Optimal facility placement is key in field development planning. However, this process should be integrated with well placement and well trajectory design which poses a real challenge especially early in the field development planning; at the concept screening phase. During that phase, multiple concepts need to be assessed within, typically, a tight project time line. The integrated workflow becomes prohibitively expensive and calls, hence, for a fast and robust method that fulfils the objectives in the context of real onshore and offshore field development planning. Facility placement optimization consists mainly of minimizing the resulting cost while honouring topological complexities and prescribed capacity and trajectory constraints. It aims at selecting the optimal number and location of the different “nodes” comprising the facility and the optimal paths of the connections (pipelines) between nodes. Although many optimization schemes are developed and widely available in the literature, these methods are prohibitively slow with exhaustive search required, or they do not account for the various topological complexities typically encountered in real scenarios. Thus, the deficiency associated with these optimization schemes are drastically limiting in the case of the concept-select phase of field development planning where many scenarios need to be considered; in a relatively short timeframe. Deterministic optimization methods hit memory and computational limitations for all but trivial scenarios and, hence, fail to address the problem. Recently, genetic algorithms were applied with promising results addressing this challenge. In this paper, particle swarm optimization (PSO) is used for facility placement optimization and coupled with the A* algorithm for pipelines layout. This paper illustrates the capability of the implemented modular PSO-based approach for facility placement optimization through various levels of problem complexity: multiple facility layers, topological complexity and prohibited areas. The proposed algorithms are apt to be used in real field development planning projects and are, to be best of our knowledge, the first published material addressing the problem with the required level of speed, robustness and integration.
Read full abstract
Jun 7, 2024
Christopher Merrick 
Open Access
