Multi-period steam turbine network optimisation. Part I: Simulation based regression models and an evolutionary algorithm for finding D-optimal designs

In this work, a methodology for building multi-period optimisation model of steam turbine network is presented. The optimisation model can estimate and evaluate the effect of changes to the thermal energy demand of processes. The subject is divided into two parts. In Part I, a method for finding regression models for steam turbine networks using a simulation model and an evolutionary algorithm for finding D-optimal designs is presented. In Part II, the method presented in Part I is used to develop and solve a multi-period MINLP model of a steam turbine network in a utility system. There are two major problems that are addressed in Part I; Firstly, the evolutionary algorithm for finding D-optimal design is applied to try to determine which values should be simulated in order to generate the data for the regression model. Secondly, a theoretical model of steam turbine performance is used to model the feasible operation of the steam turbine. This is necessary to be able to make an efficient evolutionary algorithm for finding D-optimal designs. The use of regression models makes it possible to build optimisation models of utility systems that are compact and transparent. There is a natural trade-off between the flexibility of a model and the accuracy. The major drawback of the methodology, is that the models developed must be considered ‘ad hoc’-models, and are not as flexible compared to models where all the process units are modelled in full detail. An advantage of the methodology developed in this work is that it gives more possibilities of finding an acceptable simplification of the optimisation problem, as the methodology is not bound by a certain set of thermodynamic rules or specific mathematical form of the relations in the models. Part II demonstrates how the methodology can be applied when building a multi-period optimisation model to estimate and evaluate how changes to the processes will affect the utility system.