Biography
Sangseok Yu is a Professor of Mechanical Engineering at CNU who is an expert in modeling and simulation of energy system. He was majoring transient heat and mass transfer and dynamic modeling of automotive fuel cell system at University of Michigan Ann Arbor. In particular, he has special interests in control and fault detection of automotive fuel cell system. Recently, he extended his research scope to modeling and simulation of various energy systems.
Abstract
As an alternative candidate of internal combustion engine, fuel cell vehicle has been actively studied. Compared with internal combustion engine, fuel cell vehicle is more sensitive to the operating temperature that is coupled with water management. Additionally, the durability of fuel cell is also very sensitive to operating temperature. Different from internal combustion engine, the operating temperature window of fuel cell is very narrow, it is necessary to control active control of cooling system. In this study, the control algorithm of fuel cell cooling system is developed that facilitate thermal management strategy. With non-linear comprehensive simulation model of fuel cell system, the analytic redundancy model is developed. The analytical redundancy algorithm is used to determine fault of thermal system by diagnosis of residuals. The sensor errors such as stuck offset and scale of measured value was investigated in terms of change of residual. Result shows that analytic redundancy fault detection algorithm provides decision making information of cooling module. The control algorithm is then readjusted to determine proper operating conditions so that the cooling system is effectively maintained the fuel cell temperature.
Biography
Abstract
This work was done within the MAESTRA strategic project of the University of Padua and aims at developing state-of-the-art technologies, for Vanadium Redox Flow Batteries (VRFBs) which are needed for the production of more efficient and flexible devices, by optimizing cell and stack geometries, power management units and supervisor systems. For developing the experimental investigations, a fully-monitored test facility has been built, rated 4 kW-24 kWh, provided with a 40-cell stack and two 500 L tanks. It is provided with two flow pumps powered by analog-controlled brushless motors, a bidirectional power management system, a Labview-based battery management system (BMS), multi-voltage and current meters, pressure and voltage measurements. The test program include stack voltage vs. SOC (state of charge) in charging and discharging, electric power flow, temperature analysis, distribution of stack voltage, polarization curves, electrochemical impedance spectroscopy (EIS), efficiency measurements, and aging effects. The experimental campaign is supported by extensive numerical analyses. Recently, two battery topologies have been studied and compared. In the conventional series stack, cells are connected electrically in series and hydraulically in parallel by means of bipolar plates. The alternative topology consists of cells connected in parallel inside stacks by means of monopolar plates, in order to reduce shunt currents along channels and manifolds. Channeled and flat current collectors interposed between cells have been considered in both topologies. In order to compute the stack losses, an equivalent circuit model of a VRFB cell was built from a 2D FEM multiphysics numerical model based on Comsol®, accounting for coupled electrical, electrochemical, and charge and mass transport phenomena. Shunt currents were computed inside the cells with 3D FEM models and in the piping and manifolds by means of equivalent circuits solved with Matlab®. Hydraulic losses were computed with analytical models in piping and manifolds and with 3D numerical analyses based on ANSYS Fluent® in the cell porous electrodes. The alternative topology with channeled current collectors exhibits pumping and shunt current losses one order of magnitude lower than a conventional battery rated at the same power, resulting in a round-trip efficiency 10% higher, as compared to the conventional topology.