Optimization techniques applied in a tuning process of a feedwater heater’s first-principle data-driven model

Abstract

Work related to the tuning of the first-principle model of a feedwater heater operating in a coal-fired power unit is presented, along with discussion concerning the most efficient and accurate tuning algorithms based on direct-search, first- and second-order optimization techniques. The objective of this work is to find the most efficient and accurate algorithm to tune the model parameters, that is, heat transfer coefficients based on the algorithms’ benchmarking study. The model variables (e.g. variability of the power rate of energy exchange) and estimated parameter values were used to formulate key performance indicators intended for a model-driven diagnostics approach. The computational process was organized in an iterative process of updating model parameters and indicators. The validation was successfully performed using operational data from a 225 MW coal-fired power unit.