What Happens When the UPS Fails Before the Generator Starts?

SSee · The invisible problem

The power path looks complete. Almost nobody has traced what depends on it.

JPS feeds an automatic transfer switch. A generator stands behind it. A UPS and battery bank protect the server room, and a branch point-of-sale terminal trades against what runs in that rack.

On the equipment schedule every box is ticked. What the schedule doesn't show is the chain of services the rack carries — and which parts of the power path they actually depend on.

• Equipment present — utility supply, ATS, generator, UPS, batteries and cooling all exist
• Dependencies unmapped — authentication, database, application and POS all sit behind one UPS output
• Cooling is on a different bus — the room air handler is not on the UPS

AAssess · The business consequence

Each unproven link is a small risk. Together they decide whether you trade during an outage.

A battery bank ages whether or not it has ever been needed. Generators start more slowly than the brochure says. Transfer timers are set once and forgotten. Cooling stops when the building bus drops.

None of this fails during normal operation, which is why it survives audits. It fails in the minutes that matter, and then the cost isn't the repair — it's the lost trading and the recovery work behind it.

• Degraded batteries — capacity falls with age and heat, so real autonomy can be a fraction of the design figure
• Generator and transfer delay — start-up, stabilisation and the ATS timer add up to a gap the UPS must cover
• Cooling off the UPS — servers can run hot during a long gap even if power returns quickly
• No test records — maintenance history missing or old means every figure above is an assumption

VValidate · The engineering evidence

A controlled failure test turns assumptions into a measured result.

Innovera validates continuity by testing the whole chain against the services that matter, under an approved procedure and with qualified personnel — not by inspecting each box on its own.

Run the tests below. In the as-found site the UPS rides through a short interruption, then fails in a longer outage before the generator can take over. Watch what the rack and the branch POS do.

• Test the services, not the equipment — a UPS can report healthy and still not carry the load for long enough
• Power gap ≠ outage length — services restart in dependency order, so a short gap stretches into a long outage
• Repeat after changes — the same test on the remediated site is the proof of improvement

≈ 3 min 45 srecovery time from a short power gap (as-found, synthetic) — from rack power loss to POS trading

EEvolve · The better outcome

Corrected, monitored and re-tested: the same outage becomes an event nobody notices.

Remediation is usually modest: replace degraded batteries, correct the transfer timing so the UPS covers the generator start, put battery and load monitoring in place, keep maintenance records, and schedule the test.

The point is not the equipment spend. It is that the next utility loss runs through a procedure you've already rehearsed, with a measured result you can show to the board and the insurer.

• Autonomy exceeds transfer time — battery sized and verified to cover generator start with margin
• Tested transfer — ATS timing confirmed under load, with records
• Monitored and maintained — battery health and load visible before they become a failure

0 sservice interruption in the same test (illustrative) — not a verified customer result

Outcomes shown are illustrative and not verified customer results. Timings, battery autonomy and recovery order are synthetic and time-compressed. A real test is designed around your site, load and approved procedure, and performed by qualified personnel.