Mixing battery designs in solar / powerwall system

Eddieb

New member
Joined
Aug 12, 2020
Messages
5
Hello,

I currently have ~14 kwh of 18650s in 10 packs of 7s20p configurations. Each pack has a BMS. I am charging these to a max of 28 volts and discharging to 23.5 volts. These are being charged by 4.4 kw of solar panels.

I am trying to determine if I could add LiPO4 cells to the storage setup in an 8s or 9s configuration. The cells have a 3.65 max voltage and 3.2 nominal voltage, cutoff voltage of 2.5v. In 8s max charge would be 28.8 volts and cut off would be 20v. It seems like these could operate in the voltages of my current system.

Is this a valid approach? Am I missing something?
 
Hello,

I currently have ~14 kwh of 18650s in 10 packs of 7s20p configurations. Each pack has a BMS. I am charging these to a max of 28 volts and discharging to 23.5 volts. These are being charged by 4.4 kw of solar panels.

I am trying to determine if I could add LiPO4 cells to the storage setup in an 8s or 9s configuration. The cells have a 3.65 max voltage and 3.2 nominal voltage, cutoff voltage of 2.5v. In 8s max charge would be 28.8 volts and cut off would be 20v. It seems like these could operate in the voltages of my current system.

Is this a valid approach? Am I missing something?
8s (LifePo4) is the common number in series for nominal 24v systems. 7s for lithium-ion.

Currently you're doing:
On the lithium-ion side - that's 3.36v low (23.5v/7s) and 4.0v hi (28v/7s)

And proposing:
On the LifePo4 side - that would be 23.5v/8s = 2.94v low and 28/8s = 3.5v hi.

I'm not a LifePo4 person (I'm 18650 / lithium-ion) but from what I read/understand for LifePo4 the 3.5v hi is probably not so good. The problem is that LifePo4 has a *very flat* charge/discharge curve.... so going with 3.5v high may not work well as you don't have a strong grasp on state of charge by going voltage only. Even going to 4.10v on lithium-ion side would get you 3.59 on LifePo4 side and still might be an issue.

I think you can do this - but its difficult to do it *well* - e.g its not that it can't be done, or that's it risky etc.... its just hard to control/optimize the 2 different chemistries by voltage alone and you don't get good % use of the batteries - and that's why folks don't go this route.
 
I think it should work assuming you set the correct parameters in all BMSs (both Li-Ion and LiPO) because BMS hides what's behind it (the batteries) and charging will just push amperes into all BMSs and each BMS will "pull in" what batteries require. When current comes out of the batteries it should be the same, load would be (unevenly) divided between batteries. Unevenly because even if you had Li-Ion and LiPO4 batteries of the same capacity they do have different output models.

How @OffGridInTheCity says you may not have too much control on the current flowing out of the two batteries, but I can't say how much this could be a real problem. I mean, each battery type will contribute to give current and one type could perform better than the other. In my mind it should be ok but you know how it goes when you test stuff... sometimes it just doesn't go how you expected it to go!
 
How @OffGridInTheCity says you may not have too much control on the current flowing out of the two batteries, but I can't say how much this could be a real problem. I mean, each battery type will contribute to give current and one type could perform better than the other.
To clarify - I'm not talking about current. I'm talking about SoC. LifePo4 has such a shallow voltage charge/discharge curve that at 3.55v (as an example) - you don't know the SoC with nearly the same accuracy you do for Lithium-ion. And when you combine a single/shared voltage range for both Lithium-Ion and LifePo4 in parallel its difficult to ensure some metric like '80% average DOD' applies to both chemistries at the same time. There's likely to be an imbalance and therefore you're not using both batteries at the % of capability that you might want.
 
Last edited:
To clarify - I'm not talking about current. I'm talking about SoC. LifePo4 has such a shallow voltage charge/discharge curve that at 3.55v (as an example) - you don't know the SoC with nearly the same accuracy you do for Lithium-ion. And when you combine a single/shared voltage range for both Lithium-Ion and LifePo4 in parallel its difficult to ensure some metric like '80% average DOD' applies to both chemistries at the same time. There's likely to be an imbalance and therefore you're not using both batteries at the % of capability that you might want.

SOC doesn't really matter as he is planing to use the full range of capacity anyways. Both chemistries will go from each top of curve to bottom of curve, just that the curves are different.
It is exactly as @italianuser said, sometimes one battery will deliver more current, then the other depending where in which curve you are, and which chemistry is dominating that part.
A full cycle for one, will be a full cycle for the other as well if you charge and discharge steady in one.

So technically I do not see a problem in mixing chemistries, but you need to be aware, that at some point one of the two will carry almost all load!
7s LiIon plus 8s LiFePo4, that's the only way you can mix it safely.
 
SOC doesn't really matter as he is planing to use the full range of capacity anyways. Both chemistries will go from each top of curve to bottom of curve, just that the curves are different.
However, notice that he's not running the lithium-ion pack at 'full range' - he's topping out at 4.0v... perhaps to extend the life of the lithium-ion cells but going kind of low for lithium-ion. This 'specific' range on lithium-ion (presumably for a reason) results in an unusual 3.5v hi (instead of 3.65v hi) and 2.94v low on the LifePo4. Not sure the affect doing this but for sure, a portion of the LifePo4 SoC is not utilized and the 3.5v in the flat part of the discharge curve makes things 'fuzzy' going on voltage alone.

My point is that optimizing one chemistry (4.0v hi for lithium-ion in this case) will have an affect on the other because they have very different charge/discharge curves. This complexity is not wrong but is why I think most people avoid mixing the 2 chemistries as a single battery bank. Having 2 separate batteries is a whole different discussion.
 
Last edited:
However, notice that he's not running the lithium-ion pack at 'full range' - he's topping out at 4.0v... perhaps to extend the life of the lithium-ion cells. Under this condition, that's 3.5v for LifePo4 - which is right in the flat part of the discharge curve - where the SoC is fuzzy and he's taking the LifePo4 down very low.

No, the flat part of LFP is between 3,2 and 3,4V. There is almost no extra energy from 3,4-3,65V in LFP. 3,5V is actually the best cutoff voltage for LFP cells. The range as he is stating is actually very sensible for both chemistries.

A charge and discharge on that combination would look like this at pack voltage:
27,2-28V LiIon will dominate and take most current
25,6-27,2V LFP will dominate and take most current
23,5-25,6V LiIon will dominate and take most current

Total SOC is a function of the 2 individual SOCs for the actual voltage

I think maybe @Wolf can help us out here? :)
 
The way I look at this is as @OffGridInTheCity explained. The Discharge curve of a LiFePo4 is so much flatter than the Li-ion.
Also the Li-ion battery is just about dead when the LiFePo4 just gets started. Just not a good idea.
Also the 2 chemistries in parallel will fight each other for voltage dominance causing stress on one or the other at different times.
As voltages change the batteries will try to fulfill their discharge requirements at different levels. I think the widow of equality is so narrow that you would need to stick to such a tiny voltage window there would be no practical benefit.
Wolf

1620356390030.png
 
Last edited:
Couldn't you have two sets of charge controllers, one inverter, a programmable contactor( like an ATS except for DC) That based on (time of day/voltage) switches to one dc source or the other(to the inverter)depending on a set of rules? More complex I know. The inverter wouldn't be configured to any particular chemistry. only low voltage cutoff, resume voltage, max voltage.

Later floyd
 
Back
Top