This is the second of a four-part series.
Banana 7, PARD, and the Mystery of Evan’s Missing Range
Even when the Tesla Model 3 begins to fully drive itself, it will still have humans on board. And it is best not to forget that.
The Mercury spacecraft that John Glenn rode around the earth three times in 1962 was one of the first advanced, high-technology machines designed to operate fully automatically. Since no one knew if an astronaut-passenger could survive long in space (the Russians weren’t talking, so Glenn carried an emergency morphine syringe), the entire machine was designed to operate autonomously. It could fly, orient using the space-earth horizon, send detailed telemetry and receive automated data from the ground. It could also hold its position ready for retro fire to come out of orbit at a moment’s notice.
There was no AI, but there were seven miles of wire in the capsule and a host of wildly sophisticated systems for the time. Max Faget and Paul Purser were the guiding wizards at the National Advisory Committee for Aeronautics (NACA), which became NASA. Faget, in particular, designed the Mercury capsule to be a forgiving craft that would survive malfunctions. He and the other “NACAnuts” toiled over a super-secret project called PARD — the Pilotless Aircraft Research Division. How do you design rockets and ramjets carrying nuclear warheads to fly automatically to targets thousands of miles away with nary a pilot on board? Faget and Purser learned to lean on passive systems that worked right when things went wrong. The more moving parts and active systems, the more problems. Gyroscopes were one weak link, as well as the need to be able to locate the earth horizon in space. For Project Mercury, another complicated area was providing a pressurized cabin that you could keep at a reasonable temperature and humidity for a human being in the vacuum of space.
“Full Self Driving” (FSD) in a Tesla was PARD’s ASCS in the Mercury capsule: ASCS = Automatic Stabilization and Control System. Theoretically, you could control the entire game from the ground. Small peroxide thrusters would control the capsule’s attitude and have it ready for an abort at any time. Computers on board in 1961 were still way off.
As somewhat comically shown in The Right Stuff, the astronaut was considered something of an afterthought — even a limitation — but the obsession of the original seven astronauts in getting a window into their craft was real enough. The engineers looked at all of this as distractions. Indeed, two chimpanzees flew before Alan Shepard and John Glenn, Ham and Enos. Both survived fully automated rides in the Mercury capsule. Like Shepard, Ham was only in space for 15 minutes — scarcely enough time for things to go wrong.
However, the experience of Enos on November 29, 1961, on a three-orbit mission might have been a cautionary tale about total dependence on automatic systems, at least for HVAC (Heating, Ventilation, Air Conditioning). On the morning of the launch, President John F. Kennedy held a short press conference. “This chimpanzee who is flying in space took off at 10:08 AM,” he announced. “He reports that everything is perfect and working well.” If Kennedy’s quip was amusing to the reporters, the truth was not so rosy.
It wasn’t that the automatic systems and ASCS in the Enos’ Mercury capsule (unofficially called Banana 7) didn’t work. They did — at least at first — but the HVAC system then gave trouble. Enos had been trained furiously for 1263 hours before the mission in dizzying motion trainers, the torturous centrifuge at 8 Gs, and strapped into a couch at various temperatures for hours each day. Prompted by panel lights, he was to pull a lever to demonstrate control and be rewarded with banana pellets or else be shocked. This the Chimponaut did dutifully and without hesitation, in spite of being buzzed even if he pulled the lever immediately. He was wrongly shocked 76 times! Yet, the mission was abruptly aborted after the second orbit of the earth, not for chimp abuse, but when the HVAC system was no longer able to keep the temperature in the cabin around Enos at levels that were hotter than home, Cameroon in West Africa. The automatic control ASCS system started malfunctioning too, burning fuel at an increasing rate.
Meanwhile, Enos was overheating. Project engineers called for an abort. Retro-rocketed out of orbit, Enos splashed down near Bermuda and was recovered, eager to get out of the capsule, his suit, and the slew of biomedical sensor attached and inserted into his body. To be sure, the abusive training that the chimpanzees endured during Project Mercury was inhumane in the extreme. Years later, Mercury Astronaut Scott Carpenter and Bob Crippen, the first Space Shuttle pilot, sensitive to the sacrifices of the apes nabbed from West Africa, thoughtfully toured a chimpanzee sanctuary. “I have great respect for the species that did it first,” Carpenter said. Appropriately enough, he and Crippen were showered with spouted water anytime they came close.
After Enos, when John Glenn flew three months later, it was to see how things worked for a human hominid in orbit. There was the issue of the bad warning light for the landing bag, having to do a fully manual reentry, and keeping the retro-pack on down the flaming gauntlet to the Atlantic. Yet, there were two other serious problems in Glenn’s flight overshadowed by the fatal prospect of a loose heat shield. On the first orbit, the automatic flying system — the ASCS — began to malfunction and use twice the fuel Glenn experienced doing the same thing manually. Even worse, his suit temperature became so warm that he struggled with the HVAC controls over the Indian Ocean to keep from floating in sweat. Unlike Enos, he was able to do some knob-twisting to keep it in acceptable range even though cabin humidity went up.
After the flight, Glenn suggested looking into the HVAC controls. He was also steamed that Flight Director Chris Kraft had kept him in the dark relative to the landing bag problem. “Information about my spacecraft is vital,” he complained in his post flight report. It was a poorly kept secret that Glenn’s language to his peers was worse; he was really pissed that crucial information was held back. Tell us the condition of our craft.
That Glenn flew and survived, yet had to do a large portions of his mission manually and in poor comfort, seemed to have been lost on the Project Mercury engineers. As well as the fact that both Glenn and Enos had trouble with their HVAC system. Glenn’s trouble with the automatic ASCS flight orientation system was not fully resolved.
No, the second orbital flight three months later would aim to do one scientific experiment after another in space while the automatic systems did their part to run the mission smoothly. Victim of a mild heart murmur, Deke Slayton was taken off the crew for Mercury Atlas 7. Meanwhile, Glenn’s backup pilot, Scott Carpenter, the adventurous and spirited Navy man among the original seven, was assigned. Carpenter intended to do science, evaluate the machine, and experience space. He would test out a new souped-up manual control system, but hoped to let Aurora 7 do the flying for the scientific observations.
So, here we are again going from Project Mercury to the need for Pilot Mode in the Model 3. And the segue is? Full Self Driving (FSD) as a fully autonomous platform is a near-term goal for Tesla cars. Although I have no doubt it will be achieved — and perhaps soon — we will still have human passengers who have unique needs beyond the machine’s ability to take care of itself.
Not providing effective control for the HVAC system, nor letting the occupants know what is going on or the implications of control decisions, is an oversight. There is also the other issue — if there are problems with the automated systems, are you providing the driver/passenger with the sufficient information needed to make the best decision? How will the passenger feel about the machine? Inspire confidence?
Or how will the machine feel about us? As we draw ever closer to I Robot, can we foster a good relationship?
It’s Not Just Me
To illustrate the gravity of the need for improved energy feedback and the importance of dealing with pesky HVAC energy in the Tesla Model 3, let’s cut to the experience of a friend, Evan Mills. My point in sharing is to show that this need for real-time information on the energy use of the car, whether moving or not, is not just my problem. Others are flummoxed too.
Evan contacted me about his frustration trying to achieve the nameplate range in his new Model 3. Readers should know that Evan Mills is not your average run-of the-mill (pun warning) fellow, or even an average-enthusiasm EV owner. Evan is a retired senior scientist in the energy division at Lawrence Berkeley National Laboratory. A longtime friend, he has studied the nuances of electricity use and energy efficiency in buildings, equipment, and appliances over a long and storied professional career.
Our story begins with Evan happily driving, propelled by electrons, from Mendocino to Berkeley in his new gray Model 3 (18” aero wheels, AWD, long-range battery) — with appropriate California vanity tags saying “Drop Gas.” But the “happily” part ended with a dismaying revelation: he was routinely seeing a third of his range disappear while driving 150 miles. He was needing 200 miles of battery range to go that far. After a year of help from Tesla support, he contacted me, as he knew I was driving a Model 3 as well.
He may have been dismayed to hear me announce that I could routinely get the promised range, but that it must be harder for him because of heating. What heating? “I live in Mendocino and drive to Berkeley. Both are mild climates.” Besides, he told me, he kept the heating off. Really? I was incredulous.
Evan was able to see the remaining range in the car when driving and repeatedly watched as a portion of the starting miles just evaporated over the course of a ride. And he wasn’t driving with a heavy foot. How was that possible? Was something wrong with the car? I had news for Evan. Based on tests we were doing at FSEC in which we were measuring the air conditioning and heat demand of the Model 3, its heating and cooling systems could take up to 25% of the available battery power under some driving conditions — particularly under extreme cold. At freeway speeds, the losses were much lower, I said.
Evan agreed. He knew from research done at his lab decades earlier that HVAC could make a nasty dent in an EV’s range, and he thought he was keeping it on a short leash.
Also, although he thought he had heating off, unless one is very careful, you end up heating without even knowing it. This was not, I told Evan, because Tesla Model 3 is inherently inefficient, but because the electric energy available from the batteries remains modest. While the non-conventional propulsion system of the car is incredibly efficient, the heating and cooling systems are conventional. At the same time, all of the energy used by the car for everything has to come from the batteries.
Therefore, every little bit — every hundred watts of unnecessary onboard electricity — is important.
However, at speeds around town, the problem is much worse, since the cabin conditioning losses are time-related, such that the heating, ventilation, and air conditioning (HVAC) losses are actually much greater, relative to losses from driving, when you’re going slowly. To emphasize the point, if you drive zero miles but sit in the car for three hours with the HVAC on, 0% of your “miles lost” while doing so will be from driving, and most of the “miles lost” will be from using the HVAC.
Evan described his experiences further to me. On closer inspection, he figured out that the heating system was quietly turning itself on when he periodically went into the menu simply to confirm that it was off. Such irony! On many occasions, he would turn the fan on, often with the AUTO setting on, assuming that it was just using outdoor air, unaware that with the temperature set to 65°F the heating was quietly being triggered.
He did some tests. “I can get 170–180 Wh/mile around town, with the heating off,” Evan reported, “but with the AC set to AUTO, I’ll see 270 Wh/mile.”
That was no surprise, I said. He was likely heating without knowing it — a fact made clear by the none-too-clear nature of the Model 3 HVAC controls. And he wasn’t alone.
Besides, I told him that the heating system for the Model 3 was a standard electric resistance heater. That’s the same as the heat source for the glowing coils in your kitchen toaster.
Evan couldn’t believe it. Resistance heat had been abandoned in buildings long ago. “Are you sure?” Doesn’t the Nissan Leaf have a heat pump? I was certain, I told him. I was also sure Tesla had its reasons. Heat pumps don’t work well in extremely cold conditions.
My measurements showed the cabin heater could draw at least 4000 watts on startup with full fans going, as happens with AUTO. It then backs off to about 1200 watts when the desired cabin temperature is reached. “1,200 W seems like a lot, even if resistance,” he said. I reminded him that an automobile cabin is not a well-insulated enclosure like the energy efficient buildings he was used to studying. Look at that glass roof.
In the meantime, Evan had already contacted Tesla in Berkeley, California, to be introduced to an extremely helpful and knowledgeable Tesla service technician.
“Could HVAC (in a mild climate) really be ‘eating’ a full third of my battery?” Evan wrote in an email to Tesla. “Do the Model 3’s HVAC systems have mediocre efficiencies? Is there a significant ‘base load’ that eats power when it’s on but there is little/no demand for conditioned air? ”
The HVAC system in the Model 3 was conventional, Evan learned, just as I had told him, and the heat source was electric resistance, not a heat pump.
Evan checked in to let me know he had verified my claims with Tesla. I mentioned that we had measured the energy use of the Model 3 air conditioner on low speed — it was approximately 1400 watts, but could be more than 2000 watts on startup with AUTO on and a high fan speed. Trouble was, when you were driving, you couldn’t readily see that since the power draw of the air conditioner was added together with the power use of the drivetrain. In a bit of unplanned obfuscation, a driver could not readily see how it was increasing your Wh/mile and dropping your expected range. It was a pretty much hidden effect, since it was hard to see the influences all lumped together while moving.
While that was interesting, Evan told me, he wasn’t using air conditioning. He was driving his Model 3 in a chilly autumn in Northern California — ’50s and ’60s. The sore point: electric resistance heaters potentially used a lot more electricity than the air conditioner. Plus, based on what we had seen in tests with defog activated, even when heating, one could sometimes end up with the air conditioning on at the same time as the heating and not even know it. Moreover, the Tesla displays gave no indication that any of this was happening.
The driver can easily have no idea they just chopped down their range when they get those windows clear and then leave defog on for the next half hour needlessly. They also have no idea the heat is still on in the case that A/C is off but the interior temperature is set to 65°F.
Also, the AUTO setting, as Tesla later pointed out, tries to optimize both relative humidity and temperature inside with a good amount of A/C compressor use that can be avoided. When heating, it is heating along with air conditioning when the interior’s relative humidity is elevated. When cooling, it is sometimes air conditioning with resistance re-heat (as it is referred to in the air conditioning industry).
The unknowing Tesla owner, just looking to be warm, can end up running the air conditioner and the heaters at the same time.
Understanding the Mystery of the Lost Range
Heating and air conditioning are a big deal in an EV.
While the Model 3’s electric motors are extremely efficient in comparison to their internal-combustion-engine counterparts, the HVAC system is not. (Note that this has changed to some degree in the Model Y, which now has a heat pump.)
This emerges from two factors. First, relative to a gasoline-powered car, the amount of energy carried in the Model 3’s batteries is dramatically lower than the energy stored in a conventional gasoline car in which a 10 gallon gas tank holds an energy content of about 330 kWh.
By way of comparison, the Model 3 stores about 75 kWh in its battery pack of 4,416 individual 2170 lithium-ion cells. That is only 23% of the energy of our comparative gasoline-powered car. Another way to think of it is that the batteries in the Model 3 hold about as much energy as about 2.3 gallons of gasoline. The only reason that the Model 3 can compete with conventional automobiles regarding performance is that an internal combustion engine only converts about 10–25% of the gasoline energy content into useful running energy. By way of comparison, the Tesla motor and drivetrain is extremely efficient. Its permanent magnet synchronous motor converts 90% of the battery energy into motive power — an engineering feat. If we know how efficient the car is at moving itself without any parasitic loads, we have a baseline to understand how losses and other non-propulsive loads can influence range.
So, just how efficient is the Model 3 when driving without the heating and cooling system activated?
There is little official data on this question, but Elon Musk and JB Straubel themselves published a piece on the Model S in May 2012 with exactly the kind of data one needs to examine for this question.
Alas, we could not find the same presentation for the Model 3. So I set about creating it.
To get started, the website Teslike.com has published data on the efficiency without heating or cooling of the various Tesla models. (It also has data for the Model 3 LR with 19” wheels, which is the type of car I am driving.) This website leverages EPA dynamometer testing of Tesla cars, showing range between 55 and 80 mph without HVAC. Data are located here along with details of the methodology used for the estimates:
Teslike publishes ranges for the various models at various freeway speeds, but these data can be approximately translated into Wh/mi and kWh given the speed and assumption of a 75 kWh battery. But since all of these estimates are derived, how do we know if they are reasonably accurate?
Because we can check on them with our own machine.
My son, Wade, and I used my Model 3 LR RWD here in Central Florida on long stretches of flat road to evaluate the efficiency at various speeds between 10 and 75 mph. At the time of the tests, the car had 13,050 miles on the odometer. The stock tires were inflated to 42 psi and the car was warmed up before each test, which was made at temperatures between 78°F and 87°F. Tests were only made under very low wind conditions and without traffic and rain, and all during daytime hours. The only heating, ventilating, and air conditioning (HVAC) electricity use was to have the fan set to “3” with outside air (yes, we suffered — it was quite uncomfortable for many of the tests).
A total of 5–8 tests were made at each speed, with repetition typically done in each direction to eliminate the possibility that small changes in road grade or wind were affecting results. (The Model 3 five-mile energy graph was vital for this process, as one can see the power demand smooth out during test).
The car was driven several miles until the Wh/mi values became stable and the trip counter was then reset for that run. The plot below shows the averages from our collected data. The Teslike-sourced data is superimposed over what we collected. The green circles are the actual recorded averaged data points. The yellow triangles are the Teslike data for the freeway speeds they show. It was satisfying to see how similar the data were over similar ranges.
Secondly, based on our collected data we composed a statistical model for the measured performance of the Model 3 without HVAC use. The statistical model we used to fit the data is well known in the automotive engineering industry for evaluating vehicle motive, aerodynamic, and rolling efficiency in a compact form:
Energy (kW) = A + B (Velocity) + C (Velocity)3
A = energy use regardless of speed
B= energy use increase that is linear with speed (rolling resistance)
C= energy use that increases with the cube of velocity (air resistance)
The Wh/mi were translated to kW by multiplying by speed/1000. These values were then regressed, with the results shown below:
Note that the regression was nearly perfect, with an R-squared of 0.999 and with both engineering-related terms highly significant. The 0.095 kW per mph can be loosely interpreted as the vehicle rolling resistance and the mph3 coefficient 0.0000281 as the drag coefficient — area product for the automobile.
The intercept term is also very important. It suggests that the vehicle draws approximately 380 watts regardless of speed. I caution that the values measured for the vehicle moving at only 10 mph were not well captured by the regression and the test results had higher variances than the other measurements. (These tests, so slowly rolling, were also very frustrating to do).
This regression result was then translated back into Wh/mi, which is then superimposed as a dashed blue line:
kW = 0.376 + 0.095 (mph) + 0.0000281 (mph3)
Would it have been better to do all the tests on a track? Certainly, and it would have been much faster, but as seen from our tests, they are very consistent not only with the Teslike data, but also with the earlier published data on the Model S from Tesla. Moreover, there are plenty of YouTube videos, like this one from Tom Moloughney, showing that the Wh/mi in our test is closely in agreement with other results. In his test, the Wh/mi at 70 mph was 234, against the approximately 238 Wh/mi we saw at the same speed with a RWD vehicle, but without the Aero wheel caps. How does that matter in context?
As previously described, with a gasoline engine, only a fraction of the energy moves the car, while 80–90% of the energy in the burned fuel is lost as heat. Within that inefficiency resides the second challenge for EVs. In winter, they are is at a disadvantage relative to a gasoline-powered car. There, some of massive waste heat from the gasoline engine can be used to warm the car’s cabin in winter, with little or no impact on engine winter efficiency. On the other hand, with the EV, every watt-hour must come from the battery pack, which takes away from power for propulsion.
In summer heat, both the EV and gasoline-powered cars have to run a vapor compression air conditioner. However, the energy required to operate the cooling system is a small fraction of the energy in the gasoline-powered car. Each hour that the air conditioner runs in a Tesla Model 3 drops range by about 6–12 miles depending on fan speed and how hard the compressor is pushed.
So, it is not so much that the Tesla heating and cooling systems are inefficient as it is that they have a large impact on total energy consumption relative to how much energy the car can store in its batteries. For heat, there is no heat pump — at least not yet. The Model 3 uses a resistive heater in conjunction with harvesting waste heat generated by the battery, drive units, and even some of the computers in the car to heat the cabin. Waste heat mode does a fair amount to keep the cabin warm for short periods of time, but over long distances or in colder temperatures, the heat generated by waste heat is not enough. The resistive heating element runs at up to 6,700 watts at full draw (there are two 3,350 watt heaters; one on either side), but the combination typically idles near the 1,200 watt zone depending on the temperature.
At the same time, A/C usage can be overlooked. Whenever the defroster or defog modes are activated, the climate control is set to automatic, the air conditioning will also turn on in order to reduce the humidity of the air in the car. At low speed, the A/C compressor will also use about 1,000 watts of power on average. The fans use 10 to 350 watts on top of that, depending on the setting. The A/C compressor itself is variable speed and can be highly loaded on startup. At fan speeds above 6, you typically engage the higher capacities. Power use with fans at maximum (or AUTO setting on startup) will typically be more than 2 kW or even 3 to 4 kW for a short time — a big hit.
Below, we show the measured stationary electric power for the air conditioner at 93°F outdoors using the ScanMyTesla app with a low temperature setting in cooling mode. (More about ScanMyTesla in future segments.) The power use with everything off was steady at 0.24 to 0.26 kW.
This indicates that at a moderate fan speed with A/C on, the HVAC power is only about 1.0–1.5 kW, but at high fan speed and maximum cooling, this increases to over 4 kW.
The HVAC controls are mysterious in their own ways, which we will discuss further in the next segment. For instance, when you turn defog off after using it, the air conditioner will not be turned off — you’ll need to remember to do that manually.
For heating, certainly a heat pump would be more efficient at lower heating levels than a resistive heater, but a resistance heater seems desirable in very cold climates. (Bolt and Leaf drivers in cold climates may sometimes wish they had resistive heaters on the coldest days.) At least with the Model 3, Tesla initially decided from a manufacturing-value and cold-weather-performance standpoint that not using a heat pump outweighed efficiency considerations. Still, with the heat pump now in the Model Y, we can see that Tesla is addressing the most serious engineering oversight in the Model 3.
Now, if they would just allow us to examine how our power use is affecting our consumption from the battery whether moving or not. Even if the car drives itself, it will not tell you how your decisions sitting in your mobile office will affect how far you can go when you leave. The solution?
Allow us to see battery power and HVAC status at all times.