Slipstreaming EV's - A viable pathway to extending range via OTA?

[JBee]

There was some discussion on various threads about using slipstreaming or drafting to extend EV range. This is when vehicles follow each other closely to minimize aerodynamic drag and therefore reduces energy consumption. I thought it would be interesting to discuss this in more detail, and maybe envisage ways to make it work, after considering the risks involved.

The benefit for EV's range improvement varies primarily because of:

• how large the vehicle in front is in comparison to the one following
• how close the distance is between vehicles
• how aerodynamic each vehicle is in the convoy

As can be seen in the diagram above the effect actually assists all vehicles in the convey, including the front one. The reason for this is that the first vehicle produces a slight vacuum behind it, which reduces air resistance and entrains air for each following vehicle, until the last vehicle in the convey that experiences a slightly higher air resistance as the airflow reconverges behind it and creates turbulence. On a single vehicle, this rear reconvergent turbulent airstream also creates drag, and that is why the first vehicle in a convoy also experiences a drag reduction by that amount.

This technique is also often used in sports, from Nascar, F1 to cycling like Tour de France, and even skateboarding etc. Similarly sailing uses wind shadows to slow other competitors. It's important to remember that the effect is aerodynamic, and as such the aerodynamic shape and size impact how well it performs. The figures I'm using are not accurate to all Tesla models but would fit with a CT given that it is not as aerodynamic as the other models.

In the scenario to improve EV range, there is a possibility to increase CT range by up to 40% depending on how close the cars travel together. This depends on how close the vehicles are, with closer being better, but reasonable results up to 10% at even 30m(100ft).

The intension here is to use similar if not identical Tesla vehicles to operate in the convoy. The reason to limit the initial rollout of this to a Tesla is because the integrated FSD hardware can provide substantial benefits and safety margins whilst operating in this configuration, and the slipstreaming mode could technically be incorporated in a OTA update. This mode would only be available on fairly straight main roads and highways, possibly only mutilane roads, with the right driving conditions, based on weather and traffic conditions.

Let me address the main risk of this setup, and the one that will likely be most commonly perceived as a problem.
That is braking distance versus following distance.

Now braking distance on a Tesla is in the low 30m (100ft) range, depending on speed, and at that distance a 10% range improvement could be realized. Ideally, we would reduce that distance further, but this means that the following vehicle(s) will not be able to stop in time if the vehicle in front comes to a abrupt stop. I say abrupt stop intentionally, because any controlled braking maneuver could be matched, within 10-20%, by any following vehicle. An abrupt stop would only occur should the lead vehicle hit something stationary, otherwise an emergency braking maneuver by the front vehicle could be matched by any following vehicle, within a distance safety margin.

The problem with normal human driving is the control loop time between seeing the vehicle in front braking, assessing the rate of deceleration, then modulating how much to brake as to avoid collision. In the case of the convoy it is important that the FSD and the braking of the first vehicle instantly triggers the following vehicles to brake as well, effectively closing the control loop time to ms instead of seconds. That way all vehicles act as one, essentially becoming vitual trailers of the vehicle in front, and the front vehicle FSD can relay commands down the line according to conditions, to either increase of decrease the follow distance.

The question is how do we get the front Tesla to communicate this with the following vehicles without adding any hardware. We could go down the Bluetooth, that is range limited and won't reach to the end of the convoy. Mobile cellular network is to unstable and high latency and dependent cellular towers. That leaves the vehicles wifi connection. This also has limited range, but could be used in conjunction with a mesh stack that can repeat the signal from one to the next down the convoy at low latency, and reasonable reliability. At some point a dedicated telemetry radio signal might be useful.

After thinking about a solution for this for a while I realized the solution might actually be simpler than that in that Tesla vehicles already have independent light control via it's Canbus lighting architecture. Technically this means that by switching the rear lights on and off in pulses, the front vehicle can convey braking and distance information to the following vehicles. Ideally this could be done by controlling the PWM cycling of LED drivers meaning that the coding would be imperceivable to the human eye, but easily picked up and translated by the FSD cameras and at low latency. The encoding could be by multiple lights simultaneously, or at least the three braking lights. The update to read this would be embedded into the FSD.

Following on from this, I think Tesla should possibly consider this for non-convoy driving, even manual driving as well, and to possibly extend on this by using situational awareness of the road ahead to influence other cars driving behind them. This could for example be used to combat traffic compression waves, assist in merging lanes, or to allow EMG vehicle passage and the like.

What other ideas or comments do people have regarding using CT in a convoy or automated traffic management capacity?

Sponsored

 

[MiguelAznar]

Technically this means that by switching the rear lights on and off in pulses, the front vehicle can convey braking and distance information to the following vehicles. Ideally this could be done by controlling the PWM cycling of LED drivers meaning that the coding would be imperceivable to the human eye, but easily picked up and translated by the FSD cameras and at low latency.

That’s brilliant, so simple and so effective!

 

[fhteagle]

If this isn't patented yet, you should definitely do so. Very simple and possibly quite effective method for inter car communication

 

[rr6013]

EV's range improvement varies…

1. Wind shadow varies 4:1 thus 12’ ht. Semi has an aerodynamic shadow that extends 48’ behind it. A vehicle within that 48’ draft zone will benefit.
2. Steady state speed max’s efficiency
3. Gaps kill efficiency
4. Mixed vehicles impair velocity synchronization
5. Convoys greater than six semi lengths impose traffic congestion
6. Convoy comms protocol include, speed, velocity delta(rate of change), Brake scrub, Braking delta, acceleration(+/-), grade%, grade(+/-), EBS
7. computational convoy data per vehicle Speed jitter, acceleration anticipation, Braking distance micro.sec., Velocity frequency modulation smoothing map, DraftZone%, Effective WHr/mi, RANGE gain.

This works in fleet, fleets pulling triples, fleets running point-to-point and Interstate regular route carriers in the United States.
The frequency bandwidth vehicle-to-vehicle set aside by FCC IoT was relinquished for reuse with the latest revision.
A self-network aligning router for in-vehicle will be an R&D hardware
A convoy WiFi router will be a self-select a quorum of six participants.
NHTSA will want to taxilight vehicles in-convoy to alert public when a vehicle is “in-convoy”.

The number of qualifying fleets in qualified routes able to benefit from a Convoy mode amount to corner cases you could count on one hand. That said, potential fleet subscription to a TeslaConvoy feature could expand rapidly were it quantifiable and available. Fleets such as SAFEWAY, WALMART, etc…

Ladtly, TeslaConvoy roapmaps directly to the future DOD autonomous supply convoy technology already mapped and calendared.

 

[JBee]

1. Wind shadow varies 4:1 thus 12’ ht. Semi has an aerodynamic shadow that extends 48’ behind it. A vehicle within that 48’ draft zone will benefit.
2. Steady state speed max’s efficiency
3. Gaps kill efficiency
4. Mixed vehicles impair velocity synchronization
5. Convoys greater than six semi lengths impose traffic congestion
6. Convoy comms protocol include, speed, velocity delta(rate of change), Brake scrub, Braking delta, acceleration(+/-), grade%, grade(+/-), EBS
7. computational convoy data per vehicle Speed jitter, acceleration anticipation, Braking distance micro.sec., Velocity frequency modulation smoothing map, DraftZone%, Effective WHr/mi, RANGE gain.

This works in fleet, fleets pulling triples, fleets running point-to-point and Interstate regular route carriers in the United States.
The frequency bandwidth vehicle-to-vehicle set aside by FCC IoT was relinquished for reuse with the latest revision.
A self-network aligning router for in-vehicle will be an R&D hardware
A convoy WiFi router will be a self-select a quorum of six participants.
NHTSA will want to taxilight vehicles in-convoy to alert public when a vehicle is “in-convoy”.

The number of qualifying fleets in qualified routes able to benefit from a Convoy mode amount to corner cases you could count on one hand. That said, potential fleet subscription to a TeslaConvoy feature could expand rapidly were it quantifiable and available. Fleets such as SAFEWAY, WALMART, etc…

Ladtly, TeslaConvoy roapmaps directly to the future DOD autonomous supply convoy technology already mapped and calendared.

1. Wind shadow distance is critical for slipstreaming effectiveness. The intention of the slipstreaming mode would be to reduce the the distance required to operate safely, and vary the safety margins according to conditions. Having more dedicated, or Tesla Semi optimised lead vehicles would also be beneficial, provided they don't hinder traffic flow velocities.
2. Maybe this should also be a conditional variable depending on conditions, like traffic congestion etc. Ideally it would be biased to efficiency, at the cost of time in country driving, but then more biased towards managing traffic flow in congestion.
3. As per 1. it won't be a fixed value and be as good as conditions allow.
4. Do you mean because of the mass of the vehicle affects braking distance, or because of airflow? How close they are would have to include a vehicle state, which each following vehicle would self estimate to create the appropriate distance to the vehicle it is following. The following vehicle should be capable of switching between front and following vehicle seamlessly, depending on traffic, simply by gradually reducing the following distance. There need not be a central control, apart for the instant feedback of braking information from the front, with a braking rate. Accordingly, the effectiveness of the following distance will be determined by the actual vehicle you follow, and merging traffic would be allowed as a priority, effectively making that vehicle the front.
5. If the above method works to allow other traffic to merge into, or through a slipstream lane, then length has not impact on traffic. The idea here is that traffic will only merge into a small vehicle gap if they really need too to exit or overtake, but not necessarily to cruise in that lane. If that is the case the convoy would need to overtake and restructure. FSD can identify other compatible vehicles on the road, and try to convoy with them without "asking permission" provided that vehicle is providing the necessary stopping information. It would be a fluid handover between front vehicles, not a static convoy.
6. FSD already has predictive object tracking that knows other vehicle speeds and trajectories, and already uses this to merge and turn at intersections. The follow vehicle already know all of these metrics of the car(s) in front of it that are visible. Grade can be determined by map data. Road surface conditions could result in a safety margin value being sent back, if it can't be determined by other means.

The only "necessary" and "real time" data I can see is the rate of deceleration (braking) needs to be communicated throughout the convoy as soon as possible so that all vehicles in the convoy slow at the same time and the same rate as the front vehicle, which in turn doesn't compound the braking time. It acts like a "trailer convoy". In saying that, maybe the solution is to have the signal being shared be the "following distance safety margin" of the vehicle in front. The front vehicle could continuously be sending this out via the rear driving lights and any compatible vehicle following could translate that margin to accommodate it's own requirements. Imagine a rubber band that you want to keep at a certain tension. This would also be helpful for traffic flow modulation even without slipstreaming, by allowing more controlled and smoother braking by Teslas in traffic, resulting in less compression wave buildups and congestion.

7. As above I think there needs to be some segregation between necessary and desired functionality. The low latency necessary part could be done with car lights, being continuously through daytime running lights, and supplemented by braking lights and even turn signals. But the more data intensive components, for sharing non-critical data like driving performance or efficiency can be moved to a different comms path. My understanding is that Tesla wifi hardware is imbedded into the onboard Linux computer, that is very easily changed to a host mode, or even just a UDP broadcast mode in linux itself. That means that the Teslas simply broadcast their relevant slipstreaming data to anyone that wants to hear it, and it's up to the others to then either use or discard it. For example: It could just transmit it's location, heading and velocity vectors along with state of charge, consumption, grade, vehicle state etc and a follow vehicle can use this internally to compute the desired affect. It would also reduce the amount of network requests and be more resilient to connection problems. We use this mode often for high power telemetry and video from UAV's.

Taxi lighting, or signaling that it's in convoy is another thing that could be done with the car lights, by for example letting the center brake light pulse slowly slightly so a person merging through the convoy is more aware. I don't think it's necessary for the front, seeing there won't be any interactions with the front vehicle that is different.

 

[Dids]

News traffic report says avoid I95 @845, there are 3 - 10 mile long convoys coming through.

 

[Dids]

Could use LTE direct (ProSe) for communication. Believe protocol is already built into all LTE modems and has been since 3ggp version 12 and can handle 1000 devices at ~500m range .
https://www.3glteinfo.com/lte-direct/

 

[JBee]

Could use LTE direct (ProSe) for communication. Believe protocol is already built into all LTE modems and has been since 3ggp version 12 and can handle 1000 devices at ~500m range .
https://www.3glteinfo.com/lte-direct/

That also looks like a viable option for setting up comms between vehicles. I haven't seen that variation before. The range is pretty limited for LTE, do you know if it will work at the same time as connecting to a tower to get data as well? There's obviously also the cellular telemetry path, but there's no way to guarantee package delivery, let alone latency through a tower or network.

The question I suppose is how fast it connects/reconnects, and how dependable it is. Obviously 1-2 seconds is too long at those distances and speeds (60mph = 88ft/s).

This sort of throws up a more general question and that is what is the necessary range for secondary comms in this scenario? Is it enough to have LOS (line of sight) or does it need to go further, if it's not the primary comms? I suppose I'd have to map it out a bit better first.

 

[SwampNut]

Apparently nobody is aware of V2V and V2I communications. Volvo already does V2V which can proactively slow down cars far behind the one car that sees a problem.

https://www.verizonconnect.com/resources/article/connected-vehicle-technology-v2v-v2i-v2x/

 

[JBee]

Apparently nobody is aware of V2V and V2I communications. Volvo already does V2V which can proactively slow down cars far behind the one car that sees a problem.

https://www.verizonconnect.com/resources/article/connected-vehicle-technology-v2v-v2i-v2x/

Thanks for the link it was informative. They also had a link to a truck platooning paper. These comms links are actually very similar to what I described above using Tesla wifi.

But interestingly the platooning is predominantly with human lead drivers, not robotic like FSD. Technically its of course also possible to do this without FSD, and that was sort of the aim, to provide a seamless pathway between driver, FSD and in and out of slipstreaming whilst driving down the road, by using existing Tesla hardware.

The other thing they do is pass along steering and light data through the comms. Technically FSD already has that from their object predictions and doesn't really need it, seeing that each vehicle is running FSD and the tesla convoy is not relying on the front driver for control.

 

[SwampNut]

Vehicle comms were approved on 5.9GHz as a 75MHz DSRC band. It is by design very limited in range and fairly directional. My best friend has deployed a bunch of it around Phoenix. And UofA has had test areas with it for probably 10-15 years. So it was of course designed as a way to help human drivers, but obviously we can simply feed this to the onboard computer and let it make decisions. It can easily be combines with vision AI to further refine what the vehicle sees and the live data from the vehicle it can see.

Platooning is something DOTs have been talking about forever. I worked in traffic management software for a few years and went to a lot of different DOTs and industry shows.

 

[JBee]

Vehicle comms were approved on 5.9GHz as a 75MHz DSRC band. It is by design very limited in range and fairly directional. My best friend has deployed a bunch of it around Phoenix. And UofA has had test areas with it for probably 10-15 years. So it was of course designed as a way to help human drivers, but obviously we can simply feed this to the onboard computer and let it make decisions. It can easily be combines with vision AI to further refine what the vehicle sees and the live data from the vehicle it can see.

Platooning is something DOTs have been talking about forever. I worked in traffic management software for a few years and went to a lot of different DOTs and industry shows.

I definitely didn't mean to imply in any way that I came up with the idea of drafting, I think Nascar was doing it in the 60's before I was even born.

The question is if now can be the time it can go mainstream, and if so how, given the state of technology esp. Tesla and CT.

 

[SwampNut]

Sorry, didn't mean to argue, but to back up the concepts you posted. It's coming. As usual, the greatest stumbling block is humans not "trusting" it. Or just plain old fear.

 

[rr6013]

5. If the above method works to allow other traffic to merge into, or through a slipstream lane, then length has not impact on traffic.

Primarily, the hazard that a Convoy >6 Semi trailers in length creates is in the case of faster, overtaking vehicles. The extended length train of vehicles in-Convoy moving as a block of traffic forces faster traffic to a single-lane. The hazard, different from a coned and marker safety lane, is the moving element and blocked vision. Vision is blocked to the right. And blocked vision makes the end of the convoy indeterminate.
The moving element makes the passing single-file lane an indeterminate obstruction.

Out in the West on I-80 in the U.S. it is legal to pull triples(tractor+3 trailers) the free articulation effectively a steering axel that the wheeled-5th wheel between trailers is intimidating as its always seeking CE. With three trailers dancing around behind a tractor it looks like organized chaos. To add to the chaos. Triples are LTL freight haulers so weight is always light causing the trailers to bounce down the freeway. The roadway over time on the concrete resembles wagon tracks in the R.H. slow lane. So given the trailers dancing around, pounding the road the hazard of cutting-in between a convoy is heightened by the wagon wheel ruts forcing your steering dynamics.

 

[rr6013]

4. Do you mean because of the mass of the vehicle affects braking distance, or because of airflow?

BOTH
Mass. Primarily delta V. Because a tractor trailer has more braking ability than pickup+trailer, braking isn’t the critical concern. Varying mass to power ratios in-Convoy affect acceleration specifically on change of grade.

As power units within convoy seek their power balance to grade changes the adjustments in Speed to maintain convoy are necessary or a Convoy simply can no longer maintain contact(i.e. stay together)

To maintain convoy the “whip effect” compounds the impacts upon the trailing units. That’s where the protocol packet data is most useful. It can use those data to project each trailing power unit’s degree of effort to keep in-Convoy or command it to break off.

Human truck drivers are good anticipating grade, their power and modulate throttle to keep in-contact as a convoy takes an increase in grade. Drivers know a speed-up before the grade change is maximizing an engine torque sweet spot and spreading the “gap” to afford trailing units adequate reaction time to maintain the convoy up the grade.

Air-flow.
Mixed form factors in-Convoy affect wind shadows. Every pickup that breaks into middle of a tractor trailer convoy essentially begins a sub-optimal new convoy in the chain. His little wind shadow sucks for the 12’ ht. tractor trailer behind the pickup. His acceleration is quicker and creates a gap challenge to maintain in contact.

The Convoy effect is lost. The effect comes from the aerodynamics of the six units in a chain, As the convoy begins to feel the benefit of drafting the lead unit benefits from the entirety of the convoy acting enmasse thru the wind actually reducing the lead unit effort to maintain and even increase speed – free. That’s the drafting effect pelotons use to catch a breakaway bicycle group.

Sponsored