FAST PREDICTIVE MODEL FOR DESIGN AND OPTIMIZATION OF A LOW EMISSIONS BIOMASS FUELED CHP SYSTEM

摘要

Biomass can play a key role in the development of distributed micro-generation appliances, allowing an easy and safe storage of energy in view of satisfying energy needs for the next future. As result of a recent project, called Megaris [1], a new micro CHP system has been proposed. Essentially, the system consists of a fluidized bed combustor with the heat transfer head of a Stirling engine in direct contact with the bed, where very low emissions can be achieved with a large variety of biomass sources, together with very high heat transfer coefficients towards the engine, raising its efficiency to competitive levels. To develop and optimize a configuration ready for introduction into the market, a reliable and fast model is required. In this work, a hybrid dynamic model is presented, consisting of a set of ordinary differential equations and of a Neural Network submodel introduced for those aspects that cannot be represented by lumped parameter models. The resulting model must correctly reproduce effects due to the fluidization regime that effectively establishes as working conditions are changed, both on heat transfer coefficients and on combustion efficiency. Results of the model are compared with experimental data collected on a prototype of the combustor equipped with a Stirling engine. The model is able to correctly reproduce the dynamic behavior of the system under different working conditions, with the accuracy required to perform design and optimization analysis.