Experimental Test Plan and Data Analysis Based on the Design of Experiment Methodology

Abstract

The development of fault diagnostics for fuel cell (FC) stacks and systems can increase their lifetime and therefore their overall efficiency. The development of such techniques often requires a reliable computational FC stack or system model. Given the complexity of such systems, endless time and effort has to be used to test the systems and identify the related models unless an adequate trade-off between available time and desired model accuracy is found. The Design of Experiment (DoE) methodology enables determining the best compromise between the number of experiments to conduct and the obtained information. In this work, the DoE approach is used to define a test plan that is carried out on a solid oxide fuel cell stack. The methodology is then further exploited to analyse the measured data and to determine the influence of the selected parameters (i.e. input variables) on the response extracted from the monitored variables. Specifically, a Full Factorial DoE analysis is performed and the relative results are compared. The results show which parameters have a direct effect and which affect indirectly, in coupling with other parameters, on the system response.