WIT Press


Optimization Of The Underground Gas Storage In Different Rock Environments

Price

Free (open access)

Paper DOI

10.2495/OP120021

Volume

125

Pages

12

Page Range

15 - 26

Published

2012

Size

494 kb

Author(s)

S. Kravanja & B. Žlender

Abstract

This paper presents the cost optimization of underground gas storage (UGS), designed from lined rock caverns (LRC). The optimization is performed by the non-linear programming (NLP) approach in different rock environments. For this purpose, the NLP optimization model OPTUGS was developed. The model comprises the cost objective function, which is subjected to geomechanical and design constraints. The optimization proposed is to be performed for the phase of the conceptual design. A numerical example at the end of the paper demonstrates the efficiency of the introduced optimization approach. Keywords: underground gas storage, UGS, lined rock cavern, LRC, rock mass rating, RMR, optimization, non-linear programming, NLP. 1 Introduction This paper deals with the optimization of the investment and operational costs of the underground gas storage (UGS), designed from lined rock caverns (LRC) [1–3]. The optimization is performed by the non-linear programming (NLP) approach. For this purpose, the NLP optimization model is developed. Since the optimization is proposed to be performed for the phase of the conceptual design, only some basic conditions are defined in the optimization model in order to assure enough strength safety of the rock mass and impermeability of the cavern wall and steel lining. The latter is achieved by the limitation of steel lining and concrete wall stains. Since there exist various rock masses with enough strength to support the UGS, the optimization of the UGS is proposed to be calculated in different rock environments. For this purpose, a rock mass classification – the so called rock mass rating (RMR) system is used.

Keywords

underground gas storage, UGS, lined rock cavern, LRC, rock mass rating, RMR, optimization, non-linear programming, NLP.