Leveraging LLMs in Reservoir Simulation Workflows: Opportunities and Pitfalls

This paper explores the application of Large Language Model (LLM) in reservoir simulation workflows with the aim of improving efficiency of analysing simulation data. The initial goal was to use LLM to generate code from a code repository focussing on automating reservoir simulation workflows, but it pivoted to generating insights from reservoir simulation inputs and outputs. In particular, the opportunities and pitfalls of using LLM in this context are highlighted. This paper concludes that LLM serves best as augmentation tools in reservoir simulation, offering time savings and improved accessibility while still requiring human oversight for complex or nuanced interpretations.