Algorithm-Assisted Platform Location Optmisation Using Mixed-Integer Programming for Cluster Development Strategy in the Gulf of Thailand

Abstract

The unique characteristic of gas fields in the Gulf of Thailand is the compartmentalized reservoir that requires a huge number of producing wells. The task of locating platform locations for whole field perspectives that also meet all operational criteria while minimizing the number of needed platforms is challenging. This decisional task has a critical impact on project viability, especially for marginal fields. This paper proposes an innovative solution to strengthen success in this business decision by integrating subsurface domain knowledge and optimization algorithms. This study presents an optimization algorithm for determining the optimal locations of wellhead platforms with limited numbers to maximize hydrocarbon resources. Firstly, the algorithm performs verification matching between wellhead locations and subsurface targets throughout the field under drilling criteria. Next, the optimal platform locations are optimized using mixed-integer linear programming (MILP) with the primary objective of maximizing hydrocarbon resources. The algorithm literally runs with an increment in number of platforms until there is no incremental hydrocarbon resources gain and additionally summarizes the results as the number of platforms vs. coverage resources. The algorithm has proven its viability by recommending more optimal platform locations in an actual field in the Gulf of Thailand. This algorithm-assisted workflow was able to reduce the number of required platforms. The main driver for this improved decision is that the MILP algorithm manage to improve well targeting and platform location selection under a large set of practical constraints. In contrast, manual workflow retains its limitations to consider them all. This optimization also reduces the working time required for the whole process of well targeting and platform selection. Formerly, a typical workflow takes months of equivalent man-days. Conversely, this algorithm succeeded in completing the operation within just a few hours. Further, the subsurface team saved time by eliminating some repetitive tasks, i.e., they could focus on reviewing results extracted from the optimizer. Moreover, this work demonstrated the capability to extend and scaleup to other fields with similar concepts, which ultimately could lead to more benefits. This innovative workflow translates the complicated subsurface procedure to a numerical optimization problem with a solid proven benefit from real field implementation. Apart from the positive business impact, this study shows that we can promote integration between modern data analytics and domain knowledge in oil and gas industry.