|
Improving Data Center Reliability through Commissioning and Energy Efficient Design
|
Paul Tia, LEED'® AP, San Francisco Associate Principal
Commissioning data centers constantly challenges Commissioning Agents due to the varying complexities of the critical support systems. However, regardless of the complexities, commissioning ensures that these systems are reliable and perform individually and interactively in accordance with the Basis of Design, Specifications, and the Owner's operational needs. In addition to commissioning, incorporating energy efficient designs can also improve the reliability of the critical support systems.
Reliability and Tier Classifications
To better understand reliability and data center design, it helps to recap on some industry standards. The following tier classification system for use in rating the reliability and availability of mission critical data center facilities:
Tier 1 is the lowest tier level and provides only a single path for power and cooling distribution with no redundant components.
Tier 2 is similar to Tier 1 in that it provides only a single path for power and cooling distribution, but reliability is enhanced with the addition of redundant components.
Tier 3 provides at least two paths for power and cooling distribution, with only one of the paths active and provides redundant components similar to Tier 2. A facility designed to a Tier 3 level includes infrastructure which can be be shut down for maintenance purposes without affecting the critical loads. This is referred to as "concurrently maintainable."
Tier 4 provides at least two active paths for power and cooling distribution, redundant components, concurrently maintainable, and is "fault tolerant." "Fault tolerant" means that the facility can sustain at least one failure in its infrastructure without impacting the critical loads. In determining the optimum reliability level for a data center, an Owner needs to evaluate downtime in terms of potential loss of revenue, as well as the reliability of the incoming utility power source. Infrastructure costs versus downtime is not a linear function. Costs increase exponentially as downtime decreases. Most modern data centers are designed with probable annual downtime of approximately 5 minutes, which translates to site availability of 99.999%.
Understanding a data center's tier classification enables the Commissioning Agent to develop testing procedures to ensure that the critical support systems are operating in accordance with the design intent. This includes simulating all possible failure scenarios to confirm that redundant components are able to support the critical load, and ensuring that the facilit's monitoring and alarm system provides real-time reporting (locally and/or remotely) of these failures, which will enable the Owner or operations personnel to troubleshoot the system in a timely manner. By subjecting the critical support systems to real world or "worst case" operating conditions, and exposing problems or faults that could affect the reliability and efficiency of these systems, commissioning ensures that the critical support systems are operating at optimum levels.
Energy Efficient Designs Reduce Operating Costs
Load design criteria for data centers have been in the range of 30-50 watts/sf. Presently, load densities of 150-200 watts/sf are not uncommon. Some high density mission critical facilities are being designed for 300-500 watts/sf of critical load. Issue no. 10 of Sustainability Matters included Michael Nichols' article regarding use of Computational Fluid Dynamics (CFD) modeling to help analyze increased thermal mixing resulting from higher power density designs. Manufacturers are continually looking at ways to improve efficiency of servers. More efficient equipment is more reliable, and requires less infrastructure. As sustainable engineers, Glumac is constantly looking for ways to improve energy efficiency in our data center designs, that will reduce operating costs for the facility. Energy efficient designs improve the reliability of a data center by utilizing fewer components that require preventive maintenance, repair, or replacement, or could become a single point of failure in the critical support system.
One example of a more reliable, energy efficient design is preconditioning the outside air entering the data center. In lieu of reheat coils located inside each computer room air conditioning (CRAC) unit, dehumidification occurs in an independent outdoor air dehumidification coil. The cooling coils inside the CRAC units would provide only sensible cooling. This design lowers the overall mechanical process-cooling load for the facility considerably, while maintaining the same supply air temperature along the cold aisles in the raised floor area. Depending on load density and server equipment operating temperature limitations, increasing the supply air temperature is a means of utilizing energy more efficiently.
Studies have been done on the feasibility of distributing DC power directly to server equipment. The rectifier and inverter processes of converting AC to DC, then back to AC is less efficient and increases the demand for process cooling to account for losses in the system.
Use of alternative energy sources, such as reciprocating engines or turbines, provide greater reliability by reducing a facility's dependency on the utility grid (reducing risk of planned or unplanned outages, protecting critical loads from utility disturbances), as well as utilizing cleaner energy sources and reducing dependance on fossil fuels. In the case of a co-generation plant utilizing natural gas-fired engine generators, the waste heat generated from the engines can be captured and used for other building systems (such as steam, space heating, domestic hot water, etc.). A heat exchanger may be installed for process cooling needs and refrigeration.
Utility companies, such as Pacific Gas & Electric, offer incentives for data centers that utilize energy-efficient equipment and associated controls, such as use of variable frequency drives on process cooling equipment.
Sustainable Thumbprint
As engineers for a sustainable future, Glumac is continually researching design methods to reduce the environmental impact of new construction. More "green" data centers will be emerging in the near future, facilities that utilize recycled and regionally sourced construction materials, more efficient lighting and lighting control systems, more efficient process cooling, as well as sustainable site elements. Just as important as satisfying the basis of design and the Owner's operational needs is making sure that our engineering designs and practices also satisfy the needs of future generations.
