How 110kV Oil Immersed Power Transformers Manage Thermal Loads
Thermal behaviour is the single most decisive factor in the service life of a 110kV oil immersed power unit. Electrical designers can calculate rated throughput, impedance and insulation coordination with considerable precision, yet the temperature at which the winding actually operates determines whether the cellulose insulation survives thirty years or fails in twelve. Understanding how heat is generated, transported and rejected is therefore not an academic exercise: it shapes the cooling specification, the loading policy, the protection settings and the maintenance programme for the whole substation bay.
Where the Heat Comes From
Every energized unit converts a small fraction of the power it transfers into heat. That fraction is normally divided into three families. No-load loss, sometimes called core loss, arises from hysteresis and eddy currents in the grain-oriented silicon steel laminations. It is present whenever the unit is energized, essentially independent of load, and for a 110kV class machine typically falls between 0.06 and 0.12 percent of rated capacity. Load loss, dominated by the I squared R heating of the primary and secondary windings, scales with the square of current and dominates at high throughput. A third group, stray loss, covers circulating currents in the tank, clamps, flitch plates and structural steelwork caused by leakage flux. At 110kV and above, stray loss is large enough that manufacturers use magnetic shields or non-magnetic steel in the tank wall to control local hot spots.
Designers express the sum of load and stray loss as the copper loss at rated current and reference temperature, then use it as the input for every thermal calculation. Because the resistance of copper rises by roughly 0.4 percent per kelvin, the measured load loss at 75 degrees Celsius differs noticeably from a cold resistance-based estimate. Acceptance testing therefore requires conversion to a common reference before the figures are compared with the guaranteed values in the specification.
The Thermal Path and the Hot Spot
Heat is produced inside the winding conductors, crosses the paper insulation into the surrounding oil, rises by natural convection to the top of the tank, passes through the radiators and is finally rejected to ambient air. Each step adds a temperature gradient. The standard model recognizes three quantities: top-oil temperature rise, average winding temperature rise and winding hot-spot temperature rise. IEC 60076-2 and IEEE C57.12.00 both set conventional limits, typically 60 K for top oil and 65 K for average winding rise under continuous rated load at a defined ambient, with a hot-spot rise that is higher still because the top of the innermost winding disc sees both the highest local loss density and the warmest oil.
The hot spot is the quantity that actually ages the insulation. It is not directly measurable in a conventional design; it is derived from a hot-spot factor applied to the top-oil rise, or measured with fibre-optic probes embedded in a prototype during a temperature-rise test. Specifying a fibre-optic winding temperature indicator on a large 110kV unit is increasingly common because it replaces a thermal-image estimate with a real measurement and lets the operator exploit genuine thermal headroom instead of a conservative guess.
Cooling Modes and What They Buy
The four-letter cooling designation describes the internal medium, the internal circulation mechanism, the external medium and the external circulation mechanism. ONAN, oil natural air natural, relies entirely on thermosiphon flow and radiator convection. ONAF adds fans that raise the external heat-transfer coefficient, commonly giving a 25 to 33 percent increase in continuous rating. OFAF introduces forced oil circulation, which reduces the temperature drop across the winding and is useful where radiators must be remotely located. A fourth stage, ODAF or OFWF, may be specified where space is constrained or ambient temperatures are extreme.
Each additional stage brings dependency. A fan contactor, a blocked radiator bank or a failed oil pump converts a forced-cooled rating into a natural-cooled one. The control scheme must therefore include automatic derating or tripping logic tied to cooler availability, plus alarm annunciation for loss of flow. A well-written specification requires the manufacturer to state the permissible overload duration after a cooling-stage failure, rather than leaving the operator to discover the limit during an incident.
Loading Guides and the Ageing Rule
IEC 60076-7 and IEEE C57.91 provide loading guides that classify duty into normal continuous, normal cyclic, long-time emergency and short-time emergency. The underlying physics is the thermal ageing of cellulose: the rate roughly doubles for every 6 K increase in hot-spot temperature in the moderate range, and accelerates even faster above 120 degrees Celsius. This is why a modest, short-lived overload is usually harmless while a small but persistent temperature excess quietly consumes years of life.
The practical consequence is that loading decisions should be made against a hot-spot trajectory, not against a simple ammeter reading. A unit that has been lightly loaded through a cool night can accept a significant morning peak because the oil mass is a large thermal reservoir. The same peak applied after eight hours at full load may push the hot spot past the emergency limit. Modern transformer monitoring systems compute this continuously from load current, top-oil temperature and ambient, and they are far more trustworthy than a fixed nameplate rule.
Ambient, Altitude and Installation Effects
Standard temperature-rise guarantees assume a maximum ambient of 40 degrees Celsius and a yearly average of 20 degrees Celsius, at altitudes below 1000 metres. At higher altitude the thinner air reduces convective cooling and also reduces external dielectric strength, so both the thermal rating and the insulation coordination require correction. Solar gain on a tank and radiators exposed to direct sun can add several kelvin to the top-oil reading. Enclosed installations need ventilation sized against the actual heat rejection, which for a large unit may exceed 200 kilowatts at rated loss. Radiator spacing, obstruction from nearby walls and the direction of prevailing wind all matter more than most layout drawings suggest.
Verification in the Factory and in Service
A temperature-rise type test establishes the thermal model. The unit is run under a defined loading method, usually a short-circuit or back-to-back arrangement, until the top-oil temperature stabilizes, with readings taken at fixed intervals. Winding resistance is measured immediately after shutdown and extrapolated back to the instant of switching off to obtain the average winding temperature. Dissolved gas analysis complements the thermal picture: methane and ethane indicate general thermal stress, ethylene suggests a hot spot above roughly 700 degrees Celsius, and carbon monoxide or carbon dioxide point to cellulose involvement. A rising trend in these gases is often the earliest evidence that a cooling problem has developed.
In service, the maintenance focus is unglamorous but effective. Radiator fins must be kept free of dust and vegetation, fan bearings replaced on schedule, oil level and conservator breather condition checked, and the winding temperature indicator calibrated against a traceable reference. Infrared surveys under load reveal blocked radiator sections and loose terminations quickly. Where a unit is approaching its thermal limit, the cheapest capacity increase is often a cooler refurbishment rather than a replacement machine.
Specifying for the Whole Life Cycle
Procurement documents frequently fix only the rated power, voltage ratio, impedance and insulation level. Adding a clear thermal specification is what separates a robust installation from a marginal one. That specification should state the design ambient and altitude, the required cooling stages and their ratings, the maximum permissible hot-spot rise, the loss capitalization formula used to evaluate tenders, and the monitoring provisions. Engineers comparing suppliers often begin with the published type-test dossiers; a transformer intended for a 110kV bay should arrive with a documented thermal model that can be checked against the site's actual load profile, not merely a nameplate number.
Once that model exists, the operator can make defensible decisions about overloads, cooler staging and refurbishment timing. Thermal management is not a one-time design calculation but a continuing discipline that links factory test evidence, online measurement and maintenance records. Units that are specified, verified and operated with that discipline routinely outlast their nominal design life by a wide margin.