Escolar Documentos
Profissional Documentos
Cultura Documentos
lifetime analysis
Your Name
Name of Your Department
Name of Your Institution
Date
Modelling of lithium-ion batteries for lifetime analysis October 2016
Table of Contents
Proposed project
There are many models to describe the lithium-ion battery lifetime the main idea of this
project is to analyze simulate, and compare advantages-disadvantages of each one of
the most important models, using tools such as Simulink, LT Spice, Python’s script and
Matlab. These models can be classified in:
Electrochemical models.
Electrical-circuit models.
Analytical models.
Stochastic models.
Shepered model.
Thevenin-based model.
Rakhmatov and Vrudhula model.
Stochastic modified Kinetic Battery Model (Stochastic modified KiBaM).
All of them focused in predict the lifetime battery. Will be studied with emphasis why
specific model is better to predict certain things that other model, the computational cost
involved, the range of certainty of his prediction and associated errors.
Objectives
General objective
Specified objectives
Literature review
Rui Xiong and Jinxin Fan [1] conducted a quantitative study about equivalent circuits
models of Lithium-ion battery. In this research, they describe a battery as a device that
converts chemical energy into electricity through a redox reaction. In essence, the
following excerpt represents the literature review of the article published in the energy
MDPI (Multidisciplinary Digital Publishing Institute) Journal:
The basic unit is the electrochemical cell, consisting of three main elements: the
anode, or negative electrode; the cathode, or positive electrode, and electrolyte, which
is the ionic conductor that provides the physical medium for charge transfer.
From the theoretical voltage and the theoretical capacity of the cell you can calculate
the theoretical maximum power that can provide the cell (Wh); the theoretical energy
of a battery (or cell) is usually expressed in relation to the weight (specific energy) or
volume (energy density):
In practice the energy provided by a battery (or cell) is a fraction of the theoretical
energy because is necessary to include the electrolyte volume and the no reagents
necessary for their physical construction. Is important too, understand that the battery
is not discharged to the theoretical voltage, that meaning the battery not reaches zero
volts. The cells are not ideally balanced from the standpoint of the theoretical
stoichiometric reaction; this reduces the specific battery power. Energy available from
a battery depends directly on the component materials and quantity, the more energy
required amount of materials are required.
This energy source requires sophisticated monitoring and control strategies for loading
and unloading cycles, temperature control, reading charge status and aging correction.
FACULTY NAME • UNIVERSITY NAME 4
Modelling of lithium-ion batteries for lifetime analysis October 2016
The design of such control methods demand unfailingly work with dynamic models
capable of reproducing the electro-thermal behavior of the battery, which are themselves
liable to be part of the control algorithm once. This in turn implies the need for a flexible
tool for modeling, simulation, and display algorithms microprocessor.
Lithium-ion batteries are a type of rechargeable batteries. Their good properties such
as high energy density and long life have made them play an important role in portable
electronic equipments. Currently, with the popularization of renewable generation
systems, the stationary storage is becoming a necessary tool and lithium-ion batteries
are positioning themselves as an optimal technology for these applications. On the other
hand, these batteries hold a privileged position in the market of electric vehicles, where a
vast majority of models use this technology.
While growing interest in lithium-ion batteries, does too the need to evaluate their
behavior under different working conditions within the wide range of possibilities. Without
an effective model, these tests would involve a great cost in time and money, so is
necessary have accurate models.
They have proposed different ways to model a battery according to the authors and
the type of model to be performed, since each has its advantages and disadvantages.
Some ignore transients and others cannot predict the lifetime. Some even allow to study
the internal dynamics of the battery, although they are sometimes too complex to be
implemented in a general electric model.
There have been endless studies and developed many models for characterizing the
charge and discharge cycles of batteries, as well as its lifetime. M.R. Jongerden and B.,
R. Haverkort,] Shamsi, Onori, and Rao [2][3][4][5] collected and grouped all these models
in different categories, then a brief description of these models and their most important
features is presented, all of them extracted from all works.
Electrochemical models
Directly based on electrochemical phenomena, these models predict the energy store
and provide batteries, but ignore phenomena as the change in output voltage of the
battery load, the influence of temperature and aging drifts.
𝑘
𝑉 = 𝑉𝑜 + ∙ 𝐼 + 𝑅 ∙ 𝐼 − 𝑠𝑔𝑛(𝐼) ∙ 𝐴 ∙ 𝑒 −𝐵⋅𝑆𝑜𝐶
1 − 𝑆𝑜𝐶
Usually, in this model, the exponential part is ignored because its contribution is only
to explain the initial voltage drop in the battery.
Electrical-circuit models.
There are a variety of models of equivalent circuits. Most use a capacitor to represent
the storage capacity of battery. The linear model, shown in Fig. 1, is the simplest model
and is used for simple simulations. It is an ideal source with an open circuit voltage (E battery
+ Vsense) and an equivalent series resistor Rcell, which represents the internal resistance
of the battery.
Thevenin-based model
The Thevenin equivalent model (Fig. 2) consists of a source, two resistors and a
capacitor. The EOC source has a value of the open circuit voltage, the resistor R models
the internal resistance of the battery, and the capacitor CO modeling capacity electrodes
of the cells. The resistor RO represents the nonlinear resistance between electrodes and
electrolyte. This network modeling transient response common to the batteries. In the
most basic form of this model it is assumed that the elements are constant, so it is a
model very used given its simplicity.
Various battery designs have been developed which are formulated analytical
expressions to calculate the actual capacity and the battery life time, using discharge
current values of characteristic operating environment and physical properties of the
battery as parameters. The analytical models describe the battery in an abstract way, as
well as stochastic models. The main battery properties are modeled using a smaller set
of equations. This makes this type of model easier to implement than the electrochemical
and electric models.
Many of these models can include models of constant load and the variable load, and
can capture the effect of rate of capacity and the recovery effect. They are computationally
efficient and flexible, requiring assessment simple analytical expressions, and can be
easily configured for different types of batteries.
The simplest analytical model to predict the battery life time is the Linear model where
the battery is treated as a linear current container, and this way does not consider non-
linear effects that occur in the battery. Another analytical model considered simple, that
considers part of its nonlinear properties is the Peukert Act. This act captures the non-
linear relationship between battery life and discharge rate, but does not consider the
recovery effect.
Peukert’s law
Peukert's law allows us to know how long a battery will take to download, depending
on different discharge speeds specified by the manufacturer. It is expressed
mathematically as follows:
𝐶 𝑘
𝑡 = 𝐻( )
𝐼∙𝐻
Where:
t is the time in hours that it will last the battery charge giving the current I
(discharge speed).
H is the discharge time (in hours) given to the capacity specified by the
manufacturer.
C is the nominal battery capacity given by the manufacturer.
I is the current (download speed) to which we resolved that the battery gives us
our load.
K is the exponent of Peuker. Each battery has its exponent of Peuker.
Sometimes the manufacturer will give us that we value and sometimes we get
elsewhere.
Stochastic models.
This approach is a paradigm shift in accordance to the previously analyzed, and with
him deterministic systems are abandoned, now in this models the systems are
characterized by uncertainty, where the modeling feature become a stochastic process.
This approach bases its development on Markov chains in discrete time, which is a
special type of stochastic process in which the probability of an event occurring depends
on the immediately preceding event, i.e. this type of stochastic process "remembers" the
last event and this conditions the possibilities of future events.
To test these models without the need for testing on real batteries and spending
resources and effort that entails, are designed simulators for each of the characteristics
and under different types of tools, the most popular are: MATLAB and Simulink, Python
using external bookshops (as SciPy, Pandas, scikit-learn) and developments in other
programming languages (such as C, C++, FORTRAN). Lan-Rong Dung, Ernesto
Martínez-Rosas and Cheng [6][7][8] describe different methods to do Lithium-ion batteries
simulations, based on the theory that are detailed before.
Although there are many methods to model and simulate battery behavior, almost no
study provides a comparison between these models, in order to determine: what model
is better? Under what circumstances it is feasible to use a model? The computational
cost of simulation and the precision they offer. According to all this, it is that this project
will assess a number of specific models to compare their performance and provide clear
conclusions on each of the existing methods of analysis lifetime of the batteries.
Glossary
Lithium-ion battery
It is a device consisting of two electrodes of metal or composite material (based on
carbon, for example), which are immersed in a conductive liquid (electrolyte).
Markov chain
It is a series of similar tests or observations which each test has the same finite number
of possible outcomes and where the probability of each outcome for a given assay
depends only on the result of the immediately preceding test and not any result previous.
Random variable
is any function associated each element of the sample space E with a real number
used capital letters X, Y, ... to denote random variables and respective lowercase (x, y,
...) to designate specific values.
Stochastic process
Is a collection or family of random variables {Xt, t ∈ T} with, sorted by the subscript t in
general is usually identified with time.
Thevenin circuit
Thevenin's theorem is used to convert a complex circuit, having two terminals, one
containing only simple voltage source or voltage (Vth) in series with a resistor (Rth).
Project plan
Conclusions and
0%
recommendations.
Writing final report.
10%
Milestones: M1 M2 M3 M4
Table 1: Gantt chart.
Estimated
Phase Action Start date
duration
1 Project proposal. 20/10/2016 1 week
2 Literature review. 15/10/2016 3 weeks
3 Algorithm design for models implementation. 25/10/2016 4 weeks
4 Simulink, Python and Matlab code design. 15/11/2016 3 weeks
5 Computational cost, efficiency and error rate tests design. 01/12/2016 3 weeks
6 Results evaluation. 01/12/2016 3 weeks
7 Conclusions and recommendations. 10/12/2016 3 weeks
8 Writing final report. 29/10/2016 9 weeks
Table 2: Plan table.
Milestones
References
[1] He, Hongwen, Rui Xiong, and Jinxin Fan. "Evaluation of lithium-ion battery
equivalent circuit models for state of charge estimation by an experimental
approach." Energies 4.4 (2011): 582-598.
[3] Shamsi, Mohammad Haris. "Analysis of an electric Equivalent Circuit Model of a Li-
Ion battery to develop algorithms for battery states estimation." (2016).
[4] Onori, Simona, et al. "A new life estimation method for lithium-ion batteries in plug-in
hybrid electric vehicles applications." International Journal of Power Electronics 4.3
(2012): 302-319.
[5] Rao, Venkat, et al. "Battery model for embedded systems." 18th International
Conference on VLSI Design held jointly with 4th International Conference on Embedded
Systems Design. IEEE, 2005.
[6] Handy, M., and D. Timmermann. "Simulation of mobile wireless networks with
accurate modelling of non-linear battery effects." Proc. Int’l. Conf. Applied Simulation and
Modeling. 2003.
[8] Chen, Min, and Gabriel A. Rincon-Mora. "Accurate electrical battery model capable of
predicting runtime and IV performance." IEEE transactions on energy conversion 21.2
(2006): 504-511.