Substation batteries play a vital role in overall reliability of a substation by providing dc power for protection, supervision and control of substation and line equipment. Frequently, transmission or distribution power will be lost or reduced in magnitude during system disturbances (faults) at precisely the same point in time that power is required to isolate the disturbance from the system. Power to control and supervise system components is necessary during any large-scale or system-wide loss of ac power in order to provide circuit-switching power during restoration operations. Continuous monitoring of equipment (and therefore, a continuous power source) is required to detect abnormal conditions.

To provide a continuous power source, most substation protection and control programs use dc power from batteries. These batteries can range in size from 50 Ahr at 24 V dc to more than 1000 Ahr at 250 V dc. These batteries are mostly lead-acid, although nickel cadmium (nicad) and other types may be used. The batteries must be tested regularly to ensure a constant and uninterruptible power source. Testing methods, therefore, are critical for optimum operational capability. Experience shows that reassessing and updating these methods is an essential part of any reliability control program.

For 50 years, Duquesne Light Co. (Pittsburgh, Pennsylvania, U.S.) has performed substation battery testing as part of its control program and has been attentive to establishing better methods of testing. In the past decade, however, several critical factors motivated significant changes in testing methods.

Duquesne had employed the “short circuit test,” which consisted of simply placing a load of approximately 100 A on the test battery for a few seconds and reading the voltage. A reading of less than 1.75 V meant the battery cell had reduced capacity and would not meet Duquesne's adequate reliability standards. As systems expanded, batteries increased in size to meet needs. Duquesne realized that the short-circuit test was becoming outdated. New testing methods would have to be implemented to keep up with system demands.

Table 2. Battery resistance criteria.
A-H size EXIDE micro-ohm resistance Model GNB micro-ohm resistance Model GLOBE micro-ohm resistance Model GOULD micro-ohm resistance Model AVG micro-ohm resistance * 70% limit resistance **
50 3333 2/3-DS/DSC-5 3333 5666
75 2381 2/3-DS/DSC-7 2381 4048
100 1953 3CC-5 2232 2/3-CQA/C-11 2/3-DS/DSC-9 2036 3461
125 1582 DC/DKR-11 1582 2689
150 1323 3CC-7 1351 DC/DKR-13 1337 2273
170 864 MAX/MCX-05 1190 2-MAX-170 1027 1746
175 1179 DC/DKR-15 1179 2004
190 954 MAT/MCT-05 1316 2-MAX-190 1157 1967
200 1008 3CC-9 1116 CQA/C-200 1062 1805
255 597 MAX/MCX-07 820 2-MAX-255 708 1204
285 668 MAT/MCT-07 916 MAX-285 792 1346
300 744 CQA/C-300 744 1265
320 729 EU-7 729 1239
340 451 MAX/MCX-09 622 MAX-340 536 911
380 535 MAT/MCT-09 735 MAX-380 635 1080
425 368 MAX/MCX-11 505 MAX-425 436 741
475 444 MAT/MCT-11 610 MAX-475 527 896
510 309 MAX/MCX-13 424 MAX-510 366 622
590 267 MAX/MCX-15 267 454
600 330 NAX/NCX-09 389 NAX-600 359 610
672 370 NAT/NCT-09 435 NAX-672 402 683
705 340 EU-15 340 578
750 267 MAX/MCX-15 314 NAX-750 290 493
840 298 NAT/NCT-11 352 NAX-840 325 553
900 225 NAX/NCX-13 225 383
1008 250 NAT/NCT-13 250 425
1050 234 FHC-15 195 NAX/NCX-15 214 364
1200 173 NAX/NCX-17 173 294
1344 190 NAT/NCT-17 190 323
1350 157 NAX/NCX-19 157 267
1500 145 NAX/NCX-21 145 247
1650 132 NAX/NCX-23 132 224
1680 154 NAT/NCT-21 154 262
1800 122 NAX/NCX-25 122 207
* Values are from manufacturers published data or typical test values.
** Recommended replacement limit from EPRI study.
Note: Previous short-circuit limit micro-ohm resistance = 17,500.

Short-circuit testing was time consuming and inefficient because:

  • It required at least a half-day for the crew to complete.

  • It was susceptible to inaccuracies because of manual operation of the test, sometimes requiring retesting.

  • It required special safety precautions because of the long duration of high-current flow the test required.

Duquesne's experience also revealed that battery life varied tremendously depending on service conditions such as environment and type of charger used.

In addition, Duquesne realized that voltage testing and specific gravity checks do not supply a complete account of a battery's health. The possibility of a battery failure at a major substation showed existing short-circuit test methods were not providing the high degree of reliability system control programs require. The risk of equipment damage because of loss of protective mechanisms during system faults (high-current short circuits) and efforts to increase productivity and decrease the cost of maintaining 280 substation batteries forced Duquesne to seek simplified and more effective test procedures.

Selecting a Solution

Clearly, Duquesne needed a better way to assess the overall condition of these batteries. Replacing these batteries at the optimum time near the end of their life provided opportunities for cost savings and reliability improvements. Duquesne began a study of battery life. The study involved load testing various batteries removed from service to determine capacity randomly and at scheduled replacement time. IEEE-450 (vented lead acid) and IEEE-1188 (valve-regulated lead acid) recommend replacing batteries at 80% of their optimum capacity. Testing of batteries scheduled for replacement showed that at the time of removal, capacity was 10% to 52% of their optimum (Table 1).

Table 1. Battery load tests after replacement by short circuit method.
Battery Type Battery Amp-Hour Test: % Capacity
DCU-13 150 52.6%
ETA-15 560 40%
EOP-9 160 52%
KU-9 400 10%
KU-15 577 20%
3CC-7 150 44%

Subsequently, Duquesne initiated scheduled load (capacity) testing of batteries as recommended by the IEEE standard. Load testing involved placing an external test load on the batteries (usually a resistive load box) for up to eight hours to determine their capacity. Testing showed that this process could be costly and time consuming. In fact, this testing cost over the life of the battery could exceed the cost of smaller batteries. In addition, there was a risk that testing a battery in service could either deplete its capacity or collapse individual cells. The result could cause insufficient total capacity or loss of the entire battery, because of the drain on all of the remaining battery cells in the series. Duquesne devised a plan to install a portable battery during the load test. This alternative could be time consuming and costly.

Duquesne evaluated an alternative: resistance (ohmic) battery testing. This process involved measuring the conductance, resistance or impedance of a battery. IEEE-450 and IEEE-1188 recommend ohmic testing of batteries to detect poor internal/external connections or capacity. Recent studies, such as EPRI report number TR 108826, showed a relation between capacity and resistance/conductance/impedance of a battery. Duquesne performed in-house capacity testing and compared the results to resistance testing on the same cells (Fig. 1).

Based on the EPRI report and the results of its in-house battery testing program, Duquesne concluded that a sufficient relationship existed between resistance and battery capacity. This relationship justified using resistance testing as a trigger to at least perform a load test or, in some cases, replace the battery. Table 2 shows Duquesne's criteria for load test and replacement (Note: These values are a starting point and will be revised as more data are accumulated; the best value to use is a baseline established when the battery is new and capacity is known to be 100%).

Duquesne chose to use computerized resistance testing, which has the following advantages:

  • Testing time is substantially reduced, typically about 30 minutes to test a 60-cell battery.

  • The test is automated, minimizing errors in setup.

  • It is safer than the old short-circuit test. In Duquesne's testing method, a 70-A resistive load is applied for a fraction of a second compared with 100 A for several seconds during the short-circuit test (resulting in less I2T, which could cause plate or connection failure or possibly an explosion during the test).

  • Data are collected automatically in computerized form, eliminating significant recording errors.

  • Data can be immediately displayed or archived for later analysis and placed on a common server for general access.

Implementation

Duquesne purchased seven computerized test sets from Albercorp (Boca Raton, Florida, U.S.) for use by six crews, with one set used as a backup. The cost of these test sets was relatively low compared with other substation test equipment and provided immediate savings over previous methods.

Each member of the six test crews received eight hours of hands-on classroom training in the operation of the test set. They also took part in supplemental training in safety, battery characteristics and battery maintenance. In addition, each crew member received two to four hours of field training, testing actual substation batteries (Fig. 2).

Crew members received specific training that focused on the immediate recognition of poor or defective batteries so that they could act on the spot to correct the problem. Prompt action is critical in the test program to minimize the risk of loss of system protection mechanisms (Fig. 3).

The archiving of data is important for determining trends of changes in resistance in battery life. Recording data for future statistical use also is a vital maintenance function. Presently, Duquesne is working to develop displays that highlight any data that is out of tolerance with accepted norms. This is expected to save time in the periodic review process, currently slated at once per month.

First-Round Results

Duquesne completed the first round of resistance testing for all systems by September 2000. As a result, Duquesne decided to replace several batteries based on comparison of their results with its standard reference table based on the overall test results. The best results are obtained by using actual resistance test results on new batteries with 100% capacity for a baseline reference. For example, the four cells that are in the top right quadrant in Fig. 1 are both high in resistance and capacity. These four cells are anomalies. Possible interpretations are that these were replaced at some later date with different cells or that these four cells do not show correlation between resistance and capacity. The point is that a high degree of correlation exists between resistance and capacity for the majority (90%) of the cells.

Duquesne believes that a direct benefit of this testing program is a more reliable power supply. In some cases, mainly involving smaller ampere-hour batteries, Duquesne will replace a below-standard battery without retesting because additional capacity testing is not yet cost effective. However, in the case of larger ampere-hour batteries, the additional load testing is justified to verify the capacity before replacement.

Time savings, cost savings, reduction in paperwork and future reductions were estimated at US$18,800 annually.

James P. Kane received the BSEE degree from the University of Pittsburgh in 1982. He has worked at Duquesne Light Co. for the past 19 years, primarily in maintenance of substation batteries and power circuit breakers. Over the last 10 years, he held the position of senior engineer and was responsible for the substation battery maintenance program, which included the development of battery-resistance testing at Duquesne Light. Kane is a member of IEEE.

Martin E. Gazboda received the BSEE degree from the University of Pittsburgh in 1964. He has held various positions at Duquesne Light Co. over the past 36 years. His experience is mostly in substation maintenance, including positions as substation equipment engineer, protection engineer, director of technical services, and most recently, maintenance planner in the operations and asset management division. Gazboda is a member of IEEE and a registered professional engineer in Pennsylvania.