Cronin's Workshop: Making Free Cooling Work in the Data Center
Since 2000 data centers have been migrating out of the modified corporate institutional office building into separately metered, free-standing facilities. This solution meets the CIO’s desire to utilize inexpensive real estate without being shoehorned into expensive home office space. It seemed that everyone suddenly discovered that the cost of fitting out warehouse space for a data center was several orders of magnitude cheaper than modifying less-than-ideal office space.
This trend continues today; however; as a consequence of putting data centers into their own facility, CIOs attribute the actual cost of IT power consumption to the computer processing. They can also now see the direct benefit of any energy conservation measures.
The high cost of energy combined with society’s grassroots desire for sustainability has everyone looking for reliable means to reduce the energy loads of data centers. Although consultants and experts can argue over the configuration of the “ideal” energy-efficient data center (virtualization, 64-bit processing, multi-core processors, etc), the debate tends to be too complex to produce many immediate concrete results. Taking the data center’s mechanical design and making it as efficient as possible, on the other hand, is something we have been doing for decades, although at times with questionable success.
One of the most popular solutions has been the application of a free cooling cycle in the mechanical system to take advantage of cool outside air. During the energy crisis of the mid-70s, the “Strainercycle” or “Thermocycle” concepts of removing solids from the condenser water so it could be used as chilled water were all the rage for commercial facilities.
Strainer SystemsAfter a few short years of using the Strainer/Thermocycle concept of pumping low temperature condenser water into the chilled water piping systems, many an owner learned the hard lesson of cross contamination. Engineers marveled at how well these strainer systems removed solids, but biological contaminants were considered the purview of the water treatment contractor. The limited biological growth that is treatable in the small condenser water loop could now spread across the entire chilled water loop. Lower flow rates, dead legs, and small branches in the chilled water piping systems quickly clogged these systems as condenser water with all its biological impurities began to eat away at the chilled water piping, creating pinhole leaks throughout the system. Suddenly the operational energy savings were being offset by massive capital costs to replace the chilled water piping systems.
Heat ExchangersMany an owner and designer then turned to flat-plate heat exchangers (HX). These systems required a little more pumping than the strainer systems; however; they maintained the physical isolation between the condenser water and chilled water systems. In the end, HX provided most of the energy savings of the strainers without the contamination risks.
The HX systems did not work well for facilities with data center spaces, primarily because of how they operated and the way they were piped. Typically these HX systems were piped as part of a common condenser water loop. To swing from mechanical cooling to free cooling required the condenser water temperature to be dropped substantially below the operating parameters of the mechanical cooling system. Each swing onto and off of free cooling would create a time lag during which there was no mechanical cooling available and the water temperature was too high to attain free cooling. The time lag led to large swings in data center temperatures and complaints from the IT department. As data centers became more heavily loaded the complaints grew and eventually many HX systems were mothballed.
Ironically many designers make this same design mistake today when the solution - creating a sequence of operation for the HX system wherein it is isolated to operate with a dedicated cooling tower and run in parallel with the mechanical cooling system - is very simple. This arrangement enables the HX system to immediately provide chilled water at the proper temperature without any time lapse or temperature rise in the data center. As the HX picks up the load the mechanical cooling system is gradually off-loaded. Reversing the sequence is just as smooth a transition.
Air-Side EconomizersThe latest craze seems to exploit our zeal for maximum energy efficiencies without providing adequate protection from unintended risks. In the past year there have been many data centers designed to take advantage of 100 percent outside air. This technique effectively says allows us to open up the outside walls and let Mother Nature do the cool our data centers for us while outside temperature conditions permit. If it were that easy, we could just pitch tents in cold climates to house data centers.
Does air-side economizing work? Yes, but, we must deal with a transition from the traditional “closed-and-controlled” data center environment to an open environment that includes little to no control over humidity levels or airborne contaminants. Good engineering can address humidity, which is really just a normal data center risk; however, airborne contamination creates a whole set of new risks that will vary widely from site to site.
An air-side economizer is going to pump several hundred thousand cubic feet per minute of outside air through the data center. If the data center is near the ocean, salt intrusion is a problem. In an urban location city exhaust pollutants might enter the data center? Farming creates other airborne contaminants in rural locations. And fine volcanic dust from volcanic explosions across the world can be carried to any data center location. Lastly, what about gaseous fire suppression and incipient smoke detection systems? These are all very real risks that need to be thoroughly addressed.
As a real life example, I once experienced a situation with a downtown urban data center. Approximately 12 miles away, across a river, and in an adjacent state, a fire started in a tire dump. The winds blew the smoke directly towards the data center. We promptly shut off outside air to the building, but there was sufficient penetration of the contaminants to trigger the cross zoning for the Halon systems. Fortunately, the staff had the foresight to deactivate the Halon discharge long before it would have triggered. Had the fire occurred at night (primetime for free cooling) we would not have observed the smoke cloud and may not have been as fortunate.
The ChallengeProve that airside economizers are worth the risk.
We mitigate the risk, but are the savings worth the added time and expense? Let’s hear from those who are using both HX systems and air-side economizers.
What have been your measured successes and share your experiences on pitfalls to avoid?
Next issue: Monitoring – Is it worth it?
Introducing “Cronin's Workshop”Win a Mission Critical coffee mug
The enormous creativity of our industry means that we face radical changes in the way we design, build, operate, and maintain mission critical facilities. Making this change will demand strong debate about what is right, wrong, and best practice.
We have created “Cronin’s Workshop” to foster this debate. Each month our columnist Dennis Cronin will describe a problem and pose a solution. We want to hear what you have to say about these problem/solutions posed by Dennis. In the subsequent issue we will publish the best positive and negative opinions received from readers.