When one side drops, the safest default is not to merge the floor onto the remaining UPS. In a redundant architecture, the surviving path is meant to carry the IT load through the equipment’s own redundancy, not through an improvised parallel tie between distribution paths.
What stays untouched
- Do not manually bridge the failed side into the healthy side.
- Do not bypass power conditioning just because capacity looks available.
- Do not treat spare headroom as approval to reconfigure the electrical topology.
What should be verified instead
- Confirm dual-corded systems are already drawing from the live feed.
- Confirm any ATS/STS devices actually transferred within the expected time window.
- Check that the surviving UPS path remains stable and within its intended loading range.
- Check the failed path for the root cause before any restoration attempt.
Why manual intervention is risky
- A fault on the failed side can be carried into the healthy side if the paths are bridged too early.
- Unconditioned or bypassed power can expose sensitive equipment to bad waveforms and transients.
- A transfer that is technically possible is not the same as one that is operationally safe.
For background, see Schneider Electric’s note on changeover and power quality, its UPS configuration overview, DOE’s UPS purchasing guidance, and the Uptime Institute outage analysis.
Data Center Health 