What drives the process for when to replace UPS batteries?

Posted by on Jan 15, 2014 in Up For Discussion | 5 comments

I want there to be a place where we can share our processes and learn from each other.  This is the first of (hopefully) many discussions exploring a topic of interest. Please share your experiences and best practices with everyone in the comments section below!

I have seen organizations change UPS batteries based upon several different factors. In my experience, some replace batteries on a set schedule, some upon a qualifying condition.

Since batteries represent a significant investment, I think it’s reasonable to assume that we are all following some process for making this costly decision.

What criteria or process do you use to determine when to replace all the batteries in a string/UPS, and why?

(Editor’s Note: Comments will be moderated and posted as quickly as possible.)

5 Comments

  1. The following is an excerpt from a technical paper presented at the fall 2013 Intelec conference Information on the conference can be found at http://conference.vde.com/intelec/documents/intelec2013_report1.pdf .

    This subject is controversial, both within the industry, in many technical forums and battery standards making bodies. In other words, there are no simple answers to your question.

    “Continuous monitoring and prognostics based on battery ohmic values, plus other battery measurement parameters are essential to predicting end-of-life conditions. This is true for individual cells and batteries, as well as a complete battery system. Measuring and archiving these values over time, plus prognostic and trend prediction of other measurement parameters that affect these values, allow for predictive analytics. Observation, collection of data, and the use of proven mathematical prognostic techniques and models can be combined for lifetime prediction and forecasts. This process helps maintain battery unit and system state of health (SOH), state of
    charge (SOC), and the resulting prediction of remaining useful life (RUL).
    Prognostic methods are used to monitor and track “degradation paths” in one or more battery parameters that are correlated to remaining life. Deviations between the observed and expected parameters are evaluated to determine when a monitored unit is degrading excessively. Remaining life is predicted by classifying the unit’s degradation path. in reference to degradation path models calibrated with run-to-failure data for units experiencing a similar mode of
    failure.”

  2. Technical aspects of your commentary are clearly correct and well thought out. Have you considered the mandated maintenance per OEM requirements stipulated by CMS for Health Care in regards to equipment in any data center that might warehouse critical medical information?

    • John –

      Could you send the link on “CMS for Health Care” That defines subject OEM requirements referred to in your message?

      • Center for Medicare and Medicaid requirements, I believe the paragraph references are: SC&L 04-07, “Hospital Equipment Maintenance Requirements”

  3. There is no single answer to this question. One thing is certain with any battery installation, regardless of type, chemistry or application at some time the battery will fail. Therefore the replacement policy will typically be determined by criticality of the load and just how ‘risk averse’ the end user is.
    Some will use a replace on age policy because cost of maintenance just is not justified, others will perform some maintenance but still replace on age, maybe adding a year or two to the age ‘limit’, not many will risk running a battery up to the full design life!
    Bart is right to say that a prognostic, state of health, condition based maintenance regime is the optimum way to optimise battery life. But this is an area that requires expertise and experience to get it right!
    If the users want to minimise risk while optimising service life and TCO then condition based PM is the most effective strategy.

Leave a Comment

Your email address will not be published. Required fields are marked *