Framework for embedding black-box simulation into mathematical programming via kriging surrogate model applied to natural gas liquefaction process optimization

This paper presents a framework to solve the constrained black-box simulation optimization problem that arises from optimal energy-efficient design of single-mixed refrigerant natural gas liquefaction process using reliable simulator. Kriging surrogate model is used introduce simple, computationally inexpensive, and effective algebraic formulations with derivatives objective constraints functions. The embedded into nonlinear programming (NLP) in General Algebraic Modeling System (GAMS). NLP solved efficient multi-start gradient-based CONOPT local solver determine candidate decision variables for which true functions are calculated rigorous simulation. analyzed considering one-to-three-stage expansion phase separation assess potential energy savings. present approach results show more stages can provide savings 12.02 14.70 % comparing two-stage three-stage system single-stage. consistent than Particle Swarm Optimization Genetic Algorithm given same budget evaluations considered problems, resulting 13.57 53.26