Battery thermal runaway

One of our eight Lifeline AGM GPL-8DL batteries recently went into thermal runaway, and we’ve had a few questions on the nature of the problem and why we chose to replace the full house battery bank.

The lazarette smoke/CO alarm had gone off at 3am, and upon investigating we found a rotten egg smell (hydrogen gas) and a lot of heat in the lazarette, with water dripping from the ceiling. We dug around a bit more and found two batteries were at 170F on the outside of the case, and probably well over 300F inside. A normal battery temperature on our boat is around 80F. Two of the batteries were boiling their electrolytes out–one of our eight batteries had gone into thermal runaway and taken its pair with it.

A nice Nordhavn design feature is to have battery isolation switches for every pair of batteries. We can turn a switch to isolate the failed pair, and the boat continues to operate fine, just with less house battery capacity. That night, we turned off battery pair #3, ensured the batteries were cooling, and went back to bed. The following morning they were still at 131F. One question sent to us was what if that had happened at sea? We would probably have seen it sooner with more frequent engine room and lazarette checks, but otherwise there would be no difference: we’d just turn off the battery pair isolation switch.

All our chargers are multi-step smart chargers. They go through three phases: 1) bulk charge where current is as high as can be delivered and the voltage rises as charge goes up until it hits a max of 28.6V, 2) absorption where the voltage is held constant at 28.6V and current drops as the battery gets more full, and 3) float where the voltage is maintained at 26.6V. These voltages are assuming 77F batteries. The battery problem occurred while on float charge.

Thermal runaway can occur in most battery types including flooded lead acid, valve regulated lead acid, and even non-lead/acid designs such as Lithium-Ion. The general condition is when increased temperature cause more energy to be released which yields yet more temperature and a feedback loop develops. In flooded lead acid batteries, this can be caused by plate warping or plate material sulfating, and sloughing off to the bottom of the battery. The warpage or sloughed off plate material can cause a plate-to-plate connection, which generates heat, which leads to more warpage, more current, and more heat. Absorbed Glass Matt (AGM) batteries like our Lifelines are not prone to plate shorting from sloughed off plate material, and plate warpage causing shorts is not a common fault, but they still can suffer from thermal runaway. Fortunately, it’s not a very common failure mode. Usually batteries just get old, lose capacity, and quietly fade away. But, thermal runaway does happen and, when it does, the energy released is somewhere between amazing and scary.

Dirona‘s Lifeline AGM batteries are rated for 1,100 cycles down to no less than 50% charge. They have seen far more than that, so we were getting close to replacement time anyway. We could just change the two damaged batteries since the rest continue to operate fine. But, with the use they have had, the bank was due for replacement some time back. We expected that we’d need to change them in Hawaii, but they tested fine at that time (using a Midtronics MIDMDX-640 Digital Battery Analyzer).

We now need eight Lifeline GPL-8DL batteries that list for a booming $8,264. And they are 156 lbs each, which each means we’ll be changing a half ton of batteries. With one string of two batteries disabled, we are down to 75% capacity but otherwise there is no change. So fortunately we don’t need to be in a rush to replace them.


If your comment doesn't show up right away, send us email and we'll dredge it out of the spam filter.


14 comments on “Battery thermal runaway
  1. Michael says:

    I notice that it was a smoke/CO alarm that triggered. Do you think that might mean there’s no need for a temperature sensor to catch battery failures like this? All the temperature sensors I have found for our new alarm system are pre-set to trigger at 54C/130F which I think is going to cause false alarms in the lazarette in very hot weather so I’m hoping I’m right to select just smoke/CO sensor for that compartment.
    Michael
    N4052 Coracle

    • The smoke/CO alarm did trigger on this battery thermal runaway but I’m not 100% sure that always would happen. When it did trigger the battery case was already at 160F and the only thing keeping the heat that low was the energy being dissipated off in steam. Clearly the plates were up at 212F and, once the water was boiled off, the battery case would be up there or well beyond quickly.

      To be effective, temperature sensors would need to be placed on the battery case since, even a foot away, the ambient temperature isn’t up appreciably and, as you said, you want a design that won’t produce false positives in hot weather. Putting temp sensors on each battery would require 10 temperature sensors which seems like a lot so we are relying on catching thermal runaway via smoke/CO sensors. Since these batteries hold a lot of water, there is a lot of steam released on thermal runaway and the combination of hydrogen sulfide and steam seems to trigger the smoke/CO detector.

  2. Your logic on price comparison looks reasonable to me. The only nit is that Lifelines are rated at 1,000 cycles rather than 350 if used on a 50% cycle as you propose above. In my expereince, they seem able to run far more. That doesn’t change your broad overall conclusion of +20% for a Lithium based battery system. It’s getting quite close at this point and continues to close. I’ve been watching these for other application since 2005 and meet with manufacturers every couple of years to see what progress has been made.

    Other factors worth considering after cost are: 1) charging system needs to be calibrated for the battery chemistry (not hard with a programmable smart regulator if the battery requirements are well documented), 2) Lifelines have been broadly applied for years and generally are a fairly well known quantity whereas the Sinopoly product is less well known and less broadly tested (doesn’t indicate a problem but early customers need to be pioneers), 3) Lithium based batteries are very high power density (that’s is what we want) but, as a consequence, designs need to be carefully worked through to ensure that the battery is sufficiently cooled for the expected operating conditions and there are inter-cell protection devices or fuses to prevent fire (generally not a problem but there have been numerous laptop and cell phone battery faults that lead to fire (see the Tesla Li-Ion battery design for an example of a very well thought out protection system), 4) these batteries are not available in an 8d form factor so new battery mounts need to be fabricated, 5) some changes will be needed in the boats DC wiring to accomodate 12 batteries in series.

    I’m confident that a very reliable marime battery system can be built using Lithium technology and I expect it will be the norm in 5 to 10 years. I haven’t any experience with Sinology.

    If you do decide to be an early adopter, I would love to hear more about your design and experiences over time. We will probably eventually go to Lithium based batteries on Dirona.

    P.S. It’s James rather than Matthew.

    • Rookie says:

      Dear james,
      I, too have been looking at this FePo4 technology and in the future it may be the way to go BUT the system you have on Dirona allowed you to simply switch off the troubled bank when a problem arose. If using 3 volt LiFeP04 you would not have been able to simply switch out an affected battery or pair without losing the whole bank. Add to this the very real fire risk and I feel at this point in time there is simply too much risk in using these LiFePo4 batteries. Just thrown in as food for thought.

  3. Mark McGillivray says:

    Hello Matthew, I understand you are from Vancouver Island..? Although we have lived in New Zealand since 1982, Carol and I originate from Port Alberni.
    Regarding our LiFePO vs AGM discussion, lets assume we have 8 Lifeline 8D batteries at US$610 each for N5002 12 volt house bank giving an Ah rating of 2040. Staying within a recommended AGM draw down of 50% gives a usable capacity of 1020 Ah at a cost of US$4,880.
    Now take 12 LiFePO batteries at US$500 each for the 12 volt house bank giving an Ah rating of 1200. Assuming a 85% draw down this also gives a usable capacity of 1020 Ah at a cost of US$6,000. This is only a 19% (US$1,120) higher cost for the same usable capacity but with the advantages of: higher CAR, stability of voltage through it’s discharge, lower efficiency losses during charge and discharge, and most importantly the capability of thousands of charge/discharge cycles (compared to say 370 cycles @ 50% draw down for AGMs*) this 19% cost disparity starts to look quite inconsequential. Other advantages including lower weight, freeing up more on-board space, and lower labor costs and down-times due to longer life-cycles. I contend that with all other things being equal, the time may be right to consider LiFePO.
    However, aside from faults you may find in my reasoning above, “ all other things being equal” is quite the unknown and therefore a large risk. I do not know if I could find anybody that I would have confidence in to design and implement a complete solution.
    Matthew, your esteemed critique would be greatly appreciated. Your assistance in identifying the weaknesses of my argument would be far preferable than my forging ahead with a misinformed upgrade path
    Thank you and Merry Christmas to you and Jennifer.
    *Nigel Calder, 4th edition page 17

  4. Mark McGillivray says:

    Your reputation and blog are justifiably renowned. I have studied your postings in depth and appreciate the detail you include. Would you mind providing more detail on your choice to replace your house batteries with lifeline? With the pricing of, apparently good quality Sinopoly LiFePO4 400 AH batteries at less than US$500 + freight, why did you choose AGM? We recently purchased N5002 and must not only replace the house batteries but also look at the complete system setup including inverter and charging system.

    • It’s a good question Mark. Where battery weight or size is an important consideration, Li-ION and related Lithium based chemistries are the best solution. That technology dominates mobil device, aircraft, drones, some electric cars, and many other industries where high power density needs to be combined with low wight. You asked why we didn’t go with Sinopoly LiFeP04 400 AH batteries at aroudn US$500 each. There are several reasons but the dominant reason is cost. These batteries are 3.2V so it takes 8 batteries in series to produce a 24V bank. To replace the battery capacity in Dirona, we would need 1020A@24V. Each string of 8 woudl produce 400Ah at 24V so we would need 2.5 strings.

      Let’s try to make 2 strings of batteries work. That would be 800 Ah at 24V. This isn’t quite enough power to replace the 1020 AH at 24V on Dirona. Let’s go with the 500Ah LiFeP04 battery at $675 each. That combination would give us 1000AH@24V but would require 16 batteries at $675 each. That’s $10,800 worth of batteries before shipping. It’s a nicer and smaller package than Lifeline but considerably more expensive.

      The short answer is these batteries are dropping in cost fast but they are still much more expensive than Lead Acid variants and consequently, in non-weight and space sensitive applications, lead-acide variants still win.

      Congratulations on your new boat. The Nordhavn 50 is a very strong and comforable boat. One of our favorites.

  5. I don’t follow how paralleling batteries could possibly induce thermal runaway. Can you point to more detail on that one Jeff?

    Generally, the Concord battery folks are pretty good and their manual for Lifeline batteries shows parallel and serial combinations as supported configurations.

    –jrh

  6. Jeff Stander says:

    Sorry to read about your battery difficulties.

    The problem is that you have four parallel strings of two batteries each. This number of parallel strings is a major problem as it not only leads to premature failure; it often leads to the thermal runaway you have experienced.

    What you need to do is reduce this to possibly one or two strings of 12 2V cells. A single string (12 x 2V, 900AH) costs the same and will last you easily 7 years if not more.

    You are also paying a huge premium to be in NZ. Almost double. Your 8D batteries cost $580 each in the USA and you can ship them on a pallet to NZ for about $800. Your total cost (assuming no duty) is about $5500. I did that with Lifeline AGMs sent to Australia from the factory in Azuza.

    I have done a lot of business with thesolarbiz.com. Talk to Tom Duffy, their systems engineer. I think he can design a better battery setup than you have now.

    Good luck

    Jeff

Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.