Economic studies involving the effects of load growth at Alabama Power had indicated the need to convert low-voltage distribution to higher voltages, where 12-kV systems were upgraded to 35 kV. Because the design of the 35-kV system is similar to the design of lower-voltage distribution, it was assumed that operating characteristics also would be similar.

After operating the system for almost 10 years, several problems occurred when switching underground cable, which used elbows on dead front equipment, resulting in faults. This problem was addressed with special “switchable transformers.” Another problem involved severe voltage sag on the transmission system and at the substation bus during a distribution feeder fault caused by the low impedance of 35-kV feeders.

The System

Alabama Power (Birmingham, Alabama, U.S.) uses 397 kcmil ACSR or 795 kcmil AAC feeder conductors for both 12- and 35-kV systems with loadings of 600 or 900 A on both systems. The impedance of these almost identical feeders is quite different, as shown by the per unit impedances for a given distance of circuit, where the 12-kV feeder has a per unit impedance that is 7.65 times greater than the 35-kV feeder of the same conductor. This characteristic causes more voltage drop for the 12-kV feeder during a fault, resulting in less voltage drop in the substation and transmission system.

A voltage divider calculation shows the voltage drop across the generation and transmission system is 11.13% of the total, the drop across the substation transformer is 71.10% and the feeder drop is 17.76%. Other feeders attached to the low-side substation bus will experience the same voltage sag even though the fault occurs under separate protective devices. The magnitude of the 35-kV voltage sag problem is reduced as the distance to the fault increases, but the problem is significant for faults occurring up to 7 km (4 miles) from the substation bus.

System impedance calculations take the following into consideration:

• The generating plant

• The 230-kV lines to a 230/115-kV substation

• The 115-kV transmission lines to each 115/35-kV substation

• The substation transformer impedance

• The distribution system feeder conductor impedance.

Although all three systems components' impedances are used in the calculations, the voltage drop of interest is at the power transformer low-side bus. These calculations use the positive, negative and zero sequence system impedance, per unit base ohms and per unit base current to determine the three-phase and phase-to-ground fault currents at the substation bus and at 1-km (0.6-mile) intervals along the distribution feeder. The voltage sag calculation uses the fault current for a particular fault location and the system impedance from the generating plant to the substation power transformer low side bus. A fault at the bus is a short circuit at that point, which yields a voltage sag of 100% drop, while a fault about 7 km (4.3 miles) from the substation bus results in a voltage sag of 40% drop for 60-MVA power transformer with a 397 kcmil ACSR feeder.

Alabama Power studied the possible use of feeder reactors during 1993. Although several utilities operating 35-kV systems investigated their benefits, Alabama Power chose not to use reactors because they do not completely solve the problem, they require major capital investment and their size precludes their use in existing substations due to space limitations.

While the low impedance of the 35-kV distribution feeder contributes to fault voltage sag, the high impedance of the substation transformer, generation and the transmission system also are contributors. Any increase in the feeder impedance or decrease in the generation, transmission system and transformer impedances will offer some improvement in voltage sag.

Reduced transmission impedance can be demonstrated by the new South Jefferson Transmission Substation installed during 1997 near Birmingham, where the new 230/115-kV substation offers a strong 115-kV source for the 35-kV system. The generation and transmission system impedance before the installation of South Jefferson was about twice the impedance after the new substation was installed.

The portion of the system offering the most significant contribution to voltage sag is the 115/35-kV substation transformer. Because the 60-MVA transformer offers 71.10% of the total system voltage drop during a fault 1 km from the bus on a 397 kcmil feeder, it seemed the easiest solution would be to lower the impedance of the transformer, thus dropping less voltage from the source to the bus. An investigation into the proposal to lower the impedance found that, while the low-impedance transformer could be manufactured, cost adders could make the solution too expensive. In addition, fault currents would increase above the 8000-A rating of some line devices, making it necessary to determine the appropriate impedance value for the best overall results.

Improvement Goals

Utilities know from recent experience that customer complaints also result from voltage sag occurring on the 12-kV system, although not as often as on the 35-kV system.

A study by the Computer Business and Equipment Manufacturers Association (CBEMA) provides information about various electronic equipment tolerances of voltage variations for short periods of time, which is helpful in determining the effects of system disturbances on the equipment.

Large power transformers and longer distribution feeders are common with higher voltage systems, exposing more customers to a single fault and resulting in reports of increased costs associated with voltage sags caused by faults on distribution systems. Although utilities economically justify higher voltage systems, studies indicate there may be no advantage when results may actually show a disadvantage for both customer and the utility. While economic studies may indicate 35 kV reduces substation and transmission expenses, the distribution system incurs extra cost for equipment, and the severe voltage sag causes customer complaints that do not occur as frequently with a 12- or 25-kV system.

To improve the voltage sag condition, the study recommended that the power transformer impedance be reduced from the value that had been specified for the past 20 years. Furthermore, at two locations, 25-MVA power transformers with twice the impedance of the 60-MVA units had been installed, with another planned for the future. The study also recommended that no additional 25-MVA units were to be specified. Instead, a 60-MVA unit should be installed to replace the 25-MVA unit.

Beginning in 1997, the 60-MVA transformer load rating was upgraded, before adding fans and forced oil cooling, so that the base rating of 30 MVA was increased to 36 MVA. With this new rating, the 20-year-old 35-kV impedance specification was revised to consider voltage sag improvement, while limiting maximum actual fault currents to near 8000 A. To reduce voltage sag, an impedance of 8% will be specified for all future transformers, which requires no additional manufacturing cost and results in a 20% impedance reduction. The resultant maximum fault current on the 35-kV system will be 7274 A on equipment rated at 8000 A.

Voltage sag tests under actual fault conditions on the 35-kV system were performed by closing a 10-A fuse and a 100-A fuse at the 2800-A fault location 4.8 km (3 miles) from the Elliotsville Substation. The results indicate that a voltage sag of 50% occurs for half cycle and 2.5 cycles for the 10- and 100-A fuses, respectively. The smaller fuse interrupted the fault in less than one cycle, which reduced the duration of the voltage sag. Tests using 30- and 50-A fuses also resulted in fault interruption times of about one cycle. The field tests verified calculated values and showed improvements are possible by reducing fuse sizes. Current limiting fuses tested at the field site reduced the 2800-A current to 300 A, eliminating voltage sag. However, the cost and limited size availability of current limiting fuses does not justify their use when compared with a 30-A fuse that clears within one cycle.

A fault that causes a voltage sag of 40% or less is located 5.6 to 7 km (3.4 to 4.3 miles) from the substation bus. The 35-kV feeder voltage sag for faults beyond this distance should not cause customer complaints. The results are the same as voltage sag for faults that occur beyond 1.8 km (1.1 miles) on a 12-kV feeder. Therefore, the need to improve phase-to-ground fault voltage sag conditions occurs less than 7 km from the bus. The single-phase protection on distribution feeder radial taps is typically a 100-A fuse, the maximum size that coordinates with most substation breaker relays. Fuse clearing times may be three to six cycles, allowing voltage sag to remain for this time duration. Loading on single-phase lines will allow 90% of these fuses to be reduced to 30 or 50 A, allowing clearing times of one half to one cycle. The single-phase line loading was evaluated and 100-A cutout-type switches were re-fused only within 7 km of the substations. The connected kVA was determined, doubled for cold load pick up, and 30- or 50-A fuses were installed where load allowed. Less than 10% required 75- or 100-A fuses. Current limiting fuses were too expensive for system-wide use.

Three-phase faults and the resultant voltage sag are more severe than phase-to-ground faults, where the distance from the bus is 12 km (7.4 miles) for a three-phase fault that results in a voltage sag of 40%. Lightning is the most likely cause of three-phase faults; therefore, all 35-kV feeders within 12 km require additional arresters, as well as enhanced grounding to reduce the probability of these faults.

An actual 2500-A phase-to-ground fault showed a voltage sag of 45% for 1 second before the breaker relay called for a trip. Customers served by other feeders in the substation also experienced the same voltage drop for the same duration. To address customer complaints, Alabama Power modified substation relay settings. During the early 1990s, substation breaker relays were revised to include a 12-cycle delay before tripping prior to the instantaneous reclose. A 12-cycle delay plus a 5-cycle breaker time resulted in a 17-cycle voltage sag. The instantaneous trip is usually set as low as 1200 A, which allows the breakers to trip for faults that are more remote from the substation. Raising the instantaneous trip setting to 3000 A and installing a three-phase electronic line oil circuit recloser at the 3000-A fault location will reduce the frequency of trips at the substation. After consideration of the relay timing consequences, the 12-cycle delay was eliminated, the instantaneous relay settings were raised and the electronic reclosers were used to reduce three-phase fault duration.

Conclusion

The unique characteristics of 35-kV distribution cannot be eliminated, but a thorough understanding of the system will allow voltage sag to be reduced by up to 20%. The impedance changes, fault duration reduction and three-phase fault prevention can be accomplished at minimal cost. These improvements can be incorporated into standard system design and completed over a period of time, as resources allow.

Editor's Note

This article is based on an IEEE paper, “A 35-kV System Voltage Sag Improvement,” written by M. Stephen Daniel.

M. Stephen Daniel is principal engineer at Power Delivery-Distribution Engineering Services. His experience includes 30 years of distribution engineering at Alabama Power. He served as team leader of the 35-kV Voltage Sag Study Committee, president of the National Management Association, Birmingham Division, and a loaned executive to the Greater Birmingham United Way. Daniel received BS and MSEE degrees in 1974 and 1997, respectively, from the University of Alabama at Birmingham. He is a registered professional engineer and the 2004 chair of the IEEE Alabama Section.
msdaniel@southernco.com