Most temporary-worksite power outages are visible in the data before the breaker trips. Voltage drift, phase imbalance, and slow load creep on one generator show up as trends an hour earlier. Real-time monitoring catches the three preventable failure modes in time to act. It will not see downstream of a 110V CTE transformer, and an honest platform says so.
STAT
Power outages remain, in the Uptime Institute’s words, common, costly, and preventable. In its 2024 analysis, 54 percent of serious outages cost more than 100,000 dollars and around one in five cost more than a million, with power among the leading root causes. The pattern on a worksite is the same at smaller scale: the warning was usually there. Source: Uptime Institute, Annual Outage Analysis 2024
Key takeaways
| Topic | In brief |
|---|---|
| Outages are predictable | Voltage drift, phase imbalance and load creep show as trends before the trip. |
| Three failure modes | Overload and phase imbalance, voltage drift, generator load mismanagement. |
| Alarms beat logs | Logging without predictive alarms only helps the post-mortem. |
| Honest limits | Board monitoring does not see 110V CTE tool circuits; that stays with manual testing. |
Why are most worksite outages predictable?
When a temporary distribution board trips, the data was almost always there an hour earlier. Voltage drift, phase imbalance, slow load creep on a single generator as the heating kicks in for the evening. The pattern is visible if someone is looking at it. The breaker trips because nobody was.
That is the case for real-time monitoring on temporary worksite networks. Outages are predictable, and the early signs are measurable. The problem is that nobody is in front of the screen to see them, which is exactly what an alarm fixes.
What are the three preventable failure modes?
After enough projects, the same three classes of outage recur.
Overload and phase imbalance. Three-phase systems get loaded unevenly when single-phase equipment is plugged in unevenly across phases. Voltage on one phase drops, current rises, and the protection eventually catches it. The early signal is phase imbalance crossing roughly 10 percent for more than a few minutes during a load increase.
Voltage drift. A generator running at the edge of its capacity drifts outside the roughly plus or minus 10 percent window that motorised equipment needs. Under-voltage damages compressors. Over-voltage damages electronics. Both show as a slope on a chart before either becomes a trip.
Generator load mismanagement. Two generators on site, one at 90 percent load, one at 20 percent. The 90 percent machine fails first when the next load comes on, and the 20 percent machine is burning fuel inefficiently. Load balancing between generators is the single intervention with the largest fuel saving on most builds.
| Failure mode | Early signal | Intervention |
|---|---|---|
| Phase imbalance | Imbalance over ~10% during load rise | Move single-phase loads across phases |
| Voltage drift | Voltage sloping outside ~±10% | Rebalance or add capacity before the trip |
| Generator mismanagement | One genset near capacity, another idling | Balance load between generators |
What does a monitoring platform actually need to do?
Logging the data is the baseline. Logging without predictive alarms is logging for the post-mortem.
The platform that pays back its cost is the one that:
- Sends an alarm when phase imbalance exceeds a configured threshold for a configured duration.
- Sends an alarm when voltage drifts outside the safe band, before the protection trips.
- Shows generator load against capacity continuously, so the operator can rebalance before a generator hits its limit.
- Logs every fault with timestamp, current draw, voltage at fault and local conditions, and keeps it for the project lifecycle, so the post-mortem is a five-minute review rather than a half-day investigation.
If the alarm reaches the right phone, the on-call competent person can act before the trip. That is the difference between monitoring and observation.
What can real-time monitoring not see?
There are honest limits worth stating up front. A board-level monitoring platform will not see downstream of a transformer to a 110V CTE tool circuit. That is by design. The CTE side exists to isolate the operator from the supply, and the monitoring at the board is on the supply side. Faults at the tool stay in the domain of the manual test routine and the competent person’s inspection.
A platform that claims to monitor downstream tool circuits directly is either using a different topology than UK practice or over-claiming. A fair question on any vendor demo: show me the live current on a 110V CTE tool circuit at the tool end, not at the transformer. If they cannot, take that as honesty.
When does monitoring pay for itself?
The first time a site avoids a Saturday evening outage that would have cost three crews half a shift, the platform has paid for its quarter. The cumulative value is harder to measure: every fault prevented, every generator that ran at the right load instead of 90 percent all weekend, every hour of build time the data logger turned into an export rather than a phone call asking what went wrong.
The case is sharpest where the schedule cannot move. An outdoor event has no second Saturday, so the same three failure modes carry a much higher cost for the same electrical fault. The planning that follows from that is covered in our checklist for temporary power supply at UK outdoor events.
Real-time monitoring is not a guarantee against outages. It is the difference between an outage that catches you and one you saw coming and prevented.
A breaker trip is a data point you ignored an hour too long. The job of monitoring is to move that data point from the post-mortem to the alarm.TSR-Elsite
Summary
Most worksite outages are predictable from voltage, phase and load trends, and the three recurring failure modes all announce themselves before the trip. A monitoring platform earns its place when it turns those trends into alarms that reach a person in time, and when it keeps the fault history for the project lifecycle. It will not see the CTE tool side, and that is fine. If you want to walk through what real-time monitoring looks like on a live distribution board, see Spine for rental fleets or get in touch: sales@tsr-elsite.fi or +358 9 4555 588.
Frequently asked questions (FAQ)
Are worksite power outages really predictable?
Most are. The common causes, phase imbalance, voltage drift and generator overload, develop as measurable trends minutes to an hour before the trip. They are predictable in principle; the gap is that nobody is watching the trend, which is what alarms solve. Exception: a sudden hard fault, like a cable cut by a machine, gives no warning.
What is the difference between monitoring and alarms?
Monitoring records what happened. Alarms tell a person something is going wrong now. Logging without alarms only helps the post-mortem. The value is in a threshold breach reaching the on-call competent person’s phone in time to rebalance or shed load before the protection trips.
Can monitoring see a fault at a 110V tool?
No. Board-level monitoring sits on the supply side of the CTE transformer and sees supply-side faults. A fault at the tool stays with the manual test routine and competent-person inspection. Treat any claim to monitor CTE tool circuits directly with caution.
What is phase imbalance and why does it matter?
It is uneven loading across the three phases when single-phase equipment is plugged in unevenly. The overloaded phase heats and its voltage drops, eventually tripping. Catching imbalance over about 10 percent early lets you move loads in a five-minute job instead of losing a shift.
Does monitoring reduce fuel use?
It can, mainly through generator load management. Keeping each generator in its efficient band and shedding idle machines is where the saving sits. Monitoring provides the live load-versus-capacity view that makes that possible. The platform measures it; the operator acts on it.
How long should fault data be kept?
For the life of the project at least. The Spine platform records all measurable data throughout the history of the site, so a fault review is a quick export rather than a reconstruction from memory.