You are on page 1of 9

Copy of Paper Presented to:

CD

11tsLsI: 94

Vancouver, Canada .

October 30 - November 2,1994

Analysis and Interpretation of Conductance Measurements used to Assess the State-OfHealth of Valve Regulated Lead Acid Batteries

Part III: Analytical Techniques

by:

Mark J. Hlavac Midtronics, Inc.

Dr. David O. Feder

Electrochemical Energy Storage Systems Inc. Madison, New Jersey, USA

MIDTRONICS

Midtronics, Inc.

8230 S. Madison Street Burr Ridge, Illinois 60521 U.S.A.

Tel: (630) 323-2800 Fax: (630) 323-2844

Midtronics b. v. Noord IJsseldijk 24 3402 PH - IJsselstein The Netherlands

Tel: +31 306 873 100 Fax: +31 306 889 015

ANALYSIS AND INTERPRETATION OF CONDUCTANCE MEASUREMENTS USED TO ASSESS THE STATE-OF-HEALTH OF VALVE REGULATED LEAD ACID BATTERIES.

Part III: Analytical Techniques.

David O. Feder, Electrochemical Energy Storage Systems, Inc.

Mark J. Hlavac, Midtronics Inc.

Abstract:

This paper presents data showing conductance aging of new cells and capacity/conductance aging of older cells. It presents data showing the comparative accuracy of several techniques for selection of a cell conductance value suitable for use in field determination of cell pass/fail conditions based on either capacity failures below 80% or below 50% of rated value.

The accuracy of the various selected conductance values are determined by actual comparison to cell by cell capacity/conductance data. This data represents 192 cells of 1000 AH (Ampere Hour) size in 48 volt telecom use and 96 cells of 600 AH size in 24 volt strings in another type of telecom usage.

The data presented should allow the user to select a technique which appears most suitable to his needs and capabilities.

Introduction/Background:

During the last several years, the authors have presented the results of a series of studies showing the relationship between measured VRLA (Valve Regulated Lead Acid) cell capacity and measured cell conductance. These papers included capacity/conductance data on several thousand cells and monoblocs, ranging from 25 to 1000 ampere-hours in capacity and representing the products of five different battery manufacturers 2,3,4.

These results have documented the close correlation between measured cell capacity and measured cell conductance. More recent publications have demonstrated the initial application of techniques utilizing these conductance values to detect individual cell capacity failures (less than 80% of rating), with an accuracy generally exceeding ninety percent 4,6,8,9.

Those studies have all enjoyed the benefit of actual capacity test data obtained in conjunction with conductance values

and hence may be considered as the first step in confirming the utility of the conductance test in detecting low capacity VRLAcells.

For the user to utilize this technology effectively in the maintenance and surveillance ofthe condition of his battery plant, he must have available some type of "signature" information for the specific type of cell involved in order to determine the appropriate conductance value corresponding to the capacity failure criteria (usually 80%) at which some remedial action must be taken.

In this paper, the authors present several techniques for obtaining these "signature" conductance values, explain their derivation and their Significance, demonstrate the procedures involved in their utilization and present and analyze typical predictive results obtained.

1. Conductance and Discharge Capacity Trending

Ideally, the starting point for any decision on the utilization of conductance to determine possible degradation of VRLA (Valve Regulated Lead Acid Battery) cells performance would be the availability of published conductance values for each cell type, when new, as provided by the manufacturer. Unfortunately, this type of data is just beginning to be determined by several manufacturers and is not generally available, nor have industry standards been established as to sample size, use of average, maximum or minimum values, standard deviation, etc. Recognizing that different VRLA processing techniques (Tank vs. Jar Formation, Calibrated Fill vs. Fill and Dump) may all affect initial conductance values of new cells, studies are appropriate for the determination of a standard "settling in or burn in" time (i.e ... 3, 6, 9 months in service) in order to establish a reference or starting base value, from which conductance degradation can be followed throughout cell life and appropriate end of life failure criteria established.

A very desirable way to maintain cells/batteries is to monitor the decrease in conductance and capacity with time as the cells/batteries age. To date we have been able to obtain

VRLA 1000 AH Cells, Conductance/Capacity Trend Analysis @ 4.5 Year arid 6.0 Year Test Interval.

+ lII.urtd ConduCllnctlDilchlrgt ClpllclIy@4.SVII/I

• .....Itd ConductlncllDilchlrg. ClpllClty@6.0V ....

y= 80.29 + 2.56. R squared = 0.811

'ii'

c CD

Iii CD

iii ~



CD U C II

ti

::J '0 C o U



I~~~ __ ~~~ __ ~~ __ ~ __ -J

o 10 20 30 40 50 60 70 80 90 100 110 120

% Discherge Capacity

Figure 1

Conductlnce/Capaclty Trondlng Data For lGOAH. 6 Volt lIonanblocl.

Readlngl 12 lIonthl Aplrt 1 Hr Aat. to 5.75 VPB.

1400

a;

C CD

E

• 1200

~

Regression for this data:

V = 43<1.7 + 13.9>< A squared = 0.954

1300

1100

1000

• U

C

~

::J '0 C o U

900

Regression from previous publication "'-'0' the enlire population 1-9 ye.r monoblocs:

V= 381.6 + 15.ax R squared = 0.889

800

700

• Measured Conductance/Capacity @ 7 V ..... + Measured Conductance/Capacily @ 8 Years.

Discharge Minutes

Figure 2

VRLA 750 AH, Tracking of New and 14 Month Cell Conductance.

i'
c

E 3.0

0
~ ~9

0 ~8
:
:I
II ~7

0 ~6
c

..
0 ~5
~
11 0
C
0 .... 0 .. ·
I) New (G) 141.tJn1h (G)

Minimum ~520 VOO

Maxinum ~830 ~960

Range 0.310 0.260

Mean ~611 tm

Variance 0.007 0.006

SID DEV. 0.082 0.181

SID ERROR. 0.024 0.023

, , , ,

" ,.,p

... 0"'0

0 .. ·0 .. ·0 .. ·0 .. ·0'··... :

'0"'0·· ..

4 10 12 14

New Ceil Conductance ValJes ---r- Conductance VaIles@ 14 Months

Cell Numbers

Fi ure3

only a limited amount of data suitable for trend analysis. Fi~ ure 1 shows a combined capaclty/conductance correlatio plot tor a 48 volt string of 1000 AH VRLA cells measure, after 4.5 years service and 18 months later. Note that al though both capacity and conductance have decreased, botl data sets fall on the same correlation plot, with the later dati merely lower on the line. Overall, the cells have suffered at average capacity loss of 27%, with nine cells losing mon than 40% of rated capacity, while only four cells showed nc loss after the 18 months of additional service. Likewise, thE conductance loss averages 22% ranging from 5% to 29%

Another set of trending data has been published by Britist Telecom 7 and is referenced in Figure 2. Capacity and con ductance measurements were made 12 months apart on i year old monoblocs. Here the data shows the same charac teristic, i.e .... both sets of data fall on the same regressior curve, regardless of the initial discharge capacity. Again here we see a Significant capacity loss averaging 28% of ratec capacity or 36% of the 7 year capacity during this 12 montt interval. It is significant that this data falls almost exactly or: the capacity/conductance regression line previously estaoIished for this entire population ranging from one to nine years in age8.

Field Trending Of New Cell Conductance Values.

In this section, descriptive statistics are used to show the general trend of conductance values on new product as installed after several weeks on float and after 14 months in service. Data from one manufacturer is shown in Figure 3 . This data shows an overall upward shift or increase in the range of conductance and a 6% increase in the mean conductance. One might speculate the increase of the conductance might be caused from the formation of previously unformed active material, improved distribution of electrolyte or increased element compression and grid/active material contact from formation of the Pb021ayer on the positive grids over the 14 months in service, or some combination of all of these. Tear down and material analysis would be necessary to quantify these possibilities. As data continues to be gathered for these cells a determination of the peak and eventual drop-off of conductance values with respect to time will be evaluated. Conductance data accompanied with periodic discharge capacity testing will be performed to better understand when these cells reach peak capacity/conductance level and then begin their descent in service.

While the trend appears to be significant the limited data available do not answer the many questions posed earlier which would be needed to understand the applicability of "New Cell" conductance values. However, it is reasonable to hope that battery manufacturers will now recognize the usefulness of this kind of information and users will begin to measure conductance values of newly installed cells and then follow changes in conductance (capacity) through out the life of the cell. The final approval and publication of IEEE (Institute of Electrical and Electronics Engineers) draft standards 1188 and 1187, hopefully by early 1995, should

provide the catalyst for the accumulation of this type of information.

2. Use of conductance values derived from correlation plots and regression analysis.

However, for most users, the problem they face is directly related not to newly manufactured products, but to batteries which have been in service for several years. For these cells no new cell reference conductance values were available and if the cells are greater than three years old, data on product currently being manufactured is probably different due to design and process changes over the last several years. The extraordinarily successful joint efforts between telecommunication users and Midtronics, Inc. which established the validity of the conductance/capacity relationship, has resulted in a massive data base of capacity/conductance values9, 10. This data can now be used to test the validity of analytical techniques which may be used to predict cell capacity, or more importantly, potential cell failures from measured conductance values.

In one technique, which has been utilized in previous publications 11, actual individual cell capacity data usually derived from individual battery string discharges are plotted vs. conductance values measured before the discharge. The "best fit", or linear regression line is then calculated, and provides a weighted average line which most accurately reflects the overall trend of the data. The accuracy of this fit is identified by the correlation coefficient (R2=1.0 indicates perfect fit). An example of a previously published plot11 for one string of cells of a 48 volt telecom battery is shown in Figure 4. Here discharge time to an end of discharge voltage is plotted vs. conductance, measured in "Cell KMhos", with a correlation coefficient of 0.93. To establish the conductance value equivalent to 80 % capacity (150' =100% capacity to 1.80 VPC at 263 amperes; 80% = 0.8 x 150' or 120') a horizontal line is drawn at 120 minutes until it intersects the regression line. The conductance value at the intersection reflects the "best value" of conductance for 80% capacity for that string. Dividing the plot into four quadrants, as in Figure 5, allows use of the "box score" technique described in previous publications6,8,9,11 to determine the overall accuracy of using this conductance value to predict cells which passed or failed the 80% capacity end of life criteria. The results for this string are shown in Figure 6 and indicate an overall accuracy of 100%, 100% accuracy in detecting failed cells and 100% accuracy in identification of good cells. This technique has been applied to seven additional strings of 48 volt batteries in both this and a second telecom plant which uses the same type of cells. The overall results, on a string by string basis are shown in Table 1 for discharges to 1.80 volts per cell. If instead of plotting discharge time vs. conductance in KMhos, one were to plot percent capacity vs. percent conductance, the same plot now appears as shown in Figure 7. The percent conductance values are determined by their respective intersection with the regression line and the resulting box score for 80% capacity shown in Figure 8 is, as expected, identical to Figure 6. Note that 80% capacity equates to 89% conductance while

Discharge Capacity va Conductance
VRLA 1000 AH, 24 cen.,
String .5 Data,
263 Amps To 1,80 Volts Per Cen
200
U y = • 138,32 + 102,30. R'2 c 0,835
II.
>
0
II! 150
~ •
~


~ 100
e
i

!!'

.t:.
u 50

C
0
I
Conductance KMhos (KSlemena) Figure 4

Discharge Clpaclty va Conductance
VRLA 1000 AH, 24 Clna,
String 15 Dlta,
200 263 Amp. To 1,80 Volta Per Cell
c.i y • • 131,)2 + 102.3Ox R·Z • 0.135
A.
> (A)
0 (e)
"! 150
~ aD% Capacity 1120 Minutes)
0
l-

! 100 •
::1 (8)
e
i •

.. 5D

.t:. II



s

"Box Score" Analysis of String #5 24 Cells.

Conductance Test Outcome

Bad Good

Figure 5

o~--~~·--~~------~~~

1 2

Conductance KMhos (KSlemens)



CI)

E

~

o

~

.~ Bad ~

c. ctI U

Good

Correct

......- --, -, Assessment

Overall:

24

- = 100% 24

Correct Assessment of Failures:

20 =100%

20 F' 6

Igure,

o

4

20

o

Plrclnt Dllchlrgl Clpully v.. Plrcant Conductancl
VRLA 1000 AH. 24 CIIII.
U 100 263 Amp. to 1.80 VOItI Per CIII
Q.
> •
90 Y = . 91.716 + 1.9320x R'2 = 0.935
0
II! 80
..
0 70
..
~ 60
'il
II
Co 50
II
U
40
II
III 30
..
II
.c
u
.,
Q
C •
II 0
U 50 60 70 80 90 100 110
..
II
Q.
Percent Conductance Figure 7 "Box Score" Analysis of String #5 24 Cells.

Using Conductance and Capacity Percentages

Conductance Test Outcome

Bad Good

Correct

,-----r------. Assessment Overall:

24

- =100% 24

Correct Assessment of Failures:

20

=100%

o

4

20

Figure 8

20

o

Correct

r-----,..------, Assessment Overall:

21 -=88% 24

Correct Assessment of Failures:

14

- =100%

14 Figure 9

"Box Score" Analysis of String #5 24 Cells.

Using 80% Conductance (63% Capacity)

Conductance Test Outcome

Bad Good

Q) E

o Good

S

o

~

~ Bad '0

10

C.

10

o

3

7

14

o

80% conductance intersects the regression line at 63% capacity. Figure 9 shows the box score using 80% conductance (63% capacity) as the selection criterion.

While these figures show the overall accuracy of the technique, it would require the end users to perform both conductance and discharge capacity tests on 100% of their batteries in order to obtain the results shown.

3. Use of conductance values derived from correlation plots and regression analysis of a single pilot string .

Fortunately there is sufficient similarity in the capacity/conductance relationships for a given cell design to allow the 80% capacity/conductance regression value derived from a single string discharge to be applied to other strings of similar cells in similar telecommunication usages. Using conductance equal to 2.55 KMhos, the conductance value derived from 80% capacity in string #5 (Figure 5) and applying it as a "typical" or "pilot" value to the other seven strings previously tested, results in the data shown in Table 2, with accuracy in detecting failed cells ranging from 82% to 100%.

While the "pilot" string regression intersection/box score technique appears to combine both a technical approach and acceptable accuracy in detecting failed cells, itdoes require a minimum of at least one single string discharge with significant spread in conductance/capacity values for each particular cell type. This effort may be impractical for some users.

4. Simple Analytical Method:

A Simple technique which requires no discharges, measures the conductance of all cells in a battery string, assumes that the cell or monobloc with the highest conductance is equal to 100% capacity. This assumes that 80% of this conductance value equals 80% capacity and uses this "80% conductance" value to determine pass/fail conditions. This technique has been used for the same eight strings with the results shown in Table 3. As might be expected, these results are more variable than with the regression procedure, but may be acceptable for many users and allow them to detect potential catastrophic failures. The users experience with this technique has proven extremely effective particularly when conductance on all cells within a battery string indicate a wide cell to cell variation after "burn in" of 25-50% or more. This wide range of conductance and capacity within a Single string has been reported more often than not in the U.K., Canada and United States7,8,9.

Another variation of the above technique using 80% of the 100% conductance value derived from capacity/conductance regression analysis is shown in Table 4 (this typically correlates to a regression value of 60-70% capacity). Here the results are similar to those obtained by using 80% of the highest conductance per string (Table 3), but the technique requires at least one set of capacity/conductance data in order to obtain the necessary regression information.

Table #1- VLRA 1000 AH Cells, Conductance Prediction Accuracy Using Pass/Fail Conductance From 80% Capacity/Regression Intersection.

)ffice String G (KMhosJ@ Overall Assessment Percent Percent
\lumber Number Intersection of Accuracy of Bad Bad Missed Good Missed
80% Capacity Percent Accuracy
and Regression Percent
1 2 2.51 79 89 11 50
1 5 2.55 100 100 0 0
1 7 2.60 100 100 0 0
3 4 2.40 100 86 14 0
3 5 2.60 100 100 0 0
3 6 2.90 100 100 0 0
3 8 2.60 87 100 0 21
3 9 2.50 83 75 25 13 Discharges to 1.80 volts per cell.

Table 1

Discharges to 1.80 volts per cell.

Table #2- VLRA 1000 AH Cells, Conductance Prediction Accuracy Using "Pilot" Conductance 2.55 KMhos From 80% Capacity/Regression of String #5/0ffice #1.

Pttice String "Pilot" G (KMhos) Overall Assessment Percent Percent
~umber Number Office #1 Accuracy of Bad Bad Missed Good Missed
String #5 Percent Accuracy
Percent
1 2 2.55 75 89 II 67
1 5 2.55 100 100 a a
1 7 2.55 100 100 0 0
3 4 2.55 87 87 13 a
3 5 2,55 95 94 6 0
3 6 2.55 83 82 18 0
3 8 2.55 87 100 a 21
3 9 2.55 75 88 13 31 Table 2

Table #3 - VLRA 1000 AH Cells, Conductance Prediction Accuracy For 80% Capacity Pass/Fail Using Highest Conductance in String.

Office String Highest 0.8 x Highest Overall Assess- Percent Percent
Number Number Conductance Conductance Accuracy ment Bad Missed Good
(KMhos) In In String Percent of Bad Missed
String Accuracy
Percent
1 2 2.98 2.38 92 89 11 0
1 5 2.88 2.30 88 85 15 0
1 7 3.14 2.51 100 100 0 0
3 4 2.90 2.32 79 67 33 0
3 5 2.92 2.34 79 72 28 0
3 6 3.40 2.72 96 95 5 0
3 8 3.16 2.53 83 90 10 21
3 9 3.49 2.79 63 100 0 56 Discharges to 1.80 volts per cell.

Table 3

Discharqes to 1.80 volts per cell.

Table 4

Table #4- VLRA 1000 AH Cells, Conductance Prediction Accuracy For 80% Pass/Fail Using 80% of Regression Value of 100% Conductance.

Oflice String 100% 80%01100% Overall Assess- Percent Percent
Number Number Conductance Conductance Accuracy mentol Bad Good
Capacity Capacity Percent Bad Missed Missed
Intersection Intersection Percent
1 2 2.83 2.26 63 7R 22 0
1 5 2.62 2.26 68 85 15 0
1 7 2.96 2.37 100 lno 0 0
3 4 2.73 2.18 67 <17 53 0
::I 5 ·297 ?::I1l 7!l 7? 28 0
3 6 3.31 2.65 91 91 9 0
3 8 3.09 2.47 91 100 0 14
3 9 3.10 2.48 83 75 25 13 Discharges to 1.80 volts per cell.

Table 5

Table #5 - VLRA 1000 AH Cells, Conductance Prediction Accuracy For 80% Capacity Pass/Fail Using 70% of Highest Conductance in String.

Office String Highest 0.7 x Highest Overall Assess- Percent Percent
Number Number Conductance Conductance Accuracy ment Bad Missed Good
(KMhos) In In String Percent of Bad Missed
String Accuracy
Percent
1 2 2.98 2.09 56 47 53 0
1 5 2.68 2.02 54 45 55 0
1 7 3.14 2.20 !:If; 94 6 0
3 4 2.90 2.03 46 13 87 0
3 5 2.92 2.04 42 22 78 0
3 6 3.40 2.38 52 50 50 0
3 6 3.16 2.21 88 70 30 0
3 9 3.49 2.44 83 75 25 13 In what might seem to be a more conservative approach, instead of 80% conductance the user might select 70% of the highest conductance value in hopes of more accurately predicting cells which show less than 80% capacity failure. Results are shown in Table 5, and while generally accurate in predicting good cells, the technique appears to be a poor indicator of failed cells.

However, if one were to reduce the pass/fail requirements from 80% capacity to 50% capacity and utilize the 70% conductance pass/fail criteria, now the results, shown in Table 6, become more attractive in detecting cells with less than 50% capacity. While this appears attractive, limited trending data in section 1 of this paper suggest that a 50% capacity cell can drop to less than 20% capacity within only one year, which poses a real risk to the user's confidence in the reliability of his reserve battery plant.

In Table 7, the results of all of the above procedures are summarized and collected into a single table for comparison of their accuracy in detecting capacity failure.

In this table overall accuracy, ie: percent of good plus bad, percent of bad only, percent of bad cells missed and percent of good cells missed, are displayed for each of the techniques which have been discussed.

The string by string capacity/regression technique shows the greatest overall accuracy and in detecting failed cells. It is followed closely by the "pilaf' conductance technique. Both the techniques using 80% of the highest conductance per string and 80% of 100%

conductance show acceptable results to detect 80% capacity failures. In stark contrast, lowering the conductance criterion to 70% of the highest conductance per string shows poor overall results, fails to detect approximately 50% of the less than 80% capacity failures, but shows the greatest accuracy in detecting good cells. If the user can accept the possibility of only 50% of his engineered reserve, then the last technique utilizing 70% of the highest conductance per string to detect capacity failures below 50% provides acceptable results. Almost identical results have been obtained for these same techniques applied to these same eight strings but discharged to 1.75 volts per cell. Those results have been omitted due to space limitations.

Many of these same conductance value selection techniques have been utilized on eight 24 volt strings representing another cell size and a different usage. Results are shown in Table 8.

Table #6 - VLRA 1000 AH Cells, Conductance Prediction Accuracy For 50% Capacity Pass/Fail Using 70% of Highest Conductance in String.

Office String Highest 0.7 x.Hlghest Overall Assess- Percent Percent
Number Number Conductance Conductance Accuracy ment Bad Missed Good
(KMhos) in In String Percent of Bad Missed
String Accuracy
Percent
1 2 2.98 2.09 _96 89 11 0
1 5 2.88 2.02 83 73 27 8
1 7 3.14 2.20 100 100 0 0
3 4 2.90 2:03 96 67 33 0
3 5 2.92 2.04 83 67 33 14
3 6 3.40 2.38 96 92 8 0
3 8 3.16 2.21 75 67 33 24
3 9 3.49 2.44 67 100 0 33 Discharges to 1.80 volts per cell.

50% Capacity = 75 minutes to 1.80 volts per cell.

Table 6

Table #7 - VLRA 1000 AH Cells, Overall Conductance Prediction Accuracy Summary

Table Number Analytical Percent Overall Assessment Percent Percent
Number of Technique Capacity Accuracy of Bad Accu- Bad Good
Strings Pass/Fall Percent racy Percent Missed Missed
to 1.80 vpc
80% Capacityl
Table 1 8 Conductance 80 94 94 6 11
Intersection I
Regression
80% Capacityl
Table 2 8 "Pilot" Conduc- 80 88 93 8 15
tance
~ ..... 80% of Highest
Table 3 8 Conductance in 80 85 87 13 10
String
80% of 100%
Table 4 8 Regression 80 85 61 19 3
Value of Conduc-
tance
Table 5 8 70% of Highest 80 65 52 48 2
Conductance in
String
Table 6 8 70% of Highest SO 87 79 21 10
Conductance in
String Table 7

Table '8· VLRA 600 AH Cells, Overall Conduct.nee Prediction Accuracy tor 8·24 VoU String Discharged er tne 4 Hour Rate to 1.75 vens Per Cell.

Number Numbei'o' PustF.1I PasIF.11 Ow.,.11 ........ P~nl P .... m
. , 24 VOI1 CrUtf1a Cancluclanco Accu_y .... , B .. Goad
01'_ Slrtngs C8p8eHyl (K ...... ) ....... ...... Miss8>CI W, ....
Mtnl.ft'"s <:m_ Accuroc:y
"'"",m
'"PiIoI-80%
3 8 80"4/192 CapJRegAl8 82 9< 9 23
SIrings
1.9.
80% Highest
3 • 80%/192 ConducIonco 70 ., 58 '5
perStmg
IS. ·1.92
eO%of 100%
3 8 80%/192 ConducIo nee 75 O. 0 35
204
70% Highest
3 8 80041192 Canduaance 69 17 83 0
perSIMg
1.3.·1.7D
70% Highest I
3 8 50% 1120 Conductance 9S 100 0 ,
pel String
1.34· t.70 Table 8

Conclusions:

1. While trend analysis which follows conductance and percent capacity of new cells as they age is technically the most satisfactory, available data are too limited to allow definitive description of the aging process or to define the ideal conductance loss which will signal cell replacement.

2. Limited trend analysis of older cells reveals a surprisingly sharp decrease of capacity in 12 to 18 months, but indicates that the capacity loss is accurately reflected by conductance.

3. Several different procedures are presented and evaluated in an attempt to select the most effective technique for determining the conductance value which most accurately signals capacity failure.

a. The most accurate involves both conductance/capacity testing and regression analyses of significant numbers of aged battery strings and would not be practical for many users.

b. A technique requiring only one discharge test with prior conductance measurements appears to be of adequate accuracy and would be more generally usable.

c. The simplest technique, using 80% of the highest conductance per string is accurate and is within the capabilities of any user.

d. A similar technique using 70% of the highest conductance per string to detect cells with less than 50% of rated capacity combines accuracy both in detecting failed cells and in minimizing the number of good cells improperly characterized.

e. However, in applying the technique described above in (d), the user must consider whether his engineered reserve can tolerate the very steep capacity aging curve, i.e.: a 50% capacity cell can degrade to only 10-20% within a one year period.

Acknowledgment:

The authors would like to acknowledge contributions of tes data by Mr. D. Ogden, Ogden Power, Inc; Mr. B. Jones British Telecommunications and Mr. J. Urban, Battery Shop

References:

1. K. S. Champlin: "Dynamic Method for Storage Battery Diagnostic Testing" Talk presented to 1975 SAE Off-Highway Vehicle Meeting, Milwaukee, WI, Sept. 1975.

2. D.O. Feder; T.G. Croda; K.S. Champlin; S.J. McShane; M.J. Hlavac: "Conductance Testing Compared to Traditional Methods of Evaluating the Capacity of Valve-Regulated Lead Acid Batteries and Predicting State-of-Health". Journal of Power Sources, 40 1992. pp 235-250.

3. D.O. Feder; T.G. Croda; K.S. Champlin; M.J. Hlavac: "Field and Laboratory Studies to Assess the State-of-Health of Valve Regulated Lead Acid Batteries: Part Conductance/Capacity Correlation Studies." Proceedings of the 19921NTELEC Conference, pp 218-233.

4. M.J. Hlavac; D.O. Feder; D. Ogden: "Field Application of Conductance Measurements used to Ascertain Cell/Battery State-of-Health and inter-cell Connection Integrity in Electric Power Utility Applications". Proceedings of the American Power Conference 1993, Volume 1, pp 44-57.

5. M.J. Hlavac; S.J. McShane: "Conductance Testing of Standby Batteries in Signaling and Communications Applications for the Purpose of Evaluating Battery State-of-Health". Proceedings of the 1993 Association of American Railroads.

6. D.O. Feder; M.J. Hlavac; W. Koster: "Evaluating The State of Health of Lead Acid Flooded and Valve-Regulated Batteries: A Comparison of Conductance Testing vs. Traditional Methods". Proceedings of the 1993 International Conference on Lead-Acid Batteries, Varna, Bulgaria.

7. B. Jones, British Telecommunications "Conductance Monitoring Of Recombination Lead Acid Batteries" 1993 proceedings of the 11 th International Lead Conference, Venice Italy.

8.0.0 Feder; M.J. Hlavac; S.J. McShane: "Updated Status of Conductance/Capacity Correlation Studies To Determine the State of Health of Automotive and Standby Lead Acid Batteries:' Journal of Power Sources, 48, 1994 pp135-161

9. M. J. Hlavac; D.O. Feder; T. G. Croda; K.S. Champlin"Field and Laboratory Studies to Assess the State-of-Health of Valve-Regulated Lead Acid and Other Battery Technologies:

Using Conductance Testing: Part II: Furthe Conductance/Capacity Correlation Studies". Proceedings of the 1993 INTELEC Conference, Paris, Volume II, pp 375-383

10. D.O. Feder; "VRLA Cell Failures' The 60's Revisited":

Batteries International, Issue 20, July 1994, pp44 to 48.

11. D.O. Feder; M.J. Hlavac; "Setting Conductance Cut-off Values," Batteries International, Issue 18, January 1994, pp60-61.

You might also like