MTech has been calculating the effects of maintenance on data center reliability for over 20 years. Our analyses have consistently shown that the risk of downtime is significantly increased by most maintenance practices. This message has not been welcomed by many in the industry, but some clients have reduced the frequency of maintenance on systems where our calculations and evidence show that the risks are especially high. Costs are always reduced, while reliability is increased or unchanged. The response to Covid-19 forced many data center operators to severely limit or curtail maintenance. Facilities adopted a “break/fix” strategy, where only failed components were serviced. This is a variant of condition based maintenance. While determining the condition of an operating component is often difficult, it is simple when the unit has failed. Steve will discuss MTech’s approach to incorporating maintenance into data center reliability calculations and review the results for key components. Then he will compare the results from incremental reduction in maintenance frequency with the effects of the 2020 shift in maintenance strategy. The inexorable increase in the size and power of data centers has resulted in moving medium voltage distribution systems from the utility interface to deep inside many data centers. Medium voltage has unique risks and requirements that are not widely known. Steve will review case studies of medium voltage failures and discuss the implications for maintenance and downtime.

Steve Fairfax (bio)
President
MTechnology