Application of Multivariable Regression Model for SOFC Stack Temperature Estimation in System Environment

Abstract

The applicability of multivariable linear regression (MLR) models to estimate the maximum temperature inside a SOFC stack is investigated experimentally. The experiments were carried out with a complete 10 kW SOFC system. The behavior of the maximum temperature measured inside a SOFC stack with respect to four independent input variables (stack current, air flow, air inlet temperature and fuel flow) is examined following the design of experiments methodology, and MLR models are created based on the retrieved data. The practical feasibility of the MLR estimate is investigated experimentally with the 10 kW system by evaluating the accuracy of the estimate in two test cases: (i) a system load change where the stack temperature is regulated by a closedloop controller using the MLR estimate and (ii) during operator-imposed disturbances in the fuel system (a variation in the methane conversion in the fuel pre-reformer). Finally, the performance of the MLR estimate is evaluated with another, 64-cell stack operated at higher current density.