Addressing low and part load (in)efficiencies in data centers - real world examples from operating data centers

Abstract

As data center designs continue to evolve, performance and reliability improves resulting in higher levels of uptime in the facilities. Often, this results in systems with additional capacity in power and cooling equipment to allow for concurrent maintenance, protection against equipment failures and can allow for more efficient operation. In addition to higher levels of reliability, data center designs are also pushing the envelope on efficiencies from the supporting infrastructure. While the design intentions are generally in the best interest of the buildings performance, data center designs can often suffer from overprovisioning power and cooling compared to the near-term or even long-term loading of the building. Further, the commissioning of systems often utilizes load banks to simulate ITE loads in the data center, typically done at design loads. Due to this, the low-load conditions in the data center may not be fully realized until the facility goes live. The low-load conditions can cause considerable inefficiencies to arise in the power and cooling systems and often the design conditions may not be met for a number of years, if at all. However, through careful planning and proper commissioning of the systems, design efficiency levels can be achieved and even exceeded.