With fuel cells making their way into the maritime sector, cooling systems are becoming a more important aspect of ship design with regard to power saving. Because fuel cells generate heat at a lower temperature, they require a cooling system that is comparatively larger tha
...
With fuel cells making their way into the maritime sector, cooling systems are becoming a more important aspect of ship design with regard to power saving. Because fuel cells generate heat at a lower temperature, they require a cooling system that is comparatively larger than that of conventional power trains. Ship cooling system research mainly focuses on the implementation of variable-speed pumps, with little research focusing on the control of cooling systems. For this reason, this research aims to examine the performance of different rule-based and optimisation-based control systems applied to the cooling system of a fuel cell-powered ship.
In this research, a cooling system based on the FPS Waal is modelled in Simulink using Simscape. The cooling system consists of two coolant pumps, two river water pumps, a three-way valve, and two heat exchangers that cool six fuel cell units.
Four rule-based control methods and three optimisation-based control methods were tested. Two rule-based control methods, standard rule and temperature difference, have a rule-set based on just temperature measurements. The other two, load-based flow and load-based pump, have an additional basic algorithm where the control system knows the current cooling system configuration. Two optimisation-based control methods, steady-state optimised and lookup, are based on steady-state operation, and current-state optimised is based on the current system state. The optimisation models used are Mixed Integer Quadratic Constrained Programs (MIQCP). The control methods are tested in three types of simulations. A constant heat load is used to compare temperature control, pump on/off cycles, and power consumption; a step load is used to compare temperature control, and scenario simulations are used to compare temperature control and power consumption.
Optimisation-based control methods perform better, with regard to temperature control, pump on/off cycles and cooling system power consumption, in constant load despite suffering from nonlinearities. With step load, the rule-based control methods have better temperature control because they have spare cooling capacity. In the scenario simulations, the optimisation-based control methods have significantly lower cooling system power consumption than rule-based control methods. Load-based flow has an average power consumption over the scenario simulations, which is 39.2% higher than the average power consumption of steady-state optimised. load-based pump, temperature difference, and standard rule perform significantly worse with 96.4%, 179.5%, and 197.0%, respectively. Current-state optimised only has a 0.85% higher average power consumption and lookup a 5.85% higher average power consumption.
While the overall performance of the optimisation-based control methods was superior, simplifications in the optimisation models resulted in steady-state errors in temperature control. Further research is required to implement optimisation-based control methods for cooling systems to reduce the errors from simplifications and take into account cooling system component degradation.
