phidauex
Well-Known Member
- First Name
- Sam
- Joined
- Dec 8, 2020
- Threads
- 16
- Messages
- 844
- Reaction score
- 1,548
- Location
- Colorado
- Vehicles
- 2021 MachE 4EX, 2006 Prius, 1997 Tacoma
- Occupation
- Renewable Energy Engineer
- Thread starter
- #1
I've been capturing some data files using Carscanner Pro and a Viecar ODBII reader, and while I haven't had time to really extract a lot of knowledge out of them, I've finally been successful at capturing some good data and getting it cleaned up with some Python scripts. It is now in a state that is fairly easy to look at and work with.
The "drive" datafile is from a ~30 minutes spirited drive out of town on fairly twisty, up and down roads, 1pd, Unbridled, 90F ambient temp, climate control on automatic low, 72F. The actual sample rate was about 600ms, and I've resampled the data to 500ms resolution and used some rolling average filters to ensure that all the data points are filled.
One or two insights: The car doesn't like to use the front motor much at all - I'm only seeing it come in when acceleration needs to be over 0.10gs. Above that and it starts blending in about 40% front. I'd go so far as to say that this car isn't "RWD biased" in the way that a Subaru is "FWD biased" but that it is straight up a RWD car that can call for a short burst from the front motor if needed.
During regen, however, it pulls the front motor in aggressively (or doesn't pull it in at all). Regen is either 100% rear motor, up to 90kW, but in some cases it mixes the front motor in at about 50%. There may be a speed relationship there that I haven't sifted out yet.
Charging was using P&C at SigNet manufactured Electrify America location. P&C worked fine - plugged it in, waited a moment for it to negotiate, and it started charging. Finished the charge using the unlock button by the charge port and it ended without errors. I charged from 45% to 81% indicated, though it appears that the datapoint I'm reading is actually the "real" SoC of the battery, not the adjusted SoC reported to the user. This is handy and could be used to check for degradation. This file was resampled to 1s resolution.
A few small insights. The charge curve looked as expected - a few minutes at ~122kW, then a gradual ramp down to 70kW at 80% SOC, then the cliff down to 12kW.
The charge efficiency was very good - accounting for about 1.1 kW of constant load from the car accessories (sitting with the AC running, best estimate), the battery's internal efficiency was about 97% one-way, which is quite good for a battery charging at 1C. That only meant about $0.43 spent on battery losses over the course of the charge.
Anyway, I'll be continuing to play with the data as I collect more, should get some long drives in soon. Let me know if there are datapoints you'd like to see, or if you learn anything from the files.
The two raw CSVs are straight from Carscanner, and haven't been processed. The "merged" files have been cleaned up, resampled, some calculated columns added, and some unnecessary columns removed.
UPDATE: Added another datafile, this one from 8/14/21 which roughly follows this route from Boulder to Buena Vista: https://abetterrouteplanner.com/?plan_uuid=0967e052-15cd-4e3c-82c3-e707dda11552
The trip was 131 miles, and used 44.1kWhs. In ABRP this aligns with a reference consumption of 3.55 miles/kWh. My actual consumption was 2.97 mi/kWh, but this route has a lot of big climbs up into the mountains, and I was going pretty quick. I also had the vehicle heavily loaded with gear, including a hitch rack with two bikes. This suggests that the default reference consumption in ABRP is fairly conservative.
Here is some longer run voltage information. In this case we see the voltage at 387V when the SOC is 100% reported (95.6% real). This graph is filtered for times when power is close to zero to prevent voltage drop/rise from affecting the values. Incidentally, this is why the SOC is so important for drag racing - at 100% SOC there is fully 12% more voltage available to the motors compared to 50% SOC.
Looking at it a different way, we can calculate the internal resistance of the battery and the cables feeding the inverters. Here we see Current vs. Voltage, filtered for 89% SOC. The measured voltage is in blue, and calculated voltage is in orange. The calculated voltage is just the "resting" voltage for each SOC (calculated from the above graph's linear regression), minus the voltage drop from the current * a fixed resistance (V = I*R). I varied the resistance until things looked good (chimping it). My estimate for the internal resistance of the Extended Range battery is 0.036 ohms, or 36 mOhms, which is right in line for this type of battery.
This should help people trying to troubleshoot the poor gent with the GTPE and his trap speed.
UPDATE 3:
Ok, posting a little more data, this is from a drive that included two 0-60mph pulls, totally floored, near the beginning of the logs. The second one at 190 seconds in was launched using a brake hold, Unbridled, 66% SOC (not great), 60F outdoor temp, climate control on low, no extra weight in the car.
CarScanner's acceleration timer put this at a 5.8s 0-60 (including the rollout). That is probably realistic given the lowish SOC, and it isn't a perfect timer or a prepped run.
The datalog has been resampled to 0.5 seconds, but was capturing around 0.7 seconds, this makes it a bit hard to tell what is going on at this high rate of change, but here are a few observations.
On a speed and power basis, you can see that I floored it (pedal position is dashed green line), power came up a bit, and then I launched. Power ramped up as I accelerated, and I hit peak power at about 44mph. The front motor came up to 50kW quickly and stayed there. The rear motor went to 200kW, but took a little longer to get there. It looks like it folds back a bit at the last second, but it is possible that is a sampling error. 250kW at the motors is about 335hp, not to bad given the low SOC.
On a torque and motor speed basis things are interesting. After the launch the torque slams up to max, as expected. The rear motor peaks at 431 Nm, and starts to fold back torque at 3727 RPM. The front motor peaks at 145 Nm, and starts folding back sooner, at 2718 RPM.
When I let off the pedal, both motors drop to a -75 Nm of torque. Interestingly when driving the car is heavily rear biased, but when regening it is very balanced, probably to create more controlled braking.
Data file attached, as usual (2021-09-21 18-34-33-merged.xlsx.zip)
The "drive" datafile is from a ~30 minutes spirited drive out of town on fairly twisty, up and down roads, 1pd, Unbridled, 90F ambient temp, climate control on automatic low, 72F. The actual sample rate was about 600ms, and I've resampled the data to 500ms resolution and used some rolling average filters to ensure that all the data points are filled.
One or two insights: The car doesn't like to use the front motor much at all - I'm only seeing it come in when acceleration needs to be over 0.10gs. Above that and it starts blending in about 40% front. I'd go so far as to say that this car isn't "RWD biased" in the way that a Subaru is "FWD biased" but that it is straight up a RWD car that can call for a short burst from the front motor if needed.
During regen, however, it pulls the front motor in aggressively (or doesn't pull it in at all). Regen is either 100% rear motor, up to 90kW, but in some cases it mixes the front motor in at about 50%. There may be a speed relationship there that I haven't sifted out yet.
Charging was using P&C at SigNet manufactured Electrify America location. P&C worked fine - plugged it in, waited a moment for it to negotiate, and it started charging. Finished the charge using the unlock button by the charge port and it ended without errors. I charged from 45% to 81% indicated, though it appears that the datapoint I'm reading is actually the "real" SoC of the battery, not the adjusted SoC reported to the user. This is handy and could be used to check for degradation. This file was resampled to 1s resolution.
A few small insights. The charge curve looked as expected - a few minutes at ~122kW, then a gradual ramp down to 70kW at 80% SOC, then the cliff down to 12kW.
The charge efficiency was very good - accounting for about 1.1 kW of constant load from the car accessories (sitting with the AC running, best estimate), the battery's internal efficiency was about 97% one-way, which is quite good for a battery charging at 1C. That only meant about $0.43 spent on battery losses over the course of the charge.
| Min SOC | 45.97 |
| Max SOC | 78.30 |
| Total kWhs in Battery | 99 |
| Energy to Battery | 32.0 |
| Energy from Charger | 33.5 |
| Vehicle Overhead Energy | 0.54 |
| Charge Efficiency | 97.0% |
Anyway, I'll be continuing to play with the data as I collect more, should get some long drives in soon. Let me know if there are datapoints you'd like to see, or if you learn anything from the files.
The two raw CSVs are straight from Carscanner, and haven't been processed. The "merged" files have been cleaned up, resampled, some calculated columns added, and some unnecessary columns removed.
UPDATE: Added another datafile, this one from 8/14/21 which roughly follows this route from Boulder to Buena Vista: https://abetterrouteplanner.com/?plan_uuid=0967e052-15cd-4e3c-82c3-e707dda11552
The trip was 131 miles, and used 44.1kWhs. In ABRP this aligns with a reference consumption of 3.55 miles/kWh. My actual consumption was 2.97 mi/kWh, but this route has a lot of big climbs up into the mountains, and I was going pretty quick. I also had the vehicle heavily loaded with gear, including a hitch rack with two bikes. This suggests that the default reference consumption in ABRP is fairly conservative.
Here is some longer run voltage information. In this case we see the voltage at 387V when the SOC is 100% reported (95.6% real). This graph is filtered for times when power is close to zero to prevent voltage drop/rise from affecting the values. Incidentally, this is why the SOC is so important for drag racing - at 100% SOC there is fully 12% more voltage available to the motors compared to 50% SOC.
Looking at it a different way, we can calculate the internal resistance of the battery and the cables feeding the inverters. Here we see Current vs. Voltage, filtered for 89% SOC. The measured voltage is in blue, and calculated voltage is in orange. The calculated voltage is just the "resting" voltage for each SOC (calculated from the above graph's linear regression), minus the voltage drop from the current * a fixed resistance (V = I*R). I varied the resistance until things looked good (chimping it). My estimate for the internal resistance of the Extended Range battery is 0.036 ohms, or 36 mOhms, which is right in line for this type of battery.
This should help people trying to troubleshoot the poor gent with the GTPE and his trap speed.
UPDATE 3:
Ok, posting a little more data, this is from a drive that included two 0-60mph pulls, totally floored, near the beginning of the logs. The second one at 190 seconds in was launched using a brake hold, Unbridled, 66% SOC (not great), 60F outdoor temp, climate control on low, no extra weight in the car.
CarScanner's acceleration timer put this at a 5.8s 0-60 (including the rollout). That is probably realistic given the lowish SOC, and it isn't a perfect timer or a prepped run.
The datalog has been resampled to 0.5 seconds, but was capturing around 0.7 seconds, this makes it a bit hard to tell what is going on at this high rate of change, but here are a few observations.
On a speed and power basis, you can see that I floored it (pedal position is dashed green line), power came up a bit, and then I launched. Power ramped up as I accelerated, and I hit peak power at about 44mph. The front motor came up to 50kW quickly and stayed there. The rear motor went to 200kW, but took a little longer to get there. It looks like it folds back a bit at the last second, but it is possible that is a sampling error. 250kW at the motors is about 335hp, not to bad given the low SOC.
On a torque and motor speed basis things are interesting. After the launch the torque slams up to max, as expected. The rear motor peaks at 431 Nm, and starts to fold back torque at 3727 RPM. The front motor peaks at 145 Nm, and starts folding back sooner, at 2718 RPM.
When I let off the pedal, both motors drop to a -75 Nm of torque. Interestingly when driving the car is heavily rear biased, but when regening it is very balanced, probably to create more controlled braking.
Data file attached, as usual (2021-09-21 18-34-33-merged.xlsx.zip)
Last edited:
