In short. Transient response is how a generating set behaves in the seconds after load is applied or removed: how far voltage and frequency deviate, how long they take to settle, and whether anything trips. Voltage behaviour is governed by the alternator and AVR, frequency behaviour by the engine and governor, and the two interact. Specify it by ISO 8528-5 performance class and the actual load steps, and prove it with a recorded step test at commissioning.
What happens during a load step
When a block of load closes onto a running set, several things happen within about one second. Current rises instantly and the alternator’s internal reactance drops the terminal voltage. The AVR increases field current, but the rotor winding is highly inductive, so the voltage recovers over a few tenths of a second. At the same time the electrical torque on the shaft rises, the engine slows, the governor opens the fuel rack, the turbocharger spools up, and frequency recovers over one to several seconds. Load rejection is the mirror image: voltage and frequency overshoot and settle back.
The two responses are linked. A deep voltage dip reduces the power drawn by resistive loads and the torque of motors, which momentarily eases the engine. A slow frequency recovery lowers voltage further on machines whose AVR has under-frequency roll-off, a deliberate protection that reduces voltage when speed falls. Reading one trace without the other misleads, which is why both are recorded together. The voltage dip and frequency dip guides take each part in turn.
What each component contributes
| Component | Governs | Design choices that improve response |
|---|---|---|
| Alternator | Depth of voltage dip, sustained fault current | Lower transient reactance, larger frame on the same engine, 2/3 pitch winding |
| Excitation and AVR | Voltage recovery time, stability, behaviour under distortion | PMG or auxiliary-winding excitation, digital AVR with correctly tuned gain and stability |
| Engine | Load acceptance, frequency dip, smoke | Adequate rating margin for the step, turbocharger matched for transient duty, warm engine |
| Governor | Frequency recovery, droop or isochronous operation, hunting | Electronic governor tuned on site for the actual load |
| Rotating inertia | Initial rate of speed fall | Heavier flywheel where specified by the manufacturer for the duty |
| Load and controls | Size and order of steps | Staged transfer, soft starters, restart delays, UPS settings matched to generator |
A package is only as good as its weakest contributor. We have seen excellent alternators let down by a sluggish mechanical governor, and well-governed engines paired with high-reactance alternators that dip 30 percent on a modest motor. The AVR guide covers the excitation side.
Specifying transient performance
Specify response in two parts. First, name the ISO 8528-5 performance class: G2 for general industrial and commercial plant, G3 where UPS, telecoms, medical or drive loads are significant. The class fixes the permitted transient deviation and recovery time for a defined step. Second, state the actual load steps the site will impose, including the largest motor start with load already on the set, the transfer block load and the UPS with battery recharge, and require the supplier to show from manufacturer data that each stays within class. A specification that only says “voltage dip not more than 15 percent” without a step size is unenforceable, because any set meets it for a small enough load.
Typical class limits
Under class G2 the transient frequency deviation on load increase is held to around 10 percent with recovery within about 5 seconds, and transient voltage deviation to around 20 percent with recovery within about 6 seconds. Class G3 tightens these to roughly 7 percent and 3 seconds for frequency and 15 percent and 4 seconds for voltage. The figures are typical of the standard’s structure and should be confirmed against the current edition; what matters is that the class, the step and the acceptance values are written down before the order is placed.
Testing and acceptance
Transient performance is demonstrated, not assumed. A proper test uses a resistive load bank, or a resistive-reactive bank where the site load is motor-heavy, applied in defined steps from defined initial conditions: typically no load to 25, 50 and 75 percent, then the largest site-specific step, and the corresponding rejections. A power quality recorder captures voltage and frequency at a cycle or better, and the record notes temperatures, initial load, step size, minimum and maximum values, recovery time within the agreed band and any alarms. Repeat with the actual plant load where possible, because a load bank does not reproduce motor starting or UPS behaviour. Agree acceptance criteria in writing beforehand, and record any governor or AVR settings changed during testing so the demonstrated result matches the installed configuration. The load bank testing service and the commissioning checklist describe the procedure.
Paralleled sets and transient sharing
Where two or more sets share a bus, transient response also decides how a step is divided between them. Sets with different governor response take unequal shares for the first second, and sets with different AVR response share reactive load unequally, which can trip a machine on reverse power or overcurrent even though the total is within capacity. Matching settings and using load-sharing controls designed for the mix of sets keeps the transient within limits, and a set still recovering from a step cannot accept a partner because synchronising needs stable frequency. The kW and kVAr load sharing guide covers the controls.
A worked example (illustrative)
A pharmaceutical formulation unit near Hyderabad specified two 750 kVA sets to class G3 because its HVAC drives and a 300 kVA UPS fault on voltage excursions beyond 12 percent and frequency beyond 3 percent. The supplier’s data showed each set accepting a 55 percent step within G3 from no load. The site’s transfer scheme, however, closed 480 kW onto a single set before the second had synchronised, a 78 percent step. During commissioning the recorded frequency fell about 9 percent and the UPS went to battery. The remedy was in the controls: the ATS was re-sequenced to close the HVAC bus first, then the UPS bus 8 seconds later, and the second set was programmed to start on mains failure rather than on load demand so that it was ready to synchronise as the second block arrived. Retested, the largest step seen by one set was 42 percent, the frequency deviation about 5 percent and the voltage deviation under 10 percent, and the UPS stayed on generator. The generators were adequate; the sequence had been the problem.
Common mistakes
- Specifying a dip percentage without a step size or performance class, so the requirement cannot be checked or enforced.
- Judging response from a panel meter during commissioning, which averages the transient and hides excursions that drives and UPS respond to.
- Tuning the AVR or governor to pass the test and not recording the settings, so a later service visit restores defaults and the problem returns.
- Testing only on a resistive load bank and never on the real load, so motor starting and UPS interaction are first seen during an actual outage.
What to send us
Send the generator and alternator nameplates, the controller, AVR and governor types, a single-line diagram showing how load is transferred and in what steps, the voltage and frequency tolerance windows of any UPS and large drives, and any alarm logs from previous events. For a new set, send the load inventory grouped by transfer step and the performance class you believe you need. Send your application engineering enquiry.
Questions engineers ask
What is the difference between transient response and steady-state regulation?
Steady-state regulation is the voltage and frequency accuracy once the set has settled, typically within 1 percent for voltage and 0.25 to 1 percent for frequency on an electronic governor. Transient response is the excursion and recovery in the seconds after a load change. A set can have excellent regulation and poor transient response.
Which ISO 8528-5 class do I need?
G2 suits most industrial and commercial plant. G3 is appropriate where UPS, telecoms, medical equipment or sensitive drives are a significant part of the load. G1 is for simple resistive loads and G4 is for special cases agreed with the manufacturer.
Can AVR and governor tuning fix a poor transient response?
Tuning can shorten recovery and remove hunting, and it should be done on site with the real load. It cannot reduce the initial voltage dip set by alternator reactance or create load acceptance the engine does not have, so it complements correct selection rather than replacing it.
Why does my generator pass the load bank test but trip on real load?
A resistive load bank applies unity power factor steps with no inrush. Real plant applies motor starting current at low power factor, UPS harmonic current and a transfer block load, which stress the alternator and engine differently. Test with the actual load sequence as well.
Technical content reviewed by D. Srinivas Reddy, Managing Partner and Service Engineer, Reddy Generators. Last updated 19 October 2026.