How an Africa Colocation Facility Cut Downtime with Online UPS

An African colocation facility cut its unplanned downtime by 91 percent in the first year after replacing its ageing standby units with online double-conversion UPS systems sized to the real load and backed by lithium battery strings. The result came not from a single purchase but from fixing three specific failure points that had been causing repeated, predictable outages for its tenants.

The Starting Point: A Fragile Power Chain

The facility sat on a regional grid with frequent voltage sags and multi-minute interruptions, and its original protection was a bank of standby units that only engaged once mains had already failed. Those units passed raw utility power through to the racks during normal operation, so every sag and brownout reached the tenants' equipment directly, and the transfer to battery took several milliseconds that some servers did not survive.

Tenants were experiencing brief resets and corrupted writes during events the facility considered minor. The pattern repeated weekly, and the operator's support queue was dominated by the same complaint: intermittent, hard-to-prove power quality issues that no single outage log captured cleanly.

Root Cause One: No Conditioning in Normal Mode

The standby topology's defining weakness is that it does nothing to the supply until the supply is already gone. On a weak grid the more damaging events are the sags and spikes that never trip a transfer, and those were reaching the load unchecked. The fix was to move to online double-conversion, where the inverter feeds the load continuously and the output is regulated independent of the input.

After the swap, the same sags that had caused resets simply never reached the racks. The measured count of tenant-impacting events dropped sharply in the first month, before any other change was made.

Root Cause Two: Slow Battery Recharge Between Events

The original sealed lead-acid batteries recharged slowly and, after a cluster of outages in a single week, arrived at the next event partially discharged. A unit that is still recharging is not really protected. Replacing the strings with lithium iron phosphate cut recharge time from hours to minutes and roughly tripled the service life.

The fast recharge mattered more than the extra capacity. In a region with repeated daily events, the ability to be full again within minutes, not hours, is what keeps the site protected through a bad week rather than just a bad day.

Root Cause Three: No Remote Health Visibility

The old units reported little beyond on-battery and low-battery. The operator learned about a degrading battery only when it failed during an event, which is the worst possible moment. The new systems report state of health remotely, so the central team now replaces a weak string on a schedule rather than during an outage.

This turned an invisible risk into a managed one. Across a colocation with many tenants, knowing the true condition of each string is the difference between planned maintenance and a tenant-facing incident.

The Measured Result

Over twelve months the facility recorded a 91 percent reduction in unplanned downtime affecting tenants, measured against the prior year's incident log. The bulk of the improvement came from eliminating the small, frequent events that standby topology had been letting through, with the remainder from faster recovery and proactive battery replacement.

The operator also reported fewer support tickets, higher tenant renewal intent, and the ability to advertise a stronger uptime commitment. The capital cost was recovered within the first year through reduced truck rolls and retained contracts, after which the savings were effectively margin.

Lessons Any Operator Can Apply

• On a weak grid, conditioning in normal mode matters more than transfer speed during a full outage.
• Fast recharge is a resilience feature, not a convenience, where events repeat often.
• Remote battery health reporting turns outages into scheduled maintenance.
• Size the UPS to the real load with headroom rather than to the cheapest frame.

The power architecture behind this deployment, including online double-conversion and lithium storage options for demanding grids, is described at https://www.upsboss.com/data-center/. If your site faces a similar outage profile, the common questions and answers are at https://www.upsboss.com/faq/.

Key takeaway: the 91 percent downtime cut came from fixing three specific weaknesses, not from buying a bigger unit. Online double-conversion conditioned the supply continuously, lithium batteries recharged fast enough to survive repeated events, and remote health monitoring turned battery failure from a surprise into a scheduled task.

Frequently Asked Questions

Would a line-interactive unit have worked as well?
For this grid, no. Line-interactive still passes sags through in normal mode, which were the most common tenant-impacting events. Online double-conversion was chosen specifically because it regulates the output continuously.

Why did lithium matter more than extra capacity?
Because the outages repeated within a single week. Fast recharge, not a larger bank, is what kept the site protected through a cluster of events, and the longer life also removed repeated truck rolls.

How was the 91 percent figure measured?
By comparing the tenant-impacting incident log over twelve months after the upgrade against the prior twelve months, counting events that caused resets, corruption or downtime rather than counting raw utility blips.