This paper explores the use of divide and conquer algorithms to solve production optimization problems. Exploitation of inherent structure facilitates efficient formulations and solutions to production optimization problems that frequently appear in the upstream oil and gas sector. We present the daily production optimization, which gives rise to network optimization problems, and also add some comments to the closely linked reservoir optimization problem. Subsequently, a common non-invasive optimization approach is presented prior to three concepts that apply decomposition and structure exploitation. The first approach splits a network simulator into interconnected parts and exploits the local components and their couplings in devising efficient formulations and algorithms. This opens for a range of opportunities in terms of formulations and algorithms, several of which will be discussed. This part is accompanied by a realistic numerical study. The second approach exploits a decomposable network structure, which frequently appears in large production systems, by applying a Lagrangian type decomposition. The last approach deals with reservoir optimization where there is no apparent structure to exploit. However, by recognizing this as a dynamic optimization problem, it may be split in the time domain. The paper ends with some conclusions.
展开▼
