Edited by Robert Repas
Manufacturers test thousands of distribution transformers at worldwide locations each week. The primary incentive is to make sure the transformers meet manufacturing specifications. But there is an economic factor as well. Companies that make large distribution transformers that fail to meet minimum load-loss specifications can face penalties of up to $5,000 for every kilowatt of load loss exceeding the guaranteed value. In addition, a transformer that suddenly jumps in load loss indicates the device is starting to break down and should be replaced before total failure occurs.
Load loss is the small amount of energy consumed by the transformer as it moves main power from primary to secondary windings. It can come from resistance loss in the windings due to load current, magnetic loss due to stray fluxes in the windings, core clamps, circulating currents in parallel windings, and other factors. The term “small” is relative to the overall power that the transformers handle. The U.S. Environmental Protection Agency (EPA) estimates that transformer losses account for 60 to 80 billion kW-hr annually. That’s the equivalent of nine days of U.S. generating capacity annually at a cost of $3 to $4 billion.
It used to be that the way to test electrical qualities of distribution transformers was with motor-generator (MG) sets or tap-changing transformers. But these manual methods of transformer testing are giving way to automated test facilities using solid-state ac-power sources with computerized controls that offer near instantaneous change over test parameters.
A single automated test set can replace a wide range of test equipment including applied test sets, induced step-up transformers, core loss, and load-loss step-up transformers. They can also supply auxiliary power for accessories such as load tap changers (LTCs) or fans that let them test a gamut of power transformers in a multitude of kilovolt-ampere ranges and frequencies. This built-in level of control and flexibility in one instrument translates directly into efficient tests and cost savings.
What makes these new test sets possible today is the advances in high-power solid-state electronics. Solid-state variable-voltage and frequency power supplies have been around for over 30 years. But these units were small, bench-top systems with maximum power output measured in volts-ampere. On the other hand, some distribution transformer tests need kilovolt voltage levels at power levels into the kilovolt-ampere range. The test set must supply these levels for proper testing. For example, the AMX series of solid-state power supplies from Pacific Power has a rated output of up to 12 kVA at frequencies from 20 to 5,000 Hz.
Traditionally, transformer test floors have relied on using several motor generator sets or a combination of sliding contact and under- load tap-changing transformers and motor-generator sets as variable voltage and frequency sources. These sources of power would normally be used to vary the input voltage to an applied test step-up transformer, load-loss step-up transformer or core-loss step-up transformer. However, there are drawbacks to each type of variable-voltage or frequency supply.
MG sets are expensive and can deliver voltage only at one frequency. They also need separate excitation systems. If the excitation system is another small motor-generator set, the voltage stability of the main MG output will be poor. The hunt-and-seek phenomena is a complex interaction between the MG set and transformer load that sets up a voltage/current oscillation. This oscillation is at a lower frequency than the 60-Hz output of the MG set with a net effect of making the generator voltage drift higher and lower without operator intervention. Even though most modern separate-excitation systems are now solid state and the voltage stability is fairly good, MG sets are still large, noisy, and require maintenance of the bearings, input air filter, and brushes. Most are located in soundproof rooms because of difficulty hearing spoken words when these systems are placed on a test floor. Additionally, MG sets generate a significant amount of partial discharge. This partial discharge from the generator commutator brushes makes it difficult to perform RIV/ PD tests without in-line noise filters or large impedances between the motor generator set and transformer winding under test.
These motor-generator sets need large starters and breakers for operation and time to spin up or spin down. An induced motor-generator set is also prone to self-excitation. The self-excitation phenomenon can cause an overvoltage in the generator and transformer under test. A 1,000-kVA, 4,160-V generator failed because of self-excitation while testing a 300-MVA power transformer.
Under-load regulators come in several types. The toroidal variac regulator is used for small loads such as an applied test set. These regulators can be sliding or rolling- contact design where the moving contact runs up and down the face of the coil. The regulator can also be an under-load tap-changing (ULTC) transformer. All of these regulators have a defined limit to the smallest voltage step they can move which is the volts/turn of the coil, typically 10 to 20 V. Ac-power sources can supply voltages with much smaller steps, such as 0.1 V, because there is no physical volts/turn lower limit. The ULTC regulators cannot supply voltage at different frequencies unless a variable-speed motor generator is connected to supply the different frequencies and the core of the regulator is designed tp handle different frequencies.
These regulators need mechanical maintenance of the moving contacts. Typically, this means annual or semiannual inspections for ULTC regulators. However, sliding contact transformers need more frequent inspections. Regulators also need breakers on the input side to minimize damage should a fault occur on the output side of the transformer. The low-impedance design of a regulating transformer minimizes voltage drop across the regulator under full load. This makes the regulator susceptible to damage if not protected by a correctly sized breaker.
The sliding contact and toroidal regulators can generate a significant amount of partial discharge because of their moving brushes, again interfering with PD/RIV testing. Regulators also take time to run up and down their voltage range whereas solid-state ac-power sources can reach a given voltage in a matter of seconds. This lets users check voltages and currents quickly while ramping up and turn the voltage off instantly once a test is complete. This is very important when performing load loss so that the resistive heating caused by the test does not throw off the losses measured during the test.
When using plant power feeds, sliding contact, or under- load tap-changing transformers a manufacturer must add high-voltage/high-current capacity breakers. The reason for this is that motor-generator sets and plant power feeds can supply tremendous amounts of fault current should a transformer under test fail. Test-floor failures of transformers are relatively frequent occurrences. A distribution- transformer plant may have one or more per day while a large power transformer plant may experience one every month.
These high current breakers are expensive and inherently dangerous in a test-floor environment because they are designed for utility operation, not test-floor usage. A typical test floor may open and close a circuit breaker 10 or more times per day. In a utility environment, this same breaker may toggle only several times a year. Mechanical wear along with dirt buildup on vacuum, air, and oil breakers can cause external flashovers and explosions. For this and other reasons, most test floors require more than one of these breakers in series in case one fails. These breakers require maintenance and inspection every few thousand operations. Solid-state ac-power sources do not require these breakers and can be fed from and feed out to standard breaker panels. The breakers in these panels do not need to be operated because the power supply can be turned on and off with a computer command and only serve as protection in the event of a wiring short. The only requirement for the solid-state ac-power source is a set of contactors so that the power-supply output can be switched to power up the correct test equipment.
These solid-state ac-power sources offer new technical and speed advantages for transformer test floors. Electronic protection and shutdown circuits make the units immune to load-induced damage caused by transformer failures on the test floor. In many applications, the power sources are “hardened” to withstand ground transients generated during high-voltage lightning impulse tests. Additionally, these power sources have the advantage of thermal/electronic tripping that limit the amount of short circuit current they deliver in the event of a test failure. Limiting short-circuit current makes core loss failures less catastrophic and eliminates the need for high-voltage/ high-current breaker protection on the test floor.
Power sources may be paralleled to boost total kilovoltampere during initial facility design, or they can undergo expansion later in the field. Paralleling requires that the power supplies each share part of the load. This is usually accomplished with special connections between the supplies to balance the power levels. Paralleling offers the test-floor designer flexibility in sizing the floor for existing power demands, while providing the means to expand future test capacity. When used with a fully compensating capacitor rack, the units have no practical limit to the size of power transformer under test. The solid-state power sources needs to supply only the real kilowatt losses by the transformer under test.
A typical test floor may use the 13,800-V 3Φ plant feed to power the entire test floor through a 1,500-kVA step-down transformer. The output of this step-down transformer is usually 480 V. The 480-V power runs through a 2,000-A main breaker in a panel with sixteen 100-A breakers. The 100-A breakers feed the ac-power sources of the test sets.
The 0-to-208-V output of the ac source feeds a distribution panel with sixteen 250-A breakers and a 4,000-A main automatic breaker. The output of the distribution panel routes through three load contactors that power the main step transformer, applied test set, and auxiliary loss transformer. The fully compensating capacitor rack is directly connected to the transformer under test.
Variable-frequency operation is possible because of solid-state power sources that can change their frequency of operation through a control panel or computer setting. The same ac source can perform either 50, 60, or 400-Hz core loss tests by sending it the proper computer command. Another computer command transitions the power source from core loss frequency to induced voltage test frequency and then ramps up the voltage. The variable frequency power supply lets the operator or computer program find the minimum excitation current by varying frequency. As stated earlier, what sets these supplies apart from their earlier brethren is their power-output level.
Determining the minimum excitation current is important because all transformers become capacitive loads at some frequency. By programming the solid-state power source to shift frequencies, it’s possible to determine which multiple of the core-loss frequency draws the lowest current. For example, an LV:480-V, 500-kVA transformer needs an induced test per ANSI standards to approximately 2× rated voltage. The solid-state power source changes from 60 to 120 Hz and then ramps the frequency up from 120 Hz until the minimum excitation current is found. As test frequency rises, the excitation current drops until the capacitive current of the transformer becomes greater than the magnetizing excitation current. At that point excitation current begins to rise as the now capacitive transformer begins to conduct.
To vary the frequency of an MG set the motor rpm had to change — higher rpm means a higher frequency. However, higher rpm also meant the output voltage of the generator would rise as well. This new control method eliminates the dangers of an overvoltage condition on the unit under test when the capacitive-transformer load self excites a motor generator. Only variable-frequency solid state power supplies let an operator perform the test this way, improving safety.
Because of output capacitive filtering, ac-power sources generate less than 5 pC of partial discharge (PD) so that they will not interfere with the PD and radio interference voltage (RIV) testing of transformers. Motor-generator sets often require the use of output PD/RIV filters so that the PD/RIV measurements made at the transformer does not pick up the generator brush contact PD noise. All of the frequency and voltage changes required to meet the ANSI/IEEE or other customer-specified test requirements are programmed on a host computer. This process eliminates operator intervention and significantly shortens test time.
Test floors can now be automated because the operator has full control of the test voltages and frequency through a computer and the electronic power supply remote- control unit. Therefore, standard tests such as load loss, core loss, induced voltage, applied voltage, heat run and soak tests (extended core loss tests) can be fully automated requiring no operator intervention during testing. The operator has only to connect the test leads and the computer-controlled solid-state ac-power sources can ramp up check voltages and currents against target values and query measuring equipment to take final measurements. Test reports with these values can then be printed and sent with the transformer to the customer. Without the ability to accurately and completely control the power supplies, the operator would have to manually ensure phase-to-phase voltage and current balance. In the future, whether they test motors, transformers, breakers, or other power equipment, test floors will use solid-state ac-power sources to automate standard load loss, core loss, applied, induced, and heat-run tests.