Modeling the Cost and Performance of
Lithium-Ion Batteries for Electric-Drive
Vehicles
Final Report
&EPA
United States
Environmental Protection
Agency
-------
Modeling the Cost and Performance of
Lithium-Ion Batteries for Electric-Drive
Vehicles
Final Report
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
Prepared for EPA by
Argonne National Laboratory
Chemical Sciences and Engineering Division
Contract No. DE-AC02-06CH11357
Submitted July 15, 2011
NOTICE
This technical report does not necessarily represent final EPA decisions or
positions. It is intended to present technical analysis of issues using data
that are currently available. The purpose in the release of such reports is to
facilitate the exchange of technical information and to inform the public of
technical developments.
&EPA
United States
Environmental Protection
Agency
EPA-420-R-12-023
August 2012
-------
TABLE OF CONTENTS
LIST OF FIGURES vii
LIST OF TABLES ix
ABBREVIATIONS x
LIST OF SYMBOLS xii
ACKNOWLEDGEMENTS xv
EXECUTIVE SUMMARY xvi
PREFACE TO THE FINAL REPORT xvii
1. Introduction 1
2. Battery and Cell Design Format 3
2.1 Cell Design 4
2.2 Module Design 5
2.3 Battery Pack Design 6
3. Modeling of Battery Design and Performance 9
3.1 Criteria for Power, Energy, and Life 9
3.2 Voltage at Maximum Power 11
3.3 Governing Equations 17
3.4 Calculation of the ASI 18
3.4.1 Current Collection Resistance 20
3.4.2 Potential and Current Distribution 22
3.4.3 Determination of Module Terminal Size 24
3.5 Calculation of Battery Dimensions 25
3.5.1 Cell Dimensions 25
iii
-------
3.5.2 Module Dimensions 26
3.5.3 Battery Pack Dimensions 26
3.6 Additional Considerations 26
3.6.1 Maximum Electrode Thickness 27
3.6.2 Accounting for Parallel Cell Arrangements 33
3.6.3 Accounting for Parallel Module Arrangements 33
4. Thermal Management 34
4.1 Heat Generation Rates in the Battery Pack during Driving 34
4.2 Heating under Adiabatic Conditions 35
4.3 Active Cooling Systems 35
4.3.1 Heat Transfer from Cell to Module Wall 36
4.3.2 Heat Transfer from Module Wall to Flowing Coolant 38
4.4 Cooling and Heating Required to Maintain Pack Temperature 42
4.5 Heat-up from Cold Ambient Conditions 43
5. Modeling of Battery Pack Manufacturing Cost 44
5.1 Approach 44
5.2 Materials Costs and Purchased Items 45
5.2.1 Battery Specific Materials Cost 45
5.2.2 Purchased Items Cost 50
5.2.3 Pack Integration Cost 50
5.3 Baseline Manufacturing Plant 53
5.3.1 Receiving and Shipping 54
5.3.2 Electrode Materials Preparation 56
IV
-------
5.3.3 Electrode Coating 56
5.3.4 Calendering 57
5.3.5 Inter-Process Materials Handling 57
5.3.6 Electrode Slitting 58
5.3.7 Final Electrode Drying 58
5.3.8 Control Laboratory 58
5.3.9 Cell Stacking 59
5.3.10 Current Collector Welding 59
5.3.11 Enclosing Cell in Container 59
5.3.12 Electrolyte Filling and Cell Sealing 60
5.3.13 Dry Room Management 60
5.3.14 Formation Cycling, Final Cell Sealing, etc 60
5.3.15 Module and Battery Assembly 61
5.3.16 Rejected Cell and Scrap Recycle 62
5.3.17 Baseline Plant Summary 63
5.4 Adjustment of Costs for Rates 63
5.5 Plant Investment Costs 66
5.6 Unit Costs for Battery Pack 66
5.6.1 Variable Costs 66
5.6.2 Fixed Expenses 67
5.6.3 Profits 68
5.6.4 Battery Pack Warranty Costs 68
5.7 Summary of Baseline Battery Cost 68
-------
6. Description of Spreadsheet Model and Instructions for Use 71
6.1 Background 71
6.2 Instructions 71
6.2.1 Enabling Calculation 71
6.2.2 System Selection Worksheet 73
6.2.3 Battery Design Worksheet 73
6.2.4 Remaining Worksheets 77
6.3 Battery Design Format Requirements 80
6.4 Troubleshooting and General Advice 80
6.5 Suggested Number of Cells, Modules, and Performance Inputs 80
6.6 Entering a New Material Couple 81
7. Illustrated Results 83
7.1 Number of Cells in Series 83
7.2 Cathode Materials 84
7.3 Parallel-Connected Cell Groups and Electrode Thickness 84
7.4 Manufacturing Scale 86
8. Future Work 89
8.1 Initial Power Designed at Differing Fractions of the Open-Circuit Voltage...89
8.2 Optimum Battery Voltage for Minimum Drivetrain Cost 89
8.3 Multipurpose Battery Manufacturing Plants 90
8.4 Stand-Alone Graphical User Interface for Model 90
References 91
VI
-------
LIST OF FIGURES
2.1 Prismatic cell and module design for battery packs 3
2.2 Cell sandwich inside of prismatic pouch cells 4
2.3 Coated current collector foil for prismatic electrodes 5
2.4 Hermetically-sealed module 6
2.5 Insulated battery jacket with enclosed modules that are cooled on their upper and lower
surfaces by ethylene glycol-water solution 7
3.1. Summary flow of the design model 10
3.2 a) Required change in [V/U] to maintain rated power with increases in internal
resistance over the life of the battery, b) Increase in current due to lowered [V/U] 13
3.3 Change in heat rejection requirement from increases in resistance for batteries with
different designed voltages at maximum power 14
3.4 Efficiencies for batteries designed to achieve maximum power at different fractions of
their open-circuit voltage 16
3.5 The change in current and potential within the positive and negative foils. The current
collection design results in a uniform current distribution along the length of the foil ....23
3.6 Cell capacity simulated at the C/l and C/3 rate as a function of electrode thickness
(loading) for NCA-Gr 29
3.7 Normalized electrolyte salt concentration at the end of discharge at the C/l and C/3
discharge rates 29
3.8 Calculated ASI from a simulated 10-s, 5C discharge pulse for the NCA-Gr cell couple at
60%SOC 30
3.9 The potential of the negative electrode versus a hypothetical lithium reference electrode
located in the center of separator during a 5C charge pulse for the NCA-Gr couple 31
4.1 Plot comparing the estimated resistance to heat transfer from the cell center to the
cooled surface of the module to that calculated by the FlexPDE model 38
4.2 Heat transfer from the module wall to the laminar flow heat transfer fluid. The
temperature profile of the fluid is shown at different lengths down the path 39
Vll
-------
4.3 Temperature profile in the heat transfer fluid for various fractions of the dimensionless
path length 41
4.4 Correlation of model simulation results relating the Graetz number and mean Nusselt
number for laminar flow between an insulated surface and the module casing 41
5.1 Metal ingot cost contribution to the current collector foils over a 20 year period 49
5.2 Baseline lithium-ion battery manufacturing plant schematic diagram 54
5.3 Breakdown of installed capital equipment costs for the baseline plant 65
5.4 Breakdown of unit costs for baseline battery with total price to OEM of $2428 70
6.1 Iteration must be enabled for the spreadsheet model to function 72
6.2 The specific cell chemistry for the battery design is selected on the System Selection
worksheet 73
6.3 System Selection worksheet 74
6.4 Top portion of Battery Design worksheet 75
6.5 Middle portion of Battery Design worksheet 76
6.6 Bottom portion of Battery Design worksheet 78
6.7 Summary of Results worksheet 79
7.1 The effect of the number of cells for NMC441-Gr, 60-kW, PHEV25 packs with 10.7
kWh total energy (70% useable) 83
7.2 Mass and volume of electric vehicle battery packs with lithium iron phosphate (LFP),
lithium manganese-spinel (LMO) and lithium nickel-manganese-cobalt oxide (NMC441)
positive electrodes versus graphite designed to deliver 150 kW of power at 360 V (25%
SOC) 85
7.3 Battery pack price to OEM for LFP-Gr, LMO-Gr and NMC441-Gr battery packs for
same designs as in Fig. 7.2. NMC441-Gr and LMO-Gr result in nearly the same price...85
7.4 Battery pack cost as a function of number of parallel cells and for different maximum
electrode thicknesses 87
7.5 The effects of manufacturing rate on the price calculated by the model for battery packs
of various cell chemistries, power capabilities and vehicle types 88
Vlll
-------
LIST OF TABLES
3.1 Criteria for designing batteries for a specific end-use application 9
3.2 The effect of electrode loading on the price of a 17 kWh NCA-Gr PHEV40
battery with 96 cells 32
4.1 Sample calculations of composite thermal conductivities of cell structures
across layer and parallel to layers 37
4.2 Range of parameter values for calculating heat transfer rates in FlexPDE model 37
5.1 Details of stated costs for cathodes, anodes, electrolyte, and separator 46
5.2 Cost equations for purchased items 50
5.3 Costs to integrate battery pack into vehicle drivetrain 51
5.4 Summary table of the baseline plant 55
5.5 Materials yields during electrode and cell fabrication 62
5.6 The effect of processing rate (R) on cost for various scale factors 64
5.7 Battery pack manufacturing investment costs 66
5.8 Unit cost of battery pack 67
5.9 Summary of results for cost of baseline battery and that of similar batteries with
double the power and double the capacity of the baseline battery 69
6.1 General suggestions for range of input parameters that change with battery type 81
IX
-------
ABBREVIATIONS
AST area specific impedance
BOL beginning of life
DMC dimethyl carbonate
EC ethylene carbonate
EMC ethyl methyl carbonate
EOL end of life
EV electric vehicle
Gr graphite
GSA General, Sales, and Administration
HEV hybrid electric vehicle
HEV-HP high-power assist hybrid electric vehicle
LCO lithium cobalt oxide
LFP lithium iron phosphate
Li lithium
Li-ion lithium-ion
LMO lithium manganese spinel
LMR lithium and manganese rich
LTO lithium titanate spinel
microHEV micro or mild power assist hybrid electric vehicle
MW molecular weight
NCA lithium nickel cobalt aluminum oxide
NMC lithium nickel manganese cobalt oxide
-------
NMP N-Methyl-2-pyrrolidone
OCV open-circuit voltage
OEM original equipment manufacturer
PE polyethylene
PET polyethylene terephthalate
PHEV plug-in hybrid electric vehicle
PP polypropylene
R&D research and development
SOC state of charge
USGS United States Geological Survey
XI
-------
LIST OF SYMBOLS
Section 3, 4 and 6
a
ASI,
energy
ASI,
power
c
du
E
E
F
G
h
Hj
H/W
T lor
him
In
'total
k
ratio of interfacial area to electrode volume, cm"1
area of the positive electrode, cm2
area of the terminal, cm2
area specific impedance for energy, ohm cm2
area specific impedance for power, ohm cm2
cell capacity, Ah.
parameter
specific heat capacity, J/g K
hydraulic radius, cm
total energy, Wh
energy usage rate, Wh/mile
Faraday constant, 96485.3 C/mol
mass flowrate, g/s
heat transfer coefficient, W/cm2 K
height of j, cm
aspect ratio of pouch cell
exchange current density related to the interfacial area, A/cm2
average current density, A/cm2
ionic limiting current density, A/cm2
local current density, A/cm2
total current density, A
thermal conductivity, W/cm K
Xll
-------
Ij length of j, cm
LJ thickness of j, cm
m.j mass of j, g
n parameter
Nj number of j
[N/P] negative to positive capacity ratio
P battery power, W
Pbatt maximum designed battery power, W
q heating rate, W
Q specific capacity of the electrode, mAh/g
rc C-rate, h"1
rc,iim limiting C-rate, h"1
TJ radius of j, cm
R universal gas constant, 8.3144 J/mol K
Rj resistance of j, ohm
t time, s
T temperature, K
U0cv,p open-circuit voltage at SOC for power, V
UOCV.E open-circuit voltage at SOC for energy, V
v square root of dimensionless exchange current
Vceii cell voltage, V
[V/U] fraction of the open-circuit voltage
Wj width of j
Xlll
-------
X
y
•^
a
P
Pj
Section 5.1
C
C0
Xi
MWi
MW
Section 5.4
C
C0
P
R
Ro
Cartesian coordinate, cm
compression factor
constant, ohm cm3
constant, ohm cm2
volume fraction of active material
first cycle efficiency
fluid viscosity, g/cm s
metal potential of foil k, V
density of j, g/cm3
conductivity of j, S/cm
final cost of lithiated oxide, $/kg
cost of raw material for component i, $/kg
baseline cost of lithiated oxide, $/kg
molar stoichiometry of component i
molecular weight of component i, g/mol
molecular weight of lithiated compound, g/mol
capital cost of an installed equipment for the designed battery, $
capital cost of an installed equipment for the baseline plant battery, $
power factor
designed battery processing rate for specific process step
baseline plant processing rate for specific process step
XIV
-------
ACKNOWLEDGEMENTS
Support from the Vehicle Technologies Program, Hybrid and Electric Systems, initially under
Tien Duong and now David Howell, at the U.S. Department of Energy, Office of Energy
Efficiency and Renewable Energy, is gratefully acknowledged. The submitted manuscript has
been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory
("Argonne"). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated
under contract no. DE-AC02-06CH11357. The U.S. Government retains for itself, and others
acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to
reproduce, prepare derivative works, distribute copies to the public, and perform publicly and
display publicly, by or on behalf of the Government. We especially thank Danilo Santini of
Argonne's Transportation R&D Center for his support and suggestions in carrying out this study.
Ralph Brodd reviewed our baseline plant and made several suggestions which we have
incorporated in the present design. Fritz Kalhammer and Haresh Kamath of the Electric Power
Research Institute have reviewed our work over several years and made suggestions that resulted
in improvements. The work was done under the direction of Dennis Dees and Gary Henriksen of
Electrochemical Energy Storage who provided guidance in carrying out the work and preparing
this manuscript.
XV
-------
EXECUTIVE SUMMARY
This report details the Battery Performance and Cost model (BatPaC) developed at Argonne
National Laboratory for lithium-ion battery packs used in automotive transportation. The model
designs the battery for a specified power, energy, and type of vehicle battery. The cost of the
designed battery is then calculated by accounting for every step in the lithium-ion battery
manufacturing process. The assumed annual production level directly affects each process step.
The total cost to the original equipment manufacturer calculated by the model includes the
materials, manufacturing, and warranty costs for a battery produced in the year 2020. At the time
this report is written, this calculation is the only publically available model that performs a
bottom-up lithium-ion battery design and cost calculation.
The purpose of the report is to document the equations and assumptions from which the model
has been created. A user of the model will be able to recreate the calculations and perhaps more
importantly, understand the driving forces for the results. Instructions for use and an illustration
of model results are also presented. Almost every variable in the calculation may be changed by
the user to represent a system different from the default values pre-entered into the program.
The distinct advantage of using a bottom-up cost and design model is that the entire power-to-
energy space may be traversed to examine the correlation between performance and cost. The
BatPaC model accounts for the physical limitations of the electrochemical processes within the
battery. Thus, unrealistic designs are penalized in energy density and cost, unlike cost models
based on linear extrapolations. Additionally, the consequences on cost and energy density from
changes in cell capacity, parallel cell groups, and manufacturing capabilities are easily assessed
with the model. New proposed materials may also be examined to translate bench-scale values to
the design of full-scale battery packs providing realistic energy densities and prices to the
original equipment manufacturer.
The model will be openly distributed to the public in the year 2011. Currently, the calculations
are based in a Microsoft® Office Excel spreadsheet. Instructions are provided for use; however,
the format is admittedly not user-friendly. A parallel development effort has created an alternate
version based on a graphical user-interface that will be more intuitive to some users. The version
that is more user-friendly should allow for wider adoption of the model.
XVI
-------
PREFACE TO THE FINAL REPORT
Changes were made to the draft report in direct response to the peer-review report generated by
ICE International on March 31, 2011 for the U.S. Environmental Protection Agency. Other
changes were also made in response to additional private peer-reviews. Below is an incomplete
list of where changes have been made in response to the peer-reviews. Only the most important
changes are detailed below.
1. Implemented liquid thermal management on all battery packs. A new section was added to
explain the thermal management strategy Section 4. Costs of the thermal management are
addressed in 5.2.3.
2. Changed the default maximum single-sided electrode thickness to be 100 microns while
continuing to allow the user to override with any value. A greatly expanded discussion of the
transport issues, supported with validated physics-based models as well as a sensitivity analysis,
was placed in Section 3.6.1. Additional instruction on how to change the default value was
explicitly stated in Section 6.2.2.
3. Changed allocation of overhead costs based on reviewers comments to more closely represent
a Tier 1 auto supplier. Exact values with discussion are in Section 5.6.
4. Added cost estimates for automatic and manual disconnects necessary for safe operation of the
battery, Section 5.2.3.
5. Added and increased costs of battery management system to represent the complete
monitoring and control needs, Section 5.2.2 and 5.2.3.
6. Increased cost of tabs to account for polymer sealant and use of isolation tape, Section 5.2.2
7. Changed capital cost of the materials preparation (lower) and coating (higher) in response to a
more detailed review of the cost inputs from an industrial partner, Sections.3.
8. Reference and discussion on the validation of the model in Sections 1, 3.4, and 3.6.1.
9. The definition of the designed fraction of the open-circuit voltage at maximum power and its
connection to existing vehicle battery systems is discussed in Section 3.2.
10. Breakdown of the cathode material price into raw materials and baseline costs (processing,
profit, utilities, other materials) for the oxides based on nickel, manganese, and cobalt is in
Section 5.2 with some sensitivity to cobalt prices.
11. Included discussion of what the predicted model price represents and situations where it
might be conservative or optimistic, Section 1.
12. Increasing clarity on how to use the spreadsheet, Section 6 & 7.
XVll
-------
13. Clarification of the production volume scaling methods, Section 5.4.
14. Discussion of where waste management costs are included in the estimation, Section 5.5
15. Explanation of approximate method to account for modules, composed of the same cells,
connected in parallel, Section 3.6.3
16. Cell thickness changed, moving to thinner values, Section 3.1.
XVlll
-------
1. Introduction
The recent penetration of lithium-ion (Li-ion) batteries into the vehicle market has prompted
interest in projecting and understanding the costs of this family of chemistries being used to
electrify the automotive powertrain. The model described here-in is a calculation method that
was developed at Argonne for estimating the manufacturing cost and performance of Li-ion
batteries for electric-drive vehicles including hybrid-electrics (HEV), plug-in hybrids (PHEV),
and pure electrics (EV). To date, a number of cost models of various levels of detail have been
published in different forms.1"11 The cost of a battery will change depending upon the materials
chemistry, battery design, and manufacturing process.12"14 Therefore, it is necessary to account
for all three areas with a bottom-up cost model. Other bottom-up cost models exist but are not
available to the general public and have not been explicitly detailed in an open document. The
motivation for this work is based on a need for a battery cost model that meets the following
requirements:
1. Open and available to the entire community
2. Transparent in the assumptions made and method of calculation
3. Capable of designing a battery specifically for the requirements of an application
4. Accounts for the physical limitations that govern battery performance
5. Based on a bottom-up calculation approach to account for every cost factor
The Battery Performance and Cost model (BatPaC) described here-in is the product of long-term
research and development at Argonne National Laboratory. Over a period of years, Argonne has
developed methods to design Li-ion batteries for electric-drive vehicles based on modeling with
Microsoft® Office Excel spreadsheets.12"20 These design models provided all the data needed to
estimate the annual materials requirements for manufacturing the batteries being designed. This
facilitated the next step, which was to extend the effort to include modeling of the manufacturing
costs of the batteries. In the following sections of this document, a model is presented that meets
the above criteria and may be used to analyze the effect of battery design and materials
properties on the cost of the final battery pack. Use of BatPaC requires some basic knowledge of
battery packs; however, a user does not need to be an expert. For instance, the number of cells
and thus battery pack voltage must be specified by the user. However, default values are
available for more specific requirements such as experimentally measured values. In this way, a
person with reasonable knowledge of batteries may be able to conduct cost comparisons and
"what if studies.
The battery pack design and cost calculated in BatPaC represent projections of a 2020 production
year and a specified level of annual battery production, 20,000-500,000. As the goal is to predict
the future cost of manufacturing batteries, a mature manufacturing process is assumed. The
model designs a manufacturing plant with the sole purpose of producing the battery being
modeled. The assumed battery design and manufacturing facility are based on common practice
today but also assume some problems have been solved to result in a more efficient production
process and a more energy dense battery. Our proposed solutions do not have to be the same
methods used in the future by industry. We simply assume the leading battery manufacturers,
those having successful operations in the year 2020, will reach these ends by some means.
-------
Establishing the validity of the model calculation is important to justify the conclusion drawn
from exercising the model. The design methodology used has been previously validated against
cylindrical wound cell formats.15 The calculated materials quantities agreed with the actual
values within 3 %. Moving to a prismatic format simplifies the current collection calculation
while leaving the governing equations unchanged. The new approach developed for calculating
the cell impedance has been validated against experimental measurements from electrodes up to
100 (im in thickness.20 The module and battery jacket construction is of lighter construction
compared to contemporary designs. The battery pack energy density calculated in BatPaC is
higher than the battery packs used in the first versions of the Nissan Leaf and Chevrolet Volt for
equivalent cell designs (calculated value of 100 Wh/kg compared to reported value near 84
Wh/kg for the Volt).21 Significant engineering advances are necessary to minimize current
inactive material burden in the commercial pack designs thereby reducing the cost, mass and
dimensions of future automotive battery packs. We have assumed a design that we believe will
be representative of the engineering progress achieved by successful manufacturers in the year
2020.
Validation of the input material and capital costs are more difficult to achieve as few values are
publically available. We have relied, to a large extent, on private communications amongst
equipment manufacturers, materials suppliers, cell manufacturers, and original equipment
manufacturers. Variation does exist amongst the communicated values and we have maintained a
practical level of skepticism for their accuracy. Experts from all aspects of battery development
have reviewed the model both privately and as part of a formal peer-review process. While the
largest uncertainty in calculated values will exist in point cost estimates, the most instructive
information may be gained by examining ranges in parameter values and relative changes
between material properties (e.g. the advantage of moving to a manganese spinel cathode from a
layered-oxide material or from increases in cell capacity, etc).
The battery pack price to the OEM calculated by the model inherently assumes the existence of
mature, high-volume manufacturing of Li-ion batteries for transportation applications. Therefore
the increased costs that current manufacturers face due to low scale of production, higher than
expected cell failures in the field, and product launch issues are not accounted for in the
calculation. The model results for year 2020 could be considered very optimistic if the
transportation Li-ion market fails to develop as a result of insufficient investment in product
research and development, reduced motivation for lowering petroleum consumption and
greenhouse gas emissions, and/or a series of high-profile safety incidents.
-------
2. Cell and Battery Pack Design Format
Various cell and battery design concepts are under development at battery manufacturers. Based
upon experience gained from extensive previous work, we have found the exact design of the
battery does not have an important effect on the cost for a set cell chemistry; the amounts of
electrode materials and the number, capacity and electrode area of the cells, are the determining
cost factors. The most common cell designs for batteries nearing large-scale production are
cylindrical wound cells, flat wound cells, and prismatic cells with flat plates. Cylindrical cells
probably have a slight advantage for the assembly of the electrode-separator unit because of the
ease of making a cylindrical winding. For the different cell designs, there are small differences in
the weights of the terminal extensions and the procedures for connecting these extensions to the
current collector sheets, with a small advantage for flat plate cells. The flat-wound and flat-plate
cells form a more compact module and have better heat rejection capabilities than the cylindrical
cells. These small differences would have minor effects on the cost of batteries produced in high
volume in a mature, automated production plant and all of the cell designs can be adequately
cooled for most applications. We conclude that the BatPaC cost calculations would be relevant
for batteries differing considerably from the selected design approach.
To provide a specific design for the calculations, a prismatic cell in a stiff-pouch container was
selected (Fig. 2.1). For this design, calculations of the current collector and terminal resistances
are easily done with a one-dimensional model, because the terminals are almost the same width
as the electrodes.
Polymer Seal of
Cell Container
to Terminal
Ultrasonic Welds
of Terminal to
Collector Foils
Terminal Seal
Aluminum
Conduction
Channel
_
Cell Cross-Sections
\l
After Edge Sealing
After Edge Shaping and
Addition of Aluminum
Conduction Channel
Cell with Stiff, Multi-
Layer Container
Figure 2.1 Prismatic cell in stiff pouch container with aluminum conduction channel added for
heat rejection
-------
The cells are hermetically sealed in module containers, which are then enclosed in an insulated
jacket. The module enclosure provides added protection of the cell seals from the diffusion of
moisture from the external environment into the interior of the cells or alternatively loss of
electrolyte solvent from the cells. The exterior surfaces of the modules are cooled by ethylene
glycol-water solution. The calculated electrical performance of a battery of this construction is
near optimum and the configuration is compact and light-weight. We have not likely selected the
most viable design in this short study; there may be serious flaws in some details. However, the
calculated overall performance and low cost for the selected design will be challenging to match
in actual production and will only be met by the most successful manufacturers, those that will
dominate the market.
2.1 Cell Design
The prismatic cell of this design embodies individual positive and negative electrodes consisting
of current collector foils coated with electrode materials on both sides. The current collectors are
usually solid copper and aluminum foils for the positive and the negative. An illustration of a
segment of the cell is detailed in Figure 2.1. Each electrode is made up of active material
particles held together by a polymeric binder. A conductive additive, carbon black and/or
graphite, is added to the positive electrode and sometimes to the negative electrode. The
electrodes and separator each have porosity that is filled with the electrolyte solution. During
discharge, the Li-ions move from the electrode particles into the electrolyte, across the separator,
and then insert into the particles composing the opposite electrode. The electrons simultaneously
leave the cell through the current collection system and then enter through the opposite side after
doing external work. The materials currently used in Li-ion cells are based on an intercalation
process. In this process, the Li-ion is inserted into or removed from the crystal structure of the
active material. The oxidation state of the active material, or host, is concurrently changed by
gaining or losing an electron. Other electrode materials based on conversion reactions or
electrodeposition could be implemented into the model if the user desired.
positive
electrode
separator
Figure 2.2 Cell sandwich inside of prismatic pouch cells.
-------
The electrodes are easily and efficiently prepared by coating wide sheets of foil (up to 2-meters
in width) with uncoated strips running the lengths of the foil being coated. The individual
electrodes can be cut from these sheets with little waste of electrode coating material or foil (Fig.
2.3). The separator for these cells can be handled as a single sheet that is folded back and forth as
the electrodes are inserted. The electrodes are inserted so that all of the positive tabs extend
beyond the separator sheet in one direction and the negative tabs extend in the opposite direction.
The design model selects the number of electrodes to meet a set cell thickness determined by the
type of cell: HEV, 6 mm; PHEV, 8 mm; EV, 12 mm. These cell thicknesses are default values
and may be changed to suit the designer. The cell terminals are formed from flat stock to be
almost as wide as the entire cell. They are bent to the shape shown in Fig. 2.1 and ultrasonically
welded to the current collector tabs. The cell stack is then sealed between the two halves of the
cell container. The cell housing material is a tri-layer consisting of an outer layer of polyethylene
terephthalate (PEP) for strength, a middle layer of 0.1-mm aluminum for stiffness and
impermeability to moisture and electrolyte solvent vapors and an inner layer of polypropylene
(PP) for sealing by heating.22'23 The two halves of the cell container are pre-shaped to facilitate
assembly. The aluminum foil in the cell container material provides stiffness and it may be
increased in thickness to assist in conducting heat to the module container. After sealing the
edges of the cell, the edges are flattened along the sides of the cell to form a compact shape and
an aluminum conduction channel is added to assist in heat rejection at the sides of the cells.
._-,_
I .'. .
:- •
Electrode
Coating
~ Uncoated Area
for CC Tabs
Lines Showing
Places to Cut for
Electrodes
Figure 2.3 Illustration of coated current collector foil showing four rows of prismatic electrodes
before slitting or stamping into individual electrodes
2.2 Module Design
The module format is based on a casing of 0.5-mm thick aluminum that is sealed by double
seaming, a process that is well established and inexpensive because it is automated, rapid, and
uses low-cost equipment that is common in the container industry. The sealing of the module
provides an additional barrier to the loss of electrolyte solvent from the cells and the entrance of
water vapor. These deleterious transfers through the seals of pouch cells may shorten their lives
to less than the desired fifteen years.22
-------
Heat Transfer Surfaces
on Top and Bottom of
Container in Contact
with Cell Conductors
Cell Terminal
Connections
Module
Terminal
Double-Seamed
Module Closure
Figure 2.4 Hermetically-sealed module
The cells are placed on their sides in the module and the terminals of adjacent cells are connected
either mechanically with small bolts and flat springs to maintain contact or by laser welding.
Space is provided within the module casing on the left side, as sketched in Fig. 2.4, for an
electronics package that includes cell monitoring for malfunctions (temperature and voltage) and
for state-of-charge (SOC) control. The SOC control is activated when ever the battery is at rest
and it diverts charge from the cells at highest voltage to those at lowest voltage.
2.3 Battery Pack Design
The model designs the battery pack (Fig 2.5) in sufficient detail to provide a good estimate of the
total weight and volume of the pack and the dimensions of the battery jacket so that its cost can
be estimated. The modules are arranged within the battery jacket either in a single row, with the
terminals facing the same side of the pack, or in an even number of rows with the terminals in
one row facing the terminal of an adjacent row. For a pack with a single row of modules, a
busbar must be provided to carry the current to the front of the battery pack. This feature results
in an additional cost for the busbar. For batteries with more than one row of modules (Fig. 2.5),
the terminals are laid out on the module so as not to interfere with those on the opposite row of
modules, thus conserving space in the battery pack. The modules in a row are interconnected,
negative to positive terminals, by copper connectors. The modules casings are compressed
together by two steel sheets bound with steel straps at the front and back of the battery pack. The
compression is necessary to ensure intimate contact between the active layers that make up the
pouch cells that are tightly fit into the modules. The compressive force also serves to add
structural support to the module casings.
-------
Coolant
Outlet
Compression
Plate and Straps
Polymer Foam
to Block Flow
Pack Cover
Attached to
Module Tray
Section A-A
Coolant
Inlet
Section B-B
Pack with Two Rows of Modules
Section A-A
Section B-B
Pack with One Row of Modules
Figure 2.5 Insulated battery jacket with enclosed modules that are cooled on their upper and
lower surfaces by ethylene glycol-water solution
The modules are supported by a tray that provides space for the heat transfer fluid (ethylene
glycol-water solution) to flow against the top and bottom of each module. All connections to the
pack terminals that lead to the exterior of the pack and signal wire feedthroughs can be made
-------
before inserting the attached modules into the jacket and making the final closure. The bolts
depicted in the diagram (Fig 2.5) for making this closure are for illustrative purposes only.
The battery jacket consists of a sheet of aluminum on each side of a 10-mm thick layer of ridged,
light-weight high-efficiency insulation. The thickness of each of the aluminum layers is selected
by the modeling program to be 1- to 2-mm thick, depending on the total volume of the modules.
The insulation slows the interaction of the battery with the external environment that cools the
battery in the winter and heats it in the hot summer weather.16
Although the main purpose of the battery pack design for the model is to provide a plausible list
of materials to estimate the manufacturing cost of the battery, the overall design approach
permits the battery to be shaped by the designer to fit dimensional objectives. If there is a height
restriction for the battery pack, a high ratio of height-to-width for the positive electrode will
result in a battery of low height (the cells are placed on their side in the pack).
-------
3. Modeling of Battery Design and Performance
The design portion of the model calculates the physical properties of a battery based on user-
defined performance requirements and minimal experimental data. An illustration of the model is
shown in Figure 3.1. The user is asked to enter a number of design parameters such as the battery
power, number of cells and modules, and target voltage at maximum power, etc. In addition, the
user must enter one of the following three measures of energy: battery pack energy, cell capacity,
or vehicle electric range. Defining one of these values will determine the value of the other two.
An iterative procedure then solves for the user defined energy parameter (energy, capacity, or
range) and remaining battery properties by varying the cell capacity and electrode thickness. The
result is the dimensions, mass, volume, and materials requirements for the cells, modules, and
battery pack.
The model has been designed to allow the user to enter as many customized values as desired. In
this way, the model allows flexibility in the battery chemistries studies and some of the cell,
module, and battery design aspects. Hence, the focus of this report is on the method of
calculation and not the exact values chosen for a specific capacity or cell thickness. However, the
default cell design parameters as well as experimental data measured at Argonne National
Laboratory, for a number of different battery chemistries both commercial and developed at
Argonne, are available for use within the model. There are five governing equations for battery
performance that calculate the current density, battery energy, electrode area, electrode
thickness, and resistance. The voltage at maximum power and the area specific impedance (ASI)
are two important parameters in the design model for calculating the battery performance. Most
of the discussion will be spent on these two properties.
3.1 Criteria for Power, Energy, and Life
In order to fully specify a battery design, the user of BatPaC must supply criteria for power,
energy, and life. These criteria will depend on the application for which the battery will be used.
While the user may change some of the settings as they prefer, we list our suggestions in Table
3.1. The battery type is defined by the end-use application. Hybrid electric vehicles (HEVs),
plug-in hybrid electric vehicles (PHEVs), and electric vehicles (EVs) have increasing levels of
electrical energy storage for use by the vehicle drivetrain. The model will use Table 3.1 or the
user's explicit inputs to size the battery correctly for the chosen application.
Table 3.1 Criteria for designing batteries for a specific end-use application
Battery Type
SOC for Rated Power, %
Power Duration, sec
SOC Range for Useable Energy, %
Cell Thickness, mm
microHEV
50
2
40-65
6
HEV-HP
50
10
40-65
6
PHEV
25
10
25-95
8
EV
25
10
15-95
12
-------
Battery Design Model
Pack Requirements
• power
• energy or range
• number of cells
Key Constraints
• max electrode thickness
• target cell potential, V, at
peak power
• cell/module format
Iterative Spreadsheet
Solves for cell capacity
and designs battery pack
by varying:
1. Cell area
2. Electrode thickness
3. Internal resistance
Cell Chemistry
Measured Properties
• pulse power ASI
• discharge ASI
• mAh/g, g/cm3
• electrode porosity
• SOC window
• physical properties
ASI = area specific impedance
Figure 3.1 Summary flow of the design model
The microHEV is a micro or mild-hybrid that provides a moderate power level, -25 kW, for two
seconds. This design is best suited for cell chemistries capable of very high power-to-energy
(P/E) ratios. The HEV-HP is a power-assist hybrid that provides the rated power for a full 10
second pulse. The power for both HEV applications is rated at 50 % state-of-charge (SOC). The
energy available for discharge and charging is 25 % of the total energy to ensure long cycle life.
As the capacity of the HEV cells is typically small, a cell thickness of 6 mm is used. The PHEV
utilizes a much larger portion of the total energy, 70 %. At the end of discharge, the PHEV
battery is operated in a charge sustaining mode. Therefore, the power rating for the battery is
Calculated Battery
Pack Properties
• dimensions
• volume & mass
• specific energy,
power
• materials required
10
-------
determined at 25 % SOC. PHEV cells should be much larger than HEV cells and thus a cell
thickness of 8 mm is assumed. Finally, EV batteries use 80 % of their total energy with their
power rated near the end of discharge. Cell thicknesses are set to 12 mm to accommodate a high
capacity design. As noted later in the report, selecting a parallel arrangement of cells
automatically assumes a cell thickness of 6 mm regardless of the end-use application.
Additionally, the use of negative electrodes operating at potentials high above the lithium metal
potential may extend the upper end of the available SOC range from 95 to 100 %. The lithium
titanate spinel, Li4Ti5Oi2 (LTO), negative electrode is an example of intercalation electrode with
almost no risk of plating lithium metal during a charge pulse. On this basis, the available energy
for LTO-based Li-ion chemistries is suggested to be 75 % for PHEVs and 85 % for EVs.
In an established factory, the fixed design parameter is most likely the electrode area for a single
layer rather than a set cell thickness. To make higher capacity cells, more layers of the
predetermined footprint are stacked, thus increasing the cell thickness. In our model, the plant is
constructed for the sole purpose of building the battery being designed. Flexibility to produce
other products is not taken into account. Most importantly, the model calculates very little
difference in cost whether the cell has large-area layers that are few in number as compared to a
high number of layers smaller in area. For ease of calculation, we use a set cell thickness to
determine the number of layers. The area of each layer is set by the cell capacity requirement.
Accounting for capacity and power fade in the battery requires the user to design the battery with
the appropriate excess energy and power at the beginning-of-life (BOL). Defining the voltage at
which maximum power is achieved at BOL is one way to set the allowable power fade over the
life of the battery. This is discussed in detail in the following section. Capacity or energy fade
must be accounted for by over-sizing the battery at BOL. If the user has a certain fade
requirement, then the BOL energy may be increased to meet the end-of-life (EOL) target. The
design model does not attempt to predict fade rates or even suggest an allowable fade for a
specific application. It is our view that many aspects of materials chemistry, cell design, and
battery use directly affect the decay of the battery pack. Hence, we allow the user to make
accommodations for decay as he or she believes is necessary.
3.2 Voltage at Maximum Power
The voltage at which a cell reaches the designed maximum power is one of the most important
factors in the design of a battery. However, this specification is one of the least discussed aspects
of battery design. The voltage at maximum power, Vceii, is a measure of the largest polarization
the cell will undergo during operation at the BOL. This initial value has a direct effect on round-
trip battery efficiency, heat removal requirements, cold-cranking power, and allowable power
fade. A basic calculation demonstrates the maximum achievable power for a battery at BOL is at
50 % of the open-circuit voltage (OCV). Operating at these conditions would result in an
inefficient battery and require a significant cooling system to reject heat. More importantly, the
battery will never be able to reach this power level after any increase in impedance occurs. With
all certainty, the impedance of a battery will rise with time and the power rating of battery will
no longer be accurate.
11
-------
We design the battery to achieve EOL power capabilities at a specified fraction of the open-
circuit voltage, [V/U], at BOL. This approach is unique when compared to current design
practice of OEMs and cell manufacturers. However, a characteristic value of [V/U] exists for all
batteries regardless of the battery design process. One may determine this value for an existing
system in a straightforward manner. The potential at maximum power is measured at the end of a
10 s pulse at the EOL power rating and the SOC used for the power rating of a specific battery
type (HEV, PHEV, EV). The designed [V/U] value is the measured potential at the end of the
pulse divided by the open-circuit potential reached long after the pulse. This design point then
captures the degree to which the battery has been oversized to enable long-life, cold-start, and
efficient operation. The remainder of this section presents a discussion for setting the BOL
voltage at maximum power at no less than 80 % of the open-circuit voltage, [V/U] = 0.8.
Defining the voltage as a fraction of the OCV, allows for direct calculation of all the necessary
battery properties (see for example Eq. 3.6 or 3.8 in the section 3.3).
The allowable increase in battery resistance over the life of the battery is a function of the
designed voltage for maximum power. In general, designing the battery to achieve maximum
power at a higher [V/U] allows for larger resistance or impedance increases over the lifetime of
the battery. Figure 3.2 created from Eq 3.1 displays how the voltage at maximum power will
change to meet the designed power as the internal resistance of the battery increases. Clearly,
achieving BOL power at a high fraction of the OCV allows for greater degradation within the
usable lifetime of a battery. RI is the initial resistance of the battery at BOL while R2 is the
resistance as the battery ages. If the minimum voltage is 55 % of the OCV, the allowable
increase in resistance for batteries designed for BOL max power at 70, 80, and 90 % OCV is 18,
55, and 175 %. The consequence of achieving the power at lower and lower fractions of the
open-circuit voltage is that both electric current and heat generation will increase over the
lifetime of the battery, Figure 3.2b and Figure 3.3. The proper design of a battery will account
for the changes over the entire lifetime and not just desired behavior at BOL.
V_
U
-y-
U _
(
U-
ll
"y"
[/
A
J
(3.1)
The level of heat production is significantly different at BOL for batteries designed to meet
maximum power at differing fractions of the open-circuit voltage. We may compare the
differences in designed [V/U] by assuming the resistive heating (joule heating) is the most
significant factor in determining the heat generation, Eq. 3.2. This assumption is true for
moderate to high rate applications. We also reasonably assume the ASI will not change
significantly in the range of current densities and electrode thicknesses we vary in the
comparisons. From this point, we can analyze the difference in heat generation from different
designed [V/U] values, Eq 3.3.
12
-------
I
o
O.
re
>
O
O
u
re
Designed
BOL [V/U] at
max Power
50
80
70
60
50
40
Designed
BOL [V/U] at
max Power
0 25 50 75 100 125 150 175 200
Increase in Resistance, %
Figure 3.2 a) Required change in [V/U] to maintain rated power with increases in internal
resistance over the life of the battery, b) Increase in current due to lowered [V/U].
13
-------
U2 1-
qj=(Aposl)2RJ=ItotalU\l-
V_
~u
U
(3.2)
y
U
(3.3)
ouu
A Kn -
4OU
^9
0^
•* /inn -
C T'UU
o
t; i^n -
co oou
3_
0
c onn -
0 ouu
O
*- ocn _
CO ^OU
0
onn -
_ ^uu
0) 1c-n _
(0 1OU
CO
0
*- mn -
O 1 vlvl
_c
en _
ou
n _
Designed
BOL [V/U] at
max Power /
90%
80% /
---- 70% /
/
/
/
I /
J^
0 25 50 75 100 125 150 175 200
Increase in Resistance, %
Figure 3.3 Change in heat rejection requirement from increases in resistance for batteries with
different designed voltages at maximum power.
The ratio of resistances may be found by equating the power for the two cases. Then the
resistances, and areas if the ASIs are equivalent, are determined solely by the fraction of the
open-circuit voltage at which they achieve maximum power, Eq 3.4. Then substitution will give
14
-------
the ratio of heat production at maximum power for the two cases, Eq. 3.5. A battery that achieves
maximum power at 80 % of OCV will have a heat production at maximum power that is 2.3
times higher than one designed at [V/U] = 90%. A battery producing power at 70 % of the OCV
will have 3.9 time higher heat generation than at [V/U] = 90 %.
owerlA
powerl pos2
"y"
u _
"y"
u _
(
,( ~
(
2\ ~
"y"
u _
"y"
u _
J
1
J
(3.4)
"y"
u _
"y"
_c/_
(>-
i-
"y"
"y"
j
j
(3.5)
Two straightforward design changes will enable operating a battery at 90 % of OCV compared to
80 % while maintaining the same power output. First, a second identical battery may be
connected in parallel to the original battery. This will lower the resistance of the battery pack by
one half but will also double the energy and cost of the battery. A more realistic approach is to
reduce the electrode thickness by coating a larger separator area. The capacity of the cell is
maintained while minimizing increases in cost from a larger separator, current collector and
packaging area. This approach is feasible as long as the reduced electrode thickness is above the
increase in AST, > 20 microns as discussed in detail below.
The efficiency of a battery defines the heat rejection requirements and may be measured or
calculated. Measurement of round-trip efficiency of a battery is best performed by using a
calorimeter to measure the heat given off during the cycling of the battery. The calorimeter
removes the requirement of knowing the exact SOC of a battery during the entire drive cycle.
Calculation of the round-trip efficiency of a battery requires a detailed transient battery model
within a vehicle simulation program to exercise the battery over the many acceleration and
deceleration periods that occur during a drive cycle. The interesting result is that the same battery
will have different power ratings depending on what level of round-trip efficiency the user is
willing to accept.
Figure 3.4 shows the efficiency of a battery as a function of the designed potential at which the
battery reaches maximum power. The figure is created using Equations 3.1 and 3.4 above. Each
line may be considered a different drive cycle, or duty load, for a battery with the same energy
but different impedance (changing separator area). The straight, solid black line represents the
efficiency of the battery operated only at maximum power, P/Pmax = 1. In example, a battery
designed at [V/U] = 0.8 will have 80 % efficiency for a single discharge pulse at maximum
power. Likewise, a battery designed at 0.9 will be 90 % efficient at maximum power. Batteries
are normally operated in the area above the line of the maximum power. Therefore, the other
curves represent the efficiency of discharging a battery at power levels below maximum power
15
-------
(typical driving conditions). Consider two batteries each designed for a maximum power of 100
kW although one achieves this power at a [V/U] = 0.9 and the other at 0.7. If the two batteries
are discharged at 45 kW, P/Pnw* = 0.45, the battery designed at [V/U] = 0.9 will be 6.4 % more
efficient. This is significantly less than the 20 % efficiency improvement realized when operated
at maximum power. The efficiency penalty is reduced as the battery operates less and less near
the designed maximum power.
100%
50%
0.5
0.6
0.7
0.8
0.9
Designed [V/U] at Max Power
Figure 3.4 Efficiencies for batteries designed to achieve maximum power at different fractions
of their open-circuit voltage. Comparative efficiency lines are shown for equivalent power
demands over a period of battery operation.
16
-------
3.3 Governing Equations
The five coupled, algebraic equations that govern the battery design are presented in this section.
While these equations are perhaps the most important, many other equations are used to fully
define the battery mass and volume. These other equations will be specified where necessary in
the following subsections.
The user of the model specifies the required maximum power, Pbatt, of the battery. This power is
translated to a current density, /, in Eq 3.6 using the area of the positive electrode, Apos, the
number of cells, NCeii, the open-circuit voltage at the SOC for power, C/0cv,p, and the fraction of
the open-circuit voltage at which maximum power is achieved, [V/U].
/=-
batt
v_
u
(3.6)
The relationship between capacity and battery energy is described by Equation 3.7. Formally, the
energy of a battery is the product of the capacity and the average voltage at which the energy is
obtained. The average cell voltage is approximated in Eq. 3.7 by subtracting the polarization
from discharging the battery at a C/3 rate from the open-circuit voltage at the SOC for energy,
f/ocv,E- The energy for all batteries designed by the design model is calculated at a C/3 rate and
the average open-circuit voltage at 50 % SOC. The remaining necessary values are the capacity
of the cell, C, ASI for energy, ASIenergy, number of cells, and area of positive electrode. Either the
battery energy or capacity may be specified. The energy may alternatively be determined from a
stated range, fraction of total energy available, and energy usage rate for the vehicle (Wh/mile).
r ASI
en
ergy
3 A
(3.7)
pos j
The area of the positive electrode in Eq. 3.8 is determined largely by the area specific impedance
for power, ASIpower, and resulting voltage drop. The voltage of cell at max power, Vceii,p, is found
from the product [V/U] £/0cv,p, where Uocv,p is the open circuit voltage at the SOC for max power.
In general, the area of the electrodes will increase if the ASI for power increases. The areas of
the negative electrode and separator are determined from the area of the positive electrode. The
negative electrode is taken to be 1 mm larger than the positive electrode in both height and width
to alleviate concerns of lithium plating during charge pulses. The separator area is slightly larger
than the negative electrode to prevent the electrical shorting of the two electrodes.
P
power rbatt
u
(3.8)
The positive electrode thickness, Lpos, in Eq. 3.9 is determined from the capacity of the cell, C,
specific capacity of the electrode, Q, volume fraction of active material, sact, bulk density of the
17
-------
active material, p, and the positive electrode area. The negative electrode thickness is determined
by its specific reversible-capacity and the designed excess-capacity to prevent lithium plating
during charging. We have chosen a ratio of 1.25 negative to positive reversible-capacity (N/P
ratio) for the default value for the cells with graphite negative electrodes. LTO negative electrode
based cells are designed at a 1.1 N/P ratio because of the previously mentioned minimal
possibility of lithium deposition. The maximum allowable electrode thickness is a user defined
value. The calculation for the electrode area changes when the designed thickness is greater than
the maximum allowed (Section 3.6.1).
. (3.9)
Finally, the AST for power (and for energy to a lesser extent) is calculated using an expression
that is based on the electrode thicknesses, the current density, and the C-rate. The exact
expression will be discussed in the next session. The AST in Eq 3.10 shows the basic
dependencies with a and P being constant valued parameters.
pos
3.4 Calculation of the ASI
In most battery design scenarios, the ASI directly determines the electrode thickness to meet a
specified power-to-energy (P/E) ratio. From this electrode thickness, the area of the electrode is
set to meet the capacity requirements. Clearly, the ASI plays a significant role in the design of a
battery and particularly in the case of the P/E ratios required by automotive applications.
However, the ASI is not an inherent constant of a specific battery chemistry or cell design. The
measured value of the ASI is a complex combination of resistances within the battery resulting
from the physical processes occurring at different length and time scales. Consequently, the
measured value is a function of many factors (state of charge, pulse length, current density, C-
rate, particle size, transport and kinetic parameters, etc). The calculation used for the ASI in this
battery design model has been discussed in detail and validated against experiments elsewhere.20
The physical meaning of the equation will be discussed but those interested in the derivation are
directed to the separate publication. We note that the ASI described here is slightly different than
the one addressed in the paper. The thermodynamic component is removed that originated from
the change in open-circuit potential with concentration for the intercalation materials. Equation
3.11 contains the definition of the ASI used in this document. 7ti is a positive valued current
density for a discharge pulse. I& is equal to zero as it is during the relaxation period after the
pulse. Time 0, tO, is the time just before a current pulse begins, time 1, tl, is the time just before
the current pulse ends, and time 2, ?2, is the time long after the current pulse when the cell is at
open-circuit and the concentration gradients have relaxed. Therefore, this ASI measurement is
not troubled by accounting for a change in open-circuit voltage with the passage of current. In
general, the ASI measured with this definition is similar, although smaller, in value to those
produced using the more standard definition used elsewhere.
18
-------
(3.11)
The AST for the electrochemical charge and discharge process is referred here-in as ASIechem. Our
calculation approach for both the AST for power and for energy involves adding three
components together to reach the ASIechem, Eq. 3.12. The first two factors include impedance that
arises from the interfacial charge transfer and transport. The third factor is a lumped parameter
used to capture the remaining impedance.
A S7 — A <\JP°S -L A
-------
I = rcQpeaaLpos (3.15)
The cell AST for energy, ASIenergy, and power, ASIpower, are determined by adding the ASIechem to
that of the current collectors, ASICC, as discussed in the next subsection. The difference between
ASIenergy and ASIpower is that the limiting currents are not important during the C/3 discharge for
energy and the ASIconst is a different value for two cases. ASIenergy will always be higher than
ASIpoWer if a battery is operated far from the limiting current. The higher impedance is due to the
formation of significant concentration polarizations during the longer time scale of the energy
discharge. A reasonable rule-of-thumb is that the ASIconst for energy is 2.2 times the value for
power in layered oxide materials such as LiNio.soCoo.isAlo.osOi.
3.4.1 Current Collection Resistance
The resistance from the conductors used to collect the current must be accounted for as they can
contribute significant ohmic drop to the battery. The AST used to calculate the required cell
separator area, ASIpower, is larger than the AST for the electrochemical charge and discharge
processes, ASIechem,p, as shown in Equation 3.16. The ASIechem value is typically measured from
experiments and must be added to the external resistances that arise from the materials used to
conduct the electric current. These resistances come from current collection in the cell and also
those on the module and battery pack level.
ASIpower = ASIechemf +ASICC +ASl£ +^^L (3.16)
Ncells
The current collector foil impedance, ASICC, is determined from an analytical expression, Eq.
3.17, which accounts for the coated and uncoated region of the foil, labeled act for active and tab
respectively. The resistance factor, R/, and the resistance of the current collector foils, Rcc, are
also shown for clarity in Eq 3.18 and 3.19. The factor of 2 in the R/term is due to assuming half
of the foil thickness carries the current produced on one side of the foil. While all of the current
passes through the tab region, the magnitude of the current varies along the height of the coated
foil as the reaction area continually contributes current to the foil. An equivalent length for the
resistance calculation may be determined so that multiplication by the total current for a cell will
give the correct ohmic drop. This equivalent length is H/3 if the current density is relatively
constant over the entire area. The derivation of this equivalent length as well as an in-depth
discussion of the voltage and current distribution in the foils may be found in subsection 3.4.2.
Also in the later subsection, the assumption of constant current density is verified with numerical
modeling.
ASIcc=HaaWaaRcc=Rf
(3.17)
Rf=\ 7 + 7 1 (3-18)
V foil ,neg foil ,neg foil, pos foil, pos
20
-------
(3.19)
The cell terminals are ultrasonically welded to the ends of the current collector foil tabs. While
the welding removes this contact resistance, the AST of the terminal must be included in the total
cell resistance. The AST of the cell terminals, ASI^, is the summation of the positive and
negative cell terminals as shown in Eq 3.20. The dimensions for these terminals are set by the
calculated width of the cell and the user defined terminal thickness and height.
1.11 #_ (3_2Q)
(7 (7 \W J
term,neg term,pos J term term
pos
The AST for connection losses is the last term in the AST summation stated in Eq. 3.16. This AST
value is calculated by multiplying the ratio of cell positive electrode area to number of cells by
the summation of the resistances, ^cnct, for cell terminals, module terminals, module
interconnects, and batteries terminals. In this way, each cell shares in the burden of overcoming
the system losses from carrying the electric current. The calculation of ^Cnct is detailed in Eq.
3.21 with the individual sources of connection losses shown. The voltage drop resulting from
cell-to-cell contact resistance, Rc^ct , is taken to be 10~4£/OCv,E in Eq. 3.22, a small fraction of the
open-circuit voltage. A battery manufacturer would only tolerate a minimal voltage drop from
cell-to-cell contact. One connection method is to physically press the two cell terminals together.
This resistance could be lowered by increasing the physical pressure and contact area, or by laser
welding the terminals together. Regardless, the value used in the model is left to the choice of the
user to leave as is or to change to a different value.
modR^ + (Nmod - Ifel + R%L (3-21)
= 10"4 Nc*U°">'E (3.22)
The module terminal resistance, R™^ , calculation in Eq. 3.23 is shown as an example of how the
terminal and interconnect resistances are calculated for the module and battery pack. The size of
the terminals and thus their resistance are determined from a calculation based on a pre-
determined allowable rate of temperature rise for the conductor. This approach is explained in
more detail in subsection 3.4.3.
mod " term AT mod
term = 7 Nterm
21
-------
3.4.2 Potential and Current Distribution in the Current Collection Foils
The designed current collection system was evaluated using a numerical simulation package.
Equations 3.24-3.26 were solved for a steady state, isothermal, and 1-D simulation. Here, the
conductivity, 03, is the effective conductivity of l/2 of the foil (the other half carries the current
from the opposite side). The bulk conductivity value, 03°, is multiplied by the thickness of the
conductor, L/2, to lower the dimension of transport.
In=— H^ ^^^ (3-24)
-In (3-25)
ln (3-26)
The boundary conditions were set for both ends of each foil. The tab ends of the foils were set to
a specified voltage and the opposite ends of the foils were restricted to a no flux condition. The
simulation was performed using the foils defined in our battery design: 12 micron thick copper
foil and 20 micron thick aluminum. The cell length was 20 cm, the ASIechem was 30 ohm cm2, and
the Uocv and Vceu were set to 3.72 and 3.57 V respectively. Figure 3.5 shows the current and
potential distribution in the foils and in the cell resulting from the simulation. The cell potential
along the length of the foil varies only by 1.5 mV from maximum to minimum difference. The
0.4 % variation in voltage results in a 0.9 % variation in current density. This verifies the current
density is uniform along the length of the foil. This is also obvious from the linear relationship of
current with foil height in Fig. 3.5. The assumption of constant current density was tested in cell
heights up to 100 cm and found to be satisfactory. The assumption should be reasonable as long
as the ASIechem is at least twice the value of ASICC. The simulated resistance of the foils is found
to raise the ASIechem by 0.7 for an ASIpower of 30.7 ohm cm2. Additionally, the numerical result
verified that H/3 is the correct equivalent length to represent the ASICC for the cell. This may also
be found analytically, Eq. 3.27-3.29, if you assume an even current distribution. We have shown
that to be a reasonable assumption.
(3.27)
21 gs -<&lne
ASIechem + ASICC = — I : - —
H 0 ln
22
(3.28)
1 l ' (3.29)
-------
3.5750
„ 3.5740
.2
"^
c
0)
+•>
o
Q.
"55
o
c
(0
3.5730
3.5720
O 3.5710
0)
"^
'(75
o
Q.
o
o>
Q.
0)
i_
O
3.5700
3.5690
0.12
0.10
0.08
0.06
0.04
0.02
0.00
0.0000
•Cell
•Positive foil
•Negative foil
Negative foil
Positive foil
•current density
0.2
0.4
0.6
0.8
x/H
a>
-0.0040 '•*=
O)
0)
-0.0050
-0.0060
0.00492
0.00491
O
0.00486
Figure 3.5 The change in current and potential within the positive and negative foils. The current
collection design results in a uniform current distribution along the length of the foil.
23
-------
An analogous problem has been solved by Euler and Nonnemacher and then communicated
repeatedly by Newman et al.24' 25 The analytical solution they presented may be used after a
slight alteration to dimensionalize the current density to the geometry of our concern, Eq. 3.30
and 3.31. This solution was reached assuming linear polarization behavior and is valid for cases
where the current density varies along the height of the current collector foil. Thus, this approach
is a more general solution than the one we use in the design model.
H2
a
•"
+ -
1 + -
v sinh v
(3.30)
H2
echem V pos neg
(3.31)
3.4.3 Determination of Module Terminal, Battery Terminal, and Module Inter-connect Size
An important factor for setting the resistances of a module terminal, battery terminal, or module
interconnect is the allowable rate of temperature rise in the conductor at full power. We set the
acceptable rate of temperature rise, dT/dt, at 0.2 °C/sec or a 2 °C rise for a 10-sec power burst
under adiabatic conditions. The heating rate, q, is then used to determine the mass, m, of the
terminal required for the designed battery in Eq. 3.32. Since the heating rate may also be
determined by Eq. 3.33, we may determine the cylindrical terminal radius and mass by assuming
a length, //term- In this way, the size of the module terminal is redesigned during each simulation
to meet the specified power requirements and allowable temperature rise, Eq. 3.34. The mass of
the conductor is found to be inversely proportional to the allowable temperature rise.
r dT
mCp~T
p dt
(3.32)
term term
(3.33)
A = 7A
term pos
dT_
r'term " term ^ p ,
-1/2
(3.34)
A copper busbar must also be sized for batteries using a single row of modules. We have
somewhat arbitrarily assumed a AV&, = 30 mV drop across the busbar to be allowable at
maximum current. This value maybe easily changed by the user. Equation 3.35 is used to
calculate the mass of the busbar, m^,. The complicated expression for the volume of the busbar is
24
-------
derived from the voltage drop, conductivity, busbar width, Wbb, and required busbar cross-
sectional area.
ltOt\Wbb)
nlbb ~ rbb
3.5 Calculation of Battery Dimensions
The goal of the model is to quantify how the various components of a specific battery design
sum to make the mass and volume of the battery pack. In this way, a true energy and power
density can be calculated as well as the exact materials requirement to meet this design.
Summing the mass of the components is relatively straight forward. Determining the total
volume that contains the components and required free volume is not as obvious. The exact
calculations used in the design model are detailed below for the cell, module, and battery pack.
3.5.1 Cell Dimensions
The number of layers in each cell is approximated in Eq. 3.36 by accounting for the compression
factor, .Xcomp, and the individual thicknesses of the current collector foils, Lf0a, electrodes, Lpos
and Lneg, separator, Lsep, and container, Lcont. Xcomp is usually taken to be 0.97. The Li-ion battery
chemistries this model was designed for are assumed to undergo negligible volume change on
the cell level. No effort was made to address possible changes in electrode volume upon cell
discharge or charge.
T _2/~ +T"eg
AT _ y cell ^^cont ^ ^foil ,~ ,-, -^
layers ~ comp j mg j pos ,J,T ~j ~j 7 ^ ' >
^foil ^ ^foil ^ L\^sep ^ ^neg ^ ^ pos >
The Niayers approximation is necessary as the cell thickness is a user defined parameter. The
aspect ratio of the cell is also user defined; therefore, solving for the width also determines the
height of the cell as seen in Eq. 3.37. The width is calculated from the number of layers and the
aspect ratio, H/W. The factor of 2 enters the denominator as both sides of the foil are assumed to
be coated.
(3-37)
Having determined the width and height of the electrode, the rest of the cell dimensions are
relatively straightforward, Eq. 3.38 and 3.39. The width of the cell, Wceii, is 8 mm wider than the
positive electrode to allow for the larger separator area and pouch seals. The pouch seals are
folded up, pressing along the inside wall of the module casing. The height of the cell, Hceu, is the
height of the positive electrode in addition to the distance for the terminals and connections to
the foil tab, Lterm>cnt. Our assumed design requires 15 mm for this distance. The volume of the cell
is the product of the three dimensions.
25
-------
WaU=Wpa+* (3.38)
Hcdl = Hpos + 2Lmt (3.39)
3.5.2 Module Dimensions
The module dimensions are defined by Eq. 3.40-3.42. The height and length of the module are
both just 2 mm wider than the cell dimension. The width of the module is related to the total
thickness from all of the cells with allowance for a SOC controller at one end.
Hmod=Wcdl+2 (3.40)
Lmod=Hcell+2 (3.41)
Wmod=Lcell(Ncell/mod+l) + l (3.42)
3.5.3 Battery Pack Dimensions
The battery pack volume includes all of the modules, spacing for connections between modules,
channel for the cooling air to flow, Hair, thickness of the module compression plates, Lcomp, and
the battery pack jacket, Ljack (Eq. 3.43-3.45). Ljack includes a 10 mm thick insulation layer
sandwiched between two aluminum walls for the container. The thickness of the aluminum wall
increases from 1 to 1.5 to 2 mm as the battery volume increases from < 20 L to < 40 L to larger
dimensions. The layout of the modules, number per row, Nmod/row, and number of rows, Nrow, is
also included. The final volume of the battery is the product of the three dimensions. The space
left for connections between modules, Lgap, is a function of the number of rows of modules. Lgap
is equal to 8, 10, or 20 depending if there is one, two, or four rows of modules. Three rows of
modules are not allowed as the positive and negative terminal for the battery would be on
opposite ends and thus not very practical. A number greater than four rows of modules is deemed
unnecessary.
Hbatt = Hmod + 2Halr + 2Ljack (3.43)
^batt = N mod/ row "mod + "air + ^^comp + ^^ jack (3-44)
Wbatt=NrowLmod+Lgap+2Ljack (3.45)
3.6 Additional Considerations
A few situations may arise that require a change in the calculation method. These situations are
addressed in the subsections below. The inclusions of these calculations into the model allow for
a more realistic depiction of limitations often encountered by cell manufactures.
26
-------
3.6.1 Maximum Electrode Thickness
A practical limitation exists for the maximum achievable electrode thickness. This limitation
may be set by manufacturing capabilities, ionic and electronic current transport within the porous
electrode, susceptibility to plating lithium on the negative electrode, or aging characteristics
related to adhesion to the current collector. Some of these challenges are discussed in great detail
in the following subsection. When the maximum electrode thickness, LmaX, has been reached on
either the positive or negative electrode, the electrode area equation is modified as shown in Eq
3.46. The electrode thickness, Ltgt, is the largest electrode thickness, negative or positive,
calculated at the targeted fraction of the OCV [V/U].
ALpr=Y^Apos (3.46)
max
The area of the electrode is now determined by the cell capacity requirement to meet the battery
energy demands and not the target voltage at maximum power. As a consequence, the battery
pack will operate at a higher [V/U] than originally selected by the battery designer. The new
[V/U] may then be calculated from Eq. 3.47 which is the solution to the quadratic found in Eq.
3.48.
2C7
cell
v
' -*- •»- T \l-r-r \2, A UU.ll I}UWP.t~ /,-* A l—l\
(3.47)
-r j ^* -r -T CKtt •» I V UKll ' -» T i \ '
power
V N A
cell cells pos
(3.48)
The maximum electrode thickness may have a large impact on the energy density and cost of
cells designed for high energy and range. Nelson et al. demonstrated this concept in 2009
assuming a 100 micron maximum electrode thickness.12'19 In 2010, Santini et al. relaxed this
assumption to 300 micons; although, the thickest electrode discussed in the paper was a 225
micron graphite electrode in the LMO-Gr EV with 100 mile range.13 In conversations with
manufactures, 100 microns appears to be the general electrode thickness used for EV type cells
at the present time. However, Santini et al. has shown substantial increases in energy density and
decreases in cost if larger electrode thicknesses may be utilized. The challenges to achieving
thick electrodes, in addition to those already mentioned, relate to fast charging while avoiding
lithium metal deposition, utilizing all of the materials reversible capacity, removing gases formed
during formation cycling, wetting the full porosity of the electrode, achieving defect free
coatings, and drying the thick electrode at high rates. Our opinion is that the successful cell
manufacturers will engineer ways to overcome these challenges to increase energy density and
lower cost.
3.6.1.1 Challenges of Large Electrode Thicknesses
Dependent upon the battery chemistry and designed P/E ratio, the maximum achievable
electrode thickness (loading) may have a significant effect on the end cost and energy density of
27
-------
a battery pack. For batteries designed at low P/E ratios or for cell chemistries with low
volumetric capacities, the designed electrode thickness based on the target efficiency is often
larger than what is feasible during operation in a transportation environment. This subsection
explores some of the challenges that arise in the electrochemistry when larger electrode
thicknesses are utilized.
Argonne gained a wealth of experience in the NCA-Gr in 1.2 M LiPFe 3:7 EC:EMC cell during
the Advanced Technology Development program sponsored by the US Department of Energy
(DOE). Dees and coworkers developed a world-leading parameter set for a numerical model
through exhaustive electrochemical measurements, ex-situ characterization techniques, and
multi-scale modeling activities.26"35 The resulting phenomenological cell model founded on the
methodology originating from John Newman (UC Berkeley) will be used to evaluate the
electrochemical behavior of cells using thick electrodes.36 The coupled, non-linear partial
differential equations are solved with the finite element method using FlexPDE.
Simulated discharge capacity for the C/l and C/3 discharge rate is shown in Figure 3.6 as a
function of electrode thickness. For reference, the target positive electrode thicknesses for this
cell operating at a 5C-rate and a [V/U] = 0.8 is 142 microns. The line of 100 % capacity
utilization is also shown as a means to judge the deviation from theoretical capacity. As
expected, the C/l rate deviates more strongly than the C/3 rate with increasing electrode
thickness. The loss in capacity is a result of the cell hitting the discharge voltage cut-off, 3.3 V,
before all of the lithium has been transported from the negative to the positive electrode.
Figure 3.7 displays the normalized concentration profile of the electrolyte salt, LiPFe, at the end
of a C/l and C/3 discharge for an electrode thickness of 245 microns. The C/l discharge results
in a positive electrode starved of electrolyte salt. This transport limitation results in the cell
prematurely reaching the voltage cutoff. In order to overcome this limitation, the electrode would
need to be engineered with significantly reduced tortuosity37 or utilize an electrolyte with better
mass transfer characteristics. This behavior is exacerbated by lower temperatures, such as those
experienced during winter driving conditions. The fraction of theoretical discharge capacity
begins to lower significantly at thicknesses greater than 100 microns, 3.4 mAh/cm , at the C/l
rate and 175 microns, 6.4 mAh/cm , at the C/3 rate. The electronic transport properties of the
cathode material also play an important role in determining the current distribution within the
electrode. While the NCA material has a reasonably high conductivity, other cathode materials
have lower valued electronic conductivities and, depending on the conductive additive
properties, may have different current distributions and limitations within the electrode.
28
-------
10
• 6
o
re
a.
re
-------
The simulated AST for a 5C, 10-s discharge pulse at 60 % SOC is shown in Figure 3.8 as a
function of electrode thickness. The initial decrease in the AST is a mathematical result of
diminishing significance of the interfacial impedance as more current is passed in the same
geometric area. The AST then remains constant from 75 microns to nearly 400 microns. The
constant AST results from ohmic losses that behavior linearly with applied current. The dramatic
increase in the AST at the largest electrode thicknesses results from limitations in electrolyte
transport within the porosity of the positive electrode. This is similar to what is displayed in Fig
3.7 above during the constant discharge at the C/l rate for an electrode thickness of 245 microns.
The most significant issue for pulse power operation with thick electrodes occurs on the negative
electrode during a charge or regen pulse, Figure 3.9. The potential of the negative electrode may
drop below that of a hypothetical lithium reference electrode during a charge pulse, inferring an
undesirable side reaction of lithium plating on graphite.38 This behavior is exacerbated by
increasing electrode thickness. Operation at higher SOC and lower temperatures will also
increase the probability of lithium plating. The lithium reference electrode is taken to be in the
center of the separator layer. The two times shown in the graph, 1-s and 10-s, represent different
polarization measurements for the electrode. The 1-s value includes all of the interfacial
impedance and minor contributions from concentration polarization. The longer time value
includes additional changes in potential due to the concentration gradient in the electrolyte. The
1-s time is the more accurate valuation of the tendency of the electrode to plate lithium.
50
40 •
so •
E
o
E
£
O
J 20
10
•Full Cell
Positive Electrode
Negative Electrode
.1"
100 200 300 400
Electrode Thickness, microns
500
Figure 3.8 Calculated AST from a simulated 10-s, 5C discharge pulse for the NCA-Gr cell
couple at 60% SOC. Positive and negative electrode thicknesses are similar in value for this cell
design. Transport within the electrolyte is not limiting until the electrode thickness approaches
450 microns for these simulation conditions.
30
-------
0.1
0.0
E
= -0.1
>
>
-0.2
-0.3
-0.4
1s Pulse Time
10s Pulse Time
50
300
350
100 150 200 250
Electrode Thickness, microns
Figure 3.9 The potential of the negative electrode versus a hypothetical lithium reference
electrode located in the center of separator during a 5C charge pulse for the NCA-Gr couple.
The tendency for lithium plating appears to be significant at electrode loadings greater than 3.4
mAh/cm2 under these specific conditions. Lower continuous and pulse charge rates are most
likely necessary to ensure the safe operation and long life of this battery. Fast charging of PHEV
and EV batteries is an oft discussed value-added characteristic necessary to increase the
attractiveness of electric vehicles to the consumer. However, this fast charge requirement may
require a lower loading design to prevent lithium plating from occurring. This would not be true
for batteries based on LTO negative electrodes, or possibly even non-graphitic carbon electrodes
(e.g. hard carbon). The consequence of a lower loading design is that higher quantities of
inactive materials are used resulting in a more expensive and less energy dense battery.
A limit of 100 microns has been chosen for the default maximum electrode thickness. This
thickness represents a graphite electrode balanced to a positive loading of 3.5 mAh/cm2 and is
the largest thickness that AST measurements have been validated at Argonne. However, a low
volumetric capacity electrode, such as LMO, will result in a lower area-specific capacity as the
limit will be determined by the positive electrode thickness. One domestic OEM has suggested
that at the time of this publication, state of the art electrode loadings for PHEV applications are
less than 2 mAh/cm . This low loading level was selected based on cold-start performance, life
testing, and rate capability studies. We note that different electrolytes will likely have the most
significant effect on the transport limitations and result in different optimum electrode loadings.
Gel-based electrolytes or standard carbonate electrolytes mixed with low molecular weight
polymers are often utilized in pouch cell based batteries. While the ionic conductivity of these
systems may approach standard carbonate electrolytes used, diffusion of the salt is restricted.
31
-------
This decrease in mass transport will result in large increases in impedance during longer
discharges (constant speed highway driving). Hence, companies using an electrolyte with
sluggish mass transport will require thinner electrodes than what the model would normally
calculate based on pulse power applications. Greater electrode thicknesses may be achievable in
the future as manufacturers expend significant engineering efforts to minimize the inactive
material that lowers energy density and raises cost. The concerns over lithium plating may
remain unless new, less susceptible negative electrodes are developed that still enable high
energy density.
Table 3.2 displays a calculated sensitivity analysis for designing the battery at different target
electrode loadings to achieve 17 kWh of total energy for the NCA-Gr cell couple for two power
levels, 110 kW and 60 kW of power. The consequence of varying electrode loading at constant
power and energy is to change the fraction of open circuit voltage at which maximum power is
achieved at BOL. This quantitative comparison of electrode thickness with [V/U] is only valid
for the specific battery and cell chemistry designed in the table, while the qualitative results
would hold for all systems. An alternate but equivalent way to create Table 3.2 is to maintain a
constant value of [V/U], but vary the designed power level. As mentioned previously, private
communications suggest that current electrode thicknesses used in PHEV applications are near
50 microns or 2 mAh/cm2; however, the exact details of that cell configuration are unknown and
thus direct comparison should be conducted with caution. For the cell chemistry shown in the
table, a loading of 2 mAh/cm with 110 kW of power correlates to achieving the maximum
power at BOL at [V/U] = 0.84. Our default suggestion of using 80% of open-circuit voltage
results in a moderately less expensive battery, albeit similar in value. If a lower P/E ratio battery
was designed, then the electrode thickness (loading) would be much larger for the same designed
fraction of open-circuit voltage. The sensitivity to [V/U] decreases with lower [V/U] values
owing to the inverse proportionality of electrode loading to the P/E ratio. The calculated
designed electrode loadings for lower P/E ratios will differ the most significantly from those
used in industrial practice today. This is in part the motivation to limit electrode thicknesses to
100 microns as well as the transport limitations discussed earlier. While the standard practices of
today are important, the goal of the calculations is to evaluate the potential cost of Li-ion
batteries in the future years after improvements have been made resulting from the competitive
marketplace. Therefore, calculating a range of values will be the most instructive approach to
determining where future battery costs may fall.
Table 3.2 The effect of electrode loading on the price of a 17 kWh NCA-Gr PHEV40 battery
with 96 cells
17kWh 110kW 60 kW
[V/U]
68
72
76
80
84
88
92
mAh/cm 2
3.74
3.45
3.09
2.67
2.19
1.65
1.04
microns
96
88
79
68
56
42
27
Price, $
4305
4358
4442
4572
4780
5163
6045
mAh/cm2
7.38
6.80
6.11
5.30
4.38
3.36
2.22
microns
188
174
156
135
112
86
57
Price, $
3954
3977
4014
4091
4188
4364
4749
32
-------
3.6.2 Accounting for Parallel Cell Arrangements
The user of the design model may wish to use a parallel arrangement of cells within the larger
series arrangement of the battery pack. Several motivations exist for a parallel cell arrangement.
For example, a battery supplier may wish to only produce cells of a specific capacity. The
manufacturer may only have the equipment to produce a certain size cell or they may encounter
engineering design problems for very large cells ( > 60 Ah). Thus, a cell group composed of
parallel connected cells may be necessary to meet the energy requirements while staying within
battery pack voltage and current requirements.
When the user chooses to have cells connected in parallel, the design model calculation includes
the appropriate factors necessary to account for changes in the resistance, volume, and mass of
the battery. The end result is a lower energy density for the battery from including the cell group
interconnects and additional inactive material for each cell. Furthermore, the thickness of the
cells is reduced to 6 mm to better suit the smaller cell format.
3.6.3 Accounting for Parallel Module Arrangements
Designing a battery composed of modules in parallel cannot be explicitly simulated in the model.
However, one may approximate this scenario by simply doubling the number of modules, and
thus battery pack voltage. The simulated battery pack is nearly the same as if constructed of
parallel packs. A simple wiring change would lower the pack voltage with a proportional
increase in capacity. The overall energy and power will be the same. Subtle errors will exist in
the size of the module terminals and conductors as the calculated total current will be less in the
series arrangement than that in a parallel configuration.
33
-------
4. Thermal Management
The power and life of lithium-ion batteries are more drastically affected by the battery
temperature than they are for most other batteries including those based on lead-acid, nickel-
cadmium, and nickel-metal hydride systems. It is important that the temperature of a lithium-ion
battery be controlled at all times, even when the battery is at rest. Developing schemes for
effectively controlling the pack temperature at minimum cost will certainly be important in the
success of this technology. The most difficult problem is the removal of heat generated within
the battery, principally by ohmic heating. Avoiding excessive temperature rise during idle
periods in hot ambient conditions is also a problem. Either of these conditions might raise the
temperature to well above 40°C, which enhances degradation reactions and shortens the battery
life. Thus, maintaining the battery temperature as near the minimum temperature for adequate
power will prolong battery life. Because the battery has poor power at low temperatures, heating
the battery from a very cold condition is necessary and especially difficult for large EV battery
packs for which no assistance is available from the engine. For electric-drive vehicles to be
competitive in the market with conventional vehicles, these thermal control problems must be
solved at moderate costs and by means that do not compromise the safety of the vehicle or
battery system.
The BatPaC model has a separate worksheet for estimating the thermal parameters and
requirements for lithium batteries and for designing the thermal management system. The results
are transferred to the Battery Design worksheet to calculate the mass, volume and materials
requirements for the battery pack.
4.1 Heat Generation Rates in the Battery Pack during Driving
During driving, the heat generation rate depends on the driving cycle and the power of the
battery relative to the demands of the cycle for the vehicle being driven. As discussed below, the
heat capacity of the battery pack smooths out fluctuations in the heat generation rate. The rate
that the cooling system must handle is the average rate for the most difficult driving conditions to
which the battery pack will be subjected.
The best way to determine the maximum cooling rate requirement for the battery pack is by
vehicle simulation studies. These studies require a battery impedance algorithm that makes
possible accurate estimates of internal heating and the use of vigorous driving cycles and high-
speed driving patterns. The results of vehicle simulation studies of battery heating can be entered
on line 19 in the Thermal worksheet to override the estimated default values. However, pertinent
results are not always accessible and therefore we have provided some initial, although
dramatically simplified, estimations for heat generation.
For microHEV and HEV-HP battery packs, heat generation is intermittent and substantial
periods of little or no heat generation exist in the load profile. The model estimates the heat
generation rate for 25-kW micro-HEV batteries at 100W. For the HEV-HP battery packs, the
maximum average battery power is derived from the energy requirement, default of E = 300
Wh/mile, entered on the Battery Design worksheet (line 157). The energy use requirement for
HEV-HP batteries is estimated to take place at an average driving speed of 40 mph and involve
34
-------
the battery 50% of the time. These assumptions calculate an average estimated battery power
level of 6.0 kW. For a 50-kW battery pack, the heat generation is estimated as 5% of the average
battery power. The heat generation for a HEV-HP battery of different maximum power is
estimated to be inversely proportional to the battery pack rating as shown in Equation 4.1.
?HEV-HP = £*40*0.5*0.05* — (4.1)
"batt
For PHEV battery packs, the maximum heating condition is deemed to occur during the
declining charge period when essentially all of the vehicle power is supplied by the battery.
Thus, the factors controlling heating during operation of PHEVs and EVs are similar and may be
treated in the same way. The highest rate of continuous heat generation will occur on a very
vigorous driving cycle or when driving at sustained high speed. Continuous discharge of the
battery at constant power results in increasing impedance and increasing heat generation rate
because of solid-state diffusion overvoltage. The area-specific impedance (AST) will reach 2-3
times the 10-second AST in about five minutes accompanied by a similar increase in the internal
heat generation rate. Whether or not steady high speed driving will result in greater internal heat
generation than a vigorous driving cycle will depend on many factors including vehicle speed,
battery power, and electrode thickness. The energy usage requirement, default of E = 300
Wh/mile, is assumed to occur at a steady discharge of 60 mph in PHEVs and EVs. The vehicle
speed and energy usage assumptions provide an estimate of the maximum long-term discharge
rate. The heat generation is calculated from this power requirement and AST for energy. Finally,
the maximum heat generation rate is assumed to be 1.3 times the value at 60 mph. The heat
generated in the battery pack for these conditions is inversely related to the pack power.
4.2 Heating under Adiabatic Conditions
A factor to be considered in thermal management is the substantial heat capacity of the battery
pack. This smoothes out temperature fluctuations resulting from power bursts so that the heat
dissipation system need only handle the average heat generation rate for the most extreme
driving profiles the battery is likely to encounter. For large PHEV and EV batteries the heat
capacity of the battery will limit the temperature rise of the centerline of the cells by distributing
the heat throughout the battery until steady state is reached. For a large EV battery with power of
120 kW and energy of 50 kWh, the temperature rise under adiabatic conditions may be only
15°C or less for a complete discharge and certainly less with a cooling system even if it is only
moderately effective. For HEV batteries, which have high power-to-energy ratios, the main
effect of the heat capacity will be to smooth out temperature fluctuations.
4.3 Active Cooling Systems
There are several choices of coolant that have been considered for cooling battery packs
including air from the cabin, which may be heated or cooled, water-ethylene glycol solutions and
dielectric liquids such as transformer coolants. Air is the least expensive, but it is less effective
than the liquids because of its poor conductivity, the need for large flow passages and high
pumping power. Dielectric liquids are expensive, but have the advantage of being compatible
35
-------
with terminals and other parts at electrical potential. Water-50% glycol solution is inexpensive
and has good conductivity; we have selected it as the coolant for this study.
We selected a general cell and battery design that can be adapted to all of the electric-drive
batteries from micro-HEVs packs to EV packs (section 2). This design incorporates a
hermetically sealed module closure. Unfortunately, the enclosure does not have sufficient surface
area to be cooled effectively by air. This design can be effectively cooled by liquids and requires
that only the module terminals and connectors be protected from contact with a conducting
coolant, thus accommodating water-glycol coolant. In contrast, cooling individual cells would
require flow passages between the cells, which would add to the pack volume. That design
feature would permit air cooling for microHEV and HEV-HP batteries. For larger batteries
requiring liquid cooling, flow passages between the cells would probably require the use of a
dielectric coolant because of the difficulty of protecting so many cell terminals from a
conducting coolant. Alternatively, elaborate flow channels with engineered seals may be used to
contain an aqueous based heat transfer fluid.
4.3.1 Heat Transfer from Cell to Module Wall
As described in section 2, the cells transfer heat to the cooled walls of a hermetically sealed
module with the aid of an aluminum heat conduction channel. Some of the heat is transferred
through the sides of the cell to the channel and from there to the module wall. The remainder is
transferred directly through the seal edge of the cell to the conduction channel flange which is in
contact with the module wall. Calculation of heat transfer in this two-dimensional array through
several materials is complex requiring a numerical model. The spreadsheet iterates several
hundred times in reaching a solution, each resulting in a slightly different cell design. Thus, it
would be impractical to imbed a numerical model directly, which may increase the total
calculation time to many minutes. Instead, a software program based on the finite element
method, FlexPDE 6.15 by PDE Solutions Inc., was employed to calculate heat transfer rates for
70 cell configurations. The resulting simulations were empirically correlated so that simple
equations occupying a few cells in the spreadsheet could rapidly calculate the heat transfer rate
with only a small error.
An important requirement for calculating heat transfer rates within the cell is to estimate the
composite conductivities of the cell layers both parallel to the layers and across the layers. The
resulting conductivities vary considerably with the relative thicknesses of the layers as shown in
Table 4.1, for which the results are consistent with the literature.39"43 These values for
conductivities and a range of cell dimensional parameters (Table 4.2) were employed in selected
arrangements for calculating heat transfer rates with the FlexPDE model for 70 representative
cells that covered a broader range of variables than is needed for practical cells. For each of these
cells, the FlexPDE model calculated the temperature difference between the cell center and the
module housing per unit of heat generation, AT/q (°C/W), and the fraction of the total heat that
was transferred through the edge of the cell, qe/q. The balance was transferred through the side of
the cell to the aluminum conductor, qs. The division of the heat transfer into two routes is
represented by the equation:
q/AT = qe/AT + q/AT (4.2)
36
-------
Estimated values for g/Ar and g/Ar were determined by empirical correlation of the results
obtained for the calculation of the 70 cells by the FlexPDE model with the result shown in Fig.
4.1. Empirical values of these estimated values resulted in the equations:
°-58
9r L2nr°-75
Lceii W
/ A-T 7/cior 0.55, 0.58T -0.21-,T,r0.7T 0.72
qs/AT = 1628ky kx Lceii W LAi
(4.3)
(4.4)
Table 4.1 Sample calculations of composite thermal conductivities of cell structures across layer
and parallel to layers
Layer Thicknesses, microns
Positive foil
Negative foil
Positive coating
Negative coating
Separator
Total bicell structure
Thermal Conductivities, W/cm-K
Aluminum
Copper
Positive coating
Negative coating
Separator
Across layers, kx
Parallel to layers, ky
Celll
20
12
30
40
20
212
2.0
3.8
0.013
0.013
0.0020
0.00689
0.4127
Cell 2
20
12
75
100
20
422
0.2.0
3.8
0.13
0.13
0.0020
0.00689
0.4127
Cell 3
20
12
150
200
20
772
0.2.0
3.8
0.13
0.13
0.0020
0.01045
0.1228
Cell 4
20
12
220
300
20
1112
0.2.0
3.8
0.13
0.13
0.0020
0.0.01112
0.0892
Table 4.2 Range of parameter values for calculating heat transfer rates in FlexPDE model
Conductivities, W/cm-K
Across layers, kx ~~l (a)
Parallel to layers, ky J
Cell edge, ke
Cell Dimensions, cm
Cell thickness, Lceii
Cell width, W
Cell edge thickness, Le
Aluminum conductor thickness, LAI^
Parameter Levels Evaluated
1
0.00689
0.4127
0.10
0.6
8
0.1
0.03
2
0.00689
0.4127
1.0
12
0.06
3
0.01045
0.1228
1.4
18
0.10
4
0.0.01112
0.0892
kx and ky values are calculated as in Table 4.1 and were, thus, paired together in the Flex
PDE model calculations.
^The total conductor thickness consists of the conductor thickness itself plus twice the thickness
of the aluminum layer within the pouch material.
37
-------
2,0
1.5 -
a
u
C
IA
OJ
a*
(IJ
i.o -
0.5 -
0.0
0.0
0,5 1.0
Calculated Resistance, AT/q (oC/W)
1.5
2.0
Figure 4.1 Plot comparing the estimated resistance to heat transfer from the cell center to the
cooled surface of the module to that calculated by the FlexPDE model.
The average error in the estimated AT/q compared to the values calculated by the FlexPDE
model for the 70 cases studied was 6.0% and the maximum error was 13.0%. This accuracy was
deemed to be satisfactory in that for all practical battery designs, the error will be only a fraction
of a degree Celsius.
4.3.2 Heat Transfer from Module Wall to Flowing Coolant
Heat may be transferred to and from the modules by flowing fluid directly across the module
casing. For ease of communication, the focus of this discussion will be on cooling of the module
rather than heating for cold-climate operation. In theory, both liquid and air may be used as the
heat transfer fluid. However, the stacked cell design we have selected has minimal exposed area
relative to the overall volume of the cells. Our calculations have shown that liquid cooling is the
only feasible option for this particular module design. The superior heat transfer of liquids
(density and heat capacity) allows for implementation of this compact design without exposure
of the individual cells to the heat transfer fluid. A design of this kind should result in a lower cost
and higher energy density battery than a different design that cools individual cells.
38
-------
The model directly calculates the temperature drop between the module wall and the heat
transfer fluid for a set pressure drop, fluid (coolant) temperature rise, and fluid physical
properties. A 50/50 ethylene glycol, deionized water (EG/HiO) mixture was selected based on
the low cost and contemporary use in coolant systems. The default pressure drop was taken to be
10 millibar, but may be changed by the user if desired. The gap in which fluid flows is sized to
maintain the target pressure drop without going below a minimum gap height of 3 mm. A
coolant temperature rise of 1 °C was selected to establish a mass flow rate, but also may be
changed by the user.
Calculation of the heat transfer coefficient allowed for determination of the temperature
difference between the module and average coolant temperature. A schematic of the flow
passageway and change in temperature profile with distance is shown below in Figure 4.2. The
outer wall of the flow passage is assumed to be perfectly insulated. The inner wall (module
casing) is assumed to have a constant heat flux perpendicular to the wall. Laminar flow was
assumed to simplify the calculation of the velocity profile (parabolic).
t
t
t
Figure 4.2 Heat transfer from the module wall to the laminar flow heat transfer fluid. The
temperature profile of the fluid is shown at different lengths down the path.
Frequent use of dimensionless numbers was necessary to adequately correlate the numerical
results into a generally useable form. We define the Reynolds, Prandlt, Graetzl, and Nusselt
numbers here for completeness.44 The Reynolds number, Re, is the ratio of inertial to viscous
forces. The Reynolds numbers were always less than 1000 confirming laminar flow. The Prandlt
number, Pr, is the ratio of the momentum diffusivity to thermal diffusivity. The Prandlt number
for the EG/H2O mixture is approximately 38. The Graetz number, Gz, is directly proportional to
the product of the Reynolds and the Prandlt numbers. Moreover, the Gz value is inversely
proportional to the distance down the fluid flow path, /, resulting in higher values near the start
of the flow path. Finally, the Nusselt number, Nu, is the ratio of the convective to conductive
heat transfer. Here uave is the average fluid velocity, dn is the hydraulic radius (twice the flow
gap), and // is the viscosity. The heat capacity, cp, thermal conductivity, k, and heat transfer
coefficient, h, are the critical heat transfer values. The mass flow rate, G, and the width of the
channel, W, are the remaining parameters.
Re =
Pr =
(4.5)
(4.6)
39
-------
Gz = 2-L = 2RePr (4.7)
kWl I
Nu=-^ (4.8)
k
Coupled momentum and heat transfer has been solved previously by determining a number of
the eigenvalues for a series solution of a vast number of various geometrical configurations
related to pipe, duct, and parallel plate flow.45'46 We have chosen to reach the solution
numerically and then fit a correlation between the Graetz number and the mean Nusselt number.
The empirical form provided by Nickolay and Martin provides an accurate means of correlating
the results over many orders of magnitude.47 The correlation, shown in Equation 4.9, relates the
Graetz number and the limiting solution, Nuoo = 5.385, to the mean Nusselt number. Then the
mean heat transfer coefficient may then be directly calculated from Nu. Here n and C; are fitting
parameters.
Nu = [(Nu J* + (CjGz173 )" f (4.9)
The numerical model was solved with the finite element method using FlexPDE software. We
note that the bulk or "cup mixing" fluid temperature in Equation 4.10, the average temperature of
the fluid normalized by the fluid velocity profile, was necessary to reach the proper values.
(4.10)
The following important assumptions were used to reach a solution.
1. Flow of incompressible heat transfer fluid is laminar
2. Thermal diffusion is allowed up and down stream of the heat transfer (for convergence)
3. Boundary conditions: dT/dy = 0 at insulation; q = constant at module casing
4. Negligible radiative energy transfer
5. Steady state conditions reached
Figure 4.3 displays the temperature profile between the module casing and the insulated wall for
various distances along the flow channel. The average temperature of the fluid has risen 1 °C at
the end of the flow path even though the maximum and minimum temperature is separated by
nearly 5 °C. The simulated change in average temperature down the length of the flow channel
allows the calculation of the average heat transfer coefficient and thus Nusselt number. The
correlation, Eq 4.9, determined from various simulations conditions is shown in Figure 4.4. An
excellent fit is obtained allowing for implementation of the correlation into the design and cost
model. This correlation now enables efficient and accurate calculations of the heat transfer
coefficient to be made in the spreadsheet informing the user of the effectiveness of the thermal
management in the design.
40
-------
Dimensionless distance
down fluid path
-1 -0.5 0 0.5 1
Dimensionless distance from centerline
Figure 4.3 Temperature profile in the heat transfer fluid for various fractions of the
dimensionless path length.
100
0)
.0
E
3
C
= 10
0)
C
«
Q)
• Calculations
— Correlation
Nu« = 5.385
n = 3.592
G! = 2.255
0.1
1.0 10.0 100.0
Graetz number, Gz
1000.0
Figure 4.4 Correlation of model simulation results relating the Graetz number and mean Nusselt
number for laminar flow between an insulated surface and the module casing.
41
-------
In general, the heat transfer from the module is improved by increasing the contact area and
increasing the fluid flow rate. The contact area may be increased by using cells with a higher
aspect ratio. This also results in a smaller temperature gradient within the cell as discussed
previously in section 4.3.1. Increasing the fluid flow rate is accomplished by using a lower
temperature rise and/or a larger target pressure drop. The gap height may prevent a change in a
single parameter from having a significant affect on the temperature drop. Physical limitations of
implementing a cooling system should be considered when moving to higher flow rates and
pressure drops. The user should note that raising both of these parameters will increase the cost
of the battery design in ways that the model does not consider (e.g. more expensive pump,
increasing structural integrity, etc).
For high speed driving or very aggressive driving cycles, the temperature difference between the
surface of the cooled module surface and the bulk of the coolant may become fairly large
(>10°C). This coupled with the temperature rise within the cells could result in too high cell
centerline temperatures. This result can be avoided by controlling the inlet coolant temperature
as a variable that is adjusted in a classic cascade automatic control system to control the module
wall temperature at the desired value. Thus, the temperature rise at the center of the cells will be
essentially held to that resulting from conduction within the cell and will not be greatly
influenced by the temperature rise in the coolant.
4.4 Cooling and Heating Required to Maintain Pack Temperature
When parked in the sun for several hours, the internal vehicle temperature and, thus, that of the
battery may become so hot that the life of the battery is reduced. To avoid this, the vehicle air
conditioning system may be actuated intermittently to cool the battery. By allowing the
temperature of the battery to fluctuate by several degrees, it is only necessary to actuate the
cooling system about once per hour for a few minutes. For a set of target temperatures with a
difference of 25°C and with the default insulation thickness (10 mm) and default thermal
conductivity (0.00027 W/cm-K), BatPaC calculates the average cooling requirement to be about
60 W for PHEV-40 batteries. The performance coefficient of the vehicle air-conditioning system
might reduce the actual energy draw to less than half that, but heating of the system outside of
the battery during the hour-long downtime periods would be a counter-acting factor. The BatPaC
model calculates the energy required for cooling of all types of electric-drive vehicle batteries.
However, most HEVs may not have electrically driven air-conditioning units and some other
method might be needed to avoid very high battery temperatures during parking such as thicker
insulation and fan cooling.
If the battery is to deliver full power at startup, it must be at a temperature of at least 5°C. This
minimum temperature can be maintained by heaters and circulation of the glycol solution.
BatPaC calculates the amount of power required to maintain the battery temperature for any set
of battery and ambient temperatures. PHEV-40 batteries would require about 50 W of heat to
maintain the battery temperature at 20°C above that of the ambient under steady-state conditions.
During recharging this should be easily done for 20 hours at a cost of $0.10 for an energy cost of
$0.10 per kWh. If the vehicle is not at a source of power for recharging, limited energy (say 1-2
kWh) can be drawn from the battery (if not blocked by a switch actuated by the driver) and then
42
-------
automatically shut off after maintaining the battery temperature for one to three days, depending
on the ambient temperature.
4.5 Heat-up from Cold Ambient Conditions
All of the batteries for the various types of electric-drive vehicles will occasionally be exposed to
very cold temperatures, which will require special heat-up procedures. All but the EV batteries
can be heated with the aid of the engine. This can be done with electric heaters operated from
power taken from the generator or from glycol solution from the engine cooling system. If the
latter, it might be prudent to isolate the engine coolant from the battery coolant by means of a
plastic heat exchanger.
Another method of heating the battery is by means of the electric heaters that should be available
for maintaining the battery temperature (section 4.4). BatPaC calculates the amount of heat
needed and the time required with suitable heater power. For PHEV-40 batteries, about 15
minutes is required with 2-kW heaters. This method of heating will be slower than with the
engine coolant and even the latter would result in some delay before the battery is capable of full
power.
To avoid delay in starting vehicles from a cold startup, the driver could initiate heating by means
of a remote device, which in the future may be a telephone. By this means, heating could be
initiated either from heat drawn from the engine or electric heaters. Remote initiation of heat-up
would be especially important for an EV away from a charging station in that no engine is
available to assist heating and the large size of the battery would result in a long heating period
with electric heaters. The BatPaC model estimates the time under these conditions.
43
-------
5. Modeling of Battery Pack Manufacturing Cost
5.1 Approach
The manufactured cost of a battery pack is calculated with input from the design information
generated in modeling the cell and battery pack performance. The design modeling determines
the annual materials and purchased items requirements. The manufacturing cost is then added to
these materials costs, along with a warranty cost, to reach the unit cost of a single battery pack.
The manufacturing costs for the designed battery are scaled from a baseline plant. The baseline
plant was designed for a battery of intermediate size and production scale so as to establish a
center-point for other designs. The baseline plant accounts for the size, speed, number of units,
direct labor, and depreciation of the capital cost for each processing step. These costs are
adjusted to meet the requirements for a plant producing the battery under study. The process
expenses are summed with the additional costs of operating the manufacturing facility. These
costs include launch costs, working capital, variable overhead, general, sales, administration
(GSA), research and development, depreciation, and profit. Additionally, the costs for the
thermal management, battery management system, and disconnects have been estimated to
provide the total cost to the OEM for the integrated battery pack.
In this analysis, all costs are evaluated for 2020 when large battery manufacturing plants are
built. All dollar values are brought back to 2010 with allowance for inflation. In other words, all
costs and prices are in 2010 dollars. Some materials and battery manufacturing costs are lower
than recent values, where we judged that processing improvements for high volume production
of materials would reduce costs.
The baseline manufacturing plant was calculated for an annual production rate of 100,000
batteries. The cost model accounts for different scales of manufacture by recalculating the costs
of each individual step in the manufacturing process. The changes in capital and operating costs
will change the calculated unit cost of the battery pack. The parameters were determined to
provide reasonable estimates for manufacturing rates of 20-500 % of the baseline rate. Thus, for
a plant that is far different in size from the baseline plant, for instance a pilot plant having an
annual production of only 5,000 battery packs per year, the estimate from this study would be
expected to be less accurate than if determined in a study dedicated to that purpose.
To simplify the cost calculations, it was assumed that all hardware items for the cells, modules
and battery will be purchased from a vendor specializing in similar products. The costs for these
items were estimated to be a fixed value plus an additional value proportional to the weight of
the item, which is calculated during the battery design. In mature manufacturing plants in 2020,
toward which this study is directed, some items which are assumed to be purchased in this study
might actually be internally manufactured from raw materials. This would increase the number
of processing steps needed in our manufacturing simulation and thus complicate the cost
calculations. Assuming that some parts would be purchased if they would actually be produced
from raw materials would tend to underestimate capital and labor costs and overestimate
purchased items expenses. However, the net effect would be a very small change to the overall
unit cost of the battery pack.
44
-------
5.2 Materials Costs, Purchased Items, and Pack Integration
The end battery pack cost depends significantly on the cost of both the active and inactive
materials that compose the design. In this subsection, the assumed material costs and the
rationale behind them are presented. We provide for means to scale the materials cost with
production volume using the same method used for processing rates as discussed in section 5.4.
In general, the materials costs will be largely insensitive to production volume since we have
assumed a high volume market already exists. Only the negative and positive electrode active
materials are assumed to have a minor benefit for larger scales of production. While we state
suggested materials costs and sensitivity to production scale, the users of the cost model may
enter any value that he desires.
5.2.1 Battery Specific Materials Cost
The largest contributions to the materials cost of the battery are from the following components:
positive and negative electrode active material, separator, electrolyte, and current collector foils.
The choice of the materials often defines the size and performance of the battery as well as the
cost. Many different variations of materials are possible in the Li-ion family of chemistries.
However, we have chosen to focus on the different available positive electrode materials with
less attention on the negative electrode. This reflects the current research and manufacturing
activities. The separator and the electrolyte are also both active areas of development. However,
the following battery designs are based on a single electrolyte and separator combination.
Including the cost and effect of additives and enhanced separators is beyond the scope of this
work. The user is always able to modify the dimensions, cost, and ASI that may be required to
account for changes in these materials.
The price of specific battery materials is of some debate. The values presented in Table 5.1
compare our suggested costs to those reported recently in the open literature. Our values, as well
as the others in the table, are derived from conversations with material, cell, and original
equipment manufacturers. The sources are commonly anonymous and the accuracy of the values
is generally unknown. We present the comparison of published values so that the user of the cost
model may appreciate the accepted range of values for commonly used materials.
5.2.1.2 Positive Electrode Active Materials
The cost of positive electrode materials is driven to a large extent by the cost of the raw materials
from which it is made. The archetype Li-ion positive electrode material, lithium cobalt oxide
(LCO), was the original material commercialized in Li-ion batteries for consumer electronics.
LCO has many excellent characteristics but is not considered a viable choice for use in Li-ion
batteries for automotive applications. One of the largest drawbacks of LCO, other than safety
concerns, is the high and volatile cost of the cobalt. While tolerable in the consumer electronics
market, the cost is too high for use in an automobile battery. Many other materials are in a
commercially viable state of development and are currently utilized in Li-ion batteries produced
today (Table 5.1) such as lithium manganese spinel oxide (LMO) and lithium nickel manganese
cobalt oxide (NMC).3'6 The relative advantages and disadvantages of each material will not be
discussed here.
45
-------
S ^ 2
Q ^ OO
*
in
(N
(N
(N
in
in
i
^
o
ON
5
o
(D
ON (D .-., S
(D
toOtoOtoOtoOtoOtoOtoOtoOtoOtoOtoO ^ ^g ^-H
o
in
•°,
•o"
_o
•_
a
(U
CO
£H
'o
-J— »
8
o
CO
D
T3
O
-4— *
cd
O
£
-4— »
CO
O
O
jO
CO
O
CO
-4— »
(U
Q
2
CS
H
Abbreviation
hemistry
U
Material
0
s
'si'
o
X
~S
a"
j
CD
o
cd
1
CO
CD
CO
CD
03
toO
03
s
LNMO
0
wo
1
wo
0
Z
,_)
*
CD
O
03
O
13
CO
in
!
2
CD
,_)
CD
O
-s
03
O
CD
Phospholivi
U
Z
(N
0
0
CD
^
CD
O
U
0
OO
CD
Z
,_)
*
CD
1
O
CD
O
-a
§
U
S
z
(N
0
wo
O
6*
U
CO
a
S
s
q
,_)
*
CD
"S
.g
s
o
CD
O
1
i— H
^^
^^
U
S
z
(N
0
wo
O
U^
si
q
,_)
*
CD
"S
4^
s
o
CU
O
1
03
U
S
z
1
(N
O
N
U
N
j?
CO
O
1
7<
*
CD
O
03
O
"S
(H
CD
03
^
1
O
(N
0
o
y
3
CD
1
13
o
CD
X
O
1
03
O
u"
1
Graphite an
O
(N
<5
wo
E^
3
•S
o
03
'co
Titanate spi
U
U
a
£*
°""^ Cu
H^ ^^
5 ^
^ SH
(N Q"
~ £
Electrolyte
Separator
CD
&
&
O
U
r^
o
o
8
u
luminum
<;
r^
o
o
8
u
o
U
1
o
o
'o
o
"O
-Q
O
U
o
CC
1
2
'ecipitation o
OH
6
C*
Ik
.3
o5
CC
"O
o
( \
* The cost of (
_u
3
'c/3
&
o
r
O
G°
'G
u
** Cost repres
-------
The amount of cobalt and nickel, as well as ease of manufacture, controls the end price for a
positive electrode material. For example, the NMC-441 is less expensive than the NMC-333 as
the cobalt quantity is significantly reduced. The market price for cobalt and nickel metal varies
dramatically from year to year. Reducing the quantities of these materials in the positive
electrode will reduce the total price and price volatility. Researchers at TIAX LLC have treated
this variation and shown the significant effect on end battery cost. The average traded metal
prices for the last 20 years is 48 $/kg and 15 $/kg for cobalt and nickel respectively. These
numbers are based on historical prices for the metals as collected by the United States Geological
Survey (USGS).48 The metal prices are indicators for how the intercalation material cost will
relate when compared to one another. The fact these materials are not earth abundant means they
will not benefit as much as other materials from increased scales of production.
We employ the relationship in Equation 5.1 to systematically calculate the cost of the transition
metal based spinel and layered compounds. The final cost, C, of the lithiated oxide depends on
the baseline cost, Co, and the contributions of the lithium and transition metal raw materials, Q.
The molar stoichiometry, Xi, is transformed to a mass basis with the molecular weight of the raw
material, MWi, and the final product, MW. The baseline cost is the sum of the cost for processing,
additional raw materials, and profit margin associated with the manufacture of the materials. We
assume a baseline cost of $7/kg for single metal containing oxides (LMO and LCO) and $16/kg
for the co-precipitated metal oxides such as NMC-333 and NMC-441. NCA is known to have a
slightly lower yield and requires additional raw materials resulting in an assumed Co = $20/kg.
The costs for Li, Ni, Mn, and Co are taken to be 0.22, 0.85, 0.15, and 5.00 $/mol respectively.
Aluminum is assumed to be similar in cost to manganese for these calculations. One may directly
translate these numbers to raw materials costs resulting in $6/kg for I^COs, $5.5/kg for NiSC>4,
$32/kg for CoSCM, and $l/kg for MnSO/t. Calculations are also shown in Table 5.1 using
$2.5/mol for cobalt as a simple demonstration of the effect of cobalt on the end material cost.
x{C{MW{ (5.1)
In general, earth abundant elements should be the dominate transition metals used if a low
positive electrode cost is desired. Both iron and manganese are abundant and inexpensive
transition metals for intercalation materials. Comparison of the iron phosphate, LFP, to
manganese spinel, LMO, reveals how processing costs contribute to the end price of a material.
LMO is relatively easy to manufacture. In contrast, LFP requires a reducing atmosphere and a
carbon coating step to reach the end product. The increased complexity in the manufacturing
process is realized in the price. However, one could argue that the manufacturing cost will
decrease with increased knowledge from larger scales of production.
5.2.1.2 Negative Electrode Active Materials
While several negative electrode materials exist for Li-ion batteries, carbon materials in the form
of graphite and/or hard carbon are still used in the vast majority of commercial cells. Graphite
offers the greatest energy density while hard carbon is said to enable high rate capability with
decreased risk of lithium plating (an undesired side reaction) during high charge rates. We have
chosen synthetic graphite as a generic carbon electrode in our model. The price of graphite is
47
-------
much better understood than that of most of the positive electrode materials. However,
significant differences in cost and performance will exist between synthetic, natural, and coated-
natural graphite. The method of production and necessary heat-treatment will control the end
cost. Graphite, although in different purity grades or microstructure forms, is used in many
industries. This is in stark contrast to the positive electrode materials.
The lithium titanate electrode, LTO, offers an interesting option compared to graphite. Unlike
graphite, LTO operates within the stability window of the electrolyte. The higher electrode
potential, 1.5 V vs Li, dramatically reduces or eliminates the formation of the solid electrolyte
interphase (SEI). As a result, nanoparticle-based LTO may be implemented without concerns of
increased side reactions with the electrolyte. The reduced dimensions increase the available
surface area for reaction while simultaneously shortening the diffusion length. Both of these
factors combined with the lack of SEI dramatically reduce the impedance of the electrode.
5.2.1.3 Electrolyte and Separator
The electrolyte used in this model is based on a lithium hexafluorophosphate salt, LiPFe,
dissolved in a carbonate based solvent system. The carbonate solvent system is a blend of
ethylene carbonate, EC, and a linear carbonate such as ethyl methyl carbonate, EMC, or dimethyl
carbonate, DMC. Other chemical additives may be used to lower the capacity and power fade of
the battery over time. Polymers may be added to the electrolyte as either a minor or major
component. This is not discussed in any further detail in this work. The price of 18 $/kg, about
22 $/L, is only for the base electrolyte (i.e. no additional additives).
The separator is typically a porous membrane based on polypropylene (PP) and sometimes
includes a polyethylene (PE) middle layer. PP and PE are very inexpensive raw materials and
thus the suggested cost of $2/m2 is in large part due to the manufacturing process required to
form the porous network in the membrane. As competition and scale of manufacture increase,
the prices of the separator may fall closer to $l/m . However, the cost of improved technology
ij
may offset some of this cost reduction, so we have retained our cost estimate of $2/m .
As safety is a major concern for Li-ion batteries, the separator plays a key role in isolating the
oxidant from the fuel. If the two charged electrodes contact each other (short), then a run-away
reaction is possible. Separators have been designed to "shut-down" or melt at key temperatures.
The middle PE layer is the shut-down feature in our proposed separator. Ceramic coatings have
also been used to ensure structural integrity. Many other approaches are being developed to
increase the safety of Li-ion batteries. The user of the cost model should account for the
increased technology in the price and dimensions of the separator as needed.
5.2.1.3 Current Collector Foils
The current collector foils are based on copper metal for the negative electrode and aluminum for
the positive electrode. However, the LTO anode material, because of its high voltage relative to
lithium, enables the use of aluminum as the negative electrode current collector. The price of
these foils is based on raw materials and manufacturing costs. The aluminum foil is produced by
rolling of thicker stock foils into thinner and thinner sheets. On the other hand, copper foil is
48
-------
more likely to be produced through an electrodeposition process. The foils are 12 microns and 20
microns thick for the copper and aluminum current collectors respectively. The foils used in
batteries have additional requirements beyond the cheapest product available. Surface treatments
are often necessary to promote adhesion of the electrode to the foil surface. In addition, alloying
of the foil may be necessary to achieve the required material properties for long life.
The raw material contributions to the foil price will vary with the volatility of the market price
for the metals. Figure 5.1 displays the metal ingot price contribution on a $/m basis. These
numbers are based on historical prices for the metals as collected by the USGS.
48
The values for both aluminum and copper tend to vary significantly over the time period
examined. The price for copper is more volatile and always more expensive than aluminum.
Analysis of Figure 5.1 reminds the user of the cost model that cost quotes are only valid for a
short period. As the market price for raw materials changes, so will the price for the finished
product.
r\
Conversations with manufacturers and suppliers lead us to take a price of 1.80 and 0.80 $/m for
battery grade copper and aluminum foil respectively. We point out that the current metal ingot
price is only a small contribution to the end foil price being about 16 % of the aluminum foil
price and 23 % of the copper foil price. Thus, a doubling of the ingot prices would only
moderately increase the foil prices.
u.au •
Don _
.OU
*E
t/3- n vn -
O
2 n en -
0>J U.DU
_c
•** n £n -
^\>CrO^c,^^JJ-^
85 1990 1995 2000 2005 20
Year
10
Figure 5.1 Metal ingot cost contribution to the current collector foils over a 20 year period. The
average for that period is also shown. All costs are in 2010 US$.
49
-------
5.2.1.4 Additional Electrode Components
The binder and conductive additive in the positive and negative electrodes add a small but real
cost to the battery. The conductive additive, more common for the positive electrode, was priced
at 6.80 $/kg for a high purity and moderate surface area carbon black material. The binder,
perhaps PVDF or CMC based, is assumed to be 10 $/kg. The W-Methyl-2-pyrrolidone (NMP)
solvent for the PVDF binder is estimated to be 3.20 $/kg. Most of the NMP is recovered after
evaporation and recycled as discussed in section 5.3.3. Only the small amount lost in processing
need be replaced. No cost is assumed for water used to in the electrode slurry processing.
5.2.2 Purchased Items Cost
Table 5.2 lists the purchased items for the cell module and battery jacket. The cost of a SOC
controller for each cell, or group of parallel cells, is $2.50 plus a small factor for the cell capacity
(Ah), which allows for higher cell balancing currents for larger cells. The other components cost
a fixed amount plus an additional factor, which is proportional to their mass, mi. The cell
negative terminal and parallel cell group connection are both made from nickel plated copper
sheet and thus have the same cost equation. The costs shown for the terminals include an
allotment for isolation tape that is necessary to protect the electrical connection. The bus bar is a
fixed cost and is only charged if a single row of modules is used. A single row of modules
requires a bus bar in order to locate the positive and negative terminals at the same end of the
battery.
Table 5.2 Cost equations for purchased items
Component, /
SOC controller
Cell positive terminal
Cell negative terminal
Cell container
Aluminum heat conductor
Parallel cell group connection
Module terminals
Balance of module (casing)
Module interconnect
Battery terminals
Bus bar for one module row
Battery jacket
Cost Equation, $/unit
2.50 + 0.01C
0.25 + 4m/
0.25 + 6nn
0.20 + 3mf
0.10 + 4mr-
0.25 + 6mt
0.75 + 5nn
1.00 + 3mr-
1.00 + 5mr-
15.00 + 0.02/toto/
20.00
30.00 + Imi
Cost per unit
cell or parallel cell group
cell
cell
cell
cell
parallel cell group
module
module
module
battery pack
battery pack
battery pack
5.2.3 Pack Integration Cost
Various additional components and thus cost are necessary to integrate the battery into the
electric drive system, which adds cost. While it is not clear what should and should not constitute
the cost of the "battery pack," we present these additional items in Table 5.3 in an attempt to be
complete. The model treats these values as a cost to the OEM for integrating the battery into the
vehicle. After all, the price of the entire system is of interest to the final consumer of the product.
50
-------
The general conclusion is that the pack integration costs have the largest consequence for the
smallest batteries. The worst case is perhaps that of a small PHEV10 battery. The integration
costs of a PHEV10 battery carry the burden of charging from the grid, but provide only a modest
electric drive benefit. The fixed cost of pack integration may amount to 25 % of the battery pack
total even without considering the costs of additional powertrain components. Clearly,
understanding the entire cost of the electric drive system is of importance to evaluating the true
value of the electrified vehicle to a consumer.
Table 5.3 Costs to integrate battery pack into vehicle drivetrain. $/kW numbers reflect maximum
kW of cooling or heating required.
Battery Management System MicroHEV HEV-HP PHEV & EV
Current and voltage sensing, $ 40 70 100
Module controls, $/module 10 10 20
Disconnect Units
Auto, disconnect, $ 50 70 200
Manual disconnect, $ 15 15 15
Thermal Management System
Baseline thermal system, $ 30 80 120
Additions to AC system, $/kW 40 40 40
Heating system, $/kW 20 20 20
5.2.3.1 Battery Management System
The battery management system (BMS), in our assumed battery design, integrates the modules
and battery into the overall electric drive system. The BMS includes measurement and control
features such as the following:
• Measurement of battery pack current and voltage
• Balancing of the module voltages (cell balancing done within module)
• Estimation of battery pack state-of-charge (SOC) and state-of-health (SOH)
• Estimation of module SOC and SOH
• Monitoring and signaling of battery thermal management
The cost of the BMS will scale with magnitude of battery current and with the need to charge
from the electrical grid. Therefore the PHEV and EV batteries will have a higher burden from
the BMS. The micro-HEV is assumed to have less complicated management and thus less cost
than the HEV-HP.
5.2.3.2 Manual and Automatic Disconnects
The manual and automatic disconnects integrate a high-level of safety and electrical management
into the electric drive system. The manual disconnect breaks the current flow pathway from the
high-voltage terminals to the outer system allowing for the safe service of the vehicle and battery
pack. This disconnect is designed to be operated when the electrical system is de-energized. The
automatic disconnect is much more complex. This unit contains the connections for the high-
voltage system to the rest of the vehicle's electrical system: drivetrain, grid charging (if
applicable) and accessories (high and low voltage). Fuses are present as a hard-wired safety
51
-------
device to prevent unusually large current spikes from damaging the battery or drivetrain.
Multiple contactors are used to appropriately channel electrical current depending upon normal
operation or grid-charging. Engaging the contactors requires that multiple safety interlocks are
established including isolation of the high voltage bus from the vehicle chassis and an inertia
based sensor (crash protection). Finally, a small circuit is provided to prevent arcing of the
current across the high-voltage contactor when the high-voltage circuit is closed.
The relative cost of the automatic disconnect amongst the various battery designs is driven by the
pack voltage, maximum battery current, and the need for charging from the grid. The voltage of
the pack has a significant effect if a 42 V micro-HEV pack is considered. For this system,
electrical safety regulations allow a less complicated system to be used. Requiring higher battery
currents generally increases the cost of electronics and conductors. The additional complications
arising from grid-charging adds a significant additional cost to the PHEV and EV systems. It is
unclear to the authors at this time what other factors may enable a lower burden of external
safety controls. These additional costs in the automatic disconnect unit have the most
pronounced effect on the cost of smaller batteries, as the burden amounts to a significant fraction
of the total cost.
5.2.3.3 Balance of Thermal Management System
The thermal management of the battery is crucial to meeting the life and safety requirements of
transportation applications. The complexity of this system must be minimized to reduce the cost
and size burden on the vehicle. Our assumed design format uses liquid thermal management for
all vehicle battery types. In practice, current microHEVs and HEV-HPs are more likely to be
cooled by blowing air. Air-cooling is generally less expensive than liquid cooling, but is less
effective, requires larger system volume, and may result in substantial background noise.
Furthermore, directly cooling the pouch cells with air increases the magnitude of oxygen and
water permeation through the seals resulting in deleterious effects to the fifteen year life of the
battery. Air cooling is not feasible with our current assumed battery format (stacked cells sealed
within module). The cells would need to be separated to allow air to flow past at least one side to
achieve sufficient heat transfer as earlier versions of this model had used.1 Future versions of the
model may include an option to select either air or liquid cooled. However, the main goal of this
model is to explore the effect of battery performance and materials chemistry on the price of the
battery to the OEM.
A single refrigerant compressor is used for both the cabin air and the battery cooling
applications. Likewise, the same radiator and fan as the cabin cooling will also be used for the
battery cooling refrigerant. Most OEMs appear to use an electric compressor for all full-HEVs
(HEV-HP) and PHEVs/EVs that are liquid cooled. The incremental cost for an electric
compressor at high volume in year 2020 will likely be $200-300 more than the commonly used
$100 belt driven compressor. We do not include this incremental cost in our thermal
management system cost; however, we state it here for completeness. The additional cost to the
compressor for the battery cooling capacity is insignificant compared to burden of transitioning
to the electric compressor. Experts in the field have informed us that the electric motor and high-
voltage invertor are the largest contribution to the incremental cost of the electric compressor.
52
-------
An expansion valve on the refrigerant line and a heat exchanger (chiller) transfers the thermal
energy from the heat transfer fluid to the refrigerant loop. A 50/50 Dl-water/ethylene-glycol
solution is selected as the heat transfer fluid. The assumed battery design has the heat transfer
fluid pumped over the module casing to transfer heat from and to the cells. The battery may be
heated by either a positive thermal coefficient (PTC) or flexible mat heater. The PTC heater
would directly raise the temperature of the heat transfer fluid in a reservoir while the matt heater
would be placed under the battery jacket insulation.
The PHEV and EV batteries will likely have both active and passive thermal management modes
requiring some additional monitoring and an electrically actuated valve. We have assumed
decreasing cooling costs for the HEV-HP and microHEV systems without explicitly dictating
where the savings originate. In general, one would expect smaller batteries to have a less
complicated control system, lower flow rates and possibly even direct cooling of the cells with
the evaporator.
5.3 Baseline Manufacturing Plant
The baseline plant is designed to produce 100,000 NCA-Gr baseline battery packs per year. The
baseline battery pack produced by the plant has sixty, 40-Ah capacity cells, providing a total
pack power of 50 kW and total energy of 8.7 kWh. The battery will power 20 miles of vehicle
travel at 70% of the pack energy and 300 Wh/mile. The schematic diagram of the plant (Fig. 5.2)
is designed to illustrate the flow of materials through the plant and the relative floor areas for the
processing steps rather than representing a realistic plant layout. The overall manufacturing rate
of 100,000 battery packs per year is achieved by operating for three shifts at the equivalent of
300 days per year of fully effective production. There will be more than 300 days of operation,
but some days will have less than 100% effectiveness. The exceptions to three-shift operation are
the Receiving and Shipping sections, which are active for only two shifts per day. The cost
factors for the individual manufacturing steps in the baseline plant are summarized in Table 5.4
and discussed in detail in the sections that follow. Most of the operations are carried out with
normal factory atmosphere, but the cell assembly process steps are completed in a dry room
atmosphere.
53
-------
Receiving
Electrode Materials
Preparation
>
Electrode
Coating
Solvent r.v.--1.-"-- Solvent
Evaporation
i
Recovery
Calendering
Electrode
Slitting
Vacuum
Drying
Shipping
Cell and
Scrap
Recycling
Battery Pack
Assembly
and Testing
Control
Laboratory
Module
Assembly
Charge-
Retention
Testing
Formation
Cycling
Final
Cell
Sealing
Air
Locks
Materials
Handling
Cell Stacking
Enclosing
Cell in
Container
Current
Collector
Welding
Electrolyte Riling
and Cell Closing
Assembly Route
Dry Room
Outdoor dry-room air
processing equipment
The areas in this diagram for each processing step are approximately proportional to
the estimated plant areas in the baseline plant.
Figure 5.2 Baseline lithium-ion battery manufacturing plant schematic diagram. Manufacturing
rate: 100,000 NCA-Gr battery packs per year, 50-kW pack power, 40-Ah capacity, 60 cells per
battery. Operating year: 300 days with three 8-h shifts (two shifts for receiving and shipping)
5.3.1 Receiving and Shipping
These operations require the moving equipment and storage facilities common to any such
factory facilities. The Receiving section handles slightly less than 6,000,000 kg of materials per
year and also has facilities to handle and store some of the electrode materials in a dry
atmosphere. The Shipping section is required to enclose the battery packs in crates, which
requires some automated equipment and more labor than is required for Receiving. Shipping also
handles about 400,000 kg of scrap each year, which is broken down and prepared for shipping in
the Rejected Cell and Scrap Recycle section. The estimated resources needed for the Receiving
and Shipping sections are shown in the table below.
Receiving
Off-loading
Moving
Storage
Shipping
Rate Factor
870,000 kWh/y
870,000 kWh/y
Direct Labor
3 per shift
6 per shift
Capital Equip.*
3.60 mil$ total
0.60
1.20
1.80
5.0 mil$ total
Plant Area, m2
900
900
''Total cost including installation
54
-------
£>
cp
03
CD
CD
CL
J/3
"CD
O
0
•e?
CO
Q.
cO
O
<(
i
i_"
CD
O
CL
B
CQ
>
"^
i
O
in
CO
/IN
'u>
c
'Q.
Q.
Ic
CO
T3
CO
;|
CD
£>
£
M—
fr
T3
CD
«
lc
CO
o
i
CO
jtz
~«
i_
o
I
oo
Q)
Q)
£
CQ
0)
^
c
CQ
Q.
^j
s
CZ
Q
'3
w
d
O
k.
o
0)
k.
Q
0)
CQ
O
CO
c
^
Si
CQ
m
CD
75
0)
=
CQ
m
CD
CQ
O
CO
0)
"cu
CQ
CO
7s
CD
J5
P
^
0.
^
O LO CDCD OOOOCD CDCDCDCDCDCD OOOOCDCD^OOCDCDCDCDCDCD
tt O OO OOO OOOOOO OOOOOOOOOOOO
CQ
"•
CM O OO OOLO LOLOOOOO OOOOOOOOOOOO
co oo LO LO eg eg eg o o o o OOOOOOLOOOOOO
„
CO
CD
0. « {
c\T c\T 5-~ 5-~ 5-^
O co r- r- do'co r~- r~- r~- r~- r~- r~- "oor-r-co"r-r-r-c'?i--i--
tt d do "-'"-'d dddddd S-ddddS-dddSdd
tl 00 00 on on ~
l£ °' °' odd
£8 88 888 888888 8888888K8888
:= c\j co
I
„
O^" LOLO LOLO1^ LOLOh-LOLOLO h-h-LOLO^I^LO^LOLOh-LO
•JJO OO OOO OOOOOO OOOOOOOOOOOO
^ o o o o o o o o o o o o o o o o o o o o o o o o
^» o o o o o o o o o o o o o o o o o o o o o o o o
*- •*" •*" •*" oo" oo" •*" •*" i-~" oo" oo" •*" oo" CD" CD" ^- CD" •*" i-~" •*" ••- co" co" CD" oo"
O
X
Ji
|CCO CNJCNJ •^••^•CNJ C\JT— •^••^•CNJ'^- LOLOCOLOCNJOOCNJCOCDCDLOCD
C/)
"^n *t
DC 03 03 (Tj
CD B B ™ w
LO
eg
-
1
3
OJ
o
CD
8
£*~
Q
8^
CJ5
CO
eg"
o
w
CC^
„
"ro
i
| .
$
O
15
C
§
.c
"CD
|
.c
°
o
Q
§•
CD
r-
c
-^
CD
(0
o
• w
1 0°
J2
1
"5.
"5
E
W
0
O
(-
CO
QI
w
3
03
O
03
a p
CD ^
•^ cc
1 I
" 00
£ 2.
— SE
O) ^
•i O°
^
£ "
"to
.^ o
B
ra =5
[^
Q
CD
i 1
^_,
£
si
2
"0
O
^
O^
+-
Q
0)
.^
"0
w
"S
.c
"o
0
.0
«
o
<
o
<^-
^
w
_c
"Q
g1
g
p
£
,2
rn
CO
P
1^-
03
"
p
c|
^
o
II
to
3
w
[I
^
5
Q_
d
CD
£
•2
"CD
_w
co
_^-
0
_1_I
C
0
"u
03
.±±
£•
_Q
CO
0
4-I
§
CD
.g
"CD
_ra
CD
.C
'o
§
o
.2
^
"§
E
o
£
g
£^
£
£
03
Q
**—
.B
'S
CO
2
^
^
^
of
g^
o
O
o
O
CD
a
O"l
d
^
"^
T3
CD
•|
2.
"§
~-^
CD
-g
O
^
2
_Q
E
c
CD
1
Illll sssssss
CU
•i-H
13
05
X)
^
o
r^H
3
^: "o
O o
CO O
Q_ CQ
CO
§.
CO
CQ
O CD
°o
81
CO
m CD
a a o o z
Q) Q) °i- °i- Q)
.^ ^ E E ^
•^ ^ (3) (3) t—
eg r- co co eg
LO co ooo
5! 8
Is- O
O O O O O -C
"E "E "E "E "E i
O) O) O) O) O) CD
CO CO CO CO CO T—
OOOOOO
cncncncnocncncn
ooooooooooooooo
cncD
oo-^
cg
oo"
O)O>O>O)O>O> IDl^l^LOOJLOLOLOOOLOO)
O O O O O CD i— i— i— i— i— i— i— O O i— CD
OJOJOJOJOJOO COCOCOCO CO CO CO O T— CO 00
oo" oo" oo" oo" oo" CD" CD" CD" CD" CD" CD" CD" CD" CD"
CD ^
w o
o
=i « ° «
-21 s 1 a .^ -e i
;5-2o
~ "-
CD
CD ±=
E & | c
Q. O Q. E
^_ CD "5 w
8
If &~! f *f J
11=1^ISS
C/5
—
2
ca
H
Receiving (two-shift o
Materials preparation
Positive electrode
Negative electrodf
Electrode coating
Positive electrode
Negative electrodf
Solvent recovery
Calendering
Positive electrode
Negative electrodf
Materials handling*
Electrode slitting
Vacuum drying
Control laboratory
Cell Assembly in Dry
Cell stacking
Current collector wi
Enclosing cell in cc
en
c
l^
B
>,
Q
'o
-CD
(5
2
"c
0
o
1
£•
Formation cycling
Final cell sealing
Charge retention test
Module assembly
_l_i 1—
P O
^ w
CD
W T3
W C
03 03
v ^z
0 "K
<5 «
Q-T3
£-B
B 8
!_L.
t±
Ic
w
O)
c
"o. _
Q. 03
Q.
W
CD
.C
'o
•E
!c
CD
*The baseline cost fo
the positive electrode
CD
2
$
o
w
"ra
^
**The baseline costs
capacity, the cost is
***The baseline cost
number of modules (
55
-------
5.3.2 Electrode Materials Preparation and Delivery to Coating
The electrode materials, which consist of active material, carbon (if necessary), binder and
binder solvent, are well mixed in small batches in portable tanks. At the design production rate
in the baseline plant, each shift requires three tanks each holding about 1000 liters of positive
electrode material mix and three tanks each holding about 900 liters of negative electrode
material mix. The section must be capable of exceeding this design rate of production by at least
25% to catch up in case of unscheduled downtime in Materials Preparation or in some of the
section immediately following that section. The tanks of prepared materials are moved to the
Coating section and pressurized to push the coating paste into the coating mechanism. The
estimated resources needed are the following:
Materials Prep.
Positive
Materials
Storage tanks
Mixing tanks
Moving equip.
Negative
Materials
Rate Factor
1,71 0,000 kg/y
active material
1,2 10,000 kg/y
active material
Direct Labor
2 per shift
2 per shift
Capital Equip.*
2.0 mil$ total
1.00mil$
0.50
0.50
2.0 mil$ total
Plant Area, m
600
600
*Total cost including installation
5.3.3 Electrode Coating on Current-Collector Foil
The positive and negative electrode structures are formed by coating both sides of the current
collector foil. In the baseline plant, the coating lines are 1.5 meter wide continuous roll-to-roll
coating processes carried out at a line speed of 10 m/min. The first set of coating and drying
stations coats one side of the current collector foil, drives off the solvent in a heated oven, and
turns the foil over while transferring it to a second set of stations. The second set of coating and
drying stations applies and dries the remaining coating before the coated foil is wound into a
large roll at the end of the line. An advanced alternative would be to run the foil directly into the
calendering process. The negative and positive coating lines are very similar. However, some of
the negative material is coated only on one side to provide the electrodes at the end of the cell
stacks. For the baseline plant, a total of 8,170,000 m2/y of coating (annual cell area) is required
for the positive electrode (slightly more for the negative electrode), which allows for the 5% loss
of cells expected to fail testing and inspection. A larger area of foil than the coated area must be
fed to the coaters to allow for the part of the foil that is not coated so as to provide tabs for
welding to the terminals (about 10%) and to allow for trimming losses during electrode slitting
(8%). Also, about 30% excess coating capacity must be provided to allow for unscheduled
downtime. Only one coating line is needed for each electrode type to meet these needs. If one
coating line breaks down, the other coating line may change over temporarily to coat the other
electrode material.
The oven sections of the coating line are designed to dry coatings about 100 microns thick at the
coating speed of 10 m/min. A thicker coating will require longer ovens at additional capital cost
which is provided in the adjustment of costs discussed in section 5.4. For the same annual area
56
-------
throughput, a coating line that coats both sides with a 300-micron coating would cost $9,500,000
rather than the $8,000,000 cost for the 100-micron coater. The binder solvent for the positive
electrode in the baseline plant is NMP, which must be recovered by condensation and recycled.
About 0.5% of the binder solvent is combusted with a thermal oxidizer and must be replaced. For
the negative electrode the binder is water, which need not be recovered. The estimated resources
to meet these needs are the following:
Electrode Coating
Positive Electrode
Uncoated area
Width of coater
Coating speed
Number of coaters
Maximum rate
Excess capacity
Negative Electrode
Solvent Recovery &
Oxidation
Rate Factor
8,170,000
m2/y cell area
8,170,000
m2/y cell area
1,527,000kg
NMP/y
Direct Labor
4 per shift
4 per shift
2 per shift
Capital Equip.*
8.0 mil$ total
18%
1.5m
10 m/min
One
13,000,000 m2/y
30%
8.0 mil$ total
3.0 mil$ total
Plant Area, m2
750
750
225
*Total cost including installation
5.3.4 Calendering
The materials leaving the coating lines may be stored on large rolls (see next section). However,
typically the materials leaving the coaters would go directly to the calendering process in which
the coatings are compressed by rolling to meet the specified void volume fraction, which will
later be filled with electrolyte. The calendering equipment must match the output of the coating
equipment producing 8,170,000 m2/y of cell area with a maximum rate of 13,000,000 m2 of foil
per year to meet contingencies as in coating. We estimate three workers are necessary to
collectively operate the two pieces of equipment. The estimated resources to meet these needs
are the following:
Calendering
Positive Electrode
Negative Electrode
Rate Factor
8,170,000
r\
m /y cell area
8,170,000
m2/y cell area
Direct Labor
2 per shift
1 per shift
Capital Equip.*
1.0mil$ total
1.0mil$ total
Plant Area, m2
225
225
*Total cost including installation
5.3.5 Inter-Process Materials Handling
For all processes (Fig. 5.2), work in progress must be transported and occasionally stored to
permit nearly-continuous operation of the equipment. Storage areas must be provided both inside
and outside of the dry room. Raw materials must also be moved to the processing sites, which for
those in the dry room means through a separate air lock for materials transfer. One-third of the
57
-------
total space for Inter-Process Materials Handling is within the dry-room for the baseline plant and
also for all other plants. The estimated resources to meet these needs are the following:
Materials Handling
Rate Factor
8,170,000
ij
m /y cell area
Direct Labor
4 per shift
Capital Equip.*
1.5mil$total
Plant Area, m2
900
*Total cost including installation
5.3.6 Electrode Slitting
The coated electrode foils are slit into strips between the coated sections and then into individual
electrodes as shown in Fig. 2.3. The estimated scrap loss of foil for this process is about 8%. The
estimated resources to meet these needs are the following:
Electrode Slitting
Rate Factor
8,170,000
r\
m /y cell area
Direct Labor
4 per shift
Capital Equip.*
2.0 mil$ total
Plant Area, m2
300
*Total cost including installation
5.3.7 Final Electrode Drying
In the absence of electrolyte, no harm is done by exposing the electrodes to normal factory air;
however, the electrodes must be dried by heating under vacuum prior to cell assembly.
Maintaining extremely low moisture conditions during cell assembly is believed to be very
important in achieving long battery life. The final drying step coupled with dry room conditions
ensures a minimal quantity of moisture will exist in the final product. The pertinent processing
rate in determining the resources necessary for drying is the total amount of active materials
processed per year (other electrode materials are approximately proportional), which for the
baseline plant is 2,950,000 kg/y or 3,275 kg/shift. The individual electrodes exiting from the
electrode slitting process are separated into stacks by polarity, loaded into vacuum drying ovens,
dried for several hours, and unloaded directly into the dry room. The estimated resources to meet
these needs are the following:
Electrode Drying
Dryer capacity
Number of dryers
Maximum rate
Rate Factor
2,950,000 kg/y
active material
Direct Labor
2 per shift
Capital Equip.*
1.6mil$total
600 kg/shift
8
4,320,000 kg/y
Plant Area, m2
300
*Total cost including installation
5.3.8 Control Laboratory
The purpose of the control laboratory is to ensure that the raw materials and the electrodes being
fabricated meet specifications. Laboratory personnel collect or supervise collection of samples
and carry out analyses. The estimated resources to meet these needs are the following:
58
-------
Control Lab
Rate Factor
869,000 kWh/y
Direct Labor
4 per shift
Capital Equip.*
1.5 mil$ total
Plant Area, m
300
*Total cost including installation
5.3.9 Cell Stacking
The cells are assembled in four steps, which are carried out in a dry room. The first of these steps
is cell stacking. The primary rate factor that determines the cost for all steps in cell assembly is
the number of cells assembled per year. For cell stacking an additional cost factor is the capacity
of the cells; large cells usually require more electrodes of larger area and thus a more capable,
faster cell stacking machine. The method used to determine the extra costs of stacking equipment
is detailed in Table 5.4. The capacity of the cells is deemed to have only a minor effect on the
other steps in cell assembly and is not taken into account for those steps. The electrodes are
inserted in a folded separator sheet, the positive electrodes tabs protrude on one side and the
negative electrodes tabs on the other. As in other parts of the plant, excess capacity is provided to
allow catching up after unscheduled downtime. The estimated resources to meet these needs for
the baseline plant are the following:
Cell Stacking
Stacking rate
Number of units
Maximum rate
Rate Factor
6,320,000 cells/y
Direct Labor
5 per shift
Capital Equip.*
4.0 mil$ total
5 cells/min
4
8,640,000 cells/y
Plant Area, m2
600
*Total cost including installation
5.3.10 Current Collector Welding
The current collector tabs for the negative and positive electrodes are welded to their respective
terminals by ultrasonic welding. This procedure achieves a connection of near-zero resistance
and avoids overheating the electrodes during the welding process. The estimated resources to
meet these needs are the following:
Tab Welding
Cell rate
Number of units
Maximum rate
Rate Factor
6,320,000 cells/y
Direct Labor
5 per shift
Capital Equip.*
4.0 mil$ total
5 cells/min
4
8,640,000 cells/y
Plant Area, m2
600
*Total cost including installation
5.3.11 Enclosing Cell in Container
The aluminum foil in the pouch container is sufficiently thick (100 microns default thickness) to
permit the use of stiff, pre-shaped pouch halves. The pouches are assumed to be purchased as
finished parts. Each cell is enclosed in these containers, which are then partially sealed prior to
injecting electrolyte. The estimated resources to meet these needs are the following:
59
-------
Enclosing cells
Cell rate
Number of units
Maximum rate
Rate Factor
6,320,000 cells/y
Direct Labor
3 per shift
Capital Equip.*
3.0 mil$ total
5 cells/min
4
8,640,000 cells/y
Plant Area, m2
600
*Total cost including installation
5.3.12 Electrolyte Filling and Cell Sealing
At this station, the cells are evacuated, filled with electrolyte and temporarily sealed. The
estimated resources to meet these needs are the following:
Filling & 1st Seal
Cell rate
Number of units
Maximum rate
Rate Factor
6,320,000 cells/y
Direct Labor
5 per shift
Capital Equip.*
5.0mil$total
5 cells/min
4
8,640,000 cells/y
Plant Area, m2
900
*Total cost including installation
5.3.13 Dry Room Management
Excellent dry-room atmosphere is required for lithium-ion cell assembly. A maximum dew point
temperature of -40 °C is maintained in the room. The load on the dry-room drying apparatus is
determined by diffusion of water vapor through the walls, entry of air through the air locks, the
number of workers in the room, and the need to admit some fresh air to limit the build up of
contaminants such as electrolyte solvent vapor. These load factors are approximately a function
of the room area. Because of the importance of the proper functioning of the dry room, two
workers are on duty at all times to monitor its performance. The equipment for circulation and
purification of the dry air will be located outside of the plant building, adjacent to the dry room.
The estimated resources to meet these needs are the following:
Dry Room
Operating
Area
3,000 m2
Direct Labor
2 per shift
Capital Equip.*
20.0 mil$ total
Air Locks, m
100
*Total cost including installation
5.3.14 Formation Cycling, Final Cell Sealing and Charge Retention Testing
Formation cycling is expensive because it takes considerable time and each cell must be
monitored separately. For plants to be operated in 2020, we expect some improvements from
present day operations because of the urgency to improve and thus save cost. We project that the
entire formation cycling and testing can be done in two shifts. These operations consist of
charging the cell, discharging to full depth to measure capacity and impedance, followed by fully
recharging the cells. These tests will be carried out in large temperature controlled cycling units
60
-------
that test 500 cells simultaneously, monitor each cell and automatically identify failed cells. The
capital cost of the cycling equipment is primarily a function of the annual number of cells to be
tested, but to a lesser extent on the capacity of the cells.
The short-term testing described above does not detect cells that have self-discharge rates that
are slightly above normal, which could lead to catastrophic failures later. To detect such defects,
the cell charge is topped off and the cells are stored for two weeks and then checked for loss of
charge. Most of the test period is spent in large racks in compact arrays, without electronic
monitoring. Incidentally, the two-week long self-discharge testing requires less floor space than
for formation cycling, which lasts only two shifts.
The final cell sealing occurs between the formation cycling and charge-retention storage test.
Gas generated during formation cycling may accumulate in the reservoir space that was created
during the temporary sealing step. This gas is removed by creating the final seal below the
reservoir and trimming off the unwanted portion.
The estimated resources to meet these needs are the following:
Formation Cycling
Cell capacity
Number of cyclers
Cells per cycler
Length of test
Testing capacity
Final Cell Sealing
Charge Retention
Testing rack capacity
Racks per stack
Number of racks
Length of test
Testing capacity
Rate Factor
6,320,000 cells/y
6,320,000 cells/y
6,320,000 cells/y
Direct
Labor
8 per shift
2 per shift
3 per shift
Capital Equip.*
30.0 mil$ total
40 Ah
35
500
2 shifts
7,875,000 cells/y
2.0 mil$ total
4.75 mil$ total
500 cells
5
750
14 days
8,040,000
Plant Area,
m2
2200
450
900
*Total cost including installation
5.3.15 Module and Battery Assembly
Approximately 5% of the cells are expected to fail the formation cycling and charge-retention
tests and these are sent to the Rejected Cell and Scrap Recycle section. The accepted cells
(6,000,000 finished cells per year) are assembled into modules by attaching the terminals
through laser welding or mechanical joining with spring loaded devices. Electronic circuit packs
are attached that occupy about the same volume as a cell. An aluminum heat conductor is placed
around every cell. These operations are carried out at four automated stations each capable of
handling about 280 cells per hour. For the module design being cost estimated in this model, the
61
-------
module is enclosed in an air-tight aluminum container by double seaming. The processing rate
that determines the cost of module assembly is the number of finished cells that must be handled
per year.
The finished modules are assembled into battery packs with the aid of automated stations. The
total cost of these stations is dependant mainly on the number of battery packs to be assembled
per year (100,000 for the baseline plant), but to a lesser extent on the number of modules per
pack. After assembly, the packs are moved to testing stations where they are discharged as a final
check of impedance and to lower the state of charge to a level suitable for shipping. The
estimated resources to meet these needs are the following:
Module Assembly
Number of stations
Cells/h/station
Capacity
Battery Pack Assembly
Modules/pack
Number of stations
Packs/h/station
Capacity
Battery Pack Testing
Rate Factor
6,000,000 cells/y
100,000 packs/y
100,000 packs/y
Direct
Labor
6 per shift
3 per shift
3 per shift
Capital Equip.*
6.0 mil$ total
4
280
8,060,000 cells/y
3.0 mil$ total
4
3
6
130,000 packs
3.0 mil$ total
Plant Area,
m2
600
450
450
5.3.16 Rejected Cell and Scrap Recycle
Scrap is generated in preparing the electrodes and by the rejection of 5% of the cells that go
through formation cycling and charge-retention tests. This scrap is gathered and packaged for
shipment for recycling of the materials having value. No credit is taken for the value of the scrap
in this model except that the costs of gathering, sorting, packaging and shipping are understated
by about that value. The main factor in determining the cost of scrap recycle is the number of
cells rejected, which have to be disassembled to recover the scrap, a labor intensive process. The
yields of materials in the various processing steps are shown in Table 5.5.
Table 5.5 Materials yields during electrode and cell fabrication
Material
Positive Electrode
Negative Electrode
Positive Current Coll.
Negative Current Coll.
Separator
Electrolyte
Material
Mixing
99
99
Coating
95
95
99
99
Electrode
Slitting
99
99
92
92
Cell
Stacking
99
99
99
99
98
Electrolyte
Filling
94
Total
92.2
92.2
90.2
90.2
98.0
94.0
The estimated resources needed for scrap recycle are the following:
62
-------
Scrap Recycle
Scrap rate
Rate Factor
6,320,000 cells/y
Direct
Labor
5 per shift
Capital Equip.*
2.5 mil$ total
441 kg/shift
Plant Area, m2
600
5.3.17 Baseline Plant Summary
The processing rates and the primary cost factors for the baseline plant are summarized in Table
5.4. The main cost-determining rate of processing for each step is shown in the second column.
The requirements for direct labor, capital equipment and plant area, which are shown in detail in
the subsections above, are summarized in the table. It is seen that the plant requires a total of 90
workers per shift, $127,450,000 worth of capital equipment, and 15,425 square meters of plant
area to manufacture the baseline battery at a rate of 100,000 battery packs per year.
5.4 Adjustment of Costs for Varying Production Volumes
Production volume may affect the end price of the battery in two distinct ways. First, the user of
the model may change the annual production volume and every processing step will be affected.
Somewhat differently, as the performance requirement and thus design is changed, the
production of individual steps will change in non-uniform ways. As noted in Table 5.4, there are
many processing rates that must be considered in addition to the overall number of battery packs
manufactured per year. Each of these rates affects the costs of one or more steps in the process
and may have no effect upon the costs of other steps in the process. For instance, when the user
of the model increases the power of the battery packs without increasing the number of cells or
their capacity, the model increases the area of the cells and decreases the electrode coatings
thicknesses. Such changes would result in an increase in the cost of the coating equipment, the
floor area occupied by the equipment, and in the direct labor for that step in the process. It would
have no effect on the cost of mixing the materials to be coated because the amounts of these
materials per battery back are unchanged under the assumed conditions.
The general approach to cost estimation of multiplying a known cost by the ratio of processing
rates raised to a power has also been applied to the capital cost of individual items of
49
equipment.
C = C0(R/Rof
(5.2)
Here, C0 is the capital cost of an installed equipment item designed for the baseline processing
rate, ^0. The power factor, p, relates the capital investment cost and the processing rate for the
manufacturing step.
If the value of p were 1.0, it would imply that the cost of the equipment item, or the equipment
items if there are several in parallel, would be directly proportional to the processing rate.
However, the value of p for the cost of equipment is frequently about 0.6 to 0.7 for many
manufacturing process steps because the equipment is larger for the higher processing rates and
63
-------
its cost is less than if it were directly proportional to the processing rate. For process steps
requiring the addition of many identical pieces of equipment for scale up, such as may be true for
formation cycling of battery cells, the value ofp may be as high as 0.9. The value ofp is unlikely
to reach 1.0 because the equipment cost includes installation, for which there is some savings
even in installing multiple units of the same processing capacity. The relationship between cost
and processing rate for two-fold and three-fold rate changes is illustrated in Table 5.6.
Table 5.6 The effect of processing rate (R) on cost for various scale factors
CICo = (R/R0f
Scale Factor, p
0.25
0.3
0.4
0.5
0.6
0.7
0.8
0.95
1.0
Cost Ratio, C/C0
R/R0 = 2
1.19
1.23
1.32
1.41
1.52
1.62
1.74
1.93
2.00
R/RO = 3
1.32
1.39
1.55
1.73
1.93
2.16
2.41
2.84
3.00
Similar equations have been applied for determining the effect of processing rate on the annual
hours of labor and the plant area required for a manufacturing step. In general, the value of p is
low for the labor equation, usually only 0.4 to 0.5, because only a relatively small addition to the
labor crew permits operation of larger equipment or of operating several more units of the same
processing capacity.24 The value of p for the plant area required for a processing step is slightly
less than that for equipment. The floor area required for larger equipment or for more equipment
items of the same size is proportionately less than the increase in the processing rate because of
the more efficient use of the space occupied by the equipment and the savings in aisle area.
The value of the scale factors (i.e. p factors) for labor, capital equipment, and floor area were
estimated for each of the processing steps (Table 5.4). The scale factors selected for the direct
labor requirement are usually only 0.4 to 0.5, which indicates considerable unit cost reduction for
increasing the plant throughput.
For most processing steps, increasing the processing rate beyond that in the baseline plant would
result in a decision to increase automation or use faster equipment to mitigate the costs of higher
levels of throughput. Decreasing the processing rate would have the opposite effect. Some steps
in the process such as cell stacking, welding of current collectors, and formation cycling do not
appear to be easily automated beyond the level intended in the baseline plant and, thus require a
higher value for the scale factor of 0.8. This higher scale factor results in achieving fewer
reductions in the cost per battery pack with increasing production volume. Additionally, a higher
64
-------
p factor results in a less severe penalty for lower production scale for an individual step in the
process.
There are five steps for which the cost of the capital equipment is affected by other factors than
the main processing rate for the process step. These are discussed in the footnotes at the bottom
of Table 5.4. For these steps, the costs that have been adjusted for the changes in the processing
rate from the baseline rate are further adjusted to take into account the other cost factors. The
cost of the coating equipment is adjusted for the amount of solvents to be driven off of the
positive and negative electrodes; thicker coatings need longer, more expensive ovens to drive off
the additional binder solvent or the coater most be operated at lower speeds. The cost of the cell
stacking equipment and that of the formation cycling equipment, for which the main cost factor
in both cases is the number of cells to be fabricated annually, are also adjusted for the capacity of
the cells; larger cells require more expensive equipment. The cost of the capital equipment for
battery assembly is primarily a function of the number of cells in the battery, but it is also a
function of the number of modules that must be interconnected. This dependence is accounted
for in the model with an additional multiplying factor.
A breakdown of the baseline plant capital equipment costs listed in Table 5.4 is illustrated in Fig.
5.3. The largest costs for capital equipment are for formation cycling and testing, cell assembly
in the dry room and electrode coating. These capital costs are likely to be dominant in any
lithium-ion battery plant in the near future.
2% 7%
15%
29%
28%
D Receiving and shipping
• Materials preparation
D Electrode coating
D Calendering
• Materials handling
D Electrode slitting
• Vacuum drying
n Control laboratory
• Cell assembly in dry room
• Formation cycling and testing
n Module and pack assembly
n Rejected cell and scrap recycle
Figure 5.3 Breakdown of installed capital equipment costs for the baseline plant
65
-------
5.5 Plant Investment Costs
In this model, the calculated investment costs are defined as those directly related with building
and operating the plant (Table 5.6). Other costs that may require investment, such as research
and development, are added separately to the unit cost of the battery. The largest investment cost
is for the installed capital equipment. Each cost item for the battery under design is adjusted from
the estimate of the baseline plant. The plant cost is done in a similar way with a cost of $3,000
per square meter ($280/sq. ft) including land and utilities. The high cost for land and utilities
accounts for both the area of the manufacturing facility as well as other land requirements such
as office buildings and waste water treatment requirements. Launch costs include plant start-up,
employee training and materials that are lost or recycled in early stages of production, beyond
the normal amounts. Launch costs are estimated to be 5 % of annual materials costs plus 10 % of
annual direct labor and variable overhead (Section 5.6). Working capital is needed to cover the
costs of payroll, receivables, and the inventories of raw materials, work in progress and finished
product. These working capital costs are partially offset by bills that are payable. We estimate
the working capital to be 15 % of the annual variable costs.
Table 5.6 Battery pack manufacturing investment costs
Investment Costs
Capital Equipment
Plant Floor Space
Launch Costs
Working Capital
Description
Equipment costs including
installation
Space includes aisles and space
for unfinished processing
inventory plus land and utility
costs.
Plant start-up, training, out-of-
spec product.
Cash to meet payroll,
receivables, inventories of raw
materials and of unfinished and
finished product, minus
payables.
Method of Calculation
Estimates of costs for each
processing step at baseline rates
adjusted for actual rates.
Estimates of costs for each
processing step at baseline rates
adjusted for actual rates.
5% of annual materials cost,
10% of direct labor plus
variable overhead.
15% of annual variable costs.
5.6 Unit Costs for Battery Pack
The unit costs of the battery pack are calculated as summarized in Table 5.7.
5.6.1 Variable Costs
The costs of the materials and purchased items are based on the costs discussed in section 5.2,
and the annual amounts of materials are adjusted for the yields of materials (section 5.3) and
yield of cells. The direct labor is the sum of the labor cost for each step in the process, which are
66
-------
each calculated for the baseline plant and adjusted for the rate associated with the battery under
study. Variable overhead is the cost of indirect materials and labor, utilities, and plant
maintenance. It is estimated to cost 40 % of direct labor costs and 20 % of depreciation.
5.6.2 Fixed Expenses
Fixed expenses include General, Sales, and Administration (GSA), research and development,
and depreciation. The cost of GSA includes the plant office, taxes on income and property, cost
of sales and insurance. It is estimated by the model as 25 % of direct overhead and depreciation.
Research and development (R&D) must be carried out to ensure that the battery packs that are
produced in the plant and the means of production continue to be competitive in the world
market with respect to performance and price. The greater the investment in the plant and its
equipment, the greater is the need to be successful in the R&D effort. Thus, the expenditure has
been set at 40 % of the depreciation expense. Depreciation expense provides funding available
for future investment in this plant or another venture to replace deteriorating plant and
equipment. The equipment and plant are depreciated at straight-line rates for 6-year life (16.7 %
per year) and 20-year life (5 % per year).
Table 5.7 Unit cost of battery pack
Variable Costs
Materials and Purchased
Items
Direct Labor
Variable Overhead
Fixed Expenses
General, Sales, and
Administration (GSA)
Research and Development
Depreciation
Profit
Warranty
Description
All materials and purchased
items in finished product and
lost in processing.
Labor costs for operations and
immediate supervision.
Indirect materials, labor,
utilities, plant maintenance
Plant office, taxes on income
and property, cost of sales and
insurance expenses.
On-going research needed to
upgrade product and maintain
competitive position.
Provides funds for new
investments to replace those in
current equipment and plant.
Return on invested capital after
taxes.
Funds set aside for reimbursing
customers for battery pack
failures.
Method of Calculation
Based on prices of materials,
cost equations for purchased
items and yields.
Estimates of costs for each
processing step at baseline rates
adjusted for actual rates.
40% of direct labor cost plus
20% of depreciation
25% of direct labor and variable
overhead plus 25% of
depreciation.
40% of depreciation
16.7% of capital equipment cost
plus 5% of plant floor space
cost.
5% of total investment costs.
5.6% added to price based on
present worth of projected
payments.
67
-------
5.6.3 Profits
The profit goal for this type of venture varies with the financial structure of the company,
especially regarding long-term debt. For the model, the profit is set to provide a 5 % return on
the total investment, which is an approximate average for mature manufacturing as vehicle
battery production is expected to be in 2020. In general, the chosen cost structure and the
resulting margin are similar to a Tier 1 supplier in the automotive industry.
5.6.4 Battery Pack Warranty Costs
If a battery module or an entire pack fails, the replacement will cost much more than the original
price paid by the OEM. It is important that such events are rare, but provision must be made to
reimburse the vehicle owner, especially in the early years of the projected battery life. The extra
costs of replacing the battery will result from labor for testing and replacing the battery,
inventory costs for stocking replacement batteries, and servicing the battery controller if the new
battery is slightly different than the old battery. It is likely that the battery manufacturer will be
responsible for the cost of the new battery, which we assume will be equal to the cost of the
original battery. The other costs of replacing the battery, to the extent that they are covered by
the warranty, are assumed here to be covered by the automobile manufacturer and the dealer.
The goal for average battery life is 15 years and a warranted life of 10 years, with full
replacement in the first five years and shared cost of replacement for the last five years seems
appropriate. The vehicle owner would pay an increasing share of the cost from between 0 % at 5
years to 100 % at 10 or more years. With these assumptions, the cost to the battery manufacturer
will be equal to the present worth of the future costs of the new battery or modules as provided in
the warranty. The rate of battery failure will vary over the life of the battery with a slightly
higher rate early in life, then a low failure rate followed by a gradually increasing failure rate.
For purposes of calculation we assume a failure rate of 1.0 % per year throughout the warranty
period. With an internal rate of return of 8 % and calculated on a monthly basis, the present value
of the future costs would be about 5.6 % of the price of the battery before adding the warranty
cost.
5.7 Summary of Baseline Battery Cost
The spreadsheet version of the model, which is discussed in more detail in sections 6 and 7,
provides a summary sheet which is illustrated in Table 5.8 for the cost of the baseline battery and
that of two others. This breakdown of the battery costs, with a brief summary of the design
values, illustrates the effects of the cost factors. The second battery has twice the power of the
baseline battery and the third battery has the same power as the baseline battery, but twice the
capacity. The number of cells is the same for each battery. The energy storage is slightly higher
for the battery with double power because the voltage would be slightly higher during the
discharge to determine capacity. The battery with double the capacity has fewer electrodes which
are longer and wider, because the cell thickness is maintained, resulting in higher resistance in
the current-collector structure. The higher impedance lowers the voltage during the discharge
capacity measurements and results in slightly less than twice the energy storage of the baseline
battery.
68
-------
Table 5.8. Summary of results for cost of
double the power and double the capacity
Calculated Battery Parameters
Battery energy storage, kWh
Battery power at target % OCV, kW
Required battery power, kW
Capacity, Ah
Number of cells
Battery weight, kg
Battery volume, L
Weight and Volume of Components Exterior to Battery
Weight, kg
Volume, L
Cooling system power requirement, W
Vehicle electric range, miles
Investment Costs
Capital equipment cost including installation, mil$
Building, Land and Utilities
Area, m2
Cost, $/m2
Building investment, mil$
Launch Costs
baseline battery and that of
of the baseline
Baseline
8.7
50.0
50.0
40
60
59.1
31.2
11.0
5.6
1,760
20.3
127
15,374
3,000
46.1
battery
Double Power
8.8
100.0
100.0
40
60
72.4
37.1
9.0
4.0
649
20.5
149
18,165
3,000
54.5
similar batteries
with
Double Capacity Double Modules
17.3
86.9
50.0
80
60
108.2
55.2
11.0
5.6
857
40.5
161
19,204
3,000
57.6
17.4
100.0
100.0
40
120
116.0
59.7
9.0
4.0
724
40.6
212
24,284
3,000
72.9
Rate: 5% of direct annual materials + 10% of other annual costs
Total, millions
Working capital (30% of annual variable costs), mil$
Total investment, mil$
Unit Cost of Battery Pack, $
Variable Cost
Materials and Purchased Items
Cell materials
Cell purchased Items
Module
Battery pack
Total
Direct Labor
Electrode processing
Cell assembly
Formation cycling, testing and sealing
Module and battery assembly
Cell and materials rejection and recycling
Receiving and shipping
Control laboratory
Total
Variable Overhead
Total Variable Cost
Fixed Expenses
General, Sales, Administration
Research and Development
Depreciation
Total Fixed Expenses
Profits after taxes
Total unit cost per battery not including warranty, $
Summary of Unit Costs, $
Materials
Purchased Items not including cooling system
Direct Labor
Variable Overhead
General, Sales, Administration
Research and Development
Depreciation
Profit
Warranty (includes battery pack only)
Price to OEM for battery pack, $
Pack integration (BMS & Disconnects), $
Estimated cost to OEM for thermal management, $
Total cost to OEM for complete battery system, $
10.62
28.78
212.53
1,302
63
202
148
1,714
35
26
17
16
6
8
5
113
92
1,919
110
94
235
438
106
2,463
1,302
412
113
92
110
94
235
106
138
2,601
395
360
3,356
13.11
35.80
252.58
1,739
66
205
141
2,151
51
26
17
16
6
8
5
129
107
2,387
128
110
276
514
126
3,027
1,739
413
129
107
128
110
276
126
170
3,196
395
240
3,831
17.05
47.53
283.32
2,452
74
237
164
2,927
47
26
17
16
6
11
7
130
112
3,169
135
119
297
551
142
3,861
2,452
475
130
112
135
119
297
142
216
4,078
395
280
4,753
19.30
53.28
357.74
2,547
110
399
187
3,243
50
39
26
22
11
11
7
165
144
3,552
175
156
390
721
179
4,452
2,547
696
165
144
175
156
390
179
249
4,701
475
240
5,416
69
-------
Doubling the power does not add as much cost to the materials and purchased parts as doubling
the cell capacity. Most of the labor costs for the three batteries are similar with the major
difference being for the labor cost for electrode processing. The double power battery requires
greater labor costs principally for coating the larger electrode area. Capital equipment and
depreciation costs are higher for both the high power and high capacity battery packs. The
increases in capital equipment cost for the high-power battery are for coating, calendering,
materials handling and vacuum drying equipment. For the high-capacity battery, the main
additional capital equipment costs are for the materials mixing, binder solvent recovery, cell
stacking and formation cycling steps in the process.
Overall, doubling the power of the battery increases the price by only 23 %. Doubling the
capacity of the cells increases the cost by 57 %, considerably more than for doubling the power.
Alternatively, doubling the number of baseline cells and modules within a larger battery jacket
(two rows of modules instead of one, twice the voltage, energy, and power) would increase the
cost by 81 %.
The summary of unit costs for the baseline battery pack, which is shown at the bottom of Table
5.8, is illustrated in Fig. 5.4. The materials and purchased items are the largest costs for the
battery. For larger levels of production, these costs are even more dominant because the scale
factors for these items are close to one.
50%
16%
n Materials
• Purchased Items not including
cooling system
D Direct Labor
D Variable Overhead
• General, Sales, Administration
D Research and Development
• Depreciation
D Profit
• Warranty (includes battery pack
only)
Figure 5.4 Breakdown of unit costs for baseline battery with total price to OEM of $2600. The
total cost to the OEM, including pack integration components, is $3,360.
70
-------
6 Description of the Spreadsheet Model and Instructions for Use
6.1 Background
Historically, the model has been based on Microsoft® Office Excel spreadsheets. The flexibility
afforded by a spreadsheet approach has been extremely useful to the development of the
calculations. Until now, the model had been in a constant state of development. Changes to
parameters and equations were made rapidly and frequently. The publication of this report
represents the first time a version of the model will be "frozen" for open distribution to the
public. Advances will continue to be made with the model, such as those discussed in the last
section of this report. However, distributions of the revised model will be made in an orderly
fashion rather than the continuous improvement approach taken over the last number of years.
6.2 Instructions
The following subsections are a brief explanation of how one may operate the spreadsheet based
model. The user is advised to save the original document separately as a back-up copy.
Corruption of the calculation is possible and will likely occur during use by someone unfamiliar
with the model.
6.2.1 Enabling Calculation
This Microsoft® Office Excel workbook requires the use of iteration. To enable this feature in
Office 2003, go to the "Tools" drop-down menu and select "Options." On the calculation tab,
check the box next to "Iteration" and change the maximum number of iterations to 1000 (Figure
6.1). Perhaps most importantly, ensure the calculation is set to automatic and not manual. If the
iteration is not turned on, the software will present an error complaining about circular
references. If the model is opened while a different Excel spreadsheet is in use, the software will
also warn of an error. Simply close all Excel windows except for the model; alternatively, one
could re-enable the iterative function as discussed above. In newer versions of Excel such as the
Office 2010 edition, the iterative function may be enabled by going to File > Options >
Formulas.
71
-------
] 9= Edt S»S" fserc
look Eafa »ld™ t
* -a a - ,/ 1 o •
-si *i
u
> •/, . -as iff _ -
i * ' !>• s> i -3 <; I ./. » I J 4 «! £ & J
1
2
3
4
5
6
7
ToT
11
12
A
B
C D E
Cell
Positive Electrode
Active material capac
Weiaht %
Active material
Carbon
Binder
Binder sol\
Density, a
13 I Active mat
~U\ Carbon
15 I Binder
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
. <
Negative
N/P capac
Active mat
Weiaht %
Active mat
Carbon
Binder
Binder sol\
ent
_%
cm3
erial
Electrode
ty ratio aft
erial capac
erial
ent
Void. Vol% %
Density, q/cm3
Active material
Carbon
Binder
Positive Foil
Material
Thickness
urn
Negative Foil
••iftsssi'sd.
Selected System: LMO-G
ty, mAh/g: 100
F
G
Chemistry
NCA-G NMC-G
1160
89 89
6
5
NMP
32
1.825
1.77
5r formation 1.25
ty, mAh/g: 290
95
6
5
NMP
32
4.78
1.825
11.77
1.25
290
95
0 0
5 5
Water
34
2.24
1.95
1.1
Water
34
2.24
1.95
1.10
Aluminum
20
Aluminum
20
175
89
6
5
NMP
32
4.65
1.825
1.77
1.25
290
95
0
5
Water
34
2.24
1.95
1.10
H
l
J
Default Values
LFP-G
155
89
6
5
NMP
50
3.45
1.825
1.77
1.25
290
95
0
5
Water
34
224
1.95
1.10
Aluminum
20
Aluminum
20
£sante
LMO-LTO
inn
K
L
M
N
LMO-G Other
Override
Values
•t nn
cda | Intematwnal | save | Errachedang | SrjelSng | EecurRy
f? startup
^Nona
window opt*
r Page b,
Task Pane P Eprmulaha P 3atusbar P Widows In Tasltar
*„,»,*,»„* r ««„„,,„
aks F Row & cdlenn headers ^ HorEontd scral bar
p'Se.ovakjes F sheet tate
| OK | Cancel
~89 yt> \
6
5
Water
40
3.40
1.95
1.10
0
5
Water
34
2.24
1.95
1.10
Aluminum
20
Aluminum
20
0
P I
=|
'-m
nn ' M.,4,l= ' unwind 4EMHMH&8F ly
Excel Options
General
| Formulas
Ft oof ing
Save
Language
Advanced
Cuitomtje Ribbon
Quick Autss Toolbar
Trust CtnW r
^-^j Change options related to formula calculation, performance, and error handlmg.
CakuiaLon options
Workbook Calculation . J Enable iterative calculation
® Autonwtk MasimumHtritioni; 1,000 -s-
Automatic except lor jjata tables
Waiimum Change1 0.001
.1 Manual
0 ftecakulatewoifcboot Defer* saving
WoiUngwtthlannulH
P4CI itltttntt Jtyte •
0 formula AutoCompiete
J: U?*labl< names in formulas
IE Use GetP>
•J Incomiitent calculated columniormula in tables J Unlocked cells containing formulas
151 Mumoers formatted as text or preceded by an apostrophe -f Data entered in a table is invalid
QK "] | Caind
Ssna Find a
»r" Filter- S*l*n-
Editing
Q
R
S
T *z
Ready Calculate | | l-3f 3 EJ 100% -
+
Figure 6.1 Automatic iteration must be enabled for the spreadsheet model to function. The top
screen shot was taken from Microsoft
the 20 10 edition.
®
Office Excel 2003 edition and the lower image is from
72
-------
6.2.2 System Selection Worksheet
The cell chemistry is selected by copying the system designated at the top of a column, for
instance NCA-G in cell F3, pasting it into cell E3 (Figure 6.2). Any of the values in row E can be
overridden by entering the desired value in column L. For example, the maximum electrode
thickness may be overridden by placing a new value in cell L53. The selection of the cell
chemistry also includes the associated prices at the bottom of the page. These prices can also be
overridden by entering the desired values in column L. A full screen shot of the system selection
worksheet is in Figure 6.3. An alternative cell couple, NMC333-G, is pasted into column O as an
example of another commercially relevant battery chemistry.
i [Compatibility Mode] - Microsoft E«e!
:±1 _jj ffl a a
|«™«i^ "S^"- =««" ^
1
2
3"
4
5
e
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
"27
28
29
E3 t
A B
t3»«. BF—I», ' ^ j ii^l
h/1 Gndtmei B HMdlnQi Zoo"* IM% loom to
selection
NCA-G
c
'D 1 E
S B
N(W Arrang
Window All
FW s spin
~Htd«
RWIH- a unM>*
F G
Cell Chemistry
Selected System:| NCA-G
Positive Electrode
Active material capacity, mAh/g: 160
Weight %
Active material
Carbon
Binder
Binder solvent
Void. Vol% %
Density, a/cm3
Active material
Carbon
Binder
Negative Electrode
N/P capacity ratio aft
Active material capac
Weight %
Active material
Carbon
Binder
Void. Vol% %
Active material
Carbon
Binder
Positive Foil
" system Selection Battei
=r formation
ty, mAh/g:
r
-------
Ji | NCA-G
D
Selected System:[
Positive Electrode
Active material capacity mAh/g:
Weight %
Active material
Carbon
Binder
g/cm:
13 Active material
14 Carbon
15 Binder
16 Negative Electrode
17 N/P capacity ratio after formation
18 Active material capacity mAh/g
19 Weight %
20 Active material
21 Carbon
22 Binder
23
24
2S
26 Active material
27 Carbon
28 Binder
29 Positive Foil
30 Material
31 Thickness. u.m
32 Negative Foil
33 Material
34 Thickness, ^m
35 Separator
36 Thickness, urn
37 Void, Vol% %
38 Density, g/cnr
39 Electrolyte density, g/cm3
40 Cell Voltage and Resistance Parameters
41 OCV at full power (at SOC Tor power rating). V
42 Open circuit voltage at 50% SOC V
43 Solid state diffusion limiting C-rate (10-s). A/Af
44 Negative electrode cm2/cm3
45 Positive electrode cmVcm
46 Electrode system ASI for power, ohm-cm'
47 Selected ASI value
_48 i At 50% SOC. 2-sec burst
49~] At 50% SOC. 10-sec burst
50 1 At bottom of SOC range. 10-sec burst
_51^ ASI correction factor for limiting current. %
52 Electrode system ASI tor energy, ohm-crrf
53 Maximum electrode coating thickness. u.m
54 Available battery energy, % of total
55 Selected % energy
56l microHEV and HEV-HP
57 PHEV
58 EV
59 Cell Materials Costs
60 Positive Electrode. $/Kg
61 Active material
62^ Carbon
~6§1 Binder PVDF
64 Binder Solvent (NMP)
65 Negative electrode material S-'k
66 | Active Material
J»7_ Carbon Black
68 Binder PVDF
69 Binder Solvent
70 Positive current collector foil. S/i
71 Negative current collector foil. S
72 Separators. S/nf
73 Electrolyte. S/L
74
75
1.25
330
5
Water
3.551
3.68
27
74000
890
30
18
23.6
30
3
51.92
100
Cell Chemistry
Default Values
-G j NCA-G
D 160
:','
5
NMP
32
4.78
1.825
7 1 77
5 1 25
D 330
95
0
5
er Water
34
4 224
5 1 95
• 1.10
mm Aluminum
20
>er Copper
12
20
50
6 046
> 1 20
51 3551
S 3 680
27
DO 74000
0 BQOO
18
6 236
30
3
)2 51 9
3 100
25
70
80
NCA-G
&;:;;! 3700
0:;;;:; 680
fl:!;!:! 1000
-------
'aC Model Beta [Compatibility Mode] - Microsoft
L^ New Window 3 Spirt
jjg Arrange All "1 Hide
Freeze Panes - 3 Unhid
Save Switch
ij workspace Wind
f* \ PHEV
ram for Calculating Performance and Materials Requirements
LiNi0.80Co0.15AI0.0502-Graphite
4 System Chemistry Input
5 Finished Cell Materials
G Positive Electrode g
7 Active material
8 Carbon
9 Binder
Battery 1 Battery 2 Battery 3 Battery 4 Battery 5
Void
Total
Vol %
12 Negative Electrode, g
13 Active material
14 ^Carbon
15 Binder
17 Total
18 Baance of Ceil
19 Positive foil, m;
20 Negative foil, m2
21 Separator, m2
22 Electrolyte . L
23 Positive terminal assembly, g
24 Negative terminal assembly, g
25 Thickness of cell container aluminum layer, um
26 Thickness of cell container (PET-AI-PP). urn
27 Density of cell container. g/crn3
28 Cell container (PET-AI-PP). g
29 Cell mass, g
30 Length-to-width ratio for positive electrode
31 Cell thickness, mm
32 Thickness of cell edge from positive electrode to outside of fold, mm
34 Top of positive electrode to top of terminal, mm
35 Cell Capacity Parameters
36 Positive active material capacity, mAh/g;
37 Positive electrode capacity. Ah/cm3
38 Negative active material capacity, mAh/g:
39 Negative electrode capacity. Ah/cm3
40 Negative-to-positive capacity ratio after formation
41 Cell Voltage and Resistance Parameters
42 OCV at full power, V
43 Open circuit voltage average for discharge. V
44 Electrode system ASI for energy, ohm-cm*
45 Excess negative area, %
46 Maximum allowable electrode coating thickness, ^m
47 Cell terminal contact voltage loss, % of cell OCV
48 Rate of terminal temperature rise at full power. °C/sec
49 Target % OCV at full power
50 % OCV at full power adjusted for thickness limit
51 Battery Input Parameters
J2 Vehicle type (microHEV, HEV-HP, PHEV, EV)
53 Duration of power burst (10 or 2). s
54 Battery power, kW
55 Number of cells per module
56 Number of cells in parallel
57 Number of modules in row
58 Number of rows of modules
59 Number of modules per battery
60 Cells per battery
61 Voltage drop for bus bar for packs with one row of modules. V
62 Battery pack insulation thickness, mm
63 Battery jacket total thickness, mm
64 Number of batteries manufactured per year
65 Cell Chemistry Input
66 Battery Performance and Design Input
67
63
69
70 Calculated Cell Parameters
71 Capacity, Ah
72 Cell group capacity
73 Cell capacity
74 ASI Calculation
75 Limiting current density, mA/cnr1
75 Limiting C-rate. A'Ah
77 Electrode system ASI for power, ohm-cm'
78 Current collector resistance parameter, ohms
79 Current collector ASI. ohms-cm2
Weight %
(89
I
100
Weight %
35
!
100
Thick., tim
20
12
20
r ym
jm
Density
478
1 825
1.77
2749
Density
224
1 95
1 10
1406
Density
270
892
046
120
192.95
13.01
1084
21680
11986
631
12617
0476
0517
0929
00693
9.5
31.2
100
150
2.2
23.0
579
192.39
12.97
10.81
21617
11947
629
125.75
0.571
0615
1 116
0.0711
96
31 7
100
150
22
236
599
191.80
12.93
1073
215.51
119.03
626
12530
0.726
0.774
1.419
0.0742
9.8
32.5
100
150
22
246
632
191 40
12.90
1075
215-05
11871
625
124.96
0893
0946
1 747
0.0775
101
33.4
100
150
22
257
668
191.10
1288
10.74
214.72
118.46
623
124.70
1.077
1 135
2 108
0.0813
104
34.3
100
150
22
26.9
708
1 30
80
1.0
1 00
15
160
0392
330
0 441
125
3551
3680
51.9
234
100
001
1 30
8.0
1 0
1 00
15
160
0392
330
0 441
125
3.551
3680
51 9
228
100
001
Program for Calculating Performance and Materials Requirements
LiNI0.80Co0.15AI0.0502-Graphite
Battery 1 Battery 2 Battery 3
005
80
80.0
PHEV
10
95
16
1
3
2
6
96
003
10
13
0 100,000
snts
0.05
SO
800
PHEV
10
110
16
1
3
2
6
96
003
10
13
100,000
meters
Wcm'
309
30.9
85
27
30.8
308
85
27
307
30.7
85
27
30.6
306
85
27
306
306
85
27
Figure 6.4 Top portion of Battery Design worksheet.
75
-------
Page
Layout IUJ Full Screef
Workbook Views
[7] Rulei [7] Formula Ba
G/j Gridlines J7| Headings
Show
f* j PHEV
B
D
H
J2_ Vehicle type (microHEV, HEV-HP, PHEV, f
53 Duration of power burst (10 or 2), s
54 Battery power. kW
55 Number of cells per module
56 Number of cells in parallel
57 Number of modules in row
58 Number of rows of modules
59 Number of modules per battery
60 Cells per battery
PHEV
61 Voltage drop for bus bar for packs with one row of modules. V
62 Battery pack insulation thickness, mm
63 Battery jacket total thickness, mm
64 Number of batteries manufactured per year
65 Cell Chemistry Input
65 Batteiy Performance and Design Input
67
66
69
10
50
16
1
3
2
6
96
003
10
13
100.000
PHEV
10
66
16
1
3
2
6
96
003
10
13
100.000
PHEV
10
80
16
1
3
2
6
96
0 03
10
13
100.000
PHEV
10
95
16
1
3
2
6
96
003
10
13
100.000
PHEV
10
110
16
1
3
2
6
96
0 03
10
13
100,000
Program for Calculating Performance and Materials Requirements
LiNi0.80Co0.15AI0.0502-Graphite
Battery 2
meters
iA/cm2
30.9
30.9
35
27
70 Calculated Celt Parameters
71 Capacity, Ah
72 Cell group capacity
_73 Cell capacity
74 ASI Calculation
75 Limiting current density. mA/cm*1
76 Limiting C-rate. AWi
77 Electrode system AS1 for power, ohm-cm2
73 Current collector resistance parameter, ohms
79 Current collector ASi. ohms-cm2
80 Total cell terminal AS! ohms-cm'
81 Cell and battery terminal connections, ohms
82 Total cell hardware and battery resistance, ohm-cnr1
83 Total cell ASI for power, ohm-cm4
84 Total cell ASI for energy fC/3 rate), ohm-cm2
85 Electrode Coating Thickness Calculation
86 Positive electrode thickness parameter, urn
87 Negative electrode thickness parameter, jim
88 Positive electrode thickness at adjusted % OCV,
89 Negative electrode thickness at adjusted % OCV.
90 Cell Area Calculation
91 Area determined at target % OCV
92 Area limited by max allowed electrode thickn
93 Cell area based on total ASI for power, cm2
94 Cell Dimensions
95 Number of bicel! layers (97% packing density)
96 Width of positive electrode,
97 Length of positive electrode,
98 Length of current collector tabs, mm
99 Width of terminals, mm
100 Length of terminal materia
101 Width of cell, mm
102 Length of cell, mm
103 Volume of ceil, cm3
104 Module Parameters
105 Weight of each cell group interconnect (copper), g
106 Module state-of-charge regulator assembly, g
107 Terminal heating factor. W/g
108 Terminal resistance factor A-ohms/cm
109 Module terminals, if more than one module (each 2.0-c
110 Module terminal resistance both terminals, oh
111 Module wall thickness (aluminum), mm
"112. Length of aluminum conductor, rnm
113 Thickness of aluminum conductors, mrr
114 Total weight of aluminum conductors, g
115 Balance of module materials, g
116 Module length, mm
117 Module width, mm
118 Module height, mm
119 Module volume. L
120 Module weight, kg
121 Calculated Battery Parameters
122 Total battery energy storage. kWh
123 Useable battery energy storage, kWn
124 OCV at full power. V
125 Nominal battery voltage (OCV at 50% SOC)
126 Battery power at target % OCV and SOC. kW
127 Maximum current at full power. A
128 Maximum current density at full power. mA/cm'
129 C-rate at full power, A/Ah
130 Coolant space above and below modules and at end of jacket, mm
131 Thickness of module compression plates (steel), mm
132 Battery pack le
H « > M »HV4J-J.L
Battery 3
298
0 005409
0781
0.073
0.000372
17
143
185
16
I. mm
135
26
1*5
215
249
hm-cm2
3CV, |im
OCV, urn
ness.
O.BS7 0.918
306 31.4
55.2 55.7
99.7 747
1107 830
90.0 747
100.0 830
7911 10524
8 758 8 732
8.758 10524
0.975
324
56.6
58 5
650
58.5
65.0
13.401
8.705
13.401
1.036 1 102
33.7 35 1
57.5 58.5
473 39.1
52.6 43.5
473 391
52 6 43.5
16523 19961
8.687 8.674
16523 19961
Ready Calculate
Battery Design Summary of Results Manufacturing Cost Calculations
Figure 6.5 Middle portion of Battery Design worksheet
76
-------
The Battery Design worksheet automatically receives input from the System Selection
worksheet. These values are shown in purple (Figures 6.4 and 6.6) and must not be altered on the
Battery Design worksheet. As explained above, cell chemistry values may be adjusted on the
System Selection worksheet. The operator provides battery design input in the aqua colored cells
(Figures 6.4 and 6.6). The battery input parameters on lines 54 to 58 (Figure 6.4) and lines 162
to 164 (Figure 6.6) are the only input values that the operator is required to provide to study a
group of batteries. The type of vehicle battery (microHEV, HEV-HP, PHEV, or EV) on line 52
in Figure 6.4, is another important variable to be specified. One performs the selection by typing
the name of the vehicle battery type in cell F52. While the correct spelling is important,
capitalization is not. This selection automatically determines the state of charge at which full
power is designated (thus, the open-circuit voltage and AST for full power) and the length of the
power burst (2 seconds for microHEV and 10 seconds for all others). It is expected that the
majority of the remaining default values should serve well for most batteries; however, the user
may also change to their exact specifications.
The cell capacity (lines 162 to 164 in Figure 6.6) can be set in any of three ways: (1) directly
specifying the capacity (Ah) on line 162, (2) specifying the total battery energy on line 163 or (3)
specifying the electric range of the vehicle (miles). Only one of the three lines should be filled in
and the others should be blank. The model will follow the directions of the top-most line with
non-zero values.
The number of batteries manufactured per year is selected on line 64 in Figure 6.4. Changing this
value from the default value of 100,000, which is the manufacturing rate for the baseline plant,
will change the manufacturing cost.
If it is desired to study more than five batteries in the same workbook it is only necessary to add
additional columns by copying the battery 5 column to the right as many times as desired. Care
should be taken that the appropriate values are maintained when the cells are copied over. The
aqua colored cells are typically the source of any problems. The same column additions most
also be done for all other worksheets containing calculations.
6.2.4 Remaining Worksheets
The cost calculations are done on the Manufacturing Cost worksheet and the results for the
model are shown on the Summary of Results worksheet (Figure 6.7). No parameters need to be
entered on these worksheets by the operator; all of the input for these worksheets is from the
Battery Design and the Cost Input worksheets. Tables for presentations or for preparing graphs
of the data can be assembled at the bottom of either the Battery Design or the Summary of
Results worksheet. These tables can be transferred to a blank worksheet for more complex
studies. For instance, results for different cell chemistries can be copied and pasted (special
paste, values and numbers formats) to a blank worksheet. On the last worksheets, the cell,
module, and battery design, as well as the baseline plant are sketched.
77
-------
: Model Beta [Compatibility Mode] - Microsoft E.
*V Home Insert Page Layout
Formulas Data Rev ew vie^
& Q = rgl y
! |7J1JJ Pag. Break Previ™ f ^^ ^ O PS Fg] ^ N™ Window E3 Spirt JJ |~ 1 CD
^ 13 Custom Wews « «™ W 9 arrange All — \ Hide j' "-•
Normal Page @ Gridlines El Headinss Z°°"i K»* Zoom to Sa.e Switch Macros
Layout U-l Full screen Selection gll Freeze panes * :, workspace Windows '
Workbook Views
Show Zoom Window Macros
F90 W- *| |*
ABC
D E F
90 |Cell Area Calculation
91 Area determined at target % OCV
92 Area limited by max. allowed electrode thickness.
93 Ce area based on total ASI for power, cm
94 Cell Dimensions
95 Number of bicell layers (97% packing density)
96 Width of positive electrode, mm
97 Length of positive electrode, mm
98 Length of current collector tabs, mm
99 Width ofterminals. mm
100 Length of terminal material, mm
101 Width of cell, mm
102 Length of cell, mm
103 Voumeofcell. cm*
104 Module Parameters
105 We ght of each cell group interconnect (copper), g
7911
8.768
8,758
17
143
185
16
135
26
145
215
249
0.0
106 Module state-of-charge regulator assembly, g 128
107 Terminal heating factor W/g 0.019
108 Terminal resistance factor. A-ohms/cm 0.00054
109 Module terminals, if more than one module (each 2.0-cm long), g 24
110 Module terminal resistance both terminals, ohms 0.0000120
1 1 1 Module wall thickness (aluminum), mm 0.5
112 Length of aluminum conductor, mm 185
113 Thickness of aluminum conductors, mm 0 40
114 Total weight of aluminum conductors, g
115 Ba ance of module materials, g
116 Module length, mm
117 Module width, mm
118 Module height, mm
119 Module volume. L
120 Module weight, kg
121 Calculated Battery Parameters
122 Total battery energy storage. kWh
123 Useable battery energy storage kWh
124 OCV at full power. V
125 Nominal battery voltage (OCV at 50% SOC)
126 Battery power at target % OCV and SOC. kW
127 Maximum current at full power, A
547
227
217
144
146
458
10.20
1071
750
340.9
3533
55.4
178
128 Maximum current density at full power. mA/cm2 20.31
129 C-rate at full power. A/Ah 5.8
130 Coolant space above and below modules and at end of jacket, mm 0.6
131 Thickness of module compression plates (steel), mm 1.5
132 Battery pack length, mm 461
133 Battery pack width, mm 471
134 Battery pack height mm
174
135 Battery volume, L 37.7
136 Weight of each module inter-connect (5-cm long), g 30
137 Weight of both battery terminals (each 5 0-cm long), g 74
138 Weight of module compression plates and steel straps, g 1537
139 Weight of bus bar for packs with one row of modules, g 0
140 Resistance of module interconnects if more than one module, ohms 0.0000903
141 Resistance of battery terminals
142 Power of battery heaters kW
143 Weight of battery pack heaters (0.1 kg'kW), kW
144 Battery jacket weight parameter, g/cm2
145 Battery coolant weight within jacket, kg
14G Battery jacket weight, kg
147 Battery weight kg
148 Pack integration (BMS & disconnects), kg
149 Pack integration (BMS & disconnects). L
150 Battery Cooling System
0.0000181
20
02
0.84
276
10.5
71.7
4.0
40
151 Heat generation rate for pack, W 3871
152 Cooling System Refrig 4
153 Weight and Volume of Cooling System Exterior to Battery
154 Weight kg 70
155 Volume. L 2.8
156 Vehicle Electric Range
157 Energy requirement. Wh/mile
158 Available battery energy. % of total
159 Vehicle range, miles
160 Cell Capacity Calculation
161 Select capacity, battery energy, or vehicle rang*
162 Capacity {Ah}
163 Battery energy jkWh)
164 Vehicle range (miles)
165 Capacity at C/3 Ah
166 Capacity holding
167 Positive electrode thickness
168 Positive electrode thickness holding
169 Convergence parameter
1701
1711
300
• 70
25.00
. but only one.
| 25.0
30.872
30.872
" 900
90.0
0.3
Restart (0/1) 1
"*t N 'BffiSsJH*' Battery i>esign bummary ot nesufcs rianura
G
10.524
8.732
10.524
19
145
188
16
137
26
147
218
256
00
123
0.019
000054
32
0.0000090
0.5
188
040
562
232
220
144
149
470
10.54
1071
750
3409
353.3
650
238
22.65
77
0.5
15
461
476
176
386
40
99
1577
00000674
00000135
2.0
0.2
084
258
10.3
735
4.0
40
3062
Refrig 4
70
2.8
300
70
25.00
25.0
30.782
30782
' 74.7
747
0.3
1
turing Cost C
H 1 1
13.401 16523
8.705 8 687
13.401 16.523
23 28
148 152
193 198
16 16
140 144
26 26
150 154
223 228
268 280
00 00
128 128
0019 0.019
0.00064 0.00054
39 46
00000073 0.0000061
05 05
193 198
0 40 0 40
588 616
239 246
225 230
144 144
152 156
4.91 5.14
11 10 11 72
1071 1071
7.50 7.50
340.9 340.9
3533 353.3
80 0 95 0
293 348
2189 21.08
96 114
0.5 04
15 1.5
461 461
485 495
179 183
40 0 41.7
49 58
122 145
1645 1719
0 0
00000547 00000461
00000109 00000092
2.0 2.0
02 02
0 84 0.84
239 226
103 10.5
77.0 80.8
40 40
40 40
2303 1829
Refrig 4 Refrig 3
70 50
2.8 2.0
300 300
70 70
25 00 25.00
25.0 25.0
30 688 30 624
30.688 30.624
' 58.5 ' 473
58.5 47.3
0.3 0.3
1 1
ilcuiations Cost Input
J K
19.961
8.674
19961
32
156
203
16
148
26
158
233
294
00
128
0.019
000054
54
0.0000053
05
203
0.40
648
255
235
144
160
640
12.41
10.71
750
340.9
353.3
110.0
403
20.21
132
0.4
15
461
506
187
435
67
168
1803
0
00000398
0 0000080
20
02
0.84
216
107
851
40
40
1501
Refrig 3
50
2.0
300
70
2500
25.0
30.576
30.576
1 39.1
39.1
0.3
L #f
B
1
Price ot Modules AiHuHFAHillslSBli] < >_Q
Ready Calculate
Figure 6.6 Bottom portion of Battery Design worksheet
78
-------
lodelBeta [Compatibility Mode] - Microsoft Excel
"^1 JJ Page Break Preview
l3 Custom Views
Normal Page g] rj
Workbook View!
idline! 0 Heading! Z°°"i ™* Z°°
Sele
Show Zoom
^ q£ New Window
~A „_
•* 3 Arrange All
nto
ton P Freeze Pane! -
23 split _u n nn
Save Swrtch
Ij Workspace Windows
Window
Macros
Macros
F5 - £ ='Battery Design'!F122
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
1B
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
31
32
A B • ~C D
E F G
H
Summary of Results
LiNi0.80Co0.15AI0.0502-Graphite
Battery 1 Battery 2 Battery 3
Calculated Battery Parameters
Battery energy storage kWh I 10 71 107
Battery power at target % OCV. kW 55 4 " 65.0
Required battery power, kW
Capacity. Ah
Number of cells
Battery weight, kg
600 650
31 31
96 96
71 7 73 5
Battery volume. L 377 38 S
We ght and Volume of Components Exterior to Battery
Weight kg 10 11 0
Volume L 5.6 5 6
Cooling system power requirement. W 1.551 1226
Vehicle electric range, miles 25.0 25 0
Investment Costs
Capital equipment cost including installation, mils 162 166
Bu Iding. Land and Ut iities
Area, m2 18.948 19433
Cost. S.'m! 3.000 3 000
Building investment. mil$ 56.8 58 3
Launch Costs
Rate" 5% of direct annual materials + 10% of other annual costs
Total, million! 1291 1332
Working capital (30% of annua variable costs), mill 3497 3613
Total investment mils 26647 273 36
Unit Cost of Battery Pack, $
Variable Cost
Materials and Purchased Items
Cell materials
Cell purchased Items
Modue
1.561 1.619
87 87
306 306
Battery pack 138 140
Total 2.081 2 153
Direct Labor
Electrode processing 37 40
Cell assembly 34 34
Format on cycling, testing and sealing 22 22
Modu e and battery assembly 20 20
Cell and materia s rejection and recycling 9 9
Receiving and sh pping 9 9
Control laboratory 6 6
Total 136 139
Variable Overhead 114 117
Total Variable Cost 2.332 2 409
General. Sales. Administration 137 140
Research and Development
119 122
Depreciaton 298 305
Total Fixed Expenses 554 567
Profits after taxes 133 137
Total unit cost per battery not ncluding warranty. S 3.019 3113
Summary of Unit Costs, S
Materials 1.551 1.619
Purchased Items not including cooling system 530 534
Direct Labor 136 139
Variable Overhead 114 117
General Sales. Administration 137 140
Research and Development
Depreciation
Profit
Warranty I ncludes battery pack only)
119 122
298 305
133 137
169 174
Price to OEM for battery pack. $ 3188 3287
Pack integration (BMS & Disconnects). S 435 435
Estimated cost to OEM for thermal management- S 320 320
Total cost to OEM for complete battery system, S 3943 4.042
Price to OEM for modules for one pack. $ 3014 3111
Chart Values
Vc
Ready Calculate
Range 25.00 25.00
Thickness 100 83
Weight 71.7 735
Volume 37.7 38.6
Cost 3014 3111
Cell Area 09 11
Itage (%OCV) 82 80
107
800
800
31
96
770
40-0
10
56
923
250
172
20177
3.000
60.5
1399
38.00
284 13
1.732
88
307
142
2.269
44
34
22
20
9
9
6
143
121
2.533
145
127
316
588
142
3.263
1 732
537
143
121
145
127
316
142
183
3.446
435
280
4.161
3.267
2500
65
77.0
40.0
3267
1.3
80
1
Battery 4
107
95.0
950
31
96
808
41.7
9.0
40
733
250
178
20.934
3000
628
14.72
40.06
295.40
1856
89
308
147
2398
48
34
22
20
9
9
6
148
125
2.671
150
131
328
609
148
3.427
1.855
544
148
125
150
131
328
148
192
3.619
435
240
4.294
3.436
2500
S3
808
41 7
3436
1.7
80
J
Battery 5
10.7
1100
110.0
31
96
85 1
43.5
90
4.0
601
250
134
21.722
3.000
652
15.52
4233
30738
1.991
90
309
152
2541
53
34
22
20
9
9
6
152
129
2822
155
136
340
631
154
3,607
1.991
551
152
129
155
136
340
154
202
3.809
435
240
4.484
3.620
2500
43
85.1
43.5
3620
20
80
K L M
"
N
.i.u...iaF^.i.
mrannfTU tITi
',n3D 100% - ~H±)
Figure 6.7 Summary of Results worksheet
79
-------
6.3 Battery Design Format Requirements
As the battery design is based off an assumed format (Section 2), certain design requirements are
necessary to ensure the modeled battery is physically realistic. The dimensions of the calculated
battery pack should be examined. Some final designs may benefit from changing the cell aspect
ratio, H/W, to fit the end-use application. One example would be, to change the height of the
battery pack. Also, for a set number of cells in the pack, changing the number of modules, thus
cells per module, allows for adjustment of the pack dimensions.
6.4 Troubleshooting and General Advice
The spreadsheet iterates to find the solution and this sometimes causes error messages to appear
after an entry is changed. These errors can usually be removed by first correcting any erroneous
entries (non-numeric, two decimal points, etc.). Then the cells may be reset to default values by
entering a "0" (i.e. zero) in the restart cell, F170 in Figure 6.6. Finally, entering a "1" in F170
restarts the iteration process leading to a successfully converged answer.
At some point, a user will ask the model to design a battery that is outside the bounds of what is
allowable for the selected cell chemistry. The most common error is when too large of a P/E ratio
is requested. Two different physical limitations are approached with increasing P/E ratio. First,
the electrode thickness is shrinking. At some point, the value will become unrealistic and
eventually approach 0 crashing the calculation. At the same time, the C-rate for the active
material is approaching the limiting C-rate defined in the Cell Chemistry Worksheet. As this
value is approached, the AST will increase to larger and larger values, which thus demands
smaller and smaller electrode thicknesses. Eventually, the calculation will crash.
Common sense approaches to resolve these issues are to use lower designed power or higher
designed energy. The C-rate and electrode thickness are easily viewed in the model output.
These are found on the Battery Design worksheet in row 129 for the C-rate and rows 88 and 89
for the electrode thickness. Therefore, the user may try designs of increasing P/E ratios and
watch to see how the electrode thickness and C-rate is changing. Different cell chemistries will
have different sensitivities to the P/E ratio depending on the defined limiting C-rate and
calculated AST for power. What is possible with the LMO-G system will not always be possible
with the NCA-G system. P/E ratios that satisfy the expression in Eq. 6.1 generally result in
successful battery designs. Higher P/E ratios are allowable in some situations. Note that selecting
the microHEV design doubles the allowable C-rate since only two second pulses are used. The
limiting C-rate, rc,um, may be found in cell E43 on the System Selection worksheet and is carried
over to row 76 in the Battery Design worksheet.
(6.1)
E 1.35
6.5 Suggested Number of Cells, Modules, and Performance Inputs
Table 6.1 presents some suggestions for the required inputs into the design model that might
change depending on the type of vehicle battery being designed. These values are only
80
-------
suggestions, but tend to be similar to practices used today or projected to be used in the future by
industry. If the calculated cell capacity is higher than 60 Ah, the user should consider the
inclusion of parallel cells as an additional parameter to examine. The default energy usage rate in
BatPaC is 300 Wh/mi. This rate may be used to size the energy requirement based on a desired
electric range for the vehicle by specifying the distance in row 164 on the Battery Design
worksheet.
Table 6.1 General suggestions for range of input parameters that change with battery type
battery type modules/battery OCV @ 50% SOC Power (kW) Energy (kWh)
HEV-25
HEV-HP
PHEV
EV
1-4
2-4
4-6
4-6
40 - 200
160 - 260
290 - 360
290 - 360
25
25-80
40 - 160
80 - 160
0.6-1.5
1-2
4-30
20 - 200
6.6 Entering a New Material Couple
The user of the model may wish to examine an electrochemical couple that is not included as one
of the options available in the model. We list below a brief explanation on how to properly enter
new materials into BatPaC. Various properties may be calculated, found in literature or measured
in the laboratory. The self-consistency of the data used is very important.
Experimentally measured values required:
1. Half cell formation cycling data from positive and negative electrode
2. Half cell cycling data from positive and negative electrode at C/3 rate
3. Full cell open-circuit voltage measurement at 50 % and 25 % SOC
4. Full cell ASI measurement for 5C pulse at 50 % and 25 % SOC
5. Full cell ASI at 50% SOC during a C/3 discharge
6. Electrode void fractions, active material densities, electrode component weight percent
7. Estimated interfacial area from surface area (preferred) or particle size measurements
The ASI calculation includes some additional parameters that become important as the designed
P/E ratio increases above 10 h"1 or the electrode thicknesses decrease below 30 microns. The user
is referred to section 3.4 and the supporting manuscript from Gallagher et al.20 for the parameter
estimation process. An exchange current of 0.15 mA/cm2, normalized to the surface area
calculated using the BET method from nitrogen absorption experiments, is used in the model in
row 77 in the Battery Design worksheet. While the exact value of the exchange current will vary
from the material to material, the general behavior of the ASI will be preserved with this
assumption. If lower P/E ratio designs are desired, the exact valuation of the exchange current,
interfacial area, and limiting C-rate are less important. However, the experimental ASI
measurement should then come from a cell with similar P/E ratio (electrode loading). Electrode
thicknesses 40 microns or larger should be used to minimize the contribution of interfacial
impedance to the ASI measurement. Otherwise the "ASI correction factor" may not accurately
remove the interfacial component. A reasonable approach for a first approximation of the ASI
parameters may be to select the same values for a similar material. For example, if the new
material is based on nano-sized primary particles, then the parameters for LFP or LTO may be
81
-------
close enough. The AST does depend on a large number of factors and a full determination of the
parameters is important to capture all of the physical behavior of the cell couple. If the desire of
user is to reproduce or to reverse engineer an existing cell, then many of the parameters may be
estimated from the electrochemical characterization of the full cell data. These electrochemical
results typical require a teardown of the cell to measure the area of the electrodes and their
loadings/thicknesses, assuming no legal contract has been established prohibiting the analysis of
the cell.
The available lithium for cycling in a full cell configuration may be calculated from half cell
measurements. The calculation method may be found in the Capacity Calculator worksheet in the
spreadsheet model and is detailed below. Alternatively, the reversible capacity of the full cell in
the experiment may be normalized to the mass of the positive and negative electrodes while
carefully accounting for the negative to positive capacity ratio. The first cycle efficiency of a
positive or negative electrode based half-cell may be defined as the ratio of the first discharge
capacity divided by the first charge capacity, Equation 6.2. We have assumed the first discharge
capacity is equivalent to the reversible capacity when measured against lithium foil (half-cell
arrangement).
Qrev
*= (6'2)
The quantity of lithium consumed from the positive electrode in the negative electrode SEI, gs
may be calculated from Equation 6.3. Here, [N/P] is the negative to positive capacity ratio.
Qsel =
After one full cycle, the remaining lithium in the positive electrode available for cycling, Qpaci, in
a full-cell configuration (positive electrode versus non-prelithiated negative electrode) may be
calculated by choosing the minimum value determined in Equation 6.4 below. Here we see the
possibility that the positive electrode is unable to accept the full amount of lithium released
during the first charge cycle. This so called "irreversible capacity" of the positive electrode
results in lithium residing in the negative electrode. While this excess lithium may require
additional negative electrode capacity, it also provides some beneficial aspects to cycle and
calendar life. ° We have chosen to set the [N/P] = 1.25 for layered oxides positive electrodes
(NCA, NMC441, NMC333) due to their tendency to have a lower first cycle efficiency ~ 88%.
The lithium manganese spinel and lithium iron phosphate cells have a high first cycle efficency
and thus we selected a [N/P] = 1.20. For positive electrodes with a high first cycle efficiency, the
reversible capacity of the cell is reduced by the lithium consumed in the graphite electrode SEI
during the formation cycle. Conversely, the lithium titanate spinel negative electrode does not
form an SEI and is significantly safer than the graphite electrode as discussed in Chapter 5.
Thefore the [N/P] ratio is set to 1.1 for the cells based on lithium titanate spinel.
{
Qf = MIN 27 i + —- - N/P —- 27 (6.4)
82
-------
7. Illustrated Results
The BatPaC model may be used to study the effects of battery parameters on the performance
and the manufactured cost of the designed battery packs. A few examples are given below for the
effects of various parameters on battery pack volume, weight and cost.
7.1 Number of Cells in Series
For a set battery pack power, the number of cells in the pack has substantial effects on the price
of the pack, the pack voltage and the maximum current. These effects are illustrated (Figure 7.1)
for NMC441-Gr PHEV25 batteries (providing 25-mile electric range) with 60-kW power at a
[V/U] = 0.8. The price of the pack increases by 17% in changing the number of series-connected
cells in the pack from 32 to 96 and the entire pack integrated cost increases by 15.7%. The
integrated cost includes additions to the vehicle air-conditioning system to provide for battery
cooling and the battery management system with disconnects. The change in the maximum
current, resulting from differing pack voltages, would also affect the cost of the motor and the
electronic converter and controller, but in the opposite direction. As a result of these offsetting
effects on the total cost of the electric drivetrain, a study is required to determine the optimum
current at maximum power as a function of the total battery pack power and other parameters
(see the Future Work section).
_ 4000
-g 3500 -
s.
-O
v
M
ta
3000 -
2500
2000 -
o
1500 -
O 1000 -
- 200 Q)
- 100 £
0 >
U
E
3
E
1
•o
E
re
20 40 60 80 100
Number of Series Connected Cells
120
Figure 7.1 The effect of the number of series-connected cells for NMC441-Gr, 60-kW, PHEV25
packs with 10.7 kWh total energy (70% useable).
Current PHEV battery technology uses battery packs containing 80-96 series connected cells.
However, these series connections are often composed of parallel cell groups. For instance, the
83
-------
battery used in the first production model of the Chevrolet Volt is in a 3P-96S configuration.21
Three low capacity cells are connected in parallel forming a parallel cell group. Then 96 parallel
cell groups are connected in series. The cost savings from moving to larger format cells with
only series connections is discussed later in this section.
7.2 Cathode Materials
Lithium-ion batteries for PHEVs and EVs do not require a high P/E ratio or low AST to meet
their goals. The most important material properties for performance are high specific capacity
(mAh/g), high cell voltage, and high electrode density. Since the graphite electrode is almost
universal in commercial cells (although not all graphite is the same), changes in the cathode
result in dramatically different calculated batteries. To compare the performance of EV battery
packs made from various Li-ion chemistries, we designed the packs to provide 150 kW at 360 V
(25% SOC) for a [V/U] = 0.8. Each pack consisted of six modules containing 16, 16, and 18
cells for the cell chemistries LMO-Gr, NMC441-Gr, and LFP-Gr, respectively. This calculation
assumes that large capacity cells may be reliably produced. Moving to a parallel connection of
smaller capacity cells would result in higher cost as discussed later in the section.
The NMC441-Gr system has excellent energy density and low cost (Fig. 7.2 and 7.3). The LMO-
Gr system is less energy dense than the NMC441-Gr couple, but equivalent in calculated price to
the OEM. The LFP-Gr system results in a battery that is larger and more expensive than the other
two chemistries. The mass-specific cathode raw material prices are 29, 20, and 10 $/kg for
NMC441, LFP and LMO respectively. The differences in initial cost do not directly translate to
the end cost of the battery. The performance (exhibited by specific capacity and voltage) affect
the quantity of both active and inactive material required. The NMC441 material achieves 175
mAh/g at a good cell voltage and is representative of an advanced, although close to
commercialization, layered oxide cathode.51 The combination in voltage and capacity results in a
superior energy density compared to the other cathodes. The LMO cathode has similar cell
voltage to NMC441 and low raw material cost but also a low specific capacity of 100 mAh/g.
This low capacity results in a positive electrode loading limited by the maximum achievable
electrode thickness -100 microns. The LFP electrode has moderate capacity, 150 mAh/g, and
raw material cost, but exhibits a lower cell voltage and electrode density. These poor
performance characteristics result in a low energy density battery with a high price.
7.3 Parallel-Connected Cell Groups and Electrode Thickness Limits
BatPaC also allows the user to create parallel cell groups and to set a maximum electrode
thickness. The effect these two unique design factors have on battery price are illustrated below
in Figure 7.4 for the LMO-Gr and NMC441-Gr systems. In this illustration, the PHEV battery
pack design parameters are 100 kW of power at a [V/U] = 0.8 and 17 kWh of total energy. The
nominal battery pack voltage (OCV at 50% SOC) is around 360 V from 96 cell groups connected
in series. The number of cells in the parallel cell group is varied from a single cell (no parallel
connections) to four. Two maximum electrode thicknesses of 100 and 200 microns are shown for
the LMO-Gr chemistry. In contrast to the NMC441-Gr, the LMO-Gr chemistry benefits form
allowing larger electrode thicknesses. The thickness of the positive electrode is limiting the
LMO-Gr chemistry while the thickness of the negative electrode limits the NMC441-Gr
84
-------
700
5? 600 -
500 -
_
,2,400
01
E 300
_3
o
^.200
s
ra 100
DO
-Volume NMC441
-MassNMC441
-Volume LMO
•MassLMO
-Volume LFP
-MassLFP
50
100
200
250
150
Vehicle range (mi)
Figure 7.2 Mass and volume of electric vehicle battery packs with lithium iron phosphate (LFP),
lithium manganese-spinel (LMO) and lithium nickel-manganese-cobalt oxide (NMC441)
positive electrodes versus graphite designed to deliver 150 kW of power at 360 V (25% SOC).
18000
•— • 16000 -
•l/V
14000 -|
u
O
O 12000 -
re
re
EQ
10000 H
sooo H
6000 -I
4000 -
2000 -
-NMC441
-LMO
-LFP
50
100
200
250
150
Vehicle range (mi)
Figure 7.3 Battery pack price to OEM for LFP-Gr, LMO-Gr and NMC441-Gr battery packs for
same designs as in Fig. 7.2. NMC441-Gr and LMO-Gr result in nearly the same price.
85
-------
chemistry. The calculated value for the NMC441-Gr system never exceeds 100 microns for this
P/E ratio. The LMO-Gr is the least expensive in all cases. However, the difference between the
two chemistries lessens with smaller limiting electrode thickness. The costs will become even
closer for lower designed P/E ratios. In general, thicker electrodes reduce the cost of the battery
pack by lessening the amount of inactive materials used (separator, current collector, etc).
Moving to 300 microns allows for greater savings in the LMO-Gr design but not the NMC441-
Gr design. However, a lower P/E ratio design for NMC441-Gr would take advantage of electrode
thicknesses greater than 200 microns.
&*-
The cell capacity is shown for the NMC441-Gr case limited to 100 microns. While the exact
values will change with cell chemistry, they will all be similar. The cell capacity is reduced by
one half as a single cell is added in parallel. This approach is commonly used by cell
manufacturers and OEMs that cannot reliably produce or successfully operate cells of high
capacity for transportation applications. However, this approach also increases the price of the
battery pack. In this example, the price is increased by ~ $500 when an additional string of cells
is incorporated in a parallel arrangement.
The model calculations show that the lowest cost battery pack will utilize thick electrodes and
large capacity cells. In current practice, these two approaches have yet to be successfully
implemented within the entire community. In the challenge of lowering costs, it is useful to point
out the largest gains come from the initial advances (e.g. moving from 100 to 200 micron limit).
After that point, the benefits are diminishing.
7.4 Manufacturing Scale
The effects of manufacturing scale come into the cost calculation even if the annual number of
packs produced is unchanged, but the design is altered (e.g. power is increased). For a fixed
design, the effect of changing the scale of operations depends on the fraction of the total price
that is made up of materials costs. Unit materials costs change little with scale whereas the costs
per pack for labor, capital and plant area may decline substantially with increasing production
rates, especially at low production rates, Fig. 7.5.
The lines in the graphs are for the best-fit power relationships through the data with power
factors of -0.076, -0.077, -0.147, and -0.211 from the top curve to that at the bottom. The least
negative power factor is for the battery pack with the highest fraction of materials cost in the
total pack cost. The more negative power factors result from a decreasing contribution of
materials cost as a fraction of the total pack cost. These power factors for equations of the cost of
a single unit can be converted to factors relating the total annual cost of manufacturing similar to
Eq. 5.2 by adding 1.0 to each power factor. Thus the factors become 0.924, 0.923, 0.853, and
0.789. These large factors show only a small to moderate effect of scale. When the power curves
are compared to the points in each of the graphs of Fig. 7.5, it is apparent that the scale factors
approach one as the scale increases. This is because the model assigns a value of 0.95 for the
active materials and 1.0 for the balance of the materials. As the production level increases and
the materials costs become a larger fraction of the total price of the battery, the scaling power
approaches 1.0 and the effect of scale become very small. Likewise, the effect of scale on battery
price is much larger for HEV batteries than for EVs because materials costs constitute a smaller
86
-------
portion of the total cost for HEV batteries. Increasing the production rate for HEV batteries will
result in a more dramatic reduction in cost than increasing the production rate for EV batteries.
6000
5500
_ 5000 -
•u>
2 4500 -
LJJ
O
O 4000
01
r: 3500
D.
3000
2500
2000
NMC441-Gr
1234
Number of cells in parallel group
65.0
55.0 ~
h 45.0 -f
TO
CL
TO
\- 35.0 ^
-------
LLI
14000
13000 -
12000 -
11000 -
10000 -
gjj 9000
8000 -
7000 -
6000
y=15609x
-0.076
y=13935x-°-077
EV150NCA-Gr
EV150LMO-Gr
10
100
1000
Manufacturing rate (1000's/year)
3500
3000 -
vv
uu
O
2
OJ
2500 -
2000 -
1500 H
1000 -
500
PHEV20LMO-Gr
HEV-25 LMO-Gr
= 4931.1x
-°-147
y=2462.3x
-°-211
10 100 1000
Manufacturing rate (1000's /year)
Figure 7.5 The effects of manufacturing rate on the price calculated by the model for battery
packs of various cell chemistries, power capabilities and vehicle types. The EV and PHEV
batteries are composed of 96 cells and the HEV-25 is composed of 48 cells.
-------
8. Future Work
8.1 Initial Power Designed at Differing Fractions of the Open-Circuit Voltage
The objective of this study is to evaluate the various benefits and costs associated with over- or
under- sizing the initial power of the battery utilizing Argonne National Laboratory's BatPaC
(battery performance and cost model) and Autonomie (vehicle model). The fraction of the open-
circuit voltage at which initial designed battery power is achieved has a direct impact on heat
generation during operation, performance under cold conditions, and acceptable power fade. The
vehicle focus will be on PHEVs for a mid-size vehicle glider and also a small SUV. Coupling the
vehicle and battery models will allow for a complete study evaluating the benefits and tradeoffs
associated with this critical design parameter. This work will be carried out in a cooperative
effort with Argonne's Transportation Technology R&D Center, which will perform vehicle
simulation tests to determine the rate of heat generation in the battery pack.
The end result of this study will be quantitative justification for designing a battery at a set
fraction of the open-circuit voltage. The justification will likely depend on battery chemistry and
powertrain type as differing architectures lead to differing benefits. These benefits may lead to
changes in the system design to maximize net present value. A secondary, but just as important,
benefit will be extrapolation of thermal management requirements from heat generation and cold
temperature performance calculated from various levels of sizing the battery.
8.2 Optimum Battery Voltage for Minimum Drivetrain Cost
For a set cell chemistry and set battery pack power, the cost of the pack increases as the number
of cells and the pack voltage are increased (Fig. 7.1). The additional cost results primarily from
the cost of additional state-of-charge equalization circuits and the additional number of cells
needing formation cycling and testing. The increase in battery pack cost is almost linear with the
increase in the number of cells. As the number of cells is decreased, the pack current at
maximum power becomes very high and the cost of the balance of the drivetrain increases at an
accelerating rate. Thus, there must be a number of cells and an associated pack current at
maximum power for which the total cost of the drivetrain is at a minimum. This minimum would
be for the current at full power for which the slope of the cost curve for the balance of the
drivetrain versus current was equal to the negative of slope of the cost of the pack versus current.
This optimum current will increase with the pack power because the slope of the cost-versus-
current curve increases with increasing power and, therefore, the optimum current will also
increase.
To represent these phenomena in illustrating the model in Section 7, we used an equation (Eq.
7.1) for selecting the current at full power as a function of the battery pack power. This equation
is just an estimate and it does not provide for differences in the optimum current that would
result from differences in cell chemistries, which are known to affect the battery cost versus
power function.
89
-------
A study is needed to determine the cost of electric motors and the electronic control equipment
required for the vehicle and battery pack as a function of power and maximum current capability.
Once appropriate cost curves are established, we will determine equations relating the optimum
battery current to the desired battery power taking into account the battery chemistry. We intend
to do this study with the cooperation of Argonne's Transportation Technology R&D Center.
8.3 Multipurpose Battery Manufacturing Plants
Our cost modeling is based on the concept that a manufacturing plant is constructed to produce a
single type of battery pack at a predetermined level of production. In practice, manufacturers will
have to produce several types of batteries within the same manufacturing plant and the levels of
production may fluctuate. We intend to investigate how this increased flexibility requirement
will affect the various manufacturing costs. For example, the cell filling and sealing equipment
may be required to handle cells of different dimensions. This would most likely increase the
capital cost for this equipment. Alternatively, additional packaging and sealing lines might be
needed. We intend to evaluate combinations of vehicle battery packs that are easily integrated
into the same plant. The manufacturing cost will most likely increase after these considerations
are built into the model.
8.4 Stand-Alone Graphical User Interface for Model
The spreadsheet program described here-in allows versatility in designing the battery and in
calculating the costs, but like all complex spreadsheet programs it is not user-friendly to those
unfamiliar with the details of calculation. In addition, the model is easily corrupted by a poor
choice of input parameters. As a result, the final spreadsheet program will be converted to a
stand-alone user-friendly application, primarily with the efforts of Ira Bloom. Visual Basic for
Applications will be used to hard code in the model calculations and to also create the graphical
interface. The new user interface should allow for a wide distribution of the model while
maintaining the ability to change the vast majority of input parameters. The retention of this
flexibility should make the model a valuable tool for those interested in batteries regardless of
the specific material property or manufacturing cost structure the user seeks to analyze.
90
-------
REFERENCES
1. M. Anderman, F. Kalhammer, and D. MacArthur. "Advanced Batteries for Electric Vehicles:
An Assessment of Performance, Cost, and Availability." California Air Resources Board, June
(2000). Available from http://www.arb.ca.gov/msprog/zevprog/2000review/btap report.doc
(accessed on December 17, 2010).
2. B. Barnett, D. Ofer, C. McCoy, Y. Yang, T. Rhodes, B. Oh, M. Hastbacka, J. Rempel, and S.
Sririramulu "PHEV Battery Cost Assessment," 2009 DOE Merit Review, May (2009). Available
fromhttp://wwwl.eere.energy.gov/vehiclesandfuels/pdfs/merit_review_2009/energy_storage/
es_02_barnett.pdf (accessed on December 11, 2010).
3. B. Barnett, J. Rempel, D. Ofer, B. Oh, S. Sriramulu, J. Sinha, M. Hastbacka, and C. McCoy,
"PHEV Battery Cost Assessment," 2010 DOE Merit Review, June (2010). Available from
http://wwwl.eere.energy.gov/vehiclesandfuels/pdfs/merit_review_2010/electrochemical
_storage/esOO l_barnett_2010_o.pdf (accessed on December 11, 2010).
4. A. Dinger, R. Martin, X. Mosquet, M. Rabl, D. Rizoulis, M. Russo, and G. Sticher, "Batteries
for Electric Vehicles: Challenges, Opportunities, and the Outlook to 2020." The Boston
Consulting Group, Available from http://www.bcg.com/documents/file36615.pdf (accessed on
December 11,2010).
5. L. Gaines and R. Cuenca. "Costs of Lithium-Ion Batteries for Vehicles," Center for
Transportation Research, Energy Systems Division, Argonne National Laboratory, ANL/ESD-
42, Argonne, IL May (2000).
6. F.R. Kalhammer, B.M. Kopf, D.H. Swan, V.P. Roan, M.P. Walsh, "Status and Prospects for
Zero Emissions Vehicle Technology: Report of the ARE Independent Expert Panel 2007,"
California Air Resources Board, April (2007). Available from http://www.arb.ca.gov/
msprog/zevprog/zevreview/zev_panel_report.pdf (accessed on December 17, 2010).
7. M. A. Kromer and J. B. Heywood, "Electric Powertrains: Opportunities and Challenges in the
U.S. Light-Duty Vehicle Fleet," Sloan Automotive Laboratory, Laboratory for Energy and the
Environment, Massachusetts Institute of Technology, LFEE 2007-03 RP, Cambridge, MA
(2007).
8. P. Mock, "Assessment of Future Li-Ion Battery Production Costs," presented at Plug-in 2009,
Long Beach, CA, (2009).
9. National Research Council of the National Academies, "Transitions to Alternative
Transportation Technologies - Plug-in Hybrid Electric Vehicles," The National Academies
Press, Washington, D.C. (2010). Available from http://www.nap.edu/catalog.php?record_id=
12826 (accessed on December 17, 2010).
10. TIAX LLC, "Cost Assessment for Plug-In Hybrid Vehicles (SOW-4656)," Report to US
DOE Office of Transportation Technology, October (2007).
91
-------
11. M. Anderman "The Plug-In Hybrid and Electric Vehicle Opportunity Report: A critical
assessment of the emerging market and its key underlying technology: Li-Ion batteries,"
Advanced Automotive Batteries, May (2010).
12. P.A. Nelson, DJ. Santini, J. Barnes, "Factors Determining the Manufacturing Costs of
Lithium-Ion Batteries for PHEVs." International Electric Vehicles Symposium EVS-24,
Stavanger, Norway, (2009).
13. DJ. Santini, K.G. Gallagher, P.A. Nelson, "Modeling the Manufacturing Costs of Lithium-
Ion Batteries for HEVs, PHEVs, and EVs," International Electric Vehicles Symposium EVS-25,
Shenzhen, China, (2010).
14. K.G. Gallagher, P.A. Nelson, D.W. Dees, "PHEV battery cost assessment" 2011 DOE Merit
Review Presentation Washington, DC May 9-12 (2011). Available from http://wwwl.eere.
energy.gov/vehiclesandfuels/pdfs/merit_review_201 l/electrochemical_storage/esl 1 l_gallagher_
201 l_o.pdf (Accessed on July 12, 2011).
15. P. Nelson, I. Bloom, K. Amine, G. Henriksen, "Design modeling of lithium-ion battery
performance," Journal of Power Sources, 110, 437 (2002).
16. P. Nelson, D. Dees, K. Amine, G. Henriksen, "Modeling thermal management of lithium-ion
PNGV batteries," Journal of Power Sources, 110, 349 (2002).
17. G.L. Henriksen, K. Amine, J. Liu, and P.A. Nelson, "Materials Cost Evaluation Report for
High-Power Li-Ion HEV Batteries," Electrochemical Technology Program, Chemical
Technology Division, Argonne National Laboratory, ANL-03/05, Argonne, IL (2002).
18. P. Nelson, K. Amine, A. Rousseau, H. Yomoto, "Advanced Lithium-Ion Batteries for Plug-in
Hybrid-Electric Vehicles", International Electric Vehicles Symposium, EVS-23, Anaheim, Ca
(2007).
19. P. Nelson, "Modeling the Manufacturing Costs of Lithium-Ion Batteries for PHEVs,"
presented at Plug-in 2009, Long Beach, CA, (2009).
20. K.G. Gallagher, P.A. Nelson, and D.W. Dees, "Simplified Calculation of the Area Specific
Impedance for Battery Design," Journal of Power Sources, 196 2289 (2011).
21. "Chevrolet Volt: Development Story of the Pioneering Electrified Vehicle" ed. Lindsay
Brooke, SAE International, Warrendale, PA pgs 215 (2011)
22. A.N. Jansen, K. Amine, and G. L. Henriksen, "Low-Cost Flexible Packaging for High-Power
Li-Ion HEV Batteries," Electrochemical Technology Program, Chemical Technology Division,
Argonne National Laboratory, ANL-04/09, Argonne, IL (2004).
92
-------
23. P.A Nelson and A.N. Jansen, "Comparative Costs of Flexible Package Cells and Rigid Cells
for Lithium-Ion Hybrid Electric Vehicle Batteries," Electrochemical Technology Program,
Chemical Technology Division, Argonne National Laboratory, ANL-06/43, Argonne, IL (2006).
24. J. Euler and W. Nonnenmacher, "Stromverteilun in Porosen Elektroden," Electrochim. Acta,
2268(1960).
25. J.S. Newman and C.W. Tobias, "Theoretical Analysis of Current Distributions in Porous
Electrodes," J. Electrochem. Soc., 109 1183 (1962).
Electrode thickness limitations
26. "Diagnostic Examination of Generation 2 Lithium-Ion Cells and Assessment of Performance
Degradation Mechanisms," ANL Report, ANL-05/21, July 2005.
27. D. Dees, E. Gunen, D. Abraham, A. Jansen, and J. Prakash, "Electrochemical Modeling of
Lithium-Ion Positive Electrodes during Hybrid Pulse Power Characterization Tests" J.
Electrochem. Soc., 155 (8) A603 (2008).
28. D. Dees, E. Gunen, D. Abraham, A. Jansen, and J. Prakash, "Alternating current impedance
electrochemical modeling of lithium-ion positive electrodes," J. Electrochem. Soc., 152
(7),A1409 (2005).
29. D. Abraham, S. Kawauchi, and D. Dees, "Modeling the impedance versus voltage
characteristics of LiNio.gCoo.isAlo.osCV Electrochim. Acta, 53 2121-2129 (2008).
30. D. P. Abraham, S. D. Poppen, A. N. Jansen, J. Liu, and D. W. Dees, "Application of a
lithium-tin reference electrode to determine electrode contributions to impedance rise in high-
power lithium-ion cells," Electrochim. Acta, 49 4763 (2004).
31. D. P. Abraham, E. M. Reynolds, E. Sammann, A. N. Jansen, and D. W. Dees, "Aging
characteristics of high-power lithium-ion cells with LiNiO.8CoO.15A10.05O2 and Li4/3Ti5/3O4
electrodes," Electrochim. Acta, 51, 502 (2005).
32. D.P. Abraham, J. Liu, C.H. Chen, Y.E. Hyung, M. Stoll, N. Elsen, S. MacLaren, R. Twesten,
R. Haasch, E. Sammann, I. Petrov, K. Amine, G. Henriksen, J. Power Sources, 119-121 511
(2003).
33. D. P. Abraham, R. D. Twesten, M. Balasubramanian, J. Kropf, D. Fischer, J. McBreen, I.
Petrov, and K. Amine, "Microscopy and spectroscopy of lithium nickel oxide-based particles
used in high power lithium-ion cells," J. Electrochem. Soc., 150 (11) A1450 (2003).
34. A.M. Andersson, D.P. Abraham, R. Haasch, S. MacLaren, J. Liu, K. Amine, "Surface
characterization of electrodes from high power lithium-ion batteries," J. Electrochem. Soc., 149
A1358 (2002).
93
-------
35. D. Abraham, J. Knuth, D. Dees, I. Bloom, and J. Christophersen, "Performance degradation
of high-power lithium-ion cells - Electrochemistry of harvested electrodes," J. Power Sources,
170 465 (2007).
36. J. Newman and K. E. Thomas-Alyea, "Electrochemical Systems," 3rd ed., Wiley
Interscience, New York (2004).
37. Wei Lai, Can K. Erdonmez, Thomas F. Marinis, Caroline K. Bjune, Nancy J. Dudney, Fan
Xu, Ryan Wartena, and Yet-Ming Chiang, "Ultrahigh-Energy-Density Microbatteries Enabled
by New Electrode Architecture and Micropackaging Design" Advanced Materials, 22, E139-
E144 (2010).
38. Pankaj Arora, Marc Doyle, and Ralph E. White, "Mathematical Modeling of the Lithium
Deoposition Overcharge Reaction in Lithium-Ion Batteries Using Carbon-Based Negative
Electrodes," J. Electrochem. Soc. 146 (10) 3542-3553 (1999).
39. Y. Chen and J.W. Evans, "Thermal Analysis of Lithium-Ion Batteries" J. Electrochem. Soc.
143(9)2708-2712(1996).
40. S.C. Chen, C.C. Wan, and Y.Y. Wan, "Thermal Analysis of lithium-ion batteries" J. Power
Sources 140 111-124(2005).
41. H. Maleki, S.A. Hallaj, J.R. Selman, R.B. Dinwiddie, and H. Wang, "Thermal Properties of
Lithium-Ion Battery and Components" J. Electrochem. Soc. 146 (3) 947-954 (1999).
42. M. Sievers, U. Sievers, and S.S. Mao "Thermal modelling of new Li-ion cell design
modifications" Forsch Ingenieurwes 74 215-231 (2010).
43. T.M. Bandhauer, S. Garimella, and T.F. Fuller "A Critical Review of Thermal Issues in
Lithium-Ion Batteries" J. Electrochem. Soc. 158 (3) R1-R25 (2011).
44. R. Byron Bird, Warren E. Stewart, and Edwin N. Lightfoot, "Transport Phenomena," 2nd Ed.
John Wiley & Sons, Inc, New York, NY (2002).
45. R.K. Shah and A.L. London, "Laminar Flow Forced Convection in Ducts" Advances in Heat
Transfer Supplement, Academies Press, New York, NY (1978).
46. R. Siegel, E.M. Sparrow, and T. M. Hallman, "Steady Laminar Flow Heat Transfer in a
Circular Tube with Prescribed Wall Heat Flux," Appl. sci. Res. Section A, 7 386-392 (1958).
47. M. Nickolay and H. Martin, "Improved approximation for the Nusselt number for
hydrodynamically developed laminar flow between parallel plates," Int. J. Heat Mass Tran. 45
3263-3266 (2002).
48. United States Geological Survery, Commodity Statistics and Information, Available from
http://minerals.usgs.gov/minerals/pubs/commodity/ (accessed on December 17, 2010).
94
-------
49. R.H. Perry, D.W. Green, and J.O. Maloney, Perry's Chemical Engineers' Handbook, Sixth
Edition, pp: 25-68 (1984).
50. M. Broussely, Ph. Biensan, F. Bonhomme, Ph. Blanchard, S. Herreyre, K. Nechev, and RJ.
Staniewicz, "Main aging mechanisms in Li ion batteries," J. Power Sources 146 90-96 (2005).
51. Sun-Ho Kang, Wenquan Lu, Kevin G. Gallagher, Sang-Ho Park, and Vilas G. Pol, "Study of
Lil+x(Mn4/9Col/9Ni4/9)l-xO2 Cathode Materials for Vehicle Battery Applications" J.
Electrochem. Soc. 158 (8) A936-A941 (2011).
95
------- |