Planning a region-wide integrated transit system (part I)

Part I - The hierarchy of services

Last July I made a trip across the Alps, starting from the French Jura region, to rejoin my family who was vacationing in the Pustertal/Val Pusteria, in the German-Speaking Italian region of Sud Tirol/Alto Adige. It was a long, complex trip that involved multiple legs over 3 days. But, as soon as I was in Switzerland, it became smooth and effortless. The last leg of the trip, from Zurich to Olang/Valdaora involved taking five different trains and one bus across the border. Despite so many connections, I spent most of the time travelling and very little waiting for a connecting trip. Despite taking the scenic but circuitous route via the Albula Tunnel on quite a slow train, my trip was only slightly longer than driving along the same route. How so? Many of you already know the answer: the Swiss transit network has been redesigned over the years around the idea of seamlessness, based on a country-wide integrated timetable (integraler taktfahrplan), achieved through regular interval schedule (takt) and timed transfers at designated connection nodes (knoten).

The main goal is to minimize waiting times at connections in order to make trips involving multiple transfers less onerous. In a complex regional transit system, especially in polycentric regions, it’s impossible to provide frequent one-seat rides from any point to any point. However, a system designed around the principle of timed transfer can make the trips between most if not all Origin-Destination (O-D) pairs almost as smooth as one-seat direct services.

But how does this work on the planning side and what are the trade-offs that need to be considered? How should planners proceed to design a timed transfer-based regional transit system? What are the main principles and how do planners can better engage with inevitable trade-offs? These are some of the questions I will try to answer in a series of posts dedicated to some of the principles of integrated timetable planning that are seldom covered in the available online sources in English.

This first post will briefly cover the core notions of regular-interval (clockface) and symmetric schedule and how these apply to the hierarchy of services within a wider region.

Two guiding principles: Clockface and Symmetry

There are two fundamental closely related planning principles underpinning the implementation of a fully integrated regional timetable a la Suisse:

1.      Clockface or regular interval schedules (taktfahrplan in German, orario cadenzato in Italian). As the term suggests, trains and buses run at fixed intervals that are a fraction of the hour (commonly every 15, 20, 30 or 60 minutes), departing at the same minute(s) past the hour from each stop along a given line.

2.      Symmetric schedules. That means that trains and buses running in opposite directions meet each other at designated stations that become the “nodes of symmetry”. The minute(s) past each hour when they meet is called the symmetrieminute(n), that is the time of symmetry.

The best way to explain this plainly is through a simple real-life example. In this twitter thread, I tried to illustrate in a concise way these two principles using the Pustertal/Val Pusteria line. Thanks to a clockface and symmetric timetable, connections between all Origin-Destinations are ensured at the nodes of symmetry. This animated image summarizes how the Pustertal line works (sorry for the typo).

Hierarchy of services

Now let’s make things a bit more complex. Let’s apply these principles to a large region with multiple rail lines connecting at different nodes. Let’s imagine there is a main trunk rail line that connects the most important urban centres of that region. How do we organize service in that theoretical region in order to optimize the service around timed transfer? What is the relationship between the tiering of service patterns and their specialization, between the number of nodes, the optimal number of Symmetrieminuten and service frequency? Let’s dive into it.

Here is our theoretical region: a few urban centers of various size along a main rail line, with various connecting lines and a network of intercity buses linking towns and villages not served by rail.

First, for a logically organized construction of a region-wide integrated timetable, there must be a clear hierarchy of lines, and, furthermore, a hierarchy of services within some of those lines. In other words, there are hierarchically superior services dictating the schedule pattern of hierarchically subordinated services.

In our theoretical region, we can imagine having 3 types of lines:

1. A trunk rail line, connecting all the major urban centres;

2. Feeder local rail lines, connecting with the trunk line at major stations;

3. Feeder interurban (rural) bus lines connecting either with the trunk line or with the local rail lines.

N.B.: planning of urban and suburban transit in and around the major centres is somehow unlinked from our exercise, as probably in the more urban cores transit is frequent enough (< 15 minutes) that its coordination with the rest of the regional network doesn’t really matter.

Multiple levels of service on the trunk line.

It is important to plan for a tiered service pattern along the main regional spine, with both faster expresses and slower locals and/or suburban trains, while the service on local feeder rail lines and bus ones can be essentially made of local “omnibus”, that is services calling at all stops. A tiered service on the trunk line is necessary because when one’s trip is longer than 30-40 km, it is critical to have a service that reaches a higher average speed to achieve competitive travel times; and the trunk lines are where longer regional trips will pass through.

The regional trunk line of our example has 2 (or 3) levels of service:

• Local service, calling at all stops. It can be an R - Regio train (stops every 3-5 km) and/or, in case of larger metro areas, also an S - Suburban train, that is an S-Bahn like service stopping every 1.5 – 3 km and reaching out some 30-40 km around the major urban area. Beware: the differences between these two types of local service can be blurred depending on the local human geography and in some cases it’s purely about branding. They can also coexist, with the R- Regio acting as an outer local + inner express within the area served by the S - Suburban, as it happens around Milan and Munich, for example

• Internode express service, stopping at major centres with high demand and also in nodes where local feeder lines connect. That is an express regional service with maximum speeds of 140-160 km/h and an average speed of 70-90 km/h.

The Internode (hereinafter RV) service is the Structuring Service, that is the service around which the timetables of other, hierarchically subordinated services must be built around as to ensure timed connections along most O/D pairs at major nodes. It’s the service that determines the rhythm (that is, the takt) of the region-wide pulse, a sort of “master takt to rule them all”. Hierarchically, the timetable of the trunk line is what determines the feeding lines’ timetables. Hence, it is more important to devise first the timetable and identify the nodes of symmetry and the symmetrieminute(n) of the trunk internode “master” service and then, only in a second time, organize the other lines’ timetable around it.

In Switzerland, the structuring services are the IC (Inter City) and the RE (Regio Express). In Northern Italy, that role is (or, better, should be) covered by the RV (Regionali Veloci).  In Austria, it’s mostly the Interregio (IR). Regardless of the name and of the variations in the real-life examples, we are talking about an express regional service that, in the real-world’s available examples, stops every 15-25 km or every ~15 minutes. Why 15 minutes? Because in a 30-minute clockface-based system, 15 minutes and its multiples are the “magical numbers” of symmetry.

This is not a normative rule. The currently planned nation-wide Deutschlandtakt in Germany is supposed to be built around an hourly ICE service, and Italian planners have briefly floated the idea of a nationally integrated timetable coordinated around long-distance HSR + Intercity service dubbed “Alta Velocità di Rete”. But for the sake of simplicity, this post will focus of the RE/RV type of regional-scale structuring service, at the scale of a region spanning a few hundred kilometers.

Choosing between a 30’ or a 60’ RV takt. The relationship between “symmetrieminuten” and frequency.

First, it’s important to remember that frequency is not only freedom for the user, but also for the planners. Frequency is the single most important factor determining how many “nodes of symmetry” there can be along a line and whether they are “simple” or “double”, that is, whether they have one or two symmetrieminuten per hour. Thus, the frequency of the structuring service matters a lot for how flexible planners can be in organizing the hierarchically subordinated services. Let’s see why and how by seeing the difference between running an hourly vs and half-hourly clockface service on the structuring RV service of our hypothetical region.

If we plan to run the RV internode service hourly, it means that the RVs running in opposite directions will meet at a node of symmetry every 30 minutes. Assuming a top speed of 160 km/h, we will have a symmetry node every 60-70 km. In our theoretical region of closely knitted medium-sized cities this means missing some points of symmetry (B, D, F). Moreover, all our nodes of symmetry (A, C, E) will be “simple”, that is, they will offer only one minute of symmetry (symmetrieminute) per hour.

If the connecting R - trains or bus services run half-hourly, half of them will miss the timed connection. That means that the frequency of the hierarchically superior RV determines the frequencies of the hierarchically subordinated connecting services. For example, that is the situation at the nodes A and E, where the symmetry minute of the hourly RV is set at around :15.

If we plan to run our RV internode service half-hourly, we have the opportunity to create nodes of symmetry much more closely spaced, some 15 minutes apart. In our model, it means all the the stops served by the RV (A,B,C,D,E,F). Assuming again a top speed of 160 km/h, there can be a node every 25-30 km. Moreover, we are not only getting more closely spaced nodes of symmetry, we are also getting them as "double" nodes of symmetry, each having two symmetry minutes per hour.

Deciding between 30- or 60-minute headways does not only depend on the expected demand for the trunk internode service, but probably more importantly on the will of creating more or less opportunities for all O-D timed connections along our regional trunk line. Running an hourly service means that we only get bidirectional timed-transfer once an hour in only half of the nodes in our hypothetical model. Instead, running a half-hourly takt means that we get two bidirectional timed transfer-windows per hour in all of relevant nodes. It’s a difference between 3 minutes of symmetry per hour across 3 out of 6 potential nodes for the hourly service versus 12 minutes of symmetry per hour for the half-hourly one. The increase in seamless connectivity opportunity going from an hourly to an half-hourly takt is fourfold, even if the amount of service is only the double. Of course this proportion depends on the geography of the region. But as a general rule, the closer the relevant nodes are, the shorter the headways need to be to increase regional connectivity.

There is also room for compromise solutions: planners, for example, can opt for a combination of hourly long-run RV services stretching along the whole trunk line alternated with hourly “short-run” services, covering the sections where demand is greater and opportunities for timed-transfer are higher. This will ensure a 30 minutes clockface on the “core” section of the trunk line and, in our theoretical example, add two more double symmetry nodes. The trade-off (there is always one) is that some timed-transfer connections will be assured only in one direction at the nodes where the short-run hourly service terminates (B,E) while others get no symmetry at all (F) or just a simple one (A).

Where does this model could be applied?

The ideal setting for an integrated regional transit system are polycentric regions. It is not by chance that the idea has been initially developed and thoroughly applied in Switzerland, a country with many medium-sized and closely knitted cities and towns. Yet, the same model is now being applied in Austria, a country with a dominant capital and only a few medium-sized centres lined along two main corridors. The Po Valley, the Tuscan plain and Medioetruria region, Puglia, Campania are all ideal candidates for an integrated regional timetable, if only Italian authorities were remotely interested in pursuing a consistent approach, putting service planning before infrastructure.

But that planning approach could be useful also outside of the Old Continent specifically in some regions of North America. The Northeast Corridor is obviously the most promising place to implement an integraler taktfahrplan. The Keystone, and the Empire Corridor services already have the spacing and average speed of an RV Internode Structuring Service (see chart below). Along the NEC, the Acela Northeast Regional can be used as a faster Internode Structuring Service for the broader region (similar to the ICE in the proposed Deutschlandtakt), while additional shorter RV-like services along some sections of the route can act as the Structuring Service within shorter subsections of the corridor, such as NYC to Philadelphia and Philadelphia to DC, assuring timed-connections with local suburban and regional service. The potential for seamless regional integration is immense across the NEC, if only States, Amtrak and local transit agencies could make an effort toward integrated planning.

Similarly, the Golden Horseshoe region could definitely plan its own regional network around the same principles. Metrolinx’s RER GO plan falls short of fully understanding the potential of better tiering its regional rail service based on a clearer distinction between S-Bahn-like Suburbans and Regional Expresses (RE/RV). The proposed Trillium Rail Network by TRBOT, on the contrary, stresses the importance of having a tiered service along some trunk lines radiating from Toronto. This express service, especially if progressively extended to the entire Southern Ontario, can become the backbone of a fully integrated regional transit system spanning between Toronto, Hamilton, Kitchener, London, etc. covering the region with a web of RV Structuring Services giving the rhythm to the whole regional transit.

What’s next?

In a future post I will focus on a related topic: how planners can approach conflicting needs for multiple types of timed connections across complex metropolitan nodes, using Bologna as a case study. Here is a sneak peak: