Embed Size (px)
Citation preview
PREFACE
In pursuance of the press restructuring approved by the Prime Minister and the
proposal made by the Ministry of Education and Training, the Ministry of Culture and
Information issued the Decision No 519/GP-BVHTT dated November 25th, 2002,
licensing the publication of the University of Danang journal entitled “Science and
Technology”.
The Press Bureau under the Ministry of Culture and Information issued the official
dispatch No 816/BC dated August 10th, 2006, permitting the University of Danang to
publish the “Science and Technology” every two months instead of every three months.
The National Scientific and Technological Information Centre under the Ministry
of Science and Technology issued the official dispatch No 44/TTKHCN-ISSN dated
February 6th, 2007, agreeing to grant an international standardized code, namely ISSN
1859-1531 to the UD journal “Science and Technology”.
The Press Bureau under the Ministry of Information and Communications issued
the official dispatch No 210/CBC of March 5th, 2008, allowing the UD journal “Science
and Technology” to be published in English and French in addition to the Vietnamese
publication.
On September 15th, 2011, the Ministry of Information and Communications issued
the Decision No 1487 /GP-BVHTT, granting an additional licence which permits the UD
journal “Science and Technology” to increase the publication period frequency from
every 2 months to every month and to raise the number of pages from 80 to 150 pages.
On January 7th, 2016, the Ministry of Information and Communications issued
Decision No 07 /GP-BVHTT, granting a print licence which permits the UD journal
“Science and Technology” to operate with a publication frequency of 15 issues per year
(of which 3 issues are published in English).
The UD journal “Science and Technology” has been established in view of
announcing and introducing scientific researches in the fields of teaching and training,
offering information on the domestic and foreign scientific research results that serve the
education in colleges and universities and propagating the State and Party’s guidelines
and policies in education and training as well as in scientific and technological researches.
Over the past 40 years, the UD journal “Science and Technology” has been
published as the continuity and development of the scientific information journals and
conference proceedings published at the University of Danang and its member
universities.
The journal’s editorial staff wishes to receive many-sided contributions of the
scientists and educators from within the university and from other colleges and
universities inside and outside the country so that the journal’s quality will be more and
more improved in the future.
Journal’s Editorial Staff
CONTENTS ISSN 1859-1531 – THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018
ENGINEERING AND TECHNOLOGY
Chemical composition of essential oil extracted from leaves of vitex negundo linn from Binh Thuan province by hydrodistillation and microwave hydrodistillation Nguyen Thi My Dung, Vo Thi Dieu Hoa, Do Thi My Lien, Phung Van Trung, Pham Hong Ngoc,
Le Ngoc Hung 1
A comparative analysis of passivity-based control approaches with application to linear dynamical systems Hoang Ngoc Ha 4
Performance analysis and assessment of a transformer different protection relay SEL387 at 110kV Lang Co substation Le Kim Hung, Vu Phan Huan 8
A study on the reductive dechlorination of chloroform with nano Fe/Cu bimetallic particles in aqueous solution Phan Kim Nguyen, Bui Xuan Vung 13
A study on CF3I-Ar and CF3I-Kr mixture gases substituting SF6 in high voltage equipments Tran Thanh Son, Do Anh Tuan 17
Evaluation of shear strength of reinforced concrete structural walls of ACI 318-14 and eurocodes Tran Anh Thien 21
Fully resolved simulation of the phase change process of a liquid drop Vu Van Truong, Truong Viet Anh, Tran Xuan Bo, Truong Van Thuan 26
Multi-period linearized optimal power flow model incorporating transmission losses and thyristor controlled series compensators Pham Nang Van, Le Thi Minh Chau, Pham Thu Tra My, Pham Xuan Giap, Ha Duy Duc, Tran Manh Tri 31
NATURAL SCIENCES
Applying semismooth newton method to find fixed points of nonsmooth functions of one variable Pham Quy Muoi, Phan Quang Nhu Anh, Duong Xuan Hiep, Phan Duc Tuan 37
Polynomial solution of descriptor system Le Hai Trung 41
Fast gaussian distribution based adaboost algorithm for face detection Tuan M. Pham, Hao P. Do, Danh C. Doan, Hoang V. Nguyen 45
SOCIAL SCIENCES
The empirical study about e-CRM: a case study of Vietnam Airlines Nguyen Thi Khanh Chi 51
Teachers’ perception towards the use of ict in Vietnam: using activity theory to identify contradictions Huynh Ngoc Mai Kha, Pham Thi To Nhu 55
Improving structures of students’ argumentative essays through genre pedagogy Dinh Thanh Liem 59
The use of sociolinguistically rich pedagogical dialogues in teaching conversational English Nguyen Ngoc Nhat Minh 64
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 1
CHEMICAL COMPOSITION OF ESSENTIAL OIL EXTRACTED FROM LEAVES
OF VITEX NEGUNDO LINN. FROM BINH THUAN PROVINCE BY
HYDRODISTILLATION AND MICROWAVE HYDRODISTILLATION
Nguyen Thi My Dung1, Vo Thi Dieu Hoa4, Do Thi My Lien2, Phung Van Trung3
Pham Hong Ngoc4, Le Ngoc Hung4,* 1National University HCM City, Ho Chi Minh City; [email protected]
2Sai Gon University, Ho Chi Minh City; [email protected], 3Institute of Chemical Technology, Vietnam Academy of Science and Technology
4Center for Research and Technology Transfer, Vietnam Academy of Science and Technology.
[email protected], [email protected], [email protected],
Abstract - Essential oils from fresh and dry leaves of Vitex negundo (HD-Fresh, HD-Dry) were obtained by traditional hydrodistillation (HD) and microwave-assisted hydrodistillation (MHD) (MHD-Fresh, MHD-Dry). The chemical constituents of essential oil of leaves are analyzed by GC/MS technique. The results indicate that the major compound of four essential oil contains the same dominant
components β-caryophyllen (23.5%, 16.3%, 16.4% and 16.8%), eremophilene (18.9%, 15.1%, 14.4% and 14.2%), eucalyptol (16.2%, 16.3%, 13.6% and 19.6%), α-terpinyl acetate (10.8%, 7.6%, 9.2% and 8.8%), and sabinene (7.3%, 8.6%, 8.5% and 10.3%), respectively in oils obtained by MHD, HD from fresh leaves, MHD and HD from dry leaves. The total amount of sesquiterpenoid hydrocarbons (51.5% and 45.3%) is higher than monoterpenoids (44.8% and 43.5%) in essential oil obtained by MHD, respectively in oils from fresh and dry leaves. In contrast, the essential oil obtained by HD shows the greater concentration of monoterpenoids (45.3% and 53.6%) than sesquiterpenoids (44.3% and 41.0%), respectively in oils from fresh and dry leaves. By using MHD method, it is superior in terms of saving energy and extraction time although the total composition decreases with this method.
Key words - Vitex negundo; essential oil; microwave-assisted hydrodistillation; GC/MS; hydrodistillation.
1. Introduction
Essential oils are composed of a wide range of bioactive
chemical compounds. They traditionally found application
as flavour, fragrances and medicinal aroma. Vitex negundo
Linn. belonging to Verbenaceae family is an important
herb with a broad spectrum of pharmacological activities,
medicinal properties and applications. Its essentional oil
extract has been analyzed elsewhere [1].
All parts of the Vitex negundo are used as medicine,
however, the leaves are specially considered to be the most
potent for the isolation of medicinal constituents. It has
been used for the treatment of eye-disease, inflammation,
leucoderma, and toothache, skin-ulcers, in catarrhal fever,
rheumatoid arthritis, gonorrhea, sinuses and bronchitis [2].
The main techniques to obtain essential oils from the
medicinal herbs are hydrodistillation (HD), steam
distillation, steam and water distillation, maceration,
expression. Among these techniques, HD has been the
most common method to extract the essential oils from
plants. The HD method has several drawbacks such as long
extraction time, high energy use and so on. Hence, in order
to increase the extraction yield, save energy and time
extraction, new approaches are improving. In recent years,
the use of microwave-assisted hydrodistillation (MHD)
method has been increasing, especially for extraction [3],
[4]. By using microwave energy, the materials reach their
boiling point rapidly, leading to short extraction or
distillation time and saving energy.
Based on using Vitex negundo as flavor and medicinal
products, the aim of this study is to compare and evaluate
HD and MHD for their effectiveness in the extraction of
essential oils leaves, and to determine and compare the
composition of the essential oil obtained by HD and MHD.
2. Experiment
2.1. Plant materials
The fresh plants of Vitex negundo were collected in
Binh Thuan province, Vietnam. One part of the healthy
matured leaves of V. negundo is thoroughly washed with
distilled water, dried by centrifuge, preserved in low
temperature fridge of 50C and finally cut into small pieces
of 3 mm before hydrodistillation. The second part is also
thoroughly washed with distilled water, shade dried in dust
free condition for 2 weeks and finally cut into small pieces
of 3 mm before hydrodistillation. Moisture content of the
sun-dried samples is measured by a moisture analyzer.
2.2. Hydrodistillation (HD)
The Vitex negundo leaves (300 g fresh sample versus
300 g dried sample) are placed in a 1L round bottom flask
and connected to a Clevenger-type apparatus. The
evaporation is condensed by a condenser combined with a
chiller at 10oC. Hydrodistillation is completed for 2hs. after
boiling. The Vitex negundo hydrodistillation oil is
collected after water separation, stored in a culture tube.
2.3. Microwave-assisted hydrodistillation (MHD)
First MHD of 300 g fresh Vitex negundo leave sample
is carried out without adding water in a 1L glass tube flask
put in a special MHD equipment (Milestone ETHOS X,
Italia). Second experiment used 300 g dried sample with
adding 250 ml distilled water in the same glass tube flask
and MHD equipment. Parameter setting is microwave
energy of 1800W, running time of 20 mins, condensation
temperature at 10oC. The Vitex negundo oil is collected
using separation funnel and stored in culture tube.
2.4. GC/MS analysis
GC/MS data is obtained on the Gas Chromatography-
Mass Spectrometry (GC/MS: SCION 456 equipped SQ
mass spectrometer) using RXi5-ms (30 m×0.25 mm, film
2 Nguyen Thi My Dung, Vo Thi Dieu Hoa, Do Thi My Lien, Phung Van Trung, Pham Hong Ngoc, Le Ngoc Hung
thickness 0.25 µm). The mass range is 50 to 500 amu.
Carrier gas is nitrogen at a linear flow rate of 1.5 ml/min;
injector volume for all samples is 0.1μl. Temperature
programming is from 50ºC to 280ºC. Column oven
temperature is held isothermal at 50ºC for 3 minutes then
heated at 35ºC/min to 100ºC, again it is heated at 7ºC/min
to 220ºC, continue heat at 50C/min to 2800C and is held
isothermal for 3 minutes. The total program time of the
instrument is 34.57 min. The injector and detector
temperatures are 270ºC and 280ºC respectively. The oil is
injected neat with split injection mode having split ratio of
1:50. Quantitative results are mean data derived from GC
analysis. The final confirmation of constituents is made by
computer matching of the mass spectra of peaks with the
Wiley and NIST libraries mass spectral databases. Relative
amounts of individual components are based on GC peak
areas [5].
3. Results and discussion
Sensory properties and yield of Vitex negundo leaf
oils: The moisture content of all dried Vitex negundo leaves
after sun drying is around 13.8 %. The essential oils from the
Vitex negundo leaf obtained by (MHD) and classical
hydrodistillation (HD) are compared in terms of yield and
chemical composition. All the essential oils are pale yellow
liquids with strong characteristic odor and dried over
anhydrous Na2SO4. The HD dry leaves essential oil has
strongest odor. The essential oils obtained by HD give a
yield of 0.05 % and 0.35 % (w/w) on a dry weight basis from
fresh and dry leaves, respectively. When extracted by MHD
0.04 % and 0.30 % (w/w) on a dry weight basis are obtained
for fresh and dry leaves, respectively. Each of the extract is
stored in a sealed glass bottle in a refrigerator until analysis.
The data shows that MHD technique produces lower oil
yield in comparison to HD.
Chemical analysis of Vitex negundo leaf oils
The chemical composition of the essential oils achieved
from leaves Vitex negundo collected from two methods
(microwave-assisted hydrodistillation and hydrodistillation)
are represented together with the retention time in Table 1.
The GC–MS analyses of four samples reveal the presence of
a total of 30 components including monoterpenoids
(44.78%, 43.50%, 45.25% and 53.62), sesquiterpenoids
(51.50%, 44.40%, 43.46% and 40.48) and diterpenoids
(3.72 %, 10.45%, 8.22% and 5.00) from MHD (fresh and dry
leaves) and HD (fresh and dry leaves), respectively. This
result shows the major component of essential oil obtained
by MHD is sequiterpenes, but monoterpenoids is the main
compounds in essential oil obtained by HD method.
From Table 1, it is clearly found that there are
significant differences in the essential oils composition
isolated by two methods (HD and MHD). The essential oil
using MHD for dry leaves detects 30 compounds and then
22 compounds for dry and fresh leaves, while 27 and 29
compounds are detected in HD method for dry and fresh
leaves, respectively.
According to results in current study, the major
compounds are found to be β-caryophyllen (23.5%,
16.3 %, 16.4%, and 16.8%), eremophilene (18.9%, 15.1%,
14.4%, and 14.2%), eucalyptol (16.2%, 16.3%, 13.6% and
19.6%), α-terpinyl acetate (10.8%, 7.6%, 9.2% and 8.8%),
and sabinene (7.3%, 8.6%, 8.5%, and 10.3%) in oils
obtained by MHD, HD from fresh leaves, MHD and HD
from dry leaves, respectively. The major compound in
essential oil obtained by MHD from fresh leaves and HD
from fresh leaves is β-caryophyllen, but eucalyptol is the
major compound in essential oil obtained by HD from dry
leaves. Moreover, β-caryophyllen is the highest (23.5%) in
oils extract from fresh leaves by MHD method. It has
shown anti-bacterial activities by similar chemical
composition [1] and anti-inflammatory [6] and anesthetic
[7] effects.
While the total number of compounds in fresh leaves
essential oil achieved from MHD is less than that from HD
and the oil yield is lower than HD method, the total number
of compounds in dry leaves essential oil from MHD is
more than from HD method. MHD method is important in
terms of saving energy and extraction time (20 min
compared to 120 min with HD method) and the essential
oil with higher content of monoterpenes exhibits better
antibacterial activities [8].
Figure 1. Total ion chromatogram (obtained by GC-MS
analysis) of the Vitex negundo from fresh leave essential
oil extracted by HD method
Table 1. The retention times and chemical composition of
essential oils of Vitex negundo
No RT* Compound
%
MHD
Fresh
HD
Fresh
MHD
Dry
HD
Dry
1 4.68 3- carene 0.88
±0.21
1.87
±0.20
1.64
±0.20
2.32
±0.20
2 5.44 sabinene 7.34
±0.89
8.56
±0.75
8.53
±0.91
10.34
±0.85
3 5.53 (-)-β-pinene - 0.67
±0.20
0.70
±0.29
0.84
±0.31
4 5.75 β-myrcene - 0.51
±0.12
0.57
±0.19
0.57
±0.05
5 6.43 α-terpinene - - 0.35
±0.11
0.52
±0.17
6 6.79 β-phellandrene 3.74
±0.37
3.73
±0.32
3.88
±0.43
4.33
±0.35
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 3
7 6.86 eucalyptol 16.21±
0.56 16.29
13.57
±0.44
19.57
±0.63
8 7.25 β-ocimene 0.63
±0.11
0.44
±0.23
0.69
±0.20
0.62
±0.32
9 7.61 γ-terpinene 0.71
±0.73
0.48
±0.94
0.73
±0.76
0.98
±0.84
10 8.90 β-linalool 0.70
±0.39
0.45
±0.32
0.62
±0.28
0.72
±0.34
11 11.56 δ-terpineol 0.57
±0.17
0.48
±0.20
0.36
±0.05
0.53
±0.12
12 12.02 (-)-terpinen-4-
ol
2.04
±0.20
1.95
±0.07
1.26
±0.17
2.07
±0.12
13 12.55 terpineol 1.16±
0.78
1.26
±0.67
0.88
±0.52
1.37
±0.72
14 15.43 lavandulol
acetate -
0.43
±0.03
0.46
±0.10 -
15 16.73 α-terpinyl
acetate
10.81
±1.01
7.55
±0.85
9.24
±0.92
8.83
±0.83
16 18.28 β-caryophyllen 23.50
±0.86
16.33
±0.55
16.42
±0.34
16.79
±0.65
17 18.95 α-caryophyllen 1.16
±0.52
0.72
±0.47
0.97
±0.77
0.73
±0.44
18 19.70 eremophilene 18.92
±0.39
15.09
±0.20
14.37
±0.25
14.20
±0.18
19 19.86 (+)-
bicyclogemacrene
2.34
±0.20
1.38
±0.12
2.05
±0.09
1.50
±0.21
20 21.65 caryophyllene
oxide
0.54
±0.19
1.26
±0.16
0.75
±0.34
0.50
±0.42
21 23.07 α-cadinol 0.55
±0.42
1.14
±0.54
1.24
±0.39
0.90
±0.57
22 27.62 cambrene 0.69
±0.77
0.87
±0.64
1.41
±0.86
0.56
±0.43
23 28.78 widdrol 4.50
±0.20
6.68
±0.89
8.59
±0.75
5.86
±0.38
24 28.94 geranyl-α-
terpinene
0.66
±0.23
1.11
±0.20
1.73
±0.15
0.88
±0.26
25 29.14 α-guainene - 0.88
±0.03
0.91
±0.12
0.50
±0.03
26 29.47 epimanool - 0.55
±0.01
0.48
±0.02 -
27 29.60 cis-3,14-
clerodadien-13-ol
0.53
±0.54
2.41
±0.81
2.10
±0.72
1.20
±0.71
28 30.26 kaur-15-ene 1.85
±0.98
1.97
±1.20
4.33
±1.01
2.37
±1.00
29 30.86 phenanthrene - 1.71
±0.54
0.73
±0.51
0.40
±0.47
30 31.94 kolavelool - 0.43
±0.30
0.40
±0.41 -
Monoterpenoids 44.78 44.66 43.50 53.62
Sesquiterpenoids 51.50 42.59 44.40 40.48
Diterpenoids 3.72 7.33 10.45 5.00
Total percentage 94.44 83.37 85.74 90.28
Yeild (%) (w/w) 0.04 0.05 0.3 0.35
Extraction time (mins) 20 120 20 120
*RT: Retention time
4. Conclusion
With MHD method, time extraction is significantly
shorter than with HD method. MHD results in a reduced
extraction time and a substantial energy saving compared
to the conventional HD technique. After 20 minutes of
MHD extraction, it is possible to collect almost all the
existing essential oils of the Vitex negundo leaves.
However, the essential oils achieved from the two
methods have a strong characteristic odor and the essential
oil obtained by HD from dry leaves is the strongest. This
probably causes headache when we smell it for a long time.
In the future, the study on the composition contributing to
this strong characteristic odor will continue.
REFERENCES
[1] Khokra S. L., Prakash O., Jain S., Aneja K.R. and Yogita D., Essential Oil Composition and Antibacterial Studies of Vitex
negundo Linn. Extracts, Indian Journal of Pharmaceutical Sciences,
70(4), 522-526, (2008).
[2] Ladda PL. and Magdum CS, Vitex negundo Linn.: Ethnobotany,
Phytochemistry and Pharmacology- A Review, International
Journal of Advances in Pharmacy, Biology and Chemistry, 1(1),
111-120, (2012).
[3] Golmakani M. T., Rezaei K., Comparison of microwave-assisted
hydrodistillation with the traditional hydrodistillation method in the
extraction of essential oils from Thymus vulgaris L. Food Chem 109, 925–930, (2008).
[4] Khanavi M, Hajimehdipoor H, Emadi F, Kalantari Khandani N.,
Essential oil compositions of Thymus kotschyanus Boiss. Obtained
by hydrodistillation and microwave oven distillation. J. Essent Oil
Bear Plants 16,117–122, (2013).
[5] Adams R. P., Identification of essential oil components by gas
chromatography/ mass spectroscopy. Allured Publishing Corporation, Carol Stream. (1995).
[6] Marin S., Padilla E., Ocete M.A., Galvez J., Jimenez J., Zarzuelo A.,
Anti-inflammatory activity of the essential oil of Bupleurum
fruticescens, Planta Med. 59(6), 533-536, (1993).
[7] Ghelardini C., Galeotti N., Di Cesare Mannelli L., Mazzanti G.,
Bartolini A. Local anaesthetic activity of β-caryophyllene, Farmaco.
56, 387-389, (2001).
[8] Medeiros J. R., Campos LB, Mendonça SC, Davin LB, Lewis NG.
Composition and antimicrobial activity of the essential oils from invasive species of the Azores, Hedychium gardnerianum and
Pittosporum undulatum. Phytochemistry, 64,561-565, (2003).
(The Board of Editors received the paper on 03/4/2018, its review was completed on 26/4/2018)
4 Hoang Ngoc Ha
A COMPARATIVE ANALYSIS OF PASSIVITY-BASED CONTROL
APPROACHES WITH APPLICATION TO LINEAR DYNAMICAL SYSTEMS
Hoang Ngoc Ha
Duy Tan University; [email protected]
Abstract - This study focuses on linear dynamical systems whose dynamics are affine in the control input. Such dynamics are extensively considered to be rewritten into a canonical form, namely the passive port-Hamiltonian representation in order to further explore some structural properties such as interconnection and damping matrices, Hamiltonian storage function and dissipation term. On this basis, the passivity-based control design approaches including proportional controller and energy shaping controller are proposed for the purpose of stabilization. Interestingly, the energy shaping controller seems to be better since the controller gain accepts a larger domain of validity and can even be negative. A mass-spring-damper system is used to illustrate the proposed approach. Besides, numerical simulations are included in both the open loop and closed loop to compare the results.
Key words - Port-Hamiltonian representation; modeling; passivity-based control; proportional controller; energy shaping controller.
1. Introduction
This paper deals with the port-based modeling of
general nonlinear dynamical systems [1–3] whose
dynamics are described by a set of Ordinary Differential
Equations (ODEs) and affine in the input u as follows:
( ) ( ) ; ( 0) init
dxf x g x u x t x
dt= + = = (1)
where ( )x x t= is the state vector in the operating region
nD ; ( ) nf x expresses the smooth (nonlinear)
function with respect to the vector field x . The input-state
map and the control input are represented by ( ) nxmg x
and mu , respectively. It is worth noting that many
industrial applications of electrical, mechanical or
biochemical engineering belong to this kind of systems [4–7].
Many control methodologies have been developed for
the stabilization of the system (1) at a desired set-point *x .
This is for instance the case for sliding mode control,
adaptive control and model predictive control, etc. to cite a
few. Recently, passivity-based control (PBC) methodology
which is recognized as an extension of Lyapunov approach
has attracted much attention from researchers and
practitioners. In the PBC framework, it is always important
to transfer the (original) dynamics (1) to the
port-Hamiltonian (pH) representation prior to developing
state feedback laws for control [6, 8]. The control design and
control scenarios proposed for the simulations are main
contributions of this work.
This paper is organized as follows. Section 2 gives a
brief overview of the pH representation of (affine)
dynamical systems. Section 3 is devoted to the case study
of a mass-spring-damper system modeled within pH
framework. Passivity-based control designs (including
proportional controller and energy shaping controller) and
comparative simulations are given in Section 4. Section 5
ends the paper with some concluding remarks.
Notations: The following notations are considered
throughout the paper:
• is the the set of real number.
• is the matrix transpose operator.
• m and n ( m n ) are the positive integers.
• *x is the set-point.
• initx is the initial value of the state vector.
2. The passive port-Hamiltonian (pH) representation
Assume that if the function ( )f x verifies the so-called
separability condition [7, 9], that is, ( )f x can be
decomposed and expressed as the product of some
(interconnection and damping) structure matrices and the
gradient of a potential function with respect to the state
variables, i.e., the co-state variables:
( ) ( ) ( )( )H x
f x J x R xx
= −
(2)
where ( )J x and ( )R x are the n n skew-symmetric
interconnection matrix (i.e., ( ) ( )J x J x
= − ) and the
n n symmetric damping matrix (i.e., ( ) ( )R x R x
= ),
respectively while ( ) : nH x → represents the
Hamiltonian storage function of the system (possibly
related to the total energy of the system). Furthermore, if
the damping matrix ( )R x is positive semi-definite,
( ) 0R x (3)
Then, the original dynamics described by (1) is said to be
a pH representation with dissipation [4, 5]. Equation (1) is
then rewritten as follows:
( ) ( )( )
( )
( )( )
H xx J x R x g x u
x
H xy g x
x
= − +
=
(4)
where y is the output.
It can be clearly seen for the pH models defined by (3)
(4) that the time derivative of the Hamiltonian storage
function ( )H x satisfies the energy balance equation [5]
below:
( ) ( )( )
( )dH x H x H xR x u y
dt x x
= − +
(5)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 5
Thanks to (3), (5) becomes:
( )
supplied power
stored power
TdH x
u ydt
(6)
From a physical point of view, inequality in (6) implies that
the total amount of energy supplied from external source is
always greater than the increase in the energy stored in the
system. Also, equality in (6) holds only if the damping
matrix ( )R x , which is strongly related to the dissipation
term, is equal to 0. Hence, the pH system (4) is said to be
passive with input u and output y corresponding to the
Hamiltonian storage function ( )H x [3]. This is one of
advantageous features of the pH representation and has
been applied to the control design, even for the stabilization
of infinite dimensional systems (see, e.g. [10, 11]).
We shall not elaborate any further on the pH
representation here (for example, the concepts related to
the cyclopassive/ passive property or Dirac structure, etc.)
and refer the reader to [4, 6, 9] for more information.
3. A case study: Mass-spring-damper system
To illustrate the concepts proposed in Section 2, we
illustrate our main points with a simple case study, which
is the mass-spring-damper system. Originally, the port
Hamiltonian representation has been first considered for
electrical or mechanical systems as seen in the literature
(see, e.g. [2, 12]).
Buildings or suspension structure of a vehicle traveling
over a bumpy road can be modeled as a mass-spring-damper
system in a vertical position1 as shown in Figure 1 [12].
Figure 1. A mass-spring-damper system
The following equation is derived using Newton's
second law [14]2:
( )( )
( )2
2
d z t dz tM F kz t c
dtdt= − − (7)
where:
• M is the mass of the body;
• F is the external force;
• k is the stiffness constant of the (linear) spring;
1 Fixed-base or base-excited configuration [13] can be handled similarly. 2 This belongs to the so-called (generalized) Euler-Lagrange equations of classical mechanics [2]. Equation (7) is still valid even in the presence of
friction where the friction force is assumed to be proportional to the velocity.
• c is the damping constant.
Let x be the vector consisting of the position ( )z t and the
momentum of the body ( )
,dz t
Mdt
i.e.
( ) ( )( )
1 2, , ,dz t
x x x z t Mdt
=
(7) can be rewritten as
follows:
11
22
0 1 0
1 1
dxkx
dtFx
dx cM
dt
= + − −
(8)
The system dynamics (8) lead to a pH representation (4)
with:
( )0 1
1 0J x
=
− (9)
( )0 0
0R x
c
=
(10)
( )0
,1
g x u F
= =
(11)
( )2
dz txy
M dt= (the velocity) (12)
and, ( )2
2 2
1
1 1
2 2
xH x kx
M= + (13)
In this case, the Hamiltonian storage function ( )H x (13)
is equal to the total energy of the system, (i.e., it
characterizes the amount of the elastic potential energy of
the spring and the kinetic energy of the body, respectively).
It therefore has the unit of energy. The damping matrix
( )R x (10) is symmetric and positive semi-definite.
Remark 1. The system states 1x and 2x converge to the
nonzero values at steady state (i.e., *
1
Fx
k= and
*
2 0x = ) if
F is different from 0.
Remark 2. As an analogy between mechanical and
electrical systems [15], it is worth noting that a second
order ordinary differential equation of the series RLC
circuit operated under a voltage source ( )V t can be written
as follows:
( ) ( )( )
( )2
2
1d i t di t dV tL R i t
dt C dtdt+ + = where ( )i t is the
electric current. This is clearly equivalent to (7) in some
sense.
6 Hoang Ngoc Ha
4. Passivity-based control design and simulations
4.1. The proportional controller design
Let us state the following proposition.
Proposition 1. Under a zero state detectability condition3
and the boundedness from below of the Hamiltonian storage
function ( )H x by 0, it follows that an explicit proportional
static output feedback law of the form
pu K y= − (14)
with y given by (12) and 0pK a so-called damping
injection gain, renders the controlled pH system (4) with
(9)–(13) dissipative and therefore asymptotically stabilized
at the (singular) equilibrium ( )* 0,0x
= 4.
Proof. From (6) and (14), one obtains:
( )0p
dH xy K y
dt
The proof is followed immediately by invoking La
Salle's invariance principle [1, 6]. A complete version of
the proof can be found in [11].
Remark 3. The convergence speed of the controlled
system goes faster by increasing the controller gain pK .
Better performance of the controller can be proposed with
the gain pK which is derived from the Ziegler-Nichols
tuning method.
4.2. The energy shaping controller design
Proposition 2. A state feedback law
*
1 1
1 1u x k x
K K
= − + +
(15)
with 1
0kK
+
asymptotically stabilizes the system (4)
with (9)–(13) at the (nonsingular) equilibrium ( )* *
1 ,0x x
= .
Proof. From (15), we consider 1
1
( )adH xu
dx− which
leads to 2 *
1 1 1 1
1 1( )
2aH x x k x x
K K
= − +
. On the other hand,
let ( )dH x be the (closed-loop) Hamiltonian storage
function, i.e., 1( ) ( ) ( )d aH x H x H x= + 5. We can easily
check that the function ( )dH x admits a global minimum
at ( )* *
1 ,0x x= since its Hessian matrix
10
01
0
kK
M
+
3 This condition is a version weaker than the observability condition. 4 If a nonzero equilibrium ( )* *
1 ,0x x= is considered as desired set-point, the proposed result can also be deduced similarly using a coordinate
transformation given by *x x x= − .
5 Hence, the open loop Hamiltonian storage function has been shaped by 1( )aH x to become ( )dH x .
6 It can be shown that the damping factor 1
2
c
kM equals 0.2. The open loop system is therefore underdamped.
and the time derivative
2
2( )
0ddH x xc
dt M
= −
. The latter
completes the proof.
4.3. Numerical simulations
The simulations are carried out using MATLAB &
SIMULINK. The model parameters are given with
0.25k = (N/m), 0.5c = (N/(m.s)) and 6.25M = (kg)
(see also [13])6. The input force imposed on the system is
a unit step, i.e., ( ) ( )u t S t= where ( )S t is the unit step
function. The initial conditions are chosen to be
1 1,( 0) 3initx t x= = = and 2 2,( 0) 0initx t x= = = . Figure 2
shows the time evolutions of the states and the storage
function. It is shown that the storage function is bounded
from below by a positive scalar since steady states are
different from 0 (see (13) and Remark 1).
Figure 2. The states and storage function w.r.t. time
In what follows, without loss of generality we propose
to stabilize the system at the natural equilibrium
( )* 0,0x
= . The closed loop simulations with the
proportional feedback law (14) are implemented with
0.1pK = which is derived from the Ziegler-Nichols tuning
method. On the other hand, the stability condition for the
gain K of the energy shaping controller (15) is either
0K or 14K
k − = − . Let us choose 10K = − so that
1pK
K= for the purpose of comparisons. Since the
controlled Hamiltonian storage functions converge to 0 as
t →+ (Figure 3), it implies that the system (4) with
(9)–(13) is globally stabilized with (14) and (15) (see the
closed loop phase plane in Figure 4). In both cases, the
global convergence of the controlled states x to x* is
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 7
guaranteed. However, the exponential spiral orbit via (15)
is embedded by the orbit generated by (14) (i.e., it seems
to be better for the practical implementations). In addition,
the manipulated inputs u given by (14) and (15) are
physically admissible in terms of amplitude and dynamics
as seen in Figure 5.
Figure 3. The controlled Hamiltonian storage function
Figure 4. The closed loop phase plane
Figure 5. The manipulated input w.r.t. time
5. Conclusion
In this paper, a mass-spring-damper system is used to
introduce the so-called port-Hamiltonian representation. In
this presentation, some structural properties such as
interconnection and damping matrices, Hamiltonian storage
function and dissipation term are highlighted from a physics
point of view. The feedback designs (including proportional
controller and energy shaping controller) and control
scenarios proposed for the comparisons are main
contributions of the paper. Interestingly, the energy shaping
controller seems to be better for the practical implementations
since the controller gain accepts a larger domain of validity
and can even be negative. It remains now to adapt the
proposed results to nonlinear multiphysics multiscale systems.
REFERENCES
[1] H.K. Khalil, Nonlinear systems. Prentice Hall, Upper Saddle River,
3rd edition, 2002.
[2] R. Ortega, A. Loría, P.J. Nicklasson, and H. Sira-Ramírez, Passivity-
based control of Euler-Lagrange systems: Mechanical, electrical and
electromechanical applications. Springer London, 1st edition, 1998.
[3] A. Van der Schaft, L2-gain and passivity techniques in nonlinear
control. Springer, 3rd edition, 2017.
[4] B. Maschke, R. Ortega, and A. Van der Schaft, “Energy-based
Lyapunov functions for forced Hamiltonian systems with dissipation”, IEEE Transactions on Automatic Control, 45(8), pp.
1498-1502, 2000.
[5] R. Ortega, A. Van der Schaft, I. Mareels, and B. Maschke”,Putting
energy back in control”, IEEE Control Syst.Mag., 21(2), pp. 18-33, 2001.
[6] R. Ortega, A. Van der Schaft, B. Maschke, and G. Escobar,
“Interconnection and damping assignment passivity-based control
of port-controlled Hamiltonian systems”, Automatica, 38(4), pp. 585-596, 2002.
[7] M. Guay, and N. Hudon, “Stabilization of nonlinear systems via
potential-based realization”, IEEE Transactions on Automatic
Control, 61(4), pp. 1075-1080, 2016.
[8] T.S. Nguyen, N.H. Hoang, and M.A. Hussain. “Feedback
passivation plus trackingerror-based multivariable control for a class of free-radical polymerization reactors”, Int. J. of Control, 2018.
https://doi.org/10.1080/00207179.2017.1423393
[9] H. Hoang, D. Dochain, F. Couenne, and Y. Le Gorrec, “Dissipative
pseudo-Hamiltonian realization of chemical systems using
irreversible thermodynamics”, Mathematical and Computer Modelling of Dynamical Systems, 23(2), pp. 135-155, 2017.
[10] A.A. Alonso, and B.E. Ydstie, “Stabilization of distributed systems
using irreversible thermodynamics”, Automatica, 37(11), pp. 1739-
1755, 2001.
[11] N.H. Hoang, and D.T. Phan, “Nonlinear control of temperature profile
of unstable heat conduction systems: A port-Hamiltonian approach”,
Journal of Computer Science and Cybernetics, 32(1), pp. 61-74, 2016.
[12] C. Batlle, “Passive control theory I and II”, II EURON/GEOPLEX
Summer School on Modeling and Control of Complex Dynamical Systems, July 18-22, 2005, Bertinoro, Italy.
[13] R.G. Longoria, “Modeling and experimentation: Massspring-
damper system dynamics”, Department of Mechanical Engineering,
The University of Texas at Austin, July 20, 2014.
[14] M.W. McCall, Classical mechanics: From Newton to Einstein - A
modern introduction. Wiley, 2nd edition, 2010. ISBN:
9780470715741
[15] F.A. Firestone, “A new analogy between mechanical and electrical
systems”, The Journal of the Acoustical Society of America, 4(3), pp. 249-267, 1933.
(The Board of Editors received the paper on 29/01/2018, its review was completed on 11/5/2018)
8 Le Kim Hung, Vu Phan Huan
PERFORMANCE ANALYSIS AND ASSESSMENT OF A TRANSFORMER
DIFFERENT PROTECTION RELAY SEL387 AT 110KV LANG CO SUBSTATION
Le Kim Hung1, Vu Phan Huan2 1University of Science and Technology – The University of Danang; [email protected]
2Center Electrical Testing Company Limited; [email protected]
Abstract - Based on the influences of current transformer connection type, CT errors, magnetizing inrush current, errors because of tap changing and fault conditions on differential protection function, the paper establishes and assesses the performance of a numerical relay SEL387 model concerning the protection of the 115/24kV transformer at Lang Co Substation by Matlab/Simulink. The paper also calculates the setting value of two actual slope characteristics (O87P = 0.3, U87P = 10, SLP1 = 25%, SLP2 = 50% and IRS1 = 3). The results can be applied to increase the accurate and reliable performance of the differential transformer protection relay against internal faults. Simulation has simplified the process of selecting relay and protection system. This can improve the quality of the protection system design early, thereby reducing the number of errors found later in the operation.
Key words - Different protection relay; transformer; two slope characteristics; Matlab/Simulink; SEL387.
1. Introduction
Nowadays, there are a variety of numerical transformer
different protective relays on the market such as Siemens
7UT613, SEL387, Schneider P632, Toshiba GRT200,
ABB RET670, which include many functions in one unit,
as well as providing metering, communication, and
transformer protection. These protective relays help us to
simplify implementation of the protection in circuit design
and setting calculations.
The connection diagram is used for the numerical
protective relay SEL387 (shown in Figure 1) that provides
protection of two transformer windings (HV, LV) as well
as differential function (F87T) for sensitive detection of
inter turn faults within the transformer winding. Both HV
CTs and LV CTs are wye connected. The F87T obtains
three phase current inputs from them. This function
compares the currents entering and leaving the protected
zone of the windings of the transformer.
Figure 1. Secondary current in HV and
LV side at normal condition
As with most false trips involving F87T in Central
Power Grid, checking the relay should be done first, and
can be done by inspecting the LED indicators, cable
connections, auxiliary relay and so on. The main cause
shown in Figure 2 is CT secondary wire connected to the
incorrect tap on the CT, Crossed phases, Incorrect CT
polarity in design or construction [2]. In addition, there is a
general lack of understanding the ground differential
protection principle. In most cases, inadequate or no
verification test is performed to check the correctness of the
secondary current circuit. So, these errors depend on skill of
testers. If the hardware has no issues, then it is very likely to
be a setting problem. Unlike hardware issues, setting issues
cannot be assessed by the naked eyes, so the universal relay
test set and commissioning tool are required. It performs to
check wiring and setting of relays, by using
primary/secondary injection of currents from the test set.
Figure 2. Incorrect CT Ratio, CT Polarity, or Crossed Phases
But even in a no fault situation, the magnitude and the
phase of the currents in both sides of the transformer will
not have the same value. This is often the case that mis-
operation on relay does not become apparent immediately.
One possibility is CT errors, magnetizing inrush current
during initial energization, CTs mismatch and saturation.
Another possibility is that the transformation ratio changes
due to Tap changer. These have already been introduced to
the different currents by devices that have not yet caused
any problems, but will cause significant disruption to the
transformer in the future. Besides, the possibility is that
someone with unauthorized access infiltrates the relay and
reconfigures incorrect setting to a relay, instructing it to
release a false trip signal without the existence of any fault.
When these types of mis-operation risks go undetected, it
is very easy for substation operators to mistakenly believe
that their relay protection is secure. The question substation
operators need to ask is, “How confident am I that my relay
protection is reliable and secure?”
After this introduction, the rest of the paper is organized
as follows. The section 2 describes the transformer
different protection function and provides instructions for
setting calculation SEL387 in Lang Co substation. The
section 3 builds the power system and the relay protection
on Matlab Simulink. The section 4 simulates the testing
normal/fault conditions. The section 5 gives the
conclusions related to the transformer different protection.
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 9
2. Transformer different protection function
The main operation of a current differential protection
relay is made by comparing the vector current in both sides
of the transformer: IDIFF = |I1 + I2| (1)
Restraint current:
IBIAS = |I1| + |I2| (Siemens, Sel, Abb) (2)
IBIAS = 0.5(|I1| + |I2|) (Schneider) (3)
a. Siemens 7UT613 characteristic b. SEL387 characteristic
c. ABB RET670 characteristic d. Schneider P632 characteristic
Figure 3. Differential protection characteristic
Based on these values of IBIAS and values of IDIFF, the
trip/restrain characteristics offer from vendors of
protection relay such as Siemens, Sel, Abb, and Schneider
which has a three-step shape (two slopes and one pickup
level) as in Figure 3 and defined by the following settings
as in Table 1. The F87T operates when IDIFF exceeds a
minimum operate current threshold and a percentage of
IBIAS, defined by a slope setting (slope 1, slope 2). Consider
matlab code of this matter in subsection 3 below.
Table 1. Parameter characteristic of relay vendors
Parameter Siemens Sel Abb Schneider
Minimum
pickup IDIFF> 87OP Idmin IS1
Slope 1 Slope 1 Slope 1 Slope Section 2 K1
Slope 2 Slope 2 Slope 2 Slope Section 3 K2
Unrestraint
tripping IDIFF>> 87UP ID>>
Points of
intersection Base Point 2 IRS1
End Section1,
End Section 2 IS2
To help us understand a setting calculation for relays,
we use SEL 387 to protect a 25MVA, 50Hz, (115/24) kV,
Y/y0 Vina Takaoka transformer in Lang Co Substation that
has CTHV = 200/1, CTLV = 800/1, and an OLTC with
tapping range from 1 to 19 positions. It has satisfied the
following requirements from Decision No. A3-06-
2015/LCO110 by the Central Region Load Dispatch
Centre of Vietnam [3, 6].
Windings 1 and 3 are validated for differential
protection. Settings will be: E87W1 = Y, E87W3 = Y.
The voltages for winding 1 and 3 are 115kV and 24kV,
respectively: VWDG1 = 115, VWDG3 = 24.
The internal compensation (ICOM = Y) for the
wye-wye transformer with a wye-wye CT can be set to
12 to remove the zero sequence currents.
Winding 1 CT Conn.Compensation W1CTC =12
Winding 3 CT Conn.Compensation W3CTC =12
The following settings refer to the CTs connection and
to the current ratio for each winding: W1CT = Y;
CTR1 =200; W3CT = Y; CTR3 = 800.
The secondary current of CT HV side under normal
operating condition is:
IHV = MVA/(1.732 × VWDG1× CTR1)
IHV = 25×106/(1.732×115×103×200) = 0.628 [A]
and requires ratio compensation TAP1 = 1/0.628 = 1.593
Under normal condition, secondary current in LV side is:
ILV = MVA/(1.732 ×VWDG3×CTR3)
ILV = 25×106/(1.732×24×103×800) = 0.752 [A]
and requires ratio compensation TAP3 = 1/0.752 = 1.33
a. Setting characteristic b. Test point in
Normal/ Fault conditions
Figure 4. SEL 387 setting characteristic in Lang Co Substation
As shown in Figure 4a, dual slope characteristics can
be used with a minimum pickup setting. This can be
mathematically represented as follows:
The minimum pickup O87P should be set as sensitively
as possible while considering the steady state CT error and
transformer magnetizing current. The O87P setting must
yield an operating current value of at least 0.1×IN, at the
least tap. In this case O87P ≥ 0.1IN/TAPMIN = 0.1x1/
1.33 = 0.0752. The typical O87P range is 0.3 to 0.5.
Therefore, the O87P setting of 0.3 is valid.
The instantaneous unrestrained current element is
intended to react quickly to very heavy current levels that
clearly indicate an internal fault. Set the pickup level (U87P)
about 10 times TAP. The unrestrained differential element
only responds to the fundamental frequency component of the
differential operating current. It is not affected by the SLP1,
SLP2, IRS1, PCT2, PCT5, or IHBL settings. Thus, it must be
set high enough so as not to react to large inrush currents.
Slope 1 region is used between the minimum pickup
region and the slope 2 breakpoint. Slope 1 provides
security against false tripping due to the following factors:
Excitation current = 2 %, CT accuracy = 3%, NLTC = 5%,
LTC = 10%, Tap mismatch = 0%, and Relay
accuracy = 5%. All these percentages sum to 25 %, thus a
setting of SLP1 = 25% can be used.
IDIFF/IN
IBIAS/IN
Idmin
End Section 1
Slope Section 2
End Section 2
Slope Section 3
IDIFF/IN
IBIAS/IN I
K1
IS2
K2
TAP Pos is 18
IDIFF/IN
IBIAS/IN
O87P
IRS1
CT Error
U87P
TAP Pos is 9
External Fault
Internal Fault
Magnetization
SLP1 = 0.25
SLP2 = 0.5
IDIFF/IN
IBIAS/IN
O87P
Sum
IRS1 = 3
Saturation
TAP
changer
U87P
10 Le Kim Hung, Vu Phan Huan
Slope 2 is used to prevent false tripping caused by
saturation of the CTs. A setting of SLP2 = 50 % for slope
2 covers all the situation.
IRS1 = 3 is restraint current slope 1 limit.
Operate time (restrained function): 20 to 35 ms.
Operate time (unrestrained function): 5 to 20 ms.
PCT2 = 15% (the F87T is going to be blocked if the
second harmonic is higher than 15% from fundamental).
PCT5 = 35% (the F87T is going to be blocked if the
fifth harmonic is higher than 35% from fundamental) and
TH5P = OFF (the 5th harmonic alarm is deactivated).
3. Building of the differential protection function using
Matlab Simulink
For the purpose of testing reliability of the relay
protection from SEL vendor to test the algorithm of
different protection, the power system model has been
simulated in the Matlab Simulink and it is depicted in
Figure 5. It consists of a 115 kV, 1000 MVA system, a
25MVA, 50Hz, (115/24) kV, Y/y0 OLTC regulating
transformer, two CT (200/1A and 800/1A), 24 MW /1Mvar
loading and SEL 387 relay protection. All fault conditions
are created to transformer via the three phase fault block.
A relay SEL387 model shown in Figure 6 combines
functions of vector group compensation, TAP factor
compensation, different and bias current calculation,
inrush harmonic blocking and slope characteristics. Firstly,
current signals have been simulated in Matlab, which
combines vector group value of current throw S-function,
which is used to correct the phase shift across the YNy0
transformer. Since the HV, LV side of the transformer are
wye connected, they require and will use the same Ɵ = 00.
The identity matrix is shown below:
)cos()120cos()120cos(
)120cos()cos()120cos(
)120cos()120cos()cos(
3
2)(0
00
00
00
−+
+−
−+
=CTC
C
B
A
C
B
A
COMPC
COMPB
COMPA
I
I
I
I
I
I
CTC
I
I
I
−−
−−
−−
==
211
121
112
3
1)0(0 0
_
_
_
After that it sends to the subsystem combined TAP
factor compensation (TAP1 = 1.59, TAP3 = 1.32).
Secondly, the subsystem different value and bias of current
are calculated for each phase separately according to the
relation of the equation (1) and (2). Similarly, in the
harmonic subsystem the F87T is going to be blocked if the
second harmonic is higher than 15% from fundamental.
Finally, S-Function Builder checks the position of operating
point described by currents (for each phase separately) with
respect to the pick-up characteristic IDIFF = f(IBIAS) and
decision tripping the pulse, which opens a circuit breakers
located on both sides of the protected transformer. The
following Matlab code is written for phase A:
Figure 5. Matlab/Simulink Model of the proposed system
Figure 6. Overview of the function blocks of the F87T
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 11
double IRT = (*O87P) / (*SLP1);/* starting point of
SLP1*/
double IRT2 = (*U87P) / (*SLP2); /* ending point of
SLP2*/
if ((*Ibias_A <= IRT) && (*Harmonic_Block == 0))
{
if (*Idiff_A > (*O87P))
*Output = 1;
else
*Output=0;
}
else
{
if (*Idiff_A > (*U87P))
*Output = 1;
else
*Output=0;
}
if ((*Ibias_A > IRT) && (*Ibias_A <= *IRS1) &&
(*Harmonic_Block == 0))
{
if (*Idiff_A > ((*Ibias_A) * (*SLP1)))
*Output = 1;
else
*Output=0;
}
if ((*Ibias_A > *IRS1) && (*Ibias_A <= IRT2) &&
(*Harmonic_Block == 0))
{
if (*Idiff_A > ((*Ibias_A) * (*SLP2)))
*Output = 1;
else
*Output=0;
}
4. Simulation results and discussion
The main goal of the simulation is either to obtain or
calculate waveforms such as currents on both sides of
transformer, TAP position, voltage at bus C41, IDIFF, IBIAS,
trip signal, and harmonic block signal during normal/fault
conditions for analysis of the behavior of relay.
4.1. Case.1. Normal Condition
When the transformer is operating normally, TAP
position is 9 and the resulting voltage at bus C41 is
0.99pu. The differential currents in all the phases
(IDIFF = 0.028) are well below pick up value O87P = 0.3,
IBIAS = 1.35 and the relay does not issue any trip signal.
Figure 7 shows IDIFF and IBIAS in any one phase (phase
A) and relay output.
Figure 7. TAP position at 9
As the transformer taps further from the balance
position (9), i.e. TAP position is 18 and voltage at bus C41
is 0.855pu, so the magnitude of the different current
increases IDIFF = 0.14, IBIAS = 1.08. However, the
differential current is still smaller than 0.2, the relay will
not trip (shown in Figure 8).
Figure 8. TAP position at 18
There are other ways to increase IDIFF by the CT errors.
It makes secondary current on two sides, not equation
under healthy conditions; for example a 15VA - 5P20 CT
has a guaranteed error of less than ± 5% when it is
subjected to 20 times its nominal current and delivers into
its nominal load (15 VA to In). At TAP = 9, current in HV
side is (CT error +5%) and current in LV side is (CT error
-5%), then IDIFF = 0.11, IBIAS = 1.35. Relay does not issue
any trip signal as shown in Figure 9.
Figure 9. TAP position at 9, CTHV error +5%, CTLV error -5%
At TAP = 9, the transformer is energized from the HV
side, magnetizing currents appear due to its core
magnetization and saturation. Figure 10 shows the waveform
of a magnetizing inrush current with transformer energized at
0.1s. The IDIFF is = 0.88, IBIAS = 0.44 but the relay does not
issue any trip signal because harmonic blocking signal is = 1.
12 Le Kim Hung, Vu Phan Huan
Figure 10. Harmonic block for energization condition
4.2. Case.2. Fault Condition
There are various types of faults, such as single phase to
ground, double phase, double phase-to-ground, and three
phases. If a fault is detected, i.e. the start signals will be set
by the differential protection (the measured IDIFF > O87P),
and at the same time the internal/ external fault discriminator
will determine the relative phase angle between them.
Figure 11. Internal fault at phase A to ground
For an internal AG fault is performed on F1, it is located
within the differential protection zone. Therefore, the fault
currents will flow out from the faulty power transformer on
both sides. The fault currents on the HV and LV sides will
have the same direction as shown in Figure 11. In the figure
immediately after the fault is applied, we can observe that
fault current in HV side is increased enormously,
IDIFF is = 4.35, IBIAS = 3.1 and the trip signal occurs at 0.12s.
Figure 12. External fault at phase AB
For an external AB fault is performed on F2, it is
located in the LV side of the transformer model. The fault
current contributions from the HV and LV side are
180 degrees out of phase as shown in Figure 12.
IDIFF is = 0.106, IBIAS is = 3.76 and relay does not trip.
Reviews: By using numerical relays, problems like CT
ratio mismatches and phase shift compensation can be
solved mathematically in the software of the relay. Besides,
the test point results of relay SEL387 (shown in Figure 4b)
demonstrate the stable operation during cases of normal
conditions (CT error, the change in tap position of a power
transformer, and magnetizing inrushes), external fault and
higher sensitivity during internal faults.
5. Summary
This paper provides a detailed description of a
transformer different protection function based on a
two-slope characteristic. It also provides valuable tips on
how to guide the setting calculation and troubleshooting
process. Furthermore, the power system model simulates
numerous test cases for an existing power transformer of
Lang Co Substation, Viet Nam using Matlab/Simulink
software package. These test cases save time by
immediately indicating whether the issue has occurred on
the SEL387. As a result, protection engineers can easily
analyze the mis-operation to determine the root cause and
can fulfill very demanding requirements set by power
utilities.
REFERENCES
[1] Sandro Gianny Aquiles Perez, “Modeling relays for power system
protection studies”, the Degree of Doctor of Philosophy in the Department of Electrical Engineering University of Saskatchewan
Saskatoon, Saskatchewan Canada, July 2006.
[2] Casper Labuschagne and Normann Fischer, "Relay-Assisted
Commissioning", 59th Annual Conference for Protective Relay
Engineers College Station, Texas, April 4–6, 2006.
[3] SEL, “SEL-387 Relay Current Differential Instruction Manual”,
2010.
[4] Zoran Gajić, “Differential Protection for Arbitrary Three-Phase
Power Transformers”, Doctoral Dissertation Department of Industrial Electrical Engineering and Automation, Lund University,
SWEDEN, 2008.
[5] M. Tanveer Ahmad, “Differential Protection of Transformer using
Harmonic Restraint Circuitry”, The 12th GCC Cigre International
Conference, Doha, Qatar, 8-10 November, 2016.
[6] Decision No. A3-06-2015/LCO110 by the Central Region Load
Dispatch Centre of Vietnam.
(The Board of Editors received the paper on 31/01/2018, its review was completed on 26/02/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 13
A STUDY ON THE REDUCTIVE DECHLORINATION OF CHLOROFORM WITH
NANO Fe/Cu BIMETALLIC PARTICLES IN AQUEOUS SOLUTION
Phan Kim Nguyen, Bui Xuan Vung
University of Education – The University of Danang; [email protected]
Abstract - In this work, nano-Fe/Cu bimetallic particles are synthesized and used to reduce chloroform to methane in aqueous solution. The synthesized particles are characterized by X ray diffraction (XRD) pattern, Transmission electron microscopy (TEM) images and energy dispersive X ray (EDX) analysis. Such key parameters on the reduction of chloroform as pH, nano-Fe/Cu dosage, treatment time have been investigated. Closed batch experiments have been conducted for this investigation. Experimental results show that the de-chlorination of 50 mL of 20 ppm chloroform aqueous solution has the highest degradation efficiency of 88.93% under the experimental conditions such as pH = 3, reaction time of 30 minutes and nano-Fe/Cu dosage of 0.05 gram. GC-MS analysis for a 20 ppm chloroform aqueous solution before and after treatment has shown that there is no formation of such products containing chlorine as CH2Cl2 and CH3Cl.
Key words - Nano-Fe/Cu; bimetallic particles; chloroform; de-chlorination; degradation; aqueous solution.
1. Introduction
Trihalomethanes including mainly chloroform (CHCl3)
are disinfection by-products formed when using chlorine
for disinfecting drinking water [1] and treated wastewater
before it is conveyed into water distribution systems [2].
Chlorine is by far the most widely used chemical
disinfectant in water and wastewater treatment. These by-
products are linked to a direct health risk such as liver and
kidney cancer, nervous system and reproductive effects.
The recommended concentration value by WHO for
chloroform in drinking water is 0.3 mg/L [1].
Many technologies such as advanced oxidation, air
stripping, and physical adsorption have been applied to the
removal of chloroform in water [3-6]. A reductive system
with zero- valence iron and the reductive process coupled
with Fenton’s reagent were also used for such a purpose.
However, the destruction of chloroform requires additional
treatment [7].
Another efficient approach for degrading a variety of
contaminants is that using nano-Fe0 coated with another
metal such as Ag, Pd, Pt, Ni or Cu because the rate of
reduction by bimetallic particles is significantly faster than
those observed for Fe0 alone [8]. An investigation shows
that nano-Fe/Cu particles increase the rate of reduction
1,1,1-trichloroethane related to Fe/Ni combination and the
bimetals show a dramatically faster rate than Fe0 alone [9].
In this regards, Fe/Cu combination was chosen to degrade
chloroform in aqueous solution. In this study, nano Fe/Cu
particles are firstly synthesized and characterized and then
used for the investigation of effects on the removal of
chloroform from aqueous solution.
2. Experimental
2.1. Synthesizing nano-Feo and nano-Fe/Cu particles
To synthesize nano-Fe0 particles, two solutions of A
and B were respectively prepared by dissolving 4 gr of
FeSO4.7H2O (99%, China) into 50 mL of distilled water
and 0.4 gr of NaBH4 (99%, Merck) into 10 mL of distilled
water to form the solution that was then added with 10 mL
of 1% w/v starch solution. Solution B was added slowly in
the rate of 3-4 mL.min−1 to solution A at ambient
temperature and vigorous stirring. All aqueous solutions
removed dissolved oxygen by bubbling argon gas for 20
min. During this reaction, ferrous ion (Fe2+) was reduced
into black particles by sodium borohydride reductant in the
following reaction:
4Fe2+ + 2BH4- + 3H2O → 4Fe0 + 2H2BO3
- + 8H+ + 2H2
The black precipitates were filtered by vacuum
filtration and then, washed with distilled water and ethanol
at least three times. The prepared Fe0 particles were mixed
with 10 mL of 1% w/v starch solution, and then distilled
water was added to obtain 50 mL of solution C.
Bimetallic nano-Fe/Cu particles were prepared by
adding drop by drop 10 mL of aqueous solution D containing
0.500 gr CuSO4.7H2O to solution C in vigorous stirrer and
ambient temperature. After a few minutes, a redox reaction
occurred between Cu2+ and nano-Fe0 as follows:
Fe0 + Cu2+ → Fe2+ + Cu0
The resulting nano-Fe/Cu particles were washed with
distilled water, and stored in ethanol. The whole process
above was carried out under the condition of bubbling the
solutions with clean argon gas [10,11].
2.2. Effect on the CHCl3 de-chlorination
To find out the de-chlorination capacity of the
synthesized material, experiments were set up to investigate
effects of pH, material dosage, and treatment time on the
degradation of chloroform. These de-chlorination
experiments were performed in a closed batch system.
Determinations of pH were carried out by using the pH
meter (Sension+ PH31, Hatch (UK)) that was daily
calibrated at pH 4.00 and 7.00 using commercial buffers. In
most cases, each bottle received 50 mL of 20 ppm CHCl3.
2.3. Analytical methods
Chloroform degradation was analyzed by gas
chromatography coupled with mass spectrometry (GC-MS
Triple Quad 7098A-7001B Agilent, USA). The injection
temperature and detector temperature of the GC were set at
110 and 230oC, respectively, and a gradient program was
applied in the oven with an initial temperature of 50oC held
for 1 min and then gradually increased to 230oC at a rate of
15oC min-1, and remained at 230oC for 1 min. Chloride ion
concentration was determined by spectrometry (Lamda
650 UV-VIS spectrometer, USA) at a wavelength of
460 nm after reaction with mercury thiocyanate to form an
14 Phan Kim Nguyen, Bui Xuan Vung
orange-red compound [12]. The remaining chloroform
concentration after treatment was calculated based on the
chloride formation, which was quantitatively analyzed by
UV-VIS spectrometry. Finally, the efficiency of
chloroform degradation was calculated using initial
chloroform concentration and its remaining concentration
after the reduction.
3. Results and Discussion
3.1. Characterization of synthesized Fe/Cu nano particles
The synthesized nano-Fe/Cu particles were
characterized by XRD, TEM and EDX. Figure 1A shows X-
ray diffraction of the synthesized Fe/Cu nano particles were
obtained by a D8 Advanced Bruker diffractometer. It can be
seen from Figure 1A and 1B that there is a similarity of XRD
spectra obtained from this study and from the work of Chien-
Li Lee and Chih-Ju G Jou [13]. Figuge 2A presents TEM
image of nano-Fe/Cu particles which have been recorded by
a JEOL JEM-1010 transmission electron microscope. It is
found that the diffraction patterns indicate the state of
chemical combination of the bimetallic nanoparticles, and
the TEM image shows the particles are well-combined and
crystalline sizes are less than 1000Ao (or 100nm). Elemental
analysis performed by energy dispersive spectrometry
(EDX) with Horiba EMAX EDS detectors were presented in
Figure 2B. The weight percentage of Fe and Cu in the
synthesized nano-Fe/Cu particles obtained from the
elemental analysis is 81.72% and 13.04% respectively.
Figure 1. (A) Fe/Cu nano particle XRD patterns of this study and (B) of Chien-Li Lee & Chih-Ju G Jou
Figure 2. (A) TEM image and (B) EDX spectrum of fresh Fe/Cu particles
3.2. Effect on the chloroform de-chlorination
3.2.1. Effect of pH
Figure 3. Effect of initial pH on the CHCl3 degradation
The effect of initial pH on the reduction reaction of
chloroform by the synthesized nano-Fe/Cu is shown in
Figure 3. In each experiment 0.025 gr of the material was
added to 50 mL of 20 ppm CHCl3.
As can be seen from Figure 3 in the pH range from 3 to
7 the more acidic medium, the faster rates of chloroform
reduction are achieved. When the initial pH is 3.0, the
degradation efficiency of chloroform at 10 min reaction
reaches the maximum at 85.35%. When the initial pH is
7.0, the degradation efficiency of chloroform at 10 min
reaction decreases to 44.49%, much smaller than that at pH
3.0. However, at the initial pH 2 the degradation efficiency
is only 35.50 %, a much smaller value as compared with
that at pH 3.0. This abnormal issue could be explained by
the dehalogenation mechanism suggested by Leah J.
Matheson and Paul G. Tratnyek [14]. Alkyl halides, RX,
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 15
approach to the surface of the reductive material and then
can be reduced by iron according to the following reaction:
Fe0 + RX + H+ → Fe2+ + RH + X-
The increase in pH favors the formation of iron
hydroxide precipitates that may eventually form a surface
layer on the metal, which leads to inhibiting further
dissolution of the metal. Otherwise, at a more acidic pH,
there is an additional reaction between Fe0 and H+ to form
H2. In the absence of an effective catalyst such as Pd or Pt,
H2 is not a facile reductant, and this reaction will not
contribute directly to dehalogenation. In fact, excessive H2
accumulation at the metal surface inhibits the continuation
of reduction reactions in organic synthesis.
3.2.2. Effect of treatment time
For each experiment to investigate the effect of
treatment time on the chloroform de-chlorination, 0.025 gr
of the material was added to 50 mL of 20 ppm CHCl3 at pH
3 in the reaction time intervals of 5, 10, 30, 60 minutes.
Figure 4 shows that the degradation efficiency rises up
from 45.15% for 5 min treatment to 87.88% for 30 min
treatment. The degradation efficiency of 88.76% for
60 min treatment implies that there is an insignificant change
in the degradation efficiency after 30 min treatment.
Figure 4. Effect of treatment time on the CHCl3 degradation
3.2.3. Effect of nano-Fe/Cu dosage
In order to investigate the effect of synthesized nano
Fe/Cu dosage on the chloroform de-chlorination, the
dosage of 0.01, 0.025, 0.05 and 0.1 gr was respectively
added to 50 mL of 20 ppm CHCl3 at pH 3 with the
treatment time of 30 min.
Figure 5. Effect of nano Fe/Cu particle dosage on the CHCl3
degradation
From Figure 5 we can see when the material dosage
increases from 0.01 to 0.05 gr, the degradation efficiency
of chloroform increases from 79.33% to 86.84% and then
when adding 0.1 gr of the material, the degradation
efficiency is almost unchanged any more. So the material
dosage of 0.05 gr per 50 mL of 20 ppm CHCl3 can be
optimum for the investigation.
3.3. GC/MS analysis of chloroform degradation
In order to investigate whether such fewer chlorine
intermediate products as CH2Cl2, CH3Cl were formed from
chloroform de-chlorination by the synthesized nano Fe/Cu
particles, GC-MS analysis was performed for 20 ppm
chloroform solution before and after the treatment with the
reaction conditions of pH 3, treatment time of 30 min,
nano-Fe/Cu dosage of 0.05 gr, and illustrated by GC-MS
chromatograms in Figure 6E and 6F respectively. In
addition to this purpose, based on GC-MS analysis, the
degradation efficiency was also evaluated to compare with
that calculated by UV-VIS method as mentioned above. A
comparison between Figure 6A and 6B shows that the peak
of CH2Cl2 impurity at the retention time of 6.623 min
almost disappeared after the treatment. Meanwhile, the
area of the CHCl3 peak at the retention time of 8.037 min
is decreased from 4344995 to 482577, corresponding to the
degradation efficiency of 88.93%. This efficiency shows a
resemblance to the one calculated by using UV-VIS
method. From the comparison of Figure 6A and Figure 6B,
there is no evidence for the formation of fewer chlorine
intermidiate products from the treatment.
Figure 6. (A) GC-MS chromatograms of the sample before treatment and (B) the sample after the treatment
16 Phan Kim Nguyen, Bui Xuan Vung
4. Conclusion
Under the conditions of our experiments, chloroform
undergoes rapid reductive dehalogenation in the presence of
the synthesized nano-Fe/Cu particles with the degradation
efficiency is nearly up to 90%. Dehalogenation efficient of
chloroform is higher in a more acidic medium for the pH
range from 3 to 7, but this trend will be reversed at a pH of
2. This indicates that a strongly acidic medium is not in favor
of the chloroform de-chlorination. Results of GC-MS
analysis show that chloroform is completely transformed
into methane without forming products containing chlorine
such as CH2Cl2 or CH3Cl. Further investigations on the
effect of ions present in aqueous solution such as sulfate,
nitrate, phosphate, and of dissolving oxygen should be
conducted to approach the practical conditions.
REFERENCES
[1] W.H.O. Guidelines for drinking water quality (Chloroform),
incorporating 1st and 2nd addenda, 3rd ed. (2008), Vol(1), 451-453.
[2] Hua G.; Yeats S. Control of Trihalomethanes in Wastewater
Treatment. The Florida Water Resources Conference. (2009), USA.
[3] Alavi, N.; Tahvildari, K. Removal of Trihalomethanes in Tehran
Drinking Water by an Advanced Oxidation Process (2015), Nature Environment and Pollution Technology Vol.14, No.1, pp. 211-216.
[4] Wu, F.; Wu, S. Removal of Trihalomethanes from Drinking Water
by Air Stripping (2009). 2009 international conference on energy
and environmental technology, Vol2, pp. 695-698.
[5] Babaei, A. A.; Niknam, E.; Ansari, A.; Godini, K. Removal of
trihalomethane precursors from water using activated carbon
obtained from oak wood residue: kinetic and isotherm investigation
of adsorption process (2017). Desalination and Water Treatment, 92, 116–127.
[6] Lu, C.; Chung, Y.L.; Chang, K.F. Adsorption of trihalomethanes from
water with carbon nanotubes (2005). Water Res.,39(6),1183-9.
[7] Arruda, T. L. D.; Jardim, W. F. Treatment of groundwater
contaminated with chlorinated compounds using elemental iron and
Fenton's reagent. Quim. Nova. (2007), 30, 1628-1632.
[8] Wang, C.Y.; Chen, Z.Y. The preparation, surface modification, and
characterization of metallic nanoparticles. Chin. J. Chem. Phys.
(1999), 12, 670–674.
[9] Fennelly, J.P.; Roberts, A.L. Reaction of 1,1,1-trichloroethane with zero-valent metals and bimetallic reductants. Environ. Sci. Technol.
(1998), 32, 1980–1988.
[10] Zin, M.T.; Borja, J.; Hinode, H.; Kurniawar, W. “Synthesis of
Bimetallic Fe/Cu Nanoparticles with different Copper loading ratios”,
International Journal of Chemical, Molecular, Nuclear, Materials and Metallurgical Engineering (2013), Vol:7, No:12, 1031.
[11] He, F.; Zhao, D. Preparation and characterization of a new class of
starch-stabilized bimetallic nanoparticles for degradation of
chlorinated hydrocarbons in water. Environmental Science and
Technology, (2005), Vol. 39, No. 9, pp. 3314–3320.
[12] “Methods for chemical analysis of water and wastes” (1983),
Environmental Monitoring Support Laboratory (EMSL), Cincinnati, Ohio, pages 325, 2, 1-2.
[13] Lee, C. L.; & Jou, C. J. G. Integrating Suspended Copper/Iron
Bimetal Nanoparticles and Microwave Irradiation for Treating
Chlorobenzene in Aqueous Solution. Environment and Pollution
(2012), Vol. 1, No. 2, 159-168.
[14] Matheson, L.J.; Tratnyek, P. G. Reductive Dehalogenation of
Chlorinated Methanes by Iron Metal. Environ. Sci. Technol. (1994), 28, 2045-2053.
(The Board of Editors received the paper on 27/3/2018, its review was completed on 15/6/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 17
A STUDY ON CF3I-Ar AND CF3I-Kr MIXTURE GASES SUBSTITUTING SF6
IN HIGH VOLTAGE EQUIPMENTS
Tran Thanh Son1, Do Anh Tuan2 1Electric Power University; [email protected]
2Hung Yen University of Technology and Education; [email protected]
Abstract - The present binary mixtures of the SF6 gas with Ar and Kr gases have not been used in many industries as long-term measures for totally eliminating the potential contribution of SF6 to global warming. In order to gain more insight into electron transport coefficients in mixture gases as substitutes for SF6 in high voltage equipment, transport coefficients such as electron drift velocity, density-normalized longitudinal diffusion coefficient, ratio of the longitudinal diffusion coefficient to the electron mobility, Townsend first ionization coefficient, electron attachment coefficient, and density-normalized effective ionization coefficient in CF3I-Ar and CF3I-Kr mixture gases are calculated and analyzed in the wide E/N range of 0.01 – 1000 Td using a two-term approximation of the Boltzmann equation for the energy. These calculated coefficients are analyzed and compared to those in pure SF6 gas. The limiting field strength values of E/N, (E/N)lim, of these mixture gases are also derived and compared with those of the pure SF6 gas at different percentages of CF3I and SF6. The mixture gases of 70% CF3I with Ar and Kr have (E/N)lim values greater than those of the pure SF6 gas. Therefore, these mixture gases could be considered to substitute SF6 gas in high voltage equipment.
Key words - Trifluoroiodomethane; CF3I; SF6; Boltzmann equation analysis; electron transport coefficients; gas mixture
1. Introduction
Sulfur hexafluoride (SF6) has been widely used as an
isolated gas in high voltage equipment. The Kyoto
Protocol, however, has listed the greenhouse gases as CO2,
CH4, N2O, hydrofluorocarbons (HFCs), perfluorocarbons
(PFCs) and SF6, and we need to regulate the emissions and
the utilizations of those gases in the many industries [1]. In
recent decades, the conventional gases such as N2, CO2,
and air and the rare gases such as Ar, Kr, Xe, He, and Ne
have been considered to mix with the SF6 gas as a potential
to reach those attempts [2]. However, the present binary
mixtures of the SF6 gas with other gases have not been used
in many industries as long-term measures for totally
eliminating the potential contribution of SF6 to global
warming [2].
Recently, much research has been concentrated on
trifluoroidomethane (CF3I) gas because of its low global
warming potential, very short atmospheric lifetime and
relatively low toxicity gas [3]-[5]. It is a gas that is a
substitution candidate for the SF6 gas and as a candidate to
the replacement of potent greenhouse affects. This gas has
also been considered to be a candidate replacement for
bromotrifluoromethane (CF3Br), which is used in aircraft
for fuel inertness and for fire-fighting [3]. The boiling point
of CF3I gas is higher than that of the SF6 gas [4]. At an
absolute pressure of 0.5 MPa, CF3I becomes liquids at
about 260C, whereas the SF6 gas becomes liquids at about
-300C [4]. On the other hand, the SF6 gas is used in gas
circuit breakers at 0.5 to 0.6 MPa. Therefore, it is
impossible to use CF3I gas if this gas is used at this pressure
level [4]. However, in order to reduce the liquefaction
temperature of CF3I gas, Taki et al. [4] decreased partial
pressure by mixing it with other gases such as N2 and CO2.
For example, the boiling point can be reduced from about
260C (pure CF3I) to about -120C at 0.5 MPa by using a 30%
CF3I-CO2 mixture [5]. Therefore, it is necessary to mix the
CF3I gas with different buffer gases.
Moreover, the sets of electron collision cross sections
and electron transport coefficients for atoms, molecules,
and binary mixture gases are necessary for quantitative
understanding of plasma phenomena. Some gases, such as
rare gases (Ar, Kr, Xe, Ne, and He), N2, CO2, air, and O2
mixed with each of F2, Cl2, and SF6, are also necessary for
many applications, such as rare-gas halide laser, plasma
etching, and gaseous dielectric materials [2]. On the other
hand, the collision processes and electron transport
coefficients of the binary mixtures of CF3I gas with other
gases have been scarce so far. To the best of our
knowledge, neither measurements nor calculations of the
electron transport coefficients in the binary mixtures of the
CF3I gas with the Kr gas with the entire CF3I concentration
range have been performed previously.
In the present study, in order to gain more insight into
the electron transport coefficients, the electron transport
coefficients (electron drift velocity, density-normalized
longitudinal coefficient, and density-normalized effective
ionization coefficient) in the E/N range(ratio of the electric
field E to the neutral number density N) of 10 - 1000 Td
and the limiting field strength of E/N, (E/N)lim, for the
CF3I-Ar and CF3I-Kr mixtures are calculated by a two-term
approximation of the Boltzmann equation for the energy.
The negative differential conductivity (NDC) phenomena,
that is, decreasing electron drift velocity with increasing
electric field strength, in these binary gas mixtures are
suggested. The electron transport coefficients calculated
are also compared with those of pure SF6 gas and the
(E/N)lim values in those mixtures are also compared
respectively with those of SF6 mixtures with correlative
gases (Ar and Kr) in the experiments. The binary mixtures
of CF3I gas with Ar and Kr gases with CF3I concentration
equal to about 65 - 75%, are considered for use in high
voltage and many industries.
2. Calculation method of electron transport coefficients
in CF3I-Ar and CF3I-Kr mixtures
The electron transport coefficients are calculated by
sets of electron collision cross sections for gases and a two-
term approximation of the Boltzmann equation for the
18 Tran Thanh Son, Do Anh Tuan
energy given by Tagashira et al. [6]. The accurate electron
collision cross section sets for each gas in mixture are
chosen for calculation to obtain the reliable electron
transport coefficients. The electron energy distribution
function (EEDF) can be computed by solving the
Boltzmann equation. In this study, a two-term
approximation is applied as successfully used in our
previous article [7]. Based on the EEDF, f(ε, E/N), the
electron drift velocity, W, the density-normalized
longitudinal diffusion coefficient, NDL, the Townsend first
ionization, α, and the electron attachment coefficient, η,
can be calculated as following equations:
1/2
m0
1 2 eE df ( ,E / N)W d .
3 m N q ( ) d
= −
(1)
where ε is the electron energy, m is the electron mass, e is
the elementary charge, and qm(ε) is the momentum-transfer
cross section.
( )
121
1 2L 1 0
0 0T T
0 2 1 1 02
VND E (F )d F d
3N q q
A A .
− = +
− − −
(2)
where V1 is the speed of electron, qT is the total cross
section. Fn and n (n = 0, 1, 2) are respectively the electron
energy distributions of various orders and their
eigenvalues.V1, n , 0n , and An are given by
1/2
1
2eV
m
=
; 1
20 1 i 0
0V N q F d
= ;
11 2
1 0 0 1 010
T
V E(F )d ( A )
3N q
− = − + −
;
12
0n 1 i n0
V N q F d
= ; n n0
A F d .
=
where qi is the ionization cross section.
1/2
1/2
i
I
1 2/ N f ( ,E / N) q ( )d .
W m
=
(3)
where I is the ionization onset energy and qi(ε) is the
ionization cross section.
1/2
1/2
a
0
1 2/ N f ( ,E / N) q ( )d .
W m
=
(4)
where qa(ε) is the attachment cross section.
The electron collision cross sections for CF3I
determined by Kimura and Nakamura [8], Ar determined
by Nakamura and Kurachi [9], and Kr determined by
Hayashi [10] are used throughout the present study. The set
of electron collision cross sections for the CF3I molecule
[8] includes one momentum transfer, one attachment, three
vibrational excitations (threshold energies of 0.032 -
0.134 eV), five electronic excitations (threshold energies
of 4.7 - 9.6 eV), and one total ionization (threshold energy
of 10.2 eV) cross sections.
The set of electron collision cross sections for Ar atom
[9] includes one momentum transfer, five electronic
excitations (threshold energies of 11.6 - 13.9 eV), and one
total ionization (threshold energy of 15.69 eV) cross
sections. The set of electron collision cross sections for Kr
atom [10] includes one momentum transfer, fourteen
electronic excitations (threshold energies of 9.915 -
13.437 eV), and one total ionization (threshold energy of
14 eV) cross sections. The accuracy of the electron collision
cross section set for each gas is confirmed to be consistent
with all electron transport coefficients in each pure gas.
3. Results and discussions
The results for the electron drift velocities, W, as
functions of E/N for the binary mixtures of CF3I gas with
Ar and Kr gases calculated in the E/N range 10 < E/N <
1000 Td by a two-term approximation of the Boltzmann
equation are shown in Figures 1-2, respectively. Slight
regions of the NDC phenomena in these gas mixtures are
observed in the E/N range 15 < E/N < 170 Td. The NDC is
relatively shallow for all mixtures. The occurrences of
these phenomena are due to the Ramsauer-Townsend
minimum (RTM) of the elastic momentum transfer cross
sections of the Ar and Kr atoms, and the CF3I molecule.
These suggestions are analyzed and explained thoroughly
by Chiflikian [11]. In the binary mixtures of the CF3I gas
with the Ar and Kr gases, the values of W are suggested to
be between those of the pure gases over E/N > 100 Td and
these values grow linearly over E/N > 200 Td. For the sake
of comparison, the electron drift velocity obtained by
Aschwanden [12] for the pure SF6 gas is shown in Figures
1-2. The calculated electron drift velocities in 70%
CF3I-Ar in the E/N ranges of E/N < 600 Td are very close
to those of the pure SF6 gas.
Figure 1. Electron drift velocity, W, as functions of E/N for the
CF3I-Ar mixtures with 10%, 30%, 50%, and 70% CF3I.
The solid line and symbols show present W values calculated using
a two-term approximation of the Boltzmann equation for the CF3I-
Ar mixtures. The solid curves show present W values calculated for
the pure CF3I molecule and pure Ar atom. The star symbol shows
the measurement value of the pure SF6 [12]. The inset figure shows
these results calculated in the E/N range of 200 - 1000 Td
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 19
Figure 2. Electron drift velocity, W, as functions of E/N for the
CF3I-Kr mixtures with 10%, 30%, 50%, and 70% CF3I. The solid
line and symbols show present W values calculated using a two-
term approximation of the Boltzmann equation for the CF3I-Kr
mixtures. The solid curves show present W values calculated for
the pure CF3I molecule and pure Kr atom. The star symbol shows
the measurement value of the pure SF6 [12]. The inset figure
shows these results calculated in the E/N range of 200 - 1000 Td
The results for the density-normalized longitudinal
coefficients, NDL, as functions of E/N for the binary
mixtures of CF3I gas with Ar and Kr gases calculated in the
E/N range 10 < E/N < 1000 Td by a two-term
approximation of the Boltzmann equation are shown in
Figures 3-4, respectively.
For each E/N value, the NDL values of the binary
mixtures of the CF3I gas with Ar and Kr gases decrease with
the increase in the CF3Icontent in the mixture. This behavior
is due to the growing influence of the electron-CF3I
interaction as the CF3Icontent increases. In these figures, on
the other hand, these NDL curves have minima in the E/N
range of 15 - 170 Td for these binary mixtures. The same
process responsible for the NDC region in the electron drift
velocity curves in these binary mixtures caused the
occurrence of these minima. Urquijo et al. [13] also
observed the similar behavior for the C2F6-Ar mixtures. The
density-normalized longitudinal coefficient for the pure SF6
obtained by Aschwanden [12] is also shown in Figures 3-4
for the sake of comparison. The NDL values of the pure SF6
are greater than those of these binary mixtures.
The results for the density-normalized effective
ionization coefficients, (α - η)/N, as functions of E/N for the
binary mixtures of CF3I gas with Ar and Kr gases calculated
by a two-term approximation of the Boltzmann equation are
shown in Figures 5-6, respectively. In the binary mixtures of
the CF3I with the Ar and Kr gases, the values of (α - η)/N are
also suggested to be between those of the pure gases,
respectively. For the sake of comparison, the density-
normalized effective ionization coefficient obtained by
Aschwanden [12] for the pure SF6 gas is also shown in
Figures 5-6. The (α - η)/N values for 70% CF3I mixtures
with the Ar and Kr gases are very close to those of the pure
SF6 gas over E/N < 450 Td and E/N < 470 Td, respectively.
Because of the accuracy of the electron collision cross
sections for the present gases and the validity of the
Boltzmann equation, the present calculated results are
reliable. More experiments of the electron transport
coefficients for the binary mixtures of the CF3I gas with
these buffer gases need to be performed over the wide
range of E/N in the future. In general, when the percentage
ratio of the CF3I gas in binary mixtures increases, the
values of the electron transport coefficients increase
progressively to those of the pure CF3I.
The limiting field strength values of E/N, (E/N)lim, at
which α = η for the binary mixtures of CF3I gas with Ar
and Kr gases are derived at 133.322 Pa and shown in Figure
7. These values are also compared respectively with those
of the binary mixtures of the SF6 gas with the Ar [14] and
Kr [15] gases shown in Figure 7. The (E/N)lim value
calculated for the pure CF3I gas is equal to 437 Td greater
than the (E/N)lim of the pure SF6 gas (361 Td) [12]. It can
be considered as a prospective substitute for the SF6 gas. In
Figure 7, the CF3I concentration in the binary mixtures of
CF3I gas with Ar and Kr gases equal to about 65 - 75%, is
considered for use in high voltage and many industries if
other chemical, physical, electrical, thermal, and
economical studies are considered thoroughly.
Figure 3. Density-normalized longitudinal coefficient, NDL, as
functions of E/N for the CF3I-Ar mixtures with 10%, 30%, 50%, and
70% CF3I. The solid line and symbols show present NDL values
calculated using a two-term approximation of the Boltzmann
equation for the CF3I-Ar mixtures. The solid curves show present
NDL values calculated for the pure CF3I molecule and pure Ar atom.
The star symbol shows the measurement value of the pure SF6 [12]
Figure 4. Density-normalized longitudinal coefficient, NDL, as
functions of E/N for the CF3I-Kr mixtures with 10%, 30%,
50%, and 70% CF3I. The solid line and symbols show present
NDL values calculated using a two-term approximation of the
Boltzmann equation for the CF3I-Kr mixtures. The solid curves
show present NDL values calculated for the pure CF3I molecule
and pure Kr atom. The star symbol shows the measurement
value of the pure SF6 [12]
20 Tran Thanh Son, Do Anh Tuan
Figure 5. Density normalized effective ionization coefficient,
(α - η)/N, as functions of E/N for the CF3I-Ar mixtures with
10%, 30%, 50%, and 70% CF3I. The solid line and symbols
show present (α - η)/N values calculated using a two-term
approximation of the Boltzmann equation for the CF3I-Ar
mixtures. The solid curves show present (α - η)/N values
calculated for the pure CF3I molecule and pure Ar atom. The
star symbol shows the measurement value of the pure SF6 [12]
Figure 6. Density normalized effective ionization coefficient,
(α - η)/N, as functions of E/N for the CF3I-Kr mixtures with
10%, 30%, 50%, and 70% CF3I. The solid line and symbols
show present (α - η)/N values calculated using a two-term
approximation of the Boltzmann equation for the CF3I-Kr
mixtures. The solid curves show present (α - η)/N values
calculated for the pure CF3I molecule and pure Kr atom. The
star symbol shows the measurement value of the pure SF6 [12]
Figure 7. Limiting field strength values of E/N, (E/N)lim, as
functions of the percentage of CF3I gas for the binary mixtures
CF3I-Ar and CF3I-Kr. The solid line and solid symbols show
present (E/N)lim values for these binary mixtures calculated
using a two-term approximation of the Boltzmann equation.
The dotted curves and the open symbols show (E/N)lim values for
the binary mixtures SF6-Ar [14] and SF6-Kr [15]
4. Conclusion
The electron drift velocity, density-normalized
longitudinal coefficient, and density-normalized effective
ionization coefficient in the binary mixtures in CF3I with
Ar and Kr gases are calculated using a two-term
approximation of the Boltzmann equation for the energy in
the E/N range of 10 - 1000 Td for the first time. The NDC
phenomena in these binary gas mixtures are suggested. The
electron transport calculated coefficients are also
compared with those of the pure SF6 gas in experiments.
The limiting field strength values of E/N for the binary
mixtures of 70% CF3I gas with Ar and Kr gases are
determined and greater than those of the pure SF6 gas.
Therefore, these binary mixtures with CF3I concentration
equal to about 65 - 75% are considered for use in high
voltage and many industries. For the purposes of
justification of the accuracy of our results, more
experimental data for electron transport coefficients for the
binary mixtures of CF3I with these gases need to be
performed over a wide range of E/N.
REFERENCES
[1] Conference of the Parties, Third Session Kyoto, Kyoto, Japan (1997).
[2] L. G. Christophorou and R. J. Van Brunt, IEEE Trans. Dielectrics
and Elec. Insulation, 2, 952 (1995).
[3] M. K. M. Jamil, Ph.D. dissertation, Dept. Elect. Electron. Eng.,
Kyushu Institute of Technology, Fukuoka, Japan (2007).
[4] M. Taki, D. Maekawa, H. Odaka, H. Mizoguchi, and S. Yanabu,
IEEE Trans. Dielectrics and Elec. Insulation, 14, 341 (2007).
[5] H. Takagari, H. Kasuya, H. Mizoguchi, and S. Yanabu, IEEE Trans.
Dielectrics and Elec. Insulation, 15, 1424 (2008).
[6] H. Tagashira, Y. Sakai, and S. Sakamoto, J. Phys. D, Appl. Phys.,
10, 1051 (1977).
[7] D. A. Tuan, J. Korean Phys. Soc., 64, 23 (2014).
[8] M. Kimura and Y. Nakamura, J. Phys. D, Appl. Phys., 43, 145202-
1 (2010).
[9] Y. Nakamura and M. Kurachi, J. Phys. D, Appl. Phys., 21, 718 (1988).
[10] M. Hayashi, J. Phys. D, Appl. Phys., 15, 1411 (1982).
[11] R. V. Chiflikian, Phys. Plasmas, 2, 3902 (1995).
[12] T. Aschwanden, Ph.D. dissertation, Eidgenössische Technische
Hochschule Zürich, Zurich, Germany, 1985 (in German).
[13] J. de Urquijo, A. A. Castrejón-Pita, J. L. Hernández-Ávila, and E.
Basurto, J. Phys. D, Appl. Phys., 37, 1774 (2004).
[14] J. de Urquijo, J. L. Hernández-Ávila, E. Basurto, and F. Ramírez, J.
Phys. D, Appl. Phys., 36, 1489 (2003).
[15] Y. Qiu and D. M. Xiao, J. Phys. D, Appl. Phys., 27, 2663 (1994).
(The Board of Editors received the paper on 27/02/2018, its review was completed on 14/3/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 21
EVALUATION OF SHEAR STRENGTH OF REINFORCED CONCRETE
STRUCTURAL WALLS OF ACI 318-14 AND EUROCODES
Tran Anh Thien
University of Science and Technology - University of Danang; [email protected]
Abstract - Reinforced concrete structural walls are very effective in resisting lateral loads due to their high strength and stiffness. While Vietnamese Standard TCVN 5574-2012 does not provide detailed provisions for design of structural walls, the shear strength of reinforced concrete structural walls according to various building codes are very different. The paper investigates the design shear strength of reinforced concrete structural walls using provisions from ACI 318-14, Eurocode 2 EN 1992-1:2004, and Eurocode 8 EN 1998-1:2004. The theory used in these building codes to determine wall shear strength is analyzed and numerical comparison is carried out to evaluate the influence of key parameters, including compressive concrete strength, axial load level, and shear span ratio, on the wall shear capacity, for both non-seismic and seismic design.
Key words - reinforced concrete; structural wall; shear wall; shear strength; building code
1. Introduction
Reinforced concrete structural walls are commonly
used to resist lateral loads, such as wind or earthquake
loads, due to their high strength and stiffness. Although
being important structural members, detailed provisions
for behavior and design of structural walls are not provided
in Vietnamese Standard TCVN 5574-2012 [1]. In fact, the
design of reinforced concrete structural walls for shear has
become an issue as the shear capacity according to various
building codes are very different due to the complex stress
redistributions that occur after cracking. Shear transfer
mechanisms have been proved to be influenced by various
parameters.
Several factors have an influence on the shear capacity
of reinforced concrete structural walls. The most
influenced parameters are known as wall cross section,
concrete strength, axial force and shear span ratio. There
has been much research on the shear performance and
design of reinforced concrete structural walls according to
different building codes [2], [3], [4], [5], [6]... However, no
investigation has been done on direct comparison of design
provisions from ACI 318 and Eurocodes with respect to
specific parameters. This study investigates the design
shear strength of reinforced concrete structural walls using
provisions from ACI 318-14 [7], Eurocode 2 EN
1992-1:2004 [8], and Eurocode 8 EN 1998-1:2004 [9] for
both non-seismic and seismic design.
2. Shear provisions of structural walls in ACI 318-14,
Eurocode 2 EN 1992-1:2004, and Eurocode 8 EN
1998-1:2004
2.1. ACI 318-14
In ACI 318-14, design provisions for shear of reinforced
concrete structural walls are presented in Sections 11.5.4 and
18.10.4. These two semi-empirical equations are based on
the modified truss analogy approach and can be used to
predict the peak shear capacity of reinforced concrete walls.
According to this modified truss analogy approach, the peak
wall shear strength is the summation of two shear forces, one
resisted by concrete and the other resisted by horizontal web
reinforcement. Both sections assume a diagonal tension
failure mechanism with a 45-degree crack and diagonal
compression failure is prevented by controlling an upper
limit for the wall shear stress. While Section 11.5.4 provides
requirements for non-seismic design (NSD) of structural
walls, Section 18.10.4 provides provisions for seismic
design (SD) of structural walls.
The procedure to predict the shear capacity of
reinforced concrete walls in Section 11.5.4 is given by the
following equations.
'10n c s cV V V f hd= +
(1)
1 2min[ , ]c c cV V V=
(2)
'
1 3.34
uc c
w
N dV f hd
l= +
(3)
'
'
2
0.21.25
0.6
2
uw c
w
c cu w
u
Nl f
l hV f hd
M l
V
=
+
+
−
(4)
v yt
s
A f dV
s=
(5)
where Vn (lb) is the nominal shear strength of the wall;
Vc (lb) is the nominal shear strength provided by concrete;
Vs (lb) is the nominal shear strength provided by horizontal
web reinforcement; is the modification factor to reflect the
reduced mechanical properties of lightweight concrete
relative to normal weight concrete of the same compressive
strength, =1.0 for normal concrete; f’c (psi) is the specified
compressive strength of concrete; h (in.) is the thickness of
the wall; lw (in.) is the length of the wall; d (in.) is the
distance from the extreme compressive fiber to centroid of
longitudinal tension reinforcement and assumed to be 0.8lw
unless a larger value is determined by a strain compatibility
analysis; Mu (lb-in) and Vu (lb) are the factored moment and
shear force at the critical section, respectively; Nu (lb) is the
factored axial load that is positive in compression and
negative in tension; Av is the area of horizontal web
reinforcement within spacing s; fyt is the specified yield
strength of horizontal web reinforcement.
The value for u uM V is evaluated at the critical section
above the base of the wall, and the location of that section
is described in Figure 1.
22 Tran Anh Thien
Figure 1. Location of critical section for checking flexural-
shear strength [10]
Equation (4) does not apply if:
02
u w
u
M l
V− (6)
In order to avoid diagonal compression failure, the wall
shear stress is limited to '10 cf . The horizontal web
reinforcement has the minimum ratio of 0.25%:
0.0025t
(7)
The minimum vertical web reinforcement ratio is given by:
( )0.0025 0.5 2.5 0.0025wl t
w
h
l
+ − −
(8)
In SI units, the three equations (1), (3), (4) are replaced
by the three following equations (9), (10), (11),
respectively,
'0.83n c s cV V V f hd= +
(9)
'
1 2.74
uc c
w
N dV f hd
l= +
(10)
'
'
2
0.20.1
0.05
2
uw c
w
c cu w
u
Nl f
l hV f hd
M l
V
=
+
+
−
(11)
where f’c is (MPa), h (mm), lw (mm), d (mm), Mu (Nm), Vu
(N), and Nu (N).
The nominal shear strength of reinforced concrete
structural walls per Section 18.10.4 is given as follows.
( )' '10n c c t yt cv c cvV f f A f A = +
(12)
where Vn (lb) is the nominal shear strength of the wall;
c is a function of wall aspect ratio, equal to 2.0 for
2.0w
w
h
l , 3.0 for 1.5w
w
h
l
and varies linearly for
1.5 2.0w
w
h
l ; f’c (psi) is the specified compressive
strength of concrete; t is the horizontal web
reinforcement ratio; ytf (psi) is the yield stress of the
horizontal web reinforcement; Acv (in2) is the gross area
of the wall bounded by the web thickness and the wall
length; hw (in) and lw (in) are the height and length of the
wall, respectively.
In SI units, Equation (12) is rewritten as the following
equation.
( )' '0.83n c c t yt cv c cvV f f A f A = +
(13)
where Vn (N); c is equal to 0.17 for 2.0w
w
h
l , 0.25 for
1.5w
w
h
l
and varies linearly for 1.5 2.0w
w
h
l ; f’c (MPa);
ytf (MPa); Acv (mm2); hw (mm) and lw (mm).
The major difference between sections 11.5.4 and
18.10.4 is the calculation of wall shear strength provided
by concrete. In Section 11.5.4, the two values of the
concrete contribution corresponding to different cracking
conditions are determined. Equation (3) corresponds to
the occurrence of web shear cracking at a principal tensile
stress of approximately '4 cf at the centroid of the shear
wall cross section. Equation (4) corresponds
approximately to the occurrence of flexure-shear
cracking at a flexural tensile stress of '6 cf at a section
above the section being investigated. In Section
18.10.4, the concrete contribution is determined using an
empirical factor c that is a function of wall aspect ratio.
The design shear strength of reinforced concrete
structural walls is Vn, where is the strength reduction
factor for shear.
2.2. Eurocode 2 EN 1992-1:2004
The non-seismic design provisions for shear of
reinforced concrete structural walls in Eurocode 2 EN
1992-1:2004 (EC2 (2004)) are presented in Section 6.2.
According to EC2 (2004), horizontal shear reinforcement
of structural walls is calculated on the basis of a variable
inclination truss model, which is similar to the theory
applied to beams. It should be noted that in this model, all
shear is resisted by the provision of links with no direct
contribution from the shear capacity of the concrete itself.
The shear resistance RdV in Section 6.2 is determined as
follows.
, ,max;minRd Rd s RdV V V =
(14)
, cotsw ywd
Rd s
A f zV
s=
(15)
1,max
cot tan
cw cd wRd
f b zvV
=
+ (16)
where ,Rd sV is the shear resistance of the links which are
horizontal web reinforcement; ,maxRdV is the maximum
2
wl
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 23
design value of the shear which can be resisted by the
concrete strut; swA is the cross-sectional area of the shear
reinforcement; s is the spacing of horizontal web
reinforcement; ywdf is the design yield strength of the
horizontal web reinforcement, 1.15
yk
ywd
ff = ;
ykf is the
ultimate design yield strength of the horizontal web
reinforcement; z is lever arm between the upper and lower
chord members of the analogous truss, which can be taken
as 0.8 wl ; is the angle between the diagonal concrete
compression struts to the wall axis perpendicular to the
shear force; cw is a coefficient taking account of the state
of the stress in the compression chord, which is taken as 1
for non-prestressed structures; cdf is the design concrete
compressive strength, 1.5
ckcd
ff = ;
ckf is the ultimate
design concrete compressive strength; 1v is a strength
reduction factor for concrete cracked in shear
1 0.6 1250
ckfv
= −
.
In EC2 (2004), the angle has a value between 21.8
and 45 degrees, or:
1.0 cot 2.5 (17) 2.3. Eurocode 8 EN 1998-1:2004
The seismic design provisions for ductile reinforced
concrete structural walls in Eurocode 8 EN 1998-1:2004
(EC8 (2004)) are presented in Section 5.5. The following
provisions are required to prevent diagonal tension failure
of the wall web due to shear.
If the shear span ratio 2.0Eds
Ed w
M
V l = , the previous
provisions in EC2 (2004) are applied with the angle
taken as 45 degrees.
If the shear span ratio 2.0s , the shear resistance RdV
is determined as follows.
, , ,0.75Rd s Rd c h yd h wo s wV V f b l = +
(18)
where ,Rd cV is the design value of shear resistance for
members without shear reinforcement, h is the
reinforcement ratio of horizontal web bars, bh
wo h
A
b s = ;
,yd hf is the design yield strength of the horizontal web
reinforcement; wob is the thickness of the wall.
The design value ,Rd cV of shear resistance for members
without shear reinforcement is given by:
1/3
, 1 1 w ,min(100 )
c Rd c ck cp cV C k f k b d V = +
(19)
The shear resistance Vc is not less than:
,min min 1 w(v )
c cpV k b d= + (20)
where fck (MPa) is the characteristic compressive cylinder
strength of concrete at 28 days;
2002,01k
d
= +
(21)
with d (mm); 1 is the longirtudinal tension with d (mm)
1
1
w
0.02s
A
b d = 22)
Asl is the area of longitudinal tension reinforcement,
which extends (lbd+d) beyond the section considered; k1
is 0.15; cp (MPa) is the compressive stress in the concrete
from axial load,
c
0.2Ed
cp
Nfcd
A = (23)
NEd (N) is the axial force in the cross-section due to
loading; to be taken as positive for compression and
negative for tension; Ac (mm2) is the area of the concrete
cross section; fcd (MPa) is the design value of concrete
compressive strength; bw (mm) is the smallest width of the
cross-section in the tensile area; d (mm) is the distance
from the extreme compressive fiber to centroid of
longitudinal tension reinforcement, vmin is determined as: 3/2 1/2
min 0.035 ckv k f= (24)
3. Comparison of design code provisions
The shear resistance of a reinforced concrete shear
wall is investigated based on the following building code
provisions: ACI 318-14 (NSD), ACI 318-14 (SD),
EC2 (2004) and EC8 (2004), where NSD and SD stand
for non-seismic design and seismic design, respectively.
The structural wall has the rectangular cross section of
25×300cm and is subjected to a concentrated force at the
top of the wall. Horizontal web reinforcement consists of
two layers of 14mm-diameter bars at spacing of 200mm,
and has the design yield strength of 280MPa.
3.1. Shear strength versus concrete compressive strength
level
Figure 2. Comparison of *
nV versus B for walls with
the axial load of 2400kN
24 Tran Anh Thien
Figure 3. Comparison of *
nV versus B for walls with
the axial load of 6000kN
Figures 2 and 3 present the relation between the design
shear resistance *
nV
and concrete strength grade B
mentioned in TCVN 5574-2012. *
nV stands for nV in ACI
318-14 and RdV
in Eurocodes. The structural wall has
aspect ratio of 2.0. The axial load acting on the wall is kept
constant as 2400kN and 6000kN, which is equal to '0.2 c cvf A and '0.5 c cvf A for B15 concrete, in Figures 2
and 3, respectively.
As can be seen from the figures, *
nV calculated for non-
seismic wall design is significantly larger than that for
seismic design in Eurocodes, especially with high concrete
compressive strength. That is because ,maxRdV is
proportional to cdf , leading to that ,max .Rd Rd sV V with
high concrete compressive strength for all 21.8 45 = degrees. At B45 or above, *
nV of EC8 (2004) is 2.5 times
larger than that of EC2 (2004). This difference in wall
shear strength is also obvious between ACI 318-14 (NSD)
and ACI 318-14 (SD), especially with high axial load.
In EC8 (2004), as the angle is set constant as 45
degrees and , .max( 45)Rd s RdV V = , the design shear
resistance does not change with varied concrete strength. *
nV based on ACI 318-14 (SD) and EC8 (2004) are quite
closed together for walls with low concrete compressive
strength, EC8 (2008) always gives more conservative shear
strength compared with ACI 318-14 (SD).
3.2. Shear strength versus axial load ratio
Figures 4 and 5 show the relation between the design
shear resistance *
nV
and axial load ratio '
c cv
N
f A. The
structural wall also has aspect ratio of 2.0. The concrete
strength grade is B20 and B50 in Figures 4 and 5,
respectively. Of these building code provisions, only the
design shear resistance calculated in ACI 318-14 (NSD)
depends on the axial load. *
nV in ACI 318-14 (NSD)
increases much more significantly with higher concrete
compressive strength. For walls using B20 concrete,
*
nV at '
0.5c cv
N
f A is 1.7 times larger than that at
'0.2
c cv
N
f A. This difference increases up to 2.5 times for
walls using B50 concrete.
Figure 4. Comparison of *
nV versus axial load ratio for concrete B20
Figure 5. Comparison of *
nV versus axial load ratio for concrete B50
3.3. Shear strength versus shear span ratio
When the wall shear span ratio increases from 2.0, the
shear strength based on ACI 318-14 (NSD) decreases,
while that based on the other building code provisions
remains the same. Figure 6 presents the relation between
the design shear resistance *
nV from ACI 318-14 (NSD)
and concrete strength grade B, corresponding to the shear
span ratio of 2.0, 3.0, and 4.0. The figure indicates that *
nV
decreases when the shear span ratio increases. However, *
nV decreases substantially at low shear span ratio and
decreases slightly at high shear span ratio.
Figure 6. Comparison of *
nV versus B at shear span ratio
ranging from 2.0 to 4.0
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 25
4. Conclusion
The study is carried out to investigate the influence of
various parameters on the shear resistance of structural
walls in the following building codes: ACI 318-14,
Eurocode 2 EN 1992-1:2004, and Eurocode 8 EN 1998-
1:2004. The structural wall has the rectangular cross
section of 25×300cm and is subjected to a concentrated
force at the top of the wall. Horizontal web reinforcement
consists of two layers of 14mm-diameter bars at spacing of
200mm, and has the design yield strength of 280MPa.
From the study, the following conclusions can be drawn.
- ACI 318-14 provides semi-empirical equations to
determine the wall shear strength which are based on the
modified truss analogy approach. The peak wall shear
strength is the summation of two shear forces, one resisted
by concrete and the other resisted by horizontal web
reinforcement. ACI 318-14 assumes a diagonal tension
failure mechanism with a 45-degree crack.
- According to EC2 (2004) and EC8 (2004), the wall
shear strength is calculated on the basis of a variable
inclination truss model, in which all shear is resisted by the
provision of horizontal web reinforcement with no direct
contribution from the shear capacity of the concrete. The
angle between the diagonal concrete compression struts
to the wall axis perpendicular to the shear force varies from
21.8 to 45 degrees in EC2 (2004), while it is set constant
as 45 degrees in EC8 (2004).
- Generally, the wall shear resistance increases with
concrete compressive strength, except for that calculated
from EC8 (2004). In EC8 (2004), as the angle is set
constant as 45 degrees and , .max( 45)Rd s RdV V = ,
the design shear resistance does not change with varied
concrete strength.
- EC8 (2008) provides more conservative wall shear
strength compared with ACI 318-14 (SD). Shear strength
calculated from ACI 318-14 (SD) is almost 1.5 times larger
than that from EC8 (2008) for walls with shear span ratio
of 2.0 and B60 concrete.
- Of these building code provisions, only the design
shear resistance calculated in ACI 318-14 (NSD) depends
on the axial load. *
nV in ACI 318-14 (NSD) increases much
more significantly with higher concrete compressive
strength.
- The shear strength based on ACI 318-14 (NSD)
decreases when the wall shear span ratio increases from
2.0, while that based on the other building code provisions
remains the same. The shear strength decreases more
slightly at higher shear span ratio.
REFERENCES
[1] TCVN 5574-2012 Concrete and reinforced concrete structures - Design standard.
[2] ASCE-ACI Committee 426 (1973). The Shear Strength of
Reinforced Concrete Members, Journal of Structural Division,
ASCE, V. 99, No. 6, pp. 1091-11872.
[3] ASCE-ACI Committee 445 (1999). ACI 445R-99 Recent
Approaches to Shear Design of Structural Concrete.
[4] Rangan, B. V. (1998). Shear design of reinforced concrete beams, slabs
and walls, Cement and Concrete Composites, No. 20, pp. 455-464.
[5] Gulec, C. K., and Whittaker, A.S. (2009). Performance-based
assessment and design of squat reinforced concrete shear walls,
Technical Report MCEER-09-0010, 660 pp.
[6] Dashti, F. and Dhakal R. (2013). Comparative performance of RC
shear walls designed by different standards. Advances in Structural Engineering and Mechanics, pp. 1084-1103.
[7] ACI 318-14 Building code requirements for structural concrete and
commentary. American Concrete Institute. Farmington Hills.
[8] Eurocode 2 EN 1992-1:2004. Design of concrete structures - Part 1-
1: General rules and rules for buildings.
[9] Eurocode 8 EN 1998-1:2004. Design of structures for earthquake
resistance – Part 1: General rules, seismic actions and rules for
buildings.
[10] Wight, J.K., and MacGregor, J.G. (2012). Reinforced concrete –
Mechanics and design, Pearson Education, Inc., 1157 pp.
(The Board of Editors received the paper on 29/5/2018, its review was completed on 13/6/2018)
26 Vu Van Truong, Truong Viet Anh, Tran Xuan Bo, Truong Van Thuan
FULLY RESOLVED SIMULATION OF THE PHASE CHANGE PROCESS OF
A LIQUID DROP
Vu Van Truong, Truong Viet Anh, Tran Xuan Bo, Truong Van Thuan
School of Transportation Engineering, Hanoi University of Science and Technology;
[email protected], [email protected], [email protected], [email protected]
Abstract - This paper presents a numerical investigation of the solidification process of a liquid drop on a cold solid surface. The drop is immersed in the computational domain with the presence of three phases: solid – liquid – gas. The Navier-Stokes and energy equations are used to solve the problem in which the interface separating different phases is represented by a front-tracking method. The interpolation technique is used to impose the non-slip boundary condition on the solid surface. The cold surface on which the drop is placed is the cause of solidification with the phase change interface propagating from the cold surface to the top of the drop. The numerical results of some typical cases that are compared with the available experimental ones indicate the accuracy of the numerical method used in this study.
Key words - Numerical simulation; phase change; liquid drop; front-tracking; three phases
1. Introduction
Understanding of the dynamics of a drop including
phase change heat transfer is very important because of its
wide range applications such as atomization, crystal
growth, food processing and so on. In nature, one can find
the solidification of water drops attaching to leaves, cable
lines, wind turbine blades and aircraft wings. Accordingly,
there have been many studies concerned with this
solidification problem.
Experimentally, Enríquez et al. [1] dripped a water drop
on a subzero temperature plate that caused the drop to
freeze. Jin and co-workers [2] used a molecular tagging
thermometry technique to capture the motion of the water–
ice phase change interface during the solidification of a
water drop on a cold plate. Recently, Zhang et al. [3] also
paid attention to the freezing process of a water drop on a
plate. Itoh and co-workers [4] used molten silicon, a
semiconductor material, as a phase change material to
growth crystallized silicon drops for solar cell applications.
Satunkin [5] used molten silicon, germanium and indium
antimonide to form solid drops and to find the growth
angles at the tri-junction (where the three phases meet).
A common interesting feature of all above-mentioned
works is the formation of an apex at the top of the drop
after complete solidification because of volume expansion
and the effect of the growth angle.
Theoretically, Sanz [6] and Nauenberg [7] developed
models to reproduce the evolution of the phase change and
drop fronts of a drop solidifying on a cold plate. In another
recent work, Zhang et al. [8] also theoretically investigated
the freezing process of a water drop on a plate.
Concerning numerical simulations, a few studies have
been done for the solidification problem of a liquid drop on
a plate. For instance, the boundary integral method [9], the
Galerkin finite element method [10] and the front-tracking
method [11,12] have been used to investigate the dynamics
of the phase change heat transfer of the drop solidification
problem.
It is evident that the problem of a liquid drop solidifying
on a cold plate has been getting more and more attractive.
Thus, an accurate method for simulating the problem
becomes very important. Accordingly, the present study
presents a direct numerical method for simulations of a
liquid drop solidifying on a cold plate. The methods used
here are the front-tracking technique to represent the
interface and an interpolation technique to deal with the
non-slip boundary condition.
2. Numerical problem and governing equations
We consider an axisymmetric liquid drop placed on a
cold plate whose temperature Tc is below the fusion
temperature Tm of the drop liquid (Figure 1). Because of
the cold plate, the solidifying front formed on the plate
surface propagates upwards to the top of the drop. Initially,
the liquid drop is assumed to be a section of a sphere with
a volume denoted by V0. The contact angle 0 is then
defined at the plate. The growth angle at the tri-junction
(i.e. triple point) at which three phases meet [12,13] is
defined as
gr s l = − (1)
where s is the angle between the tangent to the solid–gas
interface and the horizontal, and l the tangent to the
liquid–gas interface and the horizontal. Assuming all fluids
(i.e. gas and liquid) to be incompressible, immiscible and
Newtonian, the one-fluid formulation gives
( ) ( ) Tt p ( ) + = − + +u uu u u
( )f f
f
dS+ − + x x n f g (2)
( ) ( ) ( )p pC T t C T k T + = u
( )f
f
q dS+ − x x (3)
0 =u (4)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 27
Figure 1. A liquid drop solidifying on a cold plate: (a) computational domain and (b) front-tracking representation
Here, u is the velocity vector, p is the pressure, g is the
acceleration induced by gravity, T is the temperature and
f is the forcing term used to enforce the non-slip velocity
boundary condition at the solid surface. , , k and Cp are
respectively the density, viscosity, thermal conductivity
and heat capacity that are assumed constant in each phase.
The Dirac delta function δ(x − xf) is zero everywhere
except a unit impulse at the interfaces xf with f denoting
interface. The superscript T denotes the transpose. At the
liquid–gas interface, and are the interfacial tension
coefficient and twice mean curvature, respectively. q is the
heat source at the solidification interface, given as
s l
s l
T Tq k k
n n
= −
(5)
where the subscripts s, l and g (when available) represent
solid, liquid and gas, respectively.
For some phase change materials such as water, silicon,
and germanium, the density of the solid phase is different
from that of the liquid phase, and thus change in volume
occurs during solidification. Accordingly, Eq. (4) is
rewritten as follows [12]
( )1 1 1
f
h s l f
qdSL
= − −
u x x (6)
where Lh denotes the latent heat. The boundary conditions
are as follows: symmetry at the left, full slip at the right,
open at the top and non-slip at the bottom (Figure 1a).
3. Numerical method
To solve the problem, we use a front-tracking method
for three phase simulations with an interpolation technique
to treat the presence of the solid phase within the
computational domain on which a uniformly distributed
rectangular grid is constructed [14]. The spatial derivaties
are descritized by a second-order central difference
approximation. The time integration is approximated by a
predictor-corrector scheme. The interface separating
different phases is represented by connected points moving
on the fixed grid (Figure 1b). We update the position of the
liquid–gas and solid–liquid front points by
1n n
f f f fV t+ = + x x n (7)
where the superscripts n and n+1denote the current and
next time levels. nf is the unit vector normal to the
interface. Vf, the velocity magnitude of the front point, is
given as [15]
( ) ( )
( )
1 0 1
where 1 1 0
0 1
f f x y ijV d r d r
r, r ,
d r r, r ,
, r
=
−
= + −
n u
(8)
for the liquid–gas front, and ( )f s hV q L= − (9)
for the solid–liquid front with q calculated by a normal
probe technique
( ) ( )1
s s m l m lq k T T k T Th
= − − − (10)
where h is the grid spacing. Ts and Tl are the temperature
of the solid and liquid near the phase change front at the
solid and liquid points that are h away from the front. At
the triple point, we impose a constant growth angle,
gr = constant [12]. Thereby, the solid–gas points are
constructed.
To identify the phases as well as their fluid and thermal
properties, we construct two indicator functions Is (from
the solid–liquid and solid–gas interfaces) and Il (from the
liquid–gas and solid–gas interfaces) whose values are
specified as
0 in solid, and 1 in liquid and gas
0 in solid and liquid, and 1 in gas
s s
l l
I I
I I
= =
= = (11)
Accordingly, the values of the material properties such
as , , k and Cp (represented by ) at every location in the
domain are given by
( )1l s s s l g lI I I I = + − + (12)
A more detailed description of the method used in this
study can be found in [12,14].
4. Numerical parameters
We choose an equivalent radius of the drop
Rc = R = ( )1 3
03 4V as a scaling length. Here, V0 is the
28 Vu Van Truong, Truong Viet Anh, Tran Xuan Bo, Truong Van Thuan
volume of the initial liquid drop. An alternative scaling
length is the wetting radius Rw (Figure 1), i.e. Rc = Rw. The
characteristic time scale is 2c l l c lC R k = . The
characteristic velocity scale is taken to be .c c cU R =
With these above choices, the dynamics of the problem is
governed by the following dimensionless parameters
(Prandtl number Pr, Stefan number St, Bond number Bo,
Ohnesorge number Oh, dimensionless initial temperature
0, density ratios sl and gl, viscosity ratio gl, thermal
conductivity ratios ksl and kgl, heat capacity ratios Cpsl and
Cpgl):
2pl l pl m c l c l
l h l c
C C (T T ) gRPr ,St ,Bo ,Oh
k L R
−= = = = (13)
00
g gc ssl gl gl
m c l l l
T T, , ,
T T
−= = = =
− (14)
g ps pgs
sl gl psl pgl
l l pl pl
k C Ckk , k , C , C
k k C C= = = = (15)
Figure 2. Evolution of the solidifying front at different stages of solidification with the temperature contours (color) plotted every
= 0.1 and the velocity field normalized by Uc. Rc is equal to R. The parameters are shown in the text
The temperature is non-dimensionalized as
( ) ( ).c m cT T T T = − − The dimensionless time is = t/c.
5. Results and discussion
Figure 2 shows the temporal evolution of the
solidification front with the temperature contours and the
velocity field. The parameters for this calculation are
Pr = 7.25, St = 0.104, Oh = 0.2, Bo = 2.0, sl = 0.9,
gl = gl = 0.05, ksl = 4.0, kgl = 0.05, Cpsl = 0.5, Cpsl = 0.24,
gr = 00, 0 = 1 and 0= 900. At = 0.016 (Figure 2a), the
gravity results in a downward flow at the center of the drop
and deforms the drop, inducing counter clockwise
circulations around the drop. This downward flow causes a
reduction in the temperature field within the liquid phase
while the temperature in the solid phase increases from the
wall value ( =c = 0) at the wall to the fusion value
( =m = 1) at the solid–liquid interface. At a later time
= 0.4, the gravity balances with the surface tension force
holding the drop in a spherical shape, and thus no flow
appears at this time (Figure 2b). Accordingly, the
temperature in the liquid phase is almost at the fusion value.
At = 1.25 (Figure 2c), almost all liquid has solidified, and
the solidified drop has an apex at the top because of volume
expansion (i.e., sl = 0.9) [12,14].
Figure 3 shows the temporal variation of the drop
height and the drop volume for the case shown in Figure 2.
This figure clearly shows the effect of the gravity at the
beginning of solidification with a decrease in the height of
the drop. Then, the surface tension force acting on the
liquid–gas interface pushes the liquid–gas interface up to
be against the gravity. Because the liquid is denser than the
solid (sl < 1.0), the drop height and volume increase in
time as the solidification proceeds as shown in Figure 3.
After complete solidification, the volume of the solidified
drop is about 1.1 times the volume of the initial liquid drop.
A few phase change materials having such a feature, i.e.
volume expansion upon solidification, include water,
silicon and germanium [5]. However, most metals have the
solid denser than the liquid and thus experience shrinkage
upon solidification.
Figure 3. Temporal variation of the solidifying front and the
drop volume for the case shown in Figure 2
To evaluate the capability of the method for
computations of the solidification process of a drop, we
perform a simulation of a water drop freezing on a cold
plate and compared the results with the experiment of
Zhang et al. [3], as shown in Figure 4. Experimentally,
Zhang and co-workers [3] placed a water drop of 10 L
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 29
on a horizontal wall that was kept at Tc = –16.50C, and
used a photographic technique to capture the evolution of
the ice–water phase change interface during freezing
(bottom row in Figure 4). The corresponding Prandtl,
Stefan and Bond numbers for this case are Pr = 7.5,
St = 0.209 and Bo = 0.25, based on the properties of water
and ice and Rc = R. The growth angle gr is set to 00 [5].
The top row in Figure 4 shows the time sequence of the
freezing of the drop yielded from the present computation.
The comparison indicates that the computation result
agrees quite well with the experimental data of
Zhang et al. [3].
Figure 4. Evolution of the water–ice front at different stages of freezing in comparison with
the experiments of Zhang et al. [3]. Rc is equal to R
Figure 5 shows the evolution of the average height Hs
of the freezing front for the case shown in Figure 4. The
variation with respect to time of Hs indicates that the
freezing rate is high at the beginning and near the end of
the freezing process. This tendency is in accordance with
Nauenberg’s theoretical analysis [7]. As previously
mentioned, water is a phase change material whose liquid
density is higher than that of ice, and thus the drop expands
in volume as the freezing process progresses, with a
volume increment of about 10% after complete
solidification, as shown in Figure 5. Figure 5 also confirms
that the increase in the volume of the computational drop
is in good agreement with the experimental result of Zhang
et al. [3].
Figure 5. Temporal evolutions of the average height of the
water–ice front Hs and of the volume of the drop Vd normalized
by the volume of the initial liquid drop V0. The cirles are data
from the experiment of Zhang et al. [3]
Silicon, a semiconductor material, also has a liquid
phase denser than the solid phase, and thus experiences
volume expansion upon crystallization. Figure 6 shows the
evolution of the solid–liquid interface during the
crystallization process of a molten silicon drop attaching to
a cold wall. The main parameters for this computation are
Pr = 0.013, St = 0.116, sl = 0.91 and gr = 120, based on
the properties of silicon [5]. Because of volume expansion
and the effect of the growth angle, the crystallized silicon
drop is very different from the initial molten one with an
apex at the drop top. This crystallized drop shape agrees
very well with the experimental drop shape (the most right
frame in Figure 6) reported in Satunkin’s work [5]. For
more details of the comparison, see [11].
Figure 7 shows the evolution of the solidifying
interface during the solidification process of an indium
antimonide drop whose main parameters are Pr = 0.0255,
St = 0.278, sl = 0.8 and gr = 250. The corresponding
solidified drop yielded from Satunkin’s experiment [5] is
present at the right of Figure 7. Because of the significant
effects of the growth angle and volume change, the initial
molten drop with a spherical cap solidifies to a very conical
solid drop with a large increase in its height (the final
height is more than two times higher than the initial height).
This shape of the solidified drop is in good agreement with
the experiment one reported by Satunkin [5]. For more
details of the comparison, see [11].
Based on the comparisons with the available
experiments, we can conclude that the method presented
here can accurately capture the evolution of the solidifying
drop attached to a cold wall.
6. Conclusion
We have presented a numerical method to simulate the
solidification process of a liquid drop. The method is the
front-tracking technique representing the interface by
connected elements laid on a fixed rectangular grid. The
drop with the presence of three phases (solid, liquid and
gas) is immersed in the computational domain. The Navier-
Stokes and energy equations are used in the entire domain
with the solid treated by the interpolation technique. The
propagation of the solid–liquid interface is calculated by
the normal probe technique. The liquid drop initialy treated
as a section of a sphere is placed on the plate whose
temperature below the fusion value of the liquid causes the
30 Vu Van Truong, Truong Viet Anh, Tran Xuan Bo, Truong Van Thuan
solidifying front to form and move upwards. The numerical
results show that the solidification process progresses fast
at the begining, and the solidified drop forms a cone at the
top of the drop in the case of volume expansion. The
comparisons with the available experiments for various
phase change materials including water, silicon, and
indium antimonide show that the numerical method
accurately captures the evolution of the solidification
interface as wells as the drop shape after complete
solidification.
Figure 6. The solidification process of a molten silicon drop in comparison with the crystallized drop of Satunkin [5]
Figure 7. The solidification process of a molten indium antimonide drop in comparison with the crystallized drop of Satunkin [5]
Acknowledgments
This research is funded by Vietnam National
Foundation for Science and Technology Development
(NAFOSTED) under grant number 107.03-2017.01.
REFERENCES
[1] O.R. Enríquez, Á.G. Marín, K.G. Winkels, J.H. Snoeijer, Freezing singularities in water drops, Phys. Fluids. 24 (2012) 091102-
091102-2.
[2] Z. Jin, X. Cheng, Z. Yang, Experimental investigation of the
successive freezing processes of water droplets on an ice surface,
Int. J. Heat Mass Transfer. 107 (2017) 906–915.
[3] X. Zhang, X. Wu, J. Min, Freezing and melting of a sessile water
droplet on a horizontal cold plate, Exp. Therm. Fluid Sci. 88 (2017) 1–7.
[4] H. Itoh, H. Okamura, C. Nakamura, T. Abe, M. Nakayama, R.
Komatsu, Growth of spherical Si crystals on porous Si3N4 substrate
that repels Si melt, J. Cryst. Growth. 401 (2014) 748–752.
[5] G.A. Satunkin, Determination of growth angles, wetting angles,
interfacial tensions and capillary constant values of melts, J. Cryst.
Growth. 255 (2003) 170–189.
[6] A. Sanz, The crystallization of a molten sphere, J. Cryst. Growth. 74
(1986) 642–655.
[7] M. Nauenberg, Theory and experiments on the ice–water front
propagation in droplets freezing on a subzero surface, Eur. J. Phys. 37 (2016) 045102.
[8] X. Zhang, X. Wu, J. Min, X. Liu, Modelling of sessile water droplet
shape evolution during freezing with consideration of supercooling
effect, Appl. Therm. Eng. 125 (2017) 644–651.
[9] W.W. Schultz, M.G. Worster, D.M. Anderson, Solidifying sessile
water droplets, in: Interactive Dynamics of Convection and Solidification, Kluwer Academic Publishers, 2001: pp. 209–226.
[10] A. Virozub, I.G. Rasin, S. Brandon, Revisiting the constant growth
angle: Estimation and verification via rigorous thermal modeling, J.
Cryst. Growth. 310 (2008) 5416–5422.
[11] T.V. Vu, G. Tryggvason, S. Homma, J.C. Wells, H. Takakura, A
front-tracking method for three-phase computations of solidification
with volume change, J. Chem. Eng. Jpn. 46 (2013) 726–731.
[12] T.V. Vu, G. Tryggvason, S. Homma, J.C. Wells, Numerical
investigations of drop solidification on a cold plate in the presence of volume change, Int. J. Multiphase Flow. 76 (2015) 73–85.
[13] T.V. Vu, Three-phase computation of solidification in an open
horizontal circular cylinder, Int. J. Heat Mass Transfer. 111 (2017)
398–409.
[14] T.V. Vu, J.C. Wells, Numerical simulations of solidification around
two tandemly-arranged circular cylinders under forced convection,
Int. J. Multiphase Flow. 89 (2017) 331–344.
[15] G. Tryggvason, R. Scardovelli, S. Zaleski, Direct numerical
simulations of gas-liquid multiphase flows, Cambridge University Press, Cambridge; New York, 2011.
(The Board of Editors received the paper on 25/02/2018, its review was completed on 28/3/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 31
MULTI-PERIOD LINEARIzED OPTIMAL POWER FLOW MODEL
INCORPORATING TRANSMISSION LOSSES AND THYRISTOR CONTROLLED
SERIES COMPENSATORS
Pham Nang Van1, Le Thi Minh Chau1, Pham Thu Tra My2, Pham Xuan Giap2, Ha Duy Duc2, Tran Manh Tri2 1Hanoi University of Science and Technology (HUST); [email protected], [email protected]
2Student at Department of Electric Power Systems, Hanoi University of Science and Technology (HUST)
Abstract - This paper presents multi-period linearized optimal power flow (MPLOPF) with the consideration of transmission network losses and Thyristor Controlled Series Compensators (TCSC). The transmission losses are represented using piecewise linear approximation based on line flows. In addition, the nonlinearity due to the impedance variation of transmission line with TCSC is linearized deploying the big-M based complementary constraints. The proposed model in this paper is evaluated using PJM 5-bus test system. The impact of a variety of factors, for instance, the number of linear blocks, the location of TCSC and the ramp rate constraints on the power output and locational marginal price (LMP) is also analyzed using this proposed model.
Key words - Multi-period linearized optimal power flow (MPLOPF); mixed-integer linear programming (MILP); transmission losses; Thyristor Controlled Series Compensators (TCSC); big-M
1. Introduction
Electricity networks around the world are experiencing
extensive change in both operation and infrastructure due
to the electricity market liberalization and our increased
focus on eco-friendly generation. Managing and operating
power systems with considerable penetration of renewable
energy sources (RES) is an enormous challenge and many
approaches are applied to cope with RES integration,
mainly the management of intermittency. In addition to
increasing power reserves, energy storage systems (ESS)
can be invested to mitigate the uncertainty of RES. The
increasing application of ESS as well as problems
including time-coupled formulations such as power grid
planning, N-1 secure dispatch and optimal reserve
allocation for outage scenarios have led to extended
optimal power flow (OPF) model referred to as multi-
period OPF problems (MPOPF) [1]-[2].
Typically, the MPOPF problem is approximated using the
DC due to its convexity, robustness and speed in the electricity
market calculation [3]. To improve the accuracy of the
MPOPF model, transmission power losses have been
integrated. This is significant because the losses typically
account for 3% to 5% of total system load [4]. When power
losses are incorporated in the MPOPF model, this model
becomes nonlinear. To address the nonlinearity, reference [3]
deploys the iterative algorithm based on the concept of
fictitious nodal demand (FND). The disadvantage of this
approach is that the MPOPF problem must be iteratively
solved. Reference [5] presents another approach in which
branch losses are linearized. The branch losses can be
expressed as the difference between node phase angles or line
flows [4]. The main drawback of this model is that it can lead
to “artificial losses” without introducing binary variables [5].
Moreover, the TCSC is increasingly leveraged in power
systems to improve power transfer limits, to enhance
power system stability, to reduce congestion in power
market operations and to decrease power losses in the grid
[6]. When integrating TCSC in the MPOPF problem, this
model becomes nonlinear and non-convex since the TCSC
reactance becomes a variable to be found [7]. At present,
there are several strong solvers like CONOPT, KNITRO
for solving this nonlinear optimization problem [8].
However, directly solving nonlinear optimization
problems cannot guarantee the global optimal solution.
References [9]-[10] demonstrate the relaxation technique
to solve the nonlinear optimization problem in power
system expansion planning considering TCSC investment.
Furthermore, the iterative method is used to determine
optimal parameter of TCSC in reference [11].
The main contributions of the paper are as follows:
- Combining different linearized techniques to convert
the nonlinear MPOPF to the mixed-integer linear MPOPF.
- Analysing the impact of some factors such as the
number of loss linear segments, the location of TCSC as
well as the ramp rate of the units on the locational marginal
price (LMP) and generation output.
The next sections of the article are organized as
follows. In section 2, the authors present general
mathematical formulation of multi-period optimal power
flow (MPOPF) model incorporating losses and TCSC. The
different linearization techniques are specifically presented
in section 3 and 4. Section 5 demonstrates multi-period
linearized optimal power flow (MPLOPF) model. The
simulation results, numerical analyses of PJM 5-bus
system are given in section 6. Section 7 provides some
concluding remarks.
2. General mathematical formulation
For normal operation conditions, the node voltage can
be assumed to be flat. A multi-period optimal power flow
(MPOPF) considering network constraints can be modeled
for all hour t, all buses n, all generators i, and all lines (s, r)
as follows:
( ) ( )( )
,min , . ,
i
gi giP
t T i I b G t
b t P b t
(1)
Subject to
( )( )
( )( )
( ): , : ,
, 0
,
g d
gi dj n
i i n M j j n M
P t P t P t
n N t T
− − =
(2)
( ) ( ) ( )max , ; , ; , ,ub lsr rs srP t P t P s r t T (3)
( ) ( ) ( )0 , , ; , ,ubgi gi iP b t P b t i I b G t t T (4)
32 Pham Nang Van, Le Thi Minh Chau, Pham Thu Tra My, Pham Xuan Giap, Ha Duy Duc, Tran Manh Tri
( ) ; ,lb ubgi gi giP P t P i I t T (5)
( ) ( )1 ; ,upgi gi iP t P t R i I t T− − (6)
( ) ( )1 ; ,dngi gi iP t P t R i I t T− − (7)
The objective function in (1) represents the total
system cost in T hours (here, T = 24 h). The constraints
(2) enforce the power balance at every node and every
hour. The constraints (3) enforce the line flow limits at
every hour. The constraints (4) and (5) are operating
constraints that specify that a generator’s power output as
well as power output of each energy block must be within
a certain range. The other constraints included in the
formulation above are the ramp-up constraints (6) and
ramp-down constraints (7).
If the reactance of branch xsr is taken as a variable due
to TCSC installation, in the range of min max[ , ]sr srx x , it yields
a new model:
( ) ( )( )
, ,min , . ,
sri
gi giP x
t T i I b G t
b t P b t
(8)
Subject to
min maxsr sr srx x x (9)
( ) ( )2 7− (10)
The above general model is nonlinear. Sections 3 and 4
present different linearization methods to convert this
model to the linear form.
3. Linearization of the network losses
In this section, the subscript t is dropped for notational
simplicity. However, it could appear in every variable and
constraint. Additionally, the expressions presented below
apply to every transmission line; therefore, the indication
( ), ls r will be explicitly omitted.
The real power flows in the line (s, r) determined at bus
s and r, respectively, are given by
( ) ( ) ( ), 1 cos sinsr s r sr s r sr s rP G B = − − − − (11)
( ) ( ) ( ), 1 cos sinrs s r sr s r sr s rP G B = − − + − (12)
The real power loss in the line (s, r), ( ),losssr s rP can
be attained as follows:
( ) ( ) ( ) ( )2
, , ,losssr s r sr s r rs s r sr s rP P P G = + − (13)
In the lossless DC model, the real power flow in the line
(s, r) at bus s is approximately calculated as in (14):
( ) ( ) ( )1
,sr s r sr s r s rsr
F BX
− − = − (14)
Substituting (14) in (13), the real power loss in the line
(s, r) is expressed as in (15):
( ) ( )( )
2 2
2,
1 /
loss srsr s r sr sr sr sr
sr sr
RP G X F F
R X = =
+ (15)
Equation (15) can be further simplified. The resistance
Rsr is usually much smaller than its reactance Xsr,
particularly in high voltage lines. Consequently, (15) can
be further reduced to (16)
( ) 2,losssr s r sr srP R F = (16)
The first advantage of (16) compared to (13) is that
power flows in lines neither built nor operative are zero.
Another advantage of (16) is its possible application to
model losses in HVDC lines.
The quadratic losses function (16) can be expressed
using piecewise linear approximation according to
absolute value of the line flow variable as follows:
( ) ( ) ( )1
,L
losssr s r sr sr sr
l
P R l F l =
= (17)
To complete the piecewise linearization of the power
flows and line loss, the following constraints are necessary
to enforce adjacency blocks:
( ) ( )max. ; 1,..., 1sr sr srl p F l l L = − (18)
( ) ( ) max1 . ; 2,...,sr sr srF l l p l L − = (19)
( ) ( )1 ; 2,..., 1sr srl l l L − = − (20)
( ) 0; 1,...,srF l l L = (21)
( ) 0;1 ; 1,..., 1sr l l L = − (22)
Constraints (18) and (19) set the upper limit of the
contribution of each branch flow block to the total power
flow in line (s, r). This contribution is non-negative, which
is expressed in (21) and limited upper by max /ubsr srp P L = ,
the “length” of each segment of line flow (18). A set of
binary variables ( )sr l is deployed to guarantee that the
linear blocks on the left will always be filled up first;
therefore, this model eliminates the fictitious losses. Finally,
constraints (22) state that the variables ( )sr l are binary.
A linear expression of the absolute value in (17) is
needed, which is obtained by means of the following
substitutions:
sr sr srF F F+ −= + (23)
sr sr srF F F+ −= − (24)
( )0 1 ubsr sr srF P− − (25)
0 ubsr sr srF P+ (26)
In (24), two slack variables srF +and srF −
are used to
replace Fsr. Constraints (25) and (26) with binary variable θsr
ensure that the right-hand side of (23) equals its left-hand side.
Moreover, the slopes of the blocks of line flow ( )sr l
for all transmission lines can be given by Eq. (27).
( ) ( ) max2 1sr srl l p = − (27)
It is emphasized that the number of linear segments will
radically affect the accuracy of the optimal problem
solution. Moreover, this linear technique is independent of
the reference bus selection and thereby eliminating
discrimination in the electricity market operation.
Using the above expressions, the real power flow in line
(s, r) computed at bus s and r can be recast as follows,
respectively:
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 33
( ) ( )
( ) ( )1
1, ,
2
1
2
losssr s r sr s r sr
L
sr sr sr sr
l
P P F
R l F l F
=
= +
= + (28)
( ) ( )
( ) ( )1
1, ,
2
1
2
lossrs s r sr s r sr
L
sr sr sr sr
l
P P F
R l F l F
=
= −
= − (29)
The power withdrawn into a node n, ( ),nP t can be
written as
( ) ( )1:( , )
1
2l
L
n nk nk nk nk
lk n k
P R l F l F=
= +
(30)
A linear substitution for the function in (3) can be found
by the following equivalent constraints without increasing
the number of rows
( ) ( )1
1
2
Lub
sr sr sr sr sr
l
R l F l F P=
+ (31)
Rewriting Eq. (31), the constraints (3) are expressed as
follows
( ) ( )1
11
2
Lub
sr sr sr sr
l
R l F l P=
+
(32)
4. Linearization of a bilinear function
When xsr is taken as a variable, constraint (14) also
makes the MPOPF model nonlinear since this constraint is
a bilinear function. To overcome the nonlinearity of this
constraint, we introduce a new variable Fsr, instead of
variable xsr. After obtaining the optimal solution with
variable (P, F, δ), the optimal reactance can be uniquely
determined according to Eq. (33)
s rsr
sr
xF
−= (33)
Therefore, the constraint (9) becomes:
min maxs rsr sr sr
sr
x x xF
− = (34)
It is noted that the sign of Fsr cannot be determined
beforehand. Moreover, if the denominator Fsr is zero, the
numerator s r − must be zero. As a result, (34) can be
converted into the expression (35) depending on the sign
of Fsr.
min max
max min
0 .
0 0
0 .
sr sr sr s r sr sr
sr s r
sr sr sr s r sr sr
if F F x F x
if F
if F F x F x
−
= − =
−
(35)
These condition constraints can be combined by
leveraging binary variables ysr and big-M based
complementary constraints as follows [12]. In our model, M
is taken to be / 2 due to system stability requirement [13].
( ) ( )
min max
max min1 1
sr sr sr s r sr sr sr
sr sr sr s r sr sr sr
My F x F x My
M y F x F x M y
− + − +− − + − + −
(36)
It is important to stress that linear technique using the
above binary variable is exact while the linearized
technique in Section 3 is approximately presented.
5. Multi-period linearized optimal power flow
(MPLOPF) model with losses and TCSC
The MPLOPF model with losses and TCSC has the
following form:
( ) ( )( )
, ,min , . ,
i
gi giP F
t T i I l G t
b t P b t
(37)
Subject to
( )( )
( )( )
( ) ( ) ( )
( ) ( )
: , : ,
1
:( , )
1 1
1, ,
2; ,
, ,
g d
l
gi dj
i i n M j j n M
L
nk nk nk nk
l
L Lk n k
nk nk
l l
P t P t
R l F l t F l t
n t
F l t F l t
+ −
=
+ −
= =
− =
+
+ −
(38)
( ) ( ) ( )1
11 , ,
2
Lub
sr sr sr sr sr
l
R l F l t F l t P + −
=
+ + (39)
( ) ( ) ( )max, . , F , ; 1,..., 1sr sr sr srl t p F l t l t l L + − + = − (40)
( ) ( ) ( ) max, , 1, . ; 2,...,sr sr sr srF l t F l t l t p l L+ −+ − = (41)
( ) ( ), 1, ; 2,..., 1;sr srl t l t l L − = − (42)
( ) ( ) ( ) , 0; , 0; , 0;1sr srF l t F l t l t+ − = (43)
( ) ( ) ( )1
0 , ; , ,L
ub lsr sr
l
F l t t P s r t T+
=
(44)
( ) ( ) ( )1
0 , 1 ; , ,L
ub lsr sr
l
F l t t P s r t T−
=
− (45)
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
min
max
max
min
1
1
sr sr sr s r
s r sr sr sr
sr sr sr s r
s r sr sr sr
My t F t x t t
t t F t x My t
M y t F t x t t
t t F t x M y t
− + − − +− − + −
− + −
(46)
( ) ( )4 7− (47)
Regarding the computational complexity of the model,
the number of continuous variable is 24. .GEN GENiN N
( )24. 1 2.24. .BUS LINN N L+ − + and the number of binary
variables is ( )24. . 1 2.24.LIN LINN L N− + .
After the MPLOPF problem is solved, the marginal cost
at the node i in hour t can be determined by the following
expression [3]:
. .i E i E l i l
l
LMP LMP LF LMP SF −= − + (48)
6. Results and discussions
In this section, the multi-period linearized optimal
power flow model is performed on the modified PJM 5-bus
system [3]. The MPLOPF problem is solved by CPLEX
12.7 [15] under MATLAB environment.
34 Pham Nang Van, Le Thi Minh Chau, Pham Thu Tra My, Pham Xuan Giap, Ha Duy Duc, Tran Manh Tri
6.1. System data
The test system is shown in Figure 1. The total peak
demand in this system is 1080 MW and the total load is
equally distributed among buses B, C and D. The daily load
curve is depicted in Figure 2. Two small size generators on
bus A have the capability to quickly start up. The ramp rate
for the other generators is 50% of the rated power output [14].
E D
A
B C
Limit=240 MW
Brighton
Park
City Load
CenterSolitude
Sundance
110MW
$14
600MW
$10
200MW
$35
520MW
$30100MW
$15
Figure 1. PJM 5-bus system and generation parameters
Figure 2. Daily load curve for PJM system
6.2. Impact from the number of linear blocks
Table 1. The effects of number of linear blocks
Linear blocks Objective ($) Total losses (MW) Time (s)
2 3844.43 316.69 1.71
4 3824.04 244.83 2.97
6 3822.96 238.56 5.28
8 3820.70 230.41 8.42
10 3820.55 229.49 12.35
11 3820.51 229.49 14.61
The number of linear blocks can significantly affect the
solution time as well as the model accuracy listed in Table
1. The key idea in this paper is to find the number of linear
blocks which give the best balance between the model
accuracy and the solution time. In this case, 10 is an
appropriate number in terms of objective value, total losses
and calcultaion time.
6.3. Impact from losses
Table 2 compares the results of power output at 10 AM
using the proposed model. These results are also compared
with those of POWERWORLD software using the ACOPF
model [16]. When comparing to POWERWORLD
software, the calculated results using the proposed model
considering losses are more accurate and less different than
that of the model neglecting losses.
Table 2. Generating output results at 10 AM
Bus Lossless (MW) Losses (MW) POWERWORLD (MW)
A1 110 110 110
A2 100 100 100
C 19.95 30.1 27.83
D 195.05 194.8 197.2
E 600 600 600
Figure 3. LMP at bus B at different hours without losses
and with losses
The results of LMP calculations at node B for 24 hours
using the proposed model with and without losses are given
in Figure 3. This figure illustrates that the effect of power
losses on LMP is very little. This result is consistent because
the power losses account for about 1% of the total load for
this PJM 5-bus system, therefore the marginal generating
units as well as congested lines are the same in both cases.
6.4. Impact from TCSC location
It is assumed that power losses are not considered and
the ramp rate of the generating units (not including units
at node A) are taken as 25% of the maximum power
output. Also, the compensation level of TCSC varies from
30% to 70%.
Figure 4 depicts the power output of generator at node
C for 24 hours for different locations of TCSC. During the
period from 1 AM to 3 AM, the power output of the unit at
node C nearly remains when the location of TCSC varies.
In addition, the power output of this unit is highest in 24
hours when TCSC is located in line A-B.
Figure 4. The dependence of Generating output of
Unit at bus C on TCSC location
6.5. Impact from ramp rate constraints
Figure 5 shows the power output of generator located
at node C when changing the ramp rate of generators and
it is assumed that TCSC is not applied to the power grid.
From the 5 AM to 24 PM, the power output of this unit is
the same for ramp rates of 25%, 35% and 50%. At the same
time, the output of this unit is the highest for ramp rate
100% of the maximum power.
900
950
1000
1050
1100
0 5 10 15 20 25
Load
(M
W)
Hour
20
25
30
35
0 5 10 15 20 25
LM
P (
$/M
Wh
)
Hour
Losses
Lossless
0
200
400
0 5 10 15 20 25
Gen
era
tio
n (
MW
))
Hour
Line A-B
Line B-C
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 35
Figure 6 depicts the effect of TCSC placement on the
power output with different ramp rate scenarios at 10 AM.
We see that the power output of generator at node C does
not change as the ramp rate of the units changes in case of
placing TCSC on line A-B. However, when TCSC is not
installed, the ramp rate of units has a significant effect on
the unit's output, increasing from 30,097 MW for the ramp
rate of 50% to 223,37 MW for the ramp rate of 100%. Thus,
using TCSC also reduces the impact of the ramp rate on the
power output.
Figure 5. The dependence of generating output of
Unit at bus C on Ramp rate without TCSC
Figure 6. The dependence of power output of
Unit at bus C on Ramp rate with TCSC in line A-B at 10 AM
7. Conclusion
This paper presents multi-period linearized optimal
power flow (MPLOPF) model based mixed-integer linear
programming (MILP). This MPLOPF integrates line losses
and Thyristor Controlled Series Compensator (TCSC). The
different linearization techniques, such as piecewise linear
approximation and big-M based complementary
constraints are deployed to convert multi-period nonlinear
OPF problem to multi-period linearized OPF model. The
calculated results using the proposed model are compared
to those of the commercial POWERWORLD software and
this proves the validation of the proposed model.
Additionally, the influences of the number of linear blocks,
line losses, location of TCSC and ramp rate are analyzed.
The results reveal that these factors can importantly impact
on LMP, generating output of units as well as revenue of
participants in electricity markets.
NOMENCLATURE
The main mathematical symbols used throughout this
paper are classified below.
Constants:
( )sr l Slope of the lth segment of the linearized power flow
in line (s, r)
( ),gi b t Offered price of the bth linear block of the energy bid
by the ith generating unit in hour t
srB Imaginary part of the admittance of line (s, r)
srG Real part of the admittance of line (s, r)
srR Resistance of the line (s, r)
srX Reactance of the line (s, r)
( )djP t Power consumed by the jth load in hour t
L Number of the blocks of the loss linearization ubsrP Transmission limit of line (s, r)
ubgiP
Upper bound on the power output of the ith producer
lbgiP
Lower bound on the power output of the ith producer
upiR Ramp-up limit of the ith unit
dniR Ramp-down limit of the ith unit
minsrx Lower bound of the reactance of the line with TCSC
maxsrx Upper bound of the reactance of the line with TCSC
BUSN Number of nodes
GENN Number of generators
LINN Number of transmission lines
GENiN Number of energy blocks of unit i
Variables:
( ),giP b t Power output corresponding to the bth block of the
ith unit in hour t
( ),nP t Power withdrawal at bus n in hour t
( ),srP t Power flow in line (s, r) at node s in hour t
( ),rsP t Power flow in line (s, r) at node r in hour t
( )s t Voltage angle at node s in hour t
( )srF t Power flow in line (s, r) in hour t without losses
( ),losssrP t Power losses in line (s, r) in hour t
( )sr l Binary variable relating to the line flow linearization
( )sry t Binary variable corresponding the big-M based
complementary constraints
( )srx t The reactance of the line with TCSC in hour t
iLF Loss factor at bus i
l iSF − Sensitivity of branch power flow l with respect to
injected power i
l Shadow price of transmission constraint on line l
Sets: I Set of indices of the generating units
( )iG t Set of blocks energy bid offered by the ith unit in
hour t N Set of indices of the network nodes
l Set of transmission lines
ACKNOWLEDGMENT
This research is funded by the Hanoi University of
Science and Technology (HUST) under project number
T2017-PC-093.
REFERENCES
[1] D. Kourounis, A. Fuchs, and O. Schenk, “Towards the next
generation of multiperiod optimal power flow solvers”, IEEE Trans.
0
200
400
1 3 5 7 9 11 13 15 17 19 21 23
Gen
era
tio
n (
MW
)
Hour
Ramp rate 25%
Ramp rate 35%
Ramp rate 50%
0
50
100
150
200
250
25% 35% 50% 100%
Gen
erat
ion (
MW
)
Ramp-rate
TCSC in line A-B
No TCSC
36 Pham Nang Van, Le Thi Minh Chau, Pham Thu Tra My, Pham Xuan Giap, Ha Duy Duc, Tran Manh Tri
Power Syst., vol. 8950, pp. 1–10, 2018.
[2] P. N. Van, N. D. Huy, N. Van Duong, and N. T. Huu, “A tool for
unit commitment schedule in day-ahead pool based electricity
markets”, J. Sci. Technol. Univ. Danang, vol. 6, pp. 21–25, 2016.
[3] F. Li, S. Member, R. Bo, and S. Member, “DCOPF-Based LMP
Simulation : Algorithm, comparison with ACOPF and sensitivity”, IEEE Trans. Power Syst., vol. 22, no. 4, pp. 1475–1485, 2007.
[4] D. Z. Fitiwi, L. Olmos, M. Rivier, F. de Cuadra, and I. J. Pérez-
Arriaga, “Finding a representative network losses model for large-
scale transmission expansion planning with renewable energy
sources”, Energy, vol. 101, pp. 343–358, 2016.
[5] J. M. Arroyo and A. J. Conejo, “Network-constrained Multiperiod auction for a pool-based electricity market”, IEEE Trans. Power
Syst., vol. 17, no. 4, pp. 1225–1231, 2002.
[6] P. N. Van, N. D. Hung, and N. D. Huy, “The impact of TCSC on
transmission costs in wholesale power markets considering bilateral
transactions and active power reserves”, J. Sci. Technol. Univ. Danang, vol. 12, pp. 24–28, 2016.
[7] G. Y. Yang, G. Hovland, R. Majumder, and Z. Y. Dong, “TCSC allocation
based on line flow based equations via mixed-integer programming”, IEEE
Trans. Power Syst., vol. 22, no. 4, pp. 2262–2269, 2007.
[8] Alireza Soroudi, Power System Optimization Modeling in GAMS.
Springer, 2017.
[9] O. Ziaee, O. Alizadeh Mousavi, and F. Choobineh, “Co-
optimization of transmission expansion planning and TCSC placement considering the correlation between wind and demand
scenarios”, IEEE Trans. Power Syst., vol. 8950, no. c, pp. 1–1, 2017.
[10] M. Farivar and S. H. Low, “Branch Flow Model: Relaxations and
Convexification (Parts I, II)”, pp. 1–11, 2012.
[11] P. N. Van and L. M. Khanh, “The optimal location and
compensation level of Thyristor Controlled Series Compensator
(TCSC) in Wholesale Electricity Markets considering Active Power Reserves”, J. Sci. Technol. Tech. Univ. Vietnam, 2017.
[12] T. Ding, R. Bo, W. Gu, and H. Sun, “Big-M Based MIQP Method
for Economic Dispatch With Disjoint Prohibited Zones”, IEEE
Trans. Power Syst., vol. 29, no. 2, pp. 976–977, 2014.
[13] T. Ding, R. Bo, F. Li, and H. Sun, “Optimal Power Flow with the
Consideration of Flexible Transmission Line Impedance”, IEEE Trans. Power Syst., vol. 31, no. 2, pp. 1655–1656, 2016.
[14] Y. Wei, H. Cui, X. Fang, and F. Li, “Strategic scheduling of energy
storage for load serving entities in locational marginal pricing
market”, IET Gener. Transm. Distrib., vol. 10, no. 5, 2016.
[15] IBM, “IBM ILOG CPLEX Optimization Studio Community
Edition”.
[16] https://www.powerworld.com/
(The Board of Editors received the paper on 18/4/2018, its review was completed on 04/5/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 37
APPLYING SEMISMOOTH NEWTON METHOD TO FIND FIXED POINTS OF
NONSMOOTH FUNCTIONS OF ONE VARIABLE
Pham Quy Muoi, Phan Quang Nhu Anh, Duong Xuan Hiep, Phan Duc Tuan
University of Education – The University of Danang;
[email protected]; [email protected]; [email protected]; [email protected]
Abstract - In this paper, we investigate the problem of finding a
fixed point of the nonsmooth function,1 2
max ( ), ( ), , ( ) .n
f x f x f x
First, we recall the definition of Newton derivative and examine some basic properties. Then, we investigate the Newton
differentiability of function 1 2
max ( ), ( ), , ( ) .n
f x f x f x We give the
necessary and sufficient conditions for Newton differentiability of
this function in two cases: A special case:1 2
max ( ), ( )f x f x and the
general case: 1 2
max ( ), ( ), , ( ) .n
f x f x f x We emphasize that, the
sufficient condition for the special case is much weaker than that of the general case. After that, we apply the semismooth Newton method to find a fixed point of the above function. The local quadratic order convergence of the method is proven. Finally, we present the numerical results for some specific examples.
Key words - Newton Derivative; Newton differential; Fixed point; Semismooth Newton method; Nonsmooth function
1. Introduction
Thought out the history, fixed point theory has been
widely considered by numerous researchers both
domestically and internationally. There is a large amount
of published research including Banach, Browder and
Borel’s fixed point theory, ... [1, 4]. Fixed point theory has
wide applications in abundant areas such as partial
differential equation theory, economics (game theory), ...
[3, 4]. In numerical programming, the familiar method
which has been used is fixed point’s iteration as well as
advanced one [2]. As we know, fixed point iteration
converges in linear speed.
Recently, in optimization theory for non-smooth
problems, there are more advanced algorithms with high
convergent speed that have been studied and improved. In
these algorithms, semismooth Newton’s method for fixed
point problems is widely researched and applied. In this
paper, we investigate semismooth Newton’s method for
fixed point problems and provide quadratic convergence of
our algorithm. Specially, we study this method for fixed
point problems:
1 2( ) max ( ), ( ), , ( ) ,
nF x f x f x f x (1)
where :if C with C ( 1, ,i n ) is
smooth, continuous functions.
Problem (1) appears in different areas, especially in
constrained optimization problems. Some special cases are
most likely to find in the necessary condition of solution in
sparsity problems [7] and non-sparsity problems [8].
Noticeably, fixed point problem of F is equivalent to
finding solution to the following equation
( ) : ( ) 0.G x x F x (2)
We want to emphasize that function F is continuous
function but non-smooth so that G is non-smooth function
too. Hence, the efficient method to find solution to this
function like gradient method and Newton method, ... that
are not utilized. In this paper, we investigate Newton
differentiable of F and apply semismooth Newton to find
solution to equation (2). Finally, we apply this method in
some examples.
The other parts of this paper are organized as follows.
In section 2, we first re-define Newton derivative for one
variable and recall some its properties. In section 3, we
present Newton differentiable of F with 2n . In
section 4, we state Newton differentiable of F for 2n
in general. In section 5, we introduce and prove the
convergence of semismooth Newton method for equation
(2). In conclusion, we also present many numerical
examples.
2. Newton derivative
Definition 2.1. Let U be a open subset of ,
:F U be a function that defines in U . Function F
is Newton differentiable at x U if exist function
: ( , )G U U so that
0
| ( ) ( ) ( ) |lim 0.
| |h
F x h F x G x h h
h
where ( , )U is the space of all linear bounded mappings
from U to .
The function G is called one of Newton derivative F at x .
From the previous definition, we imply that Newton
derivative of F at one point is a function and it is not a real
number like Fréchet derivative.
Remark 2.2. In [7], authors indicated that:
a) Newton derivative is not singular.
b) If function F is continuous and Fréchet
differentiable in ( , )a b then F is Newton differentiable in
( , )a b . Moreover, F is one of Newton derivative of
function F .
c) If function f and g are Newton differentiable at x
then , ,f f g fg are Newton differentiable and one of their
Newton derivative is equivalent to Fréchet derivative [5].
Next, we consider Newton derivative of function F
that was given by (1). Because of the distinction of Newton
differentiable of function F when 2n and general case
( 2n ) we will present its differentiable in individual
38 Pham Quy Muoi, Phan Quang Nhu Anh, Duong Xuan Hiep, Phan Duc Tuan
cases. In addition, the sufficient condition to reach Newton
differentiable of F when 2n (that is presented in [5])
is weaker than this one in general case.
3. Newton derivative of F when 2n
Newton derivative of F when 2n was presented in
[5]. For the convenience for readers, we recall some
important properties in [5] by the following theorem.
Theorem 3.2. Let f and g be continuously Fréchet
differentiable in then function ( ) max{ ( ), ( )}F x f x g x
is Newton differentiable for all x and one of Newton
derivative of ( )F x is G which is defined by
( ),( )
( ), ,
f x x PG x
g x x Q
where,
{ | ( ) ( )}P x f x g x and { | ( ) ( )}.Q x f x g x
4. Newton derivative of F when 2n
Theorem 4.3. Let 1( ),..., ( )
nf x f x be n continuously
differentiable function in which has finite intersections.
Denote 1( ) max{ ( ),..., ( )},
nF x f x f x
( ) { {1,2,..., } | ( ) ( )},
iI x i n f x F x
( ) min ( ).A x I x
Then, ( )F x is Newton differentiable for all 0x
and one of Newton derivatives of ( )F x at 0x is
( )( ) ( ),
A xG x f x x .
Prove. For all 0x , we consider two cases:
Case 1: 0( ) { }, {1,2,.., }.I x k k n
Then 0 0( ) ( ), {1,.., }, .k jf x f x j n j k
Since if is continuous in for all 1,..,i n then for
all x , there must be a 0 that satisfies
0 0( , )x x x and we have
( ) ( ), {1,.., },k jf x f x j n j k .
For h so that 0 0 0( , )x h x x , we have
0 0 0| ( ) ( ) ( ) |
0| |
F x h F x G x h h
h
0 0 0| ( ) ( ) ( ) |
| |k k kf x h f x f x h h
h
0 0 00 0
| ( ) ( ) ( ) || ( ) ( ) |
| |k k k
k k
f x h f x f x hf x f x h
h 0, 0.h
The last one follows from the continuity and
differentiability of function ( )kf x in .
Case 2: 0( )I x contains more than one element.
Because of the finite intersections, there must be a
0 so that 0 0
( ) { }, ( , )I x k x x x
and
0 0( ) { }, ( , ).I x l x x x
+ For h so that 0 0 0
( , )x h x x we have
0 0 0| ( ) ( ) ( ) |
0| |
F x h F x G x h h
h
0 0 0| ( ) ( ) ( ) |
| |k k kf x h f x f x h h
h
0 0 00 0
| ( ) ( ) ( ) || ( ) ( ) |
| |k k k
k k
f x h f x f x hf x f x h
h
as0, 0.h
The last one follows from the continuity and
differentiability of function ( )kf x in .
+ For h so that 0 0 0( , )x h x x we have
0 0 0| ( ) ( ) ( ) |
0| |
F x h F x G x h h
h
0 0 0| ( ) ( ) ( ) |
| |l l lf x h f x f x h h
h
0 0 00 0
| ( ) ( ) ( ) || ( ) ( ) |
| |l l l
l l
f x h f x f x hf x f x h
h
0, 0.h
The last one follows from the continuity and
differentiability of function ( )lf x in .
5. Semismooth Newton’s method for finding fixed point
Consider function 1( ) max{ ( ),..., ( )}
nF x f x f x where
( ), 1,if x i n is continuously Fréchet differentiable
which has finite intersections.
Denote ( ) ( )G x x F x and consider the following
problem
( ) 0G x (3)
Denote *x is a solution to function (3) then the iteration
of semismooth Newton method for problem (3) is given by
1
1( ),
( )n n n
n
x x G xG x
(4)
where G is one of Newton derivative of G .
Theorem 5.4. Assume *x is one solution to problem
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 39
(3) and function G is Newton differentiable with G the
Newton derivative of G at *x . If there is a neighborhood
U of *x so that ( ) 1, , 1,if x x U i n then Newton
iteration (4) converges to *x when *
0| |x x is small
enough.
Prove. We have *( ) 0G x .
*
1 *
1| | | ( ) |
( )k k k
k
x x x G x xG x
* *1[ ( )( ) ( ) ( ))]
( )| |
k k k
k
G x x x G x G xG x
* *1( ) ( ) ( )( )
( )| || |
k k k
k
G x G x G x x xG x
(5)
Since *( ) 1, ( ), 1,if x x U x i n
it follows that *
( )( ) 1 0, ( )
A xG x f x U x .
This implies that 1
( )G x is bounded in U .
Choose 0 so that * *( , )x x contains U and
0M so that
* *1, ( , ).
( )| | M x x xG x
Since G is Newton derivative of G at *x then exists
(0, )r so that
* * *| ( ) ( )| |
( ) | ,2
G x h G x G x h hM
h (6)
for all | | .h r In addition, by choosing 0x so that
*
0| |x x r then from (5), (6) and
*
kh x x we
deduce that
* *( , ),kx x x k and
* *
1
1 1| | . | | | | .
2 2k kx x M h x x
M
Therefore, iteration (4) is completely defined and *.
kx x
Example 5.1. Find the fixed point of function 3 2( ) max{2 , }F x x x x .
Denote ( ) ( )G x x F x .
According to the prove in previous theorems, G is
Newton differentiable. Noticeably, * 0x and * 1x
are zero-point of ( )G x . Moreover, Newton derivative G
satisfies assumption in Theorem 5.1. Therefore, we apply
semismooth Newton method that is given by
1
1( ).
( )n n n
n
x x G xG x
The iterations of semismooth Newton method are
showed in Table 1 with 0
1x . We see that kx speedily
converges to *x within 6 iterations. Importantly, if 0x
closes 0 or 1 then kx converges to its, respectively.
Table 1. The iterations of semismooth Newton method with 0
1x
k xk G G’
0 -1 -2 3
1 -0,333333333 -0,592592593 1,3333333
2 0,111111111 0,098765432 0,7777778
3 -0,015873016 -0,031738033 1,9984883
4 8,00455E-06 8,00448E-06 0,999984
5 -6,40738E-11 -1,28148E-10 2
6 0 0 1
Example 5.2. Find the fixed point of function 2 3 4( ) max{ , , }.F x x x x
Denote ( ) ( )G x x F x . Noticeably, * 0x and
* 1x are zero-point of ( )G x . As well as Example 5.1,
we apply semismooth Newton method that is given by
1
1( ).
( )n n n
n
x x G xG x
The iterations of semismooth Newton method are
showed in Table 2 with 01,2x . We see that k
x speedily
converges to *x within 5 iterations. Importantly, if 0x
closes 0 or 1 then kx converges to its, respectively.
Table 2. The iterations of semismooth Newton method with 01,2x
k xk G G’
0 1,2 -0,8736 -5,912
1 1,052232747 -0,225585556 -4,6431625
2 1,003648284 -0,014673094 -4,0438593
3 1,000019796 -7,91882E-05 -4,0002376
4 1,000000001 -2,35129E-09 -4
5 1 0 -1
6 1 0 -1
6. Conclusion
In this paper, we have investigated Newton
differentiable of function
1 2
( ) max ( ), ( ), , ( )n
F x f x f x f x
in special case 2n as well as in general case. The
sufficient condition of each case is also different. That is,
the condition in the case 2n is weaker than one in
general case.
40 Pham Quy Muoi, Phan Quang Nhu Anh, Duong Xuan Hiep, Phan Duc Tuan
Next, semismooth Newton method is applied to find
fixed point for function ( )F x , the convergence and the
quadratic rate of convergence for this problem. All
examples have slightly showen that semismooth Newton
method reach fast convergence within several iterations to
reaches exact solution.
REFERENCE
[1] Ravi P Agarwal, Maria Meehan, and Donal O'Regan. Fixed point theory and applications, volume141. Cambridge university press,
2001.
[2] Vasile Berinde. Iterative approximation of fixed points, volume
1912. Springer, 2007.
[3] Kim C Border. Fixed point theorems with applications to economics
and game theory. Cambridge university press, 1989.
[4] Andrzej Granas and James Dugundji. Fixed point theory. Springer
Science & Business Media, 2013.
[5] Duong Xuan Hiep, Pham Quy Muoi, Phan Duc Tuan. Some basic
properties of Newton derivatives of one variable function. Journal of Science and Technology, The University of Danang, Number 9
(118):94-98, 2017.
[6] Mohamed A Khamsi and William A Kirk. An introduction to metric
spaces and fixed point theory (volume 53). John Wiley & Sons,
2011.
[7] Pham Quy Muoi, Dinh Nho Hào, Peter Maass, and Michael Pidcock.
Semismooth newton and quasi-newton methods in weighted l1 - regularization. Journal of Inverse and Ill-Posed Problems,
21(5):665–693, 2013.
Pham Quy Muoi, Dinh Nho Hào, Peter Maass, and Michael Pidcock.
Descent gradient methods for non-smooth minimization problems in ill-posed problems. Journal of Computational and Applied
Mathematics, 298:105–122, 2016.
(The Board of Editors received the paper on 21/04/2018, its review was completed on 25/05/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 41
POLYNOMIAL SOLUTION OF DESCRIPTOR SYSTEM
Le Hai Trung
University of Education - The University of Danang; [email protected]
Abstract - The aim of article is to prove that it is possible to find state function x(t) and controllability function ( )u t of the descriptor
systems '( ) ( ) ( )Ex t Bx t Du t= + in which E, B, D are real matrices
with size equivalent to state function and controllability vector in the type of polynomials of degree 2 1.p + The basis of the
theory is a method to prove the cascade splitting to transform the original system into an equivalent system in the type
' ( ) ( ) ( ).p p p p px t B x t D z t= + In the final step, we obtain function
( )px t satisfying the condition and substituting this in the previous
step. Hence continuing this process, we can find out the functions
( )x t and ( )u t of the initial descriptor system.
Key words - Descriptor systems; controllability function; state function; polynomial; differential algebraic equations
1. Rationale
Consider the descriptor system, also known as the
differential algebraic equation, as follows:
( ) ( ) ( )Ex t Bx t Du t= + (1)
with , ( , ),n mE B L ( , ),l mD L ( ) ,nx t ; ( )x t
is
the state function and ( )u t is the controllability function.
The system is called controllable in the interval [0, ]T
if for any a, b in ,n it exists the control function ( )u t so
that its root ( )x t satisfies the following condition:
(0) , ( )x a x T b= = (2)
The problem of descriptor system has received the
attention of many mathematicians around the world, such
as Amit Ailon (see [8], [9]), S. P Zubova (see [2], [3], [4],
[10]), Раецкая Е. В. (see [7]). Amit Ailon (see [9]) has
demonstrated that existed a polynomial function ( )u t of
the system (1) with degree not large than M where
2 1, ,M r r n l + = − can be computed so that the solution
( )x t of the system (1) is a polynomial trajectory of
degree M with (0) , ( ) .x a x T b= = In this article, the
author proposes a method for constructing solution )(tx
of the system (1) in polynomial form which satisfies the
condition (2) in polynomial form of degree 2 1p + , where
p is the step at which we obtain a linearly ceasing system
satisfying the controllable criterion.
Use the following property of the matrix :: sk RRC →
ker Coim ; im Cokerk sC C C C= + = + (3)
and contracted matrix C~
on CoimC, : Coim im ,C C C→
and its inverse matrix .~ 1−C
In addition, we use P − the projection matrix on ker ,C
Q − the project matrix on Coker ,C in which kerC and
CokerC are described in the formula (3),
( ).C C I Q+ = −
2. Results and Survey Research
Consider the following lemma (см [7]):
Lemma. The equation ,,, sk RvRuvCu = is
equivalent to system:
+=
=+ ,
0
PuvCu
Qv
in which −Pu is an element in .ker C
Apply this lemma to equation (1) when C D= , then
(1) is equivalent to the system:
( ) '( ) ( ) ( )
'( ) ( ),
u t D Ex t D Bx t Pu t
QEx t QBx t
+ += − +
=
(4)
in which ( )Pu t − is an element in ker D . The second
equation of (4) is rewritten as ,QE G= hence we have the
equation:
'( ) ( ).Gx t QBx t=
Using the lemma with ,C G= the final equation is
equivalent to system:
'( ) ( ) '( )
( ) 0,
G
G
x t G QBx t P x t
Q QBx t
+ = +
=
(5)
in which GQ and
GP − are projection matrices on
CokerG and kerG respectively.
The first equation of the system (5) after
transformation becomes:
( ) '( ) ( ) ( ) ( )G GI P x t G QBx t I P QBx t+− = = − (6)
Using the properties of the matrices , ,G GQ PGI P− we
have:
( ) ( ) ( ) ( )( ) ( )
( ) ( ).
G G G G
G G G
I P QBx t I P QB I P I P x t
I P QBP P x t
− = − − −
+ −
We denote:
( ) ( ) ( ), ( ) ( ),
( ) ( ) , ( )
G G
G G G G
I P x t x t P x t u t
I P QB I P B I P QBP D
− = =
− − = − = (7)
Then equation (6) becomes:
'( ) ( ) ( )x t B x t Du t= + (8)
And the condition (2) becomes:
0(0) ( ) (0) ( ) ;
( ) ( ) ( ) ( ) .
G G
G G T
x I P x I P a a
x T I P x T I P b b
= − = − =
= − = − = (9)
The problem in equation (1) – (2) becomes the problem
(8) – (9).
Again we apply the lemma to equation (8) with
.C D= Hence (8) is equivalent to system:
42 Le Hai Trung
( ) '( ) ( ) ( )
'( ) ( ),D
D D
u t D x t D Bx t P u t
Q x t Q Bx t
+ + = − +
=
(10)
Our goal is to find ( )x t satisfying the system of
equations (10) and the condition (9).
We have obviously:
( ) ( ) ( ) ( )D D
x t Q x t I Q x t= + −
Using the properties of the matrices ,D D
Q I P− we
have:
'( ) ( ( )) ( )( ) ( )D D D D D D D
Q x t Q BQ Q x t Q B I Q I Q x t= + − − (11)
We denote:
1 1
1 1
( ) ( ), ( ) ( ) ( ),
, ( ) ,
D D
D D D D
Q x t x t I Q x t z t
Q BQ B Q B I Q D
= − =
= − =
The equation (11) becomes:
'1 1 1 1 1( ) ( ) ( )x t B x t D z t= + (12)
The condition (9) becomes:
1 1,0 1 1,
1 1
(0) (0) , ( ) ( ) ,
(0) ( ) (0), ( ) ( ) ( )
TD D
D D
x Q x a x T Q x T a
z I Q x z T I Q x T
= = = =
= − = − (13)
From formula (12) we get the condition:
'1 1 1 1 1 1,1
'1 1 1 1 1 1,1
(0) (0) (0) ,
( ) ( ) ( ) .
x B x D z a
x T B x T D z T b
= + =
= + =
(14)
Then (8) – (9) becomes:
1 1
'1 1 1 1 1
( ) '( ) ( ) ( )
( ) ( ) ( ),
( ) ( ) ( )
Du t D x t D Bx t P u t
x t x t z t
x t B x t D z t
+ + = − +
= +
= +
(15)
with the conditions (13) – (14).
Again we use again the lemma for the third equation
in (15) with 1C D= to obtain:
1
1 1
'1 1 1 1 1 1 1
1 2 2
'2 2 2 2 2
( ) '( ) ( ) ( ),
( ) ( ) ( ),
( ) ( ) ( ) ( ),
( ) ( ) ( ),
( ) ( ) ( )
D
D
u t D x t D Bx t P u t
x t x t z t
z t D x t D B x t P z t
x t x t z t
x t B x t D z t
+ + = − +
= +
= − +
= +
= +
(16)
in which:
1 1
1 1 1 1
1 2 1 2
1 2 1 2
( ) ( ), ( ) ( ) ( ),
, ( ) ,
D D
D D D D
Q x t x t I Q x t z t
Q B Q B Q B I Q D
= − =
= − =
And the new conditions:
1 12 1 2,0 2 1 2,
'2 2 2 2 2 2,1
'2 2 2 2 2 2,1
'' ' '2 2 2 2 2 2,2
'' ' '2 2 2 2 2 2,2
(0) (0) , ( ) ( ) ,
(0) (0) (0) ,
( ) ( ) ( ) ,
(0) (0) (0) ,
( ) ( ) ( ) .
TD Dx Q x a x T Q x T a
x B x D z a
x T B x T D z T b
x B x D z a
x T B x T D z T a
= = = =
= + =
= + =
= + =
= + =
(13)
In this second step appear 2.2 + 2 = 6 conditions for
2 ( ),x t in which two new conditions appear by '' ''
2 2(0), ( ).x x T
Keep using the lemma for 2C D= and so on…
At j-th step we obtain the system:
1
1 1
'1 1 1 1 1 1 1
1
'1
'
( ) '( ) ( ) ( ),
( ) ( ) ( ),
( ) ( ) ( ) ( )
....
( ) ( ) ( ),
( ) ( ) ( ) ( ),
( ) ( ) ( )
j
D
D
j j j
j j j j j j jD
j j j j j
u t D x t D Bx t P u t
x t x t z t
z t D x t D B x t P z t
x t x t z t
z t D x t D B x t P z t
x t B x t D z t
+ +
+ +
−
+ +−
= − +
= +
= − +
= +
= − +
= +
and appear new (2j+2) conditions (similar (17)):
es es( ) ( )
,0 ,(0) , ( ) , 0, .d d
v vj k j k Tx a x T a v j= = = (18)
According to the content in [7]1 we receive an
equivalent system to (8) as follows:
' ( ) ( ) ( )p p p p px t B x t D z t= + (19)
with (2p+2) new conditions (similar to (18) for the
function ( )px t , at the p-th step, we can affirm the
controllability of the system ' ( ) ( ) ( )p p p p px t B x t D z t= +
and from this step, we are able to infer the controllability
of the system (8). As a result, we have the following
system of the equations:
1
1 1
'1 1 1 1 1 1 1
1
'1
'
( ) '( ) ( ) ( ),
( ) ( ) ( ),
( ) ( ) ( ) ( )
....
( ) ( ) ( ),
( ) ( ) ( ) ( ),
( ) ( ) ( )
p
D
D
p p p
p p p p p p pD
p p p p p
u t D x t D Bx t P u t
x t x t z t
z t D x t D B x t P z t
x t x t z t
z t D x t D B x t P z t
x t B x t D z t
+ +
+ +
−
+ +−
= − +
= +
= − +
= +
= − +
= +
with new conditions:
es es( ) ( )
,0 ,(0) , ( ) , 0, .d d
k kp k p k Tx a x T a k p= = = (20)
Whether we can continue to find the function ( )px t
under the polynomial form (2p+1) or not, the following
arguments will answer this question by proving the
existence of the vector factor , 0,2 1i i p = +
in the
representation2 1
0
( ) .p
i
p i
i
x t t+
=
=
Substitute (19) to this representation ( )px t we have:
,0
1, 0, .
!i ia i p
i = = (21)
Using conditions (18) we obtain the system of
equations:
1Theorem. There exits the number *p so that either the matrix
pD = or pD ís surjective.
The meaning of the above theorem in the first case is that the system (8)
is uncontrollable, while in the second case the system (8) is controllable.
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 43
1 2 2 11 2 2 1 0,
0
11 2
2 12 1 1,
0
21 2
2 1 ,
... ,
( 1) ( 2) ...
(2 1) ,
..............,
( 1) ...2 ( 2)( 1)...3 ...
(2 1)2 ...( 1) (
pp p p i
p p p T i
i
p pp p
pp i
p T i
i
p p
pp p T
T T T a T
p T p T
p T a i T
p p T p p T
p p p T a i j
+ + ++ + +
=
++ +
++
=
+ +
+
+ + + = −
+ + + +
+ + = −
+ + + + +
+ + + = − −
1
0 0
) .
ppi j
i
i j
T
−−
= =
(22)
The determinant of the system (22) is defined as:
1 2 2 1
1 2
2
...
( 1) ( 2) ... (2 1)
...
( 1) ...2 ( 2)( 1)...3 ... (2 1)...( 1)
p p p
p p p
p
T T T
p T p T p T
p p T p p T p p T
+ + +
++ + + =
+ + + + +
This is the Wronxki determinant, i.e 0. Combine this
with (21) we can imply the existence of , 0,2 1.i i p = +
Hence, the following theorem is proved:
Theorem 1. If the system (17) - (19) - (20) is
controllable, then there exists the state function ( )px t in
polynomial form of degree (2p + 1).
According to the instruction above, when ( )px t is
determined we can determine1( ),px t−
when 1( )px t−
is
determined we can determine2 ( ),px t−
... So on, we can
determine ( ).x t With the determination of ( )x t of the
system (8) - (9), ( )x t and ( )u t of (1) - (2) are
determined via (7) and (4), respectively. Then we have:
Theorem 2. If the system (17) - (19) - (20) is
controllable, then there exists the control function ( )u t
and the state function )(tx of (1) - (2) in polynomial form
of degrees not exceeding (2p + 1).
3. Example
Consider the system of equations:
'1 1 2 3 1
'2 1 2 3 2
1 2 3 3
0.
0 0.
x x x x u
x x x x u
x x x u
= + − +
= − + + = − − +
(23)
with initial conditions:
(0) (1), (1) (2)x x= = (24)
The system (23) is written in matrix form as follows:
,
000
000
001
111
111
111
000
010
001
3
2
1
3
2
1
'
3
'
2
'
1
+
−
−
−
=
u
u
u
x
x
x
x
x
x
with: 1 0 0 1 1 1 1 0 0
0 1 0 ; 1 1 1 ; 0 0 0 .
0 0 0 1 1 1 0 0 0
E B D
−
= = − = −
We have:
2
3
1 0 0 0 0 0
0 0 0 ; 0 1 0 ;
0 0 1 0 0 1
1 0 0 0
0 0 0 ; .
0 0 0
P Q
D Pu u
u
+
= =
= =
Using the projection matrices ,Q P , (see lemma) the
system (23) is equivalent to systems: '11 1'
2 2 2 2
'3 3 33
1 0 0 1 0 0 1 1 1 0
0 0 0 0 0 0 1 1 1 ,
0 0 0 0 0 0 1 1 1
xu x
u x x u
u x ux
− = − − + −
or
'1 1 2 3 1
2 2
3 3
( )
.
x x x x u
u u
u u
− + − =
= =
(25)
And:
'1 1'2 2
'33
0 0 0 1 0 0 0 0 0 1 1 1
0 1 0 0 1 0 0 1 0 1 1 1 ,
0 0 1 0 0 0 0 0 1 1 1 1
x x
x x
xx
− = − −
'2 1 2 3
1 2 3
0 0
0 .
x x x x
x x x
=
= − + = − −
(26)
From the last equation of the system (26) we get
1 2 3 ,x x x− = substitute the final equation into the second
equation we obtain '
2 2 12 2 (*).x x x= − + This equation has
the form of (8) with 1x as control function, because
1 2 0D = therefor this "system '
2 2 12 2x x x= − + " is
controllable (In this step we can write ' '
2 1 2 1 1 1, , .x x x x x z= = = ). According to theorem 1 we can
define 1,p = we can determine subsequently2 ( )x t (now
serves as the pseudo-state function of the "system" (*)) in
polynomial form of degree 2 1 2.1 1 3.p+ = + = Denoting:
2 32 0 1 2 3( )x t a a t a t a t= + + +
and using conditions (24) we obtain:
2 0 2 1 2 3(0) 1; (1) 2 1x a x a a a= = = = + + + (27)
Using (*) we obtain:
'2 2 1
'2 2 1
' 22 1 2 3 1 2 3
(0) 2 (0) 2 (0) 0;
(1) 2 (1) 2 (1) 0
( ) 2 3 0; 0 2 3 .
x x x
x x x
x t a a t a t a a a
= − + =
= − + =
= + + = = +
(28)
Using (27) and (28) we obtain system:
2 3 2
2 3 3
1 3
2 3 0, 2,
a a a
a a a
+ = =
+ = = −
i.e 2 32 ( ) 1 3 2 ,x t t t= + − then 2
1 1 3 2 .x t t= + − We finally
obtain 2 3
3 3 5 2 ,x t t t= − + 2 3
1 1 4 6 4 .u t t t= − − +
44 Le Hai Trung
4. Conclusion
This paper demonstrates that the solution ( )u t and the
control functions ( )u t of (1) - (2) can be found in
polynomial form of degree not exceeding (2p +1). The
basis for the above problem is based on the fact that we
can transform the descriptor control system (1) to the
ceasing linear system (8), then performs the substitution
and transformation, then prove the possibility to define
the degree of the system’s solutions in polynomial form.
We then use the initial conditions to draw out the
conclusions for ( )x t and ( )u t .
Acknowledgement
This research was supported in part by Foundation for
Science and Technology Development of Vietnams
Ministry of Education and Training. No. B2017.DNA.09
REFERENCES
[1] Андреев Ю. Н, Управление конечномерными линейными
объектами. М. Наука, 1976.
[2] Зубова С.П., Раецкая Е.В., Ле Хай Чунг, “ О полиномиальных
решениях линейной стационарной системы управления”.
Автоматика и Телемеханика. No 11. стр: 41-47. 2008.
[3] Зубова С.П., Ле Хай Чунг, “Полиномиальное решение
линейной стационарной системы управления при наличии контрольных точек и ограничений на управление”.
SpectralandEvolutionproblems. No: Vol. 18. стр 71 – 75. 2008.
[4] ЗубоваС.П., Ле Хай Чунг, “Построение полиномиального
решения линейной стационарной системы с контрольными
точками и дополнительными ограничениями”. Системы управления и информационные технологии. No 1.2(31). стр:
225 – 227. 2008.
[5] Красовский Н. Н, Теория управления движением. М. Наука.
1968.
[6] Ле Хай Чунг, Полиномиальное решение задач управления для
линейной стационарной динамической системы. Дисс...канд. физ. – мат. наук. Воронеж 2009.
[7] Раецкая Е. В, Условная управляемость и наблюдаемость
линейных систем. Дисс...канд. физ. – мат. Наук. Воронеж 2004.
[8] AilonA., Langholz G, “More on the controllability of linear time-
invariant systems”. Int. J. Contr. 1986. 44. № 4. P. 1161 – 1176.
[9] Ailon A., “Polynomial controllability in linear time-invariant
systems: some further results and applications to optimal control”.
15th Triennial Word Congress, Barcelona, Spain. P. 449 – 454.
[10] S. P. Zubova, Le Hai Trung, “Construction of polynomial controls
for linear stationary system with control points and additional
constrains”. Automation and Remote Control. Volume 71, Number 5. Pages 971-975. 2010.
[11] Lê Hải Trung, “Về hàm trạng thái đa thức cho hệ dừng động học
tuyến tính”. Tạp chí Khoa học và Công nghệ Đại học Đà Nẵng.
Số: 11(60). Trang 53 - 57. 2012.
(The Board of Editors received the paper on 07/05/2018, its review was completed on 29/05/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 45
FAST GAUSSIAN DISTRIBUTION BASED ADABOOST ALGORITHM
FOR FACE DETECTION
Tuan M. Pham1, Hao P. Do2, Danh C. Doan2, Hoang V. Nguyen2 1University of Science and Technology - The University of Danang; [email protected]
2Hippo Tech Vietnam; {haodophuc, danhdoan.es, nguyenviethoang.25}@gmail.com
Abstract - In the past few years, Paul Viola and Michael J. Jones have successfully developed a new face detection approach which has been widely applied to many detection systems. Even though the efficiency and robustness are proved in both performance and accuracy, there is still a number of improvements that we can apply to enhance their algorithm. This paper inherits face detection framework of Viola-Jones and introduces two key contributions. First, the modification is used to apply integral image so that features are more informative and help increase detection performance. The second contribution is the new approach to utilize AdaBoost that uses Gaussian Probability Distribution to compute how close to the mean positive and negative distributions are, then classify them more efficiently. Furthermore, by experiments, we also prove that a small fraction of a feature set is far enough to develop a good strong classifier instead of the whole feature set. As a result, the memory required as well as the time for training is minimized.
Key words - face detection; Gaussian distribution; AdaBoost; Haar-like pattern; weak classifier
1. Introduction
In recent decades, along with the rapidly advanced
improvement in technology, face detection has now
become the most popular topic that can be applied to many
fields in industries or in real life. Algorithms for face
detections are developed quickly and become more
enhanced to support complicated applications like multi-
view face detection [1-4], occluded face detection [3, 5],
pedestrian detection [6, 7], ... In this paper, we inherit the
work of Viola-Jones [8] which has been proved successful
in accuracy as well as in performance.
Thanks to their great work, the number of practical
real-time applications and systems are built for face
detection or related topics. We will propose some new
methods for feature extraction and implementation so that
the system can train and detect images faster than previous
one from Viola-Jones and it also utilizes less memory
storage. Besides, new Haar-like patterns are proposed to
improve the efficiency of detection.
There are two main contributions in our face detection
systems and they are briefly introduced below and in detail
in next sections.
First, we utilize integral image representation as the
main component to quickly compute feature values.
Nevertheless, it is more efficient when applying non
integer-sized pattern as described in section 3. In this way,
given a feature, we can obtain much more information that
is necessary for classification process. In this step, Viola-
Jones system pre-calculates feature values and stores them
in hard drive. This method is useful in training process
because all information is already computed. In contrast,
the significantly long time is used for this calculation and
working with hard drive. Instead of following their method,
we introduce another approach. Given the size of a data set,
size of image as well as the features patterns, a lookup table
used for feature indexing can be generated separately
before training procedure proceeds. We have to compute a
specific feature value when needed. It is more efficient than
the previous work [8] not only in performance but also in
memory consumption.
The second enhancement is how AdaBoost [9] is
applied. Viola-Jones used threshold to classify positive and
negative example images. Multiple weaker classifiers are
combined to form a strong one which can divide data
perfectly when learning, but in testing, it might fail. The
detection quality depends much on the correctness of the
AdaBoost function to classify data. In case the positive and
negative distribution overlap, it is apparently difficult to
choose a good threshold between those regions. In this
paper we apply Gaussian probability distribution for the
classification task. Why we use and how we apply this
method to AdaBoost process is described in this section,
and pseudo-code is also provided. Besides, the number of
operations for Gaussian is less than using threshold, thus
the training and detection time is reduced.
a. Overview
We start by review Viola-Jones systems and point out
some functions that we need to improve. The review is in
section 1. In section 2, we describe and explain how we
choose and implement our new algorithms and why they
are effective in face detection system. After that, the
experiments and results are clearly shown in Section 3.
2. Review of Viola-Jones Algorithms
a. Haar-like patterns
In every vision system, the efficiency and accuracy
depend strongly on the features it uses and the quality of
those features. Feature design and its calculation is the key
to the success of a computer vision or machine learning
system. To extract features from example images, Viola-
Jones used Haar-like rectangle features in their systems as
shown in Figure 1.
Figure 1. Haar-like rectangle patterns in Viola-Jones system
Feature value for a certain rectangle is the difference
between white and black pixel values. If doing this task by
common methods, it would take the complexity of O(HW)
where H and W are height and width of a pattern.
46 Tuan M. Pham, Hao P. Do, Danh C. Doan, Hoang V. Nguyen
Integral image representation is one of their
contributions in their paper. It is applied to rapidly compute
feature values. Its formula is described below:
𝑖𝑖(𝑥, 𝑦) = ∑ 𝑖(𝑥′, 𝑦′)
0 ≤𝑥′≤𝑥,0 ≤𝑦′≤𝑦
Where 𝑖(𝑥, 𝑦) is the image intensity at pixel (𝑥, 𝑦) and
𝑖𝑖(𝑥, 𝑦) is the value of integral image at pixel (𝑥, 𝑦).
By using integral image method, we are able to
calculate any rectangular sum by pre-computed referenced
rectangles. Thus, the complexity is approximately O(1).
In their detection system, they trained and tested by
24x24 Grayscale PNG images. For each image, they
applied those 5 rectangle features. Heights, widths and
positions of each rectangle are also varied. Because
information of images and feature patterns are given, the
number of features which can be applied to an image is
known. There were 43200, 43200, 27600, 27600 and
20736 features for each rectangle of category (a), (b), (c),
(d) and (e) respectively, thus 162336 features in total.
b. AdaBoost algorithm
From that, there were a huge number of features
corresponding to each image sub-window. Due to this fact,
it is still a lot of work even when those features are
calculated quickly and efficiently. However, by using a
very small set of features, detection system can form an
effective classifier.
Thanks to the invention of AdaBoost [9], this method
can be used to select the essential features as well as to train
for a strong classifier. AdaBoost is an algorithm for
constructing a strong classifier from a linear combination
of weak classifier.
𝐹(𝑥) = ∑ 𝑎𝑡ℎ𝑡(𝑥)
𝑇
𝑡=1
Where 𝑥 is an example (19x19 image in our system),
ℎ𝑡(𝑥) is a weak or basis classifier. Normally, the set
ℋ = ℎ(𝑥) is finite.
A weak classifier is a function of a feature (f), a
threshold (𝜃) and a polarity (p) that denotes the direction
of the inequality:
ℎ(𝑥, 𝑓, 𝑝, 𝜃) = {1 𝑝. 𝑓(𝑥) < 𝑝. 𝜃0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
For a weak learner, we do not expect the best
classification. After each round of AdaBoost, a weak
classifier with a smallest weighted error is chosen:
ℎ̂𝑡 = arg𝑚𝑖𝑛
ℎ𝑗 ∈ 𝐻𝜀𝑗 = ∑ 𝑤𝑖|ℎ𝑗(𝑥𝑖 , 𝑓, 𝑝, 𝜃) − 𝑦𝑖|
𝑖
Where 𝑦𝑖 is the correct label for example 𝑥𝑖, 𝑦𝑖 is 1 if
𝑥𝑖 represents a face, otherwise it is 0.
In additions, every example is re-weighted so that it is
emphasized in the next training round. Clearly, an example
which is incorrectly labeled in the current round will have
the greater weight compared to correct ones.
𝑤𝑡+1,𝑖 = 𝑤𝑡,𝑖𝛽1−𝑒𝑖
Where 𝑒𝑖 = 0 if 𝑥𝑖 is correctly classified, otherwise it
is 1, and 𝛽 =𝑒𝑖
1−𝑒𝑖
AdaBoost is very simple to implement and efficiently
extract good features from a very large set. One of the
disadvantages of AdaBoost algorithm is that over fit is the
result of choosing a very complex-training model, turning
this to the key challenge to applying this method.
Graphic visualization of AdaBoost process after each
round is shown in Figure 2 below. In this example, we need
to detect and classify blue dots from red ones. We apply
AdaBoost to this problem and try to find the best weak
classifier to classify these two regions in each round.
Figure 2. Visualization for AdaBoost process after t=1 and t=3
After completing the first round, 1 weak classifier is
chosen as described by the black straight line. The total
detection quality is now very low. However, by combining
3 weak classifiers, the accuracy is significantly improved.
In Viola-Jones system, they used AdaBoost and for
each weak learner, tried to find the optimal threshold
classification function. Supposing that there are N image
examples and K features for each image, so they had KN
combinations for feature and threshold. For the data set
used by Viola-Jones, K was 162336 features and N was
6977 images to train. In a training round of AdaBoost
procedure, it needs to iterate the whole data set to evaluate
the training error for a feature/threshold combination. It
means, the complexity required for each round was
O(NKN) to find a weak classifier. By setting the number of
weak classifiers to M, the total complexity to train their
system is O(MNKN). With M = 200, at least 1.58x1015
operations needed to be processed in any training machine.
Even when working with a super computer, it is still a
tremendous procedure and takes a significantly long time
to finish.
To improve the system as well as reduce the training
time, they proposed the modified algorithm. With a
specific feature, they could find the optimal threshold by
using current example weights without generating all
possible combinations of this feature and every image
example. To apply this algorithm, with each feature,
examples should be sorted by their feature values, and the
complexity for this process is O(Nlog2N). Thereafter, it
only requires O(N) to find the optimal threshold for current
feature. Hence, the complexity to this sub-task is O(max(N,
Nlog2N)) = O(Nlog2N). This algorithm led to the reduced
complexity of O(MKNlog2N) and sustainable decrease in
the number of operations to 2.89x1012.
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 47
Pseudo-code for Viola-Jones' algorithm is shown as
below:
1. Giving example image (𝑥1, 𝑦1), … , (𝑥𝑛 , 𝑦𝑛) where
𝑦𝑖 = 0, 1 for negative and positive examples
respectively.
2. Initializing weights 𝑤𝑡,𝑖 =1
2𝑚,
1
2𝑙 for 𝑦𝑖 = 0, 1
respectively, where m and l are the number of
negatives and positives respectively.
3. For 𝑡 = 1, … , 𝑇:
a. Normalizing the weights 𝑤𝑡,𝑖 ←𝑤𝑡,𝑖
∑ 𝑤𝑡,𝑗𝑛𝑗
b. Selecting the best weak classifier with
respect to the weighted error
𝜀𝑡 =𝑚𝑖𝑛
𝑓, 𝑝, 𝜃∑ 𝑤𝑖|ℎ(𝑥𝑖 , 𝑓, 𝑝, 𝜃) − 𝑦𝑖|
𝑖
a. Defining ℎ𝑡(𝑥) = ℎ(𝑥, 𝑓𝑡 , 𝑝𝑡 , 𝜃𝑡) where
𝑓𝑡 , 𝑝𝑡 , and 𝜃𝑡 are the minimizers of 𝜀𝑡
b. Updating the weights:
𝑤𝑡+1,𝑖 = 𝑤𝑡,𝑖𝛽1−𝑒𝑖
Where 𝑒𝑖 = 0 if example 𝑥𝑖 is classified
correctly, 𝑒𝑖 = 1 otherwise, and 𝛽 =𝑒𝑖
1−𝑒𝑖
4. Combining strong classifier
3. Proposed Method to Improve Viola-Jones system
The main purpose of this paper is to propose the new
method that can perform detection faster than the
traditional Viola-Jone system, so we choose the same
Haar-like rectangle features in their system. Besides, we
also introduce our new features which are more efficient
for face detection systems when the complexity level
increases, such as detecting rotated faces.
3.1. New feature selection
In Viola-Jones system, they used integer-size of
rectangle. In our research, we again use those rectangle but
with non-integer size. With this method, features can be
more informative, thus the detection performance is higher
than that by the Viola-Jones system. Figure 3 shows some
examples of new non integer-sized rectangle. This is 2x2
sub-window from an image and the height of rectangle
feature is 3.
Figure 3. 2x2 sub-window image with non- integer-sized feature
Because the size is a non -integer number, feature
values are represented by floating-point number. It results
in new difficulties when systems use complicated pattern
of features. For this problem, we also figure out the
approach which can quickly compute the feature values in
few operations with the complexity of O(1). Compared to
the traditional feature calculation, processing time is now
nearly the same.
To apply new rectangle features, users only need to
clearly specify new pattern before training their models.
Below is an example for pattern (d) in Figure 1. The matrix
shows color map of features with 1 for white and -1 for
black.
1 1 1
-1 -1 -1
1 1 1
By using the color map above, we can quickly compute
the feature value for each rectangle with non-integer size.
The size and position of a feature can vary
correspondingly to the 19x19 image. Given the size of a
pattern, we can manage the number of arising features.
This point leads to another improvement for our system,
that is pre-calculating and storing for feature values are no
longer required. Back to Viola-Jones approach, they have
to use hard drive to store the whole set of feature values.
Apparently, the way costs a tremendously long time to get
the training data available. Besides, it consumes huge
memory storage which is now not essential in our system.
There are 29241, 29241. 23409, 23409 and 29241
features for category (a), (b), (c), (d) and (e) respectively.
Thanks to the constraint of rectangle sizes, we
separately make a lookup table from the given information
about those rectangles. It means that when it requires any
single feature value, our system can immediately find the
exact feature as well as its size and position. After that we
easily compute them by our formula as we have mentioned
before. To generate this lookup table, we just apply brute-
force algorithm to iterate and find feature information.
3.2. Gaussian distribution as classification function
Even though the training time is significantly reduced,
it is still a long time. In our research and experiment, we
propose a new way to train our detection system by
applying Gaussian probability distribution instead of
finding optimal threshold for each feature.
Starting with the point that positive and negative
distribution of Haar-like feature for an image are very hard
to classify by a single threshold. By combining multiple
weak classifiers with many thresholds, the number of
operations exponentially increases without guaranteeing
the increase of detection accuracy. Besides, it can result in
over-fitting problem on the training process.
Figure 4. Histogram of a specific feature for face and
non-face images
48 Tuan M. Pham, Hao P. Do, Danh C. Doan, Hoang V. Nguyen
Figure above shows histogram of feature values for
face and non-face images computed from a specific feature.
Blue region denotes feature values for face images, and
non-faces are drawn in orange. The x-axis is the feature
values; meanwhile y-axis shows the frequency after
normalization of each value. From the figure, it is clear that
2 histograms are overlapped, leading to the difficulty to
select a threshold. Moreover, in those situations, a weak
classifier's performance is poor but the training time is
longer and finally the testing result is poor as a
consequence.
In this paper, we propose the AdaBoost algorithm that
uses Gaussian distribution of feature values. Gaussian
distribution is one of the most important probability
distributions for continuous variables and it is really useful
in natural sciences. From theory, the averages of samples
of a variable from independent distributions converge in
distribution to the normal. In other words, it becomes
normally distributed when we have enough observations.
Below is the formula for a Gaussian distribution of a single
real-valued variable x:
𝑓𝑔(𝑥|𝜇, 𝜎2) =1
√2𝜋𝜎2𝑒
−(𝑥−𝜇)2
2𝜎2
Where 𝜇 is the mean and 𝜎2 is the variance of the
sequence of feature values for the data set with a specific
feature.
To overcome this overlapping problem, we apply
Gaussian distribution to calculate and compare the
difference to 2 means of positive and negative distributions.
The classification function is applied as follows:
ℎ(𝑥, 𝑓, 𝜇𝑝, 𝜎𝑝2, 𝜇𝑛, 𝜎𝑛
2)
= {1 𝑓(𝑥|𝜇𝑝, 𝜎𝑝
2) > 𝑓(𝑥|𝜇𝑛, 𝜎𝑛2)
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Where 𝑓(𝑥|𝜇, 𝜎2) is the Gaussian function that is
mentioned above. 𝑥 is a feature value of an example image.
𝜇𝑝, 𝜎𝑝2, 𝜇𝑛, and 𝜎𝑛
2 are means and variances for positive
and negative distributions respectively.
Our method proceeds as follows. For each feature, all
corresponded values to image examples are computed at a
given feature. Next , we calculate means and variances for
2 distributions. After that, formula of Gaussian distribution
is applied to find the distance to the means before
comparison. If an image example is more likely to be a face,
so the closer it is from the means of a positive region.
Otherwise, it is closer to negative distribution. With this
method, the difficulty of overlapping is overcome because
the means of distributions are all always separated.
After finishing the current classification, error is
computed to select the best feature of the current AdaBoost
round. Clearly, the weights of image examples are
re-computed to emphasize incorrect classification for later
training.
Our procedure is described in the pseudo-code below.
The only difference between our algorithm and Viola-
Jones' is the classification function with Gaussian method.
1. Giving example image (𝑥1, 𝑦1), … , (𝑥𝑛 , 𝑦𝑛) where
𝑦𝑖 = 0, 1 for negative and positive examples
respectively.
2. Initializing weights 𝑤𝑡,𝑖 =1
2𝑚,
1
2𝑙 for 𝑦𝑖 = 0, 1
respectively, where m and l are the number of
negatives and positives respectively.
3. For 𝑡 = 1, … , 𝑇:
a. Normalizing the weights 𝑤𝑡,𝑖 ←𝑤𝑡,𝑖
∑ 𝑤𝑡,𝑗𝑛𝑗
b. Selecting K’ features from full set of features
c. For each feature:
i. Computing feature values for every
example
ii. Calculating means and variances for
positive and negative distributions.
iii. Selecting the best weak classifier that
minimizes the error:
𝜀𝑡 = 𝑚𝑖𝑛𝑓 ∑ 𝑤𝑖|ℎ(ℎ(𝑥𝑖 , 𝑓, 𝜇𝑝, 𝜎𝑝2, 𝜇𝑛, 𝜎𝑛
2 ) − 𝑦𝑖|𝑖
iv. Defining ℎ𝑡(𝑥) = ℎ(𝑥, 𝑓𝑡) where 𝑓𝑡 is the
minimizer of 𝜀𝑡
d. Updating the weights:
𝑤𝑡+1,𝑖 = 𝑤𝑡,𝑖𝛽1−𝑒𝑖
Where 𝑒𝑖 = 0 if example 𝑥𝑖 is classified
correctly, 𝑒𝑖 = 1 otherwise, and 𝛽 =𝑒𝑖
1−𝑒𝑖
4. Combining strong classifier
In our method, it takes linear time to compute mean and
variance for each positive or negative distribution with
complexity of O(N) and all simple operations. O(MKN) is
the total complexity for our algorithm. However, due to the
use of Gaussian distribution, the floating-point operations
are obligated. Indeed, exponential operation for floating-
point numbers is far complicated than simple arithmetic
ones. In our system, this expression is used to find 𝑒𝑥 for
classification function 𝑓(𝑥|𝜇𝑝, 𝜎𝑝2). Although the current
complexity is O(MKN), it costs even longer time than
O(MKNlog2N) does if those operations are not well
computed. Thus, for this step, instead of exponentials we
use inverse operation which is natural logarithm 𝑙𝑛(𝑥). By
our experiment, it takes much less time to compute
𝑙𝑛(𝑒𝑥) compared to 𝑒𝑥. By simplifying expressions to find
Gaussian function, the number of operations is also
remarkably reduced.
If an image is a face, ratio of Gaussian functions for two
distributions should be greater than 1:
1
√2𝜋𝜎𝑝2
𝑒
(𝑥−𝜇𝑝)2
2𝜎𝑝2
1
√2𝜋𝜎𝑛2
𝑒(𝑥−𝜇𝑛)2
2𝜎𝑛2
> 1
or,
√𝜎𝑛2
√𝜎𝑝2
> 𝑒
(𝑥−𝜇𝑝)2
2𝜎𝑝2 −
(𝑥−𝜇𝑛)2
2𝜎𝑛2
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 49
Because all elements are non-negative numbers, the
inequality remains unchanged when taking square of both
sides:
𝜎𝑛2
𝜎𝑝2
> 𝑒
(𝑥−𝜇𝑝)2
2𝜎𝑝2 −
(𝑥−𝜇𝑛)2
2𝜎𝑛2
Moreover, natural exponential is a one-to-one and
increasing function, we can apply natural logarithm to the
inequality:
ln (𝜎𝑛
2
𝜎𝑝2
) >(𝑥 − 𝜇𝑝)2
2𝜎𝑝2
−(𝑥 − 𝜇𝑛)2
2𝜎𝑛2
At this point, we can use the above inequality to make
comparison; all of those operations can be calculated in a
short time. We have reduced a lot of operations and the
computation is now much simpler.
The following shows pseudo-code for our approach:
1. Computing feature values for the whole data set
2. Finding mean 𝜇 and variance 𝜎2for positive and
negative distributions
3. Using natural logarithm to find: 𝑎 = ln (𝜎𝑛
2
𝜎𝑝2)
4. Finding the value of 𝑏 = (𝑥−𝜇𝑝)2
𝜎𝑝2 −
(𝑥−𝜇𝑛)2
𝜎𝑛2
5. Comparing 𝑎 and 𝑏, return 1 if 𝑎 > 𝑏, otherwise 0
4. Experiments and Results
a. Experiments
As mentioned before, to train and test this system, we
use data set from MIT cbcl Face Data [10] and it contains
19x19 grayscale PGM images.
In this procedure, we conduct some experiments to
evaluate and compare the processing time between original
method with our proposed one. Before preparing lookup
table and training, we normalize the whole data set for both
train and test images [11] Thus, those images are now in
the same standard. Samples of images before and after the
normalization process are listed below:
Figure 5. Original images before normalization
Figure 6. Images after normalization
In the experiment, we implement our algorithm from
scratch in Java. After that, the system runs in a normal
computer with Mac OS, memory 8 GB 1600MHz DDR3
and processor 2.6 GHz Intel Core i5. The amount of hard
drive is not specified due to the fact that we only use Ram
to experiment our system. It means, hard drive is not
mandatory as in Viola-Jones's.
b. Results
Theoretically, for each round of AdaBoost process,
there are totally 134541 features used for testing to choose
the best weak classifier. By analysis and experiments, it is
unnecessary to test all of those features, instead, a small
number of them can be used to reduce the training time but
still maintain the accuracy level. From experimenting
different numbers of features for training each round,
we have found that K' = 5000 features are sufficient for
2 criteria above.
The graph in Figure 7 shows the comparison between
choosing K' = 5000 features versus the whole 134541
features. Both of two AdaBoost algorithms that are used
Viola-Jones with threshold and proposed method with
Gaussian distribution are involved. Viola-Jones’
approaches are figured by blue and gray dashed lines.
Figure 7. Processing time and accuracy between 2 methods
with different number of chosen features on training image set
In this experiment, we do not follow Viola-Jones
method which requires hard drive to store feature values
due to the long processing time with memory storage.
Hence, in our proposed method, lookup table is utilized to
reduce the training time.
The x-axis is processing time measured in second and
the y-axis is the corresponding accuracy by percentage.
When Viola-Jones method using threshold is applied to
classify face/non-face images, it takes approximately
760 seconds for a weak classifier if we test the whole set of
features. In contrast, 30 seconds is needed if this procedure
is performed by 5000 features. When applying Gaussian
distribution, the processing time decreases. It requires
600 seconds for the full feature set and 22 seconds if
5000 features are chosen.
Figure 8. Result from experimenting with testing image set
By conducting this experiment, it is proved that the
detection system can still obtain the high performance
50 Tuan M. Pham, Hao P. Do, Danh C. Doan, Hoang V. Nguyen
without choosing the whole set of features. From the above
figure, applying Gaussian distribution is better than
original Viola-Jones method in both cases. This result is
gained by testing with the training image set, we have the
similar result with the testing image set and it is shown in
Figure 8.
In those experiment processes, we compute the
accuracy of detection in the fixed processing time of about
1 hour. With this period of time, AdaBoost by Viola-Jones’
algorithm can produce 5 and 125 weak classifiers for the
whole set and 5000 features set respectively. Similarly,
6 and 160 weak classifiers are chosen with Gaussian
distribution algorithm.
We also conduct another experiment that sets the fixed
value of T - the number of weak classifiers. For T = 200,
if we apply Viola-Jones system that pre-compute all feature
values and store to hard drive, then use those values for
training, it takes 46 seconds to compute and choose 1 weak
class. Approximately, 153 minutes or 2 hours 33 minutes
is required for the complete training process. By setting the
same value for T, but keep using threshold for AdaBoost
procedure, our new system requires 30 seconds for each
weak classifier even though our computation is more
complicated by using floating-point numbers. The total
training time is now about 96 minutes or 1 hours
36 minutes, 2/3 of the previous time if we compare that to
Viola-Jones system.
The training time is significantly reduced if Gaussian
probability distribution is applied. For each weak classifier,
the processing time decreases to 22 seconds. Clearly, the
training process costs half of the original time with
76 minutes or 1 hour 16 minutes.
By using Gaussian probability distribution, the number
of operations is reduced and now the speed of training is
2 times faster than that of the previous work. However, in
Viola-Jones method, they had to use hard drive to
pre-compute training data and this process took a
significant time as described in previous section. If this
factor is taken into account, our new method is proved to
be far efficient not only in processing time but also in
memory usage.
5. Conclusion
In this paper, we propose a new way to apply Haar-like
patterns as well as how to use integral image technique for
computing feature values. For the same feature, much more
informative values can be extracted and hence, detection
rate is better as well.
The more important contribution is how we apply
Gaussian probability distribution to AdaBoost to improve
its performance. By utilizing this function, we can avoid
the difficulty to choose optimal thresholds for each round
of AdaBoost. Classification problem becomes simpler and
more straightforward by determining how far to the mean
positive and negative distributions are. Besides, the
detection speed is also faster because of classification rate.
Those two contributions have been characterized into
the success of our paper. By applying this system or this
idea about implementation, face detection system can be
run in a normal computer or machine. From the advance in
performance, this method can be used in other real-time
detection systems in practice.
Acknowledgement
This research was funded by Vietnam Ministry of
Science and Technology Research Project in 2017-2018,
No. CNTT-10
REFERENCES
[1] Bo Wu, Haizhou AI, Chang Huang and Shihong Lao, “Fast Rotation
Invariant Multi-View Face Detection Based on Real Adaboost”, IEEE FGR'04, 2004.
[2] Paul Viola, Michael J. Jones, “Fast Multi-view Face Detection”,
Mitsubishi Electric Research Lab TR-2003-96, 2003.
[3] Shengcai Liao, Anil K. Jain, and Stan Z. Li, “A Fast and Accurate
Unconstrained Face Detector”, 2015.
[4] T. Mita, T. Kaneko, and O. Hori, “Joint Haar-like Features for Face
Detection”, ICCV 2005.
[5] X. P. Burgos-Artizzu, P. Perona, “Robust Face Landmark
Estimation Under Occlusion”, ICCV, 2013.
[6] B. Leibe, E. Seemann, and B. Schiele, “Pedestrian Detection in
Crowded Scenes”, CVPR, 2005.
[7] S. Zhang, R. Benenson, M. Omran, J. Hosang, and B. Schiele,
“Towards Reaching Human Performance In Pedestrian Detection”,
IEEE PAMI, 2017.
[8] Paul Viola, Michael J. Jones, “Robust Real-time Face Detection”,
International Journal of Computer Vision, 2004, pp. 138-143.
[9] Robert E. Schapire, “Explaining AdaBoost”, In Empirical Inference,
2013.
[10] CBCL Face Database. Retrieved from http://cbcl.mit.edu/software-
datasets/FaceData2.html
[11] Dwayne Philips, “Image Processing in C 2nd”. R&D Publications,
2000.
(The Board of Editors received the paper on 03/01/2018, its review was completed on 03/4/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 51
THE EMPIRICAL STUDY ABOUT E-CRM:
A CASE STUDY OF VIETNAM AIRLINES
Nguyen Thi Khanh Chi
Foreign Trade University, Ha Noi; [email protected]
Abstract - Customer relationship management is now the tools and strategies for airlines to increase service quality and customer satisfaction which in turn gain customer loyalty on the age of IT. The objective of the present study is to analyze the effect of electronic customer relationship management (e-CRM) on customer’s loyalty in Vietnam Airlines (VNA). Research methodologies used in this paper are qualitative and quantitative methods. Qualitative method is expert interview conducted by questionnaires from 5 key managers to examine the e-CRM features and factors affecting customer loyalty from e-CRM. Quantitative method is qualified by a survey given to 401 customers randomly selected. Data is collected and examined by CFA and SEM through AMOS 20.0. The results of the study show that loyalty is significantly and positively affected by the largest determinant Service quality, followed by e-CRM features and Customer satisfaction.
Key words - e-CRM; service quality; customer satisfaction; loyalty; Vietnam Airlines
1. Introduction
E-CRM is Electronic Customer Relationship
Management and is defined in different ways by many
scholars, administrators and researchers to help businesses
attract more customers and attend customer loyalty. From
a technological perspective, e-CRM is a technology that is
used to create value when defining, implementing,
integrating, and concentrating on the business capabilities,
based on customers’ need. The aim of e-CRM is to create
customer value in the long run to segment existing
customers and improve customer profitability through the
use of software and the internet (Blery and
Michalakopoulos, 2006; Starkey and Woodcok, 2002; Xu
et al., 2002; Frow and Payne, 2004). From a strategic
perspective, e-CRM is a business strategy that builds and
develops long-term relationships with customers (Blery và
Michalakopoulos, 2006; Reichheld. F.F, 1996; Reichheld.
F.F & Sasser, 1990; Winer, 2001; Chen, L., & Sukpani, N.,
1998; Chan, S., & Lam, J., 2009).
The biggest benefit of e-CRM is to help enterprises
provide good services/products at levels above customer
expectations in order to attract more customers, retain
existing customers, improve customer profitability,
increase customer advocacy and engagement, gain
customer insight, increase customer lifetime value and
interaction. In return, customers will be loyal to the
enterprises for longer periods of time. Especially, Airlines’
effort to provide better loyalty programs and customization
of service (Ahodmotlghi, E. and Pawar, P., 2013). In
Vietnam, Vietnam Airlines (VNA) not only has e-CRM
features in e-CRM program such as Apps on mobile
devices, Website to provide service instantly to customers
but also has customer care like Frequent Flyer Program
(FFP), Email and Call Center to solve customer problems
and understand customer needs. VNA also offers
customers promotional fares and special discounts to
partner businesses based on FFP. The objectives of this
research are (1) to examine the factors affecting customer
loyalty, (2) to find the impact of factors affecting customer
loyalty, (3) to recommend VNA measures for improving
customer loyalty.
2. Literature review
2.1. E-CRM features
Anton and Postmus (1999) defined e-CRM features and
identified 25 features in their analysis and study of e-CRM
in retailing. E-CRM features are described as a site
customization, alternative channels, local search engine,
membership, mailing list, site tour, site map, introduction
for first-time users, chat, electronic bulletin board, online
purchasing, product information online, customization
possibilities, purchase condition, preview product, external
links, problem solving, complaining ability, spare parts,
customer service page (Anton, J. and Postmus, R., 1999).
In 2000, e-CRM features were added 17 features by
Feinberg et al. (2002). Several studies have attempted to
determine both empirical and conceptual relationship
between e-CRM features and customer satisfaction, and
e-CRM features and customer loyalty. There was a
statically significant positive relationship between the
number of e-CRM features and customer satisfaction
(Feinberg and Kadam, 2002; Anton and Postmus, 1999;
Feinberg et.al, 2002; Fragouli and Noutrixa, 2014; Kim
et.al, 2003; Mithas et. Al., 2005).
2.2. Service quality
Service quality has been defined in services marketing
literature as an overall assessment of service by the customers
(Ganguli, S. and Roy, S.K., 2011). According to Parasuraman
et al. (1985), service quality is the differences between
customer expectations and perceptions of service.
Parasuraman et al. introduced a formal service quality model
including 5 dimensions: tangibleness, reliability,
responsiveness, assurance, empathy. Gronroos (1984) offered
a service quality model with dimensions of technical quality,
functional quality and corporate image. Service quality is
found to be a strong predictor of customer satisfaction (Cronin
and Taylor, 1992; Dabholkar et al., 2000)
2.3. Customer satisfaction
Satisfaction was considered to be transaction-specific
construct which resulted from immediate post purchase
judgment or affective reaction (Oliver, 1993). Customer
satisfaction is also considered from a cumulative
satisfaction perspective and is defined as customer’s
overall experience to date with a product or service
provider (Johnson et al., 2001; Krepapa et al., 2003). Many
52 Nguyen Thi Khanh Chi
researchers reveal that service quality influences customer
loyalty indirectly through customer satisfaction (Anderson
and Sullivian, 1993; Caruana, 2002; Beerli et al., 2004;
Lewis and Soureli, 2006).
2.4. Loyalty
According to Kandampully and Suhartanto (2000),
loyal customer is a customer who intends to repurchase
from the same service provider, to keep a positive attitude
towards the service firm, and willing to refer the service to
others. Customer loyalty is defined as a deeply held
commitment to rebuy or patronize a preferred product or
service consistently in the future, despite situational
influences and marketing efforts having the potential to
cause switching behaviour (Oliver, 1997).
3. Methodology
3.1. Questionnaire design
Expert interview was conducted directly by
questionnaires from five key managers of VNA including
IT manager, Commerce manager, Service manager,
Investment and Strategy manager, and manager in human
resource department. The qualitative method was used to
examine the e-CRM features applied at VNA and factors
affecting customer loyalty from e-CRM at VNA. Basing on
42 e-CRM features identified by Anton and Postmus (1999)
and Feinberg et al. (2000) to interview three managers
(IT, commerce and service department), the result showed
that seven e-CRM features used at VNA were identified to
directly affect service quality and customer loyalty. They are
Email, Online purchasing, FAQ, Postal address, Problem
solving, Information online, and Perceived rewards.
Especially, when five managers were asked about the effect
of these e-CRM features on loyalty, they confirmed that
e-CRM features have had impact on service quality,
customer satisfaction and loyalty.
Therefore, this study also examined two more factors
(service quality and customer satisfaction) outside e-CRM
features affecting customer loyalty. An empirical research
model is developed in Figure 1. There are four constructs
such as e-CRM features (EC), Service quality (SQ),
Customer satisfaction (CS) and Loyalty (LO).
Figure 1. The empirical research model
The questionnaire was designed in three parts. The first
part contained 6 questions related to general information of
customers such as kind of passenger, the times of passenger
flying with VNA, kind of light route, kind of seat class, the
way to buy tickets and the reason to choose VNA. The
second part contained 19 questions about e-CRM features,
Service quality, Customer satisfaction and Loyalty. The
third part was designed to identify demographic attributes
of the respondents and consisted of 4 questions such as
gender, age group, nationality and occupation. The total
questions were 29. The form of survey was set online by
using Google docs. The languages of questionnaires were
set in two languages (English and Vietnamese).
Table 1. Measurement scale of research
The questions in Part 1 and Part 3 were set as fixed-
alternative which were designed as multiple-choice
answers. Questions in Part 2 were structured which were
designed as Likert with 05 scale from “1” to “5” where “1”
Strongly Disagree, "2" Disagree, "3" No Comment, "4"
Agree, "5" Strongly Agree.
3.2. Sample size
The population of research consists of international
passengers travelling to Vietnam using VNA service and
domestic customers. According to the report of CAPA,
VNA reached 20 million arrivals in 2016 (Hoang, 2017).
In the first six months of 2017, VNA had nearly 10.3 million
arrivals, up 6% over the same period last year. In order to
collect data, a random sample was conducted in this
research. The target group was VNA customers who are the
member of VNA’s Frequent Flyer Program. It is almost
impossible to collect all passengers because there are a large
number of passengers in Viet Nam and over the world.
Therefore, random sampling method was necessary to select
a small group from the list of customers who are member of
Vietnam Airlines Frequent Flyer Program. The sample size
is 600 including international and domestic passengers.
The questionnaires were sent to 600 customers including
Vietnamese and Foreigner ones. and the respondents were
401, the return rate was at 66.83%.
3.3. Statistical procedures
Data was collected by questionnaires and took place
over 56 days. The data was recorded firstly in Excel
program from 401 online answers from automatic excel in
Google document. SPSS 20.0 was conducted secondly for
analyzing Cronbach’ Alpha and EFA. The results of EFA
analyzing was used for CFA and structural equation
modelling analysis (SEM) in AMOS 20.0 program in order
to test hypotheses
4. Findings
The results show that out of 401 respondents, 214 of
whom are male (53.3%) and 187 are female (46.7%). The
respondents concentrated on the age group of 23 to 45 with
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 53
the total number of 240 (69.8%) which was under the
working age while the other is lower (<23 is only 8% and
>45 is 22.2%). Regarding to occupation, the respondents
were mainly officers, which could make the result more
applicable towards the officer sample. The number of
officers is 212 (52.9%), of senior officers is 112 (27.9%),
others are 52 (13%) and retired/house wives are only
25 (6.2%). For nationality, there are a number of
nationalities such as Vietnamese, Hong Kong, French,
Australian, Korean, Taiwanese, Japanese and Singapore.
Most of them are Vietnamese. In terms of passenger card,
184 (45.9%) were Silver card passengers, 128 (31.9%) was
Titan card passengers, 48 (12%) was Gold card passengers
and the remainder 41 (10.2%) was Platinum card passengers.
The major reason why customer choose Vietnam Airlines
was its service quality (29.9%), next was suitable flight
schedule (26.9%), followed was better connectivity to
various destination (22%), ticket price (13.9%), Brand
image (7%), the last was others (0.3%). The flight frequency
of passengers mostly concentrates on more than
20 times/year (40%). This is the reasonable representation of
flight frequency due to the subjects of this research are
member of Frequent Flyer Program. The major routes which
passenger selected were both (domestic route and
international route) and mark up 41.2%. The economy class
was mostly chosen by passenger (50.2%). The business class
and premium economy class were the same with
24.9% each.
Using the Cronbach’ Alpha coefficient to measure the
reliability of data with 4 constructs and 19 observed
variables, the Cronbach’ Alpha values of EC, SQ, CS and
LO are 0.867, 0.850, 0.805 and 0.801 respectively (>0.6) and
the Corrected items (Total Correlation coefficient) of
19 observed variables are higher than 0.3. It can be
concluded that there are 19 good reliability variables from 4
constructs.
By conducting an EFA with the principal axis factoring
of component method and the Promax with Kaiser
Normalization method, the first result has not reached
convergence value even though KMO is high at 0.900 and
sig is 0.000. Continuously removing fours inappropriate
variables from the model (EC7, SQ3, SQ7, CS3), the model
has been conducted secondly and the results get convergence
factor at four group components (EC, SQ, CS and LO) with
KMO of 0.879 and sig of 0.000. In the extraction sums of
squared loadings, the percentage of cumulative is 59.620%
and the total of initial eigenvalues is 1.005.
Confirmatory factor analysis (CFA) was conducted to
identify the relationship between 4 constructs and
19 observed variables and was examined two times. The
results for the first running CFA show that Chi-square/df
was 4.317, GFI was 0.893, TLI was 0.883, CFI was
0.907 and RMSE was 0.091. Meanwhile CFI and GFI
values greater than 0.90 indicate good model fit (Hu, L.T.
& Bentler, P.M. , 1999). GFI values greater than
0.70 indicate good model fit (Schumacker, R.E. & Lomax,
R. G., 2004). RMSEA (root mean squared error) values
less than 0.06 also indicate a good model fit (Hu, L.T. &
Bentler, P.M. , 1999) while values ranging from 0.08 to
0.10 indicate mediocre model fit and those greater than
0.10 indicate poor fit (Byrne, 2001). Therefore, model fit
of the first CFA running is not good, but the covariance of
M.I (modification indices) between two exogenous
variables (e1 and e2) was highest at 61.240, followed by
the covariance of e5 and e6 at 24.534. Thus, CFA was
conducted the second time to adjust the absolute value by
linking e1with e2 and e5 with e6. Consequently, the GFI
(goodness-of-fit index), TLA, CFI (comparative fit index)
had higher value than the first one.
Structural Equation Modelling (SEM) procedures were
used to determine the impact of e-CRM features (EC),
service quality (SQ) and customer satisfaction (CS) on
loyalty (LO) and were also used to determine the
relationship between EC and SQ (H1); EC and CS (H2);
EC and LO (H3); SQ and CS (H4); SQ and LO (H5);
CS and LO (H6); the interrelationship among EC and SQ
on CS (H7); EC, CS, SQ on LO (H8).
Figure 2. The SEM analysis result
The SEM showed the total effect between CS and LO
(0.204), SQ and LO (0.343), EC and LO (0.208) whereas SQ
has strongest impact on LO. However, the result of SEM
analysis did not show an indirect effect link from EC and CS
to LO, an indirect effect link from SQ and CS to LO. It can
be concluded that e-CRM features, customer satisfaction and
service quality have direct impact on loyalty.
The finding indicates that e-CRM features (EC),
service quality (SQ) and customer satisfaction (CS) have
significant and positive influence on Loyalty (LO) with
Beta at 0.215, 0.391 and 0.175, p value <0.05. Thus, H3,
H5 and H6 are asserted. Besides, EC and SQ, SQ and CS,
EC and CS have interaction at 0.224, 0.206 and
0.321 respectively whereby e-CRM features and customer
satisfaction have largest impact (Beta at 0.321). Thus, H1,
H2, H4 are supported. Moreover, H7 is also asserted
because Customer satisfaction (CS) is predicted positively
by e-CRM features (EC) and Service quality (SQ).
Additionally, Loyalty (LO) is positively influenced by EC,
SQ and CS, hence, H8 is asserted.
Consequently, eight hypotheses are examined by
confirming the presence of a statistically significant
relationship in the predicted direction. Loyalty is
significantly and positively impacted by the largest
54 Nguyen Thi Khanh Chi
determinant Service quality, followed by e-CRM features
and Customer satisfaction in case of VNA.
5. Conclusion
The results show that e-CRM features, service quality
and customer satisfaction are factors affecting VNA
customer loyalty. Particularly, e-CRM features have large
impact on service quality and customer satisfaction which in
turn affect loyalty. Service quality has great influence on
customer loyalty. The results of this study have important
implications to VNA for obtaining customer loyalty in the
long time. In all empirical research, this study has limitations
that need to be identified. Firstly, the sample is limited to
VNA customers. Secondly although the results from this
research are useful for describing the characteristics of a
large population of passengers, the generalizations of the
results are limited to all VNA customers who are not in the
list of FFP. In our future work, we will try to examine new
variables in context of applying e-CRM at VNA.
REFERENCE
[1] Anton, J. and Postmus, R., 1999. The CRM performance index for Web based business. unpublished paper, available at:
www.benchmarportal.com
[2] Byrne, B. M., 2001. Structural equation modeling with AMOS.
Mahwah, NJ: Lawrence Erlbaum Associates.
[3] Caruana, A., 2002. Service loyalty, the effects of service quality and
the mediating role of customer satisfaction. European Journal of
Marketing, Volume 36, pp. 811-828.
[4] Chen, I.J and Popovich, K., 2003. Understanding customer
relationship management: people, process and technology. Business Process Management, 9(5), pp. 672-687.
[5] Cronin, J.J. Jr and Taylor, S.A. , 1992. Measuring service quality: a
reexamination and extension. Journal of Marketing, 56(3), pp. 55-68.
[6] Dabholkar, P. and Bagozzi, R.P, 2002. An attitudinal model of
technology-based self-service: moderating effects of consumer traits
and situational factors. Journal of the Academy of Marketing
Science, 30(3), pp. 184-201.
[7] Feinberg, R. and Kadam, R., 2002. E-CRM web service attributes as
determinants of customer satisfaction with retail websites. International
journal of service industry management, 13(5), pp. 432-451.
[8] Gronroos, C., 1984. A service quality model and its marketing
implications. European Journal of Marketing, 18(4), pp. 36-44.
[9] Hoang, N., 2017. Vietnam Airlines là nhà vận chuyển hành khách lớn thứ 2 ĐNA. [Online] Available at: https://news.zing.vn/vietnam-
airlines-la-nha-van-chuyen-hanh-khach-lon-thu-2-dna-
post716293.html [Accessed 1 1 2018].
[10] Kandampully, J., and Suhartanto, D, 2000. Customer Loyalty in the
Hotel Industry: The Role of Customer Satisfaction and Image. International Journal of Contemporary Hospitality Management,
Volume 12, p. 346.
[11] Kim, J., Suh, E. and Hwang, H., 2003. A model for evaluating the
effectiveness of CRM using the balanced scorecard. Journal of
interactive Marketing, 17(2), pp. 5-19.
[12] Krepapa, A., Berthon, P., Webb, D. and Pitt, L, 2003. Mind the gap:
an analysis of service provider versus customer perceptions of market orientation and the impact on satisfaction. European Journal
of Marketing, 37(2), pp. 197-218.
[13] Lewis, B., and Soureli, M, 2006. The antecedents of consumer
loyalty in retail banking. Journal of Consumer Behaviour, Volume
5, pp. 15-31.
[14] Mithas, S., Krishnan, M.S. and Fornell, C., 2005. Why do customer
relationship management applications affect customer satisfaction?. Journal of Marketing, 69(4), pp. 201-209.
[15] Parasuraman, A., Zeithaml, V.A. and Berry, L.L, 1985. “A
conceptual model of service quality and its implications for future
research. Journal of Marketing, 49(4), pp. 41-50.
(The Board of Editors received the paper on 06/4/2018, its review was completed on 18/5/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 55
TEACHERS’ PERCEPTION TOWARDS THE USE OF ICT IN VIETNAM:
USING ACTIVITY THEORY TO IDENTIFY CONTRADICTIONS
Huynh Ngoc Mai Kha1, Pham Thi To Nhu2 1University of Foreign Language Studies – The University of Danang;[email protected]
2Universiti Malaysia Sabah; [email protected]
Abstract - This paper presents the findings of a study which was undertaken at five different primary schools in Vietnam. The participants were 5 teachers of English aged 28 to 35 in Quang Ngai province. This quantitative study aims to explore teachers’ perception towards the use of ICT in their schools in Vietnam. Activity Theory is employed as the framework for guiding the study owing to the fact that its focus is on the identification of contradictions occurring in the activity system. From Activity Theory analysis, several contradictions could be located. The study provides implications for future research in terms of teachers’ adoption of ICT at different levels as well as recommendations for ICT use improvements.
Key words - activity theory; contradictions; ICT; perception; teacher
1. Introduction
ICT (Information and Communications Technology) has
played an important role in English language teaching and
learning in the world. The use of ICT has been emphasized
and spread all over the world in this field. Developed countries
in Asia strongly endorse and support ICT as an essential
component of innovative student-centered pedagogy (Albion,
Tondeur, Forkosh-Baruch, & Peeraer, 2015). As Peearer and
Van Petegem (2011) mentioned, with directive 55 (MOET,
2008), an educational reform rationale puts the emphasis in
that direction in Vietnam too. In the context of the movement
for friendly schools and active students (MOET, 2009) the
role of ICT is conceptualized as to support educational
renovation towards a creative learning society. In addition,
educators in Vietnam were encouraged to reasonably
implement ICT applications as part of new and innovative
methods of teaching and learning (MOET, 2008). However,
whether ICT has been adopted effectively or not is still
controversial. In fact, few studies have consistently shown that
technology integration shows disappointing levels of
penetration and success (Cuban, Kirkpatrick& Peck, 2001;
Bauer & Kenton, 2005; Dang, 2013). Recently, it has been
pointed out that there are crucial teacher attributes including
perceptions, beliefs and attitudes which play an important part
in the acceptance or rejection of ICT (Vandelinde, 2011;
Veen, 1993; Mumtaz, 2000, Jimoyiannis& Komis, 2006).
Moreover, according to Loveless (2003), teachers are aware
of the ubiquitous presence of ICT in their teaching
environment, but may not perceive the link to their teaching
practices. In reality, teachers’ perception towards the use of
ICT in teaching is very important as it forms a tendency that
makes them feel favorable or unfavorable towards the use of
modern technology in teaching (Qasem, 2016). In Vietnam,
this situation is not different with poor penetration of ICT in
teaching (Hong, 2014). On the other hand, Activity Theory
(AT) is thought to be the best kept secret in academia
(Engestrom, 1993) due to its popularity for use in the field of
education. Besides, AT can be used as a lens to understand the
important issues related to a certain matter (Murpy &
Rodruguez - Manzanares, 2008). In fact, in simple terms,
ATis all about ‘who is doing what, why and how’ (Hasan &
Kazlauskas, 2013). In AT, the relationship between subject
(the doer) and doer (the thing being done) forms the core of an
activity (Figure 1)
The Core of an Activity
Subject object outcomes
(the doer) (the deed)
Figure 1. The core of an activity
(Adapted from Activity Theory: who is doing what, why
and how (Hassan &Kasslauskas, 2013)
As can be seen from Figure 1, the subject of an activity
encompasses the activity’s focus and purpose while the
subject, a person or group engaged in the activity,
incorporates the subject’s/s’ various motives. The outcomes
of an activity can be the intended ones, but there can also be
others that are unintended (Hassan &Kasslauskas, 2013).
In addition, according to Hardman (2005), the basic
unit of analysis for AT is an activity system which refers to
a group of people who share a common object (or problem
space) and who use tools to act on that object, transforming
it. In Figure 2, the object is represented as a circle
indicating that this space is subject to change and is in a
state of flux, making it difficult to pin down (Hardman,
2005). This author also maintains that relationships in this
system are driven by rules, which both afford and constrain
behaviors. Rules are the norms and sanctions that specify
and regulate the expected correct procedures and
acceptable interactions among the participants (Cole &
Engerstrom, 1993). Division of labour within the system
describes both a horizontal division among community
members, as well as a vertical division between power-and
status-holders which then can be understood as related to
power within and between systems (Hardman, 2005).
Figure 2. An activity system
(Adapted from Activity Theory as a Framework for
Understanding Teachers’ Perceptions of Computer Usage at a
Primary School Level in South Africa (Hardman, 2005))
Using the above-mentioned model to analyze teachers’
perception on the use of ICT, the elements of AT of the
56 Huynh Ngoc Mai Kha, Pham Thi To Nhu
present study can be mapped as follows:
Subject Teachers
Object The goal of teaching (quality
communication skills both written and spoken)
Tools Methods and facilities used for teaching
Rules Rules of the schools and relevant authorities
Community students, teachers, and relevant
administration as well as staff
Division of Labour The roles and
responsibilities of members of the community
(Adapted from English Teachers’ Perceptions about
Their Teaching: Using Activity Theory to Identify
Contradictions (Marwan, 2009)).
In this study, AT is used to examine the contradictions
within teachers’ perception on the use of ICT in their
teaching. Therefore, the following research question is
used to guide this study:
Research question: What are the contradictions within
teachers’ perception on the use of ICT in their teaching?
1.1. Literature review
The term 'ICT' is defined as “forms of technology used
for creating, displaying, storing, manipulating, and
exchanging information” (Donnelly, McGarr, & O’Reilly,
2011). This definition seems to be general, thus, within the
scope of this study, ICT is defined as computer, and the
internet-based technologies which can be categorised into
two types: i) generic software applications, e.g., word
processors, presentation software, email packages, and
web browsers; and ii) CALL software applications and
useful websites with a focus on purposeful language
teaching and learning (Sarkar, 2012).
1.2. Teachers’ Perception on ICT Use
Hepp, Hinostraza, Lavaland Rehpain (2004) suggest
that teacher beliefs and attitude to ICT influence the rate of
ICT adoption. In fact, they identify those who recognize
the potential of ICT will quickly explore tools in their
practice and perceive computers as a ‘valuable tool’ and
‘useful’ (Hepp et. al, 2004).
In addition, Loveless (2003) claims that teachers’
perceptions on ICT use are fashioned by their identity and
participation in wider cultural and social sphere which
influence the professional areas and settings in which they
practice. Besides, in Rogers’ Diffusion of Innovation Theory
(2003), perceptions on ICT use include perceived ease of use,
perceived usefulness and satisfaction. According to Roger,
perceived ease of use refers to the degree to which a teacher
believes that using ICT will be free from effort. Teachers may
believe that technology is useful and at the same time, they may
perceive the use of technology to be too difficult and therefore
performance benefits of usage are outweighed by the effort of
using the technology. This in turn affects the actual use of ICT
in education. Perceived usefulness is the degree to which a
teacher believes that using ICT will enhance his or her job
performance at school (Rogers, 2003). It has been proved that
teachers tend to use ICT when they think that it will help
enhancing their teaching (Ma, Anderson &Streith, 2005). This
is also supported by Knezek and Christenen (2002) who
reasons that perception of potential usefulness of the computer
could influence attitude towards use of ICT. Also, according to
Huang and Liaw (2005), teachers’ attitudes towards ICT and
their perceptions on ICT use play an important role in their
making use of ICT in their teaching activities. Such perceptions
are significant as they may influence the teachers’ future ICT
pedagogical practices (Von Konsky et al, 2009; Al-Zaidiyeen
et al, 2010). Last but not least, Harris (2002) holds the
viewpoint that the benefits of ICT will be gained when
confident teachers with high perceptions on ICT use are willing
to explore new opportunities for changing their classroom
practices by using ICT.
2. Method
Five full time teachers at 5 different primary schools in
QuangNgai province located in the Central of Vietnam
were invited to take part voluntarily in this research. Prior
to the recruitment, fliers containing information about the
study, including the research contents, the research aim and
the researcher’s contact details were sent to potential
teachers. After some time, five teachers contacted the
researcher and expressed their interests in sharing their
perception on the use of ICT in their teaching context.
Details of the 5 teachers (already coded as T1, T2, T3,
T4 and T5) are shown in Table 1.
Table 1. List of teachers in this study by genders, highest
academic achievements, number of years teaching English
No. Participant
Code Gender
Highest
academic
achievement
Number of
years teaching
English
1 T1 Male BA in TEFL 5
2 T2 Male BA in English 4
3 T3 Female BA in English 1
4 T4 Male BA in TEFL 4
5 T5 Female BA in TEFL for
Primary Education 3
Semi-structured interview is chosen as an instrument
for data collection. This tool is used because it allows the
researcher to probe for views and opinions of the
participants (Corbetta, 2003). Furthermore, it gives the
researchers the chance to gain deep information about the
phenomena being investigated (Cresswell, 2005). The
interview questions were used to explore all the issues
related to the contradictions within the activity system at
such schools in Quang Ngai provice. All the questions in
this kind of focus group interview were intended to give a
full description of each component of the Activity system
of the teaching context in terms of Activity Theory as a
framework of this study. (See Appendix for the questions)
Data is analyzed using Nvivo 8 (2008). The tree nodes
(tho (T4) ughts and definitions about data, together with
selected passages of text) are developed to create ideas,
concepts, categorize the data. Walsh (2006) claims that this
software is useful as it helps the researcher organize raw
data and links them with memos and data biting where the
researcher can make codes and analytical notes, and then
edit and rework ideas as the project progresses. For
preserving the participants’ anonymity, the participants are
referred to as T1, T2, T3, T4 and T5.
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 57
3. Results
The findings presented below are some of
contradictions found from participants’ responses.
In this research, teachers are willing to promote the use
of ICT in their teaching to help students improve their
communications skills, but find they are incapable of
designing suitable activities with ICT as they think such
designing activities would take time and much effort. In
addition, lacking facilities in teaching is another problem.
This is a contradiction related to Subject and Tools (1).
This is acknowledged in the following comments:
Nowadays using technology in teaching also gives
teacher some difficulties. Teachers need a lot of time to
find material to prepare lesson plans as well as it also
depends on many extent factors.
(T1)
I know that using ICT in my teaching is very useful as it may
help me to save my time and it will have great effect to students
but my ICT use is quite limited because I am not confident in using
ICT in my class. I can only use powerpoint and some audio.
(T5)
My school is in a poor village and far from town so I
only use my computer to prepare lesson plan, check
pronunciation and play recording for my students to listen
to. As a result during English lessons some students feel
bored and difficult to learn. (T2)
It is very difficult for me to teach English. At my school,
there is nothing in it although it is in the central of Duc pho
town. There isn't computer room, interactive table and
many different facilities. I do everything myself. I buy a
cassette to teach listening for my students and I only use
cassette to teach English at school
(T3)
The inadequacy of technical staff leads to a
contradiction between Subject and Division of Labour (2)
as teachers need technical support when they encounter
difficulties when using ICT in their teaching but usually
there are few staff available when teachers are in need. For
example, a teacher commented as follows.
Teacher needs a lot of time to prepare electric lesson
planning such as search pictures, fun games, .... Teaching
technology depends on many things: electricity,
computers, machines but I often have some difficulties in
using them. Unluckily, whenever I have problems, I can’t
find any technical staff to help me.
The workload teachers suffer from other work besides
teaching hinders teachers from raising good quality in using
ICT for their teaching. This is one form of contradiction
between Tools and Object (3). It is true that if teachers are
given a fine workload where there are chances for them to
fulfill, they could provide quality teaching to students
(Dison, Scott, & Dixon, 2007). T4 said:
Using ICT is time consuming and there are not enough
facilities. Moreover, the new curriculum contains too many
lessons. Besides teaching, I still have a lot of things to do
such as doing the task as a form teacher. You see, I have to
attend many meetings.
The lack of coordination and support among teachers
and administrative staff (i.e. those in charge for timetable
management and curriculum management) leads to the
contradiction between Division of Labor and Object (4)
because teachers’ willingness to diversify their teaching to
motivate their students so that their learning will become
better due to the fact that teachers sometime cannot use the
language laboratory or the contents of the lesson are too
long for teachers to cover in one period. One teacher
expressed his opinion as follows:
(T2)
Sometimes I cannot use the lab as it is overlapped and
my lesson has too many contents. I cannot let students play
some games with ICT though I know they will be very
interested in it.
(T1)
A contradiction also occurs between the Subject ad
Rules (5) since teachers’ effort in using ICT to create
exciting activities may not be supported by a number of
schools’ regulations due to the workload or textbook (T3).
One teacher said during the interview:
The curriculum is so long, it is hard for me cover all the
contents. If I create more products, it will take not only my
time at home but also my time in class. It is very hard for
me in my situation
Figure 3. Contradictions within the activity system
4. Conclusions and Suggestions
The good adoption of ICT into teaching requires many
different factors. To understand which factors may facilitate
or constrain the adoption process requires a good
understanding of activity system in which they are located.
Engesstrom (2001) claims that an analysis of contradiction
using Activity Theory is a useful way. Thanks to such analysis
approach, the researcher could understand the whole process
within its activity system, teachers’ use of ICT in their
teaching in this context. In addition, she could also identify
problems that need to be addressed so that the situation would
be improved. Hence, the contradictions in this research should
not be merely seen as the problems, but they are considered as
the useful sources for improvements (Nelson, 2002).
In this study, there are several contradictions within
teachers’ perceptions on the use of ICT in their teaching context.
First, the use of compulsory textbooks puts some constraints on
teachers as its long contents hinder teachers from performing
some other kinds of activities that bring good motivation and
58 Huynh Ngoc Mai Kha, Pham Thi To Nhu
interest to their students. Consequently, the view that requires
teachers teach the whole contents of the books should be
reviewed so that teachers could base on the main contents of the
textbooks and choose what and how to teach so that their
teaching will bring the best quality to their students.
Second, the contradiction related to the lack of
technical support for supporting teachers in their use of
ICT in teaching should not happen in the future. This can
be done in different ways. On the one hand, all the facilities
must be checked regularly to make sure that they are in
good conditions. On the other hand, each school should
have enough technicians who are present at the language
laboratory so that teachers could be supported in time. This
will, in turn, encourages teachers use ICT in their teaching
as teachers feel more confident.
Third, the contradictions triggered by teachers and other
staff in administration due to workload and timetable
management can be solved if the teaching quality is considered
a priority and teachers should be facilitated in their teaching.
In summary, this study has provided some
recommendations for the improvements based on the findings
regarding to the contradictions happening within the system
using AT as the framework. The study cannot avoid some
limitations due to the limited number of participants and
therefore the results might hardly be generalized.
Furthermore, the study is carried out at primary school level;
further study can also be conducted at different levels.
Appendix
Questions for the focus group interviews
i. What is your viewpoint about ICT use in teaching? Do you
often use ICT in your teaching? When teaching speaking skills, what
advantages and disadvantages do you encounter if you use ICT?
ii. Do your school and relevant authorities support your use
of ICT in teaching? In what ways?
iii. Do you use ICT in teaching speaking skills? Which ICT
use do you adopt in teaching speaking? What are the students’
attitudes toward such use of ICT?
iv. Do you think it is necessary to integrate ICT in your
teaching? What will facilitate this integration?
v. In what ways could the skills you presented in your
products be applied into real life setting or workplace?
vi. How did educational technology (eg. Web Tools 2.0) help
in teaching and learning?
vii. Do you think it was important to integrate educational
technology into curriculum-based learning? Why do you think so?
viii. What were the changes you wish you could have
made during the planning, designing, and implementation stage
and how could the changes affect teaching and learning?
ix. How could you further improve your classroom
management/teaching method?
x. Did your product(s) benefit your learners in terms of their
knowledge and soft skills? Why do you think so?
xi. Do you think your teaching approaches work well with the
application of educational technology? If so, in what way? If not, why?
xii. How did your instructions and product(s) impact the
learners? If so, in what way? If not, why?
xiii. What difficulties do you encounter when developing
technology based materials for teaching speaking skills?
xiv. Could you suggest some solutions to those problems?
REFERENCES
[1] Bauer, J., & Kenton, K. (2005). Toward technology integration in
the schools: Why it isn’t happening. Journal of Technology and
Teacher Education, 13(4), 519-546.
[2] Bingimlas, K. A. (2009). Barriers to the successful integration of
ICT in teaching and learning environments: A review of the literature. Eurasia Journal of Mathematics, Science and Technology
Education, 5(3), 235-245.
[3] Creswell, J. W. (2005). Educational research: Planning, conducting
and evaluating quantitative and qualitative research (2nd ed.). Upper Saddle River: Merrill Prentice Hall.
[4] Dang, X. T. (2013). ICT in Foreign Language Teaching in an
Innovative University in Vietnam: Current Practices and Factors
Affecting ICT Use. La Trobe University: Australia.
[5] Donnelly, D., McGarr, O., & O’Reilly, J. (2011). A framework for teachers’ integration of ICT into their classroom practice. Computers and Education,
57(2), 1469–1483. https://doi.org/10.1016/j.compedu.2011.02.014
[6] Engestrom, Y. (1993). Development Studies of Work as a Testbench
of Activity Theory: The Case of Primary Care Medical Practice. In S. Chaiklin & J. Lave (Eds.) Understanding practice: Perspective on
Activity and Context. Cambridge: Cambridge University Press.
[7] Hardman, J. (2005). Activity Theory as a framework for understanding
teachers' perceptions of computer usage at a primary school level in South Africa. South African Journal of Education. Retrieved from
https://vn.search.yahoo.com/yhs/search?hspart=visicom&hsimp=yhs-
weathernow&p=Activity+Theory+as+a+framework+for+understanding+teachers%27+perceptions+of+computer+usage+at+a+primary+sc
hool+level+in+South+Africa&type=150_VN_vn
[8] Hassn, H., & Kazlauskas, A. (2013). Activity Theory: who is doing what,
why and how. Theori Research Group. University of Wollongong.
[9] Loveless, A. (2003). The Role of ICT. Continuum: London & New York.Knezek, G. & Christensen, R. (2002). Impact of New
Information Technologies on Teachers and Students. Education and
Information Technologies, Vol. 7, no. 4, 369-376.
[10] Krashen, S. (2013). Second Language Acquisition: Theory, Applications, and Some Conjectures. Cambridge University Press.
[11] Levy, M. and Stockwell, G. (2006). CALL dimensions: Options and issues
in computer-assisted language learning. Mahwah, NJ: Lawrence Erlbaum.
[12] MOET. (2009). Report on 2009 national university entrance
examinations. Retrieved November 20, 2009, from http://moet.gov.vn/?page=1.26&view=1368.
[13] Murphy, E., & Rodruguez- Manzanares, M.A. (2008). Using
Activity Theory and Its Principle of Contradictions to Guide Research in Educational Technology. Australia Journal of
Educational Technology, 24 (4): 442: 457.
[14] Peeraer, J., & Van Petegem, P. (2011). ICT in teacher education in an
emerging developing country: Vietnam’s baseline situation at the start
of “The Year of ICT”. Computers and Education, 56(4), 974–982
[15] Rogers, E.M. (2003). Diffusion of Innovation Theory. Macmilan
Publishers: London.
[16] Sarkar, S. (2012). The Role of Information and Communication
Technology (ICT) in Higher Education for the 21st Century. The Science Probe, 1(1), 30–41.
[17] Vanderlilnde, R. (2011). School-based ICT Policy Planning in a
Context of Curriculum Reform (Doctoral Disseratation, Faculty of
Psychology & Pedagogic Sciences, University of Gent)
[18] Veen, W. (1993). The Roles of Beliefs in the Use of Information Technology: implications for teacher educaiton, or teaching the right
thing at the right time, Journal of Information Technology for
Teacher Education, 2: 2, 139-153.
[19] Von Konsky, B.R., Ivins, J. & Gribble, S.J., 2009. Lecture attendance and web based lecture technologies: A comparison of
student perceptions and usage patterns. Australasian Journal of
Educational Technology, 25 (4), 581-595.
[20] Vu, N.T.T. & Burns. A. (2014). English as a medium of instruciton: Challenges for Vietnamese tertiary lecturers. The Jounal of ASIA
TEFL, 11(3), 1-31.
(The Board of Editors received the paper on 29/5/2018, its review was completed on 13/6/2018)
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 59
IMPROVING STRUCTURES OF STUDENTS’ ARGUMENTATIVE ESSAYS
THROUGH GENRE PEDAGOGY
Dinh Thanh Liem
University of Foreign Language Studies - The University of Danang; [email protected]
Abstract - This paper analyses the structures of argumentative texts produced by second-year students who major in English for Tourism at the Department of English for Specific Purposes, University of Foreign Language Studies – the University of Danang. It seeks to examine whether the students’ text structures can be improved after they complete a ten-hour writing course informed through genre pedagogy. The article is underpinned by genre theory developed within Systemic Functional Linguistics. As the finding indicates, the students show an improvement in rhetorical structures of texts, which allows them to achieve communicative purposes in the cultural and situational contexts thanks to the adoption of genre pedagogy.
Key words - genre pedagogy; learners; teachers; argumentative writing; text structure
1. Introduction
Argumentative writing is one of the key skills in
curricula at many educational levels in many countries. In
Australia, the argumentative genre is foregrounded in the
school curriculum where students are expected to
demonstrate the ability to write effective argumentative
essays and appraise diverse opinions. In EFL contexts such
as Vietnam, the importance of argumentative writing is
also emphasized in the curricula of many disciplines,
including the English major.
Argumentative writing is prominent not only in the
curricula of many countries, but is also emphasized in a
range of high-stakes testing and assessment contexts. It is
assessed in the National Assessment Program – Literacy and
Numeracy (NAPLAN) in Australia as persuasive writing.
Internationally, the ability to persuade others is privileged in
the high-stakes standardized exams and is tested in both the
International English Language Testing System (IELTS)
and the Test of English as a Foreign Language (TOEFL),
both of which act as gate-keeper for admission to an English-
speaking university. In China, argumentative writing is
included as a major component of the college test
administered upon completion of tertiary studies. Similarly,
this argumentative genre is also a core requirement of three-
level Vietnamese Standardized Test of English Proficiency
(VSTEP) in Vietnam. Although the ability to write well-
constructed arguments with persuasive supporting evidence
is critical for academic success, little attention has been paid
to building the capability of student writers to meet high
demands of the academic world as well as to satisfy the
requirements of future professions upon graduation.
On the other hand, previous traditional approaches to
writing instruction do not generate significant benefical
impacts on students’ writing skills. One of these approaches
is product-oriented, which has been dominant in Vietnamese
classrooms for years. This approach has been criticised due
to the failure to prepare students with necessary linguistic
resources to make meanings and raise student awareness of
text in its social context. Writing in this approach is seen just
as a different way of learning about grammar and the job of
the teacher is confined to designing, assigning and assessing
student writing. As a response to the shortcomings of the
traditional model, educational institutions have, in recent
years, adopted a process-oriented approach. This approach
has been long applauded for its contributions to developing
cycles of planning, drafting and revising. One serious
concern with this tradition, however, is its efficacy as an
instructional approach in EFL classrooms to deal with the
complicated nature of the writing task. This approach places
an over-emphasis on ones’ personal experience and one’s
own writing process without a theoretical base of the way
language is patterned to make meanings in human interaction
and underestimates the role of the teacher who should be
empowered to teach rather than raise metacognitive
awareness (Hyland, 2003). This approach also does not give
enough attention to L2 learners, particularly struggling ones
who have limited understanding of the rhetorical structures
of target text (Cope & Kalantzis, 1993).
Although the above-mentioned orientations have been
criticized by many scholars, genre-based tradition has been
developed as an innovative pedagogy to develop
knowledge of text structure for learners through explicit
teaching of genre. The genre pedagogy views the social
nature of language use and prototypical structures of texts
as a central point to make meanings and accomplish the
purposeful act of writing. The fundamental tenet of the
pedagogy is based on “guidance through interaction in the
context of shared experience” (Rose & Martin, 2012,
p. 58). It aims to develop conscious understandings of
genre and build up a repertoire of language to enable
successful written performance. The pedagogy not only
focuses on the writing product, but also the process of how
to write to achieve communicative goals.
The genre pedagogy is developed based on the theory
of systemic functional linguistics (SFL) (Halliday &
Matthiessen, 2013). In SFL, texts with similar social
purposes and linguistic patterns are considered to be
instances of the same genre (Humphrey, Droga, & Feez,
2012). Genre can be defined as ‘staged, goal-oriented,
social activity’ (Rose & Martin, 2012, p. 53) which meet
various social purposes in the cultural and situational
contexts. Depending on the social purpose of a text in its
cultural context, it has its own text organizations with
Stages and Phases to achieve its purpose and depending on
the situational context including field (topic), tenor (who is
involed in the communication) and mode (written or oral),
a text has its own distinctive linguistic features approroriate
with the genre. Based on the purpose of persuasion,
argumentative texts are classified into four distinct genres:
60 Dinh Thanh Liem
hortatory exposition, analytical exposition, discussion and
challenge (Derewianka, 2016).
In this article, the structure of analytical exposition is
chosen for research because learners often encounter this
genre in important exams such as VSTEP. The structure of
analytical exposition has three main Stages: Thesis,
Arguments and Reiteration (Martin & Rose, 2012). In the
thesis, the writer states his or her views on the topic of
discussion and previews key points. In the body
paragraphs, the writer analyzes and supports their
arguments with persuasive and reliable evidence,
explanation and counter-arguments. In the conclusion, the
writer reiterates his/her position on the subject of debate to
persuade readers to change their thoughts on the issue.
The following section deals with genre-based pedagogy
developed within the theoretical underpinning of SFL. As
seen from the figure below, the genre approach offers
useful instructional cycles enabling the teacher to organize
classroom sequences through three main stages:
deconstruction, joint construction and independent
construction. Throughout the three stages, the teacher
continually provides students with knowledge about
contexts: context of culture and context of situation, which
aims to help them understand the social purpose of writing
and linguistic characteristics typical of genres. At the same
time, the teacher also builds up the subject-matter
knowledge for learners through a variety of activities such
as reading newspapers and magazines. Through these
activities, the teacher helps students develop vocabulary
and background knowledge about the subject.
Figure 1. Instructional cycles of genre pedagogy (Rose, 2012)
In Stage 1, sample texts are analysed to help learners
grasp rhetorical structure of written discourse and relevant
language patterns used to achieve the social purpose of
genres. During the course of analyzing model texts, stages
and phases unfolded in the texts are made explicit to
students and guidance is given to develop learners’ deeper
understanding of meaning-making choices made by
competent authors at each stage and phase of sample texts.
Guidance is also given to lead learners to color main
linguistic features that appear in sample texts and
understand the metafunctions of language including
expressing ideas, connecting ideas, interacting with others,
and creating a well-linked text. Before the analysis of
linguistic characteristics, students are supported to increase
their knowledge about field, tenor and mode, since all three
register factors have significat impacts on language choices
being made to fulfil communicative purposes.
In Stage 2, students are guided to write a similar text
with shifted field. They are asked to negotiate meaning,
discuss and give their ideas under the guidance of the
teacher. While they are voicing their ideas, the teacher is
writing them up on the board. During collaborative writing,
the teacher helps students to select linguistic choices
appropriate to field, tenor and mode and to apply the
knowledge learned in the first stage to produce a new text.
In Stage 3, students write a similar text on their own.
Before writing, they are supported to develop knowledge
of the topic and discuss with other peers about outlining of
ideas. After that, they begin to write either individually or
in groups. Once their texts are completed, they will be
collected for feedback by the teacher.
The genre approach has been very successful in
developing writing skills for learners in many countries
around the world. However, little is known about the
impact of this pedagogy on the writing performance of
students of English for Tourism at University of Foreign
Languages - University of Da Nang. In this article, an
attempt is made to compare two argumentative texts
written at two points in time by a full-time student. One
text is written before the student partakes in a genre-based
writing program and the other is produced after the
program. The purpose of this paper is to determine the
effect of genre pedagogy on the improvement of the
discourse structure of student writing.
2. Content
2.1. Object of the study
The study examines the impact of genre pedagogy on the
improvement of argumentative text structures of second-
year students who major in English for Tourism at
Department of English for Specific Purposes - Danang
University of Foreign Language Studies. The study analyzes
the structure of students' texts produced before and after they
undertake a writing course underpinned by genre pedagogy.
In this article, the detailed analysis of only one text written
by a student from the Department - Lan will be presented.
Lan's text is chosen as a case study because she has difficulty
presenting her ideas during her course participation.
Although Lan is a student who demonstrates spoken English
fluency, intelligible pronunciation, rich vocabulary and good
grammar knowledge she has great challenges in expressing
herself in writing. She does not know how to arrange ideas
in a logical well-sequenced manner because of her illogical
and unsystematic way of thinking. She has problem seeing
relationships of ideas and categorizing them into a system
network. In addition, she usually makes inaccurate choices
of meanings and this thus results in the production of an
ambiguous or obscure text. Her text also lacks coherence and
close connection between ideas, making it difficult for the
reader to grasp the main ideas and supporting details of her
writing; to recognize the link between the preceding and
following ideas in the same paragraph and between the
previous paragraph and the following paragraph in a full
single text. It can be said that Lan is a typical case
representing many students of English for Tourism. The
students in this discipline have relatively good oral
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 61
communicative skills but have huge difficulty with the
writing skills, especially text organization, arrangement of
ideas, meaning-making choices and logical critical-thinking.
2.2. Procedures
Students enrolled in a five-week writing course with
two hours per week. The stages and phases of the course
were developed based on the above-mentioned teaching
cycle with three main stages.
Stage 1 - Deconstruction - includes the following main
phases. Phase 1, eveloping field knowledge. In this course,
the topic "The Impact of Tourism on Economy, the
Environment and Cultures" was selected. To develop
relevant topic and vocabulary knowledge, students were
guided to watch a video about the impact of tourism on local
communities in the city of Cornwall, England and then to
discuss it in groups. In their discussion, they were requested
to relate what they had seen in the video to the current
situation of tourism development in their local areas. Phase
2, analyzing sample texts. In this phase, the students were
supported to understand register variables: field, tenor and
mode of communication. Phase 3, the students were step-by-
step guided to analyze the structures of sample texts,
recognize the stages and phases of texts. They were also
required to color key language features typical of the genre.
These linguistic characteristics include norminalization (the
process of translating verbs, adjectives, adverbs, and
conjunctions into nouns) to produce abstract, formal and
informative texts, constrastive conjunctions (although,
however...) to help the author present his or her views and
counter-arguments, Tenses (Present Simple, Present Perfect)
to express temporal meanings, language used to indicate the
causal relationship of ideas as well as cohesive links to
connect ideas together as a whole.
Stage 2 - Joint Construction – the students were
directed to collaboratively write a new text responding to a
task prompt about "The Impact of Tourism on the
Environment" with the teacher. Before collaborative
writing, open-ended questions were asked to activate
students’ prior knowledge. In the following phase, the
students were gradually instructed to complete the stages
and phases of the new text, and to apply the knowledge
they had learned in the Deconstruction stage to construct a
new text. While writing, the teacher had the role of asking
guiding questions, listening to the students' ideas, and
writing their responses up on the board. The teacher also
provided qualitative feedback on their writing product and
guided the whole class to make changes to their orginal text
in order to ensure the quality of their text.
Stage 3 - Independent Construction. In this stage,
students began to write their own text on "The cultural
Impact of Tourism”. This stage includes the following
phases. Phase 1, develoing field knowledge. The students
were guided to understand reading texts related to the topic.
Phase 2, they were supported to outline a plan for their text.
Phase 3, they began to write their own essays and
submitted them. In the last phase, they received the
teacher’s written feedback for their texts. After receiving
the feedback, students were asked to revise their essays
according to comments supplied by the teacher.
2.3. Data collection
The article analyses Lan's essay collected at two points
in time: before and after her participation in a writing
course adopting genre pedagogy for the organization of
learning activities. Prior to her involvement, Lan was
assigned to a writing task asking her to express her views
on the impact of tourism on local communities. As
requested, she needs to demonstrate her position on
whether the positive impacts of tourism on the local
community are superior to the negative effects. Upon
completion of the program, Lan was asked to present her
views on the issue of whether “the environment and local
cultures are often ignored by the tourism industry”
3. Findings and Discussion
3.1. Lan’s text structure
3.1.1. Lan’s text written before her course participation
The analysis of the structure of Lan's essay (Table 1),
indicates that although the task prompt asked her to state
her opinion on whether the positive impact of tourism on
communities outweighs negative effects, she failed to do
so at the opening paragraph. Rather than writing an
analytical exposition, she decided to write a discussion text
which is to discuss the benefits and drawbacks of tourism
development. Thus, her essay did not meet rhetorical
purposes of argumentative genre.
As she did not organize her ideas according to an
analytical exposition, but in a discussion right at the
beginning, she focused on developing two main parts of a
discussion paper: advantages and disadvantages of tourism
development in the subsequent body paragraphs. In
general, she made strong attempts to explain ideas in the
body sections. However, her explanation was somewhat
problematic. In her presentation about the impact of
tourism on security and safety in the local community, she
provided an illustrative example of Hoi An case, but this
example did not clarify her point. She claimed that tourism
has generated detrimental impacts on Hoi An security, but
her example failed to do its job.
Take Hoian ancient town as an example, that place is
regarded as renowned for its old city, path, temples or some
traditional crafts. This is a main cause to bring in a lot of
problems about overpopulated or untouched city.
Lan did not make the relationship of a densely
populated city with the security and safety of the local
community clear and obvious to the reader. Therefore, the
conclusion that tourism has an impact on local security
drawn at the end of the paragraph is illogical.
In the last paragraph, Lan provided a conclusion.
However, this conclusion was not that of a discussion
paper, but of an analytical exposition. Even though she
pointed out her point of view, it was just a balanced view
without adopting her own position. In the next phase, she
wrote her suggestion:
Besides, not only headquarters but also citizens or habitants
need to put much effort into improving some drawbacks to make
our countries better than ever.
At the end of the essay, she gave her personal
62 Dinh Thanh Liem
assessment of travelling. However, the assessment was still
unclear and obscure.
Table 1. Lan’s text written before her course participation
Structure Content
Thesis,
but no
position
Accompanying with dominant development of technology, modernization or even industrialization, tourism renovation
which plays a crucial part in growth of countries exists two
specific factors: drawbacks and benefits.
Side 1:
Benefits
There is no doubt that the development of tourism which is
a breakthrough step all over the world not only generates a
lot of job opportunities, but also promotes our countries’s
cultures and customs.
Evidence
for job
opportunitites
According to statistics, the percentage of unemployed people has decreased dramatically in some recent decades
thanks to tourism industry. It means that that is the first
successful step to enhance the standard of life in some
developing or poor countries.
Explanation
Another novel good effect, exchange experience or culture is
regarded as a dominant prediction for cooperation and development of numerous countries in the world. We will
have plenty of chances to promote some attractions,
incredible views, some traditional crafts and so on which
attracts a bunch of visitors to our fabulous countries.
Side 2:
Drawbacks
It is undeniable that besides incredibly various advantages, tourism development emerges a few of bably little-known
aspects such as environment or local habitants.
Environm
ental
impact
More and more domestic people or foreigners have taken a
trip to other countries. A lot of headquarters or
environmentalists ensure that pollution is inevitible. By dint of unawareness, holiday makers from other countries who
may not obey some rules or laws relating to environment
may do harm to the health of local people more or less.
Impact on
security
Moreover, the security of communities is more and more
difficult to ensure.
Irrelavant
Example
Take Hoian ancient town as an example, that place is
regarded as a renowned for old city, path, temples or some
traditional crafts. This is a main cause to bring in a lot of
problems about an overpopulated or untouched city.
Link It has a dramatical impact on a lot of inhabitants’s safety
in terms of enhancing development of some countries.
Conclusio
n
In a nutshell, tourism development has two sides: superb
or even bad influences.
Position In my view, I am mutual between two standpoints because
of unpredictable convertion.
Proposal
Besides, not only headquarters but also citizens or habitants
need to put much effort into improving some drawbacks to
make our countries better than ever.
Evaluation What is not a responsibility, is mission of a lot of people
to go further than we have gone in tourism industry.
3.1.2. Lan’s text written after her course participation
The analysis of the post-intervention text (Table 2)
showed that Lan structured her text quite well to meet the
requirement of the writing prompt. The prompt asked her
to express an opinion on whether the local environment and
culture are often ignored by the tourism industry. In the
opening section, Lan introduced the topic of argumentation
by acknowledging the contribution of the tourism sector to
the economic development. After the introduction, she
wrote a counter-argument that denies the role of the
tourism sector in environmental protection and in the
preservation of cultural values and immediately argued
against it and adopted her position. Her position is that
tourism has made great efforts in preserving cultural values
1 Bold words in the structure column of Table 2 indicates differences in Lan’s texts written before and after her course participation.
as well as contributing to an increase in state revenues used
to pay for environmental protection.
Table 2. Lan’s text written after her course participation
Structure1 Content
Orientation Tourism is one of the most developed industries in the world that contributes to the development of an
economy.
Counter-
argument
Although the tourism sector puts a strain on the preservation of cultures and protection of the
environment,
Position
it has been making high efforts to preserve cultural
values through historical and cultural exhibits and
raise the government’s environmental revenues.
Counter-
argument and
Point 1
Some may say that the tourism industry might be the
main factor that could cause the negative changes in
values and customs, but it may not be the case.
Explanation
The tourism sectors heads towards to preserve local
traditions and cultures by some practical ways such as
organizing some historical artifacts and architecture exhibits.
Example
For example, annually, in Danang city, the tourism
industry usually holds large-scale cultural heritage exhibits for local people and tourists to discover. This
makes people more understand and aware of
importance of cultural preservation.
Link It is obvious that the tourism industry contributes to
the preservation of cultural values.
Point 2
Another contribution is that the tourism sector helps
increase the government’s revenues spent on the
protection of environmentally sensitive areas.
Counter-
argument
Some people argue that the tourism sector rarely pays
attention to these areas.
Rebuttal
In reality, this industry contributes to the
government’s revenues from park entrance fees and similar sources which can be spent on maintenance
and management of these places.
Example
For example, the funds from those sources can be used for overall conservation programs and activities
such as park ranger salaries. This leads to the
improvements of park maintenance.
Link
There is no doubt that the tourism sector makes big
efforts to enhance the protection of these sensitive
areas through revenues.
Conclusion
(Restatement
of a Position)
In conclusion, although the damages of tourism industry
have been emerging continuously, the tourism sector has
attempted to save cultures and the environment.
In the body, Lan presented her point about culture
through the use of counter-argument.
Some may say that the tourism industry might be the main
factor that could cause the negative changes in values and
customs, but it may not be the case.
While she wrote the travel industry as a major factor
making negative changes to the local cultures and customs in
the counter-argument, she immediately rejected it. Next, she
gave her explanations for her argument that the tourism
industry has implemented strategies to preserve cultural
values through cultural exhibitions. She provided an example
of tourism-related events organized in Danang and argued for
the role of these events in raising public awareness and
understanding of the importance of the maintenance of
cultural identity. In the last sentence, she reaffirmed the
contribution of tourism to the preservation of cultural values.
In the second paragraph of the body section, she
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 63
presented a point towards the role of the travel industry in
raising revenues used for the protection of the
environment. After presenting her argument, she
recognized a counter-argument and rebuttal. In the counter-
argument, she wrote:
Some people argue that the tourism sector rarely pays
attention to these areas.
In the rebuttal, she wrote:
In reality, this industry contributes to the government’s
revenues from park entrance fees and similar sources which can
be spent on maintenance and management of these places.
In the next phases, Lan provided an example of financial
sources from tourism earnings allocated for conservation
activities such as salaries for conservationists. She also
included a conclusion about significant efforts made by the
tourism industry to preserve environmentally sensitive areas.
In the iteration, she restated her position about big
attempts made to protect the environment and culture.
3.2. Discussion
Through the interpretation of Lan’s texts, it is found
that Lan develops an understanding of text structure and is
able to apply her understanding in structuring and
sequencing her post-intervention text in a more persuasive
and argumentative way.
Before undertaking the intervention program, Lan could
not make differences between different genres such as
analytical exposition and discussion and thus she structures
her text in a way that does not meet communicative
functions required by the writing task. Instead of organizing
ideas according to analytical exposition, she decides to write
a discussion. Thus, her text does not meet the requirement of
the task prompt and does not achieve communicative
purposes in the cultural and situational context.
After the intervention, Lan develops a better and deeper
understanding of the structure of written discourse as
manifested in the quality of her text. She clearly
understands that an argumentative text needs to undergo
major stages and phases. These stages and phases are to
argue for a position supported by points followed by
explanation, example and link in order to persuade the
reader. She understands that an argumentative text must
include the writer’s position in the orientation, points,
counter-arguments, rebuttal and persuasive evidence
through specific illustrations and a restatement of the
writer’s position at the end of the text.
That the student develops knowledge about text
structure is thanks to genre pedagogy. This pedagogy is
advanced based on the principle of guidance through
interaction in the context of shared experience. According
to this principle, students are step-by-step guided to
understand and apply knowledge gained through quality
interaction with the teacher and other peers to complete a
new task. At the beginning stage, the teacher is a
knowledge transmitter and simultaneously a person who
offer favorable conditions for students to absorb new
kmowledge through interactions so that both the teacher
and students share the same understanding. At the
following stage, the teacher is a guide and a participant
who collaboratively works with students to co-construct a
new text. In the last stage, once new knowledge has been
achieved and students demonstrates abilities to do similar
task on their own, the teacher will act as a supervisor and
evaluator who organizes learning activities and provides
feedback and evaluation of students’ performance. Genre
pedagogy is a balanced approach which both focuses on the
writing process and the product enabling students’ success
in the academic world.
4. Conclusion
The student shows a significant improvement in text
structure upon completion of a writing course informed by
genre pedagogy. The pedagogy is considered effective to
be adopted for the delivery of writing lessons because of
several reasons. This pedagogy is a balanced approach that
focuses on developing the writing ability for learners which
allows them to produce meaningful argumentative texts. It
is also a balanced approach in terms of the role of the
teacher and students. The pedagogy appreciates the
controlling role of the teacher in the first stage when
students need the teacher to impart new knowledge but the
teacher becomes a facilitator who offers needed help at
later stages. Once students become more independent, the
teacher begins to hand over control to students who take
full charge of completing new tasks on their own.
The genre pedagogy is regarded as useful for teaching
writing skills. However, there are some considerations that
need to be taken. In the process of analyzing structures of
sample texts, the teacher should make learners aware that
stages and phases in the samples should not be treated as a
rigid formula, but as a flexbile frame that can be altered or
changed. If the teacher intentionally sees text structure as
fixed, the teacher is reinforcing the notion of formulaic
writing and restricting creativity and originality of learners
in the presentation of their arguments. Only when attention
to the flexibility of text structures has been drawn, can the
pedagogy act as a fruitful tool enabling teachers to organize
useful classroom activities and supporting the development
of writing skills for learners.
REFERENCES
[1] Cope, B., & Kalantzis, M. (1993). The Powers of Literacy: a Genre
Approach to Teaching Writing. London: Falmer Press, 1993.
[2] Derewianka, B., & Jones, P. (2016). Teaching Language in Context.
South Melbourne, Victoria Oxford University Press, 2016.
[3] Halliday, M. A. K., & Matthiessen, C. M. I. M. (2013). Halliday's
Introduction to Functional Grammar. London: Routledge.
[4] Humphrey, S., Droga, L., & Feez, S. (2012). Grammar and
Meaning. Australia: PETAA.
[5] Hyland, K. (2003). Genre-based pedagogies: A social response to
process. Journal of Second Language Writing, 12, 17-29.
[6] Martin, J., & Rose, D. (2012). Pedagogic Discourse: Contexts of
Schooling. International Journal of Language and Communication
(38), 219-264.
(The Board of Editors received the paper on 31/03/2018, its review was completed on 26/04/2018)
64 Nguyen Ngoc Nhat Minh
THE USE OF SOCIOLINGUISTICALLY RICH PEDAGOGICAL DIALOGUES IN
TEACHING CONVERSATIONAL ENGLISH
Nguyen Ngoc Nhat Minh
University of Foreign Language Studies - The University of Danang; [email protected]
Abstract - This paper discusses the potential use of sociolinguistically rich pedagogical dialogues, i.e., textbook dialogues replayed by real speakers of English. Due to the increasing recognition for English language varieties, exposing learners to naturalistic language and language variations in real life can contribute to the effective use of textbook dialogues in communication/conversation classes. To create such dialogues, teachers can guide learners to record people performing textbook dialogues and then compare the language used in the recordings to that introduced in the textbook. When being recorded, the interlocutors rely on the given prompts, which show the speech acts of the sample dialogue, to reproduce the conversation in their own language. This technique was trialed by the researcher and the results are discussed in this paper to examine how the speakers’ identities influence their language use and how this might benefit learner language repertoire, from which conclusions are drawn regarding the use of sociolinguistically rich pedagogical dialogues. The analysis of the replayed dialogues reveals that language use is affected by the speaker’ identities including gender, age, etc., and that some language elements which do not appear in textbook dialogues, such as hedging or some spoken expressions, are actually useful for learner language repertoire.
Key words - sociologically rich dialogue; naturalistic language; language variations; identity; language repertoire
1. Introduction
Due to the increasing importance of intercultural
communication, language variations are more and more
welcomed in addition to the so-called standard English. The
language that speakers use reflects their identities and what role
they decide to take in communicative social situations. There
are cases in which English language learners may not
understand the language used in real-life communication, or
they produce the language they have learned from the textbook
dialogues in communication/conversation classes. Therefore,
it is useful to make the textbook dialogues more real with
varieties of the English language that embrace the
interlocutors’ identities when they are conversing. In this way,
textbook dialogues would become sociolinguistically rich and
therefore benefit learner active language repertoire more by
preparing them for English variations they might encounter
outside the classroom. One way of creating such
sociolinguistically rich pedagogical dialogues is to record a
textbook dialogue role-played by real speakers of English. In
implementing this technique, the speakers do not memorize the
content and turns of the conversation. Instead, they use their
own language following the given prompts to re-produce the
speech acts performed in the sample dialogue. To visualize this
technique and explore its potential benefits, the researcher
trialed it and reported the results in this paper. The trial aimed
to examine: 1/ how the interlocutors’ identities influence their
language use in the conversations; and 2/ how the speakers’
language use (influenced by their identities) can benefit learner
active language repertoire, or in other words, what we can learn
from the use of sociolinguistically rich pedagogical dialogues.
2. Literature review
Although Lammers (2005) believes that textbook
dialogues and drills serve as the primary source of acquiring
“spoken command” of the target language, he agrees that they
are “limited in variety” and seem to provide the language
which is probably much more polite than the variations we are
actually exposed to on the street. Such variations might be
language use associated with interlocutors’ identities since “in
our use of language we represent a particular identity at the
same time that we construct it” (Hall, 2003). They might also
be varieties of the English language.
Thanks to the global need of using English as a medium
of communication, we are now using English in multicultural
contexts. As a result, the Englishes used by non-native
speakers are being added to our linguistic repertoire (Gallowa,
2017). Gallowa (2017) suggests that those variations greatly
influence how English should be taught to our students,
raising questions like “What grammatical, pragmatic, and
cultural norms should they learn?” or “How can I ensure they
are prepared to use the language as a lingua franca?” Kendall
(2011) also recognizes the importance of spoken language
varieties, assuming that they constitute the sociolinguistic
richness of communication in that language. These authors’
views have inspired the attempt on making textbook dialogues
become more sociolinguistically rich. Listening to textbook
dialogues played by real English users (who can be native or
non-native speakers), learners are expected to have their
linguistic repertoire improved.
3. Methodology
3.1. Materials
The original dialogue was selected from the Student
book Solutions Intermediate (Solution series), involving 14
turn-taking moves between two speakers. The conversation
(see Table 1), which is somewhat stilted, took place
between a female speaker calling to ask the male speaker
out. To ease the data analysis and discussion, the recipient
(man) was named A as the first person to start the dialogue
by picking up the phone, and the caller was B. The
conversation was tried and recorded with three pairs of
speakers, referred as A1-B1, A2-B2, and A3-B3, in
dialogue 1, 2, and 3, respectively.
The conversation was first divided into specific social
interactions, which were then broken into smaller speech
acts as the prompts for the real interlocutors.
3.2. Participants
As Norton & Toohey (2011) proved, identity categories
including race, gender, and sexuality had interacted with
language learning and teaching, the dialogue was tried on very
different pairs of speakers to see how they would respond in
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 65
that context and the language they used in terms of the
identities they took/chose. The first pair included two non-
native speakers of English, who were close friends in real life.
They actually found the situation embarrassing about the
invitation for a date out made by the woman, but they still
played the assigned roles as in the textbook (the male student
named A1, and the female B1). However, the second pair of
speakers asked to switch the roles as the female participant did
not agree to take the initiative as person B did in the given
dialogue. As a result, A2 was female who received the
invitation and B2 was male asking her for a date. These
speakers were native speakers of English and were actually in
a relationship. The last pair was two women who were
classmates. Interestingly, the fact that they had the same
gender probably did not enable them to recognise that the
social actions on the “invitation”, in fact, suggested some
ways to ask out a person. This pair, certainly, did not have any
problems taking the roles. A3 was a non-native speaker, living
in the US before to study her first master, so she was fluent in
English, and B3 was an American student. However, these
participants, who were not interested in parties, decided to
make the invitation for a farmer market instead. The recorded
conversations of the three pairs can be found in the results and
discussion of each dialogue.
Table 1. Social interactions and speech acts observed in the original dialogue
Social actions Speech acts Turns
I. Initial greetings Greeting on the phone 1. A: Hello?
Greeting & Introduce name 2. B: Hi! It’s Grace.
II. Identifying the
caller
Expressing hesitation because you don’t remember B 3. A: Er… right. Grace?
Introducing yourself as X’ friend & Reminding that you
met each other at the cinema
4. B: Abigail’s friend. We met last Saturday, at the
cinema.
Remembering & Asking if B got your number from X 5. A: Oh, yes, I remember! Did Abigail give you my
number?
Agreeing 6. B: Yes, she did.
Accepting (this turn can be neglected) 7. A: Oh, OK.
III. Making an
invitation
Asking if A has free time 8. B: Harvey, I was wondering. Are you doing
anything tonight?
Agreeing & Wanting to know why B asked 9. A: No, not really. Why?
Asking if A would like to join a party with you/do
something together
10. B: I was wondering … do you fancy going to a
party?
IV. Asking for more
information
Agreeing & Asking information on the party 11. A: Yes, why not? Whose party is it?
Answering & Confirming if A will come 12. B: A friend of mine from school. So, you’ll come,
then?
V. Accepting Accepting/Confirming the invitation 13. A: Yes.
Showing excitement 14. B: Great.
4. Findings and Discussion
4.1. The influence of identities on language use
In terms of overall structure, the original dialogue consists
of 14 turns whereas the recorded ones were all longer. The
first pair had the same number of turns but produced more
words in each turn-taking move. The two other pairs took
many more turns (23 turns in dialogue 2, and 28 turns in
dialogue 3), both of which included longer information
clarification and conversation closing. Despite the variation in
the length of the dialogues, the speakers still performed
necessary speech acts analyzed from the textbook version.
Language variation revealed much more clearly in
regard to the correspondence between the identities taken
by the participants and the language they produced.
Dialogue 1
From Table 2, we can see that the structures and phrases
used by both interlocutors, especially A1 (the male),
clearly defined them as non-native speakers of English.
Their language, in general, is relatively controlled, i.e. each
turn exactly serves a specific speech act given in the
prompts. Also, they barely used indirect language as native
speakers often use as a tool to mitigate Face-threatening
acts (FTA), which was noticeable in many of A1’s turns.
He picked up the phone starting with “Hey, who’s
calling?”, and insisted on identifying the caller (“but who’s
calling?”) when she had not answered his question. His use
of “I don’t quite remember…”, “So what?” or “I will be
there as long as we will have fun.” can be considered very
rude and highly threatened the other’s face in an actual
situation. At the same time, this linguistic feature
highlights A1’s identity as a man with the strong directness
adopted in his language, and gives an impression that the
speaker was a young man who might have been arrogant
and self-centered. However, the way he talked might
possibly come from the fact that he knew the other speaker
quite well, therefore was relaxed and not conscious of his
language use.
Despite B1’s limited use of indirect speech in the
conversation, she performed occasional hedging, which
helps mark her femininity (“I’m calling and wonder if … I
wonder if you wanna join.”, or “…if you have time.”).
Linguistic devices used for hedging signal the speaker’ lack
of assertiveness which is found to have high correlation with
femininity and therefore characterizes women’s speech
(Lakoff, 1973). How she made the invitation with the
repetition of “if you have time” also revealed her hesitation
and lack of confidence asking the man out.
66 Nguyen Ngoc Nhat Minh
Table 2. Dialogue 1 of the pair A1 – B1
Social
actions First pair
I. Initial
greetings
1. A1: Hey, who’s calling?
2. B1: Hi. Is it Oumar speaking?
II.
Identifying
the caller
3. A1: Yes, but who’s calling?
4. B1: Hi. It’s Amber. Do you remember me?
5. A1: Amber? I don’t quite remember who is
Amber.
6. B1: We met last Sunday at the cinema. I was
with Minh. I’m her friend.
7. A1: Oh, I see, I see. Where did you get my
number from?
8. B1: From Minh.
9. A1: Okay. So what?
III. Making
an invitation
10. B1: So I’m calling and wonder if you have
time this Saturday. We’ ve got a party, and I
wonder if you wanna join.
IV. Asking
for more
information
11. A1: I have time, but what type of party are
you talking about?
12. B1: It’s a bonfire party…uhm..around the
Del Monte beach, if you have time.
V. Accepting 13. A1: Ok. Ok. I will be there as long as we will
have fun.
14. B1: That would be great.
Dialogue 2
The language produced in this conversation presented
as Table 3 was much less controlled than that of the first
one, and contained typical linguistic components by
English native speakers.
A2’s spoken English featured the identity as a female
speaker due to the excessive use of hedging language
including fillers (“oh”, “um”, ok”, “yeah”) and indirect
requests – a “superpolite” form (Lakoff, 1973). While the
male native speaker in the original conversation also
sought the information indirectly (“Did you get my number
from Abigail?”), A2, as a female, sounded friendlier and
softer when she showed hesitation “Uhm …” followed by
“Can I ask how…?”
As a native speaker, B2 produced very natural
English, using phrasal verbs (“coming up”, “pick you
up”) and currently common spoken expressions (“it’s up
in …”). Interestingly, his maleness was strongly revealed,
although not explicitly, in his saying “I thought I’d give
you a call.” Traditionally, men have often said this when
asking another out, but women normally do not. It does
not mean a woman cannot say that in a similar context
(actually, some women have said that way in this modern
time), but this expression, which seems somehow
“powerful,” supposedly bears “a male’s voice/position”.
Compared to his partner, B2 did not have many filled
pauses. Most of his utterances were quite direct and took
fewer words to perform speech acts. In addition, B2
brought his real identity to the conversation with the
phrase “have got” in turn 14. His parents, who are South
Africans, speak British English and its very similar
version – South African English, so his English was
sometimes mixed with British English even though he
was raised and grown up in the US.
Table 3. Dialogue 2 of the pair A2 – B2
Social
actions
Second pair
I. Initial
greetings
1. A2: Hello?
2. B2: Hey Lisa. This is Mike!
3. A2: Oh. Hey Mike.
4. B2: Hey, how are you?
5. A2: I’m good. How are you doing?
II.
Identifying
the caller
6. B2: Oh, you don’t remember me?
7. A2: Um…
8. B2: Oh, ok. Well, we met at the cinema
recently. And I thought I’d give you a call.
9. A2: Oh. Oh, Ok. Ok, yeah, I remember. Uhm.
Can I ask how you got my number from?
10. B2: Oh, yeah, I got your number from John at
the cinema last week.
11. A2: Oh, ok, ok, yeah. How are you?
12. B2: Oh, I’m doing well. How are you?
13. A2: Good.
III. Making
an invitation
14. B2: Uhh. This weekend we’ve got a party
coming up on Friday. Would you like to go?
IV. Asking
for more
information
15. A2: Um, what time?
16. B2: Um, it’s about 7.
17. A2: Ok. Where is it?
18. B2: Uh, it’s up in Ford Collin’s.
V. Accepting 19. A2: Oh, that sounds good.
20. B2: Alright, ah. so I’ll pick you up at 6?
21. A2: Great. Yeah, that sounds good.
Dialogue 3
Generally speaking, the speakers’ choice of grocery
shopping over party already constituted part of their
identities as female speakers, and possibly marked their
age (nearly 40) as well (see Table 4).
Table 4. Dialogue 3 of the pair A3 – B3
Social
actions
Third pair
I. Initial
greetings
1. A3: Hi. Who is it?
2. B3: Hi. This is Janelle.
II.
Identifying
the caller
3. A3: Oh.. Janelle...uhm.. Janelle? Like..are you
– like - from my Financial Administration class?
4. B3: No, remember we had Marketing together?
We were in a group with Lucy and Dash?
5. A3: Ohh.. Yeah, I remember. Yeah,
Dash..yeah, that was a great project.
6. B3: I know. And a lot of work.
7. A3: Yeah, it did, but finally, it’s over. So,
yeah, so I I remember you, Janelle, but like, I
don’t remember, like, giving you my phone
number? Did you get it from Dash?
8. B3: Well, sort of. I just got it from our group
work together, when we all exchanged numbers. I
just kept it in my phone.
9. A3: Oh, yeah, I remember at the very beginning
of the semester, yeah, we did, like exchange
numbers. And I remember that we also had like this
post-it notes and the Google documents.
III. Making
an invitation
10. B3: Yeah, all that work. Uhm, I was calling to
see if you’re free tonight.
11. A3: Uh, maybe I will have to check my
schedule. Like, you know, we have so many
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 67
things, but yeah, but I have some time, but like, I
was wondering like why, why do you ask?
12. B3: Well, I was just wondering if you wanted to
come with me to the Farmer’s market for dinner later.
IV. Asking
for more
information
13. A3: Oh, yeah, like, that would be awesome.
Like, but, where is it? Like…
14. B3: It’s downtown, on Alvarado street.
V. Accepting 15. A3: Oh..so yeah, I think that I can come. Like, but
like, then I would just want to know where will we
meet. At Alvarado, you know, it’s a very long street.
16. B3: Yeah, I was thinking we could meet at
Plumes, and, and go from there. There’s – they
have all different kinds of food, and produce, if
you need produce for the week.
17. A3: Oh, so I might want to, like, bring a
grocery bag?
18. B3: Sure.
19. A3: Do they take cash?
20. B3: They do. And there are a few ATMs
around if you don’t have cash.
21. A3: Oh, ok. So or maybe like I should even
better, you know, carry some cash with me.
22. B3: Yeah.
Not much pausing and a large amount of hedging
helped identify A3 as a native-like, female speaker.
Noticeably, her overuse of “like” in the conversation might
signal quite different aspects of the identity she aligned
herself with. Many native speakers agree that they used
“like” quite often (even on a sentence basis), especially
when they were younger, like in their 20s. “Like” was
observed one of the fillers found more common among
women and younger participants (Laserna, Seih, and
Pennebaker, 2014). This is also a lexical hedge which
shows a lack of the speaker’s assertiveness and is
frequently encountered in women’ speech (Lakeoff, 1973).
However, it is quite certain that her use of “like” does not
mark her age as “young”. Therefore, the abused “like” in
dialogue 3 might be a linguistic feature that portraits A3’s
lack of confidence and comfort in the conversation, or her
desire to sound like a native speaker of English, either of
which support her native-like fluency and “female” voice.
The language B3 used to proceed the conversation also
highlighted her femininity. She answered A3’s questions
with many details (“I just got it from our group work
together, when we all exchanged numbers. I just kept it in
my phone.” at turn 8) and even added some information not
required in the questions (“…they have all different kinds
of food, and produce, if you need produce for the week.” at
turn 16), which is “being considerate and caring” – a
quality often recognised in females. Additionally,
B3 showed her personality as a direct person giving
unambiguous information. Although she did use hedging
language (“I was just wondering…”, “I was calling to
see…”), she answered directly almost all the questions and
included sufficient directions in her answers (e.g. turn 12)
without taking many turns.
4.2. Lessons from sociolinguistically rich pedagogical
dialogues
4.2.1. From the perspective of discourse analysis
Compared to the dialogue taken from the textbook, the
recorded ones contained more use of linguistic features that
might be considered “problematic” for prescriptive
grammar. One easily identifiable category is the
appearance of filler words, or in a broader term,
non-fluency features. These features, mentioned as
“disfluencies”, are linguistic phenomena that “interrupt the
flow of speech” and barely contribute to the spoken content
(Gilquin, & De Cock, 2013). Filled pauses appeared from
time to time in the original dialogue (“Er…right” – turn 3,
“Oh, OK.” – turn 7) but occurred with high frequency in
the conversations made between real speakers. They were
used in almost every turn in dialogues 2 and 3; however,
they did not disrupt the structures of expressions produced
by the second pair (“Well, we met at the cinema recently”,
“Ok, yeah, I remember. Uhm…Can I ask how you got my
number from?”), but they seemed distracting in the last
dialogue (“Like…are you – like – from my Financial
Administration class?”, “… but like, I don’t remember,
like, giving you my phone number?”, or “maybe like I
should even better, you know, carry some cash with me.”).
These disfluency features might be erroneous if they were
abused as in dialogue 3, but is acceptable and natural as in
dialogue 2. It can be said that although they do not provide
any “propositional content” to utterances or even
sometimes distracting for listeners, they have been used
constantly in the flow of natural speech and “deliberately
for intelligibility and rhetorical affects” very often in daily
speech and therefore identified as inevitable and
indispensable (Gilquin, & De Cock, 2013).
Another group of “potential erroneous” language
behaviours is the spoken form of certain patterns used in
the recorded conversations. These look “problematic”
because they missed some sentence elements or were not
produced in accurate form. For example, at turn 15, A2
asked “What time?” instead of “What time is the party?”,
or at the turn 22, B2 said “See you soon” instead of “I’ll
see you soon.” Some other noticeable examples would be
the common use of yes-no questions in statement form with
question mark:
Dialogue 2:
“Oh, you don’t remember me?” (turn 6)
“…so I’ll pick you up at 6?” (turn 20)
Dialogue 3:
“We were in a group with Lucy and Dash?” (turn 4)
“…I might want to, like, bring a grocery bag?” (turn 17)
These uses are actually shortened question forms
known as declarative yes-no questions in casual speech,
which means the speaker’s being surprised or intention of
checking information (Cowan, 2014).
From the utterances discussed above, it should be
understood that some elements of real-life communication
which are often considered grammatically inaccurate are
acceptable in the communicative discourse. However,
there are several structures produced by the participants
that were actual errors. Dialogue 1 involving two non-
native speakers of English revealed some incorrect
linguistic forms that characterize English language learner
(ELL) problems. For example, in turn 5, A1 did not
68 Nguyen Ngoc Nhat Minh
produce the right form of reported speech (“I don’t quite
remember who is Amber”), or he used “… as long as we
will have fun” instead of “as long as we have fun” for
adverb clause form. These phenomena would require
interference and treatment in order not to be repeated
(Cowan, 2014).
4.2.2. For learner active language repertoire
Both the original dialogue and its recorded versions
have the desirable lexical sources and structures for
language learners’ productive repertoire. Those linguistic
features can fall into following categories:
Overall structure
The textbook dialogue, claimed to be authentic by the
book series authors, are still slightly unrealistic. In a real-
life situation, the call or the conversation would not just
stop at where the recipient accepts the invitation. Normally,
(s)he would seek further information of how and when to
meet like the speakers did in dialogues 2 and 3. Also,
speakers would naturally take turns to signal the
conversation closing. For example:
“Yeah, that sounds good.” → “Ok. See you soon.”
(Dialogue 2)
“Like, well, sounds like we have a plan.” → “Ok. I’ll
see you then.” (Dialogue 3)
As learners might encounter similar communicative
situations at some points in real life, they should stay aware
of the natural structure in implementing such events to
produce more native-like language and reduce awkward
interactions in the target language.
Patterns and hedging
The dialogues are about making an invitation on the
phone, so they all contain useful expressions that help
achieve the communicative purposes. The usefulness of the
linguistic devices in use comes from not only its intended
functions but also its embedded cultural element – hedging
- which is valued in Western culture to avoid potential face-
threatening acts.
The target language intended in the dialogues is the
structures to make an invitation. This language function,
involving either making a normal invitation or an invitation
for a date, takes place very often in our daily interactions
and is therefore useful learners’ effective communication.
Several structures students should be exposed to are:
“Are you doing anything tonight?” / “I was calling to
see if you’re free tonight.”
“I was wondering…”
“This weekend we’ve got (the event). Would you like to go?”
How to respond to an invitation is equally important. In
the discussed context, expressions for “accepting”
language/purpose were given:
“Yes, why not?”
“Oh, that sounds good.”
“Yeah, that works for me.”
“sounds like we have a plan.”
Besides, students should stay informed of relevant
expressions so they would be able to successfully
communicate in similar social actions, including phone
language and closing phrases. For example:
“Hello?” / “Hi. Is it (name) speaking?
“Hey/Hi. This is (name).”
“See ya.”
“Well, cool.”
Learners’ active repertoire can also be enriched given
access to real-life use of phrasal verbs and common spoken
expressions by native speakers (“I’m doing well.”, “Well,
sort of.”, “I just kept it (your number) in my phone.”,
“if you need produce for the week.”, “then” in “You’ll
come, then?”, etc.).
Last but not least, part of the hedging language that
would well benefit learner’s oral language ability is filler
words. As mentioned before, filled pauses help with the
flow of speech, when speakers need more time to think
about what to say next. Laserna, Seih, and Pennebaker
(2014) found out that conscious participants reflect high
frequency of fillers. Despite the recognised indirectness of
this type of non-fluency linguistic features, learners should
be advised against the overuse of fillers to limit distractions
for listeners.
The above-discussed findings have shown what and
how students can learn from sociolinguistically rich
pedagogical dialogues, which helps highlight their
potential benefits if used in teaching communication/
conversation classes. Teachers are therefore advised to
adopt the technique of re-playing textbook dialogues with
real English speakers in their classroom, at intermediate
level and above. This language level is recommended for
using the technique since textbook dialogues at this level
might contain more helpful communicative expressions.
When using the technique, teachers should sample the
procedure and guide their learners in creating such
sociolinguistically rich dialogues and exploring which
language elements of the recorded dialogues should be
added to their language repertoire.
5. Conclusion
Analyzing the collected data allowed a chance to
examine how the speakers’ identities were captured in the
language they used, and saw how different dimensions like
gender, age, background, personality, etc. shaped their
language choices. There were some insights into certain
linguistic features that are constantly adopted in our daily
speech. It was interesting to be informed of different
perspectives to look at and discuss the speakers’ identities
through their using “like” or “you know”. Using
sociolinguistically rich dialogues with the same social
actions performed by different groups of speakers enables
students to become aware of language variations and of
how these can fit in their current linguistic repertoire.
Specifically, they can have their active (and perhaps
passive) English develop with regard to the overall
structure of a real conversation in English and
communicative strategies involving hedging and spoken
structures. However, challenges might arise as there are
ISSN 1859-1531 - THE UNIVERSITY OF DANANG, JOURNAL OF SCIENCE AND TECHNOLOGY, NO. 6(127).2018 69
structures that students need to understand but should not
produce. From one of the recorded dialogues appeared
some language components that were not really authentic
or useful. For example, the phrases produced by speaker
A1 in dialogue 1 could potentially make the listener,
especially someone from Western culture, lose face, hence
lead to conversational breakdowns. As language teachers,
we can try to handle this challenge, need to explain and
notify our learners of these linguistic behaviors.
REFERENCES
[1] Cowan, R. (2014). The teacher’s grammar of English. Cambridge
University Press.
[2] Falla, T., & Davies, P. A. (2012). Solutions Intermediate: Student's
book with multirom pack. Oxford: Oxford University Press.
[3] Gallowa, N. (2017). Global Englishes and change in English
language teaching: Attitudes and impact. Abingdon: Routledge.
[4] Gilquin, G., & De Cock, S. (2013). Errors and disfluencies in spoken
corpora. Amsterdam: John Benjamins Publishing Company.
[5] Hall J.K. (2003). Teaching and Researching Language and Culture.
London: Pearson. (Applied Linguistics in Action Series)
[6] Kendall, T. (2011). Corpora from a sociolinguistic perspective.
Revista Brasileira de Linguística Applicada, vol. 11(2).
http://dx.doi.org/10.1590/S1984-63982011000200005
[7] Lakoff, R. (1973). Language and woman's place. Language in
Society, 2(1), 45-80.
[8] Lammers, W. P. (2005). Japanese the Manga way: An illustrated
guide to grammar and structure. Berkeley: Stone Bridge Press.
[9] Laserna, C. M., Seih, Y. T., & Pennebaker, J. W. (2014). Um... Who like says you know: Filler word use as a function of age, gender, and
personality. Journal of Language and Social Psychology, 33(3),
328–338.
[10] Norton, B., & Toohey, K., (2011). Identity, language learning, and
social change. Language Teaching, 44(4), 412-446.
(The Board of Editors received the paper on 08/5/2018, its review was completed on 11/6/2018)
