August 16th, 2020 by Danny Parker
To Measure is to Know
How do we find out how we are doing in any endeavor in a complex machine? The need is for specific performance data that allows an informed decision — at least, if we wish.
Our continuing case for providing key energy data for Tesla Model 3 is well illustrated by our continuing example from the earliest days of human spaceflight.
On 24 May 1962, Mercury Astronaut Scott Carpenter attempted to duplicate John Glenn’s three orbits of the earth — at that time an amazing feat. The tiny Mercury spacecraft, while designed to be fully automatic, did not always live up to that aim. Over a hundred measurements were taken in the capsule in real time, but the engineer-designers, assuming that the human being aboard was little more than a passenger, did not provide much of the data to the astronaut. That proved a mistake.
The lack of onboard data made for dangerous developments for Scott Carpenter and Aurora 7. On the very first orbit, the HVAC system in Aurora 7 began to malfunction, telemetry being received at mission control showing the cabin rising above 104°F. Another display on the ground showed Carpenter’s core body temperature rose to over 102°F. Carpenter did not have that data available and there ensued a disagreement between the pilot and the flight directors as to how hot the interior really was? As Carpenter fumbled with the HVAC settings, there was no immediate feedback in the spacecraft. Bad again.
While there was a gauge showing remaining hydrogen peroxide fuel quantity, there was no indication of the rate of fuel burned when thrusters were fired. Anxious controllers could see gulping rate of consumption from data on the ground. However, Carpenter had no idea he was inadvertently leaving on a thrust system that needlessly wasted fuel with every maneuver. He couldn’t see that and it bit him. More next time…
Readers can guess my direction: human drivers (pilots) in any vehicle that is not operating flawlessly need informative and effective HVAC controls as well as information on the rate which fuel (power) is being used. Even if the vehicle is fully autonomous, it still has human passengers that need comfort and need to know how to most efficiently achieve it if one wishes to travel far. If a human is driving, they need power use information too — whether they are moving or not.
Performance of Tesla Model 3s in the Wild
Meanwhile, let’s circle back to the experience of my friend Evan Mills and what it can tell us about the importance of better energy feedback and HVAC controls in the Tesla Model 3.
One thing Evan wanted to know in one of our weekly séances seemed at first like a difficult question: “How efficiently am I driving the car compared with others?” How does all this play out?
We measured the performance of a rear-wheel-drive (RWD) Model 3 Long Range on the road, but without heating or cooling operating. In contrast, earlier this year, Car and Driver performed tests examining Model 3 energy use under harsh circumstances in winter, and with possibly the least favorable control mode (AUTO).
Those are two ends of the spectrum. What would performance be like “in the wild?” What is typical performance? It turns out that we have data. Although, Tesla itself has much more. Here I surface what we know from the non-proprietary data mining by others.
A Better Route Planner (ABRP) already has crowdsourced data from hundreds of Model 3s that have signed up to share data within that platform, which provides an enhanced navigation environment. These data show the performance of 233 Model 3 cars that were accumulated with the vehicles remaining constantly at a given speed for 30 seconds or more. 220 of the Model 3s were Long Range (LR) versions, mostly at that point with RWD. There were also data for the dual-drive Performance model (P3D). The data are a bit dated, and at the time, there was not a good sample of the newer all-wheel drive (AWD) non-performance LR version. Still, the comparison with our collected data seems quite valid. The key insight from this source is all power used by the cars, including HVAC and other electrical systems.
The graphs below, courtesy of ABRP, show the performance of the measured Model 3 LR and dual-motor performance version (P3D) of the model “in the wild.” The data reflect thousands of miles from those contributing data to ABRP. Each of the blue points in the data represents the power demand of one of the cars for a 30 second period at a particular speed. As one might expect, there is large variation in the data. Although, ABRP has fitted curves to the median values (yellow circles) seen at each speed. Multiply m/s by 2.237 to obtain miles per hour.
What does all of this mean for typical Model 3 drivers at highway speeds? At 65 mph, these data boil down as follows:
- Model 3 LR (mostly RWD): 15.08 kW; 232 Wh/mi
- Model 3 P3D (Performance, Dual Drive): 17.35 kW, 267 Wh/mi
The Model 3 Performance uses about 15% more energy at the 65 mph speed. I fit the same statistical model to the ABRP data in the following figures that I used for characterizing tests of my Model 3 in Florida:
The statistical results below agree very closely with what we measured in my road tests, save one key difference. The intercept kW that doesn’t seems to change with speed was about 0.4 kW with everything off in our road tests. However, it is about 1.8 kW in the ABRP data:
ABRP data for the LR Model 3:
kW = 1.76+ 0.098 (mph) + 0.0000254 (mph3)
ABRP data for P3D:
kW = 1.79 + 0.124 (mph) + 0.0000276 (mph3)
Our road test of my LR with no HVAC for comparison:
kW = 0.376 + 0.095 (mph) + 0.0000281 (mph3)
Now, we revise our earlier graph from our road test and the Teslike data to include the ABRP data with the performance of Model 3s in the wild.
In our tests in Florida without HVAC on flat roads at 65 mph, we measured 13.81 kW or 212 W/mi. The ABRP data suggests that at 65 mph the typical Model 3 LR driver is using about 1.4 kW more than we show. Although this can come from many factors (grade, tire inflation, aero wheels vs. not), it is likely more than coincidental that this is approximately the average power measured for the car AC system operating or the heating system idling. What would all this mean for range? If we assume perhaps the 75 kWh battery is available, we end up with the relationships seen below.
- Model 3 LR (ABRP data): 15.08 kW; 319 miles
- Model 3 LR (No HVAC; flat roads): 13.81 kW; 348 miles
Presenting the same data rendered as range yields the relationship shown below.
One might legitimately question the Wh/mi of current AWD Tesla Model 3s and how they would compare to my test data (RWD) as well as the early ABRP data. It turns out that this is readily available at speeds of 55–75 mph courtesy of the clever analysis behind the Teslike.com website.
They leverage published EPA dynamometer testing and use this to interpolate performance across models. For readers’ convenience, I have translated these data into Wh/mi against speed as for the other comparisons, shown in the graphic below.
The EPA dynamometer data used by Teslike clearly shows that when Tesla moved to the AWD configuration, there was a reduction in the efficiency versus the single motor RWD (which has been discontinued — except now for production in China). However, as can be seen, a large part of the efficiency loss can be made up for with the 18” wheels with the aero covers.
Although not always considered aesthetically pleasing, these data clearly show the rather striking advantage of the aero wheels at highways speeds. Their increase to efficiency has been variously tested to be on the order of 3–5%. Teslike assumes 4%, which seems appropriate, as Car and Driver in separate testing saw a 3.4 % efficiency improvement — and 10 miles of additional range — under controlled track test conditions between 50 and 90 mph. Teslarati performed a similar test in West Covina, California, at a constant 70 mph and found a 4.3% efficiency improvement.
Although the Model 3’s very low Cd is advanced, that is almost assuredly not the end of it. Aerodynamic efficiency will always remain fundamental to machines moving through the atmosphere. The Cd of the Model 3 is only 0.23, but some very detailed after-market CFD simulation work has shown that stick-on front and rear spoilers can potentially reduce overall Cd to 0.20 — another 10% drop in aerodynamic drag at 70 mph. The rear view mirrors are a real drag sore spot and Tesla is clearly working towards being able to obviate them at highway speeds, likely using cameras.
Penetrating the Darkness: HVAC Controls
Pre-heating the Model 3 in winter before driving or pre-cooling in summer before driving is a key method to avoid immediate HVAC power use and associated battery drain on the road. It’s unlikely using pre-heating will drain a lot of power from the battery, but it does put the HVAC into AUTO mode. So if you weren’t expecting the change, it could elevate consumption. So, it is generally wise to place the HVAC back into manual mode with the fan speed you desire when you depart. Obviously, with heating, recirculating the interior air will reduce the heating load for the drive.
With the Model 3, since you set the system to manual control, it will remain that way unless changed. Two events can inadvertently change this on the unwary: using the phone app to pre-heat/pre-condition the car, or when the driver selects automatic pre-conditioning (AUTO). AUTO will also leave the fan speed in what it last determined when you change to manual, so you will have to change it back to what you want rather than assume it is at a low number
When you are using manual controls, if you set the A/C off (greyed out), it will not turn on to cool the cabin under any circumstance. However, there is another dark point of unanticipated energy use for the unsuspecting:
The silent PTC resistance heaters will turn on if the temperature is set high enough to need its assistance. For instance, if a driver has the temperature set to “LO” while in manual control, no resistance heat will be used. However, if you set it to say 65 degrees, then the vehicle will blend in heat to make the air 65ish degrees coming out of the vents. Of course, you will have no signal to let you know this is happening
One rule of thumb, for moderate conditions in Northern California, is that setting the cabin to 72 degrees and below will not use the resistive heater once the car has had some time to warm up the coolant, maybe 3 miles or so. But until the coolant has warmed up, it will use the resistive heater to warm the air at the start of the trip.
Helpful service personnel with Tesla Berkeley were kind enough to help us with a few tips from their own experience and training on efficient EV driving habits. I was able to add some of my own.
First, only sparingly use the automatic climate control (AUTO). “The system simply tries to do too much. It will have that A/C running in the background even if you have it set to 80 degrees just to make what is likely air that is plenty dry even more dry.”
The key point is to run the space conditioning system in the manual mode and select what you wish from the options there.
In colder conditions, it is important to pre-heat the vehicle before leaving the garage, which is then largely done with house power. However, when that is done, remember to turn the HVAC system back from AUTO to manual when you take the wheel.
When departing in the pre-heated car in winter, plan to use the seat heaters to the maximum possible to avoid excessive use of the energy-hungry cabin heat. Seat heaters are a low energy option (110 watts per seat with 3 bars of heat) that put warmth directly onto your body, instead of trying to heat all the air in a car to a comfortable temperature.
When using cabin heat, setting the heat on manual at temperatures 72 degrees or lower maximizes the use of waste heat and minimizes the use of the resistive heating elements. For most, 72 degrees on manual with the A/C turned off is plenty of heat, at least in mild, coastal California winters.
In Florida, it is almost not needed. Pre-heat and a sweater work best here. But in the cold north, both cabin heat and seat heaters will be frequently needed.
Tesla’s own tips are here: https://www.tesla.com/support/range
Keep in mind that the air conditioner in the Tesla Model 3 is a fully variable speed compressor. I’ll repeat that so you can really hear it: the cooling system in the car is fully variable speed. This means anything you can do to make it run more slowly will save power — although, not in a linear fashion (like other variable speed compressors).
The compressor’s power use can be only ~1 kW at low speed (4 range miles lost per hour) against up to 6 kW (24 lost miles of range per hour) with the compressor running very hard at full fan and at high compressor speed.
So, the key is to keep the compressor speed down and not to let the cabin temperature get really hot inside the car before you get in. We measured it with the ScanMyTesla app, going through several repetitions to get the numbers right. The compressor speed is proportional to:
- The distance the cabin temperature is from the temperature setting.
- The fan speed, which is the indicator for the compressor fan speed as well.
Sonically, you can hear the higher compressor speeds, particularly when you choose AUTO on entering the car after it has been sitting in a hot place. If you have ScanMyTesla, you can see in a stationary car how power increases when the fan speed is increased with the air conditioner on. Of course, we shouldn’t need the ScanMyTesla app and hardware — the car should provide this critical information, which it already has. Tesla only has to make it available.
The excess energy needed to pull down the interior temperature to the set point can be controlled to a certain extent by:
- Parking in the shade.
- Expert window tinting for all windows.
- Shade screen insert for top window, and also adding top insulation.
- Using windshield solar shade.
Also, the fact that the AC compressor is fully variable speed means that occupants are rewarded for controlling heat gain while driving (window tint, window screens) as well as choosing thermostat set points that are not too low. Generally, drivers and passengers will find the interior feels cooler at a given setting if the car has been parked in the shade. This is because human comfort is strongly related not only to the air temperature, but also to the global mean radiant temperature (MRT) from automobile interior surfaces.
Energy Use of Other Accessories
Beyond HVAC, we also measured the power use of accessory systems in the Model 3, which I have then graphed below. The measured idling power of my Model 3 is approximately 250 watts sitting still in a parking lot. Placing the car in drive, this number goes up about 100 watts.
Note that most of the big numbers seen above (defrost and cabin heat on Max and AUTO as well as air conditioning when “pulling down to thermostat”) show the initial startup power and not the power draw over time as the loads are met. Remember that that air conditioner is fully variable speed, so the setting on the thermostat relative to inside temperature will matter a lot along with the fan speed, which is the signal for the compressor for how hard to work. The compressor is somewhat noisy at high speeds and it gives you a sonic indicator of what you see in the chart above that we measured with ScanMyTesla. If you hear the compressor running, it is drawing a good amount of power from the battery.
I ran checks on these numbers for fans speeds and AC power use for three repetitions and used the median for the plot. There will be some variation in what you see for your own individual circumstance, in particular where one starts relative to the thermostat setting. The uncertainty of the measurements, although repeated, is about +20 watts given the noise on the battery power signal.
Nevertheless, this chart gives an idea of “the big knobs.” For cabin heat, it matters whether both PTC (Positive Temperature Coefficient) heaters are engaged, left and right. For cooling, your thermostat setting matters a good amount along with the fan speed you set.
I also evaluated a lot of things that hardly matter. For instance, once the car is on the sound system doesn’t show up at any available resolution. Windshield wipers draw about 100 W during the upstroke, which is similar to the watt draw of the power windows as well. Placing the car in park (0.26 kW) to drive (0.34 kW) shows about 800 watts of stuff associated with preparing to move or sit waiting in traffic. The energy use of headlights is low (~110 watts), but Tesla could help us with an option to have this go off when Park is engaged, rather than waiting for the driver to get out of the seat.
I made a particularly careful evaluation of fan speeds for the HVAC system using ScanMyTesla, as everyone will be making choices with these settings to yield good comfort. Removing the idle speed of the car, one finds the following in increased wattage from the fans. The good news is that choices of fan speed 6 or below for fan only have very low power. Above 7, power increases quickly and is much higher for speeds 9 and 10. And as seen in the above chart, if the AC or heat is on, the change in power from those choices is much larger.
Fan Speed → Added Watts
- 2 → 25
- 4 → 35
- 6 → 50
- 8 → 150
- 9 → 225
- 10 → 350
If you want to know such data at high resolution for yourself, you can use the same nifty aftermarket device I did, which came from e-mobility-driving-solutions.com out of Germany. It can access the OBD to the car via a Bluetooth OBDLINK to read the many logs that are produced by the car. You can read them in real time on a Samsung or Apple device using their ScanMyTesla app on Google Play. Indeed, I found this invaluable to produce the data shown in portions of this report.
That done, it is possible to gain access to much more information (full screens of numbers) than just about anyone could want. While useful for diagnostics for service technicians and über nerds, most of this is overkill.
Indeed, such detail is contrary to what I am advocating. A few vital indicators are more important to provide effectively in a user interface than a slew of numbers that distract from the important ones. Yes, battery kW is vital information for Tesla to optionally provide to drivers. If you want to understand how to preserve range in your Tesla, at a minimum, you need to know overall kW when the vehicle is not moving.
However, you likely don’t need to know twenty battery temperature numbers, shaft torque, coolant loop flow, or any one of scores of other parameters. Providing too much info affronts a key aspect of an effective user interface design: concision and parsimony.
Keep in mind that one of the key facets of an effective user interface (UI) is that it does not face you with too many numbers and trends — but only those which are most important. Indeed, one of the reasons not to have flashy displays is that the human mind can only fully pay attention to one parameter at a time and distractions are not helpful.
Also note that the data we wish for — instantaneous kW to and from the battery is readily available from the OBD link to the car.
My point in this article is that one should not have to buy and install an OBD-linked device with the Model 3 just so you can get the car power draw whether moving or not. It can be provided by a simple software update, as the data are readily available to Tesla software engineers.
How am I Doing?
Finally, at the end of this segment, I return to one of Evan’s fundamental questions: “How am I doing in my driving efficiency (and use of HVAC) relative to other Model 3 drivers?”
Not only does Tesla and ABRP collect real-world data on how many kWh per mile are being used by drivers, but very useful third party apps also provide the same information. One I have been enjoying is Stats: For Tesla Model S/X/3, as it shows exactly what Evan wanted. It indicates how you are doing in your driving efficiency compared with a frequency distribution of more than 10,000 other drivers using the same app over the same time. Here is where I am currently for the last 100 miles or so.
A driving efficiency number of 100% is 241 Wh/mi for the Model 3, and my efficiency over the last week came in at 115%. This is about 210 Wh/mile, which has gotten better as I have learned the various ways to improve efficiency that I’ve been revealing within this series. My lifetime efficiency driving the car some 14,000 miles has been 229 Wh/mi, but I’m getting better now that I’ve learned subtle nuances we’re covering.
The second plot available from Stats allows you to see how all drivers using the app have been doing over the full year. Note that at mid-summer, the average Model 3 driver is obtaining an efficiency of 94% of the rating or about 256 Wh/mi.
Clearly, driving efficiency is lower in winter than in summer for reasons we’ve covered briefly, but will dive into later. In winter 2020, Model 3 drivers were achieving a driving efficiency of about 83%, or 290 Wh/mi. Even though I am not in the cold north, I have been collecting information from many drivers who are, and am assembling this for those of you who might be interested.
Note that Stats can provide one’s driving efficiency against temperature, but mine for this summer under the pandemic is not too instructive — other than that I air condition more when it gets hot. I compare it with another for a driver in a colder climate where one can see their trends as well as the variance due to driving speed and a host of other factors. A good thing for Stats to provide would be the regression line for the average for the overall population as reflected in the sample of thousands reflected in the figure above (perhaps a quadratic fit). Nevertheless, these data do allow one to roughly see how HVAC is affecting things for an individual Model 3 driver.
But for now, we are still facing the dog days of summer, and with a pandemic. As you can also see, the average Model 3 driving efficiency has been a little lower this summer than last summer, which likely has something to do with driving habits during the COVID pandemic compared with “the normal” before.
I’ve got a lot of tricks coming up to help drivers in both summer and winter.
Yet, it is still important for me to keep pounding the point to those above.
Hey, Tesla: please give drivers optional information on battery kW and HVAC status so we can most efficiently drive and do base camp.
Danny Parker is Research Scientist at the FSEC Energy Research Center, where he has worked in the energy efficiency field for the last 30 years.* Beyond better machines, he has a keen interest in low-energy cooling technologies, zero-energy homes, rockets, and good coffee. His sister lives in Fremont, where he became familiar with Tesla. Neighbors and other pals in Cocoa Beach work for SpaceX. Tesla investor.
*Disclaimer: Note that the author’s information and opinions do not imply recommendation, endorsement, or favor of specific products or services by the University of Central Florida or FSEC, its research institute.
Have a tip for CleanTechnica? Send us an email: firstname.lastname@example.org