Professional Documents
Culture Documents
com
ScienceDirect
Solar Energy 108 (2014) 467–478
www.elsevier.com/locate/solener
Received 10 May 2014; received in revised form 20 July 2014; accepted 2 August 2014
Abstract
Photovoltaic (PV) arrays consist of series and parallel connections of PV modules. Difference in current–voltage (I–V) characteristics
among a batch of modules form an array causes power losses in PV systems referred to as mismatch losses. These power losses are
conventionally reduced by module sorting techniques which sort modules based on an I–V parameter such as short circuit current,
current at maximum power or maximum power. This work introduces a new method that employs genetic algorithm (GA) to find an
arrangement of modules in an array which minimizes mismatch losses more effectively than conventional methods do. Extensive
simulations are applied to adapt a GA to the problem of mismatch losses, find the arrangement and demonstrate its superiority over
module sorting techniques in terms of mismatch losses decrement and energy yield increment. Instructions for practical application
of the suggested method are also provided.
Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2014.08.005
0038-092X/Ó 2014 Elsevier Ltd. All rights reserved.
468 S. Shirzadi et al. / Solar Energy 108 (2014) 467–478
Publication Citation
No
No
Year Year
Fig. 1. Publications on mismatch power losses (left) and citations of mismatch power losses (right).
The same way PV modules are composed of PV cells. This (BL) and total cross tied (TCT) Ramaprabha and
modular nature of PV systems, is advantageous when it Mathur, 2012. SP configuration which is the most practi-
helps to wire the system up to desirable level of current, cally used configuration is considered in this study. In this
voltage and power. But the fact that PV modules with configuration connected modules in series form strings
the same brand and same ratings are not exactly identical, and connected strings in parallel form an array. Obviously
turns the modularity of PV systems to be disadvantageous all modules per each string work at the same amount of cur-
when it causes sort of power losses known as mismatch rent and all strings in an array work at the same amount of
losses which are recognized by several research works voltage at a time. So it is simply concluded from Kirchhoff
(Bucciarelli, 1979; Chouder and Silvestre, 2009; Picault laws that the string voltage equals the summation of mod-
et al., 2010). Since PV modules are fabricated in the ules voltage in strings, the array current equals the summa-
factory, further investigation and modification in cell level tion of strings current and array’s voltage equals the voltage
requires damaging the module encapsulation. So investi- of every string. An array of 40 modules and 4 strings with
gating mismatch losses mitigation techniques among cells SP configuration is depicted in Fig. 2 as an example.
inside modules is of PV module manufacturers’ interest
whereas such investigation among modules in arrays is of 3. Mismatch losses
system operators and installers interest. This paper deals
with mismatch losses among modules at array level. As previously mentioned, differences in PV modules
Module sorting techniques are regular methods for mini- characteristics together with modularity of PV arrays cause
mizing mismatch losses in PV arrays in which modules are mismatch losses in PV arrays. These differences are dis-
sorted in arrays by one of their characteristic parameters. This cussed in first subsection and second subsection explains
paper elaborates these techniques and reviews the mechanism how these differences cause mismatch losses.
of mismatch losses to propose a more effective solution. The
proposed solution applies a GA to find the optimal arrange- 3.1. Differences in PV modules characteristics
ment of modules in an array, considering array output power
as an objective function to be maximized. The arrangement of A group of modules of the same brand and same nominal
modules obtained by the GA is then compared to the ratings are not exactly identical. Their differences are under-
arrangements obtained by the sorting techniques in terms of stood by comparing and contrasting their characteristic
mismatch losses and energy yield. parameters such as fill factor (FF), maximum power
Understanding electrical configuration of PV arrays, (PMPP), current at maximum power (IMPP), voltage at max-
difference in PV modules characteristics and mismatch imum power (VMPP), short circuit current (ISC) and open
losses mechanism are inevitable steps toward comprehend- circuit voltage (VOC). Difference in module characteristic
ing the problem of mismatch losses in PV arrays and parameters is called I–V mismatch, since it results in differ-
subsequently devising solutions to this problem. Second ent electrical performance. I–V mismatch comes from either
and third sections elaborate these steps. Section four temporary or permanent sources as classified in Fig. 3.
reviews the conventional treatments for mismatch losses. Shading or non-uniform illumination might happen by
Section five explains the hypothesis of this study. Sixth fallen leaves of trees, scattered clouds moving over the PV
section is devoted to the methodology of the work. Sections
seven and eight provide results and conclusions respectively.
2. PV array configurations
Manufacturing Tolerance
Light-induced Degradation
Soiling
Permanent Factors
Discoloration
Delaminataion
Sources of I-V mismatch
Cracking
Temporary
Shading or non-uniform illumination
Factors
array, shadow of an object situated around the PV array or previously explained. Simply put, under central MPPT,
another reason (Sonnenenergie, 2008). This factor can two modules with different IMPP that are connected in ser-
temporarily result in mismatch losses in an array of mod- ies, compromise to work at the lower IMPP and similarly
ules, whereas there are other factors which permanently two modules with different VMPP that are connected in par-
cause mismatch losses such as manufacturing tolerance, allel work at the lower VMPP. Generalization of these con-
light-induced degradation, discoloration, soiling, delamina- ditions to a large array of PV modules connected in series
tion and cracking. Referring to datasheet of PV modules are and parallel, results in the following equation (Chamberlin
currently available in the market, it is found that despite all et al., 1995):
advancements of PV modules production technology, there !
P mod;i arr
still exist a manufacturing tolerance of ±3% to ±5% in their P
iX MPP P MPP
rating PMPP, IMPP and VMPP. Light-induced degradation, MML% ¼ 100 mod;i
ð1Þ
i
P MPP
discoloration, soiling and delamination are all matters of
aging (Smith et al., 2012). Cracking in cells is a module where MML% is the percent of mismatch losses, P mod;i
MPP is
defect that can happen during shipment, installation or the maximum power produced by ith module if it works
further happen due to hail (Ton et al., 2007) so it also can independently, and P arr
MPP is the output power of the whole
be considered as a matter of aging. All these permanent array.
factors cause dispersion in PV module characteristic
parameters. This variation in module characteristic param- 4. Conventional solutions
eters is the root of mismatch losses (Zilles and Lorenzo,
1992; Reis et al., 2002). This paper focuses on finding a According to the mechanism of mismatch losses
solution for mismatch losses occurs due to all permanent explained in Section 4, the more variation exists in charac-
factors. Thus, hereinafter term mismatch losses addresses teristic parameters of modules forming an array the
the mismatch power losses coming from permanent factors greater the resulting MML% of that array will be. This
and term mismatch refers to permanent I–V mismatch in PV fact is the basis of some conventional mismatch mitigation
modules. techniques called module sorting techniques which stand
It should be pointed out that aforementioned permanent for sorting modules in arrays based on one of the modules
factors (excluding manufacturing tolerance) have another characteristic parameters such as IMPP, ISC or PMPP
impact on the PV array performance that is always studied (Sonnenenergie, 2008; Webber and Riley, 2013; Bakas
separately entitled “performance degradation” (Smith et al., 2012) for arrays with SP configuration. Variation
et al., 2012; Coello, 2011). Performance degradation is of maximum power current of modules ðI mod MPP Þ in strings
not subject of this study. Performance degradation is mea- and variation of maximum power voltage of strings
surable by comparing the performance parameters of an
ðV str
MPP Þ in the array are in fact sources of mismatch losses
aged PV system to its fresh condition whereas mismatch
(Bucciarelli, 1979; Kaushika and Rai, 2007). Sorting mod-
losses is measurable by comparing array output power to
ules by ISC and IMPP reduces mismatch losses by greedily
the summation of individual modules output power.
minimizing the variation of I modMPP in strings (Webber and
Riley, 2013; Bakas et al., 2012) but it is further demon-
3.2. Mismatch losses mechanism strated that it neglects the variation of V str
MPP if it is com-
pared to sorting by PMPP. It is also demonstrated that
In practical PV generators, a central Maximum Power sorting modules by PMPP returns a balancing outcome
Point Tracker (MPPT) system is applied to lead the system in terms of minimizing the variation of I mod str
MPP and V MPP ,
to work at its’ possible maximum power. In the presence of but it does not result in lower mismatch losses in compar-
such a system all modules are led to work at their IMPP and ison with sorting by ISC or IMPP.
VMPP which are not exactly identical for all modules as
470 S. Shirzadi et al. / Solar Energy 108 (2014) 467–478
5. Proposed solution and mismatch losses and energy yield must be calculated
for each one respectively. Following subsections cover the
The number of overall possible arrangements of methodology of the work step by step.
modules in an array is calculated as follows:
6.1. PV module modeling
m m ð1 lÞ m ð2 lÞ
X¼
l l l This is a common practice in electrical engineering to
ð2Þ
m ððn 2Þ lÞ model an electrical element with some mathematical equa-
... 8m; n; l 2 N & n > 1
l tions and their equivalent circuit. Without such a model,
participation of that element in any credible calculation
where X is the number of possible arrangements, l is the or simulation would be impossible. One-diode model
number of modules in series per each string, n is the num- (Chan and Phang, 1987) is chosen to represent PV modules
ber of parallel strings, and m is the total number of mod- in different simulations which are carried out in this study.
ules in the array. Table 1 shows how dramatically Respective equations are as follows and its equivalent
number of overall possible arrangements for an array circuit is shown in Fig. 4.
increases by increasing the number of modules. For exam-
ple, the second array in Table 1 with 40 modules organised V þ Rs I V þ Rs I
I ¼ I ph I 0 exp 1 ð3Þ
in 4 parallel strings of 10 modules, has 4.7054 1021 pos- ns V t Rp
sible arrangements and sorting modules by IMPP, ISC and akT
PMPP are only three of them. Since module sorting tech- Vt ¼ ð4Þ
q
niques reduce mismatch losses through minimizing the var-
iation of I mod str
MPP and V MPP by simply sorting modules and
where I and V are the module current and voltage, Rs and
performances of other possible arrangements are unknown, Rp are series and parallel resistances, Iph is the photo-
the idea that there might be an arrangement with lower generated current, I0 is the diode saturation current, ns is
resulting mismatch losses than what is obtained by those the number of cells in series per each module, a is the diode
few arrangements of module sorting techniques seems wor- ideality factor, k represents the Boltzmann constant
thy of investigation. Finding such an arrangement is the (1.381 1023J/K), T is the module temperature and q is
hypothesis of this study. Instead of trying all the possible the electron charge (1.602 1019 C).
arrangements to find the best one which is apparently inap-
plicable due to magnitude of the search space (all possible 6.2. Data sets of modules
arrangements), GA which is highly qualified among prob-
abilistic search algorithms is applied to find it. The stochastic process introduced in Bakas et al. (2012) is
used to generate 4 data sets of 18, 40, 65 and 90 PV modules.
6. Methodology Studies on the statistical distribution of PV modules charac-
teristic parameters shows that Gaussian distribution
In order to examine the suggested mismatch mitigation explains the distribution of VOC, ISC and PMPP (Reis
technique and compare it to conventional ones in terms et al., 2002; Damm et al., 1995). This stochastic process
of its capability of mismatch losses reduction and energy assumes Gaussian distribution for VOC, ISC and PMPP and
yield improvement, some simulations are applied in MAT- generalizes it to IMPP and VMPP as well by considering their
LAB environment. The arrays listed in Table 1 require datasheet values as mean values and 10% of datasheet values
datasets of modules with dispersed characteristic parame- as their standard deviations. Datasheet values of CSUN
ters to represent arrays with mismatch losses. Each module 095-36M mono-crystalline solar module which are listed in
needs to be modeled to be implemented in simulations. A Table 2 are used in this study. Next step of the stochastic
GA must be adapted to the problem of mismatch losses process is to randomly choose values within the range of
to find the new arrangement of modules in the arrays. ±10% of mean value for I MPP , V MPP , I SC and V OC , and within
Finally the GA based arrangement and three other the range of ±3% of mean value for P MPP . These two allow-
arrangements need to be applied to each of the arrays able selection ranges are in accordance with the manufactur-
ing tolerance reported in the datasheet and standard
Table 1
4 Different arrays and respective number of overall possible arrangements. Rs I
Array/dataset Array Number of Number of all possible +
number dimension modules arrangements (size of
search space)
Iph D Rp V
1 36 18 1.7153 107
2 4 10 40 4.7054 1021
3 5 13 65 8.8090 1041 -
4 5 18 90 1.3811 1059
Fig. 4. PV module equivalent circuit based on one-diode model.
S. Shirzadi et al. / Solar Energy 108 (2014) 467–478 471
Table 2 Sera et al. (2007), Can and Ickilli (2014). Stochastic process
Datasheet values of CSUN 095-36M mono-crystalline solar module. that is applied in this study is summarized in Fig. 6.
Parameter Value Unit Fig. 7 shows the current voltage curves (I–V curves) at
PMPP 95 (W) STC for 40 modules that comprise the array with 40 mod-
VOC 22.5 (V) ules. I–V curves show more dispersion in I SC and VOC than
ISC 5.56 (A) what they do at the knee point of the curves. This condition
VMPP 18.3 (V)
IMPP 5.21 (A)
caused by applying the restriction of ±3% of mean value for
Performance deviation of PMPP ±3 % PMPP which equals the multiplication of IMPP and VMPP.
Performance deviation of VOC, ISC, VMPP and IMPP ±10 % The absolute error between generated characteristic
Voltage temperature coefficients 0.307 %/K parameters and those acquired from respective module
Current temperature coefficients +0.039 %/K model are calculated by Eq. (5) and their maximum amounts
Power temperature coefficients 0.423 %/K
Number of cells in series per module 36 Pieces
among all the modules are demonstrated for each of the
arrays in Table 3 to verify the accuracy of modeling process.
E x ¼ jxmodeled xgenerated j ð5Þ
0.8
where E_x is the absolute error, xgenerated is the generated
Probability Density
Assuming Gaussian distribution for , , , and with a mean value equals to their datasheet values and a standard deviation equals to
10% of their datasheet values
Choosing random values within the range of ±10% of datasheet value for , , , and
The extra condition for choosing and is that the multiplication of each selected pairs of VMPP and IMPP must be in the range of ±3% of
datasheet value of
Extracting equivalent circuits parameters ( ℎ , 0, R s , R p , and a) for each group of selected values
Using one-diode model for each module and finding , , , and and comparing them with the generated values to verify the
accuracy of PV module modelling
Table 3
Maximum absolute error for generated characteristic parameters relative to those acquired from one-diode model among all modules of 4 datasets.
Array/dataset number Array dimension Number of modules E_ISC (A) E_VOC (V) E_IMPP (A) E_VMPP (V) E_PMPP (W)
1 36 18 2.94E05 5.51E06 4.99E04 1.75E03 3.00E04
2 4 10 40 1.99E05 4.44E06 5.08E04 1.92E03 2.54E04
3 5 13 65 2.94E05 5.51E06 5.08E04 1.92E03 3.00E04
4 5 18 90 2.69E05 6.15E06 5.08E04 1.93E03 3.32E04
Table 4
Format of dataset matrixes for each array.
Module number Characteristic parameters Equivalent circuit parameters
ISC VOC IMPP VMPP PMPP Rs Rp a I0 Iph
1 5.85 22.13 5.35 17.35 92.75 2.35E01 4.16E+02 1.49 7.04E07 5.85
2 5.35 23.26 4.92 18.77 92.36 1.49E01 4.91E+02 1.55 5.12E07 5.35
3. 6.08
.. 21.08
.. 5.63
.. 16.90
.. 95.13
.. 1.70E01
.. 7.17E+02
.. 1.30
.. 1.71E07
.. 6.08
..
.. . . . . . . . . . .
m 5.94 23.38 4.91 19.71 96.84 1.78E05 2.59E+01 1.33 3.30E08 5.94
S. Shirzadi et al. / Solar Energy 108 (2014) 467–478 473
Fig. 8. P–V curves for 4 different arrangements of 40 modules of the second array.
well-known problem called Traveling Salesman Problem considered a random answer for FF. Mutation concerns
(TSP) that has been solved by GA Liu (1402), Ahmed swapping the position of two modules in the array. For
(2014), Changdar et al. (2014), Wang (2014). Imagining a each of the arrangements that are created at the initializa-
group of cities with specified distances between them, tion, during the GA process as random answers or as the
TSP is to find the shortest path that passes all the cities final optimal answer P arr k;i
MPP is calculated by the method that
and returns back to the starting point. Main similarity is explained in Section 6.4.
between finding optimal arrangement of PV modules in Flowchart of the GA is demonstrated in Fig. 10. Where i
arrays and TSP is that both problems deal with searching counts the GA procedure rounds and indicates one gener-
and finding one specific arrangement of large number of ation among all generations, FF A ðiÞ is the average fitness
objects with some specific qualities among numerous possi- score of all answers of ith generation, j indicates rounds
ble arrangements. Other promising similarity is that PV of GA operation process at which the population of
modules in arrays are as unique as the cities in TSP are. answers stabilizes and stops anymore improvement and
According to Eq. (1) rearranging the PV array affects FFS(j) is the respective value of fitness function. The
P mod
P arr
MPP with i P MPP being constant. In other words minimi-
FF(Ak,i) in Eq. (6) that calculates the fitness score of answer
zation of MML% equals the maximization of array output k of the generation i should not be confused with FFA(i) as
power. Therefore, in order to adapt a GA to the problem the average fitness score of generation i, or FFS(j) as fitness
of mismatch losses, fitness function is defined as the array score of the stabilized generation. Maximum fitness score
output power to be maximized as formulated in Eq. (6). of each generation that is plotted in Fig. 10 is not a neces-
sity to the algorithm flow but an important part of the
FF ðAk;i Þ ¼ P arr
MPP
k;i
ð6Þ results. The GA starts with generating random answers
to form the initial population. Evaluation of these answers
where FF is the fitness function, Ak,i is a 1 m vector that is applied by calculation of fitness function for each of
shows the arrangement of m modules in the array, P arrMPP
k;i
is them (FF(Ak,i)). Then parents’ selection is performed by
the array output power for k th arrangement at ith gener- means of a probabilistic procedure called casino roulette
ation, k indicates an answer among the population of wheel which is explained in Iglesias et al. (2006). After that
answers and i indicates one generation of all generations PMX crossover is applied to the selected parents and
produced during the algorithm operation. In order to con- resulting offsprings are replaced by the current generation.
sider positions of modules as genes, different arrangements Evaluation, parents selection, crossover and generation
of modules in the array as chromosomes and apply par- replacement are applied sequentially and continuously
tially mapped crossover (PMX) Naveen Kumar and until stabilization in FF among population of current gen-
Rajiv Kumar, 2012, it is needed to assimilate the matrix eration is achieved which means all answers are converged
shape of the PV array to a vector shape of chromosome. to one answer and stop anymore improvement. At this
To remedy this strings of each PV array are lined up stage, mutation is applied to all the population of current
sequentially to form a chain of modules as depicted in answers. This mutated population goes through the evalu-
Fig. 9 for the array of 40 modules and 4 strings as an exam- ation, parent selection, crossover, and generation replace-
ple. Positions of modules in the array are kept unchanged ment to reach the stability again. This process continues
before and after this transformation to prevent miscalcula- whilst average fitness (FFA), fitness of stabilized genera-
tion of array output power which is further explained. tions (FFS) and maximum fitness per each generation are
Since modules are labeled by unique numbers, every ran- preserved. Finally procedure stops whenever several stabi-
dom permutation of numbers from 1 to m in Ak,i gives a lized populations of answers with same fitness score are
random arrangement of modules in the array which is observed consecutively and at least one mutation has been
474 S. Shirzadi et al. / Solar Energy 108 (2014) 467–478
Fig. 9. Transformation applied to make vector shape of chromosomes out of matrix shape of PV array.
Initialization:
Generating 30 Random Answers
= 1, = 1 and (1) = 0
No
Crossover Yes
( )= ()
Replacing Offspring with Previous
Generation (Selected Parents) No
Mutation Yes
No
Is
= +1 ( )= ( − 1)?
Yes
End
applied since the beginning of GA process. Fig. 11 shows be. In order to appropriately size the initial population for
the evolution of the GA. Dropping points in the left curve each simulation, at first an initial population between 100
of Fig. 11 are the average fitness score breakdowns after and 500 (almost in accordance with what has been experi-
every mutation. enced for solving TSP by GA (Julstrom, 1996) was selected
The introduced GA process that takes a data set of PV for each simulation and valid and stable answer (which
modules with determined SP configuration as input and shows crossing the knee point in the GA evolution curves
returns an optimal arrangement of the modules in the array e.g. Fig. 10 and does not vary for multiple times of simula-
as output, is applied to the 4 different datasets that are tion repetition) was achieved. Then the initial population
generated through stochastic process regarding 4 arrays was decreased step by step to see if we could get the answer
specified in Table 1. Choosing the right size for initial popu- with smaller initial population. Table 5 lists all simulations
lation is important because the bigger the initial population that are performed for the 4 arrays in this work. Respective
is the more time consuming the simulation program would results are discussed in the next section.
S. Shirzadi et al. / Solar Energy 108 (2014) 467–478 475
Fig. 11. Evolution of GA for the 4 10 array (a) average fitness of each generation and (b) maximum fitness of each generation.
Table 5
Format of data set matrixes for each array.
Array number Number of modules Array dimension (strings modules per string) Initial population Mutation percentage Crossover method
1 18 36 20 2 PMX
2 40 4 10 30 2 PMX
3 65 5 13 50 2 PMX
4 90 5 18 70 2 PMX
Table 7
Specifications of UPM solar site.
Location Longitude Latitude Daily average Average monthly Daily maximum
radiance level (W/m2) ambient temperature (°C) ambient temperature (°C)
UPM, Selangor, Malaysia 2.99 101.72 253–512 29.60 30.2–36.6
Table 8
Energy yield for each of the arrays during one day and one month.
Arrangement Energy yield (kW h)
Array #1 Array #2 Array #3 Array #4
Day Month Day Month Day Month Day Month
Sorted by I SC 14.02 277.12 31.47 622.08 51.02 1008.40 70.87 1400.78
Sorted by I MPP 14.05 277.72 31.49 622.43 51.12 1010.31 70.91 1401.42
Sorted by P MPP 13.99 276.59 31.45 621.57 51.02 1008.37 70.78 1398.88
Arranged by the GA 14.07 278.13 31.62 624.87 51.29 1013.79 71.19 1407.02
Table 9
Recoverable energy for each of the arrays during one day and one month.
Arrangement Recoverable energy
Array #1 Array #2 Array #3 Array #4
Wh % Wh % Wh % Wh %
Day 50 0.36 141 0.45 270 0.53 320 0.45
Month 1010 0.36 2792 0.45 5390 0.53 6240 0.45
Table 10
Comparison of different arrangements in terms of MML%, variation of I mod str
MPP and V MPP at STC for the second array of 40 modules.
(a) (b)
(c) (d)
Fig. 12. Different arrangements of 40 PV modules in the 4 10 array (a) Sorted by ISC, (b) sorted by IMPP, (c) sorted by PMPP and (d) arranged by the GA.
Reis, A.M., Coleman, N.T., Marshall, M.W., Lehman, P.A., Chamberlin, Ton, D., Tillerson, J., McMahon, T., Quintana, M., Zweibel, K., 2007.
C.E., 2002. Comparison of PV module performance before and after Accelerated Aging Tests in Photovoltaics Summary Report, US
11-years of field exposure, in Photovoltaic Specialists Conference. Department of Energy Efficiency and Renewable Energy.
Conference Record of the Twenty-Ninth IEEE, 1432–1435. Wang, Y., 2014. The hybrid genetic algorithm with two local optimization
Sera, D., Teodorescu, R., Rodriguez, P., 2007. PV panel model based on strategies for traveling salesman problem. Comput. Ind. Eng. 70, 124–
datasheet values, In: IEEE International Symposium on Industrial 133.
Electronics. ISIE, 2007, pp. 2392–2396. Webber, J., Riley, E., 2013. Mismatch loss reduction in photovoltaic
Smith, R.M., Jordan, D.C., Kurtz, S.R., 2012. Outdoor PV module arrays as a result of sorting photovoltaic modules by max-power
degradation of current-voltage parameters, in World Renewable parameters. ISRN Renew. Energy.
Energy Forum. Denver, CO, USA. Zilles, R., Lorenzo, E., 1992. An analytical model for mismatch losses in
Sonnenenergie, D.G.f., 2008. Planning and Installing Photovoltaic PV arrays. Int. J. Sol. Energy 13, 121–133.
Systems: A Guide for Installers, Architects and Engineers, Earthscan.