In short. Load shedding removes lower-priority loads so the remaining sources stay stable. It must act faster than the plant collapses, shed the right amount in the right order and never shed life-safety loads. Tests with real source trips prove it works.
Types
| Type | Trigger | Speed |
|---|---|---|
| Fast event-based | Loss of a source or breaker, known in advance | Tens of milliseconds |
| Slow overload-based | Measured overload or limit exceeded | Seconds |
| Under-frequency | Frequency falls below thresholds in stages | Fractions of a second to seconds; backup to the others |
| Under-voltage | Voltage falls in a weak system | Context dependent |
Priority tiers
- Life safety and emergency systems: never shed.
- Critical process and IT.
- Essential mechanical systems.
- Production loads that can pause.
- Comfort and non-essential loads: shed first.
See load prioritisation.
Shed table
For each credible event, such as the loss of generator 1 or utility, the shed table lists which loads to trip, by how much power and in what order. Its size must match the capacity left and the real-time load, which is why it relies on metering of the shed feeders.
Design checks
- Shed capacity is greater than the shortfall for the worst event.
- Breakers and contactors that shed are fast and reliable.
- Tripped loads do not restart automatically until allowed. See restoration.
- Under-frequency relays back up the PMS.
- Shedding is tested with a real or simulated source loss.
See also load shedding.
Illustrative example
A plant runs three 1,000 kW generators with 2,200 kW load. If one trips, the remaining two (2,000 kW) are overloaded by 200 kW. The shed table trips 300 kW of chillers and process pumps within 100 ms, frequency stays above its limit, and the standby set takes over in 20 seconds.
Why shedding must be automatic
When a generator is overloaded, frequency falls within seconds, protections trip and the whole bus can black out. Operators cannot respond in that time. Automatic load shedding removes selected loads quickly so the remaining supply survives.
| Trigger | Speed | Use |
|---|---|---|
| Static (capacity-based) | Fast, on loss of a source | Shed before frequency falls, based on available kW |
| Dynamic (frequency-based) | Hundreds of ms | Backup when something unexpected happens |
| Rate of change of frequency | Fastest on large disturbances | Large step loads or source trips |
Priority table (illustrative)
| Priority | Load class | Shed order |
|---|---|---|
| 1 | Comfort HVAC, water heating | First |
| 2 | General process, auxiliary pumps | Second |
| 3 | Important process | Third |
| 4 | Critical and life safety | Never |
Capacity-based logic
Shed required = connected load − available generation × safe loading factor
If three 500 kW sets are online with 1,400 kW load, and one trips, available is 1,000 kW × 0.9 = 900 kW, so at least 500 kW must be shed immediately.
Design points
- Use hardwired signals for breaker status and trip commands so shedding is fast.
- Avoid restoring load too soon, since it can overload again.
- Test with real breakers at commissioning.
See load shedding and load restoration.
The master engineering question
For each credible loss of source, will the load that is shed arrive before frequency falls to the point of collapse, and is it the load the site would choose?
What to send us
Send feeder list, priorities and generator data. Send your application details.
Questions engineers ask
Is under-frequency shedding enough?
Often not, because it reacts after frequency has already fallen. Use it as a backup.
How much load to shed?
Enough to restore balance, plus margin, calculated for each event.
Who sets priorities?
Operations and safety owners, reviewed by engineering.
How is it tested?
By simulated or controlled trips with monitoring.
Values are illustrative. Shed tables must be project-specific.