1.1 Basic data – the line

I will base my calculation primarily on the Cologne–Frankfurt high-speed line.

The travel distance from station to station is 180,005 m. Along this, the new line is only 163.6 km, but it has also double-tracked branches 15.2 km, 13.2 km, and 4.8 km long – a total of 196.8 line-km. (The surface area taken up by tracks is thus around 2 million m².)

The total length of tunnels along the lime is 50,020 m (some added to original plans upon local demands, some with construction problems, contributing to a well over-budget €6.0 billion total cost), and even bridges add up to 6,012 m. While there are high-speed lines with much more superstructure (one of the first German lines is more than 90% tunnels and bridges, while Japan's and Taiwan's trunk lines are not far behind), most have around this relative amount or less, some much less – say French lines or Berlin–Hanover.

The amount of concrete used to line the tunnels was 3 million m³. I have no overall data for the bridges, but from the amount of concrete / length of some of the largest (for example, the below pictured Hallerbachtalbrücke = Haller Creek Valley Bridge: 992 m long with 17,320 m³ concrete), it should add up to no more than 0.1–0.2 million m³. Most of the track was built as fixed track: instead of rails on sleepers on a bed of broken stone, rails were fixed to a concrete base. For this, 0.55 million m³ was used. Summing up and rounding up, I'll calculate with 4 million m³ of concrete. (This is around 10 million tons.)

From data given for some tunnels and bridges, it seems that the superstructures contain steel with a mass around 1/20th of concrete, most of that in reinforced concrete. That's 0.5 million tons. Rails are 60 kg/m, so two tracks over 200 line-km are 48,000 tons. All other uses (catenary, signal masts etc.) are dwarfed by these. I'll go with 0.6 million tons of steel.

In addition to concrete and steel, various vehicles had to transport the 7.5 million m³ earth that was removed. After adding all up, I'll go with 30 million tons to move.

1.2 Basic data – specific energy and emissions

Steel
A 2000 German source claims a global average of 19.8 GJ⁄t(steel), and from its totals, a global average specific emission of 1.7 t(CO2)⁄t(steel) can be calculated, and they say it's less for Germany. However, many other sources, including an quarry in Austria [pdf!] and a German wind turbine energy balance calculation [pdf!], use 1.85–2 t(CO2)⁄t(steel).

Now checking with the German Federal Environment Agency's database of energy/emissions models, I find various data. The average German mix, which includes about two-thirds steel from iron ore and one third electro-steel from scrap metal (much less energy intensive process), is cited at 19.9 GJ⁄t(steel) resp. 1.39 t(CO2)⁄t(steel). Oxygenated steel at 22.8 GJ⁄t(steel) resp. 1.6 t(CO2)⁄t(steel). What I find strange is the even higher figure for raw iron: 23.8 GJ⁄t(steel) resp. 1.64 t(CO2)⁄t(steel). I note that clearly almost all of the energy and emissions comes from smelters, for example, mining contributes only around 0.5 GJ⁄t(ore) and 0.05 t(CO2)⁄t(ore).

Very much on the safe side, I shall use 25 GJ and 2 t(CO2)⁄t(steel).

Concrete
For concrete itself, one can find strongly different figures. That quarry in Austria [pdf!] gives 78 kg(CO2)⁄t(concrete). A paper on ecological home-building [pdf, German!] evaluates two different types of concrete at 73 and 131 kg(CO2)⁄t(concrete) – and 495 resp. 764 MJ⁄t(concrete). US industry site Concrete Thinker's figures convert to 103 kg(CO2)⁄t(concrete). That wind turbine calculation [pdf, German!] even gives 171 kg(CO2)⁄t(concrete). Even the German Federal Environment Agency database has one set of data with 131 kg(CO2)⁄t(concrete), another with 170 kg(CO2)⁄t(concrete) and 988 MJ⁄t(concrete). In different units (1 m³ concrete is roughly 2.5 t), a German industry brochure [pdf!] gives 251 kg(CO2)⁄m³(concrete) and 1,750 MJ/m³(concrete). A study of Swiss concrete production facilities [pdf!] found primary energy uses of 905–2,370 MJ/m³(concrete).

Why so much variation? If one checks figures for actual concrete production, say those in Swiss construction giant Holcim's energy-emissions report [pdf!], they are about half a kg(CO2)⁄m³(concrete) resp. below 35 MJ/m³(concrete) – negligible compared to all figures above. Almost all of the energy and emission comes just from making cement. As Concrete Thinker says, concrete can consist of anything between 7–15% cement, which explains the above variation. For cement production, 40% of emissions come from combustion, and 60% from the chemical process itself: calcination releases CO2.

Concrete Thinker gives 0.9 t(CO2)⁄t(cement) for cement production. The German power plant ash association's brochure [pdf!] cites two figures, 0.95 and 1.01 t(CO2)⁄t(cement). As for energy, the US industry site cites 4.65 MBTU/(short) ton(cement), which is around 5.45 GJ⁄t(cement) in civilised units. The German Federal Environment Agency's database has 0.944 t(CO2)⁄t(cement) and 4.88 GJ⁄t(cement). (Note that for the above-mentioned concrete with 170 kg(CO2)⁄t(concrete) emissions, they calculated with 16.9% cement.)

So if for concrete, it's all about cement content, I have to check that. A construction research site's lexicon [German!] gives, as high mark, 270–300 kg(cementJ⁄m³(concrete) for surface reinforced concrete. Cement content in the special concrete for the fixed track trackbed is somewhat higher, in more mundane filling concrete lower, in shotcrete 50% higher, so I feel justified to use the maximum for surface reinforced concrete as overall average. With the above German data, that would mean up to 1,465 MJ and 0.28 t(CO2)⁄m³(concrete) from the cement.

I will use 2,000 MJ/m³(concrete) and 0.3 t(CO2)⁄m³(concrete).

As postscript, I note that net CO2 emissions of concrete are significantly less: concrete recarbonates over time, absorbing CO2 from the air. But I hand the question of how significant that could be for bridge/tunnel/trackbed concrete to margouillat.

Materials transport
I will assume that all the material to be moved is transported on trucks. Thus I can take the worse figure from Part 1 (the one for >3.5 t trucks on average) of 3.6 l(diesel)⁄100 tkm. When calculating total energy, I will convert to Mega-Joules with the standard figure for volumetric energy density used in Germany, 35 MJ/l(diesel). For CO2 emissions, I will round up the higher figure in Part 1, to 100 g(CO2)⁄tkm.

I will further assume average transport distances of 100 km, one way. I will account for the return journeys as if those were loaded, too.

Tunnel boring
The tunnels on the Cologne–Frankfurt line are mostly large-diameter single-tube ones built with the drill&blast method. However, I think the use of tunnel boring machines (TBMs) is more typical and will become even more typical, and two tubes is now standard for longer tunnels. Thus I chose another German high-speed line tunnel as standard, one with rather large-diameter bored tubes and cut through varied stone under not too high mountains.

The Katzenbergtunnel, just north of the Swiss city of Basel, will open in 2011. The bored part of its two tubes is 8984 m long. The Marion/east tube took 838 days (14 June 2005–20 September 2007), that's 10.72 m/day; the Inken/west tube 733 days (23 September 2005–1 October 2007) or 12.25 m/day.  The two TBMs used were rated at 3,200 kW. Generously assuming TBMs running at full power throughout (in truth they stopped for holidays and regular checks), we get 0.121 TWh for both tubes, that's 435 TJ. Dividing with 8,984 m, one gets 48.5 TJ/km(line).

The above figure is end-use energy. From DB's figures in the previous diary, one can estimate traction end-use electric energy as around 35% of primary energy. I'm going to assume that TBM electricity supply is rather similar to locomotive electricity supply, thus: around 140 TJ/km(line).

For CO2, one can just divide the figures for electricity-related emissions and end-use energy in Part 1 (about 5/3 t(CO2)⁄TJ), and apply that to the above – getting at around 8,000 t(CO2)⁄km(line).

2.1 Total construction primary energy consumption

Now I apply the specific figures to the total volumes on the Cologne–Frankfurt line, e.g. multiply the numbers in 1.1. and 1.2:

• steel: 0.6 million t(steel) x (25 GJ/t(steel)) = 15,000 TJ,
• concrete: 4 million m³(concrete) x (2,000 MJ/m³(concrete)) = 8,000 TJ,
• material transport: 30 million t(load) x 2 x 100 km x (3.6 l(diesel)⁄100 tkm) = 216 million l(diesel) = 7,560 TJ,
• tunnel boring: 50 km(line) x (140 TJ/km(line)) = 7,000 TJ.

2.2 Total construction CO2 emissions

• steel: 0.6 million t(steel) x (2 t(CO2)⁄t(steel)) = 1.2 million t(CO2),
• concrete: 4 million m³(concrete) x (0.3 t(CO2)⁄m³(concrete)) = 1.2 million t(CO2),
• material transport: 30 million t(load) x 2 x 100 km x (0.0001 t/tkm) = 0.6 million t(CO2),
• tunnel boring: 50 km(line) x (8,000 t(CO2)⁄km(line)) = 0.4 million t(CO2).

Let's try one cross-check on the truck data, using a standard number for CO2 emissions from burning Diesel oil including fuel production  share: 216 million l(diesel) x (2.84  kg(CO2)⁄l(diesel)) = 613,440 t(CO2), that's a good fit.

3 From construction totals to contributions for train travel

Now comes the trickiest part. On just what amount of passenger-kilometres should I distribute the construction-related totals?

One part of this question is time. In how many years should one expect the replacement of the entire material of the line, or decommissioning from service? With the oldest high-speed line (Tokyo–Osaka) in service for only 43 years, and high-speed fixed track not in use for two decades, there is no hard data. Tunnels on the older Japanese lines got major overhaul, but that still wasn't complete material replacement. On normal rail mainlines, concrete in tunnels and bridges can survive over a century. Rails are changed much faster, 1–3 decades. With development in technology, it is unpredictable how energy efficient the next replacement or even maintenance will be.

So, while it could be much linger as per above, I will calculate with a depreciation time of 50 years for all main contributing factors.

With that, the annualised construction-related primary energy consumption for each contributing factor, expressed in Tera-Joules, as well as in litres gasoline, with the German standard of 32 MJ/l(gasoline) (figure for Superbenzin blend):

• 12,000 TJ/50 = 240 TJ or 7.5 million l(gasoline) from steel,
• 8,000 TJ/50 = 160 TJ or 5 million l(gasoline) from concrete,
• 7,560 TJ/50 =~ 150 TJ or 4.7 million l(gasoline) from materials transport,
• 7,000 TJ/50 = 140 TJ or ~4.4 million l(gasoline) from tunneling.

The annualised construction-related emissions:

• 1.2 million t(CO2)⁄50 = 24,000 t(CO2) from steel,
• 1.2 million t(CO2)⁄50 = 24,000 t(CO2) from concrete,
• 0.6 million t(CO2)⁄50 = 12,000 t(CO2) from materials transport,
• 0.4 million t(CO2)⁄50 = 8,000 t(CO2) from tunnel boring.

Now what about annual travel volume? This is the trickier half of the trickiest part.

Traffic volumes on high-speed lines over the world differ by more than an order of magnitude. Even if we would go for network averages, the large difference remains (say between Japan's and Spain's). What's more, traffic volume usually runs up over the time scale of a decade, or eve decades, so calculations based on the first few years might be too conservative.

I will not even attempt to find a magic average. Instead, I will calculate five different scenarios, with five different traffic volumes:

1. 50 million passengers x 200 km = 10 billion pkm (should be typical for connecting close megacities, like in East Asia),
2. 20 million passengers x 200 km = 4 billion pkm (say line between a megacity and another major city, the busiest European and US lines have/can reach this level),
3. 10 million passengers x 200 km = 2 billion pkm (connecting two major cities; it's the level of Cologne–Frankfurt in 2007: 11.7 million passengers),
4. Connecting less major cities: 5 million passengers x 200 km = 1 billion pkm (connecting less major cities),
5. 2 million passengers x 200 km = 0.4 billion pkm (connecting to local centres, say Córdoba–Málaga).

4.1 Contributions to specific primary energy consumption of train travel

So construction adds to the specific energy consumption, in terms of litre gasoline / 100 passenger-kilometre (l(gasoline)⁄100 pkm):

4.2 Contributions to specific CO2 emission of train travel

So construction adds to emissions, in terms of grams CO2 / passenger-kilometre (g(CO2)⁄pkm):

5 Comparison with other modes

The rival modes for high-speed are cars on highways, and short-range airplanes between airports. To be frank, I don't know enough about highways and airports to make a quantitative comparison, so here only some qualitative considerations.

High-speed lines have (or connect to) stations, airports have terminals, highways have tank stops and rest houses and exits – I think those are roughly comparable, and suspect they don't make much of a difference in a comparison.

Because cars can climb higher grades, highways have less tunnels and bridges than high-speed rail. However, they are much wider: the standard highway road-top cross section for Germany is 29,5 m, as opposed to 9 m for the fixed track on the Cologne-Frankfurt line (and only a little more at the bottom of the crushed stone ballast bed for conventional track). But I don't know how much concrete is involved, nor the CO2 and energy figures for the frequently changed asphalt cover – only suspect that overall, it's the same volume.

For a comparison with car traffic, one would need to separate short-range commuters near cities and longer travels along the highway. German highways see 100,000+cars/day in the busiest city sections, and I found average figures for the state of Rhineland-Westphalia (includes the Ruhr Area conurbanism) around 50–60,000 cars a day. From that, I conlude that long-range highway traveller volume along Cologne–Frankfurt is probably multiple billion pkm, anything from half to four times the high-speed rail traffic.

While airplanes don't need infrastructure to fly, they need to start/land and park: the landing strips, taxiways and apron on airports. Some airports need total volumes comparable to my above high-speed line: the new Berlin Brandenburg International airport (rebuilt from Berlin-Schönefeld) needs 1.35 million m³ concrete and 0.25 million m³ asphalt. But this involves some fixed installations which I put aside before, and for short-range flights, smaller airports should be the comparison. There is the new South landing strip at Leipzig [pdf, German!]: the 3,600 m landing strip, its taxiways, and the new apron cover altogether 870,000 m², on which 0.6 million m³ concrete was poured. That's an order of magnitude less than for my high-speed line, though I suspect it must be replaced much more frequently.

I note that the number of nodes and Cologne–Frankfurt-scale lines in a hypothetical fully-built German network are around the same (25–30), and airports for domestic flights should be in the same range. Thus one medium-size airport to one high-speed line seems a reasonable approximation.

:: :: :: :: ::

In a third episode, to be posted maybe next week, I shall look into more studies of full-chain or partial-chain railway and other transport emissions.

:: :: :: :: ::

Check the Train Blogging index page for a (hopefully) complete list of ET diaries and stories related to railways and trains.