Optimal cost and design of an underground gas storage by ANFIS

Abstract

We present an optimal cost and design prediction of an underground gas storage (UGS) system, which is proposed to be constructed from one or more lined rock caverns. The adaptive network based fuzzy inference system ANFISUGS was generated to predict minimal investment costs and optimal UGS design. Since a safe and impermeable UGS system requires a rigorous calculation, three steps were proposed to solve this task: the first is solving the geotechnical engineering problem for different UGS designs, the second is the cost/design optimization of the UGS structures, and the last is the generation of an ANFIS system for optimal cost and design prediction of the UGS. While the geotechnical problem was solved with a series of finite element analyses in order to define special geotechnical constraints to be put into the optimization models, a parameter non-linear programming (NLP) optimization approach was used for a variety of different UGS design parameters. The ANFISUGS system was then constructed on the basis of data sets defined from previous NLP optimization results. A case study demonstrates the effectiveness and the prediction capability of the proposed ANFISUGS system.