Improving Data Center Availability with Single-Phase UPS Systems

The mission-critical role that data centers play in supporting global business operations cannot be overstated. The rapid adoption of hybrid IT and cloud services, the growth of artificial intelligence (AI), and the ability to support a remote workforce have made data centers essential for storing and transmitting the “1s and 0s” the world economy runs on. The connectivity of home automation devices, as well as the dramatic increase in e-commerce shopping and virtual conferencing during the COVID-19 pandemic, has also bolstered data centers’ significance in everyday life.

To give a sense of the volume of information being processed by data centers, consider eBay. It processes 300 billion data queries each day, with a data footprint exceeding 500 petabytes (4e+9 GB). That is enough to back up the American Library of Congress more than 300 times.

While the economy depends on data centers for a variety of reasons, the more than seven million data centers operating around the world depend on one common factor for their 24/7/365 availability: the reliable supply of electrical power.

Downtime costs

Data centers that are experiencing regular power outages risk their temperature control systems shutting down, potentially leaving the estimated 18 million servers deployed worldwide within them vulnerable to the dangers of overheating and condensation changes. Both can limit a server's capacity to safely store data and may permanently damage its operation. According to a 2018 survey by Uptime Institute, 31% of data centers experienced a power-related downtime incident or severe degradation that year, and 48% reported at least one outage at their site or at a service provider in the previous three years. This problem has only grown more severe ― and more expensive ― since then.

Downtime due to a power outage has grave consequences for data centers. According to a study by Gartner, “Ensure Cost Balances Out With Risk in High-Availability Data Centers,” data center downtime costs $5,600 per minute (based on the study) or between $140,000 and $540,000 per hour (based on another study conducted by Ayaya, “Network Downtime and Complexity Results in Job and Revenue Loss Plus Missed Business Opportunities”), depending on the organization. Based on an average reported incident length of 90 minutes, the typical cost of a single downtime event was approximately $505,500.

These costs are calculated applying factors such as data loss or corruption, employee and equipment productivity reductions, equipment replacement, root-cause detection and recovery actions, legal and regulatory repercussions, revenue loss, and the long-term repercussions on reputation and trust among key stakeholders. While business disruption and lost revenue are cited as the most significant cost consequences of downtime, other less obvious costs also have a significant impact on an average downtime event. For example, every second a data center experiences downtime, their clients experience downtime as well, creating a cascading effect of losses.

UPS battery-related failures

A survey conducted by Emerson that polled 63 data center organizations in the United States that had experienced an outage in the past 12 months showed that more than 39% of data center outages were attributed directly to vulnerabilities in the facility’s power. Among the general root causes of data center downtime related to power, uninterruptible power supply (UPS) battery failures proved to be the costliest ($687,700) and accounted for one-quarter of all such events. The battery in a UPS proved to be the most vulnerable part of the system.

Having UPSs is essential in any data center, large or small, to ensure service continuity. A UPS provides critical backup power if the primary power source goes offline due to a disruption or a catastrophic emergency. It bridges the short-term loss between electric utility power failure and backup generator operation. In addition to backup power, a UPS offers protection against surges, sags, and other disruptions that can severely damage connected devices, reduce their lifespan, or affect their overall performance.

Why do UPS batteries fail?

Like every battery, UPS batteries have a built-in lifespan (approximately three to six years) and require replacing when they can no longer supply 80% of rated capacity in ampere-hours. However, UPS battery life may be compromised by several factors other than time. For instance, if ambient temperatures are too warm (or if the battery is subjected to wide temperature variances) it is put at serious risk of degrading. In general, if the temperature drops below 22°C, UPS batteries will underperform or sustain damage. On the other side of the ledger, operating above 25°C will result in increased battery capacity but reduced battery life. As a guideline, every 8°C rise in temperature above 25°C cuts the battery life in half.

Over-cycling is another culprit. “Cycling” refers to when a battery is discharged and recharged. If this happens too often, it reduces the capacity of a battery and causes deterioration of the battery contacts. Constant discharging of batteries will send them to premature end-of-life. In addition, UPS batteries may fail due to incorrect float voltage, leaving them in storage too long without recharging, or as a result of simple human error.

UPS battery overheating

Overheating is by far the most frequent cause of UPS battery failure and the acceleration of the aging process. Even batteries designed for high-temperature chemical reactions are not immune to heat-induced failures due to parasitic reactions within the cell. Overheating will happen in a data center due to:

Air conditioning not producing a sufficient volume of chilled air. This event commonly happens when an older UPS is replaced with a larger capacity UPS. Hot summer days can also lead to air conditioning not providing the needed amount of cooling. Air conditioning units must be serviced regularly to ensure the proper cooling of each UPS system.
Dust build-up inside the UPS is an unseen villain that causes overheating of not just the battery but also of all the components of the UPS.
Fans used to cool the UPS may fail. In a multi-fan system, a single fan failure may go unnoticed.
An overloaded UPS continuously operating at 100% or higher will overheat.
A UPS that is installed in an area without proper ventilation will overheat.

Single-phase UPS units

A single-phase UPS is generally used for smaller loads within the data center, such as for HVAC and control equipment, network workstations, VoIP, rack or distributed servers, and safety/security systems. Smaller remote IT centers and edge data centers may rely entirely on single-phase UPS units to keep infrastructure operational.

Until recently it was thought that only loads less than 20kVA could safely use a single-phase UPS. Yet most quality industrial-grade single-phase UPS units are now fully capable of handling much higher kVA applications in a broader range of direct current voltage without failure. For low- to mid-power requirements, single-phase UPS units are more efficient than 3-phase UPS units, striking the right balance of price, energy density, power, and resiliency. Correctly specified and installed, an industrial single-phase UPS will intercept virtually any potential power disruption the data center may experience while maintaining electricity at a consistent rate and providing critical power if the commercial power source is down.

Types of single-phase UPS units

In general, classifying single-phase UPS units comes down to three basic characteristics.

Topology: For the most part, UPS topology comes down to a choice between standby (basic), line-interactive, and online. Each has its advantages and disadvantages. In single-phase applications involving a control system, a good choice is to use a standby or line-interactive UPS with a simulated sine wave in battery mode, a technology based on approximated sine wave output waveform. This type of UPS delivers a combination of low cost, light weight, and a small footprint for non-sensitive loads. A UPS that utilizes heat-sinks with further increase reliability, especially in dusty environments
Form factor: When choosing a form factor, the main consideration is where the UPS is going to be installed. Standard UPS form factors are desktop, tower, rack/tower, rackmount, and DIN-rail. In data centers, it is best practice to install a small single-phase UPS on a DIN rail, which is a metal rail of a standard type widely used for mounting circuit breakers and industrial control equipment inside equipment racks. The lighter the UPS and the smaller its footprint directly translates into more space for additional equipment on the DIN rail with the ventilation needed to keep mounted equipment cool.
Battery type: In industrial and data center areas, a good choice for a single-phase UPS battery is sealed, valve-regulated lead-acid (VRLA) technology. VRLA batteries are rechargeable and are considered low-maintenance. In recent years, Lithium-ion (Li-ion) UPS systems have become available in some UPS designs. Being volatile, Lithium makes certification difficult in an industrial or data center area. Also, Lithium has strict battery management system requirements that measure each cell; if one exceeds ~50°C to 55°C, then the whole system shuts down without warning. Most commonly, Li-ion is found within three-phase UPS systems deployed to provide resiliency in industrial plants or large data center facilities.

Selecting a single-phase UPS

Specifying a UPS solution with optimal power protection is essential for data center availability. To ensure that the UPS matches your needs, consider these factors during the selection process:

Capacity: UPS capacity is simply how much power a UPS system can provide. To calculate the load, create an equipment list based on the total watts each device requires to run. The higher the UPS capacity, the more devices it can support. If requirements exceed 16,000W, consider a 3-phase UPS, or divide equipment into groups to be supported by several single-phase UPS units. When replacing an existing UPS, keep in mind the IT load may have changed since it was originally installed.
Runtime: Backup runtime refers to the duration the UPS can bridge power to the devices it is supporting during an outage.
Space: Data center real estate is always at a premium for IT and facility managers. Specify a UPS with the smallest footprint and lowest weight without compromising protection.
Energy efficiency: While the efficiency of a typical UPS ranges from 94% to 95%, that rating plunges as the load decreases.
LCD panel: LCDs display critical data points at a glance such as voltage, low battery, frequency, and backup time, permitting easier system management. There are also audible alarms that notify the facility manager of status conditions.
Redundancy: To meet uptime requirements for data centers, UPS are often deployed with redundancy systems. There are three main UPS redundancy architectures: N+1, 2N, and 2(N+1). N is the full UPS capacity required to handle the total load or the same as non-redundant.
Connection ports: UPS may have several connection ports for your application. Serial ports connect a UPS unit to a computer. USB ports are used for communication. An RJ45 port or a network management card can be used to control and configure the UPS remotely via a web browser or network management system.
Ratings: The three major data center design and infrastructure standards developed for the industry are Uptime Institute’s Tier Standard, ANSI/TIA 942-A 2014, and EN 50600 (International). There are also operational standards for day-to-day processes. Specifically for UPS units, there are IEC 62040 and UL-1778. UPS units can also be rated explosion-proof (Class 1, Div. I), corrosion-resistant, and sealed to prevent moisture ingress.
Temperature and humidity: ASHRAE revised its acceptable operating range for data centers upwards from 18°C to 27°C (64°F to 81°F). While this saves power and money, it gives facility managers less time to react to escalating temperatures. ASHRAE’s 2016 guidelines for data center humidity is 50%; minimum humidity is set at 20%, while maximum humidity is 80%. You should not purchase a single-phase UPS assuming nothing could go wrong with environmental systems. The UPS you specify must have a wide operating temperature range outside of ASHRAE’s recommendations to prepare your data center for a worst-case scenario.

UPS network monitoring

Some IT departments manually assess UPS health by seeing if an alarm is sounding or a fault indicator light is on. Upgrading to using a network to monitor UPS units on a website dashboard is a major step forward in that it minimizes labor costs and, if properly used, can dramatically reduce battery problems.

Warning signs, such as deteriorating performance or an overheating battery, result in the sending of real-time notifications by text or email. Technicians can make repairs or battery replacements before serious breakdowns have a chance to occur.

Is the future underground?

One of the most dramatic trends in data centers is the move away from traditional “clean room” environments to damp, underground sites. Data storage requirements are growing exponentially, yet available real estate is not. The massive fleets of hard drives and servers contained within data centers require hundreds if not thousands of acres of land. Increasingly, this simply isn’t cost-effective, leading companies to shift to new underground facilities. Rather than constructing a building from scratch or occupying space in an existing building on a lease, it is far cheaper to use an abandoned bunker, cave, or mine.

Besides reducing real estate costs, subterranean data centers have geographical and geological advantages, such as zero solar heat gain, low ambient temperature, natural geothermal cooling, and solid rock surrounded structures. These factors dramatically reduce cooling costs, plus may improve physical security.

The downside to subterranean data centers is moisture. Damp walls and elevated humidity levels can cause havoc on ordinary location UPS systems. While exposure to moisture might not immediately cause a UPS failure, it facilitates corrosion of the cabinet and its internal components, leading to UPS failure. Simple dehumidification equipment may not be sufficient where humidity is a problem.

Using an industrial-grade, IEC 60068-rated UPS will provide the optimum backup protection of critical data center infrastructure from moisture. IEC 60068 offers guidance for the environmental testing of electronics, including UPS units, such as heat, cold, and humidity. Specify UPS models with maximum permissible relative humidity levels of 90% or more. Because standard UPS units may not have been subjected to IEC 60068 testing, they are not appropriate for underground data center use.

Conclusion

Compared to large 3-phase floor UPS systems, single-phase UPS units are less visible yet can play a vital role in maintaining modern data centers. Behind the scenes, single-phase UPS units protect systems such as the controls for the facility’s HVAC, workstations, VoIP, and safety/security. For low- to mid-power requirements, single-phase industrial-grade UPS units are more efficient than 3-phase UPS, offering an optimal ratio of price to resiliency.

Matt Gurreri works at the Rosemont, Ill., Emerson Appleton Division Headquarters as product marketing manager. He can be reached at [email protected].