I'm sorry if this is an obvious question, but I've been scratching my head with it for a while.

I am used to UPS backup batteries that have an autonomy time between 2 to 4 hours, so I have always discharge tested for 3 hours down to a cutoff voltage of 1.8 Vpc. I have had issues going lower than that where batteries started to swell during recharge, although I think that in theory I could go down to 1.7 Vpc.

Recently started working on a datacentre with an autonomy time of 15 minutes. During the commissioning autonomy test the batteries discharged down to 1.83Vpc, and have been designed assuming an 8 year service life so will eventually drop down to 1.6Vpc. The manufacturer gives a 12 year life, but there are design factors for the fact the temperature is uncontrolled (!).

I have checked with the manufacturer who have told me that these batteries are OK to discharge lower, however some other people at work insist this will irreversibly damage all of the batteries without really being able to explain why. The manufacturer hasn't been so clear on why it is OK either, but they are a very large and reputable company.

My understanding has always been that a number of chemical reactions in a lead acid happen at a fixed rate independent of the rate of discharge, so that

  • the amount of available energy is lower, so 1.6 Vpc at C0.25 is a lower % discharge than 1.6 Vpc at C3
  • the equivalent change in internal resistance is lower, so although at C0.25 the internal resistance starts higher, over the course of 3 hours, the instantaneous i2r loss at a lower discharge rate may end up higher

My personal thoughts are that I tend to ask for a discharge test once every 3 years, and using a cutoff of 1.7 Vpc gives a decent margin of error. If they don't last for 8 years then that's on the designer's head.

  • \$\begingroup\$ What does think a chemist about it ? Chemistry reactions when this occcurs ? \$\endgroup\$
    – Antonio51
    Commented Nov 17, 2021 at 13:56
  • \$\begingroup\$ Given the price of LiFePO4 these days and their vastly superior performance, I'm a little surprised lead-acid performance is still relevant. \$\endgroup\$
    – user16324
    Commented Nov 17, 2021 at 14:01
  • \$\begingroup\$ @user_1818839 in my experience a lot of customers are highly conservative and won't touch anything with the word "lithium" in it because of the perceived fire risk \$\endgroup\$
    – scc268
    Commented Nov 17, 2021 at 14:10
  • \$\begingroup\$ @scc268 yes ... even though the fire risks of completely different lithium chemistries don't transfer to LFP. \$\endgroup\$
    – user16324
    Commented Nov 17, 2021 at 14:14
  • \$\begingroup\$ The lower you discharge a battery composed of multiple cells in series, the higher the risk of overstressing a cell. That depends on how equal the individual cells behave versus each other. The industry (including battery makers) advice to limit discharge at 50% of capacity to maintain them in good health. Personally I'ld monitor how the discharge profile to 50% evolves over time to determine if the battery is still good or not - if you have more than one system you can compare them to find the outliers. \$\endgroup\$
    – le_top
    Commented Mar 25 at 8:42

1 Answer 1


This question reminds me of another I answered recently), about why there's several lines in the battery's datasheet Ampere Table (it's depicting different life-cycle results (min to max) according to discahrge-Voltage treatment (because discharge cutoff voltage is the most commonly used technique, rather than the more ideal coulomb-counting).

It's not clear to me what you mean by "C0.25" - do you mean "0.25C" aka "C/4"? That's a relatively high discharge rate for PbA; datasheets tend to default to C/20 discahrge rates (i.e. yielded capacity over a 20 hour run-time); outside of that, you're 'on your own'.

Also unknown in your question is whether you're quoting 1.83Vpc disrcahrge cutoff voltage when the battery is under load (in which case you have to take into account I^2.R voltage drop at a minimum, including at the system/installation level, or after the battery has 'recovered' during a period of no-load, when battery V recovers significantly (the former is of course the most common reality, but accounting for known internal resistance is less common ;) )). The two different measurement techniques yield completely different outcomes and resulting battery life-cycle results.

1.8Vpc discharge cutoff is reasonably safe and should yield a cycle life at the higher end of the manufacturer's spec. 1.6Vpc is fairly brutal, and will dramatically lower the life cycle count.

Specifying a battery lifespan in "years" is manufacturer-speak, and is only part of the picture. 8 years, with what treatment? (i.e. how many cycles, to what depth-of-discharge, over those 8 years???) Maybe the manufacturer specified this in their datasheet, maybe you have to ask for it. I wouldn't promise a customer anything in terms of warranty until I had that completely known at the time of designing a solution leading to the signing a contract for installation :-)

It's impossible to tell from your question if there's "specification implications" in the distinction between your ~3 hour UPS run-time example, and this new 15-minute run-time example (is max C discharge rate being respected in the latter/15-minute example?); but there's a tendency amongst too many insufficiently educated UPS/battery installers to 'flog' batteries harder than one otherwise might if it's only for 15 minutes, rather than 3 hours. It's not rational, but that's the way some people think about the problem.

A battery that's bulging during recharge (assuming it's a sealed type) means H2 gas is being generated and bulding up pressure, probably because it's being recharged too quickly (too high of a C rate). If you're lucky the H2 will recombine later; if you're not, the battery pops (hopefully a pressure relief valve) & H2 escapes forever (and what safety implications of escaped H2 might be, are of course installation-dependent), so the water/acid-solution level drops, never to recover.

Re: One of the comments to your question, LiFePO4 doesn't suffer the high risk of thermal runaway that's commonly associated with the NMC & NCA "lithium" (as commonly used in gadgets, and thanks to Tesla, also in grid-scale battery solutions).

  • \$\begingroup\$ The batteries are Exide Sprinters that appear to be optimised for short discharge times. You're right, 0.25C ie 15 minutes. The 1.83Vpc comes from the terminal voltage of the batteries at the end of a commissioning autonomy test, where the UPSes support design maximum load (using a load bank) for 15 minutes. Because they are UPSes in the UK with reasonably good power quality TBH I would not expect the batteries to discharge at all, or maybe once or twice in their life, plus annual autonomy test and three yearly discharge test. \$\endgroup\$
    – scc268
    Commented Nov 19, 2021 at 10:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.