Modelling and Control of Photovoltaic Inverter Systems with Respect to German Grid Code Requirements

Transcription

1 1 Modelling and Control of Photovoltaic Inverter Systems with Respect to German Grid Code Reqirements Tobias Nemann, Stdent Member, IEEE, István Erlich, Senior Member, IEEE Abstract The increasing share of renewable energies leads to the point where system operators reqire accrate modelling of generation nits for planning extensions in their power system. Conseqently manfactrers of renewable energy sorces like wind power, photovoltaic or biomass are obliged to prove the electrical properties of their nits in field test as well as validate the performance of simlation models based on the measrements. In Germany standards and technical gidelines were defined which describe the procedre for the testing of the generation nits and the validation of the simlation models. Independent certification bodies have to spervise the testing and analyze the performance of the models to determine whether or not the generation nit flfills the German Grid Code Reqirement. In this paper the athors describe the different reqirements of the German grid code with focs on the dynamic reqirements. Additionally the testing setp and the methodology for comparing measrements with simlation are explained. Frthermore the control of a three-phase inverter system for photovoltaic application is derived. Finally simlation reslts of a photovoltaic inverter system are presented in order to demonstrate the behavior dring short term grid distrbances. Index Terms Control, Falt Ride Throgh, Grid Codes, Three-Phase Inverter, Modelling, Photovoltaic, Renewable Energy L I. INTRODUCTION arge scale photovoltaic (PV) systems are one part of the efforts of national governments to increase the share of renewable energy sorces in the energy mix. Most of the existing PV systems are small nits installed in hoseholds and connected to the low voltage level. By contrast large scale PV nits are connected to the medim or even to the high voltage level. As a conseqence large scale PV systems affect the power flow in the interconnected network and so they have to flfill certain reqirements regarding their electrical properties. According to German reglations these reqirements for PV application are eqivalent to all renewable energy based generation nits like wind power or biomass. The reglations contain reqirements concerning the Tobias Nemann ( tobias.nemann@ni-disbrg-essen.de) and István Erlich ( istvan.erlich@ni-disbrg-essen.de) are with the Institte of Electrical Power Systems at the University Disbrg-Essen, 4757 Disbrg, Germany. active and reactive power control, the power qality, the system secrity fnctions and behavior dring grid distrbances. These reglations are designed to achieve a reliable and secre grid integration of renewable energies sorces. Usally detailed system stdy takes place before large nits are connected to the grid. For these investigations accrate simlation models are needed. The prpose of the German certification rles is to gather information abot electrical properties of a nit by measrements dring field or laboratory testing and to tilize these measrements for creating simlation models which describe the behavior of the nit as exactly as possible. Based on certificates for certain nits or types simlation models for larger systems will have to be developed. Sch system can be an onshore or offshore wind farm consisting of a certain nmber of wind trbines or a large scale PV Systems consisting of several PV inverter systems. Unit and system certificates are highly desired by the manfactrers these days de to the necessity for the grid integration as well as for marketing interests. II. TECHNICAL REQUIREMENTS Grid codes with new reglations for generation nits had to be pblished all over the world in the last years. In Germany it is distingished between conventional generation nits with directly connected synchronos generators and generation nits of renewable energy nits like wind power, PV or biomass. The reglations contain reqirements concerning the active and reactive power control, the power qality, the secrity fnctions and the behavior dring grid distrbances. The active power reqirements inclde the verification of maximm active power which can be provided by the nit, the verification of active power redction in case of a defined setpoint change or in case of overfreqency and the verification of the active power gradient after restarting the nit. The reactive power reqirements imply the verification of the reactive power limits of the nit. This will be tested for the indctive as well for the capacitive limit in step changes of the active power configration. Additionally the verification of reactive power provision in case of a defined setpoint change, the reactive power step response of the system and the steady state voltage control de to reactive power provision are reqired in the technical reglations. Regarding the power /1/$31. 1 IEEE

2 qality and the interaction between the power system and the generation nit reglations and limits for flicker and harmonics exist in order to obtain a reliable operation of the power system. Secrity fnctions needed to be implemented and tested dring different operating points of the generation nit. Therefore the behavior of the generation nit below or above certain voltage and freqency thresholds is part of grid code reqirements. Generation nits shold stay connected to the power system dring distrbances. Additionally, the reactive crrent control of the generation nit mst be sed to spport the grid voltage in case of short term voltage drop or increase. In order to spport the grid voltage, there are strict stiplations concerning the time variation of the injected reactive crrent following a step voltage change. Fig. 1 describes the Falt Ride Throgh (FRT) reqirements of the German grid code which crrently forms the basis for testing by the certification body. It is reqired that the nit stays connected to the power system for 15 ms in case of a voltage dip to p.. (so called Zero Voltage Ride Throgh). With modern protection devices the falt drations are normally in a range of some hndred milliseconds or less. In spite of this voltage dips with drations of p to 1.5 s according (the red crve in Fig. 1) will be tested. Usally the FRT testing procedre is separated into 4 parts with the following residal (falt) line-to-line voltages a) U res.5 p.., b). U res.5 p..,.45 U res.55 p.. and.7 U res.8 p.. with corresponding dration for each. All tests shold be performed for symmetrical, three phase falt as well as for nsymmetrical, two phase falts. Line-To-Line Voltage [p..] No Tripping Tripping is allowed nder definite circmstances Time / ms Fig. 1. Falt Ride Throgh Characteristic range for the gain of...1 p.. with a defalt vale of p.. To allow the controller also to react to small voltage deviations and to speed p the response a continos voltage control withot dead band can be favorable. Units mst be able to inject at least 1 p.. reactive crrent dring symmetrical falts according to the proportional characteristic of the voltage controller. For nsymmetrical falt the generation nit has to provide at least.4 p.. positive seqence reactive crrent. The two characteristics in Fig. describe two different German grid codes. The red crve explains the reqirements according to the Transmission Code 7 [5] and the Medim Voltage Directive of 8 [6] while the ble crve illstrates the demands of the Ordinance on System Services by Wind Energy Plants (SDL Wind 9) [7]. These reqirements for symmetrical and nsymmetrical falts are related to the positive seqence of the crrent. Concerning the negative seqence crrent injection there are no reqirements yet. Hence the negative seqence crrent is sally sppressed to zero. crrent deviation ΔI [p..] Dead band 1 -- = Transmission Code 7 -- = SDL Wind capacitive Reqired reactive Fig.. Voltage Control Characteristic Q k indctive Voltage drop or increase ΔU [p..] Δ U = U U 1 st + ΔIQ Gain: k = -1 Δ U k = 1 p.. Fig. 3 shows the reactive crrent injection as a fnction of time dring the short term voltage spport. Hereby a rise time of less than 3 ms and a settling time of less than 6 ms are reqired. A tolerance band between + % and -1 % of the dynamic crrent reference vale is considered. Rise time describes the time at which the reactive crrent reaches the tolerance band for the first time while the settling time implies the time needed for the reactive crrent to remain within the limits of the tolerance band. This timing characteristic only refers to the positive seqence qantity. Additionally to the FRT capability the German grid code reqires the spport of grid voltage dring falt conditions by the generation nits. Fig. describes the reqired behavior dring voltage drop or increase. In case of overvoltage the nit has to inject indctive reactive crrent while dring voltage drops the nit has to feed in capacitive reactive crrent. The voltage ΔU is defined as difference between the pre-falt voltage and the voltage dring the falt. The voltage control may inclde a dead band of ±.1 p.. and has a setting

3 3 Step Response Final Vale % -1% Rise Time 3 ms Settling Time 6 ms Fig. 3. Step Response of the Reactive Crrent Injection dring FRT III. TESTING SETUP AND DATA PROCESSING Time Regarding the behavior dring grid distrbances FRT tests of PV inverter systems can be performed by a testing setp described in Fig. 4. Hereby an indctive voltage divider is sed to create defined voltage dips to the generation nit. Frther testing setps are possible, which are already developed and can be sed for the FRT testing (e.g. controlled grid simlators). The residal voltage depends on the ratio between X1 and X. This testing device is designed to redce the voltage at the terminals of the generation nit to a defined level within a short period of time. The effect of FRT testing to the connected power system shold be limited de to the configration of the voltage divider coordinated with the short circit capacity of the interconnected power system. Dring normal operation, switch S1 is closed and switch S is open. In order to redce the impact of the short circit on the grid, switch S1 is opened and the series reactance X1 is connected between the grid and generation nit in order to redce the short circit capacity at the point of interconnection. After all transients have died down, the short circit impedance X is connected in parallel to the generation nit by closing switch S. This cases the voltage dip. After a certain time period depending on the desired dration of the dip S is opened again. The voltage at the generation nit recovers and after decaying all transients S1 is closed again and normal operation is restored. Three-phase PV inverter systems are connected to the low voltage side. The testing device can also be configred for the low voltage level bt can also be designed for the medim voltage level. Therefore an additional transformer between medim and low voltage side is necessary. The measrement point for analyzing the electrical qantities of the generation nit is always located at the low voltage terminals of the PV inverter system. The DC sorce can be implemented as a rectifier controlling and emlating the power flow coming from the PV cells. Fig. 4. Principle setp for FRT testing of a PV inverter system Three phase measrements of the phase voltages and crrents at the measrement point MP are necessary for evalating the dynamic behavior of the generation nit. To verify the system response to the aforementioned grid reqirements, the measred data mst be frther processed to calclate the qantities of voltage, active and reactive power and active and reactive crrent as positive seqence fndamental freqency components. One procedre to calclate the positive seqence based on the measrement of instantaneos voltages and crrents is explained in IEC Annex C [8]. Hereby the following eqations can be sed t = () t cos( π ft)dt (1) k,cos k 1 T tt t = () t sin( π ft)dt () k,sin k 1 T tt where k = a,b,c and f 1 is the fndamental freqency. 1 [ 3( )] 6 + = (3) 1,cos a,cos b,cos c,cos c,sin b,sin 1 [ 3( )] 6 + = (4) 1,sin a,sin b,sin c,sin b,cos c,cos The positive seqence crrent components i 1+,cos and i 1+,sin can be calclated by the same eqations like the voltage components 1+,cos and 1+,sin. The active and reactive powers of the fndamental positive seqence are then 3 ( ) i i P + = + (5) 1 1,cos 1,cos 1,sin 1,sin 3 Q1 ( 1,cos i1,sin 1,sin i1,cos ) + = (6). The root mean sqare (rms) vales for active and reactive crrents of the fndamental positive seqence are P 1+ I P1+ = 3 U1 + Q 1+ I Q1+ = 3 U1 + (7) (8).

4 4 The rms phase to phase voltage of the fndamental freqency component can be calclated by the following eqation 3 U1 + ( 1+,sin 1 +,cos ) = + (9). The integration over one period of the fndamental freqency of (1) and () reslts in a considerable delay which will be compensated by sbtracting the cycle time of the fndamental from the calclated reslts for rise and settling time. IV. CONTROL OF THE THREE-PHASE INVERTER A three-phase PV inverter sally consists of an IGBTbased six plse bridge. Fig. 5 shows the basic configration of sch inverter system. Each IGBT modle is controlled by Plse Width Modlation signals coming from the controller. The inverter and its control are mainly responsible for the electrical behavior of the nit. Fnctionalities which needed to be performed by the inverter are the Maximm Power Point Tracking and the decopled control of active and reactive crrent and ths active and reactive power. In the following sbchapter a classical two-dimensional crrent control aligned to the grid voltage for the positive seqence with an oter DC voltage controller is described. Hence there are no reqirements for the negative seqence the target of the control is to sppress the negative seqence crrent totally. Therefore the negative seqence of the voltage has to be calclated qickly and combined with the positive seqence of the inverter voltage. Rewriting eqation (1) in complex qantities the following eqation (11) for the positive seqence with the index + can be derived where the reference frame of each qantity is chosen arbitrarily with the notation and transformed into the per-nit system. d l i dt g,+ g,+ LSC,+ = + (11) The reference frame of the described control is aligned to the grid voltage. Therefore the angle Θ g can be calclated by the trigonometric fnction of eqation (1) where α, β represents the alpha and beta components of the space vector of the positive seqence grid voltage. arctan g,α,+ g,β,+ Θ g = (1) Eqation (11) will be transformed in a reference frame rotating with the angle Θ g described by the notation. The g following two eqations (13) and (14) already separated into the direct and qadratre axis components where ω is the radian freqency of the grid voltage. g di g g,d,+ g g g,d,+ ω g,q,+ LSC,d,+ = l l i + (13) dt g di g g,q,+ g g g,q,+ ω g,d,+ LSC,q,+ = l + l i + (14) dt It follows from above (13), (14) that the reference voltages of the inverter can be described by the following eqations g g g g LSC,d,+ g,d,+ ' LSC,d,+ ω g,q,+ = + l i (15) L1 L L3 i g L C DC g g g g LSC,q,+ g,q,+ ' LSC,q,+ ω g,d,+ = l i (16) where the two crrent PI-controllers can be derived as follows g 1 g g ' LSC,d,+ = KP 1 + ( ig,d_ref,+ ig,d,+ ) Ts (17) N g LSC Fig. 5. Principle arrangement of a PV inverter From the loop eqation for the voltages in the circit described in Fig. 5 the following eqation (1) is derived. The grid voltage is represented by g while the controlled otpt voltage of the converter is LSC. The voltage drop over the choke can be calclated sing the indctance of choke L and the crrent i g. Hereby the resistive part of the choke will be neglected becase the Q factor of the choke is ordinarily small. g 1 g g ' LSC,q,+ = KP 1 + ( ig,q_ref,+ ig,q,+ ) Ts (18). The PI controllers have to adjst the otpt voltage of the inverter in case of a setpoint change qickly. In steady state the feed forward and the grid voltage terms of (15) and (16) lead already to stable conditions in the control loop. Only errors in system parameters need to be compensated by the PI controller in steady state. The corresponding principle bloc diagram of the inner crrent control loop of the inverter is described in Fig. 6. g,l1 i g,l1 LSC,L1 d g,l = L ig,l + LSC,L dt g,l3 i g,l3 LSC,L3 (1)

5 5 i g,d_ref,+ i g,d,+ i g,q_ref,+ i g,q,+ 1 KP 1+ Ts ω l ω l 1 KP 1+ Ts g,d,+ g,q,+ Fig. 6. Bloc diagram of the positive seqence crrent control LSC,d,+ LSC,q,+ The principle positions and strctre of the complex phasors of a three-phase PV inverter system is illstrated in Fig. 8. g j i g,q j q g i g g i g,d j β LSC g Θ g g,α jωl i g g,β α d g In order to sppress the negative seqence crrent in nbalanced steady state or dynamic conditions the negative seqence grid voltage with the index - has to be merged with the positive seqence reference voltage of the inverter. Fig. 7 illstrates the copling of the negative seqence space vector with the space vector of the positive seqence. The reslting reference inverter voltage contains the positive and the negative seqence voltage. LSC,d,+ LSC,q,+ g,d,- g,q,- j g e Θ j g e Θ LSC,α,+ LSC,β,+ LSC,α,- LSC,β,- Fig. 7. Calclation of the reference inverter voltage LSC,α LSC,β Positive seqence active and reactive power of the inverter can be written in the following eqations where p + and q + are expressed as rated vales. g g g g g,d,+ g,d,+ g,q,+ g,q,+ p i i + = + (17) g g g g g,q,+ g,d,+ g,d,+ g,q,+ q i i + = (18) De to the alignment of reference frame on the grid voltage the qadratre components of g is eqal to zero in case of symmetrical conditions. It follows that direct component i g,d,+ and the qadratre component -i g,q,+ are proportional to the active and reactive power. The power balance between the DC and the AC circit can be described by the following eqation withot considering losses inside the inverter system g g DC DC g,d,+ g,d,+ p = i i (19) It is obvios that the DC voltage is inflenced by the direct component i g,d,+ of the inverter system. The reference vale of the direct component crrent controller is conseqently the otpt of the oter DC voltage controller where the reference vale of the DC voltage reslts from the MPP Tracking. 1 1 ( ) Ts g ig,d_ref,+ = KP + DC_ref DC () Fig. 8. Principle phasor diagram of a PV inverter V. SIMULATION RESULTS In the following chapter reslts carried ot by EMT simlation models based on Matlab/Simlink describe the dynamic behavior of PV inverter systems dring short term grid distrbances. Therefore an inverter system with the nominal active power of 15 kw was designed. The system is connected to the low voltage level with a nominal voltage of 7 V which means that the nominal crrent of the system is 3 A. The connected low voltage grid has a short circit power of 1 MVA. The grid distrbances are created by an indctive voltage divider described in chapter IV. The short circit ratio between FRT testing device and generation nit is 1. The instantaneos vales are measred at the point between generation nit and testing device. Based on these instantaneos vales the positive and negative seqences are compted according to the calclation method of IEC described in chapter IV. In this chapter three different falt conditions and the response of the inverter system are visalized. In Fig. 9 a symmetrical three-phase falt to 5 % of the nominal voltage with a dration of 15 ms is illstrated. Usally this FRT testing reqires a longer dration, bt de to the intention of showing the instantaneos vales of the system the dration of the falt is redced. From the figres it is clear that the conditions dring the falt are stable so that the inverter can withstand this falt for a longer period of time. In Fig. 1 a direct symmetrical falt (with a residal voltage of % of the nominal voltage) is shown. Hereby the inverter stays connected to the power system and is injecting capacitive reactive crrent. The calclation method according to IEC has the disadvantage that a very small voltage system leads to an error in separating active and reactive power components as well as the active and reactive crrent component. So in Fig. 8 only the positive seqence of the apparent power and crrent is visalized. Fig. 11 describes an nsymmetrical two-phase falt and the response of the inverter system. The voltage of the two affected phases is redced to % of the nominal voltage. So the positive seqence of the voltage is redced to 5 % while

6 6 3 Grid Line-To-Grond Voltage 3 Grid Line-To-Grond Voltage Crrent / A Crrent / A Crrents Positive & Negative Seqence Voltage Positive Seqence Active & Reactive Crrent U Negative Seqence Active & Reactive Crrent U 1- I P1+ I Q1+ Crrent / A Crrent / A Crrents Positive & Negative Seqence Voltage Positive Seqence Apparent Crrent U Negative Seqence Apparent Crrent U 1- I S1+ Crrent / A 1-1 I P1- I Q1- Crrent / A 1-1 I S1- Power / kw,kvar Positive Seqence Active & Reactive Power Time / s Fig. 9. Simlation reslts of a FRT Test with a symmetrical voltage drop to 5 % of the nominal voltage P 1+ Q 1+ Power / kva Positive Seqence Apparent Power Time / s Fig. 1. Simlation reslts of a FRT Test with a symmetrical voltage drop to % of the nominal voltage S 1+

7 Grid Line-To-Grond Voltage a negative seqence of 5 % occrs dring this nsymmetrical falt. Althogh there is only a reqirement to inject.4 p.. of the positive seqence capacitive reactive crrent the inverter has the capability to provide p to 1 p.. of reactive crrent dring nbalanced conditions. The negative seqence crrent is sppressed totally althogh a negative seqence of approximately 5 % occrs in the voltage system. 7 Crrent / A Crrent / A Crrent / A Power / kw,kvar Crrents Positive & Negative Seqence Voltage Positive Seqence Active & Reactive Crrent Negative Seqence Active & Reactive Crrent Positive Seqence Active & Reactive Power U Time / s U 1- I P1+ I Q1+ Fig. 11. Simlation reslts of a FRT Test with an nsymmetrical -phasefalt with voltage drop to % of the nominal voltage in the affected phases I P1- I Q1- P 1+ Q 1+ VI. CONCLUSION In this paper the athors have introdced the crrent reqirements of German grid codes which are valid for generation nits of renewable energy sorces like wind power, PV or biomass. Frthermore the procedre for performing the FRT testing for PV inverter systems and the methodology for calclating the positive seqence qantities based on instantaneos vales were described. The control of the inverter system is to a large extent responsible for the behavior in steady state as well as in dynamic operational mode of PV application. Hence the athors introdced the basic control strctre of the inverter systematically. Conseqently EMT simlation reslts illstrate the dynamic behavior of a PV inverter system dring grid distrbances and show the good performance of this control. In the three case stdies the simlation reslts illstrate that modern PV application systems can meet the reqirements of the German grid code. Dring symmetrical falts the PV inverter can ride throgh any voltage dips. Even zero voltage ride throgh is possible with an appropriate control strategy. Additionally the inverter is capable of injecting a considerable positive seqence capacitive reactive crrent to spport the dipping voltage. Dring nsymmetrical falt events the PV inverters are also able to ride throgh the nbalanced conditions and sppress a negative seqence crrent totally. The injected positive seqence capacitive reactive crrent dring nsymmetrical falts can be 1 p.. Regarding the timing of the voltage control the simlation reslts emphasize that PV inverter system can meet the reqirement of the rise and the settling time of German grid codes by sing fast control actions dring falt conditions. VII. REFERENCES [1] J.Fortmann, S.Engelhardt, J.Kretschmann, C.Feltes, I.Erlich, Validation of an RMS DFIG Simlation Model According to New German Model Validation Standard FGW TR4 at Balanced and Unbalanced Grid Falts, 8th International Workshop on Large-Scale Integration of Wind Power into Power Systems as well as on Transmission Networks for Offshore Wind Farms, October 9 Bremen, Germany [] C.Feltes, S.Engelhardt, J.Kretschmann, J.Fortmann, I.Erlich, Dynamic Performance Evalation of DFIG-based Wind Trbines regarding new German Grid Codes Reqirements, IEEE PES General Meeting, Jly 1 Minneapolis, Minnesota, USA [3] S.Engelhardt, J.Kretschmann, J.Fortmann, F.Shewarega, I.Erlich, C.Feltes, Negative Seqence Control of DFG based Wind Trbines, IEEE PES General Meeting, Jly 11 Detroit, Michigan, USA [4] T.Nemann; C.Feltes, I.Erlich, Response of DFG-Based Wind Farms Operating on Weak Grids to Voltage Sags, IEEE PES General Meeting, Jly 11 Detroit, Michigan, USA [5] VDN: Transmission Code 7, Netz- nd Systemregeln der detschen Übertragngsnetzbetreiber, Version 1.1, Agst 7

8 8 [6] BDEW: Technische Richtlinie Erzegngsanlagen am Mittelspannngsnetz, Jni 8 [7] Verordnng z Systemdienstleistngen drch Windenergieanlagen (Systemdienstleistngsverordnng SDLWindV), BMU, Germany [8] IEC Edition, Wind Trbines, Part 1: Measrements and assessment of power qality characteristic of grid connected wind trbines [9] FGW: Technical Gidelines for Power Generation Units - Part 3, Determination of electrical characteristics of power generating nits connected to MV, HV and EHV, Rev.1, [1] FGW: Technical Gidelines for Power Generation Units - Part 4, Demands on Modeling and Validating Simlation Models of the Electrical Characteristics of Power Generation Units and Systems, Rev.5,.3.1 [11] FGW: Technical Gidelines for Power Generation Units - Part 8, Certification of Electrical Characteristics of Power Generating Units and Systems in the Medim-, High- and Highest-voltage Grid, Rev.5, VIII. BIOGRAPHIES Tobias Nemann (1977) received his Dipl.-Ing. degree in electrical engineering from the University of Disbrg-Essen/Germany in 9. Since Janary 1 he is doing his Ph.D. stdies in the Department of Electrical Power Systems at the same University. His research interests inclde wind and PV power generation, mainly focssing on control and modelling as well as DSP programming. He is stdent member of IEEE. Istvan Erlich (1953) received his Dipl.-Ing. degree in electrical engineering from the University of Dresden/Germany in After his stdies, he worked in Hngary in the field of electrical distribtion networks. From 1979 to 1991, he joined the Department of Electrical Power Systems of the University of Dresden again, where he received his PhD degree in In the period of 1991 to 1998, he worked with the conslting company EAB in Berlin and the Franhofer Institte IITB Dresden respectively. Dring this time, he also had a teaching assignment at the University of Dresden. Since 1998, he is Professor and head of the Institte of Electrical Power Systems at the University of Disbrg-Essen/Germany. His major scientific interest is focsed on power system stability and control, modelling and simlation of power system dynamics inclding intelligent system applications. He is a member of VDE and senior member of IEEE.