TRW Document No. 18353-6006-RO-OO
APTD-1345
Environmental Protection Agency Contract No. 68-04-0028
APRIL 1972
FINAL REPORT
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF AIR PROGRAMS
Advanced Automotive Power Systems Development Division
Ann Arbor, Michigan 48105
TRW
SYSTf/US GROUP OF TRW IMC.
ONE SPACE PARK • REDONDO BEACH, CALIFORNIA 9O278
-------
TRW Document No. 18353-6006-RO-00
DIEVIElOP IHIllCGIHI CIHIMCGIE
AND
Dlls)CIHIA~CGIE M1rIE
lIEAD/ACHD IaA1r1rIERY 1rIECIHINOlOCGY
Environmental Protection Agency Contract No. 68-04-0028
APRIL 1972
FINAL REPORT
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF AIR PROGRAMS
Advanced Automotive Power Systems Development Division
Ann Arbor, Michigan 48105 .
TRW
SYSTEMS GIIOUI' M rRW: INC.
ONE SPACE PARK. REDONDO BEACH. CALIFORNIA 9021B
-------
FORWARD
This report represents a summary of the work performed by the TR\~
Systems Group, TRW Inc.
The work was performed for the Environmental
Protection Agency (EPA), Office of Air Programs, Advanced Automotive Powe~
Systems Development Division under Contract Number 68-04-0028.
period was 28 April 1971 through 18 March 1972.
Performance
Dr. H.P. Silverman, Program Manager, was responsible for the overall
direction of the project.
Dr. E. T. Seo was responsible for the TRW tasks.
Other major contributors from TRW were N.R. Garner, Dr. G.H. Gelb, R.S.
Margulies, M.L. McClanahan and Dr. R.R. Sayano.
designed the test equipment.
B. Berman and H. K. Gehm
Dr. R.E. Biddick, Manager, Battery Development, Gould Laboratories,
Gould Inc., was responsible for the work performed by Gould Inc. on Sub-
contract Number 029DHl with TRW Inc.
Other major contributors from Gould
were Dr. M.H. Little, Dr. F.L. Marsh, G.A. Mueller, R.J. Rubischko and
E.J. Sobcjzak.
The Environmental Protection Agency Proj~ct Officer was W.A. Robertson
of the National Aeronautics and Space Administration, Lewis Research Center.
Mr. Robertson worked for the EPA under a special technical assistance
agreement between NASA and EPA.
manager was Dr. J.T. Salihi.
The Environmental Protection Agency program
-iii-
-------
SUMMARY
TRW adopted two approaches to the development of a lead-acid battery
compatible with the requirements of a hybrid heat-engine/electric vehicle.
The first approach, a near term plan, was designed to have in production a
suitable battery within two years. The goals of this near term approach '-'ere
to develop a lead-acid battery capable of producing a peak power of 200 W/lb
without seriously altering the conventional manufacturing processes, cost
or lifetime of the battery.
The approach adopted, therefore, was to optimize
conventional pasted-plate technology for power.
A state-of-the-art Gould 22F-GP-61 SLI battery (Group Size 22F, 61 A
hr, 12 V) was evaluated by tests using both direct current (dc) and the TRW
electror.\echanical transmission (EMT) chopper modes. Discharge currents
during dc testing were 50 and 365 A and charge currents were 25 and 180 A.
During the EMT testing with a nominal I-kHz chopper frequency, the average
discharge currents were 54 and 350 A at a 50% duty cycle and the average
charge currents were 27 and 40 A.
Overall energy efficiency at steady
state was 79% for both dc and EMT tests.
However, during start-up, more gas
was evolved in the EMT tests resulting in a lower initial efficiency for the
EMT mode.
By modifying current pasted-plate design and refining present produc-
tion techniques it appeared probable that a battery meeting the performance
and cost objectives of the Environmental Protection Agency could be in
production by the end of 1972.
An analysis of state-of-the-art batteries
showed that significantly higher specific power could be achieved by
modifying cell components.
In order to obtain quantitative data which
would permit.a closer approach to an optimum design, component parameters
were tested in statistically designed experiments.
The experiments were
designed to indicate optimum plate thickness, paste density, grid design
and material.
The results established that thin light-weight plates using
a iow density paste and a lithium-lead alloy provided an optimum power
configuration.
Based on these results, a computer program was used to pre-
dict the current and potential distributions in various plate configurations
and to optimize the power to weight ratio.
As a result of these studies the cell element was redesigned.
Grid
resistance was reduced by redistributing the load.
More vertical structure
-v-
-------
elements were provided and tapered elements were used to provide more lead
in the areas carrying the most current.
The current collecting lug was re-
located, the plate was shortened and a more conductive alloy was used.
Elec-
trolyte resistance was reduced by the use of thinner plates and separators.
Lead-acid cells using the modified design were built that demonstrated
specific powers of 150 W/lb for 75 see (3 W hr/lb) and 204 W/lb for 20 see,
double the output of conventional prismatic cells. This improvement in power
was accomplished without requiring any major change in present production
methods and without seriously impacting costs. Modified batteries based on
these cells can be delivered on a production basis by 1973 at a cost com-
parable to the cost of conventional batteries.
The performance of the test cells exceeded the power requirements of
EPA specifications for a 550-lb battery system and approached those for the
450-lb goal with respect to both average and peak power.
Service life for these test cells was estimated by subjecting them to
a stringent, accelerated life test based on the EPA power requirements and
compatible with the performance time period of this program. The test cycle
contained 365- and 50-A discharges that corresponded to the 70.5- (and 55-)
and 10-kW rates of the EPA service-life requirements.
Jhe ratio of high-
rate to low-rate discharges was 1 to 29 and the charge rate was one-half
the discharge rate with the low-rate charge time adjusted to provide a 5%
net overcharge. Based on 1.5 V as the failure voltage for the high-rate
discharges, our test cells lasted 8,000 to 10,500 total cycles or 260 to
400 high-rate cycles.
Failure analysis indicated shorting through the separator.
Incom-
plete charging due to the shorting through the separators resulted in shed-
ding of the active material. Further tests will be carried out using a
more conservative separator material which i8 thicker and less porOus to
prevent shorting.
In fact, single-plate test cells, using a more conserva-
tive separator have survived 600 cycles at twice the current density of the
high-rate discharges of the cell tests.
On the basis of the test results, a cell of identical design, except
using thicker separators and 17 instead of 21 plates, is expected to meet
EPA specifications.
The full-size battery would ~onsist of 22 six-cell
units connected in series to give a nom~nal 264 V.
Weight and volume would
-vi-
-------
3
be 525 lb and 5.5 ft . respectively.
be three to five years.
Life of the battery is expected to
TRW's second, or longer term approach, was to consider advanced tech-
no logy that could lead to batteries producing 300 W/lb.
The bipolar plate,
because of its minimum electrical resistance,
absence of intercell connec-
tors, maximum current capability, and minimum volume and weight appeared
to be a good choice for further development.
The feasibility of the lead-
acid bipolar battery had been shown in earlier studies by Gould which demon-
strated average power levels in the range desired.
However, several tech-
nical developments were required before it could be used.
the development of
These included
.
a light-weight electronically conductive substrate to which
active material could be firmly attached and which would be
inert in the cell environment,
.
a method of attaching the active material utilizing Plante
formation or pasting techniques, and
.
a method of constructing a bipolar battery so that individual
cells are sealed from each other.
2 2
The performance goals were to sustain a 2 A/in. (0.3 A/cm ) discharge rate
for 60 sec with a cell voltage >1.5 V and a recharge time of twice the
previous period.
Achievements under this program included:
.
Fabrication of thin (0.06 cm), conductive vitreous carbon-
epoxy substrates, chemically inert to lead-acid cell en-
vironment, a~d with a resistivity of <1 ~ cm and a density
of ~l.4 g/cm ,
.
method for applying (pasting) the active material onto the
substrate,
.
negative bipolar plates t2at exceeded our performance tar-
get-of 2 A/in.2 (0.3 A/cm ) at 1.5 V for 60 sec and out-
performed standard pasted plates on the basis of a figure
of merit, and
.
positive bipolar plates that met the performance goal of
2 A/in.2 (0.3 A/cm2) at 125 of (52°C).
-vii-
-------
Pasted positive bipolar plates performed about the same as average
conventional pasted-plate positives. Cycle life of the best bipolar posi-
tive plate was 6,000 cycles including 600 deep discharges to 1.0 V. This
is believed to be about equivalent to what would be expected from a good
conventional positive plate.
The Plante formation of positive active material in a sufficient quan-
.22
tity to meet performance requirements of 2 A/in. (0.3 A/cm ) for 60 sec is
a major technical obstacle. The difficulty is shedding of the positive
material after only a limited thickness is achieved. Efforts to produce
more adherent material have failed, thus far. However, it is expected
that procedures now under consideration will produce a pasted positive bi-
2 2
polar plate that will have a long life capability at 2 A/in. (0.3 A/cm ).
This would lead to a projected battery producing 230 W/lb. It is estimated
that a prototype battery with these characteristics could be built by 1975.
i.
-viii-
-------
1.
CONTENTS.
INTRODUCTION. . . . . . . . .
. . . . .
. . . . . .
Page
1
1
. . . .
1. 1 PROGRAM GOALS AND APPROACHES
1.2
. . . . .
. . . .
. . . . . .
3
ACHIEVEMENTS. . .
7
2.
. . . . .
. . . .
. . . . . .
. . . . .
ANALYSIS OF PROBELM . .
7
. . . . .
. . . . . . .
2.1 HYBRID-VEHICLE BATTERY REQUIREMENTS. .
. . . .
. . . . . .
2.2
STATE-OF-THE-ART OF CONVENTIONAL (PRISMATIC) BATTERIES. . . . . 10
2.2.1
2.2.2
2.2.3
2.2.4
Factors Affecting Specific Power Output. . . . . . . . . 10
2.2.1.1 Effect of Current on Terminal Voltage. . . . . . 10
2.2.1. 2 Influence of Design of Battery Requirements. . . 11
Influence of Duty Cycle on Battery Performance. . . . . . 14
2.2.2.1 Overcharging. . .. . . . . . . . . . . . 14
2.2.2.2 Undercharging. . . . . . . . . . . . . . . 16
2.2.2.3 Charge Rate . . . . . . . . . . . . . . . . . . 16
2.2.2.4 Depth of Discharge. . . . . . . . . . . . . . . 17
Analysis of an Available SLI Battery.
. . . . . . 18
. . . .
2.2.3.1
2.2.3.2
Performance Characteristics. . . . . . . . . . . 18
Testing of a 61-A hr Group Size 22F Battery. . . 22
Discussion. . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.4.1 Identification of Technical Problems. . . . . . 33
2.2.4.2 Conclusions. . . . . . . . . . . . . . . . . . . 36
2.3
ADVANCED (ALTERNATE) CONCEPT BATTERIES. . . .
. . . . . . .
. . 37
2.3.1
2.3.2
Bipolar Battery. . .
... . 37
. . 38
. . . . .
. . . . . . .
Tubular Plate Battery. . . .
. . . . . .
...........
. . . . . . .
. . . . 40
L.4 APPROACH TO SOLUTION
3. . OPTIMIZATION OF CONVENTIONAL (PRISMATIC) BATTERIES.
. . 43
. . . . .
. . . . . . .
. . . . . .
. . 43
3.1 GRID STRUCTURE
3.1.1
3.1. 2
. . . .
Mathematical Modeling. . . . . . . . . . . . . . . 43
3.1.1.1 The Model. . . . . . . . . . . . . . 43
3.1.1. 2 Results. . . . . . . . . . . . . . . . . . 45
Accelerated Corrosion Tests of Selected Alloys. . . . . . 46
3.1.2.1 Approach and Selection of Alloys. . . . . . . . 46
3.1.2.2 Experimental Procedure. . . . . . . . . . . . . 48
3.1. 2. 3 Results and Conclusions. . . . . . . . . . . . . 51
-ix-
-------
3.2
CONTENTS (Continued)
Page
COMPONENTS-LEVEL TEST PLATES
..........
. . . .
. . . . 62
3.2.1
3.2.2
3.2.3
3.2.4
3.3
Concept Requirements and Performance-Test Design.
. . . . 62
Test Plate Fabrication. . . . . . . . . 66
Test Equipment. . . . . . . . . . . . . . . . . . . . . . 68
Performance Tests and Results. . . . . . . . . . . . . . 71
3.2.4.1 Preliminary Check-Out. . . . . . . . . . . . . . 71
3.2.4.2 Performance Tests. . . . . . . . . . . . . 78
3.2.4.3 Cycle-Life Tests. . . . . . . . . . . . . 89
TEST CELLS. .
. . . . .
. . . . . .
3.3.1
3.3.2
3.3.3
3.3.4
4.
. . . . . .
. . . . . . . . 89
Cell Design and Fabrication. . . . . . . . . . . . . . . 89
. . . . . . . . . . 96
. . . . . . . . . . . . . 99
3.3.3.1 Performance. . . . . . . . . . . . . . . . . . . 99
3.3.3.2 Cycle Life. . . . . . . . . . . . . . . .104
Summary and Recommendations. . . . . . . . . . . . . . .107
Test Procedures. . . . .
Tests and Results.. . . .
. . . . . .
. . . .109
DEVELOPMENT OF BIPOLAR BATTERIES.
. . . . . .
ELECTRODE STUDIES. . . . . . . . . . . . . . . . . . . . . . . .109
4.1.1 Objectives. . . . . . . . . . . . . . . . . . . . .109
4.1. 2 Conductive Substrates. . . . . . . . . . . . . . .109
4.1.3 Electrode Development. . . . . . . . . . . . . . . . . .116
4.1
4.2
. . . . .
. . . . . . . .125
. . . . .125
. .127
.131
ELECTRODE AND BATTERY EVALUATION. . .
. . . . .
4.2.1
4.2.2
4.3
Fabrication. . . .
Tests and Results.
. . . . . .
. . . . .
SUMMARY AND RECOMMENDATIONS. . . . . .
. . . . .133
5.
. . . . .
. . . .
PROJECTIONS AND RECOMMENDATIONS. . . .
. . . .
. . . .
5.1 PRISMATIC CELLS. . . . . . . . . . . . . . . . . . . . . . . . .133
5.1.1 Duty Cycle and Design Requirements. . . . . . . . .133
5.1.1.1 Battery Power Requirements. . . . .133
5.1.1.2 Battery Weight and Volume. . . . . . . . .138
5.1.2 Projected Characteristics of Cell and Full-Size Battery .138
5.1.3 Availability and Development. . . . . . . . . . . . . . .138
-x-
-------
CONTENTS (Continued)
Page
5.2
BIPOLAR BATTERY. . .
. . . . . . . . . . . . . . . . . . . . . .140
5.3
FlITURE WORK. .
5.3.1
5.3.2
Program Plan.
Work Statement.
APPENDIX A. .
APPENDIX B.
. . . . . . . . . . .
. . . .
. . . .
. . . . . . .
.141
. . . . . .
. . . .
. . . . . . . .141
. . . . . . . . . . .142
.147
. . . . . .
. . . . . . . . . . . . . . . . . .
.153
....... .... .......... " "
-xi-
-------
Number
10
11
12
13
14
15
16
17
18
19
LIST OF TABLES
1
Plate Capacity in Terms of Electrolyte Specific Gravity
for Several Temperatures. . . . . . . . . . . . . . . .
2
Page
. . 15
Weight Breakdown for 6l-A hr Group Size 22F Battery. . . . . 19
3
Electrolyte Distribution in 6l-A hr Group Size 22F
Battery Cell. . . . . . . . . . . . . . . .
. . . . . 20
4
Resistances of Components in a 6l-A hr Group Size 22F
Ba t t ery . . . . . . . . . . . . . . . . . . . . . . . . .
5
Percent of Total Resistance for Components in a 6l-A hr
Group Size 22F Battery. . . . . . . . . . . . . . . . .
6
Steady-State DC Performance of 6l-A hr Group Size 22F
Batt;ery. . . . . . . . . . . . . . . . . . . . . .
7
Losses Due to Resistance during DC testing for a 61-A hr
Group Size 22F Battery. . . . . . . . . . . . . . . . .
. . 23
. . 24
. . 27
. . 32
8
Discharge and Charge Cycles for a 55-kW Battery. . . . . . . 41
9
Resistances of Pasted-Plate Cells during Accelerated
Corrosion Test at 71 °c. . . . . . . . . . . .
Capacities of Pasted-Plate Cells during Accelerated
Corrosion Test at 71 °c. . . . . . . . . . . .
Growth of Pasted-Plate Specimens during Accelerated
Corrosion Test at 71 °c. . . . . . . . . . . .
. . . . 52
. . . . 53
. . . . 54
Growth-Rate Constants for Pasted-Plate Specimens at 71 °c. . 55
Estimated Growth-Rate Constants and Life Times for
Pasted Plates. . . . . . . . . . . . . . . .
. . . . . 59
Comparison Data for Rod, Bare-Grid and Pasted Plate
Specimens Obtained during Accelerated Corrosion Test. . . . 61
Test Plan Layout for Components Level Test Plates at
One Temperature. . . . . . . . . . . . . . . . . .
Test Sequence of Test Plates.
......
. . . .
. . 63
. . 65
75-Second Discharge Currents for Pb 660 Positive Plates. . . 74
Summary of i75V Values for Pb 660 Positive Plates. . . . . . 75
Average i7SV Values for Pb 660 Positive Plates. . . . . . . 76
-xii-
-------
:;umber
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
LIST OF TABLES (Continued)
Page
Analysis of Variance for Power (Pb 660 Positive Plates). . . 77
Properties of Components-Level Test Plates.
. . . .
. . . . 79
Distribution of Components-Level Test Plates between
Replicate Groups. . . . . . . . . . . . . . . . . . . 80
Figures of Merit for Specific Power Based on Plate
Weight, WIg. . . . . . . . . . . . . . . . .
. . . . . 81
Figures of Merit for Specific Power Based on Paste
Weight, WIg. . . . . . . . . . . . . . . . . . . . . .
. . . 82
Figures of Merit for Charge Acceptance for Components-
Level Test Plates. . . . . . . . . . . . . . . . . . .
. . . 83
Summary of Burn-In and Temperature Effects for Components-
Level Test Plates. . . . . . . . . . . . . . . . . . . . . . 84
Analysis of Variance for Specific Power for Components-
Level Test Plates. . . . . . . . . . . . . . . . . . . .
. . 85
Figures of Merit for Specific Power for Components-Level
Test Plates from Test Run 2 by Grid Type and Temperature. . 87
Sunnnary of Components-Level Performance Tests on
Negat~ve Plates. . . . . . . . . . . . . . . . . . . . . . . 88
Cycle-Life Test Data for Components-Level Test Plate
P089 at 52 °c. . . . . . . . . . . . . . .
. . . 90
Group Weights for Type IA Cells. .
. . 92
. . . . .
. . . . . .
Group Weights for Type IB Cells. .
. . . 94
. . . . . .
. . . .
Summary of Performance Test Data for Pasted-Plate Cells. . .100
Summary of Peukert-Plot Test Data for Pasted-Plate
Test Cells. . . . . . . . . . . . . . . . . . .
. . .101
Averaged Specific Power Values for Pasted-Plate Cells. . . .102
Anodic Corrosion Test Data on Substrate Material. . . . . .112
Results of Conductive Plastic Fabrication Experiments. . . .113
Formation and Performance of Plante-Type Positive Plates
Formed on Conductive Epoxy-Vitreous Carbon Substrates. . . .117
-xiii-
-------
Number
39
40
41
42
43
44
45
LIST OF TABLES (Continued)
Fabrication of Pasted-Plate Positives with Epoxy-Vitreous
Carbon Substrates. . . . . . . . . . . . . . . . . . . .
Page
. .119
Formation and Performance of Pasted-Plate Positives with
Epoxy-Vitreous Carbon Substrates. . . . . . . . . . . . . .120
Performance Data for Bipolar Positive Electrode PA12 . . . .123
Single Bipolar Electrode Weights. . . . . .
. . . . . .
Type II (Bipolar) Battery Electrode Weights.
. .126
. . . . . . . .128
Performance Test Data on Bipolar Plates. . .
Battery-Weight Constraints for Hybrid-Vehicle Systems
(92.5-kW Peak Power Demand). . . . . . . . . . . . . .
-xiv-
. .130
. . .139
-------
Number
10
11
12
13
14
15
16
17
18
19
20
21
LIST OF FIGURES
Page
1
Power Profile for Revised EPA Battery Performance
Requirements. . . . . . . . . . . . . . . .
. . . . .
9
2
Time ~. Voltage Trace during DC Test of 6l-A hr,
Group Size 22F Battery. . . . . . . . . . .
. . . . . 26
3
EMIT Transient Waveforms during Test of 6l-A hr, Group
Size 22F BAttery. . . . . . . . . . . . . . . . . .
. . . . 29
4
Gas Evolution during Test of 61-A hr, Group Size 22F
Battery. . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5
Outline of Grid for Pasted-Plate Test Cell. .
. . . .
. . . 47
6
Accelerated Corrosion Test Assembly for Bare Grids and Rods. 49
7
Growth Rate Plot for L1 and Sb-As Alloys. .
. . . . .
. . . 56
8
Second-Order Growth-Rate Plots for Sb-As Alloy. . . . . . . 57
9
Outlines of Grids Used in Components-Level Test Plates. . . 67
Block Diagram of Components-Level Tester. .
. . . . . .
. . 69
Schematic of Component-Level Test-Cell
. . . . .
. . . . . . 72
Type IB Test Cell. . . .
. . . .
. . . 9S
. . . . .
Battery Test Equipment. .
. . 98
. . . . .
. . . .
Typical Peukert Plot for Type IA Test Cell. . .
. . . .
. .103
End-of-Discharge Voltage for Cycle B vs. Cycle Number
for Test Cells. . . . . . . . . . . . . . . . . . . . .
IR Loss in Bipolar Substrate at 0.31 A/cm2 ~. Substrate
Resistivity at Various Substrate Thicknesses. . . . . .
. .105
. .110
Resistivity vs. Vitreous Carbon Content of a CIBA
1139/972 Substrate. . . . . . . . . . . .
. . .115
Two-Cell Bipolar Battery. . .
. . . .
. .129
Wheel Power Demands for a 4,000-lb . .
. . . .
. . . . . . .134
EPA Battery Power Profile for a 4,000-lb Series Hybrid
Vehicle. . . . . . . . . . . . . . . . . . . . . . . .
. . .135
Typical Histogram of Battery Current during a Driving
Cycle. . . . . . . . . . . . . . . . . . . . . . . . .
. . .137
-xv-
-------
1.
INTRODUCTION
The objective of this program is to develop a high charge rate, high
discharge rate lead-acid battery compatible with the requirements of a hy-
brid heat-engine/electric vehicle. Because the power requirements and charge-
discharge profiles for a hybrid-vehicle battery are sufficiently different
than those for a conventional lead-acid battery, a development program was
undertaken to optimize the lead-acid battery for hybrid-vehicle use.
The
goals of the program are to maximize specific power without significantly
affecting cost, lifetime or maintenance, As a result of this program ten
test cells producing specific powers of 150 W/1b for a 75-sec discharge and
204 W/lb for a 20-sec discharge were developed and tested.
1.1
PROGRAM GOALS AND APPROACHES
As a result of examining the lead-acid battery problem, a program
plan was developed which comprised two approaches to developing batteries
that can satisfy the hybrid-vehicle requirements. One approach was to adapt
conventional pasted-plate technology to a 55-kW battery (original EPA speci-
fications) with further optimization leading to a 200-W/lb battery system,
an anticipated requirement confirmed during the course of the program by
EPA's revised specifications. This approach was selected to provide a near
term solution with batteries being ready for production in two years. (EPA
specifications are detailed in Section 2.1.) The second approach, a longer
term effort, was to develop a bipolar battery which ultimately could lead
to a 300-W/lb battery.
The revised power-time specifications were evaluated in terms of vehicle
velocity-time characteristics for a 4,000-lb family car. The EPA specifica-
tions allow a series hybrid vehicle to accelerate to 60 mph in slightly
over 12.5 sec and reach almost 90 mph in 25 sec. This performance is in
line with EPA's vehicle design goals.
A state-of-the-art Gould 22F-GP-6l 8LI battery (Group Size 22F, 61 A
hr, 12 V) was evaluated by tests using both direct current (dc) and the TRW
electromechanical transmission (EMT) chopper modes. Discharge currents
during dc testing were 50 and 365 A and charge currents were 25 and 180 A.
During the EMT testing with a nomi~al I-kHz chopper frequency, the average
-1-
-------
discharge currents were 54 and 350 A at a 50% duty cycle and the average
charge currents were 27 and 40 A.
was 79% for both dc and EMT tests.
Overall energy efficiency at steady state
However, during start-up more gas was
evolved during the EMT tests, resulting in a lower initial efficiency for
the EMT mode.
By adjusting conventional pasted-plate design and by refining present
production techniques it appeared probable that the performance and cost
objectives of the original EPA specifications could be met by 1972. However
analysis showed that by optimizing components, higher specific power was
realizable. In fact, TRW had tested pasted-plate batteries with a specific
power exceeding 200 W/lb but their recharging characteristics were not satis-
factory. In order to obtain data which would permi~ a closer approach to
an optimum design, various design concepts, component modifications, and
parameters were tested in statistically designed experiments on single elec-
trodes and cells. Those experiments were designed to indicate optimum plate
thickness and paste density, and elucidate to some degree a preferred grid
design and material.
Accelerated corrosion tests were used to evaluate selected, more con-
ductive alloys. Both extensive and intensive tests were performed to find
improved alloys and evaluate their life by means of accelerated tests.
Computer programs were used to predict the current and potential dis-
tributions in various plate configurations, with the objective of optimiz-
ing the weight, uniformity of potential, and ease of fabrication of. the
battery.
Development of the bipolar battery which was recognized to be a longer
term effort, was undertaken because it offered the advantages of minimum
electrical resistance, absence of int~rcell connectors, maximum current
capability, and minimum volume and weight for use in a hybrid heat-engine/
electric family automobile. The feasibility of the lead-acid bipolar battery
had been demonstrated in rather long-term tests lasting from several months
to over a year at average power levels in the range required for the hybrid
car.
-2-
-------
1.2
ACHIEVEMENTS
Prismatic lead-acid cells that produce specific powers of 150 W/1b
for 75 see and 204 W/lb for 20 see were built and tested. This is almost
double the output of conventional prismatic cells.
The improved specific
power is achieved by accommodating a larger number of thin, light-weight
plates with the volume and weight of a conventional battery.
These thin
light-weight plates are the result of the development of highly conductive
alloys which are more corrosion resistant than the standard grid alloy, and
a new grid structural design.
The optimization studies which led to the
improved design were accomplished using a factorial experiment and a TRW-
developed computer program. No major change in present production methods
or serious impact on costs is anticipated. Modified batteries based on these
cells can be delivered on a production basis by 1973 at a cost comparable
to the cost of conventional batteries.
The performance of the test cells exceeds the power requirements of the
for the
EPA specifications for the 550-lb battery system and approaches those
450-lb goal with respect to both the 70.5-kW average and 92.5-kW peak
The EPA power requirements translate into the following specific
revised
power.
powers and currents:
Power, kW Weight, lb Minimum Voltage, V Specific Power, W/lb Current, A
92.5 550 200 168 462
92.5 450 200 206 462
70.5 550 200 128 352
70.5 450 200 157 352
55 550 150 100 367
Our test cells sustained currents of 418 to 500 A for 75 see, which corre-
sponded to specific powers of 143 to 170 W/lb for a 75-sec discharge and
peak currents of 600 A for 20 see which corresponded to 204 W/lb (4.7-lb
packaged cells). Thus in a 550-lb system the test cells would exceed the
original 55-kW and the revised 70.5-kW requirements for average power for
a discharge time which is 25-sec longer than the two 25-sec discharges of
the EPA requirement. The 204-W/lb peak power for 20 see at an average cell
voltage of 1.6 V approximates the two 9-sec discharges of the EPA specifica-
tions.
Both average and peak specific power values approach the goal of a
-3-
-------
45D-lb battery system.
Service life of the test cells was estimated by subjecting them to
a stringent, accelerated life test based on the EPA power requirements and
compatible with the performance time period of this program. The test cycle
contained 365- and 50-A discharges that corresponded to the 70.5- (and 55-)
and IO-kW rates of the EPA service-life requirements. The ratio of high-
rate to low-rate discharges was I to 29 and the charge rate was one-half
the discharge rate with the low-rate charge time adjusted to provide a.5%
overcharge. Based on 1.5 V as the failure voltage for the high-rate dis-
charges, the test cells failed after 8,000 to 10,500 total cycles or 260 to
400 high-rate cycles. The 10-kW level did not place a burden on the cell
performance.
Failure analysis indicated shorting through the separator and
softening and shedding of the positive active material.
Softening resulted
from incomplete charging due to the shorting through the separators and the
very stringent cycling conditions which are useful for comparing various
cell designs but hardly permit estimation of battery life in a hybrid vehicle.
Further tests should be carried out under conditions more representative
of hybrid vehicle duty, with more shallow cycles, and a more comp.lete charge
after an occasional high-rate discharge.
Because thin, porous separators were used, we feel confident that the
shorting through the separator can be corrected by a more conservative test
cell design using thicker, less porous material to prevent shorting. In
fact, a components-level single-plate test cell survived 600 cycles at twice
the current density of the high.rate discharges during the test cell cycling.
Test cells are now being retrofitted with more conservative separator material.
However, the testing of the cells will not be completed in time to be included
in this report.
On the basis of the test results, a cell of identical design, except
using 17 instead of 21 plates, is expected to meet EPA specifications.
full-size battery would consist of 22 six-cell units connected in series
to give a nominal 264 V. Weight and volume would be 525 lb and 5.5 ft3,
respectively. Life of the battery is expected to be three to five years
based on accelerated corrosion tests on new grid alloys and past experience.
The
-4-
-------
The develppment of bipolar batteries is a longer term effort than
the optimization effort, but could lead to a long-life, 300-W/lb battery.
The objectives of this program were:
.
a light-weight electronically conductive substrate to which
active material could be firmly attached and which would be
inert in the cell environment,
.
a method of attaching the active material utilizing Plante
formation or pasting techniques, and
a method of constructing a bipolar battery so that indivi-
dual cells are sealed from each other.
The performance goals were to sustain a 2 A/in.2 (0.3 A/cm2) discharge rate
.
for 60 sec with a cell voltage >1.5 V and a recharge time of twice the pre-
vious discharge period.
Achievements included:
.
Fabrication of thin (0.06 cm), conductive vitreous carbon-
epoxy substrates, chemically inert to lead-acid cell envi-
ronment and with a resistivity of <1 Q cm and a density of
~1.4 g/cm3. -
.
method for applying (pasting) the active material onto the
substrate,
.
negative bipolar plates that exceeded our performance tar-
get of 2 A/in.2 (0.3 A/cm2) at 1.5 V for 60 sec aud out-
performed standard pasted plates on the basis of figure of
merit ::lnd
.
positive pasted bipolar plates that met the goal of 2 A/in.2
(0.3 A/cm2) for 60 sec at 125 of.
Positive pasted bipolar plates performed about the. same as average
conventional pasted-plate positives.
Cycle life of the best bipolar positive
plate was a total of 6,000 cycles including 600 deep discharges to 1.0 V at
This is believed to be about what could be expected from
room temperature.
a good conventional positive plate.
The current density goal of 2 A/in.2 (0.3 A/cm2) for 60 sec was met
by the pasted positive bipolar plate in tests at 125 of. Plante formed
positive bipolar plates were not tested at high temperature because tests
at room temperature indicated that a sufficient quantity of Plante-formed
-5-
-------
positive active material could not be made to adhere to the plate. Active
material shedding caused short cycle life and efforts to produce a more
adherent structure by alloying the lead base material failed.
Based on single-plate results, it is conservatively estimated that a
prototype bipolar battery with pasted plates of the type tested will have
a specific power of 164 W/lb and 21 kW/ft3 for a 75-sec discharge to 1. 5 V
per cell. We estimate that a prototype battery of this type weighing 430 lb
and occupying 3.3 ft3 could be built in 1973, with production in 1975.
Over a longer term it is expected that the overall long-life current
2
capab~lity can be increased to 2 A/in. , resulting in a 33% reduction in
size, to four modules instead of six. This leads to a projected battery
3
weighing 300 lb and having a volume of 2.2 ft. Specific power would be
230 W/lb and 31 kW/ft3. It is estimated that a prototype battery with these
characteristics could be built in 1975 or earlier.
The organization of this report is essentially by task number, ex-
cept that for the sake of clarity the material covering the development and
testing of bipolar cells has been separated from the material concerning con-
ventional (prismatic) cells. Bipolar battery development is covered separately
. in Section 4.
-6-
-------
2.
ANALYSIS OF PROBLEM
2.1
HYBRID-VEHICLE BATTERY REQUIREMENTS
The impact of a hybrid-vehicle's operating profile on its battery
system was discussed in Section 1.1. Design goals for the battery and the
program plan to achieve the goals were dictated by EPA's original require-
ments in Exhibit I, RFP No. EHSD 7l~Neg. 100 and TRW's r~alization that a
much higher specific power (~200 W/lb) was needed and attainable. During
the course of the program, revised EPA requirements were provided by W.A.
Robertson, Project Officer, in Technical Direction No.3, 1 December 1971,
which approximated our original ~200-W/lb estimate.
Both EPA requirements of a lead-acid battery for a hybrid heat
engine/electric automobile are given below.
Exhibit I, RFP No. EHSD 7l-Neg. 100 (original requirements):
The full size vehicle battery shall be capable of the following
performance:
Power:
1.
55-kw discharge rate sustained for 25 sec and performed for t,,,o such
25-sec periods within a 60-sec interval.
2.
30-kW recharge rate sustained for 90 sec directly following the
two above discharges of Item 1.
Voltage:
200 to 220 V open circuit; 150 V minimum working voltage.
Life:
The battery should be capable of operating at the above power rates and
durations after operating for 5 years and 200,000 cycles which include
the following number and types.
Energy transferred
Number of Discharge Charge to and from
Cycles Rate, kW Rate, kW battery, W hr
500 55 30 380
3000 55 30 130
3000 55 30 80
balance of
200,000 10 5 30
-7-
-------
Weight:
550 Ib maximum.
Cost:
The battery should be capable of a mass production cost of less than $550.
ODeration:
.
Suitable for safe use in a family automobile without undue care and
maintenance.
Technical Direction No.3, 1 December 1971 (revised requirements):
The full size vehicle battery shall be capable of the following
performance:
Power:
1.
The power profile of Figure 1 with peak power of 92.5 kW and performed
for two such 25-sec periods within a 60-sec interval.
2.
A minimum recharge rate of 39 kW sustained for 90 see directly
following the two above discharges.
VoltaRe:
Minimum working voltage of 200 V.
Minimum cell voltage of 1.5 V.
Life:
The battery should be capable of operating at the above power rates and
durations after operating for 5 years and 200,000 cycles which include
the following number and types:
Energy transferred
Number of Discharge Charge to and from
Cycles Rate, kW Rate, kW battery, W hr
500 70.5 39 490
3000 55 30 130
3000 55 30 80
balance of
200,000 10 5 30
Weight:
550 lb maximum. (Goal 450 Ib max.)
-8-
-------
Figure 1.
?;
~
..
~
w
?;
o
a...
PEAK POWER = 92.5 KW
80
AVERAGE POWER = 70.5 KW
- -- - ---- ---------
----------
60
40
20
o
o
20
25
5
10
TIME, SEC
15
Power Profile for Revised EPA Battery Performance Requirements
-9-
-------
l.:ost:
The battery should be capable of a mass production cost of less than $550.
Operation:
Suitable for safe use in a family automobile without undue care and nlain-
tenance.
2.2
STATE-OF-THE-ART OF CONVENTIONAL (PRISMATIC) BATTERIES
2.2.1
Factors Affecting Specific Power Output
2.2.1.1
Effect of Current on Terminal Voltage
Specific power output is determined by all the factors that govern
the battery's terminal voltage under load, and by the weight of the battery
itself. The following equation developed by Shepherd (C.H. Shepherd,
"Theoretical Design of Primary and Secondary Cells,!; U.S. Naval Research
Laboratory, Report No. 6129, AD 617 333, 1964) gives the relationship
between the terminal voltage of a cell and certain fundamental Darameters
E == P.
8
- iK ( Q.) - iL
Q - It
where E is cell voltage, V; E is the open-circuit voltage, V; K is a
2 s. . d. A/ 2 Q. ]
polarization constant, n cm ; 1 15 current ens1ty, cm; 18 tle
limiting electrode's available capacity per unit area, C/cm2; L is the
internal resistivity per unit area, n cm2; and t is time, sec.
This equation was analyzed to determine the relative importance
of its factors.
The current density, i, should be as low as possible.
This means
a maximization of the electrode surface area, requiring, therefore, the
use of as thin a plate as possible consistent with maintaining physical
integrity for long life under the stipulated cycle conditions. Physical
. integrity is also governed by the paste density.
The internal resistivity, L, should be as low as possible.
This
constant represents the net resistivity of the separator, grids, electro-
-10-
-------
lyte, terminal~ and intercell connectors.
Through-the-partition connec-
tors in which the intercell connection is through the cell case wall,
r.ather than across the top of the battery, effectively minimize this
contribution to the overall intenlal resistance. The necessity of using
thin grids disallows any freedom in changing grid resistance except
through the antimony content
For example, pure lead at 20
4% antimony-lead alloy, 24 x
-6 .
x 10 u cm. The interelectrode
and possibly by rearranging structural members.
°c has a resisitivty of 21.2 x 10-6 u em,
10-6 Q cm, and 8% antimony-lead alloy 26.~
spacing is governed by the separator,
and separator resistance depends on the construction material, the thick-
ness of the backweb, size and pore-size distribution.
The factor Q/(Q - it) is essentially a measure of the state-of-
charge, and is a term that modifies the current density, i.
This means
that as active material is electrochemically consumed, the current
density becomes effectively larger.
In the case of the present application, .
hOT_lever J thestate-of-charge is always at a comparatively high level, and
the gross electrolyte concentration remains essentially unchanged during
the shallow cycles.
Th~ factor, K, is a coefficient of polarization, a highly critic~l
term involving such parameters as paste densities (porosities) and elec-
trolyte concentration.
Concentration polarization is a source of
voltage drop when the discharge current is high and prolonged, since the
concentration of sulfuric acid in the immediate vicinity of the active
material surface decreases.
2.2.1.2
Influence of Design on Battery Performance
Separators
The primary purpose of the separator in a lead-acid battery is to
prevent electronic conduction between the plates of opposite polarity
while permitting electrolytic conduction.
A wide variety of separators
is presently used in the industry.
These include types made of wood
veneer, perforated and slotted separators of hard rubber, microporous
rubber. and fibrous glass mats; additionally, separators made of micro.-
-11-
-------
porous p12stics, fibrous materials impregnated with resins, regenerated
cellulose films, and fabrics such as Saran, Dynel, and Vinyon are available.
As a general rule, ribbed separators are used in SL1 batteries.
This type of separator is flat on one side and ribbed on the other;
in the construction of the battery Eclements, the flat surface is placed
against the negative plate and the ribs against the positive.
The
ribbed separator construction, since it ensures the presence of free
electrolyte in the vicinity of the positive Dlate, ontimizes the high-
rate discharge characteristic of the battery.
The positive electrode discharge is in accordance with this
equation:
Pb02 + S04- + 4H+ + 2e
=
PbS04 + 2H20.
As can be seen, the reaction is such that not only is sulfuric acid con-
sumed during the discharge but water is created serving to dilute the
concentration further.
By the use of the ribbed separator, which
. establishes a region of free..flowing electrolyte near the positive
electrode, the transfer of liquid species is enhanced.
A resin-impregnated paper separator with embossed ribs coated with
polyvinyl chloride is typically used in SL1 batteries. It has an overall
thickness of 0.052 in. (0.132 cm) with a resistivity of approximately 0.23
~ cm2 measured in accordance to the AABM test methods. In this test the
separator resistivity is calculated by subtracting the resistivity between
battery plates spaced a fixed distance apart without a separator from the
resistivity measured with the separator between the plates; the specific
gravity of the sulfuric acid for this test is 1.280.
Other separators of
lower resistivities are currently available, for example, 0.052-in. (0.132 cm)
thick porous polyethylene separator recently tested was found to have a re-
sistivity of 0.12 ~ cm2.
Paste Density of Plates
Failure (incapability of the battery to deliver a stipulated percent-
age of capacity at a specified rate) occurs, in most instances, because of
\\
-,
,.\
-------
the shedding of material from the positive plate, and grid corrosion. This
condition is due, for the most part, to persistent overcharging and is accel-
erated by the use of plates not sufficiently dense for the service require-
ments. High-rate discharge polarization is particularly sensitive to the
porosity of the positive plate (J.F. Dittman, "The Optimum Performance from
the Lead-Acid Battery," SAE Paper 269C, 1961). As stated in the cited ref-
erence: ". . . the same amount of positive material at a low density of
50 g/in.3 would be capable of initially cranking an automobile twice as
long at 0 °c as would the same element design, but with the positive active
material at a high density of 70 g/in.3." However, since plates of low
density tend to limit cycle life, a compromise must necessarily be made
by adjusting the density of the positive plate to achieve the desired
life performance.
Specific Gravity of Electrolyte
The appropriate specific gravity of sulfuric acid must be selected
to optimize performance.
Since the capacity of. a positive plate is
sensitive, particularly at high rates, to the concentration of acid
immediately available to it, the concentration selected should be as
high as possible, but restricted by three considerations:
.
In general, separators deteriorate faster at higher acid
concentrations and temperatures.
The maximum specific gravity
to ensure separator integrity is 1.300 where the electrolyte
temperature is maintained below 43°C ("Battery Service Manual,"
6th Ed., The Association of American Battery Manufacturers, Inc.,
East Orange, New Jersey, 1964, p. 32).
.
At 25°C the minimum resistivity of sulfuric acid is at
31.1%; at this point the resistivity is 1.213 Qcm; 31.1% is
equivalent to a specific gravity of 1.227 at 25 DC. For
comparison, at 25 DC the resistivities of sulfuric acid of
specific gravities of 1.185 and 1. 265 are 1. 261 and 1. 231 Q em,
respectively.
.
The capacity of each of the plates varies with the specific
gravity and the temperature.
-13-
-------
The percent capacity with respect to the capacity at 25°C of posi-
tive and negative plates at three electrolyte specific gravities (G.W. Vinal,
IIStorage Bat'teries,1I 4th Ed., John Wiley and Sons Inc., New York,.l966,
pp. 224-225) is presented as a function of temperature in Table 1.
In summary then, the density of the positive plate should be
adjusted to optimize the specific power and to maximize cycle life.
If the battery is operated at elevated temperatures, e.g., 60°C, a
separator must be selected which will not deteriorate at that temperature
under the specified cycle conditions, and finally, the specific gravity
of the electrolyte should be adjusted so that its resistivity is at a
minimum at the anticipated operating temperature.
2.2.2
Influence of Duty Cycle on Battery Performance
Cycle life in traditional service is affected by charge and discharge
conditions.
2.2.2.1
Overcharging
Battery life is shortened by overcharging through the following
mechanisms:
.
Increased potential caused by overcharging a lead-acid battery
corrodes the positive grids.
.
The gases evolved by the decomposition of water tend to dislo4ge
active material from the plates.
.
The hi3her concentration of acid that results from the decom-
position of water is harmful to the separators.
.
Overcharging increases internal temperatures which accelerate
the deterioration of the cell components.
.
Overcharging, particularly, at high rates, by forcing electro-
lyteout of the cells can damage external components and
associated equipment.
-14-
-------
Table 1. Plate Capacity in Terms of Electrolyte Specific Gravity for
Several Temperatures
£apacity/Capacity at 25 DC, %
Specific Gravity Positives Negatives
10 DC 0 DC -10 DC 10 DC 0 DC -10 DC
1.315 87 78 67 65 43 25
1. 240 84 72 58 75 57 37
1.140 82 68 30 83 68 30
,
,
,.
-15-
-------
2.2.2.2
Undercharging
A battery that is repeatedly operated in an undercharged condition
over a ]ong period of time may develop a type of lead sulfate in the
plates, which, being dense and coarsely crystalline, cannot be readily
electrochemically converted into lead and lead dioxide.
Frequently, the
density of the lead sulfate is such that strains are set up, particularly
in the positive plate, causing distortion and buckling.
2.2.2.3
CharRe Rate
During overcharge, oxygen is evolve~ at the positive, and hydrogen
at the negative plate. This can cause loss of active material and shorten
the cycle life.
In addition, in cells with antimony alloy grids, stibine
is also liberated. An analysis was performed by Durant and co-workers
(K. Peters, A.I. Harrison, and W.H. Durant, "Charge Acceptance of the Lead-
Acid Cell at Various Charging Rates and Temperatures," in "Power-Sources 2,"
D.H. Collins, Ed., Sixth International Power Sources Symposium, Brighton,
Sussex, Pergamon Press, Ltd., Oxford, 1968, pp. 1-16) on the charge accep-
tance of positf'.Te and negative electrodes at three temperatures, 0, 25,
and 40°C. The highest charge rate employed in their experiments was 0.8C,
a rate substantially lower than what would be employed in the hybrid
vehicle application. For example, with a 60-A hr battery charged at 180 A,
as required after a 55-kW discharge, the charge rate would be 3C.
In the above study the charge efficiencies for the positive plate
at 25°C at the 0.8C rate were 99.1% at half-charge~ and 71.1% at near
100% charge.
For the negative plate under the same conditions, the
efficiencies were 100% at half-charge and 83.8% at near 100% charge.
At
O°c the charging efficiency of the positive plate was reduced to 34.3%
and the negative to 70% at near 100% charge.
The authors suggest that
the charge acceptance at the ?ositive plate is influenced by the produc-
tion and subsequent decomposition of persulfuric acid.
I t is apparent
that charge acceptance at high rates of charge may present a serious
problem in this application.
-16-
-------
Some poss~ble solutions to this problem would be:
1)
Maintaining the state-of-charge of the battery at a level lo~
enough to ensure a charge efficiency of close to 100% (about
75% state-of-charge).
2)
Charging the battery in such a fashion that gas evolution
is minimized. An example would be the imposition of high
currents at the beginning of charge, tapering to lower levels
as the state-of-charge increases.
The control would be an
empirically determined voltage cut off to ensure minimal
gas evolution, probably on the order of 2.35 V per cell plus
the iR voltage component.
2.2.2.4
Depth of DischarRe
Although no specific information is currently available on the rela-
tionship between depth of discharge and cycle life for automotive batteries,
an example takeu from the testing of Gould Motive Power Batteries is useful.
GSA Federal Specification W-B-133a for batteries of this type requires a
minimum of 1000 cycles when discharged at the 6-hr rate to 80% of rated
capacity and 2500 cycles when discharged to 45% of rated capacity at the
same rate. Tests conducted by Gould's Industrial Battery 'Division have
shown that the cycle life for the first mentioned test exceeds 1500 cycles,
and for the second, 3500 cycles. The total time for each cycle in the first
test is 12 hr, for the second, 8 hr.
An equation for expressing a semi-logarithmic relationship between
cycle life and depth of discharge has been reported by Voss and Huster (E.
Voss and G. Huster, "The Effect of Depth of Discharge on the Cycle Life of
Positive Lead-Acid Pla'tes," VARTA, Germany, unpublished connnunication) where
the depths of discharge are 40 to 100%. If a 6l-A hr 5L1 battery were used
in a hybrid-vehicle application when the single-pulse discharge is about 2.5
A hr or 4.1% depth of discharge, the life of the 6l-A hr battery would be
calculated as 300 cycles, compared to 24 and 127 cycles at 100 and 40% depths
of discharge, respectively~ A more typical life at 40% depth of discharge
would be 200 to 250 cycles at 40°C. This would lead us to expect a cycle
-17-
-------
life of at least the 500 cycles'required for a hybrid vehicle.
No serious difficulty is envisaged in complying with the cycle life
requirements for this application, provided that: 1) overcharge is kept
to a minimum, 2) electrolyte temperatures are maintained at such a level
that the separators are not deleteriously affected and corrosion is not
excessive, and 3) the charge rate does not cause excessive gassing.
2.2.3
Analysis of an Available 8L! Battery
2.2.3.1
Performance Characteristics
A battery must have minimum weight and ohmic resistance for optimum
performance.
Capacity, as such, is not important.
With this in mind,~the
resistances and weights of various components of a 6l-A hr, super premium c;
grade Group Size 22F battery (Gould 22F-GP-6l) were analyzed. Overall case
dimensions of this unit are: 9 7/l6-in. long x 6 7/8-in. wide x 7 5/8-in.
high (24.0 em x 17.5 em x 18.7 cm) for a volume of 480 in.3 (7870 cm3) and
a weight of 35.9 lb (16.3 kg). The specific energy and energy density are
18.7 W hr/lb (41.2 W hr/kg) and 1.52 W hr/in.3 (0.093 W hr/cm3), respec-
tively.
Weight
Each cell element has six negative plates, 0.053-in. (0.135-cm) thick
and five positive plates, 0.073-in. (0.185-cm) thick. The weight breakdown
of various components of the battery (6 cells) is given in Table 2. The
volume and weight distribution of electrolyte per cell in the 61-A hr
Group Size 22F battery is given in Table 3. Using these values, a weight
reduction of 2.04 lb (929 g) per battery could be. achieved by designing
the battery case so that sediment well and excess electrolyte above the
separators is eliminated.
In the current SLI batteries the sediment well
is necessary in order to accommodate material which sheds, and the excess
electrolyte provides for reserve capacity (at low-rate discharge). In
hybrid application, due to low depth of discharge, shedding of active
material should be small and reserve capacity is not required. Positive
electrodes can be encased in a separator envelope to ensure against
possible shorting due to some shedding which will occur.
-18-
-------
Table 2. Height Breakdown for 61-A hr Group Size 22F Battery
Cornponen~ Quantitv Mean Height, 1b (g) Total Height, lb (kg) % of Total
. . . .. ""- ,
Positive Paste 30 0.209 (95) 6.28 (2.9) 17.5
Negative Paste 36 0.174 (79) 6.26 (2.8) 17.4
Positive Grid 30 0.125 (57) 3.75 (1. 7) 10.4
Negative Grid 36 0.106 (48) 3.82 (1.7) 10.5
Terminal Strap 2 0.187 (85) 0.37 (0.17) 1.04
Internal Strap 10 0.116 (53) 1.16 (0.53) 3.2
I
~
\D Cover 1 0.469 (213) 0.47 (0.21) 1.3
I
Container 1 1.52 (691) 1.52 (0.69) 4.2
Separators 60 0.014 (6.4) 0.84 (0.38) 2.3
Electrolyte 11.4 (5.2) 31.8
35.87 (16.3) 100
-------
Table 3.
Electrolyte Distribution in 6l-A hr Group Size 22F Battery Cell.
Space WeiStht , lb. (g) Volume, 3
em
Total 1. 90 (864) 690
Sediment Well 0.168 (76) 66
Above Separators 0.271 (123) 98
-20-
-------
Resistance
The intercell-connector resistance measured 0.08 ron, and since there
are five connectors per battery, the total contribution for intercell
connectors is 0.40 m~.
resistance of 0.08 ron.
Each of the two terminal connectors has a
Resistance contributions from interplate electrolyte and separators
were calculated for 1.265 sp gr acid at 24°C (which has a resistivity of
1.23 Q cm), sep~rator resistivity of 0.23 ~ cm2, plate area of 177 cm2
and plate separation of 0.135 cm. The total resistance of the separators
and the electrolyte between two plates is 2.2 roO and since there are 10
separators in a cell, the resistance for a cell is 2.2 ron/lO or 0.22 ~1.
The total of the terminal and intercell connector resistances and
the electrolyte resistance of the battery is 1.9 ron. The measured resis-
tance of a typical Group Size 22F battery at room temperature is 10.8 ~;
the difference of 8.9 roO must be attributed to the grids, the active material,
and the pores.
The effective resistance of a grid is 2.5 roO or a grid re-
sistance of 0.5 roO per grid group.
Thus the sum of the resistance of the
positive and negative grids is 1.0 roO per cell.
Since the thickness of the positive plate is 0.19 em, by assuming a
porosity of 40% and a tortuosity factor of 2, the electrolyte resistance
per plate is 2.0 roO or 0.20 ron per cell. It should be noted that.this cal-
culation is not particularly accurate because of the assumption of the
tortuosity factor.
The only remaining resistance is that of the active material. A typi-
cal pellet of active material is 0.19-cm thick and the total cross-sectional
area of pellets in a plate is about 54 cm2. Since on the average the current
travels half the maximum distance through the pellet, and assuming a path
. tortuosity factor of 2.5 and porosity of 40%, the pellet resistance is esti-
mated as 1.8p per cell, where p is the resistivity of the positive active
material. If the positive plate is completely formed and charged, the active
material is Pb02 and the resistivity p, is about 10-2 ~ em. This would yield
-21-
-------
an estimated positive
likelihood the actual
fall in the range of
pellet resistance of 0.108 mQ for a battery. In all
pellet resistance is greater than this and should
0.1 to 1 mQ.
By comparison, resistance of negative active material should be
entirely negligible, provided the battery is not entirely discharged, be-
cause of the high conductivity of metallic lead.
The calculated battery resistances are summarized in Tables 4 and 5.
The greatest uncertainty in these calculations is believed to be
the estimate of active material resistance, but this is not likely to
be in error by more than 0.5 mQ.
It can be seen that the greatest source of resistance in a fu1ly-
charged battery at room temperature is the grid, amounting to about 60%
of the total. Electrolyte resistance contributes another 30%, leaving
only 10% contribution from connections and active material.
2.2.3.2
Testing of a 6l-A hr Group Size 22F Battery
Testing was performed on the Gould 22F-GP-61 SLI battery (Group Size
22F, 61 A hr, 12 V) in order to establish performance behavior of lead-acid
batteries cycled in a manner similar to EPA specifications.
The Group Size
22F battery was selected for this test because it represents a "state-of-the-
art" SLI battery with the specific power output and performance characteristics
most nearly approaching the EPA requirements.
Test Method
The batteries were tested using the cycle routine developed from the
original EPA specifications which require discharges at two power levels, 55
and 10 kW, based on a 200- to 220-V battery system. Furthermore, to sim-
plify the cycling routine the shallow 55- kW discharges were replaced wi th
deep (25-sec)
The batteries
tromechanica1
discharges resulting in a much more stringent cycling routine.
were cycled using both direct-current (de) and the TRWelec-
transmission (EMT) chopper modes. The cycle routines were as
-22-
-------
Table 4.
Component
Connectors and Terminals
Grids
Interplate Electrolyte
and Separators
I
N
VJ
I
Electrolyte in Pores
of Active Material
Active Material (est.)
Total
Measured
Battery.1
Fully Charged
"-'75% Charged
Battery 2
Fully Charged
"-'75% Charged
Resistances of Components in a 61-A hr Group Size 22F Battery
~OC (-40 oF)
0.44
Lf. 7
9.0
15.0
0.50
29.6
Resistance, Inn
-180C (0 OF) 240C ( 7,5 ~11 52°C (125 oF) 60 °c (140 ° F)
, ,
0.49 0.56 0.62 0.65
5.4 6.0 6.6 6.9
3.4 1. 32 0.90 0.47
5.3 2.02 1.38 0.72
0.50 0.50 0.50 0.50
15.1 10.4 10.0 9.2
10.8
11.1
10.2
10.8
-------
Table 5.
Percent of Total Resistance for Components in a 61-A hr Group Size 22F Battery
Resistance %
Components -400C (-40 OF) -18 0c (0 OF) 24°C (75 OF) 52 °c (125 OF) 60°C (140 OF)
~'. . .
Metal (Connectors, 17 38 63 82 88
Grids, Terminals)
Electrolyte 81 58 32 13 6
Active Material (est.) 2 4 5 5 6
I
N
~
I
-------
f0110\-1s:
Direct Current
EHT
Cycle A
Discharge at 50 A for
10.8 sec; charge at
25 A for 22 sec.
50 A for 10.8
sec; 25 A for
22 sec.
Cycle B
"Discharge at 365 A
for 25 sec; charge
at 180 A for 50 sec.
365 A for 25 sec;
40 A for 225 sec.
The sequence involved thirty of Cycle A followed by one of Cycle B with the
entire sequence then being continuously repeated.
Due to the inability or
the EMT to deliver a l80-A charging current for 50 sec, the Cycle B charg-
ing period was changed to 40 A for 225 sec.
parameters were monitored:
The following performance
o
Battery discharge voltages,
battery charge voltages,
a
o
currents during charge and discharge,
cell or electrolyte temperatures,
a
o
outer skin temperature (the temperature of the outer surface
of the battery case), and"
gas evolution rates.
o
A typical recorder chart showing terminal voltage during these cycles under
dc testing is presented in Figure 2.
Results and Discussion
The steady-state test results for a battery cycled under dc conditions
are shown in Table 6.
The data represent a total of 6859 cycles (sum of
IIAII and "BII cycles, total test time 65 hours).
Since the EMT had to be
operated manually to achieve the average constant currents of the test cycles,
the sequence was repeated only three times (total time, 62 min).
Test re-
suIts indicate that steady-state operation was achieved during this time.
EMT results are similar to those obtained under dc conditions.
That is,
with the EMT the average voltages during a Cycle A charge and discharge were
13.8 and 12 V, respectively. During Cycle B, voltages of 14.2 and 9.2 V
were observed for charge and discharge, respectively.
-25-
-------
Figure 2.
15 . 33 V
\
77V
.,
. 12. 10 V
10 MIN
"B" CYC LE
L
30 "A" CYCLES
Time~. Voltage Trac~ during DC Test of 6l-A hr,
Group Size 22F Battery.
-26-
-------
Table 6.
Steady-State DC Performance of 61 A-hr
Group Size 22F Battery
Cycle A Cycle B
Dischar~ Char~ Discharge Charge
Voltage Range, V 12.1-12.4 13.1-13.2 9.0-9.1 15.2-15.4
Average Voltage, V 12.3 13.2 9.0 15.3
Current Range, A 50.5-51. 5 25-27 350-39:' 166-200
Average Current, A 51 26 373 182
Equilibrium Cell Temperature, of
Cell 1 130-136 130-136 130 135
Cell 2 137-144 137-144 137 142
Cell 3 140-147 140-147 140 146
Cell 4 142-149 142-149 142 148
Cell 5 138-148 138-148 138 142
Cell 6 128-134 128-134 129 134
Equilibrium Skin Temperature, °F*
Cell 4 124-126 124-126 124 125
Gas Evolution per Cycle, cm 3 0.60 17
Cou!ombic Efficiency, % 99.4 99.0
* Center of container side.
-27-
-------
The average discharge and charge currents under EMT conditions were
54 and 27 A, respectively, for Cycle A, and 350 and 40 A for Cycle B.
Typical peak currents during Cycle A were 105 A at 50% duty cycle during
discharge and 90 A at 30% duty cycle during charge. Peak currents during
Cycle B were 480 A at 73% duty cycle during discharge and 90 A at 45% duty
cycle for charge. Figure 3 shows the transient waveforms observed under
EMT conditions with a nominal I-kHz chopper frequency.
When cycling of the battery was initiated at room temperature, be-
cause of the higher resistance of the battery (10.2 ~ at 23 °c compared
to 8.6 ~ at steady-state operating temperature) the charge voltages were
higher and the discharge voltages lower than at steady state.
Initially,
under dc conditions battery voltage during Cycle A was about 11.8 V on dis-
charge and 14.2 V on charge; during Cycle B the battery voltage was 8.6 V on
discharge and 18.3 V on charge.
Similarly under EMT conditions, battery volt-
age for Cycle A charging decreased from 17.5 to 13.8 V during the first two
sequences while the discharge voltage changed from 11.2 to about 12 V. Cycle 3
discharge resulted in a battery voltage of 8.9 V in the first sequence and
9.2 V for the next two sequences.
The average electrolyte temperature at steady state was higher during
Cycle B than during the "A" Cycles.
for both dc and EMT test conditions.
This temperature change was 3-4 °c
The temperature of the electrolyte
during testing was 60-64 and 43-47 °c under dc and EMT conditions, re-
spectively. The average differences between the electrolyte and skin temp-
eratures were 8 and 5 °c for dc and EMT conditions, respectively. The
tests were conducted under natural convection with the batteries set on
a wooden board.
The average specific gravity after the charge portion of each cycle
for dc tests was approximately 1.250. By comparison with the full-charge
specific gravity of 1.280 the batteries were estimated to be operating
at about 75% state-of-charge.
Even though the specific gravity of the
battery was not checked under EMT cycle test ~onditions, past experience
with SLI batteries operating under EMT conditions indicated 90-95% state-
of-charge.
-28-
-------
CYC LE A - CHARGE CYCLE B - CHARGE
300 300
40 30
« « <
o < 0
.. 0
I- .. r-
Z r- I- -i
UJ 20 -i Z 20 :t>-
~. --.J--,~~ » UJ (;)
G) 0:::
0::: m
m G -300 ..
u -300 ..
~ < <
10 ~ 10
-600 -600
TIME, 10-3 SEC/DIV TIME, 10-3 SEC/DIV
I
N
\.0 CYCLE A - DISCHARGE CYC LE B - DISCHARGE
I
0 J ~I 0 40
« 20 < « <
.. 0 .. 0
I- ~ r I- r-
Z -i Z -i
UJ -300 » UJ - 1 200 20 »
0::: J j G) 0::: G)
0::: m 0::: m
:) t- 10" :) ..
U ~ < U <
-600 -2400 - 0
0
TIME, 10-3 SEC/DIV TIME, 10-3 SEC/DIV
Figure 3.
EMT Transient Waveforms during Test of 61-A hrt Group Size 22F Battery
-------
----.----- - ---
The gas evolution during both the high and low rate charges at steady
state :'/as found to be surprisingly low. In both cases the coulombic effi-
cien~y was greater than 99%.
3
The rates of gas evolution under dc conditions were 0.60 and 17 em
per cycle during Cycles A and B, respectively. Under EMT tests the rates
3
were 3.7 and 50 cm per cycle during Cycles A and B, respectively. This
leads to a coulombic efficiency of greater than 99% for Cycles A and B
for both dc andEMT test conditions.
The gas evolution during start-up
is shown in Figure 4.
Coulombic efficiency during start-up was deter-
mined to be 91 and 78% for dc and EMT operations, respectively.
Overall energy efficiency at steady state for a single test sequence
(30 Cycle A's and one Cycle B) was 79% for both dc and EMT.
Average
energy efficiencies for Cycles A and B were 90 and 61%, respectively, for
dc conditions and 78 and 58%, respectively, for EMT conditions.
Power inputs for Cycles A and B under dc conditions were 0.34 and
2.8 kW, respectively, and the outputs, 0.63 and 3.4 kW.
The specific
power for Cycles A and B averaged 17 and 92 W/lb (38 and 206 W/kg), respec-
tively. Similarly for EMT conditions, power inputs for Cycle A and B
. .
were 0.37 and 0.57 kW, respectively, and outputs were 0.65 and 3.2 kW,
respectively. The specific power under EMT conditions becomes 18 and 88
W/lb (39 and 193 W/kg) for Cycles A and B, respectively.
The energy losses
due to resistance that occurred during the dc cycling regime were deter-
mined for the Group Size 22F battery as shown in Table 7. The internal
resistance of the battery at the operating temperature of 63°C was 8.6 mQ.
The total resistive energy losses are 0.10 arid 12 W hr for Cycles A and
B, respectively.
These losses represent 5 and 32% of the total
energy inputs of Cycles A and B, respectively.
These values are large compared to coulombic inefficiency losses .of
0.55 and 0.97% for Cycles A and B, respectively.
In order to subject the batteries to more realistic operating condi-
tions, the EMT was a1so operated. under the following conditions:
Acceleration:
20 to 30 mph in 10 sec (average discharge current
45-95 A)
-30-
-------
6
EMT
~ 0 @
c::ra---
--I 4 ~
...
£:)
w /
>
--I
0
>
w
I VI
W <
I-' 0
i
--I
< DC
~ 2 .
o . . .
~
o
o
30
60
90
C YC LES
Figure 4.
Gas Evolution during Test of 61-A hr, Group Size 22F Battery
-------
Table 7.
Losses
Voltage, V
Power, W
Energy, W hr
I
w
N
I
Losses Due to Resistance during DC Testing of a 6l-A hr Group Size 22F Battery
Cycle A Cycle B
Discharge (10.8 see) Charge (22 see) Discharge (25 see) Charge (50 see)
0.44 0.22 3.21 1. 57
22 5.8 1197 285
0.067 0.036 8.3 3.9
-------
Vehicle Weight:
Test Sequence:
30 to 20 mph in 20 sec (average charge current
20-65 A)
75 sec (average charge current 45 A)
30 ft lb at 30 mph
4,000 lb
accelerate-decelerate 15 times followed by idling
(repeated 7 times, total time - 41 min).
Deceleration:
Id.ling:
Drag:
At steady-state conditions the charge voltage was 13.6 to 13.3 V and
discharge voltage was about 11 V. The cou1ombic: efficiency was 95% (based
on average disctarge current of 55 A). Once again, the largest change in
voltage and gas evolution was observed at the beginning of the test.
2.2.4
Discussion
2.2.4.1
Identification of Technical Problems
We have searched the literature to uncover information on the state-
of-the-art of the lead.-acid batteries germane to this application.
Ho
information was discovered that would permit us to predict with any degree
of certainty the behavior of existing lead-acid batteries under the cyclic
conditions specified.
However, by analyzing the physical characteristics
and behavior of existing batteries we were able to identify in general
terms those parameters that would be critical.
Possible problem areas which
offer room for improvement are discussed in the following paragraphs.
Internal Resistance
Specific power for a particular battery can be improved by decreasing
the internal resi.stance.
Our component-by-component analysis of the
internal resistance has identified the grids as the major contributory
source of the overall resistance of the battery.
This resistance can be
reduced by a change in grid design and plate size and by use of a more
conductive alloy.
A second significant contributor to the internal
resistance is the electrolyte which soaks into the active materials of
the plates.
This resistance is related to the porosity of the plates and
a tortuosity coefficient.
The more porous the plate the greater the
conductivity: unfortunately we are unable at this time to relate the
tortuosity coefficient to the porosity.
-33-
-------
Our analysis also revealed that the cell interplate connectors, ter-
minal straps, and the resistance of the electrolyte between the plates
constitutes only about 17% of the battery's overall resistance.
Separators
The two most important characteristics of the separator are its re-
sistivity and its physical integrity under the anticipated operating condi-
tions. The resistivity can be lowered by increasing the pore size, but
larger pores increase the chance of shorting through the separator. The
resin-impregnated paper separator with embossed ribs has a resistivity of
2 .
0.23 Q cm and tends to deteriorate when operated at temperatures greater
than 43°C.
Further, with an electrode spacing of 0.135 cm, it contributes
about 58% of the overall resistance between plates.
If, because of the high
power drains, the anticipated opeFating temperature of the battery is greater
than 43°C, a separator must be selected capable of maintaining its physical
integrity at that temperature, and further, in order to minimize the internal
resistance of the battery, its resistivity should be as low as possible.
Improved separators are available.
Weight of Electrolyte
Since present SLI batteries are desi8ned for high reserve capacity,
they contain approximately 14.5% of their electrolyte outside the element.
Since this electrolyte is of no benefit in high-rate discharge,much
of it could be removed thereby reducing the battery weight.
Charge Acceptance
Our exploration of the literature indicated that we would experience
difficulty in charging a lead-acid battery efficiently at high rates (on
the order of 3C).
However, our experiments with 61 A hr Group Size 22F
batteries revealed a surprisingly high coulombic efficiency, greater
than 99% at the equilibrium temperature of 60°C under the cyclic
conditions employed.
At lower operating temperatures we would expect
the efficiency to. be lower.
-34-
-------
Energy Eff~ciency
Data from the experiments on the batteries mentioned above
sh(>'J!E:d an overall energy efficiency of 80% as an average for one SS-kh'
dee~ cycle and 30 lO-kW shallow cycles, with 63% for the SS~kW deep dis-
charge, and 90% for the lO-kW shallow discharge.
The major energy 10ss2s
occurred in resistive heating. Using the cycle frequency routine of the
EPA specifications, we estimate the average energy eff~ciency would be 86%.
Operating Temperature
The batteries were tested under a more severe routine than that
specified by EPA, since all SS-kW discharges \vere of 2S"'sec duration,.
whereas for the EPA routine 'the average duration of the 5S-kW discharges
is only eight seconds.
Under the test conditions used, the difference
between the battery case temperature and ambient was about 36°C at
steady-state.
We estimate that this difference would be reduced to
approximately 14 °c under the EPA routine.
Further, under the anticipated
operating conditions of the hybrid vehicle, convective cooling would un-
doubtedly prevail and the temperature of the batteries would be kept at
a temperature differential considerably below 14°C, and probably below that
of SLI batteries as they are currently used in automobiles where they are
exposed to the engine's heat.
Cycle Life
It is known that in currently produced lead-acid batteries cycle
life is related to the depth of discharge; the shallower the discharge:
the longer the life. However, the literature search revealed no information
on the relationship of depth of discharge to cycle life at depths less than
40%. Assuming, for example, a 6l-A hr battery would be used in this applica-
tion and the maximum capacity withdrawn on a single-pulse discharge would be
2.4 A hr, the depth of discharge would be approximately 4%. Any extrapolation
from the relationship given in the literature for depths of discharge
ranging between 40 and 100% and cycle life, to the levels anticipated
for this application, is of questionable validity.
Cycle life is also affected by the corrosion rate of the positive
grids which is highly sensitive to temperature.
As noted above, we expect
-35-
-------
the operating temperature of the battery pack in the hybrid vehicle to
be at a level lower than that of present SLI batteries in automobiles
so the corrosion rate would be lower.
Because of the expected lower
corrosion rate and the shallow discharge cycles, we feel batteries of
the SLI type can meet the five-year life requirements.
However, to insure
meeting this requirement, a search should be carried out for alloys more
resistant to corrosion than the 4.5% antimonial lead now used.
2.2.4.2
Conclusions
Our examination of the state-of-the-art of lead-acid batteries (which
included the actual testing of super premium grade SLI batteries) considered
two aspects of developing a battery system for hybrid-vehicle applications.
One, to satisfy the original EPA specifications for a 100-W/lb battery,
and two, to develop a 200-W/lb battery, the latter being a more realistic
requirement based on TRW's experience with hybrid-vehicle systems. Problem
areas which offered room for improvement were discussed in Section 2.2.4.1.
The SLI battery studied embodies, in our opinion, the application of the
highest level of technical knowledge to mass produced units. However, our
detailed analysis of this battery showed that it could be improved and
that the problems in designing a battery for the hybrid vehicle application
are soluble.
With respect to the original EPA requirements, using 17 of these bat-
teries in series yields a weight of 610 Ib (277 kg), with a power output
of 57 kW, at 1.53 V per cell and 365 A at the high-pulse load. With 16
batteries, the weight would be about 575 Ib (261 kg), and the power output
would be 53.5 kW.
The specific power yield with these batteries is about
93.5 W/lb (206 W/kg); what is required to meet the weight requirement is
100 W/lb (220 W/kg) , a difference of 6.5 W/lb (14.3 W/kg). A second
sive high power pulse of 25-sec duration can be delivered readily.
succes-
An increase in specific power can obviously be effected by a decrease
in weight anu by a reductiori in internal resistance.
By reducing the weight
of each of the Croup Size 22F batteries by 1.5 lb (0.68 kg) and by using
16 of these batteries in series, the weight would be brought to an accept-
able level.
A reduction of this amount could be obtained simply by re-
moving sorae of the electrolyte outside thE element.
Howe'ler, a more funda-
-36-
-------
mental approach lies in attempting to design a battery, by a careful con-
sideration of all factors involved, in order to maximize the specific power.
As was mentioned above, the grids are the major contributor to the internal
resistance of the battery; however, merely increasing the conductivity of
the grids by the addition of more cross-members, obviously does not solve
the problem in optimum fashion because of the resultant increase in weight.
Plate size, active material density, and separator should be optimized also.
In other words, th~ optimization of the battery for specific pO\ver involves
the use of a IIsystemsl1 approach, preceded, naturally, by a thorough evaluatioD.
of all the individual parameters that affect specific power and their inter-
relationships.
2.3
ADVANCED (ALTEfu~ATE) CONCEPT BATTERIES
In Section 2.2 attention was focused on the pasted-plate SLI (starting,
lighting and ignition, or automotive) battery.
A bipolar battery with Plante-
formed plates and a tubular-plate battery were considered as possible alter-
natives to the pasted-plate SLI battery.
2.3.1
Bipolar Battery
A Plante plate is manufactured electrolytically by anodic oxidation
of a pure lead substrate surface to active lead dioxide.
ordinarily is deeply grooved to enhance its surface area.
The lead substrate
The greatest advan-
tage of Plante plate batteries is their very long 'life of up to 40 years, due
to the corrosion resistance of the pure lead substrate and the exceptional
adherence of the active material to the substrate.
However, because of
their great weight they are presently used only in stationary applications.
In 1924, Kapitza [P. !Zapitza, "A Method of Producing Strong Magnetic
Fields,!: Proc. Roy. Soc., Ser. A, 1-°5,691-710 (1924)] described a bipolar
battery from which a very high-power output was obtained for 10-20 msec.
A bipolar battery utilizes positive and negative electrode~ of adjacent
cells as an integral electrically conducting sheet, which serves also as
the partition between the cells.
This arrangement provides the shortest
possible current path between cells. Kapitza stacked thin lead sheets
between U-shaped rubber gaskets and then formed a very thin layer of active
material on either side of each sheet by the Plante process. He obtained current
2 . 3 3
densities of 2 to 5 A/cm at a power density of 500 kW/ft (17.7 W/cm ),
withdrawing a capacity of about 0.047 A sec/cm2 (0.3 A sec/in.2) in 20 msec.
-37-
-------
In 1968, Biddick and nelson [R.E. Biddick and R.D. Nelson, "Lead-
Acid Bipolar Battery for Multisecond Pulse Discharge," IECEC '68 Record
(Proceedings of the Intersociety Energy Conversion Engineering Conference,
Boulder, Colorado, Paper 689006) 47-51 (1968)] developed a higher capacity
Plante bipolar electrode which sustained currents of 1 to 3 A/in.2 (0.16
2
to 0.48 A/cm ) for several seconds. They also showed that the bipolar
battery could be quickly and efficiently charged. A high state-of-
charge \olaS maintained during prolonged cycling with a discharge pulse of
five seconds at 1 A/in.2 (0.16 A/cm2) and a charge period of 20 sec. The
battery employed a unique gasket design which allowed water to be added
to all cells at once through a filling trough.
Under Navy sponsorship, a 300-V, 3.6-kW battery was constructed. It was made
up of five 60-V modules ("Design and Fabrication of 300 Volt, 3.6 Kilowatt
Pile-Type Bipolar Lead-Acid Battery for Pulse Duty,iI Final Report,
29 December 1967 to 29 September 1968, Contract No. N00123-68-C-0862, U.S.
Navy Undersea Warfare Center). One of these modules was cycled 140,000
times at a discharge current density of 1 A/in.2 (0.16 A/cm~) for one second
and a charge period of 15 seconds.
In addition, a three-cell bipolar
battery was cycled 200,000 times at a five-second discharge current density
of 1 A/in.2 (0.16 A/cm2) and 20-second charge. A single cell was cycled
for 425,000 times on the same duty cycle.
These results established the
feasibility of the bipolar battery concept for storing electrical energy
to meet high power requirements. Power outputs exceeding 200 W/1b
(440 W/kg) and 50 kW/ft3 (1.77 W/cm3) were obtained at an overall ~ower
efficiency of 80% using 0.032-in. (0.08l-cm) thick lead sheet electrodes.
It appears likely that specific power output can be increased significantly
by use of a thinner and lighter electrode. . However, the long-term life
of these batteries has not been established.
2.3.2
Tubular Plate Battery
Tubular plates have three primary advantages over the pasted variety:
1)
Tubular construction is very effective in preventing loss of
the active material during cycling,
-38-
-------
2)
the weight of the grid is only about 50% of the weight of a
pasted-plate grid giving equal life, and
3)
a higher energy density is attained in comparison with pasted
plates. (This was no longer true at the end of this program
because of advances in pasted-plate technology.)
A conventional type of tubular plate utilizes either cylindrical or
prismatic tubes to contain the active material.
The positive active
material is retained on the grid by means of a fabric sheath which is
most often a woven Dynel or polyester material, but in some cases, may
consist of braided fiberglass with a perforated polyvinyl chloride outer
shield. The grid for this type of structure consists of a single "spine"
of lead alloy along the axis of each tubular sheath. These suines are
connected to a rather massive header at the top of the plate which is both
a bus bar and a strengthening frame.
cell.
The tubular-plate battery is exemplified by Gould's 85T Motive Power
An improved tubular retainer recently developed has increased
cycle life to about 3000 cycles at 40% depth of discharge and over 1000
cycles at 80% depth of discharge.
This improvement is due to the use
of a polyester material, specially tailored for resistance to sulfuric
acid, resulting in excellent retention of active material ~/ithin the
cylindrical fabric sheath.
Hith less resistant retainer materials,
corrosion of the polymeric fiber results in reduction of thread diameter
and increased porosity of the sheath so that small particles of lead
dioxide permeate the sheath and collect in the sediment well.
This
results in loss of capacity and eventual shorting of the cell when active
material accumulates to the top of the sediment well and contacts adjacent
positive and negative plates.
The specific energy of this type of battery
at the six-hour rate is 11 W hr/lb (24 W hr/kg).
The tubular cell has been developed for ruggedness, dependability,
and long life in motive power service (e.g., in fork-lift trucks where
high specific power is not a requirement). At the present time, however,
the smallest diameter tube is 0.275 in. (0.7 cm) as compared to an overall
-39-
-------
plate thickness of 0.075 in. (0.19 ern) or less in pasted-plate SLI
batteries. In order to achieve s~ecific powers comparable to what is
possible for pasted plates it would be necessary to radically modify
the present design by reducing the diameter of the tube in order to
decrease the mean distance bet~yeen the active material and the spine.
Additionally, since the spine diameter must also be reduced, mechanical
problems associated with the introduction of active material into the
tube can be anticipated.
Consequently, we did not feel that the tubular construction should
be actively explored at this time. The pasted-plate design and the
bipolar electrode concept appear to be the most promising for the hybrid-
vehicle application.
2.4
APPROACH TO SOLUTION
As a result of examining the lead-acid battery problem, a program
plan was developed which comprised
two approaches to developing batteries
that can satisfy the hybrid-vehicle requirements.
One approach aimed at
adapting conventional pasted-plate technology to a 55-kW battery (original
EPA specifications) with further optimization leading to a 200-W/lb battery
system.
The second approach involved the development of a bipolar battery.
By adjusting present pasted-plate design and by refining present
production techniques it appeared probable that the performance and cost
objectives of the original EPA specifications (discharge and charge cycles
summarized in Table 8) could be met by 1972. This design, was in all
probability, not the optimum design for a pasted-plate battery. Higher
specific power was realizable. TRW had tested pasted-plate batteries with
specific power exceeding 200 W/lb but lifetime, cycle life and other charac-
teristics had not been examined. It did appear probable that significantly
higher specific power could be achieved by modifying cell components. In
order to obtain data which would permit a closer approach to an optimum de-
sign we proposed to test various design concepts, component modification,
and parameters in statistically designed experiments on single electrodes
and cells as discussed in Section 3. Those experiments were designed to
indicate plate thickness and paste density, and elucidate to some degree
-40-
-------
Table 8.
Discharge and Charge Cycles for a 55-kW Battery
DischarRe Charge
Nwnber Rate, Energy, . Time/Cycle, 'rota1 Time, Rate, En~rgy, Time/Cycle, Total Time,
of Cycles kW W hr see sec kW W hr* sec sec
500 55 380 25 12,500 30 475 57 27,500
3,000 55 130 8.5 25,500 30 163 20 60,000
3,000 55 80 5.2 15,600 30 100 12 36,000
193,500 10 30 10.8 2,090,000 5 375 27 5,200,000
I 2,143,600 5,323,500
~
......
I = 595 hr = 1480 hr
* Assuming 80% energy efficiency.
-------
a preferred grid design and material.
Other factors to be studied included
separator thickness and ribbing to provide good gas release from the cell
on overcharge and the effect of electrolyte concentration on performance~
An important factor in determining battery life is the composition
of the lead alloy used to fabricate the grid.
Both extensive and intensive
tests were planned in order to find improved alloys and evaluate their life
by means of accelerated tests.
Computer programs were used to predict the current and potential dis-
tributions in various plate configurations, with the objective of optimizing
the weight, uniformity of potential, and ease of fabrication of the battery.
The bipolar battery offered the following advantages for use in a
hybrid heat engine/electric family automobile:
. Minimum electrical resistance
. Absence of inter cell connectors
. Maximum current capability
. Minimum volume
.
Minimum Weight.
Since the feasibility of the lead-acid bipolar battery had been demonstrated
. .
in rather long-term tests lasting from several months to over a year at
average power levels in the range required for the hybrid car, its d~velop-
ment was certainly worth pursuing.
The specific tasks required to develop the bipolar battery were:
.
Develop a lightweight substrate material onto which thin layers
of active material can be formed.
.
Develop a method of forming active material on the substrate.
.
Develop a method of sealing a plastic frame around the substrate,
e.g., ultrasonic welding.
.
Develop a method of sealing the plastic frames together to
produce the battery.
Development of the bipolar battery is discussed in Section 4.
-42-
-------
3.
OPTIMIZATION OF CONVENTIONAL (PRISMATIC) BATTERIES
3.1
GRID STRUC'r'JRE
3.1.1
Mathematical Modeling
3.1.1.1
The Model
A mathematic~l model was constructed for the optimization (maximizatiop)
of a figure of merit for the pasted~grid configuration of the positive
plate.
The figure of merit for the plate was defined as
i[E - i(R + R + R ))/W
g p e
where i is the plate current, E is the half-cell potential of the positive
plate (vs. Hg/HgZSu4)' Rg is the grid resistance, Rp is the paste resis-
tance, R is the electrolyte resistance, and W is the weight of the plate.
e
Thus, the figure of merit is seen to be the ratio of delivered power to
fm
weight.
This figure of merit was optimized with the values of i and R
e
remaining fixed, while R , R , and W varied with grid geometry.
. g p
of course, a physical constant.
E is,
A fairly detailed model of the current flow in the plate was constructed
and implemented.
Each paste element (pellet) was regarded as independent of
all of the others.
The current density flowing from electrolyte to paste
was assumed constant, and Poisson's equation was set up and solved
analytically for each rectangular pellet.
This equation describes the
steady-state distribution of current and voltage in an extended resistive
medium.
The result of the analysis of Poisson's equation was a set of
expressions for:
a)
b)
The equivalent resistance of a pellet,
the average voltage (iR) drop in a p~llet,
c)
the fraction of the pellet current which is delivered to each
of its four bounding grid elements, and
d)
the resulting voltage drop in each of the bounding grid elements.
The development of these expressions is sho~~ in Appendix A.
-43-
-------
Kirchh::>ff's laws were written for the distribution of voltage through-
out the grid and of the currents delivered to the grid by the pellets.
These laws express mathematically the conditions that the sum of the
(positive and negative) currents 2ntering any intersection of grid elements
must be zero, and that the sum of the voltage drops around any closed path
in the grid must also be zero.
Calculation showed that collateral current
flow in the paste elements and also in the electrolyte was small.
A computer program was developed, embodying the Kirchhoff and related
equations.
In the present model, the currents flowed to a single tab
of arbitrary size and location on one boundary of the grid.
The voltage
distribution relative to this tab was computed,
and its average value
set equal to iR in the formula for figure of merit. Although the program
g
contains the capability of specifying the current flow into each pellet
individually, no use was made of this feature, due to the restricted
scope of the present study.
The heart of the computer program is a set of Sparse Matrix
Inversion subroutines developed at TRW.
These are used to solve, in a
highly efficient manner, the hundreds of simultaneous linear algebraic
equations that arise from the use of KirchhofCs laws.
machine used was the CDC 6500.
The computing
Although all of the most important physical effects have been
included in the model, it was unfortunately necessary to omit from
consideration a good many others.
These include:
variation of physical
parameters with state of charge, temperature, current and age; any
consideration of the negative plate or of the cell as a whole; variation
of electrolyte composition (specific gravity) and volume; variation of
separator parameters; variation of the current from pellet to pellet
caused by voltage drops across the grid; reaction rates (mass action)
and their dependence and effect on electrolyte concentrations; and effects
of collateral current paths in the pellets and electrolyte.
It would
be desirable to include as many of these effects as possible in future
modeling studies.
-44-
-------
3.1.1.2
Results
A series of computer runs was made to optimize a grid design based
on a square plate 4 1/4 in. (10.8 cm) on a side and 0.040 in. (0.102 cm)
in thickness, having 24 vertical and 8 horizontal elements, ~ pair of
tapered vertical elements (the 9th and 16th counting from one side) and
a current-collector tab extending from the 8th to the 10th element at the
top of the plate.
The plate chosen was smaller than a conventional plate
such as a Group Size 22F to minimize iR drop at high current den3ities.
The pellet size \vas the same as in conventional SLI battery grids.
The ':lidth of the side and bottom frame members and the small end of the
stiffeners was O.040-in. (0.102-cm), corresponding to a 0.0016-in.2
(0.01-cm2) cross-sectional area., The cross-sectional areas of the
vertical and horizontal elements were 0.0004 and 0.0008 in.2 (0.0026
2
and 0.0052 cm ), r2spectively.
Parameters which were allowed to vary in
the optimization were the taper of the 9th and 16th vertical elements
(stiffeners)
and the design of the upper horizontal portiori of the frame.
Corrosion of the elements was considered in the calculation.
The half-cell potential, E, was taken as 1.68 V and a plate current
of 60 A was used.
This is the appro~imate current that would be required
for a Group Size 22F plate to meet the EPA specifications.
Computer ru~s were made on the uncorroded plate first.
The figure-
of-merit values for all of the tapered designs tried were found to lie
within 1% of each other.
This independence of the figure of merit on
design results from the fact that at 60 A the iR term in the figure of
merit is small compared to E. In the present case, a 10% decrease in
grid resistance, R , produces only a 0.4% increase in the term
g
E - i(R + R + R ) in the figure of merit. This increase is usually
g p e
more than offset by the attendant weight penalty. This situation is
characteristic of small plates where the figure of merit is governed
largely by plate weight. Therefore, our figure of merit could be
improved by adopting a plate with a larger (perhaps fiber reinforced)
I
pellet size which would contain less lead and weigh less.
-45--
-------
In the case of corroded plates, however, the situation is somewhat
different. Corroded plates were simulated by decreasing all grid element
cross-sectional areas, so that all of the vertical grid elements, except
the frame and stiffeners, essentially disappear. Consequently, a
comparison series of calculations was made on a plate with 4 vertical
and 8 horizontal elements.
The taper of the two inner vertical elements
and dimensions of the uppet frame were varied.
Here, because of the much-
increased value of the grid resistance, it was found possible to maximize
the figure of merit. The maximum was "broad!:, however ~ and several grid
designs that were tested had figure-of-merit values within 1% of the optimum.
The optimum design was found to be a 0.13-in.(0.33-cm) width for the
top of the tapered stiffeners and the top frame element between the stiffeners,
and a tapered width for the top element outside of the stiffeners. The out-
line of a grid actually cast is shown in Figure 5. The pellet size in the
grid is the same as in standard SLI grids but the vertical and horizontal
dimensions are interchanged.
3.1. 2
Accelerated Corrosion Tests of Selected Alloys
3.1.2.1
Approach and Selection of Alloys
Six lead test alloys were prepared and cast into grids and rods.
Bare grids and rods and positive plates incorporating test-alloy grids
were exposed to 1.260 sp gr sulfuric acid electrolyte. Corrosion was
accelerated by maintaining test cells at 71°C (160 OF) and impressing upon
the test specimens a potential of 0.075 V
tential of the alloy in the electrolyte.
relative to the equilibrium po-
Dimensions of the test specimens
and the capacity of the positive plates were measured periodically.
Rod
specimens were examined metallographically to determine corrosion patterns.
Compositions and selection criteria for the lead test alloys were
as follows:
1.
3.08% Cd-2.8% Sb-O.15% Ag -
High mechanical strength;
feasible for using as
extremely thin grids.
2.
0.013% Ll
High electrical conductiv-
ity; superior electro-
chemical behavior.
-46-
-------
Figure S.
-
2 em
Outline of Grid for Pasted-Plate Test Cells.
-47-
-------
3.
4.5% Sb
Standard alloy for battery
grids.
4.
0.017% Li-0.90i, Sn
Improved castability and
mechanical strength over
0.013% Li.
5.
2.5% Sb-0.45% As
-- Contains less antimony than
4.5% Sb (standard alloy), .
but retains equivalent mechanical
strength through use of arsenic.
6.
0.066% Ca-0.90% Sn
-- Good corrosion resistance and
improved castability.
The 4.5% Sb alloy (standard alloy) served as the comparison reference
standard.
3.1.2.2 Experimental Procedure
Lead alloys of the compositions described above were cast into grids
and rods. Grids were Group Size 22F size, 12.4 em x 14.7 cmxO.14 cm (4.9
in. x 5.8 in. x 0.055 in.) with the vertical and horizontal bars on 1. 55-em
(0.6l-in.) and 0.46-cm (0.18-in.) centers, respectively.
The positive plates were prepared by using conventional techniques
and in the same manner as plates fabricated for component testing.
These
techniques are described in Section 3.2.2.
The formation of plates was
carried out with three-plate cells; i.e., one positive and two negative
plates per cell with the cells connected in series.
dry charged.
Plates were not
Twelve corrosion test cells were set up, six of which contained two
bare grids and two rods of one alloy in each cell. The other six cells each
. .
contained two posit.ive plates prepared from one alloy. Each grid or plate
was assembled in a three-plate electrochemical group (two standard negative
plates with each positive plate and two bare standard Sb-alloy grids with
each bare grid). Each test cell contained two grid (or plate) groups. The
rods were placed between the two grid groups in each grid cell. A test cell
containing two bare-grid groups and two rods- is shown in Figure 6.
Tests were conducted at 71 °c (160 OF) in a manner similar to that
used by Willihnganz [E. Willihnganz, "Accelerated Life Testing of Stationary
-48-
-------
I
.po
\D
I
l~~-
- WATER COOLED CONDENSER
HEA TER
THERMOREGULATOR
ROD SPECIMEN
ST IRRIN G MOTOR
SIDE VIEW
FRONT VIEW
Figure 6. Accelerated Corrosion Test Assembly for Bare Grids and Rods
-------
. .
Batteries," Electrochem. Tech., .Q.~ No. 9-10, 338-341 (1968)].
On the expec-
tation that positive grids might warp during th~ test and could come in
contact with the negatives, negative grids were enclosed with perforated
Koroseal envelopes.
Each positive plate was separated from its two negative plates by
standard PVC separators (Ethyl Corporation) and each plate group was clamped
together as a unit to avoid excess warpage.
The cell containers used were oversize so that adequate stirring of
the electrolyte could be accomplished by ~eans or a magnetic stirring
assembly.
All test components in an individual cell were connected in parallel
and the test cells were in turn connected in parallel, each with its own
current trinuner. Each cell was wired throllgh a central panel so that cell
voltage, positive electrode potential ~. a saturated Hg/Hg2S04 reference
electrode containing 1.260 sp grHzS04) and current cpuld be monitored
through selection switching without the need for manual probes and without
disturbing the wiring. Positive plate potential was maintained at 0.075 V
relative to equilibrium plate potential at 71 °c in the 1.260 sp grelec-
trolyte.
Potential of the bare grids was monitored with respect to the
reference electrode and was maintained 0.075 V above the open-circuit
value.
Cells were constructed so that they could be easily dismantled for
growth and weight measurements on the test specimens.
Growth measurements
represent the mathematical averages of horizontal and vertical measure-
ments made at three locations each (in center and ~l em from each edge).
Rods were fitted into a Teflon (TFE) fixture which in turn was screwed
into the cell cover.
The length of the rod was measured from the end of
the rod to the fixture.
Diameter measurements were made on samples cut
from the end of the rods for metallographic analysis.
Plate capacities were measured, normally at two-week intervals.
The
cells were allowed to cool to room temperature before the capacity and
dimension measurements. Discharge rate was 60 A (0.16 A/cm2) to a O.95-V
-50-
-------
(cell voltage) cutoff. Recharge rate (at room temperature) was 3 A (0.008
Alcm2) until 110% of required charge was returned to each cell.
3.1.2.3
Results and Conclusions
Pasted-~late and Bare-Grid Specimens
The accelerated corrosion test of pasted plates was terminated whpo th0
last surviving grid (Sb-As alloy) cracked during handling preparatory to
measuring growth (138 days at 71°C). The last pasted plate (Li alloy)
failed after being on test for 128 days at 71°C. Tables 9 and 10 summarize
resistance and capacity data on the pasted-plate specimens.
alloys had the best overall performance.
Sb-As and Li
Cannone ~nd co-workers [A.G. Cannone, D.O. Feder and R.V. Biagetti,
"Lead-Acid Battery;
Positive Grid Design Principles,;1 Bell System Tech.
J., ~, No.7, 1279-1303 (1970)], have shown that plate growth can be
described by one of two equations,
G = k t
1
2
G = k t
2
where G is the percent growth, t is the time and k is the rate const3nt.
Growth data for pasted plates are summarized in Table 11. Grotvth
was linear for the Li, Sb and Li-Sn alloy plates and quadratic for the
Sb-As alloy plate. Rate constants, obtained graphically, are given in
Table 12. Growth curves for plates of the longest-lived alloys, Li and
Sb-As, are shown in Figure 7.
is shown in Figure 8.
A quadratic plot for the Sb-As alloy plate
Vertical growth of plates is greater than horizontal growth on a per-
centage basis. Grid design (longer horizontal pellet section) and gravi-
tational effects may account for this difference.
For the purpose of these measurements, growth was defined as enlarge-
ment of the grid (or plate) without cracking.
Once a crack occurred in
the grid frame, the frame was subject to further cracking during handling;
a crack was not an isolated local occurr~nce. None of the bare grids
cracked at the frame bar.
For example, the interior members of the Cd-
-51-
-------
Table 9. Resistances of Pasted-Plate Cells during Accelerated Corrosion Test at 71 °Ca
Resistance, m.12
Alloy 7 Jul 21 Julb 4 Aug 1 Sep 14 Sep 24 Sep 22 Oct 27 Oct 11 Nov
Cd-Sb-Ag 4.8 6.0
Li 4.3 4.4 5.0 5.3 5.6 7.6 8.8 6.Sc
Sb 5.7 6.4 8.0 8.7 9.2 8.6c 8.9c
Li-Sn 4.3 4.4 5.0 5.6 6.1
I Sb-As 4.5 5.0 5.5 6.1 6.5 8.4 8.9
VI
N'
I Ca-Sn 4.6 4.8 5.7 6.9 7.9
aN' plates made with standard antimonial grids.
egatl.ve
b made.
No measurements
c second plate broken.
Result from one plate;
-------
Table 10. Capacities of Pasted-Plate Cells during Accelerated Corrosion Test at 71 °Cc
Capacity, A hra
Alloy 7 Ju1 21 Ju1 4 Aug 1 Sep. 14 Sep 24 Sep 15 Oct 27 Oct 11 Nov
Cd-Sb-Ag 3.87 3.94 3.78
Li 4.34 4.45 3.37 3.53 4.08 3.88 3.01 2.16 1.17b
Sb 4.29 4.44 4.12 2.56 2.10 2.37b 2.08b 1. 73b
Li-Sn 3.63 3.79 3.66 3.21 3.33 3.06 1. 62b ---
Sb-As 4.04 4.05 4~15 4.42 4.16 3.88 2.22 1.83
I
VI
\..oJ Ca-Sn 4.30 4.32 4.08 2.25 2.38 1.84
I.
a 2
At 60-A (0.16-A/cm ) rate to 0.95 V.
.b second plate broken.
Based on one plate only,
c Corrosion test carried out at 71 0C. but capacity measurements made at room temperature.
-------
Table 11. Growth of Pasted-Plate Specimens during
Accelerated Corrosion Test at 71 QC
Growtht %
Alloy Time, days Vertical Horizontal
Li 39 1.3 0.9
63 1.6 1.3
84 2.0 1.9
114 3.1 2.6
128 3.6 3.5
Sb 39 2.0 1.2
63 2.7 1.9
84 3.6 2.8
114 6.0 3.9
Li-Sn 39 0.9 0.5
63 1.5 0.7
84 1.8 1.3
Sb-As 39 1.2 0.6
63 3.3 2.1
84 5.5 4.1
114 10.4 6.5
-54-
-------
Table 12. Growth-Rate Constants for Pasted-Plate
Specimens at 71 °Ca
Rate Constant x 102 %/day
.
Alloy Vertical Horizontal
Li 2.8 2.4
Sb 5.0 3.3
Li-Sn . 2.3 1.4
4 %/day2
Rate Constant x 10 ,
Sb-As 8.1 5.3
a
Estimated graphically from Figures 7 and 8.
-55-.
-------
6
;:R.
0
..
:I:
~
:;
0
~
0
4
10
. Sb-As, VERTICAL
o Sb-As, HORIZONTAL
. li, VERTICAL
6 li , HORIZONTAL
8
2
00
50
100
TIME, DAYS
Figure 7.
Growth-Rate Plot for Li and Sb-As Alloys
-56-
-------
6
~
0
..
:r:
~
3
0
IX'
<-'
4
8
10
. VERTICAL
. 0 HORIZONTAL
2
o
o
5000
TlME2, DAYS2
10,000
Figure 8.
Second-Order Growth-Rate Plots for Sb-As Alloy
-57-
-------
Sb-Ag alloy grid warped and pieces fell out, but the integrity of the
frame was retained.
Growth measurements on the Cd-Sb~g alloy grids were
discontinued because of the warpage, difficulty of making measurements and
failure of the corresponding plates.
Capacity, resistance and growth measurements were made at room tempera-
ture.
The cooling, measuring and reheating period took 24 hours.
This
testing period is not included in the time scale of the growth curves.
Since
the Cd-Sb-Ag and Ca-Sn alloy plates were the first to fail, data for them
were too limited for analysis.
Direct growth measurements on bare-grid specimens were difficult to
make because waves or ripples had formed in the plane of the grid during
expansion.
Growth estimates were therefore made indirectly on projected
grid images.
These values, although erratic, seemed much less than those
for pasted plates.
Expansion of the pasted plates led to uniform bowing
or dishing.
Only the Sb-As alloy plate developed ripples which led to
cracked and loosened pellets.
Estimates of growth rate constants and life times (times to failure)
for 52°C (125 OF) and 25 °c (77 OF) were made using the Arrhenius plot
(log k vs. lltemperature, OK) data of Cannone for pure lead (second-order
rate: 162 kJ/mol) and 6% Sb-Pb alloy (first'-order rate; 62.8 kJ/mol).
calculated values are summarized in Table 13. The approximations assume
that the activation energy is the same for 1) lead alloys with the same
The
reaction-'rate order and 2) our test conditions.
(Cannone used 1.210 sp gr
sulfuric acid and maintained the plates at 0.080 V above the reversible
Pb02/PbS04 potential.) Growth rate is assumed to be independent of grid
design.
These estimates of life times are optimistic because they do not take
into account the effect of positive plate potential on corrosion-caused
grO\vth.
Our tests were carried out at the minimum growth potential of
0.075 V with respect to open circuit.
At open circuit or under cycling
conditions the corrosion rates may be several fold greater. Therefore,
grid corrosion is a likely cause of failure of a hybrid-vehicle battery.
Perhaps the best way of estimating corrosion life is by examining the life
-58-
-------
Tab13 13.
Estimated Growth-Rate Constants and Life Times for Pasted Plates
at 52 and 25 °Ca
Sb-As
Temperature. °c Rate Constant. b Rate Constant. %/day 2c Life Time, mo
%/day
71 -2 4.0
2.4 x 10 .
52 -'3 14.5
6.6 x 10
25 8.1 x 10-4 119
71 -2 3.3
3.3 x 10
52 9.2 x 10-3 11.9
25 11.2 x 10-4 97
71 -2 3.5
1. 4 x 10
52 3.9 x 10-3 12.5
25 4.7 x 10-4 103
71 -4 3.8
5.3 x 10
52 -5 20.0
1. 9 x 10
25 -8 300
8.4 x 10
Alloy
Li
Sb
I
V1
\0
I
Li-Sn
a
Horizontal growth.
b
Based on activation energy of 62.8 kJ/mo1.
c .
Based on activation energy of 162 kJ/mo!.
-------
of present-day 5LI batteries which utilize antimony-alloy grids that are
)erhaps 50% thicker than grids projected for use in hybrid-vehicle batteries.
Our corrosion te'sts indicate that lithium alloy is 20% more resists.nt to
corrosion than antimony alloy, and antimony-arsenic alloy is perhapslOO%
more resistant than antimony alloy.
On this basis, the antimony-ars~nic
alloy would be expected to yield a life time in a hybrid-vehicle battery
comparable to that of present SLI batteries.
Rod Specimens
Rod specimens were kept on test for 138 days at 71°C until all plate
and bare grid tests were terminated.
Specimens did not show significant
growth or pronounced curvature.
Three alloys, Cd-Sb-Ag, Sb, and Ca-Sn,
revealed casting defects (voids) not evident when rods were prepared for
testing.
Weight loss (per unit of original exposed surface area) and change
(decrease) in diameter are summarized in Table 14. Included in the table
are the times to failure for pasted-plate and bare-grid specimens.
Agree-
ment among the results from the three types of corrosion tests was obtained
only with the Cd-Sb-Ag alloy.
appeared the best overall.
However, in general~ Sb-As and Li alloys
The alloys tested can be divided into eutectic (Cd-Sb-Ag, Sb and
Sb-As) and noneutectic (Li, Li-Sn and Ca-Sn) alloys. Metallographic exa-
mination showed that corrosion in eutectic alloys is concentrated along
the interdendritic network, whereas in noneutectic alloys, attack occurs
preferentially along grain boundaries, the matrix being resistant to
corrosion.
Corrosion is more uniformly distributed in eutectic alloys be-
cauSe the interdendritic network is much finer than the grain boundary net-
work.
It is therefore less destructive because it is distributed over a
large area, although it progresses more rapidly and results in greater
\"eight loss. The behavior of Sb-As and Li alloys (Table 14. demonstrates the
relationship between material loss and plate (or grid) life time for the two
classes.
-60-
-------
Table 14.
Comparison Data for Rod, Bare-Grid and Pasted-Plate Specimens
Obtained during Accelerated Corrosion Test
WeiRht Loss, R/cm 2 Diameter Decrease, cm Time to Failure, days
a a Plates b c
Alloy Rods Rods Grids
--
Cd-Sb-Ag 0.36 0.071 36 70d
Li 0.07 0.010 114 117
128 132
Sb 0.32 0.058 84 99d
114
I Li-Sn 0.12 0.015 104d 99d
0\
..... 114d
I Sb-As 0.15 0.028 132
138
Ca-Sn 0.14 0.018 84d 99d
a 138-Day test; nominal diameter, 0.25 in. (0.64 cm).
b Based on capacity determined at 60-A (0.16-A/cm2) discharge rate to 0.95 V; failure
defined as capacity of 2.00 A hr or less.
c Broke apart while measuring growth.
d . .
Both specimens failed at the same time.
J
-------
3.2
CO~1PONENTS-LEVEL TEST PLATES
3.2.1
Concept Requirements and Performance-Test Design
Plate thickness, T, paste density, P, and grid design were identified
as variables in the component plates most likely to influence the perfor-
shown in
mance of lead-acid batteries in a hybrid vehicle.
For each matrix point
Table 15 four each of both positive and negative plates were
fabricated with the following selected values (nominal) for the parameters:
Grid Design; Gl:
G2:
Thickness, Tl:
T2:
T3 :
Paste Density, PI:
P2:
P3:
Normally spaced members
Closely spaced members
0.040 in.
0.060 in.
(0.10 cm)
(0.15 cm)
(0.20 cm)
0.080 in.
3
3.3 g/cm
3
4.0 g/cm
3
4.7 g/cm
Based on the fabricated plates and the test condition variables, tempera-
ture(s), electrolyte specific gravity(ies) and replication, a test plan
for positive plates was developed which included both grid designs, all
three thicknesses, two paste densities (PI and P3)' two temperatures
(27 and 52°C) and replication. These variables were g~ouped to establish
a statistically designed minimum test matrix. Evaluation of several
implementation approaches resulted in selection of the test plan layout
shown in Table 15.
information.
Table 16 shows the sequencing of tests used to obtain the optimum
Evaluation Period 1 represents two test runs on each of eight
plates representing cell numbers shown in Table 15.
(Four of these plates
are run first at tl followed by a run at tz and four plates are run first
at t2 followed by a run at tl.) For Evaluation Period 2, eight new plates
(duplicates of the plates from Evaluation Period 1) are selected and the
runs of Evaluation Period 1 are duplicated, except that the initial tempera-
tures are reversed. In Evaluation Period 3, eight new plates are selected
(two each for Cell Numbers 9, 10, 11 and 12 in Table 15) and run with the
-6Z-
-------
Table 15.
Test Plan Layout for Components-Level
Test Plates at One Temperature.
Gl G2
PI P3 PI P3
Tl 1* 2 3 4
. T2 5 6 7 8
T3 9 10 11 12
* Cell number denotes one of twelve matrix
points for sequencing the test run (Table 16).
-63-
-------
temperature sequencing shown in Table 16.
It should be noted that the
design of these evaluation periods is such that options are provided for
reducing the total number of test runs. For instance, if all of the effects
of temperature, paste density, plate thickness, and grid geometry are well
defined after Evaluation Period 1, there is no need for Evaluation Periods
2 and 3.
Likewise, Evaluation Period 3 is not necessary if the thinnest
plate (Tl) is shown to be best in Evaluation Periods 1 and 2.
Figures of merit for specific power and charge acceptance were
developed to provide a measure of battery performance on the components
test plate level.
The figure of merit for specific power is defined as
(fm)
sp
i75Y/W
where i75 is the current required to discharge the test plate to 1.5 V
in
75 se~, V is the average cell voltage for the 75-sec discharge, and W
is the plate w~ight.
Current rather than current density was used in the
figure of merit because all the test plates have identical geometrical
surface areas.
Furthermore, the average cell voltage, Y, rather than
the plate potential (vs. Hg/Hg2S04) was used since preliminary performance
tests on Gould Pb 660 positive plates of comparable size indicated that the
difference in the cell voltage and plate potential values was only 25 mV.
The current value required for 75-sec discharge was obtained from a Peukert
plot (log discharge time vs. log discharge current).
The figure of merit for charge acceptance was defined as
Capacity after cycling
Capacity before cycling
(fm)
ca
where the capacities are measured at constant-current discharge of 20 A
(0.28 A/cm2) to 1.5 V.
Performance data is obtained with the following test sequence:
Peukert Plot (Specific Power)
1.
2.
Charge at I A (0.014 A/cm2) for 1 hr (2.7 V limit).
2 .
Discharge at 30 A (0.42 A/cm ) to 1.0 V.
Charge at one-half the discharge rate to 2.7 V; at 2.7 V
3.
for 5 min.
-64-'-
-------
Table 16. Test Sequence of Test Plates
Test Condition
Evaluation Period Test Run .!.l .!.2 Total Plates Total Tests
2 8
1 1 6
7 4
Ib 3 5
8 2
2a 6 1
4 7
5 3
8 2
6 1
1 4 7
2b 5 3
.2 8
2a 1 6
7 4
.3 5
10 12
1 9 9
11 10
12 11
3
12 10
2a 9 9
10 11
11 12
a Test Run 2 is a rerun of Test Run 1 with the cell temperature tl and
t2 reversed.
b Test plates used in Evaluation Periods 1 and 2 correspond to plates
in Replicate Groups 1 and 2, respectively, in Section 3.2.4.
-65-
".
I, . ;"
-------
4.
Repeat Steps 2 and 3 for discharge currents of 20,10 and
2
5 A (0.28, 0.14 and 0.07 A/cm ).
Charge Acceptance
5.
Discharge at 20 A for 10 sec.
Charge at 10 A for 20 sec.
Repeat Steps 5 and 6 nine times.
6.
7.
8.
9.
Discharge at 20 A to 1.0 V.
Charge at 2.5 A to 2.7 V; at 2.7 V for 5 min.
3.2.2
Test Plate Fabrication
Grids with the nominal thicknesses and separation of vertical members
described in Section 3.2.1 were prepared from 4.5% Sb alloy. Those with nor-
mally spaced members (Gl) and 0.080- and 0.040-in. (0.20- and O.lO-cm) nominal
thicknesses were cast in corresponding grid molds. One-half each of the
O.20-and O.lO-cm molds were combined for casting the 0.15-cm thick grid.
A set of grids with closely spaced grid members (G2) was cast after both
sets of molds were recut to reduce pellets to one-half of their original
size; i.e., three additional vertical members were added. In the Gl and G2
grids, the vertical members are on 3/4- and 3/8-in. (1.9- and 1.0-cm) centers,
respectively.
(Measurement note:
the frame bar measurement is made from
the outer edge.) Outlines of the fabricated grid types are shown in Figure 9.
The actual grid thicknesses were 0.040, 0.066 and 0.088 in. (0.10, 0.17 and
0.22 cm) for the corresponding nominal thicknesses of 0.040, 0.060 and 0.080
in.
(0.10, 0.15 and 0.20 em), respectively.
Pastes were hand-mixed in the laboratory using standard automotive
oxide, Gould SPl02, and two standard automotive paste formulas each for posi-
tive and negative plates.
The lowest-density paste in each case was obtained
by adjusting a standard paste formula with water and sulfuric acid until the
desired wet paste desity was obtained. Positive paste formulas were PF-62,
PF-66, and PF-7lG which correspond to the nominal densities of 3.3, 4.0,
and 4.7 g/cm3, respectively. The actual wet paste densities were 3.81,
4.08 and 4.36 g/cm3, respectively. Negative paste formulas were PF-69,
PF-73, and PF-76FG which correspond to actual wet paste densities of 4.24,
3
4.47 and 4.69 g/cm , respectively.
Densities of negative paste are normally
-66-
-------
Figure 9.
.
.
Outlines of Grids Used in Components-Level Test Plates.
-67-
-------
slightly higher than positive densities.
proprietary Gould formulas.)
(All paste designations refer to
A total of 72 positive and 72 negative plates were fabricated.
I nd i-
vidual grid weights, paste density, and dried (unformed) plate weights were
obtained for those plates actually tested and are tabulated in Table 17,
Section 3.2.4. The dry paste weights were obtained by subtracting the grid
weights from the plate weights.
After pasting,the plates were cured and dried using conventional tech-
niques. Pasted plates were interleaved with pasting paper, wrapped in plastic
and allO\ved to cure at room temperature for 24 hr. Interleaved papers were
removed and the stacked plates were rewrapped in plastic and curing was
continued another two days.
to dry overnight in air.
The plates were then unwrapped and racked
Plates were then tank-formed.
After formation,. positives were washed
in running water and then were racked and dried' in room air' for three hours.
The negatives were first soaked in hot Stoddard solvent at 99°C. When
bubbling action ceased, the temperature rose to 127 °t and plates were removed.
Plates were allowed to cool in the air.
dry active material was not measured.
Actual density or porosity of the
3.2.3
Tes t Equipm'ent
Four components-level test stations were assembled.
These stations
were developed for controlled testing of the plate parameters defined with-
in the statistically designed experimental matrix.
The testers are capable
of providing accurate current and voltage control through various charge-
discharge sequences to provide data for calculating the figures of merit
for test plates.
One of the stations contains' additional programming
capability required for performing preliminary tests on plates to determine
the effects of cycling and past history on the figures of merit.
Each components tester consists of the following major elements (Fig-
ure 10):
.
A dc power supply with independent current and voltage controls
for charge and discharge,
-68-
-------
MANUAL
OVERRIDE
I
0-
\0
I
.. TIMERS
...
.., ~,
- SEQUENCER AND .... DATA
LOGIC CONTROL.. .... RECORDI NG
~~
. . .~. ... P.',
... ..
--... POWER CONT ACTOR .... TEST CELL
POWER SUPPLY r- AND LOAD ....
r.
Figure 10.
Block Diagram of Components-Level Tester.
-------
.J",
.
load resistors to absorb discharge power,
.
power contactor to transfer power from power circuitry to
the test cell and to se1~ct modes of operation (charge
or discharge),
.
sequencing and control logic, and
.
data recording.
The programmed sequence for the components-level performance test was
the following:
Step
Function
1
2
Off
Charge at 1.0 A for 1 hr (voltage limit
set at 2.7 V)
Discharge at ~O A to 1. 0 V
3
4
5
6
7
Charge at 15 A to 2.7 V
Charge at 2.7 V forS min
8
9
Discharge at 20 A to 1.0 V
Charge at 10 A to 2.7 V
Charge at 2.7 V fpr 5 min
Discharge at 10 A to 1.0 V
10
11
12
Charge at 5 A to 2.7 V
Charge at 2.7 V for 5 min
Discharge at 5 A to 1.0 V
Charge at 2.5 A to 2.7 V
Charge at 2.7 V for 5 min
13
14
15-20
21
22
23-40
41
42
43
44-60
Advance
Discharge at 20 A for 10 see
Charge at 10 A for 20 see
Repeat Steps 21 and 22 nine times
Discharge at 20 A to 1.0 V
Charge at 2.5 A to 2.7 V
Charge at 2.7 V for 5 min
Reset.
-70-
-------
.'
A schematic of the cell designed for components-level plate testing is
shown in Figure 11. All surfaces exposed to the sulfuric acid electrolyte
are Teflon (TFE) or glass except for rubber bands used to secure the plates
and separators.
Ports are provided in the Teflon cover for a thermometer,
a mercury/mercurous sulfate reference electrode (No. 288, Koslow Scientific
Co.) and addition or removal of electrolyte.
is lead welded to the plate tab.
Electrical lead to the plate
3.2.4
Performance Tests and 'Results
3.2.4.1
Preliminary Check-Ou~
Tests were conducted on standard lead-acid positive plates to check-
out the procedures developed for determining the figures of merit. The
plates were those us~d in a co~~rcially available battery (Gould Pb 660,
3-A hr capacity at the 20-hr rate) and which have nominal dimensions of
2 1/4 in. in width x 2 7/16 in. in height x 0.080 in. (5.7 em x 6.2 em x
0.20 em). The plate area is 11.0 in.2 (71 cm2). The test cells consisted
of two negative plates and'one positive plate with paper separators. A
mercury/mercurous sulfate reference electrode (at the same temperature 25
the electrolyte) was used in measuring the plate potentials.
The cell
was maintained at a constant temperature with a thermostated water bath.
Specific graviti of the sulfuric acid was 1.260.
After checking out the test equipment and the procedures for deter-
mining the figures of merit, tests were performed
to determine the effect
of previous cycling (discharging. and charging) on a plate's Peukert plot.
These tests were done by carrying out the performance test using several
permutations of the discharge current sequence, 30, 20, 10, 5 A. The
permutations selected for this check-out were the following:
Permutation
1
Discharge Current Sequence
30, 20, 10; 5 A
20, 10, 5, 30 A
10, 5, 30, 20 A
5, 30, 20, 10 A
2
3
4
-71-
-------
TEFLON CELL TOP
AND BINDING POST
REFERENCE ELECTRODE PORT
GlASS lEAKER
Figure 11. Schematic of Components-Level Test Cell.
-72-
-------
Preliminary results from tests on used Pb 660 positive plates indi-
cated that the Peukert plots were independent of the permutation used.
However, to eliminate any doubt, four new plates were used in a test of
broader ~cope 10 which each plate was run f.:lve times with the permutation::;
sequenced as follows:
Permutation
Run Plate 1 Plate 2 Plate 3 Plate 4
1 1 2 3 4
2 2 3 4 1
3 3. 4 1 2
4 4 1 2 3
5 1 2 3 4
Note that Run 5 ends the test for each plate with the starting discharge
current sequence and that although the permutation sequence is the same
for all plates, the permutation for Run 1 for each plate is different.
Altogether, twenty runs were made to establish the influence .of previous
cycling.
The plates were charged after e~ch discharge at one-half the
discharge current until the test cell voltage reached 2.7 V, then for
5 min at 2.7 V. There was no additional charging between runs. The
specific gravity of. the sulfuric acid was 1.260 and the tests were
conducted at 38 ae.
The results in terms of i75 and i75V values are listed in Tables 17
and 18, respectively. The i75V values averaged according to plate, permuta-
tion and run groups are listed in Table 19. The value of the product i75V
was used as the measurab1e.output parameter and the results for the analysis
of variance are given in Table 20. Sample calculations are given in Appen-
dix B.
Based on the F-ratio test it was concluded at the 95% confidence
level that test runs and order of current application have no significant
influence on the power output. Accordingly, all data were pooled and the
best estimate of the standard error of experimental variation was calculated
to be 2.02.
Since the overall average was 37.4 W, the coefficient of varia-
tion was found to be 5.4%.
Such
a small coefficient. of variation denotes
a well run experiment.
The data do indicate that there may be some influence
of past history at low power levels.
-73-
-------
Table 17. 75-Second. Discharge Currents for Pb 660 Positive Plates
Discharge Current, A
Run Plate 1 Plate 2 Plate 3 Plate 4
1 24.0 22.6 22.3 20.2
2 24.2 22.6 22.2 .18.9
3 22.3 22.5 21.8 23.3
4 21. 8 23.1 23.5 22.0
5 21. 7 23.6 23.5 22.5
-74-
-------
Table 18. Summary of i7SV Values for Pb 660 Positive Plates
i7SV' W
Run Pl.:tte 1 Plate 2 Plate 3 Plate 4
1 39.8 36.8 37.0 33.9
2 40.6 37.5 37.5 31.9
3 37.7 36.7 36.8 38.9
4 36.8 37.0 39.9 37.0
5 35.6 39.6 38.3 38.2
-7S-
-------
Table 19.
Average i7SV Values for Pb 660 Positive Plates
Plate
Average Power, W
--
1
38.1
2
37.5
.3
37.9
4
36.0
Permutation
Order
1 36.2
2 39.2
.3 37.5
I, 36.6
Run
1 36.9
2 36.9
3 .37.5
4 37.7
5 37.9
-76-
-------
Table 20.
Analysis of Variance for Power (Pb 660 Positive Plates)
Degrees of Mean
Source of Variation Freedom Squares
Plates 3 7.83
Runs 3 0.75
Permutations 3 6.74
Experimental variation 6 3.79
F-Ratio
2.06
1. 78
-77-
-------
3.2.4.2
Performance Tests
The physical properties of all the positive and negative plates that
were tested are listed in Table 21.' The positive plates were placed in
two replicate groups according to the matrix shown in Table 22.
Each
replicate group was further divided into two test runs with one test run
being made first at 27°C (80 OF) and the other being made first at 52 °c
(125 OF).
Upon completion of the first test run, the temperatures of the
cells were reversed and the test run was repeated.
For the second repli-
cate group, the initial temperature for each matrix point was reversed
from that of the first replicate group.
Positive plates to b~ tested were assembled into cells using two stand-
ard negative plates of the same size as the positive plates and paper sep-
arators. .The specific gravity of the electrolyte was 1.260. Results of
the tests are shown in Tables 22, 24 and 25. Tables 23 and 24 show the
figures of merit for specific power calculated as i75V divided by the plate
weight and paste weight, respectively. Table 25 summarizes the figure
of merit for charge acceptance.
A "burn-in" effect was observed in that the second test run performance
was found to be better than the first.
In order to check the stabilization
of the burn-in, two plates were subjected to additional tests. The results
summarized in Table 26 demonstrate that no significant change occurs after
the second run.
An analysis of the results shows that the "burn-in" f:.ffect
does not influence the selection of the optimum electrode configuration.
The
figure of merit was shown to improve with increasing temperature.
The re-
suIts of the analysis of variance for specific power are shown in Table 27
which indicates that all main effects are highly significant.
A preliminary
analysis showed that there were no significant interactions between the
design parameters.
Therefore, the optimum conditions were determined by
a simple regression equation on the main effects.
given in Appendix B.
Sample calculations are
It was recognized that pellet area (size) is correlated to grid type,
Therefore rather than use dummy variables for grid type, the regression
.
equation was run using pellet area as the variable for grid type. In addi-
tion~ although temperature is not a design parameter, it was also included
-78-
-------
Table 21. Properties of Components-Level Test Plates
Plate a
Grid Grid Paste Density, Paste Replicate
Plate ~ Wei~ht, ~ Thickness, cm ~/cm3 WeiKht. K WeiKht, K Group
P058 G1 14.5 0.17 4.36 15.9 30.4 1
P059 G1 14.6 0.17 4.36 15.9 30.4 2*
P062 G2 17.0 0.17 4.36 15.8 32.8 1
P063 G2 17.0 0.17 4.36 15.9 32.9 "2*
P065 G1 8.6 0.10 4.36 11.7 20.3 1*
P068 G1 8.7 0.10 4.36 12.3 21.0 2
P069 G2 10.1 0.10 4.36 11.7 21.8 1
P070 G2 10.4 0.10 4.36 11.6 22.0 2*
I P081 G1 14.4 0.17 3.81 14.4 28.8 1
...... P082
\0 G1 14.5 0.17 3.81 13.6 28.1 2*
I
P085 G2 16.8 0.17 3.81 13.7 30.5 2
P086 "G 17.0 0.17 3.81 14.0 31.0 1* "
2
P089 G1 8.6 0.10 3.81 10.6 19.2 1*
P092 G1 8.7" 0.10 3.81 10.1 19.8 2
P093 G2 10.1 0.10 3.81 9.1 19.2 1*
P094 G2 10.4 0.10 3.81 9.5 19.9 2
N089 G1 8.7 0.10 4.24 10.3 19.0
N092 G1 8.8 0.10 4.24 10.8 19.6
,
a first run at t1 (270C), then at t2 (520C); others run in
Asterisks denote plates
reverse order.
-------
i
00
o
Table 22. Distribution of Components-Level Test Plates between Replicate Groups
pl
P3
First Replicate
Tl
Gl
P089*
P065*
G2
P093*
P069
T2
Gl
P081
P058
G2
PD86*
P062
Second Replicate
Tl
Gl
P092
P068*
G2
P094
P070
T2
Gl
P083*
P059*
G2
P085
P063*
* Denotes plates run first at t.. .
LEGEND:
TI = 0.10-cm thickness
!„ = 0.17-cm thickness
2
G.. = normally spaced grid members (0.86-cm pellet area)
2
G_ = closely spaced grid members (0.39-cm pellet area)
3
PI = 3.81-g/cm paste density
1 3
P, = 4.36-g/cm paste density
-------
Table 23.
Figures of Merit for Specific Power Based on Plate Weight, Wig
Test Run 1 for Each Plate
tl t2
T1 T2 T1 T2
Gl G2 Gl G2 Gl G2 Gl G2
PI 1.27* 1. 04* 0.97 0.78* 1. 61 1.40 1. 26* 1.18
P3 1. 00* 0.92 0.65 0.66 1.15 1.04* 1.14* 1.13*
I
00
I-'
I
Test Run 2 for Each Plate
tl tz
T1 TZ T1 TZ
G1 G2 Gl G2 G1 GZ G1 GZ
P1 1.45 1.10 1. 05* 0.93 1. 77* 1.52* 1.42 1.20*
P3 1.04 1.00* 0.89* 0.84* 1. 37* 1. 25 0.99 1.01
* Denotes plates from Replicate Group 1.
-------
Table 24.
Figures of Merit for Specific Power Based on Paste \.Jeight, \~/g
Test Run 1 for Each Plate
tl 't
Z
Tl T2 Tl T2
Gl G2 Gl GZ Gl G2 Gl G2
PI 2.29* 2.20* 2.00 1. 73* 3.00 2.92 2.53* 2.62
P3 1. 73* 1. 74 1. 25 1.36 1.96 1. 93* 2.17* 2.42*
I
00
,N
I
Test Run 2 for Each Plate
,
tl t2
.,
Tl T2 Tl T2
Gl G2 . Gl G2 Gl G2 Gl G2
PI 2.70 2.31' 2.10* 2.06 3.21* 3.20* 2.93 2.66*
P, 1.77 1. 86* 1. 69~ 1. 75* 2.37* 2.37 1.90 2.09
3
* Denotes plates from Replicate Group 1.
-------
. Table Z5.
Figures of Merit for Charge Acceptance for Components-Level Test Plates
tl tz
TI TZ TI T2
G1 GZ G1 G2 G1 G2 G1 G2
PI 0.85* 0.22* 0.95 0.41* 1.02 1.11 1.00* 1.03
P3 0.35* .0.23 0.26 0.59 1.10 1.10* 0.40* 0.83*
Test Run 1 for Each Plate
I
00
W
I
Test Run 2 for Each Plate
t1 t2
Tl T2 T1 T2
Gl G2 G1 G2 G1 G2 G1 G2
.
PI 0.86 0.42 0.87* 0.89 1.02* 0.93* 0.96 1. 06*
P3 0.82 1.06* 1. 01* 1. 09* 1.07* 1.13 1.05 1.05
* Denotes plates from Replicate Group 1.
-------
Table 26. Summary of Burn-In and Temperature Effects for Components-Level Test Plates
. Test Run
Plate Figure of Merit 1 2 3 4 5 6a 7a
.
P089 Specific Power b 1. 27c 1.77 1.71 1. 75 1.72 1. 22 1. 37
Charge Acceptance 0.85c 1.02 1.01 0.95 0.90 0.91 0.85
P092 Specific powerb 1. 61 d 1.45 1.87 1.90 1.88 1.52 1.58
Charge Acceptance 1. 02d 0.86 1.00 0.98 1.01 0.87 0.86
r
00
~
I
a Additional runs made at 27°C.
b
Based on plate weight, Wig.
c First run made at 27°C, next four at 52 °c.
d
First run made at 52°C, second at 27 °c and next three at 52°C.
-------
Table 27.
Analysis of Variance for Specific Power for Components-
Level Test Plates. Figures of Merit Based on Plate Weights,
WIg (Test Run 2 for Each Plate).
Sum of .. Degrees of Sum of Mean
Variation F.reedom Squares Squares F-ratio a
Temperature (t) 1 0.312760 0.312760 >39
Thickness (T) 1 0.296208 0.296208 >37
Grid type (G) 1 0.079665 0.079665 > 9
Paste density (P) 1 0.262400 0.262400 >32
Experimental 11 0.090537 0.008230b
variation
a F-ratio r~quired for significance at 5% probability is 4.84.
b The standard error per measurement is 0.0082301/2 = 0.0287.
average measurement was 1.1764, the coefficient of variation
indicating a well run experiment.
Since the
is 2.44%,
;..
.,
-85-
-------
in the regression analysis to improve the predictive capability.
suIting equation is
The re-
! .
(fm)
sp
The observed and predicted values are given in Table 28.
=
2.977 + 0.0112t - 3.887T - 0.4657P + 0.3003G.
Since no inter-
actions wer~, detected within the region of experimentation, we can only say
that the region of optimum specific power is for a thin plate with low
paste density on a grid having normally spaced members.
If it is assumed that specific power is functionally related to these
parameters in a linear manner only, then the optimum combination is the
thinnest plat~, largest pellet size and minimum paste density possible con-
sistent with design constraints of weight and cost, and fabrication capa-
bilities. Since the thin plate provides the optimum specific power, it was
decided not to run the additional tests with the thickest plates (0.20 em).
In summary, within the scope of the experiment, the design for a posi-
tive plate that provides optimum specific power is the following:
Paste Density:
Thickness:
Normally spaced members,
pellet area size)
3.81 g/cm3
0.10 em
2
G1 (0.86-cm
Grid Type:
Further experimentation around these points is necessary to more accur-
ately pinpoint the ,combination of design parameters which would provide op-,
timum specific power.
Components-level testing of negative plates was limited to the
with physical parameters corresponding to those of the best positive
The physical parameters of the negative plate were given in Table 21.
plate
plate.
The
results of the test, Table 29, show that for negative plates'the' figures of
merit for specific power are slightly larger and those for charge acceptance
slightly lower than those for the corresponding positive plate.
The effect of operating temperature on negative plate performance
(Table 29) is a reduction in both specific power and charge acceptance with
decreasing temperature.
-86-
-------
Table 28.
Figures of }1erit for Specific Power for Components-Level Test Plates
from Test Run 2 by Grid Type and Temperature
I
CIO
'.J
I
Gl G2
tl t2 tl t2
Tl T2 TI T2 Tl TZ Tl T2
, Pl 1.45b 1.05 1.77 1.42 1.10 0.93 1. 52 1. 20
1. 37 1.10 1. 65 1. 38 1. 23 0.96 1. 51 1. 24
P3 1.04 0.89 1.37 0.99 1.00 0.84 1. 25 1.01
1.11 0.84 1. 39 1.12 0.97 0.70 1. 25 0.98
a Specific power in Wig based on plate weight.
b
Top, observed value; bottom) predicted value.
-------
Plate
Table 29. Summary of Components-Level Performance Tests on Negative Plates
Test Run a
Figure of Merit 1. 2 3 I, 5 6 7 8
Specific Power b 1.96 1.90 1.69 1.39 1.46 0.87 1. 02
Charge Acceptance 0.92 0.92 0.86 0.66 0.74 0.48 O. L~3
Specific Pow.er b 1. 90 1. 82 1. 81 1.35 1. 36 0.89 1.06 1.03
Charge Acceptance 0.88 0.88 0.89 0.88 0.81 0.50 0.40 0.38
N089
N092
I
00
00
I
a Test Runs 1 and 2 made at 52°C, 3, 4 and 5 at 27°C and 6, 7 and 8 at 0 °c.
b Based on plate weight, Wig.
-------
3.2.4.3
Cycle-Life Tests
Cycle-life tests were made on the best positive plates (P089 and P092).
A cycle consisted of a 20-A (0.28-A/cm2) discharge for 15 sec followed by a
2
10-A (0.14-A/cm ) charge for 30 sec. Before the cycle test, bot~i plates
were charged at 1 A (0.014 A/cm2) or 2.7 V which~ver limited, for one hour.
The capacity of one plate (P089) was measured after each 100 cycles.
Capacity was determined using a 5-A (0.070-A/cm2) discharge. After
- 1.
measuring the capacity, the plate was charged at 2.5 A (0.035 A/cm ) to a
2.7-V cut-off, then held at 2.7 V.for 5 min. The other plate (P092)
was subjected to 600 cycles after which the cell was charged at 1 A (0.014
A/cm2) 01:" 2.7 V for one hour. The test sequence was repeated until the
plate failed. Plate failure was defined as the point at which the dis-
charge voltage decreased to 1.5 V. The tests were conducted at 52°C.
current densities were ~ntentionally chosen to accelerate failure.
High
The results of the cycle-life test of Plate P089 are shown in Table 30
Even though the plate failed after 589 cycles, the plate ~ould be restored
by charging at 2.5 A (0.035 A/cm2). However, the number of cycles to
f&ilure decreased with each recharge. After 900 cycles, the plate could
not be reactivated. Plate P092 failed after 868 cycles. Changing the
charge time from 30 to 32 sec to provide 15% overcharge restored the
ulate for an additional 125 cycles.
Examination of the plates indicated
that failure was due to loss of active material.
Approximately 15 to 50%
of the active material was lost from Plates P089 and P092, respectively.
The cell containing Plate P092 showed evidence of an electrical short
caused by bridging between the plates by active material accumulated at
th~ bottom of the cell.
3.3
TEST CELLS
3.3.1
Cell Design and Fabrication
T~o ty~es of pasted-plate test cells were designed using the results
of the components level tests and of the mathematical modeling of grid
geometry.
One (Type IA) emphasized high specific power while the other
(Type IB) compromised power for the sake of life.
Grid alloys selected for
. the test cells \"rere 0.013% Li (Type IA cell) and 2.5% Sb-O. 45% As (Type IB
-89-
-------
Table 30. Cycle-Life Test Data for Components-Level Test Plate P089 at 52 °c
a
Cycle
b
Capacity. A.hr
Remarks
1.14
Initially charged at
1 A (0.014 A/cm2).
100
0.64
200
0.63
300
0.58
0.92
Stored (fully charged)
over weekend at 52 °c
- 400
0.58
500
0.56
589
Failed. <1. 5 V.
600
0.42
672
Failed, <1. 5 V.
700
0.32
0.83
Stored (fully charged)
overnight at 52 °c.
750
Failed, <1. 5 V.
800
0.26
825
Failed., <1. 5 V.
900
0.11
a Discharged at 20 A (0.28 A/cm2) for 15 sec, charged at
10 A (0.14 A/cm2) for 30 sec.
b Determined at 5-A (0.070-A/cm2) di~charge rate, then.
charged at 2.5-A (0.035-A/cm2) rate.
-90-
-------
cell). Separators were sintered polyvinyl chloride (PVC) membranes (Ethyl
Corporation), 0.020-1n. (~.051-cm) thick with a 0.013-in. (0.033-cm)
backweb. In the Type IB (long~life) cells, pairs of separators were heat-
sealed together to form envelopes in which the positive plates were placed,
whereas in the Type IA (high-specific-power) cells the separators were used
singly in the conventional manner.
Additives were not used in either cell.
Design parameters and projected operating characteristics of the
pasted-plate test cells are the following:
Number of Positive Plates
10
11
Number of Negative Plates.
Single Plate Area (10.8 em x 10.8 cm)
Grid Thickness
233 cm2
0.10 em
Naximum Current
600 A
Maximum Current Density
Minimum Cell Voltage
2
2.6 A/cm
1.5 V
Approximate Cell Stack Size
4 cm x 17 cm x 15 cm
A grid mold was fabricated for the grid design that resulted from
the mathematical modeling study described earlier in Section 3.1.1.
Grids
of O.lO-to O.ll-cm thickness were cast from Li and Sb-As alloys.
(An out-
line of an actual grid was shown in Figure 5, Section 3.1.1.) Positive
grids were pressed to 0.10 em and negative grids to 0.089 em before pasting.
Connector straps and terminal posts were also cast and a burning comb for
a 2l-plate element was fabricated.
Five Type IA (Li~alloy grid) cells were fabricated.
Their positive
and negative group weights are sununarized in Table 31. Typical grid weights
were in the range of 28-33 g. The lighter grids were segregated for use in
negative plates and the heavier for positive plates.
Pastes were hand mixed
in the laboratory using standard automotive oxide. The positive paste for-
mula was PF-62 and the negativt paste formula was PF-69. Wet densities of
the pastes were 3.88 and 4.20 g/cm3 for the positive and negative, respec-
tively.
Five Type IB (As-Sb-alloy grid) cells were fabricated. Typical grid
weights were in the ranges of 27-32 'g for negatives and 32-35 g for positives,
-91--
-------
I
I
Table 31. . Group Weights for Type IA Cellsa
Positive Group Weights, g Negative Group Weights. g
Cell Grids Formed Plates Active Haterial. Grids Formed Plates Active Haterial
.
1 295 663 368 300 665 362
2 657 362 660 360
3 651 356 656 356
4 653 358 657 357
5 644 349 656 356
I
1.0
N
I
a Each cell contains 10 positive and 11 negative plates.
Weight of straps and posts not included.
-------
with averaged weights of 30.4 g for negatives and 33.3 g for positives.
Averaged group grid weights, group formed wiehgts, and active material per
cell are given in Table 32. The positive and negative paste formulas were
PF-66 and PF-73, respectively. Average wet paste densities were 4.0 and
4.45 g/cm3 for the positive and negative, respectively.
Construction and assembly were the same as with the lithium-alloy
cells except that the polyvinyl chloride separators were heat-sealed on
three edges to form pockets (two separators per pocket) in which the posi-
tive plates were inserted.
The foot was removed from the gtid because,
with the envelope separator construction, there is no danger of sediment
collecting on the bridge ridge and shorting beneath a separator which may
net be properly seated on the ridge.
Plates were cured, dried and tank formed in the same way as were
the components-level test plates.
The formed plates were dry charged as
before, then joined to connecting straps to form positive and negative
groups and assembled into cells. All containers were cut from Group Size
22F automotive battery cases.
Because the plates were not wide enough to
fill the container, foamed polystyrene (closed pores) was used to fill the
extra space.
A photograph of a Type IB cell is shown in Figure 12.
The
cells were heavier than necessary because of excess volume in the cell
compartment of the Group Size 22F container and because the containers were
cut oversize for stability in handling and testing. Excess weight
correction for calculation of specific power is estimated as 161 g as
shown below.
Excess weight
Extra plastic outside cell
Extra plastic in cell (excess
length and width)
Styrofoam (space filler)
Extra electrolyte in sediment well
90 g
30
5
19
Extra connections
37
Total
181 g
-93-
-------
Cell
1
2
3
4
5
I
1.0 a Each
~
I .
Table 32.
a
Group Weights for Type IB Cells
Positive Group Heights, g Negative Group Weights, g
Grids FormedP1ates Active Material Grids Formed Plates Active Hateria1
-
333 691 358 334 654 320
691 358 654 320
692 359 654 320
692 359 656 322
692 359 656 322
cell contains 10 positive and 11 negative plates.
Weight of straps and posts not included.
-------
,
.",
.
-
-
..,
....
~,-::-
-::'<
3.1
';I~~
~i
~"i:
:::f'
.......
Figure 12.
Type IB Test Cell.
-95-
-------
Deficient \.Jeight
Top terminals and bushi~~s
20
8__.+_-
Net Excess Weight
161. g
.3.:1.2
L;s t Procedures
The cell test procedure was designed so that it would provide infor-
mation for predicting the performance of a full-scale battery system.
overall test sequence conducted at 52°C (125 OF) included three major
The
steps:
Step 1.
Performance Test for Specific Power.
Peukert plots were
obtained to establish initial performance characteristics of
the test cells.
Tests were made on all cells to establish
manufacturing variability for each type of cell and were run
three times on each cell to determine the extent of any burn-in
effect. Discharge currents used were 600, 400, 200 and 100 A.
Since each cell contained ten 233-cm2 positive plates, the
currents corresponded to current densities of 0.26, 0.17,
0.086 and 0.043 A/cm2, respectively. Charge \vas at one-half
the discharge currents to 2.7 V; then 5 min at 2.7 V.
SteD 2. Cycle Test. One cell of each type, preferably the one
\vhich exhibited the highest specific power in Step 1 was
to be subjected to 20,000 charge-discharge cycles. The
cycling profile and power levels approximated those in
Exhibit I of RFP No. EHSD 71-Neg. 100:
Cycle A
50-A (0.021-A/cm2) discharge for 10.8 see
25-A (0.011-A/cm2) charge for 23 see
Cycle B
365-A
2
(0.16-A/cm)
2
(0.08-A/cm )
discharge for 25 see
charge for 50 see
l80-A
Cycle A was repeated 29 times followed by one Cycle B.
The dis-
charge and charge times were calculated to provide a 5% over-
charge for the total of 29 "A"'ind 1 "B" cycles.
Note that the
-96-
-------
current densities in the cycle test (Ste~ 2) were significantly
lower than the 0.31-A/cm2 discharge/0.15-A/cm2 charge profile
used in the accelerated cycle-life tes~s on the components level.
Although the selected current densities were of reasonable mag-
nitude, it s~ould be noted that the test sequence was still a
severe one since the cells were cycled continuously at a high
ambient temperature (52°C) and were not provided any additional
charging at a lower rate.
Step .3. Performance Test. A repeat of Step 1 following 20,000
cycles in Step 2 was included ~n the test. sequence to establish
the effect of cycling on the figure of merit for specific OO~Jer.
The figure of merit fo( the performance tests (both Steps 1 and 3)
was the same as the pne for specific power used iri the components level
tests:
(fro)
8P
=
i75V/W
where i75 is the current required to discharge the test cell to 1.5 V i~
75 sec, V is the average cell voltage for the 75-sec discharge and W is tIle
cell weight. The current value required for 75-sec discharge, i75! is
obtained from the Peukert plot (log discnarge time ~. log discharge
current).
Tests were implemented by using a cell tester which was designed
and constructed around a 600-A, 40-V power supply (Christie Electric
Corp., Model IC040-600E245). The tester design was identical to the
block diagram shown in Figure 10 for components-level tester, except for
higher out~ut power levels. Removable programming drums were used to
program the tester for either the performance or cycle test schedule as
described previously. The bipolar cells and batteries were tested using
the components-level tester adjusted for the proper current levels. The
tcst cells were thermostated at 51°C (125 OF) using a constant temperature
"later bath. A view of the test station is shown in Figure 13.
-97-
-------
Figure 13.
Battery Test Equipment. Left, cell tester; right,
components-level tester; thermostated baths with
cells in fume hood.
-98-
-------
3.3.3
Tests and Results
3.3.3.1
Performance
The perforIT~nce test was run at least three times for each cell and the
results of the last test for each cell are assumed to be free of burn-in
effects.
Table 33 is a summary of the results for specific power whereas
Table 34 shows the results of the last run for each celL
(Actual cell
weights, uncorrected for excess weight, were used.)
A test of a 61-A hr,
Group Size 22F cell is included for comparison. It can be seen from Table
34 that the specific power of both Type IA and Type IB cells is approximately
twice that of the Group Size 22F cell.
To evaluate cell differences, the last
run fo~ ~ach cell (Table 34) was used to analyze the standard errors and
specifi~ power output. Ther~sults are:
Mean Specific Power, W/lb
Standard Error
Type IA
147
2.24
Type IB
140
11.04
The range of specific powers for Type IA cells is significantly larger than
that for the Type IB cells. With this large variation and few test sample, the
~-test does not indicate that the cell types are different in performance.
A comparison of the specific power for the separate runs on
individual cells (Table 33) shows a large variat~on.indicating a large
test error. Components of this error may be a burn-in problem, changes in
test procedures, inefficient test equipment, and unknown sources of variation.
Ignoring an unequal number of runs, and assuming all tests are compatible,
the overall average specific power for each cell was computed.
The results
are given in Table 35.
Since the largest value for a Type IB cell is
lower than the lowest Type IA cell value, it is concluded that the Type IA
cell has higher specific power than the. Type IB cell. The figures of
merit in the above discussion are based on a 75-sec discharge time.
For
discharge times shorter than 75-sec the figure of merit increases because
of the current-time relationship shown in Figure 14 which is a typical
Peukert plot for a Type IA test cell.
a current of 600 A can be maintained.
For example, for a 20-sec discharge,
This discharge ~ime corresponds
-99-
-------
Table 33.
Summary of Performance Test Data for Pasted-Plate Cells
IA-2
W/1bb End-of-Discharge
Specific Power, Voltage, V
145 1.0
141 1.0
134 1.0
162 1.5
. 160 1.5
156 1.0
152 1.0
154 1.0
155 1.5
152 1.0
151 1.0
146 1.0
149 1.5
155 1.0
149 1.0
144 1.0
147 1.5
134 1.5
154 1.0
150 1.0
145.. 1.0
138 1.5
130 1.0
136 1.5
142 1.5
150 1.0
138 1.5
141 1.5
141 1.0
130 . 1.5
141 1.5
133 1.5
137 1.5
143 1.0
133 1.5 .
141 1.5
144 1.0
144 1.0
138 1.5
141 1.5
Cell
IA-1a
IA-3
IA-4
IA-5
IB-1
IB-2
IB-3
.IB-4
IB-5
a Denotes Type-IA cell, Serial Number 1.
b Specific power based on discharge time to 1.5 V.
-100-
-------
Table 34. Summary of Peukert-Plot Test Data for Pasted-Plate Test Ce11sa
Cell i7~ ~ (fro) ,W/1b
------sp
IA-1 500 1.64 160
IA-2 480 1.63 155
IA-3 462 1.63 149
IA-4 .418 1.60 134
IA-5 430 1.63 138
IB-1 452 1.63 142
IB-2 442 1.63 141
IB-3 428 1.63 137
IB-4 448 1.64 141
IB-5 440 1.65 141
22Fb 270 1.56 70.4
a Last run.
b Group Size 22F, Gould Power Breed 22F-GP-61.
-101-
-------
Table 35.
Averaged Specific Power Values for Pasted-Plate Cells
Cell Specific Power , W/lb
IA-1 148
IA-2 154
IA-3 150
IA-4 146
IA-5 147
IB-1 136
IB-2 143
IB-3 136
IB-4 139
IB-5 142
-'-102-
-------
10.0
6.0
z
~
..
w
~
~
2.0
1.0
0.6
20
75 SEC
50
60
100 200
. CURRENT I A
500
1000
600
Figure 14.
Typical Peukert Plot for Type IA Test Cell
-103-
-------
roughly to the two 9-sec.. 90.2-kW peak discharges for the power profile of
the [PA specifications. Th~ average voltage during this discharge is
approximately 1.6 V so the specific power for a 20-sec discharge is approxi-
mately 188 W/lb or 205 W/lb based on corrected cell weights as discussed in
Section 3.3.1.
3.3.3.2
Cycle Life
Cycle-life data was analyzed by plotting the end-of-discharge vol-
tage for the liB" cycle as a function of the total number of cycles.
These
voltages were obtained every 300 cycles.
When the end-of-discharge vol-
tage reached 1.5 V the cell ,vas considered to have failed since this was
the minimum c~ll voltage given in the revised EPA specifications.
The comparison of the results for Cells IA-3 and IB-l to those for
a cell f '~om a 6l-A hr, Group Size 22F battery is shown in Figure 15. The
cycle 1:Ue to 1. 5 V was approximately 10,500 cycles for Cell IB-I and 8,000
cycles for Cell IA-3. Thi$ corresponds to 400 and 260 high power "B" cycles
for Cell IB-I and IA-3, respectively.
Even though the revised EPA
specifications call for 500 cycles at 70.5 kW, the power level for "B"
cycles actually used in the cycle testing was 72.6 kW.
Since the "B"
cycle ~laS repeated after 29 rather than 200 lmv power "A" cycles as
required by EPA, the test used represents a more stringent cycle-life
test. Although the end-of-discharge voltage for the Group Size 22r cell
had decreased to 1.5 V at approximately 90 cycles, the cycle test was
continued because the failure was attributed to the large iRdrop due to
the high resistance of 1.25 m~ (compared to resistances of 0.58 and 0.62 ~
for Cells IA-3 and IB-l. respectively). Degradation of the Group Size 22F
. .
cell beyond that point was slow, with the end-of-discharge voltage reaching
1.28 V at approximately 30,000 cycles (1,000 "B" cycles).
The end-of-discharge voltage for "A" cycles (lQ-kW discharg-a) re-
mained above 2.0 V for all the cells tested for the entire duration of each
test represented in Figure 15.
Therefore, the cycle life of the cells was
affected more by the h~gh-power rather than low-power cycling. The results in
Figure 15 were obtained without any additional low-rate charging. Cell IB-2
was run with attempts made to increase cycle life by introducing low-rate
-104-
-------
2.0
>
..
w
0
«
;- 1 ,.
....J ..:I -
I 0 -
t-' >
0
VI ....J
I ....J
W
U
1.0
Figure 15.
-
167 "B" CYCLES
5000
- ---lB-1
--
1A - 3
GROUP SiZE 22F, 61 A HR
--......,
"
,
-'
,
\
\
-
-
-
334 "B" CYCLES
500 "B" CYC LES
10,000
TOTAL "A" AND "B" CYCLES
15,000
End-of-Discharge Voltage for Cycle B~. Cycle Number for Test Cells
-------
charging after approximattly 2,000 and 9,000 cycles, but the end-of-
discharge voltage still decreased to 1.5 V after 12,000 cycles. Cell
IA-3 recharged after 8,000 cycles, also exhibited similar results with
end-of~discharge voltage falling below 1.5 V several hundred cycles
after resumption of cycling. Cell IA-2 was lost during test after 5,000
cycles because of a malfunction in the cell tester.
1\ post-mortem of both Type IR cells shm.,ed that the positive active
material had become s6ft and that shedding had occurred.
Insufficient
charging in the cycling program probably contributed to the softness of
the active material.
Shorting through the separator was observed as
evidenced by several spots of metallic lead being present on the positive
plates.
The lead spots were about 1.3 to 2 em in diameter and about four
were found in each test
c:~ll.
The actual shorting through the separator
(lead treeing) could be observed only by microscopic examinations.
As
expected, the grid was not noticeably corroded.
The separator bagR which
enclosed the positive r:-lates were not intact along the bottom edge, also
providing a path for s~orting. Shorting also contributes to shedding
of the active material by causing the cell to run at a decreased state
of charge.
Since the main failure mode was caused by shorting through the
separator, it is felt that the results were not a realistic test of the
new plates and great improvement would result from the use of other
separators.
Use of microporous separator materials such as Darami~ (micro-
porous, extruded polyethylene, W.R. Grace, Polyfibron Division), which
have proven performance i.n heavy-duty industrial batteries, could prevent
shorting and thus increase cycle life of the cell.
Paper separators
as used in Group Size 22F cells tested can also be investigated.
Previous
high-current cycle test results on the components level; Section 3.2.4.3,
indicate that plates with paper separators which were cycled with a 2-A/in.2
(0.3l-A/cm2) discharge compared to 1 A/in.2 (0.16 A/cm2) for the "B" cycles
in the cell tests ran for 600 cycles before failure. Considering that the
Group Size 22F cell has a long cycle life, but insufficient specific power,
artdthat both Type I cells had sufficient specific power but short cycle
life, a trade-off in design between these cells should result in a high-power
cell with a long cycle life.
-106-
-------
3.3.4
Sununary and Recommendations
It is recommended that development of conventional lead-acid batteries
be continued.
This work should be directed toward a better understanding
of the causes of failure of lead-acid batteries in hybrid vehicle duty and
of the life that can be obtained.
The effect of separator and the method of testing on battery life
should be investigated.
The failures that have been observed so far in
the program have all been caused by separator failure and the softening
of the positive active material.
This is greatly affected. by the manner in
which charging is accomplished, and especially by incomplete charging, which
fails to re-cem~nt the active material and leads to softening.
The very
stringent cycling conditions that have been used are probably useful as an
accelerated test for comparing various cell designs but they hardly permit
estimation of battery life to be expected in a hybrid vehicle.
At least one life test should be carried out under cycling condi-
tions more representative of hybrid-vehicle duty, with more shallow cycles,
which will allow more time for complete recharge after an occasional deep,
full-power discharge.
Also, the present cycling regime should be tried
with provision. for more complete recharge.
For example, an occasional
charge for a few hours could be interspersed in the cycling to assure com-
plete recharge.
State of charge should be monitored by measurement of elec-
trolyte specific gravity.
Further development of conventional cells should stress the effect
of various types of separators, alloys, lead. oxides, and perhaps addi-
tives to the electrolyte (e.g., phosphoric acid, which has been observed
to reduce softeoing of positive active material).
Present grid design
and size should be retained.
Accelerated corrosion tests of this grid
should be carried out at three temperatures to ascertain the temperature
effect; it is recommended that the two lead alloys already s2lected
(0.013% Li and 2.5% Sb-0.45% As) be compared with the conventional 4.5% Sb
and that an improved quaternary alloy (0.07% Ag-l.5% Sb-l.6% Cd) also,
be studied.
Separators, oxides, and additives should be studied in
component tests.
Results of these studies should be utilized in the
construction of improved test cells in which performance and life can
be evaluated.more definitively.
-lG7-.
-------
4.
DEVELOPMENT OF BIPOLAR BATTERIES
4.1
ELECTRODE S'!'UDIES
4.1.1
Ob;ectivzs
The objectives in d~velopment of a bipola1 lead-acid battery were to
develop
.
a light-weight electronically conductive substrate to which
active material could be firmly attached and which would be
inert in the cell environment,
.
a method of attaching the active material utilizing Plante
formation or pasting techniques, and.
a method of constructing a bipolar battery so that individual
cells are sealed from each other.
The performance goals were to sustain a 2-A/in.2 (0.3-A/cm2) discharge rate
.
for 60 sec with a cell voltage .::.1.5 V and a recharge time of twice the pre-
vious discharge period.
A specific power goal of 300 W/lb (137 W/kg) and
a target life of 5 years was desired.
4.1. 2
Conductive Substrates
Several approaches to develop a light-weight conductive substrate were
evaluated.
Considerations were given to conductive plastics and gold-plated
titanium. Calculations of iR losses in a bipolar substrate due to its re-
sistivity and thickness at 2A/in.2 (0.31 A/cm2) were made and are depicted
in Figure 16. At the target current density, the iR loss will be 47 mV
per cell if the substrate resistivity and thickness are 5 Q cm and 0.010
in.
(0.0254 cm), respectively.
At a resistivity of 1 Q cmand thickness
of 0.025 in. (0.064 cm), iR loss will be 20 mV.
Gold-plated titanium was eliminated from considertion early in the
program due to difficulites in achieving a pore-free, adherent gold plate
and expected difficulties in working the titanium into a usable configura-
tion.
Cornmerically available conductive epoxies and thermoplastic materials
were also found to be unsuitable because the usual
conductive fillers
(e.g., carbon, graphite, copper and silver) would react with the cell en-
viroronent.
Results of subjecting various conductive plastic materials
to anodic polarization at the oxygen evolution potential in 1.260 sp gr
-109-
-------
>
..
V1
V1
o
-J
0::
0.4
0.3
0.2
O. 1
0.0
o
5
10
RESISTIVITY, n CM
Figure 16.
IR L688 in Bipolar Substrate at 0.31 AI cm2
vs. Substrate Resistivity at Various
Substrate Thicknesses
-llO-
115
I
-------
sulfuric acid at room temperature are shown in Table 36.
It was clearly
evident from these results that graphite or carbon black was definitely
not acceptable as conductive fillers and that glassy or vitreous carbon
~ould be acceptable.
Vitreous carbon (a product of Beckwith Carbon Corp. ,
Van Nuys, California 91406) is a distinct form of carbon which is amorphous,
has a true density of 1.47 g/cm3, a hardness of 7 on the Moh scale, and
a resistivity of ~10-3 Q em.
The development of a substrate material was therefore concentrated
on the fabrication and evaluation of various plastics filled with finely
divided vitreous carbon powder.
Initially, substrates were fabricated
using epoxy, TFE and FEP Teflon, polypropylene, polyethylene and fluoro-
silicone rubber filled with varying quantities of vitreous carbon.
Poly-
propylene and polyethylene were eliminated from further consideration
because their mechanical properties were seriously degraded when they were
filled with vitreous carbon to the extent that the resistivity was accep-
table.
TFE and FEP Teflon and fluorosilicone rubber were also eliminated
due to difficulties encountered in fabrication and expected difficulties
in developing adequate seals with these materials.
However, epoxy-vitreous
carbon systems appeared promising and further efforts were concentrated
on their development.
It was found that the lowest resistivities could be achieved by hot
pressing the epoxy-vitreous carbon mixtures during the gel stage and oven
curing.
Some early results with CIBA 6010 resin and CInA DP-152 and 951
CTETA, triethylenetriamine) are listed in Table 37. Data for CIBA 7079/
972 and 1139/972 are included. Depending on the epoxy system used, temp-
eratures of 100 to 175 °c and pressures of 2 to 10 tsi with press times
of 5 to 15 min were required.
The optimum hardner was found to be a
methylene dianiline (MDA, supplied by CIBA Products Co. as CIBA 972).
Several resins were found to be acceptable from a chemical resistance view-
point; these included phenol Novolac liquid resin (CIBA EPN 1139) and con-
ventional EPI-BIS resins (CIBA Araldite 6005 and 6010 liquid resins and
CIBA Araldite 7072 solid resin).
With the solid resin, the proper pro-
portions of resin, hardner and vitreous carbon were mixed and ground to
a fine powder and then compression molded.
When liquid. resins were used,
-.111-
-------
Table 36. Anodic Corrosion Test Data on Substrate Materia1a
Weight
Plastic Conductor Conductor, Vol % Resistivity, Q cm Change, % Comment
Epoxy Vitreous carbon 67 16.1 1.7 No visual effect, resis-
tivity increased to 43.8
Q cm.
Fluorosilicone Vitreous carbon 48 12.8 0 .l~ No visual effect, resis-
rub b er tivity ir..creased to 20.8
Q cm.
Polypropylene Vulcan XC72 23 3.1 -8.0 Sample corroded through
I graphite at test clip contact.
f-' -4
f-' Graphoil 181 Original material is 61%
N --------- 8.5 x 10
I sample swelled
porous;
from 0.025 to 0.06 cm
in thickness.
Polypropylene Vitreous carbon 38 65 Sample broke due to
mechanical force.
Polyethylene Carbon black 2.8 -1.4 Visible corrosion at the
liquid line.
TFE Graphite 85 0.4 Sample corroded through
at test clip contact.
a
Test run 4 days.
-------
Table 37. Results of Conductive Plastic Fabrication Experiments
Vitreous
Epoxy Carbon. Wt % Cure Resistivity, ~ cm
CIBA 6010/152 40 Gravity, 40 DC, 4 hr 9.3 x 103
CIBA 6010/152 45 Gravity, 40 DC, 4 hr 3
2.4 x 10
CIBA 6010/152' 50 Gravity, 40 DC, 4 hr 697
CIBA 6010/152 50 Gravity, 40 DC, 4 hr 179
CIBA 6010/152 55 Gravity, 40 DC, 4 hr 143
CIBA 6010/152 60 Gravity, 40 DC, 4 hr 65
CIBA 6010/152 60 Gr'avi ty, 40 DC, 4 hr 125
CIBA 6010/951 50 Gravity, 40 DC, 4 hr 6 x 105
CIBA 6010/951 55 Gravity, 40 DC, 4 hI' 2.3 x 103
CIBA 6010/951 60 Gravity, 40 DC, 4 hr 4.8 x 103
CIBA 6010/152 71 Rolled, 50 DC, 2 hr 31.4
CIBA 6010/152 71 Rolled, 50 DC, 5 hr 16.1
CIBA 6010/152 80 Pressed, 60 DC, 1 hr 0.18
CIBA 6010/152 70 Gravity, 60 DC, 40 hr 6.6
CIBA 6010/152 80 Pressed at 1 tsi,
60 DC, 1 hr 0.13
CIBA 6010/152 80 Pressed at'l tsi,
60 DC, 1 hr 0.2
CIBA 6010/152 75 Pressed at 3 tsi,
100 DC, 10 min 0.2
CIBA 6010/152 80 Pressed at 3 tsi,
1000C, 10 min 0.2
CIBA 6010/152 80 Cold pressed at 6 tsi; "
"
press cured at 1/2 tsi,
100°C, 10 min 0.12
CIBA 6010/152 77.5 Cold pressed at 6 tsi;
press cured at 1/2 tsi,
100°C, 10 min 0.29
CIBA 7079/972 60 Pressed at 10 tsi;
heated to 150°C, 15
min; hot pressed at 2
tsi, 1500C, 10 min;
cured at 1000C, 16 hr;
cured at 1600C, 4 hr;
annealed at 500C, 16 hr 1-2
CIBA 1139/972 60 Same as above 0.85
'-113-
-------
the proper proportions of resin. hardner and vitreous carbon were mixed
with a solvent (acetone). the solvent evaporated. and the epoxy partially
cured ("B-staged").
The resulting material was then ground 'to a ,fine
powder and compression molded.
The initial substrat~s were compression molded using an aluminum
mold where one die had a rib pattern and the other was a flat surface.
These substrates contained 80 wt % vitreous carbon were 0.020-in. (0.05l-cm)
thick and had a 0.015- or 0.020-in. (0.038- or 0.051-cm) high raised
grid pattern on one side.
A O.030-in. (0.076-cm) thick lead back plate
to serve as the current collector was placed in the flat surface meld and
bonded directly
to the ~poxy-vitreous carbon substrate.
Lead tabs were
burned onto the lead back plate and a layer of epoxy coating applied
over the exposed lead surface.
The overall dimensions of these bipolar
half-plates were 2.25 in. x 2.44 in. (5.72 cm x 6.20 cm) and the rib width
was 0.025 in. (0.064 em).
The epoxy-vitreous carbon substrate (without
lead backing) weighed 4.0 g. A 0.125-in. (0.32-cm) wide edge was proVided
for sealing which left an active material area of ~5 in.2 (32 cm2).
The epoxy-vitreous carbon substrates as
first developed had very high
conductivity (resistivity <0.2 ~ cm) but attempts to form positive active
material on these structures were unsuccessful.
The problem was traced
to corrosion at the interface between the substrate and the lead backing
plate. It was found that the substrate contained pinholes or large pores
which allowed sulfuric acid to contact the interface where corrosion
occurred during formation of the active material.
The corrosion product
led to high resistance at the interface and caused buckling of the plate.
The problem was eliminated by changing the material mixing procedure
and reducing the vitreous carbon loading to 60 wt %.
A small increase
in ~esistivity occurred with reduction in carbon loading.
The relation
between carbon loading and resistivity is shown in Figur~ 17 for a CIBA
1139/972 substrate. At 60 wt % loading. a resistivity of 0.85 Q cm was
found which results in an iR loss of 13 mV for a 0.020-in. (O.05l-cm)
thick substrate at a current density of 0.31 A/cm2. No further develop-
ment of the basic conductive plastic was pursued.
-114- '
-------
0.8
O.6~
~
u
c:
..
~ 0.4
:>
l-
V)
V)
w
~
0.2
0.0
50
Figure 17.
60 70 80
VITREOUS CARBON CONTENT, WT %
90
Resistivity ~. Vitreous Carbon Content of
a CIBA 1139/972 Substrate
-115-
-------
4.1. 3
Electrode Development
Substrates for positive electrodes which were to be of the Plante
type were fabricated in the same manner as described except that a 0.010-
in. (0.025-cm) thick lead sheet was bonded to the front surface of the
substrate during molding in place of the molded grid pattern.
The O.OlO-in.
(0.025-cm) thick lead sheet was then cleaned and Plante formed in the usual
manner. Several sets of Plante-type positive electrodes were produced.
All of these electrodes were Plante formed to produce the active positive
material, Pb02' in an identical manner except for the formation time.
A flooded cell consistir.g of a bipolar half-plate (electrode) and either'
another bipolar half-plate or a standard Pb 660 plate of opposite polarity
was used to evaluate the performance of the electrode.
The separator was
microporous polyvinyl chloride with an overall thickness of 0.050-in.
(0.13-cm) and 0.030-in. (O.08-cm) high ribs (Porvic-l, Electric Power
Storage Limited, Essex, England).
Electrolyte was 1.260 sp gr sulfuric acid.
Results are listed in Table 38.
The poor performance results listed for
Plate PLI were attributed to insufficient active positive material.
A
significant capacity increase was achieved by increasing the Plante forma-
tion time of the next two plates.
During the testing of Plate PL3, it was observed that increasing
charge rate and
concentration from 35% H2S04 (1.260 sp gr) to 42% (1.320
capacity decrease of 13% at the 2.5-A(0.08-A/cm2) dis-
26% at the 5.0-A (0.16-A/cm2) discharge rate. Charging
the electrolyte
sp gr) caused a
at a
constant voltage with a maximum current limit of 2.5 A,
. . 2
compared to a charging at constant current of 2.5 A (0.08 A/em ), resulted
in discharge times to 1.5 V at 5.0 A of 70 and 30 see, respectively.
2.71-V
The improved performance of Plate PL4 is attributed to charging at a
constant voltage of 2.71 V with a 4.5-A (0.14'-A/cm2) limit. The longest
discharge time to 1.5 V at 5 A was 2.0 min and occurred at Cycle 42. The
discharge rate was then changed to 10' A (0.31 A/cm2), which resulted in the
maximum discharge time (31 see) occurring at Cycle 50.
Discharge time or
capacity then began to decrease (Cycle 100 -- 20 see, Cycle 250 -- 4 see,
Cycle 440 -- <1 see).
. -116-
-------
Table 38. Formation and Performance of Plante-Type Positive Plates Formed
on Conductive Epoxy-Vitreous Carbon Substrates
Formation Increase in Dischar~e to 1.5 V
Plate Time. hr Thickness. cm Current Density. A/cm 2 Time. min
PL1 (Pure Pb) 20 0.016. 22.1
0.031 8.7
0.077 1.5
0.16 0.5
PL3 (Pure Pb) 46 0.020 0.071 2.4
0.14 1.2
I PL4 (Pure Pb) 46 0.020 0.14 2.0
t-'
1-'
'..,J 0.28 0.5
I
PL5 and PL6 (Sb) 69d 0.030
. PL7 and PL8 (Li) 50d 0.025
Total Nwnber
of Cycles
14a
141b
440c
a Charged at constant current; capacity decreased rapidly with cycling.
at end of test.
b Charging at constant current yielded lower capacity than charging at constant voltage; higher sulfuric
acid' electrolyte concentration decreased capacity. Active material loose and shedding at end of test.
c Charged at constant voltage of 2.71 V; capacity decreased with cycling. Active material loose and
shedding at end of test.
Active material was adherent
d
Active material loose and shedding at the end of formation.
-------
Further attempts to increase the performance of Plante positives
by using lead alloys (to produce a more adherent active material) and
longer formation times (to increase the quantity of active material)
proved unsuccessful.
It is believed that Plates PL3 and PL4 are close
to the ultiw4te that can De achieved by this method.
The Plante forma-
tion process is a delicately balanced empirical procedure which does not
permit a great amount of variation.
The only known modification which might be attempted to improve the
adherence and performance would be to deeply groove the lead sheet to enhance
the surface area. This has never been done with lead sheets as thin as those
being used in this project and great difficulty is expected. Even with 0.025-
cm thick lead sheets a bipolar plate of the Plante type would weigh 26.6 g;
that is 30% greater than a bipolar plate with 0.05l-cm thick layers of
pasted active material. The development of bipolar electrodes utilizing
Plante formation was terminated and the effort concentrated on electrodes
utilizing pasted active material.
Pasted positive plates were prepared by plating a thin film of lead
on the grid structure of t3e conductive epoxy surface. The plates were
then pasted with oxide, cured and electrochemically formed in sulfuric
acid to produce the active Pb02 material. . Paste formulas and fabrication
data are summarized in Table 39. The pasted positives initially fabricated
could not be completely formed and exhibited poor performance character-
istics, particularly at high rates as shown by the data in Table 40.
electrodes were tested using the same type cell described previously.
Variations in oxides, curing conditions and forming conditions all failed
The
to significantly improve the situation.
The difficulty was finally traced
to pinholes in the substrate material which permitted sulfuric acid to
penetrate to the lead backplate, forming a high resistance interface by
corrosion.
Decreasing the carbon content of the conductive epoxy eliminated
this problem.
Plate PA9 with 73 wt % vitreous carbon in the substrate
exhibited greatly improved high rate capacity ~nd longer life.
Comparison
of the performance of Plates PA9 and PAlO indicates that cycling at lower
rates to "condition" the electrode may be helpful.
In an effort to
further improve the performance and life of the positive electrodes, a
-118-
-------
Table 39. fabrication of Pasted-Plate Positives with Ep0xy-Vitreous Carbon Substrates
Vitreous Carbon Lead Layer Paste Formula Cure Time,
Plate Content , wt % Thickness, cm Number/Oxide hr
PAl 80 a PF-66/SPI02 89b
o . 0013
PA2 80 None PF-66/SPI02 8gb
PA3 80 O.0013a PF-66/SPI02 61b
PAl. 80 O. 0013a PF-35-A/SPl03 60b
PAS 80 O. 0013a K2S208 + PbO Sb
PA6 80 0.0013a
PA7 & PA8 80 O. 0013d PF-66/SPI02 72e
I 0.005ld 72e
f-' PA9 73 PF-66/SPI02
1-'
'" 0.005ld 72e
I PAlO 73 PF-66/SP102
PAll & PAl2 60 0.0025d Special leady oxide 16£
a rib height, 0.038 cm~
Substrate
b In air at room temperature and humidity.
c negative converted chemically into positive.
Formed
d
Substrate rib height, 0.051 cm.
e Wrapped in plastic.
£ At 100% relative humidity and 60°C.
-------
Table 40. Formation and Performance of Pasted-Plate Positives
with Epoxy-Vitreous Carbon Substrates
Formation 2 Discharge to 1.5 V Total Number
Plate Rate , mAl cm Current Density, A/cm 2 Time , min of Cycles
PAl 1. 7a, b 0.00062 490 10
.0.078 <0.05
PA2 1. 7a, c 0
PA3 1 _a,b 0.078 0.43 12
. I
PA4 0.6a,b 0.00062 465 7
I PAS' 0.6a,d 0.00062 351 9
t-'
N
a PA6 0
I
PA7 & PA8 7f,g 0.0078 0.17
PA9 7f 0.016 46 628h
0.031 20
0.078 5.2
0.16 1.5
PAlO 7f 0.031 0.25 l6i
PAll 7f 0.016 63 9j
PAl 2 7f 6086
a
Electrolyte, 1.060 sp gr.
b white
Poor formation, spots or layer.
-------
Table 40. Formation and Performance of Pasted-Plate Positives
with Epoxy-Vitreous Carbon Substrates (Continued)
c Poor formation. X-ray diffraction analysis results, a-Pb02' 10%;
. a-pbo2' 38%; PbO (tetr.), 6%; PbS04' 23%.
a .
Poor formation; additional formation in 1.260 sp gr sulfuric acid at
50 °c unsuccessful.
e Chemical conversion incomplete.
I
:...
1'-'
1--'
I
f Electrolyte, 1.080 sp gr.
g Appeared well formed.
h Charged at a constant voltage of 2.55 V (0.16 A/cm2).
600 cycles at 0.16 A/cm2; discharge time ~1.5 min for
then decreasing to <1 sec. Discharge time for Cycle
was 11 min.
Discharged through
first 20 cycles, 2
628 at 0.031 A/em
j. Same charging mode as for Plate PA9. All discharges at 0.031 A/cm2; dis-
charge time for Cycle 16 down to 6.6 min.
j Charged at constant-voltage of 2.70 V for 2 hr. Test discontinued after
9 cycles (0.53 A hr, 39.8% of theoretical capacity) because of a short.
-------
special leady oxide was used to fabricate Plates PAll and PA12.
A
significant increase in capacity was observed for Plate PAll;
However,
testing was discontinuec after a short developed during the ninth recharge.
The performance hiscory of Plate PA12 is given in Table 41.
All the
performance testing was carried out at room temperature, and it is believed
that no significant temperature increase occurred due to the testing.
It
should be noted that the discharge times [except at a current density of
1.5 A/in.2 (0.23 A/cm2)] actually increased after 45 deep discharges. The end-
.2. 2
of-discharge voltage in the "B" cycles [discharge at 1.0 A/in. (0.16 A/cm )
for 25 see] began at 1.75 V and then decreased with cycling. It appears that
a fl111 recharge \I1as not possible in 50 sec, even though. charging at constant
voltage was used.
After about 850 continuous "B" cycles, during which the
end-of-discharge voltage had deteriorated to 1.0 V in <25 see, another 100
"B" cycles wi th a constant end-of-discharge voltage of 1. 65 V was achieved
by increasing the recharge .time to 75 sec.
"A"-cycle performance was better; after 2800 cycles of discharge'at
a constant current density of 0.16 A/in.2 (0.025 A/cm2) for 10.8 sec, and
charge at a constant current density of 0.08 A/in.2 (0.012 A/cm2) for 22
'see, the end-of-discharge voltage was 1.89 V. However, during the next
2000 cycles the end-of-discharge voltage had deteriorated to 1.0 V in
<10.8 sec of discharge. After a boost charge, the end-of-discharge volt-
age of the "A" cycle returned to a higher value (1.95 V).. After this
treatment, the cell was cycled with a 0.25-A/in.2 (0.039~A/cm2) discharge and
a 2.7-V constant-voltage charge for 2 hr. A capacity of about one-half that
found initially in the same cycle mode was achieved.
Upon disassembling
the cell, it was noted that the active positive material was soft and porous,
i.e., swelling had occurred and there was evidence of some material shedding.
Initially the plate contained 9.5 g of active material, 1.0 g of lead
plating and 2.0 g of' conductive substrate assignable to the positive bi-
polar component. The i75 value (75-sec rate to 1.5 V) artd the average cell
- 2 2..
voltage, V, were estimated to be 1.4 A/in. (0.22 A/cm ) and 1.6. respec-
tively.
The negative pasted bipolar electrodes were made with automotive leady
oxide and paste formula PF-73.
The electrodes were cured in air or wrapped
-122-
-------
Table 41.
Performance Data for Bipolar Positive Electrode PA12
Discharge Time, sec
Cycle Current Density, A/cm 2 1. 75 V 1.50 V 1.00 V
to to to
1 0.31a '\.1 36 62
6 0.23 27 64 78
2 0.16 74 124 135
3 0.088 255 286 297
7 0.022 1165 1198 1244
8-52 0.039
53 0.039 1282 1365 1397
54 0.31 0 13 65
55 0.23 62 127
I
:..... 58 0.16 43 216
N 157
w
I 59 -0.088 297 474.
442
60-110 . Bb >1.75 V at end of discharge
-
. 111-896 B Performance deteriorated to 1.0 V in
<25 see of discharge
897-910 Various low-rate cycles ---------------------------------
911-1011 Modified BC End of 25-sec discharge constant at
1.65 V
1012 Ad 2.05 V at end of discharge
1013-2947 A ---------------------------------
2948 A 1. 95 V at end of discharge
2949-3849 A ---------------------------------
3848 A 1. 89 V at end of discharge
-------
Table 41.
Performance Data for Bipolar Positive Electrode PA12 (Continued)
. Cycle Current Density, A/cm 2
3850-5850 A
.5851 Low-rate boost charge
6064 A
6065-6086 lJ.039
to L 75 V
Dischar2e Time, sec
to 1. 50 V
to . 1. 00 V
Performance deteriorated to 1.0 V in
<10.8 see of discharge
---------------------------------
1.95 V at end of discharge
---------------------------------720
a
Charged at 2.70 V for 2 hr.
I
......
N
~
I
b
"~" Cycle: 2 2 .
Discharged at 0.16 A/cm for 25 see, charged at 2.71 V (0.11 A/cm maximum) for 50 sec.
c Modified "B""";Cyele:
Charge time increased to 75 sec.
d "A"-Cycle: .
. Discharged at 0.025 A/cm2 for 10.8 sec, charged at 0.012 A/cm2 for 22 sec.
-------
in plastic. Electrolyte used during formation was 1.060 or 1.080 sp gr sul-
furic acid and f~rmation rates varied from I to 7 mA/cm2. All of the nega-
tive electrodes. produced had a very uniform appearance.
Single-electrode
measurements of assembled test cells indicated that the cells were limited
by the positive bipolar electrode.
No signficant changes were observed
after the negative electrodes were cycled.
The capacity to total discharge
of a negative electrode prepared by pasting ~4.5 g active material onto a
substrate with O.038-cm high ribs was 0.375 A hr at the 20-hr rate, approx-
imately 56% of theoretical capacity.
A wide latitude is permissible in the
processing of negative electrodes and no performance problems with the nega-
tive electrodes are expected.
4.2
ELECTRODE AND BATTERY EVALUATION
4.2.1
Fabrication
Bipolar'half-plates (electrodes) and batteries were fabricated in order
to determine their performance figures of merit on both the components and
test cell levels as was done with the conventional (prismatic) plates and
batteries. Two negative electrodes (N3 and N4) and two positive electrodes
(P4 and P6) were fabricated for single-electrode tests.
The positive elec-
trades for single-electrode testing were produced in the 'same manner as Plates
PAll and PAIZ (SeGtion 4.1.3) and used the same special leady oxide.
After
formation, the positive plates were washed and air dried. The negative
electrodes used the same substrate material (conductive substrate composed
of CIBA 7072 epoxy resin and CIBA 972 hardener with 60 wt % vitreous carbon)
as the positiv~ electrodes. The negative paste was automotive leady oxide
and paste formulQ PF-73. Negatives were formed under the same conditions as
the positive electrodes. After formation the negatives were washed, the
water removed in boiling Stoddard solvent, and then air dried. Each electrode
contained 4.0 g of conductive substrate (of which half is assignable for
bipolar electrode considerations), a 0.030'-in. (0.076-cm) thick pure lead
backplate and a tab. The conductive substrate was plated with a O.OOI-in.
(0.002S-cm) thick layer of lead which weighed 1.0 g. The backplate and lower
portion of the tab of the positive electrodes were lacquer-coated to prevent
oxidation during testing.
Table 42.
Weights of the single electrodes are listed in
-].25-
-------
Table 42. Single Bipolar Electrode Weights
Weights, g
Electrode
Active Material
Bipolar Portion
N3
4.9 7.9
4.8 7.8
7.4 10.4
7.7 10.7
N4
P4
P6
-126-
-------
A test cell was constructed by sealing the edges of a positive and
negative electrode, separated by a Teflon (TFE) spacer, 0.060-in. (O.lS-cm)
thick, with epo~!. The epoxy used was CIBA 6010 resin with TETA hardener
with Thixin added in sufficient quantity to produce a non-flowing mixture.
The electrodes were the same as described above.
The PJrvic-l separator
(described in Section 4.1.3) was inserted after the epoxy had cured and the
Teflon spacer had been removed. The positive bipolar portion of the cell
weighed 10.6 g and the negative portion 8.0 g.
Two tWo-cell batteries were fabricated in essentially the same manner
except the active material on the electrodes was unformed.
The end elec-
trodes were the same as the lead-backed electrodes described previously. The
center electrodes of the batteries were true bipolar elements composed only
of conductive vitreous carbon-epoxy substrate with active material pasted
on both sides. The electrodes were separated by Teflon spacers and the edges
were sealed in the same manner as the cell.
Porvic-l separators were inserted
after the epoxy had cured and the Teflon spacers had been removed.
A lead
",
wire was sealed in place with epoxy to contact the center bipolar plates
for voltage readings.
listed in Table 43.
Weights of the electrodes in the batteries are
A photograph of the two-cell bipolar battery is shown in Figure 18.
4.2.2
Tests and Results
The bipolar half-plates (electrodes) were assembled in "flooded" cell
configuration using a Porvic-l separator and Pb 660 plates of opposite
polarity as auxiliary electrodes. The cell was filled with 1.260 sp gr
sulfuric acid and tests conducted at 51°C (125 OF). The test sequence was
similar to that used for the components-level plate test (Section 3.2).
results of the test are listed in Table 44.
The
For the negative electrode, the i75 value was >0:33 A/cm2 for both
electrodes tested. The figure of merit for specific power was 2.30 Wig.
This may be compared to a best value of 1.90 Wig for the pasted plates de-
scribed in Section 3.2. The target performance was 60 sec of discharge at
0.31 A/cm2 with a cell voltage greater ,than 1.5 V; the bipolar negative
electrodes yielded 90 sec at that rate, 50% better than the target value.
-127-
-------
Table 43. Type II (Bipolar) Battery Electrode Heights
1
Weight, g
Electrode Paste a Bipolar Portion
End negative 5.1 8.1
End positive 6.9 9.9
Bipolar element 9.9 14.8
End negative 5.0 8.0
End positive 8.5 10.5
Bipolar element 9.3 15.6
Battery
2
a Paste weight includes both negative and positive pastes on the
bipolar elements.
-128-
-------
~~~'~
~~
"
~~
Figure 18.
Two-Cell Bipolar Battery. Extra lead for
monitoring center bipolar plate.
-129-
-------
Table 44. Performance Test Data on Bipolar Test Plates
Discharge to 1. 5 V i7S' (fm)sp
. a 2 2
Electrode Resistance, mQ Current, A Current Density, A/cm Time, min A/cm IT/'
,. g
P4 >100
P6 25.5b 15 0.47 0.34 0.26 1. 26
10 0.31 1.0
5 0.16 2.6
2.5 0.08 5.6
I
....... N3 20 15 0.47 0.7 0.36 2.30
w
0
I 10 0.31 1.6
5 0.16 4.1
2.5 0.08 12.7
N4 21 15 0.47 0.64 0.33 2.30
10 0.31 1.4
.5 0.16 3.8
2.5 0.08 12.0
a
At 1 kHz.
b 50 ~ on standing overnight at open circui t.
Increased to
-------
The positive electrode yielded 60 sec at 0.31 A/cm2 discharge with a
. cell voltage >1.5 V. This also meets the target value. The i75 value for
the positive electrode was 0.26 A/cm2 and it had a figure of merit
of 1.26 Wig.
This r.lay be compared wi th the bes t figure of meri t 0 fl. 77 \\ / g
for the pasted plates described in Section 3.2.
It was noted that the active
material became soft and sbme shedding occurred and that electrode resistance
increased.
The test cell described in Section 4,2.1 could not be evaluated
because of high internal resistance (150 mQ). The high internal resistance
was probably due to a high contact resistance at the interface between
active material and the substrate.
Difficulties were encountered when attempts were made to form the
electrodes in the two-cell bipolar batteries.
One cell in each of the
batteries would not accept a charge and a possible short in the unformed
cell was suspected.
In an effort to repair the short and prevent a recur-
rance of the problem, the separators were removed and the battery was rinsed
in water and Stoddard solution at 105°C and dried.
New separators were
then inserted ar.~ sealed with epoxy at the bottom of the cells to prevent
bridging of the sediment,
Even though this procedure eliminated the short
problem, the unformed cell could not be completely formed.
I t is believed
that sulfation of the paste had occurred while the active material was im-
mersed in the forming acid during the initial attempts to form the plates.
4.3
SUMMARY AND RECOMMENDATIONS
The following conclusions can be reached:
.
Conductive plastic substrates can be fabricated using vitreous
carbon and various epoxy systems which will be chemically inert
in a lead-acid cell environment. These substrates have a re-
sistivity of ~1 D cm and a density of ~l.4 g/cm3 and can be
fabricated in a thickness of ~0.025 in. (0.064 cm).
.
A sufficient quantity of active material cannot be formed by
known Plante methods to meet the performance requirements of
2 A/in.2 (0.31 A/cm2) for 60 sec. The difficulty is funda-
mental because shedding of the active material occurs after
some point and efforts to produce a more adherent structure by
alloying the lead base material failed.
-131--
-------
a
Perfarmance .of negative bipalar-type plates exceeded .our target
.of 2 A/in.2 (0.21 A/cm2) far 60 sec. Figure .of merit far the
pasitive bipalar-type plates was 30% lawer than far the best
canventianal pasitive plate tested. Cycle life .of the best
bipalar pasitive plate was 6000 cycles including 60 deep
discharges ta 1.0 V. This is believed ta be abaut equivalent
ta what wau1d ~p. expected fram a gaad canventianal pasitive
plate. Mast bipa1ar pasitive plates had shart lives due ta
saftening .of the active material.
In view .of these canclusians,it is recammended that:
.
A variety .of thermaplastic materials be evaluated as alter-
nates far epaxy in a canductive plastic substrate. The .ob-
jective wauld be ta reduce manufacturing time and ta fabricate
a thinner substrate with greater strength. Candidate materials
are chlarinated palyvinyl chlaride, chlarinated palyethers, paly-
vinylidene fluaride, palyphenylene .oxide and ather therma- .
plastics. .
.
Develapment should continue in the area.of p.ositi ve .oxides,
pasting and f.ormati.on techniques, and adhesi.on .of the active
material to the substrate.
.
Meth.ods of sealirig bip.olar plates into a cell ~nd of sealing
the cells .of a battery tagether should be devel.oped further. .
Candidate methads are a) use .of rubber gaskets and campressing
between end plates, and b) sealing .of bipolar substrates t.o
plastic frames fallawed by sealing .of frames t.o each .other by
ultras.onic, solvent .or thermal welding techniques.
;-132-
-------
5.
PROJECTIONS AND RECOMMENDATIONS
5.1
PRISMATIC'CELL3
5.1.1
Duty Cycle £nd DesiKn Requirements
Several considerations impact the performance specifications for a
hybrid-vehicle battery. They are vehicle acceleration and cruise perfor-
,mance, vehicle weight and voltmle, and battery durability. Many of the bat-
tery requirements can be generated from the design criteria developed by EPA
and stated in ''Vehicle Design Goals - Six Passenger Automobile," Revision
B, Advanced Automotive Power Systems Development Division, Environmental
Protection Agency, 11 February 1971.
The following discussion is based on
those goals and is augmented by information generated during a parallel TRW
effort, "Cost and Emission Studies of Heat Engine/Battery Hybrid Family Car,"
Contract No. 68-04-0058, Environmental Protection Agency.
5.1.1.1 Battery Power Requirements
.
Peak Power
Figure 19 shows the road horsepower and cruise road horsepower for a
4,OOO-pound vehicle as a function of vehicle velocity.
These curves were
generated from performanc~ goals described in Paragraph 8 of the EPA design
goals doctmlent.
In order to translate the vehicle performance goals into power levels
at the battery terminals, it was necessary to adjust the loads in Figure 19
for the effects of power train inefficiencies between battery and road.
Using data ge~erated in the Cost and Emission Studies, it was possible to
construct vehicle velocity-battery power-time relationships for various
hybrid configurations, engine power settings and battery characteristics.
Figure 20 shows a case in which the vehicle velocity varies as a function
of time according to the revised EPA battery specifications (detailed in
Section 2.1).
The particular vehicle modeled in this case was a series-
connected hybrid, with a single gear ratio between the motor and wheels.
The car has the capability of meeting or exceeding all the vehicle design
goals and has a top speed (engine power only) of close to 90 mph which is
reached in 25 sec after the start of the acceleration.
The battery power-
time relationships for specific velocity-time schedules, or the reverse,
. are small functions of the specific configuration chosen; however, the
values of Figure 20
are quite represen.t8:tive of hybrid-vehicle demands.
-133-
-------
I
......
W
.I:-
.1
. 125
Q.,
::I: 100
~
o
Z
«
~
w
o
0::
W
~
o
Q.,
-I
W
w
:I:
~
150
MERGI NG TRAFFIC
75
~~~Q
~S ~
>(-.,0 ~~Q
(,
~o1o
50
30% GRADE CLIMB
25
\J\S~
S) (,~
~O~ .
\.. ~ '\j ~ \..
o
o
80
90
50
60
70
30
40
10
20
VEHICLE SPEED, MPH
Figure 19.
Wheel Power Demands for a 4,OOO-lb Car
-------
:r:
~ 60
..
>-
~
~ 50
-.J
W
>
i
I-'
W
\J;
I
~4O
u
:r:
w
> 30
90
80
.70
92.5 KW, 200 V
20
10
23 KW, 239 V
o
o
5
10
Figure 20.
o KW, 240 V
39 KW, 224 V
63.5 KW
212 V
71.5 KW, 210 V
92.5 KW, 200 V
- ~ 70 K'lv, 290 V
-90 KW, 269 V
-40 KW, 253 V
15 20
TIME, SEC
25
30
35
EPA Battery Power Profile for a 4,000-lb Series Hybrid Vehicle
-------
The revised EPA battery specifications identify a 39-kW recharge
rate for 90 sec.
In reviewing the modes of power train operation, it is
difficult to develop a ~ircumstance in which this situation
,
could arise.
For example, Figure 20 shows that a panic stop from 90 mph in 10 sec could
produce charge power levels at the battery in excess of 200 kW, but for
periods of only a few seconds.
Extended hill descents could last for 90
sec but the 39-kW power level for charging is difficult to construe.
Average Power
Analyses conductei during the Cost and Emission Studies contract con-
tributed to an understanding of hybrid power-train components performance
on a representative driving cycle.
Average battery power and current are
9.4 Z 0.6 kW and 3.9 Z 0.2 A, respectively, during charge; and 11.3 Z 2 kW
and 50 Z 10 A, respectively, during discharge. The averages are defined
in terms of the average of all similar events; that is, the 9.4-kW charge
power means the average power during all charging events is 9.4 kW.
A histogram for the battery charge and discharge currents during a
typical driving mission is shown in Figure 21.
During .approximately 70%
of the vehicle operating time the battery is being charged.
The mos t prob-
able battery current is approximately 50 A charge. Extremely high charge
or discharge rates are not experienced. Such data are useful in constructing
battery test cycles to simulate actual operating experiences.
While the ratio of charge to discharge time in the revised EPA speci-
fication is in agreement with the histogram ratio experienced in hybrid-
vehicle operation, the Lime spent in high-rate charging seems unrealistic.
There is some question as to the need for a full recharge after every dis-
charge.
For example, it is not clear that each 70.5-kW discharge should be
associated with a 39-kW recharge which restores the battery to its original
energy state.
Rather, TRW's experience in operating its. electromechanical
transmission system is that the higher power level charge events occur in
a quasi-tapered charge manner.
With reference to Figure 20,
the charging
level during deceleration rapidly decreases as the vehicle slows down.
Fur-
ther time correlations are necessary to completely define the discharge-
charge cycles which would represent actual battery usage.
-136-
-------
-;R
0
..
I-
Z 40
w
~
~
....,
U
U
u.
u 30
w
Q,.
VI
I-
<
w
:E
i I-
I~ 20
w 0
.....!
I Z
>
~
0
10
50
--
~
- ... CHARGE DISCHARGE .
I-
-
I I I I ,- I I
300
200
200
400
300
100
o 100
BATTERY CURRENT, A
Figure 21.
Typical Histogram of Battery Current during a Driving Cycle
-------
5.1.1.2
Battery Weight and Volume
Paragraph 2 of the EPA design goals specifies a maximum power-train
weight of 1,600 lb and ~ desirable weight of 1,300 lb, the latter consis-
tent with current automotive practice. Analyses of the series and parallel
hybrid systems suggest power-train weights of 1,090 and 840 lb, respec-
ti vely, wi thout battery, Table 45 compares the battery weights and peak
specific powers for series and parallel systems for maximum and conventional
power-train weights. It appears that the battery specific power require-
ments for the 1,600-lb power train can be met for both systems, although
the series system may be marginal.
Both systems, especially the series, will
find it difficult to meet the lower power-train weight of a conventional
vehicle without substar.tial improvement in battery specific power.
Due to the high density of lead-acid batteries no particular volume
problems are anticipated. Furthermore, the ability to modularize the
"
battery package into small distributed packages will aid in vehicle design
flexibi 11 ty.
5.1. 2
Projected Characteristics of Cell and Full-Size Battery
. .
On the basis of results for test cells, it is estimated that a cell
of identical design except with 17 plates instead of 21 plates would meet
the revised EPA power and capacity requirements.
Weight of a six-cell
battery in a custom-designed polypropylene container would be about.
23.5 lb (10.7 kg) and the container would be about 8-in. long ,x 5.5-in.
wide x 6.75-in. high (20.3 cmx 14.0 em x 17.1 em);'
The full-size battery would consist of 22 six-cell units connected in
series to give a nominal 264 V. Weight and volume would be 525 lb (239 kg)
and 5.5 ft3 (0.16 m3), r~spectively.Life of the battery is expected to
be three to five years.
5.1. 3
Availabili ty and Development.
It is estimated that the cost of. this battery to the original equip-
ment manufacturer at a production rate of 100 batteries/day (2,200 l2-V
units) would be approximately $440. Production could begin in 1973.
basis.
A full-size battery could be produced in 1972 on a developmental
The biggest problem would be acquisition of polypropylene containers,
-138-
-------
Table 4.'5.
Battery-Weight Constraints for Hybrid-Vehicle
Systems (92.5-kW Peak Power Demand)
Power Train Battery Specific Power,
Hybrid System Weight, 1L Weight, lb W/lb
Series 1,600 510 180
Parallel 1,600 760 120
Series 1,300 210 435
Parallel 1,300 460 200
-139-
-------
for \",hich a mold would have to be designed and fabricated.
An existing
I
container could be used, as was done in the case of th~ test cells, with
considerable sacrifice in space" utilization and with a weight penalt~ of
perhaps 10/';.
5.2
HIPOLAK BATTERY
Results achieved in the present program have demonstr.ated that' a light-
wei~ht conductive substrate can be pasted and formed to produce a bipolar
~latc having current capability of 2 A/in.2 (0.31 Alcm2) for one minute to
a 1.5-V per cell cutoff. Discounting this by one-third to 1.4A/in.2 (0.22 AI
')
em"") to allow for losses normally encountered in scale-up, leads to anesti-
mate of a total weight of 430 lb (195 kg) and a total volume of 3.3 ft3 for a
full-size battery system.
The battery would consist of six modules, 6.6 in. x
7.8 in. x 15.5 in. (16.8 em x 19.8 em x 39.4 em) which contain 134 cells each.
The entire module would be served by addition of water to. a single trough
serving all cells.
Other details are the following:
Bipolar Plate Size
.
fIctive ,i"lateria1
6.2 in. y 6.2 in. x 0.020 in.
(15.7 em y 15.7 em x 0.051 cm)
.
Substrate
6.4 in. x 6.4 in. x 0.025 in.
(16.3 em x 16.3 em x 0.064 cm)
.
Plastic Frame
6.6 in. x 7.8 in. x 0.115 in.
(16.8 em x 19.8 cm x 0.292 em)
Contains reservoir, separator and vent.
268-V Module
.
Size
6.6 in. x 7.8 in. x 15.5 iri.
.
\-lei ght
(16.8 crn x 19.8 cm x
3 J
"800 in. (0.013 m )
" .
70 Ib (32 kg)
39.4 crn)
.
Volume
-140-
"I
,
-------
Complete Batt~ry
Prototype Projected
. Nodules 6 4
. 75-Sec Cur-rent 320 A 304 A
. Average Voltage 220 V 227 V
. 75-Sec Energy 7000 kJ 6900 kJ
. Volume 0.093 m3 0.062 m3
. Weigh t 195 kg 136 kg
We estimate t~at a prototype battery of this type could be built
in 1973, with production in 1975.
Over a 10:1ger term it is expected that current capab.ility can be
increased to 2 A/in.2 resulting in a 33% reduction in size, to four
modules instead of six. This leads to the projected battery, weighing
3 3
300lb (136 kg) and having a volume of 2.2 ft (0.062 m). It is
estimated that a prototype battery with these characteristics could be
built in 1975 or earlier.
5.3
FUTURE WORK
5.3.1
Program Plan
It is recommended that development of both conventional and bipolar
lead-acid batteries be continued.
Work on conventional batteries should
be directed toward a better understanding of the causes of failure of
lead-acid batteries in hybrid-vehicle duty and of the life that can be
obtained.
Limited optimization studies of the components of a conventional
lead-acid cell have resulted in the fabrication and testing of cells that
will produce an average power of 150 W/lb for 75 sec.
This is almost
twice the perfoLrnance of conventional cells.
While we do not believe we
can continue to ilnprove performance at this rate we do believe that addi-
tional optimization can result in a battery producing 200 W/lb for 75 see
without seriously changing curreht ll~nufacturing methods or impacting
the cost.
The most serious question really concerns the lifetime of these
high specific ~ower batteries.
Therefore we propose to perform life
-141-
-------
studies at the components level to obtain the best compromise between
lifetime and performance.
The failures that have been observed so far in the program have all
been caused by softening of the positive active material.
This is -
greatly affected by the manner in which charging is accomplished, and
especially by incomplete charging, which fails to re-cement the active
material and leads to softening. The very stringent cycling conditions
that have been used are probably useful as an accelerated test for com-
paring various cell designs but they hardly permit estimation of
battery life to be expected in a hybrid vehicle.
Some life tests should
be carried out under cY2ling conditions more representative of hybrid
vehicle.. duty, with more shallow cycles, which will allow more time for
complete chargeback after an occasional deep, full-power discharge.
We also believe that the longer term development of bipolar
electrodes eventually will result in a long life lead-acid battery,
comDatible with hybrid-vehicle demands, which will produce 300w2tts
per pound.
Thus our p~oposed program is divided into three parallel'
tasks, Tasks 1, 2 and 4 plus a design task, Task 3:
Task 1
Task 2
Optimization Studies of Conventional Cells
Life Studies of Conventional Cells
Task 3
Task 4
Design and Performance Projections
Development of Bipolar Battery
A detailed design will be made of a lead-acid battery capable of
meeting the revised EPA requirements for a full size hybrid heat-engine/
electric automobile.
5.3.2
Hork Statement
TASK 1.
OPTIMIZE CONVENTIONAL CELLS
1.1
Screen Cell Components
The highly successful program of optimization of the cell components
instituted in the current program will be continued.
-142-
-------
The effect of separator composition, thickness and design on perform
ance will be tested along with increased pellet size (fewer structural
members in the grid), use of inert binders .in the pellets and the ne\y
grid design (4.25 in. x 4.25 in.) developed under the current program.
Theperforrnance test will be the same as used in the current program and
the same figure of merit,
(fm)
sp
=
i75V/W
where i75 is the c~rrent required to discharge a cell to 1.5 V in 75
seconds, V is th~ average cell voltage for this discharge, and W is the
plate weight, will be used for evaluation.
The statistically designed performance test will consider information
obtained during the current program.
For example, interactions will be
assumed negligible and use will be made of the estimate of experimental
variation obtained to minimize the total number of tests necessary to
achieve the test objectives.
use of a Latin Square Design.
Preliminary considerations indicate the
As in the current program, tests will be
performed in sequenU.al steps to provide monitoring points so that the
effectiveness of the test can be assessed as it proceeds.
Separators to be used in this study will be of various designs and
will be acquired. by purchase or by manufacture. These separators will be
of various thicknesses no larger than about 0.020 in. and will be of
various materials selected for low resistance and/or long battery life.
Candidate materials are polyvinyl chloride and polypropylene with and
without ribs.
One or more separators may contain glass mats to aid in
retention of active material.
1.2
Perform Coreputer-Assisted Modeling of Cells
A computer model developed under the current program for the plates
will be extended to include the whole cell.
The model of the plate will
be refined to include the effect of pellet resistance, electrolyte resis-
tance, state of charge, separator resistance and the number and size of
plates.
.-143-
-------
TASK 2.
LIFE-TEST CONVENTIONAL COMPONENTS N~D CELLS
2.1
Perform Cycle-Life Study of First Generation Test Cells
Cycle-life test of a test cell started under the current program
(first generation test cell) will be continued.
The test program will be
revised to include a recharge cycle and to match more closely the discharge
cycle described by EPA.
The test cell will be cycled 200,000 times or
until failure whichever cames first.
A cell will be considered to have
failed when the cell voltag2 drops to 1.5 V.
A Group Size 22F automotive
SLI battery will be submitted to an identical test for comparison.
Although no catastrophic failure is anticipated, there is possibility
of minimal degradation.
As an example of data analysis, the end-of-
discharge voltage can b~ plotted as a function of the number of cycles.
The
data will be plotted for each cell type and if a degradation trend is noted,
the data will be subjected to a standard regression analysis to determine
the curve which best describes the trend.
This curve will be used to
predict the number of cycles to an end-of-.discharge voltage of 1.5 V,. If
the voltage reaches this v:=tlue, the cell is consldered to be in a failed
condition. The magnitude of the coefficient of variation,
o lx, between cells obtained from the specific power performance tests in
w
the current program can be assumed to be the magnitude of variation
L,~t\.:eer. cells on the cycle axis so that a probability distribution of
failure can be ascertained, if degradation is detected.
2.2
Perform Cycle-Life Tests on Components
Plates and separators which have been subjected to the performance
test (Task 1.1) will be cycled using a form of an accelerated life test.
Each component will be tested for the same number of cycles after which
its performance will again be determined as described in Task 1.1
Difference in perform~nce will be used as a criteria for measuring degra-
dation as a function of cycling.
2.3
Perform Accelerated Corrosion Tests on Selected Gtid Alloys
Four alloys will be prepared and cast into grids and test rods,
Grids will be of the special design developed in the current program and
performance tested in optimized cells. Grids will be pasted and tested in
sp gr 1.260 sulfuric eciJ electrolyte.
Corrosion. rate will be accelerated
-144-
-------
by maintaining tre test cell at elevated temperature and by maintaining
the test specimen at a potential of 0.075 V relative to the equilibrium
potential of the alloy in the electrolyte.
Plate dimensions will be
measured periodically in situ by means of a cathetometer to determine rate
and amount of growth.
Plate capacity will be measured periodically and
the test will be discontinued when capacity has decreased to 50% of its
initial value. Tests will be carried out at 160, 140, and 120 of. Life
to 50% capacity will be expressed in terms of equivalent time at 75 and
125 of. Test rods will be examined meta1lographical1y to determine
corrosion pattern.
The following alloys will be tested:
1-
2.
3.
4.
0.013% Li-Pb
2.5% Sb-4.S% As-Pb
0.07% Ag-l.5% Sb-l.6% Cd-Pb
4.5% Sb.-J!b
TASK 3.
PROJECT PERFORMANCE AND DESIGN FULL SCALE BATTERY
3.1
Fabricate and Test Optimized Cells
The results of Tasks 1 and 2 will be used to design second generation
test cells.
These cells will be fabricated and tested using a test proce-
dure similar to the one used in the current program.
The emphasis will be
placed on perfcrmance data although accelerated life tests will be
considered.
3.2
Design Full-Scale Battery System
Based on the above approved design requirements and the results of
the cell tests performed in Tasks 1, 2 and 3.1, the characteristics of a
full scale battery system for a hybrid automobile will be projected with
emphasis on weight and cost reduction consistent with specified life
requirement. A preliminary design of a full scale battery system for a
hybrid heat-engine/electric vehicle shall be performed. Full considera-
tion shall be given to operational and maintainability requirements as
well as cost and size reduction.
-.145-
-------
TASK 4.
DEVELOP LIGHTWEIGHT BIPOLAR BATTERY
Methods of forming bipolar electrod~s using conductive plastic sub-
strates \-1il1 be investigated. Both pasting and direct formation of ac-
tive materials will be studied.
Methods of incorporating bipolar electrodes into a battery will be
investigated. Performance objectives are a maximum discharge current
density of2 A/in. 2 sustained for one minute to 1.5 V per cell cutoff
and recha~ge capability suitable for a hybrid power train. Target specifi~
power is 300.W/lb and target life in a hybrid power train is five years.
Electrodes will have at leastlOO-cm2 active area.
Initially, epoxy-glassy carbon substrates previously developed
will be used. Methods of sealing bipolar electrodes into a battery will
be studied.
Candidate methods are 1) use of rubber gaskets and compressing
the electrodes between endplates, and 2) sealing bipolar electrodes to
plastic frames followed cy sealing the frames to each other.
design will permit easy servicing.
Battery
Alternate conductive substrates offering lower cost and greater
st~ength will be investigated.
Candidate substitutes for epoxy in the
substrates include chlorinated polyethers, chlorinated polyvinyl chloride,
polyvinylidene fluoride a~d polyphenylene oxides.
-146-
-------
APPENDIX A
~~THEMATICAL MODELING-PASTE ELEMENT
The basic assumptions used in this discussion are:
1.
Current flowing from electrolyte to paste element moves in the
z-direction and is uniformly distributed with respect to x and
y.
2.
The paste element occuplies the region 0 ~ x < a, 0 ~ y ~ b of
the x-y plane'- and is "thin" in the z-direction.
Its conducti-
vity, a, is small compared to that of its boundary which is taken
to be an equipotential, at electrostatic potential $ = O. How-
ever, a is large compared with the conductivity of the electro-
lyte.
.The starting point for the analysis is the divergence theorem for
. +
the current desnity, J, in the pellet. This current moves in the x-y plane
(1. e., it has no z"component). The theorem states that
J Jon dS
=
J V.j dV
(1)
in which fi is. the unit normal to the surface element of area dS, and dV is
a volume element in the paste.
The two integrals in Equation (1) are readily
evaluated.
The i~tegral on the left is the total current flowing out of the
volume element of the paste.
This is of course equal to the total current
entering the volcme element from the electrolyte.
If the area of the element
which faces the ele~trolyte is dA, and the current density in the electro-
lyte is J,
then.
J+~
Jon dS
= J dA
(2)
Since, by assumption, J is independent of x and y, and since the plate is
+
thin, VoJ is constant, and therefore
r ,
J Vo} dV = 8 dA VoJ
(3)
~l47-
-------
where e is the thickness of the plate.
Combining Equation (1). (2) and (3)
-+
I7.J = J/e.
(4)
The relation between j. the electric intensity, E, and electrostatic po-
tential, 4>, is
-+. -+
J == aE = -al7cj1.
(5)
Combining Equation (5) with Equation (4),
2 .
v 4> = -J/ea.
(6)
This is Poisson's equation for the electrostatic potential, and is to be
solved with the condition 4> = ° on the boundaries x = O,a, and y = O,b.
The solution, which can be effected by, for instance, the method of separa-
tion of variables, is
""
4>(x.y) "= k I2:
sin (mx1T/a) sin (nY1T/b)
2 2 2 2
mn(m /a + n /b )
(7)
m,n=1
m,n oJd
where
4
k = l6J/7r ea
(8)
The average value, 4>, of 4> is
a b
~ = (1/ab) f dx f dy 4>(x,y) = 4ka2 II
° °
1/ [m2n2 (m2+ n2t2;] (9)
m,n odd
where t = a/b. By symmetry, it is necessary to tabulate the sum in Equation
(9) in the range ° -=:. t -=:. 1 only. The double sum is. reduced to a single sum
by means of the identity
11/[m2(m2 + u2)]= (1T14) [7r/2} - (1/u3) tanh 7Tu/2]
m odd
(10)
"":148-
-------
Combining Equation (9) and (10), with u = nt,
$ = (kaZI,)[,5/19ZtZ - (1/t3) ~ (1/n5) tanh 'ntIZ].
n odd
(11)
In deriving Equatioll (11), use has been made of the identity
~ 4 4
L l/n = 7T /96.
n odd
In terms of the average potential, $, defined by Equation (9), the
equivalent series resistance of the pellet may be defined as
R = 4>/abJ'
(12)
that is, the series resistance is equal to the ratio of average potential
to total current, &bJ.
The current leaving the pellet at its boundaries is next considered.
The current per unit length through the edge y = 0 is
Jo(x) ~ ea (d$/ay)y = 0
which becomes, on using Equations (7) and (8),
(13)
Jo(x)
=
2 3 ~ . 22 22
16a J/7T b L [s1n (mx7T/a)]/[m (m + n t )].
(14)
m,n
odd
The total current through the boundary y = 0 is the integral of Equation
(14) :
I
Y
fa Jo(x) dx = 32a3J/7T4b L 1/[m2(m2 + n2t2)].
o
(15)
m,n
odd
The total current~ Ix' through the boundary at x = 0 can be obtained from
Equation (15) an& the conservation equation,
2(1 + 1 ) = Jab
x y
(16)
~149-
-------
By the use of Equation (10), 1 may be written as a single sum:
y
1y = (abJ/2) [ 1 - 16/,\ L
-------
The double sum in Equation (21) is the same as that in Equation (15).
The
result is, thertiore,
v ~ labJR/4) [1 - 16/.3t
~ (1/n3) tanh .nt/Z].
(22)
n odd
The sums whi~h appear in Equation (11), (18) and (22) were evaluated
on the Hewlett-Packard 9100B calculator, for a variety of values of t in
the range 0 < t < 1. Parabolic fits to these results were used in the
plate-resistance computer program.
-151-
-------
APPENDIX B
SAMPLE CALCULATIONS FOR ANALYSIS OF VARIANCE
The Mean Square is equal to the Sum of Squares divided by the degrees
of freedom asso~i&ted with the source of variation (see Tables 20 and 27).
Using data for Runs 1 through 4 of Table 18. the Sums of Squares were
computed as follows:
a.
Total Sum of Squares (TSS)
TSS = Sum of Squares of Individuals (SSI) - Correction Factor (C)
= I x2 - (I x) 2/n
- 39.82 + 36.82 + ... + 39.92 + 37.02 - 22186.10
: 22254.84 - 22l86~10
=: 68.74
Note:
x denotes i75V,
b.
Sum of Squares for Runs (SSR)
SSR = [( 2: x) 12 + (2: x) 22 + (2:~) 32 + (I x) ~ 2] /4 - C
= (147.52 - 147.52 + 150.12 + 150.72)/4 - 22186.10
= 2.25
Note:
subscripts denote Run numbers.
c.
Sum of Squares for Plates (SSP)
SSP = [("2 x) 12 + (2: x) 22 + (2: x) 3 2 + (2: x) 42] /4 - C
= (154.92 + 148.02 + 151.22 + 141.72)/4 - 22186.10
= 23.48
Note:
subscripts denote Plate numbers.
d.
Sum of Squares for Permutation Order (S80)
SSO :;: [( 2: x) 1 2 + (2: x) 22 + (2: x) 32 + (2: x) 42] /4 - C
= {145.52 + 156.22 + 149.22 + 144.92)/4 - 22186.10
= 20.23
Note:
subscripts denote Permutation Order number (Section 3.2.4.1).
-153-
-------
e.
Sum of Squares for Experimental Variation (SSE)
SSE = TSS - SSR - SSP - SSO
= 68.74 - 2.25 - 23.48 - 20.23
= 22.78
Since no significance was detected, all data ( that is, all five runs)
. were used to obtain the averages given in Table 19.
Using data for Te5t Run 2 of Table 23, the Sums of Squares given in
Table 27 were computed as follows:
a.
Total Sum of Squares (TSS)
TSS = SSI - C
= 2 x2 ( L x) 2/n
= 1.452 + 1.102 + ... + 0.992 + 1.012 - 22.144083
= 23.185653 - 22.144083
= 1. 041570
Note:
x denotes (fm) .
. sp
Sum of Squares for Thickness (SST)
SST = [(2x)12 + (Lx)/]/8 - C
= (10.5002 + 8.3232)/8- 22.144083
= 0.296208
b.
Note:
subscripts 1 and 2 denote thicknesses Tl and T2' respec-
tive1y.
c.
Sum of Squares for Temperature (SSt)
SSt = [( L x)/ + (L X)/]/8 - C
= (8.2932 + 10~5302)j8 - 22.144083
= 0.312760
Note:
subscripts 1 and 2 denote temperatures t1and t2' respec-
tively.
d.
Sum of Squares for Grid Type (SSG)
SSG = [(Lx)/ + (Lx)/J/8 - C
-154-
-------
= (9,9762 + 8.8472)/8 - 22.144083
= 0.079665
Note:
subscripts 1 and 2 denote grid types G1 and GZ' respectively.
Sum of Squares for Paste Density (SSp).
e.
SSP = [(2x)/ + (~x)22]/8 - C
= (10.4362 + 8.3872)/8 - 22.144083
= 0.262400
Note:
subscripts 1 and 2 denote paste densities PI and P3' respec-
ti'lely.
f.
Sum of Squares for Experimental Variation (SSE)
SSE = rss - SST - SSt - SSP - SSG
= 1.041570 - 0.296208 - 0.312760 - 0.262400 - 0.079665
= 0.090537
'.
-155-
I~
.
------- |