Some years ago, ComEd began experiencing a marked increase in failures of its low pressure fluid-filled transmission system. Analysis showed four lines were near to the ends of their lives and needed replacement. Staff report

Like most major conurbations around the world, Chicago’s electricity infrastructure was developed piecemeal by a plethora of small competing supply companies which never were fully co-ordinated and optimised.

Modernising and upgrading inner city networks in concentrated high value property areas has always been a phenomenally expensive and politically tortuous undertaking, and there has been very little commercial incentive to invest in these systems in recent years.

However, Unicom subsidiary Commonwealth Edison has recently invested more than $1.5 billion in a massive overhaul of the utility’s transmission and distribution system as a whole.

Among the key elements of this ambitious undertaking are substation renewal in Chicago (see Modern Power Systems, May 2000) and replacement of four 138 kV transmission circuits in the city. Maintenance of the existing paper insulated fluid filled cable network had become increasingly laborious and costly.

Preserving existing infrastructure

The cable replacement project was undertaken in partnership with Pirelli Cables and Systems Oy, Finland (formerly Nokia Cables, NK Cables and NK Energy).

The retrofit project covered 99 miles of cable, over 400 joints, 48 cross-bonding boxes, 23 grounding boxes, 1600 fibre optic splices, 18 fibre optic organiser boxes and 30 terminations. Commonwealth Edison was responsible for engineering design, removal of the old cables and cleaning of manholes and old concrete ducts. Pirelli was responsible for delivering the new cable, which was manufactured at the Pikkala factory in Finland, delivering accessories, manufactured at the Delft factory in the Netherlands, and for overall installation, which it has done on a turnkey basis, using local sub-contractors.

The total project period was from 1996 to 2000. The first circuit, 9.8 circuit miles in length, was put in-service on 21 December 1998. The second circuit, 5.8 circuit miles, was energised on 5 April 1999. The third circuit was placed in service on 21 December 1999. The fourth and final circuit was complted on 1 May 2000.

An important requirement of the cable replacement project was to preserve the existing infrastructure and use the existing conduit and manhole system. This avoided unnecessary construction work in a crowded area but posed many other challenges. Around 45 years old on average, the existing fluid-filled cable system was housed in 4in stone ducts and manholes, some dating back the 1920s.

Being near the lake, the ground water level is high and the manholes are always filled with water.

Having housed a failing fluid-filled system, the ducts and manholes were extremely dirty. They were filled with residual dielectric fluid from failures, silt and grime dropped from traffic and construction activities, and chemicals from leaks or spills in the area.

Removal of the old cable was very messy, with residues of dielectric fluid, lead jacketing and old pulling lubricant (hydrocarbon grease). To make the conduits ready to receive the new cable an intensive programme of cleaning, proofing for obstructions and alignment, and remote camera inspection was undertaken.

The original grease aided removal of the very heavy old cables, but much of it remained in the existing stone conduits.

A cleaning process was developed which included pulling through a series of three rubber disc swabs and multiple passes with a washer head supplied with hot water and degreasing agent at 1000-4000 psi.

A 12in long mandrel 1/4in less in diameter than the conduit was also pulled through the ducts to proof them for correct diameter, alignment of joints, etc. The mandrel was pulled with graduated tape to locate problems.

Pilot project

  About three months prior to the start of work, a pilot project was instituted, consisting of six pulls estimated to be particularly hard.

The aim of the pilot was to put some practical experience behind the pulling calculations, finalise the cable installation methodology and verify that the cable design was sufficiently robust.

In the event, actual pulling tensions were found to be as low as 3000 lb for a 1400 ft run.

As a result of the pilot, new tools were developed to facilitate installation of cable and joints.

For cable installation a variety of specialised rollers and pulling blocks were designed. For jointing a hydraulic cable pulling frame was used.

The frame has two clamps to hold the cables. On one end there is a half-open ring that holds the joint insulator when it is pushed on the first cable end. The hydraulic cylinders of the frame are driven by a small pump. To ease operations in the manhole, a commercially available mini-lift was modified to lift cables and frame to the required level. A second hydraulic frame was used to pull the cable into the pull-through manhole.

Cable system design

With only 10 mm (0.25 in) clearance between the new cable and the existing ducts, and with many of the ducts having bends, pulling tensions were initially a concern. But the low weight of the new cable, approximately 10 lb/ft, meant these tensions turned out to be much less than expected.

Because of the restrictions of the duct system, the outer diameter of the cable had to be 3.5in (90 mm). However, the new cable was required to have a rated AC voltage of 138 kV and a current ampacity of 920 A, forcing it to its load limits. Integrated optical fibres were thus needed to allow real-time on-line temperature monitoring.

Characteristics of the new cable are: 800 mm2 copper, 4 segment stranded conductor, 650mil XLPE insulation thickness, copper concentric wire screen able to withstand 63kA of fault duty, copper laminate moisture barrier longitudinally sealed with adhesive and two optical fibre units between the copper screen wires.

Thermal expansion of the new cable system was another concern and was investigated in Europe on a full scale manhole model built specifically for the project, appropriately named “Little Chicago.” As a result, the engineering groups designed a flexible racking system such that all three phases are supported from the manhole roof, allowing them to move as one unit while keeping each splice rigid.

The cable design used, with XLPE insulation, pre-moulded joints etc, was new to Chicago as well as to other utilities in the USA. But Pirelli found Commonwealth Edison to be open-minded, curious, receptive to innovation and a good team player. The project proved to be an important step in introducing this new technology to the USA.

Integrating fibre optics

As already noted, an important feature of the new cable is fibre optic distributed temperature sensing (DTS), allowing dynamic thermal circuit management.

Power transmission decision models are becoming more complex as the unpredictability of deregulation takes hold and transmission assets are increasingly used in ways not originally envisaged. One of the major variables in the equation is the temperature of underground cables and how and where hot spots may be developing under constantly changing climatic and loading conditions.

Historically, in the absence of a real-time thermal profile covering the entire length of an underground transmission cable, a system operator has had to base his emergency ratings largely on inferred rather than actual temperatures for the cable. With the numerous variables and their associated uncertainties, the margins required to ensure system safety can stack up and result in overly conservative “book” rating calculations. The Electric Power Research Institute (EPRI) estimates that most transmission lines have significantly greater power transfer capacity (20-30 per cent) than their static ratings indicate.

In July 1999, Systems & Processes Engineering Coporation (SPEC) entered into a partnership with Commonwealth Edison that has resulted in the integration of DTS technology developed by SPEC’s Sensor Trans division into ComEd’s 138 kV underground transmission grid. The first systems were installed at two of ComEd’s Chicago substations in May 2000.

A distributed temperature sensor makes measurements by detecting and analysing the Raman scattering produced by a pulsed laser source introduced into an optical fibre. The Raman scattering produces frequency shifted wavelengths known as the Stokes and anti-Stokes lines. The intensity of the Stokes line is temperature independent. The anti-Stokes line intensity varies as a function of temperature. The ratio of these two lines thus gives a direct and very accurate measure of absolute temperature.

Typical accuracies of ±1°C and spatial resolutions of 0.5m are readily achievable on runs as long as 8-10 km.

The SPEC technology, developed in SPEC’s SensorTran division, has its origins in the US space programme. Under a 1996 NASA contract, SPEC developed a space-qualified distributed temperature sensing system using fibre optics embedded in the outer shell of spacecraft fuel tanks. The space system is designed to detect and accurately locate cryogenic fuel leaks before they can escalate to a Challenger-like disaster.

In the Chicago application, the SPEC/SensorTran DTS system introduces laser light into the fibre sensor at a rate of 10 000 pulses per second at a wavelength of 1 064 nm. Over a 10 km length of fibre and with 0.5 m resolution, that amounts to 200 million (20 000 x 10 000) individual data points for which Raman scattering data must be gathered and manipulated each second to produce a thermal profile of the fibre. For every measurement point, three or more individual parameters are collected or derived. In short order, continuous DTS measurements along a transmission cable can produce a mountain of data, which by itself would be unwieldy and largely meaningless.

Critical to a useable DTS system, therefore, is intelligent software that can mine the mountain and extract just the information of value to the operator. This involves comparing real time data with historical thermal and circuit loading databases to pinpoint unexpected or potentially anomalous temperature excursions in time to take corrective action.

DTS at ComEd

The most common calls ComEd rating personnel receive from Operations each summer relate to whether a cable circuit can withstand loading beyond established emergency book ratings. Open-access and the introduction of independent power producers have further complicated power flows beyond conventional planning. Traditionally, where the line of interest is uninstrumented, this involves obtaining loading history from the dispatcher, and a quasi-dynamic rating would be calculated using empirical ground temperature data. ComEd has a substantial body of ground temperature data and these temperature profiles do not change significantly from year to year. The ground temperature during July-August varies between 24° and 13°C as depth of installation changes from 2 ft below ground to 15 ft below ground. Use of actual DTS data, however, removes the estimation error component entirely.

In April 1999, SPEC made distributed temperature measurements on an energised phase cable, Line 0705, at ComEd to verify baseline data. Line 0705 is a DTS-ready cable that starts out from ComEd’s Washington Park substation in a 4×3-conduit package. The temperature profile shown in the upper diagram is a plot of DTS data for the first 5 000 meters of the line. The only other line in the package is another 138kV line that was not energised at the time of measurement. The conduit package is about 10 feet below grade as it exits the substation.

Several pieces of information can be extracted from this graph. First, the deeper installation and sparse conduit occupancy are responsible for cooler cable temperatures. As the circuit moves further south from the substation the conduit run is shallower and ground temperature warmer. The conduit configuration is generally 4×3, but at times it changes to smaller packages; however, away from the substation, load-carrying cables occupy most of the duct runs. The higher cable temperatures can be attributed to the heat loss of other cables in the duct affecting Line 0705. The load on the line at the time of temperature measurement was about 100 MVA (less that 50 per cent of its rated capacity). If the entire temperature profile is extrapolated to July/August, the cable temperature can be expected to be around 48°C at 100 MVA.

Eventually, all of these data will be gathered and analysed on a real-time basis to accurately assess the thermal capacity of the cable