The long read: How can yield be increased, and by how much?


From pv magazine 09/2020

How, and how much, can project developers and O&M service providers boost their earnings? Finding the answer to this question, or at least experts to help us answer it, was the task we set for ourselves in the run-up to the pv magazine Roundtables Europe on Asset Management, which took place in June – online, of course, due to the pandemic.

There has been no shortage of talk and reporting on this subject in recent years, including within the pages of this magazine. Vendors of monitoring and asset management software such as Meteocontrol, Solytic, Envision and others are developing methods using monitoring data to identify faults automatically and calculate their cost over a given timeframe. This figure can then be compared with the cost of repairs to decide when a fix is worthwhile. But this is not as easy as it sounds.

Virtual Roundtables USA

We will be hosting the second edition of our successful Virtual Roundtables this year on November 17. The program will be focused on two of the hottest topics in the U.S.: Battery Storage: Storage technologies assume a bigger role in the global transition to renewable energy, and we will have a look at the trends and standards. Distributed Generation and Utility Scale are in focus. Quality: More than ever it is important to cut down costs and optimize your LCOE. We will discuss the importance and challenges of smart O&M. Eric Wesoff, Editorial lead of the U.S. platform, and his fellow pv magazine editors will moderate the Virtual Roundtables USA to provide a great virtual stage for our speakers. Apart from the in-depth quality program sessions, you can also look forward to a multitude of virtual networking opportunities.

“Defining the reference value for the performance ratio alone is no mean feat,” says Boris Farnung, until recently head of the PV Power Plants group at Fraunhofer ISE. Without this, however, it is impossible to know how much a system falls short of expectations. There is also no clear method of defining an up-to-the-minute performance ratio. For instance, whether to deduct inverter failures or not depends on the purpose of the calculation, which is why there is still debate over how monitoring platforms are currently helping to establish the correlation between action and returns.

Data dilemma

A case study from the Mediterranean region shows why it is not always easy to determine when repairs make sense. When Lucie Garreau-Iles first visited the site, its 1.5-meter-high grass probably made the same enchanting impression that has captured the hearts of so many tourists around the Mediterranean.

Yet, for all its charm, the high grass was one of the reasons why analysis of the monitoring data from the 3.5 MW solar plant offered few insights. Garreau-Iles works as a technical regional manager at DuPont Photovoltaic Solutions, as part of a team investigating the condition of solar installations. She came to the site to see if the yield of the eight-year-old system could be improved.

Based on the monitoring data, it was difficult even to judge whether the plant had a yield problem at all, since the weather sensors had failed repeatedly in the past. During the first two years of operation there was a complete lack of data, and after that the quality was poor, with some of the data either missing or incorrect. In this case, perhaps it was of little concern to the investor, because a high feed-in tariff ensured adequate returns.

Performance losses

During her site visit in 2019, Garreau-Iles found many modules with snail trails and delaminated EVA film. “There were numerous visual anomalies,” she says. Determining the extent to which the performance of modules is affected by snail trails, delamination and microcracks is not straightforward.

Microcracks do not affect performance, until with time and certain field conditions, they open into full fledged cracks that damage the cell interconnection, causing yield to drop. In such cases, the challenge is in determining performance loss with the least effort – on the one hand, so as not to spend too much money before it is clear whether the expense is worthwhile, and on the other, the parties involved often require accurate figures, even when it is clear that modules will have to be replaced anyway.

The next step was to have the plant inspected using drone thermography. This clearly showed that some 20% of the modules had hot spots distributed throughout the entire array.

Representative results

Measuring current-voltage characteristics (I-V curves) is a good way to determine power loss, but it is fairly labor-intensive, which is why the experts selected a representative sample of strings for characteristic curve measurements based on the IR images. This made it possible to measure performance with a high degree of accuracy, despite only half of the strings being tested. Some strings were tested before and after cleaning, and thus determined a performance drop of around 1.6% due to soiling. In addition, the thermography showed that 3.5% of the solar modules were not connected. Measurement of the characteristic curve itself showed that the output of the remaining connected modules had decreased by 19.3% on average. All told, the overall performance reduction was 22.8%.

“From this, we were able to conclude that repairs were worthwhile,” says Garreau-Iles. The team calculated that replacing modules and upgrading the installation would increase revenues by 17%.

The repairs are currently underway. All of the modules are being replaced, even those without hotspots, as they may have microcracks. These can develop into fractures with or without hot spots and thus lead to a drop in output. “Replacing only some of the modules is therefore not cost-effective,” says Garreau-Iles.

Big boost

The question of whether it is possible to quantify the relationship between maintenance expenditures and increased yield can only be answered on a case-by-case basis. In some installations, one knows almost nothing in advance, similar to the example described above. In such cases, a study can be commissioned to clarify the situation.

However, experts say that it is often impossible to judge in advance whether the cost of such a study will be recouped by an increase in revenue, and service providers make no guarantees in this respect. Then there are installations for which such an investigation has already been carried out, in connection with a sale, for example, and have been inspected for this purpose. Decisions are easier to make with such plants.

Duncan Bott has extensive experience with investments in the secondary market. He works for asset management company LCF Alliance, and has purchased 73 solar PV plants over the past few years, most of them five to six years old. His experience is that many investors are passive and do not make the most of their investments. He, on the other hand, has subjected his investments to an optimization process. Although every system is different, he says that with time it is possible gets a feel for what you are buying, like an experienced car mechanic who can get a fairly good idea of the condition of a car by looking under the hood.

“On average, we’ve been able to increase yield by 4.8%,” he says. He simply measures the performance ratio before and after the optimization process.

Optimizing performance

A study led by Fraunhofer ISE’s Boris Farnung investigated degradation in 44 solar plants on the roofs of Aldi Süd buildings with capacities between 0.5 MW and 1.5 MW (Degradation in PV Power Plants: Theory and Practice, 36th European EU PVSEC). Plants aged 10 years and older had an average annual degradation rate of 0.7%, or around 7% over 10 years. And even that figure can be improved, according to the study’s authors. Testing showed that the modules, which were studied separately, had annual degradation rates of less than 0.24%, many even less than 0.1%.

Reasons for system degradation in the plants studied were mainly inverter faults, partial inverter failure, and soiled modules. “If you work on it, you can improve the performance of the systems accordingly,” says Farnung. Also of interest is the fact that failures were less frequent with central inverters, but when they do fail they have a much greater impact on performance, often leading to output that lags behind that of plants with string inverters.

It is important to remember, however, that the equipment studied in the paper seems to have been built and maintained with great care, says Lucie Garreau-Iles. The fact that 99.5% of the data were available is not the normal situation. Neither is the fact that the modules had already been thoroughly evaluated prior to installation. The research thus shows what improvements can be made even in well planned and maintained systems. With a typical plant on the market, the potential should be even greater.

This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: