In short. Restoring everything at once can undo the recovery. A good restart sequence returns loads in stages sized to what the source can accept, waits for recovery between stages and checks that each step has worked before moving on.
Principles
- Restore the most critical loads first.
- Limit each step to the source’s block-load capability. See load-step response.
- Wait for voltage and frequency to settle between steps.
- Delay motors until residual voltage has decayed.
- Start the largest motors when there is most capacity available.
- Provide manual hold for loads that need operator checks.
Sample stages
| Stage | Loads | Why |
|---|---|---|
| 1 | UPS, control power, emergency lighting | Safety and visibility first |
| 2 | Ventilation, fire and life safety pumps | Protect people and the building |
| 3 | Cooling and process support | Prevent damage |
| 4 | Large motors, one at a time | Controlled starting |
| 5 | Comfort and secondary loads | Only when capacity is confirmed |
Permissives for each step
- Source stable for a set time.
- Previous step completed and settled.
- Available capacity above the step plus margin.
- Load not locked out or in maintenance.
- Process permissive, if any.
Transformers and capacitors
Energise transformers one at a time to limit inrush. See inrush studies. Reconnect power-factor correction last, if at all, on generator supply.
Verification
The sequence should check that each step produced the expected state and abort or hold if frequency or voltage leaves limits. Record the time for each step so the sequence can be tuned. See state machines.
Illustrative example
After a blackout at a hospital wing, the generator takes the emergency bus in 10 seconds. The sequence restores stage 2 after 10 seconds, stage 3 after another 15, and starts the chillers one at a time at 20 second intervals. One start is held because frequency dips below the limit, and resumes after recovery.
Restoring is as careful as shedding
After an outage, every motor wants to restart together. The inrush from simultaneous starts can be several times the running load and can collapse a generator or trip the supply again. Restoration is sequenced by priority and by electrical limits.
Restoration logic
- Verify the source is stable: voltage and frequency inside window for a set time.
- Restore the highest priority group first.
- Wait for current and frequency to settle before the next group.
- Start large motors one at a time, with soft starters where fitted.
- Stop adding load if frequency or voltage falls below a threshold.
Example (illustrative)
| Step | Group | Delay after previous |
|---|---|---|
| 1 | Lighting, controls, IT | 0 s |
| 2 | Fire pumps, lifts | 5 s |
| 3 | Process pumps, one by one | 10 s each |
| 4 | Chillers and compressors, one by one | 30 to 60 s each |
| 5 | Comfort HVAC | After generator loading is below the set limit |
Compressors and process loads
Some loads need pressure to bleed down or a minimum off time before restart. Set restart delays accordingly. See the process-specific equipment guidance.
When mains returns
Transfer back also causes a momentary outage or paralleling. Re-sequence the restart after transfer, not all at once.
See load shedding automation and motor starting.
The master engineering question
At each step, is there enough capacity and stability to accept the next block, and what stops the sequence if there is not?
What to send us
Send your load list and start methods. Send your application details.
Questions engineers ask
Why not restore everything together?
The combined start surge may stall the generator.
How long between steps?
Long enough for voltage and frequency to recover, usually seconds.
Should restoration be automatic?
For critical loads yes; some loads need operator approval.
What after a shed event?
Hold shed loads until capacity returns. See load shedding.
Values are illustrative. Tune on site.