Self-Driving Trucks: Will They Disrupt Railroads

[Broeb22]

Does anyone think that self-driving trucks, if they ever get here, would seriously challenge the current ground transportation supply chain? Could self-driving trucks eventually gain a cost advantage on rail?

 

 

[Cardboard]

Rail will always have the advantage in terms of cost per mile due to physics.

 

Cardboard

[LC]

Rail will always have the advantage in terms of cost per mile due to physics.

 

Cardboard

 

Pretty much. Trucks will continue to handle excess demand, short-distance logistics, etc., but the bulk will still be transported by rail for reasons Cardboard mention.

[Grant]

Could self-driving trucks eventually gain a cost advantage on rail?

 

I'm not familiar with the costs in these industries, but they cannot beat rail at energy efficiency. Platooning only results in about a 20% reduction in drag (http://www.extremetech.com/extreme/241442-mit-study-driverless-truck-platoons-will-save-fuel-money-especially-tailgate). Rail is more efficient mostly because of the coefficient of rolling resistance of the best pneumatic tires on tarmac is about 2.3x that of rail.

 

A fully-loaded semi uses about half its power overcoming drag, and about half overcoming rolling resistance.

[rb]

There was a thread on this last year about self driving trucks doing convoys. The conclusion: it's not even close.

[WayWardCloud]

I hear you guys on the vehicle energy efficiency comparison but I'm also thinking if you use trains you still need a truck to carry merchandise from the factory to the train and then another one at arrival from the train-yard to the destination. I imagine there are additional costs involved in moving cargo on and off trains and trucks. If those are big enough could it make sense to skip the whole "train" step even at higher cost per mile driven just to save the hassle of loading and unloading?

 

it's a question I've been asking myself in regards to Berkshire's investments in railroads.

[Grant]

Trains, trucks and ships can use standardized shipping containers so loading and unloading isn't necessary. Though yes I'm sure there's still some switching costs involved, making rail more attractive for longer trips.

[StubbleJumper]

Could self-driving trucks eventually gain a cost advantage on rail?

 

I'm not familiar with the costs in these industries, but they cannot beat rail at energy efficiency. Platooning only results in about a 20% reduction in drag (http://www.extremetech.com/extreme/241442-mit-study-driverless-truck-platoons-will-save-fuel-money-especially-tailgate). Rail is more efficient mostly because of the coefficient of rolling resistance of the best pneumatic tires on tarmac is about 2.3x that of rail.

 

A fully-loaded semi uses about half its power overcoming drag, and about half overcoming rolling resistance.

 

 

Interesting.  So, if you accept the proposition that the speed of a driverless truck could be dropped from 65 mph (110 kmh) to 50 mph (80 kmh) without really affecting delivery times, you would save what, 20% in fuel use from the lower speed?  And then in a convoy you are saying that there would be another savings from reducing drag?

 

That 4-to-1 fuel advantage that WEB referenced in years past might drop to 2.5-to-1 or something like that?

 

Interesting.

 

 

SJ

[Hielko]

Trains, trucks and ships can use standardized shipping containers so loading and unloading isn't necessary. Though yes I'm sure there's still some switching costs involved, making rail more attractive for longer trips.

You still have to move the containers from the trucks to trains and/or vice versa. That requires manpower, time and machinery. Besides that, adding a train step might add considerable delay to delivery times as well. It's not like cargo trains run every 10 minutes to every destination, and a lot of destinations might also add a bit of a detour. So adding the train step will only make sense if the two destinations are sufficiently far apart.

[gfp]

It will be interesting to see how the use of public infrastructure / implied government subsidy of the trucking industry shakes out vs. the owned and privately built and maintained rail infrastructure.  Obviously there are road use and diesel taxes on trucking currently, but it is nowhere near the level it will likely be in the future when states and governments are even more indebted and rail lobbyists are pointing to these driverless convoys clogging up the public roadways, blocking on-ramps for a thousand feet at a time, etc...

 

That said, I certainly feel better about owning the railroad out in the big wide open west, vs an eastern-only railroad.  As mentioned, long distances help and there are some pretty long distances to travel from western ports, mines, etc...

 

Rail will never have a place delivering your Wayfair furniture to your town, but it will likely be delivering the container of Wayfair furniture from the port of long beach to the distribution centers for a long time.

[SharperDingaan]

Rail will always have the advantage in terms of cost per mile due to physics.

 

Cardboard

 

Pretty much. Trucks will continue to handle excess demand, short-distance logistics, etc., but the bulk will still be transported by rail for reasons Cardboard mention.

 

Self-driving trucks will disrupt the trucking industry, but it's not going to displace rail.

 

In the supply-chain use, finding/retaining long-haul drivers has always been a problem. Most would expect that the operating cost of a self-drive will only be marginally lower than for a truck+driver (dont cut cost if you dont have to); but it will go a long way to alleviating the driver problem.

 

In the mning use, driver cost is the issue (200K+ to drive a 300 ton Heavy Haul truck). It's boring work, you need a lot of them, and its prone to injuries through inattention. Driverless is both a real cost-reducer and safety-enhancer. There are also advantages to deep hard-rock mining where temperatures are high & cooling systems are at/past practical limits.

 

SD

[oddballstocks]

The closer and closer you get trucks to operate like trains the more they'll need dedicated lanes and roads.  And then suddenly trucks aren't efficient anymore.

 

One of the biggest advantage for a truck is they operate on public roads.  They spend some in gas taxes, but that's it.

 

There are two other factors:

1) The rolling efficiency.  Stop by a rail museum sometime and examine a train wheel on a track.  For how large the wheel is the contact area is about the size of a dime.  And it's steel on steel.  It doesn't take much to roll.  Now compare that to the giant rubber tires on a truck, that travel on cement or asphalt roads full of bumps.

2) Trains are already employing diesel-electric engines, they have been for decades.  The engine is a diesel powerplant (essentially a large generator) that powers electric motors that turn the wheels.  The driving is done with electricity.  This is much more efficient than the trucks we have.  Beyond this railroads are investigating how to employ hybrid engines to further reduce energy costs.  And not that long ago much of the railroad was electrified. 

 

Consider that a typical train has 100-140 cars.  Imagine a truck convoy that long, it might stretch 1.5 miles unbroken.  How do you exit the highway, pass, do anything? 

 

[Dustin T]

I would almost expect them to work together more in the future. Shipping containers are pretty standard and at the point that trucks are self driving I expect the exchange of the shipping container to the truck could also be done without an operator. This would shift more of the long haul trucking to train. I think the logistics make this a little more complicated now with the need to schedule around people. I believe the future favors rail.

[StubbleJumper]

The closer and closer you get trucks to operate like trains the more they'll need dedicated lanes and roads.  And then suddenly trucks aren't efficient anymore.

 

One of the biggest advantage for a truck is they operate on public roads.  They spend some in gas taxes, but that's it.

 

There are two other factors:

1) The rolling efficiency.  Stop by a rail museum sometime and examine a train wheel on a track.  For how large the wheel is the contact area is about the size of a dime.  And it's steel on steel.  It doesn't take much to roll.  Now compare that to the giant rubber tires on a truck, that travel on cement or asphalt roads full of bumps.

2) Trains are already employing diesel-electric engines, they have been for decades.  The engine is a diesel powerplant (essentially a large generator) that powers electric motors that turn the wheels.  The driving is done with electricity.  This is much more efficient than the trucks we have.  Beyond this railroads are investigating how to employ hybrid engines to further reduce energy costs.  And not that long ago much of the railroad was electrified. 

 

Consider that a typical train has 100-140 cars.  Imagine a truck convoy that long, it might stretch 1.5 miles unbroken.  How do you exit the highway, pass, do anything?

 

 

A couple of observations:

 

1) You wouldn't need a convoy of 100 trucks because you could just as easily have 15 convoys of 7 trucks.

 

2) If trucks ever get automated, there is an excellent opportunity to *decrease* their individual weight because there's no meaningful labour advantage to having one truck instead of two.  That might actually reduce the damage that heavy trucks are currently doing to the road.  Bullshitting just a little bit more, charging a per-mile toll for trucks becomes relatively straightforward if they are fully computerized and that might even give states the opportunity to put in place time-of-day road pricing (ie, if you want to send your autonomous truck down the turnpike at 8am, it'll cost you 40 cents per mile, but if you do it at 1am, it'll be 1 cent per mile).

 

 

SJ

[PullTheTrigger]

1) You wouldn't need a convoy of 100 trucks because you could just as easily have 15 convoys of 7 trucks.

 

Yes, the term the industry is using is "truck platooning." A small number of vehicles "linked" together with autonomous systems. There is a lead vehicle that controls all the following vehicles. Still lots of testing to do. It would likely just be a small number of vehicles (less than 10), but I'm sure they'll figure out the optimal number based on the roads, etc.

 

https://www.fhwa.dot.gov/research/truck_platooning/

 

 

 

 

[Cardboard]

"There are two other factors:

1) The rolling efficiency.  Stop by a rail museum sometime and examine a train wheel on a track.  For how large the wheel is the contact area is about the size of a dime.  And it's steel on steel.  It doesn't take much to roll.  Now compare that to the giant rubber tires on a truck, that travel on cement or asphalt roads full of bumps.

2) Trains are already employing diesel-electric engines, they have been for decades.  The engine is a diesel powerplant (essentially a large generator) that powers electric motors that turn the wheels.  The driving is done with electricity.  This is much more efficient than the trucks we have.  Beyond this railroads are investigating how to employ hybrid engines to further reduce energy costs.  And not that long ago much of the railroad was electrified."

 

Very true. Plus, a train has a few engines and a very large number of wagons (50:1?).

 

They have tried 3 trailers behind a truck and it was banned due to becoming highly unstable: wind, turns, etc. Whether it drives on its own or not, this issue does not go away.

 

Cardboard

[oddballstocks]

In PA and OH they allow trucks to carry three trailers.  I see this all the time on the turnpike.

 

The issue with trucks is their cargo capacity is so much smaller than a train car.  Look at this: http://business.tenntom.org/why-use-the-waterway/shipping-comparisons/

 

A truck can carry 26 tons, a train car 100 tons, so four trucks to a train.  Or where a train gets an effective 200 mpg.

 

The advantage of a truck is always short haul, and that they can navigate cities quicker.  If you need to move things in bulk, and need to move them fast you put them on a train.

 

I guess where I'm struggling on this is why is an automated chain of 20 trailers better than a train of 100 train cars (eq of 400 trucks) operated by two people? On a train the labor is so tiny compared to the cargo.  On current trucks I get it, you have a single driver for a single trailer.  But to out edge a train seems like a crazy high hurdle to get over, and that's if trains never improve.  I mean truck companies send their trucks on trains if it's over a certain distance.

[SharperDingaan]

Trains are stuck to rails, and go only from point A to point B. Trucks are not, and it's why we have inter-modal transport.

You build a regional distribution centre next to a rail line, and transport containers to/from port via rail. Unload the containers, unload the contents, & distibute to warehouse &/or sales locations by truck. Where possible automate the warehousing to maximize reliability and minimize cost. The 'last mile' truck isn't driverless, as cost exceeds benefit.

 

The main cost driver for self-drive trucks is JIT inventory requirements. Pay more for the greater control and reliability of trucks, and offset the cost against lower working capital costs. The more limited contents of 3-4 trucks 2-hours late because of border delays will not shut down a production line, the much bigger quantities in a delayed train will. Take the truck drivers out, to further improve control and reliability.

 

The big benefit is greater safety on the highways.

Fewer fatigued long-haul drivers and more fresher local drivers, with rail and self-drive doing more of the long-haul.   

 

SD

 

 

 

 

[Spekulatius]

I think we will have self driving trains before self driving trucks, since trains have already a dedicated lane, by definition. It should be much easier to implement and in fact there are already several systems operating.

[StubbleJumper]

I think we will have self driving trains before self driving trucks, since trains have already a dedicated lane, by definition. It should be much easier to implement and in fact there are already several systems operating.

 

 

 

I don't doubt that at all. But, what will be the potential savings?  A typical train might move 10,000 tonnes of freight and it requires how much labour to operate?  Are there three guys in a train these days?  So, if you eliminated those three guys, you'd save perhaps $100/hour each, or $300/hour total?  So, you might save $300 for every 600,000 tonne-miles?  Those savings don't strike me as a game changer.

 

Eliminating human error from railway operations, however, is probably a good thing.

 

 

SJ

[Spekulatius]

I think we will have self driving trains before self driving trucks, since trains have already a dedicated lane, by definition. It should be much easier to implement and in fact there are already several systems operating.

 

 

 

I don't doubt that at all. But, what will be the potential savings?  A typical train might move 10,000 tonnes of freight and it requires how much labour to operate?  Are there three guys in a train these days?  So, if you eliminated those three guys, you'd save perhaps $100/hour each, or $300/hour total?  So, you might save $300 for every 600,000 tonne-miles?  Those savings don't strike me as a game changer.

.

 

Eliminating human error from railway operations, however, is probably a good thing.

SJ

 

 

 

 

I think improving the efficiency and reducing errors will be the largest benefit. Unions will probably be trying to block progress, but the same is going to happen with respect to autopilot trucks

[Cigarbutt]

The historical evolution of railroad employment shows a long term trend that started well before the 1980 deregulation Staggers Act.

 

https://www.railserve.com/employment.html

 

Freight railways have continued to decrease manpower but it is hard to see big savings going forward at that level.

 

Since the Staggers Act, freight railway has produced a widening and staggering growth in productivity in comparison to air, trucks and private business in general but this growth has also shown a tendency to become flatter because the low hanging fruits are gone. From the input side, there are still potential promising areas (increased locomotive efficiency, innovations in car design and capacity, longer train lengths, improvements in operating practices, and technological innovations in train control). From the input side, I would submit that the moat that freight railways have for long-haul transportation will continue to widen versus trucks because the roads are essentially public infrastructures that need to be shared. IMO, driverless truck technology and convoys will not compensate this growing disadvantage.

 

Would like to add though that the major reason for increased productivity and improved operating ratios (BTW most of the gains have been harvested by shippers (and consumers?) although that may be changing? {railway pricing power versus potential re-regulation}) has come not from rationalizations and decreased employment but from increased economies of densities (increasing output in ton-milles per network) and the two most important factors behind this phenomenon have been 1-western coal and 2-intermodal shipments related to containerization and increased global trade. Interesting to note that increased energy prices would tend to magnify the moat between freight trains and trucks (even driverless) but economies of densities can go both ways.

 

[John Hjorth]

Aren't there already self driving trains around the world? The Copenhagen metro, partly inaugurated in 2014 - still not fully completed - but since then at least connecting Malmö [sweden], Copenhagen Airport and Copenhagen Central Rail Station is running by use of driverless trains. It feels like you're some kind of residue slag on a conveyor belt, while the reality most likely is more that you're in a drone train.

[Cigarbutt]

Aren't there already self driving trains around the world? The Copenhagen metro, partly inaugurated in 2014 - still not fully completed - but since then at least connecting Malmö [sweden], Copenhagen Airport and Copenhagen Central Rail Station is running by use of driverless trains. It feels like you're some kind of residue slag on a conveyor belt, while the reality most likely is more that you're in a drone train.

 

The technology adoption is spreading. Closer to home (Montreal), there will be a major upgrade to provide a complete driverless light metro system (to be completed in 2021) led by Alstom. When you think of it, commuting metro/trains in urban centers are conceptually similar to elevators. And now, elevator operators are a thing of the past.

http://www.elevatordesigninfo.com/why-elevator-operators-went-extinct

 

Freight trains may be next on the list but security issues will be raised. It is possible that resistance (union or otherwise) will be met. Anyways, as mentioned before, I think that potential cost savings would be minimal but human errors due to fatigue (real problem) could be eliminated.

 

Apparently planes would also be a realizable target. Would you board on a plane without a pilot?

I have a feeling that my grandkids may find it funny when I will tell them that there was actually a pilot sitting in the plane.

The pace of change is accelerating. Buckle up!

 

[VAL9000]

This debate reminds me of packet switching vs circuit switching.

 

I think that eventually packet switching (self driving trucks) will win in overall cost vs circuit switching (trains).  But the economic model of self driving trucks will be very different from today's trucking model.