International onferene on Power Systems Transients IPST in New Orleans, US omparative Study etween Power System lokset and PS/EMT for Transient nalysis of ustom Power evies ased on Voltage Soure onverter W. reitas, and. Morelato ept. of Eletrial Energy Systems, State University of ampinas, ampinas, Sao Paulo,.P. 6, 8-97, razil (e-il: walmir@ieee.org) bstrat This paper presents a omparative study between two ommerial programs onsidering transient analysis of ustom power devies based on voltage soure onverters. The programs investigated were the Power System lokset for use with Matlab/Simulink, whih employs state-variable analysis, and PS/EMT, whih is based on nodal analysis. The objetive is to determine the in differenes between them onsidering omputation time, easiness of implementation of the neessary models, evaluation of the existent libraries and aurateness of results. The ustom power devies analyzed were the STTOM and the VR. In all studies presented, suh devies were simulated by using detailed models, i.e. the swithing elements IGTs/diodes and the PWM signal gener were expliitly represented. In the end of work, the in advantages and disadvantages of eah program are argued. Keywords ustom power devies, STTOM, VR, Power System lokset, PS/EMT. I. INTROUTION Reently, various power eletroni devies have been proposed espeially to be applied to medium voltage networks, generally named ustom Power [], []. mong these new devies, speial attention has been given to the family based on the voltage soure onverter tehnology due to several attrative features, suh as faster response, output little influened by network variables and possibility of utilization together with energy storage devies, allowing ative and reative power ompensation simultaneously. Two devies belong to this kind of equipment are the STTOM (istribution Stati Synhronous ompens) and the VR (ynami Voltage Restorer). Generally, suh devies have been used to improve power quality and reliability aspets. Whatever the appliation being onsidered, it is neessary to arry out eletrogneti studies to predit the dynami behavior and to projet suitably these devies. Nowadays, these transient studies are usually aomplished through digital simulation. In this ontext, various approahes have been developed and implemented for the formulation and solution of the network equations. roadly, suh approahes an be lassified on methods based on state-variable analysis or nodal analysis. Power System lokset (PS) for use with Matlab/Simulink employs state-variable analysis [], whereas, PS/EMT is based on analysis nodal [4]. Thus, in this work, suh tools were tested during transient analysis of the previously mentioned ustom power devies. It is important to delare that all models and ases investigated in the next setions are the same presented in [5]. Suh fat beome easier to validate the results obtained here. II. USTOM POWER EVIES ustom power onept has been proposed to ensure high quality of power supply in distribution networks using power eletronis devies [], []. dditionally, various ustom power devies are based on the voltage soure onverter tehnology [6], [7]. Thus, two voltage soure onverter-based devies were investigated in this work. The hosen devies were the STTOM and the VR. In distribution voltage level, usually, the employed swithing element is the IGT (Integrated Gate ipolar Transistor), due to its lower swithing losses and redued size. Moreover, the onverter rating employed in these devies is relatively low. Hene, the output voltage ontrol an be exeuted through PWM (Pulse Width Modulation) swithing pattern, reduing the low order harmoni generation. urthermore, here, the onverters are indiretly ontrolled, i.e. only the output voltage angle is ontrolled and the gnitude reins proportional to the d voltage [],[6].. STTOM STTOM (istribution Stati Synhronous ompens), whih is shetially depited in ig., onsists of a voltage soure onverter onneted in shunt to the distribution network through a oupling transformer []- []. Suh onfiguration allows the devie to absorb or generate ontrollable ative and reative power. The STTOM has been utilized inly for regulation of voltage, orretion of power fator and elimination of urrent harmonis [], [5]. Vi V d Z tr Z ij onverte V j ig STTOM struture.
International onferene on Power Systems Transients IPST in New Orleans, US Here, suh devie is employed to provide ontinuous voltage regulation using an indiretly ontrolled onverter [6]. The ontroller is shown in ig.. The ontroller input is an error signal obtained from the referene voltage and the value rms of the terminal voltage measured. Suh error is proessed by a PI ontroller and the output is the angle δ, whih is provided to the PWM signal gener. This ontroller is the same employed in [5]. It is important to note that in this ase, indiretly ontrolled onverter, there is ative and reative power exhange with the network simultaneously [], [5]-[7]. V set V PI T ig Indiret ontroller.. VR VR (ynamial Voltage Restorer) also onsists of a voltage soure onverter onneted to the distribution network through a transformer. On the ontrary to the STTOM, the transformer is onneted in series with the distribution line. In general, the VR has been employed to protet ritial and sensitive loads against short duration voltage dips and swells as well as to eliminate voltage harmonis [], [5]. The ontroller struture employed here is the same presented in ig.. The objetive is to avoid voltage sags during short-iruits in the network. Moreover, it is onsidered that the VR ats only during fault period, on the ontrary, it is onsidered by-passed. V i Z tr V d V j onverter δ. Power System lokset Power System lokset for use with Matlab/Simulink is based on state-variable analysis and employs either variable or fixed integration-step algorithms []. In this software, the dynamis of the linear part of the eletrial iruit are represented by ontinuous or disretized timedoin state-spae equations. dditionally, the non-linear part of eletrial iruit is solved separately using predefined models, and ombined with the solution of the linear part. In this simulation pakage there are various eletrial system apparatus, eletrial hines and power eletronis omponents. urthermore, suh tool has a graphial interfae very friendly. The version. was adopted in this work.. PS/EMT PS/EMT employs the well know and established nodal analysis together with trapezoidal integration rule with fixed-step algorithms [4]. The version., whih is employed in this work, is written using ortran 9, it has allowed a speed-up of the omputing time omparing to older versions. Suh software present various eletrial system apparatus, eletrial hines and power eletronis omponents. The graphial interfae is also very well developed. IV. OMPUTER SIMULTIONS. STTOM The test system employed to arry out the simulations onerning the STTOM atuation is shown in ig. 4, whih is the same system presented in [5]. Suh system is omposed by a kv, 5 Hz transmission system, represented by a Thévenin equivalent, feeding a distribution network through an -winding transformer onneted in Y/Y/Y, // kv. To verify the working of a STTOM, a variable load is onneted at bus and a three-phase apaitor bank at bus. uring the simulation, in the period from to 6 ms, the swith is losed and from 9 to ms the swith is losed, whih rein losed until of the end of the simulation V s Z s ig VR struture. S S III. ELETROMGNETI TRNSIENT NLYSIS Several eletrogneti transient analysis omputational tools have been developed during the last deades. onsequently, there are various approahes for the formulation and solution of the network equations. In a simplified way, suh approahes an be divided in state-variable analysis and nodal analysis. ased on suh fat, two programs are analyzed here. Power System lokset for use with Matlab, whih employs state-variable analysis, and PS/EMT, whih is based on nodal analysis. apaitor (Y) vd ig 4 One-line diagram of the test system. The implementation of this system using PS and PS/EMT are shown in ig. 5 and 6 respetively. The in omponents are desribed to follow. onverter L L
International onferene on Power Systems Transients IPST in New Orleans, US rk STTOM Swith. rk powergui isrete system Ts=e-5 9 kv 75. rk - pulses z N kv Swith. rk a b a b MV // kv Measures Swith. rk Load Load Vinv_ref PulsesSignal(s) isrete PWM Gener (b) STTOM. Vinv f PI Referene onverter Voltage Vinv _ref 4 ontrol System ab. Vref () indiret ontroller. (a) test system. pi/8 si n degrees -> rad -*(pi/8) si n Vinvf :4 *pi*5 *(pi/8) si n Gain (d) phase-modulation of the ontrol angle δ.. ig 5 STTOM and assoiated ontrollers implemented in PS. RK reaker losed@t rk Gn Gn Gn Gp Gp Gn Gn 75. PWM Signals Gn Gn Gp Gn ontrol System Gp Gp Gp Gp Gp rk reaker Open@t rk Gn Gn rk RMS Load Graphs R=....758.758.758 rk.... [MV] # #.. [kv]. rk reaker Open@t rk reaker Open@t rk..96..96..96 Load.5.59.5.5.59.59 ontrol System * Vref OMP TIME I P ompar OMP. (b) indiret ontroller. * (a) test system. rtl rtl rtl rtl rtl ompar ompar ompar Gp Gn Gp Gn Gp Gn Gp Gn Gp Gn Gp Gn Gp Gn Gp Gn Gp Gn rtl 5 (b) PWM signal gener. ig 6 STTOM and assoiated ontrollers implemented in PS/EMT.
International onferene on Power Systems Transients IPST in New Orleans, US ll neessary elements to implement the network; for example, three-phase transformers, iruits breakers, apaitors; are existent in the PS library, ig. 5(a). urthermore, a three-phase onverter model is also inluded, ig. 5(b). Suh onverter an employ various kinds of swithes, e.g. GTO, IGT, MOSET. The PWM signal gener is also available in the PS library, whih an be used to generate signals for 6 and pulses onverters, ig. 5(b). The implementation of the indiret ontroller is well easy using the Simulink bloks, ig. 5(). Moreover, the phase-modulation of the ontrol angle δ is presented in ig. 5(d). PS/EMT has also all neessary omponents to represent the network, ig. 6(a). Regarding the STTOM and assoiated ontrollers, PS/EMT does not have a pre-defined onverter model in its library. However, suh omponent an be implemented very easily, ig. 6(a). The implementation of the indiret ontroller is also very simple using only existent omponents, ig. 6(b). There is a PWM signal gener in the PS/EMT library, however, it has not been used in this work. The PWM signal gener adopted here is shown in ig. 6(). In ig. 7, the voltage response of bus for the events previously desribed is shown. In this ase, there is no STTOM into the network. It an be verified that the results obtained through the PS and PS/EMT are very similar. The differene is inly due to methodology employed to alulate the three-phase rms voltage. In PS, the three-phase rms voltage is alulated using ourier analysis over a sliding window of one yle of the threephase instantaneous voltage measured []. On the other hand, in PS/EMT, the three-phase rms voltage is alulated transforming the three-phase a instantaneous voltage measured in d and after smoothing it through a filter [4]..6.4 PS EMT mention that there is no filtering equipment in this simulation. Moreover, the results obtained using PS and PS/EMT are very lose and the results obtained here are very similar to the results presented in [5].6.4.. PS EMT..4 ig 8 Voltage response of the test system using PS and PS/EMT ase with STTOM.. VR The test system employed to arry out the simulations onerning the VR atuation is shown in ig. 9, whih is the same system presented in [5]. Suh network is omposed by a kv, 5 Hz generation system, represented by a Thévenin equivalent, feeding two transmission lines through an -winding transformer onneted in Y/ /, /5/5 kv. Suh transmission lines feed two distribution networks through two transformers onneted in /Y, 5/ kv. To verify the working of a VR employed to avoid voltage sags during short-iruit, a fault is applied at point X via a resistane of 6 Ω. Suh fault is applied from to 6 ms.. Vs Z s 5 4 X onverter S Xtr L. vd...4 ig 7 Voltage response of the test system using PS and PS/EMT ase without STTOM. In ig. 8, the voltage response of bus in the presene of a STTOM is presented. It an be verified that the STTOM an keep the terminal voltage approxitely onstant during the transients. The apaity of the STTOM d element was adopted equal to 9 kv, and suh apaitor is previously harged through the iruit breakers shown in ig. 5(b) and 6(a). It is important to 6 L ig 9 One-line diagram of the test system. The implementation of this system using PS and PS/EMT are shown in ig. and respetively. The ins omponents have already been desribed previously. In suh figure, the insertion transformer onnetion is expliitly shown. 4
International onferene on Power Systems Transients IPST in New Orleans, US a b N a b a b a b ault - pulses powergui V Ia z PulsesSignal(s) Vinv _ref 4 isrete system Ts=e-5 (a) VR. ontrol System Vab (pu) VR (a) test system. ig VR and assoiated ontrollers implemented in PS..5 86.5 86.5 86. [MV] # # 5..... Va_inj P_VR Graphs.. [MV] # #. 5. [kv] 5...5..5..5 ault ULTS. [MV] # # 5.. RK Va_inj RK reaker losed@t RK RK Q_VR RMS....96.96.96 # # # # # # Gn Gn Gn. rtl ompar Gp Gn. Vref Gp I P Gp Gp TIME. TIME P_VR.. Power P Q ompar ompar 5 Q_VR * * rtl rtl rtl rtl ompar ompar rtl Gp Gn Gp Gn ig VR and assoiated ontrollers implemented in PS/EMT. In ig, the voltage response of bus 4 is presented; in this ase there is no VR into the network. It an be verified that during the fault ourrene the voltage is very affeted. urthermore, again, the simulation results obtained through PS and PS/EMT are very similar. The voltage response of bus 4 when there is a VR into the network is shown in ig.. It is possible to note that the VR is apable of keeping approxitely onstant the terminal voltage. In this ase, the VR oupling transformer is onneted in in the VR side and the leakage reatane is equal to %. dditionally, the apaity of the d element was adopted equal to 5 kv [5]. The other data an be obtained in [5]. 5
International onferene on Power Systems Transients IPST in New Orleans, US The in differene between the results obtained from PS and PS/EMT ours when the VR omes in and out. Suh fat is inly due to different methodologies employed for eah software to alulate the three-phase rms voltage, as has already been previously disussed. This has large impat on the ontroller perforne. However, it an be onsidered that the results are very similar. It is important to remember that different measures y also our using real eletri measurement apparatus... PS EMT....5.7.9 ig Voltage Response of the test system using PS and PS/EMT ase without VR.. PS EMT to improve the PS perforne. In this ase a ode is generated. In both programs an integration step equal to 5 µs was utilized. Easiness of implementation of the models: both program have all neessary elements to implement the STTOM and VR models easily. Moreover, the graphial interfaes of these programs are very friendly. Evaluation of the existent libraries: in this ase the two programs shown to be very omplete. dditionally, the non-existent models an be very easily implemented. urateness of results: it an be verified that the results obtained from PS and PS/EMT are very lose. Therefore, it is possible to onlude that both programs present the same kind of aurateness. omplishment sequential studies for parameter adjustments: both programs have tools to arry out multiple-simulations for determination of parameters; e.g. PI onstants. However, PS an be utilized with several toolboxes and bloksets existent in Matlab/Simulink environment, for example Nonlinear ontrol esign lokset. In general terms, both programs are suitable for transient analyses of ustom power devies and very easy to use. The in advantage of the PS is it be developed into Matlab/Simulink environment, suh fat beome possible to utilize it together with several other ontrol design tools. On the other hand, the in advantage of the PS/EMT is omputing time. In this software the simulations run very fast......5.7.9 ig Voltage Response of the test system using PS and PS/EMT ase with VR. V. ONLUSIONS onsidering the appliability of the PS and PS/EMT for transients analysis of ustom power devies and based on results presented and disussed in previous setions, the following onlusions an be obtained: omputation time: in this ase the PS/EMT is quiker beause it is implemented in ortran 9 and the PS is based on a sript language. However, it is possible to use the Simulink eler and the Real-Time Workshop KNOWLEGMENTS The authors would like to aknowledge the finanial support of PESP, razil (Pro. /598-). The authors would also like to aknowledge Mr. Gilbert Sybille from IREQ for his help in the development of the STTOM model using the Power System lokset. REERENES [] N.G. Hingorani, Introduing ustom Power, IEEE Spetrum, Vol., pp. 4-48, 995. [] E ha, V.G. gelidis, O. naya-lara, T.J. Miller, Power Eletroni ontrol in Eletrial Systems, st edition, Newnes,. [] TEQSIM International In., Power System lokset User's Guide,. [4] Manitoba HV Researh entre, EMT User's Manual, 99. [5] O. naya-lara, E. ha, Modeling and nalysis of ustom Power Systems by PS/EMT, IEEE Trans., Power elivery, Vol. PWR-7 (), pp. 66-7,. [6] N.G. Hingorani and L. Gyugyi, Understanding TS: onepts and Tehnology of lexible Transmission Systems, st edition, The Institute of Eletrial and Eletronis Engineers,. [7] Y.H. Song, and.t. Johns, lexible a transmission systems (TS), st edition., The Institution of Eletrial Engineers, 999. 6