January, 2007

Reducing Hot-Spot Temperatures in Transformers

Over the past several years, Consumers Energy ("Consumers") has come to rely strongly on external oil coolers to delay scheduled transformer capacity increases, or to cool transformers that experience marginal high top-oil temperatures. A transformer experiencing a top-oil temperature of 90 to 100 C or more would be a likely candidate for such an installation. These types of external coolers are installed in close proximity to the transformer using flexible hoses that are typically connected to existing 11/2" taps near the top and bottom of the transformer.

Now that Consumers has acquired more than 20 oil coolers, questions frequently are being asked regarding the effectiveness of these units in actually limiting the loss of insulation life. Although the cooler reduces the oil temperature, there is a concern that it may be disrupting the natural convective oil flow inside the transformer and the hot-spot cooling effect may not be as great as expected or indicated by the top oil temperature.

Under normal conditions, the temperature gradient between the top and bottom of a transformer produces an internal oil circulation that acts to remove heat from the coils through convection. An external cooler can diminish this normal temperature gradient, resulting in reduced convective currents and, in theory, create pockets of stagnant oil and induce local overheating. To avoid this situation, some utilities have reportedly removed OEM-installed oil pumps from transformers where there has been no internally directed oil flow.

Equipment description
Study One…
The transformer selected for Study One was a unit being rewound for Consumers by Siemens Westinghouse of Hamilton, Ontario. This 5/6.25 MVA circular-core unit was originally manufactured by Allis Chalmers in 1952. Design changes by Siemens Westinghouse increased the OA rating to 6 MVA and the FA rating to 7.5 MVA. Six Luxtron fiber optic sensors were implanted near the top of the transformer's secondary coils—two in each winding with one located between the first and second disk and one between the second and third disk. The sensors were installed as near to the mid-point of the disks as feasible and in contact with the copper conductor. These locations are thought to closely represent the transformer's hot-spot location. All other temperatures recorded in this study were taken from standard thermocouples.

A 50 kW external oil cooler was obtained from Unifin of London, Ontario. This cooling unit consists of a 1 HP Cardinal pump, two 4.0 HP fans and a heat exchanger. The pump used by Unifin is designed for a variety of applications,with the desired oil flow for a given application achieved by throttling the flow with a valve on the discharge side of the pump.Nominally, this combination of components is rated by Unifin for a flow rate of 20 GPM, but the pump can produce a much higher flow, as was observed in this study.

Study Two…
The transformer selected for Study Two was a unit being rewound for Consumers by Ohio Transformer of Tallmadge, Ohio. This 5 MVA base circular-core transformer was originally manufactured by GE in 1963.

Six FISO fiber optic sensors, two per phase, were implanted in the coils of the transformer and a FISO Nortech-6 monitor was installed to record the readings. The hotspot locations were determined by the design team at Ohio Transformer, and the sensors were installed during the rewind process. All other temperatures recorded in this study were taken from standard thermocouples.

A 100 kW external oil cooler was obtained from SD Myers. This cooling unit consists of a 3 HP pump, 5.0 HP fans and a heat exchanger. The cooler is mounted on a portable trailer and includes hoses configured with check valves and quick connect fittings. The desired oil flow is achieved by throttling the flow with a valve on the discharge side of the pump.Nominally, this combination of components is rated by SD Myers for a flow rate of 50 GPM, with a capability of removing 340,000 BTU/hr.

An industry standard mineral oil and an ultra pure mineral oil manufactured by Petro-Canada with the trade name of Luminol were obtained from Ohio transformer. The transformer was first filled with standard mineral oil, tested, drained, refilled with Luminol, and then retested to obtain the efficiency comparison between the insulating oils used in combination
with and without the external
auxiliary oil cooler.

Study conditions and results
Study One…
Heat runs were initially conducted on the Allis Chalmers transformer (which had undergone design changes and was being rewound by Siemens Westinghouse) at the OA and FA ratings and then at 150% of the FA rating, or 11.25 MVA.While still at the 11.25 MVA level, the oil cooler was connected and temperatures were recorded until temperature stabilization was achieved. The cooler's oil flow rate maintained for the initial run was 45 GPM. The observed temperature differential between the cooler's inlet and outlet was consistently about 10 C degrees.

One of the fiber optic sensors stopped working early in the first heat run. The instrument displaying the fiber optic temperatures is capable of displaying four readings at a time. The temperatures recorded were taken one each from the outside windings and two from the center phase winding.

The warmest hot-spot temperature recorded while loaded to 11.25 MVA, and without the cooler operational, was 112 C on the center phase winding.When temperature stabilization was reached after the cooler was operational, this temperature had been reduced to 100 C. The magnitude of this temperature reduction was fairly consistent across all the sensors.

At the end of the first heat run with the cooler connected, the pump flow rate was increased to its maximum (estimated to be about 60 to 65 GPM) for one hour.No appreciable change was noted in the hotspot temperatures as a result of this, although there was a reduction of two degrees in the top-oil and average-oil rise temperatures. Had the test continued at this higher flow rate for a longer period, it is expected that the hot-spot temperature would have registered a similar decline.

The flow rate was then reduced to 20 GPM for a four-hour period. This resulted in an increase in the hot spot temperatures of approximately 4 C degrees.

Study Two…
Heat runs were conducted on the GE transformer (that was being rewound by Ohio Transformer) at the OA and FA ratings and then at 150% of the FA rating, or 10.5 MVA, initially with the transformer filled with standard industry mineral oil and then repeated after draining the oil and re-filling with Luminol.While at the 10.5 MVA level and after the temperature stabilized, the oil cooler was connected and temperatures were recorded until they stabilized again. The cooler's oil flow rate maintained for this study was 24 GPM.

 

The average hot-spot temperature recorded while loaded to the FA rating of 7 MVA, and without the cooler operational, was 92 C, using standard oil, and 87 C, using Luminol after stabilizing.When
temperature stabilization was reached after the cooler was operational, this temperature was reduced to 83 C, using standard oil, and 80 C, using Luminol. The magnitude of this temperature reduction was fairly consistent across all the sensors. The observed temperature differential between the cooler's inlet and outlet varied between 8 and 14 C degrees, using standard oil, and between 11 and 18 C degrees, using Luminol.

The load was increased to the 10.5 MVA level, the oil cooler was connected, and temperatures were recorded until temperature stabilization was achieved. At this point, it was observed that the average hot-spot temperature of 140 C, in both cases, had been reduced to 127 C, using standard oil, and 115 C, using Luminol. The magnitude of this temperature reduction was fairly consistent across all the sensors. The observed temperature differential between the cooler's inlet and outlet varied between 12 and 15 C degrees, using standard oil, and between 21 and 28 C degrees, using Luminol. (See Tables I & II and Figs. 2, 3, 4, 5, 6, 7 on the previous pages.)

Conclusions
This study substantiates the benefit of employing an external oil cooler and the added benefit of using an ultra pure mineral oil (Luminol) in reducing a transformer's hot spot temperature, thus preserving the life of the unit's paper insulation. The relatively large internal oil quantities and large heatexchange surfaces of the transformers in this study result in relatively low internal oil and hot-spot temperatures.

Conversely, for a more modern unit with higher design temperatures, the expected temperature reduction with an external oil cooler could be even more impressive. However, the possibility of disrupted internal convection currents or diversion of oil from the transformers' own radiators also would seem to be more likely because of the characteristically lower internal oil volumes. Consequently, a lower oil flow rate in the external cooler might be needed to avoid disrupting the transformer's normal internal cooling pattern.

The transformer in Study One contained 1,920 gallons of oil, or 0.32 gallons per OA rated kVA, and the transformer in Study Two contained 1,300 gallons of oil, or 0.26 gallons per OA rated kVA. In a spot check of six transformers recently purchased by Consumers Energy, the lowest amount of oil found was 0.205 gallons per OA kVA rating. The SD Myers transformer maintenance guide reported in 1981 that some transformers had as little as 0.02 gallons per kVA.

In light of the significant variations in transformer oil volumes, flow to the external cooler may need to be tailored for the particular transformer involved. Besides possibly needing to modify the internal oil-cooling pattern, there also is a concern for creation of a vortex at the top hose connection. This would lead to air being sucked in and air bubbles being injected into the bottom of the transformer. A minimal oil level above the top hose connection must be maintained to avoid this or other possible measures must be adopted.

Noel Staszewski is a senior engineer in the Network Services Department of Consumers Energy.He has over 25 years of engineering experience in asset management and equipment maintenance in the utility industry, combined with additional experience in technology and product development, evaluation, reliability engineering and failure analysis of electronic components and systems in the automotive and computer industries. Telephone: (810) 760-3237; E-mail: nnstaszewski@cmsenergy.com

Mike Walker, a registered Professional Engineer in Michigan, spent 33 years in a number of engineering positions with Consumers prior to retiring in 2003. Since then, he has worked as an independent contractor for various companies. E-mail: mkwalker16@hotmail.com