For decades, chiller plants have been treated as the primary chilling method. If a data center needed cooling, it needed a chiller plant sized to handle the entire thermal load. That assumption made sense for previous densities, but it doesn’t anymore.
The Old Cooling Design
Traditional chiller-based cooling exists to close one specific gap: the difference between what a data hall (or a chip) needs and what the outside air can provide on its own. For most of the industry’s history, that gap has been wide. Legacy IT equipment ran cool enough, and ambient temperatures rarely got close enough to useful, so operators built mechanical cooling to do the entire job.
However, that approach comes at a cost – financial, environmental and community. Chiller plants rely on compressors and refrigerants, some of which are still being phased out for environmental reasons. Many designs are evaporative, consuming significant volumes of potable water to reject heat, and chillers are expensive to build, expensive to run, and inefficient compared to simply using the air or water that’s already outside the building.
The industry has lived with these tradeoffs because the alternative, closing that thermal gap without mechanical help, wasn’t available. Until recently, it wasn’t a realistic design option.
How to Close the Thermal Gap with Trim Chilling
The size of that thermal gap comes down to one number: how close a cooling system can get to ambient conditions before mechanical intervention is required. This is typically referred to as approach temperature, and most of the industry has designed around a fairly wide margin.
Nautilus’s cooling distribution units are built to run at a meaningfully tighter approach temperature than conventional designs. That difference sounds small on paper, but it changes the entire equation. A tighter approach temperature means a facility can rely on ambient air or available water sources to do most of the cooling work, and only needs mechanical chilling to cover the remaining, much smaller gap.
We call this trim chilling: using mechanical cooling to trim the last few degrees rather than carry the entire thermal load.
How A Trim Chilling Approach Changes Design
Trim chilling isn’t a single feature; it’s a shift in how facilities can be designed and scaled.
Chiller capacity becomes flexible. Cooling distribution units (like the Nautilus EcoCore® FCD) are built to carry most of the thermal load on their own, rejecting heat to whatever water or air source is available on site. Instead of committing to a fully sized chiller plant on day one, operators can size mechanical cooling to the small gap that’s actually left, and add capacity in increments as workload or conditions demand it.
Geography becomes a design input. A facility in a cool, northern climate may need close to zero mechanical chilling across most of the year. A site in a hot, arid climate will lean on trim chilling more heavily, but still far less than a traditional 100%-chilled design. Either way, the mechanical cooling is sized to the actual gap, not to a worst-case assumption at the start.
Workload becomes part of the equation. As chip water tolerances shift, and as newer generations of hardware are able to run on warmer supply water, the size of the gap changes again, often shrinking further. A design built around trim chilling can absorb that shift without a wholesale rebuild of the mechanical plant.
Flexibility for Dynamic Loads and Temps
The point of trim chilling isn’t to eliminate chillers everywhere. Some sites, some climates, and some workloads will always need more mechanical support than others. The point is that operators are no longer locked into a single answer decided at the design table years before a facility opens.
Nautilus’s cooling technology, first developed and proven on our own first-of-a-kind build, was engineered from the outset to close that gap physically, in the hardware, rather than manage around it after the fact. That’s what makes trim chilling possible: not a control layer bolted on top of a conventional system, but a cooling architecture designed from day one to need less mechanical help in the first place.
From 0% to 100% mechanical cooling (which isn’t really a thing anymore), and everywhere in between, the goal is the same: give operators a cooling plant that matches what their site and their workload actually require, not what a worst-case assumption demanded a decade ago.
