City Transport
Fixing Transport

Dutch double-decker trains can carry lots of passengers

The Four Unwritten Laws

Four assumptions are made almost universally regarding public transport, and all four are wrong. These assumptions must be examined and discarded if public transport is to compete successfully with the car.

1. Charge Fares

While some systems operate with as little as 20% of revenues coming from the fare box, virtually all systems charge a fare.

In the Third World, public transport is profit-making. Almost everybody uses it. Although it consumes a considerable amount of personal income (though still far less than a private automobile), public transport in these areas is demand-driven and works well up to the limits of road capacity and until automobile usage increases to the point of choking the network.

Bangkok is a terrible example of what happens to a large, developing city lacking a metro when private auto usage begins to increase. The bus network functioned reasonably well until around 1980, when the number of private automobiles began to increase rapidly. Bangkok is now a throat-burning polluted nightmare where traffic stands still during rush-hour. A small metro has finally been constructed.

In the West, I think most systems should abandon fares entirely. Urban automobile usage should be taxed to the level required to support good public transport without a fare. The usage fees imposed on automobiles would be much less than the farebox revenue that would be lost, as the cost of collecting fares is high due to loading delays, larger fleet requirements, higher operating personnel costs, and the costs of vending and checking tickets. Better service would attract more users, further reducing the per-rider subsidy.

2. Passengers Stand

Automobilists are assured of one thing: there will always be a seat. Public transport passengers are likely to have to stand under jerky and crowded conditions during rush-hour and often at other times as well.

Public transport should provide a seat at all times. The use of higher-capacity vehicles makes this possible without increased labor costs. Bus routes with enough traffic to require the use of high-capacity articulated buses are ripe for metro construction (or at least the installation of a tram system). A heavy-rail metro can carry virtually unlimited traffic. Trains can run as frequently as every 60 seconds. Trains as long as 12 cars are common, and even longer trains are possible. Double-decker trains (such as the RER system in Paris) can further increase capacity on systems with large-bore tunnels. It should be possible to virtually guarantee a seat.

3. Driving Is Faster

In almost every city, over almost every route, driving is faster than taking public transport. In a few cities with good public transport, this is not always the case. San Francisco's BART is one example, at least during rush-hour. New York subways are often faster than taxis. Buses, however, must compete with private automobiles for right-of-way, so the automobile is faster, since buses stop to load and unload. Only if sufficient priority were given to buses could they become as fast as cars. Cities must adopt strategies to make public transport faster than driving, certainly during rush hour. The necessary measures will further reduce average speeds of single-occupancy vehicle (unless the budget for construction of new high-occupancy vehicle lanes is very large), but public transport ridership will soar.

4. It Will Be Unpleasant

Taking the bus or metro is not usually a pleasant experience. It needs to become more like an afternoon aboard a lake steamer on a beautiful summer's day. The crown jewel of Russia is underground: the Moscow subway, renowned for its chandeliers and art. A real commitment is required to make public transport a positive experience.

Improving Speed

A bus struggling cross-town in Manhattan can often be overtaken by a fit pedestrian. If public transport is to succeed, the transport system needs to achieve at least the average speed of a bicyclist and preferably better. This is actually a fairly difficult design goal - in many urban areas, bicycling is the fastest door-to-door transportation mode. There are several essential prerequisites.

No Traffic Delays

Unimpeded public transport service can only be assured by a dedicated, grade-separated right-of-way. Two grade-separated rights-of-way are in any case nearly essential for an efficient transport system in any large city. This is even true of Los Angeles, where the freeway network is separated from the network of streets by elevation or depression. Venice, Italy, is a nice example - the streets form the pedestrian network and the canals serve the passenger and freight boat network. They operate independently, without interference. Manhattan has its streets and its subways. Grade separation allows greater capacity in a given area, because two grade-separated networks encounter far less cross-traffic than a single, at-grade network.

A single, at-grade network could perform quite well in cities up to about 500,000 if all personal transport were on foot, by bicycle, or aboard trams. With dedicated tramways free of competing traffic, quite good speeds can be achieved. Freight could still be moved by truck over streets of normal width. With larger cities, however, the need for two grade-separated networks becomes nearly unavoidable.

Compact Network

The shorter the distance, the faster the journey. In compact, dense cities the city center is never far away.

Few Stops

Stops bring the average speed down quickly. The system needs to have the minimum possible number of stops while keeping all destinations within easy walking distance of a station. I regard ten minutes as the absolute outer limit of "easy walking distance." Five minutes is much better.

Few Transfers

Transfers cost a tremendous amount of time. Route systems requiring more than one transfer must be avoided.

Short Headways

Waiting for the bus or train is often the largest single component in a public transport journey, and one that does not arise when traveling by car. If the service is frequent enough, one need not even bother to consult a schedule. This also relieves some of the problem with transfers.

High Acceleration

Vehicle acceleration should be as high as possible within the limit of what is safe and comfortable for riders. This limit is well below the traction limits of steel wheels on steel rails. In addition, it has been shown that it is jerk, not acceleration, that is the real limiting factor in passenger comfort and safety. With precise computer control of motors it is possible to achieve quick, smooth starts and stops. If jerk is low then relatively high acceleration is quite acceptable, leading to a large improvement in average speeds.

PCC streetcars (developed in the 1930s) could accelerate smoothly to 45 MPH (73 km/hr) in 10 seconds (an average acceleration of 6.6 feet/sec/sec or 0.2 G). This is close to the upper limit for a vehicle with standing passengers.

Short Dwell

Dwell time (the duration of a station stop) must be kept short. Large articulated buses with front-door-only boarding and fare payment sometimes experience dwell times of several minutes when a large number of passengers board. The steep steps found on most buses are also a serious impediment. Some new bus designs incorporate low floors. This improves dwell times and also makes life much easier for the infirm, those with baby strollers, and anyone carrying packages.

With metros, dwell can be kept short by providing level-loading platforms and plenty of doors. It should be possible to achieve an average dwell time of around 10-12 seconds, much better than the usual 15-30 seconds. At busy stops, dual platforms allow passengers to exit from one side of the train while others board from the other side. This arrangement is already in use in a few locations, and leads to greatly reduced congestion at doors.

Door Holding

It is the most natural and neighborly thing in the world - hold the door for a few seconds for a passenger running to catch a train. Unfortunately, it wastes the time of 1000 other passengers and breaks the schedule. With controlled platform access, it is possible to close access to the boarding platform a few seconds before the train departs.

Early Door Opening

This is still occasionally seen on older equipment - the motorman can enable door opening at a slow speed once the vehicle is committed to a stop. The doors are fully open when the train stops. Used correctly, this is an entirely safe technique and saves about two seconds per stop.


Paradoxically, this has become a more serious problem in the last 30 to 40 years. One might think that a door that failed to close properly once in a million operations would be sufficiently reliable, but this is not the case. Such a failure rate leads to system delays about once a week, which is unacceptable. Reliability figures at least ten times better than this are required. With the engineering advances of in the last century, it should not be difficult to reverse the decline in reliability of the last few decades.


Occasional delays cannot be entirely prevented, if only because medical emergencies are inevitable. If a train must wait while a passenger is evacuated, service is going to stop. With the topology proposed elsewhere, it would be possible to catch a train going in the opposite direction and arrive at one's destination about 10 minutes late. (The closed-loop design provides an alternative route.) Should both tracks be blocked, systems should be designed so that trains can shuttle back and forth among several stations (running in both directions on both tracks). Capacity remains nearly the same, but due to the need for numerous transfers, travel time will increase. Such outages should be rare.


Newark, New Jersey, used to have a bus line known by its riders as the "banana line" because the buses always came in bunches. This was a very heavy line with frequent service. Something would delay one bus just a minute or two with the result that there would be abnormally many passengers at the next halt, who would take longer than usual to board, thus further delaying the bus. Delays would snowball and sometimes as many as six buses would arrive after a long wait.

The solution is to keep adequate capacity on the route and to prevent vehicles from being delayed in the first place. On a metro with turnstile control, it would be possible to hold back passengers from the platform in the case of a late train with a heavy load, thus keeping the schedule from further deteriorating. The passengers not served by the first train would board another train shortly after the first departs.

Reducing Metro Operating Costs

Operating costs for metro services have sky-rocketed in recent decades. This is in part due to the cost of maintaining good service on a system with declining ridership and in part due to a failure to take advantage of known methods of cost control. Some of the techniques proposed below can be applied to existing systems; others can only be built into new systems.

Track Maintenance

Track maintenance costs can be reduced by several techniques:

  • Adopt Modern Construction. The use of continuous welded rail on concrete ties and the provision of good drainage can reduce maintenance costs. Maintenance-related outages are less frequent, the track does not deteriorate so quickly, and a smoother ride is provided.

  • Prevent Wheel slip. Wheel slip occurs whenever the motorman applies too much power and the wheels start to slip (overspeed). Computer control can eliminate this problem, which causes premature rail wear.

  • Minimize Curvature. Curved track wears far faster than straight, and the sharper the curve the worse the problem.

  • Reduce Ton-mileage. Track wear is directly correlated to traffic and weight. Reducing vehicle weights helps.

Energy Efficiency

Most modern rapid transit vehicles are surprisingly wasteful of energy. Most electric rail vehicles use dynamic braking - the motors are used to slow the train, with the energy dissipated to heating grids atop the cars (which leads to even hotter tunnels in the summer). This saves brake wear but does not recover any of the energy used to accelerate the train, which, in metro service with frequent stops, is the largest component of energy usage. Full regenerative braking can recover most of the energy otherwise lost to braking. In addition, most vehicles are heavier than need be. New designs and advanced materials should permit significant weight reductions.

Wheel Truing

In all rail operations, wheel sets are routinely removed from the vehicle and turned on a lathe to restore roundness and the correct profile. This is a very expensive operation, and after wheels have been turned several times they must be discarded or reworked. There are several causes for wheels going out of true:

  • Emergency stops. These can be nearly eliminated on a grade-separated right-of-way. With computer control of motors, faster, non-sliding stops can be achieved than by simply locking up the wheels.

  • Sharp curves. Curves are the bane of all railroading. In a city designed around its transport system, curves can be kept gentle, greatly reducing wear of the wheel profile.

  • Disc brakes. There is no longer any excuse for building passenger rail vehicles with service brakes that run on the wheels. This practice is dangerous (leading to wheel failure if a brake drags) and very expensive in operation. The entire problem can be avoided simply by using regenerative braking, which can provide smooth stops without wear. (With advanced computerized motor controllers it is possible to stop a train without use of the brakes.)

Reducing Personnel Costs

Fully grade-separateed systems can run without drivers. While it took nearly three decades to perfect the technology, several systems are now operating routinely in fully-automated mode. Two other approaches also reduce labor costs.
  • High Speed and Capacity. The higher the average speed of the vehicle and the greater the number of seats behind the driver, the lower the per-seat-mile personnel operating costs. City buses are terrible in this respect. Even articulated buses have only 80 seats behind the driver, and average speeds are very low due to traffic and slow loading and unloading up and down steps.

  • Tight Route Network. A compact route network with frequent service saves money and serves riders better. Sparse networks with the necessarily infrequent service are expensive to operate and provide slow service. A concentration of origins and destinations is a necessary requirement of a compact route network; the suburban pattern of habitation makes such a network impossible.


A change in attitude on the part of transport managers will be required before public transport systems provide comfort levels equivalent to the private automobile. None of the required measures is difficult to achieve, but the perception of the need and the will to make the necessary changes is certainly required. Hard seats, lurching vehicles, and excessive noise can all be overcome with existing technology. Stations can be kept clean, advertising can be removed from vehicles and stations, and decent lighting can be installed. Artists can be asked to help make stations attractive and interesting. The required attitude is that public transport is a first-class service and not merely for those who cannot afford to drive.

Continue on to Fixing the Car
Back to More
Return Home
About the book: Carfree Cities

Copyright ©1996-2002 J.Crawford