In “Battery Life” (Spring 2004), Dennis DeCoster recommends that in a multiple battery string system of N+1 type, we should maintain different life batteries for reliability. A few months ago, when I was in the process of changing the batteries in more than 10 systems, I recommended the same thing. However, I was told by the battery consultant and the battery vendor that we should not have different life battery. According to both of them, the older and supposedly weaker battery string can take down the whole UPS. For in-creased reliability, both the vendor and the consultant recommended all batteries should have similar age in all strings.
Jag Bansal, Prudential Financial, Newark, N.J.
Dennis DeCoster responds: My article dealt with common “5-year” design life AGM-UPS batteries, not the “20-year” premium VRLAs, which have considerably more negative discharge issues (hydrogen imbalance) and varying chemistry (catalysts).
If your battery people are referring to paralleling of individual batteries within a string, they would be correct. It is also never a good idea to replace old battery strings with different brands or models of VRLA batteries in parallel strings on the same charger. The impedance mismatch can lead to charging imbalances which may shorten the life of batteries.
Regarding staggering the replacement dates, which is the central issue of this article, one can also make a case for the same thing about replacement of different ages of batteries in a multi-string arrangement. In an ideal world, where cost is not an issue, a truly mission-critical site might change all the VRLAs out no longer than every three years. In fact, if cost were no issue, a truly mission-critical site would be using flooded batteries which are an order of magnitude more reliable. But we don’t live in an ideal world. We routinely replace random bad batteries in strings as they age anyway (mixing ages and impedances), and we routinely see many VRLA string sites pushed to 5, 5.5 or 6 years, or even more.
If one staggers string replacement, it does not relieve the need for good maintenance and proper design. Assuming a target string service life of 4 or 4.5 years rather than 3 years (with 70
The predominant failure mechanism for VRLA batteries remains dryout—an open circuit condition causing loss of the entire string. This risk clearly increases with age. At some point and with some mix of operating conditions, it goes up near exponentially. Pushing all three VRLA strings in an “N+1” arrangement to 4 or 5 years before group replacement carries great risk.
If one does decide to target a 4- or 4.5-year string life rather than, say, 3 years, there will be undeniable added risk for the cost savings achieved. If trading between three strings all replaced at year 4.5 but with good charging balance, or a rotating replacement schedule where at worst case only one string was older than perhaps three years but there may be some level of undesirable impedance imbalance, I would without doubt take the latter.
Dennis DeCoster, Mission Critical, West Redondo Beach, Calif.