Well Placement Optimization under Uncertainty with CMA-ES Using the Neighborhood

Abstract

In the well placement problem, as well as in many other field development optimization problems, geological uncertainty is a key source of risk affecting the viability of field development projects. Well placement problems under geological uncertainty are formulated as optimization problems in which the objective function is evaluated using a reservoir simulator on a number of possible geological realizations. The existing approaches to cope with geological uncertainty require multiple reservoir simulations (on the possible realizations) to estimate the expected field performance for a given well configuration, which is often very demanding in CPU time and impractical when dealing with real field cases.In this paper, we present a new approach to handle geological uncertainty for the well placement problem with a reduced number of reservoir simulations. The proposed approach uses already simulated well configurations in the neighborhood of each well configuration for the objective function evaluation. We use thus only one single reservoir simulation performed on a randomly chosen realization together with the neighborhood of each well configuration to estimate the objective function instead of using multiple simulations on multiple realizations. This approach is combined with the stochastic optimizer CMA-ES (Covariance Matrix Adaptation -Evolution Strategy).The proposed approach is shown on the benchmark reservoir case PUNQ-S3 to be able to capture the geological uncertainty using a smaller number of reservoir simulations. This approach is compared to the reference approach using all the possible realizations for each well configuration. It is shown that the proposed approach is able to reduce significantly the number of reservoir simulations by more than 80% for the reservoir case in this study. Now with TOTAL. ECMOR XIII -13th European Conference on the Mathematics of Oil Recovery

Covariance Matrix Adaptation Evolutionary Strategy with Worst-Case Ranking Approximation for Min--Max Optimization and its Application to Berthing Control Tasks

“…, 𝑦 𝑠 } by discretizing the space Y. In this case, the min-max problem can be formulated as min 𝑥 ∈X max 𝑦 ∈𝑆 𝑓 (𝑥, 𝑦), and this formulation is employed in many applications, particularly in geo-science field [Bouzarkouna 2012;Miyagi et al 2019Miyagi et al , 2023Yeten et al 2003]. However, in this formulation, the worst-case function 𝐹 𝑠 := max 𝑦 ∈𝑆 𝑓 (𝑥, 𝑦) significantly depends on the discretization method of the space Y and the number of scenarios |𝑆 |.…”

“…In other words, a simulator such that ℎ sim (𝑥) ≈ ℎ real (𝑥) must be developed. However, owing to some real-world uncertainties, the predetermined conditions often contain errors and hence ℎ sim (𝑥) does not approximate ℎ real (𝑥) well [Bouzarkouna 2012;Chen et al 2013;Oberkampf et al 2002;Scheidegger et al 2018;Walker et al 2003]. In such situations, there is a risk that the optimal solution obtained in simulation-based optimization, 𝑥 sim = argmin 𝑥 ∈X ℎ sim (𝑥), does not perform well in the real-world and results in ℎ real (𝑥 sim ) ≫ ℎ sim (𝑥 sim ).…”

An efficient optimization process for hydrocarbon production in presence of geological uncertainty using a clustering method: A case study on Brugge field

“…This problem in petroleum community has been considered by selecting a finite number of N r realizations and modeling the objective function as below (Bouzarkouna et al, 2012):…”