Torque sleep not functioning properly on some cars?

I am wondering if torque sleep is not functioning properly on some cars. In particular, I'm interested in determining if torque sleep is functioning properly or not on my P85D. I have some pretty solid evidence that it is not. I have been trying to discuss this with Tesla, but not having much success. Before I make even more of a nuisance of myself than I already have, I thought I would present my case for torque sleep not working properly in my car here, to see what some of you think.

I have a P85D, with 19 inch Cyclone wheels and the Pirelli Sottozero Series II winter tires. They have been inflated properly. Last night I checked them and they were all at about 43.5 PSI and I increased them all to 46 PSI.

When I first contacted Tesla about this it took a couple of days and a few phone calls, but eventually I received the following:

--
One of our field engineers reached back out to me regarding torque sleep on your vehicle.
He said that torque sleep is active on your vehicle. When he looked at the logs for your vehicle, here is the information the car told us.

1/23 through the end of 2/1 (before torque sleep update was pushed to your vehicle)
Total Avg consumption: 417wh/mi
Avg HVAC (A/C and Heater) Consumption: 47wi/mi
Avg 12v consumption: 11wh/mi
Total avg – avg HVAC – avg 12v = 359wh/mi

2/2 through 2/12 (after torque sleep update was pushed to your vehicle)
Total Avg consumption: 434wh/mi
Avg HVAC (A/C and Heater) Consumption: 68wi/mi
Avg 12v consumption: 15wh/mi
Total avg – avg HVAC – avg 12v = 351wh/mi

So it looks like any efficiency increase you gained from torque sleep, was canceled out by the increased usage of the HVAC system and the 12v battery (increase in 12v consumption could come from the seat heaters, radio, lighting, etc). He said your car is operating normally and torque sleep is enabled and working. He said that there are a lot of variables that can affect your efficiency, like speed, temperature, tire PSI, acceleration rate, cargo in the vehicle, and terrain.

So at this point in time everything look like it is working properly on your vehicle. If you have more questions, let me know.
--

Note that the conclusion that torque sleep is working is in no way supported by the numbers, which show a decrease of 8 wh/mi after adjusting for HVAC ad 12V usage. 8wh/mi is basically noise over such a short time period. It is also roughly 2% of the 359 wh/mi original figure.


I have photos of trip data I took on February 4, just a few days into torque sleep, and again today. Using the February 4 information as "before", I get the following:

This shows a very slight reduction of 14 wh/mi in the month we've been driving with torque sleep. But before we get too excited about this 14 wh/mi, let me point out a couple of things. One is that those first 2358 miles include the first 1000 miles, on brand new tires, which are supposed to be at worse efficiency. Also I certainly did a lot more Insane Mode 0-60 launches in those first 2000 miles than I've been doing lately. So without any benefit from torque sleep we would have expected to see some improvement in this last month. Also before 02/04 we never used range mode. We have been using range mode almost exclusively once we learned that torque sleep worked more effectively in range mode, with the exception of a few trips while we still had .167 installed and knew that we should not be using range mode in .167. Even before torque sleep existed, there were efficiency gains from using torque sleep, so some of the gains of the last month should be attributed to that. If any of the 14 wh/mi improvement is due to torque sleep itself, it's probably only a small part of the improvement.





In addition to the evidence above, I've been logging all of our highway trips, which are the vast majority of our trips, and comparing them to EV Trip Planner expectations. The result is that we're typically near or above EV Trip Planner energy usage estimates. Other P85Ds that have compared trips to EV Trip Planner are typically 8-12% better than EV Trip Planner expectations. (I started a thread on this here: Comparing P85D Torque Sleep efficiency (versions .139 and .140) to EV Trip Planner)

The vast majority of the driving we do and all the logged trips include a highway segment of at least 30 minutes as well as a good bit of time on rural roads with speed limits of 45-55 MPH. This would seem to be the perfect environment for torque sleep to kick in. Whenever possible, meaning whenever weather conditions permit, we use TACC on the highway portion of the trips, so driving style does not come into play. And as for driving style, both my wife and I are probably on the very mellow end of the curve. She sets the TACC speed to 68 and I set it to 70 (the speed limit is 65) and we both set the following distance to the max of 7. These trips are also all taken in sport mode. Here are all those trips, with the comparisons to EV Trip Planner:


What do you think? Is it reasonable for me to believe torque sleep isn't working correctly on my car, based on my efficiency numbers as compared to other P85Ds, and as compared to my numbers from before torque sleep was implemented? And if it's not working correctly on my car, there's a chance it's not working correctly on others as well.

Thanks for any input.

mspohr said:

You can believe what you want but I would believe the Tesla engineer.
Your anecdotal evidence is really irrelevant because there are too many uncontrolled variables.

Click to expand...

You'd believe the Tesla engineer even though the Tesla engineer's numbers did not support the Tesla engineer's conclusion? I don't believe the engineer did an exhaustive investigation into the logs of my car and determined torque sleep was functioning properly. The engineer confirmed that torque sleep was "active on my vehicle." For all I know he may have just confirmed that I had a version of the firmware that included torque sleep installed, and then forwarded along the numbers he pulled. It's those numbers that I have an issue with, and they are not anecdotal. Those numbers, directly from that same Tesla engineer, indicate an improvement of 8 wh/mi, after adjusting for HVAC and 12V. 359 wh/mi before and 351 wh/mi after. 8/359=.0222. That's a little over 2%. In his blog post JB Straubel said:

"The software update to implement torque sleep will be downloaded to the dual motor fleet by the end of January 2015 and will substantially improve the range of dual motor vehicles by roughly 10%."

2% is not roughly 10%. In fact, it is roughly 1/5 of 10%, or in layman's terms no where near 10%. So that is one reason I don't want to just "believe the Tesla engineer."

As for my anecdotal evidence being irrelevant, I take issue with that as well. I have been collecting data quite carefully and then comparing it very meticulously against EV Trip Planner data. I've been comparing that to what other people here have recorded. There are significant differences between my numbers, which remain close to EV Trip Planner, or are worse than EV Trip Planner predicts, and others' numbers that typically beat EV Trip Planner by a good margin.

stevezzzz said:

I can't say for sure, but given that the temperatures you're driving in are consistently below freezing and often in the teens or single digits F., my gut tells me that it's the weather that is killing your efficiency. Wait until you see some temps consistently in the high 40s and above, then report back.

Click to expand...

I've wondered about that too. And if Tesla said "Torque sleep just doesn't work when it's really cold" that would be a sufficient answer for me. But they haven't said that. And some people here who have posted data from trips in the cold have seen an improvement in efficiency. So while I'm still wondering about the cold, I'm definitely not sure enough to just wait until spring and see if everything settles out as I'd hope it would then.

gpetti said:

Some thoughts.
I haven't been able to hold onto my car long enough to do any analysis - though I could probably match you for cold.
I guess the question with the Tesla engineer would be whether he/she actually saw a log that showed the disconnection (sleeping) of the rear motor, or if as you suspect they simply assumed that you had the right software, or looked at the numbers an assumed TS because they were slightly better.

Click to expand...

This was one of the frustrating aspects of dealing with Tesla on this.

I started out by just calling the 800 number. It took a few calls and follow-ups with a rep there, who seemed to be treating this as a low priority. (One of the times when I called back after a few days he acknowledged that he simply had forgotten to forward my email.) This rep was serving as the intermediary between me and engineering. That might not have been so bad if there hadn't been so much lag in the process. But when I responded to the engineer's e-mail with specific questions about when the car was torque sleeping, just as you suggest, instead of following up with the engineer the case somehow got switched over to a service center, to "confirm" that what the engineer had told me was correct. The service center had no way to do that. So I was basically starting from scratch, but now with the service center manager acting as go between to engineering. The most recent response was a canned response about how torque sleep works in general, with no specifics about whether or not it is working on my car and when.

It seems as if everyone at Tesla is happy to just push this aside, since efficiency is so variable, by using that as an excuse, since it's a lot easier to do that than to actually dig into the issue and figure out if something might be wrong or not. Since until about a month ago motors weren't torque sleeping at all, is it that ridiculous to think that perhaps there could be some issue where some of them don't torque sleep properly? I would think Tesla might have an interest in finding out if that might be happening. But instead I've been met with a lot of patronizing explanations of the factors that affect efficiency. It's quite frustrating.

gpetti said:

think in another post I was suggesting that you might be able to use energy measurements, sounds, feel to figure out if your car is in torque sleep or not. That is probably easier said than done as I imagine Lolachampcar's sensitivity to driving response is probably finer that yours (or most drivers). In one of the threads (possibly wk057s comparison thread) I think a couple of people commented on how they can tell by the energy usage reduction.

Click to expand...

I have tried this a bit, and not gotten anywhere. I never see the energy usage suddenly drop, to indicate one of the motors has entered torque sleep, and I certainly can't hear anything to be able to tell. I'm driving with snow tires, which probably isn't helping. Or perhaps my car isn't torque sleeping, so there's nothing to see or hear.

gpetti said:

With regard to the EV Trip planner comparisons, i understand why you are using this tool from another thread where you explained how comparing relative efficiency against the EV estimate allowed comparisons between cars with most of the variables removed; however, what if EV trip planner has bad cold weather prediction accuracy while being relatively accurate in other ways. It could also be accurate down to a certain temperature but there could be some temperature point where the battery heating demands become non-linear or something like that. Sorry if you've been through that particular part of the analysis already.

Click to expand...

The issues with EV Trip Planner are interesting. Just tonight I came up with something along those lines myself, but you raise good and different points. What I came up with was the possibility that there is some major error in the route that we're always plugging into EV Trip Planner, and it is somehow computing lower energy usage than it should be.

Of course the EV Trip Planner possible issues wouldn't explain the Tesla generated efficiency improvement numbers being so inconsequential, and my own before and after lifetime numbers reflecting such a small difference.

I think Stevezzz could be on the right track if torque sleep just doesn't work when it's very cold, but again, if that is the case, why wouldn't Tesla just say that? And if Tesla doesn't know that yet, perhaps they should be taking a good hard look at my data and comparing it to the cars' logs and figuring out what is going on. (For most of the trips in my spreadsheet I could give them a pretty accurate set of times when the car was on the highway. You'd think if they were interested in figuring out what's going on with torque sleep in the cold, having good historical data like the data I've been recording that could match up with the cars' logs would be pretty useful.)

I'm just not sure where to go with this next. I asked that very question of the service center manager today, via e-mail, but have not yet heard back.

BerTX said:

Just because torque sleep is implemented and functional on your car doesn't mean it is working.

Click to expand...

Exactly! That's why I was trying to get an engineer to look at the logs to see if torque sleep was actually working or not, and see if they could tell me roughly what percentage of time it was engaging. If they couldn't tell me the percentage of time it was engaging because the information was proprietary, I was hoping they could at least tell me that it appeared to be engaging the correct percentage of time. I haven't yet been able to get confirmation that torque sleep has ever engaged on my car!

BerTX said:

It seems that the torque sleep is probably very much dependent on the load on the motor. The load probably has to be consistent and fairly low. There are lots of factors that can increase load without the driver even noticing it -- things like tire inflation pressure or tire type, yadda yadda, that you are aware of and trying to eliminate.

There can be other mechanical things that can cause load, like a brake that is dragging or some other factor that isn't obvious. It may be something that, in an ICE, would not even be noticed because it would just cause a 1% drop in fuel economy. In this case, however, it is enough to disable torque sleep, which causes a precipitous drop.

Click to expand...

Again, I say yes, exactly! This is why I need Tesla's help. But they really seem more interested in "making the problem--(meaning me)--go away" than in helping me get to the bottom of what is really happening. I'm sure they are convinced that I'm just another person that doesn't understand that efficiency is very variable, and doesn't understand why I'm not seeing actual miles equal range miles, when of course that's not it at all. I could understand, perhaps, being treated like that for the first few minutes of this experience, but once I started sending them data and explaining that I was using comparisons to EV Trip Planner expectations as a measurement tool, you'd think they'd realize that I wasn't just complaining and asking "Why isn't my wh/mi 300?"

BerTX said:

Another thing I notice about your methodology is the attempt at getting some consistency by trying to maximize use of cruise control. It is possible that the cruise is what is actually causing your problem. Whereas manual speed control allows a more steady throttle position, allowing minute changes in vehicle speed, the cruise control will immediately react to any speed change. What you are looking for is probably a constant load to allow torque sleep to engage (disengage?), and the cruise control's job is to constantly manipulate throttle position -- the opposite of consistency.

Click to expand...

That would be an interesting thought, but I think I have to disagree with this, only because you don't have all the information about my trips. The trips I've been recording are very similar, and are about 90-95% the same route, with a different start/end point. When my wife is driving she is going to work in Syracuse, from our home in Ithaca, and back. When I am driving, I am going from our home in Ithaca to the place we stay in Syracuse when we need to stay up there, or back the other way. So as I mentioned, most of both routes is the same. On these routes there is a solid 30 minutes or so of 65 MPH speed limit, two-lane each way highway. This is the only portion where we engage the TACC, and the TACC generally does not have to vary the speed very much. Sure, there is some traffic, and the TACC will occasionally slow the Model S down and speed it up again, but much of the time is spent just cruising at 70 for me and 68 for my wife. Aside from that, the other half of the trip, when we never use TACC, has plenty of 45-55 MPH cruising as well. We don't use TACC because this is on rural highway that is typically one lane in each direction, and through towns where the speed limit will drop to 30 or 35, with stop lights, etc., but there will be stretches of 5-10 minutes at a time where torque sleep should be able to kick in. Finally, I'm sure most of the other P85D owners who are reporting the efficiency improvements are seeing the improvements when using TACC, so I really doubt taking it completely out of the equation is the answer. I do agree, though, that for anything other than highway driving TACC could definitely be decreasing efficiency, and probably somewhat significantly.

I'm still waiting to hear back from the service manager with respect to who I can go to next for answers.

BerTX said:

Might it be practical for you to go to a place that would have a demo P85D and do a drive of the same route in your car and the test car to see if there is a difference? You would need to drive both cars, same route, just to show that it isn't a difference in the driver. It would not be easy, since there are so many variables you'd have to try to match between the two.

Or a nearby owner who wants to help.

You need to do something to either convince Tesla that something is wrong, or convince YOU that nothing is wrong (i.e., just a weather factor).

Click to expand...

The problem is I'm a couple of hundred miles from any service center, and probably almost that far from any other D owner on the forum. So while good ideas, those kind of comparisons probably aren't practical. One thing I could try to do would be to do some sort of test of a specific loop or something with range mode on and range mode off, since Tesla says range mode on is supposed to have a greater impact on torque sleep. If I did this test with HVAC off, the stereo off, etc, in theory that should limit the difference in power consumption to any change in weather conditions between the tests, my ability to drive consistently, and the change in torque sleep behavior, if torque sleep is occurring. I may have to try that.

kennybobby said:

from J.B.'s blog noted the decreased range effect of cold temperature:
Model S Efficiency and Range | Tesla Motors

Climate Control energy usage is related to keeping the cabin at a comfortable temperature. Maintaining a desired temperature in the cabin requires energy draw from the battery, which affects range. It is challenging to generalize how much the use of climate control will affect your range. It depends predominantly on how different the cabin temperature is from the outside temperature, if there is a large heat load from the sun, and how long you operate the climate control system. If you are driving slowly for a very long time and operating climate control continually it will have the largest percentage impact on your range. In very hot or cold operating conditions with typical usage of climate control and driving at speeds around 55 mph you may see a 10-15% reduction in range.

Click to expand...

Do you have a point in there somewhere? I almost don't want to waste my time by addressing this.

Everything I'm doing controls for the temperature. The comparisons against EV Trip Planner take the temperature into account. The long-term before and after data are over long enough and similar enough periods that the temperature should equal out. The data supplied by Tesla took HVAC and thus temperature as it applies to cabin heating completely out of the equation. I'm not comparing my range now to range last summer. I'm comparing range now, when I supposedly have torque sleep to range when I didn't have torque sleep, with weather taken into account. Perhaps you were trying to be helpful, but you might as well have included a paragraph on the frunk dimensions. It would have been equally relevant.

Now if JB Straubel had said "Torque sleep doesn't work in the cold" that would be a different story, but he didn't say anything of the sort. And so far, neither has anyone from Tesla.

AmpedRealtor said:

I would wait until the weather warms up and you are better able to eliminate weather, cold, wind, etc. as variables.

Click to expand...

I may wind up having to do that. But at this point I have so much time and energy invested in this, I'd really like to get an answer, especially if there is, in fact, something wrong with my car.

qwk said:

You will never get anywhere with pursuing this problem with Tesla once the SC gets involved, and are the ones communicating with engineering. Engineering is a information black hole(even with their own SC's). Specific communication back to you, or the SC pertaining to complex problems just doesn't happen. As long as the car is working, you are SOL.

Click to expand...

I am beginning to discover this. Thanks for the comment. At least now I know what to expect.

gpetti said:

I think that when I was experimenting, I did see a drop in energy. If I had the car I would tell you some sequence of events that dependably causes this but I found this note from LolaCC:
"... I literally can accelerate to 45 mph using about 80 KW lightly lifting when I approach my target speed. The power drops for the lift and then, a second or so later when I am steady state at 45, the power falls significantly. This second fall is combined with a noticeable change in throttle response. I may be crazy here but I am reasonably sure I can tell exactly when TS engages. "

Click to expand...

I'll have to try that exactly, and see if I can detect the change. Of course if I can't detect it, that won't prove anything, but it's still worth trying. Thanks!