A COMPARATIVE SIMULATION ANALYSIS OF MAXIMUM POWER POINT TRACKING APPROACHES.

Transcription

1 A COMPARATIVE SIMULATION ANALYSIS OF MAXIMUM POWER POINT TRACKING APPROACHES Chee Lim Nge 1,2, Georgi Yordanov 1,2, Ole-Morten Midtgård 1, Tor Oskar Sætre 1, Lars Norum 2 1 University of Agder, Faculty of Engineering and Science, Grooseveien 36, 4876 Grimstad, Norway 2 Norwegian University of Science and Technology, Department of Electric Power Engineering, 7491 Trondheim, Norway chee.l.nge@uia.no ABSTRACT: The occurrence of multiple local maxima due to partial shading conditions poses a serious challenge to MPPT techniques. This paper compares the popular real-time MPPT algorithm with the scanning approach using PSPICE simulation. The MPPT methods are compared at dynamic partial shading conditions. The compromises between real-time and scanning methods are investigated. It is shown that the scanning approach is able to search for global MPPs but incurs power loss during the scanning period. A new strategy that reduces the scanning period by skipping parts of the voltage range is presented in this paper. Keywords: shading, inverter, photovoltaic 1 INTRODUCTION There have been a large number of publications on maximum power point tracking (MPPT) methods and the research interest has continuously grown in the recent past [1]. MPPT contributes greatly to the overall efficiency of photovoltaic (PV) inverters, which in turn is an important factor for the economic viability of the system. The occurrence of multiple local maxima due to the partial shading conditions poses a serious challenge to MPPT techniques. This problem is more profound in urban installations, which recently have increased in demand due to government subsidies. The reductions in energy yield due to the limitations of MPPT techniques under partial shading conditions are confirmed in reports based on field data [2-5]. If we consider only true MPPT techniques that are not dependent on a specific PV module, they can be grouped into two categories: the real-time and the scanning approaches. The real-time MPPT techniques include the popular perturb-and-observe [6-8] and the incremental-conductance (IncCond) [9, 10] methods. They have fast responses as they are able to continuously track peak power under rapidly changing atmospheric conditions. However, such approaches have no means of detecting global peak power and they only move the operating point to the nearest local maximum point. The scanning techniques are capable of tracking the global peak power because they periodically search throughout the full possible operating range. One straightforward approach is to characterize the I-V curve using current sweep waveform [11]. The converter is disabled during the sweep in order to reduce the sweep period. Advanced search algorithms using a Fibonacci sequence [12] and particle swamp optimization technique [13] are proposed to reduce the amount of operating points for MPP evaluation. However they do not necessarily reduce the scanning period because such algorithms need to randomly search through the possible range before reaching steady state. Since scanning techniques are initiated periodically, they lose track of the MPP between the scanning periods. Reference [14] tracks MPP using perturb-and-observe in real-time but scans for global peak power occasionally when sudden change in irradiance occurs. Since it stops the scanning when the subsequent peak is lower than the previous peak, this approach would miss a distant global maximum point at higher voltages. On the other hand, a two-stage approach that first moves the operating point to the vicinity of the global peak power on the PV load line tends to miss global peak power that lies to the left side of the load line [15]. It is obvious that there are compromises amongst the MPPT approaches under dynamic atmospheric and partial shading conditions. This paper provides a comparative analysis between real-time and scanning approaches. Moreover, a new scanning strategy is proposed, which reduces the time for tracking global peak power by skipping parts of the voltage range. 2 SIMULATION SETUP Figure 1 is the popular two-stage inverter configuration suitable for grid-connected PV systems. The boost converter stage sets the operating point of PV module and it is controlled by the MPPT control unit. The full-bridge inverter keeps the dc-link voltage between the two stages constant by adjusting power flow into the grid. For this simulation analysis, the power stage is simplified to a boost converter with constant voltage load. This is assuming that the full-bridge inverter has fast response. Photovoltaic Boost Converter MPPT Control Full-bridge Inverter Control Figure 1: Two-stage grid-connected PV inverter Grid The PV system is modeled using the circuit-oriented simulation software, PSPICE. It is widely used in power electronics applications because of its effectiveness in simulating analog circuits. In order to illustrate the tradeoffs amongst various MPPT approaches, the following techniques are compared: a. IncCond is selected to show the fast dynamic response of real-time approaches. b. Combination of IncCond with a simple scanning approach that utilizes voltage sweep waveform to periodically characterize the I-V curve and track the global peak power.

2 c. Combination of IncCond with the new proposed scanning method, which skips parts of the voltage range 2.1 Boost converter The schematic consisting of the main blocks of the boost converter PSPICE model is shown in Figure 2. The power stage is considerably simplified to optimize the computational efficiency. The power devices are represented by ideal diodes and switches. The full-bridge inverter stage is modeled as a voltage source connected to an R-L filter that simulates the transient response. Ctrl1 VIN MPPT VREF Ctrl2 VREF PWMA2 PWMA2 shading condition as depicted in Figure 4. The characteristic curves correspond to the PV string model in Figure 3. The global peak shifts from 189W at 97V to 207W at 74V when I PV1 increases from 2.5A to 3.0A. This corresponds to irradiance variation of 100W/m 2 within 100ms. The results are presented in Section 4. IPV1 TD = 500m TF = 100m TR = 100m I1 = 2.5 I2 = 3.0 IPV5 2Adc Dpv1 Dpv2 Dpv3 Dpv4 Dpv5 Figure 3: PV string model Dbp1 Dbp2 Dbp3 Dbp4 Dbp5 Rsr1 PARAMETERS: vocm = 22.3 is = {5/nVT} Rsh1 nd = 1 ns = 36 Rsr5 Rsh5 PV_POS PV_NEG PV1 PV VIN PV_NEG PV_POS Lin Cin Rcb Rlin SW VoltMode S1 + D1 D1_ D2_ VOUT S2 1 2 PWMA2 + D2 Rinv Linv Cout Vdc-link 0 Figure 2: Main schematic of the PSPICE model The VOLTMODE hierarchical block is the single loop voltage mode controller that generates PWM signals by comparing the input voltage, with the reference signal, VREF. The MPPT control unit measures the current and voltage of the PV modules, and tracks the MPP by sending VREF to the VOLTMODE controller. Considering that each power switch is connected to an anti-parallel diode, the synchronous boost converter always operates in the continuous conduction mode. The input voltage control of the boost converter becomes similar to buck converter output voltage control if the switching duty ratio of S1 is used as the control output. As a result, the controller is considerably simplified since the duty-cycle to input-voltage transfer function is always linear. 2.2 PV modules Figure 3 depicts the PSPICE model of a PV string with 5 modules connected in series. Each PV module consists of 36 cells connected in series. It has an opencircuit voltage of 22V and a short-circuit current of 5A. The PV module is modeled as a current source, I PV connected in parallel with a diode, D PV. For simplicity, the irradiance level is directly proportional to the photocurrent, I PV. The series and shunt resistances, R SR and R SH model the fill factor losses. It is assumed that the first 4 modules have identical characteristics and irradiance hence I PV1, R SR1 and R SH1 are lumped parameters. For simplicity, there is only one bypass diode (D BP1-5 ) connected in parallel with each module since the irradiance is assumed to be uniform on each module. The MPPT is evaluated under the dynamic partial Figure 4: Power-voltage characteristic of a PV string 3 PROPOSED MPPT METHOD As previously discussed, it is necessary to scan the P- V curve of the PV modules in order to find the global peak. Figure 5 shows the simplified P-V curves that consist of two local maxima under partial shading conditions. It represents a PV string that is illuminated with different irradiance levels. The blue dashed lines correspond to constant current where the photocurrent I PV1 is larger than I PV2. The valid input voltage range of the boost converter is between V MIN and V MAX.

3 P MPP I PV1 I PV2 are shown in Figures 6 and 7: a) Incremental-conductance, IncCond. b) Combination of IncCond with voltage sweep. c) Combination of IncCond with the new scanning method proposed in the present paper. JUMP to V EST V MIN (a) V MAX P MPP I PV1 V EST >V MAX I PV2 V MIN V MAX (b) Figure 5: Proposed MPP voltage estimation approach The proposed scanning trajectory is indicated as the green dashed line in Figure 5. The sweep starts at V MIN and the peak power is continuously updated while the operating point moves towards higher input voltage. The instantaneous PV current, I PVx decreases as the input voltage increases. Once a local maximum power P MPP1 is detected at lower voltage level, parts of the P-V curve at higher voltage can be avoided. For a given I PVx the lowest voltage at which a higher MPP is possible can be estimated from equation (1). This is to assume the ideal condition where I PVx remains constant up to the MPP. V P MPP1 EST = (1) IPVx As shown in Figure 5a, parts of the curve below P MPP1 are avoided using the JUMP to V EST operation. It should be noted that JUMP to V EST operation does not necessarily occur when current is equal to I PV2. As long as I PVx is less than I PV1, JUMP to V EST operation might occur. The operating point continues to move towards higher voltage level as long as V EST does not exceed the upper voltage range of the converter, V MAX. Referring to Figure 5b, the scan stops when V EST is larger than V MAX. The lower level local MPP at higher voltage is avoided in this case. It is important to note that the proposed algorithm assumes static P-V curve during the scanning. 4 RESULTS PSPICE simulation results that compare different MPPT methods on the system presented in Section 2.2 Figure 6: MPPT comparison: I PV1 increases to 3.0A Method b) and c) search for the global peak using a scanning approach and then they move the input voltage to the vicinity of the global MPP voltage after the sweep. They track MPP in real-time using the IncCond method between the scanning periods. The scanning mechanisms are initiated every 1s and they sweep the input voltage at the slew rate of 0.15V/ms. The boost converter minimum input voltage is selected at 70V and this is where both scanning operations start. The voltage sweep method of b) is a simple periodic sweep of PV voltage that stops when the PV current drops to less than 0.5A. The proposed method in c) JUMPs to V EST every 40ms and it stops scanning when V EST is larger than V MAX.

4 proposed method reduces the energy loss during scanning period by approximately 65%. IV. JUMP to V EST is initiated 40ms after t 2 and this is where V IN waveforms of the proposed method lead those of method b) V. For method b), P IN drops to approximately 50W, when the operating point moves towards opencircuit voltage before scanning stops at t 4. This results in a considerable power loss. For the proposed method, P IN does not need to deviate significantly from MPP. There are tradeoffs in selecting the interval of the scanning mechanism. Hypothetically, it is necessary to initiate scanning immediately after global peak shifts to another local maximum point. It needs to scan again if the global peak varies during scanning or immediately after that. On the other hand, there are power losses during scanning as P IN deviates from MPP. Also, scan should be avoided if it is not under partial shading conditions. One possible solution is to initiate the scanning mechanism adaptively when a considerable variation in local MPP is detected. This is an indication of the shift of global peak. However, the MPP variation due to irradiance change could be too small and it is difficult to distinguish from IncCond perturbation. The P IN increment in Figure 6 at t 1 is less than 10W. 5 CONCLUSIONS Figure 7: MPPT comparison: I PV1 decreases to 2.5A Figures 6 and 7 show that method c) yields the highest average input power P IN to the boost converter. Other important observations from the comparative simulation results are: I. After the irradiance (i.e. I PV1 ) changes at t 1, the global MPP voltage shifts from 74V to 97V or vice versa. Both methods b) and c) are able to calibrate the tracking voltage to the global peak but method a) does not. II. The scanning starts at t 2 after I PV1 changes. The three MPPT techniques are identical between t 1 and t 2 because they track MPP using the same approach i.e. the IncCond method. III. The power P IN deviates from global peak during the scanning periods. For method b) the scanning period is between t 2 -t 4 while the proposed method is between t 2 -t 3. At I PV1 = 2.5A, there is about 100ms difference between method b) and the proposed method. At I PV1 = 3.0A, the proposed method completes scanning faster than method b) by about 150ms. Compared to method b), the The comparative simulation results verify that the real-time approaches such as IncCond are not able to track global MPP. It is shown that IncCond locks to local maximum point as it has no means of detecting global peak power. The global peak can be found by scanning through the voltage range. However, if the power losses of the scanning mechanisms are significant, the time interval between scanning instants needs to be relatively large. This in turn results in poor response if the global peak shifts among local maxima rapidly. On the other hand, it is necessary to scan immediately after global MPP moves to another peak. The proposed MPPT method is evaluated using a simplified PV string model that operates under dynamic partial shading condition. The capacitances of the PV modules and the interconnect inductances are not considered in the modeling. Compared to the simple voltage sweep approach, the proposed method reduces the energy loss during scanning period by approximately 65%. The new method prevents the scanning from drifting to open-circuit voltage level where the power is low, and this is done by comparing the next possible MPP voltage, V EST to V MAX. 6 REFERENCES [1] T. Esram and P. L. Chapman, "Comparison of Photovoltaic Array Maximum Power Point Tracking Techniques," Energy Conversion, IEEE Transactions on, vol. 22, pp , [2] R. Bruendlinger, B. Bletterie, M. Milde, and H. Oldenkamp, "Maximum power point tracking performance under partially shaded PV array conditions," in 21st European Photovoltaic Solar Energy Conference, Dresden, Germany, Sept. 2006, pp

5 [3] Pablo Sanchis, Jesús López, Alfredo Ursúa, Eugenio Gubía, and Luis Marroyo, "On the testing, characterization, and evaluation of PV inverters and dynamic MPPT performance under real varying operating conditions," Progress in Photovoltaics: Research and Applications, vol. 15, pp , [4] A. Woyte, J. Nijs, and R. Belmans, "Partial shadowing of photovoltaic arrays with different system configurations: literature review and field test results," Solar Energy, vol. 74, pp , [5] A. Catani, F. Gómez, R. Pesch, J. Schumacher, D. Pietruschka, and U. Eicker, "Shading losses of building integrated photovoltaic systems," in 23rd European Photovoltaic Solar Energy Conference, Valencia, Spain, Sept. 2008, pp [6] E. Koutroulis, K. Kalaitzakis, and N. C. Voulgaris, "Development of a microcontroller-based, photovoltaic maximum power point tracking control system," Power Electronics, IEEE Transactions on, vol. 16, pp , [7] D. Sera, T. Kerekes, R. Teodorescu, and F. Blaabjerg, "Improved MPPT Algorithms for Rapidly Changing Environmental Conditions," in Power Electronics and Motion Control Conference, EPE-PEMC th International, 2006, pp [8] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, "Optimization of perturb and observe maximum power point tracking method," Power Electronics, IEEE Transactions on, vol. 20, pp , [9] O. Wasynezuk, "Dynamic Behavior of a Class of Photovoltaic Power Systems," Power Apparatus and Systems, IEEE Transactions on, vol. PAS-102, pp , [10] K. H. Hussein, I. Muta, T. Hoshino, and M. Osakada, "Maximum photovoltaic power tracking: an algorithm for rapidly changing atmospheric conditions," Generation, Transmission and Distribution, IEE Proceedings-, vol. 142, pp , [11] M. Bodur and M. Ermis, "Maximum power point tracking for low power photovoltaic solar panels," in Electrotechnical Conference, Proceedings., 7th Mediterranean, 1994, pp vol.2. [12] M. Miyatake, T. Inada, I. Hiratsuka, Z. Hongyan, H. Otsuka, and M. Nakano, "Control characteristics of a fibonacci-search-based maximum power point tracker when a photovoltaic array is partially shaded," in Power Electronics and Motion Control Conference, IPEMC The 4th International, 2004, pp Vol.2. [13] M. Miyatake, F. Toriumi, T. Endo, and N. Fujii, "A Novel maximum power point tracker controlling several converters connected to photovoltaic arrays with particle swarm optimization technique," in Power Electronics and Applications, 2007 European Conference on, 2007, pp [14] H. Patel and V. Agarwal, "Maximum Power Point Tracking Scheme for PV Systems Operating Under Partially Shaded Conditions," Industrial Electronics, IEEE Transactions on, vol. 55, pp , [15] K. Kobayashi, I. Takano, and Y. Sawada, "A study of a two stage maximum power point tracking control of a photovoltaic system under partially shaded insolation conditions," Solar Energy Materials and Solar Cells, vol. 90, pp , 2006.