Automotive Fuel Cell System simulation, component and compressor modelling

Future fuel cell (FC) systems for automotive applications will significantly depend on the development and realization of efficient and reliable components. The screw compressor for charging of the FC stack constitutes one important component in the system. In the European project NFCCPP (Numerical Fuel Cell Component Performance Prediction tool) within the EC 5 FP, a modular simulation environment has been developed which allows virtual testing of components of fuel cell systems. The objective of this development was to create a useful tool for automotive component suppliers to test their component or subsystem models for performance prediction in a realistic FC system environment, for relevant driving cycles. MATLAB/Simulink, being well established in industry, was chosen as software platform for the modelling. The system model structure, interface definitions and data flows, as well as different approaches to the modelling of screw compressors, will be presented in this paper. Component models of different degree of detail will be possible to plug into the system model. A standard reference model is based on simplified component models. In particular, there have been two main issues for the development of this system: • Lumped models or characteristics approach. Simplified component models will be required to allow for acceptable computing time for system simulation. This is even more important if the system should be used with hardware in the loop, i.e. for real time simulation. • Code protection. Preventing access to confidential information and data of other component suppliers was considered necessary. Even competitors can then test their components together in such an environment. Three different approaches have been taken to this: • Centralised simulation with remote user control. • Localised simulation with simulation-time model usage control. • Parallel distributed simulation. Compressor models. The complete screw compressor unit, also including electric motor and control system for capacity (speed) control has been modelled in NFCCPP. For a simple compressor model, empirical test data in the form of screw compressor maps for one specific machine type and size could be used. To generalize, screw compressor maps could be transformed by means of similarity laws for scaling. A more detailed screw compressor model should be based on the lumped approach according to the since long well established model structure; a number of compression chambers connected by leakage paths. Building a lumped screw compressor model in Simulink has been tested. Compressor flow field models of CFD element type are however not considered realistic as component models for the complex FC system simulation. In some FC systems, screw expanders are also used, for energy recovery. The FC system model can be configured to include also an expander. The commercialisation of this program system is being planned for. 1 Fuel Cells, background After the fuel cell’s first appearance in space industry to power satellites and spacecraft, fuel cells for stationary power generation and for automotive application were identified as a potential saviour from the ever increasing debate around reduction of green-house gases and thus protection of the earth’s atmosphere. Using fuel cells in the automotive industry for replacing the internal combustion engine as prime mover by a fuel cell system is however quite challenging as far as reliability, cost, safety and Europe-wide operability are concerned. Fuel cell systems for automotive applications are complex power systems consisting of different component groups as shown in Table 1. Table 1: Component Group Definition Group Components Air Management Compressor/expander, humidifier, filter, mass flow sensor, water separator Auxiliaries Pumps, piping, valves, pressure regulator Control Supervisor, anode, cathode, thermal, power electronics Driveline Model Motor, transmission, vehicle Fuel Processing Reformer (partial oxidation, steam, autothermal, ammonia cracker), reformate cleaning (shift reactor, preferential oxidation) Fuel Storage Liquid fuel tank, compressed H2, electrolyte storage Material Properties Fuels, mixtures, thermodynamic equations and data Power Management DC/DC converter/inverter, battery Stacks PEM, AFC Thermal Management HEX, radiator, start-up burner, off-gas burner, cooling fan These components interact closely together and influence the entire system behaviour. The general challenges in the development of components for fuel cell systems for different fuel cell applications are: • Understanding and defining development goals (i.e. component specification) • Improvement of system efficiencies • Reduction of production costs • Improvements of availability and reliability • Reduction of weight and required space • Optimisation of dynamic behaviour (incl. start/stop strategies and charging system) • Further component optimisation (e.g. noise reduction) 2 NFCCPP Simulation program for automotive Fuel Cell systems In order to develop and optimise single components it is necessary to understand the interaction between the system and the component under development. Thus a "Numerical Fuel Cell Component Performance Prediction Model NFCCPP" [1] has been developed as a project within the EC 5 FP, to fulfil the particular needs of component manufacturers for a fast and effective simulation tool to evaluate component performance within the fuel cell system. This model is going to be used as a "Reference Model" for fuel cell systems. This reference model contains all known components and sub-systems to date which have been modelled with non-confidential information and modelling knowhow, featuring • High degree of modularisation for fast and easy access to the components • Standardised I/O signals for the components to enable the use of each simulation block at different positions (if physically possible) • A graphical user interface to allow a wide range of experts to interact, even if one is not a specialist in the specific software environment • Easy changes to the system parameters (e.g. model and control parameters) 2.1 Using a visual programming environment MATLAB/Simulink, being well established in industry, was chosen as a feasible software platform for the modelling. For solving systems of differential equations, using the Simulink visual programming environment is easy compared to conventional programming languages, since the program automatically sorts out the integration order and dependencies. It is easy to get started and to get results quickly. The drawback of using Simulink is that it is not as flexible as an object oriented programming language. For complex models, many function blocks and lines have to be connected in the graphical user interface, and many complex equations have to be entered. The overview can then somehow be lost when too many blocks are interconnected. 2.2 Standardisation of simulation modules and interfaces To enable ease of component connection and interchange, a standard connection methodology was devised. The basic premise of this was to use generic connections in the form of mechanical, electrical and fluid connections (Table 2). Since Simulink is a casual simulation system the inputs and outputs to each component have to be strictly defined, as opposed to other simulation packages which calculate their own causes and effects from given equations (e.g. Flowmaster, Hopsan) [2]. A component-based approach was used, i.e. the components were modelled the way component manufacturers were used to model them. The mechanical and electrical connections both work the same way: If the Speed/Voltage is selected as an input, Torque/Current is used as an output and vice versa. In all components the most convenient input and output were selected to guarantee a proper cause and effect action/reaction for the system [3]. For the fluid connection, the Pressure and Mass flow have about the same functions as Torque/Speed. In addition, the Temperature and Mass fractions are included in this connection to calculate the fluid properties, including density, heat capacity, and more, inside each component. The Temperature and Mass flow information are both assumed to always be in the same direction as the main flow direction. The requested mass flow always should be used as an input together with the downstream pressure. Using this method for all the fluid connections, all components will have the same input and output, and can then be connected in any order. In addition to the Mechanical, Electrical and Fluid connections, there are also control signals, e.g. “requested torque” or “measured mass flow” (electric control signals, e.g. 010V, in a real system). They can be used only as input or only as output from the system. Table 2: Generic Component Connections Mechanical Electrical Fluid 1 Fluid 2 Torque (Nm)/ Current (A)/ Temperature (K) Pressure (Pa) Speed (rad/s) Voltage (V) Total mass flow (kg/s) Mass fraction , element 1 Mass fraction, element 2, .... Mass fraction, element n Through use of the connection scheme of Table 2, a component may be defined as shown in Figure 1. Component Electrical connection in Electrical connection out Mechanical connection in Fluid connection out Fluid connection in Mechanical connection out Control signal out Pressure to upstream Downstream fluid Control signal in Fig. 1: Component connections Such a modularised system model enables the user to remove his component module within the existing system model (Reference model), and to put a very detailed, selfdeveloped module into the system model and generate information about the component behaviour in the complete system. There will be a growing acceptance of the results if component manufacturers and their customers use the same reference model for their performance predictions. It is assumed that especially smaller and medium sized companies will profit from such a standard tool since they presumably have little interest in developing a complete fuel cell system model with their own internal resources, but would profit from the ability to assess the performance of their components in the system.