Assessing green hydrogen production via offshore wind in the Dutch North Sea
Complementing techno-economic simulation with machine learning and optimization
Justin Starreveld (Universiteit van Amsterdam)
Laurens Frowijn (TU Delft - Energy and Industry)
Riccardo Travaglini (University of Florence)
Renske van ’t Veer (TU Delft - Energy and Industry)
Alessandro Bianchini (University of Florence)
Kenneth Bruninx (TU Delft - Energy and Industry)
Dick den Hertog (Universiteit van Amsterdam)
Zofia Lukszo (TU Delft - Energy and Industry)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This study analyzes the production of green hydrogen using dedicated offshore wind power in the Dutch North Sea region. The analysis is based on a detailed techno-economic model that simulates physical flows and estimates the levelized cost of hydrogen (LCOH). However, the model’s outputs depend on user-provided inputs and evaluating all possible inputs is computationally infeasible. To this end, “optimization with constraint learning” is employed, where surrogate machine learning models are trained on simulation data and embedded in mixed-integer optimization problems. The surrogate models are trained on 4096 simulation runs and achieve a mean absolute percentage error of ≤[jls-end-space/]3% for physical flow-related outputs, and an error of ≈[jls-end-space/]10% for the LCOH-related outputs. Once trained, these surrogates enable one to solve stakeholder–specific problem instances in sub-second solve times, supporting rapid scenario analysis and trade-off exploration.
help_outline