The earliest transit systems were organic parts of the cities they served.  They grew from the densely populated city center (central business district, CBD) into the emerging suburbs that were created by privately-operated public transit.  Those initial streetcar lines were like valves releasing the pressure on crowded and unpleasant CBDs which housed financial districts, sometimes slaughterhouses, factories, and lots of people. 

Once—and I mean 60, 80 years ago—the warehouses, factories, movie theaters, produce and meat markets, bakeries, apothecaries, haberdasheries, and other sources of shopping and entertainment, all of which are also called jobs, were all located in the CBD.  It has been more than 40 years since Petula Clark bayed about downtown in her 1965 hit of that name (“Downtown, everything’s waiting for you.”) and today those functions of daily living are mostly located somewhere other than downtown.  But as the share of regional employment has dipped to or below 10% in most American CBDs, transit agencies have continued to vigorously pursue the decreasing number of people going downtown as the primary target for transit services.  They do this with what we call a radial or hub-and-spoke system.

Pros of the Radial System.  This is not a foolish policy and is, in fact, quite understandable.  The historical roots of radial systems are compelling.  CBDs are still pretty dense and those higher densities are always attractive to transit providers because CBDs were constructed in a fashion that is very convenient for transit users: the buildings are near the road and, along with street trees, provide shade; awnings and balconies protect pedestrians from the rain and sun.  In downtown Minneapolis, to use an extreme example, there is a system of interior and/or underground sidewalks to keep people out of those frozen Midwestern winters.  Downtown walking distances are relatively short and visually interesting; people will walk great distances where there are things to see.  And there are often amenities like sidewalks, benches, and quality lighting as part of the street scene.  CBDs are, in many ways, very good places to drop passengers off.

Above all, radial systems are easy to manage, plan, and operate.  Buses or trains leave a central point (plaza, terminal, station, whatever we call it), go to an end-of-line point, turn around and return.  At the central point we keep information and fare sales booths, bathrooms, security, supervisors, bike stations, lost-and-found, and vending machines.  Maybe even a daycare.   As transit providers this is a very comfortable and familiar scenario. 

The downside.  Unfortunately, radial systems are increasingly obsolete.  The jobs are not downtown anymore.  To use one sad example, in Minneapolis the Target Corporation is building its new world headquarters 15 miles northwest of downtown while the $715 million dollar Hiawatha Light Rail Line starts downtown and runs southeast.  Considerable research by Greg Thompson and Jeffrey Brown has demonstrated the effectiveness of decentralized systems in getting people to the multiple destinations passengers wish to reach while Brookings has documented the difficulties of passengers moving between suburban destinations on transit in places with high-quality, rail transit focused on the CBD (Missed Opportunity: Transit and Jobs in Metropolitan America, May 2011).

As development has moved away from downtown two things are working against the success of the radial.  First, as already noted employment has moved out of the downtown, especially the service and retail jobs that provide many customers to transit agencies.  Secondly, in response to the movement of those jobs transit providers are traveling greater distances from the central point to reach those jobs and new housing as graphic 1 below shows.  If you study old maps from a transit agency in your area you may find routes that 30 years ago were 30- or 40-minute round trips are now 60- or 90-minutes.  The geographic space of the city has expanded and the employment is more likely located at the periphery, but transfers in a radial system are still made largely downtown.

Graphic 1. Somewhere County Densities, then and now

These conditions lead to several phenomena:

Making the Pie Piece (Out-of-direction travel).  Referencing graphic 1, a passenger from point D going to point H must first go to the central point.  Stopping at the hub is not much of a concern for this cross-town trip, but someone coming from I to H would prefer to head directly south (as he would in a car and as shown by the dotted gray line) but instead is artificially forced to go downtown (A) to make the pie piece or out-of-direction travel.  It is frustrating and time consuming.

Redundancy.  Radial systems are based on convergence, so the radial system sends multiple buses along the same roads often at the same times on the way to the central point.  This is particularly common for systems with high headways (60 minutes) that operate on clock-face time (that is, many/most buses are in the central plaza at 30 minutes after the hour, for example).   The resources used to move several buses along a corridor at nearly the same time—driver time, fuel, and maintenance miles—could be used elsewhere outside of the constraints of the radial system.

Redundacy and funnelling are part of a radial system as seen here in Birmingham, AL.

Funneling:  One of the great disadvantages of a radial system is that it is a system of collection or accumulation.  From the historical perspective this is a positive:  buses would travel in and out of neighborhoods to collect as many passengers as possible and then drop them at destinations in the downtown.  But as the urban core became more and more a place to make transfers in order to travel to destinations well outside the downtown that accumulative system started to show its age spots.  

There are two types of funneling: geographic and time funneling.  Geographic funneling is related to the Pie Piece phenomenon.  Staying with Somewhere County a person coming from an apartment at I and going to a regional mall at H would first go to the CBD, A.  Of course people from I are not the only ones going to the mall so passengers from C,D, E, F, and G also go to A and cram onto a bus, causing overcrowding. 

Then there is time funneling.  The redundancy noted above consumes resources and thus contributes to low frequencies on routes.  Assuming the not-uncommon scenario of a city with a majority of its routes running one once an hour, people wanting to travel at multiple times during the course of an hour can only use the single travel opportunity provided by hourly frequencies.  For example, persons wanting to travel at 7:55, 8:05, 8:15, and 8:25 must all use the 7:45 trip in order to get to the desired destination on-time.  Again this contributes to overcrowding on buses. 

One-Box Dilemma:   A clock-based radial system depends on most buses being at the central terminal at a given time, say the bottom of the hour, to facilitate transfers between vehicles.  Every route has a certain number of minutes to travel out and back.  On an hour route, in theory the bus runs out 30 minutes and in 30 minutes, but in practice there are destinations that must be served and buses must be turned around at appropriate locations.  Therefore the end-of-line point has great influence on the complete route.  A route that is cut short to serve a near destination consumes driver time (the driver still has an hour for the complete trip); a route that is extended a little too far causes delay.  Because of the connection requirements (all buses together at one point on the clock) , the One-Box Dilemma can cause a major destination (or just a good turn around point) to either cause a route to have too much or too little time to be correctly executed.

Route dependency:  Lastly and perhaps most importantly when many low-frequency routes (hourly routes) meet at a central point, connections must be made.  Typically transit agencies in this type of system will hold all buses at the terminal for a late bus.  This makes every bus a little late which is seen as a better alternative to putting the penalty (an hour wait) entirely on the passengers of a late bus.  This means each route is dependent on all the others and an accident or other delay on one roadway results in the immediate delay of the entire system making high on-time performance extremely difficult and subject to vagaries of the auto accidents, construction delays, etc.