La plupart des projets photovoltaïques s'appuient sur une seule valeur P50 pour représenter le rendement énergétique annuel prévu. Or, un seul chiffre ne suffit pas à rendre compte de la situation dans son ensemble et ne permet certainement pas de cerner l'éventail des résultats auxquels votre projet pourrait être confronté.
Dans ce webinaire, nous analysons en détail les facteurs qui contribuent réellement à l’incertitude du rendement photovoltaïque et expliquons pourquoi il est essentiel de la comprendre pour la rentabilité du projet. Vous découvrirez les trois principales sources d’incertitude dans les calculs de rendement : la modélisation du rayonnement solaire, la simulation photovoltaïque et la variabilité interannuelle. Vous apprendrez également comment la réduction de cette incertitude permet de prendre de meilleures décisions de conception et d’obtenir des conditions de financement plus avantageuses.
Regardez l’enregistrement pour découvrir comment Solargis calcule l’incertitude liée au rendement photovoltaïque en s’appuyant sur des données solaires et météorologiques de référence, des approches de modélisation éprouvées et une méthodologie transparente.
In projects we see in the deserts, one of the highest concerns is soiling. Dust accumulation in such environments is rapid and often underestimated at commissioning stage.
Beyond that, it is difficult to figure out without knowing full details. Which soiling model did you use? Which input dataset and simulator?
In our sample project, we have deliberately made the assumption that the P50 stays the same to evaluate the effect of uncertainty reduction. In practice, a more detailed method may refine the P50 as well, but the main benefit we wanted to highlight during the webinar is a narrower uncertainty range, meaning P75/P90 or any other Pxx values move closer to P50.
Yes, confirmed, the solar irradiance uncertainty estimate we provide in Evaluate is site-specific, not a regional average.
It is requested on demand in one of the sections and appears in the online results once the analysis is finished.
Interannual variability can also be easily extracted from PVOUT standard deviation.
Note that PV simulation uncertainty is still not covered in the platform and must be estimated separately.
It depends on when the satellite mission covering a particular region started. As a result, the available data record may begin in 1994, 1999, or 2007, depending on the location. More details can be found here.
The longer the data record, the more robust the estimate of interannual variability.
That said, because satellite observations are continuously incorporated into the irradiance model, it's generally best to work with the most up-to-date dataset available.
This allows you to benefit from a longer historical record while also capturing any recent trends that may be relevant.
That depends on the specific site and power plant complexity. PV simulation can be the dominant factor in more complex projects in well-known areas where the irradiance model performs better.
Similarly, the solar irradiance model can be the dominant factor in relatively simple projects located in areas where the irradiance model is expected to have higher deviations.
On the other hand, interannual variability can be significant in areas where cloud variability is especially challenging to capture (e.g., rapidly changing clouds within 10–15 minutes) and can even become the largest contributing factor in some cases.
We generally do not recommend this approach. It involves combining data from different models, which are likely to have different levels of consistency, methodology, and accuracy.
As a general rule, when measuring any parameter, it is better to rely on the most accurate and well-validated source available. The same principle applies to solar irradiance models. Combining results from multiple models does not necessarily reduce uncertainty and, in some cases, may even introduce additional sources of inconsistency.
All of them are used, but for different purposes. While P50 is used as the base case for financial modeling, P90 is a common metric for debt service coverage ratio (DSCR) calculations.
In general, any Pxx value can be calculated according to the requirements of the project stakeholders.
We use our own Solargis satellite model for solar irradiance and high-quality NWP reanalysis models, primarily ERA5, for additional meteorological parameters such as temperature, wind speed, and humidity, among other parameters.
More information can be found here.
As many years as you can get. The main limitation is the start date of the satellite missions covering the region.
As a practical minimum, at least 11 years of data is generally recommended to derive statistically meaningful Pxx values.
However, longer and more up-to-date datasets are preferable, as they provide more robust estimates of interannual variability while also capturing recent trends.
At a global level, Solargis GHI shows a mean bias of 0.5% and a standard deviation of 3% for annual values when benchmarked against 320 reference measurement sites.
For region-specific results, we recommend consulting the latest validation report published by Solargis, which also includes additional metrics demonstrating model performance and consistency at hourly and monthly timescales.
You can read the detailed validation statistics here.
We do use in-situ data for model validation purposes. Once the model has been validated, it is applied to the project site using PM data from reanalysis models, primarily CAMS, as an input. A full explanation of how Solargis soling model works is here.
If local in-situ soiling or dust measurements are available near your project site, we can assess whether incorporating them would improve the accuracy of the soiling loss estimate.
This is not currently offered as a standard Solargis software feature, but our consulting team can evaluate it on a case-by-case basis.
The reduction in uncertainty is based on well-established scientific methods. Uncertainty is quantified by benchmarking model performance against a large number of ground measurement stations across a wide range of conditions.
The key message for banks and lenders is that a more detailed, site-specific uncertainty assessment, rather than a generic regional one, typically results in a narrower confidence interval. This, in turn, leads to a higher P90 value and can support increased debt capacity.
The exact impact on debt sizing, for example, whether the debt ratio increases from 75% to 78%, depends on the lender's DSCR requirements and the specifics of the project's financial model.
