandem design of two layers wind turbine and guide vane

Since the VAWT, [A] is enclosed by the PAGV and wind inlet mesh, [M], there is no possibility for the blade to fly-off and cause injury to the people around.

The design solves one of the problems to locate wind turbine in urban and sub-urban areas. Compared to a conventional horizontal axis wind turbine with long blades, the vertical axis wind turbine in this invention has a smaller rotating body contained within an enclosure to reduce electromagnetic interference. The noise level of the system is expected to be reduced since the wind turbine is enclosed in PAGV.By inducting higher wind speed into VAWT, smaller and lighter rotating parts of VAWT can be used for same power output, which mean lesser load on the bearing of turbines and reduced moment inertia for better starting behaviour. Smaller parts will also result in lower material, manufacturing and installation cost.The system is designed in an easy-access approach, which is located on top of building and can be accessed from the interior of building such that it is easily for installation and maintenance. A locking device can be installed to prevent the motion of VAWT and PAGV during installation and maintenance. By this arrangement (on top of building and only can be accessed through particular entrance & exit), the risk of vandal damage will also be greatly reduced. Figure 3 illustrates the artist’s impression of wind-solar hybrid with rain collector systems which are being retrofitted on high rise buildings.

Figure 3: Artist’s impression of wind-solar hybrid with rain collector systems which are being retrofitted on high rise buildings.

ESTIMATED ANNUAL ENERGY SAVINGThe weather data were obtained from the Malaysian Meteorology Department (MMD). The data from the weather stations which are in or near to urban area, i.e. Petaling Jaya and KLIA, were used for analysis.


The data supplied include monthly variation of mean surface wind speed, solar radiation as well as rainfall over six years, from year 2002 to year 2007. Hourly variation of the parameters mentioned for year 2007 was also given. These data are being utilized for analysis of variation of mean wind speed, solar radiation and amount of rainfall. The power or energy gained from these parameters can be subsequently predicted.

The wind and solar power generation capacity, and the amount of rain water collected were estimated based on the system designed:

• Diameter (PAGV) = 30m, Height for inlet (PAGV) = 12m;

• Diameter (VAWT) = 15m, Height (VAWT) = 6m

a) Wind EnergyFrom the analyzed wind data at the Petaling Jaya weather station, more than 90% of the total wind hours per year are lower than 4m/s. Figure 4 shows that the range of wind speed from 0.5 m/s to 1.5 m/s exhibits the highest frequency, with about 40% of total hours per year. Although this speed range shows the utmost frequency, the range of wind speed between 3.0 m/s and 4.0 m/s provides the peak profile of wind energy distribution.

Mean Wind Speed (m/s)

Freq

uenc

y

109876543210

600

500

400

300

200

100

0

Shape Scale N1.696 2.022 87162.680 4.172 23601

Variab leHo u rW ind Energ y , k W h

Weibull Histogram of Hour, Wind Energy Against Mean Wind Speed

Figure 4: Weibull distribution of wind speed and wind energy at Petaling Jaya weather station (Year 2007)

The wind speed at surface increases with height rapidly when close to the surface, and the rate of increase declines with greater height. The variation of the wind speed with height can be estimated using a power exponent function, given by the following equation [8]:

( ) αrr zzVzV =)(

where )(zV : Wind speed at height z

rV : Wind speed at the reference height zr above surface

z : Height above the surface

rz : Reference height above surface

α : Exponent which depends on the roughness of the terrain and the

stability of the atmosphere


Lat.:3°06’NLong.:101°39’EHeight of wind sensor above ground: 46.2mHeight of wind sensor above Mean Sea Level (M.S.L.): 60.8m

Peak

ene

rgy

Peak

freq

uenc

y

Thus, for a typical urban area like Petaling Jaya, value of α = 0.4 [9], the wind speed at building height of 220m ( z ) is approximately 1.867 times higher than the wind speed at reference height (height of wind sensor above ground level at Petaling Jaya weather station) of 46.2m.

And, the available wind power generated after the estimated losses, windP is:

32

1 VACP WTPAGVGTRpwind ⋅⋅⋅⋅⋅⋅⋅= ρηηη

where pC : Power coefficient

TRη : Efficiency due to bearing and transmission loss

Gη : Generator efficiency

PAGVη : Efficiency due to power-augmentation-guide-vane (PAGV) loss

ρ : Air density, kg/m3

WTA : Swept surface area of wind turbine, m2

V : Wind speed, m/s

Table 1: Estimated parameters (efficiency, losses, etc) of wind system components

The wind speed is increased after passing through PAGV and VAWT experiences better flow angle (wind stream directionally guided by guide vanes). With the estimated efficiency of sub-systems or components (Table 1), the wind energy generation could be estimated. For a system (30m x 12m PAGV, 15 x 6m VAWT) installed on top of the building with 220m height, the wind energy generated is approximately 57.3 MWh per year or 157 kWh per day.Assuming pC , TRη , Gη and ρ are constant; from the analyzed data of estimated available wind energy, the system (30m x 12m PAGV; 15m x 6m VAWT) can generate 1.25 times energy compared to the 30m x 12m VAWT (without PAGV).


Generator efficiency, Gη 0.7

Efficiency due to bearing & transmission loss, TRη 0.9

Efficiency due to PAGV loss, PAGVη 0.8

Air Density, ρ ( kg/m3) 1.225

Frontal Swept Surface Area, WTA (= 6*15 m2) 90

pC 0.4

0

1000

2000

3000

4000

5000

6000

7000

8000

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Wind velocity (m/s)

Ene

rgy

(kW

h)

Estimated energy generated at reference height (height of windsensor above ground at Petaling Jaya weather station) of 46.2mEstimated energy generated at approximate building height of 220mHeight of wind sensor above ground: 46.2m

Height of wind sensor above Mean Sea Level: 60.8m

Figure 5: Estimated wind energy generated at an estimated height of building (220m above ground) compared to Petaling Jaya weather station (46.2m above ground) in year 2007

b) Solar Energy

Based on the solar radiation data analyzed at KLIA weather station located in Sepang, the solar energy generation, solarP is estimated as follow [10]:

KAGP PSsssolar ⋅⋅⋅= η

where sG : Annual mean daily global irradiance, kWm-2

sA : Array active area, m2

PSη : Module conversion efficiencyK : Solar power loss

Thus, the energy generated by solar system, solarE is given by [10]:

Hsolarsolar YPE ⋅=

where Y : Equivalent sunshine hours (12 hours)

Table 2: Estimated solar energy generated at KLIA

weather station, Sepang (year 2007)


Annual mean daily global radiation, sG (kWhm-2/day) 4.48

Solar panel area above the PAGV (=π *152 m2) 706.86

Estimated solar cell active area, sA (m2) 650

Efficiency of solar panel, PSη 0.12

Estimated power loss (electric transmission), K 0.8

Estimated solar energy generated, kWh/day 280

Rational Equation [11],

According to annual report published by Tenaga Nasional Berhad (TNB) in year 2008, the average annual electric energy consumption per unit domestic customer or per house in Malaysia for year 2008 is approximately 2.8MWh [12]. By saving 160 MWh per year, the energy saved is sufficient to supply 57 units of domestic customer yearly.

FUTURE WORKS

a) The solar concentrating photovoltaic (CPV) system to be used instead of conventional photovoltaic (PV) system. Reflectors or mirrors are located exactly on the location of solar panel, for reflecting the solar radiation to a solar target,


Month Rainfall intensity (mm)

Rainfall Amount (liters)

JAN 271 142380

FEB 139.6 73290

MAC 321 168525

APR 419.7 216090

MAY 196.2 103005

JUN 333.8 175245

JULY 212.4 111195

AUG 67.2 35280

SEPT 149.4 77910

OCT 378 198345

NOV 263.6 139020

DEC 271.6 142590

Total rain water

1582875

b) Solar thermal panel to be incorporated in place of solar panel for heating of water (provide hot water system to the building).

c) Computational fluid dynamics (CFD) simulation on the system designed (VAWT and PAGV) and structural analysis to be carried out with the final design.

b) The model of wind energy system with PAGV will be tested using wind tunnel, and its performance and capacity will be evaluated.

CONCLUSIONSAn efficient and cost effective wind-solar hybrid renewable energy generation system with rain water collection feature is aimed for green architecture and buildings in Malaysia (with the design and production technique developed locally). The design is a combination which discloses wind, solar and rain harvesting features, and its technical advantages give rise to the uniqueness of the design. The design can eliminate or minimise the problems faced by current

wind turbines. The performance of this VAWT must be optimised and competitive with existing wind turbines in market in terms of total operating hour and energy gained. With this design, the dimension of wind turbine is reduced by half, yet generates 25% more energy as compared to conventional system without PAGV. With the 30-metre diameter and 12-metre height of system, the estimated annual energy saving is160 MWh. The long-term goal is the proliferation of wind and solar energy applications in

populated urban areas or even rural regions, capable of supplying supplementary power to urban buildings and local power to remote areas.

ACKNOWLEDGEMENTSThe authors would like to thank University of Malaya for the assistance provided in the patent application of this design and the research grant allocated to further develop this design under project FS129-2008B. Thanks and appreciations are also extended to Kong Yuen Yoke and Tan Lee Ling who have contributed their efforts, information and ideas on this design.

REFERENCES

[1] W. T. Chong, “The design and testing of a wind turbine for electrical power generation in Malaysian wind conditions”, Universiti Teknologi Malaysia, Ph.D. Thesis (2006).

[2] S. Warner, R. Bareiβ and G. Guidati, Wind Turbine Noise, (Springer, Germany, 1996).

[3] N. Campbell et al., “Wind energy for the built environment (PROJECT WEB)”, Proceedings of the European Wind Energy Conference & Exhibition, July 2-6, Copenhagen, Denmark. (2001).

[4] A. S. Bahaj, L. Myers, P. A. B. James, “Urban energy generation: influence of micro wind turbine output on electricity consumption in buildings”, Energy and buildings, 39(2), 154-165 (2007).

[5] Bahrain World Trade Centre. 18 Dec 2008 <http://bahrainwtc.com>.

[6] Y. Kimura, Y. Onai and I. Ushiyama, “A demonstrative study for the wind and solar hybrid power system”, World Renewable Energy

Congress, June 15-21, Denver, Colorado, USA: Pergamon, 895- 898 (1996).

[7] W. T. Chong, Y. Y. Kong and L. L. Tan, “Wind, solar and rain harvester”, Patent pending (Malaysian Intellectual Property Office), March 2009.

[8] A. Hemmat, C. Nguyen, Bharat Singh and K. Wylie, “Conceptual design of a Martian power generating system utilizing solar & wind energy”, University of Houston, (1998).


http://bahrainwtc.com/

[9] J. Park, The wind power book, (Cheshire Books, Palo Alto, California, U.S.A, 1981), pp. 56-58.

[10] K. Kurokawa, “Realistic values of various parameters for PV system design”, Renewable Energy, 15, 157-164 (1998).

[11] D. A. Chin, Water-resources engineering, (Prentice-Hall, 2000).

[12] “Annual report 2008”, Tenaga Nasional Berhad, (2008).


AN INVESTIGATION INTO VALIDATION OF DEEPDRAWING WITH ABS

M. MORADIA ,a Assistant Professor, Department of Mechanical Engineering, Isfahan University of Technology, Isfahan, Iran.

(email: [email protected])

A. R. FATHIB

b Graduate student, Department of Mechanical Engineering, Isfahan University of Technology, Isfahan, Iran.(email:[email protected])

ABSTRACT In this paper, different features of “deep drawing with ABS (Anti-lock Braking System)” was investigated through finite element simulation and analysis. Different blank holder gap and oscillation amplitude were used in the analysis of deep drawing of a cylindrical cup. A Sine profile was considered for the oscillation of the blank holder. Circular blank used is 0.9 mm thick and made of interstitial free steel (IF Steel). Results of the analysis emphasize the fact that blank holder gap and its oscillation amplitude can significantly influence the benefits of “deep drawing with ABS”. It is demonstrated that poor adjustment of blank holder and careless adoption of its oscillation amplitude diminishes the major advantage of increased LDR (Limiting Drawing Ratio) exclusive to this procedure.

Keywords: Deep Drawing, LDR, ABS, Blank holder.

1- INTRODUCTIONDeep drawing is one of the most important operations in sheet metal forming. Considering the widespread use of deep drawing products in majority of industrial applications, it becomes clear that investigating various approaches to improve product’s desirable features is of considerable importance. The method of using anti-lock braking system in deep drawing has recently attracted attention of many researchers. Limited previous experimental works establish the effectiveness of this approach in deep drawing [1,2,3]. This approach is the same as conventional deep drawing with the exception that the blank holder switches between gripping and releasing the blank in short intervals during forming process. It looks much like the ABS systems used in braking system of autos. In this paper, the term “deep drawing with ABS” is used to refer to this method, though blank holder does not grip the blank in some specific instances (in some blank holder oscillation regimes).

2- MODELING AND THEORYModeling of the drawing instrument was done via Finite Element Analysis. Punch, die and blank holder

were modeled as rigid bodies. Different blank diameters are used in order to provide different drawing ratios in the drawing process. According to the symmetry in the components, just one quarter of the blank is modeled to decrease the analysis computational cost. Figures 1 shows the assembly of die, blank holder and punch.

Figure 1: Assembly of the parts



The gap between blank and blank holder in this assembly is adopted to be 0.78 mm. Holder mass is considered 20 kg; based on the mass of holders

conventionally used in the field. Figure 2 depicts the blank’s meshed model and the elements.

Figure 2: Blank model and elements

The round blank made of IF steel is formed to produce the cylindrical cup.

Table 1 shows the chemical composition of the steel used.

Table.1 Chemical composition of IF steel (in weight %) [4]Material Thickness C Mn Si S P Al N(ppm) Ti B NbIF steel 0.9 mm 0.0027 0.38 0.009 0.01 0.01 0.047 37 0.042 0.0007 0.046

The yield criterion of Hill [5] was adopted and applied as in Eq.(1):212

231

223

22211

21133

23322 222)()()()( σσσσσσσσσσ NMLHGFf +++−+−+−= (1)

where σij denotes the stress components and F, G, H, L, M and N are the material constants obtained from

tensile tests. These constants are related to anisotropic yield stress ratios (Rij) through Eqs.(2) and (3):


)111(21

),111(21

),111(21

233

222

211

222

211

233

211

233

222

RRRH

RRRG

RRRF

−+=

−+=

−+=

(2)

212

213

223

23

,2

3

,2

3

RN

RM

RL

=

=

=

(3)

Yield stress ratios are related to plastic strain ratios (Lankford’s r-values) through Eqs.(4) to (7) [6]:

1231311 === RRR (4)

)1()1(

900

09022 +

+=

rrrrR (5)

900

09033

)1(rr

rrR+

+= (6)

))(12()1(3

90045

09012 rrr

rrR++

+= (7)

r0, r45 and r90 are plastic strain ratios relative to the rolling direction. Table 2 shows the r-values for the IF steel used:

Table.2 r-values of the sheet metal [4]Sheet metal - thickness r0 r45 r90

IF steel – 0.9 mm 0.98 1.55 1.46

3- ANALYSIS


Explicit method was used for analysis. In order to decrease the analysis time, process is simulated 16 times faster than actual drawing process. Actual punch velocity and holder frequency are 0.417 m/min and 168.7 rad/s respectively. However in the following simulation, punch velocity and holder frequency are considered as 0.1112 m/s and 2699.2 rad/s respectively (16 times the corresponding values

in application). The analyses were checked for mesh convergence so that the effect of elemental dimensions on the results was less than 10%. The Forming Limit Diagram (FLD) damage initiation criterion was used to indicate material failure during drawing process. Failure occurs when FLD Criterion (FLDCRT) equals or exceeds unity. FLD for the alloy used in this work is depicted in Figure 3.

0

0.1

0.2

0.3

0.4

0.5

-0.2 -0.1 0 0.1 0.2

Minor true strain

Maj

or tr

ue s

trai

n

forming limit line

Figure 3: FLD for IF steel – 0.9 mm thick [4]

In this investigation, blank holder position over the blank and its oscillation amplitude was altered. In different situations, limiting drawing ratio is estimated to investigate how beneficial adopted method (oscillating blank holder) is in different situations. First simulation was conducted by setting the gap between the blank holder and the blank to be 0.78 mm; the blank holder vibrated in the downward direction. In another setting, gap distance remained the same but the blank holder vibrated upwards. Finally, the results were compared with conventional deep drawing with fixed blank holder.

4- RESULTS AND DISCUSSION

Analyses for four different settings described below were conducted and the outcomes were provided for comparison. Figure 4 (a) illustrates the first setting in which the blank holder remained fixed making a constant gap between the blank holder and the blank. “Conventional drawing” denotes this setting. For the case of gap = 0.78 mm, Limiting Drawing Ratio (LDR) obtained equals 2.3 and for the gap = 2.54 mm LDR = 2.3 could not be achieved. Figure 4 (b) depicts the state in which the blank holder

oscillates between two previous limits of (a) with a sine manner. Advantage of an oscillating blank holder is clear by comparing LDR achieved in


settings (a) and (b), thereby establishing the effectiveness of using oscillating blank holders. Figure 4 (c) shows the setting in which blank holder penetrated 0.1 mm into the blank. Results obtained in this investigation shows that this holder setting does not enhance LDR. Recall that this setting simulates

“deep drawing with ABS”. However, for IF steel with said chemical composition and in conjunction with other settings such as die-punch clearance, etc. The method did not show promising results. If the minimum gap between the blank holder and the blank was remained the same as setting (b) but amplitude reduced, another setting would be obtained. Figure 4 (d) depicts this setting. Even with oscillation amplitude halved, acceptable LDR is achieved.

Figure 4: LDR for different blank holder settings.

Figure 5 shows the blank holder oscillation path considered for holder movement in setting (b). Figure 6 (a) shows successful whereas Figure 6 (b) shows unsuccessful drawing operations in which setting (a) is used for holder assembly. Beginning to draw blanks with radius of 140 mm to one with 115 mm radius shows that LDR for this setting is 2.3.

Figure 7 (a) shows drawn cup with setting (b) adopted. Figure 7 (b) shows work piece during the drawing process with setting (b) used. Comparing Figure 6 (a) with Figure 7 (a), it should be mentioned that limiting drawing ratio in setting (b) is much larger than that of setting (a).


5- CONCLUSION

Results of this investigation establish the effectiveness of using an oscillating blank holder. However, it is worth noting that poor adjustment of blank holder can eliminate the advantages of this method.

The oscillating blank holder with appropriate.assembly could improve LDR for a round blank of IF steel and 0.9 mm thick from 2.3 to 2.9. No notable improvement in LDR was achieved in the case of poor assembly of the blank holder.

Figure 5: Blank holder oscillation scheme for minimum gap of 0.78 mm and 0.88 mm amplitude (setting


(b)).

Figure 6: Conventional deep drawing with (a) DR=2.3, successful operation and (b) DR=2.8, failed operation

Figure 7: (a) drawn cup DR= 2.9, (b) blank during drawing process


REFERENCES

[1] T.Jimma et al. “An application of ultrasonic vibration to the deep drawing process”, Journal of Processing Technology 80-81 (1998) 406-412.

[2] M.Gavas, M. Izciler, “Deep drawing with anti-lock braking system (ABS)”, Mechanism and Machine Theory 41 (2006) 1467-1476.

[3] Y.Ashida, H. Aoyama, “Press forming using ultrasonic vibration”, Journal of Materials Processing Technology 187-188 (2007) 118-122.

[4] R. Narayanasamy, C.Sathiya Narayanan.”Experimental analysis and

evaluation of forming limit diagram for interstitial free steels”, Materials and Design 28 (2007) 1490-1512

[5] R. Hill, Proc. Roy. Soc. Lond. 193A (1948) 197-218.

[6] S. Ahmadi, A. R. Eivani, A. Akbarzadeh, “Experimental and analytical studies on the prediction of forming limit diagrams”, Computational Materials Science (2008) doi:10.1016/j.commatsci.2008.08.008. (Article in press).


CHARACTERISTICS OF CONVECTIVE MASS TRANSFER UNDER INFLUENCE OF TURBULENCE CONTROL WITH WALL-RECESS IN A

PARALLEL PLATE ELECTROCHEMICAL FLOW CELL

HARINALDI, DAMORA RHAKASYWI AND HANIFA AKROM

Fluid Engineering Laboratory, Department of Mechanical EngineeringFaculty of Engineering University of Indonesia,

Kampus UI-Depok,16424, INDONESIAe-mail : [email protected]

ABSTRACTSThis research investigated the characteristics of convective mass transfer in a parallel plate electrochemical flow cell under the influence of turbulence induced by a wall-recess that produce a separating-reattached flow field. Wall-recess was in the form of a 5 mm rearward-facing step contoured-channel which was placed upstream of the electrochemical cell to serve as a control device for turbulence level. Experimental work was done to elucidate the effect of fluid dynamical parameters to the rate of mass transfer between cell electrodes made of copper plates. The experimental set-up consisted of closed loop flow channel equipped with flow rate control and solution of CuSO4 of 0.5M was used as electrolyte fluid. The measurement of rate of mass transfer was done using the limiting diffusion current technique in which local limiting current measured at micro cathodes placed in the electrochemical cell would indicate the local mass transfer coefficient (K m). The current measurement was done with a precision digital multimeter interfaced by a data acquisition system to a personal computer. Some results shows that the convective mass transfer is proportionally influenced by the increase of Reynolds numbers of main flow velocity. Within the range of Re = 300 – 3000, the mass transfer coefficient increase to Km = 3.299 x 10-4 - 3.89 x 10-4 (m/s) which give some improvement up to 25.5 % of mass transfer rate compare to the case of electrochemical process with no flow condition. Furthermore, some correlations between Sherwood and Reynolds number are also discussed.

Keyword : wall recess, turbulence control, separating-reattached flow, convective mass transfer, electrochemical cell.

1. INTRODUCTION

An abrupt change of wall contour in flow passages will change the characteristics of the flow near the wall and sometimes causes flow separation and reattachment. A typical case is the existence of wall recess in the solid boundary of a flow channel in the form of a backward facing-step. The flow separation, recirculation and reattachment such those occur behind the backstep has applications in a wide range of engineering practices. Examples include its utilizations in heat exchangers, chemical process reactors, energy system equipments, combustors, and so on. It is the alteration of the turbulence caused by the occurrence of some complex flow structure that gives significant effects to the fundamentals of transport properties in the flow field.

Eventually, the effects result in a drastic change in the performance of various fluid machinery, heat transfer devices and chemical reactors and so on.Since pioneering works on the backward-facing step flow focusing on the various fundamental features of fluid dynamics involved [1]-[3], research works on the characteristic of flow passage with turbulence control using a backstep as a kind of wall recess had been made by many researchers with various flow configurations, parameters, and objective of interests both experimentally and computationally. Yang and Tsai [4] utilized a secondary fluid injection from the base wall of channel with a backstep in the upstream position and

demonstrated that the occurrence of significant change in heat transfer coefficient along the channel. Then, Yang and Kuo [5] continued the work by



developing numerical analysis. Both studies suggested that the amount of injected mass will alter the turbulence characteristics which responsible to the heat transfer in the whole recirculation zone. Studies on local mass transfer rates had been done by Chouikhi et al. [6] at the wall of a pipe downstream of a constricting nozzle in turbulent pipe flow at various Schmidt numbers and Oduoza et al. [7] at rectangular channel with wall obstruction. Chun and Sung [8] and Yoshioka et al. [9] conducted researches on recirculation flow behind a backward facing step in channel with expansion ratio of 1.5 under influence of a periodical slot jet excitation with high frequency. The results showed a significant decrease of average length of recirculation zone. This decrease was found to be sensitive to injection frequency. Harinaldi, et al. [10] showed that mixing process in the recirculation zone was effectively occurred in within the region of three to five times of step height along the flow direction. Dejoan et al.[11] used Large eddy simulation (LES) and statistical turbulence closures to study the unsteady effect of external excitation from a 45o slot jet that showed a decrease in of recirculation zone length. Despite of various achievements in fundamentals understanding as well as practical issues previously mentioned, there still exist a lot of details that needs more

investigations, especially to further elaborate some possibilities to enhance transport properties in the flow for practical applications. This work is part of a broader study of industrial electroplating equipment involving complex geometries and complex flows. This research investigated the characteristics of convective mass transfer in a parallel plate electrochemical flow cell under the influence of turbulence induced by a wall-recess that produce a separating-reattached flow field. Wall-recess was in the form of a backward-facing step contoured-channel which was placed upstream of the electrochemical cell to serve as a control device for turbulence level. Computational as well as experimental works were done to elucidate the effect of fluid dynamical parameters to the rate of mass transfer between electrodes.

2. EXPERIMENTAL SET UP

2.1 Flow system and Parallel Plate Electrochemical Cells The experiment was conducted in a closed-loop electrolyte flow system as shown in Fig. 1. The flow system was design to enable electrolyte fluid to flow through a 1000 mm vertical channel made of acrylic plates. The electrolyte flow was driven by a pump and the control of flow rate entering the channel was done by-pass valve, control valve and a flow meter.

Figure 1. Schematics of Experimental Setup

The inlet section length was 300 mm and the exit section length was 250 mm. Following the inlet section, a test section of flow channel was constructed of an acrylic plate (length 100 mm; width 40 mm and thickness 5 mm) to serve as backstep plate and two flat plates of made of copper (length 370 mm; width 40 mm).

The plates installed in parallel at a distance of 10 mm to function as electrochemical electrodes (cathode and anode). Hence, the electrodes and the side walls formed arectangular flow channel with area of 40 x 10 mm2. The cathode was equipped with 32 surface flush


copper mini-electrodes (diameter, 1.5 mm) arranged in two axially oriented rows in the middle of the plate. In this case, the entire cathode plate with all of its mini-electrodes lay downstream of the step. The assembly of the copper mini-electrodes on the copper macro-electrode was tedious, and was performed carefully using Araldite epoxy to secure them in position, whilst ensuring complete electrical insulation from the macro-cathode. The

diagrammatic plan view of the test section and electrodes arrangements is shown in Fig. 2. The electrochemical reaction was powered by a precision DC power supplied. Data acquisition system of local electrical current measurement in the mini-cathodes was constructed of a high precision digital multimeter which was connected to a personal computer via an USB connectionandcontrolled with DMM data acquisition software. In each mini-cathode about 400 individual data of local current were sampled to ensure a sufficient amount of statistical averaging for a fluctuated flow condition.

(a)

(b)

Figure 2. Diagrammatic plan view of the (a) test section and (b) electrodes arrangements

2.2 Limiting Diffusion Current MethodIn this experimental work, the determination of local mass transfer coefficient was done by limiting diffusion current method following Chouikhi et al. [6] and Oduoza et al. [7]. Electrochemical reaction takes place between anode and cathode in the electrolyte of copper sulphate, CuSO4 (0.5 M). The local mass transfer was determined by measuring the local limiting diffusion current for the cathodic reduction of copper ionCu2+ + 2e- → Cu (1)

Table 1. Physical properties of the electrolyte at 20o C Physical properties CuSO4Density, ρ (kg/m3) 1072Viscosity, μ (kg/m.s) 0,001149Molecular weight, Mr (gr/mol) 250Concentration, C (mol/m3) 500Diffusivity of ion, D (m2/s) 4,43 x 10-10

Schmidt number, Sc 2418

Prior to each experiment, the cathode surface was cleaned and polished using different grades of wet or dry paper. Voltage was then applied beyond the limiting current region of 1.2 V for about 5 min to liberate hydrogen. This had the effect of removing


surface oxides. By applying a potential of 800 mV between anode and macro-cathodes, i.e. the mid-plateau region of the current-potential curves, limiting current values for each mini-electrode for a range of Reynolds numbers were determined. Local mass transfer coefficients were then calculated using the equation

zFACIK m = (2)

where I is the mini-electrode limiting current, z is the number of electrons exchanged (z = 2) , F is the

Faraday number (96 487 C mol-1), A is the exposed mini-electrode area and C is the bulk CuSO4

concentration.Data from the mini-electrodes were collected and processed using the data acquisition system described above. The raw and average data values for each mini-electrode were saved to disk and accessed and processed off-line. The experiment was done in a various condition of electrolyte flow rate and the corresponding Reynolds number, Re based on hydraulic diameter of cross section of flow passage upstream of the step are shown in Table 2

Table 2. Test condition of electrolyte flow

3. RESULTS AND DISCUSSION3.1. Characteristics of rate of mass transferThe effect of wall recess in the form of a backstep as a turbulent control to the convective mass transfer in

the electrolyte flow is indicated in Fig. 3 showing the mass transfer distributions at different Reynolds numbers along axial distance from the step in the stream wise location.

Figure 3. Distribution of mass transfer coefficient along axial distance from the step (at 100 mm from the inlet of test section) for various Reynolds numbers of 404, 551, 1513 and 2856.

It can be seen that the distributions demonstrate a similar characteristics for every condition of Re.

Typically, the mass transfer coefficient, Km , steeply increases just after the step to reach its maximum


Flow rate (liter/min) Re

0,8 404,81 551,42 1513,23 2856,5

(peak) value and then gradually decreases up to some axial distance of about ten step height. The mass transfer then increase again to reach another high value at far location from the step. However, compared to the case of no flow condition, the rate of mass transfer is higher only at the location of less than ten times of step height of every flow condition. This results suggest that enhancement of the convective mass transfer seems to occurs mainly within the recirculation zone up to the average location of reattachment point. Around and after the

reattachment, the fluid particles experience a highly random motion due to reattachment process. Therefore, even though the local flow still maintain high turbulence intensity, it seems that the random motions of the fluid flow causes more significant suppression effect to the mass transfer rate.

3.2 Effect of Electrolyte Flow RateThe effect free stream flow rate of the electrolyte to the convective mass transfer rate is presented in Fig.4 which shows the relation between peak mass transfer coefficients (Peak Km) to the Reynolds number of the electrolyte flow.

Figure 4. Peak mass transfer coefficient against Reynolds number

It can be seen from the figure that the peak mass transfer coefficient increases as the Reynolds number increases. The increase shows a tendency to be linear, at least within the range of Reynolds number in the present investigation. If compared with the value of no flow condition as previously shown in Fig. 3, the maximum peak Km indicates as much as 25,5 % increase of mass transfer rate obtained at Re of 2856. Current results show an agreement to those of Tagg [13] which showed an enhancement of mass transfer rate in turbulent flow in the downstream of a nozzle with a sudden expansion.

Figure 5 shows a log plot of Sherwood number (peak) Sh/Sc0.33against the jet Reynolds number, averaged over four experiments. Here the Sherwood number is based on the hydraulic diameter of cross section of flow passage upstream of the step. This plot achieves a successful correlation of the data. The equation through the points is Sh =315,0Re0.076Sc0.33. The correlation of Tagg et al. [13] in turbulent flow with an axisymmetrical cell is also shown in this plot (Sh = 0.27Re0.67Sc0.33), and is seen to lie lower than the present data for Re less than about 2000 and seems to be higher for Re more than 2000.


Figure 5. Log scale plot of peak Sh/Sc0.33 against Reynolds number at the cathode,

4. CONCLUDING REMARKSThis research work has investigated the characteristics of convective mass transfer in a parallel plate electrochemical flow cell under the influence of turbulence induced by a wall-recess in the form of a backstep that produce a separating-reattached flow field. Some results shows that the peak mass transfer downstream of the step is dependent on the Reynolds number, and lies between the flow recirculation and reattachment zones. The convective mass transfer is proportionally influenced by the increase of Reynolds numbers of main flow velocity. Within the range of Re = 300 – 3000, the mass transfer coefficient increase to Km = 3.299 x 10-

4 - 3.89 x 10-4 (m/s) which give some improvement up to 25.5 % of mass transfer rate compare to the case of electrochemical process with no flow condition.AcknowledgementsThis work was supported within the RUUI program of the University of Indonesia (project number 240C/DRPM-UI/N1.4/2008).NomenclatureA cross sectional area of mini-cathode (m2)C electrolyte concentration (mole/m3)D mass diffusivity coefficient (m2/s)d hydraulic diameter (m)F Faraday constant (A.s/mol)I Limiting diffusion current (A) Km mass transfer coefficient (m/s)Mr molecular weight (g/mol)Re Reynolds numberSc Schmidt numberSh Sherwood numberz electrons exchanged in electrode reactionμ dynamic viscosity (kg/ms)ρ density (kg/m3)

REFERENCES[1] Eaton, J.K. and Johnston, J.P., 1981, AIAA Journal, 19, pp.1093-1099

[2] Armaly, B.F., Durst, F., Pereira, J.C.F and Schonung, B., 1983, Journal of Fluid Mechanics, 127, pp. 473-496

[3] Papadopoulus, G. and Otugen, M.V., 1995 Trans. ASME, J. Fluid Engineering, 117, pp.17-23[4] Yang, J.T and Tsai, C.H., 1996 Int. J. Heat and Mass Transfer, 39, pp.2293-2301

[5] Yang, Y.T and Kuo, C.L.,1997, International Journal of Heat and Mass Transfer, 40, pp. 1677-1686.

[6] Chouikhi, S.M., Patrick, M.A. and Wragg, A.A. 1987, J. Appl. Electrochem., 17, pp. 1118-1128.

[7] Oduoza, C.F., Patrick, M.A. and Wragg, A.A., 1997, Chemical Engineering J., 68, pp.145-155.

[8] Chun, K.B and Sun, H.J, 1996, Exp. Fluids, 21, pp. 417-426

[9] Yoshioka, S., Obi, S and Masuda, S., 2001, Int. J. Heat Fluid Flow, 22, pp.301-307

[10] Harinaldi, Ueda, T. and Mizomoto, M., dalam S. Dost, H. Struchtrup and I. Dincer Eds., 2002, Progress in Transport Phenomena, Elsevier, Paris, pp. 93-98.

[11] Dejoan A., Jang, Y.J, and Leschziner, M.A, 2005, Trans. ASME, J. Fluid Eng. 127, pp.872-878

[12] Noulty, R.A., Leaist, D.G., 1987., Journal of Solution Chemistry, 16, 10, pp. 813-825.

[13] Tagg, D.J, Patrick, M.A and Wragg, A.A., 1979., Trans. I. Chem. Eng., 57, pp. 176-181



POWER QUALITY DISTURBANCE CLASSIFICATION USING WAVELET AND FUZZY LOGIC

PRAMILA P1 , S.PURUSHOTHAMAN2 AND PUTTAMADAPPAC3

1 Department of Electrical & Electronics Engineering, Bangalore Institute of Technology, Bangalore, India2Sun College Of Engineering & Technology , Sunnagar , Kanyakumari, Tamilnadu , India.

3Department of Electronics & Communication Engineering, SJB Institute of Technology, Bangalore, [email protected] , [email protected] and [email protected]

ABSTRACT This paper presents application of wavelet and Fuzzy logic (FL) for power quality disturbance classification. Features are extracted from the electrical signals by using wavelets. The features obtained from the wavelet are unique to each type of electrical fault. These features are normalized and given to fuzzy logic. The data required are generated by simulating various faults in the test system. The performance of the proposed method is compared with the existing feature extraction techniques. Simulation results show the effectiveness of the proposed method for power quality disturbance classification.Keywords : wavelets, Fuzzy logic (FL), electrical fault, Harmonics, Power quality

1. INTRODUCTIONElectrical fault arises due to sudden loads, failure in the electrical circuits, lightning, conducted or radiated low and high frequency phenomena. Due to electrical fault in the system, the power quality deteriorates which is an indication of slow failure of the equipment. Many algorithms have been developed by previous researchers for classification of electrical disturbances. Integrated Fourier linear combiner and fuzzy expert system [1] were used for the classification of transient disturbance waveforms in a power system. S-Transform and two dimensional time–time (TT) transform [2] have been implemented for electrical fault identification. Patterns generated by S-Transform and TT transform are unique and hence accuracy of identification is high. An adaptive neural network approach for the estimation of harmonic distortions and power quality in power networks are implemented [3]. A hybrid system to automatically detect, locate and classify disturbances affecting power quality in an electrical power system is presented [4]. Least absolute value (LAV) State estimation algorithm has been used to measure the flicker voltage magnitude [5]. The Simulated Annealing (SA) optimization algorithm has been used for measuring the voltage flicker magnitude, frequency and the harmonics contents of the voltage signal for power quality analysis [6]. An algorithm to detect the fundamental frequency is proposed. It is based on the chirp-z transform (CZT) spectral analysis and is able to observe all standards in force

because of its accuracy and working characteristics [7]. A wavelet-based neural classifier integrating the

DWT [8][9], learning vector quantization (LVQ) neural network, and decision-making scheme to become an actual power disturbance classifier. The classifier employed the DWT coefficients as inputs to multiple LVQ neural networks to train and perform waveform recognition, and use the decision-making schemes to classify the transient disturbance. A novel classifier using a rule-based method and a wavelet packet-based hidden Markov model (HMM)[10]. The rule-based method is employed to classify the time-characterized feature disturbances, while the wavelet packet-based HMM is utilized to categorize the frequency-characterized feature power disturbances. Wavelet-multi-resolution decomposition that combines frequency-domain with time-domain analysis for power disturbance feature extraction is proposed [11]. Extracting the features from the wavelet transform coefficients at different scales as inputs to neural networks for classifying the nonstationary signal have been demonstrated[12]. A wavelet norm entropy-based effective feature extraction method for power quality disturbance classification problem has been done by [9]. Extracting the squared wavelet transform coefficients (SWTC) at each scale as inputs to the neural networks for classifying the electrical disturbance is proposed[13][14][15]. Multi-wavelet-based neural networks with learning



vector quantization network are used for power quality disturbances as a powerful classifier [16]. The DWT coefficients as inputs to a single-layer self-organizing map neural network to train and classify the transient disturbance has been used [17]. Fuzzy ARTMAP, Back propagation algorithm and Radial Basis Function (RBF) network in combination with S Transform, Wavelet transform and Hilbert Transform (HT) for classifying power faults have been used [18]. DWT coefficients have been used as inputs to a refined neuro-fuzzy network to train and classify the power system disturbance[19].

Continuous wavelet transform (CWT) has been used to estimate the disturbance time duration and the DWT to estimate the disturbance amplitude [20]. The two features thus obtained are then used to classify the transient disturbance type. The authors have claimed HT with RBF gives more fault classification from a set of 6 different types of faults with 3000 signals generated using Matlab. Algorithms should be developed that can classify the type of harmonics and other electrical disturbances. Due to generation of non-stationary random disturbances, many existing classification algorithms are provided with intelligence using artificial neural networks, fuzzy logic and evolutionary algorithms. The fuzzy logic (FL) can classify the electrical disturbances in the presence of noise in the signal. In this work, FL has been implemented for electrical fault classification

2. PROPOSED METHODOLOGYThis research work proposes wavelets for feature extraction and FL for classification of electrical fault. In order to achieve maximum classification, proper data input, optimum topology and correct training of FL with suitable parameters is a must. A large amount of patterns are generated from different fault conditions. Twelve features are generated for each pattern. The faults considered are instantaneous interruption, instantaneous sag, instantaneous swell, momentary interruption, momentary sag, momentary swell, temporary interruption, temporary sag, temporary swell and harmonics. The features obtained are mean, standard deviation, norm, maximum and minimum of

signal for Approximation and Details at the 5 th level decomposition. These features are given as inputs for the FL along with labeling that indicates a fault. Training the FL gives final weights. The final weights are used to identify a fault during testing. Fig. 1 explains the overall sequence of proposed methodology.

Steps in feature extraction from the input signal

1. Input voltage wave form is sampled, fs=12800 Hz.

2. Decomposition of the signal by wavelets till 5 th

level.

3. Ten features are obtained from Approximation and Detail in the 5th level

4. Input the extracted features into the input layer of features.

5. Train the FL and store the final weights.

6. Test the FL with new signal having a fault.

3. WAVELET DECOMPOSITION

Wavelet is a process of analyzing the signal with scaling and shifting to obtain the details of the signal. It decomposes the signal into high frequencies (Detail) and low frequencies (Approximation). The first time getting Approximation and Detail is called level 1 decomposition (Figure 1). Subsequent decompositions are done, by using the Approximation or Detail obtained in level 1, to further get the required information. In every level of decomposition, the number of samples in the signal is reduced to 50% of the total samples to Approximation and remaining 50% of the samples to Detail. During subsequent decomposition, only 50% of the samples (Approximation will be further used for getting Approximation and Detail in the subsequent level decomposition).


Figure 1 Levels of decomposition

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-250

-200

-150

-100

-50

0

50

100

150

200

250

Time (sec)

Am

plit

ude

(V

)

S h o rt d u r a t i o n v a r ia t i o n s - I n s t a n t a n e o u s I n t e rr u p t i o n

Fig. 2 Short duration variations-Instantaneous sag with harmonics

A fault signal has been shown in Fig.2. This signal has been generated by combining pure sinusoidal wave form with harmonics and with sag as per the data available in Table-1. The Approximation and Details for the signal in Figure 2 is given in Figure 3.

0 100 200 300 400-1500

-1000

-500

0

500

1000

1500Approximation A1

0 100 200 300 400-400

-300

-200

-100

0

100

200

300

400Detail D1

Figure 3 Approximation and Detail for the signal in Figure 2

The Approximation and Details for different conditions of the signals are given in Figure 4 – Figure 6.

0 100 200 300 400-3000

-2000

-1000

0

1000

2000


0 100 200 300 400-600

-400

-200

0

200

400

600Detail D1

Fig 4 Fifth level decomposition of Instantaneous swell

0 100 200 300 400-40

-30

-20

-10

0

10

20

30

40Approximation A1

0 100 200 300 400-50

-40

-30

-20

-10

0

10

20

30

40

50Detail D1

Fig.5 Fifth level decomposition of Harmonics_signal

0 100 200 300 400-1500

-1000

-500

0

500

1000


0 100 200 300 400-300

-200

-100

0

100

200

300Detail D1

Fig.6 Fifth level decomposition of Instantaneous_interruption_point1_3cycles_no_harmon

ics


Approximation (A1)6400

Details (A1) 6400

Level 1


Details(A1)400

Level 2


Details(A1)3200

Level

5

Figure

1

Levels

of

decom

positio

n

Electrical fault signal (12800)

It can be noted from the Figures 4-6 that the wavelet decomposition really helps in identifying the presence of disturbances. In order to make it more clear features are extracted from the approximation and details.The features are obtained from the Approximation and Details of the 5th level by using the following equations

)(IFd1V1 ∑= (1)

Where d = Samples in a frame and V1 = Mean value of Instantaneous Frequency

V1)(IFd1V2 −= ∑

(2)Where V2=Standard Deviation of Instantaneous Frequency

)Maximum(IFV3 = (3))Minimum(IFV4 = (4)

2norm(IF)V5 = (5)Where V5 = Energy value of frequency

4. FUZZY LOGICFuzzy Logic was initiated by Zadeh (1984). Basically, Fuzzy Logic (FL) is a multi valued logic that allows intermediate values to be defined between conventional evaluations like true/false, yes/no, high/low. Fuzzy systems are an alternative to traditional notions of set membership.The training (Figure 7) and testing (Figure 8) fuzzy logic is to map the input pattern with target output data. For this the inbuilt function has to prepare membership table and finally a set of number is stored. During testing, the membership function is used to test the pattern.

Figure 7 Training Fuzzy Logic

Figure 8 Testing Fuzzy Logic

5. Experiment and Results

Matlab 2009 has been used to generate 300 signal patterns for each electrical fault. The equation used for generating an electrical fault signal is as follows:

φ) tf sin(2A y += π

(6)

Where f = Frequency of the signal, A = Amplitude of the signal and φ= Phase angle.

Only one wave form is created when considering pure signal,. Different harmonics with different amplitudes are generated and summed. The harmonics waveform is added to fault signals. Harmonics signals are generated with fh (1st to 25th

harmonics) and voltage of the harmonics, vh (0.004 pu to 0.09pu). Similarly for each fault, swell, sag, interruption under instantaneous, momentary and transient, 300 samples are created with varied time duration, varying amplitudes with 50 Hz as constant frequency given in Table 2. Hence, a total of (10 faults * 300) + 1 pure signals are generated using equation (6). At random, 50 signals are considered


Read the pattern and its target value

Create Fuzzy membership function

Create clustering using K-Means algorithm

Obtain final weights

Process with target values

Input a pattern

Process with Fuzzy membership function

Find the cluster to which the pattern belongs

Obtain estimated target values

from each fault for training Fl. They are given in Table 2.


Table 2 Faults LabelingNumber of patterns used

for testingNumber of patterns used for training Target labeling

Pure sine wave 1 1 0.01Instantaneous Interruption 250 50 0.02Momentary Interruption 250 50 0.03Temporary Interruption 250 50 0.04

Temporary sag 250 50 0.05Instantaneous sag 250 50 0.06

Momentary sag 250 50 0.07Temporary swell 250 50 0.08Momentary swell 250 50 0.09

Instantaneous swell 250 50 0.1Harmonics 250 50 0.11

A counter is used to make a note in how many frames have the same values of frequency and the voltage occurs continuously If duration of existence of fault is in multiples of 8, 16, 24, 32, 40, 48 cycles or more than one second then. Based upon the number of similar values in the successive frames, a fault is classified. In a signal of 1 sec, minimum 2 faults are introduced along with harmonics.

6. Conclusion:

In this work, wavelet (db2) has been used to obtain Approximation and Details at the 5 th level decomposition. As the initial sample size is 12800 in the sampled signal, in the 5th level decomposition, 400 samples for Approximation and 400 samples for Details are obtained. The features mean, std, norm, minimum and maximum of the Approximation and Details are obtained. The features are used as training and testing data for the FL. The percentage of electrical fault identification in given in Table 3


Table 3 Electrical fault identification

Faults Number of patterns identified

Number of patterns used for testing

% identification and classification

Pure sine wave 1 1 100Instantaneous Interruption 247 250 98.8Momentary Interruption 247 250 98.8Temporary Interruption 247 250 98.8


Momentary sag 247 250 98.8Temporary swell 241 250 94.4Momentary swell 247 250 98.8


Appendix Table 1 Faults Labeling


Table 2 Categorize and Characterization of Power System Faults [13]

Table 3 Faults Identification and Classification


Faults Number of patterns used for training

Number of patterns used for testing Target labeling

Instantaneous Interruption 50 450 0.01Momentary Interruption 50 450 0.02Temporary Interruption 50 450 0.03


Momentary sag 50 450 0.06Pure sine wave 1 1 0.07

Temporary swell 50 450 0.08Momentary swell 50 450 0.09


CategorySpectral

ContentDuration Voltage

Short duration variation-Instantaneous

Interruption50Hz 0.5 – 30 cycles <0.1 pu

Short duration variation-Instantaneous Sag(dip) 50Hz 0.5 – 30 cycles 0.1 – 0.9 pu

Short duration variation-Instantaneous Swell 50Hz 0.5 – 30 cycles 1.1 -1.8 pu

Short duration variation-Momentary Interruption 50Hz 30 cycles – 3s <0.1 pu

Short duration variation-Momentary Sag(dip) 50Hz 30 cycles – 3s 0.1 – 0.9 pu

Short duration variation-Momentary Swell 50Hz 30 cycles – 3s 1.1 -1.8 pu

Short duration variation-Temporary Interruption 50Hz 3s – 1min <0.1 pu

Short duration variation-Temporary Sag(dip) 50Hz 3s – 1min 0.1 – 0.9 pu

Short duration variation-Temporary Swell 50Hz 3s – 1min 1.1 -1.8 pu

Harmonics1st to

13th

Present in the entire

signal

0.0035– 0.087

pu

References[1] P.K. Dash, R.K. Jena, M.M.A. Salama, Power quality monitoring using an integrated Fourier linear combiner and fuzzy expert system, Electrical Power and Energy Systems, 21, 1999, 497–506.

[2] S. Suja, Jovitha Jerome, Pattern recognition of power signal disturbances using S Transform and TT Transform, Electrical Power and Energy Systems, 32, 2010, 37-53.

[3] P.K. Dash, S.K. Panda, A.C. Liew, B. Mishra, R.K. Jena, A new approach to monitoring electric power quality, Electric Power Systems Research, 46, 1998, 11–20.

[4] Mário Oleskovicz, Denis V. Coury, Odilon Delmont Felho, Wesley F. Usida, Adriano A.F.M. Carneiro, Leandro, R.S. Pires, Power quality analysis applying a hybrid methodology with wavelet transforms and neural networks, Electrical Power and Energy Systems, 31, 2009, 206–212.

[5] S.A. Soliman, M.E. El-Hawary, Measurement of power systems voltage and flicker levels for power quality analysis: a static LAV state estimation based algorithm, Electrical Power and Energy Systems, 22, 2000, 447–450.

[6] S.A. Soliman, A.H. Mantaway, M.E. El-Hawary, Simulated annealing optimization algorithm for power systems quality analysis, Electrical Power and Energy Systems, 26, 2004, 31–36.

[7] M. Aiello, A. Cataliotti, S. Nuccio, A Chirp-Z transform-based synchronizer for power system measurements, IEEE Trans. Instrum. Meas, 54, 2005, 1025–1032.

[8] S. Santoso, E. J. Powers,W. M. Grady, and A. C. Parsons, “Power quality disturbance waveform recognition using wavelet-based neural classifier—part 1: theoretical foundation,” IEEE Trans. Power Delivery, vol.15, pp. 222–228, Jan. 2000.

[9] , “Power quality disturbance waveform recognition using waveletbased neural classifier—part 2: application,” IEEE Trans. Power Delivery, vol. 15, pp. 229–235, Jan. 2000.

[10] J. Chung, E. J. Powers, J. Lamoree, and S. C. Bhatt, “Power disturbance classifier using a rule-based method and wavelet packet-based hidden Markov model,” IEEE Trans. Power Delivery, vol. 17, pp. 233–241, Jan. 2002.

[11]G. Zheng, D. M'x' Shi, Liu,1. Yao, Miao, Z.M, Power quality disturbance classification based on rule-based and wavelet-multi-resolution decomposition, Proceedings of 2002 International Conference on Machine Learning and Cybernetics, 4, 2002, 2137-2141.

[12] F. Mo and W. Kinsner, “Wavelet modeling of transients in power systems,”in Proc. Conf. Communications, Power and Computing Proceedings, Winnipeg, MB, Canada, May 22–23, 1997, pp. 132–137.

[13] S. Santoso, E. J. Powers,W. M. Grady, and P. Hofmann, “Power quality assessment via wavelet transform analysis,” IEEE Trans. Power Delivery, vol. 11, pp. 924–930, Apr. 1996.

[14] S. Santoso, W. M. Grady, E. J. Powers, J. Lamoree, and S. C. Bhatt,“Characterization of disturbance power quality events with Fourier and wavelet transforms,” IEEE Trans. Power Delivery, vol. 15, pp. 247–254, Jan. 2000.

[15] A. M. Gaouda, M. M. A. Salama, M. R. Sultan, and A. Y. Chikhani, “Power quality detection and classification using wavelet-multiresolution signal decomposition,” IEEE Trans. Power Delivery, vol. 14, pp. 1469–1476, Oct. 1999.

[16]Suriya Kaewarsa. Kitti Attakitmongcol. Thanatchai Kulworawanichpong, Recognition of power quality events by using multiwavelet-based neural networks, Electrical Power and Energy Systems, 30, 2008, 254–260.

[17] B. Perunicic, M. Mallini, Z. Wang, and Y. Liu, “Power quality disturbance detection and classification using wavelets and artifical neural networks,” in Proc. 8th Int. Conf. Harmonics and Quality of Power, vol. 1,Oct. 14–16, 1998, pp. 77–82.


[18]T. Jayasree, D. Devaraj, R. Sukanesh, Power quality disturbance classification using Hilbert transform and RBF networks, Neurocomputing, 73, 2010, 1451-1456.

[19] A. Elmitwally, S. Farghal, M. Kandil, S. Abdelkader, and M. Elkateb, “Proposed wavelet-neurofuzzy combined system for power quality violations detection and diagnosis,” Proc. Inst. Elect. Eng., Gen. Transm. Dist., vol. 148, no. 1, pp. 15–20, Jan. 2001.

[20] L. Angrisani, P. Daponte, M. D’ Apuzzo, and A. Testa, “A measurement method based on thewavelet transform for power quality analysis,” IEEE Trans. Power Delivery, vol. 13, pp. 990–998, Oct. 1998.

[21] A.M. Al-Kandari, S.A. Soliman, R.A. Alammari, Power quality analysis based on fuzzy estimation algorithm: Voltage flicker measurements,Electrical Power and Energy Systems 28 (2006) 723–728


MALISOUS NODE DETECTION SYSTEM FOR MOBILE AD HOC NETWORKSA.RAJARAM

Anna University Coimbatore India,[email protected]

DR. S. PALANISWAMIRegistrar, Anna University Coimbatore India, [email protected]

ABSTRACT—In this paper, we develop a trust based security protocol based on a MAC-layer approach which attains confidentiality and authentication of packets in both routing and link layers of MANETs. In the first phase of the protocol, we design a trust based packet forwarding scheme for detecting and isolating the malicious nodes using the routing layer information. It uses trust values to favor packet forwarding by maintaining a trust counter for each node. A node is punished or rewarded by decreasing or increasing the trust counter. If the trust counter value falls below a trust threshold, the corresponding intermediate node is marked as malicious. In the next phase of the protocol, we provide link-layer security using the CBC-X mode of authentication and encryption. By simulation results, we show that the proposed MAC-layer security protocol achieves high packet delivery ratio while attaining low delay, high speed and overhead.

Keywords- MANETs, MAC-Layer, Security Protocol, Encryption, authentication, Packet Delivery, Overhead, High speed.

I. INTRODUCTIONA. Mobile Ad-hoc NetworksAn ad-hoc network is a collection of wireless mobile nodes that forms a temporary network without any centralized administration. In such a environment, it may be necessary for one node to enlist other hosts in forwarding a packet to its destination due to the limited transmission range of wireless network interfaces. Each mobile node operates not only as a host but also as a router forwarding packets for other mobile nodes in the network that may not be within the direct transmission range of each other. Each node participates in an ad-hoc routing protocol that allows it to discover multihop paths through the network to any other node. This idea of mobile ad-hoc network is also called infrastructure less networking, since the mobile nodes in the network dynamically establish routing among themselves to form their own network on the fly.

B. Security Threats in MANETSThe current mobile ad-hoc networks allow for many different types of attacks. Although the analogous exploits also exits in wired networks but it is easy to fix by infrastructure in such a network. Current MANETs are basically vulnerable to two different types of attacks: active attacks and passive attacks. Active attack is attack when misbehaving node has to bear some energy costs in order to perform the threat. On the other hand, passive attacks are mainly due to

lack of cooperation with the purpose of saving energy selfishly. Nodes that perform active attacks with the aim of damaging other nodes by causing network outage are considered as malicious while nodes that make passive attacks with the aim of saving battery life for their own communications are considered to be selfish. In this the attacks are classified as modification, impersonation, fabrication, wormhole and lack of cooperation.

Attacks using ModificationModification is a type of attack when an authorized party not only gains access to but tampers with an asset. For example a malicious node can redirect the network traffic and conduct DOS attacks by modifying message fields or by forwarding routing message with false values.

Attacks using ImpersonationAs there is no authentication of data packets in current ad-hoc network, a malicious node can launch many attacks in a network by masquerading as another node i.e. spoofing. Spoofing is occurred when a malicious node misrepresents its identity in the network (such as altering its MAC or IP address in outgoing packets) and alters the target of the network topology that a benign node can either.

Attacks through Fabrication



Fabrication is an attack in which an authorized party not only gains the access but also inserts counterfeit objects into the system. In MANET, fabrication is used to refer the attacks performed by generating false routing messages Gray hole attackWe now describe the gray hole attack on MANETS. The gray hole attack has two phases. In the first phase, a malicious node exploits the AODV protocol to advertise itself as having a valid route to a destination node, with the intention of intercepting packets, even though the route is spurious. In the second phase, the node drops the intercepted packets with a certain probability. This attack is more difficult to detect than the black hole attack where the malicious node drops the received data packets with certainly. A gray hole may exhibitits malicious behaviour in different ways. It may drop packets coming from (or destined to) certain specific node(s) in the network while forwarding all the packets for other nodes. Another type of gray hole node may behave maliciously for some time duration by dropping packets but may switch to normal behaviour later. A gray hole may also exhibit a behaviour which is a combination of the above two, thereby making its detection even more difficult.

Wormhole AttacksWormhole attack is also known as tunneling attack. A tunneling attack is where two or more nodes may collaborate to encapsulate and exchange messages between them along existing data routes. This exploit gives the opportunity to a node or nodes to short-circuit the normal flow of messages creating a virtual vertex cut in the network that is controlled by the two colluding attackers.

Lack of CooperationMobile ad-hoc networks rely on the cooperation of all the participating nodes. The more nodes cooperate to transfer traffic, the more powerful a MANET gets. But one of the different kinds of misbehavior a node may exhibit is selfishness. A selfishness node wants to preserve own resources while using the services of others and consuming their resources..The following are the types of active attacks and its relevant solutions:

A. Black hole attackIn a black hole attack a malicious node advertising itself as having a valid route to the destination. With this intension the attacker consumes or intercepts the packet without any forwarding. An attacker can completely modify the packet and generate fake information, this cause the network traffic diverted or dropped. Let H be a malicious node. When H receives a Route Request, it sends back a RouteReply

immediately, which constructs the data and can be transmitted by itself with the shortest path. So S receives Route Reply and it is replaced by H->S. then H receives all the data from

B. Neighbor attackUpon receiving a packet, an intermediate node records its ID in the packet before forwarding the packet to the next node. However, if an attacker simply forwards the packet without redirecting its ID in the packet, it makes two nodes that are not within the communication range of each other believe that they are neighbors (i.e. one hop away from each other), resulting in a disrupted route. The neighbor attack and black hole attack prevent the data from being delivered to the destination. But the neighbor attacker does not catch and capture the data packets from the source node. It leaves the settings as soon as sending the false messages. C. Wormhole attackThe wormhole attack is one of the most powerful attacks presented here, since it involves the cooperation between two malicious nodes that participate in the network. One attacker, say node A, captures routing traffic at one point of the network and tunnels them to another point in the network, say to node B, that shares a private communication link with A. node B then selectively injects tunneled traffic back into the network. The connectivity of the nodes that have established routes over the wormhole link is completely under the control of the two colluding attackers.

D. DoS (Denial of Service) attackDenial of service attacks aim at the complete disruption of the routing function and therefore the whole operation of the ad-hoc network. Specific instances of denial of service attack include the routing table overflow and the sleep deprivation torture. In a routing table overflow attack the malicious node floods the network with bogus route creation packets in order to consume the resources of the participating node and disrupt the establishment of legitimate routes. The sleep deprivation torture aims at the consumption of batteries of a specific node by constantly keeping it engaged in routing decisions.

E. Information Disclosure attack


In this, a compromised node may leak confidential information to unauthorized nodes in the network. Such information may include information regarding the network topology, geographic location of nodes or, optimal routes to unauthorized nodes in the network. Attacks such as location disclosure and traffic analysis come under this category.

F. Rushing attackOn demand routing protocols that use route discovery process are vulnerable to this type of attack. An attacker node which receives a “route request” packet from the source node floods the packet quickly through out the network before other nodes which also receive the same “route request” packet can react. Nodes that receive the legitimate “route request

“ packet assume those packets to be the duplicates of the packet already received through the attacker node and hence discard those packets.

G. Jellyfish attackSimilar to blackhole attack, a jellyfish attacker first needs to intrude into the forwarding group and then it delays data packets unnecessarily for some amount of time before forwarding them. These results in significantly high end-to-end delay and delay jitter, and thus degrade the performance of real-time applications. In this a malicious node receives and sends RREQ and RREP normally. But before forwarding it delays the data packets without any reason for some time.

H. Byzantine attackHere a compromised intermediate node or a set of compromised intermediate nodes collectively carries out attacks such as creating routing loops, forwarding packets through non-optimal paths, or selectively dropping packets, which results in disruption or degradation of the routing services within the network. It is also called as impersonation attack because the malicious node might imitate another normal node. It also sends false routing information for creating an anomaly update in the routing table. In addition to this, attacker may get unauthorized admission to resources and sensitive information.

I. Blackmail attackThis attack is applicable against routing protocols which use mechanisms for the recognition of malicious nodes and broadcast the messages which try to blacklist the offender. By adding other legitimate nodes to their blacklists, an attacker might blackmail a legitimate node. Thu the nodes can be avoided in those routes.J. Sybil attackIn the Sybil attack, an attacker pretends to have multiple identities. A malicious node can behave as if it were a larger number of nodes either by impersonating other nodes or simply by claiming false identities. Sybil attacks are classified into three categories: direct/indirect communication, fabricated/stolen identity, and simultaneity.

K. Misrouting attackIn the misrouting attack, a non-legitimate node redirects the routing message and sends data packet to the wrong destination. This type of attack is carried out by modifying the final destination address of the data packet or by forwarding a data packet to the wrong next hop in the route to the destination.

L. Resource consumption attack

In this attack, a malicious node deliberately tries to consume the resources (e.g. battery power, bandwidth, etc.) of other nodes in the network. The attacks could be in the form of unnecessary route request control messages, very frequent generation of beacon packets, or forwarding of stale information to nodes.

M. Routing table or Route cache poisoningIn this attack, a malicious node sends false routing updates to other uncompromised nodes. Such an attack may result in suboptimal routing, network congestion or even make some part of the network inaccessible.

II. RELATED WORKFarooq Anjum et al. [1] have proposed an initial approach to detect intrusions in ad hoc networks. Anand Patwardhan et al. [2] have proposed a secure routing protocol based on AODV over IPv6, further reinforced by a routing protocol-independent Intrusion Detection and Response system for ad-hoc networks. Chin-Yang Henry Tseng [3] has proposed a complete distributed intrusion detection system has consisted of four models for MANETs with formal reasoning. Tarag Fahad and Robert Askwith [4] have concentrated on the detection phase and they have proposed a mechanism Packet Conservation Monitoring Algorithm (PCMA) is used to detect selfish nodes in MANETs. Panagiotis Papadimitratos and Zygmunt J. Haas [5] have proposed the secure message transmission (SMT) protocol and its alternative, the secure single-path (SSP) protocol SMT and SSP robustly detect transmission failures and continuously configure their operation to avoid and tolerate data loss, and to ensure the availability of communication. Ernesto Jiménez Caballero [6] has reviewed the possible attacks against the routing system, some of the IDSs proposed.

Yanchao Zhang et al. [7] have proposed a credit-based Secure Incentive Protocol (SIP) to stimulate cooperation in packet forwarding for infrastructure less MANETs. Liu et al. [8] have proposed the 2ACK scheme that has served as an


add-on technique for routing schemes to detect routing misbehavior and to mitigate the adverse effect Li Zhao and José G. Delgado-Frias [9] have proposed a scheme MARS and its enhancement E-MARS to detect misbehavior and mitigate adverse effects in ad hoc networks. Patwardhan et al. [10] have proposed an approach to secure a MANET using a threshold-based intrusion detection system and a secure routing protocol. Madhavi and Tai Hoon Kim [11] have proposed a MIDS (Mobile Intrusion Detection System) suitable for multi-hop ad-hoc wireless networks, which has detected nodes misbehavior, anomalies in packet forwarding, such as intermediate nodes dropping or delaying packets.Syed Rehan Afzal et al. [12] have explored that the security problems and attacks in existing routing protocols and then they have presented the design and analysis of a secure on-demand routing protocol, called RSRP which confiscated the problems mentioned in the existing protocols. In addition, RSRP has used a very efficient broadcast authentication mechanism which does not require any clock synchronization and facilitates instant authenticationBhalaji et al. [13] have proposed an approach based on the relationship between the nodes to make them to cooperate in an ad hoc environment. The trust values of each node in the network are calculated by the trust units. The relationship estimator has determined the relationship status of the nodes by using the calculated trust values. Their proposed enhanced protocol was compared with the standard DSR protocol and the results are analyzed using the network simulator-2.za Kamal Deep Meka et al. [14] have proposed a trust based framework to improve the security and robustness of adhoc network routing protocols. For constructing their trust framework they have selected the Ad hoc on demand Distance Vector (AODV) which is popular and used widely. Making minimum changes for implementing AODV and attaining increased level of security and reliability is their goal. Their schemes are based on incentives & penalties depending on the behavior of network nodes. Their schemes incur minimal additional overhead and preserve the lightweight nature of AODV. Azal et al. [12] have explored the security problems and attacks in existing routing protocols and then they have presented a design and analysis of a new secure on-demand routing protocol, called RSRP which confiscates the problems mentioned in the existing protocols. Moreover, unlike Ariadne, RSRP uses a very efficient broadcast authentication mechanism which does not require any clock synchronization and facilitates instant authentication.

Muhammad Mahmudul Islam et al. [15] have presented a possible framework of a link level security protocol (LLSP) to be deployed in a Suburban Ad-hoc Network (SAHN). They have analyzed various security aspects of LLSP to validate its effectiveness. To determine LLSP's practicability, they have estimated the timing requirement for each authentication process. Their initial work has indicated that LLSP is a suitable link-level security service for an ad-hoc network similar to a SAHN.Shiqun Li et al. [16] have explored that the

security issues of wireless sensor networks, and in particular propose an efficient link layer security scheme. To minimize computation and communication overheads of the scheme, they have designed a lightweight CBC-X mode Encryption/Decryption algorithm that attained encryption/decryption and authentication all in one. They have also devised a novel padding technique, enabling the scheme to achieve zero redundancy on sending encrypted/authenticated packets. As a result, security operations incur no extra byte in their scheme. Stefan Schmidt et al. [17] have proposed security architecture for self-organizing mobile wireless sensor networks that prevented many attacks these networks are exposed to. In addition, it has limited the security impact of some attacks that cannot be prevented. They analyzed their security architecture and they have showed that it has provided the desired security aspects while still being a lightweight solution and thus being applicable for self-organizing mobile wireless sensor networks.

III.OBJECTIVES&OVERVIEWOFTHE PROPOSED PROTOCOLA. ObjectivesIn this paper, we propose to design a Trust-based MAC-layer Security protocol (TMLS) based on a MAC-layer, approach which attains confidentiality and authentication of packets in routing layer and link layer of MANETs, having the following objectives:

lightweight in order to considerably extend the network lifetime, that necessitates the application of ciphers that are computationally efficient like the symmetric-key algorithms and cryptographic hash functions

cooperative for accomplishing high-level security with the aid of mutual collaboration/cooperation amidst nodes along with other protocols

attack-tolerant to facilitate the network to resist attacks and device compromises besides assisting the network to heal itself by detecting, recognizing, and eliminating the sources of attacks;

flexible enough to trade security for energy consumption;

compatible with the security methodologies and services in existence

scalable to the rapidly growing network size

B. Overview of the ProtocolWe propose a Trust based packet forwarding scheme in MANETs without using any centralized infrastructure. It uses trust values to favor packet forwarding by maintaining a trust counter for each node. A node is punished or rewarded by decreasing or increasing the trust counter. Each intermediate node marks the packets by adding its hash value. And forward the packet towards the destination node. The destination node


verifies the hash value and check the trust counter value. If the hash value is verified, the trust counter is incremented, other wise it is decremented. If the trust counter value falls below a trust threshold, the corresponding the intermediate node is marked as malicious.

This scheme presents a solution to node selfishness without requiring any pre-deployed infrastructure. It is independent of any underlying routing protocol.Wefocus on the CBC-X mode Encryption/Decryption algorithm to satisfy the necessity of minimum computational and communication overhead. This algorithm supports encryption/decryption and authentication of packets on a one-pass operation. The upper layers of the protocol stack are provided with security services obviously. A CBC-X mode symmetric key mechanism is devised to employ our link layer security system. Encryption/Decryption and authentication operations are included into a single step which reduces the computational overhead to half, instead of calculating them individually. The padding technique states that this method has no cipher text expansion for the transmitted data payload. Thus the communication overhead is reduced significantly.

IV.EFFICIENTMACLAYERSECURITY PROTOCOLA. Trust Based Forwarding SchemeIn our proposed protocol, by dynamically calculating the nodes trust counter values, the source node can be able to select the more trusted routes rather than selecting the shorter routes. Our protocol marks and isolates the malicious nodes from participating in the network. So the potential damage caused by the malicious nodes is reduced. We make changes to the AODV routing protocol. An additional data structure called Neighbors’ Trust Counter Table (NTT) is maintained by each network node. Let {Tc1, Tc2…} be the initial trust counters of the nodes {n1, n2…} along the route R1 from a source S to the destination D.

Since the node does not have any information about the reliability of its neighbors in the beginning, nodes can neither be fully trusted nor be fully distrusted. When a source S wants to establish a route to the destination D, it sends route request (RREQ) packets.Each node keeps track of the number of packets it has forwarded through a route using a forward counter (FC). Each time, when node nk receives a packet from a node ni, then nk increases the forward counter of node ni .

FCni = FCni + 1, i=1, 2…… (1)

Then the NTT of node nk is modified with the values of FCni .Similarly each node determines its NTT and finally the packets reach the destination D.When the destination D receives the accumulated RREQ message, it measures the number of packets received Prec. Then it constructs a MAC on Prec with the key shared by the sender and the destination.

The RREP contains the source and destination ids, The MAC of Prec, the accumulated route from the RREQ, which are digitally signed by the destination. The RREP is sent towards the source on the reverse route R1. Each intermediate node along the reverse route from D to S checks the RREP packet to compute success ratio as,

SRi = FCni / Prec (2)

Where Prec is the number of packets received at D in time interval t1. The FCni values of ni can be got from the corresponding NTT of the node. The success ratio value SR i is then added with the RREP packet.

The intermediate node then verifies the digital signature of the destination node stored in the RREP packet, is valid. If the verification fails, then the RREP packet is dropped. Otherwise, it is signed by the intermediate node and forwarded to the next node in the reverse route.

When the source S receives the RREP packet, if first verifies that the first id of the route stored by the RREP is its neighbor. If it is true, then it verifies all the digital signatures of the intermediate nodes, in the RREP packet. If all these verifications are successful, then the trust counter values of the nodes are incremented as

Tci = Tci + δ1 (3)

If the verification is failed, then

Tci = Tci - δ1 (4)

Where, δ1 is the step value which can be assigned a small fractional value during the simulation experiments.After this verification stage, the source S check the success ratio values SRi of the nodes ni.

For any node nk, if SRk < SRmin, where SRmin is the minimum threshold value, its trust counter value is further decremented as

Tci = Tci – δ2 (5)

For all the other nodes with SRk > SRmin, the trust counter values are further incremented as

Tci = Tci + δ2 (6)

Where, δ2 is another step value with δ2 < δ1.

For a node nk, if Tck < Tcthr, where Tcthr is the trust threshold value, then that node is considered and marked as malicious. If the source does not get the RREP packet for a time period of t seconds, it will be considered as a route breakage or failure. Then the route discovery process is initiated by the source again.The same procedure is repeated for the other routes R2,


R3 etc and either a route without a malicious node or with least number of malicious nodes, is selected as the reliable route.In this protocol, authentication is performed for route reply operation. Also, only nodes which are stored in the current route need to perform these cryptographic computations. So the proposed protocol is efficient and more secure.

B.MAC Frame Format

There are two types of proposals for a MAC algorithm: Distributed access protocols, which, like Ethernet, distribute the decision to transmit over all the nodes using a carrier-sense mechanism; and centralized access protocols,

Which involve regulation of transmission by a centralized decision maker. A distributed access protocol makes sense for an ad-hoc network of peer workstations. A centralized access protocol is natural for configurations in which a number of wireless stations are interconnected with each other and some sort of base station that attaches to a backbone wired LAN.The DCF sublayer makes use of a simple CSMA (carrier sense multiple access) algorithm. The DCF does not include any collision detection function (i.e. CSMA/CD). The dynamic range of the signals on the medium is very large, so that a transmitting station cannot effectively distinguish incoming weak signals from noise and the effects of its own transmission.

To ensure smooth and fair functioning of the algorithm, DCF includes a set of delays that amounts a priority scheme.

Figure.1. MAC frame format

FC- frame Control, SC- sequence Control, Oct. - Octets

D/I-duration/connection control, FCS-frame checks sequence.

Frame control indicates the type of frame and provides control information. Duration/connection ID indicates the time the channel will be allocated for successful transmission of a MAC frame. Address field indicates the transmitter and receiver address, SSID and source & destination address. Sequence control is used for fragmentation and reassembly.

C. CBC-X Mode Our proposed link layer security scheme works between the link layer and the radio layer. Our proposed method encrypts the data and computes the MAC, when the application data payload is passed from the AM layer to the radio layer. With the help of the radio channel, the encrypted message is sent out bit-by-bit. Confidentiality and authentication are

the of security services which are present in our proposed packet format.

The packet format of the proposed scheme is illustrated in Figure.2; the fields of the packet are the destination address field, the active message type field, the length field and the data field. We keep the one byte group field in the proposed scheme to make it general and applicable. We also use a 4 byte MAC field since it can provide enough security of integrity and authenticity for the mobile adhoc networks. Any error alteration during message transmission can be detected by re-computing the MAC and the error message would be discarded to improve efficiency. It uses an 8 byte initial vector (IV) and a block cipher mechanism to encrypt the data field of the packet. The fixed portions of both IVs are the destination address field, the AM type field and the length field. These fields take 4 bytes totally.

Figure.2. Packet Format

In our scheme, the generic communication interfaces are given to the upper layer and uses the lower radio packet interfaces. The nodes in the communication are not conscious of the operations on encryption/authentication because the security services are given clearly. To make the scheme easier, the encryption and authentication for every packet is carried out by our default mode in a single pass. In order to finish the message authentication and encryption concurrently before sending message, we built an authentication and encryption scheme called as CBC-X mode.

1) CBC-X Mode Operation:-

The basic steps involved in the encryption and decryption operations are illustrated in figure 3 and figure 4, respectively.If the first block has index 1, the formula for CBC encryption is

While the formula for CBC decryption is:


FC D/I Add. Add.Add.

Header

SC Add. Data

Frame Body

Trailer

FCS

The working of the present CBC mode is described below: One cipher text block will be returned for each plaintext block, if a part of the plaintext is encrypted. In encryption of the last block of the plaintext, one or two cipher text blocks can be returned. On the other hand, decryption works in the reverse order. Apart from the decryption of the last block, a one plaintext block will be returned for each cipher text block. After the decryption of the last

plaintext block, its padding is calculated and cut off, returning a valid plaintext.

2) CBC Padding Schemes. Plaintext is divided into blocks of 8 bytes (64 bits). The final plaintext block must be padded: the final a plaintext bytes 0 ≤ a ≤ 7 are followed by 8 − a padding bytes, valued 8 − a.

For example:

messagebyte1 || messagebyte2||’06’||’06’||’06’||’06’||’06’||’06’ ESP. X padding bytes 1 ≤ X ≤ 255 ‘01’||’02’||’03’||…..||’X’

V. PERFORMANCE EVALUATIONA. Simulation Model and Parameters

We use NS2 to simulate our proposed algorithm. In our simulation, the channel capacity of mobile hosts is set to the same value: 2 Mbps. We use the distributed coordination function (DCF) of IEEE 802.11 for wireless LANs as the MAC layer protocol. It has the functionality to notify the network layer about link breakage.In our simulation, 125 mobile nodes move in a 1250 meter x 1250 meter square region for 60 seconds simulation time. We assume each node moves independently with the same average speed. All nodes have the same transmission range of 270 meters. In our simulation, the speed is

varied from 15 m/s to 55m/s. The simulated traffic is Constant Bit Rate (CBR).

Our simulation settings and parameters are summarized in table 1.

Table 1

No. of Nodes 125Area Size 1250 X 1250Mac 802.11Radio Range 270mSimulation Time 60 secTraffic Source CBRPacket Size 512Mobility Model Random Way PointSpeed 15,25,35,45,55m/s Pause time 8Queuing policy FIFOPath loss model Two – ray (26)

Cipher Block Chaining (CBC) mode EncryptionFigure.3. Encryption


Block Cipher Encryption



Ciphertext

Key

CiphertextCiphertext

Key Key

Plaintext

Initialization vector (IV)

Cipher Block Chaining (CBC) mode decryptionFigure 4- Decryption

B. Performance MetricsWe evaluate mainly the performance according to the following metrics.

Control overhead: The control overhead is defined as the total number of routing control packets normalized by the total number of received data packets.

Average end-to-end delay: The end-to-end-delay is averaged over all surviving data packets from the sources to the destinations.Average Packet Delivery Ratio: It is the ratio of the number .of packets received successfully and the total number of packets transmitted.Average Packet Consumed Energy: It is the ratio of the total time taken for packets received successfully and the total number of packets transmitted.

The simulation results are presented in the next section. We compare our TMLS protocol with the LLSP [15] and RSRP [12] protocol in presence of malicious node environment.

C. ResultsA. Based On AttackersIn our First experiment, we vary the no. of misbehaving nodes as 20,40,60,80 and 100.

Figure 5 show the results of average packet delivery ratio for the misbehaving nodes 20, 40…100 scenarios. Clearly our TMLS scheme achieves more delivery ratio than the LLSP and RSRP scheme since it has both reliability and security features

.Figure 6 shows the results of average end-to-end delay for the misbehaving nodes 20, 40…100. From the results, we can see that TMLS scheme has

slightly lower delay than the LLSP and RSRP scheme because of authentication routines.

Figure 7 shows the results of routing overhead for the misbehaving nodes 10, 20….50. From the results, we can see that TMLS scheme has less routing overhead

than the LLSP and RSRP scheme since it does not involve route re-discovery routines.the LLSP and RSRP scheme since it has both reliability and security features.

Figure. 5 Attackers Vs Delivery Ratio Figure.6 Attackers Vs Delay





Plaintext

Key

Initialization vector (IV)

Key Key

PlaintextPlaintext

CiphertextCiphertext

Figure. 7 Attackers Vs Overhead

Figure. 8. Mobility Vs Delivery Ratio

Figure.9. Mobility Vs Delay

sFigure. 10. Mobility Vs Overhead

B. Based On Speed

In our Second experiment, we vary the speed as 20,40,60,80 and 100, with 5 attackers.

Figure 8 show the results of average packet delivery ratio for the mobility20, 40…100 for the 100 nodes scenario. Clearly our TMLS scheme achieves more delivery ratio than

Figure 9 shows the results of average end-to-end delay for the mobility10, 20, 30, 40, and 50. From the results, we can see that TMLS scheme has slightly lower delay than the LLSP and RSRP scheme because of authentication routines.

Figure 10 shows the results of routing overhead for the speed 20, 40….100. From the results, we can see that TMLS scheme has less routing overhead than the LLSP and RSRP scheme.

VI. CONCLUSIONIn this paper, we have developed a trust based security protocol which attains confidentiality and authentication of packets in both routing and link layers of MANETs. In the first phase of the protocol, we have designed a trust based packet forwarding scheme for detecting and isolating the malicious nodes using the routing layer information. It uses trust values to favor packet forwarding by maintaining a trust counter for each node. A node is punished or rewarded by decreasing or increasing the trust counter. If the trust counter value falls below a trust threshold, the corresponding intermediate node is marked as malicious. In the next phase of the protocol, we provide link-layer security using the CBC-X mode of authentication and encryption. By simulation results, we have shown that the proposed MAC-layer security protocol achieves high packet delivery ratio while attaining low delay and overhead.

REFERENCES[1] Farooq Anjum, Dhanant Subhadrabandhu and

Saswati Sarkar “Signature based Intrusion Detection for Wireless Ad-Hoc Networks: A Comparative study of various routing protocols” in proceedings of IEEE 58th Conference on Vehicular Technology, 2003.

[2] Anand Patwardhan, Jim Parker, Anupam Joshi, Michaela Iorga and Tom Karygiannis “Secure Routing and Intrusion Detection in Ad Hoc Networks” Third IEEE International Conference on Pervasive Computing and Communications, March 2005.

[3] Chin-Yang Henry Tseng, “Distributed Intrusion Detection Models for Mobile Ad Hoc Networks” University of California at Davis Davis, CA, USA , 2006.


http://ieeexplore.ieee.org/xpl/RecentCon.jsp?punumber=9592



[4] Tarag Fahad and Robert Askwith “A Node Misbehaviour Detection Mechanism for Mobile Ad-hoc Networks”, in proceedings of the 7th Annual PostGraduate Symposium on The Convergence of Telecommunications, Networking and Broadcasting, June 2006.

[5] Panagiotis Papadimitratos, and Zygmunt J. Haas, “Secure Data Communication in Mobile Ad Hoc Networks”, IEEE Journal On Selected Areas In Communications, Vol. 24, No. 2, February 2006.

[6] Ernesto Jiménez Caballero, “Vulnerabilities of Intrusion Detection Systems in Mobile Ad-hoc Networks - The routing problem”, 2006.

[7] Yanchao Zhang, Wenjing Lou, Wei Liu, and Yuguang Fang, “A secure incentive protocol for mobile ad hoc networks”, Wireless Networks (WINET), vol 13, No. 5, October 2007.

[8] Liu, Kejun Deng, Jing Varshney, Pramod K. Balakrishnan and Kashyap “An Acknowledgment-based Approach for the Detection of Routing Misbehavior in MANETs”, IEEE Transactions on Mobile Computing, May 2007.

[9] Li Zhao and José G. Delgado-Frias “MARS: Misbehavior Detection in Ad Hoc Networks”, in proceedings of IEEE Conference on Global Telecommunications Conference,November 2007.

[10] A.Patwardhan, J.Parker, M.Iorga, A. Joshi, T.Karygiannis and Y.Yesha “Threshold-based Intrusion Detection in Adhoc Networks and Secure AODV” Elsevier Science Publishers B. V. , Ad Hoc Networks Journal (ADHOCNET), June 2008.[11] S.Madhavi and Dr. Tai Hoon Kim “AN INTRUSION

DETECTION SYSTEM IN MOBILE ADHOC networks” International Journal of Security and Its Applications Vol. 2, No.3, July, 2008.

[12]. Afzal, Biswas, Jong-bin Koh,Raza, Gunhee Lee and Dong-kyoo Kim, "RSRP: A Robust Secure Routing Protocol for Mobile Ad Hoc Networks", in proceedings of IEEE Conference on Wireless Communications and Networking, pp.2313-2318,April 2008.[13]. Bhalaji, Sivaramkrishnan, Sinchan Banerjee, Sundar, and Shanmugam, "Trust Enhanced Dynamic Source Routing Protocol for Adhoc Networks", in proceedings of World Academy Of Science, Engineering And Technology, Vol. 36, pp.1373-1378, December 2008[14]. Meka, Virendra, and Upadhyaya, "Trust based routing decisions in mobile ad-hoc networks" In Proceedings of the Workshop on Secure Knowledge Management, 2006.[15]. Muhammad Mahmudul Islam, Ronald Pose and Carlo Kopp, "A Link Layer Security Protocol for Suburban Ad-Hoc Networks", in proceedings of Australian Telecommunication Networks and Applications Conference, December 2004.[16]. Shiqun Li, Tieyan Li, Xinkai Wang, Jianying Zhou and Kefei Chen, "Efficient Link Layer Security Scheme for Wireless Sensor Networks", Journal of Information And Computational Science, Vol.4, No.2,pp. 553-567, June 2007.[17]. S. Schmidt, H. Krahn, S. Fischer, and D. Wätjen, "A Security Architecture for Mobile Wireless Sensor

Networks", In proceedings of First European Workshop on Security in Ad-Hoc and Sensor Networks (ESAS 2004), August 2004.[18]. M. O. Pervaiz, M. Cardei, and J. Wu, “ Routing

Security in Ad-hoc Wireless Networks” Network Security , S. Haung, D. Maccallum, Springer, 2008.

[19] B. Awerbuch, D. Holmer, C. Nita-Rotaru, “ An On-Demand Secure routing protocol Resilient to Byzantine failures”, Proceedings of ACM workshop on wireless security 2003, Sep. 2003.[20] K. Sanzgir, and B. Dahill, “ A secure routing Protocol for ad-hoc networks”, Proceeding of the 10 th IEEE International Conference on Network Protocols, 2002, pp.1-10.

AUTHORS PROFILE

S. Palaniswami received the B.E. degree in electrical and electronics engineering from the Govt., college of Technology, Coimbatore, University of Madras, Madras, India, in 1981, the M.E. degree in electronics and communication engineering (Applied Electronics) from the Govt., college of Technology, Bharathiar University,

Coimbatore, India, in 1986 and the Ph.D. degree in electrical engineering from the PSG Technology, Bharathiar University,Coimbatore, India, in 2003. He is currently the Registrar of Anna University Coimbatore, Coimbatore, India, Since May 2007. His research interests include Control systems, Communication and Networks, Fuzzy logic and Networks, AI, Sensor Networks. . He has about 25 years of teaching experience, since 1982. He has served as lecturer, Associate Professor, Professor, Registrar and the life Member of ISTE, India.

A. Rajaram received the B.E. degree in electronics and communication engineering from the Govt., college of Technology, Coimbatore, Anna University, Chennai, India, in 2006, the M.E. degree in electronics and communication engineering (Applied Electronics) from the Govt., college of Technology, Anna University,

Chennai, India, in 2008 and he is currently pursuing the full time Ph.D. degree in electronics and communication engineering from the Anna University Coimbatore, Coimbatore, India. His research interests include communication and networks mobile adhoc networks, wireless communication networks (WiFi, WiMax HighSlot GSM), novel VLSI NOC Design approaches to address issues such as low-power, cross-talk, hardware acceleration, Design issues includes OFDM MIMO and noise Suppression in MAI Systems, ASIC design, Control systems, Fuzzy logic and Networks, AI, Sensor Networks.


http://www.ieeexplore.ieee.org/xpl/RecentCon.jsp?punumber=4410909

http://www.ieeexplore.ieee.org/xpl/RecentCon.jsp?punumber=4410909

http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=7755

NUMERICAL PREDICTION OF CAVITATING FLOWS IN A BALL

VALVEY. GAWAS**, DR. V. R. KALAMKAR**, V. MALI***

**Department of Mechanical EngineeringSardar Patel College of Engineering,Mumbai, Maharashtra, IndiaE-mail: [email protected]

*** Director, Centre for Computational Technologies (CCTech) Pvt. Ltd.,Pune. India.e-mail: [email protected]

ABSTRACTBall valve, a rotational motion valve uses ball shaped disk. It is used for on/off or throttling operations. It offers the minimum resistance to flow. To investigate the valve performance and its characteristics, the flow through the valve is studied using numerical technique. In this paper the numerical simulations were performed using commercial code FLUENT, to study the flow patterns and to estimate the valve sizing coefficient, torque coefficient and cavitation index for investigation of cavitating flows. These simulations were performed at different pressure drops and for varying percentage opening of the valve.

Keywords: Ball Valve, Cavitation, Multiphase Flows, Numerical Simulation, Valve Coefficients.

INTRODUCTIONBall valve is a quarter-turn valve that features a spherical closure device. As the ball move radially across the seal, the opening in the ball is exposed, which allow the flow. Ball valve is also categorized as high-pressure recovery valve. At intermediate openings, there are two throttling ports in series, one at the inlet and other at the outlet of the ball. Hence the system experiences double pressure drop, due to which ball valve has better cavitation characteristics.The flow pattern inside the valve, formation of vortices, and complex phenomenon like cavitation can be visualized by CFD techniques. The commercial code, STAR-CDTM, was used to investigate fluid flows through a ball valve and to estimate the important coefficients Chern et al.(2004). Using FLUENTTM, investigation of flow around a V-sector ball valve was performed Merati et al.(2001). A three dimensional numerical analysis, was performed to reveal the velocity field, pressure distributions in a butterfly valve by using FLUENTTM

Huang et al.(1996). Using AVL-FireTM, the flow containing the bubbles in a ball valve was analyzed by van Lookeren Campagne et al.(2002). To perform three dimensional analysis, to estimate pressure drop, flow coefficient and hydrodynamic torque coefficient in a butterfly valve ANSYS CFXTM was used by Song et al.(2007). Implementing mixture model of FLUENTTM, the cavitating turbulent flow for two dimensional NACA0009 hydrofoil was analyzed by Berand et al.(2006).

In this research the main objective is to model the fluid domain of 10 inch ball valve, along with the prescribed length of upstream and downstream piping system. Commercial package ICEM-CFD 12.0 was used as pre-processing tool, while FLUENT 12.0 was used as solver and for post-processing. FLUENT helps to study flow thru the valve and estimate the pressure drop, volume flow rate, sizing coefficient, torque coefficient, and cavitation index.


Figure 1 Trunnion type full port ball valve

VALVE SIZING COEFFICIENT, TORQUECOEFFICIENT,ANDCAVITATION INDEX

B. Valve Sizing Coefficient (Cv)It is defined as quantity of water in US gallons at 60oF that will pass through the valve each minute with a 1 psi pressure drop across it. It is also a measure of capacity of valve, which takes account of its size and natural restriction to flow through the valve. Also it is a dimensional value. It can be calculated by following equation.

Cv = 1.16 (1)

C. Torque Coefficient (Ct)Forces required to open/close a quarter-turn valves are caused by friction and hydrodynamic forces. Torque coefficients are dimensionless entities, varies with valve opening position. It is calculated by the following equation.

(2)

D. Cavitation Index (CI)Cavitation Index is a parameter derived from the dimensional analysis, which corresponds to the intensity of cavitation. It is defined as the ratio of forces trying to suppress cavitation to the forces trying to cause it.

For valves and other devices that create a pressure drop, Cavitation Index can be further defined in several ways as explained further.

Figure 2 Schematic diagram of ball valve and piping system

Pressure Tab 1 is installed one diameter upstream from valve. Pressure Tab 2 is installed five diameters downstream from valve, Figure 2. If the reference pressure in numerator is upstream pressure, Pt1 then

(3)If the reference pressure in numerator is downstream pressure, Pt2 then

(4)

Equation (4) is a much preferred form over (3), since downstream pressure is the pressure closer to zone, where cavitation actually occurs. Therefore downstream pressure, more directly influences the cavitation process. Equation (3) & (4) are related by the following equation.

(5)

Table 1 shows the typical range of Cavitation Index.

Table 1 Cavitation IndexCavitation Index Range

Intensity Of Cavitation

CI≥ No Cavitation< CI< No Cavitation< CI< Some Cavitaiton< CI< Sever Cavitation


CI≤ or negative Flashing

CFD MODELLINGE. Model Description

For this study a10 inch nominal diameter ball valve geometric model was initially provided in IGES format and then it was imported using ICEM-CFD 12.0

Figure 3 Ball valve and the disk (ball)

The required fluid domain was then extracted from the geometry and the prescribed length of piping system was added on upstream and downstream side.

Figure 4 Extracted fluid domain

To perform simulations at different percentage opening, different geometries were created as per angular rotation of ball valve. Tetrahedral mesh generated using ICEM-CFD 12.0, were converted to polyhedral mesh using FLUENT 12.0

F. Numerical Approach

Basic flow model: Working fluid is water, hence incompressible and viscous fluid flows through the ball valve. Pressure drop between inlet and outlet being very high induces high velocity flow, hence flow studied is turbulent in nature.

Turbulence model: To capture turbulence, Reynolds Averaged Navier-Stokes (RANS) equation is utilized. Its common form is written as;

Where u is the mean velocity and the subscript, i, j=1~3, refers to Reynolds-averaged components in three directions respectively. These Reynolds Stresses, , must be modeled in order to close the equation.For this study realizable k-ε turbulence model is utilized. Standard wall function approach is utilized for the near wall treatment. Standard wall function, provide reasonably accurate predictions for majority high Reynolds Number wall bounded flows.

Cavitation Modeling: 1. Introduction to Cavitation and

Flashing: When the internal pressure of liquid at some points falls below the vapor pressure and vapor bubbles are formed and at some point downstream, pressure rise above the vapor pressure again. As this pressure recovers, so the bubbles collapse and cavitation takes place.

Figure 5 Pressure variation for cavitation and flashing process


Flashing is a condition that occurs where pressure falls below vapor pressure and remains below it. There are then two phases flowing (liquid and its vapor) downstream.It is more severe as compared to cavitation. Because of these processes a very large and steep density variation occurs in low pressure region.

1. Role of Non-Condensable Gases/Nuclei: The term nuclei, is other name for gas bubbles or voids in the liquid. For cavitation to occur, there must be nuclei present. If liquid was completely deaerated (i.e. no nuclei), then liquid will sustain tension and would not cavitate when pressure dropped to vapor pressure. Liquid will cavitate far below the normal liquid vapor pressure

2. Cavitation Model in FLUENT: FLUENT’s current cavitation model is available only with Mixture Model. It accounts for all first order effects (i.e. phase change, bubble dynamics, turbulent pressure fluctuations and non-condensable gases). It provides mass transport between a single liquid and its vapor only. Cavitation model utilizes modified Rayleigh-Plesset equations for bubble dynamics and includes effects of turbulent pressure fluctuations and non-condensable gases.

3.Governing Equations Used for Cavitation Model:

8

Where = mass averaged velocity

mixture density

volume fraction, density and mass averaged velocity

(9)

Where Mixture viscosity

5. Vapor Transport Equation: Working fluid is assumed to be a mixture of liquid, its vapor and non-condensable gases. Vapor transport equation governs the vapor mass fraction ‘f’, given by

Where,

Mixture density, velocity vector of vapor phase,

γ =effective exchange coefficient, vapor generation and condensation rate

terms, derived from Rayleigh-Plesset equations and limiting bubble size consideration. Since turbulence has significant effect on cavitating flows, the phase change threshold pressures is raised from saturation pressure to a new function which is summation of saturation pressure and turbulence-induced pressure.

When p < pv

When p > pv

Where Ce = 0.02, Cc = 0.01,

local turbulence kinetic energy,


mass fraction of vapor and non-condensable gases,

saturation pressure, turbulence-induced pressure,

density of vapor, non-condensable gases and liquid.

volume fraction of vapor and non-condensable gases.

G. Working Fluid, Operating Condition, and Boundary Condition:

Table 2, shows the working fluid used in analysis, the operating pressure, type of boundary conditions used, phases involved, and vaporization pressure.

Table 2 Working Fluid, Operating Condition, and Boundary Condition

For Numerical Prediction ofValve sizing and Torque Coefficient Cavitating Flow

Working Fluid

Water Water, Water Vapor, Non-Condensable Gases

Operating Condition

101325 Pa 101325 Pa

Phases Single Phase Flow

Two Phase Flow (Water, Water Vapor)

Vaporization Pressure

3564 Pa at 300K

Inlet Boundary Condition

Pressure Inlet Boundary Condition

Pressure Inlet Boundary Condition

Outlet Boundary Condition

Pressure Outlet Boundary Condition

Pressure Outlet Boundary Condition

RESULTS AND DISCUSSIONH. Results of velocity field of a single phase fluid flow, Valve Sizing Coefficient, Hydrodynamic Torque and Torque Coefficient:

Following figures shows the contours of velocity on x-z plane for a pressure drops of 20.7 bar (300 psi) and different percentage opening of ball valve (10%, 30%, 50%).

Figure 6 Velocity contour plot for 10 percent opening

Figure 7 Velocity contour plot for 30 percentopening

Figure 8 Velocity contour plot for 50 percent openingFigure 9 shows Valve Sizing Coefficient graph. The experimental value of Valve Sizing Coefficient for 100 percent opening is 3180. The value obtained from simulation result for 100 percent opening is 3086. Hence experimental Cv values and Numerically obtained values can be compared for validation. Curve characteristic signifies that Cv values depends only on geometry and are independent of flow conditions. Cv reduces as valve is closed.


Figure 9 Graph of variation of valve sizing coefficient

Figure 10 Graph of variation of hydrodynamictorque

Hydrodynamic torque reduces with pressure drop. As valve is closed (i.e. reduction in percentage opening) hydrodynamic torque increases to a maximum value and then reduces. Though the hydrodynamic torque changes with pressure drop and valve opening, but in all cases, it is seen that peak point for hydrodynamic torque (maximum torque) is at 60% percent opening of valve.

Figure 11 Graph of variation of torque coefficient

Torque coefficient remains constant even as pressure varies. As valve is closed torque coefficient increases to a maximum value and then reduces. Maximum

torque coefficient occurs at 80% opening for this valve.

I. Results pressure variation, velocity field, volume fraction of vapour and cavitation index for two phase cavitating turbulent flow:Following figures shows the graph of absolute pressure variation along the pipe length, contour plots of volume fraction for vapor and velocity of mixture. These results are plotted for different percentage opening of ball valve (10%, and 30%).

a) Graphs of pressure variation:

The pressure variation shown in graphs is static pressure in absolute terms, caused due to variation in velocity. Double pressure drop in ball valve, one at the valve inlet and other at the valve outlet is also visible. These curves represent the occurrence of cavitation, as the local static pressure falls below vaporization pressure.

Figure 12 Graph of absolute pressure variation for 10 percent opening

Figure 13 Graph of absolute pressure variation for 30 percent opening

b) Velocity Contours on x-z plane:


High velocity flows causes reduction in static pressure, if pressure falls below vapor pressure than cavitation occurs.

Figure 14 Mixture velocity contour plot for 10 percent opening

Figure 15 Mixture velocity contour plot for 30 percent opening

c) Contour of Volume Fraction for Vapor Phase:

These contours represent the actual Cavitation or Flashing zones in the ball valve and the piping system. Also downstream piping system is severely affected by cavitating flow. Volume fraction for vapor phase varies between 0 and 1. 0 represents no vapor formation or no cavitation, whereas, 1 represents vapor flow (i.e. cavitating flow).

Figure 16 Volume fraction of vapor phase for 10 percent opening

Figure 17 Volume fraction of vapor phase for 30 percent opening

d) Results of Cavitation Index:

Cavitation Index represents the intensity of cavitation process. Following figures shows graphs for cavitation index for the downstream side of the piping system, plotted for different pressure drops and at different percentage opening.

Figure 18 Cavitation index for 10 percent opening and varying pressure drop


Figure 19 Cavitation index for 30 percent opening and varying pressure drop

CONCLUSION

• As valve is closed its sizing coefficient reduces.

• Torque coefficient raises, to a maximum point and then reduces, in this study it is maximum at 80 percent opening.

• Hydrodynamic torque increases to a peak point and falls as valve is closed. It is maximum at 60% opening.

• Though maximum torque coefficient is obtained at 80 percent of valve opening, it does not mean that hydrodynamic torque is maximum at the same percentage opening.

• Cavitation Index depends both on pressure drop and percentage opening.

ACKNOWLEDGMENT

Yogesh Gawas thanks, Dr. Vilas R. Kalamkar, Mr. Vijay Mali, and Centre for Computational Technologies (CCTech) Pvt. Ltd., Pune, India, for their assistance for the successful completion of this study by providing the necessary resources.

NOMENCLATURECt: Torque CoefficientCv: Valve sizing Coefficient (m3/hr)CI, CI1: Cavitation Index D: Nominal Diameter of Ball Valve (mm)

: Body Force (N)f : Mass Fractionk: Turbulence Kinetic Energy (m2/s2)n: Number of PhasePsat: Saturation Pressure (Pa)Pturb: Turbulence-Induced Pressure (Pa)Q: Volume Flow Rate (m3/s)

S.G: Specific GravityT: Hydrodynamic Torque (N-m)u: Mean Velocity (m/s)

: Mass Averaged Velocity (m/s)α : Volume Fractionγ : Effective exchange coefficientµ : Viscosity (Kg/m-s)ρ : Density (Kg/m3)σ : Surface Tension Coefficient (N/m)∆P: Pressure Drop (Pa)

: Reynolds Stress (Kg/m.s2

REFERENCES

[1] Chern, M. J.; and Wang, C. C. 2004. Control of volumetric flow-Rate of ball valve using V-port. Journal of Fluid Engineering, 126: 471–481

[2] Merati, P.; Macelt, M. J.; and Erickson R. B. 2001. Flow investigation around a V-sector Ball Valve. Journal of Fluid Engineering, 118: 662-671.[3] Huang, C.; and Kim, R. H. 1996. Three-dimensional analysis of partially open butterfly valve flows. Journal of Fluid Engineering, 123: 562-568.[4] van Lookeren Campagne, C.; Nicodemus, R.; de Bruin, G. J.; and Lohse, D. 2002. A method for pressure calculation in Ball valve containing bubbles. Journal of Fluid Engineering, 124: 765-771.[5] Song, X.; and Park, Y. C. 2007. Numerical analysis of butterfly valve-prediction of flow coefficient and hydrodynamic torque coefficient. World Congress on Engineering and Computer Science.[6] Bernad, S.; Susan-Resiga, R.; Muntean, S.; Anton, I. 2006. Numerical analysis of the cavitating flows. Proceedings of the Romanian Academy, Series A, 7.[7] Flowserve cavitation control. Flowserve Corporation, USA, FCD-FCENBR0068-01.[8] ANSYS FLUENT 12.0 User’s Manual, ANSYS, Inc.[9] Tullis, J.P., 1989. Hydraulics of Pipelines: Pumps, Valves, Cavitation, Transient. John Wiley & Sons, Inc.[10] Skousen, P.L., 1997 Valve Handbook, McGraw-Hill.



OPTIMAL POWER FLOW IN POWER SYSTEM NETWORK BY USING FACTS DEVICE

J.SUNILKUMAR*M.TECHS.V. University, Tirupathi Tel.: +919014666640 Email address:

[email protected] of Electrical and Electronics Engineering, Sri Venkateswara University College of Engineering, S.V.University, Tirupati- India.

CH. CHENGAIAHA** M.TECH (P.HD),Associate professor,S.V. University, Tirupathi. Tel.: +9199440403023. Email address:

[email protected] Department of Electrical and Electronics Engineering, Sri Venkateswara University College of Engineering, S.V.University, Tirupati- India.

ABSTRACT Controlling power flow in modern power systems can be made more flexible by the use of recent developments in power electronic and computing control technology. The Unified Power Flow Controller (UPFC) is a Flexible AC transmission system (FACTS) device that can control all the three system variables namely line reactance, magnitude and phase angle difference of voltage across the line. The UPFC provides a promising means to control power flow in modern power systems. Essentially the performance depends on proper control setting achievable through a power flow analysis program.The paper presents a reliable method to meet the requirements by developing a Newton-Raphson based load flow calculation through which control settings of UPFC can be determined for the pre-specified power flow between the lines. The proposed method keeps Newton-Raphson Load Flow (NRLF) algorithm intact and needs (little modification in the Jacobian matrix). A MATLAB program has been developed to calculate the control settings of UPFC and the power flow between the lines after the load flow is converged.Case studies have been performed on IEEE 5-bus system to show that the proposed method is effective. These studies indicate that the method maintains the basic NRLF properties such as fast computational speed, high degree of accuracy and good convergence rate.

Keywords: UPFC; FACTS (Statcon, TCSC, TCRP); Optimal power flow algorithm (N-R Method); Mat lab.

1. INTRODUCTION

As the power systems are becoming more complex it requires careful design of the new devices for the operation of controlling the power flow in transmission system, which should be flexible enough to adapt to any momentary system conditions. The operation of an ac power transmission line, is generally constrained by limitations of one or more network parameters and operating variables By using FACTS technology such as STATCON (Static Condenser), Thyristor Controlled Series Capacitor (TCSC), Thyristor controlled Phase angle Regulator (TCPR), UPFC etc., the bus voltages, line impedances, and phase angles in the power system can be regulated rapidly and flexibly.

FACTS do not indicate a particular controller but a host of controllers which the system planner can choose based on cost benefit analysis.The UPFC is an advanced power system device capable of providing simultaneous control of voltage magnitude and active and reactive power flows in an adaptive fashion. Owing to its instantaneous speed of response and unrivalled functionality, it is well placed to solve most issues relating to power flow control in modern power systems.The UPFC can control voltage, line impedance and phase angles in the power system[1] which will enhance the power transfer capability and also decrease generation cost (and improve the security




and stability) of the power system. UPFC can be used for power flow control, loop flow control, load sharing among parallel corridors In this paper UPFC is treated to operate in closed loop form and control parameters of UPFC are derived to meet the required power flow along the line.

2. UPFC model for power flow studies:

2.1 Principles of UPFC The UPFC can provide simultaneous control of transmission voltage, impedance and phase angle of transmission line. It consists of two switching converters as shown in fig1. These converters are operated from a common d.c link provided by a d.c storage capacitor. Converter 2 provides the power flow control of UPFC by injecting an ac voltage with controllable magnitude and phase angle in series with the transmission line via a series transformer. Converter one is to absorb or supply the real power demand by the converter 2 at the common d.c link. It can also absorb or generate controllable reactive power and provide shunt reactive power compensation.

Fig.1 Implementation of the UPFC by back-to-back voltage source converters.

The UPFC concept provides a powerful tool for cost effective utilization of individual transmission lines by facilitating the independent control of both the real and reactive power flow and thus the maximization of real power transfer at minimum losses in the line.

2.2 Power injection model of UPFC:

The two voltage source model of UPFC is converted in to two power injections in polar form for power flow studies with approximate impedances as shown in fig 2. The advantage of power injection representation is does not destroy the symmetric characteristics of admittance matrix. When formulated in polar form, the power flow equations are quadratic. Some numerical advantages can be obtained from the form. The polar form also leads naturally to the idea of an optimal power flow, which will be discussed in next section. The voltage sources can be represented by the relationship between the voltages and amplitude modulation ratios and phase shift of UPFC.In this model the shunt transformer impedance and the transmission line impedance including the series transformer impedance are assumed to be constant. No power loss is considered with the UPFC. However the proposed model and algorithm will give the solution of optimal power flow in the transmission lines this will be discussed in section 3.3.

Fig.2. Two Voltage source model of UPFC.

2.3 Steady state UPFC representation

There are two aspects in handling the UPFC in steady state analysis.

1. When the UPFC parameters are given, a power flow program is used to evaluate the impact of the given UPFC on the system under various conditions. In this case UPFC is operated in open loop form. The corresponding. The corresponding power flow is treated as normal power flow( Which is the out of the scope of the paper).


2. As UPFC can be used to control the line flow and bus voltage, control techniques are needed to derive the UPFC control parameters to achieve the required objective. In this case UPFC is operated in closed loop form. The corresponding power flow is called controlled power flow. This is the topic of this paper.

Fig3: steady state model of UPFC connected between bus l and m.

For a given control strategy, the power Sm1 on the UPFC-controlled transmission line l-m is set to constant (Pc + jQc). By means of the substitution theorem, this branch l-m can be detached as shown in Fig.3. in which Sm1, represents power from the bus m and Sm1, from the bus l. For each other additional UPFC, its corresponding branch can be dealt with similarly.

3. PROBLEM FORMULATION OF UPFC FOR POWER FLOW STUDIES.

3.1 Load flow problem

In this paper the load flow problems are solved by using N-R method in polar co-ordinate form is an iterative method which approximates the set of linear simultaneous equations using Taylor’s series expansion and the terms are limited to first approximation. In the power flow of the transmission line the complex power injected at the ith bus with respect to ground system is

= + j …. (3.1)

= …. (3.2) Where i=1, 2, 3………..n.Where is the voltage at the bus with respect

to ground and is the source current injected into the bus. +j = …. (3.3) Substituting for

= …. (3.4) Equating real and imaginary parts

=Real …. (3.5)=-imag …. (3.6)

=| | …. (3.7)

=| | …. (3.8) Real power reactive power can now be expressed as

(Real power)=| || | || |cos ( + - ); … (3.9)

(Reactive power)= | || | || |sin ( + ); … (3.10) i=1, 2, 3, 4………………..n-;i slack bus

3.2 UPFC modified Jacobian matrix elements.

In power flow the two power injections (Pi, Qi) and (Pj, Qj) of a UPFC can be treated as generators however because they vary with the connected bus bar voltage amplitudes and phases the relevant elements of Jacobin matrix at each iteration.

The formation of Jacobian matrix

= Where H, N, J, L are the elements of Jacobian matrix.

= ;

= ;

= ;


= ;

The elements of Jacobian matrix can be calculated as followscase1:

m i = = - ; = - = - ;

Where = +j ;

= -j ;( + ) = ( +j )*( -j );Case 2:m=i

=- - | |2;= + | |2;

= - | |2; = - | |2

3.3 Optimal power flow Algorithm:

In this paper Optimal power flow algorithm is adopted as it offers a number of advantages that is to detect the distance between the desired operating point and the closest unfeasible point. Thus it provides a measure of degree of controllability and it can provide computational efficiency with out destroying the advantages of the conventional power flow when used error feedback adjustment to implement UPFC model. The proposed model and algorithm as follows.

1. Assume bus voltage except at slack bus i.e. p=1, 2, 3……….n; p s Where n is the number of buses.

2. Form Y-bus matrix.3. Set iteration count k=0.4. Set the convergence criterion,

5. Calculate the real and reactive power

and at each bus where p=1, 2, 3…….n;

p s.

6. Evaluate = - and =

- at each bus where p=1, 2,

3…….n; p s.

7. Compare each and every residue with and if all of them are then go to step 13.

8. Calculate the elements of Jacobian matrix.9. Calculate increments in phase angles and

voltages.10. Calculate new bus voltages and respective phase angles = +

and = + where p=1, 2,

3…….n; p s.

11. Replace by and by

where p=1, 2, 3…….n; p s.12. Set k=k+1 and go to step 5.13. Print the final values of voltage magnitudes and corresponding phase angles.

4. UPFC implementation in power flow studies:

The Implementation of UPFC models in power flow is essentially a controlled power flow problem. The UPFC modeling needs the change of relevant elements of Jacobian matrix however the user defined power flow software do not allow users to directly modify the Jacobian matrix and only provide the facilities for the iteration between the main program and user defined model. This iteration some times diverges especially when the system is heavily loaded. 4.1 Control Strategies of UPFC

In this section the UPFC represented by two voltage sources of series path and shunt path is often transformed in to a pair of power injections (Pi, Qi), (Pj, Qj) at both sides of UPFC locations in order to be incorporated into power flow algorithm. From the view point of effects of these power injections on the system Qi can be independently regulated to support bus bar voltage connected at the shunt path Pi and Pj are used to manipulate line active power with equal magnitude but at reverse direction and Qj can control both j bus bar voltage


and line reactive power based on this analysis the philosophy of the UPFC local control strategy is described.After modifying all the UPFC connected branches, the load flow equations can be written as followsPi + jQi =

Pl+ jQl = ( )

Pm + jQm = ( )With the above modifications the load flow studies should be done, after convergence the control settings of UPFC can be determined as follows [1].

1. Use Qi to control bus voltage Vi2. Use Pi and Pj to control the line power.

3. Set Qj= 0 (i.e. better to use Qi to control Vi since it is closer to the bus). 4. The feedback information of power injections

(Pi, Qi) (Pj, Qj) derived from the above closed-loop controllers is then converted into UPFC control parameters.

5. CASE STUDY AND CONCLUSION 5.1 CASE STUDYIn order to investigate the feasibility of the proposed technique, a large number of power systems of different sizes and under different system conditions has been tested. It should be pointed out that the results are under so-called normal power flow, i.e. the control parameters of UPFC are given and UPFC is operated in an closed -loop form. All the results indicate good convergence and high accuracy achieved by the proposed method. In this section, the IEEE 5-bus system and a 14-bus practical system have been presented to numerically demonstrate its performance. It have been used to show quantitatively, how the UPFC performs. The original network is modified to include the UPFC. This compensates the line between any of the buses. The UPFC is used to regulate the active and reactive power flowing in the line at a pre specified value. The load flow solution for the modified network is obtained by the proposed power flow algorithm and the Matlab program is used to find the control setting of UPFC for the pre specified real and reactive power flow between any buses and the power flow between the lines are observed the effects of UPFC. The same procedure is repeated to observe the power flow between the buses.

(Depending on the pre specified value of the active and reactive power the UPFC control setting is determined after the load flow is converged.).

5.2 Test results for IEEE 5 bus system

The performance of UPFC on the IEEE 5 bus system shown in figure4.

Fig 4 IEEE 5 bus system

5.2.1 Test results without UPFCThe voltage profile of the system is tabulated below:

The Power flow profile of the systemBus code

Real power(p.u) (((p.u)

Reactive power(p.u)

Loss(p.u)

1-2 0.896246 0.868027 0.0287721-3 0.420055 0.192529 0.0160282-3 0.244368 -0.034026 0.0036492-4 0.276769 -0.024054 0.0046662-5 0.546338 0.053060 0.0123043-4 0.194745 0.040617 0.0004204-5 0.066428 0.009338 0.000462

5.2.2Test results with UPFC (between bus 3 & 4)


Bus code Voltage(p.u) Angle(rad) Angle(deg)

1 1.060000 0.000000 0.000000

2 0.992621 -0.033974 -1.946578

3 0.981487 - 0.080021 -4.584891

4 0.977982 -0.085585 -4.903633

5 0.964484 -0.099667 -5.710501

The voltage profile of the system with pre-specified real and reactive powerflows as 0.4 and 0.02.


1 1.060000 0.000000 0.000000

2 0.998054 -0.018069 -1.035270

3 0.990627 -0.047300 -2.710086

4 0.988389 -0.045750 -2.621289

5 0.972024 -0.075181 -4.307549

The choice of “ ” and “ ” are valid The control settings of UPFC are:

Voltage( in p.u)

Phase angle( in rad)

Phase angle( in deg)

0.013630 -1.817928 -104.159581

The Power flow profile of the system


Bus code Real power (p.u) Reactive power(p.u) Loss(p.u)

1-2 0.615899 0.858253 0.020914

1-3 0.279573 0.190009 0.008952

2-3 0.157541 -0.028911 0.001500

2-4 0.153222 -0.015307 0.001415

2-5 0.484222 0.053322 0.009603

3-4 0.400464 0.019989 0.000951

4-5 0.126755 0.002453 0.001375

5.2.3 Test results with UPFC (between bus 2 &5)

The voltage profile of the system is:

The choice of “ ” and “ ” are validThe control setting of UPFC is

Voltage( in p.u)

Phase angle( in rad)

Phase angle( in deg)

0.018444 -1.555852 -89.143778

The Power flow profile of the system

Bus code

Real power (p.u)

Reactive power(p.u)

Loss(p.u)

1-2 0.568953 0.857510 0.019900

1-3 0.329123 0.185785 0.010969

2-3 0.234625 -0.034794 0.003323

2-4 0.250920 -0.023983 0.003787

2-5 0.400231 0.019998 0.007340

3-4 0.099455 0.050129 0.000138

4-5 -0.053549 0.023245 0.000422



1 1.060000 0.000000 0.0000002 0.998937 -0.015406 -0.8827243 0.988461 -0.059093 -3.3857674 0.985640 -0.061540 -3.5260075 0.978559 -0.044272 -2.536618

CONCLUSIONSThe unified power flow controller provides simultaneous or individual controls of basic system parameters like transmission voltage, impedance and phase angle, thereby controlling transmitted power.In this thesis an IEEE 5-bus system is taken into consideration to observe the effects of UPFC. Load flow studies were conducted on given system to find the nodal voltages, and power flow between the nodes. The MATLAB program is run with and without incorporation of UPFC. The UPFC is incorporated between buses (3, 4) and (2,5) to improve the power flow between the lines to a pre-specified value. From the results it has been observed that the power flow between the lines is improved to a pre-specified value. Depending on the pre-specified value the UPFC control settings were determined. The real power losses between the lines were decreased after the incorporation of UPFC.so, it can be concluded that after the incorporation of UPFC the voltage profile and power flow between the lines improves. Also by using this program, control setting of UPFC for different pre-specified power flows can be obtained.

REFERENCES1. W. L.Feng, H. W. Nagan: “control settings

of Unified power flow controllers through a robust load flow calculations”, IEEE proceedings on generation, transmission and distribution. Vol146. No. 4, July 1999.

2. Nabavi-Naiki. A and iravani. M. R: “steady state and dynamic models of Unified power flow controller for power system studies”, IEEE Transactions power systems, vol. 11, no. 4, nov 1996.

3. C. R. fuerte Esquivel, Acha. E: “unified power flow controller; a critical comparison of Newton-Raphson UPFC algorithms in power flow studies” IEEE proceedings on generation, transmission, distribution, Vol143, no.5,September 1997.

4. C.R Fuerte Esquivel, Acha.E:”Newton-Raphson algorithms for the reliable solutions of large power networks with embedded FACTS devices “.IEEE proceedings on generation, transmission, distribution, Vol143, no.5,September 1996.

5. Ch.Chengaiah, G.V. Marutheswar and R.V.S. Satyanarayana “Control Setting of Unified Power Flow Controller through Load flow calculations “ Vol.3 , No.6 , December 2008

6. N.G.Hingorani & Naren “Understanding FACTS”IEEE Press, 2000.

7. C.L.Wadhwa “Electrical power systems”.Third edition 2003.

8. W.Stagg & A.H.El-abiad “computer method in power system analysis” International student Edition (McGrawHill Inc., 1968).

9. Hadi saadat “Power system analysis” (Tata Mc Graw Hill Edition 2002).

10. I.J.Nagrath and D.P.Kothati “Modern power system Analysis ” 2nd Edition Tata McGraw Hill., New Delhi.

Appendix 1

Data for IEEE 5-bus system

The data for the system under consideration is as follows:

Bus code

Impedance( Line charging(

)

1-2 0.02+j0.06 0.0+j0.030

1-3 0.08+j0.24 0.0+j0.025

2-3 0.06+j0.18 0.0+j0.020

2-4 0.06+j0.18 0.0+j0.020

2-5 0.04+j0.12 0.0+j0.015

3-4 0.01+j0.03 0.0+j0.010

4-5 0.08+j0.24 0.0+j0.025


AN EXPERIMENTAL ANALYSIS OF TWO SIDED ASSEMBLY LINE BALANCING USING GA

MUGALE UMESH S. 1, DR. NANDEDKAR VILAS M. 2, MUGALE RAMESH S. 3

1Asstt. Prof. in Mech. Engineering, Aditya Engineering College, Beed, MS, India.(Research Scholar, Swami Ramanand Teerth Marathwada University, Nanded, MS, India.)

Mobile: 09822929349, E-mail: [email protected] in Prod. Dept., S.G.G.S.College of Engg. & Technology, Nanded.

Mobile: 09422172302, E-mail: [email protected] Engineer, Suraj Industries Pvt. Ltd., Beed.

ABSTRACT - Assembly line balancing problems arise whenever an assembly line is configured redesigned or adjusted. The decision problem of optimally balancing the assembly work or tasks among the stations with respect to some objective is known as the Assembly Line Balancing Problem (ALBP). ALBP tries to achieve the best compromise between labor, facility and resource requirements to satisfy a given volume of production. On the one hand, research has focused on developing effective and fast solution methods for exactly solving the simple assembly line balancing problem (SALBP). On the other hand, a number of real-world extensions of SALBP have been introduced but solved with straightforward and simple heuristics in many cases. Therefore, there is a lack of procedures for exactly solving such generalized ALBP. Here, we have used Genetic Algorithm for Two sided assembly line balancing problem. The proposed Genetic Algorithm (GA) minimizes the cycle time for a given number of stations i.e. TALB-II. The proposed approach is illustrated with a real life problem in local Chassis Manufacturing Automotive Industry, and its performance analysis is tested on a set of test problems. The results show that the proposed approach performs well.

Keywords - Two Sided Assembly line; Genetic Algorithm; cycle time; TALB-II.

I. INTRODUCTIONAssembly lines are flow-oriented production systems, which are typical in the industrial production of high quantity, standardized commodities as well as in low volume production of customized products.An assembly line consists of work stations k = (1,….., m) arranged along a conveyor belt or a similar material handling equipment. The workpieces are consecutively launched down the line and are moved from station to station. At each station, certain operations are repeatedly performed regarding the cycle time. The decision problem of optimally partitioning the assembly work among the stations with respect to some objective is known as the Assembly Line Balancing Problem (ALBP). In a two-sided assembly line some tasks may be preferred to be performed at one side of the line, while others may be performed at either side (E) of the line. A pair of two directly facing stations is called as a mated-station and one of them calls the other a companion.A precedence relation (i, j) means that

task i must be finished before task j can be started and is represented as an arc in the precedence

graph. An example of a precedence graph is given in Fig. 1.

Fig. 1 Precedence graphIn this paper, we focus on TALBP-II; we have

developed software in C++ language using Genetic Algorithm for assembly line balancing problem. The proposed Genetic Algorithm minimizes the cycle time for a given number of stations (TALBP-II). The proposed approach is illustrated with a real life problem in a local automotive Chassis manufacturing


industry, and its performance analysis is tested on a set of test problems.

The reminder of this paper is organized as, after this Introduction in Section 2, TALB is explained. In section 3 literature survey is given. In section 4, objectives of the problem are explained.Methodology used is explained in section 5. In Section 6, a real life problem to illustrate the proposed approach is solved. The performance of the proposed approach is discussed in Section 7. Finally, some conclusions and future research directions are presented in the final section.

II. TWO SIDED ASSEMBLY LINE BALANCING (TALB)Assembly lines can be classified as one-sided assembly lines and two-sided assembly lines. Only one side (either left-side or right-side) of the line is used in a one-sided assembly line, whereas both left-side (L) and right-side (R) of the line are used parallel in a two-sided assembly line. Two-sided assembly lines are usually designed to produce high-volume large- sized standardized products, such as automobiles, trucks and buses. In a two-sided assembly line some tasks may be preferred to be performed at one side of the line, while others may be performed at either side (E) of the line. Fig. 2 shows configuration of two-sided assembly line.

Fig. 2. A two-sided assembly line(Simaria, A. S., & Vilarinho, P. M., 2009).

A pair of two directly facing stations is called as a mated-station (e.g. Stations 1 and 2), and one of them calls the other a companion. While balancing a two-sided assembly line, idle time is sometimes unavoidable even between tasks assigned to the same mated-station. Suppose that task h and task p are

assigned to Station 1, task i is assigned to the companion of this station, and task i is an immediate predecessor of task h. Task h can not be started unless task i completed. Therefore, the sequence-depended finishing time of tasks are taken into account.Two-sided assembly lines can provide several advantages over one-sided assembly lines in practice. These are:

(i) the assembly line length may be shorter than a one-sided assembly line;

(ii) it may reduce material handling cost, workers movement, set up time, and the amount of throughput time; and

(iii) it can also reduce cost of tools, and fixtures (Bartholdi, 1993).

Like the other ALB problems, the two-sided assembly line balancing problem is NP-hard class of combinatorial optimization problems (Bartholdi, 1993). The combinatorial structure of the ALB problems makes it difficult to obtain an optimal solution when the problem size increases. The TALB problem is categorized into two classes (Lee, Kim, & Kim, 2001); Type-I: minimization of the number of mated-stations (i.e., the line length) for a given cycle time, and Type-II: minimization of the cycle time for a given number of mated-stations.

Fig. 3. Two-sided Chassis Assembly Line(Source-REPL, Aurangabad)

III. LITERATURE SURVEYSeveral studies on the mixed-model one-sided assembly line balancing problem have been reported in the literature (Dar-El & Cother, 1975; Erel & Gökcen, 1999; Gökcen & Erel, 1997, 1998; Jin & Wu, 2002; Karabati & Sayin, 2003; Macaskill, 1972; McMullen & Frazier, 1997, 1998; Merengo, Nava, &


Pozetti, 1999; Simaria &Vilarinho, 2004; Thomopoulos, 1970; Vilarinho & Simaria, 2002, 2006).

The detailed reviews of such studies have been given by Baybars (1986), Ghosh and Gagnon (1989), Erel and Sarin (1998), and more recently by Scholl and Becker (2006), Becker and Scholl (2006), Boysen and Scholl (2007). Although many researches have been done for one-sided ALB problems, considerably few researchers studied the TALB problem. The TALB problem was addressed by Bartholdi (1993). He designed an interactive program that assists assembly line managers to assign tasks and presented an assignment rule. Kim, Kim, and Kim (2000) presented a genetic algorithm to solve the TALB-I problem with positional constraints. Lee et al. (2001) developed an assignment procedure for TALB-I and TALB-II problems with two performance criteria: Work relatedness and work slackness. Baykasoglu and Dereli (2008) developed an ant colony optimization based heuristic algorithm to solve TALB-I problem with zoning constraints. Hu, Wu, and Jin (2008) presented a station-oriented enumerative algorithm for TALB-I problem. Kim, Song, and Kim (2009) presented a mathematical formulation and a genetic algorithm for TALB-II problem. Wu et al. (2008) presented a formal formulation of TALB-I problem, and they developed a branch-and-bound algorithm to solve the problem.

IV. OBJECTIVES• To study real-life cases in Automotive Industries.

• To make expertise available to decision makers and technicians who need solution quickly.

• Design and Implementation of Approximate Procedures.

• Evaluation and Comparison of the Performance of the Developed Solution Procedures.

• To minimize the cycle time for the given number of stations.

• To improve the performance of the Assembly Line.

• To increase the line efficiency.

• To minimize the idle time on stations.

• To minimize the number of workers.

• To minimize the overall cost of assembly Line.

V. METHODOLOGY5.1 Problem definition

A two-sided assembly line is designed to carry out a set of product models with similar production characteristics in any model sequence and model mix. Each model has its own set of task precedence relationships, and they can be combined into a single precedence diagram.

The tasks in the combined precedence diagram for all models are performed on a set of mated-stations. Each of which has a pair of two directly facing stations (left-side and right-side stations). A task i on a model m is performed in a certain time (tim).

5.2 Genetic Algorithm (GA)“A population containing a number of trial solutions each of which is evaluated (to yield fitness) and a new generation is created from the better of them. The process is continued through a number of generations with the aim that the population should evolve to contain an acceptable solution.”A genetic algorithm starts with an initial population of individuals (also called chromosomes or strings) representing different possible solutions to a problem. The population is maintained by the iterations of the algorithm, called generations. At each generation, the fitness of each individual is evaluated, and the individual is stochastically selected for the next generation based upon its fitness. New individuals called offspring are produced by two genetic operators, crossover and mutation. The offsprings are supposed to inherit the good attributes from their parents, so that the average quality of solutions becomes better than that in the previous population. This evolutionary process is repeated until some stopping criteria are met. The strength of a GA is the flexibility to adapt itself to changing optimization criteria and constraints. The performance of a GA is heavily influenced by the factors, such as the representation of the individuals; the decoding methods; the initial population; the selection scheme; and the choice of genetic operators and their combinations.


VI. A REAL LIFE CASE TO ILLUSTRATE THE PROPOSED APPROACH

TABLE 1

TABLE 2 CHASSIS ASSEMBLY LEFT SIDE TASKS

CHASSIS ASSEMBLY RIGHT SIDE TASKSTABLE 2


LEFT SIDE INPUTS:

No. of Stations = 3

Minimum Time = 463 / 3 = 154.33 Sec.

Maximum Time = 463 / 2 = 231.5 Sec.

Cross over Probability = 0.8

Mutation Probability = 0.1

Generate Random Number (1 to N)

2-7-14-16-21-24-26-8-12-22-18-20-6-3

Generate Initial Population:-

2-7-14-16-21-24-26-8-12-22-18-20-6-3

14-16-21-24-26-8-12-22-18-20-6-3-2-7

26-8-12-22-18-20-6-3-2-7-14-16-21-24

18-20-6-3-2-7-26-8-12-22-14-21-24-16

16-14-7-2-3-6-20-18-22-12-8-26-24-21

8-26-21-24-16-14-7-2-12-22-3-6-18-20

RIGHT SIDE INPUTS:

No. of Stations = 3

Minimum Time = 489 / 3 = 163 Sec.

Maximum Time = 489 / 2 = 244.5 Sec.

Cross over Probability = 0.8

Mutation Probability = 0.1

Generate Random Number (1 to N)

9-10-4-5-1-13-11-23-19-17-27-25-29-28-15

Generate Initial Population:-

9-10-4-5-1-13-11-23-19-17-27-25-29-28-15

23-11-13-1-5-4-10-9-19-17-25-27-15-29-28

28-29-27-15-17-25-9-19-10-4-5-1-13-11-23

25-17-15-27-29-28-5-4-9-10-19-23-11-1-13

5-28-29-27-17-25-15-19-10-9-4-13-23-1-11

29-5-28-25-27-17-11-1-23-9-10-19-13-4-15


TABLE 5

RIGHT SIDE FINAL ITERATION

Sr. No. Y F= 1/Y Cum.A = F/FA Cum B Rand. No. Soln Reproduction1 187 0.0053 0.1666 0.1666 0.783 5 29-11-23-27-25-10-17-28-

15-9-4-19-13-1-52 187 0.0053 0.1666 0.3332 0.555 4 29-23-11-27-9-28-17-25-10-15-

4-1-13-19-53 187 0.0053 0.1666 0.4998 0.383 3 11-23-17-27-25-29-9-28-15-10-

4-19-13-1-54 187 0.0053 0.1666 0.6664 0.980 6 29-23-11-27-25-10-17-28-15-9-

4-19-13-1-55 187 0.0053 0.1666 0.8330 0.187 2 29-23-11-27-25-10-17-28-15-9-

4-19-13-1-56 187 0.0053 0.1666 1 0.154 1 29-23-11-27-9-25-19-15-28-17-

4-10-13-1-5FA = 0.0318


VII. THE PERFORMANCE OF THE PROPOSED APPROACH In order to assess the effectiveness of the proposed GA algorithm, a set of test problems are solved: a real-life two-sided Chassis Assembly Line problems in local Automotive Industries is studied. The experiments for the GAs were carried out. The processing time of task was changed from Max. to Min. Since the original value is much bigger than all the others, it limits the use of a smaller cycle time. Each experiment was repeated for number of times for every problem setting, and the average of the best values was reported in this section. The GA was coded in C++ and was run on a P IV Core 2 Duo - PC with Pentium CPU 3GB, 320GB. Various genetic parameters were first determined, and the genetic operators previously discussed were investigated. In a GA, several control parameters, such as population size, mutation rate and crossover rate, need to be set. Extensive preliminary experiments were carried out to determine appropriate parameter values. The elitist strategy, which always allows the best string in each generation to survive in the next generation, is employed. This can prohibit the best solution from disappearing by genetic operations. Mutation occurs on a string basis, not on a bit-by-bit basis as in classical GAs. The mutation rate is the probability that a mutation takes place in a string. The crossover rate is the proportion of offsprings that are newly produced by crossover in a population.

TABLE 6PERFORMANCE RESULTS OF RIGHT SIDE

TABLE 7

PERFORMANCE RESULTS OF LEFT SIDE

Fig. 7. Number of Iteration Vs Combined Objective Value


VI. CONCLUSIONS AND FUTURE RESEARCH DIRECTIONSIn this work, we have presented a Genetic algorithm for solving the assembly line balancing problem with considering minimizing the cycle time for the given number of stations (TALBP-II). The proposed approach is illustrated with a real life problem in a local automotive Chassis manufacturing industry and its performance analysis is tested. The experimental results show that the proposed approach obtains good solutions within a short computational time for every test problem. Studying techniques such as goal programming and fuzzy goal programming approaches may be good subjects for future research. In addition, the other meta-heuristic approaches, such as tabu search and efficient heuristic approaches may be developed to solve the problem for Two-sided assembly line balancing problems in future researches.

REFERENCES

[1] J. Bartholdi, “Balancing two-sided assembly lines: A case study”, International Journal of Production Research, vol.31, pp. 2447–2461, 1993.

[2] I. Baybars, “A survey of exact algorithms for the simple assembly line balancing problem”, Management Science, vol. 32, pp. 909–932, 1986.

[3] A. Baykasoglu, T. Dereli, “Two-sided assembly line balancing using an ant-colony based heuristic”, International Journal of Advanced Manufacturing Technology; vol. 36(5–6), pp. 582–588, 2008.

[4] C. Becker, A. Scholl, “A survey on problems and methods in generalized assembly line balancing”, European Journal of Operational Research, vol. 168, pp. 694–715, 2006.

[5] N. Boysen, M. Fliedner and A. Scholl, “A classification of assembly line balancing problems”, European Journal of Operational Research, vol. 183, pp. 674–693, 2007.

[6] N. Boysen, M. Fliedner, A. Scholl, “Assembly line balancing: Which model to use when?” , Internat. J. Product. Economics, vol. 111, pp. 509-528, 2008

[7] X. F. Hu, E. F. Wu and Y. Jin, “A station-oriented enumerative algorithm for two sided assembly line balancing” , European Journal of Operational Research, vol. 186, pp. 435–440, 2008.

[8] Y. K. Kim, W. S. Song, J. H. Kim, “A mathematical model and a genetic algorithm for

two-sided assembly line balancing”, Computers and Operations Research, vol. 36,pp. 853-865, 2009.

[9] Y. K. Kim, Y. Kim, and Y. J. Kim, “Two-sided assembly line balancing: a genetic algorithm approach’, Production Planning and Control, vol. 11, pp. 44–53, 2000.

[10] T. O. Lee, Y. Kim, and Y. K. Kim, “Two-sided assembly line balancing to maximise work relatedness and slackness”, Computers & Industrial Engineering, vol. 40, pp. 273–292, 2001.

[11] A. Scholl, C. Becker, “State-of-the-art exact and heuristic solution procedures for simple assembly line balancing”, European Journal of Operational Research, vol. 168, pp. 666–693, 2006.

[12] A. S. Simaria, P. M. Vilarinho, “A genetic algorithm based approach to the mixed-model assembly line balancing problem of type II”, Computers & Industrial Engineering, vol. 47, pp. 391–407, 2004.

[13] A. S. Simaria, P. M. Vilarinho, “2-ANTBAL: An ant colony optimization algorithm for balancing two sided assembly lines”, Computers & Industrial Engineering, vol. 56, pp.489–506, 2009.

[14] U. Özcan, B. Toklu, “Balancing of Mixed–Model two–sided assembly lines”, Computers & Industrial Engineering. Doi.10.1016/. Cie.2008.11.012.

[15] Y. Kim, Y.J. Kim & Y. Kim, “Genetic Algorithms for Assembly Line Balancing with Various Objectives”, Computers Ind. Engg. Vol. 30, No. 3, pp. 397-409, 1996.

[16] J. Rubinovitz, G. Levitin, “Genetic Algorithm for Assembly Line Balancing”, Int. Journal of Production Economics, vol. 41, pp. 343-354, 1995.

[17] I. Sabuncuoglu, E. Erel and M. Tanyer, “Assembly Line Balancing using Genetic Algorithms”, Journal of Intelligent Manufacturing, vol. 11, pp. 295-310, 2000.

[18] Goldberg D. E. Genetic Algorithms in Search, Optimization & Machine Learning. Addison- Wesley, Reading MA (1989).


PERFORMANCE AND EMISSION CHARACTERISTICS OF A DIESEL ENGINE FUELED WITH PALM OIL METHYL ESTER AND ITS BLENDS

WITH DIESELB. DEEPANRAJ1, P. LAWRENCE2

1Department of Mechanical Engineering, Adhiparasakthi Engineering College, Melmaruvathur-603319, Tamilnadu, India

Email: [email protected] 2Department of Mechanical Engineering, Priyadarshini Engineering College,

Vaniyambadi-635751, Tamilnadu, IndiaEmail: [email protected]

ABSTRACT

The recent energy trend offers a challenge as well as an opportunity to look for substitutes of fossil fuels for both economic and environmental benefits. Investigation and development of bio-fuels as an alternative and renewable source of energy for transportation has become a major target in the effort towards energy self-reliance. Bio-fuel commands crucial advantages such as technical feasibility of blending, superiority from the environment and emission reduction angle, its capacity to provide energy security to remote and rural areas and employment generation. Moreover, Bio fuel will also provide rich bio mass and nutrients to the soil. This paper presents the results of performance and emission analyses carried out in an unmodified diesel engine fueled with palm oil methyl ester (POME) blends with diesel. Engine tests have been conducted to get the comparative measures of brake thermal efficiency, specific fuel consumption and emissions such as CO, HC, NOx to evaluate the behavior of POME and diesel in varying proportions. The results reveal that blends of POME with diesel up to 40% by volume (B40) provide better engine performance and improved emission characteristics.

Keywords: Palm oil, diesel, biodiesel, methyl esters;

INTRODUCTION

The diesel engine dominates the field of transportation and agricultural machinery due to its ease of operation and higher fuel efficiency. The consumption of diesel fuel is several times higher than that of petrol. Due to the depletion of petroleum reserves and its increasing cost, there is an urgent need for the alternate fuels especially for the use in diesel engines. It has been found that the vegetable oils are promising fuels because their properties are similar to that of diesel.

Vegetable oils are renewable and potentially inexhaustible source with energy content close to the diesel. The major difficulties in using the crude vegetable oils in diesel engines are because of their high viscosity and the low volatility which cause poor fuel atomization, incomplete combustion and carbon deposition on the injector and valve seats. Because of the poor atomization and low volatility of

fuel, the smoke and particulate emissions are relatively more. To overcome the above difficulties, the processes like pyrolysis, transesterification, blending with diesel, etc were developed [1-4]. There are many plant species which bear seeds rich in oil. Of these some promising species produce oils like karanja, jatropha, palm, sal, neem, etc have great potential to make biodiesel for supplementing other conventional sources like fossil fuels [5]. In this analysis, the palm oil is chosen as a potential alternative for producing biodiesel and use as fuel in compression ignition engines.

ESTERIFICATION PROCESS

Transesterification is the chemical reaction between triglycerides and alcohol in the presence of a catalyst to produce monoesters. The long and branched chain triglyceride molecules are transformed to monoesters and glycerin.


Transesterification process consists of a sequence of three consecutive reactions where triglycerides are converted to diglycerides; diglycerides are converted to monoglycerides followed by the conversion of monoglycerides to glycerol. In each step an ester is produced and thus three ester molecules are produced from one molecule of triglyceride [6, 7]. The properties of these esters are comparable to that of diesel.Transesterification of palm oil was carried out by heating of oil, addition of NaOH and methyl alcohol, stirring of mixture, separation of glycerol, washing with distilled water and heating for removal of water [8-11]. The properties of palm oil methyl ester (POME) prepared by the transesterification process were compared with diesel in table 1.

EXPERIMENTAL SETUPFig. 1 shows the experimental setup and the table 2 shows the specification of the engine.

Figure 1: Experimental setup

1. Diesel engine

2. Swinging field electrical dynamometer

3. Rheostat

4. Air box

5. U tube manometer

6. Fuel tank

7. Fuel measurement flask

8. Exhaust gas analyzer

9. AVL 415 variable sampling smoke meter


TEST PROCEDURE

Testing was carried out at a constant speed rate of 1500 rpm, at no load condition to full load condition for all the test fuel. The engine was first operated on diesel fuel and then with blends of POME (bio-diesel blends) with diesel. The fuel consumption rate, air consumption rate, composition of exhaust gases were recorded for 5 different loads at steady state condition.

RESULTS AND DISCUSSIONS

It was observed that, while operating the engine with blends of POME it was running smooth at all loads. The performance and emission characteristics such as brake thermal efficiency, specific fuel consumption and the composition of exhaust gases are presented for different percentages of load for diesel oil POME blends.Fig.2 shows the variation of brake thermal efficiency with different loads for different biodiesel blends and diesel. The brake thermal efficiency is defined as the actual brake work per cycle divided by the amount of fuel chemical energy. From the figure it is observed that brake thermal efficiency increases with increase of load. But on increasing the blend percentage there is nominal decrease in the brake thermal efficiency.

Figure 2: BTE vs Load

Fig.3 shows the specific fuel consumption of POME blends with diesel. The specific fuel consumption keeps on decreasing with increasing load. But the specific fuel consumption of B50 is slightly higher when compared to that of diesel. This is due to low volatility of biodiesel.

Figure 3: SFC vs Load

Fig.4 shows the variation of exhaust gas temperature with load. The exhaust gas temperature increases with increasing in load. Due to incomplete combustion of injected fuel and part of the combustion extending into the exhaust stroke, there is a slight increase in exhaust gas temperature with biodiesel compared to diesel.

Figure 4: EGT vs Load

Fig.5 shows the variation of CO emissions with respect to different loads. The CO emissions were found to be comparable at lower loads but at higher loads it is slightly higher than that of others. From the figure it is observed that the CO value is higher for the blend B20.

Figure 5: CO vs Load

Fig.6 shows the variation of HC emission with respect to different loads. The HC value gradually increases with increase in load.

Figure 6: HC vs Load


Fig.7 shows the variation of NOx emission with respect to load. The NOx emission is almost same for all fuels at lower loads but at higher loads it is slightly higher than that of diesel fuel.

Figure 7: NOx vs Load

CONCLUSION

It was found that the blends of POME and diesel could be successfully used with acceptable performance and better emissions than pure diesel up to a certain extent. From the experimental analysis, it is concluded that blends of POME with diesel up to 40% by volume (B40) could replace the diesel for diesel engine applications for getting less emissions and better performance.

REFERENCES

[1] Ayhan Demirbas., 2008, “Relationships derived from physical properties of vegetable oil and biodiesel fuels”, Fuel, 87, pp. 1743–1748.

[2] Ramadhas, A. S., Muraleedharan, C., Jayaraj, S., 2005, “Performance and emission evaluation of a diesel engine fueled with methyl esters of rubber seed oil”, Renewable Energy, 30, pp. 1789-1800.

[3] Srivastava, A., Ram Prasad., 2000, “Triglycerides-based diesel fuels”, Renewable and Sustainable Energy Reviews, 24, pp.111-133.

[4] Sharma, Y. C., Singh, B., Upadhyay, S. N., 2008, “Advancements in development and characterization of biodiesel: A review”, Fuel, 87, pp. 2355–2373.

[5] Karaosmanoplu, F., 1999, “Vegetable oil fuels: A review”, Energy Sources, 70, pp.221-231.

[6] Ayhan Demirbas., 2005, “Biodiesel production from vegetable oils via catalytic and non-catalytic supercritical methanol transesterification methods”, Progress in Energy and Combustion Science, 31, pp. 466–487.

[7] Naik, S. N., Vidya Sagar, D., Mether, L. C., 2006, “Technical aspects of biodiesel production by transesterification-A review”, Renewable and Sustainable Energy Reviews, 10, pp.248-268.

[8] Hidekl Fukuda., Akihiko Kondo., Hideo Noda., “Biodiesel fuel production by transesterification of oils”, Journal of Bioscience and Bioengineering, 2001, 92, pp.405-416.

[9] Fangrui, M.A., Hanna, M.A., 1999, “Biodiesel production: A review”, Bio Resource Technology, 70, pp.1-15.

[10]Agarwal, A.K., “Biodiesel for CI engines: an obvious choice for next millennium”. Proceedings of the 12th international congress and exhibition on research and development. R&D Vision 21st Century, 15–16 January 1999.

[11]Canakci, M., Van gerpen, J., 2001 “Biodiesel production from oils and fats with high free fatty acids”. Trans ASME, 44, pp.1429-1436.


DURABILITY PROPERTIES OF SELF CURING CONCRETE BY ADDITION OF VEGETATIVE MATERIAL AS ADMIXTURES

1GEETHA M AND 2Dr. R.MALATHY 1Research Scholar, Faculty in Civil Engineering, CSI Polytechnic College,

Salem 636 007,Tamil Nadu, India.2Principal, Excel Engineering College, Komarapalayam ,

Namakkal 637 303, Tamil Nadu, [email protected], 2 [email protected]

ABSTRACT: Curing of concrete is maintaining moisture in the concrete during early ages specifically with in 28 days of placing concrete, to develop desired properties. Curing concrete plays a major role in developing the concrete microstructure and pore structure. Good curing is not possible in most of the cases.Internal curing is based on the hydration of cement by additional internal water that was not part of the mixing water. The additional internal water is supplied by the use of light weight aggregate or with admixtures either polymeric materials or vegetative materials. Internal curing is very effective in counteracting the self desiccation , autogenous shrinkage , cracking, reduced permeability and increased durability. Internal curing is beneficial where w/c ratio is low. Early age cracking is reduced by internal curing .On going researches come out through ideas of making good concrete better. This paper concentrates mainly on studying the properties on durability of self curing concrete using some vegetative material as internal curing admixtures. The investigations are done for saturated water absorption, porosity, modified sorptivity, abrasion resistance of self curing concrete mixes.

Key words: Durability.

INTRODUCTIONModern concretes generally possess some highly advantageous properties compared to traditional concrete, for example good workability in fresh state, high strength and low permeability. But these types of concrete are more sensitive to early age cracking than traditional cracking. Such crack result in a serious problem with regard to strength, durability and appearance. Autogenous deformation is the self created deformation of a cement paste , mortar or concrete during hardening. In traditional concrete, autogenous deformation is negligible if compared to drying shrinkage. Autogenous shrinkage must be minimized since it may induce micro or macro cracking.Conventional curing procedures (ie) curing by water ponding are not effective in case of autogenous shrinkage. They may eliminate the autgenous shrinkage in small cross sections because penetration of water from external surface is limited. Also, external curing is not possible for all types of structures. The water table at all places is going down day by day and the scarcity of water is found in many areas.

Back ground (Literature review)1 It means to provide the water to hydrate all the cement accomplishing what the mixing water alone cannot do and largely eliminates autogenous shrinkage. 2 Paper provides an overview of evidence of internal water curing and experimental methods to study them.3 Water retention for self curing is higher compared to conventional concrete mixes. Self curing concrete suffered

less self desiccation and resulted in better hydration with time.4 Chemicals is altering CSH gel morphology also, possibly by growth inhibition.5 Elastic modulus of LWA’s mix was lower than for the corresponding NWC due to lower stiffness of LWA compared to NWA. LWAC realized with expanded shale reached 28 days compressive strength similar to NWC with same matrix.6 Excellent internal curing results have been obtained by using saturated LWA’s in the size #4 (4.75mm) with 2.4 % retained on the 4.75mm7Internal curing can contribute to accelerated construction technology transfers quicker turn around times , higher performance , lower maintenance cost and more predictabilityMain Thrust of the PaperThe methods used extensively to overcome this problem are the use of light weight aggregate, use of water absorbent polymers (polyethylene glycol and paraffin wax), wood derived materials. The disadvantageous being , Light weight aggregate can negatively impact strength and can lead to variability in performance. Polyethylene glycol are more controllable but are relatively expensive compared to light weight aggregate . Wood derived materials may be an alternative to other internal curing materials , while providing consistency at a lower cost but only 50% strength is reached and micro cracks are introduced by wood derived materials.In this paper a vegetative material is used as the admixture for self curing of concrete and some durability properties are compared.



Materials Cement Portland Pozzolana cement available in local market was used in the investigation.

Coarse Aggregate Crushed angular granite metal of 20mm size from a local source was used as Coarse aggregate. Its specific gravity and fineness modulus were 2.67 and 6.97 respectively.Fine Aggregate River sand was used as fine aggregate. Its specific gravity and fineness modulus were 2.6 and 2.93 respectively.AdmixtureThe extract from palak greens was added as admixture to the concrete while preparing the concrete at 0.6% weight of cement.Experimental Programme To study the behaviour of self cured concrete with three different mix ratios on durability properties.

Testing Procedure 1.Saturated water absorptionSaturated water absorption (SWA) tests were carried out on 100 mm cube specimens at the age of 28 days curing as per ASTM C 642. The specimens were weighed before drying. The drying was carried out in a hot air oven at a temperature of 105oC. The drying process continued, till the difference in mass between two successive measurements at 24 hours interval agreed closely. The dried specimens were cooled at room temperature and then immersed in water. The specimens were taken out at regular intervals of time, surface dried using a clean cloth and weighed. This process was continued till the weights became constant (fully saturated). The difference between the measured water saturated mass and oven dried mass expressed as a percentage of oven dry mass gives the SWA. The water absorption was calculated as the ratio of the absorbed water to that of the dry weight. 2 PorosityThe saturated water absorption of concrete is a measure of the pore volume or porosity in hardened concrete which is occupied by water in saturated condition. It denotes the quantity of water which can be removed on drying a saturated specimen. The porosity obtained from absorption tests is designated as effective porosity. It is determined using the following formula :Effective porosity = ( Volume of voids/ bulk volume of specimen ) x 100 The volume of voids is obtained from the volume of water absorbed by an oven dried specimen or the volume of water lost on oven drying a water saturated specimen at 105oC to constant mass.The bulk volume of the specimen is given by the difference in mass of the specimen in air and its mass under submerged condition in water.

The test of porosity on 100 mm cube specimens was carried out in accordance with the process described above.3 Modified sorptivitySorptivity measures the rate of penetration of water into the pores in concrete by capillary suction. When the cumulative volume of water that has penetrated per unit surface area of exposure ‘q’ is plotted against the square root of time of exposure ‘“t’ , the resulting graph could be approximated by a straight line passing through the origin. The slope of this straight line is considered as a measure of rate of movement of water through the capillary pores ; this measurement is called sorptivity. In this present study, the test for sorptivity was conducted on 100 mm cube specimens as per ASTM C 642 by drying the specimens in an oven at a temperature of 1050C to constant mass and then immersing them in water after cooling the specimens to room temperature and measuring the gain in mass at regular intervals of 30 minutes duration over a period of two hours. The sorptivity was computed by considering the slope of the plot ‘p’ versus ‘“t’.

4. Abrasion resistanceAbrasion resistance of concrete shall be one of the measures of its durability. Deterioration of concrete surfaces may occur due to abrasion by sliding, scraping or action of abrasive materials caused by water. The tests were carried out on specimens of 70 mm x 70 mm x 35 mm size. The test carried out after 28 days. The specimen was kept in position by clamp in abrasion testing machine after measuring the thickness accurately. The testing machine was allowed to rotate for 300 revolutions by keeping the speed of the machine as 30 revolutions per minute. After the specified number of revolutions, the specimens were taken out and final thickness was found out. Then the average loss of thickness was determined. 5.Impact strength The impact strength tests were carried out on concret

specimens of size 152 mm diameter and 62.5 mm thickness at the age of 28 days using drop weight testing device. Drop weight test recommended by ACI committee which was originally devised by Schrader, yields the number of blows necessary blows necessary to cause prescribed level of distress in the test specimen.

The number serves as a quantitative estimate of the energy absorbed by the specimen at that distress level.The specimen was placed and centered on the base plate and free to move horizontally, 2.8 mm off between the four positioned lugs at the periphery of base plate. A bracket with cylindrical sleeve was fixed in place and the hardened steel ball was placed on the top of the specimen with in the bracket. The steel ball is free to move vertically with in the sleeve. Elastometer pads were placed between the specimen and positioned lugs, to restrict movement of the specimen during testing till the first crack is visible.


A drop hammer weighing 45 N was then placed with it on the steel ball and held vertically. The drop hammer was dropped repeatedly through a height of 457 mm and the number of blows required to cause the first visible crack and to cause ultimate failure were recorded. Ultimate failure is defined in terms of the number of blows required to open the cracks in the specimen sufficiently to enable the fractured pieces to touch three or four positioning lugs on the base plate. Each blow of the hammer represents 20.2 Nm of energy. The stages of ultimate failure are clearly recognized by the fractured specimen butting against the lugs of the base plate. Result and Discussion

1.Saturated water absorptionThe results of the saturated water absorption (SWA) tests of three concrete mixes at the age of 28 days are given in Table 1. Table 1 Saturated water absorption results of three different grades of concrete mixes

The SWA of mix A, B an C were found to be 5.04,5.06 and 4.34 for conventional concrete and 5.84, 5.5 and 3.175 for concrete with admixture , 3.86, 2.68 and 3.66 for admixture added concrete but cured for a day .From the test results, it is observed that the percentage of SWA of the mixes containing admixture with a day curing proves to be lower when compared to other cases. The saturated water absorption results for the admixture is 15.8% , 8 % and 26.8 % higher than the results of conventional concrete. When the same concrete is cured for a day the results are better when compared to conventional concrete. That is the results are 23.4 % , 47 % and 15.7 % lesser than the values for conventional curing.

Fig 1 : Saturated water absorptionThe result of SWA tests has demonstrated the superior

durability characteristics of the concrete mixes with the addition of admixture along with one day curing.

2.PorosityThe results of the porosity tests of three concrete mixes at

the age of 28 days are given in Table2. Table 2 Porosity results of three different grades of

concrete mixes

The effective porosity of the mixes A,B and C were 11.11,11.42 and 9.02 for conventional concrete . For concrete with admixture, the porosity values were 21.04, 13.04 and 7.79 and that of admixture added concrete with a days curing the values were 9.02, 9.09 and 8.13 . The results of admixture added concrete prove to be higher by 8.37% and 14.2 % for mix A and mix B where as lower by 13.6% for mix C. Where as the admixture added concrete with one day curing produced results lower than that o conventional concrete by 18.8 %, 20.4 % and 9.8 %.

Fig 2 : Porosity of Concrete

3 SorptivityThe results of the sorptivity tests of three concrete mixes at the age of 28 days are presented in Table 3.

Table 3 Sorptivity results of three different grades of concrete mixes

The sorptivity of the mixes A, B and C at the age of 28 days ranged from 0.0045 to 0.015 mm/min0.5. The values of conventional concrete varied between 0.009


and 0.007 mm/min0.5 where as the admixture added concrete varied between 0.00575 and 0.135 .The admixture added concrete with one day curing varied between 0.0045 and 0.00785. From the test results, it is observed that the sorptivity of the admixture added Concrete with one day curing produced results 12.77 % , 35 % and 7.89 % lesser when compared to the conventional concrete and prove to give the best result.

Fig 3: Sorptivity of concrete4. Abrasion resistance The results of the abrasion resistance tests of various concrete mixes at the age of 28 days are given in Table 4. The average loss of thickness is found to be between 0.2mm to 0.5mm almost for all the mix ratios. The details as shown in the figure 4. The maximum loss in thickness is found to be less than 2%.

Table 4 Abrasion resistance results of different grades of concrete at the age of 28 days

Fig 4 Abrasion Resistance of concrete5. Impact strengthThe results of the impact strength tests of concrete mixes at the age of 28 days are given in Table 5. The impact strength of A, B and C grades of trial mixes at the age of 28 days was around 1414,1717,1576,1879,1677and 2465 N-m respectively . From the test results, it is observed that the impact resistance of the admixture added concrete was more when compared with that of conventional concrete.

Table 5 Impact strength of different grades of concrete at 28 days

CONCLUSION· The saturated water absorption, porosity and

sorptivity of admixture added concrete with one day curing were lower when compared with that of the concrete of other varieties.

· The abrasion resistance of admixture added concrete with one day curing was higher at the age of 28 days.

· Abrasion resistance of all the grades were almost equal.

· Impact resistance of admixture added concrete were better compared to conventional concrete

REFERENCES:1. Bentz, D.P, Snyder, K. A., Protected paste

volume in concrete – Extension to internal curing using saturated lightweight fine aggregate, Cement and Concrete Research, Vol. 29, pp. 1863-1867 (1999).

2. Experimental observation of internal water curing of concrete Pietro Lura- Ole Mejlhede Jensen Shin-Ichi Igarashi

3. Self curing concrete : Water retention , hydration and moisture transport A.S.El-Dieb.

4. Influence of Micro structure on the Physical Properties of Self Curing of Concrete : ACI Materials Journal –Sep Oct 1996 Ravindra K Dhir, Peter C Hawlett and Thomas D Dyer.

5. Internal water curing with Liapor aggregates Heron Vol 50 No 1 (2005) Pietro Lura BYG DTU – Department of Civil Engineering Technical University of Denmark & Faculty of Civil Engineering and Geo Sciences.

6. The use of Light weight fines for the Internal curing of concrete : North East Solite corporation Aug 2002.

7. Current and future trends in the application of internal curing of concrete John W Roberts, Northeast Solite Corporation , Richmond, Virginia


INVESTIGATIONS ON EFFECT OF NATURAL SHADING FOR THERMAL COMFORT IN HOT CLIMATIC ANANTAPUR, INDIA

P. SANJEEVA RAYUDU A*, DR. K. HEMACHANDRA REDDY A

Research Scholar a*; Professor a, Dept. of Mechanical Engineering,JNT University College of Engineering, Anantapur-515 002, AP, INDIA

E-mail: [email protected]

ABSTRACT The primary function of a building is to provide a comfortable indoor environment. Traditional buildings of earlier times had many built-in architectural features for achieving comfort. Typically, urban houses in India have been copies of western models. They are also unsatisfactory at the dwelling unit level, because of improper design and scanty regard to the natural conditions of the region. The drawback of such buildings consumes an enormous amount of energy. Energy is universally recognized as one of the most significant inputs for economic growth, human development. Also in India economic growth coupled with growing population necessitates an increase in housings and energy consumption. Energy use for buildings alone accounts for nearly 40% of the total energy consumed in India. As a need of the hour, investigations are made by measuring wind parameters and by thermal comfort survey on various factors affecting thermal comfort in buildings. From the investigations it is concluded that shading and proper orientation of buildings very significantly can create thermal comfort for the occupants. Hence if the buildings are carefully designed & constructed for Natural Ventilation, the benefits of the energy saving, enhancement of health & productivity and sustainability of our natural resources can be achieved.

Key Words: Thermal Comfort, Vegetation, Natural Ventilation

1. INTRODUCTION1.1 Energy ScenarioEnergy is one of the major inputs for the economic development of any country. In case of the developing countries, the energy sector assumes a critical importance in view of ever increasing energy needs requiring huge investments to meet them. According to International Energy Agency (IEA, 2004), the global consumption of energy has increased from 4,606 Mtoe (million ton oil equivalent) in 1973, to 7,287 Mtoe in 2003. The residential, commercial and institutional building sector consumed 31 percent of global energy and emitted 1900 mega tons of carbon in 1990. By 2050, its share is expected to rise to 38 percent and 3800 mega tons respectively (IPCC, 1996). On the other hand, the fossil fuels are rapidly depleting and the era of fossil fuel is gradually coming to an end.1.2.2 Range of Environmental Impacts of Building Industry:

According to a World Watch paper entitled ‘A Building Revolution’ (Achyuthan A. et al, 2007), the building industry is responsible for:ü 40% of world’s total energy,ü 30% of consumption of raw materials,ü 25% of timber harvest,ü 35% of world’s CO2 emissions,

ü 16% of fresh water withdrawal,ü 40% of municipal solid wastes,ü 50% of ozone- depleting CFC s still in use,ü 55% of timber cut for non-fuel uses,ü 30% of the residents having sick building

syndrome2. THERMAL COMFORT2.1 Definition: Thermal Comfort is that condition of mind that expresses satisfaction with the thermal environment (Koenigsberger, 2004). Thermal environment is those characteristics of the environment, which affects a person’s heat balance. From the physiological point of view, thermal comfort occurs when there is a thermal equilibrium in the absence of regulatory sweating, the heat exchange between the human body and the environment. The best that can realistically hope to achieve is a thermal environment that satisfies the majority of people in the place, or put more simply, reasonably comfort (Crowden, C.P. 1953, Givoni B. 1976).

2.2 Factors Affecting Thermal Comfort:The various factors affecting thermal comfort are shown below (Julia Purdy, 2001).



3. INVESTIGATIONS IN A SCHOOL BUILDING3.1 About the School:A school building, Radha School of Learning, which is located in Anantapur (Latitude: l4° 40’N, 77039’E), is considered for the investigations on thermal comfort of students. This is selected because of its high vegetation, landscaping, natural ventilation adaptation nature as shown above. There are number of class rooms and no class room is used fan or

mechanical equipment for its ventilation and thermal comfort. The Anantapur occupied a status of a small town to Municipal Corporation with the growth of population. It constitutes various communities, cultures and construction of buildings improved enormously. They consume significant amount energy for needs of residential buildings. For Anantapur climate daylight may not be affected with natural external shading, if trees are carefully planted (Sanjeeva Rayudu et al, 2005).

Fig. 3.1: A School Building considered for Investigations in Anantapur

Table 3.1: Three Rooms considered for Investigation with varying features of Orientation and Shading


3.2 Experimentation:Three Rooms of equal volumes (6m x 5.4m x 3m), equal floor areas & opening areas facing different orientations and with varying natural shadings are considered. The hourly air temperatures, the air velocities, RH values in the school for working hours 8:00 to 17:00 (i.e. 8:00 am to 5:00 pm) in the three rooms at the 17 strategic locations were recorded and graphs were drawn. The strategic points are taken along Longitudinal axis (East-West:1,2,3,4,5,6,7), along Transverse axis (South-North:8,9,10,11,12,13), at each corner of the room(14,15,16,17) as shown in fig. 3.4. For the summer period i.e. March - May the hourly temperatures at the center of different rooms & ambient are also recorded and graphs are drawn.3.2.1 Instrumentation: The instruments used in the study are hot wire Anemometer with a slim probe (Lutron: AM-4204) to measure air temperatures, air velocities & Hygrometer to measure relative humidities. The hot wire anemometer can measure air velocities of 0.2 - 20m/s with an accuracy of ±5%, air temperatures of 0-500C with an accuracy of ±0.80C. The hot wire anemometer has a tiny glass bead thermistor sensor for air velocity & precision thermistor for air temperatures. Table 3.2: Recorded Values of Hourly Temperature at Center of Three Rooms & in Ambient on Typical Days of 12th March, 16th April and 5th May of the year 2006.

Fig.3.3: Variation of Temperature with Time in Three

Rooms & Ambient on Typical Days a)12th March, b)16th April and c)5th May of the year 2006.

Fig. 3.4: Strategic locations of recording temperatures for Three rooms

Fig.3.6: Variations of Air Velocities, Relative Humidities at the Center of Three Rooms on 5th May

2006.


Table 3.3: Recorded Temperatures at Strategic Locations in Three Rooms on a Typical Day 5th May 2006 at 15:00h.

Fig.3.5: Variation of Temperatures at Strategic locations in Three Rooms on a Typical Day 5 th May 2006 at 15:00h.

Table 3.4: Recorded Air Velocities, Relative Humidities at the Center of Three Rooms on 5 th May 2006.


*SL R1 R2 R3 *SL R1 R2 R31 36.5 37.4 36.8 10 37.6 37.7 36.72 37 37.9 36 11 37 38 36.63 37.1 37.2 36.8 12 38.2 38.2 35.94 37.5 37 36.7 13 39 41.2 36.45 37.1 38.3 36.7 14 37.2 37.9 36.36 37.5 38.4 37.2 15 36.6 37.8 35.67 36.9 38.5 36.6 16 37.2 37 36.18 38.4 39.8 38.4 17 38 37.3 36.6

9 37.3 37.5 36.5 * SL- Strategic Locations as shown above


5-05-06Air Velocities (m/s)Time R1 R2 R3 R1 R2 R38:00 1.0 0.3 1.0 45 44 44.59:00 1.2 0.8 1.3 43 42.5 4310:00 0.0 0.0 0.2 42 42 4211:00 1.0 0.5 0.8 41.5 41.7 4212:00 0.6 0.3 0.7 41 41.3 4113:00 0.8 0.5 0.4 41 41.2 4114:00 0.5 0.5 0.3 40.5 40.8 4015:00 2.5 0.9 1.2 40.5 40.5 4016:00 0.3 0.3 0.2 41 40.5 41.217:00 0.2 0.0 0.2 42 41.1 42.1

Thermal Comfort SurveyA survey sheet is distributed to the occupants with different age groups like 5-15, 15-25, 25-35, 35-60 years in various class rooms to determine the thermal

comfort. The questions in the survey sheet were written in local language Telugu & also in English. Most of the subjects understood English. In some cases, the researcher translated the questions. Table 3.5: The survey sheet for a case study in a school building in Anantapu

(Responses: Cft-Comfort, C-cool, W-Warm, H-Hot, Y- Yes, N-No, S-Significant, NS-Not Significant)

Also the greater adaptive opportunity of the building allows occupants to adjust certain variable conditions

and the occupants assumed to wear suitable clothing for their activities (Baker, 1995). By considering all aspects, a number of response sheets collected and a sample one is presented with the summary of responses

Table 3.6: Consolidated Responses of occupants of all age groups for the questionnaire in three rooms on 6-05-06 at 15.00h.

4. RESULTS AND DISCUSSIONS

Ø From the investigations in a school building, by recording various wind parameters and thermal comfort survey the following results inferred. The relative humidity during the period of study i.e. peak summer remained between 40 to 50%, it is in comfortable zone range specified by ASHRAE Standard.

Ø By comparing comfort survey responses (Table 3.5, 3.6) of all age groups in East-West oriented rooms i.e. Room1 & Room2, the occupants of Room 1 felt comfortable and Room 2 occupants expressed very unsatisfactory conditions. This indicates that the external natural shading by vegetation (trees) plays very critical role in creating thermal comfort in buildings.

Ø By comparing comfort survey responses (Table 3.5, 3.6) of all age groups in externally natural shaded rooms i.e. Room1 & Room3, the occupants of Room 3 felt very comfortable than Room 2 occupants. This indicates that the along external natural shading by vegetation (trees) orientation of building plays very vital role in creating thermal comfort in buildings.

Ø Also from the response sheets, it is found that the occupants felt more comfortable at the window openings than at the center of the Rooms 1&3 & felt very uncomfortable at the window openings than at the center of the Room2.


Ø From experimental results of recorded wind parameters ( Table 3.2, 3.3 & Figure 3.3-3.5), it is found that there is raise in the temperatures during the investigated period March – May in all the three rooms, but the temperature effect is more in the Room 2 which is not shaded & oriented to West.

Ø From the figures 3.4 to 3.5, it is found that the temperature variation along the East - West wall is more in Room1 & Room2 which are oriented to East-West & the temperature Variation along South - North wall is less in Room1 & Room2 than Room3 which is shaded & oriented to North-South.

Ø From the figure 3.6, it is found that the variation of air velocities with low temperatures is more in the Room 1 & Room3 which causes the reduction in the temperatures than in the Room2. The vegetation cools the surrounding air by evapo-transpiration.

5. CONCLUSIONS

From experimental results, it is concluded that the temperatures are found to be maximum in West oriented room without shading (i.e. Room2) and minimum in South oriented room with shading (i.e. Room3). An average significance temperature difference is noticed between these two Rooms. On comparing the center-line temperatures of each room, it is found that the East-West (locations 1-7) center-line of Room2 is found to be at higher temperatures at all times during a day. Further, the North-South center-line is cooler as compared to East-West line in all the rooms, all the times. The air velocities are found to be increasing in Room 2 relatively higher than in Room 1&3, resulting in higher temperatures and uncomfortable conditions to the occupants in Room2. The same has been reflected in the comfort questioner (Tables 3.5 - 3.6). From the thermal comfort survey, it is concluded that the majority (80%) of the occupants felt very comfortable in North - South oriented room with shading (i.e. Room 3). It is, therefore, concluded that north-south oriented buildings with proper shading of Roof & Walls and velocity control results in better thermal comfort and reduces the energy usage as mechanical devices were not used. Investigations indicate that building orientation & solar passive techniques like natural shading with good vegetation improves the thermal comfort of building occupants. As per vastu East & North orientations are auspicious. Many of south Indians believe in Vastu and many of illiterate people also follows Vastu. For energy efficiency and environment friendly, the North-South building orientation, longer walls of a rectangular building towards North and South can be recommended.

The passive systems may not provide 100% climatic control, but they substantially reduce the task of the active systems and hence make them more economical. To a greater extent this composition would be decided by the guidelines/byelaws followed for urban planning. In order to achieve thermal comfort & energy efficiency in the building sector for urban regions, the land development byelaws should address the natural resources to design individual buildings to accommodate local circumstances.

ACKNOWLEDGEMENTS:

The author would like to thank all students, teachers and management of Radha School of Learning, Anantapur for their voluntary participation and genuine responses.

REFERENCES1. Achyuthan,A. 1998. “Environmental Emphasis in

Vastuvidya Sthapati”, Journal of Vastuvidya Pratisthanam, Kozhikode, Vol. 1, No. 2.

2. ASHRAE Standard 55, 1995. “Thermal Environment Conditions for Human Occupancy”. ASHRAE, Atlanta.

3. Baker N. and Standeven M. 1995. “A Behavioral Approach to Thermal Comfort Assessment in Naturally Ventilated Buildings”, Proceedings CIBSE National Conference, UK. pp. 76-84.

4. Crowden, C.P. 1953. “Indoor Climate and Thermal Comfort in the Tropics”, Conference on Tropical Architecture.

5. Givoni B. 1976. “Man, Climate, And Architecture”, 2ndedn. Applied Science Publishers.

6. IEA, 2004, World Energy Outlook 2004, IEA Publications, pp 48–57.

7. IPCC, 1996. Technologies, Policies and Measures for Mitigating Climate Change- IPCC Technical Paper 1; Geneva.

8. Julia Purdy, 2001. “The significant factors in modeling residential buildings”, Seventh International IBPSA conference, Brazil.

9. Koenigsberger, O. H., Ingersoll T.G., Mayhew A. and Szokolay S.V. 1975 (reprinted 2004). “Manual of Tropical Housing and Building (Part I): Climatic Design”, India: Longman Press.

10. Sanjeeva Rayudu P., Hemachandra Reddy K. Dr. 2005. “Modeling of a high performance day lighted building using EQUEST”, National conference CAME-2005, JNT University college of Engineering, Anantapur.

11. Vastu Shilpa, 2005, Foundation for Studies and Environmental Design: “Concepts of space in traditional Indian Architecture”, Mapin Publishing, Ahmedabad.
