Optimising Electrolyser Operation for E-Methanol Production Using Renewable Electricity and Price Forecasting
Translated title
Optimering af Elektrolyserdrift Til E-Methanol Produktion Ved Hjælp af Vedvarende El- og Prisprognose
Author
Dzigurski, Filip
Term
4. term
Education
Publication year
2025
Submitted on
2025-05-30
Pages
96
Abstract
Producing e-methanol with wind and solar power is challenging because electricity prices change rapidly. This thesis develops a modelling framework that links short-term electricity price forecasts to the operation of an electrolyser, the electricity-driven unit in the process. Machine learning methods, including XGBoost and LSTM, are used to predict prices and enable predictive control. These forecasts feed a scheduling model based on mixed-integer linear programming (MILP) that chooses among three operating states (running, standby, and shutdown) to keep methanol output fixed while minimizing total production cost. The model includes realistic features such as constraints on switching between states, variable electricity pricing, and costs of sourcing CO2. Scenario analysis shows that electricity prices and electrolyser efficiency are the main cost drivers, whereas changes in CO2 cost have only a limited effect. The findings highlight that accurate short-term forecasting combined with flexible operation can improve economic viability. Future work should prioritize advances in electrolyser technology, strategies to reduce electricity costs, and real-time model integration.
Produktion af e-metanol med vind- og solenergi er udfordrende, fordi elpriserne svinger hurtigt. Denne afhandling udvikler en modelleringsramme, der kobler kortsigtede elprisprognoser til driften af en elektrolysør, den el-drevne enhed i processen. Maskinlæringsmetoder, herunder XGBoost og LSTM, bruges til at forudsige priser og muliggøre forudsigende styring. Disse prognoser indgår i en planlægningsmodel baseret på blandet heltals-lineær programmering (MILP), som vælger mellem tre driftstilstande (i drift, standby og slukket) for at holde e-metanoloutputtet fast og samtidig minimere de samlede produktionsomkostninger. Modellen indeholder realistiske elementer som begrænsninger for skift mellem tilstande, variable elpriser og omkostninger ved CO2-indkøb. Scenarieanalyse viser, at elpriser og elektrolysørens effektivitet er de vigtigste omkostningsdrivere, mens ændringer i CO2-omkostninger kun har begrænset betydning. Resultaterne understreger, at præcise kortsigtede prognoser kombineret med fleksibel drift kan forbedre den økonomiske rentabilitet. Fremtidigt arbejde bør fokusere på forbedringer af elektrolyseteknologi, strategier til at reducere elomkostninger og realtidsintegration af modellen.
[This apstract has been rewritten with the help of AI based on the project's original abstract]