New Procedure for Measuring Dynamic MPP-Tracking Efficiency at Grid-Connected PV Inverters

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters H. Haeberlin and Ph. Schaerf Berne University of Applied Sciences (BFH-I), Division of Electrical- and Communication Engineering, Laboratory for Photovoltaics, Jlcoweg 1, CH-34 Burgdorf, Switzerland Phone: +41 34 426 68 11, Fax: +41 34 426 68 13, e-mail: heinrich.haeberlin@bfh.ch, Internet: www.pvtest.ch ABSRAC: At locations, where there are often variable cloudy conditions, besides the static also the dynamic MPP-behaviour has to be considered. In the long-term monitoring projects performed by the PV laboratory of BFH- I, it could be shown that dynamic MPP-tracking problems actually occur. In a paper presented at the 21 st EU PV conference in 26, first results of dynamic MPP-tracking tests carried out with quasi rectangular variations between two different power levels were presented [3]. Such tests can easily be realised with high measurement accuracy and give a good insight in the control behaviour of the inverter. More realistic are tests with ramps with rising and falling slopes of current (proportional to irradiance G) or MPPpower. It is more difficult to realise such tests with a high measuring accuracy. Like for static tests, before the start of a dynamic MPP-tracking test an initial set-up time is needed for stabilisation of the MPP-tracker in order to have a defined starting position. his stabilisation period is followed by a few test cycles during the actual measuring period M. Inverters are always a little behind the effective MPP, therefore the offered MPP-power is not absorbed completely after a change. In this paper, these test patterns (according to the proposed new standard for inverter performance FprEN553), results of many tests with them at different inverters and possible problems are discussed. KEYWORDS: Inverter, MPP-tracking, dynamic behaviour 1 Introduction In 25 the new quantity overall or total efficiency η tot was introduced, which can describe the static operating behaviour of a grid-connected PV inverter much better than conversion efficiency η alone [1]. otal efficiency η tot is the product of DC-AC conversion efficiency η and static MPP-tracking efficiency η MPP. his quantity is increasingly used also by other institutions and professional journals (e.g. Photon). For such measurements suitable PV array simulators are needed [1], [4]. At locations, where there are often variable cloudy conditions, besides the static also the dynamic MPPbehaviour has to be considered. Inverters with a fast MPP-tracker have a somewhat higher energy yield under quickly changing irradiance than devices with slow MPP-tracking (MPP). In a study of FhG-ISE in 25, it was shown that with a correct sizing of the inverter and a fast dynamic MPP-tracking, in principle a few additional percent of energy could be obtained from the same PV array [2]. In the long-term monitoring projects carried out by the PV laboratory of BFH-I, it could be shown that dynamic MPP-tracking problems actually occur. In a paper presented at the 21 st EU PV conference in 26, first results of dynamic MPP-tracking tests carried out with quasi rectangular variations between two different power levels were presented [3]. Such tests can easily be realised with high measurement accuracy and give a good insight in the control behaviour of the inverter. More realistic are tests with ramps with rising and falling slopes of current (proportional to irradiance G) or MPPpower. It is more difficult to realise such tests with a high measuring accuracy. Like for static tests, before the start of a dynamic MPP-tracking test an initial set-up time is needed for stabilisation of the MPP-tracker in order to have a defined starting position. his stabilisation period is followed by a few test cycles during the actual measuring period M. Inverters are always a little behind the effective MPP, therefore the offered MPP-power is not absorbed completely after a change. 2 Definition of Dynamic MPP-racking- Efficiency or -Accuracy Dynamic MPP-tracking efficiency or MPP tracking accuracy η MPPdyn can be defined as: η MPPdyn = M v DC M ( t) i p DC ( t) dt ( t) dt MPP where v DC (t) = voltage at the DC input of the inverter. i DC (t) = current at the DC input of the inverter. p MPP (t) = available maximum power of PV array simulator in the respective instantaneous MPP. M = Measurement duration (start at t = ). he integral in the denominator represents the whole MPP energy that could be absorbed under optimum conditions by the inverter. In 28, the PV laboratory of BFH-I participated in a workgroup in Germany, which had to create a draft for a standard to measure overall efficiency of PV inverters including suitable test patterns with ramps for MPPtracking. Many different test patterns were proposed and discussed. In order to get practical results of tests with these test patterns, several inverters from different manufacturers were tested in order to obtain optimum test patterns that are neither too stringent nor too mild. hese test patterns were finally approved by the working group and are now included in the final draft for a provisional European standard (FprEN553). In this paper, the resulting test patterns and some test results are shown. () 1 1

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters 3 est Results 3.1 Dynamic MPP Problems Registered in Real Operation During the long-term monitoring projects of PV plants, which have been performed since 1992 by the PV laboratory of BFH-I, some dynamic MPP tracking problems in real operation could be registered (see fig. 1). Fig.1: Example of a dynamic MPP-tracking problem at a PV plant on Sept. 24 th, 27. Ggen = irradiance into the array plane Udc = DC voltage V DC Pdc = DC power P DC Pac = AC power P AC his problem was caused by several ramp-like variations of irradiance that irritated the inverter. Between 14: and 14:35 the power P DC absorbed by the inverter is sometimes too low compared with the irradiance, as the DC voltage V DC, on which it operates, has increased due to the irritation of the MPP-tracker and for some time is considerably higher than the MPP-voltage V MPP. 3.2 Possible Problems with Reproducibility already During Static MPP ests It will be shown later that sometimes there are reproducibility problems with dynamic MPP-tracking tests. In some cases there may be minor reproducibility problems already during static MPP-tracking tests. With some inverters, especially at low power levels in the range of a few percent of P DCn, under static conditions minor MPP-problems may occur, which are caused by the fact that from time to time the inverter looks for a possible new MPP not only in close vicinity of the actual MPP, but in a wider neighbourhood of it (see fig. 2) [4]. 3 25 2 15 1 5 SolarMax 1C: DC operating points at V MPP = 44 V 1 2 P-DC V-DC M = 1: η MPP = 99,% M = 2: η MPP = 96,51% 35 1 2 3 4 5 6 ime [s] M = - 6 s: η MPP = 98,71% Fig. 2: P DC and V DC versus time during operation of an inverter during a static MPP test at P MPP 2.65 kw. 65 6 55 5 45 4 During the time where P DC is < P MPP, a certain amount of energy is lost, and thus the measured value of η MPP is reduced somewhat. Depending on the duration of the measuring interval M used to determine η MPP, (e.g. 1 to 1 minutes, see [1]), 1 or 2 such search processes in the wider neighbourhood occur. herefore, depending on the state of the internal clock in the inverter at the start of the measuring period and the duration of this measuring interval M, measured η MPP may vary here between 93% and 99.9%, i.e. measurement of static η MPP is not completely reproducible at this power level for such devices. his problem can be alleviated considerably by using a sufficiently high measuring interval M. he higher M, the closer the measured values of η MPP are together and the better the reproducibility. herefore in the final draft for a European standard FprEN 553 for M 1 minutes are proposed. 3.3 Definition and Description of the Different est Ramps used for Dynamic MPP ests In the workgroup in Germany, which was already mentioned, in course of 28 different test patterns with ramps of different rise time were discussed and examined (see fig. 3 and 4). During these tests the number n of repetitions and the time intervals t 1 t 4 have to be varied in a wide range, in order to test if the MPPtracking algorithm of the inverter can follow irradiance variations with different rise time without problems. G Equivalent Normalised Irradiance G SC 5% 1% Intitial Setup ime t 1 t 2 t 3 t 4 n Repetitions Shape: e.g.12x (15/5H/15/5L) t Fig. 3: est with variations between low and medium irradiance (1% to 5% of G SC = 1 kw/m 2, some variation of V MPP ). G G SC 1% 3% Equivalent Normalised Irradiance Intitial Setup ime t 1 t 2 t 3 t 4 n Repetitions Shape: e.g.12x (15/5H/15/5L) t Fig. 4: est with variations between medium and high irradiance (3% to 1% of G SC = 1 kw/m 2, low variation of V MPP ). 2

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters he individual tests are defined by the following description code ( n x (t 1 /t 2 H/t 3 /t 4 L), see table 1): Number n repetitions, then x and in brackets the duration t 1 of the rise time in seconds, then the dwell time t 2 at high level in seconds (number + H), then the duration t 3 of the fall time in seconds and finally the dwell time at low level in seconds (number + L). According to the model for the I-V curve used in the PV array simulator, P MPP is about proportional to irradiance G, whereas V MPP varies slightly according to irradiance G. he measurements are relatively time consuming. For a measuring campaign according to table 1 including the set-up time of 3 s on the starting level for each test about 6½ hours are needed. Irradiance Variation 1 % 5% of G SC Number Slope Mode Duration n [W/m 2 /s] [s] 2.5 (8 / 1H / 8 / 1L) 354 2 1 (4 / 1H / 4 / 1L) 194 3 2 (2 / 1H / 2 / 1L) 156 4 3 (133 / 1H / 133 / 1L) 1447 6 5 (8 / 1H / 8 / 1L) 138 8 7 (57 / 1H / 57 / 1L) 1374 1 1 (4 / 1H / 4 / 1L) 13 1 14 (29 / 1H / 29 / 1L) 171 1 2 (2 / 1H / 2 / 1L) 9 1 3 (13 / 1H / 13 / 1L) 767 1 5 (8 / 1H / 8 / 1L) 66 otal ime for Setup + Measurement 15939 Irradiance Variation 3 % 1% of G SC Number Slope Mode Duration n [W/m 2 /s] [s] 1 1 (7 / 1H / 7 / 1L) 19 1 14 (5 / 1H / 5 / 1L) 15 1 2 (35 / 1H / 35 / 1L) 12 1 3 (23 / 1H / 23 / 1L) 967 1 5 (14 / 1H / 14 / 1L) 78 1 1 (7 / 1H / 7 / 1L) 64 otal ime for Setup + Measurement 6987 able 1: Planned ramp tests, number of repetitions and rise, dwell and fall times for irradiance or power variations for variation from low to medium (on top of table) and medium to high (on bottom of table) irradiance according to FprEN553. Before each measurement a set-up time of 3 s on the initial level is provided. he ramps described in table 1 are a compromise obtained after intense discussions and many tentative tests in the German workgroup DKE AK 373..3 and are now also included in the draft for a provisional EN FprEN535. At inverters with poor dynamic MPP-tracking, sometimes the resulting reproducibility of the measurements is a certain problem. Depending on the instantaneous DC voltage V DC, its actual search direction (to higher or lower values) and the state of the internal clock of the inverter at the beginning of the dynamic MPP test, for identical tests at the very same inverter sometimes different values for dynamic MPP tracking efficiency η MPPdyn are obtained. 3.4 Results of Dynamic MPP ests with such Ramps at Different Inverters 3.4.1 Overview Diagrams est patterns with ramps according to table 1 were used to perform dynamic MPP tests at six different inverters (INV1 to INV6). he MPP tracking algorithm used by the manufacturer is decisive for the dynamic tracking behaviour. Especially useful were tests with two inverters (INV3 and INV6) from two different manufactures, for which an old firmware version with relatively poor dynamic MPP tracking behaviour and a new firmware version with very good dynamic MPP tracking behaviour were available. Fig. 5, 6 and 7 show the results of such dynamic MPP tests at INV3 and INV6 with old firmware in overview diagrams indicating measured η MPPdyn for different ramp steepness. It is obvious that the devices with the old firmware have problems at certain ramp steepness and measured values of η MPPdyn are considerably below 1%. 1 98 96 94 Dynamic MPP tracking efficiency η MPPdyn for INV3 (V MPP high = 2V); old firmware, test 1 92 1% --> 5% 9 3% --> 1% 88.1 1 1 1 Fig. 5: with ramps according to table 1 at the inverter INV3 with old firmware (test run 1). he measured values of η MPPdyn are all in the range between 96% to 1%, there is no pronounced "resonance", where the device is seriously irritated by the repeated ramps. 1 98 96 94 Dynamic MPP tracking efficiency η MPPdyn for INV3 (V MPP high = 2V); old firmware, test 2 92 1% --> 5% 9 3% --> 1% 88.1 1 1 1 Fig. 6: with ramps according to table 1 at the inverter INV3 with old firmware (test run 2). Here for a slope of 2 W/m 2 /s some kind of "resonance" occurs, where the device is irritated by the repeated ramps (like in fig. 1) and therefore dynamic MPP efficiency is reduced. A comparison with fig 5 reveals that there is no exact reproducibility here! 3

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters 1 9 Dynamic MPP tracking efficiency η MPPdyn for INV6 (V MPP high = 33V); old firmware 1% --> 5% 8 3% --> 1% 75.1 1 1 1 Fig. 7: with ramps according to table 1 at inverter INV6 with old firmware. his device has no problems with very slow ramps, but has considerable reductions in dynamic MPP efficiency in the range of about 3 W/m 2 /s to 2 W/m 2 /s and has a pronounced drop at a ramp steepness of 1 W/m 2 /s. During identical tests the resulting values at a given ramp steepness may be somewhat different, i.e. the test results are not completely reproducible for inverters with deficiencies in dynamic MPP tracking. Fig. 8, 9 and 1 show the results of similar tests with new firmware that was improved by the manufacturers in the meantime. he results of the tests are now excellent. Good reproducibility was observed with all devices tested so far with good results at such tests (compare e.g. fig. 8 and fig. 9). 1. 99.8 99.6 99.4 99.2 99. 98.8 98.6 98.4 98.2 98. Dynamic MPP tracking efficiency η MPPdyn for INV3 (V MPP high = 2V); new firmware, test 1 1% --> 5% 3% --> 1% Scale expanded considerably!.1 1 1 1 Fig. 8: (run 1) with ramps according to table 1 at inverter INV3 with new firmware in the same inverter like in fig. 5 and 6. With this firmware, the dynamic MPP-tracking is excellent and measurement results are well reproducible (like with all devices tested so far with good results at such tests). Note: he scale for η MPPdyn is expanded considerably compared to fig. 5, 6 and 7. 1. 99.8 99.6 99.4 99.2 99. 98.8 98.6 98.4 98.2 98. Dynamic MPP tracking efficiency η MPPdyn for INV3 (V MPP high = 2V); new firmware, test 2 1% --> 5% 3% --> 1% Scale expanded considerably!.1 1 1 1 Fig. 9: (run 2) with ramps according to table 1 at inverter INV3 with new firmware in the same inverter like in fig. 5 and 6. Here dynamic MPP-tracking is also excellent and measurement results are reproducible (see fig. 8). Note: he scale for η MPPdyn is expanded considerably compared to fig. 5, 6 and 7. 1. 99.8 99.6 99.4 99.2 99. 98.8 98.6 98.4 98.2 98. Dynamic MPP tracking efficiency η MPPdyn for INV6 (V MPP high = 33V); new firmware (beta-version) 1% --> 5% 3% --> 1% Scale expanded considerably!.1 1 1 1 Fig. 1: with ramps according to table 1 at another inverter (INV6 with new firmware in same inverter as in fig. 7). Also here dynamic MPP-tracking is excellent. In a second test (not shown here due to space limitations) reproducibility was very good. Note: he scale for η MPPdyn is expanded considerably compared to fig. 5, 6 and 7. Fig. 11, 12, 13 and 14 show the results of further tests with test ramps according to table 1 at other inverters with some deficiencies in dynamic MPP tracking and therefore also problems with reproducibility (compare at first fig. 11 and 12 and then fig. 13 and 14). 1 9 Dynamic MPP tracking efficiency η MPPdyn for INV4 (V MPP high = 28 V); test 1 1% --> 5% 3% --> 1%.1 1 1 1 Fig. 11: with ramps according to table 1 at the inverter INV4 (test run 1). 4

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters 1 9 Dynamic MPP tracking efficiency η MPPdyn for INV4 (V MPP high = 28 V); test 2 1% --> 5% 3% --> 1%.1 1 1 1 Fig. 12: with ramps according to table 1 at the inverter INV4 (test run 2). 1 9 Dynamic MPP tracking efficiency η MPPdyn for INV5 (V MPP high = 37V); test 1 8 1% --> 5% 3% --> 1% 75.1 1 1 1 Fig. 13: with ramps according to table 1 at the inverter INV5 (test run 1). 1 9 Dynamic MPP tracking efficiency η MPPdyn for INV5 (V MPP high = 37V); test 2 8 1% --> 5% 3% --> 1% 75.1 1 1 1 Fig. 14: with ramps according to table 1 at the inverter INV5 (test run 2). 3.4.2 ime Diagrams of Specific Dynamic ests In order to get a better understanding of the behaviour of inverters during dynamic MPP-tracking tests, is makes sense to create diagrams, in which the DC power P DC absorbed by the inverter and the operating voltage V DC are displayed versus time at a certain ramp steepness (see fig. 15 to 23). In some cases also older test patterns than those indicated in table 1 were used for such tests. Fig. 15 to fig. 2 show tests with INV3, IV4, INV5 and INV6 during which the MPP tracking algorithm is considerably disturbed. 2 15 1 INV4: Dynamic DC power at V MPP high = 28 V V DC(t) V MPP Low 24 5 22 2 18 2 4 6 8 1 12 14 16 18 2 Pattern: 3x (2/1H/2/1L) ime [min] η MPPdyn = 91.69% Fig. 15: ime diagram of a ramp test according to table 1 with 2 W/m 2 /s at INV4 (1% 5%). V DC starts at the correct V MPP, but then decreases more and more despite increasing P DC. When P DC decreases V DC remains constant. Only when P DC increases the next time, V DC increases again, but afterwards during the next increase of P DC it goes again in the wrong direction. As already mentioned, during the development of the tests also other tests were carried out with ramps that are not contained in table 1 (see fig. 16, 17 and 18). 3 25 2 15 INV5: Dynamic DC power at V MPP high = 36 V 1 3 28 5 V DC(t) 26 24 1 2 3 4 5 6 7 8 9 1 Pattern: 2x (1/5H/1/5L) ime [min] η MPPdyn = 86.7% V MPP Low Fig. 16: ime diagram of a ramp test according with 4 W/m 2 /s and dwell time 5 s (instead of 1 s) at INV5 (1% 5%). During the test V DC decreases more and more and accordingly P DC also decreases more and more. Here η MPPdyn is only 86.7%. Fig. 17 shows a time diagram of INV3 with old firmware during such a test with a ramp starting at the high instead of the low level. Here the reason for the non reproducible behaviour of the MPP tracking algorithm used can be explained quite well (fig. 18). A ramp steepness of 1 W/m 2 /s seems to be quite critical. But during tests from low to high level according to fig. 5 and 6 a steepness of 2 W/m 2 /s seems to be more critical. 25 2 15 1 5 eistungsangebot High Low INV3, old FW: Dynamic DC power at V MPP high = 29 V PMPP Low PDC(t) VDC(t) VMPP Low 24 1 2 3 4 5 6 7 8 9 1 11 Pattern: 7x (4/5H/4/5L) ime [min] η MPPdyn = 8.5% Fig. 17: ime diagram of a ramp test starting on the high power level with 1 W/m 2 /s at INV3 with old firmware. If P DC decreases or remains constant V DC remains constant, too. If P DC increases, also V DC increases (more detailed explanation in fig. 18). 34 32 3 28 26 4 38 36 34 32 34 32 3 28 26 5

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters 25 2 15 1 5 eistungsangebot High Low INV3, old FW: Dynamic DC power at V MPP high = 29 V PMPP Low PDC(t) P-decrease at ΔV >, V remains constant! P-increase at ΔV >, V is increased further! - 3 Min. expanded V DC(t) VMPP Low 24 1 2 3 Pattern: 7x (4/5H/4/5L) ime [min] η MPPdyn = 8.5% Fig. 18: Expanded detail of fig. 17. As long as the power increases, the algorithm simply assumes, that this power increase is due to the previous change of V DC and therefore it continues to vary V DC in the same direction. he observation interval is about 5 s. Fig. 18 reveals that immediately before the beginning of the rising slope of P DC the voltage V DC was increased a little bit (see yellow line). he MPP algorithm now assumes, that this increase of P DC is a consequence of the increase of V DC and that it shall simply continue in the same direction, i.e. increase V DC further. he search direction only changes when V OC low (at the low power level) is reached (see fig. 17). Fig. 19 and fig. 2 (expanded detail) show a time diagram of a ramp test according to table 1 with 1 ramps (2/1H/2/1L, 1% 5%, steepness 2 W/m 2 /s) at INV3 with old firmware. 25 2 15 1 INV3, old FW: Dynamic DC power at V MPP high = 298 V P DC (t) UMPP High UMPP Low 34 32 3 28 26 3 28 26 24 Due to another position of the internal clock of the inverter at the beginning of the test (V DC has been decreased a little bit immediately before, see yellow line) now this time the device decreases V DC continuously during increasing P DC. he MPP tracking algorithm now assumes that this increase of P DC is a consequence of the reduction of V DC and that it shall simply go on in this direction. As the power P DC does not decrease as fast during a runaway towards lower voltages compared to a runaway in the direction of V OC, η MPPdyn does not drop as low as in fig.17 and 18. In fig. 21, 22 and 23 the results of some tests with INV3 and INV6 with new improved firmware are presented, where the dynamic MPP tracking is now very good. 4 35 3 25 2 15 1 INV6, new FW: Dynamic DC power at V MPP high = 34 V 5 PMPP Low V DC(t) VMPP Low 31 2 4 6 8 1 12 14 16 Pattern: 1x (4/1H/4/1L) ime [min] η MPPdyn = 99.68% VMPP High Fig. 21: ime diagram of a ramp test according to table 1 with 1 W/m 2 /s (1% 5%) at INV6 with new firmware. V DC starts at V MPP of the low power level and then increases slowly towards V MPP of the higher level. he dynamic behaviour is now excellent and η MPPdyn rises to the very good value of 99.7%. 35 34 33 32 5 22 V DC (t) 2 2 4 6 8 1 Pattern: 1x (2/1H/2/1L) ime [min] η MPPdyn = 88.44% Fig. 19: ime diagram of a ramp test according to table 1 with 2 W/m 2 /s (1% 5%) at INV3 with old firmware. If P DC decreases or remains constant V DC remains constant, too. If P DC increases, V DC decreases (more detailed explanation in fig. 2). 25 2 15 1 5 INV3, old FW: Dynamic DC power at V MPP high = 298 V P DC (t) P-increase at ΔV <, V is decreased further! - 2 Min. expanded P-decrease, V remains constant! V DC (t) V MPP Low 2 1 2 Pattern: 1x (2/1H/2/1L) ime [min] η MPPdyn = 88.44% Fig. 2: Expanded detail of fig. 19. he same explanation as in fig. 18 can be used: As long as the power P DC increases, the algorithm assumes, that this power increase is due to the previous change of V DC and therefore it continues to vary V DC in the same direction, i.e. to decrease it. 3 28 26 24 22 25 2 15 INV3, new FW: Dynamic DC power at V MPP high = 297 V 1 28 U MPP Low 5 26 P DC (t) 24 1 2 3 4 5 6 7 8 9 1 ime [min] Pattern: 1x (2/1H/2/1L) η MPPdyn = 99.59% V DC (t) U MPP High Fig. 22: ime diagram of a ramp test according to table 1 with 2 W/m 2 /s (1% 5%) at inverter INV3 with new firmware (same test pattern as in fig. 19 and 2). V DC starts a little lower than V MPP of the low level and follows very quickly the appropriate value of V MPP. Here a very good value of η MPPdyn = 99.6% is obtained. For a more detailed explanation see fig. 23. Here the MPP tracking algorithm takes into account that an increase or decrease of P DC can not only be the result of a previous variation of V DC, but may also have other causes (variations in irradiance), from time to time makes also a search into the other direction and if necessary remains on the same V DC. 34 32 3 6

24 th European Photovoltaic Solar Energy Conference, Hamburg, Germany, Sept. 29 New Procedure for Measuring Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters 25 2 15 1 5 INV3, new FW: Dynamic DC power at V MPP high = 297 V V DC(t) - 3 Min. expanded V MPP Low 24 1 2 3 Pattern: 1x (2/1H/2/1L) ime [min] η MPPdyn = 99.59% Fig. 23: Expanded detail of fig. 22 showing the operating principle of the MPP tracking algorithm. As the observation interval is only 2 s, the MPP algorithm can react much faster to changes of the MPP. 4. Conclusions Many tests performed so far at different inverters have shown, that dynamic MPP tracking behaviour can be improved considerably by an intelligent control software without affecting static MPP tracking. It could be demonstrated that at two inverters from different manufacturers by means of a mere improvement of the MPP tracking software without any changes at the inverter hardware very significant improvements of the dynamic MPP behaviour were possible. herefore every manufacturer should be able to implement analogue improvements without significant additional cost. he test procedures presented here with ramps of different steepness are appropriate to determine the quality and speed of the dynamic MPP tracking of an inverter. Due to the many different steepness values used during the test it should be possible to discover possible problems at most inverters. Somewhat disappointing is the fact, that at inverters with bad dynamic MPP tracking some problems exist in reproducing exactly the measured values of η MPPdyn at the different test patterns. It has been shown that this is a general problem, because the internal checks of MPP tracking in an inverter is always performed by means of a periodicity controlled by its internal clock. It is obvious that a synchronisation between the test equipment and this internal clock is not possible. However, during all tests performed so far reproducibility was very good at inverters with good dynamic MPP tracking. In the draft standard mentioned (FprEN553) also a turn-on / turn off test with a very slow variation of irradiance between.2% of G SC and 1% of G SC is described. Due to space limitations, in this paper the results of these tests are not discussed. With the existing and successfully commissioned equipment now available at the PV laboratory of BFH-I it is possible to perform similar dynamic MPP tests at inverters between 1 W and 1 kw as a service to interested manufacturers. 34 32 3 28 26 Important Notice Information contained in this paper is believed to be accurate. However, errors can never be completely excluded. herefore we disclaim any liability in a legal sense for correctness and completeness of the information or from any damage that might result from its use. Acknowledgements hanks go to all the institutions that gave us financial support. Without their support it would not have been possible to realise this challenging project. he work described in this paper was funded by the research fund of our school (BFH-I) and under a contract with the Swiss Federal Office of Energy (BFE) in the project "PV systems technology 27-21". Other parts of this project were co-funded by by Localnet AG (Burgdorf), GMS c/o BKW (Bern), BKW (Bern) and EBL (Liestal). References: [1] H. Haeberlin, L. Borgna, M. Kämpfer, U. Zwahlen: "otal Efficiency η tot A new Quantity for better Characterisation of Grid-Connected PV Inverters". 2 th EU PV Conf., Barcelona, Spain, June 25. [2] B. Burger: "Auslegung und Dimensionierung von Wechselrichtern für netzgekoppelte PV-Anlagen". 2. Symp. Photovoltaik, Staffelstein, 25. [3] H. Haeberlin, L. Borgna, M. Kämpfer, U. Zwahlen: "Measurement of Dynamic MPP-racking Efficiency at Grid-Connected PV Inverters". 21 st EU PV Conf., Dresden, Germany, Sept. 26. [4] H. Häberlin, L. Borgna, D. Gfeller, U.Zwahlen: "New PV Array Simulator of 1 kw: Results of first ests at a PV Inverter of 1kW". 24 th EU PV Conf., Hamburg, Germany, Sept. 29. [5] R. Bründlinger, N. Henze, H. Haeberlin, B. Burger, A. Bergmann, F. Baumgartner: "EN553 the new European Standard for Performance Characterisation of PV Inverters". 24 th EU PV Conf., Hamburg, Germany, Sept. 29. Further information and many publications about the research activities of the PV laboratory of BFH-I (former names: HI or ISB) on the internet: http://www.pvtest.ch 7