You are on page 1of 1

SIMULATION ANALYSIS OF DIFFERENT LITHIUM-ION BATTERY

MODELS FOR A BATTERY MANAGEMENT SYSTEM


The project aims to simulate some generic battery models used in the industry today for various purposes and check the accuracy of each model. It discusses the use of
these models in a battery management system. A parametric study was also performed on the models and the results were discussed in detail.

Introduction
Industry trend: Fossil fuels -> Green Cars
Better Batteries developed. Lithium-Ion trade-off between energy density &
cost. A Lithium-Ion pack is shown below.
High Cost -> Safety Importance increased.
BMS keeps battery safe. Monitors voltage, current, temperature etc. and keeps
them in check.
A battery model helps in estimating these parameters without affecting the
performance of the battery.

Battery Models
Battery models help in estimating the required parameters. The state of charge
and state of health can be estimated using battery models as well.
5 battery models were selected taking into account performance in journal
articles and availability of data.
The models that were selected area. Shepherds Model
b. Nernst Model
c. Unnewehrs Universal Model
d. Linear Model
e. RC Model (Circuit diagram on the Left)
The Linear Model was used to gather additional data required for simulation.
The models were then simulated.

Simulation of the models


Following process was followed for simulation,
Test data-Continuous Discharge Test of 3.6 Li-ion Battery.
All models were solved for their parameters using the test data.
The models were simulated and compared.(Shown below)
The curve we are trying to achieve by simulation is shown on the right, while an
example of the code used to optimise the parameters is written below.
H=horzcat(o,disc,1./SOC)
x=inv(transpose(H)*H)*transpose(H)*voltage

Shepherds Model

Unnewehrs Universal Model

Quantitative Analysis and Parametric Study of


the models
A quantitative analysis of the models was done and
comparable RMS values of errors were obtained for
each of the models, the results of which are shown in
tabular form below.
Unnewehrs
Universal
0.0911

yk E0 Rik / SOCk

yk E0 Ri k SOCk

Nernst Model

Shepherds

Nernst

RC

0.2603

0.00945

0.0001232

A parametric study was also conducted for each


model. As an example, variation of R for Unnewehrs
Model shown below.

RC Model

Conclusion
yk E0 Rik 1ln SOCk 2ln 1 SOCk

yk OCV SOCk Rc,d ik Vc e

Salman Daud 3rd Year Undergraduate Student, Engineering Department


Supervisor: Dr. Dongpu Cao

t
RC

Best performance RC Model.


Trade-off between complexity and accuracy
observed.
Parameters effect on each model as well as
results discussed.
LANCASTER UNIVERSITY
Faculty of Science & Technology

Engineering Department

You might also like