Data Center Upgrades: Temp Cooling – Not Just an After Thought
Renovations and upgrades to existing Data Center cooling systems sometimes require temporary cooling to keep the critical loads properly cooled during construction. If not designed properly, a poorly planned temp cooling system can lead to disaster. Especially when project conditions require temp cooling be operational for a prolonged period of time. Often, temp cooling is left to the Contractor’s to include in their bid and the design responsibility is delegated to them with loads and some instructions given as qualifications for a delegated design. This article stresses the importance of temp cooling and how it should be handled as a standalone design project.
We recently had the opportunity to design the upgrades to a very critical Data Center serving the needs of a global media company. The existing Data Center had grown over the years to 50% of its planned capacity when it became apparent that the cooling system could not keep up with the growing load. The shortcomings of the cooling system were primarily due to the fans inability to overcome the increasing static pressure required in the ductwork distribution system with the increased airflow demanded by the growing load. Return air issues were even more problematic because of the remote location of the cooling units and the use of a shared ceiling plenum for return air. The same ceiling plenum was shared with the office air handling systems making it impossible to separate the loads for the Data Center cooling units, reducing their cooling capacity by lowering their return air temperature with cooler, mixed return air from the office space.
The solution to cooling the Data Center properly into the future involved creating a new mechanical space adjacent to the Data Center and installing new cooling units while the existing cooling units were still connected and serving the loads. The biggest challenge was removing portions of the old air handlers and ductwork system to make way for the new, while maintaining cooling to the operating data center. To accomplish this, the design included three phases where approximately one third of the old system was removed at a time. Temporary cooling would be planned to keep the live Data Center properly cooled for each of these shut down phases. In the end, the temp cooling systems were in continuous use for nearly 7 months in order to complete all phases of the work. Adequate cooling was maintained to the Data Center throughout this entire period so that all Data Center operations could continue without interruption.
A Mixed Use of Temp Coolers
The capacity of the temporary cooling system for each phase was limited by the approximate 1/3 area that could be shut down at a time. This load required approximately 40 tons of temporary cooling capacity. Packaged DX spot coolers are only available up to 12 nominal tons capacity and each require a considerable amount of floor space to tie in the many ductwork connections that are necessary to distribute both air streams associated with DX cooling (evaporator air stream and condenser air stream):
Distribute the cool evaporator discharge air to the rack inlets.
Return the evaporator air from the hot aisles (rack outlets).
Duct both the intake and discharge of the hot condenser air stream out of the rack room.
It soon became clear that there would not be enough space within the rack room to properly install, even in a temporary condition, the quantity of DX spot coolers that would be required. There would be no room for the necessary flexible ductwork, let alone required access for the day-to-day operations of the Data Center servers.
A hybrid temp cooling system was the eventual solution. Chilled water CRAH units located in an empty rack row were combined with DX spot coolers and ductwork cross connections that shared excess capacity in the remaining live areas, to make up the remaining redundant capacity for the temporary cooling system design. Temporary chilled water piping was installed in the raised floor to the CRAH units’ bottom connections. The route for the chilled water piping passed through 150 lineal feet of critical raised floor space and were made without any joints using flexible heavy duty, high quality blend SBR/EPDM piping hose rated for 150 psig working pressure. This piping system, proposed by the contractor during the temp cooling planning and design stage, reduced the risk of leaks and helped to present a good case for using chilled water to make up the capacity we could not achieve with DX spot coolers and ductwork cross connections alone. Without the capacity of the chilled water CRAH units, we physically could not have come up with the tonnage required.
Temporary cooling was installed and commissioned before each phases’ demolition. Close coordination was required between the facilities department, trade contractors and FEA engineers to plan, test and execute each step of the project.
Commissioning Temp Cooling
To reduce risk, the shutdowns were planned to minimize the time spent on temp cooling. This meant as much of the new construction and space preparation would need to be installed and made ready prior to the shutdown, requiring close coordination between trades. Lengthy shutdowns were still unavoidable but were kept to a minimum as best as the existing conditions and building constraints would allow. Ultimately, the temp cooling system would need to be operational for a minimum of 3 to 4 weeks during the first scheduled shutdown. Thus, testing of the temp cooling system became imperative. Test scripts were prepared in advance which included failure testing redundant temp coolers, testing equipment status alarms and room high temperature alarms. The temp cooling equipment alarms including additional space temperature alarms were tied into the BAS system to monitor temp cooling and space temperature status. In order to confirm the temp cooling system would be adequate for prolonged periods of time, we included a test to watch for any steady rise in space temperature. A steady rate of temperature rise in a hot area of the Data Center for instance could mean that area, although it may appear to be properly cooled, could eventually overheat. These hot spots would be handled by making improvements to the cool air distribution (supply air) and hot air removal (return air) to those areas.
Preparations were made in advance to return cooling to the existing system in the event that the temp cooling test failed. A manual switch temporarily installed to operate each of the existing combination fire/smoke dampers (FSD) on the main cooling supply air duct, would provide quick means to return cooling to the house system.
Conclusion
With meticulous preparation and thorough testing, the temporary cooling system did not fail and all phases of the project were successful. This result could not have been possible without the planning effort and teamwork between the engineers, contractors and Owner’s facilities staff. Without proper planning, design and coordination of the temporary cooling system, the Data Center would have certainly overheated requiring shut down and operations moved to the disaster site. That unacceptable result would have been the only choice had the temporary cooling been left to just an afterthought.
