Look Inside Kona Electric Battery And See Real Capacity

Have a read of post #1 in this thread. Because the voltage varies as the cell drains measuring the energy content is not as straightforward as you have assumed. The purpose of the nominal voltage as marked on the cell is to make that easier. Aside from the basic cell rating, we don't know exactly what portion Hyundai utilise in the Kona's pack.

 

Well, I inadvertently tried to take a hit for the team this morning, and ran down to 2.5% en route to work this morning...
Some observations:
1) BMS and display SOC started matching up about 7% and below whereas there is a bigger disparity at higher charge levels.
2) A lot more variation across cell voltage as SOC got lower; just before plugging in, cell voltage ranged from 2.96-3.04 volts.

 

The only way to do this precisely is to calculate the area under the discharge curve between the voltage limits you're operating in. In practice that means charging the battery to your max operating voltage, and discharging it again to the minimum operating voltage while measuring the current x instantaneous voltage the entire time, and adding up all those datapoints.

Not nearly as convenient as just estimating it by multiplying nominal cell voltage by nominal cell amp-hours.

 

Cheers for that unexpected datapoint. The voltage values will be fractionally on the low side of OCV (<0.05 V) due to the pack being under a slight load keeping the Kona alive.

It's worth noting that the E63 datasheet considers 2.5 V to be a minimum OCV while the SoC table has 0% at 3.167 V.
Additionally the cell cycle life data only covers the range 0 to 97% SoC, so it seems that the application of the cell is intended to have buffers at both ends. That might explain why we typically see an SoC(BMS) of around 96-97% at full charge.
But from Docpw's datapoint it also seems to indicate that the Kona's 0% SoC may be lower than the cell's application working range, perhaps because the car would not be expected to be frequently used under approximately 10%.

As a side note, the SoC axis in my graph is 'any' SoC, whatever definition of SoC the particular voltage values refer to. So, until we ascertain that the car's BMS SoC matches the cell's SoC, mismatches in the graph lines are to be expected.

 

A couple of months ago, I repurposed my 8 year old Mahindra e2o battery back into my rooftop solar, and turned it into an off grid system. Works really well, and is a fantastic 2nd life option for EV battery packs.

 

I not so sure about your claim of probable useable capacity of 62 kWh. I have done a couple discharge runs from 100%- zero %SOC(BMS SOC 95.5-0%) and the actual distance traveled divided by reported consumption x .955 typically have suggested an actual usable capacity of 64kWh and I assumed the remaining 4.5% of unavailable upper BMS SOC or around 2.88 kWh was the reserved headspace. There definitely is some headspace as typically a Kona EV driver still has full regenerative capacity even at an indicated 100% SOC. Please free to correct my math or assumptions.

 

Regarding minimum cell voltage, we ran ours one day such that we came in the driveway with the car at 3%, so I took a voltage snapshot!

 

That datapoint falls exactly on LG's OCV v.s. voltage spec for 3%. Since the current draw at 3.5 A is small we can consider that close enough to OCV.

 

Sorry I am not entirely sure how your reply to my post supports your contention of 62kWh of actual available battery capacity ? I understand that regenerative braking makes more energy available but that is already accounted for by the BMS through its shunt measurements and represented in its reported consumption calculation. Again if I measure the actual distance driven(lets say Km and the odometer is correctly measuring/ GPS verified) with the total battery energy as reported 100% to 0% SOC (actual BMS start reading is 95.5%) and then divide the total distance travelled in Km by the BMS reported consumption figure in km/kW(which again takes into account captured regenerative energy) and multiple again by 0.955(correction for the actual total BMS SOC used) I will always get 64kWh of actual useable battery capacity not 62 kWh as you have suggested. I suppose the BMS could be lying and underrepresenting the actual consumption figure to hide an actual battery capacity of 62kWh but I am not entirely sure why it would want to do this?

 

Sorry If I came off too brusquely. I very much appreciate your input and am happy to defer to your expertise here.

 

Realizing that cold temp ( -5 C+/-) likely skews my values...

 

It's safe to assume that the "E63" refers to the cell raw capacity as 63 Ah. Many verification tests in the datasheet cover the entire "standard" Li-ion OCV range from 2.5 to 4.20.

For the EV application the recommended range is presumably 0-97%, also referred to in some tests and it seems that it would match the marked capacity of 60Ah, or 218Wh based on a nominal 3.63 V. I thinks it's likely that Hyundai must contractually not exceed that range and does so as SoC(BMS). As such, the SoC(displayed) range would genuinely be 64 kWh for the pack when it's all fresh and new. Why we often see a max SoC(BMS) of 95.5% is not clear however, but there are many 96s and 97s seen as well.

What's also not clear is the behaviour at the low end, OCV between the 3.167 OCV defined by LG as 0% and the absolute minimum allowable OCV of 2.5. Based the two user datapoints we have it's still unproven if Hyundai do or don't use any of that range. Based on a linear extrapolation of SoC(BMS) vs SoC(displayed) that I've tracked on a 'long' trip down to around 30%, they meet at 0% and differ only by a fixed ratio.

I'd suggest that it's plausible that Hyundai would allow some of this range under 0% to be used as an emergency buffer, but not for driving, only so the car can independently power its electrical systems to allow charging to commence.

Determining capacity based on dash consumption readouts and extrapolation has potential errors that are unlikely to produce an accurate result down to the <1% you'd need to come close to assessing pack capacity in the car. The most accurate sensor data we have from the Kona is the pack's current shunt which feeds the two cumulative current registers, which unfortunately only display a resolution of 0.5 Ah over OBD. If you read pack current directly off OBD you are limited by the sampling rate and would have to integrate that data in real time to get Ah. The BMS will be doing this itself, probably at mS intervals to accurately increment the two Ah registers, and along with the instantaneous pack voltage the two kWh registers. This is a measurement that really should be done in lab conditions.

Along with Ah you need OCV which is not easy to measure off via OBD since the system is concurrently pulling power off the pack and that drags the voltage down slightly. The only practical and accurate method I can come up with is reading the change in both Ah registers when driving the entire SoC(disp) range between 100 and 0% and comparing that with 180 Ah. But you're still stuck with all four readings introducing an error of 0.5/180, or a total of 0.3% x 4 = 1.1%.

 

67 kWh is the total energy available in the battery. As the battery discharges, the voltage decrease so does the available kWh. For the first % SoC when the battery is fully charged, you get 1/100 of 73.7 kWh, about 0.73kWh. The next % of SoC will hold a little less and so on. The last % SoC holds around 0.56 kWh. If you add the energy for each of % of SoC, you will get 67 kWh. I collected lots a data while driving my Kona in the past months. I was able to establish the amount of kWh for each % of SoC. Take note that 180Ah is constant throughout the battery discharge. See the graphic I obtained from curve fitting the data I collected. If you integrate the energy under that curve you would get 64 kWh.

 

The OBD2 actually gives a resolution of 0.1 for CCC, CDC, CEC and CED. The SoC has a 0.5 resolution.