This audio was created using Microsoft Azure Speech Services
Over the last 40 years, the world’s power grids have become more and more unstable. This is largely due to three major factors: 1) utility transmission grids have experienced very few, major infrastructure upgrades; 2) the addition of renewable energies to grid networks has made the challenge of maintaining a stable energy delivery environment more complex; and 3) an increase in catastrophic extreme weather events has added to the level of power instability and the associated risk to business resilience.
Given these external factors, commercial businesses, industrial sites and data centers are increasingly reliant on their onsite backup power generation systems. In most cases, a tandem of Uninterruptible Power Supplies (UPS) and generators provides the means for achieving reliable backup power. The generator set serves as long-term power backup (typically days) while the UPS systems serve as a bridge (typically minutes) until such time as the generators come online to support the critical load.
Investments in UPS and generator technologies are required because power anomalies, both brief and extended, can disrupt operations. In a data center environment, for example, any power interruption of more than 20 milliseconds can cause IT systems to crash. A power interruption of 60 seconds can result in a recovery period of hours or days to restart the IT systems and applications that have been affected. This can cost an organization millions of dollars (depending on the industry) and impact customer satisfaction and/or their brand and reputation.
Critical success factors to consider for power backup
The UPS is arguably the most important component in any data center and the UPS batteries are arguably the weakest link. The UPS system must be properly designed to achieve an optimized balance between capital costs and the availability of uninterruptible electrical power. The following elements represent some of the critical success factors that come into play when weighing the trade-offs of capital costs and availability:
Providing UPS power to the critical loads
Most data centers today incorporate UPS systems into their design that provide uninterruptible power to the IT equipment. However, the cooling systems that support the data center are often backed up by generator power only. With this design, upon loss of power the cooling systems will shut down momentarily until the facility is transferred to generator power. Depending on the cooling solution employed, it may take several minutes for these systems to recycle, start up, and come back up to full cooling.
In a low-density data center environment (up to 5 kW per rack, on average), the restart time gap may not be an issue. Low rack densities and plentiful cool air space (e.g., high bay ceilings) can provide a thermal ride through for a period of several minutes before IT component temperatures reach a point where the systems shut down because of thermal overload. For more information about how this works, download the APC white paper “Data Center Temperature Rise During a Cooling System Outage”.
Today, we are seeing more high-density racks being deployed in the data center. It is not uncommon to see hyper-converged racks in the 10-20 kW per rack range or blade server racks, as may be found in university research departments, in the 30-100 kW per rack range. Under these circumstances, instead of two minutes of thermal ride through, these compute racks may experience thermal overload in a matter of seconds. In these situations, it begins to make sense to have some, or all, of the HVAC system backed up by UPS power as well as the generator.
Designing for UPS reliability and availability
Given enough time, all mechanical and electrical systems will fail at some point. To counter this inevitability and achieve the desired reliability and availability, data centers are designed with redundant electrical and mechanical components and/or systems. Typically, the more redundancy you incorporate into the design, the greater the reliability and availability. However, the greater the redundancy, the greater the cost; both CapEx and OpEx. For example, a 2N data center design provides a fully redundant system, meaning, if one system can support the critical load, another fully redundant system is installed as a backup. This is a highly reliable design, but it also comes at a high cost. Most data centers today are achieving the desired reliability with an N+1 design topology.
To learn more about optimizing the design and performance of UPS systems, download our “Cost, Speed, and Reliability Tradeoffs between N+1 UPS Configurations” white paper.
Determining the appropriate UPS battery runtimes
Theoretically, you only need enough UPS battery runtime to carry the critical load until such time as the generators start and the load is transferred to the generators (typically 10-20 seconds). That said, the battery runtime will vary depending on the end user risk tolerance and the resiliency of the IT applications that are being supported by the UPS. For example, the internet giants are designing hyperscale data centers with 1-2 minutes of battery runtime. Cloud and colocation data centers are typically designed with 5 minutes of battery runtime. In the financial industry, you will typically see 10-15 minutes of battery runtime. The amount of UPS battery runtime is a customer business decision. Obviously, the more battery runtime you incorporate into your design, the higher the capital cost and the higher the annual maintenance cost.
Deciding which UPS battery technology to deploy
For decades the most common type of UPS battery technology deployed was VRLA. In recent years, a new generation of lithium ion batteries has been introduced and they are a game changer in the industry. This new technology has many benefits over the traditional VRLA battery. While these batteries may come at a slight CapEx premium, most installations will lower your Total Cost of Operation by 30%-50%.
Additionally, lithium ion batteries in many ways reduce your operational risk profile over that of traditional VRLA batteries. For example, lithium ion batteries have a life expectancy that is 2-3 times that of traditional VRLA batteries. This reduces the number of battery refreshes and the operational risk incurred to execute a battery replacement project. Test the benefits of lithium ion batteries yourself, with this comparison calculator.
UPS systems in data centers: Best practices for design and implementation
When designing and implementing UPS systems in data centers, following best practices is crucial for optimal performance. Effective data center UPS design includes selecting reliable and scalable UPS units, ensuring redundancy, and integrating efficient power management. For UPS in data centers, it is important to conduct regular maintenance and testing to prevent failures. By adhering to these practices, businesses can enhance system reliability and protect critical operations from disruptions.
Overcome power anomalies with UPSs and generators
Schneider Electric solution architects work with hyperscale and enterprise customers to help design efficient and resilient data center power backup systems. While every element is integral to a successfully run and maintained data center, the power architecture and its corresponding backup solution is critical. I hope this blog serves as an in-depth look at the necessity of power backup and equips you with the right questions to ask and important issues to consider as you make the best decisions for your data center.
