History matching through dynamic decision-making

Abstract

History matching is the process of modifying the uncertain attributes of a reservoir model to reproduce the real reservoir performance. It is a classical reservoir engineering problem and plays an important role in reservoir management since the resulting models are used to support decisions in other tasks such as economic analysis and production strategy. This work introduces a dynamic decision-making optimization framework for history matching problems in which new models are generated based on, and guided by, the dynamic analysis of the data of available solutions. The optimization framework follows a ‘learning-from-data’ approach, and includes two optimizer components that use machine learning techniques, such as unsupervised learning and statistical analysis, to uncover patterns of input attributes that lead to good output responses. These patterns are used to support the decision-making process while generating new, and better, history matched solutions. The proposed framework is applied to a benchmark model (UNISIM-I-H) based on the Namorado field in Brazil. Results show the potential the dynamic decision-making optimization framework has for improving the quality of history matching solutions using a substantial smaller number of simulations when compared with a previous work on the same benchmark.

A learning-from-data approach with soft clustering and path relinking to the history-matching problem

“…In this work, we propose a learning-from-data approach with path relinking and soft clustering to the history matching problem. Our strategy is a substantial evolution of the works presented in Cavalcante et al (2017Cavalcante et al ( , 2019. We use a similar learning-from-data approach that dynamically analyzes the data of available solutions to uncover patterns of reservoir uncertain properties that lead to good matching of the output variables considered in the history-matching process, and that can be used to guide the generation of new solutions.…”

“…We use a similar learning-from-data approach that dynamically analyzes the data of available solutions to uncover patterns of reservoir uncertain properties that lead to good matching of the output variables considered in the history-matching process, and that can be used to guide the generation of new solutions. As stated by Cavalcante et al (2017), the fundamental purpose of this learning-from-data approach is to give the algorithm the "potential to become a specialist on the structure of the history-matching problem, what according to Ho and Pepyne (2002), would definitely improve the chances of it being considered superior when compared to other strategies for the same problem".…”

“…(i) The first concerns the type of solutions considered during the process: While the strategy from Cavalcante et al (2017) analyzes input attribute patterns present in the best available solution for each well, our new learning-from-data approach, apart from the best solution for each well, also considers the best solutions for each variable involved in the historymatching problem. This characteristic broadens the learning spectrum of our algorithm, potentializing its ability to find solutions with better history-matching quality.…”

“…This characteristic broadens the learning spectrum of our algorithm, potentializing its ability to find solutions with better history-matching quality. (ii) The second difference from the approach proposed by Cavalcante et al (2017) is related to the strategy used to delimit reservoir regions changed during the history-matching process. Both works dynamically delimit these regions' clustering reservoir grid blocks considering their spatial and petrophysical properties.…”

“…Both works dynamically delimit these regions' clustering reservoir grid blocks considering their spatial and petrophysical properties. However, while Cavalcante et al (2017) adopted a hard assignment clustering approach, in this work, we propose a soft clustering strategy to delimit the reservoir regions. Our primary motivation for this proposal is to potentially consider the impact of each reservoir grid block on the performance of one or more wells.…”