Optimal boiler settings are required to ensure high efficiencies during biomass combustion and to enable compliance with emission thresholds despite lower and fluctuating fuel qualities. Thus, an affordable and simple near-infrared retrofit solution for online fuel characterization was developed and tested to allow for automatic adjustments of the boiler control with respect to actual characteristics of the used fuel assortments. To this end, a low-cost online near-infrared spectroscopy device (SCiO™, Consumer Physics), with investment costs less than 1000 € was employed. Partial least squares regression models based on 13 different spectral preprocessing methods for the prediction of moisture, ash, nitrogen and potassium content, as well as the net calorific value were developed and validated in the lab. The resulting model predictions of biomass properties agree in terms of root mean square error and R2 with the reference values obtained from laboratory analysis. Furthermore, a near-infrared retrofit solution was successfully coupled to the fuel feeding system of a commercially available small-scale biomass boiler for automatic and continuous operation. The automatically predicted values obtained by the near-infrared retrofit solution during continuous boiler operation are stable while the predicted values for moisture and net calorific value are in the range of the reference values obtained from the laboratory analysis but with a small offset. Non-reliable results were obtained for ash and nitrogen content.
Read full abstract