$FWLYLW\�5HSRUW�IURP�WKH'LYLVLRQ�RI�&RPEXVWLRQ�3K\VLFV

����±����

/XQG�,QVWLWXWH�RI�7HFKQRORJ\/XQG��6ZHGHQ

/5&3±��/XQG�5HSRUWV�RQ�&RPEXVWLRQ�3K\VLFV�����

Lund Reports on Combustion Physics, LRCP–30

Division of Combustion PhysicsLund Institute of TechnologyBox 118S–221 00 LundSweden

Telephone +46 (0)46 222 9790Fax +46 (0)46 222 4542E-mail forbrf@forbrf.lth.sewww http://www.forbrf.lth.se

Printed at KF-Sigma, LundMay 1997

&RYHU�LOOXVWUDWLRQ��,PDJH�RI�D�VSDUN�SOXJ

7DEOH�RI�FRQWHQWV� 3DJH

���,QWURGXFWLRQ �

���*HQHUDO�,QIRUPDWLRQ �

����6WDII �

����9LVLWRUV �

����$FDGHPLF�'HJUHHV�%DVHG�RQ�:RUN�&DUULHG�RXW�DW�WKH�'LYLVLRQ�GXULQJ�����±���� �

����6HPLQDUV �

����,QWHUQDWLRQDO�3URMHFWV �

����%XGJHW �

����8QGHUJUDGXDWH�&RXUVHV �

���/DVHU�'LDJQRVWLFV

����/DVHU�,QGXFHG�)OXRUHVFHQFH �3.1.1 Fundamental Research 6

Picosecond Techniques 6Absolute 2D Measurements of OH in Flames 10Two-Photon Laser-Induced Fluorescence Detection of CO at Elevated Pressures 12Two-Phase Water Detection 12Glyoxal Detection 14

3.1.2 Applied Research 6Visualisation of Fuel Distributions in Premixed Ducts, in a Low-Emission Gas

Turbine Combustor, Using Laser Techniques 14Laser-Induced Fluorescence in Small Two-Stroke Engines 16Development of Diagnostic Techniques for 2-D Mapping of Temperature and NO

Concentrations 17Planar Laser-Induced Fluorescence of H2O to Study the Influence of Residual

Gases on Cycle-to-Cycle Variations in SI Engines 18

����6RRW�&KDUDFWHULVDWLRQ�DQG�9LVXDOLVDWLRQ ��

����&RKHUHQW�/DVHU�7HFKQLTXHV3.3.1 Coherent Anti-Stokes Raman Spectroscopy (CARS) 20

3.3.2 Investigations of Two-Photon Induced Polarisation Spectroscopy (TIPS) forCombustion Diagnostics 21

���6SDUN�,JQLWLRQ�DQG�,RQ�&XUUHQW�6HQVRUV ��

����%UHDNGRZQ�0RGHO ��

����3ODVPD�0RGHO ��

����6SDUN�(QHUJ\�'HSRVLWLRQ�3URMHFW ��

����,RQ�6HQVRU ��

���6LPXODWLRQ�RI�7XUEXOHQW�&RPEXVWLRQ ��

���&KHPLFDO�.LQHWLFV ��

����0RGHOOLQJ�RI�WKH�&RPEXVWLRQ�RI�+\GURFDUERQ�)XHOV ��

����6RRW�)RUPDWLRQ ��

����12;�)RUPDWLRQ ��

����(QJLQH�.QRFN ��

���5HIHUHQFHV

��� ,QWURGXFWLRQ

The Division of Combustion Physics has been a separate division within the Department of Physicssince 1991. Its activities are directed towards phenomenological aspects of combustion phenomenaand have, by tradition, a strong background in the development of and application for laser diagnos-tic techniques. These activities are still of major importance for the division. During the last fewyears there has also been a strengthening within other areas of combustion. In the field of chemicalkinetics the division is presently running projects in the area of soot formation, NOx formation, aswell as, low-temperature oxidation phenomena, involving complex fuels e.g. autoignition and knockin gas turbines and IC engines. The division has also a long tradition in studying spark ignition phe-nomena. During the few last years this has included studies of the use of the spark plug as a sensor.In the field of modelling turbulent reacting flows, activities have been directed towards large-eddysimulation, with applications in pulsating combustion, as well as, gas turbine combustion.

The work within the division is closely connected to other research activities, in the area of com-bustion, at the Lund Institute of Technology, through the Combustion Centre. Through this, Com-bustion Physics activities are interconnected with related projects at the Schools of Mechanical En-gineering, Civil Engineering and Chemical Engineering. Many of the projects described in this bi-ennial report have enhanced by close collaboration between the departments in these schools.

During the last few years part of the financial support of research has been directed towards the ini-tiation of interdisciplinary centres. In Sweden there have been two such initiatives, one towards de-velopment of Centres of Competence, financed by the Swedish Board for Technical Developments(NUTEK), the other involves the establishment of Centres of Excellence and Graduate Schools,these financed by various foundations, the largest of which is the Foundation of Strategic Research.

The Division of Combustion Physics has been deeply involved in the establishment of two suchcentres. One Centre of Competence, entitled Combustion Processes, was initiated in 1995 and isdirected towards the development of advanced diagnostic and modelling tools. This also includesthe Division of Combustion Physics, as well as, the Department of Heat and Power Engineering.This centre is mainly financed through NUTEK and nine major companies in the field of combus-tion.The Division of Combustion Physics has also played a major role in the establishment of a nationalCentre of Excellence in Combustion Science and Technology (CeCoST) financially supported bythe Foundation of Strategic Research. In this national centre there are participants from Lund Insti-tute of Technology, and also from Chalmers University of Technology, in Gothenburg and RoyalInstitute of Technology in Stockholm.Thanks to the centres described above and to active international collaboration, the reduction of na-tional research funds has not, as yet, caused any insuperable problems for the division in fulfillingits scientific objectives. For the future, it is believed that the activities will continue to develop andthat the close interaction between diagnostics and modelling will make the division competitive onthe international perspective.

0DUFXV�$OGpQ��SURIHVVRU

+HDG�RI�GLYLVLRQ

��� *HQHUDO�,QIRUPDWLRQ

���� 6WDII

7KH�VWDII�LQFOXGHG�WKH�IROORZLQJ�PHPEHUV�

− Marcus Aldén, professor, Head of Division− Per-Erik Bengtsson, Ph.D., doc.− Clemens Kaminski, Ph.D. (951005-)− Lars Martinsson, Ph.D. (950101-960331),

present position: Albany Nordiskafilt AB, Halmstad− Fabian Mauss, Ph.D. (960115-)− Kaj Nyholm, Dr. Tech. (960916-)− Michel Versluis, Ph.D., post. doc. (950101-960331),

present position: Delft University of Technology, Netherlands− Christer Löfström, res. eng.− Mohamed Akram, grad. stud. (960220-961013),

present position: Dept. of Mathematics, Quaid-i-Azam University, Islamabad, Pakistan− Öivind Andersson, grad. stud. (951016-)− Michael Balthasar, grad. stud. (960501-)− Joakim Bood, grad. stud. (950418-)− Johan Engström, grad. stud. (961001-)− Rolf Fritzon, grad. stud. (920108-950630),

present position: L. M. Ericsson, Stockholm− Christer Fureby, grad. stud. (950101-950630),

present position: Imperial College of Science, Technology and Medicine, London− Nikola Georgiev, grad. stud. (900910-961205),

present position: Haldor Topsø A/S, Lyngby, Danmark− Greger Juhlin, grad. stud. (950101-960431),

present position: Scania, Södertälje− Bengt Löfstedt, grad. stud. (950101-960531),

present position: Opsis, Furulund− Thomas Metz, grad. stud. (960701-)− Sven-Inge Möller, grad. stud.− Daniel Nilsson, grad. stud (960701-)− Frederik Ossler, grad. stud.− Raymond Reinmann, grad. stud.− André Saitzkoff, amanuensis 50% (950901-)− Christer Berntsson, elect. techn.− Elna Brodin, secr.

���� 9LVLWRUV

Thomas Dreier, Dr., Phys. Chem. Institut, Im Neuenheimer Feld 253, D-69120 Heidelberg, Ger-many

Frank Tichy, M.Sc., Div. f. Termodynamik, Norges Tekniska Högskola, Trondheim, Norway

Antonio Trigueiros, Ph.D., Instituto de Fisica "Gleb Wataghin", Universidade Estadual de Campi-nas, Unicamp, 13087-930 Campinas, São Paolo, Brasil

Michel Versluis, Ph.D., post. doc., Dept. of Molecular and Laser Physics, Faculty of Science, Uni-versity of Nijmegen, Toernooiveld, NL-6526 ED Nijmegen, The Netherlands

Sae-Zong Oh, Dr., KIST, 136-650, P O Box 131, Cheongryand, Seoul, Korea

���� $FDGHPLF�'HJUHHV�%DVHG�RQ�:RUN�&DUULHG�RXW�DW�WKH�'L�YLVLRQ�GXULQJ�����±����

'RFWRUDWHV�

Christer Fureby, “On Modelling of Unsteady Combustion Utilising Continuum Mechanical MixtureTheories and Large Eddy Simulations”, 950606

Muhammad Akram, ”Modelling of Spark to Ignition Transition in Gas Mixtures”, 961010

/LFHQWLDWHV�RI�7HFKQRORJ\�

Nikola Georgiev, “Laser-Based Diagnostic Techniques Involving Two-Photon Absorption”, 950208

Rolf Fritzon, “Development of Two-Dimensional Laser Techniques for Combustion Flow Studies”950602

Bengt Löfstedt, “Investigations of Polarisation Spectroscopy for Combustion Diagnostics”, 960411

Greger Juhlin, “Applications and Development of Laser Diagnostics for Combustion Engine Stud-ies”, 960927

���� 6HPLQDUV

Nikola Georgiev, ”Laser-Based Diagnostic Techniques Involving Two-Photon Absorption”, 950208

Ulf Westblom, Coherent, USA, “Infinity: A Novel Approach to Pulsed Laser Technology UsingPhase Conjugate Mirrors and Diode Pumping”, 950601.

Rolf Fritzon, “Development of Two-Dimensional Laser Techniques for Combustion Flow Studies”950602

Christer Fureby, “On Modelling of Unsteady Combustion Utilising Continuum Mechanical MixtureTheories and Large Eddy Simulations”, 950606

Bengt Löfstedt, “Investigations of Polarisation Spectroscopy for Combustion Diagnostics”, 960411

Greger Juhlin, “Applications and Development of Laser Diagnostics for Combustion Engine Stud-ies”, 960927

Norbert Peters, RWTH, Aachen, Germany, “Simulation of Combustion in Diesel Engines”, 961009

Muhammad Akram, “Modelling of Spark to Ignition Transition in Gas Mixtures”, 961010

���� ,QWHUQDWLRQDO�3URMHFWV

Projects within the EU, as well as, other international projects we are engaged in are presented inTable 1. We have also received an Institutional Fellowship from the Human Capital and MobilityProgram.

7DEOH��

3URJUDP 6XEMHFW 1DWLRQDO�,QGXVWULDO�SDUWQHU(8IDEA-EFFECT (JOULE) Ignition studies VOLVO TUBRITE-EURAM Droplet/gas phase visualisation VOLVO Aero Corp.ZODIAC Development of diagnostic tech-

nique for 2-D temperature, usingatom seeding

VOLVO

Human Capital and MobilityHCM

Institutional fellowships

2WKHUVIEA Laser diagnostic developmentsAGATA (EUREKA) Temperature-O2 measurements VOLVO Aero Corp.

���� %XGJHW

The Division's budget for 1996 totalled 9.4 MSEK, of which 80% came from external sources.

���� 8QGHUJUDGXDWH�&RXUVHV

Two three credit points undergraduate courses are given by the division of Combustion Physics. Atotal of 180 credit points is required for the degree of Master of Science.

The course in )XQGDPHQWDO�&RPEXVWLRQ has been given since 1992/1993. The first years it wasgiven by four lecturers presenting different parts of the course. The literature used was combinedfrom several authors in the field of combustion. The last year course (1995/1996 was given by Per-Erik Bengtsson and the course book was Griffiths and Barnard, Flame and Combustion (3rd edition).The course consists of sixteen lectures, two practicals (chemical kinetics and flame velocity meas-urements), as well as demonstrations at different divisions and departments within the Lund Insti-tute of Technology. The course is for students on their fourth year of study in the disciplines of En-gineering Physics, Engineering Chemistry and Mechanical Engineering. In the most recent yearmore than 30 students participated in the course.

The course in /DVHU�%DVHG�&RPEXVWLRQ�'LDJQRVWLFV was held for the first time in the second studyperiod of 1996/1997. The recent book by Alan Eckbreth, Laser Diagnostics for Combustion Tem-perature and Species, was used as the course book. The course consisted of fourteen lectures andtwo practicals (LIF and CARS). The course is given for students from Engineering Physics and thisyear fifteen students attended the course. In addition, some Ph.D. students also joined the course,the total number of students was then about 20.

��� /DVHU�'LDJQRVWLFV

���� /DVHU�,QGXFHG�)OXRUHVFHQFH

As already indicated in the introduction, work on the application and development of laser diagnos-tic techniques has become a tradition in Lund, one that extends back to the late seventies. Duringthe eighties, one of the major achievements was the first instantaneous multiple–point registrationproduced by the use of laser induced fluorescence (LIF). This work has continued and today thistechnique, as extended by Stanford University and Stanford Research Institute, to encompass two–dimensional measurements, is one of the most widely applied techniques in combustion diagnostics.Another achievement during the eighties, was the detection of oxygen atoms in flames for the firsttime, using a two–photon excitation scheme. As shown below, a significant amount of work is stilldevoted to the development and application of both imaging techniques and multi–photon schemes.

The laser work undertaken within the new division has been directed at the following areas:

1. The development of new diagnostic techniques and fundamental research both on theseand on well established techniques.

1. The application of well developed techniques to the measurement of relevant parameters,

e. g. species concentrations, temperatures, velocities and particle phenomenologicalstudies (volume fraction, size, and number density), within the framework of phenome-nological studies, often in close collaboration with modelling expertise.

1. Research on the use of mature techniques for the characterisation, optimisation and con-

trol of industrial processes.

The distinctions between these different areas are more virtual than real, the links between them arevery close and there is also an exchange of people, equipment and knowledge.

������ )XQGDPHQWDO�5HVHDUFK

3LFRVHFRQG�7HFKQLTXHV

�)���2VVOHU��6��$JUXS���-��/DUVVRQ��/��0DUWLQVVRQ�DQG�7��0HW]�

The most important instruments in research involving picosecond (ps) resolution are a ps-laser sys-tem for excitation and a streak camera for detection. These are used in the study of the fluorescencedecay. In atmospheric flames the effective lifetimes lie around or below 1 ns, 100 to 1000 timesshorter than the natural lifetimes. In some instances it is not necessary to use a streak camera. A fast

��3UHVHQW�SRVLWLRQ��0DOP|�/DWLQVNROD

photomultiplier tube and a fast oscilloscope are sufficient, if a deconvolution scheme is employedfor the evaluation of the effective lifetime.

The effective lifetime of the 3d state of the H atom was studied in a H2/O2 and a CH4/O2 flame[Agrup et al., 1995]. In this case a two photon-excitation scheme at 205 nm was used. The fluores-cence decay, due to de-excitation to the 2p state (656 nm) was recorded using the streak camera.The effective lifetime in the preheat zone was low 60 ± 10 ps, but increased rapidly to 105 ± 16 psin the reaction zone. The lifetime was longer in the methane flame, although the low signals did notpermit rigorous studies. In addition to the fluorescence, stimulated emission was also observed andrecorded. The stimulated emission was short-lived and followed the shape of the laser pulse.

The effective lifetime of the 2p33p3P and the 2p33p5P states of atomic oxygen were measured to be290 ± 80 ps and 250 ± 80 ps respectively, in the recombination zone of a H2/O2 flame at atmos-pheric pressure [Ossler et al., 1996a]. Measurements were also performed in different zones of theflame and in the recombination zone of a H2/N2O flame. Laser pulses with a duration of 35 ps at awavelength of 225.6 nm, excited the atoms by a two-photon absorption process. Lifetimes were de-duced from time resolved recordings of the fluorescence at 845 nm and 777 nm, using a fast photo-multiplier tube combined with a fast oscilloscope and a deconvolution scheme.

Dopants and trace markers are often used to visualise the degree of evaporation of fuel e.g. in SI en-gines in the stages prior to ignition, or in the study of turbulent flows. In order to determine the con-centrations of these species, it is important to determine their fluorescence behaviour, at variouspressures and temperatures of importance for engine applied experiments. 3-pentanone (3P) andacetone (Ac) molecules were studied in a static cell, which could operate at pressures of air, or ni-trogen, ranging from 0.01 bar to 10 bar and temperatures between 323 K and 723 K [Ossler andAldén, 1997, Ossler et al., 1996b]. The molecules were excited with ps-laser pulses at 266 nm (4thharmonic of Nd:YAG). The fluorescence emission was measured, either temporally resolved with astreak camera, or spectrally resolved with a spectrograph and an ungated diode–array detector con-nected to an OMA. Changes in the effective lifetime of the emission were observed as a function ofpressure and temperature. The effective lifetime increased, or reached a maximum, with increasingpressure of nitrogen. Temperature was found to have a quenching effect on the emission. Acshowed a larger sensitivity to both temperature and pressure compared to 3P. Mixtures of 3P and airdid not react at temperatures lower than 548 K, when the concentration of 3P was kept low. The ef-fective lifetime varied less than ± 20 % between 323 K and 473 K, at pressures of air or nitrogenbetween 1 bar and 10 bar. The behaviour of the fluorescence emission of the ketones studied withrespect to pressure and temperature was contrary to that expected of atoms and small molecules.This can, however, be explained by internal de-excitation routes (internal quenching) which are ef-fected by intermolecular collisions.

In addition to point-like measurements, or measurements along a line, the streak camera can be usedfor multi-dimensional measurements. It has been demonstrated that ps scattering and streak cameradetection can provide three-dimensional imaging of turbulent flows [Ossler et al., 1995a]. In thatinvestigation, a ps-laser was used to capture the elastic scattering from micrometer sized droplets insix planes inside a water aerosol ((UURU��5HIHUHQFH�VRXUFH�QRW�IRXQG�). A three-dimensional mapwas obtained within a period of a few tens of nanoseconds and with a single laser pulse. However,this technique has the drawback of being sensitive to signal interference due to fluorescence emis-sion. This could be partially circumvented by using narrow interference filters.

),*������7KH�FRPSOHWH�VHW�RI���'�0LH�VFDWWHUHG�SLFWXUHV�RI�WKH�ZDWHU�DHURVRO�VKRZ���PP�×���PP�RI� WKH�VFDWWHULQJ� IODPHV��7KH�GLVWDQFH�EHWZHHQ� WKH� ODVHU±VKHHW�SODQHV�ZDV�����PP��7KHLPDJHV�RQ�WKH�VWUHDN�FDPHUD�PRQLWRU�ZHUH�WDNHQ�ZLWK�D�&&'�FDPHUD��7KH�LPDJHV�ZHUH�WKHQHODERUDWHG�DQG�VHTXHQFHG�LQ� WKH�RUGHU� WKDW�FRUUHVSRQGHG� WR�D�GHFUHDVLQJ�GLVWDQFH�EHWZHHQWKH�ODVHU�VKHHWV�DQG�WKH�UHFHLYLQJ�RSWLFV��7KH�LPDJHV�DQG�WKH�FRUUHVSRQGLQJ�UHODWLYH�LQWHQVLW\VFDOH�DUH�VKRZQ�LQ�IDOVH�FRORXU�

The ps-laser and streak camera were used to demonstrate the possibility of performing two-dimensional measurements of the effective lifetime, without scanning procedures, in the measuredvolume. Operating the streak camera in the “focus mode” (no temporal deviation) and the “streakmode” (temporal deviation) two images are obtained. The latter contains the spatial and temporalinformation on one axis which

max

min

)LJ�� ����� $Q� H[DPSOH� IRU� D� WZR�GLPHQVLRQDO� REMHFW�� $� �×�±PDWUL[� RI� G\HG� DQG� QRQ±G\HGSLHFHV�RI�SDSHU��7KH�OHIW�LPDJH��³IRFXV�PRGH´��VKRZV�WKH�LQWHQVLW\�GLVWULEXWLRQ��LQWHQVLWLHV�LQDUELWUDU\�XQLWV���WKH�ULJKW�RQH�WKH�OLIHWLPH�GLVWULEXWLRQ�FDOFXODWHG�XVLQJ�6$0��OLIHWLPHV�LQ�VHF�RQGV��

can then be separated by employing a numeric algorithm (SAM = separation algorithm method) onthe two images. This generates the two-dimensional distribution of the effective lifetime [Ossler etal., 1997, Ossler et al., 1996b]. In principle the method allows the lifetimes to be determined with aresolution set by the duration of the laser pulse and the resolution of the streak camera. In this way itis possible to determine lifetimes shorter than that expected of measurements with gated image in-tensifiers. The method was tested experimentally, as a proof of the principle, on different pieces ofdyed and non-dyed paper. By this approach SAM could be compared with traditional time resolvedlifetime evaluation along a line (TRM = time resolved method). As can be seen in Fig. 3.3 theagreement between SAM and TRM was good. Finally, experiments were performed on two-dimensional objects and the corresponding images of theAn example for a two-dimensional object: A 5×2–matrix of dyed and non–dyed pieces of paper.The left image (“focus-mode”) shows the intensity distribution (intensities in arbitrary units), theright one the lifetime distribution calculated using SAM (lifetimes in seconds).two-dimensional distribution of the effective lifetime were obtained with SAM, see Fig. 3.2.

)LJXUH� ����� )OXRUHVFHQFHOLIHWLPHV� RI� IRXU� SLHFHV� RIG\HG� DQG� QRQ�G\HG� SDSHUHYDOXDWHG� DORQJ� D� OLQHXVLQJ� WKH� WZR� GLIIHUHQWWHFKQLTXHV� PHQWLRQHG� LQWKH�WH[W�

100 200 300 4000.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

SAM TRM

Life

times

(ns

)

Channel Number

$EVROXWH��'�0HDVXUHPHQWV�RI�2+�LQ�)ODPHV

�0��9HUVOXLV��1��*HRUJLHY��6��.U|OO��DQG�/��0DUWLQVVRQ�

During the last year we have been working with a detection technique for two-dimensional spatiallyresolved absolute number density measurements of OH in flames [Versluis et al., 1996c]. In thismethod the OH concentration is determined from the ratio of two LIF images of OH, one with thelaser sheet in a forward direction and one with the sheet in the opposite direction. The OH LIF pro-files are measured in two dimensions with an intensified CCD camera. Using this bi-directional ap-proach and dividing the fluorescence profiles on a pixel-to-pixel basis, the quenching rates areeliminated. In addition, the concentration determination is not influenced by fluorescence trapping.Moreover, and this is very important in many applications, the absolute concentration is directlyobtained using this technique. No separate calibration, or estimation of for instance the collectionefficiency and detector sensitivity is therefore required.

The absolute spatially resolved OH concentration is measured in two dimensions, using absorptionof the Q1(6) rotational line in the A2Σ+(v’=0) ← X2Π(v”=0) band of OH at 309 nm. It can be shownthat the number density N(x) is given by the following expression

N(x)1

2

d

dx

F (x)

F (x)FORTH

BACK

=

σ0

ln

��$WRPLF�3K\VLFV��/XQG�,QVWLWXWH�RI�7HFKQRORJ\

where FFOURTH and FBACK are the fluorescence intensity in point x for the laser beam in the forwardand backward direction through the flame. From this equation it follows that the density at anygiven point x in the sample, can be determined by calculating the slope of the logarithm of the fluo-rescence ratio in the back and forward directions. For an absolute calibration of the concentration,only the value of the peak absorption cross section σo is required. Note that the equation does notcontain any terms associated with collisional quenching.

The experimental set-up is schematically shown in )LJXUH� ���. The laser beam intensity was re-duced to approximately 200 µJ/cm2 to avoid saturation of the transition.

One series of measurements was undertaken in the flame of a welding torch, burning methane andoxygen. The torch had a slit nozzle and the V-shaped flame was very stable.

LIF images in the back and the forward directions are shown in )LJXUH���� (top). The images areaveraged over 50 laser pulses. There is strong absorption by the flame, considering the flame is only6 mm wide at the broadest part. The ratio of the two fluorescence images was calculated and thenatural logarithm was taken for every pixel in the image. The derivative with respect to the hori-zontal direction was calculated and a contour plot of the absolute OH concentration is shown in)LJXUH����. It is worthwhile to note that the symmetry in the image is fully restored.

beam splitter

mirror

P2P3

P4

flame

P1

Intensified

CCD-camera

forth

back

Dye laser system

)LJXUH� ����� 7KH� H[SHULPHQWDO� VHW�XS� IRU� DQ� DE�VRUSWLRQ� H[SHULPHQW� XVLQJ� /,)� LQ� D� EL�GLUHFWLRQDOFRQILJXUDWLRQ�� 7KH� ODVHU� EHDP� LV� VSOLW� LQWR� EHDPVWUDYHUVLQJ� WKH� IODPH� LQ� WKH� IRUZDUG� DQG� LQ� WKHEDFNZDUG� GLUHFWLRQV�� 7KH� /,)� SURILOHV� DUH� PHDV�XUHG�ZLWK�DQ�LQWHQVLILHG�&&'�FDPHUD�

)LJXUH� ����� 7ZR� GLPHQVLRQDO� DEVRUSWLRQLQ� WKH� &+��2�� IODPH� RI� D� ZHOGLQJ� WRUFK�7KH� WZR� WRS� LPDJHV�VKRZ�WKH�/,)� LPDJHVLQ�WKH�EDFNZDUG�DQG�IRUZDUG�GLUHFWLRQ�UH�VSHFWLYHO\�� 7KH� ORZHU� LPDJHV� VKRZV� WKHDEVROXWH�2+�FRQFHQWUDWLRQ� LQ�1´� � �� DVGHWHUPLQHG�IURP�WKH�XSSHU�LPDJHV�

7ZR�3KRWRQ�/DVHU�,QGXFHG�)OXRUHVFHQFH�'HWHFWLRQ�RI�&2�DW�(OHYDWHG�3UHVVXUHV

�1��*HRUJLHY��.��1\KROP��DQG�0��5LFKWHU��

Preliminary investigations of the behaviour of two-photon laser-induced fluorescence signals fromcarbon monoxide (CO) at elevated pressures were performed. In addition, methods of eliminatingspectral interferences from other combustion species, especially from C2 molecules, were investi-gated. The purpose of these investigations was to study the applicability of two-photon excited LIFfor one- and two-dimensional imaging of CO molecules in real combustion environments, such asinternal combustion engines.

Carbon monoxide, like many other important combustion species has its first electronic resonancesin the vacuum ultraviolet region of the electromagnetic spectrum. Therefore, multiphoton excitationhas to be utilised for the detection of this molecule, in order to overcome the problem of both gener-ating and using vacuum UV light. We have applied laser-induced fluorescence by exciting the COmolecules from the ground state X1Σ+ to the excited state B1Σ+ using two photons at 230 nm and byobserving the fluorescence from the excited level to the lower lying A1Π level. The fluorescencewas detected in the spectral range between 451 nm (Y¶ 0, Y¶¶=0) and 662 nm (0, 5). Spectrally andspatially (in one dimension) resolved measurements were carried out in a cell, as well as, in a flameenvironment. The focal region of the laser beam focused to a line was imaged into a spectrometerequipped with a CCD camera.

In the cell measurements, the two-photon fluorescence signal from CO was studied as a function ofthe laser fluence and gas pressure. Measurements in the cell filled with different buffer gases werecarried out at pressures from 0.1 up to 10 bars. In the flame investigations, the behaviour of thespectral interferences from nonresonantly excited C2 molecules, produced by photodecompositionof fuel and fuel fragments were studied at various CH4/air–flame stoichiometries and laser intensi-ties.

Possibilities for two-dimensional imaging of CO molecules in real combustion environments bytwo-photon LIF were also studied. In these experiments CO distributions in different flames of vari-ous stoichiometries were recorded. Since the signal in two-photon excitation depends quadraticallyon the incident laser intensity, the laser beam profile should be as uniform as possible. In many in-stances this can not, however, easily be achieved. Different approaches to laser beam profile correc-tion were therefore investigated [Georgiev and Aldén, 1996].

7ZR�3KDVH�:DWHU�'HWHFWLRQ

�0��9HUVOXLV��*��-XKOLQ�DQG�g��$QGHUVVRQ�

The study of molecular species in different phases in a combustion environment is of great impor-tance in the understanding of the processes taking place. For example, two-phase analysis of boththe gas and liquid phases of fuels and other combustion species would provide information on flamestability and pollutant formation. Applications of this technique can also be found in spray diagnos-tics, to study the evaporation characteristics of liquid fuel droplets. Moreover, a two-phase diagnos-

��*UDG��VWXG��VLQFH�������

tic technique would be very useful in the characterization of mixture preparation in gas turbines.Here, the mixing of fuel and air and the complete vaporization of the fuel are essential, because ofthe risk of NOx formation near liquid droplets.

Within the department we have investigated a method for simultaneous two-dimensional visualiza-tion of water in both the gas and liquid phase. This laser-based diagnostic technique uses a combi-nation of two-photon laser-induced fluorescence (LIF) and spontaneous Raman scattering. A tun-able KrF excimer laser, operating near 248 nm, was used as the excitation source. The techniquewas demonstrated using single water droplets and their surrounding vapor phase content. The mo-lecular density in the liquid phase is much higher than in the gas phase. Consequently, the Ramansignal will almost exclusively contain contributions from the liquid phase. This technique will allowquantitative detection, since both LIF of the gas phase and Raman scattering of the liquid phase, willproduce intensities which are proportional to the molecular density of the specific phase in theprobed volume.

For two-dimensional imaging, the laser beam was focused with a cylindrical quartz lens( f mm= 300 ). The emission was collected from a region of 40 mm x 30 mm.�The scattered emis-sion from the droplet and the surrounding water vapor was collected and directed toward the CCDdetector using two adjustable mirrors.

a)

b)

)LJXUH������7ZR�GLPHQVLRQDO�OLTXLG�SKDVH�DQG�JDV�SKDVH�GHWHFWLRQ���D��)LOWHUHG�ZLWK�D�8*���ILOWHU��VKRZLQJ�WKH�WZR�GLPHQVLRQDO�VSRQWDQHRXV�5DPDQ�VFDWWHULQJ�VLJQDO�RI�WKH�OLTXLG�SKDVH��E��)LOWHUHG�ZLWK�D�:*������VKRZLQJ�ERWK�SKDVHV�LQ�IOXRUHVFHQFH�

These were set in such a way that two images of the region of interest, were imaged beside eachother on the camera. Both signals were spectrally filtered and the emission were then focused by anachromatic lens onto the photocathode of a CCD detector unit. In this way, one half of the CCD im-age shows the liquid-phase Raman emission at 267 nm and the other half, the image of the waterfluorescence distribution above 350 nm.In Fig. 3.6�(a) the liquid-phase spontaneous Raman scattering signal is shown, only the liquid-phasewater signal was observed. The emission signal through a WG 345 filter is shown in Fig 3.6(b). Inthis instance both the liquid phase and the gas-phase fluorescence components from water were de-tected. Consequently, it is possible to extract the individual gas/liquid phase distribution from thetwo images shown in Fig. 3.6.

*O\R[DO�'HWHFWLRQ

�3��(��%HQJWVVRQ��)��0DXVV�DQG�)��(��7LFK\�

Work by Frank Tichy at Norwegian University of Science and Technology, Trondheim, has led todetection of glyoxal (C2H2O2) in acetylene flames. Our part in this project was in the spectroscopicidentification and in chemical calculations of glyoxal [Tichy et. al., 1996, Tichy et al., 1997]. Thespectrally analysed fluorescence signal corresponded closely with the known spectra of glyoxal. Thelaser-induced fluorescence spectra from the flame was also found to contain the same characteristiclines as the LIF spectra from glyoxal vapour. Glyoxal was detected in a thin layer on the fuel-richside of the reaction zone in diffusion flames. Imaging of glyoxal is a promising tool for obtainingnew insight into flame phenomena.

������ $SSOLHG�5HVHDUFK

9LVXDOLVDWLRQ�RI�)XHO�'LVWULEXWLRQV�LQ�3UHPL[HG�'XFWV��LQ�D�/RZ�(PLVVLRQ�*DV�7XUELQH�&RP�EXVWRU��8VLQJ�/DVHU�7HFKQLTXHV

�&���/|IVWU|P�DQG�+��.DDOLQJ��

A two-wavelength laser technique has been developed and applied to qualitative visualisation ofmixing and vaporisation of the fuel in an industrial gas turbine combustor [Löfström et al., 1996c].The fourth harmonic of a Nd:YAG laser at 266 nm was used to visualise the gas phase distributionusing laser induced fluorescence and the second harmonic at 532 nm, was used to visualise thedroplets by Mie scattering. The experiments were undertaken using JET-A as the fuel. This multi-component mixture contains hydrocarbon components of differing spectroscopic and thermody-namic properties. It was therefore necessary to precede the practical studies by laboratory experi-ments, aimed at spectroscopic investigations of the fluorescence emission, as a function of both thetemperature and the influence of oxygen quenching on the fluorescence signal. The practical meas-

��92/92�$HUR�&RUS���7UROOKlWWDQ

urements were made on-site at VOLVO Aero Corporation, using a mobile system, where doubleimages, yielding simultaneous LIF and Mie scattering, were recorded for different operating condi-tions of two gas turbine ducts. The ducts were investigated in relation to the effect of temperature,different equivalence ratios and duct design. The results could then be used as a tool in the under-standing of the LPP (Lean, Premix, Prevaporised) concept (see )LJXUH� ���, ������������������/,)��YDSRXU�����������������������0LH��OLTXLG�)LJXUH����). Preliminary attempts were also made to investigate the accuracy and precision of thetechnique.

60 mm

20 mm

∅�56 mm

six pressure swirlatomizers

primary air secondary air

air swirlers

annularvaporazing/mixing film atomizer

vaporizing /mixing zone

air

fuel

Configuration A Configuration B

∅�30 mm

)LJXUH������7KH�WZR�GLIIHUHQW�FRQILJXUDWLRQV

/DVHU�,QGXFHG� )OXRUHVFHQFH� LQ� 6PDOO� 7ZR�6WURNH(QJLQHV

��g�� $QGHUVVRQ�� *�� -XKOLQ�� 0�� (NHQEHUJ�� DQG� %�� -R�KDQVVRQ��

In order to separate the HC-emissions, from two-strokeengines, into short-circuit losses and emissions due toincomplete combustion, laser-induced fluorescence(LIF) measurements were performed on the exhaustgases, just outside the exhaust ports of two engines ofdifferent design.

The difference between the two engines was the designof the transfer channels. One engine had "finger" trans-fer channels and the other "cup handle" transfer chan-nels. Apart from these they were identical. The enginewith "finger" transfer channels was already known toproduce more short-circuit losses, than the other. Thiswas confirmed by our measurements. Both two-dimensional imaging measurements and spectroscopicinvestigations were performed.Generally, the results show that there are two steps inthe exhaust process of these engines. Spectroscopic in-vestigations show differences between the two stages.Combustion products were found to be emitted in theearlier step, whilst in the later step fresh fuel is emitted.This is due to fresh fuel-air mixture short circuiting thecylinder during the scavenging process. In the cup-handle engine the two steps can be clearly separated,whereas for the finger-type cylinder there is an overlapbetween the two.

��+HDW�DQG�3RZHU�(QJLQHHULQJ��/XQG�,QVWLWXWH�RI�7HFKQRORJ\

����������������������������������/,)��YDSRXU�0LH��OLTXLG�

����������������������������������/,)��YD�SRXU�����������������������0LH��OLTXLG�

)LJXUH������$V�FDQ�EH�VHHQ��FRQ�ILJXUDWLRQ�$�LV�D�PXFK�EHWWHU�YD�SRULVHU�WKDQ�FRQILJXUDWLRQ�%�

The results show that LIF can be used to obtain information on the gas exchange processes in highspeed two-stroke engines. It has also been shown that the finger geometry is less efficient at retain-ing the fresh charge in the cylinder than the cup-handle geometry.

'HYHORSPHQW�RI�'LDJQRVWLF�7HFKQLTXHV�IRU���'�0DSSLQJ�RI�7HPSHUDWXUH�DQG�12�&RQFHQWUD�WLRQV

�&���.DPLQVNL�DQG�-��(QJVWU|P�

A problem with the detection of nitric oxide (NO) in practical combustion engines by single-photontechniques is related to the strong spectral interference caused by the presence of other molecularspecies which strongly absorb in the wavelength region required for the resonant excitation of NO.The aim of the project is to investigate the possibility of the use of a two-photon detection schemein the γ-bands (two-photon absorption wavelength around 450 nm), based on laser-induced fluores-cence (LIF), as an alternative to single-photon techniques. The sensitivity of the technique and itsability to map NO concentration distributions are to be investigated. The advantages and disadvan-tages of the technique are investigated under realistic conditions.

The possibility of using LIF from atoms seeded directly into the fuel of SI engines for two-dimensional quasi instantaneous maps of temperature is being studied. This technique is an alter-native approach to two-line temperature measurements based on either NO, or OH LIF being de-veloped by other groups. Potential advantages and disadvantages, in comparison with these tech-niques will be investigated and, if successful, the technique will be applied to practical SI enginesusing a variety of fuels.

0 50 100 150 200 250

Cup handle at crank angle=−77 degrees

20 40 60 80 100 120 140 160

0

20

40

60

80

)LJXUH� ����� ,PDJH� RI� /,)� VLJQDO� LQWHQVLW\� IRUWKH�FXS�KDQGOH�JHRPHWU\�GXULQJ�EORZGRZQ.

0 50 100 150 200 250

Finger at crank angle=−75 degrees

20 40 60 80 100 120 140 160

0

20

40

60

80

)LJXUH� ����� ,PDJH� RI� /,)� VLJQDO� LQWHQVLW\� IRUWKH�ILQJHU�JHRPHWU\�GXULQJ�EORZGRZQ�

3ODQDU�/DVHU�,QGXFHG�)OXRUHVFHQFH�RI�+�2�WR�6WXG\�WKH�,QIOXHQFH�RI�5HVLGXDO�*DVHV�RQ�&\�FOH�WR�&\FOH�9DULDWLRQV�LQ�6,�(QJLQHV

�*��-XKOLQ��0��9HUVOXLV��+��1HLM��DQG�%��-RKDQVVRQ��

Two-photon laser-induced fluorescence (LIF), with a tuneable KrF excimer laser operating at 248nm, has been used to visualise water (H2O) vapour prior to ignition, close to the spark plug, in ahigh pressure combustion cell. Two-photon absorption of water, laser sheet focusing and two-dimensional fluorescence imaging conditions were studied. The water vapour content was corre-lated with the combustion rate of the next cycle, which was determined from the measured pressuredevelopment during combustion.

��� ���� ���� ����

�����

�����

�����

�����

�����

7LPH�WR�UHDFK�FHUDWLQ�SUHVVXUH�OHYHO��PV�

:DWHU�VLJQDO��D�X��

)LJXUH������$�VFDWWHU�SORW�VKRZLQJ�WKH�FRUUHODWLRQ�EHWZHHQ�WKH�ZDWHU�FRQFHQWUDWLRQ�SULRU�WRLJQLWLRQ�DQG�WKH�UDWH�RI�FRPEXVWLRQ�GXULQJ�WKH�QH[W�F\FOH�

As can be seen from the scatter plot in )LJXUH���� the combustion rate is strongly dependent on thewater vapour content close to the spark plug prior to ignition in the cell experiments. The influenceof residual exhaust gases on the cycle-to-cycle variations in spark-ignition engine combustion wasalso investigated in a running internal combustion (IC) engine. The IC engine measurements showthat with a low inlet manifold pressure, the correlation between the residual gas fraction and thecombustion rate was as high as 0.6.

��3UHVHQW�SRVLWLRQ��$ZDSDWHQW�$%��0DOP|��+HDW�DQG�3RZHU�(QJLQHHULQJ��/XQG�,QVWLWXWH�RI�7HFKQRORJ\

���� 6RRW�&KDUDFWHULVDWLRQ�DQG�9LVXDOLVDWLRQ�3��(��%HQJWVVRQ�DQG�*��-XKOLQ�

The development of efficient engines producing low concentrations of pollutants is very importantfor the future. The Diesel engine is very fuel efficient, but the production of soot is an inherentproblem with this type of engine, when using ordinary Diesel fuels. In order to understand the com-bustion processes in a Diesel engine and for further development of Diesel technology, LQ�VLWXmeasurements of soot in the chamber of Diesel engines are required. Different laser-based tech-niques have been studied for soot characterisation in combustion environments.

The detection of both the background emission and elastic scattering have drawbacks in soot detec-tion. Emission is a line-of-sight technique of poor spatial resolution in the direction along the detec-tion path, whilst elastic scattering from soot is strong, it is difficult to evaluate in terms of sootproperties. This is because the scattering is strongly dependent on the particle size which can varyover several orders of magnitude, the dynamic range of the detector is therefore of critical impor-tance for the measurements.

There are two techniques which have in recent years received increased interest for soot characteri-sation. These are based on the heating of particles by absorption of incident laser radiation. Thetechnique of laser-induced incandescence (LII) is now widely used among researchers using laser-based soot analysis. In LII, soot particles are heated to temperatures of 4000 to 5000 K, at thesetemperatures they strongly radiate according to Planck’s law. A particle with a higher temperaturewill radiate more strongly at shorter wavelengths according to Wien’s displacement law. By detect-ing wavelengths close to 400 nm, an efficient reduction of the background emission is achieved andthe soot can then be detected. The induced incandescence signal has been shown to be approxi-mately proportional to the volume fraction of soot in the probe region. The second technique isbased on heating to high enough temperature subsequently to vaporise the particles. In this processcarbon fragments such as C2 radicals are formed. By tuning the laser to the (v’=0) d 3Πg ← (v”=1) a3Πu transition in C2 (the Swan bands), the same laser can be used not only to vaporise the soot parti-cles, but also to excite the C2 radicals. The fluorescence signal in the (v’=0) → (v”=0) band close to516 nm has been shown to be approximately proportional to the volume fraction of soot. The tech-nique has been given the acronym LIF(C2)LVS (laser-induced fluorescence in C2 from laser-vaporised soot).

The techniques LII and LIF(C2)LVS were compared in fundamental studies in atmospheric pressurelaboratory flames [Bengtsson and Aldén, 1995a]. The LIF(C2)LVS signal was of higher spectral in-tensity, which could be useful if the background flame luminosity causes severe interference. In LII,the detected signal can be time-delayed, until all the other laser-induced signals have decayed,which could be of value if PAH-fluorescence is a major problem. In this work one-dimensional im-aging measurements were demonstrated.

In the following work two-dimensional imaging using LIF(C2)LVS was carried out [Bengtsson,1996c]. Simultaneous single-shot 2-D images of LII, elastic scattering, and LIF(C2)LVS were re-corded using the same CCD in combination with a Cassegrainian telescope. This telescope was alsoused to simultaneously record two-dimensional images of LIF(C2)LVS and laser-induced fluores-cence of OH. Excitation of OH was in the band head (v’=1) A 2Σ+ ← (v”=0) X 2Π, in the regionfrom 281.1 nm to 281.8 nm, and the fluorescence was detected in the (v’=0) → (v”=0) band close to

310 nm. This simultaneous detection of the soot regions and OH (probing the reaction zone and thehot post-combustion gases), is of great interest for practical combustion studies.

In the first experimental study in Diesel engines, we investigated the possibility of using the back-ground flame emission, elastic scattering, LIF(C2)LVS, and LII for two-dimensional soot visualisa-tion in an engine. At these high pressures the LII signal was stronger in comparison to theLIF(C2)LVS signal. It can also be regarded as a much more promising technique in high-pressureapplications. The results have been summarised in [Juhlin, 1996].

���� &RKHUHQW�/DVHU�7HFKQLTXHV

������ &RKHUHQW�$QWL�6WRNHV�5DPDQ�6SHFWURVFRS\��&$56�

�3��(��%HQJWVVRQ��-��%RRG�DQG�/��0DUWLQVVRQ��

Coherent anti-Stokes Raman spectroscopy (CARS) is now a widely used method for non-intrusivemeasurements of gas-phase temperatures in practical combustion situations. The method yieldspoint measurements of high spatial and temporal resolution. In CARS, a signal is generated in themedium through the third-order interaction, when the frequency difference between two photons(normally from two different lasers) is in resonance, with a rotational, or a rotational-vibrationalresonance in a molecule. When rotational-vibrational resonances are probed, the technique is calledvibrational CARS, which is the most common CARS technique. Vibrational CARS thermometry ofnitrogen is nowadays a standard tool in combustion research. In rotational CARS the rotationalresonances are used for the spectroscopy. The dual broadband approach to rotational CARS, devel-oped in 1986, is mainly used for nitrogen thermometry in fundamental flame studies, as well asmore recently in practical situations [Bengtsson et al., 1995b,c].

The possibility of using dual broadband rotational CARS as a technique for concentration measure-ments has been investigated. The accuracy and the precision of the technique for oxygen concentra-tion measurements were studied at atmospheric pressure and at temperatures between 290 and 1410K [Martinsson et al., 1996]. Oxygen concentrations are LQ�VLWX calibrated to nitrogen, and can bemeasured very accurately when the concentrations are close to the oxygen concentration in air.However, at concentrations of a few percent the technique is inaccurate. The use of weightingmethods is therefore required to improve the accuracy. The possibility of using oxygen spectra fortemperature measurements was also investigated [Martinsson et al., 1996]. An example of a rota-tional CARS spectrum from a mixture of 20% oxygen and 80% nitrogen is shown in )LJXUH����.The evaluated temperature was 1508 K. Below each spectrum the difference between the experi-mental spectrum and the best-fit theoretical spectrum is shown.

Rotational CARS is in most cases incapable of detecting the fuel. However, very strong vibrationalCARS signals are often obtained from the fuel C-H stretching mode in hydrocarbon fuels. By com-bining the rotational CARS due to nitrogen and oxygen, with vibrational CARS of the fuel, there isa potential for measurement of the fuel-air ratio. This has been demonstrated in both gases andflames [Bengtsson et al., 1995b,c]. In this study, the beams were aligned in a double-foldedBOXCARS phase-matching geometry. The rotational CARS beam and the vibrational CARS beamwere superimposed and directed onto the spectrograph. Inside the spectrograph the two paths were

aligned, so that the rotational and vibrational CARS signals could be detected by the same diode-array. The technique is promising, but suffers from increased experimental complexity, which is adrawback in practical applications.

)LJXUH� ����� $Q� H[�SHULPHQWDO� URWDWLRQDO&$56� VSHFWUXP� IURPD� PL[WXUH� RI� ���R[\JHQ� DQG� ���� QL�WURJHQ� ZLWK� DQHYDOXDWHG� WHPSHUD�WXUH� RI� ����� .�� %H�ORZ� WKH� VSHFWUXP� WKHGLIIHUHQFH� EHWZHHQWKH� H[SHULPHQWDOVSHFWUXP� DQG� WKHEHVW±ILW� VSHFWUXP� LVVKRZQ�

In a study of engine knock, the dual-broadband rotational CARS technique has been applied to asingle-cylinder spark-ignition engine [Bengtsson et al., 1996a, Bengtsson et al., 1996b, Bood et al.,1996, Bood et al., 1997]. The measurements were performed at the Department of Thermo andFluid Dynamics, at Chalmers University of Technology in Gothenburg. In this study, cycle-resolvedtemperature measurements were undertaken prior to ignition, at different distances from the wall (inthe end-gas region), at different crank angle positions and also at different knock intensities. Prob-lems were experienced with stray light interference from the 532 nm laser beam. To reduce this anotch filter, suppressing the intensity at the wavelength 532 nm, was used, but there was a reduc-tion in experimental accuracy. The measured temperatures were used as input data for chemical ki-netic calculations, and also for comparison with chemical kinetic calculations presented in this sec-tion of the report.

������ ,QYHVWLJDWLRQV� RI�7ZR�3KRWRQ� ,QGXFHG�3RODULVDWLRQ� 6SHFWURVFRS\� �7,36�IRU�&RPEXVWLRQ�'LDJQRVWLFV

�&��.DPLQVNL��%��/|IVWHGW��5��)ULW]RQ�DQG�7��'UHLHU�

Since its first demonstration, single-photon polarisation spectroscopy (PS) has been established as apowerful tool in high resolution atomic and molecular spectroscopy. More recently, PS has gainedattention as a diagnostic technique for the probing of combustion and flame environments. In thisfield the coherent nature of the process and the high signal to background ratio that can be achievedcan provide distinctive advantages over other laser-based methods.

For most atomic and many molecular species, the first accessible energy levels lie in the vacuum-ultraviolet spectral region, which cannot be probed in single-photon steps with the majority of avail-

able light sources. The problem is exacerbated by strongly absorbing surrounding gases in thiswavelength region and in most combustion environments, such schemes are therefore not feasible.Danzmann et al. first introduced Doppler-free two-photon induced polarisation spectroscopy (TIPS)for the detection of H atoms in a dense plasma discharge. This technique overcomes these problems.Furthermore, in contrast to single-photon PS, the TIPS signal is created in the time interval whenthe two-photon absorption takes place which is instantaneous on the time scale of molecular colli-sions. It has therefore been suggested that neither ionising, velocity changing, nor populationchanging collisions affect the TIPS signal. In many combustion environments, where molecularquenching is severe, this characteristic could make TIPS an attractive alternative or a complemen-tary tool to other laser-based techniques.

283.0 283.2 283.4 283.6

0

5

10Inte

nsity

(ar

b. u

.)

Wavelength (nm)

LIF

283.0 283.2 283.4 283.6

0

5

10TIPS

)LJXUH�������8SSHU�JUDSK��7,36�VSHFWUXP�LQ�WKH� ( ) ( )a X 1 11 0Π ΣJ JY Y′ = ← ′′ =+ �V\VWHP�RI

1���/RZHU�JUDSK��������SKRWRQ�UHVRQDQW�/,)�VSHFWUXP�FRUUHVSRQGLQJ�WR�WKH�VSHFWUXP�VKRZQDERYH��%RWK�VSHFWUD�ZHUH�UHFRUGHG�VLPXOWDQHRXVO\�LQ�D�IORZ�RI�SXUH�1��DW�URRP�WHPSHUDWXUHDQG�DWPRVSKHULF�SUHVVXUH�

Present research has resulted in the first demonstration of two-photon induced polarisation spectros-copy (TIPS) detecting molecular nitrogen in atmospheric pressure flow mixtures. Laser-inducedfluorescence (LIF) from N2

+ which was created via a (2+3)-photon resonance from N2, was moni-tored simultaneously with TIPS to allow a quantitative comparison between the two techniques indifferent operating regimes. )LJXUH����� shows an example of such scans obtained in a flow of at-mospheric pressure N2 at room temperature.

Power dependence checks for both techniques were performed and the effect of collisional partnerswere investigated in different gas flow mixtures.

The results suggest that TIPS is not so severely affected by molecular quenching collisions as is LIF,[Kaminski et al., 1996a]. The high signal levels obtained prove that the technique is sensitive andmay have potential as diagnostic tool for combustive and non-combustive flows. Current effort fo-

cuses on the development of a theoretical basis for the process, which will allow the prediction andmodelling of signal lineshapes under realistic conditions. This work is undertaken in collaborationwith groups at Department of Physical Chemistry, University of Heidelberg, Germany.

��� 6SDUN�,JQLWLRQ�DQG�,RQ�&XUUHQW�6HQVRUV�$��6DLW]NRII��5��5HLQPDQQ�DQG�0��$NUDP�

We are currently working on two different applications of ionized gases. The first one concentrateson the energy deposition into a (combustible) gas from an electrically generated spark discharge.The second project concerns the use of Ion Currents for detection of the combustion process, usingthe spark plug as the sensor. The main objective of these projects is to develop models which can beuseful for the industry. As a part of this research, substantial experimental work and software devel-opment has been carried out.

In addition to these projects, two separate models of the spark discharge have, partly for historicalreasons, been developed. The models developed for the two above mentioned projects are requiredto be simple and foremost, fast. The two separate models are, however, quite complex. The first ofthese complex models calculates the breakdown of the spark whilst the second model calculates thebehavior of the plasma and the surrounding gas directly after the breakdown.

These different models will be briefly described in the sections below. They are presented in a com-bustion related order starting with the spark breakdown and ending with the actual combustion pro-cess.

���� %UHDNGRZQ�0RGHO

The ignition of a combustible gas is highly dependent on the properties of the spark. It is wellknown that the characteristics of the spark that ignites the gas influences the flame. It is thereforeimportant to understand the spark ignition process and its relation to the ignition system circuit, aswell as, the gas medium in which the spark is discharged.

The physics of a spark is a complex interaction between electromagnetic, thermodynamic and mo-lecular interaction processes. A spark is divided into several different phases. First, there is thebreakdown phase, in which the gas is ionized and a plasma channel is established, this results in avery high current and temperature. This is followed by a transient phase, where the channel expandsto the pressure of the surrounding gas, and finally, there might be an arc and/or a glow phase. Thismodel is a comprehensive study of the initial phase of a spark, the breakdown.

By applying a strong electric field over an electrode gap, an electron avalanche process will be initi-ated, and will eventually lead to a breakdown in the gas. The time ranges of the breakdown, as wellas the values of other parameters, depend on the external circuit. There are two main circuits (igni-tion systems) available, the rather slow inductive one, and the fast capacitative system. The systemwhich has been used here are the high energy capacitative circuit.

The model is based on the time-dependent Boltzmann transport equation, Maxwell’s equations andKirchoff’s equations for a circuit. It calculates the energy distribution of the electrons in the sparkgap during the pre-breakdown processes down to the sub-nanosecond time scale. The first model

has one energy dimension, but is only zero-dimensional in space. In the second model a one-dimensional space resolution has been introduced.

From the 31 approximation of the Boltzmann transport equation, transforming it into energy spaceand discretesizing it, a series of coupled differential equations are obtained. The ignition system cir-cuit is described by another set of differential equation based on Kirchoff’s equations. Using Max-well’s equations, a description of the electric field can be formulated, and a relationship between thecircuit and the gas obtained. This is a discrete model that includes most kinds of interactions, suchas ionization (of course), excitation, recombination (absorption), as well as, elastic and super-elasticcollisions. Collisions between electrons and other electrons are also included. The model basicallyworks by keeping track of the kinetic energy of the electrons, and also the electron density at a spe-cific energy. In addition to the electron densities, the electric field, the current and the voltage, anumber of other properties are also calculated. These include, the energy loss from elastic and ine-lastic collisions, the diffusion and ionization rate, the mobility and mean energy, to name a few. Acomputer program (IGNIS) has been written for the numerical solution of the model. Measurementsof the current and voltage have provided valuable information for the validation of the theoreticalcalculations.

The theoretical results show an interesting interplay between the electric circuit and the electrons inthe gas. Sub-nanosecond processes can be seen, such as fast oscillations in the current and the volt-age. In the picosecond region, the energy spectrum of the electrons is totally determined by thescattering processes and their cross sections. The system is in a far-from-equilibrium state, see ����������������������������������������E�)LJXUH���� �D�. After a couple of nanoseconds the structure of the spectrum is smoothed out, but thescattering process can still be seen. It is not until a very high electron density is reached, that thespectrum begins to resemble a Maxwellian distribution, see ����������������������������������������E�)LJXUH���� �E�. Initially the electron density changes very little, but after a few nanoseconds, thedensity begins to increase at a faster-than-exponential rate and the electron avalanche is initiated.The phase difference between the electric field and the electron density can easily be observed. Atsome point, determined by the electron density, the actual breakdown occurs, the voltage dropsdown to almost zero and the current increases to very high values. The actual values and time rangesdepends on which electric circuit is used.

��������������������������������������E�

)LJXUH������D��(OHFWURQ�GHQVLW\�DV�IXQFWLRQ�RI�WLPH�DQG�HQHUJ\�LQ�SUH�EUHDNGRZQ���E��(OHFWURQGHQVLW\�DV�IXQFWLRQ�RI�WLPH�DQG�HQHUJ\�IRU�WKH�ZKROH�EUHDNGRZQ�SHULRG�

���� 3ODVPD�0RGHO

The ignition and self-sustained propagation of a flame in a combustible gas depends upon manyphysical and chemical properties. The products (exhausts) in a combustion system not only dependon the mixture characteristics (e.g., lean or rich), but are also related to the ignition process. Leanmixtures are difficult to ignite, but once the mixture has ignited it produces less unwanted exhaustgases, such as NOX, due to a comparatively low combustion temperature. To maintain a reasonablecontrol over the ignition, as well as, over the emissions of a combustion system, it is therefore im-portant to know the reasons why lean mixtures are difficult to ignite.

This work was initiated to understand the spark to flame transition and also to answer the abovequestion. To simplify the investigation, the physical and chemical processes involved were studiedseparately. The physical processes were studied by modeling the spark behavior in chemically inertgases. One and two-dimensional models in a cylindrical coordinate systems were developed for this.Air and nitrogen were selected as the gases, because these are an important part of the real ignitablemixtures and because no significant contributing chemical reactions occur during the evolution ofthe spark. The injection of electrical energy from the ignition circuit to the gas and its re-distributionboth in space and time is successfully modeled. The flow field generated during the evolution of aspark has also been studied at length. Experimental studies have been carried out in order to validatethe predictions of the models. [Akram, 1996, Reinmann, 1996]

�D�

)LJXUH� ����� 6SHFLHV� PDVVIUDFWLRQV� DV� D� IXQFWLRQ� RIVSDUN� UDGLXV� �PP��� DORQJWKH�D[LV�RI� WKH�VSDUN�DW����QV���L��12���LL��12����LLL��2��LY��1�

These models have been proven to provide realistic initial conditions for combustion and are there-fore suitable for further studies of ignition processes. The two-dimensional model has been ex-tended by including chemical kinetics to a multi-component chemically reacting system. The earlyNO formation in the plasma channel, from oxygen and nitrogen by the Zeldowich mechanism, canbe studied together with dissociation and ionization processes. The predictions obtained from thismodel are very encouraging, as can be seen in )LJXUH����, where the concentrations of NO and NO+

are shown together with N and O. The maximum NO concentration seems to exist at a temperatureof about 4000 K and is controlled by a balance between thermal formation and dissociation. [Ak-ram, 1996]

���� 6SDUN�(QHUJ\�'HSRVLWLRQ�3URMHFW

This work has been undertaken within the IDEA-EFFECT project. We have concentrated our workwithin this project on measuring and modeling the energy deposition into the gas between the elec-trodes of a spark discharge. The quality of the combustion is dependent on the total amount of de-posited spark energy. Normally, the energies supplied by ignition systems used in present produc-tion engines, are well above the minimum ignition energy. However, with the introduction of leanburn engines this changes. The lean burn engine concept has many advantages, such as a higherthermal efficiency and less pollution emissions. The energy delivered by conventional ignition sys-tems is, however, far too low to cover the full range of requirements of a lean-burn operating en-gine. These aspects of the modeling of the energy deposition are very interesting to the car manu-facturers (Volvo, Volkswagen, Fiat, Renault, Peugeot and Rover). The resulting model will be anintegrated part of the SPEED/STAR CD multi-purpose thermofluids analysis code.

This model is based on two contributing terms, referred to as the transient-model and the glow-model. The energy contribution from the transient phase is itself divided into two parts, the first partis the breakdown energy and then follows a short relaxation period where very little energy is de-

posited. The breakdown energy originates from the stray capacitances in the close vicinity of thespark plug. The capacitances are charged to the breakdown voltage and at breakdown they will de-posit their electrical energy into the gas with very high efficiency (nearly 100%). At this time, thereexists a small cylindrical, conducting channel. In the following quasi-stationary phase, which ismodeled in the glow-model, the current initially is limited by the resistances of the cables, but at theend the spark the current is limited by the spark itself. In this phase the energy input is due to jouleheating. The Joule heating energy originates from the current that carries the electrical chargesthrough the gas. These then collide with surrounding gas molecules and thereby transfer energy tothe gas. Higher energy deposition in the gas increases the conduction of the channel and thereby theefficiency of the energy conversion is decreased. This heating is balanced by the diffusion processwhich tends to lower the temperature in the central part of the spark channel. When the temperaturehas reached a low enough value and the current is low, the spark will disappear, since the spark gapresistance rapidly increases under these conditions.

The model for the energy deposition has been found to be consistent with experimental result. Themeasurements have been carried out in a small calorimetric bomb, in which the pressure increaseoriginating from the spark can be measured. The pressure increase is proportional to the energy de-posited into the gas if the pressure is measured at a time when all pressure transients have disap-peared and before the energy has dissipated into the surrounding walls. The response of the modelto different initial conditions, such as changes in the voltage pulse, electrode distance, or the sur-rounding gas and the pressure, is in agreement with the results from the measurements. A compari-son between the experimental and the calculated spark energies are presented for a modern coil ig-nition system at various pressures in )LJXUH����. [Reinmann, 1995 and 1996]

We have used commercial ignition systems for the final evaluations, but whilst completing the vari-ous parts of the model, special ignition systems have been constructed within the department. Thismade it possible to check the validity of each part of the model. The results were very promisingsince there are no fitted parameters in the model.

)LJXUH� ����� 7KH� PHDVXUHGDQG� FDOFXODWHG� VSDUN� HQHU�JLHV� DV� D� IXQFWLRQ� RI� SUHV�VXUH�� 7KH� LJQLWLRQ� V\VWHPXVHG�ZDV�D�PRGHUQ�FRLO�LJQL�WLRQ�V\VWHP�VXSSOLHG�E\�)LDW�7KLV� V\VWHP� VHUYHG� DV� WKH� IL�QDO�WHVW�RI�WKH�PRGHO�

0 200 400 600 800 1000 1200 1400 1600 1800

3UHVVXUH��N3D�

�

�

�

�

�

��

��

��

��

Ene

rgy

(mJ)

Spark EnergyStandard deviationModel

Mean Spark Energies vs Pressure(5 ms dwell time, propane + nitrogen)

���� ,RQ�6HQVRU

Electronic engine management has been conducted in various ways during the last two decades, aselectronic controls have steadily increased. The California On-Board-Diagnostic (OBD-2) require-ments for 100 % misfire detection, in order to maintain a high performance rate of the catalyst, hasled to a more intense research in this area. The implementation of Crankshaft Velocity Fluctuation(CVF) has been adopted by several automobile manufacturers. However, CVF measurements duringrough road driving are difficult, since these conditions generate noisy signals. Complementary tothis technique, the use of in-cylinder measurements of the combustion, has been shown to be apromising approach.

Measurements of what is going on inside an internal combustion engine is a challenging task. Theuse of advanced laser techniques has been successful in many instances, but is not very practical formonitoring the inside of an engine during the regular use of motorized vehicles. Utilizing somethingalready present in the combustion chamber would be of great advantage, not only to minimize thetotal cost, but also the number of sensing probes which could disturb the combustion process.

By applying a small voltage across the gap of a spark plug after ignition, a small current is generatedand measured. An ionization sensing technique based on this has been developed. This is possiblebecause there is a relation between the state of the combustion gases and the current that can bedriven through the spark gap. It is the spark plug itself that acts as the sensor. The current is carriedby the charges produced in the high temperature conditions during and after the combustion. Theionization ratio of the species present in the post-flame region can be calculated as a thermal ioniza-tion process according to Saha’s equation. The physical background for this model can be found inthermodynamics, electrodynamics, statistical physics and atomic (molecular) physics. It is knownthat a rather minor species, namely nitric oxide (NO), is of major importance for the charge produc-tion in the post-flame region. This is due to the relatively low ionization energy of NO compared tothe other major and minor species. In the reaction zone the process responsible for the charge pro-duction is different. Here it is chemi-ionization that acts as a major contributor and the importantspecies are: H3O, C3H3, CHO and CH3.

When the current through the spark gap is investigated, three distinct phases can be identified: first alarge current that generates a spark and ignites the mixture. The spark current in the figure has beenfiltered out and only a remnant can be seen. Immediately after the spark, the first current peak oc-curs. It is related to the combustion reaction process in the flame kernel present in the vicinity of thespark gap at this time. Finally, there is a second ion current peak that is related to the thermody-namical condition of the post-flame (burned) gas in the cylinder. See )LJXUH����.

)LJXUH������$�PHDVXUHG�LRQ�FXU�UHQW�� 7KLV� LV� DQ� DYHUDJH� RI� ��F\FOHV� DW� :27� �:LGH� 2SHQ7KURWWOH�� DQG� ����� USP�� 7KHPHDVXUHPHQWV� ZHUH� GRQH� LQ� D)OLQW������9��HQJLQH��7KH�VSDUNFXUUHQW� KDV� EHHQ� ILOWHUHG� RXW�ZKDW�LV�VHHQ�LQ�WKH�ILJXUH�DW����FUDQN�DQJOHV�LV�MXVW�WKH�UHPDLQV�

This so called “Ion Current” has been used for some time by several research and developmentgroups, car manufacturers and universities, to detect, for example, knock and misfire. However,only SAAB uses it today in production cars. The work that has been undertaken here, was to de-velop a model that accurately predicts the pressure in the cylinder and theoretically relates the cur-rent to the chemical and thermodynamic states of the gas in the cylinder. The ion current contains alot of information about the chemical, thermodynamical and mechanical processes in the cylinder. Ithas for example been shown, with the help of neural networks, that it is possible to very accuratelypredict the air/fuel ratio. The crank angle at peak pressure is another factor that can be determinedwith the help of the ion current. The ongoing work is targeted towards developing a few models thatare able to determine the pressure and the air/fuel ratio with the help of the ion current. There areother highly interesting parameters that can be obtained from the ion current, like the local gas tem-perature and the shape of the mass fraction burned curve for instance. These models are also re-quired to be fairly simple. The reason for this is that the calculations have to be completed from onecycle of the engine to the next, if they are to be of any use. The use is of course to make a correctionin the engine performance (ignition timing, valve timing, fuel mixture etc.), so that the next cycle isas optimized as possible.

The work that has been published to date, consists of a model for the ionization ratio in a gas basedon statistical physics and an analysis of the thermodynamic and chemical conditions present in acombustion chamber. From these conditions the current can be calculated. The method is basicallythis: First, we calculate the temperature from both the combustion and the adiabatic compression ofthe gas. Second, the ionization ratio is calculated from the temperature which gives the current. Fora more detailed description we refer to the reference at the end of this paragraph. In order for this tobe useful, the equations are reversed and the current is used to calculate the temperature, and fromthe temperature the pressure is obtained. Since the current reflects the local conditions around thespark plug and what is required is the global pressure, the mass fraction burned ratio also has to beconsidered. [Saitzkoff, 1995 and 1996]

-100 -50 0 50 100 150

&UDQN�$QJOH��GHJ�

-20

0

20

40

60

80

100

120

140&XUUHQ

W��µ$�

6SDUN

&XUUHQW

)ODPHIURQW

=RQH

3RVWIODPH

=RQH

)LJXUH������$�FRUUHODWLRQ�EHWZHHQ�SUH�GLFWHG�DQG�PHDVXUHG�SHDN�SUHVVXUH� IRUD�IHZ�GLIIHUHQW�GULYLQJ�FRQGLWLRQV

Introducing chemistry into the models makes it possible to incorporate the effect of the presence ofthe other species, in addition to NO. Calculations reveal that the additional species contribute about5 to 7 percent of the current in the post-flame zone. Knowing the concentrations of all present spe-cies also makes it possible to estimate the effect of the electronegative species. Calculations showthat about 85 percent of the free electrons that are created, form negative ions. However, even ifonly a minority of the charged particles are electrons, they are still responsible for the major part ofthe current (more than 95%), due to their much smaller mass and hence considerably greater mobil-ity. Taking into account all possible effects, the pressure peak level can be predicted with a verygood accuracy. The correlation is above 95% when using averaged cases (500 cycles) in predictingdifferent driving conditions, see )LJXUH���� [Saitzkoff, 1997].

)LJXUH� ����� 7KH� ULVLQJ� HGJH� GHULYD�WLYH�RI�WKH�ILUVW�LRQ�FXUUHQW�SHDN�DV�DIXQFWLRQ�RI�λ�YDOXH�

It has been observed that the first ion current peak has a strong correlation with the air/fuel ratio.Parameters such as the peak level of the current and the delay of its occurrence after ignition are re-lated to the air/fuel ratio and can therefore be used to predict this ratio. The maximum value of thefirst peak is actually a function of the CH and O concentrations in the reaction zone of the flame.These concentrations show a clear dependence on the air/fuel ratio. The time delay between ignitionand the position for the fist peak maximum is connected with the ignition delay of the gas mixture.This parameter is also dependent on the air/fuel ratio. By dividing the peak value and the “ignitiondelay” an even stronger dependence is obtained. This dependence can be seen in )LJXUH����. A theo-

0 1 2 3 4 5

0HDVXUHG�3UHVVXUH��03D�

0

1

2

3

4

5

&DOFX

ODWHG�3UHVV

XUH��03D�

0

1

2

3

4

5

6

7

8

9

0.9 0.95 1 1.05 1.1

Lambda

Firs

t Pea

k D

eriv

ativ

e (u

A/C

AD

)

io13

io20

io22

io24

io25

io26

io40

ion13

ion20

ion22

ion25

ion40

retical analysis of the thermodynamic, electrodynamic and the chemical conditions is able to explainand recreate the measured current quite accurately [Reinmann, 1997].

This project has been undertaken in co-operation with Mecel, a subsidiary to Delco Electronics andGeneral Motors.

��� 6LPXODWLRQ�RI�7XUEXOHQW�&RPEXVWLRQ�6��,��0|OOHU��&��)XUHE\�DQG�(��/XQGJUHQ��

Simulation of combustion processes has become an important tool for increased understanding offundamental processes in combustion. Simulation of combustion systems can also be used in thedesign process, to reduce the number of necessary laboratory tests, when new combustion devicesare developed. Theoretical studies of combustion and development of appropriate simulation mod-els have therefore gained increased attention. The efforts described here originate from a projectconcerning Pulsating Combustion, initiated in 1989 at the Lund Institute of Technology. This col-laborative project is carried out at Division of Combustion Physics, Division of Mechanics and De-partment of Heat and Power Engineering, where pulsating combustion is being studied from boththeoretical and experimental points of view. The aim of the theoretical part of the project has beento derive a physical model and develop a simulation model.

During the progress of the project, the field of activities within the theoretical part of the project, hasbeen enlarged to encompass chemically reacting flows in general, with special emphasis on turbu-lent combustion. Pulsating combustion is still one of the applications but other test rigs includingtwo different bluff body stabilised flames, the ENSMA rig and the Validation Rig, have also beenused to verify the applicability of the model [Fureby et al., 1995a, Fureby and Möller, 1995a, Mölleret al., 1996]. The second rig is a model of an after burner for a jet engine, developed and operatedby Volvo Aero, Trollhättan, Sweden. Results from simulations and experimental measurements car-ried out on these rigs are also being used to investigate fundamental mechanisms occurring in com-bustion, e.g. flame stabilisation and the production of small scale fluctuations.

In turbulent combustion the combustion process can be characterised by the complicated interactionbetween different processes, such as fluid dynamic mixing and the chemical reactions. The theoreti-cal model, as well as, the simulation model must be derived with great care, in order to correctlytreat the non-linear, unsteady behaviour of the physical and chemical processes involved.

The theoretical model used here is derived utilising the methods and principles of the theory ofmixtures embedded in modern continuum mechanics. This model is a general model for chemicallyreacting flows and may encompass non-linear material properties, hierarchical models for speciesdiffusion and multiphase flows. A simplified version of this model has been cast into a numericalsimulation model, which is currently under development in order to include as much physics andchemistry as found relevant.

The processes in turbulent combustion occur over a wide range of spatial and temporal scales. Thegoverning equations in the theoretical model are the same for any flow, turbulent or not. The prob-lem is the numerical implementation of these equations, which are transformed from a set of con-tinuous equations to a set of discretised equations. The computer capacity available today does notallow for Direct Numerical Simulation (DNS), where all spatial and temporal scales are resolved,for simulation of combustion processes in practical engineering applications. The general approachof circumventing this problem is based on various kinds of averaging e.g. Reynolds AveragingSimulation (RAS), where the governing equations are averaged in time and the effects of turbulent

��'LYLVLRQ�RI�0HFKDQLFV��/XQG�,QVWLWXWH�RI�7HFKQRORJ\

fluctuations on the mean flow are modelled. A different more general approach is Large EddySimulation (LES), where the governing equations are filtered spatially. The large scale eddies areresolved on the computational grid, while the effect from the Sub Grid Scales (SGS) is modelled bySGS models.

In this project a numerical simulation model for chemically reacting flows based on the LES tech-nique has been developed. The chemical reactions are described by various reduced mechanisms, upto four step mechanisms, for combustion of hydrocarbons in air. The simulation model has beenimplemented in the commercially available code Phoenics, Concentration Heat and Momentum,CHAM, Ltd. (London), which here has been used as a numerical solver.

t=0.125 (measured) t=0.375 t=0.625 t=0.875

(simulated) HR

Nox

)LJXUH������0HDVXUHG��WRS��DQG�FDOFXODWHG��ERWWRP��KHDW�UHOHDVH�DQG�YHORFLW\�ILHOGV�LQ�WKH�IODWFRPEXVWLRQ�FKDPEHU�DW�IRXU�LQVWDQWV�RI�WKH�RSHUDWLQJ�F\FOH��$OVR�LQFOXGHG�DUH�WKH�FDOFXODWHG12;� � IRUPDWLRQ�LQ�WKH�FRPEXVWLRQ�FKDPEHU��2SHUDWLQJ�FRQGLWLRQV��3RZHU� LQSXW� ������N:�H[FHVV�DLU� �����DQG�IUHTXHQF\� ����+]��5HIHUHQFH�YHORFLW\�YHFWRUV�UHSUHVHQW���P�V�>/LQGKROPDQG�0|OOHU������@�

In order to understand some of the main features of the complicated physics taking place in chemi-cally reacting flows and to evaluate the predictive capabilities of the LES model, it is important to

apply the simulation model to different well known and well documented combustion systems. Onesimulation object is the Validation Rig, Volvo Aero Corporation, Trollhättan, Sweden, which hasbeen chosen because a large amount of experimental data, such as LDV measured velocities andvelocity fluctuations, CARS measured temperatures and gas analysis already exists. Various SGSmodels have been tested [Fureby, 1995a, Fureby and Möller, 1995a] and also different models forthe filtered reaction rates have been investigated at various operating conditions [Möller et al.,1996]. The ENSMA Rig has also been simulated [Fureby et al., 1995a]. The LES model seems to becapable of describing the overall behaviour, as well as such complicated behaviour as flame frontdynamics. The model also works well for different rigs at various operating conditions

Furthermore, simulations of the experimental pulse combustors at Heat and Power Engineering hasbeen carried out. Simulations have been compared with LDV measured velocity cycles. The com-puted spatially and temporally resolved heat release has been compared with spontaneous emissionfrom the CH radical [Lindholm and Möller, 1997]. The model is capable of predicting the overallbehaviour of the pulse combustor, the calculated heat release is also in good agreement, both spa-tially and temporally, with spontaneous emission from the CH-radical.

��� &KHPLFDO�.LQHWLFV�)��0DXVV��0��%DOWKDVDU�DQG�'��1LOVVRQ�

The conversion of fossil fuels into energy is the most important chemical process, used by industryand private households in stationary and mobile combustion devices. Detailed knowledge of allcombustion phenomena occurring in engines, ovens and furnaces is of great technical importance.The limited resources of fossil fuels and the undesirable formation of air pollutants demand efficientdevices which control all chemical processes. The complex interactions of not only the huge chemi-cal kinetic system describing combustion chemistry itself, but also the involvement of fluid dy-namics further complicates this control. Such a chemical system involves about 1000 chemical re-actions with 100 chemical species. Great effort has been made to understand the basic chemicalmechanism in flames. Global parameters, such as the ignition delay time of gaseous mixtures, flamevelocities or the strain necessary to extinguish diffusion flames can be calculated for a large numberof fuels and the data obtained is in good agreement with the experimental results. It is possible topredict the concentration profiles of fuel, oxidiser, both intermediate and the main products (CO,CO2, H2 and H2O) of the combustion processes accurate. The most important chemical pathwaysleading to the formation of air pollutants, NOx, soot, or dioxins are already known today. However,the level of the knowledge of the basic combustion processes required to calculate these pathwaysaccurately, is much higher, if the concentrations of the pollutants are low, as is the case in the ex-haust of modern, highly optimised combustion devices.

���� 0RGHOOLQJ�RI�WKH�&RPEXVWLRQ�RI�+\GURFDUERQ�)XHOV

One of the main research activities within the Combustion Kinetics group is the development of re-action mechanisms for the combustion of hydrocarbons. This is a major task of combustion model-ling, because the success of all simulations of the combustion processes is based on a deep under-standing of the chemical reactions occurring in a flame.

A mechanism involving over 1000 chemical reactions and more than 100 different species was de-veloped based on a mechanism provided by Chevalier et al, enhanced and revised by recent kineticrate data. The mechanism includes chemical species containing up to 14 carbon atoms and describesthe formation of benzene and the first polycyclic aromatic hydrocarbons (PAH). This mechanism isupdated periodically, with the help of the latest studies in the field of chemical kinetics, includingboth experimental and theoretical results.

This mechanism was mainly compared with the behaviour of rich C2H2 flames and is also used asthe gas phase mechanism for the calculation of soot formation within these flames. Good agreementbetween measured and calculated species profiles was obtained. To illustrate its ability to model thebehaviour of other hydrocarbon fuels, a series of simulations of flames have been performed andalso planed for the future. The starting point was the calculation of various flames burning withsmall hydrocarbons as methane, acetylene and propane. Further simulations will be performed forhigher hydrocarbons up to heptane, octane and even dodecane, as this is of greater practical and en-gineering interest. The validation of the mechanism in these flames will lead to a powerful tool forcombustion modelling. Emphasis is also concentrated on the development and validation of mecha-

nisms for new fuels. One of the most promising is Dimethylether (DME). The advantage of DMEcompared the currently used Diesel fuel is that engines running with DME do not emit soot. Amechanism developed by Dagaut et al. has been included and simulations for counter-flow flamesare planned. The results of these calculations will be compared with collaborative measurementswhich carried out within the Division of Combustion Physics at the Lund Institute of Technology.

The results of the simulation of a methane flame with equivalence ratios varying from φ=0.92 toφ=1.94 are presented below. The measurements were taken from a study by Musick et al. usingMolecular Beam Mass Spectrometry (MBMS) for the characterisation of 13 different species pro-files. The agreement between the calculations and the measured species profiles for the flame withφ=1.17 shown in )LJXUH���� was quite encouraging.

As a further improvement, a new pathway for the oxidation of the vinyl radical with molecular oxy-gen was included. This is one of the most discussed reactions in the combustion of hydrocarbons. Itrepresents the main reaction channel leading from C2-species to C1-species, e.g. CHO and CH2O,which are then successively oxidised to CO and CO2. One reason for its importance is that the vinylradical is believed to play a major role in the formation of benzene, PAHs and soot. Another reasonis the complexity of this reaction due to the change in the reaction pathway as the temperature andpressure vary. The rate coefficients for the different reaction channels and the thermodynamic dataof the intermediate species have been calculated by using QRRK and ab-initio MO methods.

These reactions were included in the existing mechanism and the resulting species profiles con-cerning the species involved were satisfactory in relation to the reaction paths, used in older mecha-nisms This new pathway for the oxidation of C2H3 is quite complicated and involves five additionalspecies.

0 5 10 15 20 2510-4

10-3

10-2

10-1

100

CH3

H2

CO

CH4

H

OH

C2H4

CO2

H2O

mol

e fr

actio

n X

i

Height above burner [mm]

1

)LJXUH������0HDVXUHG��FORVHG�V\PEROV��DQG�SUHGLFWHG��OLQHV��VSHFLHV�SURILOHV�DW�φ� �����

It is planned to employ this pathway for not only the oxidation of the vinyl radical, but also for theoxidation of hydrocarbons with a vinyl group, e.g. C4H5, C6H5 and all the PAHs, that have reactedwith C2H2. Implementing these complex reaction paths for all these species will lead to an increasein the number of chemical species and to an unacceptable increase of computational time. The com-plex reaction mechanism was therefore reduced by assuming steady state conditions for one of theintermediates. This means only one global reaction is modelling the complex reaction scheme andjust one species must then be added to the mechanism. Calculations on the same flame were per-formed, but now using the reduced reaction scheme and the results have been compared with thesimulations performed with the full mechanism. The species profiles are in good agreement forφ=0.92, φ=1.17, φ=1.42, however, for the very rich flames the results differ slightly. The results forCH2O are plotted in )LJXUH����.

0 5 1 0 1 5 2 0 2 50 .0

5 .0 x 1 0 -4

1 .0 x 1 0-3

1 .5 x 1 0 -3

2 .0 x 1 0-3

2 .5 x 1 0-3

3 .0 x 1 0 -3

3 .5 x 1 0-3

4 .0 x 1 0 -3

C H 2 O P r o f ile s φ = 0 .9 2 f φ = 0 .9 2 r φ = 1 .1 7 f φ = 1 .1 7 r φ = 1 .4 2 f φ = 1 .4 2 r φ = 1 .6 8 f φ = 1 .6 8 r

Mol

efra

ctio

n X

i

H e ig h t a b o v e b u rn e r [m m ]

)LJXUH������&RPSDULVRQ�RI� WKH�VSHFLHV�SURILOHV�RI�&+�2�IRU� WKH� IXOO� �I��DQG� WKH� UHGXFHG� �U�VFKHPH�DW�GLIIHUHQW�HTXLYDOHQFH�UDWLRV�

���� 6RRW�)RUPDWLRQ

The formation of soot is the most complex chemical system in flames. Soot particles containingseveral thousand carbon atoms are formed from simple fuel molecules within a few microseconds.Soot is one of the most unwanted products in almost all combustion processes used for the produc-tion of energy. Its emission not only reduces the efficiency of practical combustion devices, but alsohas a strong impact on the environment and human health. A method of calculating soot formationin real combustion systems, such as diesel engines, gas turbines and power plants is therefore re-quired. This should lead to a better understanding of soot formation and also a decrease of sootemission in these industrial applications. One of the major projects within the Combustion Kineticsgroup is the reduction a detailed chemical soot model to make it applicable to 3D calculations.

The most detailed soot models, can be subdivided into those which describe the size distributionfunction of the soot particles, either by a statistical momentum approach, or by subdividing the sizerange of interest ,into discrete intervals ,or classes. Both approaches can be applied to one-dimensional and laminar combustion problems with the computer capacity available. Nowadays the

dependence of soot formation on temperature and pressure, up to 20 atm, can be calculated for dif-ferent hydrocarbon fuels and compared with the experimental results.

A graphical representation of the different processes leading from small gas-phase molecules to thesoot particles is shown in )LJXUH����.

First Step Second Step n Stepth

Polymerisation of PAHs

Gas-Phase

ParticleInception

CondensationSurfaceGrowth

SOOT

Oxidation

)LJXUH������5HSUHVHQWDWLRQ�RI�WKH�GLIIHUHQW�SURFHVVHV�LQYROYHG�LQ�VRRW�IRUPDWLRQ�

The gas-phase is modelled using a reaction system containing more than 1000 chemical reactionsand over 100 species which is described more detailed in section 1. The first aromatic rings, e.g.benzene and small PAHs, are formed in the gas phase. Acetylene molecules can react with thesesmall PAHs and polymerise to larger PAHs. This process is modelled by the method of linearlumping. The first soot particles are formed when two PAHs collide and form a three dimensionalparticle (particle inception). The soot particles formed can interact with the gas-phase by the addi-tion of acetylene to their surfaces (surface growth) and by the reaction with molecular oxygen andthe hydroxyl-radical (oxidation). Another process which increases the soot mass is the collision of aPAH with a soot particle.

The formation of soot in a two dimensional diffusion flame can be calculated using a detailed sootmodel on the basis of the classical flamelet model. However, the growth of soot particles is slow incomparison with other chemical reactions, or transport processes, in the flame. Additional transportequations have to be solved in the CFD-code for both the number density and the soot volume frac-tion. The source terms of particle inception and of surface growth, scaled by the local soot volumefraction, are stored in the flamelet library.

Conventional libraries include the local scalar dissipation rate, the pressure variation with time, thecomposition of the fuel and the temperature on the air-side of the flamelet. This more accurate ap-proach is probably of interest, if the formation of soot and the formation of NOx are calculated si-multaneously. However, the libraries required for the application in 3D calculations set too highdemand on computer storage and CPU time. A reduction of the flamelet libraries, with the help ofmulti-parameter fitting, is therefore introduced resulting in simple algebraic equations and a pa-rameter library. Rates of particle inception, surface growth and oxidation are given as parameter

functions. These will be included, as the source term, in the transport equation of the soot volumefraction part of the CFD code. It can be shown that the latter processes, i.e. surface growth and oxi-dation can be calculated as functions of the local soot volume fraction, with only a slight loss of ac-curacy. The dependence of the rates of soot formation on scalar dissipation rate, pressure and tem-perature of the fuel and air, can be used to perform a multi-parameter fit. This results in simple al-gebraic equations and a set of parameters describing the dependence of the source terms of soot vol-ume fraction. This compact formulation is a major advantage in comparison to the multi-dimensional flamelet libraries. The calculations performed could be reproduced much more easilyby other scientists, by just implementing the fitted functions and parameters into their codes. A li-brary of parameters for different fuel compositions could be calculated and used as a standard tablefor the calculations of all combustion systems.

In order to test the accuracy of this approach, a laminar 2D instationary diffusion flame was studiedby providing the source terms for the soot volume fraction from both the flamelet libraries and thepoly-parameter fit. The calculations were then compared with measurements taken from a study ofsoot formation in a methane/air flame. A detailed chemical soot model using the statistical momen-tum approach, as developed by Frenklach, was used for creating the flamelet libraries for the threedifferent contributions to the source term of the soot volume fraction.

The aim of this project is to replace the soot models within current CFD codes for combustion by anmore sophisticated soot model. Application within a 3D code for the calculation of soot formationin a Diesel engine and a 3D code for the simulation of room fires will then be the first steps to vali-date the model in real combustion systems.

���� 12;�)RUPDWLRQ

Within the last few decades of this century, the environmental effects of pollutant emissions fromcombustion sources have become increasingly serious. Smog - an unhealthy mixture of sulphur ox-ides and soot particles - was first encountered in London at the beginning of this century. The roleof nitrogen oxides (NOX) did not, however, become apparent until later, when photochemical smoggradually became a problem in the Los Angeles area from the mid-1940’s onwards. Photochemicalsmog is mainly generated by the action of solar radiation on NOX from car exhaust gases. NOX alsohas an acidic effect in combination with water, thus forming acid rain. Since combustion processesare major sources of NOX species, it is of great interest to examine the NOX generation mechanismsconnected with combustion - especially high-temperature combustion of lower-grade fuels.

There are two possible sources for the nitrogen that is transformed into NOX during combustion inair; the fuel itself and atmospheric nitrogen. In the combustion of nitrogen free fuels NOX can onlybe produced in processes involving atmospheric nitrogen. Some examples of practical fuels con-taining negligible amounts of N, are natural gas and lighter liquid petroleum fractions. These fuelsare well described with the classical NOX reaction mechanisms (thermal, N2O, NO2 and promptmechanisms). It is well known that NOX emissions from these fuels increases with increasing tem-perature and decreasing fuel-to-air ratio (in the relevant range). Secondary NOX reduction processeslike reburning, thermal DeNOX and staged combustion (e.g. rich-lean and lean-rich staging) are welldocumented - also in practical use. These are all based on injection of some substance into the par-tially burned mixture. This can either be fuel or air or an additive such as ammonia or cyanuric acid.

If the fuel is contaminated with nitrogen, large amounts of NOX can be emitted. This applies to awide range of combustibles, such as crude oil, biogas and hazardous waste. For these fuels the de-pendence of NOX emissions on the different parameters is not well known and should therefore beinvestigated in the future. For the successful modelling of these kinds of fuels, information on thedecomposition of more complex nitrogen compounds is required, in addition to detailed knowledgeof the basic NOX formation mechanisms.

It is an essential task to design combustion processes for minimum NOX emissions from nitrogencontaining fuels without loss in efficiency. Staged combustion is a method which has been utilisedin heat generation applications for some time, e.g. for NOX reduction in large-scale stationary coalfurnaces and boilers. For nitrogen free fuels, staged combustion with ammonia doping of the fuelcan be used to reduce NOX emissions considerably. The experiences gained with this type of proc-ess are of course applicable to fuels containing natural nitrogen impurities as well. Often the stagingis made up of an initial stage with fuel-rich conditions keeping initial NO production low, followedby a subsequent stage, with air injection at a lower temperature, to achieve overall fuel-lean opera-tion. The reverse process (lean-rich operation) utilises the reburn process for minimisation of theNO emission. Alternatively, three-stage processes (lean-rich-lean) can be applied.

The method of staged combustion was quite recently applied to waste incineration. Investigationswere undertaken, using monomethylamine (CH3NH2) as a model species of fuel-bound nitrogencompounds found in waste products. Fuel-bound nitrogen also contains more complicated sub-stances, but in general terms, the kinetic behaviour of these is poorly understood. The use ofCH3NH2 may thus be considered as a first approximation. Previous investigations have been under-taken using ethylene (C2H4) doped with CH3NH2 burned in a scaled-down model of an industrialtwo-stage burner, consisting of a toroidal jet-stirred reactor followed by a tubular section with sec-ondary air injection. A successful theoretical model for this device is represented by a combinationof a perfectly stirred reactor (PSR) and a plug flow reactor (PFR). The jet-stirred reactor is assumedto operate far from blow-out conditions, and can therefore be closely approximated to by the PSRmodel. Current activities include the modification of existing FORTRAN code for a homogeneousbatch reactor, to incorporate models for series-connected PSRs and PFRs, as well as a general pro-cedure for combustion and re-injection of mixtures of arbitrary fuels and oxidants.

Focussing on specific aspects of kinetic modelling, there is evidence for certain similarities inemission characteristics of fuels doped with CH3NH2 and NH3 it is expected that CH3NH2 is rapidlydecomposed and that the NO formation is then mainly controlled by the concentration of NH3 andits reaction paths. Special effort will therefore made to modify the H/N/O mechanism, with the helpof experiments from investigations of NH3/NO flames, as well as H2/O2 flames doped with NH3 andNO/NH3. Well proven mechanisms for hydrocarbon combustion, NOX formation and CH3NH2 de-composition are taken as the basis for the simulations which can then be compared with relevantmeasurements. To compare the efficiency of the two staging concepts (two and three stage), addi-tional simulations will be made of the three-stage process, with C2H4/CH3NH2 as fuel, i.e. only thestaging geometry is then changed. Sensitivity analyses will be performed to cast further light on thekinetics and also to serve as the first step towards a reduced reaction mechanism. The question ofwhich reactions are relevant for the description of fuel and air re-injection is of special interest.

���� (QJLQH�.QRFN

Future demands on spark-ignition engines towards low fuel consumption and zero emission of airpollutants require detailed information of the physical and chemical processes during ignition andcombustion. Nowadays, three-way catalysts minimise the emissions of NOx, CO, and unburned hy-drocarbons from passenger cars. Technologies that reduce the formation of unwanted pollutantsduring the engine combustion process are however more efficient. Low fuel consumption and in-creased efficiency is desirable both due to considerations of carbon dioxide production (contributingto the greenhouse effect) and economy (cost effectiveness).

The most simple strategy for increasing the efficiency of an combustion engine is to increase thecompression ratio. An increase in compression ratio leads however to higher end-gas temperaturesand pressures and thereby to a higher tendency of the fuel-air mixture to autoignite. The autoignitionof the end-gas can cause engine knock that may damage the engine. The occurrence of knock limitsthe compression ratio of SI engines using today’s fuels, thus, a more detailed understanding of theprocesses governing knock is important.

As stated above is the onset of autoignition processes in the unburned end-gas of the cylinder a ma-jor cause for the occurrence of engine knock. But it is also found, that the information, that autoig-nition will occur gives no indication about the knock intensity. This is reasoned by the two stageautoignition of higher hydrocarbons at low and intermediate temperatures. For engine knock a highrate of heat release is needed and the heat of formation is released during the second stage of theignition process. It is found that the end gas temperature ranges from 900—1200 K, that is withinthe intermediate range, for knocking cycles. But before autoignition occurs passes the end-gas thelow temperature regime, thus the ignition process might be influenced by the low temperature reac-tions also.

A detailed reaction mechanism for studying the oxidation of the primary reference fuels iso-octaneand n-heptane and over a wide range of temperature and pressure is known from the literature. Thismechanism has been expanded to cetane with the help of an automatic generation tool for reactionmechanism, so that the influence of pro-knock and anti-knock additives could be analysed. It is con-cluded that the onset of knock is mainly influenced by the production or consumption of radicalsbetween the two stages of the autoignition process.

Such a detailed reaction mechanism involves more than 200 chemical species and 2000 chemicalreactions. An analysis of the basic reaction pathways is therefore even with advanced computationaltools, such as sensitivity and automatic reaction flow analysis, difficult. It has been shown that thebasic response of the ignition processes calculated from the detailed reaction mechanism to varia-tion in the pressure and temperature can be reproduced with a simple skeletal mechanism.

The temperature measurements described above will be completed by an analysis of the chemistryin the end gas of the cylinder. The end gas was modelled as an homogeneous reactor, which followsthe curve of mean pressure over about 200 cycles near to the knock limit measured within the cyl-inder. The influence of end gas compression/expansion by both the piston movement and also bythe propagating flame front can therefore be taken into account. This approach has been shown be-fore to be a useful tool in analysing the chemistry of the end gas in cylinders in both fired and mo-tored cycles.

The chemistry of the end gas has been analysed assuming that the unburnt gas mixture is homoge-neous. Non-homogeneous effects, such as hot spots, heat conductivity from the flame front and heatlosses to the walls are neglected. These assumptions lead to a set of zero-dimensional time depend-ent equations that can be solved with a higher order backward differential scheme.

ρ υ ωdY

dtWi

i i kk

N

k

r

==

∑ ,1

(1)

ρ υ ωcdT

dth W

dp

dtp i i i k kk

N

i

N rs

= −

+

==∑∑ ,

11

(2)

The set of equations (1) describes the conservation of mass for the all chemical species 1V intro-duced by the chemical reaction mechanism used.. Here ρ is the density of the gas mixture, <L are thespecies mass fractions, :L the molecular weight of the species L, νι,κ the matrix of the stoichiometriccoefficients and ωκ the reaction rate of reaction N�� Equation (2) describes the conservation of energyand FS is the heat capacity at constant pressure, KL�the enthalpy of species L and S the pressure in thecylinder. The last term on the right hand side of equation (2) is the volume work due to compressionor expansion of the end gas by the piston and by the propagating flame front. This term is used inthe calculations shown in the next chapter taken from the pressure transducer measurements.

CH2O+

CO

n-C8H18

C8H17

C8H17O2

C8H16OOH

O2C8H16OOH

HOOC8H15OOH

OC8H15OOH

OC8H15O

i-C6H13

C3H6 + C2H4 + CH3

n-C7H16

C7H15

C7H15O2

C7H14OOH

O2C7H14OOH

HOOC7H13OOH

OC7H13OOH

OC7H13O

1-C5H11

2 C2H4 + CH3

)LJXUH������6FKHPDWLF�LOOXVWUDWLRQ�RI�WKH�ORZ�WHPSHUDWXUH�R[LGDWLRQ�FKHPLVWU\�RI�SULPDU\�UHI�HUHQFH�IXHOV�

The chemical reaction mechanism used, is based on a mechanism originally developed for fuel richgases combined with a skeletal mechanism for premium reference fuels taken from the literature. Aschematic illustration of the low temperature oxidation mechanism is presebted in )LJXUH����. Fromthis the global reactions for the low temperature oxidation of iso-octane and n-heptane can be givenas follows:

iso-C8H18+O2 → C3H6+C2H4+CH3+CH2O+CO+H2+H

n-C7H16+O2 → 2C2H4+CH3+CH2O+CO+H2+H

It is obvious that the only difference in the two oxidation mechanisms is the formation of C3H6 fromthe iso-octane, instead of the C2H4 from n-heptane. The formation of C2H4 from the C3H6 radical isconsuming or chain breaking.

C3H6+O+OH → C2H4+CO+H2O+H

In addition, it is known that the overall rate for the global oxidation reaction of the branched-chainfuel iso-octane is lower than for the linear n-heptane. The abstraction of a secondary H-atom has asignificantly lower activation energy, due to the lower number of secondary H-atoms. The influenceof the fuel structure on the auto-ignition process will be discussed in the next chapter, where calcu-lations of the end gas chemistry are shown for different RON-numbers of the primary reference fuel.

)LJXUH���� illustrates the evaluated temperatures as a function of degree crank angle. The measure-ments were taken at a distance of 3 mm from the combustion chamber wall. At each degree crankangle the temperature data shown, results from one hundred single-shot spectra. The relative stan-dard deviation of the temperatures at -42° is found to be 2.5%. The precision of the temperaturemeasurements using the rotational CARS technique cannot be estimated from this value, since it isdue to the sum of the errors in the measurement and the cycle to cycle variations of the engine underknocking conditions.

It can be seen from )LJXUH���� that auto-ignition of the end gas occurs at crank angles between 9and 26 degrees. For most cycles the flame front has passed the control volume at 26 degree crankangle. An evaluation of the CARS-spectra was not possible either after auto ignition, or behind theflame front. The number of spectra which could be evaluated, decreased rapidly as the crank angleincreased.

The zero dimensional numerical calculations have been undertaken for three different initial tem-peratures within the spread of the experimental data. For these calculations the pressures have beentaken from the measurements. During the compression of the end gas, the calculated temperaturegradient agrees closely with the experimental results. The onset of auto ignition is also comparablewith the experimental findings. Typical profiles of calculated mole fractions of n-heptane and iso-octane, the product CO2, and the intermediate products CO and C2H4 are shown in )LJXUH����.

-50 -40 -30 -20 -10 0 10 20 30500

600

700

800

900

1000

T(initial) = 574 K T(initial) = 550 K T(initial) = 527 K

Tem

pera

ture

(K

)

Crank angle (degrees)

)LJXUH������0HDVXUHG�WHPSHUDWXUHV�DV�D�IXQFWLRQ�RI�GHJUHH�FUDQN�DQJOH�LQ�FRPSDULVRQ�ZLWKWKH�FDOFXODWHG�WHPSHUDWXUH�SURILOHV�

8 10 12 14 160.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0.016

n-C7H16

i-C8H

18

C2H4

CAD

mol

e fr

actio

n

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

CO2

CO

mole fraction

)LJXUH������&DOFXODWHG�PROH�IUDFWLRQV�RI�WKH�SULPDU\�UHIHUHQFH�IXHO�PL[WXUH�DW�521����RI�WKHLQWHUPHGLDWH�SURGXFWV�&�+���&2��DQG�RI�WKH�SURGXFW�&2��

��� 5HIHUHQFHV

Agrup, S., F. Ossler and M. Aldén, “Measurements of collisional quenching of hydrogen atoms inan atmospheric-pressure hydrogen/oxygen flame by picosecond laser-induced fluorescence”,Appl. Phys. B ��, 479–487 (1995).

Akram, M. and E. Lundgren, “The evolution of spark discharge in gases: I, Macroscopic models”, J.Phys. D ��, 2129–2136 (1996).

Akram, M. and R. Reinmann, “Experimental and theoretical studies of the early stages of a fastspark”, paper presented at the 18th Task Leaders Meeting of the IEA Implementing Agreement(QHUJ\�&RQYHUVLRQ�DQG�(PLVVLRQ�5HGXFWLRQ� LQ�&RPEXVWLRQ, Cranfield Univ., UK, August 6–91996.

Akram, M., “A two-dimensional model for spark discharge simulation in air”, AIAA J. ��, 1835–1842 (1996).

Akram, M., 0RGHOOLQJ�RI�6SDUN�WR�,JQLWLRQ�7UDQVLWLRQ�LQ�*DV�0L[WXUHV, Ph.D. Dissertation 1996.

Akram, M., “The evolution of spark discharge in gases: II, Numerical solution of one-dimensionalmodels”, J. Phys. D ��, 2137–2147 (1996).

Aldén, M. and N. Georgiev, “Development of single-shot imaging of combustion species using two-photon LIF”, paper presented at CLEO '95, Baltimore, USA, May 21–26 1995.

Aldén, M., R. Fritzon, N. Georgiev, and K. Nyholm, “Detection of combustion species using two-photon polarisation spectroscopy”, paper presented at CLEO '95, Baltimore, USA, May 21–261995.

Andersson, Ö., G. Juhlin, M. Ekenberg, B. Johansson and M. Aldén, “Crank angle resolved HC-detection with LIF in the exhaust of small two-stroke engines running a high engine speed”, SAEInt. Fall Fuels & Lubricants Meeting & Exposition, San Antonio, TX, USA, October 14–17 1996SAE961927.

Andersson, Ö., G. Juhlin, M. Ekenberg, B. Johansson and M. Aldén, “Laser diagnostics in smalltwo-stroke engines”, Poster presented at 18th Task Leaders Meeting of the IEA ImplementingAgreement (QHUJ\�&RQVHUYDWLRQ�DQG�(PLVVLRQV�5HGXFWLRQ�LQ�FRPEXVWLRQ, Cranfield University,UK, August 6–10 1996.

Balthasar, M. and F. Mauss, “The influence of the vinyl-oxidation on the calculated structure ofpremixed CH4/O2/Ar flames”, paper presented at Tech. Fall Meeting of the Eastern States Sec.of the Comb. Inst., Hilton Head Island, USA, December 1996.

Balthasar, M., “Reduction of a detailed soot model for the application in 3D programs”, paper pre-sented at Dept. of Mech. and Aerospace Eng., Princeton University, USA, December 11–121996.

Bengtsson, P.-E. and M. Aldén, “Soot visualisation strategies using laser techniques: Laser-inducedfluorescence in C2 from laser-vaporised soot and laser-induced soot incandescence”, Appl. Phys.B ��, 51–59 (1995a).

Bengtsson, P.-E., L. Martinsson and M. Aldén, “Combined vibrational and rotational CARS for si-multaneous measurements of temperature and concentrations of fuel, oxygen and nitrogen”,Appl. Spectr. ��, 188–192 (1995b).

Bengtsson, P.-E., L. Martinsson, M. Aldén and J. Bood, “Rotational coherent anti-Stokes Ramanspectroscopy for measurements of temperature in combustion”, paper presented at Optik iSverige, Stockholm, October 18–19 1995c.

Bengtsson, P.-E., J. Bood, K. Burgdorf, I. Denbratt, G. Fernström, S. Hajireza, F. Mauss and B.Sundén, “Experimental and modelling study of knock in a gasoline engine”, Poster presented at26th Symp. (Int.) on Comb., Naples, Italy, July 28–August 2 1996a.

Bengtsson, P.-E., J. Bood, K. Burgdorf, I. Denbratt, G. Fernström, S. Hajireza, F. Mauss and B.Sundén, “Knock in gasoline engines and the influence of heat transfer to the walls”, Poster pre-sented at 18th Task Leaders Meeting of the IEA Implementing Agreement (QHUJ\�&RQVHUYDWLRQDQG�(PLVVLRQV�5HGXFWLRQ�LQ�FRPEXVWLRQ, Cranfield University, UK, August 6–10 1996b.

Bengtsson, P.-E., “Simultaneous two-dimensional visualisation of soot and OH in flames using la-ser-induced fluorescence”, Appl. Spectr. ��, 1182–1186 (1996c).

Bood, J., M. Aldén, P.-E. Bengtsson and L. Martinsson, “Temperature measurements in a spark-ignition engine using dual-broadband rotational CARS”, Poster at XV European CARS Work-shop, Sheffield, UK, March 27–29, 1996.

Bood, J., P.-E. Bengtsson, F. Mauss, K. Burgdorf, I. Denbratt, “Knock in spark-ignition engines:End-gas temperature measurements using rotational CARS and retailed kinetic calculations ofthe autoignition process”, paper submitted to SAE Int. Spring Fuel & Lubricants Meeting &Expo., Deoborn, MI, USA, May 5-8 1997.

Engström, J., Investigation of using two-line atomic fluorescence for visualisation in combustionsystems, Diploma paper, LRCP-25 1996, /XQG�5HSRUWV�RQ�&RPEXVWLRQ�3K\VLFV.

Eriksson, L.-E., R. Rydén, H. Kaaling and C. Löfström, “OH-PLIF visualisation and LES model-ling in a lean premixed low emission combustor”, in preparation.

Fritzon R., 'HYHORSPHQW�RI�WZR�GLPHQVLRQDO�ODVHU�WHFKQLTXHV�IRU�FRPEXVWLRQ�IORZ�VWXGLHV, Lic. the-sis 1995.

Fritzon, R. and M. Aldén, “Visualisation of the mixing in interacting turbulent shearlayers throughuse of planar laser-induced fluorescence”, LRCP-20 1995 /XQG�5HSRUWV�RQ�&RPEXVWLRQ�3K\VLFV�

Fritzon, R., B. Löfstedt, N. Georgiev, K. Nyholm and M. Aldén, “Detection and visualisation ofcombustion species and temperature using polarisation spectroscopy”, paper presented at GordonConf., Plymouth, NH, USA, 1995.

Fritzon, R., N. Georgiev, K. Nyholm and M. Aldén, “Detection of combustion species using two-photon polarisation spectroscopy”, paper presented at CLEO '95, Baltimore, USA, May 21–261995.

Fureby, C., “Large eddy simulation of turbulent anisochoric flows”, AIAA J. ��, 1263–1272(1995a).

Fureby, C., 2Q�0RGHOOLQJ�RI�8QVWHDG\�&RPEXVWLRQ�8WLOLVLQJ�&RQWLQXXP�0HFKDQLFDO�0L[WXUH�7KHR�ULHV�DQG�/DUJH�(GG\�6LPXODWLRQV, Ph.D. Dissertation Lund 1995b.

Fureby, C., E. Lundgren and S.-I. Möller, “Simulations of bluff body stabilised flames”, 1st ETCMWorkshop, February 5–10 1995a, Aussois, France.

Fureby, C., E. Lundgren and S.-I. Möller, “Large eddy simulation of combustion”, presented atISTP-8, July 16–20 1995b, San Francisco, USA.

Fureby, C., E. Lundgren and S.-I. Möller, “On the formulation of the governing equations of com-bustion systems”, paper presented on ISTP-8, San Francisco, USA, July 16–20 1995c.

Fureby C. and S.-I. Möller, “Large eddy simulations of chemically reacting flows applied to bluffbody stabilised flames”, AIAA J. ��, 2339–2347 (1995a).

Fureby, C. and S.-I. Möller, “New aspects on large eddy simulations”, presented at SvenskaMekanikdagar, May 31–June 2 1995b, Lund Sweden.

Fureby, C. and S.-I. Möller, “Full numerical simulation of turbulent combustion”, presented atSvenska Mekanikdagar, May 31–June 2 1995c, Lund Sweden.

Fureby, C., “On subgrid scale modelling in large eddy simulations of compressible fluid flow”,Phys. Fluids �, 1301–1311 (1996).

Georgiev, N. and M. Aldén, “2D multiphoton multispecies imaging of combustion species”, paperpresented at 26th Symp. (Int.) on Comb., Funchal, Madeira, April 1–4 1996.

Georgiev, N. and M. Aldén, “Applications of LIF for two-dimensional multi-species imaging inflames”, Spectrochim. Acta B, in press (1996).

Georgiev, N. and M. Aldén, “Two-dimensional imaging of flame-species using two-photon LIF”,Appl. Spectr., in press (1996).

Georgiev, N. and M. Aldén, “Two-dimensional imaging of flame-species using two-photon LIF”,paper presented at 26th Symp. (Int.) on Comb., Funchal, Madeira, April 1–4 1996.

Georgiev, N., /DVHU�%DVHG� 'LDJQRVWLF� 7HFKQLTXHV� ,QYROYLQJ� 7ZR�SKRWRQ� $EVRUSWLRQ, Lic. thesis1995.

Hergart, C.-A., “On the reduction of nitrogen oxides in hybrid electric vehicles through applicationof energy management strategies and active lean NOx catalyst technology”, Diploma paper,LRCP 26-1996, /XQG�5HSRUWV�RQ�&RPEXVWLRQ�3K\VLFV.

Hajireza, S., F. Mauss and B. Sundén, “Investigation on end-gas temperature and pressure increasein gasoline engines and relevance for knock occurrence”, paper submitted to SAE Int Spring Fuel& Lubricants Meeting & Expo, Deoborn, MI, USA, May 5–8 1997.

Johansson, B., H. Neij, M. Aldén and G. Juhlin, “Investigations of the influence of mixture prepa-ration on cyclic variations in a SI-engine using laser-induced fluorescence”, SAE Tech. paper Se-ries 950108, Engine Comb. and Flow Diagnostics SP-1090, pp. 85–98.

Johansson, B., H. Neij, M. Aldén and G. Juhlin, “Residual gas visualisation using laser-inducedfluorescence”, SAE Technical paper Series 952463, Diagnostics in Diesel and SI Engines SP-1122, pp. 129–139 (1995).

Juhlin, G., $SSOLFDWLRQV�DQG�'HYHORSPHQW�RI�/DVHU�'LDJQRVWLF�7HFKQLTXHV�IRU�&RPEXVWLRQ�(QJLQH6WXGLHV, Lic. thesis 1996.

Juhlin, G., H. Neij, B. Johansson, M. Versluis and M. Aldén, “Investigation of the influence ofmixture preparation and rest gases on cyclic dispersion in IC-engines using laser-induced fluo-rescence”, poster presented at 17th Task Leaders Meeting of the IEA Implementing Agreement(QHUJ\�&RQVHUYDWLRQ�DQG�(PLVVLRQV�5HGXFWLRQ�LQ�&RPEXVWLRQ, Liège, Belgium, September 18–20 1995.

Juhlin, G., M. Versluis, H. Neij, B. Johansson and M. Aldén, “Investigation of the potential for us-ing planar laser-induced fluorescence of H 2O to study the influence of rest gases on combustionin a combustion cell”, submitted to Combust. Sci. Tech., 1996.

Kaminski, C. F., B. Löfstedt, R. Fritzon and M. Aldén, “Two-photon polarisation spectroscopy and(2+3) photon laser induced fluorescence of N2”, Optics Comm. ���, 38-43 (1996a).

Kaminski, C. F., B. Löfstedt, R. Fritzon and M. Aldén, “Two photon resonant detection of N2 usingpolarisation spectroscopy and laser-induced fluorescence”, paper presented at Laser Applicationsto Chemical and Environmental Analysis Meeting, JOSA, Orlando Topical Meeting, USA,March 1996b.

Kaminski, C. F., Ö. Andersson and M. Aldén, Report presented at Zodiac Project Midterm Meeting,PSI, Paris, France, June 27–28 1996c.

Kaminski, C. F., B. Löfstedt, R. Fritzon and M. Aldén, “2-photon DFWM spectroscopy for the de-tection of molecular nitrogen”, in preparation.

Lindholm A., S.-I. Möller, “Measurements and simulations of the fluid dynamics and heat release ina flat pulse combustor”, submitted to 16th ICDERS, Cracow, Poland, August 3–8 1997.

Löfstedt, B. and M. Aldén, “Simultaneous detection of OH and NO in a flame using polarisationspectroscopy”, Optics Comm. ���, 251–257 (1996).

Löfstedt, B., ,QYHVWLJDWLRQ� RI� 3RODULVDWLRQ� 6SHFWURVFRS\� IRU� &RPEXVWLRQ� 'LDJQRVWLFV, Lic. thesis1996.

Löfstedt, B., R. Fritzon and M. Aldén, “Investigation of NO detection in flames using polarisationspectroscopy”, Appl. Opt. ��, 2140–2146 (1996).

Löfstedt, B., R. Fritzon and M. Aldén, “Investigation of NO detection in flames using polarisationspectroscopy”, paper presented at XIV European CARS Workshop ECW '95, El Escorial, Spain,March 29–31 1995.

Löfstedt, B., R. Fritzon and M. Aldén, “Polarisation spectroscopy in combustion diagnostics”, paperpresented at Comb. Inst., Funchal, Madeira, April 1–4 1996.

Löfström, C. and M. Aldén, “Industrial applications of laser spectroscopic techniques for studies ofcombustion/flow phenomena”, paper presented at Comb. Inst., Funchal, Madeira, April 1–41996a.

Löfström, C., H. Kaaling and M. Aldén, “Visualisation of fuel distributions in premix ducts to alow-emission gas turbine combustor using laser techniques”, paper presented at 26th Symp. (Int.)on Comb., Funchal, Madeira, April 1–4 1996b.

Löfström, C., H. Kaaling and M. Aldén, “Visualisation of fuel distributions in premix ducts to alow-emission gas turbine combustor using laser techniques”, 26th Symp. (Int.) on Comb.,Naples, in press (1996c).

Martinsson, L., P.-E. Bengtsson and M. Aldén, “Oxygen concentration and temperature measure-ments in N2–O2 mixtures using rotational coherent anti-Stokes Raman spectroscopy”, Appl.

Phys. B ��, 29–37 (1996).

Mauss, F. and D. Nilsson, “A comprehensive kinetic model for staged combustion of nitrogendoped fuels”, paper submitted to 4th Int. Conf. on Tech. & Comb. for a clean environment, Lis-bon, Portugal, July 7–10 1997.

Mauss, F. and H. Pitsch, “A detailed model for surface growth of soot particles”, poster presented at26th Symp. (Int.) on Comb., Naples, Italy, July 28–Aug 2 1996.

Mauss, F., “The kinetics of autoignition during knocking SI-engine cycles”, paper presented at theDept. of Mech. and Aerospace Eng., Princeton University, USA, December 11–12 1996.

Möller, S.-I. and C. Fureby, “Large eddy simulation of the flow past a bluff body”, submitted toAIAA J. (1996).

Möller, S.-I., E. Lundgren and C. Fureby, ”Large eddy simulation of unsteady combustion”, paperpresented at 26th Symp. (Int.) on Comb., Naples, Italy, July 28–Aug 2 1996.

Neij, H., B. Johansson and M. Aldén, “Cycle-resolved two-dimensional laser-induced fluorescencemeasurements of fuel/air ratio correlated to early combustion in a spark-ignition engine”, F. Cu-lick et al. (eds.), 8QVWHDG\�&RPEXVWLRQ� pp. 383–389, Kluwer Academic Publ., The Netherlands,1996.

Neij, H., B. Johansson, G. Juhlin, M. Versluis and M. Aldén, “Visualisation of cyclic variations ofthe fuel concentration in an internal combustion engine using planar laser-induced fluorescence”,paper presented at $SSOLFDWLRQV� RI�0ROHFXODU� DQG� /DVHU� 3K\VLFV, Nijmegen, The Netherlands,December 19–20 1995.

Nilsson, D., Kinetic Modelling of NOx formation in Combustion Processes, talk given at the Dept.of Mech. and Aerospace Eng., Princeton University, USA, December 11–12 1996.

Nyholm, K., R. Fritzon, N. Georgiev and M. Aldén, “Two-photon induced polarisation spectroscopyapplied to the detection of NH3 and CO molecules in cold flows and flames”, Optics Comm.

���, 76–82 (1995).

Ossler, F., S. Agrup and M. Aldén, “Three-dimensional flow visualisation with picosecond Miescattering and streak camera detection”, Appl. Opt. ��, 537–540 (1995a).

Ossler, F., S. Agrup, J. Larsson and M. Aldén, “Picosecond laser based studies applied to combus-tion and flow phenomena”, poster presented at Gordon Conf., Plymouth, NH, USA, 1995b.

Ossler, F., J. Larsson and M. Aldén, “Measurements of the effective lifetime of O atoms in atmos-pheric premixed flames”, Chem. Phys. Lett. ���, 287–292 (1996a).

Ossler, F., J. Larsson and M. Aldén, “Application of picosecond laser-induced fluorescence formeasurements of quenching parameters of species related to combustion”, poster presented at26th Symp. (Int.) on Comb., Naples, Italy, July 28–August 2 1996b.

Ossler, F. and M. Aldén, “Measurements of picosecond laser induced fluorescence from gas phase3-pentanone and acetone: Implications to combustion diagnostics”, Appl. Phys. B ��, 493–502(1997).

Ossler, F., L. Martinsson, T. Metz and M. Aldén, “Two-dimensional visualisation of fluorescencelifetimes with picosecond laser excitation and streak-camera detection”, to be submitted (1997).

Reinmann, R., A. Saitzkoff, M. Akram, T. Berglind and M. Aldén, IDEA EFFECT, “Characterisa-tion of the early stages of spark ignition and flame development”, Contract JOU2-CT92-0162 pe-riodic report 950101-950630 presented at Warwick University, Coventry, GB, June 1995a.

Reinmann, R., A. Saitzkoff, M. Akram, T. Berglind and M. Aldén, IDEA EFFECT, “Characterisa-tion of the early stages of spark ignition and flame development”, Contract JOU2-CT92-0162,periodic report 950701-951231, presented in Lyon, France, December 1995b.

Reinmann, R., A. Saitzkoff, M. Akram, T. Berglind and M. Aldén, IDEA EFFECT, “Characterisa-tion of the early stages of spark ignition and flame development”, Final report JOULE II Programpresented at IFP, Paris, France June 25–26 1996a.

Reinmann, R. and M. Akram, “Temporal investigation of a fast spark discharge in chemically inertgases”, Accepted to J. Phys. D, July 1996b.

Reinmann, R., A. Saitzkoff, F. Mauss and M. Glavmo, “Local air-fuel ratio measurements using thespark plug as an ionisation sensor”, paper submitted to the 1997 Int. Congress and Exposition,Detroit, MI, USA, February 24–27 1997.

Saitzkoff, A. and T. Berglind, “An ionisation equilibrium analysis of the Mecel ionisation sensor”,LRCP-19 1995, /XQG�5HSRUWV�RQ�&RPEXVWLRQ�3K\VLFV.

Saitzkoff, A., R. Reinmann, T. Berglind and M. Glavmo, “An ionisation equilibrium analysis of thespark plug as an ionisation sensor”, paper presented at the 1996 SAE Int. Congress and Exposi-tion, Detroit, MI, USA, February 26–29 1996a.

Saitzkoff, A., R. Reinmann and T. Berglind, “An ionisation equilibrium analysis of the spark plugas an ionisation sensor”, SAE Tech. paper Series 960337, Electronic Engine Controls 1996 (SP-1149), 157–167 (1996b).

Saitzkoff, A., R. Reinmann, F. Mauss and M. Glavmo, “In-cylinder pressure measurements usingthe spark plug as an ionisation sensor”, paper submitted to the 1997 Int. Congress and Exposi-tion, Detroit, MI, USA, February 24–27 1997.

Tichy, F. E., T. Bjørge, B. F. Magnussen, F. Mauss and P.-E. Bengtsson, “Two-dimensional imag-ing of glyoxal in a turbulent acetylene diffusion flame”, poster presented at 26th Symp. (Int.) onComb., Naples, Italy, July 28–August 2 1996.

Tichy, F. E., T. Bjørge, B. F. Magnussen, F. Mauss and P.-E. Bengtsson, “Two-dimensional imag-ing of glyoxal (C2H2O2) in acetylene flames using laser-induced fluorescence, submitted to Appl.Phys. B (1997).

Versluis, M., G. Juhlin, B. Johansson, H. Neij and M. Aldén, “Planar laser-induced fluorescence ofthe fuel concentration and residual gas distribution for the study of cyclic variations in a sparkignition engine”, poster presented at the Gordon Research Conf /DVHU�'LDJQRVWLFV� LQ�&RPEXV�WLRQ, Plymouth, NH, USA, 1995a.

Versluis, M., N. Georgiev, A. Trigueiros and M. Aldén, “Characterisation of rich premixed methaneand hydrogen/air flames using 2-dimensional LIF of OH and CO”, presented at the Dept. ofAppl. Phys., Delft Univ. of Tech. (TU Delft), The Netherlands, November 14 1995b.

Versluis, M., G. Juhlin, B. Johansson, H. Neij and M. Aldén, “Two-dimensional imaging of waterwith applications in combustion environments”, paper presented at the Conf. Section AtomicPhysics and Quantumelectronics NNV, Lunteren, The Netherlands, November 16–17 1995c.

Versluis, M., G. Juhlin, Ö. Andersson and M. Aldén, “Two-dimensional two-phase water detectionusing a tuneable excimer laser”, submitted to Appl. Spectr. 1996a.

Versluis, M., G. Juhlin, B. Johansson, H. Neij and M. Aldén, “Planar laser-induced fluorescence ofthe fuel and residual gas distribution for the study of cycle-to-cycle variations in a spark ignitionengine”, paper presented at Comb. Inst., Funchal, Madeira, April 1–4 1996b.

Versluis, M., N. Georgiev, L. Martinsson, M. Aldén and S. Kröll, “2-D absolute OH concentrationprofiles in atmospheric flames using planar LIF in a back and forth configuration”, paper pre-sented at 26th Int. Symp. on Comb., Naples, Italy, July 28–Aug 2 1996c.

Versluis, M., N. Georgiev, L. Martinsson, M. Aldén and S. Kröll, “2-D absolute OH concentrationprofiles in atmospheric flames using planar LIF using a bi-directional laser beam configuration”,Appl. Phys. B, in press.

Versluis, M., paper presented at the Autumn Symp of the Dutch Physicists Soc. (Sec. Atomic Phys.& Quantumelectr.) Lunteren, The Netherlands, November 7–8 1996d.

,QYLWHG�7DONV

Aldén, M., Application and Developments of Laser Spectroscopic Techniques for Combustion Di-agnostics, Univ. Kaiserslautern, February 23 1995.

Aldén, M., 12 Invited Lectures at Xi’an Modern Chemistry Research Institute, PR China, May 28–June 7 1996.

Aldén, M., Development and Application of Laser-Spectroscopic Techniques for Combustion Diag-nostics, 14th Int. Symp. on Gas Kinetics, Univ. of Leeds, September 7–12 1996.

Aldén, M., Laser Applications in Combustion Research, Lund Laser Centre, September 18 1996.

Aldén, M., Laserdiagnostiska tekniker för studier av förbränningsprocessen, Förbränningsmotorn itakt med tiden?!, SVEA, Scania, Södertälje, November 21 1996.

Kaminski, C. F., B. Löfstedt and M. Aldén, Two-Photon Degenerate Four-Wave Mixing Spectros-copy of Molecular Nitrogen, XV European CARS Workshop, Sheffield, UK, March 27–28 1996.

Kaminski, C. F., Laser Applications for Applied Combustion Diagnostics, Sydkraft Tech. Meeting,Malmö, September 11 1996.

56