Control and controllability analysis of an ethanol reformer for hydrogen generation in fuel cell applications

V. GARCÍA; E. LÓPEZ; M. SERRA; J. LLORCA; M. PERRIER

Londres, UK

Simposio; Eleventh Grove Fuel Cell Symposium; 2009

One important problem related to the fuel cells as a green technologyrepresents suitable sources of hydrogen as a fuel. Ethanol is a promisingsource of hydrogen as it is a renewable source when obtained from biomass.The target solution consists in the utilization of ethanol steam reformer forhydrogen generation.In a previous contribution, García [Garcia et al, 2009] reported results addressing thedynamic modeling of a three-module reactor for fuel cell hydrogen feeding.Three specific catalysts were selected for each of the three modules of thereforming unit. A dynamic mathematical model of the three-stage reformer waspresented as a tool for design of control-oriented devices.A one-dimensional, pseudo-homogeneous, non-steady-state model has beenused to represent the dynamic behavior of the ethanol reforming in the alreadyreferenced series of monolithic reactors. 1-D, pseudo-homogeneous models areusually selected for control-oriented applications to reduce the solving time ofthe equations system.In order to improve the performance of a reformer, generate further hydrogenand eliminate CO which is a strong poison for the anode catalyst of the fuel cell,it must be controlled efficiently. We present a sensitivity and controllabilityanalysis of the non lineal model. According to the results, the most promisingcontrol structures are selected and the performance controllers evaluatedthrough simulations with the complete non linear model.A static sensitivity analysis is focused on the complete non linear model arounda nominal operating point. This operating point has been deduced from the apriori best operating conditions of the catalyst, low production of CO and thehydrogen flow needed to feed a 1kW PEM fuel cell. The static input-output nonlinearity characteristic is verified for different input levels.A controllability analysis is also performed on the system. To do this analysislinearized versions of the model are considered at different operating points,MIMO linear systems can be analyzed using different analysis tools. Thesetools are mathematic operators applied to the squared transfer functions of thelinear system that give relevant information such as stability, controllability,sensitivity, robustness, etc. They are applied to the process (without control)and characterize the controllability of the system as a property of the processitself.The performance of diagonal control structures with PI controllers at differentoperating points is also studied. A method for the tuning of the controllers isproposed and applied. The behavior of the controlled model is simulated withthe non linear model.