Underground Robotic Freight Tunnels
Pallet optimized underground tunnel logistics system outline and arguments for developing an interoperable public standard.
Last updated
Pallet optimized underground tunnel logistics system outline and arguments for developing an interoperable public standard.
Last updated
By Fletcher Hillier
Tools available for planning and estimation, follow for release
The URFT system, or Underground Robotic Freight Tunnel system, is designed to optimize the movement of pallets through sub-terranean logistics networks. With a major focus on efficiency, the URFT emphasizes standardized tunnel dimensions and transport mechanisms to ensure compatibility across different logistics operations. This system not only reduces construction costs and alleviates the dependency on proprietary technologies but also significantly increases throughput. By adhering to these standards, the URFT system aims to streamline underground freight operations, offering a scalable solution adaptable to varying logistical needs.
Interoperability maximized the utility of rail systems, and I believe it is important to apply a similar set of standards for Underground Logistics Systems (ULS).
Most logistics tunnels are too large to service all routes, too expensive to build, and highly dependent on protected intellectual properties to the point where it becomes a limitation for its own growth. A pallet optimized tunnel can achieve a throughput of over 3000 pallets per hour, per km at 80km/h. Most routes will not use anywhere near that throughput, so any extra headroom is essentially waste. If a route needs more throughput, lay more tunnels.
Max out that potential and the tunnel is earning $328,860,000/year over a 50km route, charging only $0.25/km and zero loading fees. At a projected construction cost of less than $6 million per km, the system can expand exponentially with a fraction of maximum potential. It means that the route is profitable at an incredibly low utilization. The goal is not to divert all traffic, only palletized loads.
By removing the slowest, most dangerous traffic first we target the problem pragmatically. Freight can operate in a hypoxic vacuum that is ideal for reducing degradation, while putting people into tunnels makes them inaccessible during emergencies and creates a need for life support systems among other safety requirements that are not relevant to freight.
If we assume an average travel speed of 80 kmph and an automated load transfer system that can exchange loads within a few minutes, we can achieve roughly 3 000 pallet loads per hour, per km. In one year a single machine can travel over 525 000 km, or the equivalent of about 20 000 km of full truck loads (26 pallets). The bots will push together in large convoys, allowing 500-800 transporters to push through in unison and break off without stopping the rest of the loads. That’s equal to 19-30 fully loaded transport trucks per km, per hour. If the tunnel were to maximize utilization while charging only 25 cents per km, it would generate over $6.5 million per km, per year. If more capacity is needed, speed is increased. If the capacity is not fully utilized, lower speeds can be used to lower the electrical costs.
A low cost per km is important to the growth of the road network. By simplifying all of the components and the infrastructure with interchangeable elements we can avoid bottlenecks and quickly adapt to different supply streams. A simple, low cost, modular design for all robotics platforms will keep complexity down, enabling rapid production and expansion. They don’t have to be pretty because you don’t have to look at them.
Turn waste including old laundry machines and appliances into productive assets. Used parts could be tested and implemented wherever practical to reduce costs. Redundancy and adaptability in critical components are important to the sustainability of this approach. Motor efficiency is a far less critical factor than aerodynamic pressure, and due to the fact that this system uses an incredibly low amount of energy per km it could take years to add up to the cost of a new motor. The environmental cost of creating that energy to make up for the difference in efficiency is far less than the carbon footprint of a new motor. Compared to mining the minerals, processing, transporting, and manufacturing a new motor, adaptation and operation at a lower efficiency will consume far less total energy and have a significantly lower carbon footprint overall.
Transporters in the tunnel can utilize excess energy by speeding up or operate at a lower speed to reduce power consumption. The transporters act as a kind of kinetic energy storage system with regenerative braking as well as having small on-board chemical batteries. If combined with a stable power source as well as an unstable source, this could mean better utilization of alternative energy. It can allow the system to exist within the energy headroom while creating new transmission lines protected underground.
At least one A/C conductor travels the length of the tunnel to provide a charging cable to the bots. If humans will never enter the tunnel this could be a single exposed conductor running at high voltages, otherwise it could be sealed with induction charging. The charging system depends on the access considerations and local regulations.
A modular TBM with palletized components should be used for the initial 10’ cylindrical bore. When it reaches the end of a segment, it can be packed up and brought through the tunnel for drilling in another area. Manual excavation would be used to insert the TBM at the appropriate elevation.
Smaller bots are used to remove soil material and deliver concrete down the same tunnel. They have a wheelbarrow and their job is just to move things. Water tanker bots may also be needed to transport liquids, which will use the same frame as the material movers.
Concrete ingredients are delivered by the bots to the independent concrete forming equipment at the end of the TBM. They deliver the proportions to maintain the mix ratio while helping to flatten the tunnel floor. Dry hauling allows for low complexity material hauling bots. The TBM may need to be capable of hydraulic cement injection to stabilize the soil material before the concrete forming equipment.
After the hole is dug out, the first layer of the tunnel is a rough application of aerated M20 concrete. This layer fills in the cylindrical shape to act as a buffer for ground movement. The inside layer of the tunnel is made of reinforced M40 concrete, bringing the final size down to roughly 5’x7’ with some space for the final pour of self levelling mixture.
The outer layer of aerated M20 concrete acts as a stress absorber, while the stronger inner tunnel retains a straight pathway with minimal disruption. A self levelling concrete mixture will be used to provide the smoothest pathway possible.
Many of the considerations for large transport vehicles become less relevant at a smaller scale in a tunnel with zero interference. This allows the machine to be simplified down to essentially just a frame, aerodynamic structure, motors, controllers, and some other minor components. You can get rid of useless things like airbags, headlights, seatbelts, seats, complex suspensions, inhumanly heavy components, and fancy engineer stuff with high markups.
Transport bots entering a main loading dock will pass under a conveyor system that will intercept the pallet and allow continuous movement. Smaller docks will have a hoist that will bring up the transporter with its load to avoid underground maintenance. Instead of a truck backing in and waiting to be unloaded and loaded, this interception system allows for immediate transfer without any of the delays.
The most efficient tunnel is a hypoxic environment so the parts don’t rust, and a very low air density so drag is less of a factor. The more efficient the tunnel, the more dangerous it is for people.
A tunnel is an opportunity to create the perfect environment and extend the lifespan of the system components including the tunnel itself. Under the ideal conditions, material degradation can be limited to friction and load forces. Maintenance due to dust, rust, and cracking can be almost eliminated. It allows for materials like exposed steel or PLA to potentially last for over 50 years under low stress, reducing material costs and improving the environmental footprint through extended material mileage.
Most of the materials in the tunnel prefer a temperature between 10°C to 30°C, low relative humidity, low oxygen and low Co2. When oxygen, Co2 and humidity are reduced, it extends the lifespan of the concrete, metals, and plastics significantly. The ideal environment means potentially 2 or 3 times the lifespan of these components as metal oxidation and concrete carbonation are major maintenance issues on the surface. Reducing the humidity too much, below 15% is bad for the concrete and will cause it to dehydrate and crack. The goal is to find a balance between preserving the concrete and the metals.
A temperature below 10°C metals can become more brittle, materials contract at different rates, ductility is reduced, lubricants become thicker, concrete cracks due to shrinkage and lower resistance to impact loads.
Temperatures above 30°C accelerate metal degradation, chemical reactions, and thermal expansion, leading to increased stress and deformation in metals, plastics, and concrete.
Sealed Concrete : 5°C to 30°C
Bearings & Lubricated Metal: 10°C to 40°C
Steel & Metal Components: 10°C to 30°C
Plastics: 10°C to 30°C
Copper & Electrical: 5°C to 30°C
The ideal air composition for extending the lifespan of the concrete and robotic components is < 1% oxygen and < 1% Co2. One way to create this condition is to purge the tunnel with nitrogen gas to create a hypoxic environment. This will likely also increase the relative amount of Argon gas in the tunnel because oxygen and Co2 will be displaced faster than Argon. Each of the loading stations will include low equalization vents to help purge the heavier gases. Due to the danger of hypoxia for humans in this environment, this would be implemented only in tunnels with 100% security confidence and zero human exposure risk.
The tunnel must be able to equalize with the outside air relatively quickly without collapsing in order to be safe. The initial vacuum pressure will be set at 25%, then will ramp up to 35% for the initial testing tunnel. The goal will be to sustain the pressure of an 80% reduction in air density inside the tunnel so that transport efficiency can be improved as security confidence reaches 100%. At 25-35% reduction, the oxygen levels are survivable for a person, and rapid decompression will not cause significant damage to loads or humans that may enter the tunnel for maintenance or break in due to poor security. One of the arguments for the tunnel is a reduction in transportation lethality; all parts of the underground system must achieve zero lethality potential even for law breakers. An act of sabotage leading to death will still be counted against us as a vulnerability.
This chart shows the energy cost at different speeds for each load height assuming a 0.1 average coefficient of friction in convoys. It shows that below 40 kmph there are diminishing returns and below 30 kmph the vehicle consumes more power due to constant power draw from the electronics.
40 kmph is the most cost efficient speed when onboard electronics consume a total of 100W running power. The more power that is used for communications and control, the higher this ideal speed will be.
As load height increases, drag forces require more power to overcome. This chart shows the maximum theoretical potential speed at a given load height. This assumes total available motor power is 3 kW, accounting for 85% efficiency.
It shows that just above 100 kmph is achievable for all load heights when drafting in convoys. With this data we can set the transmission maximums. It suggests a maximum speed of 100-120 would be ideal.
This chart shows that 60 inches is the maximum limit at 80 kmph and 35% vacuum with a nose cone achieving a drag coefficient of 0.3. Without a nose cone it barely achieves 80 kmph at 24” load height. As vacuum pressure is increased, these will flatten out significantly, reducing the need for aerodynamic nose cones.
It also shows that drafting at 0.1 Cd moves any load at 80 kmph and stays well under its limits. The actual average drag may be less in longer convoys.
Two sizes of nose cones may be used to reduce the coefficient of drag at higher speeds for larger loads. Calculations still need to be done to determine optimal usage of nose cones. Nose cones add weight, which reduces efficiency at low speeds significantly. Only once the load reaches a certain speed, at a certain amount of exposed frontal area, it benefits from a nose cone. Nose cones will be at the head of convoys.
Loads under 4ft may not benefit from a nose cone at normal travel speeds
Loads around 4ft tall will have a nose cone applied if the load will exceed 60kmph
Loads between 4ft and 7ft will have the 72” nose cone applied if they exceed 50 kmph
End to end chains can drastically reduce drag and allow for power sharing by pushing each other. This is an advantage of a tunnel that would otherwise cause a lot of disruption on the surface and requires coordinated control over all of the vehicles in the chain. Ideally we would have a non-stop flow of traffic without gaps. Unlike a train, each car is separate and can break off without disrupting the rest of the chain. Vehicles accelerate to fill the gap and drag is reduced as much as possible.
Linking together in a convoy end to end allows the carts to average out their power consumption and move at a faster pace while operating in an efficient range.
On the surface, trucks need to back into a docking port, trains need a lot of maneuvering just to access particular freight loads, and air shipments need to be specially packed. All of these transport methods require careful loading and waiting on full loads to optimize transport. The URFT skips all of that by automated conveyor loading onto single pallet transport vehicles. They are sent down the tunnel as a single load, without being crammed in with a bunch of other loads that move around and cause damage.
A shipping container on the surface elevated to dock height, with a loading dock on either side of the container. Loads are transferred into surface vehicles using ROVs or fully automated vehicles. A conveyor system brings up loads as needed from the tunnel. It will have a small amount of storage for loads but primarily depends on the alcove in the tunnel as a buffer. If the system is overwhelmed, loads could be shifted to the closest warehouse or branch to avoid a blockage in the line.
Loads will be transferred from the transporters onto a conveyor system, then will be stored by surface vehicles into the warehouse according to destination and ownership. Once a load is complete it can be packed into a conventional transport vehicle. These facilities also store transport vehicles to buffer machine availability and could be used as a maintenance depot for transporters. The ideal location for this facility would be at a lower elevation to reduce ramp angles to a maximum of 10°. One of these should be built at the beginning of the tunnel so it can also be used as the construction facility for the TBM and the bots. It should have some mineral processing equipment and room to store piles of excavated material. An old quarry would be ideal. The elevation could be chosen and extra materials can be dumped into the quarry for future processing.
This system offers very specific control over acceleration limits while avoiding internal pallet collisions and fall over risk during transport. Each load is an individual package with its own set of considerations which the bots can accommodate. A load of toilet paper for example could be fired down the tunnel at maximum speeds with high acceleration limits as compared to glass which might require a slow, smooth ride to stay intact. That’s something roads and conventional logistics can’t achieve due to the limitations of their physical environment and equipment.
The URFT is significantly more energy efficient per km than any on-land transportation system aside from trains. It offers the reliability of trains with the speed of trucking, without most of the inconveniences that come with those systems. Customers will choose their minimum speed which will have a major impact on cost. It can take 10x the energy to travel at 80kmph vs 40kmph depending on the drag forces, so if you can wait for it, the savings will be worth it. If you can’t wait, you’ll push everyone else faster.
If you have a critical load that needs to push through at 100kmph+, the rest of the vehicles will exit ahead of the vehicle to make way if they cant speed up. Non-critical loads are forced to wait while more important things come through. Imagine being able to say “get out of my way or move faster, my load is more important!”. Critical loads are offered at a premium to improve systems efficiency overall, improving reliability of those systems as demand increases. This non-lateral price increase helps protect the system from locking up all of the low priority loads. It also carries some of the cost of speeding up other loads, making the critical loads a potential benefit to those who were satisfied with 40 kmph but got it faster for the same price.
Pricing per km should be based on the desired speed and acceleration limits. If the load has to move slowly it should cost more because it bottlenecks the system, while demanding a high speed increases energy costs significantly to get all of the other loads out of the way. The loads that can be handled most aggressively with the most amount of time to deliver should cost the least.
Empty carts and dunnage will use a lot less power so they will share that power headroom with heavy loads ahead. Instead of a wasted return journey, dunnage loads help speed up the network while suffering minimal drag & friction losses.
It costs $0.0029/km at 85% efficiency or $0.0033/km at 65% (80 kmph, 250kg load, 0.1 cd and 35% vacuum air reduction). It has to drive 217 391 km to pay off a $100 motor or about 108 000 km to pay off a $50 motor in the cost difference. That's also assuming 80kmph, and it will often travel at 40 kmph using very little power, so the motor cost is a major factor. Using recycled or surplus motors, the overall supply of transporters will be increased and more financially stable even if they are significantly less efficient and less reliable. Redundancy, multi stream sourcing, and proper testing are important to optimizing the fleet.
Standardized testing procedures will need to be developed for maintaining standards.
The testing system for bots could be a machine that is leased to repair professionals. It would allow the bot to drive into an opening where it is subjected to an automated testing procedure.
Frame deflection
Motor efficiency
Wheel condition
Transmission tension
Electronics cooling
Tunnel testing would be carried out by a transporter fitted with testing tools. It would map out any defects for repairs.
Vibration sensors (road flatness)
Tapping device with audio sensors (void testing)
Distance sensors for linear consistency
Computer vision or AI image processing for crack detection
Aerodynamic efficiency
The system could potentially operate on private land by purchasing subterranean land rights at a given elevation. It could also use right of way passages owned by the government under contract or direct provincial ownership.
Professional criticism, programming, hardware design, promotion
For more information follow Fletcher Hillier on Linked In or X:
Fletch_h
