Battery-powered electric vehicles (EVs) and hybrid electric vehicles (HEVs) have been gaining market share as the public’s environmental concerns drive the demand for better fuel efficiency. Apart from the need to improve the mileage one can drive before recharging, battery reliability is another major concern. Effective cooling in intense charge or discharge conditions is crucial to minimising the risk of operating these batteries in vehicles. An increase in battery efficiency is one of the biggest challenges that engineers face. While Li-ion batteries are currently the best option available, its high mass reduces fuel efficiency in automobiles, and their potential flammability makes them dangerous, particularly in the event of a crash.
Exploring new options
Researchers worldwide are trying to develop a better battery to replace Li-ion technology. For now, optimising the configuration of Li-ion batteries is the best method of increasing battery performance, but batteries are complex systems that are difficult to model and test using physical prototypes. Numerical simulation is an indispensable tool for battery designers and researchers. Since the battery is a multi-physics application, its simulation involves many different disciplines, including, but not limited to, electro-chemical modelling, electrical circuit modelling, electromagnetic modelling and thermal modelling. Here we present one method of using engineering simulation solutions to optimise battery performance and function by comparing different numerical approaches.
Five-step simulation approach
Develop an equivalent circuit model (ECM) with Ansys Twin Builder.
Perform computational fluid dynamics simulations to determine the thermal characteristics of the battery. The performance of the battery will be impacted by the temperature profile.
Create a thermal reduced order model (ROM) derived from CFD simulation training data. In case of a time-constant mass flow rate, a linear and time invariant (LTI)-ROM can be derived to achieve the reduced model. Next, apply a linear parameter varying (LPV)-ROM to determine how the cooling flow rate depends on the drive cycle.
Combine the ECM and LPV-ROM in Ansys Twin Builder and perform drive cycle simulation on this combined thermo electrical battery model.
Integrate the thermo electrical battery model in a car system model or race simulator.
Creating an ECM model from Hybrid Pulse Power Characterisation (HPPC)
The performance of a Li-ion battery can be represented using an equivalent circuit model (ECM) as it’s easy to implement and consists of the open circuit voltage (Voc), internal series resistance (RS) and one or more parallel resistance or capacitance units. It can represent the state of charge (SOC) and dynamic behaviour of the battery, and also calculates the internal heat loss in the cell due to electrochemistry. The ECM parameters can be estimated based on a curve fitting of the hybrid pulse power characterisation (HPPC). The HPPC data contain the voltage measurement under current pulse for different SOC levels, temperature and the discharge rate. An accurate fitting of the model to the HPPC data is mandatory to estimate ECM parameters.
Thermal characterisation of the battery
To determine the thermal characteristics of a battery with high accuracy, transient 3DCFD calculations are needed. If complex drive cycles are performed directly in CFD, a large number of small time-steps are needed, which can make the simulations too slow to account for repeated transient analysis of the battery module. However, to characterise the thermal behaviour, it is sufficient to record the temperature responses to step inputs of the heat loads, which require fewer time steps. Therefore, this step involves the creation of training data for the ROM. The heat loads (W) are the input parameters and the temperatures (K) of the cells and other important parts are the output parameters.
Derive thermal ROM from training data in the system simulation tool
The ROM accounts for variation of the cooling flow due to car velocity by implementing an LPV-ROM. The so-called ‘local LPV-ROM’ developed here requires the same training data as for LTI-ROM generation, but it needs minimum, average and maximum values of the cooling mass flow rate. Thus, the amount of training data is three times bigger than for an LTIROM. The reason is that the cooling fluid mass flow rate is different from the heat load input parameters because it affects the thermal behaviour in a nonlinear way. Twin Builder calculates the compact LPV-ROM from the training data. The ROM results match well with those from a full CFD simulation. While a CFD model may take hours to run depending on the size of the problem, the corresponding ROM runs in seconds.
Coupling of the EMC model to the LPV-ROM
In this step, the final model involves two-way coupling of the ECM model to the LPV-ROM because the temperature of the battery depends on the internal heat losses of the cells and the electric performance of the cells depends on the temperature. Twin Builder tabulates values of the two variables that are driving this coupled system: cooling mass flow rate and electric current from a drive cycle. The ECM calculates the power dissipation and sends it to the thermal ROM, which calculates the temperatures based on power dissipation and sends them back to the battery ECM, which is sensitised to the temperatures. For the sake of a simplified display of the model, heat flows from ECM are transferred as inputs to the LPVROM by using the yellow boxes. The temperature values are directly connected from LPV-ROM to the ECM module via the green lines.
Linking ROM and ECM in a car model A model comprising the ROM and ECM for the battery can be embedded in a larger model of the car — for example, the complete drive train or the cooling system. This can be done either in Twin Builder or in other system simulation tools. In the latter case, the model from Twin Builder is exported as a functional mock-up unit (FMU). For racing applications this FMU could be a race track simulator. This is one method of using engineering simulation solutions to optimise battery performance and function.