Prognostics-based health management for free air cooling of data centers

The telecommunication industry has become conscious about its energy consumption and associated the environmental footprint of its operating infrastructure, e.g., data centers. At present, more than one half of the total energy consumption of data centers is for the power and cooling infrastructure. One way to save energy is an approach called "free air cooling," where ambient air is used to cool the equipment, thereby reducing the energy consumed in cooling and conditioning the air. Free air cooling is being adopted for data centers that are already in operation where the equipment used are unlikely to have been designed or qualified for free cooling regime. Traditional reliability evaluations depend on industry standards that assume pre-defined environmental conditions and generally perform set of environmental tests that relate to those conditions. Equipment level risk identification and mitigation methods are not practical to apply for the data center as a system. This paper provides an overview of the associated challenges posed by such an operating environment. This paper discusses the reason why traditional reliability methods cannot be used to assess the risks of free air cooling of data centers. Finally, as an alternative approach, prognostics and health management technique that can provide the users of data centers with the ability to identify impacts of free air cooling, estimate remaining useful life of the equipment, and help identify impending intermittent service interruptions and degradations is described. ©2010 IEEE.