Model-driven soft sensor for predicting biomass calorific value in combustion power plants

Abstract

An efficient way to exploit the chemical energy contained in the biomass is combustion using the state-of-the-art grate-firing systems, one of the most competitive and market-proven technology being currently used in Europe for power production: electricity and/or steam for district heating, which are offered on the market with capacities that range from 20 up to 50 MWel. A decisive variable, when operating the biomass heat recovery power plant, is the biomass calorific value, which often fluctuates between the different batches delivered to the furnace, due to harvesting, storing and transport conditions. Consequently, the operation of the grate-firing unit is complicated as a result of the uncertainty introduced by the fuel quality. Therefore, it will be advantageous, from a control point of view, to monitor online this value in order to effectively control the combustion-air system and the biomass feed rate. Thereby, increasing the plant's conversion efficiency. In this work, an estimator based on a static combustion model, along with a steam-boiler dynamic model, is developed to monitor the biomass calorific value in a grate-firing unit. The steam-boiler model predicts the amount of steam based on the released thermal power from the furnace side. Consequently, this allows for the estimation of the fuel's energy value based on measured and predicted amount of the steam produced by firing the solid biomass. Hence, no further devices, hardware calibration and additional costs are needed.