Excess power capacity drags down operating efficiency and inflates OpEx, and there are two main reasons it exists: a sub-optimal specifying process and the natural human desire for a security blanket.
As a leader in data center infrastructure, it’s not uncommon for Vertiv professionals to visit with large, respected organizations whose power system utilization is below 20 percent. Clearly, there is room for improvement, and that improvement can bring significant OpEx savings.
I participated in a series of Consultants Masterclasses recently and did an impromptu survey. Only 1 participant in a full room indicated they targeted 100% or higher redline. When I asked folks why they did not target 100% or higher, it was a mix of not wanting to operate at the equipment rating and the conservative approach of their clients.
For colos, this can represent a lost revenue opportunity. If we can help them recover, say 360 kW of capacity in a prime location, how much would that be worth? It’s not that hard to figure out. If that capacity could be sold for $1,000 per kW (to make the math easy), that’s $360,000. Certainly worth your time and effort to examine.
The most forward-thinking data centers, especially the biggest of the big, are increasingly interested in realizing these savings.
There are two ways to change the low power utilization equation:
- From the bottom up by adding IT load. We have seen some colos sell more capacity than they originally planned to increase power utilization.
- From the top down by right-sizing the power infrastructure. As an infrastructure provider, the biggest opportunity we can help with is right-sizing the power infrastructure at the front end.
Here are three main ideas for right-sized power infrastructure that will deliver the most bang for the buck:
- Size your Power Train Precisely: Optimize UPS Ratings
Purchasing organizations are well intended but tend to subscribe to a common practice that hurts utilization: buying based on lowest common denominator vs. based on optimal kW sizes. For example, specifying an 800 kW UPS system costs less per kW upfront than a 750kW system. Since the breakers in the gear are the same between 750 and 800kW, if a 750kW is selected they are leaving 50kW on the table and increasing their system cost per kW.
Increasingly, customers are evaluating the cost based on cost per kW. I’m not shy about telling customers when they are stranding capacity and increasing their cost per kW.
The UPS is just the most obvious element, but there are opportunities to rescue stranded capacity at every step in the power chain. Matching capacities across the chain – input board, output board, static transfer switch, UPS – brings you maximum utilization. Anything else strands capacity.
If you can take this step, you could save up to 7 or 8% in power system operating costs.
- Bust the Buffer Myth: Drive to 100 Percent Redline Utilization
Most UPS redlines – their “not-to-exceed capacity” – are set to less than 100%. Vertiv encourages using our UPSs to 100% and in some cases, the UPS is specified to more than that at a given temperature.
So why not use that capacity?
There are extremely IT-dependent, world-class companies that set their UPS redline at 100% or higher. And more are likely headed in that direction based on the conversations we are having.
Setting the redline at less than 100% made sense five or ten years ago, when managing your IT load to 80% maximum UPS capacity was the norm. Loads were lagging and needed more kVA than kW. But in our transformer-free world, with one degree of variability taken out of the equation, that’s no longer the case. The power system is more precise, and you can operate closer to the UPS power rating rather than the kVA.
There are lots of reasons that prevent organizations from moving their redline to 100%: lack of real-time consumption data, inconsistent full load capacity from vendors, and fear of oversubscribing. If you can align the stars, though, this can cut costs by 5 to 10%.
- Reduce Battery Runtimes
Most organizations have at least 5-minutes of battery runtime, and that’s largely because VRLA battery manufacturers historically wouldn’t provide warranties on batteries with less calculated runtime.
With greater adoption of lithium-ion battery technology and alternative smart-energy backup sources, this is changing. What’s truly needed is exactly the runtime it takes for generators to kick in and support the load. If your operations team can repeatedly prove 2 minutes is what it takes, 2 minutes is what should be prescribed.
If you cut back your battery capacity to exactly what’s needed to bridge to back-up power, you could save another 17 to 24% in power operating costs.
If It’s This Black and White, Why Isn’t Everyone Doing It?
These are just three of the areas where the largest savings can be realized. Other strategies can be adopted, such as simplifying UPS paralleling, minimizing or eliminating redundant PDUs and converting from dual-input to single-input UPS. When you take a comprehensive approach, total operating cost savings could reach up to 50% over a non-optimized system.
But even if you take just these three steps, you can save more than a third of your power system operating cost and free up significant capacity. So why isn’t everyone doing this?
There’s a simple reason: It takes guts. It’s easier and more comfortable to leave yourself extensive wiggle room. But with razor-thin margins, especially in colos, and multiple ways to achieve five nines of availability, rescuing capacity becomes a priority. We’d love to help you explore ways to both reduce your operating costs and maintain your SLAs.