Jump to Main Content
A new PI tuning method for an industrial process: A case study from a micro-cogeneration system
- Sağlam, Gaye, Tutum, Cem Celal, Kurtulan, Salman
- Energy conversion and management 2013 v.67 pp. 226-239
- algorithms, case studies, computer software, cooling, electrolytes, fuel cells, hydrogen production, models, polymers, renewable energy sources, systems engineering, temperature
- Micro-cogeneration systems are efficient and clean energy sources. For this reason, reliable control of the overall system itself and its sub-components is of great importance. This paper presents a proportional–integral (PI) controller tuning method for cooling of the hydrogen production unit within the polymer electrolyte membrane (PEM) fuel cell based micro-cogeneration system having 5kW electrical and 30kW thermal capacity. The flow rate control of the coolant water was implemented by means of the modeled temperature control system. Modeling and control system design were implemented in both MATLAB/Simulink and industrial programmable logic controller (PLC) software. The novel analytical tuning rules proposed in this paper, which were developed with the help of the digital control theory and experiences in industrial control problems, are based totally on the model parameters. The output response attained by the proposed controller is a critically damped response (a non-oscillatory system response with damping ratio: ζ=1) and the control signal is a slowly varying safe control signal. Slowly varying control signal denotes that there is no sudden or fast change in the control effort which causes a wear and tear effect in the actuators and shortens the life cycle of the relevant hardware. Besides, an evolutionary multi-objective optimization (EMO) procedure, namely non-dominated sorting genetic algorithm (NSGA-II), was used in order to simultaneously minimize the integral time-weighted absolute error (ITAE) and the integral time-weighted absolute derivative (ITAD) objectives. This results in multiple trade-off solutions that enable user to observe the overall range of possible controller parameters and to choose any option between diverse solutions. Finally, a brief post-optimality study was manually performed to find out the common relations in the PI controller parameter sets of these multiple Pareto-optimal solutions. The proposed controller was applied to the hydrogen production process model. The simulation results indicate that the proposed tuning rules are as effective as NSGA-II and eliminates the need for an iterative optimization run to get the optimal PI controller parameters.