Making my 18650 30kW Battery array

Question

I put this in the comments of my last question but I realised that might not get many eyeballs so I will ask it here.

I have have been thinking of making a battery using 18650's. the battery arrangement would be 12 parallel 18650's making the 'block' and 13 'blocks' in series, totalling 156 batteries. Each Battery has it's own PCM and the 12 parallel 'blocks'are each attached to a 48V 30A BMS device. I will have the batteries seperated (not touching) and a thermal sensor fan that activates when tempratures go beyond a certian point, the sensor will be near the middle. This makes one battery.

I then take 24 of these batteries and parallel them together to make the final array.

I attach the final large battery array to a CHarge Controller that can charge them from my solar panels thus a PV charger and it is a 48V 45A controller MorningStar TS-MPPT-45

The plan is to make a SAFE battery that can store around 30Kw of energy so the house can run purely on batteries for a few days.

I need advice on the design of this operation and when I wire the 24 batteries to a single + terminal and - terminal is it the correct way to do this? Have I taken enough procautions? Should I abandon the idea? In total there will be 3744 18650 Batteries.

Answer 1

First things first. In your previous post, you stated that your source would not provide any certification about actual battery capacity. Unless you can get some sort of guarantee, drop the whole idea. At claimed capacity and price, this http://www.ebay.com/gds/18650-Battery-Buying-Guide-test-on-all-from-eBay-below-3-/10000000178020340/g.html is likely what you will get. If you insist on dealing with these bozos, arrange for your total order to be shipped in smaller (say, 100 cells ea) with sufficient time between lots to allow for testing. Structure the order so that if a certain percentage (a few per cent or less) fails test, the entire order is cancelled. If they won't do it, don't buy from them. And good luck getting your money back if they ship you crap.

Second. If you are going to build up blocks, you're going about it backwards. Start with strings of 13 in series, with a PCM. Due to the variations between cells, you MUST not connect many cells in parallel. 2 is generally OK, and some manufacturers are OK with 4, but connecting 12 is a fabulous way to destroy your cells. The problem is that the weakest cell determines the cutoff voltage for the entire block, and it also hogs the entire charge current at the start of charge, resulting in high charge currents and battery temperature problems.

Third. You need a BMS for each string.

Fourth. For your application, you need a different package than 18650s. If you're willing to take chances, you can buy or build a spot welder and put tabs on the cells, then solder your strings. If you're thinking of using standard cell holders, forget it. Over long periods of disuse, corrosion at the contact points will be a killer.

Fifth. You need to rethink your cooling setup. The cell assemblies must be physically separated to allow airflow, and you'll need a powerful fan. If you don't know how much fan you'll need, you're in trouble. And you need a mechanism to allow the temperature sensor to disable charging as appropriate if the fan is not adequate.'

Sixth. Your calculations are off somewhere. Even allowing for the fact that you don't seem to know the difference between power and energy, the calculations for your battery array are straightforward. 3.7 volts x 4.9 Ah x 3744 equals about 86 kWh. If you want this to last "several days" (let's say two), then the allowable power is about 2800 kW, not 30kW. 30 kW / 3744 is 8 watt per cell, or about 2 amps. While this is not, by itself, an unreasonable number, it makes no sense given your stated battery capacity. You'd expect a lithium ion cell to handle at least a 1C rate, or 5 amps. If 2 amps is the number given by your prospective supplier, it indicates that he's not being honest (although the price/capacity/documentation issues pretty well establish that).

7th. Oh, feh. Enough of this. Try another approach. Do a lot more research, and start experimenting. Start small, and get experience with progressively larger arrays. Trying to jump in at the scale you're proposing is just not a good idea.

First the bucket, then the pail,

Then the laboratory scale.

Ever bigger, ever faster;

Faster, faster, then - disaster.

Answer 2

First, I agree with the above answer, the BMS goes with the series string, not a BMS for every parallel 'block' you make. The parallel block you make essentially just becomes one big or large capcity battery, while the series string becomes a larger voltage battery with the BMS controlling several things including the balance of the string which is very important in something as large as 13 in series.

However, I disagree with the above on the rule that you can only do 2 or maximum 4 lithium in parallel. If you follow appropriate safeguards, you can use more. The biggest thing would be to have all cells in parallel to be from the same manufacturer and the same lot if possible.

Also, I believe that you mentioned that you found some high quality samsung 18650 batteries that are around 2600 mAh. This sounds much more reasonable than the 4900 mAh you were originally looking at.

There are thousands of examples all over the Internet of people putting many lithium 18650 batteries in parallel with NO problems when doing it correctly. The Internet is not always correct, but experience doesn't lie. The ebike people have been doing it for some time.

The best example is Tesla. They use over 7000 batteries in their model S. 74 - 18650 in parallel. They use cooling, fuses, and careful selection of batteries, but it works. Tesla has not has problems with the weakest cell hogging the charge current. In fact, if one cell in the bank of 74 dies for any reason, they just leave it in there and let the remaining 73 continue to do the job of the 74!

You could buy a battery out of a salvaged Tesla - 85 kWh. One comes available every once in a while! Save you a lot of work.

Having said all of that, the battery you plan is a crazy amount of stored energy and if something went wrong it could mean burning down your house. You would be very wise to build some sort of special room for the battery that is fireproof just to be safe!

Answer 3

I agree with the Filek's answer and disagree with some commenter's remarks that it was unsafe, having a 3.7V short is less dangerous than a 12V or a 24V short in the other series cells setup where of course you're forced to use extensive monitoring of every cell. you can always do like the other suggestion to use extensive BMS monitoring, but that will make the battery uncompetitive, the point of using 18650 form is to have a modular, easy to fix battery, you can replace any damaged cell or rebuild it after re-balancing cells and reselecting them. for single cells battery, i advice people to check for weight information it's critical to check if power density is realistic and falls within known lithium-ion chemical formulas for their respective power-densities or not, Any battery maker can make wild claims about their batteries capacity, but they can't fake weight.

Answer 4

As far as I understand, connecting many cells in parallel can lead to chain reaction if one fails with short-circuit. Tesla also connected hundreds of 18650s in parallel, but they used tiny fuses on each cell. So, theoretically, if one cell fails, it simply vents and disconnects from remaining ones. So it is, theoretically, even possible to continue operation.... Not sure how this applies to your question, but have you considered using such "building blocks"?

https://www.energusps.com/shop/product/li-ion-building-block-3-6v-78ah-12

Apparently student formula teams are using these modules with good results. AFAIK these Samsung 18650 are all simply connected in parallel, but made very easy to use and maintain. Also they mention fuses and vent channels on each cell, so probably something like in Tesla's battery packs?

Bit pricey though. But at larger quantities, maybe...