2011 Reflections

There is a saying in the MEP consulting business: “no one ever gets sued for oversizing.” That fear-driven mentality also affects the operation of mechanical systems in data centers, which accounts for why data centers are over-cooled at great expense.  But few facility managers know by how much.  The fact is that it has been easier – and to date –safer to over-cool a data center as the importance of the data it contains has increased and with that importance, the pressure to protect it.

Last year that changed.  With new technology, facility managers know exactly how much cooling is required in a data center, at any given time. And, perhaps more importantly, technology can provide warning – and reaction time – in the rare instances when temperatures increase unexpectedly. With this technology, data center cooling can now be “dynamically right-sized.”  The risk of dynamic management can be made lower than manual operation, which is prone to human error.

In our own nod to the advantages of this technology, we re-named the company I co-founded in 2004, from Federspiel Corporation to Vigilent Corporation.  As our technology increased in sophistication, we felt that our new name, denoting vigilance and intelligent oversight of facility heating and cooling operations, was more reflective of the new reality in data center cooling management.   Last year, through smart, automated management of data center energy consumption, Vigilent reduced carbon emissions and energy consumption of cooling systems by 30-40%.  These savings will continue year after year, benefiting not only those companies’ bottom line, but also their corporate sustainability objectives.   These savings have been accomplished while maintaining the integrity and desired temperatures of data centers of all sizes and configurations in North America, Canada and Japan.

I’m proud of what we have achieved last year.  And I’m proud of those companies who have stepped up to embrace technology that can replace fear with certainty, and waste with efficiency.

Cooling Tips

Ten Tips For Cooling Your Data Center

Even as data centers grow in size and complexity, there are still relatively simple and straightforward ways to reduce data center energy costs. And, if you are looking at an overall energy cost reduction plan, it makes sense to start with cooling costs as they likely comprise at least 50% of your data center energy spend.  Start with the assumption that your data center is over-cooled and consider the following:

Turn Off Redundant Cooling Units.  You know you have them, figure out which are truly unnecessary and turn them off. Of course, this can be tricky. See my previous blog on Data Center Energy Savings.

Raise Your Temperature Setting. You can stay within ASHRAE limits and likely raise the temperature a degree or two.

Turn Off Your Humidity Controls. Unless you really need them, and most data centers do not.

Use Variable Speed Drives but don’t run them all at 100% (which ruins their purpose). These are one of the biggest energy efficiency drives in a data center.

Use Plug Fans for CRAH Units. They have twice the efficiency and they distribute air more effectively.

Use Economizers.  Take advantage of outside air when you can.

Use An Automated Cooling Management System. Remove the guesswork.

Use Hot and Cold Aisle Arrangements. Don’t blow hot exhaust air from some servers into the inlets of other servers.

Use Containment. Reduce air mixing within a single space.

Remove Obstructions. This sounds simple, but  a poorly placed cart can create a hot spot. Check every day.

Here’s an example of the effect use of an automated cooling management system can provide.

The first section shows a benchmark of the data center energy consumption prior to automated cooling. The second section shows energy consumption after the automated cooling system was turned on. The third section shows consumption when the system was turned off and manual control was resumed, and the fourth section shows consumption with fully automated control. Notice that energy savings during manual control were nearly completely eroded in less than a month, but resumed immediately after resuming automatic control.

Occam’s Razor

Data Center Energy Savings

The simplest approach to data center energy savings might suggest that a facility manager’s best option is to turn off a few air conditioners.  And there’s truth to this.  See the graph below, showing before and after energy usage, and the impact of turning off some of the cooling units.

Before & After Energy Management Software Started

But the simplicity suggested here is deceptive.

Which air conditioners?

How many?

How will this truly affect the temperature?

What’s the risk to uptime or ridethrough?

While turning things off or down is likely our greatest opportunity for significant, immediate savings, the science driving the decision of which device to turn off and when, is complex and dynamic.

Fortunately, a convergence of new technology – wireless sensors for continuous, real-time and location-specific data, along with predictive, adaptive software algorithms that take into account all immediate and known variables at any given moment – can predict the future impact of energy management decisions, taking on/off decision-making to a new level.  Now, for the first time, it’s possible – thanks to the latest AI technology – to automatically, constantly and dynamically manage cooling resources to reduce average temperatures across a facility and avoid hot and cold spots of localized temperature extremes. Simultaneously, overall cooling energy consumption is reduced by intelligently turning down, or off, the right CRACs at the right time. The result is continually optimized cooling with greater assurance that the overall integrity of the data center is preserved.