COMPARATIVE STUDY OF MAXIMUM POWER POINT TRACKINGTECHNIQUES FOR PHOTOVOLTAIC SYSTEMS
M. C. Cavalcanti, K. C. Oliveira, G. M. S. Azevedo, F. A. S. NevesDepartment of Electric Engineering and Power systems, UFPE. Recife - PE - Brasil
Emails: firstname.lastname@example.org, kleber email@example.com, firstname.lastname@example.org, email@example.com
Abstract This paper presents a comparative studyamong maximum power point tracking methods forphotovoltaic systems. The comparison takes intoaccount steady state error, dynamic response andefficiency in a large power range. In special, anextensive bibliography and a classification of manymaximum power point tracking methods is presented.Computational simulations with fast changes in the solarirradiance have been done and the best maximum powerpoint tracking technique is chosen. Experimental resultscorresponding to the operation of a photovoltaic convertercontrolled by a digital signal processor are also presented.
Keywords - Energy conversion, Photovoltaic power sys-tems, Solar energy, Tracking.
Nowadays, the requirement in generating electric energyhas lead to an intensive research of alternative ways of gen-eration. One of the possible ways of electric energy generationis the Photovoltaic (PV) energy. PV energy has great potentialto supply energy, since it can be considered a clean andpollution free source while the PV panels are generatingenergy. The main drawbacks are associated with the impacton the environment because of the high energy used duringthe panels fabrication process and the lifetime of the panelsthat is between 20 and 30 years. Other drawbacks are theinitial installation cost and the energy conversion efficiency.To overcome some of these problems, it is important tooperate the photovoltaic system near the Maximum PowerPoint (MPP) to increase the efficiency of photovoltaic arrays.If Maximum Power Point Tracking (MPPT) techniques areused in PV systems, it can be generated more power with thesame number of modules. These techniques allow the modulegenerate its maximum power, having a high level of utilizationof its generation capability. However, to improve the energycaptation it is necessary to have a converter between the PVpanels and the load or grid. This converter will have a limitedand variable efficiency in according to the load conditions andthe solar irradiance. Therefore the global efficiency (includingPV array and converter efficiencies) is a better parameter forthe whole system, but it would be necessary to compare thelosses produced by the converter switches. The purpose of thispaper is to compare the algorithms of maximum power pointtracking and the converter efficiencies are not considered.
Manuscript received October 9, 2002; revised May, 2, 2003. This area willbe used only by the Editor and Associate Editors. Please, the edition in thisarea is not permitted to the authors.
Various methods of MPPT have been considered in PVsystems. The methods may be classified as: off-line tech-niques  and on-line techniques -. The off-linetechniques require a PV array model and the measurementof temperature and solar irradiance. The on-line techniquesdo not require the measurement of temperature and solarirradiance. In addition, they do not need the PV array model.
Among the most desirable features in MPPT techniques arethe following : Stability Fast dynamic response Small steady state error Robustness to disturbances Efficiency in a large power range.The on-line techniques have been shown as more efficient
than the off-line techniques in terms of desirable features inMPPT methods. The on-line techniques may be classified as:Constant Voltage (CV), Perturbation and Observation (PO),and Incremental Conductance (IncCond). Some variations ofthese methods have been also presented in literature.
In , it was shown that the MPP voltage of a PV array isclose to a fixed percentage of the arrays open circuit voltage.In the MPPT technique, the converter is disconnected and theopen circuit voltage is measured at regular sampling rates . The energy wasted by the sampling of the open circuitvoltage is considered negligible in . This method wasdefined as Constant Voltage (CV) technique in  and it hasbeen used in some PV systems .
The PO method is often used in many PV systems -. PO techniques operate by perturbing the reference valuewith specific sampling rates . These techniques presentslow dynamic response and steady state error. A choice ofhigh values of perturbation provides a fast tracking for theMPP voltage, but it has large oscillations. If the perturbationhas a low value, the MPPT will be slower, but it will havesmall oscillations around the MPP. In addition, with fastchanges of irradiance and temperature, the PO technique cantrack a wrong point. In , it was proposed an implemen-tation of a PO method that instantaneous values of currentand voltage are used to determine the direction of the nextperturbation. This solution reduces the problems related tothe PO techniques.
A popular variation of the PO method  is based on therelationship of the PV array output power and the switchingduty cycle. This method is defined as the Hill Climbing (HC)technique in . When it happens a fast variation in theenvironment conditions, it can be tracked a wrong voltagepoint instead of a point that means the MPP, creating anerror in the algorithm. In other words, the algorithm will try
Eletrnica de Potncia, vol. 12, no. 2, Julho de 2007 163
to lead the array voltage to the MPP voltage of the curvecorresponding to the previous solar irradiance. This problemcan be also caused by a wrong choice of the sampling rate.A solution can be the best adjustment of the sampling rate and the best adjustment of the perturbation (incrementor decrement) in relation to the sampling rate , both inaccordance with the dynamics of the converter. A ModifiedAdaptive Hill Climbing (MAHC) technique has automaticparameter tuning to satisfy the requirements of fast dynamicresponse and small steady state error .
The Incremental Conductance (IncCond) technique iswidely used in PV systems -. The voltage of the MPPis tracked to satisfy dP/dV=0 . The parasitic capacitancemethod uses the capacitances of the PV array to improvethe IncCond technique . A method which improves theIncCond technique by inserting a test signal in control inputwas proposed in .
In , it was proposed a technique that determines theMPP of a PV array for any temperature and solar irradianceusing a tolerable power error. At each sample, the differencebetween the reference value and operating power of the PVarray is calculated and compared with the assumed MPP error.A MPPT control scheme for PV array based on a principleof power equilibrium at dc link is proposed in . Theproposed scheme does not need detection or calculation ofthe power. A two-mode MPPT control method combiningthe CV and IncCond techniques was proposed to improveefficiency of the PV power generation systems at differentirradiance conditions . A method of locating the MPPbased on injecting a small sinusoidal perturbation into theswitching frequency and comparing the ac component andthe average of the array terminal voltage was proposed in. A MPPT method in combination with one-cycle controlfor PV power generation was proposed in . In , itwas used an algorithm with two stages of operation. In thefirst stage, variable large steps allow fast tracking when thePV voltage is far from the MPP voltage. Around the MPPvoltage, any technique using fixed step can be used to trackthe MPP.
Due to the vast number of MPPT techniques with some-times contradictory performance claims, their comprehensivestudy still seems to be appropriate. The algorithms have beenverified on a PV system modeled in Matlab. Many simulationsresults are presented and the characteristics determined in thisstudy are summarized in comparative tables.
II. PHOTOVOLTAIC SYSTEM CONFIGURATIONS
The PV array has the equivalent circuit shown in Fig.1. Usually the shunt resistance is very large and the seriesresistance is very small . Therefore, the resistances maybe neglected to simplify the analysis. The characteristic of aPV array is given by the following equation -:
I = Ig Isat[
) 1 ] (1)where V is the PV array output voltage, I is the PV arrayoutput current, Ig is the generated current under a givenirradiance, Isat is the reverse saturation current, q is the
Fig. 1. Equivalent circuit of the PV array.
charge of an electron, A is the ideality factor for a p-n junction, k is the Boltzmanns constant and T is thetemperature (K).
The reverse saturation current and the generated currentof the PV array vary with temperature in according to thefollowing equations :
Isat = Ior[
Isc + KI(
T Tr) ] S
where Ior is the saturation current at Tr, Tr is the referencetemperature, EGO is the band-gap energy of the semiconduc-tor used in array, KI is the short-circuit current temperaturecoefficient and S is the irradiance in mW/cm2.
Two systems were used to test the MPPT techniques. Thefirst study is for a stand alone PV system using a 8.3kW PVarray consisting of twelve parallel connections of seven panelsconnected in series. Each of the solar panels has a maximumpower rating of 99W, which occurs at a rated voltage of17.7V and a rated current of 5.6A. The panels have an opencircuit voltage of 22V and a short circuit current of 6.3A.The characteristic of the PV array is shown in Fig. 2 forvariations in the solar irradiance (S). The PV array usuallygenerates energy in low voltage and depending of the powerto process, this characteristic can represent an inconvenient.In these cases a boost converter can be utilized because of itshigh efficiency and its small number of components .Figure 3 presents the stand alone PV system using a boostconverter. The second study is for a grid connected PV systemusing a 1.9kW PV array consisting of twenty four panelsconnected is series. Each of the solar panels has a maximumpower rating of 79W, which occurs at a rated voltage of 16.5Vand a rated current of 4.8A. The panels have an open circuitvoltage of 20.8V and a short circuit current of 5.2A. BothPV systems were tested in simulation by using Matlab andthe grid connected PV system shown in Fig. 4 is also used tovalidate the simulation through the experimental results. Thepanels were modeled by using equations (1), (2) and (3).
III. CONSTANT VOLTAGE
The MPP voltage (VMPP ) of a PV array is close to afixed percentage of the PV arrays open circuit voltage (VOC).The relation VMPP /VOC is usually around 76% . Initially,VOC is measured by setting the PV array current to be zero.In this way, the VMPP is adjusted for 76% of VOC . Thisvalue of VMPP is kept for a period of time until anothersample occurs. The MPPT technique samples VOC at regular
164 Eletrnica de Potncia, vol. 12, no. 2, Julho de 2007
0 20 40 60 80 100 120 140 1600
PV array voltage (V)
Fig. 2. Characteristic diagram of the PV array.
Fig. 3. Stand alone PV system using a boost converter.
samples  and the energy wasted by the sampling of VOCis considered negligible in . However, this considerationshould be evaluated. Another problem of this technique is thatthe MPP is not always located at 76% of the VOC , increasingthe steady state error . Two different sample rates are usedto estimate the efficiency. Using a low constant sample rate,the reference for VMPP is changed more frequently allowingbetter tracking while the system is connected to the PV array.However, the energy wasted by the sampling of VOC will bemore significant since the PV array current will go to zeromany times.
IV. PERTURBATION AND OBSERVATION
The PO technique compares the power of the previousstep with the power of the new step in such a way that
Fig. 4. Grid connected PV system using a three-phase inverter.
P(k) = V(k) * I(k)
P(k) > P(k-1)
V = V + Vref ref
D V = V - Vref ref
D V = V + Vref ref
DV = V - Vref ref
V(k-1) = V(k)
I(k-1) = I(k)
Sense V(k), I(k)
Fig. 5. The flowchart of the PO technique.
it can increase or decrease the voltage or current -.This method changes the reference value by a constant factorof current or voltage. It moves the operating point towardthe MPP by periodically increasing or decreasing the arrayvoltage or current. The PO method works well when theirradiance does not vary quickly with time. However, withthis method the power oscillates around the MPP in steadystate operation and it is not good when there are fast variationsof temperature and irradiance.
The flowchart of the PO technique operating by varyingthe PV reference voltage is shown in Fig. 5 . FromFig. 2, it can be seen that incrementing (decrementing) thevoltage increases (decreases) the power when operating on theleft of the MPP and decreases (increases) the power when onthe right of the MPP. Therefore, if there is an increase inpower, the subsequent perturbation should be kept the sameto reach the MPP and if there is a decrease in power, theperturbation should be reversed . A choice of high valuesof perturbation (V ) provides a fast tracking for the MPPvoltage. If the perturbation has low value, the MPPT is slower,but it has small oscillations around the MPP.
The problem of oscillation around the MPP can be mini-mized by comparing the parameters of two preceding cycles.If the MPP is reached, the perturbation stage is bypassed. This technique is named as Modified Perturbation andObservation (MPO) in this paper.
V. HILL CLIMBING
The HC method is based on the relationship of the PV arraypower and switching duty cycle -. The flowchart isshown in Fig. 6. Slope is a program variable with either 1or -1, indicating the direction to increase the output power,while "a"represents the increment step of duty cycle, whichis a constant number between 0 and 1, and D and P representthe duty cycle value for the switch in Fig. 3 and power level,respectively. With rapidly changing atmospheric conditions,the same problem of the PO can happen. The MAHC methodincludes automatic parameter tuning to have good dynamic
Eletrnica de Potncia, vol. 12, no. 2, Julho de 2007 165
P(k) = V(k) * I(k)
P(k) = P(k-1)
D(k) = D(k-1) + a*Slope
Sence V(k), I(k)
Fig. 6. The flowchart of the HC technique.
P(k) = V(k) * I(k)
P(k) = P(k-1)
D(k) = D(k-1) + a(k)*Slope
Sense V(k), I(k)
DP(k) = P(k) - P(k-1)
| P/a(k-1) | > eD
Slope = -1 Slope = 1
No No Yes
Fig. 7. The flowchart of the MAHC techniq...