Systematic Optimization of the Design of Steam Cycles Using MINLP and Differential Evolution

Abstract

The process synthesis and design optimization of energy conversion systems can be modeled as a mixed integer nonlinear programming (MINLP) problem. The nonconvexity potential and the combinatorial nature of the objective functions and constraints largely suggest the application of heuristic search methods for global optimization. In this paper, a modified differential evolutionary algorithm is applied to a MINLP problem for optimizing the design of steam cycles based on a complex superstructure, containing a variable number and varying positions of reheatings, varying layouts of the feedwater preheating train, and a boiler feedpump turbine with steam extractions. The energy-savings potential from the existing system design was studied. The optimization of a 262 bar/600 °C/ 605 °C unit with a single reheat shows that an efficiency improvement between 0.55 percentage points (PP) and 1.28 PP can be achieved. The optimal design of steam cycles over 650 °C was found to be different from those of the designs under current steam conditions: a transition throttle pressure, above which the benefits of steam temperature elevation can be completely realized, is critical and, accordingly, three design zones associated with the match of throttle pressure and the steam temperature level are clearly identified with recommended ranges of reheat pressures.