(or Climbing the Peak without Falling Off the Other Side ) Dave Edwards

Transcription

1 (or Climbing the Peak without Falling Off the Other Side ) Dave Edwards

2 Ripple Correlation Control In wind, water or solar alternative energy power conversion systems, tracking and delivering maximum power defies classic analog feedback methods because the source power is a non-monotonic function of the usual standard control inputs (such as voltage, current or power converter duty cycle). As the power peak is passed over from either direction the control function reverses its effect, turning feedback from negative to positive. With standard control methods, once the operating point goes even the slightest amount over the hump, it will rapidly slide down and away from its former maximum.

3 Ripple Correlation Control Typically software is used to implement some sort of nonlinear control method in an attempt to straddle the power peak. One of the most popular methods is the perturb and observe algorithm because it is both easy to understand and implement. But it is also very slow and may not be able to maintain its balance on a rapidly shifting peak. For grid tie applications, Ripple Correlation Control is a very fast MPPT method that makes use of the unavoidable 120 Hz perturbations about the maximum power point that result naturally from sourcing power into the 60 Hz line.

4 Ripple Correlation Control This elegant method was first studied by a Cal Tech grad student over 20 years ago using analog methods (e.g., multipliers ICs). Today this can all be done more easily digitally, but when compared to analog, this platform presents a few unique challenges (e.g., numerical dynamic range issues) as well as opportunities. This presentation will offer one possible version of the ripple correlation MPPT method that includes a new, previously unreported solution to the digital dynamic range issue.

5 Ripple Correlation Control High Pass with gain From PV Array 3m3 High Pass with gain L N The Ripple Correlation MPPT Method: Maximum power point tracking by the ripple correlation method uses the naturally existing 0% to 200%, 120 Hz perturbation of sourced power into the ac line (as filtered through the large energy storage capacitor) to dynamically detect and track the zero slope point of the power output curve of the photovoltaic panel. Anti-Alias Low Pass (10 khz) S&H 12 bit A/D Anti-Alias Low Pass (10 khz) S&H 12 bit A/D Anti-Alias Low Pass (10 khz) S&H 12 bit A/D Anti-Alias Low Pass (10 khz) S&H 12 bit A/D μp 25 khz Clk A 120 Hz, fixed phase, ac reference signal for slope detection is developed by high-pass filtering the ripple voltage that is present on the energy storage capacitor. The variable phase MPPT power ripple feedback signal is developed by sensing and multiplying together PV array current and voltage and high-pass filtering the resultant product. A ± dc phase error signal is recovered by multiplying together these two 120 Hz ac waveforms (the error output is zero at the maximum power point). Error signal strength is directly proportional to delivered power, so error gain must be divided by the output power command in order to keep overall loop-gain constant. Gain=1 2 pole HP F=100 Q= pole HP F=100 Q=0.85 Gain=1k5 Reference Waveform 2 f 90 1 pole LP F=30 Q=½ max 2kW Gain = 1 min 50W P CMD to Current Selling Routines Line Waveform Frequency Synchronization Signal Because of the inherent high efficiency of the switching power converter, the power command to the line current selling control section a priori can be set to within a few percent of measured PV array power. The small remaining error in command level is eliminated by summing this PV power feed-forward signal together with the MPPT error feedback signal. Note that if multiple maximum power points exist, this method may not necessarily lock onto the overall global peak power point, thus it will be necessary to periodically police the fast tracker solution with an additional (probably much slower) global power point algorithm. PV MPPT Section

6 But First Before we showcase tonight s big star, Let s set the stage with a little engineering background

7 MPPT What is it? What is Maximum Power Point Tracking? (and why is it used?) Alternative energy sources typically can provide only limited power so it is important to harvest the maximum available power at all times. Maximum power typically occurs only at one specific operating point (voltage and current). The maximum power point may move with time or other changing conditions.

8 The Solar Cell A Solar Cell is a typical limited power source. So how does it behave? It essentially looks like a diode... driven by a fixed (or slowly varying) current source.

9 Solar Cell Model LTspice easily models basic Solar Cell behavior. Insolation (incident sunlight) is modeled as a current source. A standard SPICE diode provides the basic I-V curve. Series and shunt resistance provide for easy adjustment of the slopes of the constant voltage and constant current portions of the curve. So show us the LTspice model!

10 Solar PV Array Model To model any number of whole panels, simply multiply the effect of a single cell (copying it over and over is unnecessarily burdensome both in terms of effort and run times). LTspice includes a series multiplier for diodes. But a generic diode multiplier is almost as easy.

11 49 Cell Panel (not real) 10A 200W 9A 8.2A 23.1V x 7.6A = 176W 180W 8A 160W 7A 140W 6A 120W 5A 100W 4A 80W 3A 60W 2A 40W 1A 20W 29.7V 0A 0V 4V 8V 12V 16V 20V 24V 28V 32V 36V V(pv) 0W

12 Performance Over Temperature 10A 9A 8A 7A -25 C (-13 F) 250W 225W 200W 175W 6A 5A 45 C (113F) 150W 125W 4A 3A 2A 1A 85 C (185F) 100W 75W 50W 25W 0A 0V 4V 8V 12V 16V 20V 24V 28V 32V 36V V(pv) Typical Single Panel Performance Curves as a Function of Cell Temperature Very Hot, Typical, and Very Cold (Starting) 0W

13 Locus of Max Power Points 2.4KW 2.2KW 2.0KW 1.8KW 1.6KW 1.4KW 1.2KW 1.0KW 0.8KW 0.6KW 0.4KW 0.2KW 0.0KW 0V 40V 80V 120V 160V 200V 240V 280V 320V 360V 400V V(pv) 14 Panel Maximum Power Point as a Function of Insolation at typical cell temperature (45 C), highlighting typical operating range

14 Worse Case MPPT Conditions 3.6KW 3.2KW 2.8KW 2.4KW 16 Panels, 635W/m 2, -5 C Voltage: 408 operate 2.0KW 1.6KW 12 Panels, 800W/m 2, 85 C Voltage: 203 operate 1.2KW 0.8KW 0.4KW 0.0KW 12 Panels, 200W/m 2, 45 C Voltage: 218 operate 16 Panels, 1000W/m 2, -25 C Voltage: 547 start (243 run) 0V 60V 120V 180V 240V 300V 360V 420V 480V 540V 600V V(pv) Panel Extreme Operating Conditions as a Function of Everything (Cell temperature indicated by color, Insolation indicated by size)

15 PV Shading Effects on MPPT Shading effects are eaily graphically illustrated with LTspice's hierarchical subcircuit capability. Run the model please!

16 PV Array - Summary The PV Array provides a limited amount of DC Power. There is only one Maximum Power Point and it can vary over a wide range (>2:1 in voltage and >10:1 in current). Fortunately, the Maximum Power Point normally changes very slowly (not necessarily true for other alternative power sources). Maximizing power harvest requires extracting dc power exactly at the maximum power point.

17 The AC Grid The Grid is more-or-less a constant voltage, single frequency ac source (60Hz). For a Grid-Tie Inverter to deliver maximum power into the grid, sourced current must be in phase with Grid voltage. Since power is V x I, sourced power will be a double frequency sine wave of zero to twice the average sourced power (this is true for any lossless inverter topology).

18 Mating the DC Array to the AC Grid To maintain PV Array harvest efficiency, the 120Hz AC portion of Grid power must be supplied by an auxiliary energy storage device (typically a very large capacitor). DC only 120Hz AC+DC Any Inverter 60Hz AC AC Grid

19 Effect of Energy Storage Capacitor Size on Power Harvest 2.490KW V(pv,x)*I(Vpv) 2.485KW 2.480KW 2.475KW C=1mF C=3mF (offset=+10w) 2.470KW 2.465KW 2.460KW C=10mF (offset=-10w) 2.455KW 2.450KW 2.445KW 2.440KW 2.435KW Effect of Cap size on power loss Voltage ripple from a 1mF cap across PV array would only cause about a 25W loss from potential peak power (about 1%) 2.430KW 9.3A 9.4A 9.5A 9.6A 9.7A 9.8A 9.9A 10.0A 10.1A 10.2A 10.3A 10.4A I(Vpv)

20 Maximum Power Point Tracking by the Ripple Correlation Method RCC MPPT was analyzed over 20 years ago. Two recent papers compare this method to many other known methods. 1) Comparison of the Performance of Maximum Power Point Tracking Schemes Applied to Single-Stage Grid- Connected Photovoltaic Systems, S. Jain and V. Agarwal, IET Electr. Power Appl., 2007, 1, (5), pp

21 Table from Paper #1

22 Table from Paper #2 2) Comparison of Photovoltaic Array Maximum Power Point Tracking Techniques, Trishan Esram, Student Member, IEEE, and Patrick L. Chapman, Senior Member, IEEE, IEEE Transactions on Energy Conversion, vol. 22, no. 2, June 2007, pp The Table follows below (next slide):

23 Major Characteristics of MPPT Techniques

24 Ripple Correlation MPPT The Ripple Correlation Method multiplies a reference double line frequency waveform (shifted 90 degrees) with the double frequency PV Array power ripple (V x I) after it passes through the energy storage capacitor (also causes 90 degrees of phase shift). At the maximum power point, these waveforms are 90 degrees out of phase, thus NO net dc results (power error is zero). Either side of the peak results in a monotonic error signal (review the double angle formulae).

25 The Ripple Correlation Method Simulated in LTspice

26 Error Signal Response to 0.5Hz 50% Cloud Event Okay, let s see the simulation.

27 Ripple Correlation Control From PV Array 3m3 L N High Pass with gain High Pass with gain Anti-Alias Low Pass (10 khz) Anti-Alias Low Pass (10 khz) Anti-Alias Low Pass (10 khz) Anti-Alias Low Pass (10 khz) μp S&H 12 bit A/D S&H 12 bit A/D S&H 12 bit A/D S&H 12 bit A/D 25 khz Clk Gain=1 2 pole HP F=100 Q= pole LP max 2kW Gain = 1 min 50W P CMD to Current Selling Routines 2 pole Gain=1k5 F=30 Q=½ HP F=100 Q=0.85 Reference Waveform Line Waveform Frequency Synchronization Signal 2 f 90 PV MPPT Section

28 Ripple Correlation Control Maximum power point tracking by the ripple correlation method uses the naturally existing 0% to 200%, 120 Hz perturbation of sourced power into the ac line (as filtered through the large energy storage capacitor) to dynamically detect and track the zero slope point of the power output curve of the photovoltaic panel.

29 Ripple Correlation Control A 120 Hz, fixed phase, ac reference signal for slope detection is developed by high-pass filtering the ripple voltage that is present on the energy storage capacitor. The variable phase MPPT power ripple feedback signal is developed by sensing and multiplying together PV array current and voltage and high-pass filtering the resultant product. A ± dc phase error signal is recovered by multiplying together these two 120 Hz ac waveforms (the error output is zero at the maximum power point). Error signal strength is directly proportional to delivered power, so error gain must be divided by the output power command in order to keep overall loop-gain constant

30 Ripple Correlation Control Because of the inherent high efficiency of the switching power converter, the power command to the line current selling control section a priori can be set to within a few percent of measured PV array power. The small remaining error in command level is eliminated by summing this PV power feedforward signal together with the MPPT error feedback signal.

31 Ripple Correlation Control Note that if multiple maximum power points exist, this method may not necessarily lock onto the overall global peak power point, thus it will be necessary to periodically police the fast tracker solution with an additional (probably much slower) global power point algorithm.

32 Thank You!