Sie sind auf Seite 1von 8

IEEE Transactions on Power Delivery, Vol. 13, No.

2, April 1998

667

A link between EMTP and MATLAB for user-defined modeling


Jean Mahseredjian' ,Gabriel Benmouyall, Xavier Lombard2, Mohammed Zouiti2, Benoit Bressac2, Luc GCrin-Lajoie3 (1)Hydro-QuCbec (IREQ) 1800 Boul. Lionel-Boulet Varennes, Qutbec, Canada J3X 1S1 (2)ElectricitC de France Direction des Etudes et Recherches 1 Ave. du GCnCral de Gaulle, 92141 Clamart Cedex (3)Hydro-Qutbec/System Planning 855 Ste-Catherine East, 18th floor MontrCal, Qutbec, Canada H2L 4P5

ABSTRACT This paper presents a programmed link between MATLAB and EMTP (Electromagnetic Transients Program). It is an interconnection where MATLAB functions can be called in from the EMTP Fortran code. The interconnection provides a user-defined modeling tool where the highlevel computation facilities of MATLAB and its powerful Toolboxes can be advantageously used in the general network simulator context of the EMTP.

Keywords: EMTP, MATLAB, user-defined models, interface, software engineering, relay models, shunt compensation

1. INTRODUCTION
There are several methods for implementing user-defined modeling interfaces for the EMTP (Electromagnetic Transients Program) [l]. These interfaces fall in two categories: internally programmed or externally connected. An internally programmed interface (internal interface) is a pre-programmed modeling language or a set of function blocks inserted with other parts of the complete program code. This is the case of the modeling functionalities currently available within the EMTP TACS (Transient analysis of control systems) module. It has limited resources that easily saturate with the increasing complexity of a user-defined model. One possible step for improving this type of interface is to program a full fledged programming language decoder [2]. The syntax of such a language can be inspired from existing languages such as Fortran or Pascal. This is close to programming a compiler and may constitute a tremendous task: the needed sophistication and high-level constructs may create a decoder requiring a programming effort comparable to the EMTP code itself. Another approach is simply to allow linking the EMTP code with externally compiled object codes. This constitutes an external interface where the external source code (Fortran

or C) must be created and connected by the user. Hooks into the TACS code for external modules are provided in the latest release (V3) of EMTP. The need to use a compiler and a lowlevel programming language, in addition to understanding interfacing details, will inherently restrict the usage of such an interface. To eliminate the compiler requirement it is necessary to use an external package that allows a direct run-time access to its computational engine. MATLAB [3] allows such an access in addition to providing a high-level model creation environment. This paper presents the implementation of a programmed link between MATLAB and EMTP. The objective is to provide the EMTP user a user-defined modeling tool based on the high-level environment of MATLAB. The user interaction with the interfacing protocol is minimized by allowing model function names created in MATLAB to be directly indicated in a standard EMTP data file. The programming facilities of MATLAB and its numerous built-in functions and Toolboxes are now widely used for advanced computations, modeling and even complete power system software [4] engineering. With the new interconnection presented in this paper, the behavior of individual models or systems created in MATLAB can be now studied within the general network simulator frame of the EMTP. Although the presented interconnection does not require compiler usage, the MATLAB C code generation Toolbox can be used to create compiled models for increasing computational speed or for creating portable model libraries. The paper starts with the presentation of the interconnection method and follows with illustrative examples which offer only a first look at the increasing number of possibilities available for creating models and inventing new structures connected to networks simulated in the EMTP.

2. THE INTERCONNECTION METHOD


PE-934-PWRD-0-05-1997 A paper recommended and approved by the IEEE Transmission and Distribution Committee of the IEEE Power Engineering Society for publication in the IEEE Transactions on Power Delivery. Manuscript submitted December 31, 1996; made available for printing May 15, 1997.

Both MATLAB and EMTP are currently available on popular computer platforms for engineering applications. This is why the developments presented in this paper remain almost automatically portable. The commonly used methods for connecting external programs to the EMTP without using a compiler rely on communications through files and interruptible memory channels. These methods will usually lack generality and portability or require excessive computer time. A much more efficient solution is achieved if the external program provides a library of

0885-8977/98/$10.00 0 1997 IEEE

668
Fortran or C routines for direct access to its internal computations engine. This is the case of MATLAB. 2.a The interconnection structure The link structure chosen in this paper is illustrated in Fig. 1. MATLAB provides a library of Fortran and C routines [SI to enable external run-time access to its computational engine. The engine control call is used to start and stop the computational engine from EMTP as a background process. All other EMTP calls are sequentially organized for data exchange with user-defined functions Fl to F , . These M-files (MATLAB function files) can call any other MATLAB based or accessible functions and can exchange data through global variable declarations. It is allowed to start more than one engine by remotely accessing separate computers on a network. Which means that user-defined functions can be grouped and executed on separate computers. If the network equations on the EMTP side are decoupled by subnetworks, then it is also feasible to parallelize engine calls if supported by the computer architecture.

2.b The EMTP internal connections


External models must be correctly interfaced with arbitrary networks assembled on the EMTP side. Two modular communication methods can be combined or used separately. The first method is based on the EMTP TACS module. Hooks into the TACS code are already available in the latest release (V3) of EMTP. As illustrated in Fig. 2 these are basically black boxes that can have any number of inputs and outputs. They have been simply adapted here to become specialized M-file (a total of p T-type functions) calling black boxes.

Figure 2: Connecting to EMTP from TACS. The black box inputs are sensors that can gather data anywhere in TACS diagrams and EMTP networks. The black box outputs are activators that can connect to an EMTP network through controlled sources and controllable switches. The activators can also return generic data to TACS diagrams. Connections between black boxes are allowed and TACS is capable of ordering its equations accordingly. The main advantage in connecting M-files to TACS is the generality in the transmitted and received data. The disadvantages are the one time-step delay between EMTP and TACS solutions and possible internal delays in TACS diagrams. The second way into the EMTP is through the compensation based solution of nonlinear elements. This type of interconnection is shown in Fig. 3. Separate compensation based M-files (a total of k C-type functions) are connected to individual EMTP subnetworks. Subnetworks are decoupled sets of network equations automatically created by the presence of transmission line delays or electrically disconnected networks. For a given subnetwork i it is necessary to solve a nonlinear set of equations:
Vth, - Rth,14L = vQ,

Figure 1: The EMTP-MATLAB interconnection: main structure The functions Fl to F, are simply identified in the EMTP data file as external models connected between an arbitrary set of nodes. In addition to identification, functions have different lists of input and output arguments. Although several function calling templates can be created and stored in the EMTP, a simple function such as outarg=funj(inurg) is applicable to most modeling cases where inarg is a vector of input data from the EMTP network to the model and outarg is a vector of output data from the model to the EMTF. Both of these vectors can have any number of cells within the fixed but adjustable dimensions coded on the EMTP side. The calling procedure of f u n j (arbitrary name) from the EMTP code requires only a few simple steps performed using the MATLAB external interface library subroutines. 1. create inarg in the workspace (full, sparse and complex matrices are allowed) 2. transfer inarg into the workspace and call f u n j 3. get outarg from the workspace It is noticed that the first step is needed only once in the EMTP time-step loop and it is not required to create outurg in the workspace before retrieving its contents.

(1)

where Vth, and Rth, are the subnetwork Thevenin voltage vector and resistance matrix respectively, IQ, and V4, are
the current and voltage Yectors for the nonlinear elements

located in subnetwork i. This system is solved using Newton iterations with the nonlinear relation G(VQ,,141) = 0 . Two interconnection methods have been programmed. A simple first method is to maintain the iterative procedure in the EMTP. In which case a given function Fc will receive Vq, at J each iteration and respond with a Norton equivalent

669
( qNJ , IoNi) where the resistance is the differential of the nonlinear function. Another method is to transmit Vth, and C Rthi to Fi which is now set to apply its own iterative solution method and return only the solution vector Iol . This atmroach is much more efficient by requiring only one call to C Fi at each solution time-point. It is also more general, since a multiport and intemally couuled device or even an entire circuit can be assembled in FF with user-defined solution methods.

in the EMTP. v,. is a reference voltage, p j and q j are J model coefficients. A linear representation is used below a minimum voltage V,in . .
J

EMTP

IEMTP

MATLAB

M-file C-type model: pzno Figure 4: Single-line diagram for test Case 1 The contents of p2no.m are given in Appendix A. It is an illustration of model creation simplicity with the high-level syntax of M-files. This compensation based model is set to receive VP and RTH which are the Thevenin voltage vector and resistance matrix respectively calculated in the EMTP for the subnetwork connected at nodes 7,8, and 9. Each nonlinear branch function has a different coefficient q j which in a practical case represents parameter drift. A simulation result is shown in Fig. 5. All simulation waveforms are perfectly identical to those found with the same nonlinear branches solved entirely in the EMTP. An interesting observation is made when the mean number of iterations required for the entire simulation of 25 ms is found to be 2.5 compared to 4.2 for the same branches represented in the EMTP. This is explained by the slightly different nonlinear solution method employed in pzno. The standard EMTP method finds the Jacobian of equation (1) after inverting R t h and replacing 14 by the required voltage dependent equations ((2)). The approach taken in pzno starts by calculating the Norton equivalents for all branches and uses the inverse of R t h only after adding it to the diagonal matrix of Norton resistors. It achieves an improved robustness for the solved system and requires less iterations for the same convergence tolerance. This observation suggests another important application for the interconnection proposed in this paper: in addition to user-defined modeling it can be used for testing algorithmic ideas. Case 1 requires 30 times the computer time of the equivalent case running entirely in the EMTP. This apparent performance deterioration is not sufficient to cancel the major new possibilities offered by the interconnection. Moreover, it is feasible to compile pzno for improving computational speed. Two possibilities are offered [7]: creating a Mex-file or a stand-alone C code. The Mex-file is a compiled version of the M-file that can be only used in the MATLAB workspace. It is created in a single command step and for pzno it will lower the previous ratio from 30 to 25. The C code of pzno is also generated in a single command step and an additional step is

Figure 3: Connecting to EMTP through nonlinear elements. It is apparent that the solution of (1) can easily accommodate linear models as a particular case of a nonlinear equation. The C-type functions of Fig. 3 benefit from a simultaneous solution with the EMTP network equations. The disadvantages of a C-type function over a T-type function (Fig. 2) are the restrictions on the type of accessed network data, but since both C-type and T-type functions can coexist and exchange simulation variables in a data case, they can be jointly used to alleviate limitations. The complete programming of the interconnection is designed to remain compatible with any EMTP data case and will be released in a future official program version.

3. TESTCASES

This section presents application examples. The first example is designed to illustrate the main numerical aspects of the interconnection and establish a performance comparison with existing EMTP based models.

The schematic diagram of this case is shown in Fig. 4. The network assembled on the EMTP side is a 230 kV network (can be found in [ 6 ] )with two 3-phase distributed parameter line models connected to network equivalents. The untransposed line is energized at 2 ms. The MATLAB M-file pzno models a 3-phase nonlinear branch identified and connected to nodes 7, 8 and 9 in the EMTP data file. Each phase j has a nonlinear branch relation given by:

This is similar to an arrester model that can be also simulated

670

needed for linking it with the EMTP object library. Compiled code usage reduces the ratio of 30 to almost 1. i
Pa 100

n
Y
0

gration and simple transfer functions. As a consequence, these building blocks can be identified and modeled using the catalogue of SIMULINK functions that belong to exactly the same nature. Mho type phase a-to-ground detection in the THR relay follows the conventional design principles of having two voltages: operation and polarization voltages (see Fig. 6).

,
l o t (ms) '5

+&@"A \L Wk
20
25
signal
n

polanzation voltage

+V,,

Figure 5: Phase a current, nonlinear element of Fig. 4 The possibility of automatic C code generation is a powerful method for creating permanently coded model libraries in the EMTP.
3.b Case 2: a SIMULINK based relay model

(I, + K i r ) Z r
, A

Figure 6: The principle of mho phase comparison mho detection.

Two different approaches are commonly used for testing relays in the EMTP. The first approach is a playback approach where waveforms are generated in the EMTP and submitted to an external relay model [8] possibly implemented in MATLAB. A major limitation of such methods is the inability to study the transient response of the protection system in conjunction with the transient response of the power system. The second approach enables feedback between the EMTP and the relay model. It can be based on a direct implementation [9] of relay models in the EMTP through available module creation resources. An EMTP include module is based on masking interconnected branch models and TACS functions and may carry a preprocessing capability. The disadvantage of this approach is that the created relay model library is limited and any needed enhancemeints or creation of new models require in-depth analysis of the existing module files and understanding of the data file based module creation language. This section innovates in presenting a static relay model created in the graphical user interface of SIMULINK and directly connected to a large EMTP network. SIMULINK [ 101 is the dynamic system simulation toolbox of MATLAB. SIMULINK facilitates model definition by using block diagram windows where models are created and edited principally by mouse driven commands. The powerful graphical user interface of SIMULINK is not its only advantage over TACS. SIMULINK has a larger library of linear functions, a
library of discrete functions and allows simple creation of

As shown in Fig. 6 the conventional operation voltage is given by (hatted variables are time functions):
n

G o , = (i,

+ K i r ) Z r - Ga

(3)

where Z r is the reach of the mho detector corresponding to a percentage (usually acound 85%) of the direct sequence impedance of the line. ir is the residual current given by:

user-defined model libraries. The relay model can be tested separately in SIMULINK and then directly connected to the EMTP. It is proposed to model the phase a-to-ground fault mho detector of the THR [ l l ] line relay. This relay belongs to the category of transistor based designs using early versions of operational amplifiers for the realization of summation, inte-

(4) i, = ia + ib + i, and K is a compensating factor related to the line zerosequence ZLO and positive-sequence Z L ~ impedances: K = (ZLO - Z , 1 ) / ( 3 Z ~ l ) . The polarization voltage comprising a cross-polarization term from phase c is given as: GPO, = G a - ?,L60 (5) These two voltages are computed in the time-domain and supplied to an integrating type phase comparator. Detection of a fault is performed when the phase angle - between the two voltages falls below 90". The phase detector is realized by the combination of an exclusive-OR logic port and an integrator followed by a voltage comparator. Of the phase a-to-ground mho detecThe tor introduced in SIMULINK is shown in Fig. 7. All transfer functions have been identified by detailed analysis of the electronic circuitry. The three phase currents pass through a transformer known as a transactor, the purpose Of which is to impress on the currents' the phase angle of positive sequence impedance of the line to be protected Three adjustments of this phase angle are offered in the THR by means of a resisat the secondary Of the transactor: 450 60" and 7 5 0 . This means that the choice of the phase angle of Z , is limited to these three values. The 75" case is shown in Fig. 7.
A n A A

673

--Vc 1 6 0 deal Transfer Fcn7 Transfer Fcn3 Transfer Fcn6 Transfer Fcn9 Transfer Fcn4

33e3*1 e-6s 33e3*1e-6s+l Transfer Fcn2

I +

1/400
l/(p1"240)~2s~+I 2/(pi"L40)s+l Transfer FcnlO

3 876s s+1412

I
Y2

L+
I

INTEG-THRI

Yl

Constant

T!ansfer Fcr!S

Figure 7: SIMULINK model of a phase a-to-ground mho type detector: thr.m The relay model of Fig. 7 is directly inserted in an EMTP network data file through a TACS M-file black box (T-type, see Fig. 2) named reluythr. The simulated network is the Hydro-QuCbec 735 kV series compensated network (see diagram in [12]). The relay is protecting a line located between buses MTG7 (Montagnais) and CHU7 (Churchill). The series compensation capacitor is shunted in this simulation since the THR has no voltage memory and cannot function in a series compensated line. The relay connections (see Fig. 7) to the transmission line are shown in Fig. 8. It is receiving the voltage vector v, ( v r =[va
Vb

vc]T, Vb is not used) and

the current vector i , ( i , =[ia i b i,IT). The trip signal y l is transmitted to the initially closed breakers Dk and Dm. The simulated fault is a phase a-to-ground fault occurring at 55 ms and located in the transmission line at 20% of length from MTG7. U7

and y3 are sent to EMTP for studying the internal behavior of the relay and y l is used as a breaker switch activator. Initial conditions for all state-variables are entered in x0, options is the vector of integration parameters and ut is the matrix of inputs defined over the interval [ODt] by corresponding EMTP variables. Since the built-in integration methods are incapable of maintaining history for repetitive external calls, it is chosen here to call thr for the fixed interval [0 Dt] while continuously saving and updating x0. It must be noticed that the integration time-step Dt used for thr can be a multiple of the main network integration time-step in the EMTP. Simulation waveform monitoring is simply achieved by inserting MATLAB plot statements in relaythr. Although the computer time of relaythr remains acceptable even with waveform monitoring, ultimate performance is achieved by generating and linking (with the EMTP object library) the C code of thr: The added overhead to the entire network simulation is negligible. The C code of a SIMULINK diagram can be generated through the Real-Time Workshop Toolbox [13]

Figure 8: Single-line diagram for a THR relay simulation It can be seen from simulation results shown in Fig. 9, that the relay output y l becomes positive at = 65 ms and resets when fault conditions are eliminated (all 6 breakers are opened). The reception of y l at Dm is delayed by 8 ms. The M-file relaythr is programmed to call the block diagram window thr of Fig. 7. Among provided integration methods, Gear is the most appropriate for this type of problem. The syntax of the calling line from relaythr is: [t,x,y]=gear('thr: [O Dt], xO,options,ut); where the left-hand arguments are the integration time-points t, the state trajectories x and the output trajectories y ( y = [yl y2 y3] in thr) returned to EMTP. The signals y2

- 4 0

50

60

70

80 t (ms)"

100

Figure 9: Breaker currents and THR trip signal y l

672

3.c Case 3: arc-furnace with shunt compensation The simplified single-line diagram of this case is shown in Fig. 10. It is a 220 kV (50 Hz) network feeding an arc-furnace. All network parameters are extracted from an actual operating network. A shunt compensator is used to reduce the flicker caused by arc length variation. The conventional network is assembled in a standard EMTP data case.

p='iT+P+ij A similar equation is written for q:

q=q+Ej+q

(1 1)

After using a low-pass filter to eliminate the dc components is and 4 , the compensator current setting ic is found from equations (8)-(11):

M-file T-type model: com M-file C-type model: arc Figure 10: Single-line diagram for Case 3 The applied arc equation IS given by a modified version of the Mayr equation, the Cassie-Mayr formula: drarc = rarc [I ;; v dt 8
2

( 6 )

where rare is the arc resistance (ohms), 8 is the arc time constant (s), vac is the arc voltage and vo is the static arc voltage. Flicker phenomenon is represented by imposing a 10 Hz sinusoidal variation on vo . It must be noticed that this version of the Mayr formula is not available in the new library of arc models [12] of EMTP-V3 and access to individual model parameters during the simulation loop is not available. The shunt compensator measures the network voltages and currents given ib by

The M-file mc.m contents are comparable to pzn0.m (see Appendix A), but it must iterate with its own model equation (6). The trapezoidal method is used to integrate equation (6) during iterations. An additional statement is needed for perturbing vo with a sinusoidal function. The nonlinear nature of (6) requires a simultaneous solution with network equations provided by the C-type interconnection, but the compensator model compens can be accommodated with a time-step delay and connected through TACS. The compesator model creation benefits directly from matrix manipulation capabilities available in MATLAB. A partial listing of compenxm is given in Appendix B. It can be seen from Fig. 11 and 12 that when the compensator is turned on at 200 ms it is able to cancel line current fluctuation.
601
l

v, =[va

Vb

vc]*

and into

"EE
a
20

i,

= [i,

i C l T respectively.

Transformation
7

a - p coordinates results in:

100 150 200 250 300 350 400 450 t (ms) Figure 11: Imaginary power q, Case 3
0

0 -1

50

According to the instantaneous power theory [141, the transmitted real and imaginary powers (p and q) are found from:

50

100

150 200

250

300 350 400 450

t (ms> Figure 12: Line current, Case 3 CONCLUSIONS This paper has contributed a new simulation tool for the EMTP by interconnecting MATLAB and EMTP. It is now

If the power p is decomposed into a dc component F , a low frequency component 6 and a high frequency component 6 then:

673 possible to share the high level modeling resources of MATLAB with the traditional power system simulation qualifications of the EMTP. Three examples have been used to illustrate provided new possibilities and benefits for practical applications. These examples offer only a preliminary exploration and various applications in such fields as statistical analysis, signal analysis, dynamic system simulation and symbolic computation should follow. ACKNOWLEDGMENTS The authors would like to thank Alain XCmard from ElectricitC de France for initiating this project and contributing to its success. The authors would also like to thank JosC Lopez Garcia from IREQ (Hydro-QuCbec) for its contributions in automatic C code generation aspects. APPENDIX A The M-file pzn0.m
function [IPHI] = pzno (VP, RTH) global icall; %preserved between calls global p q K VREF VMIN VOLD; if isempty(ical1) % this is for the first call p = [1500 1500 15001; %initialization q(l) = 26; q(2)=20; q(3)=10; K = [.5 .5 . 5 1 ; VREF = 1206600 206600 2066001; VMIN = K.*VREF; VOLD = [O; 0 ; O;]; icall=O; end
,
% guess is VOLD or VTH the first time VTH=VP(1:3); VPHI=(VOLD==O).*VTH+VOLD;
%

global icall Tc; %preserved between calls if isempty(ical1) %initialisation icall=l; %Tc is the transformation matrix Tc=sqrt(2/3)*[1 -1/2 -1/2 0 sqrt(3)/2 -sqrt(3)/2 1 ; end Valbet=Tc* [din( 4 ) din(5) din(6)1 ; %Valpha and Vbeta vector
%
% pq powers pq=[ Valbet(1) Valbet (2) -Valbet (2) Valbet (1) I *Tc*[din(1) din(2) din(3)I;

.....
pc and qc are the powers to compensate dout is the vector of compensation currents dout=-Tc*inv([ Valbet(1) Valbet(2) -Valbet(2) Valbet (1) ] ) * [pc qcl;
% %

....

REFERENCES Electric Power Research Institute, EMTP Development Coordination Group, EPRI EL-641 2-L: Electromagnetic Transients Program Rule Book, Version 2 L. Dub6 and I. Bonfanti: Models: A new simulation tool in the EMTP. European Trans. on Electrical Power Eng, (ETEP), Vol. 2, No. I , pp. 45-50, JanFeb. 1992 MATLAB, High-Performance Numeric Computation and Visualization Software. MATLAB Users guide, August 1992. The Mathworks, Inc. J. Mahseredjian and E Alvarado: Creating an Electromagnetic Transients Program in MATLAB: MatEMTP. IEEE Winter meeting, January 21-25, 1996, Baltimore, paper 96 WM 0989-4 PWRD MATLAB, High-Performance Numeric Computation and Visualization Software. External Interface Guide. January 1993. The Mathworks, Inc. E Alvarado: EMTP Workbook 11. University of Wisconsin at Madison. EL4651, Volume 2, June 1989 MATLAB, Compiler Users Guide. September 1995, The Mathworks, Inc. R. E. Wilson & J. M. Nordstrom: EMTP transient modeling of a distance relay and a comparison with EMTP laboratory testing. IEEE Trans. on Power Delivery, Vol. 8, No. 3, July 1993 A. S. Chaudhary, K-S. Tam and A. G. Phadke: Protection system representation in the Electromagnetic Transients Program. IEEE Trans. on Power Delivery, Vol. 9, No. 2, April 1994 [lO]Simulink Users Guide. April 1993, The Mathworks, Inc. [ 111Reyrolle Protection: Operating recommendations, THR

iter=O; %counter for iterations while iter < 100 VOLD=VPHI; for i=1:3 %call individual models if VPHI(i)>=O. sig =l;, else sig =-l;, end if abs(VPHI(i)) > VMIN(i), * ( (abs(VPHI( 1 )) /VREF (i)) ^q(i)) ; Ip=sig*p(i) RNMl=q(i)*p(i)/VREF(i)*(abs(VPHI(i)). .. /VREF(i))^(q(i)-1); RN(i) = l/RNMl; IN(i) =Ip-VPHI(i) /RN(i); else RN(i) =VMIN(i) /p(i)*(K(i)^q(i))); IN(i)=O; end end IPHIj=inv((RTH+diag(RN)))*(VTH+(RN.*IN) ); VPHI=VTH-RTH*IPHIj; if abs(V0LD-VPHI) < le-08 IPHI(1:3,1)=IPHIj (1:3); IPHI(4)=0; IPHI(5)=iter; return; %convergence found end iter=iter+l; end
%

=iter; IPHI (1:3,1) =IPHIj (1:3); IPHI (5) IPHI(4)=-1; %the no convergence signal

APPENDIX B
of the M-file commwsa function [dout] = compens(din) % B l n ( 4 to 6 ) : measured voltages in a, b and c
%

din(1 to 3 )

measured currents in a, b, and c

674 distance protection. February 1982. Electronics Ltd. [12]Q. Bui-Van, B. Khodabakhchian, M. Landry, J. Mahseredjian and J. Mainville: Performance of series-compensated circuit breakers under delayed current-zero conditions. IEEEPES Summer Meeting, July 28 - August I, 1996, in Denver, Colorado. Paper 96 SM 405-1 PWRD [ 13lReal-Time Workshop. For use with SIMULINK. May 1994, The Mathworks, Inc. [ 14lH. Akagi, Y. Kanazawa, A. Nabae: Instantaneous reactive power compensators comprising switching devices without energy storage components. IEEE Trans. Ind. Appl. IA-20, 1984, pp. 625-630 BIOGRAPHIES Jean Mahseredjian (M) received the B.A.Sc., M.A.Sc. and Ph.D. in Electrical Engineering from Ecole Polytechnique de MontrCal (Canada) in 1982, 1985 and 1990 respectively. At present he is a researcher at Institut de Recherche dHydroQuCbec and an associate-professor at Ecole Polytechnique de MontrCal. Gabriel Benmouyal (M) received his B.A.Sc. in Electrical Engineering and his M.A.Sc. in Control Engineering from Ecole Polytechnique de MontrCal (Canada) in 1968 and 1970 respectively. In 1969 he Joined Hydro-Qu6bec as an instrumentation and control specialist. Since 1978, he has been working at IREQ where his main field of activities is the application of microprocessors and digital techniques to substation and generating station control and protection systems. Xavier Lombard (M) graduated from Ecole Nationale SupCrieure dIngCnieurs Electriclens de Grenoble (France) in 1991 and he has been with ElectricitC de France since that date. He has spent 15 months at IREQ (Hydro-QuCbec) for a research project on harmonic initialization in the EMTP. His current research activities are concentrated in the design and modeling of control systems for advanced power conversion Mohammed Zouiti graduated from Ecole Nationale SupCrieure dElectricitC et de MCcanique de Nancy (France) in 1995, he is currently a Ph.D. student i n INPL/Nancy. He is currently involved in research activities related to new electronic device flicker mitigation. Benoit Bressac graduated from the Ecole Centrale de Lille (France) and joined the ElectricitC de France research division in 1996 where he is currently involved in research activities related to numerical methods for electromagnetic
transients simulation.

phenomena and control optimization.

Luc GCrin-Lajoie (M) received his B.A.Sc. in Electrical Engineering from Ecole Polytechnique de MontrCal in 1982. In 1982 he joined the Control and Protection department of Hydro-QuCbec in the System Planning group. His responsibilities include analytic studies related to the performance of static var compensators on the bulk transmission system, as well as stability, protection specifications, electromagnetic

Das könnte Ihnen auch gefallen