This article, co-written by DPR’s Mark Thompson and Mark Whelpley, first appeared in 7×24 Exchange’s Magazine 2019 Fall issue.
Picture this scenario: an up-and-coming data center developer is looking to expand its portfolio in the Silicon Valley data center hub of Santa Clara. The company initially casts a wide net looking for the right property on which to build its new ground-up colocation facility – only to discover that undeveloped or greenfield land is a scarce commodity in this densely developed, high tech mecca.
The developer launches a new search, this time for an existing building it could retrofit and convert to data center use. In short order it finds a candidate that seems to fit the bill: an older industrial office building that has been sitting vacant for a few years. It is priced to sell. The building’s footprint is workable, the structure is intact, and both buyer and seller are motivated. Add some extra power and cabling equipment, the developer reasons, and this dusty old office space will easily transform into a profitable data center facility. An added bonus: it’ll be up and running much quicker than building a brand-new facility, enabling the developer to move in tenants, start collecting rent and begin making a return on investment that much sooner.
The developer hires a general contractor who specializes in commercial building construction but who recently jumped into the booming data center market and now has a couple of data center projects under its belt. An architect is also brought on board, and together they devise a plan to retrofit the facility. It may not be perfect, but they assure the developer they can make it work – and that the planned retrofit will save the company time and money in the long run.
The purchase is made, and the first shovel hits the ground.
As construction gets underway, the project team quickly realizes the building’s structural capacity doesn’t support the volume of heavy equipment – including racks of servers, chillers and air handling units – that this modern data center requires. In addition, there isn’t enough land around the building’s perimeter to locate the backup generators outside. They’ll need to be installed on the building’s rooftop instead – but it turns out the roof also isn’t designed to support that amount of weight.
It’s starting to look like a complete gut and reconstruct will be required.
And then there’s the matter of the available power onsite. The contractor assumed that since this is a reuse of an existing building, power supply wouldn’t be a major issue. Now they find out it could literally take months to work with the utility company to bulk up the site’s power infrastructure in order to meet the data center’s needs. The anticipated time and cost advantages of this property are quickly evaporating, and the developer is starting to think it has made a big mistake.
The Right Approach: Steps to Success
This fictional scenario may be a bit of an oversimplification and, certainly, it represents a worst-case situation, but it’s not an entirely unrealistic depiction of what can happen when an owner doesn’t properly evaluate or conduct complete due diligence on a property that they plan to convert into a data center facility. How should this process have been approached instead? Let’s examine the steps that owners and their teams should follow to ensure their data center retrofit projects are successful.
The very first step the owner and the design and construction team should take is to clearly define what constitutes success for them on their data center project. Is speed to market most important, or do cost savings or energy efficiency take precedence? Is landing a specific tenant or providing service in a specific area the overriding concern? A building repurpose project may or may not end up being less costly than a ground-up project; depending on the circumstances, it may even cost more. The former “hidden gems” of available building flips in places like Silicon Valley, the Dallas-Fort Worth Metroplex, Loudoun County and other major data hubs are becoming fewer and farther between. Even in “edge” markets, the number of existing buildings that can be turnkey solutions for data halls are rare.
It is equally important for the owner to set early, realistic expectations of what it expects to achieve on the project and to carefully assess how easily and cost effectively a particular building could be retrofitted to new use. The time to do the homework and thoroughly evaluate candidates for a prospective retrofit/conversion is before the property is purchased, not after. Proper vetting is critical.
And that vetting process applies to selection of the design and construction team as well. While the aforementioned developer was on the right path engaging the contractor and architect prior to purchasing the property, the selected contractor that lacked historical knowledge or expertise specifically relating to the rapidly evolving data center market. As a result, the contractor didn’t anticipate some of the hidden pitfalls and “gotchas” that might have been caught by a more seasoned team. The overly optimistic “we’ll make it work” approach did not serve the owner well in this case either or help the owner to make a fully informed decision about the costs and challenges of retrofitting this property.
Bottom line? Bring a contractor and/or designer on board early in the process. Choose firms with extensive experience in data center construction, including both ground-up and retrofit projects. Ideally, they will have a decade or more worth of data center projects in their portfolio and be ranked among the Engineering News-Record’s top 5 or 10 data center contractors. A qualified general contractor or designer can skillfully guide the owner through the process of assessing prospective retrofit candidates based on a set of clear-cut criteria – and will help the owner make the best decisions.
An Objective Eye: Key Evaluation Criteria
Once the owner has selected the team and they’ve jointly scouted for and identified a few potential retrofit candidates, it’s time to objectively weigh the options. This step means taking an in-depth look at what’s “under the hood” of a given building and considering how well it meets the project goals. Think of it like bringing along a qualified mechanic to inspect the used car you’re considering buying. It may cost more up-front paying for the mechanic’s time but could well save you from making a costly decision in the long run.
There are at least 8 major criteria that should be carefully assessed on every data center candidate. They include:
- Roof Structural Capacity. Data centers require roofs with a high structural capacity since equipment and heavy systems are often hung from or attached to the roof. Depending on the building’s former use, the roof may not be up to the task and could be a big-ticket upgrade. For data centers, a roof rating of over 35 lbs./sq. ft. is best; 25-20 lbs./sq. ft. is good; and less than 15 lbs./sq. ft. falls squarely in the “bad” category.
- Floor Capacity. The racks and computer equipment that go into data centers demand a high floor capacity, something you typically won’t find when converting from an office building, call-center, multi-story structure or the like. Retrofitting this infrastructure is costly and may require tearing down and starting from scratch. For a rule of thumb, a building with a floor capacity of over 250 lbs./ft. is best; 125-200 lbs./ft. is good; and 125 lbs./ft. lands in the “bad” category.
- Structural Code. There have been three major building code revisions in the last 10 years or so, including in 2010, 2013 and 2016. This means selecting a building constructed prior to 2010 may require extensive structural changes to bring it up to current standards. Buildings constructed between 2010 and 2013 are evaluated as “good” and require more minor changes, while the “best” rating in this category are buildings designed to the latest uniform building code standards of 2016.
- Structural System. Hand-in-hand with evaluating a building’s structural code is its type of structural system. Post-tensioned or truss systems, found in buildings constructed during the 1980s and 1990s, are poor candidates for cost-effective retrofits, requiring extensive reinforcing and rebuilding. Moment frame buildings are better, while steel frame structures using buckling restraining brace frames (BRBs) are ideal candidates in high seismic zones like California. In addition, know the Importance Factor assigned to a given structure, as it will indicate how much structural redesign will be required to bring the new data center up to the necessary performance standards.
- Mechanical & Electrical Equipment and Infrastructure. Two other key evaluation criteria are the age and condition of the existing building’s MEP equipment and its MEP infrastructure. Owners should understand that a former office building’s MEP system typically will not approach what is needed for data center usage and thus will likely require complete replacement. However, conversion of a former semiconductor facility or similar technical facilities may not require such extensive changes, depending on the age of the system. The rule of thumb: mechanical/electrical systems 15 years old or older score poorly in this category; 10-15 years old may be considered good depending on the type of facility it was; and less than five years old falls into the good category.
- Watt Density. The power density per square foot of the existing building is another key measurement. The trend is to put the highest load in the smallest space. Current density trends favor more than 150 watt/sq. ft. as the best performance criteria, while 100-150 watt/sq. ft. is considered “good,” and less than 100 watts/sq. ft. is bad and will require major upgrades.
- Raised Access Floor. Raised access floors are part of most modern data centers. If the building is an older one, even if it has raised access floors, they are considered obsolete. That’s because modern rolling load capacity of the cabinets require raised access floors to be at least 36 inches high with a 3000-lb. load capacity. Replacement of raised access flooring is a big-ticket item that can run between $40-$50 per square foot on the West Coast, and $20-$25 per sq. ft. on the East Coast.
Bringing it All Together for a Successful Outcome
Armed with realistic expectations, understanding what constitutes success in meeting their project goals, assisted by a well-qualified team, and having thoroughly vetted and attained hard data on what each potential building candidate offers, the data center developer is now ready to make a well-informed decision. The savvy owner and project team also knows that since data center demands are constantly evolving, building flexibility into their project whether new or a retrofit is another essential consideration.
Technically and logistically demanding, the design and development of data centers will always present challenges as well as bottom-line opportunities for the owner. A smart approach goes a long way toward setting your next data center project up for success.