Help me figure out my battery pack numbers, great challenge

[emir-t]

Hey everyone. I drive a convert EV made by a Turkish company on a Toyota Corolla chassis the past 10-11 months and I've done around 19k miles on it, I'm loving the EV experience and am never going back.

I also love to understand its inner workings and comprehend it fully. I need help understanding the numbers of the battery to keep track of degredation, currently, I can't.

So the battery data I have is as follows;

• Rated Capacity: 36kWh
• Nominal Voltage: 358V
• Cell type: 3.2V, 100Ah LiFePo4 cells - 112 of them in series.

And the info screen of the car gives me three values;

• SoC in %
• Average consumption in km per kWh. (It is definitely a moving average of the past X miles but I don't know X)
• KMs of range left to drive. (directly relies on average consumption)

I've researched a lot about batery SoC indication and fuel gauging. I know of two techniques; coloumb counting and voltage estimation. Since this is an EV and it has irregular loads I'm assuming this is by coloumb counting since voltage would drop significantly during a fast accelaration.

Also the SoC percentage is pretty linear and realistic, it doesn't go crazy like my iPhone does (shutting down at 20%, connecting and instantly jumping +15% etc.)

How do I calculate kWh used in a charge and keep track of real degradation? I can't find a way without finding the aforomentioned moving average X km figure.

Also this pack is not thermally managed and consumption increases a lot in winter. However car shows the average consumption to fall a lot during winter months. For example now I'm getting 4mi/kWh (250Wh/mi) and in winter it gets as low as 2.8mi/kwh(357Wh/mi) Yet this implies that cold affects the consumption of the drivetrain. However we know that battery chemistry is the one that is affected hence able to deliver less capacity. Yet SoC still indicates 100% when full. This could be a clue.

Any discussion, hint is appreciated. Thanks in advance.

 

[DougIngraham]

I suspect that you don't have enough information about the instrumentation to be able to tell how much battery degradation you are experiencing.

The SOC% would need to automatically re-calibrate itself as the battery degrades or it would not be able to tell you an actual %. The assumption of 100% SOC can be made at the end of the charge+balance. The cheap way to do this would be to assume your 100 AH batteries are really 100 AH and just count coulombs (amp hours). For each amp hour counted subtract 1 % from SOC. This fails to work very well because it does not take into account the fact that when full the voltage of your pack will be around 381 volts (112 cells * 3.4 volts) and when empty will be about 336 volts. When the pack temp is warm this would be good enough for most people to drive the car and have a reasonable expectation of using SOC% as a fuel gauge. A seemingly better way to tell SOC is by keeping track of the watt hours. Not difficult but more difficult than just counting amp hours. But unless the battery is heated is quite a bit less usable as a fuel gauge.

The problem with a battery that is not heated (LiFePo4 does not need cooling under typical use conditions in an EV) is that at cold temps the battery suffers from terrible voltage sag. Your increase in wh/mile in the winter could in part be due to the way the wh/mile number is calculated. The incorrect way would be to take the current times the nominal battery voltage (not real voltage) and integrate that with the time and divide by the miles. Even if they are doing it right they are probably also including the battery load from the environmental systems (cabin heater). In my DIY EV this was 1500 watts and is barely enough to demist the windshield. In the Model S it appears to be 6000 watts. Yours is probably somewhere in between. You also have the increase in viscosity in the gearbox lubricants and the increase in stiffness of the tires. Water/snow on the road surface also has an adverse affect on rolling resistance.

Now to answer your question about how to tell battery degradation.

Before you begin let me remind you to take extreme care when working on the battery. There is no way to turn off a battery so if you drop a tool across the terminals you can suffer from severe plasma burns. The tool could be vaporized. These 100 AH cells can source upwards of 2000 amps. Remove all jewelry and make certain you have some sort of eye protection. The danger from electrocution is is possible but far less likely than a burn from a dropped item.

Buy one of the smart chargers that the radio control people use with their Lithium RC battery packs. The Power Lab or iCharger brands are ones I have used. Charge the car up completely with its regular charger. Disconnect the battery from the car. This can be done by turning off the main disconnect if there is one or just removing the big cable from one end of the battery. Let the pack rest for several hours. This allows the voltage to stabilize. Pick a cell to test. I would do at least two, and the one at the positive end of the pack and the negative end of the pack could be good choices if they are easy to reach. Measure the voltage of that cell. If it is fully charged it should be resting somewhere around 3.38 to 3.40 volts. Set the RC charger for LiFePo4 type cells and set the low voltage cutoff to 3 volts. Set the current to the highest the device can sustain. The iCharger I have will do 25 amps. Perform a discharge test on that cell. Record that number as your baseline capacity value. The discharge cycle will take 4 hours if you can do 25 amps. Recharge that cell using the RC charger. Let the cell rest for a few hours and check its voltage. You want it to be close to the predischarge test value. If low you can add a few hundred mah to the cell by going through another charge cycle. If the voltage is high set the RC charger to discharge and remove a couple hundred mah. You can then repeat this on a second cell. Reconnect everything and go drive the car a little. Charge as normal and let the bms do its balancing thing if it needs to. Check those cells a year from now and compare the capacity. This will give you an exact number

Another way would be to fully charge the battery and then drive the car until it stops. Note the mileage. Do the same again next year on a day with similar weather conditions on the same route . The problem with this approach is that if your BMS is not configured to stop you what will happen is that you will ruin the weakest cell(s) in the battery. Not quite the intended result. Another problem is that you get to tow the car to a charge station. Not very convenient.

In any case, measuring the capacity is a destructive thing to do in that it places stress on the cells. With all three variants of LiIon types of cells the damage mostly occurs at the top and bottom of the SOC. The cells last a lot longer if you stay away from the top and bottom. People tend to naturally stay away from the bottom and Tesla has provided a slider to let you select the max SOC. I set my slider at 56% and keep the battery as close to 50% as possible. I have no idea when I will do a 100% range charge next to find out what the car thinks the range is. I will certainly not do it until I need to. The very act of measuring it will affect it adversely.

Good luck and have fun with your EV!

Doug Ingraham
DIY 1985 Mazda RX-7 16kwh battery.
Multi Coat Red 2013 Model S 85 "The Woman in the Red Dress"

 

[emir-t]

I suspect that you don't have enough information about the instrumentation to be able to tell how much battery degradation you are experiencing.

The SOC% would need to automatically re-calibrate itself as the battery degrades or it would not be able to tell you an actual %. The assumption of 100% SOC can be made at the end of the charge+balance. The cheap way to do this would be to assume your 100 AH batteries are really 100 AH and just count coulombs (amp hours). For each amp hour counted subtract 1 % from SOC. This fails to work very well because it does not take into account the fact that when full the voltage of your pack will be around 381 volts (112 cells * 3.4 volts) and when empty will be about 336 volts. When the pack temp is warm this would be good enough for most people to drive the car and have a reasonable expectation of using SOC% as a fuel gauge. A seemingly better way to tell SOC is by keeping track of the watt hours. Not difficult but more difficult than just counting amp hours. But unless the battery is heated is quite a bit less usable as a fuel gauge.

The problem with a battery that is not heated (LiFePo4 does not need cooling under typical use conditions in an EV) is that at cold temps the battery suffers from terrible voltage sag. Your increase in wh/mile in the winter could in part be due to the way the wh/mile number is calculated. The incorrect way would be to take the current times the nominal battery voltage (not real voltage) and integrate that with the time and divide by the miles. Even if they are doing it right they are probably also including the battery load from the environmental systems (cabin heater). In my DIY EV this was 1500 watts and is barely enough to demist the windshield. In the Model S it appears to be 6000 watts. Yours is probably somewhere in between. You also have the increase in viscosity in the gearbox lubricants and the increase in stiffness of the tires. Water/snow on the road surface also has an adverse affect on rolling resistance.

Now to answer your question about how to tell battery degradation.

Before you begin let me remind you to take extreme care when working on the battery. There is no way to turn off a battery so if you drop a tool across the terminals you can suffer from severe plasma burns. The tool could be vaporized. These 100 AH cells can source upwards of 2000 amps. Remove all jewelry and make certain you have some sort of eye protection. The danger from electrocution is is possible but far less likely than a burn from a dropped item.

Buy one of the smart chargers that the radio control people use with their Lithium RC battery packs. The Power Lab or iCharger brands are ones I have used. Charge the car up completely with its regular charger. Disconnect the battery from the car. This can be done by turning off the main disconnect if there is one or just removing the big cable from one end of the battery. Let the pack rest for several hours. This allows the voltage to stabilize. Pick a cell to test. I would do at least two, and the one at the positive end of the pack and the negative end of the pack could be good choices if they are easy to reach. Measure the voltage of that cell. If it is fully charged it should be resting somewhere around 3.38 to 3.40 volts. Set the RC charger for LiFePo4 type cells and set the low voltage cutoff to 3 volts. Set the current to the highest the device can sustain. The iCharger I have will do 25 amps. Perform a discharge test on that cell. Record that number as your baseline capacity value. The discharge cycle will take 4 hours if you can do 25 amps. Recharge that cell using the RC charger. Let the cell rest for a few hours and check its voltage. You want it to be close to the predischarge test value. If low you can add a few hundred mah to the cell by going through another charge cycle. If the voltage is high set the RC charger to discharge and remove a couple hundred mah. You can then repeat this on a second cell. Reconnect everything and go drive the car a little. Charge as normal and let the bms do its balancing thing if it needs to. Check those cells a year from now and compare the capacity. This will give you an exact number

Another way would be to fully charge the battery and then drive the car until it stops. Note the mileage. Do the same again next year on a day with similar weather conditions on the same route . The problem with this approach is that if your BMS is not configured to stop you what will happen is that you will ruin the weakest cell(s) in the battery. Not quite the intended result. Another problem is that you get to tow the car to a charge station. Not very convenient.

In any case, measuring the capacity is a destructive thing to do in that it places stress on the cells. With all three variants of LiIon types of cells the damage mostly occurs at the top and bottom of the SOC. The cells last a lot longer if you stay away from the top and bottom. People tend to naturally stay away from the bottom and Tesla has provided a slider to let you select the max SOC. I set my slider at 56% and keep the battery as close to 50% as possible. I have no idea when I will do a 100% range charge next to find out what the car thinks the range is. I will certainly not do it until I need to. The very act of measuring it will affect it adversely.

Good luck and have fun with your EV!

Doug Ingraham
DIY 1985 Mazda RX-7 16kwh battery.
Multi Coat Red 2013 Model S 85 "The Woman in the Red Dress"

Thank you very much for your detailed response. I don't think I'll be disconnecting the batteries and measuring etc. I have no experience with anything electrcial than changing a bulb.

I thought LiFePo4 charged all the way up to 3.6V no? Also, BMW does have a control mechanism to stop me. Finally, can you write some ways the wh/mi figure could be measured so I can test too see if it is?

Thanks again.