The Part of the Project Nobody Wants to Talk About
Every data center project has a hero phase and a quiet phase. The hero phase is design the renderings, the load calculations, the vendor selection meetings where everyone feels like they’re building something impressive. The quiet phase is commissioning, and it’s where most of the actual risk in a project lives.
Ask anyone who’s run a commissioning schedule and they’ll tell you the same thing: the equipment almost never fails on its own. A transformer that passed factory testing usually works. A chiller that met spec on the data sheet usually cools. What actually causes the 2 a.m. phone calls, the missed go-live dates, the change orders nobody budgeted for it’s the space between the equipment. The handoffs. The assumptions two different engineering teams made about each other’s scope, three months apart, without ever getting on a call.
This isn’t a theoretical problem. It’s the most common, most avoidable, and most under-discussed source of delay in data center builds.
A Scenario That Plays Out More Often Than You’d Think
Picture a mid-sized colocation facility, six months from go-live. The electrical contractor has commissioned the switchgear. The mechanical contractor has commissioned the cooling plant. Both reports say “pass.” Both teams sign off. Everyone moves to integrated systems testing feeling good about where things stand.
Then, during a simulated utility failure test, the generator picks up load exactly as designed but the chillers, which were sequenced to restart on a staggered timer to avoid inrush current spikes, restart before the switchgear has fully stabilized its output. The failure sequencing logic the thing that tells each system when to act relative to the others was written by two different engineers who never compared notes, because in the org chart, “electrical” and “mechanical” were separate scopes with separate sign-offs.
That’s not a rare edge case. It’s close to the default outcome when power and cooling systems are engineered by parties who aren’t required to talk to each other until integrated testing which, by definition, is too late to be cheap.
Why This Keeps Happening
It’s not that engineers are careless. It’s that the industry’s default sourcing model actively works against coordination.
Most data center projects are still built the way construction projects have been built for decades: separate RFPs for electrical equipment, mechanical equipment, and controls, awarded to whichever vendor comes in lowest on their piece. Each vendor optimizes their own scope against their own spec sheet. Nobody on any of those teams is incentivized or sometimes even permitted, contractually to flag that their sequencing assumption might conflict with another vendor’s.
The result is that integration risk gets pushed to the very end of the schedule, where it’s most expensive to fix, discovered by the people with the least ability to change the original design.
What Actually Reduces This Risk
There’s a version of this problem that’s genuinely hard to solve physics, thermal margins, grid instability. This isn’t that. Commissioning surprises caused by sequencing and integration gaps are largely a process problem, and process problems have process solutions:
Single point of engineering accountability. When one team designs the transformer, switchgear, generator, and cooling sequencing together, failure mode assumptions get reconciled at the design table, not during a live test with a facility director watching.
Sequencing documented before procurement, not after. Load transfer logic, restart timers, and failover order should exist as an engineered document before equipment is ordered not reverse-engineered from whatever each vendor’s default settings happen to be.
Commissioning treated as a design input, not a final exam. Teams that walk through failure scenarios on paper during design catch far more issues than teams that wait for the physical test to find out what they missed.
None of this is exotic. It’s closer to basic systems engineering discipline than a technical breakthrough. But it requires structuring a project so that one party is actually responsible for how the pieces behave together which is a contractual and organizational choice, not just an engineering one.
Where Rexedge Fits Into This
This is, honestly, the problem we were built around. Rexedge designs transformers, switchgear, generators, cooling systems, and secondary fluid networks as one connected scope, with one team responsible for how the failure sequencing actually plays out not just how each component performs on its own data sheet.
That doesn’t mean every commissioning issue disappears. Real projects still surface real problems. But it means the sequencing logic gets written once, by people who understand the whole system, instead of getting assembled from four different vendors’ default assumptions and tested for the first time with the facility already under construction.
The Real Lesson
If there’s one thing worth taking from this, it’s that commissioning problems are rarely a sign that someone built bad equipment. They’re usually a sign that someone built good equipment in isolation. The fix isn’t better transformers or better chillers most of what’s on the market today is genuinely solid. The fix is making sure the people specifying those systems are talking to each other before the equipment ships, not after it’s installed.
Have a project where power and cooling scopes are still being sourced separately? Talk to Rexedge before the RFPs go out that’s the point where this problem is cheapest to solve.