What is it about?
Although robust, a photovoltaic (PV) system is not invincible nor immortal. In spite of typical operation lifetimes beyond 25 years, it may be broken, soiled, malfunction or degrade faster than expected. A standard approach to assess a system’s “health” is to model its expected generation and compare it to actual measurements. We suggest a simple modelling option: by looking at the previous two weeks, it is estimated how a PV system and a radiation sensor behave in a clear-sky day (with no clouds). Comparing the clear-sky expectation with the actual measurements can lead to two possible conclusions: when a PV system “says” it is a cloudy day, either this claim is supported by the radiation sensor, or the system is malfunctioning.
Featured Image
Why is it important?
As a long-term investment, it is very important to monitor how well a PV system is working during its lifetime. If it is malfunctioning, it is not generating energy nor profit. Standard assessment approaches are very data-intensive, requiringl technical details and weather recordings to estimate how much light reaches the PV modules and how efficiently it is converted to electricity. Gathering such data may be financially challenging for small rooftop PV systems; on the other hand, grid operators need to estimate a PV plant generation, without any access to its privately-own technical details. Thus, our methodology presents itself as a very interesting option, bypassing the use of most of these inputs. Furthermore, it is easily automated and therefore may be applied at little cost to many different PV systems.
Perspectives
This was my second PhD publication. Using clear-sky models so many times for my spatio-temporal forecasting work, it just came up that crossing this information from both irradiance and PV could be useful for performance assessment.
Rodrigo Amaro e Silva
Faculty of Sciences, University of Lisbon
This article is part of Rodrigo's ongoing PhD work focusing on solar radiation forecasting based on electrical output of spatially distributed PV systems. The main motivation is that PV system are becoming ubiquitous in contemporary urban landscapes and are producing very relevant and very large amounts of data that can be very useful for many applications. This 'PV performance assessment method' is one of these very interesting, and simple, applications.
Miguel C Brito
Lisboa
Read the Original
This page is a summary of: Data-driven estimation of expected photovoltaic generation, Solar Energy, May 2018, Elsevier,
DOI: 10.1016/j.solener.2018.03.039.
You can read the full text:
Contributors
The following have contributed to this page
