Professional Documents
Culture Documents
Connecting photovoltaic cells to form an array can cause problems when the
characteristics of the cells are not matched. This problem, known as mismatch, can
reduce the power output of the array and lead to cell degradation through localized
heating of individual cells. Such problems can arise simply through the shading of a
single cell. In this paper, which has a substantial tutorial content, three methods of
reducing the eflects of mismatch are investigated and compared; the use of integral
bypass diodes (IBDs), serieslparalleled circuit designs and reduced cell shunt resistance.
Experiments were performed on arrays incorporating cells with IBDs and their
tolerance to shading was tested and compared to arrays with other cell designs. The
potential benefits and costs of using IBDs are then assessed and conclusions drawn
about the relative eflectiveness of this strategy. The circuit design strategy known as
serieslparalleling is described and analysed and its eflect on minimizing mismatch
power loss is evaluated. Also, an expression for the optimum cell shunt resistance in a
general shaded array is derived and the technique of reducing cell shunt resistance to
minimize mismatch eflects is assessed.
IN TRODUCTION
P
hotovoltaic (PV) cells have been developed to the point where they are viable for a range of
applications. However, a major problem still affecting the performance and reliability of PV
systems is that of mismatch. Mismatch occurs due to the interconnection of cells with different
performance characteristics. This condition results in two detrimental effects: power loss and reliability
degradation.
When the characteristics of cells within a PV array differ, individual cells may operate suboptimally,
causing a reduction in array output power. This is especially the case for cells connected in series with
variations in their current outputs. Such a situation may arise simply through shading of a single cell
through leaves, bird droppings or dirt falling onto the array. '-''
Figure 1 shows a typical example of mismatch where a single low current output cell (caused by, say,
partial shading) is series connected with four identical cells having normal current outputs. The resultant
current-voltage ( I - V ) relationship shows that the current output of the shaded cell significantly limits
the current and hence power output of the array. In extreme cases, almost the entire array output can
be lost through shading of a single cell. Furthermore, when the array in Figure 1 is short circuited, the
low current output cell is severely reverse biased whilst the other cells operate near their maximum
Figure 1. Current-voltage curves showing: (1) four identical series-connected solar cells; (2) a single low current
output cell; (3) series connection of all five cells. (Adapted from Ref. 12)
power point. As the power output of the array is zero, all the power generated by the unshaded cells is
dissipated in the single shaded cell, resulting in localized heating of the cell and possible cell damage.
This phenomenon is known as hot-spot heating.13-'*
The reduction in maximum power output from an array due t o mismatch is usually expressed as a
relative power loss (RPL),defined by'9s20
RPL = Cipi-pa
Ci pi
where pi is the maximum power available from cell i when operated in isolation and Pa is the maximum
power available from entire array.
The problem of mismatch has been well recognized and several approaches for overcoming it have
been suggested, such as the inclusion of bypass diode^,^ 1-23 series/paralleled circuit design s t r a t e g i e ~ ~ . ~
and reduced cell shunt r e s i ~ t a n c e . ~In~ this
~ ~ ~paper,
* * ~ these strategies are investigated and assessed.
Bypass diodes are commonly used in PV arrays to guard against mismatch effects by limiting both
power loss and hot-spot power dissipation. The disadvantages of using large numbers of bypass diodes
can be overcome partly by using cells with integral bypass diodes (IBDs). A critique of IBD technology
is presented in the next section.
Previous publications have noted that mismatch effects can be reduced by designing PV arrays with
a large degree of parallelism and redundancy; a technique known as series/paralleling. However, little
work appears to have been done in describing the mechanisms underlying this technique and highlighting
its benefits and disadvantages. Such an analysis is presented in a later section of this paper.
A largely undeveloped approach to the mismatch problem is control of the cell shunt resistance. A
method for determining the optimum cell shunt resistance for a general array configuration subjected
to partial shading is developed in the last section of this study. It is found that, under many conditions,
a finite shunt resistance is the optimum solution. The costs and benefits of this approach are then explored.
thus preventing current limiting. This reduces power loss due to mismatch and limits the maximum
power dissipation in any cell to the power generated by the bypassed cells, thus avoiding hot-spot
heating.21-23To minimize both power loss and hot-spot heating, arrays should ideally include one bypass
diode across each cell. However, this approach is usually too expensive and requires many additional
connections. To overcome these limitations, an IBD may be fabricated into the cell structure. This
approach is potentially cheaper and requires no additional intercell connections. Several processes for
fabricating crystalline and polycrystalline solar cells with IBDs have been developed at the University
of New South Wales’ Centre for Photovoltaic Devices and Systems.26-28
1 2 3
Figure 2. Schematic representation of shading arrangements: (1) partial and complete shading of a single cell;
(2) shading along an east-west axis; (3) shading along a north-south axis
118 D . ROCHE ET AL.
7I
.4
.2
I
.8
.6
.4
.2
.4
I.6
.z
5
1 1
10
.8 .8
.6 .6 15
.4
----- 20
.4
20
.2 .2
25
"0 5 10 I5 20 25 30. 3
Voltage (V)
Figure 4. Array current-voltage curves for different cell designs. East-west shading: 0, 5, 10, 15, 20, 25%
.61
.2
.8
.6
.4
.2
"0 5 10 I5 20 25 30 3
Voltage (V)
Figure 5. Array current-voltage curves for different cell designs. North-south shading: 0, 5, 10, 15, 20, 25%
MISMATCH P O W E R LOSS IN PV A R R A Y S 119
80
20
25 50 75 I "0 5 10 15 20 25
Percentage of Single Cell Shaded ( O h ) Percentage of A m y Shaded (%)
(c)
Figure 6. Relative power loss versus percentage shading for different shading arrangements and cell designs: (a)
single cell shading; (b) east-west shading; (c) north-south shading
120 D. ROCHE ET AL.
Benejts of serieslparalleling
Increasing the degree of parallelism in a P V array has an obvious effect on the reduction of mismatch
losses. As has been noted, when a low current output cell is connected into a series string, the current
output of the string is limited by this cell. By connecting cells in parallel, this current limiting effect is
avoided and thus array power output is maintained.
In addition to the benefits of parallelism, a circuit with both series- and parallel-connected cells can
be designed in a number of ways with varying degrees of redundancy. For example, the two simple
four-cell circuits of Figure 7 display the possible circuit designs of a two-by-two cell array. The only
difference between these two circuits is the additional connection of Figure 7(b).
To understand the effect of series/paralleling on mismatch power loss, consider the situation where
two cells in both arrays are shaded by the same amount as shown Figure 7. For the two arrays, the
current in each substring can be approximated by the lowest current output of its constituent cells. For
an unshaded cell with voltage and currents outputs of u and i, respectively, a 50% shaded cell will have
voltage and current outputs of approximately u and i/2, respectively. Thus, the voltage and current
outputs of the circuit in Figure 7(a) are approximately u and i, whilst those of the circuit in Figure 7(b)
are approximately u and 3i/2. For this example, the power output is clearly greatest for the array with
greatest redundancy.
Increasingly redundancy can also improve the array fill factor as well as the short-circuit current. To
understand this effect, consider the same two four-cell arrays of Figure 7 but with only one cell shaded
in each array. Again a mismatch condition exists and the extra connection of the circuit in Figure 7(b)
is no longer redundant. However, the analysis used above shows that both circuits have similar power
outputs. Figure 8 shows I- V curves corresponding to different circuit arrangements between shaded and
unshaded cells with infinite shunt resistances. Whenever two components are connected in series the
voltages are added, whilst for parallel connections the currents are added. As can be seen from Figure 8,
there is a notable improvement in the fill factor (and hence power output) when the redundant
connection is included.
By reducing the effect on array performance of variations in cell characteristics, series/paralleling
circuit designs can reduce the need for cell sorting. Such designs can also improve tolerance to open-circuit
defects because an open-circuit defect behaves essentially like a completely shaded solar cell with an
infinite shunt resistance. Thus, the additional cost of series/paralleling designs may be partially offset by
such benefits.
The energy benefit of series/paralleling in practical installations is difficult to quantify because it
depends on the degree of mismatch between cells and the expected prevalence of open-circuit defects.
b
(a) (b)
Figure 7. Two four-cell array circuits with different degrees of redundancy and two cells shaded in each array
M I S M A TCH 121
2
Voltage (V)
Figure 8. Current-voltage curves for a number of array components, showing the effect of redundancy and how
an array I-V curve can be calculated from the I-V curves of its constituent components. Each curve is labelled
diagrammatically by the circuit it represents
For small variations in cell output arising at the manufacturing stage, the increased performance is likely
to be about 1-3%.30 However, for an array regularly subject to partial shading (such as a rooftop
application) and/or open-circuit defects, the benefit may be greater than 10% and in some cases greater
than 50%.30
for different module wiring configurations, such as in present-day modules adaptable for 6- or 12-V
operation. Another approach is to design modules with a large degree of parallelism and wiring
redundancy built in. Although this technique results in simpler wiring configurations, such modules
would have a low voltage and high current output, making them relatively inflexible and not widely
applicable to arrays of different sizes.g It is estimated that the additional cost of such designs would
need to be less than about 3% of the total module cost to be effective.
where I, is the photogenerated current, I, is the dark saturation current, q is the charge on an electron,
n is the ideality factor, k is the Boltzmann constant and T is the absolute temperature.
Figure 9 shows the calculated power output of a 36-cell series-connected array versus the normalized
cell shunt resistance (RCH),where
One cell in the array has been partially shaded by different amounts. The shading (or quality) factor is
given by a, where a = 1 corresponds to zero shading and a = 0 corresponds to complete shading.
Otherwise, all cells are assumed to be identical and the cell and array parameters of Table I have been
assumed. The array operating voltage is and the number of series-connected cells in the array (36 in
this case) is m.
lL= 4.0 A
1, = 1.0 pA
n = 1.5
T = 50°C (323 K)
= m/3 V
MISMATCH POWER LOSS IN P V ARRAYS 123
Figure 9. Power output versus normalized shunt resistance for a 36-cell array with one cell partly shaded. No
shading corresponds to a = 1, whilst complete shading corresponds to a = 0
A-
\
Y 2
P
Figure 10. An m-by-p array of cells with minimum redundancy. A number of these cells are partially shaded as shown
When no shading exists, the power output is maximum when the shunt resistance is infinite. However,
for non-zero shading levels, maximum power output is attained for a finite shunt resistance value. This
effect has been addressed previously but the procedure developed applies only to series-connected arrays
of a fixed string length.22Work was carried out to develop a procedure for analysing more general array
structures and shading patterns. Full details of the derivation are given in Refs 9 and 24.
Consider an array with both series and parallel connections, as shown in Figure 10. This array contains
mp cells incorporated into a circuit with minimum redundancy. Of the mp cells in the array, x y of them
are assumed to be partially shaded, as shown in Figure 11, so that there are equal numbers of shaded
cells in each string containing shaded cells. Otherwise, all cells are assumed to be identical. Thus, there
are three types of cells comprising the array, namely:
(i) unshaded cells in a substring containing shaded cells,
124 D. R O C H E ET A L .
20 -
10 -
0 20 40 60 RO 100
Percentage Shading of a Single Cell (%)
Figure 11. Power output of a 36-cell series-connectedarray versus percentage shading of a single cell for R,, = co.
RsH = 100RcH, R,, = ~ ~ R cRsH H , = 10RcH, Rs, = 3RcH and RsH = RsHo
Using Equations (4)and ( 5 ) together allows for the optimum shunt resistance to be obtained numerically.
By doing so, the analysis presented in here was tested against, and found to be in agreement with,
calculations made using a SPICE computer simulation.
Note that if p = y = 1, then Equation (4) reduces to the case of a purely series-connected array.
Equation (4) can also be used to analyse a series/paralleled array with maximum redundancy.
MISMATCH POWER LOSS IN PV ARRAYS 125
To do so, parallel-connected cells are treated mathematically as single large-area cells. The
series/paralleled array is then analysed as a simple series-connected array.
Consider the graphs of Figure 11. These show the power output of a series-connected 36-cell array
versus the percentage shading of a single cell, assuming the parameters of Table I and cell shunt resistances
of infinity, 100Rc,, 30RcH, 10RCH and 3RcH. The dotted curve represents the upper limit to the power
output, i.e. at each point the power output is calculated by assuming a partially shaded array and an
optimum cell shunt resistance for that shading level.
It can be seen clearly from Figure 11 that high shunt resistance levels result in maximum power output
when the degree of shading is low. At high shading levels, however, the power output falls rapidly when
the shunt resistance is high. For very low shunt resistances, the power output is quite poor but relatively
constant at all shading levels. Intermediate values, such as 1ORCH in this example, result in moderately
good power output at all single-cell shading levels. For a shunt resistance of 10Rc,, the power loss at
zero shading is approximately 5%.22
CONCLUSION
An overview of the mismatch problem has been presented and several methods have been suggested to
reduce its effect. The use of IBDs was found to be a viable technique for preventing hot-spot power
dissipation and reducing mismatch power loss, particularly when a small number of cells are heavily
shaded. Provided the additional cost of using IBD cells can be minimized sufficiently, the use of such
cells may be beneficial in some applications.
Series/paralleling has been shown to be a low-cost design technique that can significantly reduce
mismatch losses whilst also improving tolerance to open-circuit defects.
The shunt resistance optimization technique was investigated using expressions derived for determining
the optimum shunt resistance in a partially shaded array. These expressions were then applied to examples
to gain a better understanding of the interrelationships between shunt resistance and power losses. It
was found that in arrays subject to mismatch, a finite shunt resistance was usually the solution, resulting
in maximum power output. Potential problems and benefits of using low shunt resistance cells were
discussed, as well as possible methods of producing such cells.
REFERENCES
1. A. Abete, E. Barbisio, F. Cane and P. Demartini, ‘A study of shading effects in photovoltaic generators’, 9th
EC Photovoltaic Solar Energy Conference, 1989, pp. 240-244.
2. A. Abete, P. Cane, C. Rizzitano, M. Tarantino and R. Tomasini, ‘Performance testing procedures for photovoltaic
modules in mismatching conditions’, Proc. 22nd IEEE Photovoltaic Specialists Conference, 1991, pp. 807-81 1.
3. W. R. Anis, ‘Partial shadow protection without diodes’, Solar Cells, 5 183-198 (1982).
4. C. C. Gonzalez, ‘Photovoltaic array loss mechanisms’, Solar Cells, 18 195-204 (1986).
5. M. A. Green, Solar Cells: Operating Principles, Technology and System Application, pp. 115-117, Prentice-Hall,
New Jersey, 1982.
6. A. Gupta and A. G. Milnes, ‘Effects of shading and defects in solar cell arrays’, Proc. 15th IEEE Photoooltaic
Specialists Conference, 1981, pp. 1111-1 116.
7. R. Janoch, ‘Analysis of solar cell mismatch losses at standard and non-standard conditions’, Proc. 17th IEEE
Photovoltaic Specialists Conference, 1984, pp. 558-562.
8. H. S . Rauschenbach, ‘Electrical output of shadowed solar arrays’, IEEE Trans. Electron Devices, 18(8), 483-490
(1971).
9. D. M. Roche, ‘Cell mismatch in photovoltaic arrays’, Undergraduate Thesis, School of Electrical Engineering,
University of New South Wales, 213 pp., 1993.
10. J. L. Watkins and E. K. Burgess, ‘The effects of solar cell parameter variation on array power output’, Proc.
13th I E E E Photovoltaic Specialists Conference, 1978, pp. 1061- 1066.
11. L. L. Bucciarelli, ‘Power loss in photovoltaic arrays due to mismatch in cell characteristics’, Sol Energy, 23,
277-288 (1979).
12. M. A. Green, Solar Cells: Operating Principles, Technology and System Applications, p. 1116, Prentice-Hall, New
Jersey, 1982.
13. M. M. Alkaisi and N. A. Aldawody, ‘Factors affecting the hot spot efficacy in photovoltaic arrays’, Sol. Cells,
28, 11-17 (1990).
14. J. C.Arnett and C. C. Gonzalez, ‘Photovoltaic module hot-spot durability design and test methods’, Proc. 15th
I E E E Photovoltaic Specialists Conference, 1981, pp. 1099-1 105.
MISMATCH POWER LOSS IN PV ARRAYS 127
15. F. A. Blake and L. L. Hanson, ‘The “hot spot” failure mode for solar arrays’, Paper 699070, Proc. JY6Y IECEC
Conference, 1969, pp. 575-58 1.
16. C. C. Gonzalez, R. W. Weaver, R. G. Ross, R. Spencer and J. C. Arnett, ‘Determination of hot spot susceptibility
of multistring photovoltaic modules in a central-station application’, Proc. J 7th IEEE Photovoltaic specialists
Conference, 1984, pp. 668-675.
17. E. Moelenbroek, D. W. Waddington and K. A. Emery, ‘Hot spot susceptibility and testing of photovoltaic
modules’, Proc. 22nd IEEE Photovoltaic Specialists Conference, 1991, pp. 547-552.
18. P. Spirit0 and V. Albergamo, ‘Reverse bias power dissipation of shadowed or faulty cells in different array
configurations’, 4th EC Photovoltaic Solar Energy Conference, 1982, pp. 296-300.
19. H. Saha, ‘Mismatch losses in series combinations of silicon solar cells modules, Sol. Cells, 25, 143-153 (1988).
20. M. Shechter, J. Appelbaum and G. Yekutieli, ‘Quality factor of solar cell arrays’, Sol. Cells, 9,295-309 (1983).
21. A. Abete, E. Barbisio, F. Cane and P. Demartini, ‘Analysis of photovoltaic modules with protection diodes in
presence of mismatching’, Proc. 21st IEEE Photovoltaic Specialists Conference, 1990, pp. 1005-1010.
22. M. S. Swaleh and M. A. Green, ‘Effect of shunt resistance and bypass diodes on the shadow tolerance of solar
cell modules’, Sol. Cells, 5, 199-204 (1982).
23. D. M. Roche, H. R. Outhred and R. J. Kaye, ‘Optimised design of photovoltaic arrays’, Proc. Solar ‘93: Australian
and New Zealand Solar Energy Society Conference, 1993, pp. 422-427.
24. D. M. Roche, H. R. Outhred and R. J. Kaye, ‘Cell mismatch in photovoltaic arrays’, AUPEC ’93: Proc.
Australasian Universities Power Engineering Conference, 1993, pp. 329-334.
25. G. Bhattacharya and C. Neogy, ‘Removal of the spot-spot problem in photovoltaic modules and arrays’, Sol.
Cells, 31, 1-12 (1991).
26. M. A. Green, ‘Silicon solar cells with integral bypass diodes’, NERDC Report, NERDDP/EG/82/60, National
Energy Research and Development Council, Canberra, Australia, 1982.
27. S. R. Wenham, M. C. Pitt, R. B. Godfrey, M. A. Green and E. Gauja, ‘Screen printed processing of solar cells
incorporating integral bypass diodes’, Proc. 16th IEEE Photovoltaic Specialists Conference, 1982, pp. 938-942.
28. E. S. Hasyim, S. R. Wenham and M. A. Green, ‘Shadow tolerance of modules incorporating integral bypass
diode solar cells’, Sol. Cells, 19, 109-122 (1986).
29. S. Narayanan and M. A. Green, ‘Improvement in the open circuit voltage and efficiency of silicon solar cells
by rear aluminium treatment’, Sol. Cells, 26, 329-334 (1989).
30. C. Gonzalez and R. Weaver, ‘Circuit design considerations for photovoltaic modules and systems’, Proc. 14th
I E E E Photovoltaic Specialists Conference, 1980, pp. 528-535.
31. H. J. Wenger, J. Schaefer, A. Rosenthal, B. Hammond and L. Schlueter, ‘Decline of the Carrisa Plains
photovoltaic power plant: the impact of concentrating sunlight on flat plates’, Proc. 22nd IEEE Photovoltaic
Specialists Conference, 199 1, pp. 586-592.
32. R. G. Goss, ‘Technology developments toward 30-year-life of photovoltaic modules’, Proc. 17th IEEE
Photovoltaic Specialists Conference, 1984, pp. 464-472,
33. D. J. Patterson, ‘Maximum power point tracking for silicon photovoltaic systems-the development of an open
loop algorithm and a practical realisation’, Proc. Electric Energy Conference: Power for Developing Areas,
Darwin, 1991, pp. 173-176.
34. P. H. Nguyen, T. Easwarakhanthan and S. Ravelet, ‘Influence of parasitic resistances on the mismatch relative
power loss of solar cell modules’, Rev. Phys. Appl. 20, 689-693 (1985).
35. M. A. Green, Solar Cells: Operating Principles, Technology and System Applications, p. 78, Prentice-Hall, New
Jersey, 1982.
36. A. L. Fahrenbruch and R. H. Bube, Fundamentals of Solar Cells, p. 272, Academic Press, New York, 1983.
37. A. M. Ricaud, F. Forge and P. E. Sarre, ‘Solar cell failure modes under reverse voltages and reliability’, Proc.
15th I E E E Photoooltaic Specialists Conference, 1981, pp. 11 17-1 121.