Combustion optimization and KCl emission control based on MOMPC for BCFB boiler load variation

• Multi-objective model predictive controller (MOMPC) based on multiple models. • Bed temperature, flue gas O 2 , CO 2 , and emissions were analyzed under varying loads. • Based on the subspace identification method, state space submodels of different loads are established. • MOMPC can achieve low KCl emissions and high boiler efficiency simultaneously. Biomass direct-fired circulating fluidized bed (BCFB) boilers face significant challenges in coordinating chloride emission control with boiler efficiency, particularly under variable load conditions. This study aims to develop a multi-objective model predictive control (MOMPC) strategy to simultaneously optimize combustion efficiency and suppress KCl emissions in a 130t/h BCFB boiler. Based on a mechanistic model validated against operational data, subspace identification was employed to establish state-space sub-models at three load levels (60%, 80%, and 100% BMCR). A MOMPC framework was then designed to handle the nonlinear and non-minimum phase characteristics of the combustion process, with primary and secondary air valves as control inputs and boiler efficiency and flue gas KCl concentration as controlled outputs. Simulation results demonstrate that MOMPC effectively improves boiler efficiency and reduces KCl emissions under both steady and varying loads. Notably, under variable load conditions, the multi-model MOMPC outperforms single-model approaches, achieving a 3.99% reduction in KCl concentration at 80% BMCR compared to equal air distribution. This study provides a practical control solution for enhancing the operational flexibility and environmental performance of biomass-fired power plants, contributing to cleaner energy production and extended equipment service life.