Carbon monoxide emission models for small-scale biomass combustion of wooden pellets

Tighter legal emission limits require means to prevent releasing harmful substances into the atmosphere during the combustion of biomass. Economic considerations suggest to meet these restrictions by improving the ability to predict and therefore prevent emissions, which can be done by improved control algorithms. This work presents different methods to obtain models for the prediction of carbon monoxide emissions in a small-scale biomass combustion furnace for wooden pellets. The presented models are intended for an application in model based control, either as part of the underlying model or for carbon monoxide soft sensing and fault detection. The main focus is on simple structures which can be handled by the already existing hardware of the furnaces. Different black-box models and a kinetic process model are introduced and compared. The black-box models are based on the measured flue gas oxygen concentration and the combustion temperature, since these measurements are typically available even for smaller plants. The obtained models are validated with measured data in order to find the most suitable structures, of which combined fuzzy black-box models show the most promising results. The presented methodology can be readily applied to the investigated furnace. However, the model parameters have to be adapted for other plants.