BIO-ETHANOL PROCESSOR SYSTEM FOR HYDROGEN PRODUCTION. MODELING, SIMULATION AND CONTROL.

LUCAS NIETO DEGLIUOMINI; DAVID ZUMOFFEN; MARTA BASUALDO

Hydrogen Production

Nova Science Publishers, Inc.

Año: 2011;

In this chapter topics such as modeling, simulation and control of a bio-ethanol processor system (BPS) for hydrogen production are addressed. Initially, a complete nonlinear dynamic modeling of the BPS is given to be suitable integrated with a proton exchange membrane fuel cell (PEM-FC). The dynamic model is developed around the proposed operating point determined during the synthesis stage. It can be implemented thanks to the use of a specific communication protocol between HYSYS and MATLAB. The heat integration problem is preliminary solved by using the LNG tool, available at HYSYS. This approach applies the pinch analysis to approximate the best heat exchanges under specific conditions at a pseudo steady-state. It is useful for keeping the stream temperatures at their optimum values, neglecting their complex dynamic effects. On the other hand, the reactors included in the BPS were modeled dynamically in the MATLABr environment. The model of the complete process is useful to apply a new systematic approach dedicated to plant-wide control. It accounts the main objectives of the BPS based on the profitable and safety operation conditions to be connected with a particular PEM type. These main objectives basically are to maintain H2 levels on the anode of the FC, retain low CO levels at the inlet stream of the anode, and keep the temperatures of the reactors set and FC under rigorous specifications. The proper operation of the process depends on a suitable plantwide control design. In this context and considering the size of the plant is necessary to have a generalized and systematic tool to assess the benefit of different control structures. In this chapter a new methodology, called minimum square deviation (MSD), for solving simultaneously the optimal sensor location (OSL) and the control structure selection (CSS) problems via optimization is detailed, minimizing the heuristics requirements. Basically, the MSD approach tries to minimize the use of heuristic concepts and is able to work if only steady-state information is available. The OSL problem is solved by considering the sum of square deviations (SSD or the square Frobenius matrix norm) of the uncontrolled variables from their nominal values when a full internal model control (IMC) is used. The CSS problem is solved by accounting a new interaction index called net load effect (NLE) so as to deal properly with a trade-off between set point changes and disturbances effects. Based on these, different control structures (diagonal, full or sparse) would be able to assess the benefits on the selected controlled variables. A suitable parametrization of the above problems allows minimizing both the uncontrolled variables deviation and the NLE in a SSD sense via a mixed-integer optimization routine to solve this combinatorial problem. The latter, in this case, is solved via stochastic global search implemented with genetic algorithms (GA). As a result, the conceptual engineering of this novel process is presented and tested through a complete set of numerical simulations including both open and closed loop behavior of the bio-ethanol processor system for an efficient H2 production.