In short. Real power is shared through the governors. With droop control, each set lowers speed slightly as load rises, so sets settle at shared loads. With isochronous sharing, a signal or network tells each set its share while holding frequency. Settings must match across sets.
Droop sharing
Each governor has a speed-load characteristic with a small droop, typically a few percent. Two sets on the same bus run at the same frequency, so each carries the load where its characteristic crosses that frequency. Equal droop and rating give equal per-unit sharing.
Droop % = (fno-load − ffull-load) ÷ frated × 100
Isochronous sharing
Frequency is held constant. A load-sharing line or digital network communicates each set’s load so they hold equal percentages. If communication fails, sets should fall back to droop, not drift. A single set can be isochronous while others are in droop in some arrangements, but the combination needs care.
Settings that matter
| Setting | Effect |
|---|---|
| Droop percentage | Sharing accuracy and stability |
| Speed reference or bias | Load transfer between sets |
| Ramp rates | Smooth loading and unloading |
| Gain and stability | Hunting versus slow response |
| Load-sharing signal calibration | Equal sharing across different sizes |
Common problems
- Unequal sharing: different droop settings, governor faults or engine condition.
- Hunting: gain too high, mechanical lash, or interaction with the voltage control.
- Reverse power: a lightly loaded set driven by others; trips on protection. See reverse power.
- Sharing drifts with load: check calibration.
Testing
- Parallel sets at no load, check frequency match and stable operation.
- Step load in stages (25, 50, 75, 100 percent) while recording kW per set.
- Check sharing error at each step, at steady state and during transients.
- Test with one set unloaded manually, and with communication removed.
See kW and kVAr load sharing and synchronising switchgear.
Illustrative example
Two sets of 800 kVA share a 1,000 kW load. Set A carries 620 kW and Set B 380 kW. Testing finds Set B’s droop set at 6 percent versus 3 percent on Set A. Matching droop settings brings sharing to within a few percent.
Why load sharing is a control problem
Two generators on one bus have one frequency. How real power divides depends only on how their speed governors respond to that frequency.
| Method | How | Strength | Weakness |
|---|---|---|---|
| Droop | Frequency falls with load by a set percentage | Simple, no links | Frequency varies with load |
| Isochronous with load-share lines | Controllers exchange a load signal | Constant frequency | Depends on the link |
| Digital network sharing | Messages on a dedicated bus | Flexible, diagnostics | Network integrity |
Droop arithmetic
Share of load ∝ rating ÷ droop percent, for sets in parallel at the same frequency
Two sets of 500 kW and 1,000 kW with equal 4 percent droop share a 900 kW load as 300 kW and 600 kW, each at 60 percent. If the larger set has 3 percent droop, it takes more: ratios become 500/4 : 1000/3 = 125 : 333, so about 245 kW and 655 kW.
Setting up
- Match droop settings to rating.
- Test sharing at 25, 50, 75 and 100 percent load.
- Confirm the differences are within a few percent of rating.
Faults to find
- Load oscillation: gains too high or load-share line noise.
- One set always lightly loaded: different droop or speed bias.
- Reverse power on one set: speed setpoint too low.
See kW and kVAr sharing and multi-DG load sharing.
The master engineering question
If the load changes or a sharing link fails, will every set keep sharing within its limits without a trip?
What to send us
Send governor and controller types. Send your application details.
Questions engineers ask
Can different sizes share?
Yes, in proportion to rating, with matched droop in per-unit terms.
Why does one set always carry more?
Check settings, governor condition and measurement calibration.
What is a load-sharing line?
An analogue signal that lets sets compare load and adjust speed.
Does kW sharing affect kVAr?
They are controlled separately, though poor tuning can interact.
Values are illustrative. Follow manufacturer settings.