The reservoir screening standard of CO2 huff-n-puff based on orthogonal analysis method and random forest algorithm

Abstract

In heavy oil reservoirs, favorable reservoir properties have a positive impact on the production performance during CO₂ huff-n-puff. It is significant to study the screening method of applicable conditions for CO₂ huff-n-puff in actual reservoirs. To solve these problems, this paper introduced the orthogonal design method to analyze the main factors based on numerical simulation. The technical analysis and the economic evaluation were both employed to obtain the applicable conditions of selecting oil layers or injection wells during CO₂ huff-n-puff. And a new algorithms of machine learning, the random forest algorithm, was introduce to find the weighted factors and the scoring standards that were suitable for CO₂ huff-n-puff. Finally, a set of method for screening suitable reservoir conditions was established. Based on the introduction of orthogonal analysis method and random forest algorithm, a software was established to achieve the purpose of analyzing the feasibility of CO₂ huff-n-puff considering different reservoir geological parameters. This method increased the accuracy and efficiency in screening reservoir conditions that was suitable for CO₂ huff-n-puff.