Now that my giant DIY powerwall has been built using parallel battery packs, the question came up about how current is shared between the different packs. Most packs are the same, but the second style of packs has a different capacity, different 18650 cell model, and different number of cells in parallel. What does this mean for current sharing between these packs as they discharge? Let’s find out!
Going into this, I suspected that the current flow from each pack would be proportional to its capacity, because if a pack is rated at X number of amp-hours, that current all needs to be delivered over the course of discharge. While in the end this was correct on average, what I did not expect was the amount of variation within the discharge curve.
To find out how the current is shared between the packs, I took one of each type (13s5p/15.4Ah and 13s4p/12.5Ah), charged them up fully, and put them in parallel. Using a DC load to discharge tells me very accurately the total current coming out of the packs, and a current meter in line with one of them lets me know the fraction coming from each pack. (The two packs must combine to total the current being drawn by the DC load).
Test setup
First, confirming that the current measurements are accurate—they are.
 |
| Showing 4A flowing from the power supply (top) to the DC load (middle). Fluke meter (bottom) is in-line and agrees. Clamp meters (right) are always a little off; this one is decent. |
I made a time-lapse video to record the data for me so I didn’t have to sit there the whole time.
 |
| Work smart, not hard! Two packs in parallel (black and blue on left) connected to the DC load. Time lapse video lets me review all the data at the end. |
Results of the load-sharing test
 |
| Results plot of load-sharing test |
Very interesting! On average, the current from each pack was indeed proportional to their capacities, but I did not expect so much variation! Especially at the very beginning of the discharge curve when somehow the smaller pack gave more current.
I suspect this is because each pack started with a slightly different state of charge and also had different types of cells. The smaller pack has Samsung 18650INR-35E cells (3500mAh rated), while the larger pack has Panasonic NCR18650GA (3300mAh rated) cells. Different types must have slightly different discharge curves.
As a bonus, I automatically created a discharge curve for these packs at a relevant current draw for my full system. Since my system has 20 packs, and the most total current to be drawn is around 40A, about 2A per pack is a perfect place to collect this data.
 |
| Discharge curve for the parallel lithium ion packs |
As expected, the energy at the end of the discharge curve drops off quickly. The last few volts contain very few amp-hours, and the effect is even more dramatic when you consider the actual energy in Watt-hours because the voltage is lower. Going from 52V to 51V (near full) the packs contained 162Wh, but from 40V to 39V (completely discharged) there is only 21Wh.
Conclusion
For a low discharge rate (<0.2C) application like mine, putting all sorts of different packs in parallel seems to be no problem. There is some variation in the amount of current delivered from each, but on average it works out to be proportional to pack capacity. Because the current draw is relatively low, the variation due to pack differences will not create any issues for the packs like would be possible if running at a high discharge rate near the pack's design limit. In that case, temporarily having current 20% higher in one pack would not be acceptable.
Have you run Lithium ion packs in parallel? How did it go? Or what else might I be missing?
Mike
No comments :
Post a Comment
If you're having trouble commenting with your google account, try enabling 3rd party cookies. In Safari, that means un-checking "prevent cross-site tracking" under privacy settings. Super annoying I know!