Most generating sets are designed and specified at a power factor of 0.8, and the engine is therefore not capable of delivering full kVA at unity power factor. For example a 500 kVA gen-set rated at 0.8 power factor, would only be able to deliver 400 kW into a purely resistive load. Testing using a resistive only load will usually result in full load test of the prime mover (i.e. the engine), but not of the alternator, which will be tested to only 80 % of its rated current. This means that the alternator and its control system are not tested to their rated limit.
A non-unity power factor load affects the way that an alternator responds to load because, with inductive loads, the load current is not exactly in phase with the output voltage. The field within the magnetic circuit of the alternator is distorted and the automatic voltage regulator (AVR) and excitation circuit must provide a higher current to maintain the set output voltage. So the relative losses within the alternator increase when operating at non-unity power factor resulting in more heat dissipation within the alternator laminations and windings. However, that same alternator would run significantly cooler if the generating set is tested solely at unity power factor. This is because the current is lower, and is exactly in phase with the voltage (i.e. unity power factor). So the thermal performance of the generating set as a whole will not be tested as it would if the rated, non-unity power factor load were applied.
Many engineers who test generating sets consider that this is not very important, since usually the alternator is of proven design. Their main concern is to prove that the prime mover is in a serviceable condition, and is able to accept load without instability, or even stalling. There is no doubt that resistive-only tests do give valuable and useful data, but they cannot give the whole story. The electrical parts of the generating set, the alternator and ancillary components such as circuit breakers, current detectors, connections and wiring, metres and instrumentation, are clearly not being tested to their limit when a resistive-only test is done.
When professional engineers and consultants are involved in specifying a back-up power system for a project, they specify that a generator set is tested to the applicable standards, and at the nameplate rating. This means non-unity, or resistive/inductive load testing. ISO 8528 specifies that test reports should note if tests have been done at a power factor which is different than the rated one. Usually this means that tests done with a purely resistive load can be considered incomplete.