Application of inverse methodology to estimate unknown parameters of the mathematical model of biomass solar pyrolysis

Abstract The goal of the work is to determine unknown parameters of reactor heating in the process of biomass solar pyrolysis. The lab-scale reactor is heated by a xenon lamp, but the fraction of heat absorbed by the reactor is unknown, as well as the heat transfer coefficient from the reactor walls to the nitrogen flowing through internal reactor channels, and the overall heat transfer coefficient through the external reactor walls to the surroundings. The missing parameters are impossible to measure, therefore they need to be determined by solving the inverse problem. As the problem is strongly ill-conditioned, since the quantities are directly dependent on each other, two different inverse algorithms were used to retrieve them, namely, the Levenberg-Marquardt method and the Metropolis-Hastings method. The effectiveness of both methods was assessed and then they were applied to the real data. The inverse problem has been implemented in the MatLab software, while validation of the mathematical model and optimisation procedure was carried out with a CFD model built in Ansys Fluent 19.2. Calculations have shown that 14.4% of the lamp power penetrates inside the reactor and heat transfer coefficients to the flowing nitrogen equals 8.74 W/m2K and 0.965 W/m2K to the surroundings.