| dc.description.abstract | Fuel cell systems are part of a prominent key enabling technology for achieving carbon free electricity generation and can be used for stationary, mobile and portable applications. The last decade, significant research efforts have been allocated to the development of fuel cell components and integrated systems, since they constitute and efficient energy conversion technology for transforming hydrogen, and possibly other fuels, into electricity. During their operation various phenomena are evolving and their behavior is affected by many variables such as temperature, partial pressures, gas utilization and humidity. Therefore, it is necessary to be able to understand qualitatively and predict quantitatively the behavior of an integrated fue cell system in order to protect its longevity and preserve its long-term performance. Driven by this motivation their optimum operation is of great importance. Thus, it is imperative to develop appropriate control strategies and algorithms that optimize their responce so that they can accomplish certain intended functions and utilize the available resources, e.g. consumption of fuel, in an efficient manner and satisfy operating and physical constraints. The impact of control is evident not only in fuel cell systems, but also in a wide range of every day applications such as production of chemicals, automotive industry, generation and distribution of energy to name a few. Overall, control engineering provides the scientific foundation and technology for dynamically evolving systems by integrating concepts from computer science, mathematics, and systems engineering.
This thesis has a multidisciplinary scope and it is concerned with the optimal operation of an integrated Polymer Electrolyte Membrane fuel cell (PEMFC) unit and the design and development of advanced model-based control schemes which are deployed to the automation system of a small-scale experimental PEMFC unit. More specifically, a dynamic nonlinear mathematical model is developed
that describes the behavior of the PEMFC and it is experimentally validated using a formal systematic estimation procedure for the determination of the empirical parameters. Also, the automation infrastructure and the architecture of the Supervisory Control and Data Acquisition (SCADA) system is presented which is used a platform for the verification of a number of advanced controllers.
After the determination of the operational requirements of a PEMFC system a modular model predictive control (MPC) framework is designed and the PEMFC acts as a motivating system where the behavior of multivariable nonlinear MPC (NMPC) and multiparametric MPC (mpMPC) controllers are evaluated. In addition to the NMPC and mpMPC methods, a novel synergetic strategy is proposed that empowers the performance of NMPC by exploiting a multi-parametric quadratic programming (mpQP) approach. At the core of the NMPC formulation lies a nonlinear programming (NLP) problem which is solved using a simultaneous direct transcription optimization method. The performance of the NLP solver is enhanced by a warm-start initialization and a search space reduction technique. This synergy transcends the traditional problem formulation of the NMPC aiming at the reduction of the computational time for the dynamic optimization without sacrificing the quality of the obtained solution.
The interconnection between the a advanced model-based controllers and the automation system is facilitated through a custom developed software platform based on state-of-the-art industrial protocols. The establishment of such and infranstructure addresses the challenges related to the interface of control, computing and communication issues between the MPC and the integrated PEMFC unit. The MPC framework is deployed online to the industrial automation system and the performance of the controllers is assessed through a set of experimental studies, illustrating the operation of the PEMFC under varying operating conditions. | en_US |
