Used 18650 failure rate/sanity check?

[MysanthropicBuilders]

Long story short, looking to build an off-grid house with solar (Don't worry, I'm in the planning stages, construction starts in 2-3 years). Was planning on making my own ~30-40kWh battery bank with used 18650's. The problem is... We're talking about 4,000 or so individual cells. This isn't going to be a hobby, or a fun project... It will be my source of electricity, with only a gas powered generator for backup. I'm not an electrical engineer, but I'm also not an idiot... Which means I'm going to prioritize safety, and take everything slow. It also means I know really very little about this, and this post is more of a sanity check of "can this realistically be done safely" rather than "Hey, which parts do I need to make this happen?"

So... What happens when those cells inevitably begin to fail? Obviously I'll have to go around occasionally and remove any dead/dying batteries, but I'm worried about a cascade effect. One fails, puts slightly more strain on another one and it fails too, the two failing in the same pack makes that whole pack fail, suddenly all 4,000 cells are either paper weights or on fire... That kind of thing. Would a setup like this need constant monitoring and instant removal of dead cells to remain stable? Or would it just be a lot more work to set up compared to a server rack style battery pack? And if this can possibly be answered... Does anyone know about how long a 30-40kWh (assuming it was being cycled daily / every other day) pack made from recycled 18650's would last? Obviously this can't be an exact number (the pack could last 10 years or 10 days before 80% of the cells fail), but anyone with experience that could give me any sort of a ballpark would be appreciated.

 

[meetyg]

Well first, for 30-40kwh, that's gonna be alot of work.
You will need to capacity check, sort and match each and every cell. Let alone combining them into packs, soldering/spot welding etc...

Second, your worries are true: Using used cells, you never know what your gonna get in the long run. Some might fail prematurely. Some might just underperform.

Third, safety: 18650's are usually Lithium Ion (NMC/NCA etc...) which are more hazardous than say Lifepo4. A single cell failure can cause thermal runaway and cause a fire. Less so with Lifepo4.

Fourth, lifespan: Most Lithium Ion chemistries are good for a few hundred cycles, some can even reach low thousands, if used conservatively.
Having that you plan on using used cells, you really don't know how many cycles your gonna get out of them.
In contrast, Lifepo4 has a few thousand cycles (anywhere from 2000 to 6000, depending on DOD and specific cell model).

I personally would not go down the 18650 path. It might be tempting as you can get them cheap/for free, but really your time and safety isn't work it.
IMO, 18650's are great for small battery packs, but less appropriate for larger capacity packs like you are planning.
Having said that, there are some people here on the forum who have done that.
But in this day and age, with Lifepo4 prices coming down (at least Grade B), I would say it's not worth doing it with 18650s.

 

[GSXR1000]

how much are you saving compared to 32x 280ah lifepo cells?

 

[MysanthropicBuilders]

Well first, for 30-40kwh, that's gonna be alot of work.
You will need to capacity check, sort and match each and every cell. Let alone combining them into packs, soldering/spot welding etc...

Second, your worries are true: Using used cells, you never know what your gonna get in the long run. Some might fail prematurely. Some might just underperform.

Third, safety: 18650's are usually Lithium Ion (NMC/NCA etc...) which are more hazardous than say Lifepo4. A single cell failure can cause thermal runaway and cause a fire. Less so with Lifepo4.

Fourth, lifespan: Most Lithium Ion chemistries are good for a few hundred cycles, some can even reach low thousands, if used conservatively.
Having that you plan on using used cells, you really don't know how many cycles your gonna get out of them.
In contrast, Lifepo4 has a few thousand cycles (anywhere from 2000 to 6000, depending on DOD and specific cell model).

I personally would not go down the 18650 path. It might be tempting as you can get them cheap/for free, but really your time and safety isn't work it.
IMO, 18650's are great for small battery packs, but less appropriate for larger capacity packs like you are planning.
Having said that, there are some people here on the forum who have done that.
But in this day and age, with Lifepo4 prices coming down (at least Grade B), I would say it's not worth doing it with 18650s.

I'm in the process of building a 25 bay battery tester that will individually cycle each cell twice (once at a low setting, once at a "high" setting (Still nowhere near battery limits but ~1A)) and report the mAh, with a temp sensor for each bay for an auto-shutoff for the bad cells. The amount of work doesn't bother me, I was planning on spreading it out over a year or so (I understand it'll take me ~ 2-3 months just to test the batteries with a 25 bay tester). Is that timeframe realistic?

Thank you for the insight, yeah that cycle life doesn't sound good, especially rolling with used batteries... I have limited experience with 18650's so assumed my battery lifespan was always so short because of my use case scenario (High amps). Always seen their praises sung and fell into the youtube trap where people successfully built much larger systems with 18650's but... It just always didn't quite add up for me.

I'll definitely keep an eye on Lifepo4, especially since I have a few years to pick the best time to buy. I guess it would probably be a good idea to research different battery types more, but it just seemed like there were so many it was overwhelming. Any other battery types that would lend itself well towards my application and may fall into my price range or I should snatch instantly if I see a sale on within the next few years?

Thank you so much for the detailed response

how much are you saving compared to 32x 280ah lifepo cells?

I haven't done the calculations yet, since I won't be making the purchases for another year or so. I just saw 18650's were by far the cheapest right now, and also worst case scenario (in terms of amount of work, chance of failure for a first time DIYer, etc.)... So while prices might change in the next year, I can't count on a better battery becoming available for cheaper. But in the end, I need to set up a system that can generate 30kWh on a cloudy day, and a battery bank for 24 hours of storage, for ~ $10,000 USD... So I figured buying panels when the killer sales popped up and going as cheap as possible with the batteries in a set up that can be expanded on in the future would be the way to go.

 

[Modular Hippo]

Hi - I am currently putting together a 14S 40P 200A (560 x 18650) 46V and are battling with similar questions so will keep an eye on your thread. My objectives are:

1. Easy to diagnose replace dead cells without having to disable the whole cell pack.
2. Mainly Built in robust none flammable materials in case of issue.
3. Individually fused 18650 to make sure that if an issue on cell level (short) then this one should take itself out of play.
4. An extendable design so that I can connect multiple of these batteries in parallel.

I have just done 560 of the batteries and although taking time it is kind of therapeutic as well so all good Will keep you updated on my design in case of any use to you.

 

[JJJJ]

I know someone who builds batteries from Tesla type cells. When he first started building them he used a fine wire for connections which also doubled as a fuse. He uses a different method which is beyond this discussion.

Whether new or used one should consider cost versus safety. If one of these ignited you would have a whole new type of cascade on your hands. That is if using lithium ion cells.

The other consideration is the maintenance. My friend is a mechanic with many years of electrical experience. He is very methodical with his battery charging and maintenance. Lithium iron phosphate is much more forgiving and a safer technology.

I appreciate the willingness for hard work and admire the dedication. I was just wondering if there might be a more cost effective solution.

Just wanted to add my two cents. Good luck with your project!

 

[Modular Hippo]

I know someone who builds batteries from Tesla type cells. When he first started building them he used a fine wire for connections which also doubled as a fuse. He uses a different method which is beyond this discussion.

Whether new or used one should consider cost versus safety. If one of these ignited you would have a whole new type of cascade on your hands. That is if using lithium ion cells.

The other consideration is the maintenance. My friend is a mechanic with many years of electrical experience. He is very methodical with his battery charging and maintenance. Lithium iron phosphate is much more forgiving and a safer technology.

I appreciate the willingness for hard work and admire the dedication. I was just wondering if there might be a more cost effective solution.

Just wanted to add my two cents. Good luck with your project!

Hi JJJJ - You are 100% right in all your observation and appreciate the feedback. My main battery bank will most likely be a Seplos plug-and-play solution with Li-Ion Phosphate as you say (or similar) for that very reason. It is a good 24 months before I will launch my boat so a bit early to buy batteries at this point (as have a shelf life I do not want to eat in to). As far as the 18650 goes it is a challenge and a great way for me to get to understand do's and don'ts in an hands-on way. Sometimes for me it is the journey more than the destination and so far I enjoy tinkering with the battery pack and potential solutions around problems. I may never come to use it due to the inherited dangers but maybe I will, time will tell