Interstate Rail

Adapting the
Interstate Highway System
to Rail Use

By J.H. Crawford


NOTE: Unless otherwise specified, all reference is to standard North American railroad practice. European practice is substantially different, although the underlying issues are the same.


Interstate Rail (IR) is a proposal to utilize surplus capacity on the US Interstate Highway system for high-capacity, medium-speed passenger and freight rail service. As energy constraints begin to affect the USA, traffic on the Interstates will decline and freight traffic will shift from trucks to trains. It is proposed to convert the inside lanes on US Interstate highways to rail operation. A number of problems are identified and solutions proposed. This work can only be regarded as conceptual, and important questions still require definitive answers.

Increasing Demand for Rail

Rail Will Be Needed as Fossil Fuel Runs Low

Rail systems enjoy several intrinsic advantages over all other forms of transportation:

  • Low frontal area
  • Very low rolling resistance
  • Very high capacity, allowing denser (and therefore more energy-efficient) land occupancy


The USA has now reached tacit agreement that building new road capacity is never going to solve the congestion problem on urban highways. The problem is, in fact, insoluble. What we need is a way to move more traffic over limited rights-of-way. Trains are the only really good candidate for this, although certainly much progress can be made with buses. Experience shows that people are averse to riding buses but will use good train service. Cities like New York, London, and Tokyo could not possibly function without their rail systems; there is simply no other way to supply the needed capacity in very dense urban areas.

Call for More Rail Service

It seems nearly certain that as energy limitations begin to affect transportation, we will turn to trains as the most energy-efficient means of fast transportation ever devised. The problem is where these new rail lines are going to go. The solution is to convert part of the existing Interstate highway system. However, limitations of conventional rail systems preclude their use in an IR system. Hence, IR will have to be based on entirely new designs, even if the underlying technology is substantially the same. In addition, standard rail technology suffers from many serious and intractable problems, and these problems almost all lead to greatly increased operating costs for rail systems. IR thus demands a radically new approach to the design of rail transport systems. Let us begin with a detailed examination of what is wrong with conventional rail systems.

Limitations of Conventional Rail


Grades are a serious concern in standard railroading, and best practice limits grades to no more than one foot in one hundred. By contrast, grades on the Interstate Highway system are as high as 6%. While such grades have actually been reached on standard railroads (and much higher on "cog railways"), they cause serious operating difficulties and great expense. The result is that railroad construction through any terrain that is not almost perfectly level is a highly expensive proposition because of the cuts, fills, tunnels, and bridging required, all of which impose significant ongoing maintenance expenses.


Curvature is a bane of railroading. Just as with grades, highways can tolerate much tighter curves than railways. Running speeds over all but the most gentle curves must be limited to prevent the flange of the outside wheels from climbing the rail and derailing the train. Furthermore, curved rails wear very much faster than straight rails, for two reasons. First, the side load of the flanges cut into the profile of the rail head (and also greatly accelerates the wear of the wheel profile). Second, the universal application of solid axles with wheels rigidly fixed to them causes one wheel to slip on the rail when traversing a curve, because the outside wheel travels farther (and thus makes more turns) than the inside wheel. This further increases rail and wheel wear, as well as causing the familiar awful screech.

In order to minimize these problems, the outer rail of curves is normally "superelevated," which is the raising of the outer rail, so that trains take the curve on a bank. This reduces side loadings on the rails and improves passenger comfort. Unfortunately, superelevation must be limited to no more than 6" in standard practice, because loaded "mineral trains" (those composed of hopper cars carrying coal and ores) can topple into the curve if stopped on a curve with greater superelevation than that. If superelevation were unlimited, tight curves could be "balanced," so that trains running at maximum allowed speed experienced no side force from rounding the curve.

Solid axles

In standard railroad practice, wheels are shrunk-fit onto axles, which in turn are mounted into "trucks" (usually with two wheelsets per truck), and the trucks are then mounted to swivel underneath the cars, to facilitate the negotiation of curves. The use of trucks also raises the cars about three feet off the railheads, greatly increasing wind resistance and raising the center of gravity of the cars.

Head-End Propulsion

On most railroads, trains are pulled by one or more locomotives, normally at the head of the train. (Sometimes locomotives are at the rear, sometimes at both ends, and even in the middle of the train.) All propulsion is provided by the locomotives; the cars are simply pushed or pulled. Some passenger railroads use so-called "multiple-unit" (M-U) trains, in which every car is propelled (sometimes every other car). The M-U approach has never, so far as I am aware, been applied to freight service. The very high acceleration of typical subway trains derives from the use of M-U equipment—every axle of every car is powered, so, in theory, the train can accelerate at a rate of 0.25 G, and this acceleration was nearly achieved (at slow speeds) in the PCC streetcar of the 1930s. Higher acceleration than that is, in any case, hazardous to standing passengers. Only by adopting M-U trains can good acceleration be achieved. Even high-speed trains, such as the French TGV, have sedate acceleration, because the trains are propelled only by locomotives at each end.

Limited Wheel Adhesion

It is normally reckoned that the tractive effort of a powered wheel is limited to 25% of the weight on the wheel; if more power is applied, the wheel will start to slip, causing heavy wear (and "rail burn" if the train is stationary). While claims have been made for 40% adhesion in advanced locomotives, in practice these high values are not generally achieved. Further consideration must also be given to the problem of rails made slippery by ice, water, or autumn leaves, all of which at times cause serious operational problems on conventional railroads. With electrically-propelled trains, it should be a simple matter to eliminate wheel slip entirely—the necessary computerized motor controllers were perfected 20 years ago. Unfortunately, this technology is still in only limited application.

Wheel Skid

If the engineer applies too much braking force, the wheels will lock up and skid on the rails. This also causes rapid wear of the rails, but worse, it destroys wheel roundness almost instantly. Again, electric propulsion offers a way to eliminate this problem—the same computerized motor controllers can actually achieve faster, non-skidding stops than is possible simply by locking up the brakes.

Standard Cars

Standard railroad cars are about 85 feet in length and each car has one truck at each end, for a total of four axles (sometimes two trucks each with three axles). Because the span of the car between trucks is about 60 feet, the cars must be strong trusses to keep from collapsing in the middle. The result is that cars are heavier than they need to be. Cars are also generally designed to withstand tensile forces of one million pounds, which arise from locomotive tractive forces and shock loadings. Once again, the M-U system can greatly reduce these forces and permit car chassis to be made considerably lighter.


Braking distances for long freight trains are extreme because the pressure changes in the brake control lines take a long time to propagate to the end of the train. This problem affects passenger trains less because they are shorter, and M-U trains with electrically controlled brakes have rapid braking response. Even so, braking distances for trains are much longer than for cars, because braking forces normally never exceed 0.25 G, even with the wheels locked (which rapidly destroys the roundness of the wheels). However, with computer-controlled motors, emergency braking rates as high as 0.4 G should be possible on dry track. This is still far below the 0.8 to 0.9 that can be achieved with good anti-skid braking by rubber tires on dry roads. Trams can stop very rapidly if equipped with magnetic track brakes, although this often leads to passengers being thrown to the floor. The application of track brakes to conventional railroads is probably possible but better avoided.

Collisions at Level Crossings

Even at rail crossing controlled by gates and signal lights, collisions between road vehicles and trains are all too common. The only certain way to eliminate this great hazard is to eliminate level crossings. Indeed, this is why the Interstate highways are so safe (compared to other roads); there is never any crossing traffic, so the greatest single hazard is entirely eliminated.

Signal Violations

After 150 years of railroad signaling, trains colliding with each other remains the most serious hazard in railroading. Oddly, the most common cause of these collisions is the failure of the engineer to stop for a red signal. Practically everything has been tried to prevent these operating errors. The only system that has worked is so-called Automatic Train Control (ATC), which automatically enforces signal restrictions. Engineers are automatically prevented from passing red signals. Unfortunately, ATC is still not widely applied, and this omission kills people every year. The technology is more than 60 years old, although it is not cheap. (Even from a purely economic standpoint, it is probably cheaper to pay for ATC than to pay the cost of accidents, but mot railroads cling to the idea that engineers will always stop for red signals, despite all of the evidence to the contrary; in the UK alone, more than 900 red signals were passed in a single year.)

Characteristics of the
Interstate Highway System

The US Interstate Highway System (officially, the Eisenhower System of Interstate and Defense Highways) was built to a single national standard. However, a few segments of the system that were constructed before the system was formally begun were built to slightly different standards. The deviations from standard are fairly small and relatively rare. We can thus assume that, for practical purposes, the system has the following universal characteristics:

  • Grades no higher than 6%
  • Design speed of 70-80 MPH
  • Curves no tighter than can be negotiated at 70-80 MPH in a normal, low-performance automobile of the 1950s
  • Limits on the radius of vertical curvature (this prevents abrupt transitions in the grade of the roadbed) Vertical curve radius in meters = 2 x the design speed in km/hr. So for 160 km/hr (100 MPH), vertical curve = 320 meters.
  • Curves designed to 600 m minimum radius; there are known areas where this standard is not met (e.g., Providence, RI, on I-95).
  • Lanes 12 feet wide
  • Very high standard of road surface smoothness
  • Limits on the camber of the road surface
  • Axle load limits of (I believe) 10 tons per axle
  • Bridges at least 14 feet about the road surface (design standard is 15' but 14' 4" is not unusual)
  • Paved 10-foot shoulders adjacent to the outside (slow) lane
  • Narrow paved shoulders adjacent to the inside (fast) lane
  • Long acceleration and deceleration ramps to allow safe entry and exit
  • No crossing traffic (I know of at least one rarely-used rail track that crosses an Interstate)
  • No opening bridges (I believe this is violated in a few cases)
  • The system is, in general, built to a very high standard, with rock cuts well protected from falling rock, sound bridges, good drainage, and rights-of-way that are fenced off in inhabited areas.

In short, this is a resource of unusual quality.

Interstate Rail

We have seen that conventional railroad systems suffer from many serious problems, some of which could be solved but most of which are intrinsic to the basic design. At the same time, it seems likely that we are going to replace much of the passenger and freight carriage now conducted by truck or by plane with rail systems, because of limitations on energy consumption and land occupation. The cost, if conventional rail were chosen, would be very high. Suppose, however, that we start with a blank sheet of paper and design a completely new system, Interstate Rail, based on using what will be the excess capacity of the Interstate highways system as cars are trucks make way for more efficient transport.

IR would make use of the inside lanes (in both directions) on most or all Interstate highways. These lanes run continuously and mostly do not connect to entrance or exit ramps (there are enough exceptions to this rule to be troublesome, and some interchanges would have to be reconstructed at significant cost, but the scope of this problem is limited). A dedicated rail line could thus be installed running in both directions on every Interstate highway, without the need to acquire new rights-of-way or to perform heavy civil work to make these rights-of-way suitable for use by trains, provided that a new rail system can be developed that can operate within the constraints imposed by the Interstate highway system.

Proposed Interstate Rail System


Ultimately, the entire Interstate system would be converted to IR.

Alterations to Interstate Highways
to Permit Use for IR

  • In order to reduce fuel consumption, the speed limits on the Interstates (and all other roads) should be lowered to 45 MPH, certainly not more than 55 MPH. This also reduces kinetic energy and so simplifies the problem of keeping road accidents from blocking the railroad tracks.
  • Existing 4-lane highways become double-track railroads and two lane highways (probably with passing lanes every few miles, provided by widening and improving the paving of the shoulder for a distance of about a mile.)
  • 6- and 8-lane highways give up only one lane in each direction
  • Standard New Jersey Center Dividers ("Jersey Barriers") are used to separate rail and road traffic and prevent road accidents from impinging on the rail line
  • Left-side exits and entrances will have to be removed and relocated to the right side. Since the capacity of the roadways is being reduced, many high-capacity interchanges can be replaced with standard cloverleafs, which have capacity enough to handle the reduced load.

Trains would probably not change from one Interstate to another; passengers needing to change direction would change trains at stations located at interchanges where Interstate highways cross. This condition is not essential, but it eliminates the great difficulty of building rail junctions at these intersections. It may, however, prove easy to use the existing left-side entrances and exits as rights-of-way for train junctions in those places where high-capacity interchanges have been constructed.

Rail System Design


Many elements of the track design are, of course, dependent on the design of the trains—see the train design section below.

Since the system need not interface with existing railroads, the track gauge can be arranged to whatever is convenient. The current standard gauge of 4' 8-1/2" dates from Roman times and was only adopted for railroad use because it was a reasonable compromise between stability, cost of cross ties, and ability to negotiate curves. Given that the roadbed to be used is 12’ wide (and train car bodies would be about 10' 6" wide), a track gauge as wide as 9' could reasonably be adopted. Something around 8' would probably work out well, and because superelevation need not be limited except perhaps by bridge clearances in some locations (possibly requiring reduced speeds in these areas), all curves can be "balanced" for 100 MPH operation. The very low center of gravity of the train design, and the much wider track gauge, would eliminate any problem of a stopped train capsizing into the center of the curve (although it might be quite uncomfortable for passengers on a stopped train; the system should in any case be designed so that trains hardly ever stop anywhere except stations).

Superelevation would be unrestricted within the limitation of bridge clearance and should permit operation at 100 MPH through almost any curve on the Interstate system (entrance and exit ramps excluded). With curves superelevated for balanced operation at 100 MPH, rail wear would be held to an irreducible minimum. Since the trains would have no solid axles, the wide gauge presents no problems with tight curvature—each wheel would run without slipping or sliding.

Tie-plates would be bolted to the road surface using heavy expansion bolts. The tie plates are equipped with fine-pitch adjusting screws that would allow rapid, precise positioning of the rails. These adjustors could in theory permit the gauge and line of the rails to be adjusted every night by automated maintenance cars traveling over the line. These cars would detect out-of-line and out-of-gauge conditions, connect motor drives momentarily to the adjusting screws, and rapidly correct the rail locations. This could result in a base of rails that was continuously maintained in almost perfect alignment, allowing a very smooth ride. The excellent sub-grade of the Interstate highways (well drained and frost free) should remain stable under traffic. Continuous welded rail eliminates one of the few remaining sources of rail irregularities. If switches use the movable frogs pioneered on the French TGV, then the elevation of the rail heads should be almost perfectly regular.

Wheel loading must be held to values that do not exceed the design standards for the Interstate highways, which appears to be 10 tons. Trucks impose very high instantaneous loads on the road surface when they fall off the edge of one concrete section onto the beginning of the next. Over the years, this has created the familiar bump-bump-bump rhythm of concrete Interstate highways. The actual amount of the shift is quite small and it should be possible to overcome this using the screw adjustors. The absence of this shock load on smooth track might permit the use of higher tie-plate loadings that 10 tons, a point that requires research. Furthermore, because the rail itself is a fairly stiff I-beam, rail loads are distributed over two or more tie plates, probably permitting axle loads as high as 20 tons, again a point that must be verified by research.


The existing AdTranz EuroTram and the Spanish Talgo trains come the closest of any existing equipment to the designs proposed for IR.

Trucks would not use solid axles; each truck would have four independently-suspended wheels. The wide track gauge and absence of axles would permit passenger aisles to pass the trucks without rising much above their usual level, which could be as little as 8" above the rail heads. Such low floors greatly simplify the task of providing level-loading platforms at the stations.

The system design speed would be 100 MPH for both passenger and freight trains, to allow mixed passenger and freight operation. Freight trains would operate point-to-point, without the need for passing through classification yards, eliminating this expense and delay.

Trains would be manufactured almost entirely from aluminum, to save weight. Aircraft technology would be applied where practical to hold weight to a minimum.

All freight would be containerized; the containers would be loaded into tubs in the bodies of the freight cars, which should be sufficient to retain them without any fastenings. Possibly only 20-foot containers could be accommodated, but it seems probable that 40-foot and even 45-foot containers could be accommodated. The use of tubs keeps the bottoms of the containers a foot or less above rail heads, eliminating problems with the limited 14' bridge clearance (containers are only about 10' 6' high, often less). This would even leave clearance for a medium-voltage overhead centenary wire, maybe 6000 VAC.

Rail cars would not exceed 50 feet in length, to minimize bending loads and reduce structural weight

Each rail car except the first would have only a single truck that would carry one end of the car to which it was attached; the other end of the car is articulated to the next car and its weight carried by that truck. Close attention will have to be paid to wheel and rail loadings, so that the bearing capacity of the roadbed is not exceeded.

Trucks would be arranged such that the car bodies run close to the rail heads, rising up above the trucks. This holds the frontal area of the train to an irreducible minimum.

The car articulations must accommodate both the minimum vertical curvature of the Interstate highway system as well as the minimum horizontal curve. Since the cars are short, the car-to-car articulation is keep smaller than would otherwise be the case.


Initially, trains would be powered by diesel alternators located in the first and last cars of each train. Trains would be limited to 30 cars (each about half the weight of a conventional railroad car). Assuming that each car weighs 50 tons, the power required to propel such a train at 100 MPH up a continuous 6% grade is 48,000 HP (excluding wind drag and rolling resistance, which are not important factors compared to the work required to ascend the grade). This is actually excessive and would be difficult to achieve. If speed were reduced to 50 MPH over 6% grades, power could be reduced to 24,000 HP, which is high but not ridiculous. Given that grades of even 5% are uncommon, and that 16,000 HP will maintain 100 MPH over a 2% grade and 33 MPH over 6%, we can assume that 8,000 HP alternators at each end of the train would be adequate. Even these are large by prevailing standards—very powerful single-unit locomotives are now 6000 HP, but they are, of course, normally used in combinations of several locomotives. It might be possible to use a 4000 HP conventional diesel alternator for standard power generation, supplemented by a lightweight (and relatively inefficient) 4000 HP standby diesel-fuelled gas turbine alternator for those brief occasions when more power was required.

Later, the generator cars could be replaced with overhead power (which could be beefed up on steep grades to supply enough power to allow the each of the traction motors to run briefly at about 250 HP and propel the train at 65 MPH up a long 6% grade (motors can typically be run at 200% of rated power for short periods of time; 10 minutes at 65 MPH and 6% grade gives a rise of 3500 feet, which is surely higher than is encountered anywhere on the Interstate system; it should be no problem to obtain motors that have a 200% rating for 10 minutes). If advanced diesel engines are used with low-sulfur fuel, and provided that the injectors are diligently maintained, exhaust ought to be quite clean. The problem of PAH particulates needs attention, but techniques are being developed to remove PAH from diesel exhaust. These engines could, of course, be run on bio-diesel, if a large enough supply can eventually be developed.

An alternative to overhead power is hydrogen-fuel cells, although the feasibility of this is still questionable. It could give the clean operation of overhead electrical power together with the very high output required to power trains up steep grades at high speed.

Both axles on each car would be powered at 260 HP, with each wheel having its own 130 HP motor ("traction motor") which, taken together, would absorb the full 16,000 HP generated by the head-end and tail-end power units. Each motor would be individually controlled by its own computer to prevent slip and slide.


A more advanced system than the century-old track circuit should be developed. While highly reliable (if not absolutely perfectly so), the track circuit required currents of hundreds of amperes to flow over each one-mile signal block. Even though the voltage is just a few volts, the current is high enough to require very thick cables to connect the signal relays with the rails. This approach also requires insulated rail joints every mile, which are an appreciable maintenance issue and which also cause a rough rail junction every 35 seconds or so (at 100 MPH). I believe that a combination of inertial navigation systems, GPS, ground reference, and active satellites could provide dual-redundant signaling that was at least as reliable as tried-and-true track-circuit signaling, and at a fraction of the cost. This requires investigation, of course, and if it proves unfeasible, conventional signaling techniques would certainly work. Whatever system might be adopted, only cab signals would be installed. At high speeds, track signals are difficult to read, and the only safe approach is cab signaling, which is now in widespread use. Whatever system is adopted should not permit an engineer to operate a train in violation of a signal. This one measure eliminates about half of all possible wrecks.


Stations would be built at existing overpasses, with the boarding platforms extending in one direction from the bridge and completely occupying the median area between the two tracks. (This makes it easy to connect rail stations to city centers using conventional street-running trams, which can be constructed quickly and cheaply, as they were in large numbers in the late 19th and early 20th centuries.)


Junctions would exist wherever Interstate highways intersect. In most cases, the trains would cross and continue along their routes. Major stations would, of course, be constructed at these locations, as there would be a considerable amount of connecting traffic. It may be possible to use existing interchanges to switch trains from one Interstate to another, but this would mostly be unnecessary.

Potential Problems

It is not clear that the standards for vertical curvature on the Interstate highways are sufficiently stringent that trains could be operated over these vertical curves at 100 MPH. However, my experience is that the Interstates have quite gentle vertical curvature and that this would pose no difficulties. It would require early consideration, however.

The 14' vertical clearance is a significant limitation, but I believe that the system could be designed around this limitation.

Axle loadings may be the most serious limitation, and it might be that only 20 ton containers could be transported over the road, with all heavier containers being shipping by standard railroads. This would be a significant limitation that might even make it necessary to revert to the trucks-at-each-end design now used on freight railcars, and this would almost surely permit the carriage of 40-ton containers. (The Interstate highways see intense use by trucks weighing 40 tons.)

As with any highway or railroad, snow clearance can be a significant issue. In territory where heavy snowfall occurs most winters, snow-clearing trains would have to be equipped with powerful snow blowers ("rotary plows") that would throw the snow far off the rail- and roadway. This technology has existed for more than a century, and such plows keep roads and railroads across the Sierra Nevada open except during the most intense winter storms, when the roadway may close briefly.


We foresee the development of a new rail technology that could be cheaply and quickly installed on existing Interstate rights-of-way, in rapid response to suddenly worsening oil supply conditions. The principal construction material required is the roughly 215 tons of rail needed for each mile of track constructed plus comparatively small amounts of steel for tie plates and bolts to fasten the tie plates to the road surface. By far the largest amount of the work has already been done - the construction of the Interstate highways. IR merely enables their continued utility under energy-limited conditions. Many technical questions remain to be answered or at least verified, but there do not appear to be any fundamental technical barriers.

Copyright © 2001 J. Crawford. This page may be freely reproduced provided that J.H. Crawford is acknowledged as the author, his copyright is maintained, and the full text is reproduced.

By publication of this work, J. Crawford places in the public domain any original ideas contained in the work.

Return to Energy, Transport, Urban Planning & Sustainable Development
Return to
Visit J.H. Crawford's
Send e-mail