Optimizing model complexity with application to fuel cell based power systems

Chemical process simulators employ two levels of models: (1) a forest level description of models and (2) a more detailed tree level description. Reducing model order is beneficial for reducing computational complexity. However, this increases uncertainties in model prediction. This paper presents a methodology based on multi-objective optimization to find optimal model complexity in the face of model uncertainties. A case study of fuel cell power plant is presented where different level models for SOFC and PEMFC are evaluated.