Regen Experiment

[Ray B]

I have a very repeatable driving pattern and the end of my daily return journey involves a 207 ft downhill slope (according to Google maps) for a little over a mile which adds some regen energy back into the battery. I decided to employ the 'Car Scanner' app to log the data from my OBD-II scanner and I added in my observations of the EV range at the top of the hill and when I got to my house. In ECON mode I made sure to do the same speed (37 mph) on the ACC and I swapped from ACC to paddle at a specific spot where the road flattens as it approaches a stop sign. So I tried to repeat this over the past couple of weeks to observe whether things like the SOC of the battery affect the regen.

It turns out that it does, but not in the way I was expecting. I had observed earlier that my gain in EV range by the downhill stint was much greater when the SOC was close to full, compared to when the battery was closer to empty. I did confirm that, but it appears that the actual gain in SoC is greater at lower SoC levels, and the system curtails some of the regen power at high SoC (say greater than 70%). In these summer months it is not really noticeable in the feel of the regen nor in the deflection of the 'needle' on the dashboard, as it is only a ~10% difference. Temperatures for all of these tests were ~80F give or take, windows down, no AC or other big power drains.

I expect that if I repeat this test in January when temperatures are below 0 C, I will observe a much more muted regen current at high SoC and somewhat normal current at low SoC. I may also run tests in SPORT mode to determine how much it affects the regen.

Here is the graph of the results:

 

[Groves Cooke]

I expect that if I repeat this test in January when temperatures are below 0 C, I will observe a much more muted regen current at high SoC and somewhat normal current at low SoC.

My car behaves very differently in summer and winter. I live at the top of a hill (150 ft). In the winter with a full charge my car will turn on the ICE unless I have run the heater for a few minutes before going down the hill and using the brake. In the summer, going down the hill and using the brake will add a few tenths of a mile in range and the ICE does not come on!

 

[MrFixit]

@Ray B,

As usual your inquisitive nature has provided much food for thought. Here are a few comments...

I think this version of SoC ranges from ~10% (2 bars) to 100% (20 bars, or a 'full' charge). Roughly speaking, this means that a "Delta SoC" of 90% corresponds to the total EV range (call it 45 miles for simplicity). Therefore you might expect ~0.5 miles of range for each percent of SoC added. I would judge that the Delta SoC that you get from your downhill runs is pretty consistently 1% (not really a function of SoC, or only very mildly if at all). This is expected as the same hill under the same conditions should produce the same amount of energy (and should be largely independent of SoC unless you reach the point where regen shuts down). I don't think regen shuts down until above 85% SoC (again, assuming 100% is a 'full' charge). There is always the extra headroom in true battery capacity that never gets touched above this level.

So, I expect to see an EV range increase of only 0.5 miles (largely independent of the SoC). The fact that you see 1-2 makes little sense, and especially, the trend with more range gained at higher SoC. The only thing I can figure is that EV range is a function of the 'fuzzy logic' of the GOM and thus, not tightly coupled to the SoC. The GOM is somehow thinking that you are a more efficient driver when this hill is encountered and the SoC is higher. It seems like the Delta SoC is less ambiguous and more consistent than the EV Range, so I would put much more weight on Delta SoC when evaluating regen, and mostly ignore the EV range.

I am curious about the 'Max Current'... This is also just as flat as the Delta SoC (not really a function of SoC). I am trying to relate to the number though. The 1% DeltaSoC that you observe would be ~0.14 kWh. If it takes you 2 minutes to go down the hill, this charging rate would be 4.2 kW. With a current of 100A, that means a voltage of 42V. The nominal HV battery voltage is 348. Is this current somehow at a 'module' level? I can't remember enough details about the architecture to know how to relate to this.

Can you provide some more context about this "Max Current"? Going down your hill (which I assume is moderate regen) you get 100A. What extremes do you see with this? How many amps vs. the number of chevrons? How many amps does this max out at (with heavy braking, and max indication on the charge/power gauge)?

What happens with this current reading when you are not regenerating? Does it reverse polarity and indicate negative currents for battery discharge, and does it map to the 'power' range of the dashboard gauge?

 

[JFon101231]

So, I expect to see an EV range increase of only 0.5 miles (largely independent of the SoC). The fact that you see 1-2 makes little sense, and especially, the trend with more range gained at higher SoC. The only thing I can figure is that EV range is a function of the 'fuzzy logic' of the GOM and thus, not tightly coupled to the SoC. The GOM is somehow thinking that you are a more efficient driver when this hill is encountered and the SoC is higher. It seems like the Delta SoC is less ambiguous and more consistent than the EV Range, so I would put much more weight on Delta SoC when evaluating regen, and mostly ignore the EV range.

I would agree 100%, its being driven from the "fuzzy" logic

 

[Landshark]

Before and after SOC readings would be a much more accurate method of determining regenerative capability than EV range estimates. It is understood that the purpose of the experiment was to measure differences in regenerative energy when starting from various levels of SOC. I’ve seen a displayed EV range of more than 100 miles after a long descent and an overnight charge. So I know how meaningless that estimate can be.

Regenerative capability is also, approximately, infinitely variable.

For the record:
Nominal voltage is ~310 (3.7 volts x 84 cells)
Target voltage is ~344 (~4.1 volts x 84 cells)
Maximum voltage is ~352 (4.2 volts x 84 cells)

 

[Ray B]

@Ray B,

As usual your inquisitive nature has provided much food for thought. Here are a few comments...

I think this version of SoC ranges from ~10% (2 bars) to 100% (20 bars, or a 'full' charge). Roughly speaking, this means that a "Delta SoC" of 90% corresponds to the total EV range (call it 45 miles for simplicity). Therefore you might expect ~0.5 miles of range for each percent of SoC added. I would judge that the Delta SoC that you get from your downhill runs is pretty consistently 1% (not really a function of SoC, or only very mildly if at all). This is expected as the same hill under the same conditions should produce the same amount of energy (and should be largely independent of SoC unless you reach the point where regen shuts down). I don't think regen shuts down until above 85% SoC (again, assuming 100% is a 'full' charge). There is always the extra headroom in true battery capacity that never gets touched above this level.

So, I expect to see an EV range increase of only 0.5 miles (largely independent of the SoC). The fact that you see 1-2 makes little sense, and especially, the trend with more range gained at higher SoC. The only thing I can figure is that EV range is a function of the 'fuzzy logic' of the GOM and thus, not tightly coupled to the SoC. The GOM is somehow thinking that you are a more efficient driver when this hill is encountered and the SoC is higher. It seems like the Delta SoC is less ambiguous and more consistent than the EV Range, so I would put much more weight on Delta SoC when evaluating regen, and mostly ignore the EV range.

I am curious about the 'Max Current'... This is also just as flat as the Delta SoC (not really a function of SoC). I am trying to relate to the number though. The 1% DeltaSoC that you observe would be ~0.14 kWh. If it takes you 2 minutes to go down the hill, this charging rate would be 4.2 kW. With a current of 100A, that means a voltage of 42V. The nominal HV battery voltage is 348. Is this current somehow at a 'module' level? I can't remember enough details about the architecture to know how to relate to this.

Can you provide some more context about this "Max Current"? Going down your hill (which I assume is moderate regen) you get 100A. What extremes do you see with this? How many amps vs. the number of chevrons? How many amps does this max out at (with heavy braking, and max indication on the charge/power gauge)?

What happens with this current reading when you are not regenerating? Does it reverse polarity and indicate negative currents for battery discharge, and does it map to the 'power' range of the dashboard gauge?

Thanks for the comments. I'll do my best to address them, though I don't consider myself well-versed in electrical topics.

The Delta SoC that I plotted was from the app's output of the 'real' SoC meaning it represents the state of charge without the buffers added by the BMS. But that doesn't invalidate your rough estimate of 1% = 0.5 miles which is close enough. And yes, it is just fuzzy logic of the EV range algorithm that provides the disproportionate boost with a minor gain in SoC.

Max current is also provided by the app, and it is negative values during regen, though I plotted it as positive values to avoid a wonky looking graph. Based on your question, I examined one downhill event closely to run the numbers and the entire regen lasted 135 seconds, and although the max was >100 Amps (briefly), the average of the entire stint was more like ~27 Amps (just guestimating + integrating + averaging...). At 4200 W / 27 A = 155 V; you need to double it to get the system voltage which is 310V, which is pretty close to where it was during this specific test. So a lot of rough guessing going on, but the numbers do seem to work out.

And I do believe the OBD-II current measurement is correct because I also did data-logging of the gradations on the power meter, and also at the click-point of the accelerator (max EV power), which came out to ~86 kW give or take, which is close to the accepted value of 90 kW (121 hp).

I can post my values for the power meter tick marks if anyone is curious. I don't have reliable values for the regen tick marks as of yet.

As for your question on the current vs the number of chevrons on the paddle, that will have to be a separate experiment. For my test there is a specific steep grade at one point and the cruise control regen spikes to keep the car at an even speed. Shortly after the road flattens out again and when the regen power begins to lessen, that is when I release the ACC and engage 4 chevrons, but this peak regen current is normally lower than the previous ACC spike. I do it this way because (as long as there is no hindering traffic), it is easily repeatable.

 

[Landshark]

Max current is also provided by the app, and it is negative values during regen,

The negative values for current may indicate that the measurement is being taken from the traction motor rather than at the battery after being massaged by the inverter. Voltage and current measurements taken at the battery would allow us to determine regenerative capability and charge rates.

the average of the entire stint was more like ~27 Amps (just guestimating + integrating + averaging...). At 4200 W / 27 A = 155 V; you need to double it to get the system voltage which is 310V, which is pretty close to where it was during this specific test. So a lot of rough guessing going on, but the numbers do seem to work out.

Charging voltage would need to be in the 345-350V range to reach the target cell voltage of 4.1V. The ICE starting when regenerative energy is created and the HV battery is fully charged would be a good indicator that the regenerative charging voltage and current are fairly high.

The nominal voltage of the battery is 310.8V. A charging voltage of 310V would do very little to charge a battery below nominal voltage and would provide no charge at all to a battery that is above nominal voltage.