Choosing replacement batteries for an UPS

There was a great discussion with many practical methodologies at https://alioth-lists.debian.net/pipermail/nut-upsuser/2024-November/013828.html -- some excerpts follow.

General notes

One thing to remember about common PbAc batteries is that they tend to age and degrade. Speed of degradation depends on environment (e.g. temperature, electrolyte drying out) and stress (how many times was it actually used and recharged), but it is not uncommon for batteries to need replacement after 3 years of use. Some models are engineered for extended lifetimes, marketed for 5-10 years before replacement.

While a PbAc nominal voltage is 12V, this is where the battery spends most of its charge/discharge graph. A fully-charged battery may be around 13.5-14V, a "desulphurization" process can peak it around 15V, and it is effectively discharged at 10.5-11V.

Be careful when handling them, both due to sheer weight, and acid as well as poisonous lead metal inside, and due to them possibly being charged and able to give out significant current.

Notes from Harlan Stenn at https://alioth-lists.debian.net/pipermail/nut-upsuser/2024-November/013830.html

I hope the following is both correct and useful.

I have replaced a fair number (probably around a hundred) batteries in other UPSes and devices (like a Fuji X-Ray machine that takes 16 pieces of 12VDC at 22AH batteries). That's over 200VDC at a Significant current (the batteries claim 12V, but they are charged to over 13V).

I try to get high-rate batteries, and they are harder to find. As I recall, one should avoid "deep discharge" batteries for UPSes. A deep discharge battery is great for constant load stuff (like wheelchairs, golf carts, etc.), but UPSes need "shorter" bursts of lots of power.

I read the spec sheets and order the heaviest batteries I can find. More lead means heavier batteries, and longer life.

I check each battery's initial voltage. I have a note that says I expect at least 12.9V, but I haven't had a chance to verify this lately.

I check the spec'd net weight of the batteries with the actual weight of each arriving battery.

Every battery I have received from the vendors I continue to to business with has had a gross weight clearly above the spec'd net weight of the battery. Yay.

Every battery I have received from at least one other vendor has had a gross weight of at least 10% UNDER the spec'd net weight. I have immediately called for an RMA of these, with varying degrees of success in their return.

Oh, at least one vendor has offered both "regular" and "high rate" batteries, I ordered the more expensive "high rate" batteries, and what was delivered was "regular" batteries. When I asked about this, they said something like "we don't get the high rate batteries anymore" and they had no answer for why they still showed and charged me for the high-rate batteries when they knew they didn't actually have them. The RMA process here was ... bad.

Anyway, when I have a set of replacement new batteries, I then:

• put each battery on a charger for several hours
• rotating thru the entire set at least twice
• then I take my bench supply with a (self-made) octopus cable with individually-fused "hot" wires, and starting at .1V over the average voltage, start adding batteries in parallel, waiting for the current draw to reduce to a trickle. Once all of the batteries are connected in parallel and drawing a trickle, I start bumping the voltage by .1V until I get to either 13.5VDC (or the published trickle charge voltage) and I let this sit there for a day.

When all the batteries are "balanced" in the parallel string, I then install them into the battery tray(s) or directly into the unit.

Sometimes I will take the "old" batteries and run them thru the above process to see if they will "recondition". Sometimes we will use these for lighting or other non-UPS tasks.

Notes from Greg Troxel at https://alioth-lists.debian.net/pipermail/nut-upsuser/2024-November/013834.html

My own experience is that often batteries are not entirely right, both:

• capacity somewhat below
• a weak cell

I charge incoming batteries, and then do a controlled discharge at C/20h (where C is nameplate capacity) using a West Mountain Radio (WMR) CBA-II. I then have a plot of V vs Ah.

Beware that WMR software is seriously deficient because

1. it only runs on windows and
2. doesn't run under wine.

So it requires keeping a spare computer with Windows to use it. I hear that newer versions have published protocols and perhaps there is open source code to drive them.

Often, I'll have a nice discharge curve, but say 6.5 Ah vs 7 Ah spec. I call that good -- that is a clue how bad things are. And, my charging could be a little under, so I don't get upset about this.

Sometimes, it looks good for a while and then sharply decreases around e.g. 4 Ah. This I interpret as 5 ok cells and 1 low-capacity/troubled cell. Such a battery is going to be trouble.

My two strategies are:

• order 5 or 10 and sort. put matched good batteries in UPS units. Use less good batteries in less demanding applications.

  • Last time I did this I got 5, and I think 4 were ok and 1 a bit low capacity but not broken.

• order 10 from walmart. test them. return the bad ones to the store.

  • Last time I did this I think 8 were ok (> 6.3 Ah for 7!) and maybe 2 were either too low capacity or weak cell, and I just returned them. I think I've also had ok batches. I think the brand was UPG.

I note that while Harlan Stenn's tests above are vastly better than almost everyone, they haven't done controlled load testing so I think you don't really know if your preferred vendors' batteries fare better than my UPG/walmart experience. However, last time I looked at walmart and thinking about ordering, I was unhappy.

• I have about 200 Ah nominal of mostly-tired mostly-free-to-good-home batteries in parallel, solar charged, for ham use. Some of them are wheelchair pulls, when the user perceived them to be diminished, and my testing shows about half nameplate capacity. They are enormous, ~50 lbs for a 73 Ah, but multiple 35 Ah is still very useful in my application, vs someone who might be stranded if it failed. Because they are in parallel any battery sees low current. This strategy requires paying a fair bit of attention and I don't really recommend it!

I try to get high-rate batteries, and they are harder to find. As I recall, one should avoid "deep discharge" batteries for UPSes. A deep discharge battery is great for constant load stuff (like wheelchairs, golf carts, etc) but UPSes need "shorter" bursts of lots of power.

There are two types of "sealed" batteries. Both are properly called "valve regulated lead acid" or VRLA.

• Absorbed Glass Mat (AGM): electrolyte in fiberglass strands. rugged, high current. Used for starting batteries in (larger) small engines and cars. Most batteries are this type

• GEL: electrolyte in a gel. Cannot handle very high currents (C/1h bad, I think), either charging or discharging. However, if you avoid that, cycle life more like 1000 than 500 for AGM.

The word "deep discharge" is usually used to contrast the above types (as a group) with "starting" batteries. Starting batteries, almost always the "flooded" type (liquid electrolyte, some semi-sealed, some where you can add water), are optimized for very high currents, but do not do well if discharged more than a little. Typical car batteries are like this.

Anything not for starting a motor, where you intend to use the whole capacity, should be AGM or GEL. A UPS that is going to discharge in <= 1h -- which is pretty much all of them, should be AGM.

I would only consider GEL for a UPS if:

• the published and actual runtime is 4h or more
• the manufacturer says GEL is ok (partially this is about charging)

Other notes, Q&A

• put each battery on a charger for several hours

Do you have a recommendation for a battery charger? Or a list of features to look for?

• temperature compensation (float voltage varies with temperature)
• desulphation, for dealing with troubled batteries
• perhaps, higher voltage until current drops

I have https://optimate1.com/product/om3/ which I don't claim is the best, but it has been reasonable.

Anyway, when I have a set of replacement new batteries, I then:

• put each battery on a charger for several hours
• rotating thru the entire set at least twice ...

The purpose of waiting for a day is for stabilization? Waiting for one of them to die? Achieve a solid charge?

Harlan: Stabilization/get a solid charge. And it also provides extra time to make sure no battery is failing.

Greg: Lead acid batteries charge rapidly when presented with 13.5V and the batteries are discharged. The rate of charge will decrease a lot as they get close. It takes a long time at 13.5 to get to 100%. The other charging program is 14.2 or something like that until the current drops, but charging at such a high level after it is charged will damage it.

In my view, the most important thing is not to overcharge. UPS service is very rough on batteries, I think because of heat and overly aggressive charging.

I don't think it matters much if your batteries are at 80% 90% or 100% charge when inserted.

By individually-fused "hot" wires, I'm guessing that's a cable with connectors with a fuse inline.

Harlan: Yes. And one must Be Careful with these octopus cables - there can easily be unused ends, and having these live ends dangling around can be Dangerous.

When all the batteries are "balanced" in the parallel string, I then install them into the battery tray(s) or directly into the unit.

The goal there is to get them all to the published trickle charge voltage before installing them in the UPS.

Greg: I think Harlan's goal is to not have high-rate charging in the UPS, and for them to have a similar state of charge. This is particularly important if batteries are in series, as many units have 2 or 4. They all have the same charge current then. However, charging is more efficient at lower state of charge, so they tend to balance. This is much easier than lithium which seem to need explicit balancing.

Harlan: Yes, if all of the batteries are at the same level, the hope/intent is that they will all behave similarly while in use.

My (possibly incorrect) understanding is that as UPS batteries get "worked" and "used", and then recharged, the weakest battery will fail first, and as it fails/stresses the batteries to its right and left also get extra stress. The belief is that by balancing the batteries in parallel before installing them, they will all be at a similar strength (note that since there will be some variance in the amount of lead in the battery and with the chemicals, there will also be a variance in the longevity of each battery).

Sometimes I will take the "old" batteries and run them thru the above process to see if they will "recondition". Sometimes we will use these for lighting or other non-UPS tasks.

Which makes me consider my next steps: disposal. There are a number of commercial battery centers nearby. I'll start calling around before showing up with 20 batteries. ;)

I would expect that many places are happy to take them because the cost of accepting batteries and recycling is probably less than the cost of lead mining. My local Advance Auto Parts has been willing to take them in small quantities; I have not tried large quantities. Staples also will take e.g. a 7 Ah, and I have not tried larger.