Nos recherches, expériences & conceptions

Conception of a Global Digital Twin to Evaluate the Impact of Climate Change on the Production of Photovoltaic Power Plants

Photovoltaic (PV) electricity production is growing strongly worldwide, at around +20% every year.

Nevertheless, the question remains as to how the climate change effect will affect the performance of PV plants in the future.

Some studies have already demonstrated that the impact of global warming on photovoltaic (PV) energy production could result in an average reduction in annual output of 15 kWh/kWp by 2100, with potential losses of up to 50 kWh/kWp in the most severely affected regions [1].

Other studies indicate that, on a European scale, the loss of production linked to changes in solar resources should be within the range of (–14% to +2%) by the end of this century compared with estimates made under current climate conditions [2].

In both instances, the predictions are based on climate models combined with electrical models. The latter of which are frequently highly simplified, as they only consider the reduction in yield from photovoltaic panels due to the rise in temperature.

Concurrently, from an economic standpoint, it appears that the profitability margins of current power stations are becoming increasingly slim . A loss of yield of just a few percent could potentially lead to the questioning of the viability of certain installations.

In this context, the objective of the proposed project is to develop a numerical modelling tool (digital twin) that will enable the overall performance of a photovoltaic power plant to be assessed with sufficient accuracy to meet current and future economic and climatic challenges.

Ultimately, this digital twin should be able to take account of the main environmental changes (rise in temperature, changes in the solar spectrum, atmospheric pollution, etc.) in all the plant's components (modules, electronics, cables, etc.).

In order to achieve this, it is necessary to employ multi-scale modelling techniques, utilising physical models of cells, modules and power electronics.

Another challenge concerns the integration of the data needed to represent this complexity.

The combination of physical models with data-based models (of the machine learning type) can enhance accuracy while maintaining a reasonable calculation time.

The initial stage of this model’s development will entail a comparison with data from existing power plants. This will allow for an assessment of the robustness of current technologies in the context of climate disruption scenarios.

The planned digital twin will be constructed from “physical bricks,” which will enable the assessment of the impact of each of the plant’s constituent bricks on its overall behaviour.