Calculate usable battery capacity based on rated miles values

[ArtInCT]

I have something to add here based on my experiences. Here in Connecticut, especially in the Winter my Rated Miles are always optimistic and I get around .8 real miles per rated mile on average. My use case here at home is about 6 drives each being an average of 7 miles. So nothing warms up well, pack heater comes on often in the beginning of trips which makes for efficiency losses and regen is often limited for the entire use of the car. Roads are hilly, windy, narrow and environment is cold and my MPH range is about 28 on average. Ok so that is my norm. I also use my heater, seat heaters, defrosters, and steering wheel heater as the car cools down between trips. When it is really cold, one is forced, at times to use the front defroster and rear as the windows fog up if you just use the seat heaters for comfort. For example, as I write this my S's interior temp is 34 in the garage and it is 22 outside.

However, on a recent one month road trip from CT to FL where the temps were between 72 and 82. I did experience some great variances from my CT norms.... trip legs were about 140 miles on average (between superchargers). As we progressed south from CT, to NC, SC, GA and FLA I did notice that my real miles and rated miles were really becoming much closer. I attribute this to the following.... the battery pack was warm so no pack heater losses, regen was always available, the use of HVAC was diminished to at times not used at all, the terrain was nearly totally flat and the roads were smooth and excellent and I was doing 75 MPH on average. (Actually somewhat slow for the traffic).

So my actuals and rated pretty much equaled down in the South. Wh/m was an incredibly low 282 for the weeks we were in FLA. Normally the Wh/m for me is 322 here in CT and in the really cold weather 348. This is a P90DL so that big rear motor takes a toll.

The relationship of Rated Miles and Actuals may be very related to use case AND environmental variables from what I have seen. I am NOT saying Rated Miles change, just their relationship to actuals changes.

 

[Dan43]

Not sure what I can add but my experience is the discrepancy between the 'last charge' measurements and the rated miles as per the % and miles left (my chart line is 305Wh/mi an S75 in the thread should be 285Wh/mi)

Having run from genuine 100% via the last charge maths I used 53.2kW for 192 miles but when I re-charged back to 100% I had put back 62kW, mileage left was 28 based on EPA. Trip was 273kW/hr.

So a kW difference of 8.8kW from the two measurements?

 

[Missile Toad]

Cars in the South are more Green:
@ArtInCT says he gets 322 Wh/mi (348 in 'really cold weather') -- which I take to mean that is his average, for all seasons, in Connecticut. In contrast, he adds:

...282 for the weeks we were in FLA...​

I'm a refugee from Chicago, who now makes his home in Houston, so I expect the difference in driving the Model S between those towns is similar to the Connecticut/Florida differences @ArtInCT noticed. This yields a ratio of 322/282 or a 14% difference in driving efficiency. This indicates that a car purchased in these far-South locations will have the following properties (for so long as driven in those climates):

• 14% or so less carbon emissions;
• Correspondingly lower wear on the battery -- e.g. need to drive 114K miles to see same # of cycles on a similar car driven 100K miles, up North;
• Shorter visits at the Superchargers, in relation to the cross-country miles driven;
• potentially, fewer battery repair issues for the same number of miles (assuming daily temperature swings and vibrations don't age the battery significantly)

Having said that, it really is incumbent for us in the South, to encourage our friends and neighbors to participate in the EV revolution.

 

[hacer]

Not sure what I can add but my experience is the discrepancy between the 'last charge' measurements and the rated miles as per the % and miles left (my chart line is 305Wh/mi an S75 in the thread should be 285Wh/mi)

Having run from genuine 100% via the last charge maths I used 53.2kW for 192 miles but when I re-charged back to 100% I had put back 62kW, mileage left was 28 based on EPA. Trip was 273kW/hr.

So a kW difference of 8.8kW from the two measurements?

I assume you meant 273 W-hr/mi for your last figure.

Was the 62 kW-hr the AC meter power or the charge energy added reported by the car? If it is the former then the 8.8 kW-hr discrepancy is the loss due to charger efficiency plus round-trip battery efficiency. 0.94^3 would explain it. That is 6% energy loss in the charger, 6% energy loss going into the battery and 6% coming out. Of course it's not that bad because we know that the kW-hr since last charge doesn't count vampire loss, pre-heating, and parked energy consumption.

So > 94% efficiency for each of these losses isn't as good as it could be but it's also not a disaster.
https://web.archive.org/web/2009032...s/PluginHwy_PHEV2007_PaperReviewed_Valoen.pdf suggests only 80 to 90 percent efficiency for vehicle lithium ion batteries so it might actually be quite good.

If the 62 kW-hr was charge energy added then the batteries were only 92% efficient each way.

 

[Boatguy]

Not sure what I can add but my experience is the discrepancy between the 'last charge' measurements and the rated miles as per the % and miles left (my chart line is 305Wh/mi an S75 in the thread should be 285Wh/mi)

Having run from genuine 100% via the last charge maths I used 53.2kW for 192 miles but when I re-charged back to 100% I had put back 62kW, mileage left was 28 based on EPA. Trip was 273kW/hr.

So a kW difference of 8.8kW from the two measurements?

1) What RM was displayed when you departed on the trip?

2) WRT the 8.8kWh difference, used versus replaced is I think also being discussed in this thread by others who are debating the various measurement points and losses/efficiencies during the charge/discharge process.

Having said that, it really is incumbent for us in the South, to encourage our friends and neighbors to participate in the EV revolution.

Off topic, but I would absolutely appreciate that effort. FWIW, I made a small contribution to a friend in San Antonio buying a used Leaf (young couple of modest means). There are a lot of good bargains on used EVs starting to hit the market and they make great commute cars for many people.

 

[hiroshiy]

Minor question. I have X P100D and its RM is 333Wh/m, according to @wk057 's post. Now if I keep Sport mode not ludicrous, would that change??

 

[jerry33]

Minor question. I have X P100D and its RM is 333Wh/m, according to @wk057 's post. Now if I keep Sport mode not ludicrous, would that change??

It shouldn't. It's how you drive it that determines what range you will get. Of course, it's easier to use more energy with ludicrous mode on, but having it on should not cause higher energy use by default.

 

[hiroshiy]

It shouldn't. It's how you drive it that determines what range you will get. Of course, it's easier to use more energy with ludicrous mode on, but having it on should not cause higher energy use by default.

Ok, thank you for the answer. I can then keep Ludicrous mode all the time then I thought if I press the accelerator pedal, Ludicrous gives me more kW with the same amount of pedal travel and that way, I will tend to over accelerate than necessary causing higher energy consumption.

 

[jerry33]

Ok, thank you for the answer. I can then keep Ludicrous mode all the time then I thought if I press the accelerator pedal, Ludicrous gives me more kW with the same amount of pedal travel and that way, I will tend to over accelerate than necessary causing higher energy consumption.

It's up to you to educate your foot to not use that extra energy--you have to press lighter on the pedal. Kind of like how you can use more energy with a V8 Ram than you can with a 3 cylinder Insight. It used to be easier before they wrecked the display.

 

[wdolson]

I just have a standard S and it already has the potential to accelerate too quickly under normal conditions. Last summer I passed someone on the highway and thought I was up to around 80 mph. The friend in the passenger seat finally said "dude!" and I looked down, I was doing 101 mph. I didn't even hit the accelerator all that hard. If I had a Performance model, I wouldn't have Ludicrous on all the time, too much potential for getting into some serious trouble without realizing it.

Not only does the risk of speeding tickets go up, but over accelerating and having an accident is possible too. That may have been a factor in the fatal crash in Indianapolis, Indiana a few months back. I believe that was a P90DL and witnesses said the car was doing around 100 mph on a city street before hitting a tree.

A guy I worked with some years ago took a couple of weeks off and we heard his wife had been in a bad accident. I talked to him about it after he got back and he said she had always wanted a V-8 Mustang and finally got one. On a rainy day trying to merge into traffic from a driveway, she over accelerated and hit a few cars before stopping. These were the days before traction control and the accident might not happen with a newer car, but too much horsepower in regular traffic don't always mix well.

For the sort of situations you're likely going to encounter on city streets, most of the time just the acceleration of a standard Tesla is enough to impress all but the most jaded people.

 

[AWDtsla]

I just have a standard S and it already has the potential to accelerate too quickly under normal conditions. Last summer I passed someone on the highway and thought I was up to around 80 mph. The friend in the passenger seat finally said "dude!" and I looked down, I was doing 101 mph. I didn't even hit the accelerator all that hard. If I had a Performance model, I wouldn't have Ludicrous on all the time, too much potential for getting into some serious trouble without realizing it.

Not only does the risk of speeding tickets go up, but over accelerating and having an accident is possible too. That may have been a factor in the fatal crash in Indianapolis, Indiana a few months back. I believe that was a P90DL and witnesses said the car was doing around 100 mph on a city street before hitting a tree.

A guy I worked with some years ago took a couple of weeks off and we heard his wife had been in a bad accident. I talked to him about it after he got back and he said she had always wanted a V-8 Mustang and finally got one. On a rainy day trying to merge into traffic from a driveway, she over accelerated and hit a few cars before stopping. These were the days before traction control and the accident might not happen with a newer car, but too much horsepower in regular traffic don't always mix well.

For the sort of situations you're likely going to encounter on city streets, most of the time just the acceleration of a standard Tesla is enough to impress all but the most jaded people.

You know how many Subaru's are roaming the streets with 400-500HP in a 3400lbs chassis? Or for that matter GT-R's with minimal upgrades running 600-700HP?

If you can't handle it don't buy the car...

Mine stays in ludicrous all the time, and many times it's too slow.

 

[AWDtsla]

Thanks to more teslafi logging, I can say the theory that the "lost" capacity is just unmetered losses internal to the pack is just rubbish.

I had a 15 minute drive this morning at an average speed of 10mph, heat blasting, But more importantly, barely used any throttle through the drive. Probably stayed under 40kW for the whole trip, with the max logged power at 35kW. Yet there was still a discrepancy of 9.4% between trip meter reported consumption, and calculated via RM*Wh/RM loss.

 

[Boatguy]

Thanks to more teslafi logging, I can say the theory that the "lost" capacity is just unmetered losses internal to the pack is just rubbish.

I had a 15 minute drive this morning at an average speed of 10mph, heat blasting, But more importantly, barely used any throttle through the drive. Probably stayed under 40kW for the whole trip, with the max logged power at 35kW. Yet there was still a discrepancy of 9.4% between trip meter reported consumption, and calculated via RM*Wh/RM loss.

Could you expand on that with the actual numbers? What was the change in RM? How many odo miles? Trip meter kWh reported, Teslafi reported, etc.?

 

[AWDtsla]

Could you expand on that with the actual numbers? What was the change in RM? How many odo miles? Trip meter kWh reported, Teslafi reported, etc.?

2.5 odo, 5.76 RM, 660Wh/mi trip meter, 722Wh/mi calculated.

 

[hacer]

Thanks to more teslafi logging, I can say the theory that the "lost" capacity is just unmetered losses internal to the pack is just rubbish.

I had a 15 minute drive this morning at an average speed of 10mph, heat blasting, But more importantly, barely used any throttle through the drive. Probably stayed under 40kW for the whole trip, with the max logged power at 35kW. Yet there was still a discrepancy of 9.4% between trip meter reported consumption, and calculated via RM*Wh/RM loss.

I think you are misunderstanding things. There are two different meters going on here: (1) The battery state-of-charge (SOC) estimator (feeding RM), and (2) The energy consumption meter (kWh since last charge and charge_energy_added from the REST API).

The SOC uses many things to estimate the present available-for-use capacity of the battery. It includes the history, battery voltage(s), the battery temperature, and the data from the energy consumption meter. It uses a complex, proprietary algorithm. The car computes RM in linear proportion to the SOC estimate, if you display % instead of RM, you are seeing a different linear scaling of SOC.

The energy consumption meter is nothing more than an energy flow meter integrating the product of battery terminal voltage and battery current. It doesn't even operate when the car is not "ON", so for example vampire energy is not measured by it, nor is battery thermal regulation while the car is "OFF" measured by it. For certain it does not estimate the internal losses in the battery either.

So for your specific example, consider some scenarios: overnight your car was "OFF", but the there was some thermal management of the battery during this "OFF" period as well as other "vampire" drain. The energy consumption meter didn't account for these, nor did the battery SOC estimator fully comprehend them (it was sleeping too). However, as you did your drive, the SOC estimator could see that the cell voltages were lower than it had been expecting and so it adjusted its estimates accordingly. At the end of the drive you're left wondering why the SOC went down so much when you didn't use that much energy. That is only one scenario that can explain it. Another could be that you have a poorly balanced pack or some unhealthy cells in your battery so the SOC estimator has to account for that because it knows that it must limit the whole pack's available energy such that the lowest series module will maintain its required minimum voltage. Exactly how this is observed is also a function of temperature, so when the pack was cold at the start of the drive, the SOC estimator thought more was possible but as the packed warmed up, it recognized that it wasn't as good as it previously thought. Throughout your drive it continuously updated its available energy estimate which kept getting worse and worse. In either case, if you had kept driving for another hour, things would have leveled off and the overall discrepancy would have been smaller.

The SOC estimator is sophisticated and actually does a reasonable job as your fuel-gauge. It's not fantastically linear and it can make adjustments (both up and down) as it tries to estimate the battery state, it likely does most of the things you've been saying are "easy" to do in software, except that it is only trying to deal with estimating available energy, not actually estimating how far you can go with that energy because it doesn't know if you're about to go up a mountain etc. But I've (mis)understood that you've been ragging on the energy consumption meter which lacks all sophistication. It's not that good at its job because it sleeps on the job (no doubt in an effort to save energy while the car is "OFF"). It could probably be improved if a dedicated low-power processor that was always on was used for that function, but I suspect its handled by the main system which would waste too much power to keep on all the time. Even if it didn't sleep on the job, it still isn't trying to be a SOC estimator, it's job is to measure actual consumption. But if it was better, you'd see better agreement but there would still be discrepancies because of internal battery losses.

Finally, all of this is worst during cold weather when there are large changes to the battery temperature and potentially significant energy to manage them. It's likely you'll see better results when its warmer.

 

[AWDtsla]

I think you are misunderstanding things. There are two different meters going on here: (1) The battery state-of-charge (SOC) estimator (feeding RM), and (2) The energy consumption meter (kWh since last charge and charge_energy_added from the REST API).

The SOC uses many things to estimate the present available-for-use capacity of the battery. It includes the history, battery voltage(s), the battery temperature, and the data from the energy consumption meter. It uses a complex, proprietary algorithm. The car computes RM in linear proportion to the SOC estimate, if you display % instead of RM, you are seeing a different linear scaling of SOC.

The energy consumption meter is nothing more than an energy flow meter integrating the product of battery terminal voltage and battery current. It doesn't even operate when the car is not "ON", so for example vampire energy is not measured by it, nor is battery thermal regulation while the car is "OFF" measured by it. For certain it does not estimate the internal losses in the battery either.

So for your specific example, consider some scenarios: overnight your car was "OFF", but the there was some thermal management of the battery during this "OFF" period as well as other "vampire" drain. The energy consumption meter didn't account for these, nor did the battery SOC estimator fully comprehend them (it was sleeping too). However, as you did your drive, the SOC estimator could see that the cell voltages were lower than it had been expecting and so it adjusted its estimates accordingly. At the end of the drive you're left wondering why the SOC went down so much when you didn't use that much energy. That is only one scenario that can explain it. Another could be that you have a poorly balanced pack or some unhealthy cells in your battery so the SOC estimator has to account for that because it knows that it must limit the whole pack's available energy such that the lowest series module will maintain its required minimum voltage. Exactly how this is observed is also a function of temperature, so when the pack was cold at the start of the drive, the SOC estimator thought more was possible but as the packed warmed up, it recognized that it wasn't as good as it previously thought. Throughout your drive it continuously updated its available energy estimate which kept getting worse and worse. In either case, if you had kept driving for another hour, things would have leveled off and the overall discrepancy would have been smaller.

The SOC estimator is sophisticated and actually does a reasonable job as your fuel-gauge. It's not fantastically linear and it can make adjustments (both up and down) as it tries to estimate the battery state, it likely does most of the things you've been saying are "easy" to do in software, except that it is only trying to deal with estimating available energy, not actually estimating how far you can go with that energy because it doesn't know if you're about to go up a mountain etc. But I've (mis)understood that you've been ragging on the energy consumption meter which lacks all sophistication. It's not that good at its job because it sleeps on the job (no doubt in an effort to save energy while the car is "OFF"). It could probably be improved if a dedicated low-power processor that was always on was used for that function, but I suspect its handled by the main system which would waste too much power to keep on all the time. Even if it didn't sleep on the job, it still isn't trying to be a SOC estimator, it's job is to measure actual consumption. But if it was better, you'd see better agreement but there would still be discrepancies because of internal battery losses.

Finally, all of this is worst during cold weather when there are large changes to the battery temperature and potentially significant energy to manage them. It's likely you'll see better results when its warmer.

I'm not misunderstanding anything. This has nothing to do with sleeping overnight or cold or anything circumstantial. What I'm reporting is reproducible 100% of the time, in any condition, in any state of charge, in any weather, sequentially drive after drive, there is no room for mis-estimation by the BMS.

 

[hiroshiy]

I got X P100D in January 2017 and for the first time I fully charged to 100%, but the max range was 437km (273 miles) which is far from EPA 289 miles. So I thought again we (Japan) have different rated range than other parts of the world? When I had RWD P85 back then the rated was approximately 320Wh/m. Would that mean Japanese/European rated (=typical) range is different from EPA?

 

[jerry33]

I got X P100D in January 2017 and for the first time I fully charged to 100%, but the max range was 437km (273 miles) which is far from EPA 289 miles. So I thought again we (Japan) have different rated range than other parts of the world? When I had RWD P85 back then the rated was approximately 320Wh/m. Would that mean Japanese/European rated (=typical) range is different from EPA?

That would be my assumption. For tires Japan uses JTRTO, Europe uses ETRTO, and North America uses TRA. I don't know the names of the EPA equivalents.

 

[hacer]

I'm not misunderstanding anything. This has nothing to do with sleeping overnight or cold or anything circumstantial. What I'm reporting is reproducible 100% of the time, in any condition, in any state of charge, in any weather, sequentially drive after drive, there is no room for mis-estimation by the BMS.

Post your data set.

 

[AWDtsla]

Post your data set.

4 drives of varying length in widely varying temperature, from the perspective of "maybe the trip meter is right":

Trip meter reported kWh / change in reported SoC = kWh estimated usable

62.1/(.93-.12) = 76.67
53.8/(1-.28) = 74.72
21.32/(.69-.41) = 76.14
48.6/(.89-.25) = 75.94

Seems terribly consistent, even using nearly all of pack capacity. That's a nice 75kWh pack I have there.