Analyzing Hydrogen Flow Behavior Based on Deep Learning Sensor Selection Optimization Framework

Abstract

With tremendous potential to lower the carbon footprint of power generation and as an alternative energy carrier for many applications, hydrogen has emerged as a key potential energy carrier. Understanding of geological conditions and the injection and production changes over time for hydrogen storage are paramount, requiring in situ reservoir sensing options. Determining the flow behavior is of critical importance in order to estimate hydrogen volumes within the reservoir. A novel AI-driven methodology for hydrogen flow behavior and volume estimation was demonstrated. For determining the S1 through S3 predicted hydrogen storage quantities, the framework is linked to an uncertainty estimate framework. The Pohokura field in New Zealand served as the basis for the framework's evaluation, and it performed acceptably in terms of identifying the hydrogen flow behaviors and quantities inside the subsurface reservoir. The framework is a significant first step in assessing the amount of hydrogen that can be stored in subterranean reservoirs for long-term hydrogen storage.