Green hydrogen techno-economic assessments from simulated and measured solar photovoltaic power profiles

Studies estimating the production cost of hydrogen-based fuels, known as e-fuels, often use renewable power profile time series obtained from open-source simulation tools that rely on meteorological reanalysis and satellite data, such as Renewables.ninja or PVGIS. These simulated time series contain errors compared to real on-site measured data, which are reflected in e-fuels cost estimates, plant design, and operational performance, increasing the risk of inaccurate plant design and business models. Focusing on solar-powered e-fuels, this study aims to quantify these errors using high-quality on-site power production data. A state-of-the-art optimization techno-economic model was used to estimate e-fuel production costs by utilizing either simulated or high-quality measured PV power profiles across four sites with different climates. The results indicate that, in cloudy climates, relying on simulated data instead of measured data can lead to an underestimation of the fuel production costs by 36 % for a hydrogen user requiring a constant supply, considering an original error of 1.2 % in the annual average capacity factor. The cost underestimation can reach 25 % for a hydrogen user operating between 40 % and 100 % load and 17.5 % for a fully flexible user. For comparison, cost differences around 20 % could also result from increasing the electrolyser or PV plant costs by around 55 %, which highlights the importance of using high-quality renewable power profiles. To support this, an open-source collaborative repository was developed to facilitate the sharing of measured renewable power profiles and provide tools for both time series analysis and green hydrogen techno-economic assessments. • Impact of using simulated or measured PV profiles on e-fuels feasibility studies. • E-fuel techno-economic assessments amplify errors from PV simulation tools. • All PV profile simulation tools lead to an underestimated LCOF (maximum 36 %). • The LCOF error is reduced for flexible hydrogen consumers. • An open-source repository has been created to replicate the analysis at other sites.