No confidence in manufacturer tests

“PID? Not a problem with our modules!” Just about every module manufacturer has publicized this phrase in one form or another. Or: “For us, PID is a thing of the past! We’ve got the problem under control. And here’s the certificate to show it.” In making these statements, manufacturers are relying on self-chosen test methods and pass criteria. No internationally accepted standard is currently in place. IEC is working on a standard (IEC 62804), but it is in draft status. Unfortunately, our experience shows that most of the modules designated as PID-free have not undergone sufficient testing. In addition many of the modules already in use, totaling over 100 GWp worldwide, were installed without any testing at all, and this will soon lead to an increasing number of PID incidents and customer complaints once the modules have been in the field for a few years.

The phenomenon

What is PID? PID is a degradation process that occurs in PV systems when electrically conductive components are at different electrical potentials, and a potential difference builds up between the components. No one can say they knew nothing of this. The phenomenon was described by Richard Swanson [1, 2, 3, 4] in 2005. One component of the PV power plant is the solar cell, and the other is the combination of front glass and PV module frame. The crystalline solar cells are electrically insulated from the environment (frame, glass front and back) by the encapsulating material (typically EVA and the backsheet foil). As a result of solar cell operation and the wiring concept in the PV power plant, a potential difference builds up between the solar cell and the front glass and frame. In Europe this potential can amount to as much as 1,000 V depending on the PV installation. This in turn creates leakage currents, though extremely weak, as shown in Figure 1 (p. 81). Due to moisture and impurities, a conductive layer is produced on the front of the glass. Through this, potential builds up between the glass and the solar cell, generating a small current as sodium (Na⁺ ) ions move from the glass toward the cell. On the front of the solar cell, these sodium ions create an accumulation of positive charges, which then cause short-circuiting (shunting) in the solar cell.
Without moisture and dirt, there is little chance of PID occurring. Modules installed in regions in which hot and humid weather conditions are the norm, or condensation or rainfall occurs frequently [4], are very often exposed to conditions that initiate the PID process. Impurities such as conductive dust, salts, acidic or basic species or caustics and other ionic solutions further accelerate the effect, since they enhance the harmful contact formation between the glass surface and the potential to ground.
Modules in open-field installations are generally ground-mounted and therefore are not only adequately cooled and dried by the wind, but are also generally kept quite clean due to their typical mounting angle of approximately 25° in northern latitudes. In rooftop-mounted installations, the rear air circulation is generally significantly poorer. These modules take longer to dry and, in addition, condensate remains longer on the back side of the module. The trend toward east-west installation on flat roofs, brought about by the extreme drop in module prices and attractiveness from an energy-management standpoint, has resulted in installation angles of less than 15°. Such installations cannot be cleaned by the rain as effectively. In addition to dust and bird droppings, impurities include industrial pollution or salt near the ocean.

PV power plant yield

What does this mean for a PV power plant’s output? One short-circuited solar cell will not typically give rise to a noticeable drop in system output. It’s the numbers that make the difference here, as more than one cell degrades in general. A string typically consists of up to 24 modules and thus of up to 1,400 cells. Consequently, depending on the module type and wiring in the PV installation, per-module losses of over 80% can occur when significant potential differences are present.
Two aspects make this noticeable in the yield of the PV power plant: first the PID-damaged solar cell causes the total output of the system to diminish and second, every solar cell affected by PID reduces the total string voltage. The inverter switches on later in the morning and off earlier in the evening, since more solar radiation is needed to exceed the inverter’s switch-on voltage. And depending on the season and cloud thickness, additional outages occur.

Detecting PID

How can PID be detected in a PV power plant? Unfortunately, there’s no easy answer to this question. PID generally results in diminished yield in the PV system. The magnitude of the power loss depends on the degree of degradation in the system due to this phenomenon. This makes it difficult to detect PID in the early stages.
There are a number of factors, such as dirt, faulty bypass diodes and defective modules that result in reduced system output. With a suitable monitoring system, it is possible to identify some of these effects individually. For example, dirt affects the system differently than does PID. If dirt is present, there is generally a decrease in yield, initiated by a reduction in currents (Impp falls and Umpp remains more or less constant at solar radiation levels above 600 W/m² ). PID, by contrast, causes a drastic reduction in the fill factor, that is, both Impp and Umpp fall while the system is exposed to the same solar radiation levels.
If PID is suspected and the degradation is already advanced, the effect can be verified relatively easily using infrared (IR) thermography (see Figure 2, p. 82). In the IR image, PID-affected cells appear slightly warmer than normally functioning cells [5, 6]. The temperature difference, however, is just a few degrees. Similar temperature differences occur frequently when the cells are not prop

Table 1: Overview of the currently most frequently used and discussed test methods for detecting PID in PV modules. Modules are considered PID-free if they lose less than 5% of their initial power when exposed to solar radiation levels of 1,000 W/m² .erly graded or bypass diodes are not working properly. For a detailed analysis, the expert must have knowledge of the module wiring and the position of the module in the string. The potential to ground depends on the position. In PID-damaged modules, a relationship with the respective potential can generally be seen.
A common inspection method in the field is to measure the current-voltage characteristic (IU characteristic) of entire strings. But because p-type cells typically installed today are only PID affected on the negative potential side of the string, early identification of PID in IU characteristics of module strings is difficult if only a few cells are degraded. In such cases, only IU characteristic measurements for individual modules are of help. This is extremely time-consuming and requires that the system receive at least 600 W/m² of sunshine in order to compare the measured values against rating plate specifications.
There is no quick or easy way to detect PID in the initial stages, and never without shutting down the system. Anyone willing to put in more effort can employ electroluminescence (EL) imaging at dusk, or better yet, at night. This enables accurate detection of PID even in the early stages. Countermeasures can be taken as a result, and this reduces yield losses. If there is a strong suspicion of PID, the extra work can pay off.

PID test methods

What PID test methods are available? Recently a number of different methods have been developed to test for PID. Two basic methods have become established for simulating surface conductivity. One method creates the effect by means of conductive foil, such as aluminum foil. The other does this using high air humidity in a climate test chamber.
For so-called PID tests, these two methods are combined with different test temperatures and testing periods. Table 1 (see p. 81) provides an overview of the most commonly used variations. Method 1 (aluminum foil) and method 4 (air humidity) are currently being discussed in the international standards committee of the IEC, TC 82 WG2. At present, a work item proposal is the only thing on the table. Tests are considered successful when modules lose no more than 5% of their initial power. But it makes a difference whether power degradation is measured at intense solar radiation (1,000 W/m² ) or under low light conditions (200 W/m² ). The effect is much greater under low light conditions since shunting effects, like PID, are more dominant at low light conditions. Moreover, there is currently no active IEC standard in place for use in testing and issuing IEC certificates.

Comparison of test methods

What does a comparison of test methods show? The tests differ in severity, reproducibility and informative value. And regardless of method type, temperature fluctuations change the speed at which degradation occurs. There is a rule of thumb that the degradation speed doubles for every 7 to 10 °C [4, 10, 11]. This means that at 50 °C, degradation occurs approximately eight times faster than at 25 °C. Thus the 25 °C test should last eight times longer than the test at 50 °C, if it is to be equally rigorous.
It is even more difficult to compare contacting methods. Contact via metallic foil is perfectly homogeneous throughout the module, with excellent conductivity that ensures an even distribution of potential throughout the module, irrespective of module size. By contrast, contacting via air humidity is more difficult to describe, and less homogeneous under certain circumstances, as illustrated by the comparison of EL images in Figure 4 (see p. 84). Under good conditions, contacting differences can be described by a factor of around ten, as Stephan Hoffmann demonstrated [4]. Thus for tests in the climate test chamber, testing periods ten times longer are needed compared with testing using the aluminum foil method. Accordingly, the 25 °C test with aluminum foil with a test period of one week is approximately as rigorous as the test in the climate chamber at 60 °C, with 85% relative humidity (RH) and a four-day test period.
The reproducibility of the tests depends primarily on the test method employed and the environmental control accuracy of the climate chamber (temperature and relative humidity).
When using the humidity method, constant regulation and homogeneity of humidity levels at all times in the climate test chamber are key factors. Stephan Hoffmann of the Fraunhofer Institute for Solar Energy Systems ISE [4, 7] and
###MARGINALIE_BEGIN###

###MARGINALIE_END###
Peter Hacke of the National Renewable Energy Laboratory NREL [8] examined the influence of relative humidity and found a strong dependence of leakage current in the range of 70% to 90% relative humidity.
This strong dependence on relative humidity, that is, on accuracy and the ability to regulate the climate test chamber, can influence the PID test results by a factor of approximately two.
In addition, when using the climate test chamber, if the humidity is too low, only the outermost cells of the module may end up being tested against their PID susceptibility. This makes it difficult to compare results and reduces the informational value of the test. Similar results can be found in the literature [12]. This problem does not exist for the methods using conductive foil. The foil always distributes the potential evenly over the cells, as evidenced by the comparison of the two EL images in Figure 4 (see p. 84). The aluminum foil-induced PID in Figure 3 on the right shows the degraded cells randomly distributed (see also [9]) across the module. On the left, after aging in the climate test chamber, the cells are systematically more severely degraded on the edges of the PV module.
For the standard, tests 1 and 4 in particular are being discussed. We consider the aluminum foil test to be extremely well suited for routine quality measurements of framed crystalline PV modules. It is easily reproduced, simple and cost-effective. If conducted for two weeks instead of 168 hours – that’s twice as long – it is also equivalent to test 2 (the test performed at 50 °C) in terms of severity.

Real world conditions

What does the PID test reveal about real-world applications? Degradation speed depends heavily on weather conditions. High humidity in the vicinity of the modules results in a tenfold increase in degradation, and moisture condensation on the modules or rain increases it one hundredfold compared with dry air [4]. With the aid of detailed weather data that include ambient temperatures, humidity, solar radiation levels, wind speeds and rain periods, the lifetime equivalent of a PID test can be roughly estimated.
For this comparison, we assume the following average annual weather conditions: rain for 90 minutes on 80 days, with rainfall occurring during the daytime on 55 of these 80 days. The temperature of the modules during the rainfall is approximately 17 °C. In addition, every other morning the modules are covered in dew for approximately 15 minutes at module temperatures of around 10 °C. To allow for muggy summer days, let’s assume that for 200 hours during the year, the relative humidity in the absolute vicinity of the modules reaches levels greater than 80%, and the modules then warm up to 33 °C. The rest of the time, the modules are dry or too cold for PID.
With these assumptions, we can calculate that the real conditions are equivalent to a 25 °C PID test as per the aluminum test method lasting approximately 90 hours (see Table 2, p. 84). To put it the other way around: Under the weather assumptions outlined above, a successful 25 °C PID test with aluminum foil, a 1,000 V proof voltage and a 168 hour testing period resulting in a 4% power loss means that a power loss of 5% after two years can be expected for modules with maximum potential to ground (1,000 V).
It should be noted, however, that 1,000 V to ground as in the PID test occurs in the system in only the rarest of cases. This must be taken into account when estimating the reduced output and service life in the system. Current research shows that there is a roughly linear relationship between the potential to ground and the degradation speed.
If we take the example of a non-grounded system with an 800 V system voltage and 20 modules, then the potential is distributed symmetrically around the point of origin. This means that the modules are subjected to voltages between -400 V and +400 V to ground. Only the first module is at -400 V; all others are at a higher voltage by multiples of 40 V, until +400 V is reached. If we now take a look at the idealized degradation speeds, the -400 V module degrades approximately half as quickly; at -200 V it is occurring at one-fifth the rate, and at 0 V degradation is no longer occurring at all, since typical crystalline p-type silicon modules are PID-stable at positive potentials. When this is taken into account, the successful aluminum foil test reveals that it should take significantly longer than two years for PID to become noticeable
###MARGINALIE_BEGIN###

###MARGINALIE_END###
in the system’s yield, even if individual modules on the edges are damaged more quickly.
So how long does a PID test have to be? There is no standard answer to this question. The most important influencing factor is the weather and the local environment in the vicinity of the module. The warmer and more humid it is and the more often it rains, the faster PID can be observed – if the modules happen to be susceptible to this kind of degradation. Testing over several weeks certainly does make sense. But other solutions also merit some reflection here. Perhaps at a certain point it is more reasonable to design the system such that the harmful voltages do not occur, or only occur for a short time, instead of installing PID-resistant modules. This is particularly true if systems over 1,500 V are considered in the future.

Reliability

Many manufacturers test according to conditions presented in Table 1. Yet whenever we perform check measurements, we regularly find that the modules supplied fail under identical test conditions. For a pooled test from laboratories, for instance, we recently checked eleven modules from four manufacturers using the aluminum foil method. The modules from two of the manufacturers were supposedly PID-free. After 168 hours as per test procedure 1, four modules had degraded by over 20% and did not pass the test; in two others the degradation was even greater than 60% (see Figure 5, p. 85). After a test lasting twice as long, only four modules showed less than 10% degradation, with no correlation with the manufacturer data on PID resistivity. In certain cases, different modules of the same module type and manufacturer showed very different results. We suspect that different cells, EVA or backsheet foils, or manufacturing processes were used in these modules. It is therefore a major shortcoming if manufacturers merely proclaim that their modules are PID-free but do not list the materials used. After all, it is not the module type that is critical, but rather the individual components of the module and the manufacturing processes. Certificates and test reports should therefore always be provided with the module’s exact bill of materials. If the bill of materials cannot be handed out for modules upon delivery, there is reason to be cautious: the consignment should be tested for PID and the purchase should be contingent on the results.

Avoiding PID

How can PID be avoided in the field? There are various ways to lessen the effect of degradation. One obvious way is to operate on low system voltages of around 500 VDC . Humidity and temperature are usually dependent on the location and difficult to control, but a module inclination greater than 15 °C provides favorable conditions for draining water away and washing off a portion of the dirt that contributes to the surface conduction.
Besides this, transformer inverters can be used, as these allow grounding of the solar generator through galvanic isolation from the grid. Inverters with higher efficiencies typically have no transformer.
If PID does occur or if laboratory tests have identified the modules as PID susceptible, other methods must be used. PID can be reversed in certain cases by applying an opposing potential to counter the damage-inducing potential. For this reason, it would be a reasonable strategy to move the modules around in the string at regular intervals, such that the degraded modules can be regenerated. This is extremely time-consuming.

Table 2: Roughly estimated, approximately 90 test hours with aluminum foil at 25 °C (procedure 1) corresponds to one year under real world conditions in Germany.and expensive, however. Alternatively, SMA for example has developed a PV Offset Box specifically for transformer-less inverters. The box can also be retrofitted on other manufacturers’ inverters. At night, it applies a positive potential of 1,000 V to the module strings, which reverses the degradation process.
Incidentally, frameless modules with a back side mounting device are less susceptible than framed modules, since the contact between the glass surface and the potential to ground is poorer in such modules. This is supported by an appropriate module installation. But these frameless modules must be designed appropriately and matched to the substructure to prevent other faults from occurring, such as cell breakage or damage to the backsheet layers. This could result in faults that may be even more serious than PID, since all modules in the installation would be affected. But with regular inspection of the modules and a suitable system design, it is indeed possible to bring PID under control.
###MARGINALIE_BEGIN###

###MARGINALIE_END###
###MARGINALIE_BEGIN###

###MARGINALIE_END###