Removed: Dynamic Model History Matching and Testing in Petroleum Reservoir Simulation

Abstract

When transition from static to dynamic reservoir modeling, historical field performance serves as a crucial benchmark. Unfortunately, freshly constructed geological models often fall short of accurately reproducing this historical behavior. To bridge this gap, the industry relies on a practice known as “history matching.” This chapter outlines essential principles of history matching, emphasizing not only the need to fit to historical data but also, more importantly, assessing the model’s ability to predict unseen data. A discussion on sources of error is followed by a review of classical history matching techniques, along with advanced methods available in modern software packages. Additionally, the chapter briefly explores neural networks as a potential avenue for improvement. It is important to recognize that history matching remains a challenge. As an inverse problem, it involves finding model parameters to align model responses with observed data. The under-determined nature of this problem adds complexity and is often compounded by data inconsistencies due to uncertainty or error. The chapter advocates viewing reservoir simulation not as a purely scientific endeavor but as a tool for informed business decision-making. Rather than aiming for exhaustive representation, models should focus on correctly forecasting critical characteristics relevant to field development decisions and reserves quantification.