Battery Management System (BMS) For Lithium-Ion Batteries

Imagem
José Miguel Branco Marques
Battery Management System (BMS)

for Lithium-Ion Batteries
Master Dissertation
September 2014
University of Coimbra
Faculty of sciences and technology
Energy for sustainability initiative

Coimbra, 2014
Submitted in partial fulfillment of the requirements for the degree of

Master of Science in Energy for Sustainability
Faculty of Sciences and Technology, University of Coimbra
Submitted by:
Under supervision of:

Prof. Dr. Aníbal Traça de Almeida
Jury:
Prof. Dr. Manuel Carlos Gameiro da Silva
Prof. Dr. Pedro Manuel Soares Moura
Prof. Dr. Aníbal Traça de Almeida
September 2014
“People who are really serious about software should make their own hardware.”
-Alan Kay.
Acknowledgements
This dissertation would not be the same without the generosity and contribution of a large
number of people.
I would like to thank my supervisor, Professor Dr. Aníbal Traça de Almeida, for its trust
and for the possibility of developing this work on the challenging field of Battery Management
Systems for Electric Vehicles.
I am also very grateful to all my colleagues that assisted me in different ways. I appreciate
very much their suggestions and the time they spent to support me, helping to achieve this
important milestone. Special Thanks to Fábio Faria, Marco Silva, João Sousa, Luis Garrote
and André Gonçalves that worked close to me and helped to shape the final product.
I thank ISR and Professor Dr. Urbano Nunes for providing the excellent conditions and
resources that allowed me to accomplish this dissertation. This work has been supported by
project Grant "PROJECTO B: Diagnosis and Assisted Mobility for People with Special Needs"
(QREN-MAIS Centro: CENTRO-07-ST24-FEDER-002028).
I am very grateful to all my friends and to my family for their continuous support. Finally,
last but not least, I would like to express my greatest gratitude to my girlfriend, Adélia, from
who I was absent for longer times than I wished. Her support and understanding was essential,
especially in the most difficult moments.
i
Resumo
O desenvolvimento de Veículos Eléctricos (EV) é na actualidade alvo dos fabricantes de
automóveis e pode ser considerado como uma solução imediata para um sistema de transporte
rodoviário sustentável, pela sua contribuição na redução da emissão de gases geradores do efeito
de estufa.
Os EVs têm como componente principal o sistema de armazenamento de energia. Este
sistema envolve para além da bateria, todos os sistemas de gestão e monitorização da mesma,
vulgarmente denominados pela sigla BMS do termo inglês “Battery Management System”. As
baterias usadas em EVs têm exigências elevadas em termos de segurança, densidade de po-
tência (aceleração), densidade de energia (autonomia), elevada eficiência, ciclos de descarga
profundos e reduzidas taxas de auto-descarga entre outras. De entre as várias composições
químicas disponíveis para a construção de baterias, as baseadas em Lítio-Ferro-Fosfato (Li-
FePO4), Lítio-Itrio-Ferro-Fosfato (LiYFePO4) ou Lítio-Manganésio (LiMn2O4) são as mais
seguras e duradouras e como tal são estudadas no ISR-UC.
Os sistemas de gestão e monitorização de energia dedicados a EVs utilizam métodos de
estimação do estado de carga proprietários, não permitindo a sua utilização de forma simples
nem a sua melhoria. Por este motivo foi desenvolvido no ISR uma nova solução aberta e flexível
que permite estudar novos algoritmos de estimação do estado de carga.
Durante esta dissertação, instalou-se e testou-se um BMS comercial na bateria de Iões de
Lítio que alimenta uma das plataformas disponíveis no ISR-UC, o ISRobotCar. Uma segunda
plataforma, que foi re-instrumentada no decorrer deste projecto, também necessitou de substi-
tuição das suas baterias de chumbo originais por se encontrarem em fim de vida.
Devido às limitações encontradas em BMSs comerciais, desenvolveu-se de raiz um novo BMS,
designado nesta dissertação por ISR-BMS, que instrumenta uma nova bateria de LiYFePO4.
O estado de carga das baterias é uma informação fundamental para os utilizadores de
EVs. No entanto o estado de carga não é uma grandeza mensurável e apresenta uma grande
dependência da temperatura e das condições de operação da bateria. Existem várias técnicas
utilizadas para a estimação do estado de carga, que requerem a utilização de modelos das células
com diferentes complexidades. Para o ISR-BMS adoptou-se o modelo do circuito eléctrico
equivalente e modelizou-se uma célula de Iões de Lítio de 90Ah, baseado em testes laboratoriais
a diferentes temperaturas e ciclos de descarga.
Após a instalação do ISR-BMS na plataforma, efectuaram-se testes de condução real com
vista à validação das medições efectuadas e recolhidas do ISR-BMS.
Palavras-chave: Sistema de Gestão de Baterias, Baterias de Iões de Lítio, Modelização de
Células de Iões de Lítio, Estimação do Estado de Carga (SOC).
iii
Abstract
The Electric Vehicle (EV) is already on the roadmap of every important car manufacturer
and is seen as the solution to a more sustainable transport system, contributing to a reduction
of the Greenhouse Gas Emissions.
The Energy Storage System (ESS) is a key component for EVs. This includes the bat-
tery and all the management and monitoring systems that compose the Battery Management
System (BMS). Those batteries have very demanding requirements regarding safety, power
density (acceleration), energy density (autonomy), high efficiency, deep discharge cycles or
low self-discharge rates to name a few. From the available chemistries for the construction of
EV batteries the Lithium-Iron-Phosphate (LiFePO4), Lithium-Yttrium-Iron-Phosphate (LiY-
FePO4) or Lithium-Manganate (LiMn2O4) are the most safe and long lasting and because of
that they are studied at the ISR-UC.
The energy monitoring and management systems for EVs use proprietary State of Charge
(SOC) algorithms that do not allow their easy use or improvement. For this reason was devel-
oped at the ISR-UC a new BMS with an open and flexible architecture allowing the implemen-
tation of new SOC estimation algorithms, the ISR-BMS.
During this dissertation, a commercial BMS was installed and tested on a Lithium-Ion
battery pack that powers one of the electric platforms available at the ISR-UC. A new platform
was completely re-instrumented during the course of this project and required a new ESS to
replace their original lead acid batteries, which became end of life.
The SOC of the battery is a vital information to the EV user. It can be displayed as the
percentage of full charge capacity that is still available from the battery or be used to estimate
the vehicle range based on additional information from previous driving cycles. However, it can
not be measured directly from the battery and have a strong dependence with the temperature
and the operating conditions.
Several SOC estimation techniques are mentioned on the literature, that require cell models
with different complexity and computer process requirements. In this work an Equivalent Elec-
trical Circuit (EEC) model was adopted, with its parameters estimated based on experimental
data collected through cell testing at different temperatures and charge/discharge current pro-
files.
After installation of the ISR-BMS on the platform, on-road drive tests were performed to
acquire relevant information and validate the accuracy of collected measurements.
Key words: Battery Management Systems (BMS), Lithium-Ion Batteries, Cell Modeling,
State of Charge (SOC).
v
Contents
Acknowledgements i
Resumo ii
Abstract iv
List of Figures xi
List of Tables xiii
Nomenclature xv
1 Introduction 1
1.1 Motivation and context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Implementations and key contributions . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Structure of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 State of the Art 5

2.1 Battery Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Batteries for Electric Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1 Battery Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2 Lithium-Ion Battery Chemistries . . . . . . . . . . . . . . . . . . . . . . 8
2.2.3 Battery Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.4 Life Span of Lithium-Ion Batteries . . . . . . . . . . . . . . . . . . . . . 12
2.2.5 Aging Of Lithium-Ion Batteries . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.6 Performance of Lithium-Ion batteries . . . . . . . . . . . . . . . . . . . . 15
2.2.6.1 Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.6.2 Capacity Change with Temperature and Discharge Rate . . . . 15
2.3 Battery Management Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.1 Battery Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 Cell Balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3 ISR-Battery Management System 23

3.1 Preliminary Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Energy Storage System selection for the new developed platform . . . . . 26
3.2 BMS requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
vii
CONTENTS
3.3 ISR-BMS Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.3.1 Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3.1.1 The MAX14921 Analog Front End(AFE) Integrated Circuit . . 29
3.3.1.2 BMS Microcontroller . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.1.3 Current Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.1.4 Temperature Sensor . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.1.5 Communication Interface Hardware - USB . . . . . . . . . . . 33
3.3.1.6 Communication Interface Hardware - CAN bus . . . . . . . . . 33
3.3.2 Firmware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3.2.1 Real Time Operating System - FreeRTOS . . . . . . . . . . . . 33
3.3.2.2 Cells Voltage Measurement Procedure . . . . . . . . . . . . . . 34
3.3.2.3 Cell Balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.2.4 Communication Protocol for BMS Parametrization and Data
Logging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.3 The BMS Graphical User Interface for PC . . . . . . . . . . . . . . . . . 40
4 Battery Modeling 43
4.1 Modeling Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Description of Tests Performed . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.1 Pre-conditioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.2 Characterization Test Setup . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.3 Characterization Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 Cell Model Derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5 BMS Experimental Results 53

5.1 Battery Pack Current Measurements . . . . . . . . . . . . . . . . . . . . . . . . 53
5.2 Cell Voltages Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.3 Battery Balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.4 ISR-BMS Benchmarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6 Conclusion and future work 59

6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Bibliography 61
A BMS Schematics 66
B Firmware File Structure 69
C BMS-ISR Graphical User Interface (GUI) 70
viii
CONTENTS
D Winston Battery Datasheet 71
ix
List of Figures
2.1 Individual Cell and 8-Cell battery pack. . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Lithium-ion internal representation. . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Energy&power demands in batteries for different vehicle applications.[Rosenkranz
et al., ] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.4 Typical energy density for batteries and liquid fuels. [Tie and Tan, 2013] . . . . 9
2.5 Comparing power and energy density of different chemistries. [CARB, 2007] . . 9
2.6 Tradeoffs among the five principal Lithium-Ion battery technologies. [BCG, 2010] 10
2.7 Schematic drawing of shapes and components of several Lithium-ion battery
configurations: a – Cylindrical; b – Coin; c – Prismatic; d – Thin and flat.
[Tarascon and Armand, 2001] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.8 Lithium Battery from Boeing Dreamliner Airplane safety failure. . . . . . . . . . 12
2.9 Safety failure related mechanisms on cell level. [Electropaedia, a] . . . . . . . . . 13
2.10 Illustration of a battery with 80% DOD/SOC window. [Energy, 2012] . . . . . . 14
2.11 Capacity variation with temperature. [Electropaedia, b] . . . . . . . . . . . . . . 15
2.12 Capacity reduction at different temperatures and discharge rates. [Electropaedia,
b] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.13 Cell balancing topologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 ISR-UC Electric Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Lithium-Ion battery pack acquired to the company Autosil S.A. (72V, 3kWh, 24
modules in series with 4 parallel cells each - 4P24S). . . . . . . . . . . . . . . . . 24
3.3 ORION BMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4 TC Charger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.5 Battery test setup. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.6 Discharge logging of Autosil battery pack constant current discharge (15A). . . . 25
3.7 Energy Storage System of the newly developed platform . . . . . . . . . . . . . 27
3.8 ISR-BMS Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.9 Hardware development workflow. . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.10 Battery voltages with different cathode chemistries. [Electronicdesign, 2012] . . 29
3.11 ISR-BMS Architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.12 Circuit Board bottom side derived from Eagle Layout Software. . . . . . . . . . 31
3.13 HTFS200 - Hall-effect current sensor from LEM manufacturer. . . . . . . . . . . 32
3.14 MAX14921 cell voltages measurement procedure flowchart. . . . . . . . . . . . . 35
3.15 MAX14921 SPI Control of cells voltage readout. . . . . . . . . . . . . . . . . . . 36
3.16 Balancing flowchart. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
xi
LIST OF FIGURES
3.17 Smartphone display of ISR-BMS voltages and temperatures. . . . . . . . . . . . 39

3.18 ISR-BMS GUI parametrization menu. . . . . . . . . . . . . . . . . . . . . . . . . 40
3.19 ISR-BMS GUI live cell voltages plot. . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 EEC model adopted for ISR-BMS . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.2 Voltage plot of first pre-conditioning cycle. . . . . . . . . . . . . . . . . . . . . . 47
4.3 Cell characterization test setup. . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4 Characterization cell box GUI. . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.5 Measured versus simulated data at 25 ºC. . . . . . . . . . . . . . . . . . . . . . 50
4.6 EEC estimated parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.1 Comparative current measurements of ISR-BMS and Fluke i310s current probe. 54
5.2 Voltage measurements logged on the PC during a real road drive. . . . . . . . . 54
5.3 BMS debugging after installation. . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.4 Cell voltages during charge at 10A constant current. . . . . . . . . . . . . . . . . 56
5.5 Combined information of ISR-BMS, GPS and encoders from the electric platform
during a real road drive. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.6 Benchmarking the ISR-BMS with other commertial BMSs. . . . . . . . . . . . . 58
A.1 BMS-ISR Schematics 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

A.2 BMS-ISR Schematics 2/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
B.1 Firmware file structure (by Doxygen) . . . . . . . . . . . . . . . . . . . . . . . . 69
C.1 ISR-BMS GUI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
D.1 Winston battery LYP90AHA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
xii
List of Tables
2.1 Comparative analysis of commercially available BMS ICs. . . . . . . . . . . . . . 18

2.2 Advantages and limitations of existing Lithium-Ion modeling methods. [Antaloae
et al., 2012] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Comparison of different cell balancing methods. [Rahimi-eichi, 2013] . . . . . . . 22
3.1 Cost/performance comparison of possible alternative BMS architectures. . . . . 31

3.2 ISR-BMS Communication Protocol . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3 OBDII commands implemented on ISR-BMS. . . . . . . . . . . . . . . . . . . . 39
4.1 Overview of Applicable International Standards for Lithium-Ion cells or batteries 44

4.2 LFP090AHA cell specifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Test Setup Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
xiii
Nomenclature
API Application Programming Interface
BEV Battery Electric Vehicle
BMS Battery Management System
C-rate Discharge current required to fully discharge the battery in one hour
CAN Controller Area Network
CC/CV Constant Current/Constant Voltage
EEC Equivalent Electrical Circuit
EKF Extended Kalman Filter
ESS Energy Storage System
EV Electric Vehicle
FET Field Effect Transistor
GHG Greenhouse Gas
GUI Graphical User Interface
HEV Hibrid Electric Vehicle
I2C Inter-Integrated Circuit
IC Integrated Circuit
ICE Internal Combustion Engine
ISR Institute of Systems and Robotics
ISR-BSM New BMS develloped for ISRobotCar
OBDII On Board Diagnostic System
OEM Original Equipment Manufacturer
PID Proportional-Integral-Derivative
SAR Successive Approximation Register
xv
Nomenclature
SOC State Of Charge
SOH State of Health
SOL State of Life
SPI Serial Peripheral Interface
xvi
Nomenclature
xvii
Chapter 1
Introduction
This chapter presents the work developed under this Master project. It starts with the
motivations and context of the developed work. The main concepts involved are introduced
as well as the main objectives and key contributions. Finally the whole document structure is
presented.
1.1 Motivation and context

It is now unanimous that is necessary to act fast to revert the negative effects of climate
changes caused by massive Greenhouse Gas (GHG) emissions and CO2 in particular. The road
traffic is heavily dependent on fossil fuels and therefore it is one of the major contributors
for those emissions. On the other hand more and more people are living in urban areas with
negative impact on the air quality. The Electric Vehicle (EV) is seen as a potential solution to
minimize those effects due to its high efficiency and zero emissions during the use phase.
Further-more, EVs can actively interact with the electrical grid since they can be used as
an energy bank for energy storage from renewable power sources, known as vehicle to grid
interaction. Another interesting application consist in the use of EVs as controllable loads in
demand response programs.
However, there are still some restrictions related to autonomy and price, that limit its wide
acceptance. The main reason for this, relies on the batteries, the most bulky and expensive
component of an EV.
The EV is already on the roadmap of every important car manufacturer. It is expected a
continuous evolution on the different fields that will conduct to lighter materials (like composite
materials), batteries with higher energy density and longer life span, faster and more efficient
charging. The price is expected to decrease as well, promoting the adoption of efficient and
sustainable electric mobility solutions. The EV is seen as the solution to a more sustainable
transport system that can contribute to a reduction of GHGs emissions and the import of
crude oil, due to a much higher efficiency, from 28-30% of an Internal Combustion Engine
(ICE) vehicle to 74-85% of an EV, and zero tailpipe emissions [Faria et al., 2013]. In Portugal,
even with the actual energy mix, EVs can deliver an overall reduction of GHG emissions.
During the course of this work, two electric platforms were equipped with Lithium-Ion
battery packs. For one a full battery pack and a commercial BMS were installed. After testing
the system some flexibility was missing to implement different balancing algorithms. It was
1
Chapter 1: Introduction
also impossible to change the simple State Of Charge (SOC) estimation procedure based on
coulomb counting and voltage drift that needs to be fully programmable by the user.
1.2 Objectives
The objectives of this work were defined as follows:
1. Design a new BMS platform based on available dedicated integrated circuits and micro-
controllers targeting fast sampling, high accuracy, low consumption and low cost;
2. Build an open system that allows the implementation of different charging and balancing
algorithms, and that is able to be used with different battery models;
3. Install a fully functional prototype on a ISR EV;
4. Develop the required experimental testing to validate the concept;
5. Benchmark existing commercial BMSs in comparison with the ISR-BMS.
1.3 Implementations and key contributions

The main implementations and contributions can be summarized as follows:
1. Installation of a commercial BMS on a 72V, 40Ah, 3kWh Lithium-Ion battery pack in

collaboration with the colleague Marco Silva;
2. Development of a new BMS architecture, including hardware, software and Graphical

User Interface (GUI).
3. Instrumentation with the new BMS architecture of a new battery pack (72V, 90Ah,
6.5kWh) to power a new electric platform;
4. Individual cells were tested in laboratory under controlled temperature and charge/dis-
charge conditions;
5. The parameters of an Equivalent Electrical Circuit (EEC) were estimated based on the
experimental data fitting using a Matlab model [Huria et al., 2010] and estimation soft-
ware;
6. The BMS current measurement accuracy was verified against a commercial Fluke current
probe (Model i310s) on a road test drive.
2
1.4: Structure of the dissertation
1.4 Structure of the dissertation

On Chapter 1, the work developed under this Master Dissertation is presented as well as the
motivations and context of the developed work. The main concepts involved are introduced, as
well as main objectives and key contributions. On Chapter 2 the key components of a Battery
Management System are explored. The latest developments on the field of batteries, BMS and
cell modeling are presented as well. On Chapter 3 the architecture of the ISR-BMS and the
choices undertaken are justified. An overview of the whole design process is presented, as well
as the key features of the developed hardware and software. Chapter 4, explains the modeling
process of a Lithium-Ion cell and discusses the model features and the model parameters esti-
mation based on experimental data collected in the laboratory. A cell characterization setup,
built for this purpose, is also briefly described. On Chapter 5, several experimental tests made
with the EV after the ISR-BMS installation are described. The measurements taken by the
ISR-BMS are verified. Its on-line performance as well as its ability to safely charge and balance
the battery pack are also presented. Finally, on Chapter 7, the conclusions and some topics for
future work are presented.
3
Chapter 2
State of the Art

On this chapter the key components of a Battery Management System are explored. The
latest developments on the field of batteries, BMSs and cell modeling are presented.
Batteries play an important role on EVs. They are the sole power supply of the whole
system. EVs demand the use of high performance batteries that can deliver the required power
and energy and still weight as less as possible. Different battery technologies are available
nowadays but only a few are suitable for automotive applications. Lithium-Ion based batteries
are one of the most promising technologies. Battery systems require the use of a BMS that
monitors their state and guarantees their operation within safe limits and optimizes the use
of available capacity. Prior to discuss the BMS, the batteries are introduced and their key
parameters are presented with special focus on the Lithium-Ion chemistry.
2.1 Battery Introduction

A battery is an electrochemical device that converts the chemical energy contained in its
active materials directly into electrical energy. In automotive rechargeable systems the battery
is recharged by the reversal of this process. While the term “battery” is often used, the basic
electrochemical unit is the “cell”. A battery consists of one or more of these cells, connected in
series or parallel, or both, depending on the desired output voltage and capacity. Figure 2.1
shows a single cell and a battery pack with 8 cells connected in series.
The cell consists of three major components (represented in Figure 2.2):
1. The anode or negative electrode - the reducing or fuel electrode - that releases electrons
to the external circuit and is oxidized during the electrochemical reaction;
2. The cathode or positive electrode - the oxidizing electrode - that accepts electrons from
the external circuit and is reduced during the electrochemical reaction;
3. The electrolyte - the ionic conductor - that provides the medium for transfer of charge, as
ions, inside the cell between the anode and cathode. The electrolyte is typically a liquid,
such as water or other solvents, with dissolved salts, acids, or alkalis to improve ionic
conductivity. Some batteries use solid electrolytes, which are ionic conductors at the cell
operating temperature.
5
Chapter 2: State of the Art
Figure 2.1: Individual Cell and 8-Cell battery pack.
Figure 2.2: Lithium-ion internal representation.
6
2.2: Batteries for Electric Vehicles
2.2 Batteries for Electric Vehicles

EVs need a portable source of electrical energy. All this energy is previously stored on
the Energy Storage System (ESS) that should match the energy requirements of the different
car systems (traction, lighting, climate control) and guarantee a reasonable autonomy. With
existing state-of-the-art technology, most of this energy is stored in batteries (sometimes com-
plemented with super-capacitors or flywheels to improve response to peak demands).
2.2.1 Battery Requirements

EV batteries or traction batteries differ from the typical automotive battery in internal
combustion engines. The latest are also known as SLI batteries (Starting, Lighting, Ignition)
because they are used to power the start motor, the lights and the ignition system of a con-
ventional car engine. Those are typically lead-acid type, and are made of six galvanic cells in
series to provide a 12V system. They need to supply high current peaks for short time and
they work mostly close to full charge capacity.
Usually the two battery types coexist on an EV with different functions. The traction
batteries need to be designed to provide energy to the electric motor for long periods. They
need as well to provide some peak power during fast acceleration.
EV Battery Operating Requirements:
• Large capacity batteries are required to achieve reasonable range. A typical EV uses
around 100 to 150 Wh per kilometer depending on the terrain and the driving style.
• The battery must be capable of regular deep discharge (80% Depth of Discharge (DOD)
or higher) operation;
• It is designed to maximize energy content and deliver full power even with deep discharge
to ensure long range;
• A range of capacities will be required to satisfy the needs of different sized vehicles and
different usage patterns;
• Must accept very high repetitive pulsed charging currents when regenerative braking is
applied;
• Without regenerative braking, controlled charging conditions and lower charging rates
are possible ;
• Routinely receives a full charge;
• Often also reaches nearly full discharge;
• Fuel-gauging warning near "empty" point;
7
• Needs a Battery Management System (BMS);
• Needs thermal management;
• Typical voltage > 300V;
• Typical capacity > 20 - 60kWh.

Typical energy and power requirements for different vehicle applications can be found in Figure
2.3.
Figure 2.3: Energy&power demands in batteries for different vehicle applications.[Rosenkranz

et al., ]
Batteries for EVs are characterized by their relatively high power-to-weight ratio, energy-to-
weight ratio and energy density. Smaller, lighter batteries reduce the weight of the vehicle and
improve its performance. Compared to liquid fuels, nowadays, battery technologies have much
lower specific energy (Figure 2.4) and this often influences the maximum all-electric range of
the vehicles.
In automotive battery applications, as can be seen in Figure 2.5, the primary trend is
increasing energy density. This has favored a move towards Lithium-Ion chemistries, even
though other chemistries can demonstrate greater longevity, and lower specific costs.
On this basis, the remainder of this section is focused on Lithium-Ion chemistries, their
operation, potential and drawbacks.
2.2.2 Lithium-Ion Battery Chemistries

Lithium-Ion batteries comprise a family of battery chemistries that employ various com-
binations of anode and cathode materials. Each combination has distinct advantages and
8
Figure 2.4: Typical energy density for batteries and liquid fuels. [Tie and Tan, 2013]
Figure 2.5: Comparing power and energy density of different chemistries. [CARB, 2007]
9
Figure 2.6: Tradeoffs among the five principal Lithium-Ion battery technologies. [BCG, 2010]
disadvantages in terms of safety, performance, cost, and other parameters.

The most prominent technologies for automotive applications are Lithium-Nickel-Cobalt-
Aluminum (NCA), Lithium-Nickel-Manganese-Cobalt (NMC), Lithium-Manganese-Spinel (LMO),
Lithium-Titanate (LTO), and Lithium-Iron-Phosphate (LFP). The technology that is currently
most prevalent in consumer applications is Lithium-Cobalt-Oxide (LCO), which is generally
considered unsuitable for automotive applications because of its inherent safety risks. All au-
tomotive battery chemistries require elaborate monitoring, balancing, and cooling systems to
control the release of energy, prevent thermal runaway, and ensure a reasonably long life span for
the cells. The recent explosion in innovation is driven by the need to break some fundamental
compromises in battery technology. On the technical side, competing Lithium-Ion technologies
can be compared along six dimensions as shown in Figure 2.6:
1. Safety;
2. Life span (both number of charge and discharge cycles and overall battery age);
3. Performance (peak power at low temperatures, SOC measurement, and thermal man-
agement);
4. Specific energy (how much energy the battery can store per kilogram of weight);
5. Specific power (how much power the battery can deliver per kilogram of weight);
6. Cost.
Currently, as shown in Figure 2.6, no single technology wins along all six dimensions. Choos-
ing a technology that optimizes performance along one dimension inevitably means compromis-
ing on other dimensions. NCA technology, for example, is a fairly high-performance solution
but presents safety challenges, whereas LFP technology is safer at the cell level but provides a
10
Figure 2.7: Schematic drawing of shapes and components of several Lithium-ion battery con-
figurations: a – Cylindrical; b – Coin; c – Prismatic; d – Thin and flat. [Tarascon and Armand,
2001]
low specific energy. Other important characteristic of the Lithium-based technology is its flexi-
bility to build batteries in several shapes and configurations, such as cylindrical, coin, prismatic
and flat, as presented on Figure 2.7.
2.2.3 Battery Safety

Safety is one of the most important criterion for electric-car batteries. Even a single battery
fire could turn public opinion against electric mobility and set back industry development for
months or years. The main concern in this area is avoiding thermal runaway - a positive-
feedback loop whereby chemical reactions triggered in the cell exacerbate heat release, poten-
tially resulting in a fire. An overcharged battery, too-high discharge rates, or a short circuit
(internal or external) can cause thermal runaway. Chemistries that are prone to thermal run-
away, such as NCA, NMC, and LMO, must be used in conjunction with system-level safety
measures that either contain the cells or monitor their behavior. Such measures include a robust
battery box, a very efficient cooling system (to prevent the early stages of thermal runaway),
and precise SOC monitoring and cell-discharge balancing. Recently, the entire fleet of Boeing
Dreamliner airplanes was grounded because of a fire in a Lithium-Ion battery (Figure 2.8).
Despite the virtues of Lithium-Ion batteries, several factors have prevented a wider spread
of this technology, and the most significant one is, without doubt, safety. This is mainly due
to the higher energy density of those batteries, which implies more heat released in case of
failure, and to the flammability of the most common organic solvents used as electrolytes.
More than 2 million products containing Lithium-Ion batteries have been recalled because of
battery problems since October 2006. For successfully scaling up this technology to higher
11
Figure 2.8: Lithium Battery from Boeing Dreamliner Airplane safety failure.
power applications like HEVs, it is imperative to reduce these concerns, i.e. to reduce the
incidents and mitigate the effects of the incidents when they do occur. This is done by means
of various expedients, among which a battery management system, through its thermal and
charge management, plays a critical role. The dangerousness of the above-mentioned events
in a particular Lithium-Ion cell highly depends on the particular chemistry of the cell. Some
chemistries are more susceptible than others, and the more dangerous chemistries should be
avoided for high power applications, or should be modified with appropriate additives. Each
cell should be adequately insulated from the others in the battery pack, in order to avoid the
danger of chain reactions causing neighboring cells to fail due to the heat exchanged between
cells. It is responsibility of BMS to ensure that the cells of a battery operate within their
rated specifications. This can be done, for example, through cell state monitoring, charge and
discharge current limiting, cooling system management, communication between the battery
system and the vehicle, and SOH estimating. In addition, avoiding high SOC levels when the
temperature is high can be desirable. In fact, the higher the SOC of the cell is, the more likely
the cell may fail when thermally stressed. In order to minimize the risk of safety-related failures
(Figure 2.9), safeguards should be adopted redundantly and at various levels, ensuring a robust
chemical and mechanical battery design, controlling the quality of the manufacturing process,
and employing a satisfactory electronic supervisory system. Eventually, the development of
standards to evaluate the safety performance of the resulting battery systems will prove decisive
to establish Lithium-Ion batteries as safe energy storage devices, especially in the automotive
field.
2.2.4 Life Span of Lithium-Ion Batteries

Life span is an important parameter of an EV battery as it impacts strongly the total cost of
ownership (TCO) of the vehicle. Batteries are the heaviest component of an EV and their cost
can be higher than half of the final price of acquisition. During their use, they can be subject to
12
Figure 2.9: Safety failure related mechanisms on cell level. [Electropaedia, a]
different kind of stresses depending on environment conditions and driving profiles. They are
especially sensitive to extreme temperature conditions or over-charge and discharge conditions.
Therefore, it is important to understand the aging mechanisms of the battery in order to avoid
the conditions that can speed up those effects. There are two ways of measuring battery life
span: cycle stability and overall age. Cycle stability is the number of times a battery can be
fully charged and discharged (80%DoD) before being degraded to 70 percent of its original
capacity at full charge. Overall age is the number of years a battery can be expected to remain
useful. Today’s batteries do meet the cycle stability requirements of electric cars under test
conditions. Overall age, however, remains a hurdle, in part because aging accelerates under
higher ambient temperatures. Yet, it is unclear how fast various kinds of batteries will age
across a range of automotive-specific temperature conditions. To manage these uncertainties,
OEMs are specifying batteries of sufficient size to meet electric cars energy-storage needs over
the typical life of a vehicle. Most automotive manufacturers are planning for a ten-year battery
life span, including expected degradation. For example, an OEM whose electric car nominally
requires a 12-kilowatt-hour (kWh) battery is likely to specify a 20-kWh battery instead, so that
after ten years and 40 percent performance degradation, the battery will still have sufficient
energy capacity for normal operation. Of course, this approach increases the size, weight, and
cost of the battery, adversely affecting the business case for EVs.
Different battery strategies can be used to preserve the battery lifetime. A conservative
method could be to avoid a SOC above 80% keeping some margin for regenerative braking
energy and never discharge below 80% DoD. Such a method is shown on Figure 2.10. The
13
Figure 2.10: Illustration of a battery with 80% DOD/SOC window. [Energy, 2012]
drawback already mentioned is that the useful battery energy will be less than the installed
battery capacity.
2.2.5 Aging Of Lithium-Ion Batteries

Unfortunately, Lithium-Ion batteries are complex systems to understand, and the processes
of their aging are even more complicated. Capacity decrease and power fading do not originate
from one single cause, but from a number of various processes and their interactions.
Aging in Lithium-Ion cells may mainly come from:
• Reactions of active materials with electrolyte at electrodes interfaces;
• Self-degradation of active materials structure on cycling;
• Aging of non-active components (ex: electrodes binders).
These phenomena lead to energy and/or power losses. Aging mechanisms occurring at anodes
and cathodes differ significantly. The influence of the electrolyte and the aging of the electrolyte
itself (and the separator as well) mainly take place at the electrodes and in interaction with
them:
• On the negative electrode side, the main parameters are passivation layer stability, the
layer structure/composition (effect of additives in electrolyte and formation process) and
polluting agents;
• On the positive electrode side, key factors seem to be the material structure evolution
and organic solvent oxidation.
These phenomena are highly dependent of voltage and temperature.
14
Figure 2.11: Capacity variation with temperature. [Electropaedia, b]
2.2.6 Performance of Lithium-Ion batteries
2.2.6.1 Efficiency
During the charge, some of the energy is inevitably lost in the chemical conversion process
usually as heat. Similarly on the return trip, for the same reasons, some of the available charge
is lost and only part of the stored charge is available for doing work. The round trip energy
loss is about 3% for a Lithium-Ion battery. The efficiency is the ratio between the discharge
energy and the charging energy. The battery efficiency and capacity will change depending
on charging and discharging parameters. The typical charge and discharge characteristics are
usually part of the battery specifications.
2.2.6.2 Capacity Change with Temperature and Discharge Rate
Figure 2.11, shows how the capacity of a Lithium-Ion cell varies with temperature and
discharge rate. It shows that at normal working temperatures, the efficiency of the cell is very
high, but at low temperatures, there is a major drop in efficiency particularly at high discharge
rates, which can give rise to serious errors in the estimation of the SOC. This phenomenon is not
peculiar to Lithium cells as other cell chemistries also demonstrate deterioration in performance
at low temperatures.
The discharge rate (C or C-rate) of a Lithium-Ion cell is typically the current needed to
discharge a fully charged cell in one hour. Its value, in ampere units, is calculated from the
specified cell capacity in ampere-hours divided by one hour time.
The Lithium-Ion cell is working between its specified upper and lower voltage cut-off limits
15
Figure 2.12: Capacity reduction at different temperatures and discharge rates. [Electropaedia,
b]
of 4.2.Volts and 2.5 Volts, respectively. These are considered the fully charged and the empty
conditions of the cell. The "Full" line is the point at which the cell reaches full charge using the
constant current - constant voltage charging method at the corresponding temperature. Two
"Empty" lines are shown corresponding to two different discharge rates 0.2C and 1.0C. The
capacity of the cell at a given rate and temperature is the difference from the "Full" line and the
corresponding "Empty" line. Note that the cell is very inefficient at giving up its charge at high
discharge rates and low temperatures. In other words, its efficiency deteriorates dramatically
at low temperatures. Note also that the cell above could be fully discharged at the high current
rate yet could be further discharged at the low current rate by the number of milliamp-hours
between the two "Empty" points that correspond to the present cell temperature. Typical
cell specifications only give the capacity at 25°C and 0.3C rate. Figure 2.12, characterizes the
performance of a Lithium-Ion cell over two of its expected operating conditions (temperature
and discharge rate). The matrix of capacity values associated with all the possible current-
temperature combinations is useful as a look up table used by charge estimation algorithms of
a BMS.
2.3 Battery Management Systems

As EVs are evolving fast, the batteries need to improve and adapt to the stringent require-
ments of automotive EV. The BMS is a critical part of EVs, to guarantee that batteries operate
safely and last longer. In fact, the total cost of ownership of an EV is influenced by the needed
installed capacity but also by the time it remains usable. The BMS costs only a fraction of
the total investment but guarantees its lifetime. The BMS architecture can be split in two
fundamental parts:
• the hardware that acquires the state variables of the batteries: voltage, current and
temperature. It should ensure the simultaneous acquisition of the measurements in order
16
2.3: Battery Management Systems
to improve the accuracy of the estimations. It needs a high sampling rate to guarantee
that the system dynamics can be monitored and it needs to accurately measure the
electrical and thermal variables in order to avoid error propagation and to allow a reliable
estimation of all the parameters that are not measurable directly.
• the software that actively monitors all the variables, manages the battery protections
and alerts, manages the internal communication within the different hardware blocks,
manages the communication with the peripheral systems and implements estimation or
cell balancing algorithms.
There are different approaches for the hardware that can be centralized or distributed with
advantages and disadvantages for each situation.
Also the modeling of the cell is an important part of the BMS. Models can be as simple
or as complex as required. Since a simple voltage source and series resistor to electrochemical
models that represent all the interactions on the battery molecular structures.
The experience from the author reveals that most commercial BMS available rely on the
most simple estimation procedures to calculate most of the parameters. The coulomb counting
technique and voltage drift offset are used on the ORION BMS [ORION, 2014]. The main
disadvantage from coulomb counting method, is the fact that it needs always a known starting
point because the current integration over the time will accumulate errors that can become
substantial if a complete charge or discharge do not occur.
Lithium-Ion batteries are already available for many years for portable devices such as
laptops and mobile phones. Also BMSs for those devices are available from different IC manu-
facturers and are already mature. Those ICs are not suited for EV Battery pack because they
can range from tens to more than hundred cells to provide a high power system needed for
traction. Those systems with high voltage and high currents impose additional communication
signal isolation for devices interface, and are still under strong development by companies, car
manufacturers, research institutes and universities [Xing et al., 2011].
The energy crises of the 1970s and 1980s brought a renewed interest in EVs [Wikipedia,
]. They were re-introduced in the early 1990s and became mainstream in the 2000s leading
to breakthroughs in battery technology. Lithium-Ion technology has been widely adopted for
EV/HEV applications due to its high energy density and power density, low self-discharge
and long life span [Tao et al., 2011]. It faced an enormous improvement on the last decade.
Unfortunately the BMS developments could not follow those developments hindered by the
following difficulties, among others [Xing et al., 2011]:
• Battery modeling;
• Battery state estimation (SOC, SOL, SOH);
• Cell balancing.
17
The basic functions of a generic BMS can be summarized as it follows:
• Measure the battery state parameters like, individual cell voltages, pack current and
temperature;
• Trigger alarms in case of abnormal conditions or even act to disconnect the batteries in
order to avoid damaging failures;
• Control charge and discharge conditions;
• Cell balancing;
• Thermal management;
• Estimation and communication of important battery parameters like SOC, SOH, SOL.
Not all those functions are available on current commercialized BMSs. Also those functions are
not standard. Table 2.1 compares some popular marketed products.
Table 2.1: Comparative analysis of commercially available BMS ICs.

Manufac- Maxim/MAX14921 TI / BQ76PL536 AMS/AS8506
turer/Part [Maxim, 2014] [BQ76PL536, ] [AS8506, ]
Number
Description 3-16 cell analog front end 3-6 cell analog front 7-cell analog front end
scalable to any number of end plus fault scalable to any number of
cells detector scalable to cells
192 cells
Measured Cell voltage, current voltage, current voltage, current
Parameters
Measured Pack temperature, pack voltage temperature temperature
Parameters
Safety open-wire detection over and cell voltage diagnosis
Protection undervoltage/overvoltage undervoltage
alarms protection,
overtemperature
protection
Communica- SPI SPI SPI
tion
Passive cell integrated cell-balancing control outputs internal FET- Sequential
balancing FET drivers - Simultaneous with balance balancing, 50mA max.
current set by
external
components
Active cell N/A N/A option for active charge
balancing balancing with external
components - DC-DC
Flyback
Internal ADC No Yes. 14 bits SAR Yes. 12 bits SAR
18
Those products have the following limitations:
• No self data logging capabilities. This is an important feature for storing the driving
patterns in order to build up and update the SOC model;
• Lack of State Of Health (SOH) and State Of Life (SOL) estimations. This is useful
to characterize the performance of the battery and schedule the battery replacement or
repair;
• No interchangeability among current BMSs as they all have different pin-outs, different
cell balancing schemes, different communication channels between BMS ICs and between
BMS IC and host controller.
2.3.1 Battery Modeling

There are nowadays several types of battery models ranging from simple electrical equivalent
circuits (EEC) to complex electrochemical models that try to model the molecular reactions
inside the cell. Booth have advantages and limitations as shown on Table 2.2.
Table 2.2: Advantages and limitations of existing Lithium-Ion modeling methods. [Antaloae
et al., 2012]
Run-time
Modeling Method Accuracy Setup simplicity Type of method
performance
Coulomb counting offline
Direct voltage measurement offline
Impedance measurement offline
Electrode electrochemistry offline
Equivalent electrical circuit
offline
(EEC)
Adaptive methods
online
(KF,EKF,ANN,FL)
They can be categorized in two main classes:
• offline methods - derive the model parameters from analysis of available datasets. Most
of the times this data is measured in laboratory under controlled conditions. This data
should reflect the real application and environment conditions so it usually lead to exten-
sive testing. Since it is virtually impossible test all the real conditions, approximations
need to be done.
• online methods - use an adaptive approach to update the model parameters as the bat-
tery is being used. Those methods lead to improved accuracy in detriment of increased
complexity and reduced run-time performance.
19
All those methods are extensively mentioned in the literature.
The coulomb counting [Ng et al., 2009] [Piller et al., 2001], direct cell voltage measure-
ment [Coleman et al., 2007], impedance measurement [Huet, 1998], electrode electrochem-
istry [Bergveld, 2001] and equivalent electrical circuit [Schweighofer et al., 2003] [Chen and
Rinc, 2006] [Antaloae et al., 2012] [Huria et al., 2010] are examples of offline methods. The
Kalman Filter, Extended Kalman Filter [Plett, 2004], Artificial Neural Network [Parthiban
et al., 2007]and Fuzzy Logic [Salkind et al., 1999]are examples of online adaptive methods.
2.3.2 Cell Balancing
Cell balancing is an important part of any Battery Management System. There are not two
cells exactly the same. Even cells manufactured under the same conditions and coming from
the same batch can have minor changes due to their electrochemical nature. Typical battery
packs for EVs can have more than one hundred cells. Depending on how those long strings are
mounted, some cells can have slightly different temperature conditions during operation. Also in
stationary mode those differences can exist due to environmental conditions. The self-discharge
rate of Lithium-Ion cells is very dependent on temperature. So the differences among cells tend
to increase with normal cycling and aging, leading to an unbalance on the SOC of the cells.
Since the cells are in series, they share the same current. This means that the cell with
higher SOC will reach its maximum terminal voltage before the others can reach full charge.
On the other way, the cell with lowest SOC will reach its minimum voltage on discharge before
the other ones can deliver its full capacity. Continuing with the charge or discharge in those
conditions will lead to overcharge of the cell with highest SOC or over-discharge of the cell with
lowest SOC. In this way the useful capacity will be lower than the rated capacity.
Then, in order to establish safe operation, maximize battery life and optimize its useful
capacity, proper individual cell balancing should be achieved [Kuhn et al., 2005].
The balancing topologies can be categorized as passive and active balancing. The passive
balancing methods act removing the excess charge from the most charged cell(s) through a
passive resistor element, until the charge matches those with lowest charge in the pack. The
resistor element will be either in fixed mode or switched. The active cell balancing methods
remove charge from the most charged cell(s) and deliver it to lowest charged cell(s). Different
topologies are used according to the active element used for storing the energy such as capacitor
and/or inductive component as well as controlling switches or converters. The different solutions
vary in complexity, efficiency and response time. An overview of the different configurations
available on the literature [Daowd et al., 2011] can be found in Figure 2.13.
20
FixedMshuntMresistor
Passive
SwitchedMshuntM
resistor
SwitchedMcapacitor
CapacitorMbase
SingleMSwitchedM
CellMBallancing capacitor
Single/multiMinductor
TransformerM SingleMwindingsM
/InductorMbase transformer
Active
MultipleMwindingsM
transformer
CukMconverter
Buck/boostMconverter
ConverterMbase FlybackMconverter
Full-bidgeMconverter
Quasi-ResonantM
Converter
Figure 2.13: Cell balancing topologies.
A comprehensive revision on cell balancing is provided by [Cao et al., 2008] and is summa-
rized by [Rahimi-eichi, 2013] on Table 2.3.
21
Table 2.3: Comparison of different cell balancing methods. [Rahimi-eichi, 2013]
22
Chapter 3
ISR-Battery Management System

In this chapter, the architecture of the ISR-BMS is presented and the choices undertaken
are fundamented. The whole design process and key features are presented along the developed
hardware and software.
3.1 Preliminary Work

An example of the EVs used on this project is shown in Figure 3.1. Those are derived from
a Yamaha electric AGV that was re-instrumented at ISR-UC. [Silva et al., 2012]
Figure 3.1: ISR-UC Electric Vehicle
The original platforms were equipped with Lead Acid batteries that reached the end of life.
At the beginning of this work the following equipment was available to be installed on the first
re-instrumented platform:
1. Lithium-Ion battery pack aquired to the company Autosil S.A., shown in Figure 3.2.
23
Chapter 3: ISR-Battery Management System
Figure 3.2: Lithium-Ion battery pack acquired to the company Autosil S.A. (72V, 3kWh, 24
modules in series with 4 parallel cells each - 4P24S).
2. A commercial BMS from ORION manufacturer with CAN interface, shown in Figure 3.3.
Figure 3.3: ORION BMS.
3. A 1.5kW Battery charger from TC Charger, shown in Figure 3.4.
Figure 3.4: TC Charger.
The first task in this work was to integrate the ORION BMS with the Autosil battery pack
and the TC Charger. This involved all the wiring and the parametrization of the BMS to the
used cells and charger.
After integration, the system was extensively tested in the lab to check among other param-
eters the battery capacity, the BMS indications, the charge and the cell balancing. To discharge
the cells a bank of resistors with configurable resistance was used as shown in Figure 3.5.
24
3.1: Preliminary Work
Figure 3.5: Battery test setup.
After the first two preconditioning charge/discharge cycles, was concluded that the battery
pack was completely unbalanced. At discharge, 2 cell groups (11 and 18) were reaching the
cut-off voltage (2.8V) much faster than the average, limiting the useful capacity of the battery
on discharge. This can be seen on the discharge log shown in Figure 3.6.
Figure 3.6: Discharge logging of Autosil battery pack constant current discharge (15A).
The same behavior could be seen during the charge, where cell 1 clearly was reaching the
maximum charge voltage (3.65V) before all the others.
25
Due to this unbalance on the cells SOC, only 2/3 of the total battery pack capacity (40Ah)
could be effectively used. The ORION balancing feature was used for one full day without
visible results because the maximum balancing current is only 200mA. This current is enough
for regular balancing but not suitable for highly initial unbalance.
The solution was to completely charge every cell individually with a CC/CV profile. To
accomplish this, the battery pack was unmounted to access the battery terminals and power
supplies were attached sequentially to every cell controlling the charge at 2A till maximum
charge voltage could be reached (CC) and waiting the required time to the charge current fall
bellow 100mA (CV).
This procedure, besides being time consuming, was also risky since all the setup was ma-
nipulated with live voltages accessible, where the smallest unintentional mistake could give rise
to a short-circuit of the pack cells (partial or total) originating currents of several hundred
Amperes and high probability of material and personal injury.
3.1.1 Energy Storage System selection for the new developed plat-
form
The EV was equipped with 6 Lead Acid bateries of 12V (72V-78V) with a capacity of 60Ah
that became end of life and needed to be replaced.
The choice for a Lithium-Ion chemistry was taken based on their advantages comparatively
to lead acid batteries, such as high-power density, high energy density, high efficiency, no
memory effect and low self discharge.
Within the lithium chemistry, the choice was based on safety, form factor, easiness of in-
stallation, suitability for the power train requirements, cost and availability.
Since the battery pack and BMS needed to be assembled in house, single cells with higher
capacity were preferred in comparison with multiple parallel cells to reach the same capacity.
Also the prismatic shape could fit better on the compartment reserved for the battery. This
translates in less terminals (connection bars), less screws, so less probability of bad connections
or high-impedance paths within the battery pack.
The final choice was also influenced by the availability at ISR of 12 cells of LiFeYPO4
from Winston manufacturer, with a capacity of 90Ah and a nominal voltage of 3.2V. The cell
datasheet can be found on Appendix D. Those cells were used in a previous project and were
in good condition. Since the same model could be bought in Europe with fast delivery, this cell
model was chosen to power the new platform. The remaining 12 cell units needed to reach the
required power train voltage (24 cells in series) plus 4 units to be used as auxiliary battery for
the instrumentation were acquired. The Battery pack prior to the BMS installation as well as
the instrumentation battery can be seen in Figure 3.7.
The effect of mixing brand new cells with already used cells represents an additional chal-
lenge to the developed BMS. In fact this represents a real situation where one cell can become
26
3.2: BMS requirements
Figure 3.7: Energy Storage System of the newly developed platform
damaged and needs to be replaced, while the remaining cells are still in a good condition.
3.2 BMS requirements

Due to the limitations found on commercial available BMSs it was decided to develop a
new BMS module from the scratch using available blocks from main IC manufacturers and add
the required “intelligence” and flexibility to test different charge, discharge, cell ballancing and
parameter estimation algorithms.
The preliminary requirements were then defined and include:
• Interface USB;
• Graphical User Interface (GUI) for parametrization and monitoring;
• Data logging;
• Programmable CAN Interface (J1939);
• Galvanic Isolation from Power battery;
• Passive cell balancing with proprietary algorithm;
• High sampling rate;
• Voltage measurement accuracy: 1-2mV;
• Current measurement accuracy: 300mA;
• Very low consumption;
• Thermal management.
The main architecture was then outlined as shown in Figure 3.8.
27
Figure 3.8: ISR-BMS Architecture
3.3 ISR-BMS Development

The complete development can be divided in three main components:
1. The Hardware - includes the PCB, components, enclosure, cables.
2. The Firmware - includes the embedded programming for the microcontroller to perform
and acquire all the measurements and manage the communication with the PC (USB) or
the charger (CAN). To manage the complexity and optimize the task scheduling, a real
time operating system was used (freeRTOS). The program was developed using the C
language and a C18 compiler.
3. The GUI - for easy and intuitive parametrization of the BMS a PC GUI was develloped.
This software was built in QT, a cross-platform application framework for developing
graphical applications, using C++. It features real-time data acquisition and plotting.
The ISR-BMS is often referred simply as BMS during this section to avoid the prefix repetition.
3.3.1 Hardware
The hardware development workflow is represented in Figure 3.9. The main hardware blocks
of the BMS are explained in detail on the following subsections.
A complete Schematic diagram can be found in Appendix A.
28
3.3: ISR-BMS Development
Figure 3.9: Hardware development workflow.
3.3.1.1 The MAX14921 Analog Front End(AFE) Integrated Circuit
The ISR-BMS is built around an high accuracy 16-cell Analog Front End (AFE) IC, the
MAX14921 [Maxim, 2014] from MAXIM INTEGRATED. This IC is able to sample all the cell
voltages simultaneously and provides level shifting for battery packs up to 16 cells / 65V(max).
All cell voltages are level shifted to ground reference with unity gain simplifying external ADC
data conversion. The device includes a low-noise, low-offset amplifier that buffers differential
voltages up to +5V, allowing monitoring of all common Lithium-Ion cell technologies. The
resulting cell voltage error is ±0.5mV, what makes it ideal to monitor cell chemistries with very
flat discharge curves, such as Lithium Iron Phosphate (LiFePO4). A comparative analysis of
cell voltages of different chemistries can be seen in Figure 3.10.
Figure 3.10: Battery voltages with different cathode chemistries. [Electronicdesign, 2012]
29
The MAX14921 key features can be summarized as follows:
• High accuracy
– ±0.5mV(max) Cell Voltage;

– Simultaneous Cell Voltage Sampling;
– Self-Calibration.
• Integrated Diagnostics
– Open Wire and Short Fault Detection;

– Undervoltage/Overvoltage Warning;
– Thermal Shutdown.
• High Flexibility
– SPI Interface;
– Up to 16 Cell per IC;
– +6V Minimum (3 Cells) Operation;
– +0.5V to 4.5V Cell Voltage Range;
– Integrated Cell-Balancing FET Drivers.
• Low Power
– 1µA Shutdown mode;

– 1µA / 10µA Cell Current Draw.
In applications requiring more than 16 cells, multiple BMS modules can be stacked. Each
model can monitor any number of cells between 3 and 16 Cells. The stacked models can
communicate through SPI. To translate the inter-pack communication signals between different
common-mode pack voltages, digital isolators are required.
The performance of the system in which the MAX14921 is used depends highly on the
components that support and surround it [Georges, 2014].
The developed architecture requires high accuracy and resolution to accurately measure
very flat discharge voltage curve characteristics like is the case of Lithium Iron Phosphate cells.
Table 3.1 [Georges, 2014] shows a cost/performance comparison based on different possible
architectures using the MAX14921 AFE.
The chosen topology (accuracy optimized) is represented in Figure 3.11.
30
Table 3.1: Cost/performance comparison of possible alternative BMS architectures.

Relative Three-Sigma Six-Sigma
Architecture Circuit Topology
Cost Error Error
External ADC(16 bit), external
Accuracy-Optimized High 1.087mV 1.714mV
compensated voltage reference
Accuracy-Optimized, Medium- External ADC(16 bit) with internal voltage
5.014mV 7.305mV
Cost-Reduced High reference
Cost-Optimized, Medium- Microcontroller with integrated ADC(12
17.632mV 31.154mV
Accuracy-Enhanced Low bit) and external voltage reference
Microcontroller with integrated ADC(12
Cost-Optimized Low 161.849mV 251.307mV bit) and reference derived from a voltage
divider
Figure 3.11: ISR-BMS Architecture.
The key components can be identified on the circuit board developed, shown on Figure 3.12.
Figure 3.12: Circuit Board bottom side derived from Eagle Layout Software.
The MAX14921 integrates gate drivers to interface external mosfets enabling passive cell
balancing. Each cell mosfet enables balancing through a current shunt resistor. The balancing
current was set to about 200mA for a balancing voltage of 3.5V.
31
Figure 3.13: HTFS200 - Hall-effect current sensor from LEM manufacturer.
3.3.1.2 BMS Microcontroller
The microcontroller was chosen based on its available peripherals (SPI, UART, CAN),
simple 8-bit architecture, low power consumption and available program and data memory
(including its non volatile EEPROM).
The selected IC is a PIC18F2680 from Microchip Technology Inc. This is an enhanced flash
microcontrollers with ECAN™ Technology, 10-Bit A/D and nanoWatt Technology.
3.3.1.3 Current Sensor
Current measurements must be carried out with high accuracy as they represent the primary
input for most SOC estimation algorithms.
The battery pack installed on the EV is able to supply a nominal current of 270A (3C). The
traction motor on the platform is a 3.5kW DC Motor with separated excitation. In road tests,
peak powers of 11kW were observed. Considering an average voltage of 3.2V per cell (76.8V for
complete pack) this gives a peak current of 143A. This can change based on the battery pack
SOC.
To accommodate some margin also for the foreseen power train evolution from the current
DC motor to a permanent magnet DC motor with higher power, the current sensor rate was
set to 300A.
The selected sensor is a LEM HTFS200 based on the hall-effect measuring principle.(Figure
3.13 )
A typical accuracy target for current measurements ranges between 0.5 and 1%. For this
application with a current range of 150A this translates to an accuracy of 0,75 to 1.5A. Since
the average battery current consumption for this vehicle is about 30A this would lead to un-
acceptable errors and therefore the current measurement setup targets an accuracy of 300mA
over the whole range taking in account the linearity error, voltage offset error and temperature
coefficient errors from the sensor datasheet.
To accomplish this accuracy an initial current calibration procedure is run every time the
BMS is started that eliminates any voltage offset from sensor.
32
3.3.1.4 Temperature Sensor
The thermal management is an essential part of every BMS as temperature heavily affects
the performances of the battery. Parameters like internal resistance and capacity are tempera-
ture dependent.
For temperature measurements the chosen sensor was the 10kOhm NTC model ERTJ1VG103FA
from Panasonic with 1% tolerance and a measurement range of -40°C to 125°C. This resistor
varies non-linearly with temperature and this correlation is usually stored in the program
memory or other kind of non-volatile memory of the microcontroller. Another approach is to
linearize the curve for smaller temperature ranges were the error can be acceptable and use
a linear equation to let the microcontroller calculate the temperature based on the measured
voltage across the NTC as part of a voltage divider. Was followed the latest approach. And
accurately approximated the characteristic NTC curve for the temperature range of 0 to 40°C.
3.3.1.5 Communication Interface Hardware - USB
The BMS can communicate via USB, using a standard mini-USB connector.
An FT232RL IC from FTDI is used to convert USB protocol to serial UART. The commu-
nication protocol is presented on the software section.
The serial COM port should be configured for 31250kbps, 8 bits, no parity and 1 stop bit.
3.3.1.6 Communication Interface Hardware - CAN bus
In order to ensure that a BMS can take the maximum benefits from a battery charger,
a digital interface for setting instantaneous voltage and current is required. Most common
available EV battery chargers (e.g. TC charger) feature a CAN bus interface for that purpose
due to its robust characteristics.
The developed BMS also supports CAN bus interface for charging and balancing purposes.
It features a full CAN system, implementing the CAN 2.0A or B protocol as defined in the
BOSH specification.
For successful communication with the BMS the CAN bus must be configured with a bit
rate of 250kbps.
3.3.2 Firmware
3.3.2.1 Real Time Operating System - FreeRTOS
The BMS firmware is programmed with C language, using the MPLAB(v8.90) Integrated
Development Environment (IDE) and the C18 compiler from Microchip. For programing and
debug the Pickit3 development tool from the same manufacturer was used.
33
The BMS needs to manage multiple simultaneous processes like communication with CAN,
the serial communication via USB or the communication between the microcontroller and the
AFE and ADC via SPI.
To manage this complexity and the task scheduling the FreeRTOST M 1 was used.
This kernel was chosen based on the following advantages:
- Is free for use, even in commercial applications;
- Provides one solution for different architectures and development tools;
- Has a minimal ROM, RAM and processing overhead;
- Is very scalable, simple and relatively easy to use.
This system allows for multi-thread operation. In FreeRTOS each thread of execution is
called a ‘task’. Multi-tasking requires all tasks get scheduled to run on CPU according to some
pre-determined scheme.
Tasks are implemented as C functions. The only thing special about them is their prototype,
which must return void and take a void pointer parameter.
Each task is a small program in its own right. It has an entry point, will normally run
forever within an infinite loop, and will not exit.
The complete list of created tasks with a brief description is presented bellow:
1. vComRxTask - Serial communication management task.
2. vCellsReadTask - Cell parameters reading and validation task.
3. vCANRxTask - CAN bus communication management task.
4. vCANSendInfoToChargerTask - Charger management task.
5. vComserialHeartbeatTask - Heartbeat implementation task.
3.3.2.2 Cells Voltage Measurement Procedure
As the code is quite extensive, accounting for more than 5000 lines, the implementation
details are not going to be presented for all the tasks. As an example, in Figure 3.14 the cell
voltages measurement procedure flowchart using the MAX14921 is presented.
The procedure implements the scheduling required for reliable and accurate operation of
the AFE IC according to the manufacturer datasheet [Maxim, 2014] information. The relevant
timing diagram is reproduced in Figure 3.15 for easy understanding.
An illustration of the file structure with a brief description for each file, outputted from
Doxygen documentation generator tool, can be found at Appendix B.
1
FreeRTOST M is a real time operating system developed by Real Time Engineers Ltd.
34
Figure 3.14: MAX14921 cell voltages measurement procedure flowchart.
35
Figure 3.15: MAX14921 SPI Control of cells voltage readout.
3.3.2.3 Cell Balancing
The flowchart of the cell balancing procedure implemented in the ISR-BMS Firmware is
shown on Figure 3.16.
3.3.2.4 Communication Protocol for BMS Parametrization and Data Logging
To allow the exchange of information between the PC and the BMS via serial interface, a
communication protocol was developed and implemented. This protocol was designed to be
flexible and easily scalable.
With the exception of the HEARTBEAT command, the communication is always started by
the PC. All the commands and data exchanged are encapsulated between a START character
(@) and an END character (e). The syntax of the different commands is described on Table
3.2.
The ISR-BMS features also support to OBDII via CAN bus. In this way all the relevant
information from the BMS can be displayed on a tablet or smartphone with appropriated
compatible software.
An example of the codes implemented and the structure of the CAN messages are shown
on Table 3.3.
A screenshot of Torque software for Android implementing the readings of 7 cells plus the
temperatures on the cells and on the ISR-BMS is shown in Figure 3.17. To use this feature an
OBDII to bluetooth adapter based on ELM27 IC was used.
36
Figure 3.16: Balancing flowchart.
37
Table 3.2: ISR-BMS Communication Protocol
38
Table 3.3: OBDII commands implemented on ISR-BMS.
Figure 3.17: Smartphone display of ISR-BMS voltages and temperatures.
39
3.3.3 The BMS Graphical User Interface for PC

The BMS developed was designed to be flexible. It can be used with different types of
Lithium-Ion cells. The MAX14921 can operate with cells voltages in the range of 0.5V to
4.5V and a minimum of 3 Cells. Nevertheless the reference voltage of the ADC will limit the
maximum operating voltage of the ISR-BMS to 4.096V. Still the BMS is fully configurable from
3 to 16 cells per module (one master and one slave) and from 0.5 to 4.096V per cell .
It allows for real time cell voltages, current and temperature plotting and data logging to
a computer file. Also the cell balancing settings like the start balancing voltage or the allowed
delta voltage (the difference between the highest and the lowest cell voltages), can be configured
via this interface.
When multiple current sensors or temperature sensors are implemented on the BMS firmware
they can also be easily selected from the appropriated combo boxes in the GUI.
The application automatically detects the connection to the BMS and configures the port
settings for communication, when a standard USB cable is used to connect the PC with the
BMS. An HEARTBEAT command is sent every 0,5s to the PC and is used to toggle an LED
that gives a visual feedback of good communication link between the PC and the BMS.
A screenshot of the configuration screen is shown in Figure 3.18 and another one from real
time cell voltage plotting in Figure 3.19.
The GUI is documented at Appendix C.
Figure 3.18: ISR-BMS GUI parametrization menu.
40
Figure 3.19: ISR-BMS GUI live cell voltages plot.
41
Chapter 4
Battery Modeling
Users of EVs require precise and reliable information on their autonomy. This information
is of paramount importance to avoid range anxiety symptoms.
The SOC of the battery pack on an EV is comparable with the fuel gauge of an internal
combustion vehicle. The range estimation algorithms used are only as accurate as the data
provided by the BMS about the SOC of the battery pack. The SOC can not be measured
directly from the battery terminals. Therefore it needs to be estimated based on the measurable
battery parameters (individual cell voltages, current and temperature). This estimation is
important not only for optimal battery energy management but also to avoid an overcharge
or overdischarge scenario that can represent safety concerns or reduce battery life span. A
reliable and precise estimation of the remaining charge available allows the use of the battery
full capacity.
An accurate model representing the characteristics of the battery is essential for SOC es-
timation accuracy. As covered on the State-of-the-Art chapter, battery models range from
complex models that reproduce the electrochemical reactions inside the battery, requiring a
very high computational effort, to simple models like the coulomb counting that have very
good run-time performance but can be very imprecise due to error accumulation from current
measurements.
Due to the chosen architecture of the ISR-BMS, the choice of the model is conditioned
to the resources available mainly due to the RAM and ROM limitations. Therefore an offline
technique to model the cell was used, based on datasets made in Laboratory. The chosen model
is an Electrical Equivalent Circuit (EEC) [Huria et al., 2010]. For this purpose an automated
test setup was used, that allows to characterize the cell under different charge and discharge
conditions at controlled temperature. This test setup is briefly described at the end of this
chapter.
4.1 Modeling Method

The EEC models are a good trade-off between complexity and run-time performance and
allow to model certain important dynamics of the battery cells. Another important advantage
of those models is their ability to calculate both the terminal voltage and cell SOC. This model
can also be used in combination with coulomb counting to periodic calibration during rest,
based on the OCV/SOC derived correlation. By comparing the model terminal voltage and the
43
Chapter 4: Battery Modeling
Table 4.1: Overview of Applicable International Standards for Lithium-Ion cells or batteries
Standard Title
ISO 12405-1 Electrically propelled road vehicles–Test specification for
lithium-ion traction battery packs and systems–Part 1:
High-power applications
ISO 12405-2 Electrically propelled road vehicles–Test specification for
lithium-ion traction battery packs and systems–Part 2:
High-energy applications
IEC 62660-1 Secondary lithium-ion cells for the propulsion of electric road
vehicles
Part 1: Performance testing
IEC 62660-2 Secondary lithium-ion cells for the propulsion of electric road
vehicles
Part 2: Reliability and abuse testing
Figure 4.1: EEC model adopted for ISR-BMS
measured cell voltage the model error can be estimated. If enough data logging is provided and
computational resources are available, adaptive algorithms such as Extended Kalman Filter
(EKF) [Plett, 2004] can be used to improve the SOC accuracy.
The employed modeling method follows the applicable standards for Lithium-Ion cells and
battery packs whenever the bench test setup allows. The present test setup is limited to
temperatures between 0 and 45◦ C, charge currents up to 30A and discharge currents up to
20A. So the tests that require negative temperatures or higher charge or discharge rates can
not be performed with the present test setup. The relevant standards are summarized on Table
4.1.
Most models in the literature do not consider the thermal effects. It was adopted the model
presented by [Huria et al., 2010] that tackles this limitation by including the temperature as
an independent variable in the look-up tables that define the circuit elements.
The adopted model is represented in Figure 4.1.
This model, despite its simplicity can fit experimental data reasonably well for the tested
44
4.2: Description of Tests Performed
conditions. It is also suitable for embedded applications like the presented one. Of course it
can not model the different dynamics of a battery as it is limited by the use of a single RC
branch, so a single time constant.
The choice of this model requires the estimation of four independent parameters, namely
the electromotive force (Em), the series resistance (Ro), and the parallel RC branch (R1, C1),
which in turn, vary with the temperature and the SOC of the cell.
The SOC of a cell is typically presented as the remaining charge available on the cell. It is
function of the full cell capacity CQ at the temperature and discharge current considered and
the extracted charge Qe :
SOC = 1 − Qe /CQ (4.1)
Assuming the cell to be fully charged at time t=0 Qe can be defined as ,
´t
Qe (t) = 0
Im (τ )dτ (4.2)
It is usually given as a percentage of the full cell capacity. This capacity represents the
maximum extractable charge under certain conditions and depends on the following factors:
• The charge and discharge rates;
• The temperature;
• The value of the cut-off discharge voltage;
• The age of the cell.
Consequently, any SOC definition should consider the conditions under which a cell is discharged
and refer to the particular discharge current and temperature under which the SOC has been
evaluated.
4.2 Description of Tests Performed

The cells were preconditioned prior to any measurements for data fitting with the model.
The tests were performed on the model LFP090AHA from Winston manufacturer, a prismatic
90Ah of Lithium Iron Yttrium Phosphate (LiFeYPO4) chemistry.
The cell specifications are shown on Table 4.2.
45
Table 4.2: LFP090AHA cell specifications.
4.2.1 Pre-conditioning
This test is done only with new cells. It consist on some cycles of charge/discharge at room
temperature, aiming to reach a stable and replicable status of the cells.
This test is not foreseen on the IEC62660-1 but it is important to reach a stable electro-
chemical status with a predictable capacity. This procedure is foreseen on the ISO12405 and
applies to battery packs and systems.
Procedure:
• The test is performed at room temperature (25°C ± 2°C).
• The discharge is performed at C/3 until the minimum voltage recommended by the sup-
plier is reached.
• The charge is CC up to the maximum charging voltage and CV after that until the
charging current reduces to C/10.
The cell is considered “pre-conditioned” when the discharged capacity during two consecutive
discharges, does not change by a value greater than 3% of the rated capacity.
Due to the cell characterization setup the charge and discharge currents were reduced to
20A.
The charge and discharge voltage curve is shown in Figure 4.2. The cell was fully charged,
starting from its original SOC, and after that a full discharge-charge cycle was applied. The
charge cut-off voltage was set to 3.8V and the discharge cut-off voltage to 2.8V.
The cell capacity measured on the first discharge was 99,6Ah and on the second discharge
99,7Ah varying about 0,11% of the rated capacity. As a result the cells are considered pre-
conditioned.
46
4.2: Description of Tests Performed
Figure 4.2: Voltage plot of first pre-conditioning cycle.
4.2.2 Characterization Test Setup
To fully characterize the Lithium-Ion cells it is essential to measure their key parameters at
different temperatures as their performance is fully dependent on environment conditions.
There are several climatic chambers commercially available but their cost can be prohibitive,
are not easy to parametrize or do not have an appropriate interface for being controlled exter-
nally.
The required test setup needs as well a controllable DC current load to set the discharge
current on the cell and a CC/CV charger to fully characterize the charge/discharge cycle.
To achieve this goal an integrated climate chamber with temperature control and an elec-
tronic DC load were designed to be controlled by a microcontroller interfacing with a PC via
USB. In this way a rather precise, flexible and affordable setup was achieved.
The cell test setup was developped by a ISR-UC research team, and the author’s contribu-
tion in the test setup development consisted of the calibration of current measurements, the
implementation of remote charge control via PC with local cell voltage sensing and on the GUI
implementation (MATLAB) for test setup and data logging via the PC. This setup is shown in
Figure 4.3.
47
Figure 4.3: Cell characterization test setup.
Its main features are summarized on Table 4.3 .
Table 4.3: Test Setup Features

Measurements Overall Accuracy
Voltage 0.2%
Current 0.2%
Ambient Temperature ±0.25°C
Cell Temperature ±0.25°C
Sampling rate 4Hz
Max. Controlled Temperature 45°C
Min. Controlled Temperature 0°C
Max. Discharge Current 20A
The setup is composed by the following elements:
1. The charging system - driven by a 600W DC Power Supply from Delta Elektronika (Model
SM1540-D) fully controllable via analog input.
2. The discharge system - a controllable electronic DC load for voltages up to 5V and currents
up to 20A.
3. The temperature control system - The temperature is controlled with a 130 W thermo-
electric module. For temperature measurements the I2C MCP98908 digital temperature
sensor was used. The advantage of this sensor is that it is factory calibrated.
4. The measurement system - The hardware is composed by a microcontroller that interfaces

a I2C INA226 (DC accuracy: shunt voltage 2.5µV, bus voltage 1.25mV, bus voltage
range 36V). The setup interfaces with the PC via USB and allows for remote control of
temperature, voltage and current, producing charge/discharge profiles.
48
4.3: Cell Model Derivation
Figure 4.4: Characterization cell box GUI.
A screenshot of the GUI, illustrating the last 10% SOC discharge at 10°C under a constant
current of 20A, can be seen in Figure 4.4.
4.2.3 Characterization Tests

The cell under test was discharged in steps of 10% of SOC with a relaxation time of 2 hours
within discharges. The SOC was derived based on coulomb counting of the current drawn at
each step. The same discharge profile was run at 10, 25 and 45◦ C.
4.3 Cell Model Derivation

For calculating the model parameters based on available experimental data, the Matlab
Simulink GUI estimation tool, developped by Tarun Huria[Huria et al., 2010] was used to
obtain the 1 RC branch EEC parameters values lookup tables (10, 25 and 45°C).
This tool also incorporates the cell model created with SimscapeT M blocks and SimscapeT M
language allowing for simultaneous simulation of the model.
In Figure 4.5 the output of the estimation and simulation tool can be seen. The fitting
looks rather good for a model with a single time constant. The estimated model was able to
keep track of the OCV after every discharge pulse.
The experiment at the 3 different temperatures resulted in four sets of data for the 4 elements
of the EEC, that characterize the cell chemistry under test. The Figure 4.6 summarizes the
obtained data .
Some interesting results can be drawn from the experimental data:
1. The SOC-OCV model curve shows that OCV depends very little from the temperature
49
Figure 4.5: Measured versus simulated data at 25 ºC.
Figure 4.6: EEC estimated parameters.
50
4.3: Cell Model Derivation
having a strong dependance on SOC. Also the model shows a less consistent behavior at
SOC extremes.
2. Ro shows a dependance on SOC and temperature with a minimum value close to 50%
SOC for 25 and 45°C and lower values for higher temperatures. At 10°C it shows an
unexpected behavior that could not be confirmed due to the lack of time.
3. R1, the resistive component of the RC element, shows little SOC dependance on temper-
ature from 10 to 90%SOC.
4. C1, the capacitive element of the RC element does not show a predictable or consistent
behavior being difficult to attribute a significance to those values. That is one of the
drawbacks of equivalent numerical models compared to physical models that represent
the battery internal phenomena.
51
Chapter 5
BMS Experimental Results

In this Chapter several experimental tests made with the electric platform after the ISR-
BMS installation are described. The goal is to validate the measurements taken by the BMS
and evaluate its on-line performance as well as its ability to safely charge and balance the
battery pack.
5.1 Battery Pack Current Measurements

To validate the current measurements of the ISR-BMS, a simultaneous data log of the BMS
readings and a commercial current probe from Fluke (Model i310s) was performed. The current
probe used has a bandwidth of 1 kHz and was used with the selected sensitivity of 10 mA/V
featuring an accuracy of ±1% of the reading ±50mA. The current probe was connected to a
PicoScope (Model 4424) with 20 MHz bandwidth and was sampled at 10 Hz. That corresponds
to the internal sample rate of the ISR-BMS. Nevertheless the current log was made by the GUI
with a sample rate of 1HZ. The comparative result can be observed in Figure 5.1.
The results shows that the ISR-BMS data logged into the PC was not able to accurately
follow all the current peaks. This is due to the fact that the sample rate of 1 Hz is not high
enough to display the fast transients on the current. But since the internal sample rate is
indeed 10Hz the ISR-BMS “senses indeed” the current shown in red and do all its calculations
based on that information. This is very important for instance to compute the current integral
(coulomb counting) needed for the SOC estimation.
Further testing needs to be done, increasing the GUI sample rate to have a more represen-
tative information of the real currents on the PC. This can be important if this information is
needed to implement some kind of adaptive SOC estimation algorithm. If needed the ISR-BMS
internal sample rate can also be increased.
5.2 Cell Voltages Measurements

The individual cell voltages from the battery pack were also logged in the PC via the
developed software. The result can be seen in Figure 5.2.
This plot shows interesting information. As mentioned before the battery pack is composed
by two modules of 12 cells each connected in series. The cells 1 to 12 were used in a previous
project and the cells 13 to 24 were new cells acquired to complete the battery pack. The
53
Chapter 5: BMS Experimental Results
Figure 5.1: Comparative current measurements of ISR-BMS and Fluke i310s current probe.
Figure 5.2: Voltage measurements logged on the PC during a real road drive.
54
5.3: Battery Balancing
Figure 5.3: BMS debugging after installation.
difference between the two groups of cells is clearly seen on the plot where the used cells show
a higher voltage drop during discharge currents due to the increased internal resistance that
Lithium-Ion cells face with cycling and aging. It is also possible to see that during the relaxation
period or during the regenerative charging the terminal voltages converge to the same value.
This will of course bring some problems to the effective use of the capacity of the new cells
since the old cells will have a lower efficiency and need to spend more energy to supply the
needed power to the powertrain. So they will reach the end of charge threshold voltage or the
0% SOC before the new cells and no more energy can be drawn from the battery to avoid its
overdischarge that could seriously damage the cells.
5.3 Battery Balancing

In this specific case the passive cell balancing is not a perfect solution, as the used cells will
reach the start balancing voltage threshold in block and the BMS will turn on the respective
balancing resistor leading to a situation where 200mA will be drawn from each cell in order
to lower its charge and allow the other block to charge. This represents a dissipation of about
8.5W in the used BMS module that will be dissipated in the form of heat on the balancing
resistors that can be seen on the top of the ISR-BMS printed circuit board. This can be seen in
Figure 5.3. It is confirmed by the GUI voltages plot in Figure 5.4, where the cells are charging
and the terminal voltages of the used cells are reaching the balancing threshold voltage of 3.4
V in block.
The electric platform is equipped with a Robot Operating System (ROS) that allows an
55
Figure 5.4: Cell voltages during charge at 10A constant current.
easy acquisition and integration of data from different kinds of sensors. The vehicle is already
reading the data from GPS, inertial sensors and encoders placed on the wheels. Thanks to
João Sousa that created a new ROS node to read the relevant parameters from the BMS it was
possible to acquire a combined dataset with information of the ISR-BMS, GPS, and encoders
allowing to calculate the car speed and to know its position at each moment. Also the elevation
can be computed from the collected GPS data. This combined information can prove very
useful to implement range estimation algorithms for people doing research on this field. This
combined information is shown in Figure 5.5.
56
5.3: Battery Balancing
Figure 5.5: Combined information of ISR-BMS, GPS and encoders from the electric platform
during a real road drive.
57
5.4 ISR-BMS Benchmarking

The ISR-BMS was compared with two existing commercial BMSs, namelly the ORION
BMS from the American company Ewert Energy Systems, Inc and the EMUS BMS from the
Lithuanian company Elektromotus. The results are shown on Figure 5.6
Figure 5.6: Benchmarking the ISR-BMS with other commertial BMSs.
58
Chapter 6
Conclusion and future work
6.1 Conclusion
During this dissertation a lot of know-how was built around the Lithium-Ion batteries
behavior. A simple EEC model with one RC branch revealed a reasonably good capacity
to follow the cell dynamics but with limitations to represent the long time constants of the
relaxation periods.
The developed BMS hardware is fully functional and able to perform accurate measurements
of individual cell voltages, current and temperature. It can perform passive cell balancing for
the whole pack based on individual cell information, allowing this way to reduce the differences
on the SOC of the different cells optimizing in this way the useful battery pack capacity. It was
designed as well to allow the implementation of different charging, cell balancing algorithms and
SOC estimation techniques that are not possible to implement on typical commercial BMSs.
It took a long journey to reach this level of functionality and reliability. During the de-
velopment process, a lot of unpredictable situations occurred that had to be resolved in order
to progress. Problems like noise, ground loops, inconsistent communication within the BMS
blocks or between the BMS and the PC, bad contacts that lead to intermittent failures are just
a few of a long list. Most of those problems would not be present on simulations. They are
inherent problems of hardware without a generic solution. Mixed circuit boards with analog
and digital signals also place extra challenges. Some of the problems found, took days or even
weeks of intensive debugging to be overcome.
Due to the time limitation, it was not possible to implement and validate the cell model on
the ISR-BMS during this dissertation.
Working on a battery pack with a build up voltage of 80V and cells that can deliver hundreds
of Ampere during normal operation or even more during short-circuits is a dangerous work.
All the measures that can be taken to avoid any (unintentional) occurrence of a partial or total
short circuit should be taken.
6.2 Future work

As mentioned before this work will continue. In the short term the derived model should
be implemented and validated. This model have of course some limitations, therefore more
complex models can be used that can follow the different dynamics of the Lithium-Ion Cells.
59
Chapter 6: Conclusion and future work
Moving from a 1 RC branch to a 2 RC branch will increase the model accuracy but also the
complexity of estimating the model parameters and of implementing the model.
The 8-bit hardware architecture can also evolve to a 16-bit microcontroller architecture to
accomplish the higher processing power required to implement more complex models or higher
sampling rates.
Implementing active cell balancing techniques is also a field to explore.
60
Bibliography
[Antaloae et al., 2012] Antaloae, C., Marco, J., and Assadian, F. (2012). A novel method for
the parameterization of a li-ion cell model for ev / hev control applications. 61(9):3881–3892.
[AS8506, ] AS8506. Datasheet rev 1.Lithium-Ionbattery cell monitor and balancer IC. Available
online:http://www.ams.com/eng/content/download/476603/1402377/244331.
[BCG, 2010] BCG (2010). Batteries for electric cars. Challenges, Opportunities, and the Out-
look to 2020. The Boston Consulting Group.
[Bergveld, 2001] Bergveld, H. J. (2001). Battery Management Systems Design by Modelling.
[BQ76PL536, ] BQ76PL536. Datasheet rev 1.Lithium-Ion battery monitor and secondary pro-
tection IC. Available online:http://www.ti.com/lit/gpn/bq76pl536.
[Cao et al., 2008] Cao, J., Schofield, N., and Emadi, A. (2008). Battery balancing methods: A
comprehensive review. 2008 IEEE Vehicle Power and Propulsion Conference, pages 1–6.
Available from: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=
4677669.
[CARB, 2007] CARB (2007). Status and prospects for zero emissions vehicle technology. Cal-
ifornia Air Resources Board, Report of the ARB independent expert panel.
[Chen and Rinc, 2006] Chen, M. and Rinc, G. A. (2006). Accurate electrical battery model
capable of predicting runtime and i-v performance. 21(2):504–511.
[Coleman et al., 2007] Coleman, M., Lee, C. K., Zhu, C., Hurley, W. G., and Soc, A. S.-o.-c.
(2007). State-of-charge determination from emf voltage estimation : Using impedance ,
terminal voltage , and current for lead-acid and lithium-ion batteries. 54(5):2550–2557.
[Daowd et al., 2011] Daowd, M., Omar, N., Bossche, P. V. D., and Mierlo, J. V. (2011). A Re-
view of Passive and Active Battery Balancing based on MATLAB/Simulink. xx(September).
[Electronicdesign, 2012] Electronicdesign (2012). Anodes, cathodes, and

electrolytes. Available from: http://electronicdesign.com/power/
understanding-factors-lithium-battery-equation.
[Electropaedia, a] Electropaedia. Battery protection system. Available from: http://www.

mpoweruk.com/safety.htm.
[Electropaedia, b] Electropaedia. Temperature and discharge rate. Available from: http:

//www.mpoweruk.com/soc.htm.
61
BIBLIOGRAPHY
[Energy, 2012] Energy, E. (2012). Cost and performance of ev batteries. Final

report for The Committee on Climate Change. Available from: http://www.
element-energy.co.uk/wordpress/wp-content/uploads/2012/06/CCC-battery-cost_
-Element-Energy-report_March2012_Finalbis.pdf.
[Faria et al., 2013] Faria, R., Marques, P., Moura, P., Freire, F., Delgado, J., and de Almeida,
A. T. (2013). Impact of the electricity mix and use profile in the life-cycle assessment of
electric vehicles. Renewable and Sustainable Energy Reviews, 24:271–287. Available from:
http://linkinghub.elsevier.com/retrieve/pii/S1364032113002220.
[Georges, 2014] Georges, J. (2014). Getting the Most Out of the MAX14920/MAX14921 High-
Accuracy Battery-Measurement AFEs.
[Huet, 1998] Huet, F. (1998). A review of impedance measurements for determination of the
state-of-charge or state-of-health of secondary batteries. 70:59–69.
[Huria et al., 2010] Huria, T., Ceraolo, M., Gazzarri, J., and Jackey, R. (2010). High fidelity
electrical model with thermal dependence for characterization and simulation of high power
lithium battery cells.
[Kuhn et al., 2005] Kuhn, B., Pitel, G., and Krein, P. (2005). Electrical Properties and
Equalization of Lithium-Ion Cells in Automotive Applications. 2005 IEEE Vehicle Power
and Propulsion Conference, pages 55–59. Available from: http://ieeexplore.ieee.org/
lpdocs/epic03/wrapper.htm?arnumber=1554532.
[Maxim, 2014] Maxim (2014). Datasheet rev 1. High-accuracy 12-/16-cell measurement AFEs.
Available online: http://datasheets.maximintegrated.com/en/ds/max14920-max14921.pdf.
[Ng et al., 2009] Ng, K. S., Moo, C.-S., Chen, Y.-P., and Hsieh, Y.-C. (2009). Enhanced
coulomb counting method for estimating state-of-charge and state-of-health of lithium-ion
batteries. Applied Energy, 86(9):1506–1511. Available from: http://linkinghub.elsevier.
com/retrieve/pii/S0306261908003061.
[ORION, 2014] ORION (2014). Orion bms. Available from: http://www.orionbms.com.
[Parthiban et al., 2007] Parthiban, T., Ravi, R., and Kalaiselvi, N. (2007). Exploration of
artificial neural network [ANN] to predict the electrochemical characteristics of lithium-ion
cells. Electrochimica Acta, 53(4):1877–1882. Available from: http://linkinghub.elsevier.
com/retrieve/pii/S001346860701047X.
[Piller et al., 2001] Piller, S., Perrin, M., and Jossen, A. (2001). Methods for state-of-charge
determination and their applications. 96:113–120.
62
BIBLIOGRAPHY
[Plett, 2004] Plett, G. L. (2004). Extended kalman filtering for battery management systems of
lipb-based hev battery packs: Part 1. background. Journal of Power sources, 134(2):252–261.
[Rahimi-eichi, 2013] Rahimi-eichi, H. (2013). Battery management system - an overview of its

application in the smart grid and evs. (June):4–16.
[Rosenkranz et al., ] Rosenkranz, C. A., Liska, J. L., Controls-saft, J., Daney, B. A., and Bor-
deaux, F. Modern Battery Systems for Plug-In Hybrid Electric Vehicles.
[Salkind et al., 1999] Salkind, A. J., Fennie, C., Singh, P., Atwater, T., and Reisner, D. E.
(1999). Determination of state-of-charge and state-of-health of batteries by fuzzy logic
methodology. Journal of Power Sources, 80(1-2):293–300. Available from: http://
linkinghub.elsevier.com/retrieve/pii/S0378775399000798.
[Schweighofer et al., 2003] Schweighofer, B., Raab, K. M., Brasseur, G., and Member, S.
(2003). Modeling of high power automotive batteries by the use of an automated test system.
52(4):1087–1091.
[Silva et al., 2012] Silva, M., Moita, F., Nunes, U., Garrote, L., Faria, H., and Ruivo, J.
(2012). ISRobotCar: The autonomous electric vehicle project. 2012 IEEE/RSJ Inter-
national Conference on Intelligent Robots and Systems, pages 4233–4234. Available from:
http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=6386292.
[Tao et al., 2011] Tao, H., Feng, Z., Liu, H., Kan, X., and Chen, P. (2011). Reality and Future
of Rechargeable Lithium Batteries. (1):204–214.
[Tarascon and Armand, 2001] Tarascon, J. and Armand, M. (2001). Issues and challenges
facing rechargeable lithium batteries. Nature 414, 359-367.
[Tie and Tan, 2013] Tie, S. F. and Tan, C. W. (2013). A review of energy sources and
energy management system in electric vehicles. Renewable and Sustainable Energy Re-
views, 20:82–102. Available from: http://linkinghub.elsevier.com/retrieve/pii/
S1364032112006910.
[Wikipedia, ] Wikipedia. History of the electric vehicle. Available from: http://en.

wikipedia.org/wiki/History_of_the_electric_vehicle.
[Xing et al., 2011] Xing, Y., Ma, E. W. M., Tsui, K. L., and Pecht, M. (2011). Battery man-
agement systems in electric and hybrid vehicles. Energies, 4(12):1840–1857. Available from:
http://www.mdpi.com/1996-1073/4/11/1840/.
63
Appendix
64
Appendix A
BMS Schematics
66
1 2 3 4 5 6 7 8
+5V/1 X5
V+
C34 GND/2.4E 1
1uF GND/2.4E 2
IC1 VC1/2.4E 3
VC2/2.4D 4
1uF 1 16 GND
VCC1 VCC2 VC3/2.4D 5
C35 2 15
GND1 GND2 6
AGND
3 14
SCLK1 INA OUTA SCLK 7
4 13
+12V MOSI1 INB OUTB MOSI 8
5 12
A /CS1I INC OUTC /CS VC4/2.4C 9 A
6 11
IC4 MISO1 OUTD IND MISO VC5/2.4B 10
+5V/1 7 10
EN1 EN2 VC6/2.4B 11
8 9
GND1 GND2 VC7/2.4A 12
1uF
1 3
VP REG
2 ISO7241
GND_2
AGND
4
GND
C36
GND
C37
C49
TDA3664/N1
10uF
X6
100nF
+5V/1
V+
C47 VC8/2.2E 1
AGND
1uF VC9/2.2E 2
AGND
AGND
IC11 VC10/2.2D 3
VC11/2.2D 4
1uF 1 16 GND
VCC1 VCC2 5
C48 2 15
SV7
GND1 GND2 6
AGND
3 14
1
2
3
4
5
6
/CS-ADC1I INA OUTA /CS-ADC1 VC12/2.2C 7
+5V/1 4 13
/CS-ADC2I INB OUTB /CS-ADC2 VC13/2.2B 8
V+
5 12
/CS2I INC OUTC /CS2 R54 10R VC14/2.2B 9
C30 1uF 6 11
OUTD IND VC16P VC15/2.2A 10
1uF
7 10
B EN1 EN2 VC16/2.2A 11 B
10k-N.P.
C24
D20
D1
8 9
GND1 GND2 VC16P 12
AGND
OUT1OD ISO7241
AGND
DIODE-ZENERSMB
2 GND
C20
S
U1 GND
OUT1
100nF 100V
5
12 R86
13
14
2N7002BK 1u 16V X7R GND GND GND
CANRX_I
CANTX_I
OUT2
R1
1
V+
C1 17
VC16
15 100R VC16
CT16 R2 VC16/2.2A
VL
VA
100R 3 18 19
VP
R53 D CB16 VC15 100R VC15/2.2A X7
X1
G
IC5 C2 23 R3 100R
CHARGER
MOSI1
MISO1
1u VC14 R4 VC14/2.2B /CS2 1
23 14 21 27 100R
LDOWN
+VIN +VOUT CT15 VC13 2
10uF
VC13/2.2B /CS-ADC2
LED20
R5
1uF
1uF
C46 22 22 31 100R
CB15 VC12 R6 VC12/2.2C MISO 3
C33 +12V
AGND
AGND
3 C3 35 100R
KEY
OUT1
VC11 VC11/2.2D MOSI 4
1k
10uF 2 -VOUT 16 1u 25 39 R7 100R
-VIN CT14 VC10 R8 VC10/2.2D SCLK 5
DC/DC CONVERTER 26 43
C44
100R
C23
CB14 VC9 R9 VC9/2.2E 6
47
28
27
26
25
24
23
22
21
20
8
18
17
16
15
JCA0212S05-DC/DC- C4 VC8 100R VC8/2.2E SENSE_C 7
GND GND GND 1u 29 51 R10 100R
CT13 VC7 R11 VC7/2.4A T2 8
R87
AGND
V+
30 55 100R
CB13 VC6 R12 VC6/2.4B 9
+5V/1 R76 10k C5 59 100R
VC5 VC5/2.4B 10
RB4
VSS
VDD
GND 1u 33 63 R13 100R
CANTX_I AGND CT12 VC4 R14 VC4/2.4C 11
34 67 100R
C CB12 VC3 VC3/2.4D 12 C
RB1/INT1
RB0/INT0
RB7/PGD
RB6/PGC
RC5/SDO
RB5/PGM
1 8 C6 71 R15 100R
RC6/TX/CK
RC7/RX/DT
AOUT
TXD RS VC2 VC2/2.4D
RB3/CANRX
1u
V+
2 7 37 75 R16 100R
RC4/SDI/SDA
VSS CANH CT11 VC1 VC1/2.4E
R81
R78
R80
3 6 5 1 38 78 GND
VDD CANL V+ VOUT CB11 VC0 GND/2.4E
RB2/CANTX/INT2
4
RXD VREF
5 10 R89 C7
1uF
100nF
3 1u 41 16 GND
VO +IN CT10 BA16 BA16/2.1A
C50
C54
4 42 20
-IN CB10 BA15
0R
CANRX_I BA15/2.1B
C8 24
BA14 BA14/2.1B
120R
9 5 R77 0R 2 1u 45 IC9 28
V- CT9 BA13 BA13/2.1C
8 4 46 MAX14921 32
1nF
C53
CB9 BA12 BA12/2.1C
PIC18F2680-I/SO
AGND
7 3 GND IC3 C9 36
BA11 BA11/2.1D
AGND
6 2 R79 0 GND 1u 49 40
OPA365AIDBVTG4 CT8 BA10 BA10/2.1E
V+
1 50 44
~MCLR/VPP
RA0/AN0/CVREF
RA1/AN1
RA2/AN2/VREF-
RA3/AN3/VREF+
RA4/T0CKI
RA5/AN4/~SS/LVDIN
VSS_2
OSC1/CLKI
OSC2/CLKO/RA6
RC0/T1OSO/T1CK1
RC1/T1OSI
RC2/CCP1
RC3/SCK/SCL
C10 CB8 BA9 BA9/2.1E
IC6
X4 48
1u BA8 BA8/2.3A AOUT
53 52
1
2
3
4
5
6
7
9
SCLK1
IC8 CT7 BA7 BA7/2.3B
11
19
10
12
13
14
54 56
PWM X3 C28 CB7 BA6 BA6/2.3B SENSE_C
V+
C11 BA5 60
1
LED21
R82 KEY 1uF BA5/2.3C
Q2 3k9 8 DOUT 6 1u 57
BA4 64
CHARGER 2 VDD MISO CT6 BA4/2.3C T2
58 68
3 CB6 BA3 BA3/2.3D
100
C12
R17 1M
3 72
/CS1I
OUT1OD 4 GND -IN BA2 BA2/2.3E T1
LED22
1u
R500 10k
R51 10k
PWM
D 2 61 76 D
OUT2 5 AOUT +IN CT5 BA1 BA1/2.3E
1k
62
/CS-ADC2I
SCLK1 6 /CS-ADC1 CB5
LED19
Figure A.1: BMS-ISR Schematics 1/2

5 C13 7
1uF
C51
C29
100pF
MOSI1 7 ~CS/SHDN T3
R55
1k
C31
C32
100pF
7 1u 65 8
/CS2I
22pF
22pF
MISO1 8 SCLK DCLOCK CT4 T2
R83 1k 9
1
REF GND 66
CB4 T1
9
/CS1I C14
R90
AGND
/CS-ADC1I 10 ADC_REF 1u VO
AGND
+5V/1 11
4
GND
69
CT3 VO
10 R73 220R
C40
IC7 F1
10nF
R91
70
12 CB3
100n
10nF
NANO2SMF_R45201.5L
+12V
AGND 20 1 GND C15 3
VBUS VCC TXD MISO MISO
4 5 GND ADS8326IBDGKT 1u 73 2
VCCIO RXD CT2 MOSI
R85
MOSI
C42
C19
D2
D3
V+
3 74 1
1nF
/CS-ADC1I
RTS CB2 SCLK SCLK
DIODE-ZENERSMB
R84 10k
R88 10k
MTN4
MTN2
C17
C18
X2 19
RESET CTS
11 C16 /CS
80
NTC10k
AGND AGND 1u /CS

GND 2 IC10 77 GND
DTR CT1
10k
10k
V+
ID 27 9 79 GND GND GND GND GND GND
OSCI DSR EN
D+ 28 10 4 1
DGND
AGND
D+ OSCO DCD VBUS SMPL

D- 6 2 7 2
D- RI VIN VOUT ADC_REF
USB
6
1uF
VBUS 6 3
11
VOUTS
CBUS0
23 R74 1k 5 GND 4
IC1
R52
VBUS
R72
22 8 5
CBUS1 IC2
10uF
LED17 SAM_AFE
1uF
17 13 6
C22
CBUS2
MTN1
MTN3
3V3OUT EN_AFE ISR
E C39 100nF CBUS3
14 R75 1k GND GND E
15 12
LED18 1
LOGO
GND SV6
C38
D+ USBDP CBUS4 NR
C45
4u7
100nF
AGND
16
C41
C43
D- USBDM
26 4
TEST GNDS
25 3 GND GND
GND GND
AGND
AGND
7
GND
C21
18 BMS-Battery Management System
GND
100nF
21
GND
AGND
GNDGND MAX6126AASA41+ T3 is used for current Sensing with LEM HAIS50-P
FT232RL T1, T2 used for temperature measurement with 10k NTC JM
Galvanic Isolation
AGND
1 2 3 4 5 6 7 8
67
68
1 2 3 4 5 6 7 8
VC16/1.8B VC8/1.8B
LED1
18R
18R
LED9
R18
R56
R26
R64
A A
TLLG4400
TLLG4400
D 1k D 1k
R34 100R T4 IPD060N03 R42 100R T11 IPD060N03

BA16/1.7C BA8/1.7D
G S G S
VC15/1.8B VC7/1.8A
LED2
18R
18R
LED10
R19
R63
R27
R65
TLLG4400
TLLG4400
D 1k D 1k
R35 100R T3 IPD060N03 R43 100R T12 IPD060N03

BA15/1.7C BA7/1.7D
G S G S
VC14/1.8B VC6/1.8A
18R
LED3
18R
B B
LED11
R20
R62
R28
R66
TLLG4400
TLLG4400
D 1k D 1k
R36 100R T5 IPD060N03 R44 100R T13 IPD060N03

BA14/1.7C BA6/1.7D
G S G S
VC13/1.8B VC5/1.8A
18R
LED4
18R
LED12
R21
R61
R29
R67
TLLG4400
TLLG4400
D 1k D 1k
R37 100R T6 IPD060N03 R45 100R T14 IPD060N03

BA13/1.7C BA5/1.7D
G S G S
VC12/1.8B VC4/1.8A
C C
18R
LED5
18R
LED13
R22
R60
R30
R68
TLLG4400
TLLG4400
D 1k D 1k
R38 100R T7 IPD060N03 R46 100R T15 IPD060N03

BA12/1.7C BA4/1.7D
G S G S
VC11/1.8B VC3/1.8A
18R
18R
LED6
LED14
R23
R31
R59 R69
TLLG4400
TLLG4400
D 1k D 1k
R39 100R T8 IPD060N03 R47 100R T16 IPD060N03

BA11/1.7D BA3/1.7D
D G S G S D
VC10/1.8B VC2/1.8A
Figure A.2: BMS-ISR Schematics 2/2

18R
18R
LED7
LED15
R24
R32
R58 R70
TLLG4400
TLLG4400
D 1k D 1k
R40 100R T9 IPD060N03 R48 100R T17 IPD060N03

BA10/1.7D BA2/1.7D
G S G S
VC9/1.8B VC1/1.8A
18R
18R
LED8
LED16
R25
R57 R33 R71
TLLG4400
TLLG4400
D 1k D 1k
E R41 100R T10 IPD060N03 R49 100R T18 IPD060N03 E
BA9/1.7D BA1/1.7D
G S G S
VC8/1.8B
Balancing
GND
1 2 3 4 5 6 7 8
Chapter A: BMS Schematics
Appendix B
Firmware File Structure
Figure B.1: Firmware file structure (by Doxygen)
69
Appendix C
BMS-ISR Graphical User Interface (GUI)
Figure C.1: ISR-BMS GUI.
70
71
Chapter D: Winston Battery Datasheet
Appendix D
Winston Battery Datasheet
Figure D.1: Winston battery LYP90AHA.

72

Battery Management System (BMS) For Lithium-Ion Batteries

Uploaded by

Copyright:

Available Formats

Battery Management System (BMS) For Lithium-Ion Batteries

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Battery Management System (BMS) For Lithium-Ion Batteries

Uploaded by

Copyright:

Available Formats

Imagem

José Miguel Branco Marques

Battery Management System (BMS)

Battery Management System (BMS)

José Miguel Branco Marques

Submitted in partial fulfillment of the requirements for the degree of

Under supervision of:

List of Tables xiii

2 State of the Art 5

3 ISR-Battery Management System 23

3.3 ISR-BMS Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

5 BMS Experimental Results 53

6 Conclusion and future work 59

B Firmware File Structure 69

C BMS-ISR Graphical User Interface (GUI) 70

D Winston Battery Datasheet 71

2.1 Individual Cell and 8-Cell battery pack. . . . . . . . . . . . . . . . . . . . . . . . 6

3.1 ISR-UC Electric Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.17 Smartphone display of ISR-BMS voltages and temperatures. . . . . . . . . . . . 39

4.1 EEC model adopted for ISR-BMS . . . . . . . . . . . . . . . . . . . . . . . . . . 44

A.1 BMS-ISR Schematics 1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

B.1 Firmware file structure (by Doxygen) . . . . . . . . . . . . . . . . . . . . . . . . 69

C.1 ISR-BMS GUI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

D.1 Winston battery LYP90AHA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

2.1 Comparative analysis of commercially available BMS ICs. . . . . . . . . . . . . . 18

3.1 Cost/performance comparison of possible alternative BMS architectures. . . . . 31

4.1 Overview of Applicable International Standards for Lithium-Ion cells or batteries 44

BEV Battery Electric Vehicle

BMS Battery Management System

CAN Controller Area Network

CC/CV Constant Current/Constant Voltage

EEC Equivalent Electrical Circuit

EKF Extended Kalman Filter

ESS Energy Storage System

FET Field Effect Transistor

GHG Greenhouse Gas

GUI Graphical User Interface

HEV Hibrid Electric Vehicle

I2C Inter-Integrated Circuit

ICE Internal Combustion Engine

ISR Institute of Systems and Robotics

ISR-BSM New BMS develloped for ISRobotCar

OBDII On Board Diagnostic System

OEM Original Equipment Manufacturer

SAR Successive Approximation Register

SOC State Of Charge

SOH State of Health

SOL State of Life

SPI Serial Peripheral Interface

1.1 Motivation and context

3. Install a fully functional prototype on a ISR EV;

4. Develop the required experimental testing to validate the concept;

5. Benchmark existing commercial BMSs in comparison with the ISR-BMS.

1.3 Implementations and key contributions

1. Installation of a commercial BMS on a 72V, 40Ah, 3kWh Lithium-Ion battery pack in

2. Development of a new BMS architecture, including hardware, software and Graphical

1.4 Structure of the dissertation

State of the Art

2.1 Battery Introduction