Tesla NCA battery charge discharge efficiency data Victron smart shunt

[AntronX]

I have 6x 2014 Tesla Model S battery modules in 2S3P config for 43.2V 630Ah configuration at roughly 85% state of health. They are charged by Victron MPPT 150/60 controller maxing out at 60A. I had Victron smart shunt logging battery data for 18 days and the result shows 98.9 % charge/discharge energy efficiency. Typical charge rate is 0.05 C and discharge rate 0.025 C or 20 hour charge and 40 hour discharge on average.



Here is 18 days of voltage logging. Note that energy count was captured at exactly same 43.31 Volts and about the same load current of 5A or 0.008 C rate for the pack.



Charge level estimate drifted down by 2% overtime. Still tweaking capacity and efficiency setting to minimize SOC drift.

.

 

[sunshine_eggo]

Smartshunt settings?

 

[AntronX]

Smartshunt settings?

 

[sunshine_eggo]

It's no coincidence that The shunt reported 98.9% efficiency when you have it programmed for 99%.

You'll spend more time mucking about with numbers trying to get SoC to come out right. In the end, you'll eventually realize that you simply need to manually set the SoC every couple weeks when the battery is at peak voltage since you're not charging to 100%, and NCA has a strong SoC to voltage relationship.

The Smartshunt is not designed for NCA/NMC/LMO, etc., cells. It needs the lead-acid/LFP bulk/absorption charge profile to be accurate.

Alternatively, you could charge to true 100% and program the shunt to sync on that basis, but I doubt you want to do that.

 

[AntronX]

It's no coincidence that The shunt reported 98.9% efficiency when you have it programmed for 99%.

I set it for 99% after this measurement. I had it at 98% before. I would expect energy charged and energy discharged to be directly taken from shunt coulomb counting multiplied by voltage.

The Smartshunt is not designed for NCA/NMC/LMO, etc., cells. It needs the lead-acid/LFP bulk/absorption charge profile to be accurate.

Um, are you sure? I would expect it to take stated battery amp hour capacity, energy in/out measurement and count SOC based on that (taking into account peukert and stuff). At least that's how I would design it.

 

[sunshine_eggo]

I set it for 99% after this measurement. I had it at 98% before. I would expect energy charged and energy discharged to be directly taken from shunt coulomb counting multiplied by voltage.

Have you set it for 100% and confirmed your assumption? You might be right, but I'm fairly certain 98% means Ah charged is multiplied by .98 and then used to generate an efficiency number.

Um, are you sure?

Absolutely. Based on Victron's own statements, in order to maintain accuracy, the shunt must be synchronized a couple times a month.

I would expect it to take stated battery amp hour capacity, energy in/out measurement and count SOC based on that (taking into account peukert and stuff).

Coulomb counting isn't accurate enough. There will always be drift.

 

[AntronX]

Have you set it for 100% and confirmed your assumption? You might be right, but I'm fairly certain 98% means Ah charged is multiplied by .98 and then used to generate an efficiency number.

I would hate to reset it to test this. It does not make sense to do it this way if the shunt can already count in both directions and that should not drift assuming Smart shunt uses accurate midpoint voltage reference. So far I found it to be accurate with few watts during discharge. Charge current also agrees with my multimeter and MPPT 150/60 except voltage being off by 0.1V but not due to wire voltage drop (measured at no current). I did not use lab grade 10 decade multimeter to verify smart shunt accuracy.

Absolutely. Based on Victron's own statements, in order to maintain accuracy, the shunt must be synchronized a couple times a month.

Maybe that could be due to battery capacity fading overtime or incorrectly set or peukert/temperature factors?

Coulomb counting isn't accurate enough. There will always be drift.

Unless internal voltage reference drifts overtime the shunt should be accurate in both directions unless there is something else going on. Couloumb counting is simply counting Amp-hours and multiplying by Voltage and adding resultant Watt-hour cumulatively to the energy charged/discharged memory register.

 

[sunshine_eggo]

Every battery monitor I've ever seen indicates they must be periodically synched to maintain accuracy. Additionally, BMS reported SoC will also drift if never taken to 100%.

You'll gloss over a 0.1V measurement error, but you think coulomb counting will be 100% accurate?

Title is very telling:

https://ntrs.nasa.gov/api/citations/20220008848/downloads/2021.07%20-%20Energies%20-%20A%20Critical%20Look%20at%20Coulomb%20Counting%20Approach%20for%20State%20of%20Charge%20Estimation.pdf

 

[AntronX]

You'll gloss over a 0.1V measurement error, but you think coulomb counting will be 100% accurate?

That error is external to the energy in/out calculation that happens inside the shunt. As long as the shunt has the same voltage error in both directions that error will not affect relative energy loss aka battery efficiency result I am looking to measure.

From the manual:

5.1. How does the battery monitor work?

The main function of the battery monitor is to follow and indicate the state of charge of a battery, to be able to know how much charge the battery contains and to prevent an unexpected total discharge.
The battery monitor continuously measures the current flow in and out of the battery. Integration of this current over time, if it was a fixed current, boils down to multiplying current and time and gives the net amount of Ah added or removed.
For example, a discharge current of 10A for 2 hours will take 10 x 2 = 20Ah from the battery. To complicate matters, the effective capacity of a battery depends on the rate of discharge, the Peukert efficiency, and, to a lesser extent, the temperature. And to make things even more complicated: when charging a battery more energy (Ah) has to be ‘pumped’ into the battery than can be retrieved during the next discharge. In other words: the charge efficiency is less than 100%. The battery monitor takes all these factors into consideration when calculating the state of charge.

So, does this mean that battery efficiency setting is only used to better estimate the SOC ? I would hope so because using that setting to modify energy going in/out measurement does not make sense. That measurement should be absolute, not subject to modification.

State of charge. This is the actual state of charge of the battery in a percentage and is compensated for both the Peukert efficiency and charge efficiency. The state of charge is the best way to monitor the battery.

Sounds good.

Current. This is the actual current flowing in or out of the battery. A negative current indicates that current is taken from the battery. This is the current needed for DC loads. A positive current means that current is going into the battery. This is current coming from charge sources. Keep in mind that the battery monitor will always indicate the total battery current, being the current traveling into the battery minus the current traveling out of the battery.

Good.
Here it gets interesting:

Consumed Ah. The battery monitor keeps track of the Amp-hours removed from the battery compensated for the efficiency. Example: If a current of 12A is drawn from a fully charged battery for a period of 3 hours, the readout will show -36.0Ah (-12 x 3 = -36). Please note that if the "Consumed Ah" reading indicates three dashes: “---” this means that the battery monitor is in an
unsynchronised state. This mainly occurs when the battery monitor has just been installed or after the battery monitor has been left unpowered and is powered up again.

Ok, so here is where that efficiency setting starts to affect measurement. It appears to offset the Consumed Ah measurement. It does not state that discharged energy reading is also affected. For my purpose I do not care about consumed Ah reading.

Energy in kWh
• Discharged energy: This is the total amount of energy drawn from the battery in kWh.
• Charged energy: The total amount of energy absorbed by the battery in kWh.

As it should be. An absolute measurement. No mention of efficiency setting modifying things.

7.2.7. Charge efficiency factor. The “Charge Efficiency Factor” compensates for the capacity (Ah) losses during charging. A setting of 100% means that there are no losses. A charge efficiency of 95% means that 10Ah must be transferred to the battery to get 9.5Ah actually stored in the battery. The charge efficiency of a battery depends on battery type, age and usage. The battery monitor takes this phenomenon into account with the charge efficiency factor.

Does not clarify here what this corrected Ah is used for. But earlier paragraphs pointed to only Consumer Ah and SOC being affected.

10.3.4. Incorrect state of charge reading. An incorrect state of charge can be caused by a variety of reasons. Incorrect battery settings: The following parameter(s) will have an effect on the state of charge calculations if they have been set up incorrectly: • Battery capacity. • Peukert exponent. • Charge efficiency factor.

Again this ties charge efficiency with state of charge.
I conclude that the manual implies that charge efficiency factor is only used to calibrate SOC estimation and "consumed Ah" data points.

 

[AntronX]

In the end, you'll eventually realize that you simply need to manually set the SoC every couple weeks when the battery is at peak voltage since you're not charging to 100%, and NCA has a strong SoC to voltage relationship.

That's pretty much what I do now once in a few days. I do have a list of voltage/SOC points I captured by discharging a sample cell harvested from this Tesla battery pack (some modules were disassembled to harvest the cells). I captured total mAh value from 4.2 - 3v discharge then divided it by 10, charged and discharged again but stopping and letting the cell come to resting voltage 10 times. At that time the cell was 3050 mAh total so I took resting voltage measurements after each 305 mAh of discharge to get accurate voltage/SOC data at each 10% step.

 

[SeaGal]

Alternatively, you could charge to true 100% and program the shunt to sync on that basis, but I doubt you want to do that.

Sorry if going bit off topic, but do you mean it is possible to programmatically tell the SmartShunt to reset to 100% via its comms port? I'd find that very useful, TIA.

 

[sunshine_eggo]

Sorry if going bit off topic, but do you mean it is possible to programmatically tell the SmartShunt to reset to 100% via its comms port? I'd find that very useful, TIA.

No. True 100% on NCA is 4.2V/cell.