Data centers face specific challenges when it comes to addressing arc flash dangers. The only truly safe electrical system is a de-energized electrical system, but cutting power to critical servers when maintenance is required can mean losing both revenue and operational capability for connected clients. Even emergency equipment intended to ensure reliability, such as backup generators and batteries, can pose arc flash safety hazards with their own sources of available fault current. As a result, data center designers and owners need to develop a strategy that will maximize safety with a minimum impact on facility operations.
Arc flash events pose a potential for serious burn injuries or death – an arc flash can create temperatures as high as 35,000 degrees F, hotter than the surface of the sun. In essence, a plasma cloud develops around the affected equipment which could lead to a powerful explosion, sending vaporized shrapnel flying in all directions.
Over the last several years there has been an increasing amount of learning and awareness around arc flash hazards due, in large part, to the number of injuries and deaths. Investigations into these accidents have led to updated standards (including the National Fire Protection Association’s (NFPA) NFPA 70E Standard for Electrical Safety in the Workplace,) and an increase in regulatory enforcement. In particular, the data center environment is under increasing scrutiny due to the frequent need for new and upgraded branch circuits, without shutting down critical loads. Data center branch circuits are commonly added or upgraded on live electrical equipment, referred to as “hot work”, and in some cases the company has been fined. A case in point is the use of busway systems where plug-in units are inserted into a live bus, placing data center workers at risk. The Institute for Electrical and Electronics Engineers (IEEE) developed IEEE Standard 1584-2002, the IEEE Guide for Performing Arc Flash Hazard Calculations to define a range of calculations for quantifying arc flash risks at individual points in an electrical system. The results of these calculations enable managers to specify the personal protective equipment (PPE) required to address the established risk levels. But building in engineering controls to help mitigate such risks from the start can mean the difference between a “survivable injury” – what PPE is intended to enable – and no injury, at all.
For example, with arc-resistant switchgear (as defined in IEEE Standard C37.20.7-2007, IEEE Guide for Testing Metal-Enclosed Switchgear Rated up to 38kV for Internal Arcing Faults ), workers are permitted in front (with Type 1 equipment) and around the perimeter (Type 2 equipment) of switchgear with Category 0 PPE. Passive arc-resistant switchgear can still cause enough internal equipment damage to require significant repair or replacement. However, newer, more active designs incorporate fast-acting switches that can extinguish arcing faults in as little as ¼ cycle, reducing the risk to both equipment and personnel.
Of course, building data centers with explicit consideration for reducing arc flash risks can help boost both safety and uptime. One important first step is to undertake an arc flash analysis before the design is substantially complete, rather than waiting until switchgear and other electrical equipment is arriving for installation. Thinking ahead also could allow designers to incorporate redundant equipment into their plans, so shutdowns can occur without taking connected servers out of operation.
A single blog post can’t begin to provide full overview of arc flash dangers and prevention strategies. For a broader look at the specific arc flash issues data centers face, I suggest you check out a Schneider white paper on the topic, “Arc Flash Mitigation.”