CONNECTICUT LIGHT & POWER IS CURRENTLY CONSTRUCTING TWO LARGE TRANSMISSION CABLE PROJECTS along 32 miles (51 km) of heavily traveled state roads in Southwest Connecticut. Twenty-four miles (39 km) of the route have two circuits and the remaining 8 miles (13 km) are built for three circuits.

Connecticut Light & Power (CL&P; Hartford, Connecticut, U.S.) only permits one transmission-voltage extruded-dielectric cable circuit per vault so that maintenance can be performed on one circuit while the other circuit remains energized. However, there are concerns about the resulting 219 (9 ft by 32 ft [2.7 m by 9.8 m]) vaults proposed by CL&P, especially regarding installation time, traffic disruption during construction and issues involving the time that traffic may be affected during future maintenance and repair.

CL&P's first and foremost design requirement, which could not be compromised, was the safety of maintenance personnel in the unlikely event of a cable or joint failure. Next, the goal was to develop a vault that would minimize traffic disruption during installation. The resulting new vault design addresses issues of safety, operability, siting impact and reliability.


CL&P retained Power Delivery Consultants Inc. (PDC; Ballston Lake, New York, U.S.) to provide engineering services to evaluate the requirements, and to design, fabricate and test a partitioned underground vault that would provide protection for workers on one side of the wall should a cable or joint failure occur on the other side of the wall. PDC proposed a unique two-chamber vault design that would permit workers to service the extruded dielectric cable in one side of the vault while cables in the other half of the vault remained energized. The mating walls were designed to work together as a protective barrier for the safety of workers located in the de-energized side of the vault. Raymond C. Tiegs & Associates (Plymouth Meeting, Pennsylvania, U.S.) and A.C. Miller Concrete Products (Spring City, Pennsylvania) completed the structural design and fabrication of the vault scheme. The vault design consists of twin abutting vaults with a grout-filled space between them. The two-chamber design had the added advantage of making the vault easier to transport and lower into the intended site along the route of the circuit. Figure 1 shows the wide end for making the joint, and the narrow end for parking the splice assembly while preparing the joint area.

Underground explosions were simulated to determine the maximum gas pressures that might be expected in the vaults. An existing computer program that was developed and verified over an extended period of time at the Georgia Institute of Technology was used. The program has the capability of simulating the conditions during an arcing fault and a gas explosion in an underground structure. In the worst-case scenario of a cable or joint failure, the resulting high temperature arc can vaporize cable insulation materials. These compounds are hydrocarbon-based, and therefore are combustible. If they occur in proportions within their combustible range, the arc can ignite them, which can add to and magnify the amount of energy released by the arc.

During a preliminary test in which a high-energy fault was placed on an aboveground cable, the arc was accompanied by a large ball of flame and a considerable amount of soot, which confirmed that a gas explosion had actually occurred. After witnessing the initial aboveground fault and the ignition of the generated gases, subsequent tests were planned in which both the energy from the arc as well as the explosive combustion of byproducts were considered. A hypothetical worst-case scenario of a short circuit with an average root-mean-square (rms) arc voltage of 1.5 kV, an rms current of 63 kA and a duration of 18 cycles in an explosion consisting of the vault filled with 10% methane — which has more explosive energy than the combustion of byproducts from the arc — would conservatively produce a peak pressure rise inside the test vault of about 16 psi over a short duration of 300 msec. This conservative value was then used to design the vault structure to withstand that level of high-impact load without injury to personnel in the adjacent vault and to minimize structural damage. Additionally, a dynamic load factor of two was used to compensate for the short time duration of the pressure rise (impact type force).


Once a preliminary vault design was finalized, our attention turned to conducting a full-scale test of a vault prototype to assure that it met all the required design criteria.

Two separate vault tests were conducted at KEMA-Powertest Inc.'s Chalfont, Pennsylvania, testing facility. The twin vault was buried and a high-energy arcing fault was initiated inside one vault chamber. The initial test consisted of an arcing fault placed in a fully constructed high-voltage cable splice installed on cable racks. A tether restraint system was also installed on the two covers associated with the “explosion” side of the vault. The second test was identical to the first except that the arcing fault was placed in a piece of high-voltage, cross-linked polyethylene (XLPE)-insulated cable and the uplift restraint plates securing the top section of the vault to the bottom section were removed. For both vault test cases a high-energy arcing fault was introduced by applying the following test values:

Voltage = 16-kV single phase
Current = 63 kA rms symmetrical
Frequency = 60 Hz
Time = 18 cycles.

For the initial test, a 138-kV XLPE prefabricated, premolded splice with approximately 5 ft (2 m) of 2250-kcmil copper conductor, 0.850-inch (22-mm) XLPE, corrugated copper sheath, PE jacket 138-kV cable on each end was installed inside the vault. A lag screw was used to penetrate the splice for the short-circuit test. Figure 2 shows the cable splice before the test. The table summarizes data measurements recorded during Test 1: Splice Fault.

The rms arc voltage varied over the test duration, with the peak values occurring during the first two-thirds of the test. The arc voltage over the last third of the test was closer to 1500 V rms. The reduction in the arc voltage is likely due to the shortening of the arc length as the arc moved from the splice housing to the cable sheath. Figure 3 shows the splice after the fault test; note that the housing has been blown completely open. The extent of the damage was much greater than typically seen for in-service faults. This indicates that the test was more severe than in-service faults.

Large white clouds of dust-like material along with small flashes of light were observed exiting the vault openings during the fault test. No movement of the vault structure or soil covering the vault was observed. The floor, walls and ceiling of the energized part of the vault chamber had a thin layer of a light-colored “dust,” which appeared to be residue from the quarry sand from the splice housing. A layer of quarry sand and melted copper particles was found on the wall and ceiling directly over the fault location. In addition to the damage to the splice housing, the premolded components of the splice were torn and split.

For the second vault test, a piece of 2250-kcmil copper conductor, 0.850-inch XLPE, corrugated copper sheath, PE jacket 138-kV cable approximately 12 ft (4 m) long was prepared on cable racks inside the same vault chamber used in Test 1. After some calibration tests were completed, the cable was “faulted” using a steel screw installed from the sheath to the conductor several feet from the end of the cable. Figure 4 shows the cable and screw prior to the fault test. Again, a high-energy arcing fault was introduced in the cable. See the table for data measurements recorded during the Test 2: Cable Fault.

The rms arc voltage appeared to be fairly constant over the test duration. Figure 5 shows the cable after the fault test.

A larger, darker cloud was observed exiting the vault openings than was present for the splice fault, indicating additional combustion of byproducts. Also, more light flashes were visible. Again, no movement of the vault structure was observed. The cable insulation was burning after the conclusion of the test and had to be extinguished. A conical-shaped portion of the cable and conductor approximately 5 inches (12.7 cm) in diameter was missing from the cable after the test fault. A darker-colored layer was found on the vault floor, wall and ceiling following the cable fault than was found after the splice fault. Again, there was significantly more damage than typically observed for extruded-dielectric cable faults in the field.


Visual inspections of the vaults were completed after each test for signs of structural damage. With no signs of structural damage evident, it was determined that both vaults passed the test, performing within the design limits of the materials. During the first test, the uplift plates used to restrain the top section of the vault from any upward movement had no indication of being overstressed. During the second test, with the uplift restraint plates removed, no movement of the vault or disturbance of the soil was observed.

External video cameras verified that there was byproduct combustion during both tests as evidenced by the smoke and flame that exhausted from the vault chimneys. The pressures from the electrical arc and resulting combustion of gases generated by the rapid burning of decomposition products were approximately half the maximum value calculated for a high-energy fault and stoichiometric proportions of methane gas in the vault, even though the damage to the splice and cable were much more severe than found in the field.

The 9-psi pressure rise that was observed in the vault with the cable fault was caused by the high-energy electrical fault and explosion of vaporized insulation, and is the expected pressure as a result of a cable fault in the field. The test verified that the 16-psi design pressure (which was based on a theoretical, highly conservative methane gas surrogate for insulation vaporization) provides a margin of safety. On top of the 16-psi design pressure, an added factor of two was used in the vault design. The tests confirmed (a) the vaults are very capable of withstanding the 9-psi pressure expected from a high-energy electrical fault with vaporized insulation explosion and (b) the vaults protect workers in the adjacent space in the unlikely event that a high-energy electrical fault with vaporized insulation explosion occurs with the expected pressure.

CL&P plans to move forward with the new design. We believe the new design will reduce our two-vault installation from an overall footprint of about 69 ft by 20 ft (21 m by 6 m) to about 33 ft by 14 ft (10 m by 4 m). But most importantly, we believe the test data supports the conclusion that the twin vault design provides excellent protection for workers on one side of the wall in the unlikely event of a cable or joint failure on the other side of the wall. CL&P is developing work procedures to safely address induced voltages from the energized circuit.

Rachel I. Mosier received her bachelor's degree in engineering from the University of Connecticut in 1992. After working in the defense industry for several years, she joined Northeast Utilities Service Co. in Connecticut. Mosier worked in the Distribution Construction and Material Standards department before joining the Transmission Engineering department, focusing on underground transmission cable systems. She is the vice chair of ICC Subcommittee C — Cable Systems, a utility advisor on several EPRI projects and an active member of the AEIC Cable Engineering Committee. She is a registered professional engineer in Connecticut.

Victor D. Antoniello joined Power Delivery Consultants Inc. in 2003 as a senior engineer. He specializes in underground cable-system planning, design, specification, installation, and operation and maintenance. Prior to that, he worked 17 years at an electric utility providing engineering support for the installation, operation, and maintenance of underground transmission and distribution cable systems. He earned his BSEE degree from the University of Maine in 1986 and his MSEE degree from Northeastern University in 1998. He is a registered professional engineer in Massachusetts and Rhode Island.

W.Z. Black received his BSME and MSME degrees from the University of Illinois and his Ph.D. in mechanical engineering from Purdue University. He served on the faculty of the Woodruff School of Mechanical Engineering at Georgia Tech for 33 years. While there, he taught courses and conducted research in the area of thermal sciences and was the Georgia Power Distinguished Professor in Mechanical Engineering. He currently holds the title of regents professor emeritus. He is a fellow of IEEE and ASME.

Full-Scale Fault Test Conditions
Test Measurement Test 1: Splice Fault Test 2: Cable Fault
Open-circuit voltage 16 kV 16 kV
Average symmetrical current 64.9 kA 66.5 kA
Peak current 107.3 kA 172.4 kA
Current duration 308.4 msec 313.2 msec
Peak arc voltage (from voltage waveforms) ~2550 V rms ~1400 V rms
Vault pressure rise in failure vault 7 psi 8 psi to 9 psi
Temperature in “worker” vault 13.6°C (56.5°F)[1] 13.1°C (55.6°F)[1]
Sound measurement in “worker” vault ~102 dB
(meter clipped during event)
~124.7 dB[2]
Air quality in “worker” vault 21.1 ppm oxygen
(LEL not exceeded)
21.1 ppm oxygen
(LEL not exceeded)
1. No measurable temperature rise
2. 120 dB is approximately the level of an ambulance siren. Irreversible hearing loss occurs at a level of ~180 dB.


Consideration was given to restraining the manhole covers on each side of the vault so they would not fly into the air during a fault. Computer simulations predicted that trying to hold the manhole covers down, so that they could not move, was not a feasible solution. It is far better to let the cover move upward a limited distance, to vent high pressures, and maintain internal pressure at lower levels. The manhole cover was thus tethered by a short length of webbing, much like traditional seat-belt material.

The selected tether material was a combination of nylon and polyester webbing that was threaded through a metal loop cast into the concrete floor of the vault and another loop attached to the manhole cover as shown schematically here.

The tethers were attached to the vault floor and cover and sized to allow the cover to lift approximately 24 inches to 30 inches (610 mm to 762 mm) above grade due to a combination of slack and stretch when absorbing the force.

The cover-tether restraint system worked successfully during both tests performed at KEMA-Powertest's test facility. Video footage from both tests shows that the covers lifted approximately 24 inches to 30 inches above grade and then were pulled back down as the pressure declined in the vault. After both tests, a visual inspection of the tethers and attachments revealed no significant damage. A minor burn mark was found in the covering of one tether, which was likely caused by burning during the cable fault test.