To quantify the difference between the mismatch loss in partially shaded PV-modules connected in parallel and in series, experiments were conducted with nine modules alternately connected in parallel and connected in series. The measurements were done with the very same modules under virtually identical conditions, which excludes other influences on the array power than the mismatch effect.

Experiments were performed with 9 modules on sheds on a horizontal roof. The orientation of the sheds is depicted in figure 2. Figure 3 shows which 9 modules of the second shed were used for the experiments.

The 9 modules were chosen arbitrarily from the total set of modules and were not cleaned prior to the experiments. The modules have their original two bypass diodes (1 diode per string of 18 cells).

With a specially designed relay-switchbox the 9 modules were alternately connected in parallel and connected in series. During the measurement campaigns the IV-curve of the modules with the parallel connection and with the series connection were measured alternately. The pairs of IV-curves were used to calculate the corresponding pairs of maximum power point values.

Next to the PV-arrays some slim poles (5 x 6 cm) are situated with horizontal wires (6 mm). The width of the PV-cells is 12.5 cm. At the time of the measurements these poles cast a shadow over some of the 9 modules mainly in the morning period. In the afternoon this happens only sporadically. The wire always casts a shadow over the modules. The following picture shows the shadows on the modules at 10:40h. The shadow of the wire is too thin to be seen on the picture.

Figure 3: sketch of the shed with 9 x 4 modules.

The 9 modules chosen for the experiment are indicated with the dark-grey filling.

The results of the measurements on a sunny day (17/11/2003) are shown in figure 5. It shows that in the afternoon, when there is practically no shading, the parallel concept produces marginally more power than the series concept. However in the morning, when there is only little shading, the parallel concept produces significantly more.

Of course the difference between both concepts depends very much on the shading conditions and consequently no general conclusions can be formulated on the annual energy gain by the parallel concept. However the results do show that PV-systems in the built environment, which are virtually always subject to some shading, will profit by the parallel concept.

4 Accessibility

PV modules are poorly accessible when being part of a building. Usually scaffolding must be put up to replace defective modules. As this is costly it is good to consider what are the effects of a defect in the system.

Most PV-modules are equipped with one or more bypass diodes to prevent thermal destruction of cells due to hot spots in case of partial shading. Another function of bypass diodes is that failing modules in a string are bypassed and that the remaining modules of the string can still produce power. This situation however may lead to additional mismatch loss in the case of a central inverter with parallel strings. This is caused by the differences in the MPP voltage of the strings with a different number of active modules. It is also possible that the cabling that interconnects the modules fails. In these situations one failing module in a string results in the loss of the complete string. This happened recently in a Dutch BIPV system in which 6 modules out of about 400 failed. The modules were part of strings with a length of 13 modules. The 1.5% of failing modules resulted in a loss of 10% of the strings. Since energy loss of 10% was considered unacceptable the failing modules had to be replaced. Because of the low accessibility of the roof this could only be done at high costs. In case the strings would have been shorter (ultimately with a length of one module), the energy loss would have been acceptable and no repair would have been needed.

As in the previous two paragraphs this is merely an example of what can happen. It is hard to quantify the over-proportional energy loss on annual basis as averaged over all BIPV installed. Nevertheless it is recommended to use short strings in BIPV system due to limited accessibility for repairing.

It is also recommended to have a system layout that enables to mount the modules simply. An identical connection of all modules is preferred, favouring a fully parallel or series connection. In addition the mounting system must allow simple and easy installation and maintenance.

5. Consequences

In the previous chapters arguments are given for application of short strings in building integrated PV-systems. The arguments are based on the energy yield of the array. The effects of price, efficiency and reliability of the inverters were not taken into account.

The shortest possible string is a string consisting of one module. This leads to the parallel connection of many modules. Possible concepts for applying many modules in parallel are:

• All modules directly DC-connected via a low resistance cable to a central inverter (low voltage DC-bus). In the Wirefree concept [4] this cable is omitted by the dual use of metal strips for mechanical support and electrical connection.

• All modules equipped with their own module (AC-modules) with an AC-bus.

• All modules equipped with an individual maximum power point tracker:

— DC/DC module inverter connected to a high voltage DC-bus and a central DC/AC inverter

— DC/AC module inverter connected to a high voltage AC-bus and a central AC/AC inverter

6. Conclusions

In the design of a BIPV system it is necessary to take partial shading, especially at low altitudes of the sun, possible different orientations within the system and the chance of component failures into account. These factors lowering the energy output must be offset to the system cost, including engineering and maintenance cost.

In many cases a parallel connection of modules is favourable especially if it can be combined with lower component costs, such as the Wirefree concept.

7. Acknowledgement

This work was supported by the Netherlands Agency for Energy and Environment Novem (project number: 2020-02-11-11-003).

[1]