Emerging technologies, trends, and opportunities are impacting the way data centers perform the urgent, and often challenging, task of cooling their servers and other network equipment to minimize PUE and satisfy upper management at the same time.
The investment in data center cooling can be significant, and the responsibility is daunting. It’s a rare operation of any kind that isn’t wholly dependent on its data management foundation, and overheating, which can occur in a swiftly escalating domino effect at a second’s notice, can bring an entire global operation to a halt, stranding and impacting the productive time of hundreds or thousands of users.
The demand for data storage and usage was growing at a fast clip when COVID-19 dramatically changed everything, including the way we work, and the sudden increase in people working remotely rapidly escalated digital demand to record-high levels. And, even as the world returns to a more normal state, a reduction to prior levels of remote demand is unlikely.
Lowering Cooling Costs
Most commonly, the cooling strategy of choice for many data centers is similar to what has been used for decades — generating cool air to lower ambient temperatures using chilled, water-based CRAHs or CRACs. Yet, the technologies and methodologies to do so have progressed significantly in multiple directions.
For example, the cool air is often better focused, brought closer and closer to where it is needed — from the room to the row to the racks — to increase efficiency. Over the years, computers and servers have become more tolerant of higher temperatures; thus, the equipment doesn’t have to always be kept in the “human comfort zone" to maintain effective operation. In order for the cooling system to fully benefit from this change, the HVAC equipment must be able to generate higher leaving water set points. Some traditional chiller technologies, such as screw compressors, are not able to do this effectively. New computer equipment designs can operate at higher space temperatures, enabling chillers to generate leaving water temperatures as high as 82ºF, reducing compressor power consumption and helping data centers reduce operating costs. Additional technologies, such as oil-free compressors, may be introduced as well in order to realize the full benefits of this type of system.
This new chiller application capability is better enabled by the advent of oil-free compressors, which do not require oil for lubrication because the motor shaft levitates in a magnetic field. In addition to potentially providing several efficiency gains and reducing energy costs, this technology can also potentially reduce maintenance costs for data centers. For example, there is no need to periodically change the compressor oil and oil filter, and there is no mechanical wear to the system since there is no metal-to-metal contact. Operations using oil-free compressors can reduce their maintenance costs some 30% or more over those using traditional fixed-speed positive displacement compressors.
Another advantage of oil-free compressors is that because oil and mechanical wear have been eliminated, the performance remains consistent over time.
Often, companies look to improve their energy performance for environmental and/or financial reasons and are delighted by the reduction in carbon footprint that often moves in lockstep with reducing energy consumption. With an average annualized PUE of 1.57, data center losses are currently adding about 60% to the energy use of IT. And, while more new builds are designed with PUEs of 1.3 or less, it is not economically or technically feasible for many operators to perform the major overhauls needed for better efficiency in many older facilities. However, there are easy gains to be had from better airflow management, optimized controls, and equipment replacement. Further improvements might require significant change, such as retrofits with highly efficient cooling systems.
Similarly, a move to different compressor technology often allows a move to a more environmentally benign refrigerant. For example, most screw and centrifugal compressors today use R-134a, which has a global warming potential (GWP) of 1,400. Oil-free compressors can be used with several refrigerants, including low-GWP R-513A and R-515B or ultralow-GWP HFO-1234ze.
Another recent innovation takes the idea of focusing cooling closer to the heat to its logical conclusion and utilizes a cold plate placed on the server that's connected to a chilled water loop that carries the heat outside. Another alternative design concept involves specially designed servers submerged in dielectric cooling fluid that rejects heat directly from the server to the fluid.
Moving Beyond PUE
Data center HVAC systems have long been managed as a cost center with a focus on continually reducing the operating costs through various efficiency improvements. However, many companies today are benefitting from an emerging solution that replaces the cost model with an entirely new paradigm — removed heat is not dissipated into the air but is instead recovered and sold as a valuable commodity to those who need it at the time. Considering how much energy is spent to heat buildings from scratch, this is obviously a need waiting to be filled — especially in colder and temperate climates.
The general concept is familiar to many industries. In fact, many industrial plants use their waste heat in a cogeneration model, where heat removed from a process is disseminated to another area of the facility that demands heat. This reduces the amount of energy that must be otherwise generated or purchased from utility providers.
Data centers produce heat 24/7, making them a de-facto reliably consistent “generator.” Once that paradigm shift is made, there are some upfront infrastructure costs, but the concept can pay for itself and become a profit center fairly quickly based on recent demonstration projects. If the data center is close to a district heating infrastructure, which collects and generates heat for dispersion to a nearby campus or even to an entire municipality, the supply infrastructure is ready-made. But, it can also be cost-effective to create a new grid around many campus-like facilities, such as colleges or business parks, where heat must be provided to many adjacent rooms and buildings. Hyperscale and enterprise companies, with their mega-scale facilities, especially have the flexibility to locate in northern climates, which they have been doing the last several years, creating the heat recovery scale needed for these district heating systems. The higher data center operating temperatures also means the heat pumps applied to maintain cooling while also recovering heat operate at optimal efficiency, lowering the resulting heat price and justifying base-loading the heat source. Additionally, on-site backup power means a constant supply of recovered heat under demand response or other power interruption scenarios.
Oil-free compressors can help with this shift through recent advances that have expanded the operating map to support heating applications. High-lift, oil-free compressors have the ability to generate higher leaving water temperature for use in heating applications, which, in the past, have commonly used traditional oil-lubricated, positive displacement compressors. Using oil-free compressors for this application brings the benefits of reduced maintenance and no performance degradation over the life of the compressor.
In this model, the HVAC system becomes a revenue-generator for the enterprise and can ultimately provide an energy source that would otherwise be wasted if released into the atmosphere. Finally, this model can go a long way for companies looking to reduce their carbon footprints and contribute toward their decarbonization and net-zero emissions goals.