Submodule Differential Power Processing in Photovoltaic Applications

Transcription

1 Submodule Differential Power Processing in Photovoltaic Applications Shibin Qin Robert Pilawa-Podgurski University of Illinois Urbana-Champaign 1 This research is funded in part by the Advance Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award DE-AR

2 Outline Introduction PV characteristic mismatch problem Distributed power electronics Differential power processing (DPP) Control Maximum power point tracking in DPP The centralize approach The distributed approach Hardware Implementation Experimental Testbed 2

3 PV Module Current To the grid Each PV module consists of 3 submodules PV modules (submodules) are connected in series for voltage stacking Central inverter 3

4 PV Module Mismatch Current To the grid PV characteristic mismatch Partial shading Manufacturing variations Non-uniform aging Maximum power point tracking (MPPT) at submodule level Central inverter 4

5 Sub-module Level MPPT Solutions Sub-module micro-inverter Sub-module micro-inverter is prohibitive due to cost DC optimizer has to process the full power of each sub-module R. Pilawa-Podgurski, D. Perreault, Sub-Module Integrated Distributed Maximum Power Point Tracking for Solar Photovoltaic Applications, IEEE TPELS, June

6 Differential Power Processing (DPP) 30 Differenential Power Bulk Power 20 Power [W] 10 0 DPP system PV Sub-Module DPP converters process only the differential power DPP is advantageous over DC optimizer in terms of system efficiency, converter power rating and ease of integration into existing design P. Shenoy, K. Kim, B. Johnson, and P. Krein, Differential power processing for increased energy production and reliability of photovoltaic systems, IEEE TPELS, June

7 System Efficiency and Power Rating DC optimizer DPP Device DC optimizer DPP Total sub-module power [W] Total power processed [W] Average converter efficiency [%] 95% 85% Power loss in DC stage [W] DC stage conversion efficiency [%] 95% 98.9% 7

8 Ease of Integration and Scaling DC optimizer DPP 8

9 Outline Introduction PV characteristic mismatch problem Distributed power electronics Differential power processing (DPP) Control Maximum power point tracking in DPP The centralize approach The distributed approach Hardware Implementation Experimental Testbed 9

10 Maximum Power Point Tracking MPP at 100% irradiance Power [W] MPP at 50% irradiance MPP at 80% irradiance Current [A] MPPT algorithm: find (D 1, D 2, I m ) such that each sub-module operates at their MPP point (V i, I i ) 10

11 Two Loop Control D1 D2 for given Separation of two control loop Slow loop: inverter Fast loop: DPP converters 11

12 The Centralized Approach DPPs Maximize V m Slow loop: inverter Fast loop: DPP converters DPP control objective: Inverter updates I m Converter operation timeline 12

13 The Centralized Approach PV module characteristics Real-time power input to the micro-inverter Micro-inverter trapped by local maximum before DPP turned on DPP smoothed out the P-V curve and improved power output 13

14 The Distributed Approach Requiring no local current sensing Requiring only local voltage measurement and neighbor-toneighbor communication 14

15 The Distributed Approach 15

16 The Distributed Approach 16

17 The Distributed Approach 3-submodule, 2-DPP system, insolation profile:100%,80%,50% 17

18 Outline Introduction PV characteristic mismatch problem Distributed power electronics Differential power processing (DPP) Control Maximum power point tracking in DPP The centralize approach The distributed approach Hardware Implementation Experimental Testbed 18

19 PV Junction Box Integration Separate enclosure represents a significant cost 19

20 Hardware Implementation Small passive component due to high switching frequency No separate enclosure necessary for distributed power electronics Converter peak efficiency above 95% 20

21 Outline Introduction PV characteristic mismatch problem Distributed power electronics Differential power processing (DPP) Control Maximum power point tracking in DPP The centralize approach The distributed approach Hardware Implementation Experimental Testbed 21

22 Solar Emulation Controllable and repeatable indoor solar experiments Preserving the dynamics of a true PV module 22

23 Field Test Illinois Center for a Smarter Electric Grid Ongoing field test to verify the proposed technology in real irradiance Tests including MPPT tracking efficiency, system reliability, etc. 23

24 Concluding Remarks DPP Architecture Low power rating High system efficiency High reliability Easy to integrate DPP MPPT Control: Centralized and distributed solutions Minimum communication No local current sensing True MPPT Hardware Implementation Small footprint, junction box integration High converter efficiency Experimental Testbed Good experimental platform Ready to support more solar research 24

25 Questions Questions? Contact: Shibin Qin 25

26 Acknowledgement This research is funded in part by the Advance Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award DE-AR

27 PV Emulation 27

28 Sensing Small Current Change Direct signal scaling 0~5A 10bbb AAA 5A 5mm/AAA rrrrrrrrrr 1024 Moving window A 10 bbb AAA mm/aaa rrrrrrrrrr 1024 C. Barth, R. Pilawa-Podgurski, Dithering Digital Ripple Correlation Control with Digitally- Assisted Windowed Sensing for Solar Photovoltaic MPPT (Thursday 4:10pm, Session T34) Figure credit: Christopher Barth 28

29 Experimental Setup PV module: Solarworld Sunmodule 235 Poly Micro-inverter: Solarbridge Pantheon II S. Qin, K. Kim, R. Pilawa-Podgurski, Laboratory emulation of a photovoltaic module for controllable insolation and realistic dynamic performance (PECI 2013) 29

30 Hardware Specifications 30

31 Sensing Small Current Change Signal amplification is limited by DC current value Example: DC current: 5A Current change: 5mm ADC range: 3.3V 5mm 3.3mm Subtract DC value, amplify only AC value Module Current I m Bias point shift I bbbb C. Barth, R. Pilawa-Podgurski, Dithering Digital Ripple Correlation Control with Digitally- Assisted Windowed Sensing for Solar Photovoltaic MPPT (Thursday, Session T34) Figure credit: Christopher Barth 31

32 Hardware Cost Component Cost Gate Driver (IR2101SPBF) $0.50*2=$1.00 MOSFET (PSMN023) $0.14*4=$0.56 Mirco-controller(STM32F03) $0.70 Linear Regulator (LD2981AB) $0.20 Filter Capacitor $0.08*8*2=$1.28 Inductor $0.57*2=$1.14 Bypass Diode -$0.3*3=-$0.9 Total $

33 Current Sensing Circuit V sssss : scaled full current signal (DC+AC) V bbbb : DC bias provided by µ-controller PWM V wwwwwwww = G(V sssss V bbbb ): Remove most of the DC component Further amplify AC component µ-controller adjusts V bbbb to center V wwwwww in ADC range 33

34 34

35 DPP MPPT Algorithm: States DPP converter states: Converter 1 acquires x 2 k, z z [k] through neighborneighbor communication, and vice versa

36 DPP MPPT Algorithm: Input V 0 is maximized when: Define: Input:

37 Measuring Partial Derivative: P&O

38 DPP MPPT Algorithm: Update Update function:

39 Simulation 39

40 Experimental Result DPP converters' duty ratios converges from non-optimal initial values Duty ratio changes after every micro-inverter MPPT perturbation 40

41 Experimental Result PV module power loss: sum of sub-module maximum power minus micro-inverter input power PV module power loss includes: tracking losses DPP converter losses but does not include: conversion loss in the micro-inverter Irradiance Condition (normalized) PV module power loss without DPP PV module power loss with DPP 100%, 80%, 60% W 2.87 W 100%, 60%, 100% W 2.66 W 60%, 100%, 60% W 2.73 W 41