Seattle City Light (SCL) identified a thermal bottleneck that limited ampacity of 13-kV cables inside duct banks near a major substation supplying downtown Seattle, Washington, U.S. The depth of burial limited the 72 cables to 310 A each, instead of 510 A. Other cables nearby limited ampacity further. So, the existing cables and ducts would not meet near-term projected load increases. Investigations determined that new substation construction was not feasible, nor was it feasible to interrupt customers to perform capacity-improvement work.

A chilled water heat-removal system solves these problems. A prototype designed by SCL is under construction with commissioning planned for the third quarter of 2005. The ampacity will increase by about 40% for the 13-kV, 26-kV and 115-kV distribution and transmission cables near the substation. The cooling water approach to ampacity improvement has potential applications elsewhere at SCL and other utilities.

Seattle's downtown distribution system is a 13.8-kV secondary network fed from three substations with an aggregate capacity of 410 MVA serving an area of 1.33 sq miles (3.5 sq km).

Studies in the late 1990s indicated that downtown load growth might soon approach feeder capacities. Actual feeder capacities were estimated by using commercial ampacity software in conjunction with measurements of soil thermal parameters. The results led to the de-rating of many substation feeders from their original nominal ratings.

Several capital improvement programs were initiated to increase ampacity. They included site-specific evaluations prior to any capacity increases. One substation supplying downtown Seattle had serious feeder limitations. Detailed studies were made with the assistance of heat-transfer consultant Dr. Mohammed Chaaban of Montréal, Canada. The studies improved on earlier estimates by using 3-D finite element analysis (FEA), taking into account site-specific situations, including steam lines and cable crossings. Various options to increase ampacity were also modeled. Most ampacity values presented here are from the FEA. In instances where the configuration differs slightly from FEA models, ampacities provided are conservative estimates.

Thermal Bottleneck

The four parallel 13-kV duct banks shown in Fig. 2 are located just outside a major distribution substation in downtown Seattle. Many duct banks and cable crossings are closely spaced. Ampacity at this site is reduced by several thermal factors.

Each 13-kV duct bank has the cross-section shown in Fig. 3. Mutual heating occurs from within the array of six conduits and from adjacent 13-kV duct banks within about 5 m (16 ft). This heat at a burial depth of 2.5 m (8.2 ft) reduces the ampacity per cable from an expected 500 A to 307 A. Other power ducts crossing in the vicinity reduce ampacity further.

Chilled Water System

Figure 4 shows the layout for a chilled fluid heat-exchange system. Cooling pipes follow the 13-kV duct banks for the entire underground length, addressing the mutual-heating and depth problem. The pipe layout also provides cooling at each major crossing. At the substation wall, the pipes emerge from underground and join supply and return manifolds. Chillers installed in the substation then reject the heat from the circulating fluid to the atmosphere.

Cooling Pipe Details

Four cooling pipes were installed above the 13-kV duct banks, as shown in Figs. 3 and 8. They are arranged as two supply and two return pipes. The pipes are embedded in fluidized thermal backfill (FTB). This backfill provides a 50% improvement in thermal properties over most concrete. Polyvinyl chloride (PVC) pipes are used as cooling pipes instead of steel pipe. This minimizes pipe surface-current flow concerns, and it facilitates leak-free joining by means of glued sockets.

SCL discovered that placing pipes atop the 13-kV duct banks would not suffice where a 115-kV transmission line passes less than 18 inches underneath the duct bank. Therefore, some of the cooling pipes were elbowed down under the 13-kV banks to form horizontal loops a few square meters just above the transmission line.

Perpendicular Heat Source

A 26-kV feeder circuit crossing perpendicular to the 13-kV duct bank is shown in Figs. 5 and 6. The spacing was so tight that SCL could barely fit 4-inch (10-cm) cooling pipes between the circuits. Other 26-kV feeders were being completely replaced with cooling pipes built into the new conduit array.

Bottleneck Benefits

The addition of forced cooling at the bottleneck of the 13-kV cable circuit increases ampacity from less than 307 A to 520 A with a corresponding capacity gain of 110 MW. SCL found that the ampacity of the crossing 115-kV cable was limited in the region where it passed 18 inches (46 cm) below the 13-kV duct banks. With forced cooling, the ampacity of the 115-kV circuit improves from around 500 A to 800 A in the bottleneck area, depending on the loading of the other feeders.

SCL also achieved greater ampacity on the 26-kV cable circuit. Before-and-after ampacities are hard to compare exactly, but the gain seems to be 40% of the original design rating.

Alternatives Considered

All plausible approaches were examined. The objective was an ampacity of 510 A per cable while keeping the 13-kV feeders in service if possible. The following approaches were found inadequate by finite element analysis:

• Removing the cables from one duct, and using that duct to circulate cooling fluid.

• Installing thermal insulation between underground heat sources. This was counterproductive.

• Installing one cooling pipe on top of the duct bank.

FEA has determined that placing three cooling pipes on top of the existing 13-kV duct bank yielded an ampacity of 520 A. Providing a fourth pipe enhances the system redundancy.

Other approaches considered were found to be clearly deficient: backfilling with more conductive fill; reducing the burial depth; passive ventilation; and forced-air draft through a tunnel or the power conduits.

Environmental Design

SCL applied best practices with respect to environmental management and compliance. The refrigerant is R407C, a blend of three hydrofluorocarbons (HFCs). They are chlorine-free, ozone-friendly and EPA-approved under section 612 of the U.S. Clean Air Act. Similarly, the antifreeze chosen is propylene glycol (PG), which is nontoxic. Coolant leakage is a serious concern. The hydro test was performed at five times the normal system pressure to ensure the system had no fluid leaks. Other antileakage measures are too numerous to list here.

Reliability is designed into the system with redundant chillers, compressors, coolant circulation pumps and controls (Fig. 7). The duct banks have two cooling loops each. Feeders inside a duct bank cooled by only one loop should still provide an ampacity of 490 A.

Operating Scheme

System control will normally be thermostatic. One circulation pump will provide continuous flow in the loops. Sensors measure coolant return temperature. At 10°C (50°F), a chiller will kick in, providing cooling at a fluid flow of 60 gallons per minute.

The system will operate unattended under microprocessor control. Alarm conditions will annunciate at SCL's System Control Center. Indications will be available on SCL's local area network (LAN) and locally at the chillers.

Network power demand peaks in late summer and during the daytime. The cooling system may be idle much of the year and then enabled whenever power approaches a selected threshold in any of the four duct banks. The total chiller operating hours per year will be low.

Temperature and Moisture Monitoring System

Thermistors were installed at 30 critical locations to monitor the temperature of the power ducts. These temperatures (Fig. 9) directly relate to the cable temperatures, which can be estimated within a few degrees. Ampacity is then obtained empirically, not based on assumed material properties, mathematical modeling or worst-case air and soil temperatures. This data and real-time operation of our network power system will allow the cable's ampacity to be better utilized. SCL also measures soil-moisture content, which provides feedback on the thermal characteristics of the soil surrounding the duct bank. This is important because heat generated in the duct bank passes through the native soil. Because the thermal properties of soil depend on moisture content, these moisture measurements can warn the utility if soil is drying, which would result in higher soil thermal resistivities occurring.

The Unexpected

Excavation revealed some interferences and undocumented changes in the buried structures, which required ingenuity. Concrete had often been poured without forms during original construction decades ago. The duct banks were oversized, irregular and sometimes merged into a twin bank. Some control ducts were not positioned as shown on the bank cross-section drawings. Such deviations introduce uncertainty and impair heat transfer. The remedy was careful concrete chipping by SCL cable splicers. Exploratory chipping pinpointed the duct locations. Large-scale chipping restored the bank cross-section to thermally acceptable dimensions.

Cooling pipes must sometimes be diverted away from ductbanks due to interferences. SCL judged that heat will readily flow 2 m or 3 m (6.5 or 10 ft) through the copper to other points that are well cooled.

The SCL prototype-forced cooling design has many features to allow for unknowns and expansion. Several more are noteworthy.

The 13-kV and 26-kV peak loads don't normally coincide. At some loads, the circulation pumps could be used to redistribute heat from high-power ducts to low-power ducts, without even turning on the chillers. Pump pressure, flow rate and motor size allow for capacity increases.

The cooling system's volume and heat capacity are large. With coolant flow and no chilling, the system will warm up less than 5°C (9°F) per hour at maximum heat input. Many hours would be available to respond to cooling system problems or for peak loads to decrease.

Future Work

The thermal bottleneck on a 13-kV duct bank supplying downtown Seattle resulted in a circuit that sacrificed 40% of its rated supply capacity to a major distribution substation. SCL stands to regain this capacity by installing a couple thousand feet of pipe and two chillers — without even turning the power off. Some steps remain to complete and commission the system. During 2005, the chillers will be installed, some feeder cooling lines will be completed, and control and communication wiring will be connected.

Phone lines will be installed to transmit real-time data to the System Control Center for 24 hours a day monitoring. During summer peak loads, the system will be tested, and operation will be optimized based on actual values, not design values. Insights gained will allow economizing in future applications. SCL will continue thermal and capacity studies of its underground network and residential distribution systems.

Acknowledgments

We acknowledge and thank the following SCL workers for their contributions: Dr. Betty Tobin, director of Central Electrical Services, and Franklin Lu, manager of Network Engineering; Engineers Kyle Ho, Gordy Welsh, John Li, Mathew Lawson, Kosea Kalebu, Mehari Gebrewold and Irv Ogi; and crew chiefs Chuck Mahar, Gene Gihring, Keith Langi and John Hansen. We also thank SCL electrical, civil, stations and shops crews for their work in constructing this system.

Hamed Zadehgol received both the BSEE and MSEE degrees from the University of Washington, Seattle, in 1990 and 1992, respectively. He did a research assistantship with the University of Washington, and since 1990, he has worked in the Seattle City Light's Network Engineering division, most recently as an engineering supervisor. Zadehgol is a registered professional engineer. Hamed.Zadehgol@seattle.gov

Jack Prestrud earned the BSME and the MS in applied physics from the University of Washington. He joined Seattle City Light in 1999 as a senior mechanical engineer. He has previous experience as a licensed nuclear power plant test engineer; as a designer of pressure vessels and other equipment for chemical plants and refineries; in process safety management for the Alaska oil fields; and in interdisciplinary equipment troubleshooting. Prestrud is a registered professional engineer. Jack.Prestrud@seattle.gov

Design Parameters
Thermal design Heat load, 13 kV: 26 kV: Future:	30 kW 20 kW 20 kW
Fraction of heat removed by cooling water: Soil T, design max: Air T, design max:	89% 24°C (75°F) 35°C (95°F)
Coolant Supply T: Return T: Design minimum T: Composition: Freezing point:	10°C (50°F) 11°C (51.8°F) -5°C (23°F) 75% water, 25% propylene glycol -12°C (11°F)
Cooling system Volume: Pipe total length: Hydro test pressure: Operating pressure:	3300 gal 4000 ft 100 psig 20 psig
Hydraulics Coolant velocity: Reynolds number: Circulation pump capacity:	0.25 m/sec, 32 gpm/pipe 7000 500 gpm at 10 psi differential
Chiller Rated capacity: Coolant leaving T: Flow rate:	70 kW, 20.5 refrigerant tons 4.4°C (40°F) 60 gpm