SAFETY CONDITIONS FOR HYDROGEN PRODUCTION FOR FUEL CELLS

L. NIETO; D. ZUMOFFEN; M. BASUALDO

Mar del Plata

Congreso; 4º Congreso Nacional - 3° Congreso Iberoamericano - HIDRÓGENO Y FUENTES SUSTENTABLES DE ENERGíA; 2011

UTN-CNEA

In this work the optimal monitoring systems design (OMSD) for a bio-ethanol fuel processor system (FPS) connected to a proton exchange membrane fuel cell (PEMFC) is analyzed. The OMSD approach is based on detectability analysis, multivariate statistics, and optimal signal selection.  The hydrogen production plant (FPS) is based on steam reforming, followed by high and low-temperature shift reactors and preferential oxidation. This process feeds a PEMFC for automotive purposes. The computational implementation of the FPS+PEMFC plant was made by using a suitable integration of three well-know environments: MATLAB®, HYSYS® and ADVISOR®.  The control structure applied to the FPS+FC system was developed by using the minimum square deviation (MSD) methodology for plant-wide control design. In this context, the OMSD approach was applied to the overall closed-loop plant by considering typical abnormal events, i.e. sensor faults. This methodology is supported by a supervision system based on principal component analysis (PCA) with combined statistic and its corresponding detectability index. Accounting the process data base (normal and abnormal) both the fault subspace (fault direction) and detection quality can be extracted and quantified respectively. Thus, the approach optimizes the minimal detectable fault magnitude by searching the most suitable sensors/signals locations. This mixed-integer (or binary/combinatorial) optimization problem can be efficiently solved by genetic algorithms and, eventually, the functional cost may be augmented by accounting the monetary investment cost. This leads to an optimal solution with the best detectability properties and lowest investments costs.