Sunday, October 14, 2012

The Interaction of Freight and Passenger Rail

Anyone who has ever ridden a train in Europe (let alone Japan) has seen that they know how to run a railway. Their passenger trains are faster, more reliable, more extensive, and far more frequent than in North America. Unless one enjoys hopping freight cars, however, it is less obvious but no less true that North America has many lessons to teach Europe about freight rail. In fact, Canadian railways move almost as many tonne-kilometres of freight per year as moves by rail in the entire European Union – 349.1 billion tonne-kilometres in Canada versus 377 billion in Europe. Canadian freight railways benefit from long distances and shipments of bulk natural resources, but the figure remains striking given the difference in population and economic size.

In recent decades, North American railways have become increasingly reliable and economical providers of freight transportation service. They’ve expanded beyond their traditional dominance in hauling bulk goods like coal and grain over long distances and have made major gains in hauling manufactured goods since the invention of the intermodal container. Canada’s own CN is often considered the gold standard of freight railways worldwide. Meanwhile, in Europe, most bulk traffic uses the continent’s extensive inland waterway network, while lighter goods tend to move by truck.

Operating a successful freight railway presents very different challenges from passenger rail. The most important criteria of success for freight rail are cost, followed distantly by reliability. As long as their goods arrive within a few hours of promised and the bill is lower than if they had gone by truck, most shippers are quite pleased. Low cost is achieved by packing as much as possible into each freight car, and as many cars onto each train. Over the past several decades, North American freight railways have invested billions into improving their infrastructure to increase their competitiveness. Tracks have been upgraded to accommodate heavier freight cars, new technologies like locomotives distributed throughout the train have permitted longer trains, and bridges and tunnels have been raised to permit the stacking of two shipping containers on each car. European freight cars look like toys compared to their North American counterparts and double-stacked containers are unheard-of. The average European train is less than thirty cars long, while more than a hundred cars per train is the rule in North America.

North American passenger rail operators are often quite reasonably accused of trying to operate as if they were freight railways, using freight locomotives, freight track maintenance standards and freight signalling systems. Europeans make the same mistake in reverse, trying to emulate their success in passenger rail by emphasizing the importance of speed and punctuality, even going so far as to propose running freight trains on high-speed lines. This does nothing to aid in the critical factor of cost per ton-mile. The limited amount of existing freight rail traffic makes large investment seem excessive, just as the low ridership of North America’s weak passenger rail service is used to claim that passenger rail investment is uneconomic. Of course demand is low when the service is poor; demand won’t get any higher until the service is improved.

The poor state of European freight rail is a clear demonstration of the difficulty of operating both a successful passenger and freight railway. The needs of the two types of trains are simply too different. In fact, much of what makes European passenger rail so successful causes serious problems for freight. Heavy North American freight trains wreak havoc on high-quality track built for high-speed passenger service. Europe’s extensive track electrification offers enormous benefits, but it also limits overhead clearance for double-stack containers. High platforms at stations permit passengers to board without needing to climb stairs, speeding loading and providing wheelchair accessibility. Unfortunately, they can also restrict the width of passing freight trains. Some of the problems are simply the result of history, particularly in Britain where railways were built in the Victorian age to a very narrow loading gauge. New European lines are built to minimize curves, a big problem for high-speed passenger trains, but which comes at the cost of steeper grades that are a nightmare for heavy freight trains. European railways have been built over decades for passenger rail, not freight, so it isn’t surprising that European freight railways have struggled.

The needs of passenger and freight railways are so different that separation is likely the only solution. Of course, passenger trains and freight trains can still share tracks in many places, especially on low-density long distance passenger routes, but major passenger corridors should be separated from major freight routes so that they can be optimized for their respective users. Fortunately, the advent of intermodal shipping means that only a core of freight network is required; instead of picking up a few cars at each shipper, much freight traffic is now delivered by trucks to a relative handful of intermodal terminals. North America benefits from its legacy of competing private railways, which produced multiple parallel corridors that can now be divided between passengers and freight. The Amtrak-owned Northeast Corridor has almost no freight traffic, as there is a parallel corridor owned by the freight railways and dedicated to their trains. Europe needs to embark on the arduous project of clearing a core network for heavy freight cars and double-stacked intermodal containers, likely after shifting most passenger trains to new high-speed lines.

Running a decent regional rail system in a North American city will require the exclusion of most freight trains from regional rail tracks. Local customers can have freight cars picked up and dropped off at night, when regional trains are not operating, but through freight traffic should use dedicated tracks. Especially for high-speed lines, the problems posed by North American heavy freight on track superelevation, track wear, platform height, and collision safety are too problematic for extensive track sharing. (More in future posts on sharing between freight and CityRail) Fortunately, many cities including Toronto have already shifted most of their freight rail traffic to bypass routes in the suburbs. Toronto’s main intermodal terminals are in Brampton and Vaughan and there are relatively few directly served industrial customers in the city centre. The Barrie and Stouffville lines, and most of the Richmond Hill, Lakeshore and Kitchener lines, have very little through freight traffic. The Milton line and segments of other routes would simply require separate tracks for freight traffic, either in the same corridor or on a new parallel route. Most freight traffic in the GTA would not interact with CityRail at all, while separating the two track users where they do interact would not be an insurmountable task.


  1. One big difference between European and US railfreight is that US railfreight competes on price, European competes on speed. Intermodal trains in the UK typically run at 75mph (120 km/hr), and so they can provide a way to move good more quickly than on congested roads.

    I remember reading about a former short-line in the US that offered daily freight service between two cities. The freight took 9 hours, a truck took 6 hours - and could leave whenever the shipper wanted. No wonder the line failed!

    1. You're absolutely right about European freight competing on speed while North American freight competes on price. I think that's a big part of why freight railways in North America are so vastly more successful than they are in Europe. Most freight shippers are much more concerned with price than they are with speed. For the small minority that are desperate to get their goods as quickly as possible, they'll use trucks anyway.

    2. They won't use trucks if rail is quicker!

      Also, I think more and more freight is becoming time-sensitive.

    3. Rail will never be quicker than trucks. Even in Europe, freight isn't going to be quicker since trucks can deliver door-to-door. Rail competes on price.

      You're right that freight is becoming more time sensitive, but that's only because freight railways are trying to break into that market, which has been entirely reserved to trucks for the last forty years. Obviously speeding up freight rail is beneficial for some cargoes (esp. intermodal) and getting closer to truck speeds is certainly desirable, but not at the expense of increasing costs. That's why UP and BNSF have abandoned so many of their premium ultra-express services in recent years.

    4. Rail is quicker for cross-country freight trips. Truck drivers have to sleep (I know of no trucking company with a "hotseat" system, maybe there is one) and are limited to 65 mph (in most places).

      Freight trains can keep switching drivers every 8 hours and can run at 80 mph + if the freight haulers choose to.

      Really time-sensitive cargoes go by plane.

      Anything less time-sensitive is cheaper and just as fast to send by rail *over long distances*, which is why UPS and FedEx are two of the biggest customers of the freight railroads in the US.

    5. Trucks with two driver teams and sleeper cabs are actually quite common for transcontinental trips, permitting the truck to keep driving 24 hours a day. UPS uses intermodal rail for much of its transcontinental shipping, but for the most time-sensitive traffic it continues to use trucks. When the railroads tried to take some of that business, the cost to their capacity of running such an expedited service was too great. Top speeds aren't really relevant on such a long trip. You have to look at average speeds. By their own measurement, BNSF's expedited intermodal trains cover about 800 miles per day (, which is very good by freight rail standards but still only an average of 33 miles per hour. A truck with a two driver team can certainly beat that by quite a bit. The reason trains are slower (and will always be slower than trucks) is time spent sitting in yards, loading and unloading, and drayage. To be very clear, freight rail competes primarily on price, even for fast intermodal traffic, and not on speed. That's not a bad thing; the overwhelming majority of freight traffic is not so time sensitive that the difference would matter if the rail service were reliable and less expensive. Since most North American freight railways can pretty much guarantee that these days (as opposed to in the service meltdown era of the 90s), we're seeing massive growth in the intermodal sector.

    6. To be clear, I'm not saying it's impossible for freight trains to be time competitive with trucks. The problem, as UP and BNSF have found, is that operating very fast intermodal trains has huge detrimental impacts on line capacity. It just isn't worth it for the relatively small amount of freight that is that time-sensitive.

      FedEx has just started using rail for the first time, but at least for now it will be for its less time-sensitive Freight business.

  2. You say that in Europe rail is a failure, but what instead happens is that rail freight can charge higher rates because it offers better service; this is the Swiss approach, leading to North American mode shares of ton-km and higher mode shares of value. (I don't know in Canada, but in the US rail's share of the value of freight carried is 4%.) A large majority of the difference between American and European rail freight mode share in ton-km is due to physical and economic geography; lacking the opportunity for many transcontinental coal trains, Europe is instead looking for higher-value goods to carry on rails.

    Running a decent regional rail system in the US can be done alongside freight, even on the same tracks, depending on location. I don't know the details in Toronto, but in Providence, the average speed of frequent-stop EMUs as proposed by Peter Bassard almost exactly the same as that of freight trains that use the line today. Local trains have a higher top speed than freight trains but make stops, and this averages to a similar average speed; at low and medium frequencies, these trains can share tracks, and only at very high frequencies do these differences require track separation. A commenter from Germany, I forget who, once noted the same: on four-track mainlines he observed that all-stop regional trains and freight trains ran at the same average speed, so his regional train would overtake the freight train on the adjacent track in between stations while during station stops the freight train would pull ahead.

    Speaking of freight and CityRail, do you know how things work in Australia? In Australia there is no intercity rail worth mentioning, and the tracks are dominated by freight; on the transcontinental market, rail has 80% of the land market by ton-km. But within eastern Australia, rail has a much lower market share, partly because the distances are shorter and partly because within the urban areas the infrastructure is dominated by regional rail.

    1. I've always thought that ton-miles were sometimes a misleading way to measure freight rail, and value shipped is definitely also a relevant measurement. I'd be interested in seeing those figures for Switzerland and the value share in the States. That may be a bit of a special case, though. The Swiss have invested a large amount of money to create what amounts to a rail bridge across their country, aided in no small part by astronomical fees for trucks using their highways. Even if the Europeans are doing better at climbing the value chain, the difference in scale is pretty striking. DB Schenker controls the large majority of freight rail in the UK and had a revenue of 395 million pounds in 2009 (the newest figures I could easily find). CN alone had revenue in the same year of $5 billion.

      That's a very interesting point about scheduling freight trains. How do you get around all the other obstacles to mixing North American freight and rail regional rail, though? Obviously there's the FRA problem, but hopefully that could be dealt with. But then there's the catenary/double-stack clearance issue; platform height; rail wear and superelevation (though those are less of an issue for slower regional trains); and freight rail grade requirements when building new infrastructure.

      That's a good point about Australia. I'd be interested to also learn more about Russia, China and India, which all have a large amount of of both passengers and freight on rails. Japan's always interesting too, though they're not really a leader in freight rail. JR did build the Musashino Line to get freight out of Central Tokyo, but it seems like they now use it more as a circumferential passenger route.

      I have great respect for your blog and I'd be very interested in hearing more of your thoughts on the CityRail idea or anything else posted here.

    2. First, the obstacles: they tend to be problems for higher-speed trains more than for regional trains. In both India and China, double-stacked freights run under catenary, and the passenger train pantographs are modified to be taller. You can't do this at high speeds because a taller pantograph increases noise and aerodynamic drag, but for a 100 km/h regional train, or for that matter a 160 km/h regional express, it's no big deal. (In the example I brought up of Providence it's an even smaller deal because the Northeastern rail network doesn't have the loading gauge for double-stacked freight except on a few freight-dedicated lines.)

      Clearance can be dealt with in one of three ways: ban oversize freight if none needs to run on the line (this is the case on the Northeast Corridor and the Hudson Line), build platforms with retracting plates (this is done on the Worcester Line), choose a floor height lower than oversize freight such as 550 mm. In Toronto's case, all three are possibilities: Lakeshore has extra tracks so oversize freight could be routed on the express tracks if necessary, retracting plates are always an option, and 550 mm is friendlier to bilevels anyway, and the only reasons to choose high floors and platforms are compatibility with high-speed rail (which wants high platforms), and a preference for single-level trains' faster boarding and unloading times and greater ease of moving between cars.

      In the UK, Spain, and Japan, platforms are high, and I think Benelux, Norway, and Germany's height is also too high for oversize freight. But Russia is moving toward oversize-compatible 550 mm, and the rest of Europe is already committed to 550 mm, which means single-level TGVs, ICEs, Italian HSR, and so on do not have level boarding, which is not a great industry practice but is understandable in countries with so many preexisting platforms. In Canada HSR should have high platforms, since 550 and 1,220 mm are equally foreign platform heights.

      I have great praise for your CityRail proposal. Toronto has preexisting infrastructure elements making it a far easier project than in any other North American city. The only critical comment I have is that maybe (and I really mean maybe) you should include more infill stations. When I Google Earth-toured the Lakeshore lines I saw many locations for infill between Mimico and Danforth, e.g. both ends of Queen Street, Rogers Centre, Gerrard Square, Windermere.

      As for rail vs. truck freight in various countries, Canada benefits greatly from rail-friendly geography: the Trans-Canada Highway isn't an expressway and never will be, and the long distances between the urban centers benefit rail. My guess is that if you look at mode shares within southern Ontario and Quebec, trucks are dominant, just like between Sydney and Melbourne.

      I looked and couldn't find value mode shares in Europe. If you'll let me be a hack for a few sentences, let us consult page 28 of this dataset, where we'll see that of external trade involving the EU, rail accounts for 3.5% of ton-km and 1.3% of value. If the value:tonnage ratio is the same for domestic traffic (and I would guess it's higher because of lack of bulk transcontinental traffic through Russia) then the value share is also 4%. On a less hackish note, DB Schenker Rail has turnover of €4.92 billion a year, counting all of its European business and not just Britain, which has substantially less rail freight than the European average.

    3. Thanks again for the comments. The only problem with raising the catenary is that you would then presumably be dealing with incompatible catenary on slow and fast tracks of the same corridor. I suppose that's not an insurmountable problem, especially if the regional rail routes operate on a more-or-less exclusive pair of tracks like the Munich and Berlin S-Bahns.

      An exclusive pair of tracks for CityRail, which is pretty much what I modeled in that capacity post, would resolve the platform height issue as well: they could be built perhaps to the German S-Bahn standard, which would give us access to a massive pool of rolling stock without the need for modification. Even better, Bombardier is a big player in S-Bahn rolling stock, which might help in Canada.

      Still, platform heights make my brain hurt more than any other aspect of this stuff. Even the Europeans barely seem to be approaching any kind of standard. I noticed that the new Talent 2 is designed with a sloping floor so that it can provide level boarding at a range of platform heights. That would be useful if we had any existing infrastructure to deal with, but not since we'd be building everything from scratch. The only two stations in Canada that have high platforms are Central Station in Montreal and Palais in Quebec, which I believe are considerably higher than the European standard. It seems to me that it would be easiest to adopt two separate standards for CityRail and intercity trains, but that would greatly hinder opportunities for cross-platform transfers.

      It shouldn't be a major problem to remove oversize freight from most of the CityRail network. Most CN traffic uses the freight-only bypass around the city, so only the Milton corridor would need to be shared. A dedicated pair of CityRail tracks on that route would likely be necessary anyway.

      A CityRail proposal benefits greatly from fairly decent pre-existing infrastructure, with plenty more on the way. Unfortunately, the current plans are a bit ridiculous, assuming that each corridor will need its own pair of tracks to run three or four trains per hour in the peak. GO/Metrolinx's current long-term plans would therefore provide more track capacity than we are ever likely to require. Unfortunately, the same assumptions have led to planning for additional Union platform capacity (apparently it is impossible to dwell for less than 5 minutes at a platform). The new Downtown Rapid Transit study, about which I will write more soon, talks about something vaguely resembling regional rail on the Lakeshore route. Unfortunately the cost estimate is outrageous since building a cavern station under Union is assumed to be "necessary."

      You're absolutely right that additional infill stations would make a lot of sense. When designing the map, we were quite conservative. We also did not include any corridors that aren't currently used by GO or any diversions from existing corridors to serve major destinations, with one small exception. The ultimate goal would be for CityRail to serve as a rapid transit system in Toronto and the outer suburbs, with comparable bus connections to what the TTC subway already enjoys.

      In terms of freight rail, I agree that long-distance routes are by far the strongest, though CN and CP have both made pretty impressive strides in the Montreal-Toronto-Chicago intermodal segment.

    4. I forgot to mention that loading gauge is as frustrating to deal with as platform height. Obviously if we were to buy off-the-shelf European (non-Scandinavian) rolling stock, it would be considerably narrower than we could accommodate on North American track. For a project as large as Toronto CityRail, it would probably make sense to order redesigned stock with a wider profile, but that would add considerable complexity compared with an off-the-shelf order.

      I'm not particularly keen on double-deck rolling stock for CityRail. Given that Toronto is nowhere near exhausting the physical capacity limitations of its rail corridors, it seems to me that the loading and unloading benefits at Union of single-level stock greatly outweigh the capacity benefits of bilevels. I suppose at some point in the future if a CityRail corridor were actually moving as many people as RER A, we could look at something like the MI 09. For Regional Express trains serving more distant points (Kitchener, Niagara, Barrie), bilevels make more sense.

    5. Two points:
      (1) In Europe bulk freight goes by water. This doesn't work in most of the US (though actually it does happen through the Great Lakes).

      (2) Two different catenary heights are not a problem; pantographs can be built with a wide range of heights.

  3. Russia knows how to run both freight trains (one of the largest and most efficient networks in the world, comparable to the US) and passenger trains (one of the largest and most efficient networks in the world, comparable to Western Europe).

    Maybe someone should ask them how to do it.

    1. India is another interesting case. I believe separation of freight and passenger corridors is common there.