Battery Modeling

To model the electric vehicle, understanding of energy storage devices, specially battery is very important. Complex chemistry included in the battery charging/discharging mechanism makes the precise modelling of battery difficult. But for simulation of overall electric vehicle, this model has to sufficiently accurate and fast. Further to design controllers this information is extremely important. Various mathematical models provides their own advantages and disadvantages. But still battery modelling lies in its nascent stage, and the industry relies on experimental results, because of its intrinsic complexity. These models not only provide insight during the initial stages of design but also after the design and integration, for real time simulations and controller design for better performances.

In this project, Lithium ion battery was modeled in 3 different ways to understand the accuracy and computational effectiveness of different models.

• 1. Equivalent electrical model : Model consists of RC circuit and the parameters are tuned from the experimental data, capable to explaining various phenomena. Computationally fast, and appropriately accurate.

  2. Electro-chemical model :  Mathematical battery model, based on electro-chemical equations(Butler -Volmer equations) , including the transport phenomena, SEI layer growth phenomena and even more(depending on how much complexity is needed) and calculating the concentration of ions and current density, to find the voltage and state of charge, during charging or discharging, is very accurate but computationally costliest. It requires quite a number of parameters, for modeling. However there are several databases available for these parameters. 

  3. Stochastic model : Based on hidden Markov model, this model was developed to predict the battery behavior, state space equations were used to predict the behavior after each state. It is accurate and faster, but doesn't provide insight about the battery functioning. 

The experimental data from various researchers and manufacturers was used to check the modelling approach. An attempt to include aging process in the model is made carefully keeping the computational cost low. It can further be used to prepare better battery management strategy for prolong life. Several freely available software were evaluated, like Dualfoil, FastSim, ADVISOR, CoolSim, Battery Blast, etc.