MG4 51kWh LFP Battery Health

[T1 Terry]

Could this explain why the first owner of my car lost SOH around 6%?

when i bought the car
Soh:93.64%
10500 km

now(3 months later):
Soh:92.9%
21500 km

(SOH values were checked after a 100% balancing charge from a low percentage. I don't know how accurate it is.)

And is the 41.6 kWh value the value that will settle over time?
I would also like to state that I loaded 40.7 kWh energy from 3% to 83% last night. I also remember loading 45-46 kWh for balancing.

I think those energy charging figures are what was fed into the vehicle from the charging point, not what actually went into the battery. Is this correct?
Allowing a 10% inefficiency between supply and actual battery input, that comes out at around 36kWh into the battery. 80% of a 42Ah battery is 33.6kwh, fairly close, 2.4kWh difference ..... If the battery capacity was really 51kWh, then 80% is 40.8kWh, meaning the efficiency between the supply and the actual energy going into the battery was better than 100% .... I know which is more believable .......

As far as balancing, the system MG use is a loss method by burning off capacity as heat through resistors, and all the time that is happening, the inefficiency losses within the charging system remain, so the energy used during balancing is far greater than the energy actually added to the battery.

T1 Terry

 

[gdog]

I charged 19-100% last night 47.2kWh assuming charge losses of 10% that's 42.5kWh so 100% charge would be 52.4kWh capacity. So I don't buy into this conspiracy

 

[hojnikb]

I highly doubt that the actual raw capacity is 41.6kWh. That's 10kWh less than advertised. Chinese are sneaky, but not that sneaky. First of all, this would make this car way more efficient than it actually is. Secondly, this would show in charging, as considering charging losses you'd never be able to come close to 51kWh+ AC power delivered.
OK, so what is the most anyone has recorded when recharging a 51 pack?

I highly doubt that it is 'only' a ~42kWh battery pack. That's 20% less than advertised.
If that's the case than the MG4 would be MORE efficient than a Standard Range Tesla M3, which i CANNOT imagine
MG4 51kWh (or 42kWh as you say): WLTP range of 350 km, so 8,33km/kWh (350/42)
Tesla M3 Standard Range, 55kWh battery, WLTP range of 448, so 8,14 km/kWh.
Why would they have revised the battery capacity down to 49kWh? I suspect it will continue to be revised in small steps, you wouldn't want to caught out telling bold faced porky pies would you?
The vehicle weight hasn't reduced, so they didn't fit smaller capacity cells to reduce weight and make it more efficient, yet the claimed efficiency range also continues.

As far as the WLTP range .... MG MG4 Electric 51 kWh says it's only 300km, not 350, that comes out to 7.14km/kWh doesn't it?
Still, it's a bit like solar panel claimed output ..... who gets that in the real world? How many Tesla owners will admit the claimed range of the std model is work of fiction?

T1 Terry

There's two possibilites; They either downrated usable capacity to 49kWh, so there's more regen and degradation buffer, but the raw 51kWh capacity stayed. With some optimization tweaks that version retains the same sort of range, which is what you want.

Or second option; they used lower density, cheaper cells to reduce cost.

Either option is possible, but i lean towards the first, as it makes more sense.

I charged 19-100% last night 47.2kWh assuming charge losses of 10% that's 42.5kWh so 100% charge would be 52.4kWh capacity. So I don't buy into this conspiracy

Yeah, i don't buy it either. There's no factual evidence to support this conspiracy. Just (incorrect) guesstimates and incorrect assumptions.

But your estimation is pretty much bang on. Some % variation of charging efficiency and some losses due to balancing and onboard systems, that run when you charge and you come to ~50.8kWh give or take. if the pack was actually 42kWh, that would mean a ton of losses in your case or some serious measurment error.

I think those energy charging figures are what was fed into the vehicle from the charging point, not what actually went into the battery. Is this correct?
Allowing a 10% inefficiency between supply and actual battery input, that comes out at around 36kwh into the battery. 80% of a 42Ah battery is 33.6kwh, fairly close, 2.4kwh difference ..... If the battery capacity was really 51KWH, then 80% is 40.8kwh, meaning the efficiency between the supply and the actual energy going into the battery was better than 100% .... I know which is more believable .......

As far as balancing, the system MG use is a loss method by burning off capacity as heat through resistors, and all the time that is happening, the inefficiency loses within the charging system remain, so the energy used during balancing is far greater than the energy actually added to the battery.

T1 Terry

You're assuming % SoC is correct. LFPs are notoriously difficult to estimate actual state of charge. I'd wager that he was not actually at 3% and after charging the actual state was not 83%.

A much better test would be to drain the car completly to 0% and charge to 100% using DC charge and look for energy used there. There's much less losses, so calculations can be more accurate.

 

[siteguru]

I highly doubt that the actual raw capacity is 41.6kWh. That's 10kWh less than advertised. Chinese are sneaky, but not that sneaky. First of all, this would make this car way more efficient than it actually is. Secondly, this would show in charging, as considering charging losses you'd never be able to come close to 51kWh+ AC power delivered.
OK, so what is the most anyone has recorded when recharging a 51 pack?

T1 Terry

When I did my 4% to 100% and balance charge last month, the total kWh recorded in the Wallbox app was 54.08kWh. Assuming 50.8kWh battery capacity (per ev-database/manufacturer) then delivered into the pack was 0.96 x 50.8 = 48.768kWh, so 5.312kWh in "losses" or 90.18% charging efficiency. If the capacity was around 42kWh as you assert then that would be 77.66% charging efficiency, which I highly doubt.

 

[T1 Terry]

When I did my 4% to 100% and balance charge last month, the total kWh recorded in the Wallbox app was 54.08kWh. Assuming 50.8kWh battery capacity (per ev-database/manufacturer) then delivered into the pack was 0.96 x 50.8 = 48.768kWh, so 5.312kWh in "losses" or 90.18% charging efficiency. If the capacity was around 42kWh as you assert then that would be 77.66% charging efficiency, which I highly doubt.

What losses are you attributing to the balancing part of the whole charging regime, the 10% in charging inefficiency would be about right, but a total loss type balancing method used by MG and the time the charging system still running yet not really charging much, so the inefficiency there would be a much higher % between what was used and what actually went into the battery .... but you aren't allowing any losses for that part of the operation .........

T1 Terry

 

[siteguru]

I've not made any claim as to where losses are attributed. Merely quoted exactly what kWh was reported by my app, then made a calculation based on manufacturer data. However in previous versions of the iSmart app I was able to see the charge rate reported to 2 decimal places, and this typically showed 0.08kW (or 80W) which would continue until balancing completed - typically about half an hour or so, which would equate to about 40Wh of "balancing loss".

So I still don't believe that the pack is as low a capacity as you're stating.

 

[Everest]

below is from car scanner - Basic 51kW SE

So, out of fascination, I extended your plot to see what SOH might be like at 80,000 miles, assuming it follows the vaguely linear progression you have plotted so far. Looks good to me = still about 85% at 80K miles

 

[gdog]

but you aren't allowing any losses for that part of the operation

I had very little kWh associated with balancing

 

[wattmatters]

Balancing uses bugger all energy relative to a full pack recharge. Couple of handfuls of watt-hours.

 

[fnegroni]

With LFP I noticed it’s more effective to regularly take it to lower SoC before charging then balancing. The BMS seems to prefer that.

 

[T1 Terry]

So, out of fascination, I extended your plot to see what SOH might be like at 80,000 miles, assuming it follows the vaguely linear progression you have plotted so far. Looks good to me = still about 85% at 80K miles

View attachment 33933

Following the plot, how many yrs would that be? It will be interesting to see if the end of the graph settles out to a shallower curve.

T1 Terry

Balancing uses bugger all energy relative to a full pack recharge. Couple of handfuls of watt-hours.

Is that from the supply or into the battery?
Looking at it from a reverse angle, 12V to 240VAC inverters are roughly 90% efficient at close to full load.
As an example, with a 3000W inverter, at 95% load roughly 300W lost to operating "expenses" .... but that roughly 10% is not linear, it remains at the same 300W when the fridge door is opened and the little man in the fridge turns the 10W light bulb on, the draw from the battery is 310W.
Very high end inverters have a low current draw available before the main stage is woken up, sort of a 2 step set up, and generally referred to as a power saving mode, that way the clock in the microwave doesn't draw 300W continuous just to keep the display alive.

T1 Terry

 

[EVentropy]

Nominal votlage is not "arbitrary" figure. It's actually an avarage voltage you get, if you divide Wh(energy) by Ah(capacity). For example, a 280Ah LFP cell has 896Wh of energy, so 896/280 comes out to 3.2V.

896Wh for 280Ah is a contrivance just like 3.2V is the same for “average voltage”.
If you extracted 896Wh out of a 280Ah battery via wattmeter then recharged it to same "soc", then discharged it at same rate with CC, and the those figures panned out then who am I to argue with 3.2V being "average V". Another contrivance "rule of thumb"= SOC.
I prefer tail current stops dropping. (For safety, as opposed to zero if that's even possible)

 

[wattmatters]

Is that from the supply or into the battery?

From the supply - the energy has to come from somewhere (first law of thermodynamics) and there is only one supply source - the charge port. If balancing was using a lot of energy it would show up in the charging energy consumption.

The car tells you when it is balancing - it shows it on screen if you happen to capture it, and you can see how much power is being consumed at the time - which is bugger all (and likely is also partly being used for other functions as well).

In most cases balancing is done in 5-10 minutes. Only time it takes longer is if the car has gone a long time without AC charging.

Looking at it from a reverse angle, 12V to 240VAC inverters are roughly 90% efficient at close to full load.
As an example, with a 3000W inverter, at 95% load roughly 300W lost to operating "expenses" .... but that roughly 10% is not linear, it remains at the same 300W when the fridge door is opened and the little man in the fridge turns the 10W light bulb on, the draw from the battery is 310W.
Very high end inverters have a low current draw available before the main stage is woken up, sort of a 2 step set up, and generally referred to as a power saving mode, that way the clock in the microwave doesn't draw 300W continuous just to keep the display alive.

None of which has anything to do with power/energy consumption required for the car to perform cell balancing (which is bugger all).

 

[T1 Terry]

The theory for how Nom. voltage is determined is about as varied as how the universe came into being .....

The one that makes the most sense to me is nom. voltage is the voltage a charged cell drops to under its accepted test conditions load, this is from the Winston website and the graph shows the 0.5C load pulls the cell down to approx 3.2V and it can hold that voltage down to 30% SOC.

https://en.winston-battery.com/static/upload/file/20220507/1651883577893499.pdf

The other one that makes sense is the average voltage under the prescribed test load from 100% SOC to 0% SOC.

Please keep in mind, 0% SOC is not zero volts, it is when the advertised capacity has been drawn from the cell, or the cell voltage drops to the min voltage specified while under the test load. This method is the best for determining actual cell capacity at factory specs test load and at what ever load the intended use might be, as can be seen on the chart, a new cell will still deliver 100% of the advertised capacity at the 3C rate, at the factory test rate, discharged from 100% SOC to 2.8V under load in 2 hrs under constant current, the same new cell will deliver approx 115% of the advertised capacity .......

You won't find many cell manufacturers that can deliver that sort of performance .....

T1 Terry

 

[EVentropy]

The theory for how Nom. voltage is determined is about as varied as how the universe came into being .....

The one that makes the most sense to me is nom. voltage is the voltage a charged cell drops to under its accepted test conditions load, this is from the Winston website and the graph shows the 0.5C load pulls the cell down to approx 3.2v and it can hold that voltage down to 30% SOC.

https://en.winston-battery.com/static/upload/file/20220507/1651883577893499.pdf

The other one that makes sense is the average voltage under the prescribed test load from 100% SOC to 0% SOC.
.

T1 Terry

Still don't see that you can use that "nominal V" at the top end and get a meaningful result tho.

 

[T1 Terry]

None of which has anything to do with power/energy consumption required for the car to perform cell balancing (which is bugger all).

Judging by that, it seems the charger doesn't actually continue charging the battery during the balancing stage and the 80 W is used to power the monitoring side of the BMS.

The charger efficiency can only be measured while the charger is actually doing just that, then it's watts in minus watts received by the battery, minus the losses due to battery resistance converting charge current into heat ......

I'm guessing this debate will continue with no resolution until someone actually removes a battery module and tests the capacity ..... or enough time has passed to see if the SoH degradation slows at around 85% SoH, that is close to the battery capacity I'm putting forward is the true capacity.
My guess is it will be the latter and not the former that determines if I'm right or wrong ......

T1 Terry

 

[baader]

Hi, when you turn on the intelligent battery heater does the range gauge go down like when you turn on or off the air conditioning for example or does it not respond?

I when I turn on the battery heater it doesn't change the estimated range of the vehicle. But if I turn on, say, the air conditioning, it immediately changes the estimated range of the vehicle.

It seems like the battery heater doesn't even work when the range estimate doesn't respond. Yet this function should have a great enetrgy consumption...

 

[fnegroni]

Heating the battery costs some energy but depending on scenario not reduce the range

 

[Coulomb]

it seems the charger doesn't actually continue charging the battery during the balancing stage and the 80 W is used to power the monitoring side of the BMS.

I think the 80 W is needed to keep the other cell voltages from falling down to a plateau. Once the highest cell has come up to the required voltage, the BMS might search for the second highest cell... oh wait, they're all at 3.330 VPC (the main flat part of the voltage versus SoC graph). In fact, when one cell has the bleed resistor turned on, some other cell is likely to become the highest voltage cell, so its bleed resistor can be turned on as well, but that doesn't work if the charge current is so low that the other cells are a the plateau.

What gets me is why the balancing at 80% seems to take kilowatts, not tens of watts. Mine "balance" for about 30 minutes at 80% SoC, but I wonder if that's taking all cells from say 79.5% (rounded to 80% on the display) to 80.0%, and the business with the predicted end of charge equalling the current time is just poor maths, with little to no real balancing happening.

But 2 kW for half an hour is around one kWh, which is about 1/74.4 or 0.013 or 1.3%, which is a bit much for the above theory. Though with losses and overheads it might be a bit under 1%, and perhaps it goes from 79.5% to a bit over 80.0%.

Edit: Or perhaps all SoC percentages are truncated (so 79.9% displays as 79%), and when you set the limit at 80%, it actually charges to 80.99%. That makes no sense when you stop at 100% though. So maybe all display SoC% are rounded up, so 79.01% displays as 80%. That just seems weird though.

Edit 2, January 2025: I changed my limit to 90% today, since I had excess solar and am going to drive it soon. When it got to 82%, on a whim I set the charge limit in iSmart to 80% to see what it would do. Indeed, it seemed to then do an actual balance at 82% SoC, using some 150-200 W from the wall. iSmart showed 0.2 kW. It only lasted a few minutes, then stopped charging altogether.

 

[Everest]

Following the plot, how many yrs would that be?

My schoolgirl maths suggests that would depend on how many miles he is doing a year