skip to main content

→ Top Stories:
Safe Chemicals
Defending the Clean Air Act

Anthony Swift’s Blog

On the wrong track: Rail is not an alternative to the Keystone XL tar sands pipeline

Anthony Swift

Posted March 6, 2013 in Moving Beyond Oil, Solving Global Warming, U.S. Law and Policy

, , , , , , , , ,
Share | | |

Unfinished Rail 2.png

In its recently released draft environmental review of the Keystone XL pipeline that would bring tar sands from Canada to the Gulf Coast for export, the State Department attempts to make the case that rail could be a viable alternative. The State Department argues that Keystone XL would have little effect on tar sands production because rail could provide an equally feasible and economic transportation option for tar sands. This is a critical element of the draft environmental review because while State determined that tar sands is dirtier than conventional oil, it concludes that Keystone XL would have little impact on the expansion of tar sands and therefore policymakers and the public needn’t consider the impacts of that expansion. However, State’s assumptions are on the wrong track. The Keystone XL tar sands pipeline will drive tar sands expansion. Expansion depends on tar sands being able to reach the high prices of overseas markets. But pipelines to the east and west are stalled and rail – as we will show here – is not an economically viable alternative to Keystone XL. And without Keystone XL, financial analysts are already saying that the tar sands industry’s expansion plan will go off the rails.

In its most recent environmental review, the State Department is repeating its argument that the Keystone XL tar sands pipeline will have limited impact on greenhouse gas emissions because rail transport is an economically feasible alternative. State made several flawed assumptions in its environmental review, including 1) an unrealistically low cost for transporting tar sands by rail from Alberta to Texas, 2) an inaccurate estimate of tar sands production costs and 3) an unrealistic assumption that tar sands production costs will not increase with rising labor, material and energy prices.  In its analysis, State relies on statistics that pertain to rail transport of shale oil from North Dakota but that do not apply to Alberta’s tar sands. Given the infeasibility of transporting large quantities of tar sands by rail and the massive opposition to tar sands pipelines to the East and West coast of Canada, Keystone XL is the lynchpin for significant expansion of the tar sands – and industry analysts agree. Tar sands is expensive to extract and process – with breakeven prices approaching $100 per barrel – and cheap transportation is required to make new projects profitable. Without Keystone XL and the cheap transportation it provides, the tar sands industry will not reach its goal of tripling production by 2030 and the significant climate emissions that come with it.  

A cornerstone of State’s conclusion that rail is a feasible alternative to Keystone XL is the example of rail use by oil producers in North Dakota and Montana. In that region, oil producers facing price discounts and transportation constraints have turned to rail to move their crude to market. However, State doesn’t account for the reason that rail has exponentially grown in North Dakota while it has remained a highly marginal option for tar sands producers. 

Since 2010, transportation constraints have created significant discounts in prices that producers in both North Dakota and Alberta receive for their oil production. These discounts have caused producers in North Dakota to turn to rail to move their product to market. In 2009, rail only supplied enough capacity to move 30,000 bpd out of North Dakota. After significant investment in new rail terminals, rail supplied over 730,000 bpd by 2012. When folks talk about the upsurge of rail transport in the United States and Canada, this is what they’re talking about.

Meanwhile during the same time period tar sands producers have actually faced greater transportation constraints and higher price discounts than those in North Dakota – and yet, they have not seen a similar upsurge in rail. 

Discount (Bakken Cold Lake).png

Figure 1. Discounts between North Dakota’s Bakken crude and Alberta’s Cold Lake crude and similar international crudes (comparing Mexican Maya with Cold Lake tar sands and Bakken with Brent).

State Department’s 2011 Predictions about Rate of Rail Expansion from Alberta Have Proven False

In its August 2011 environmental review, the State Department’s market analysis argued that “shipment of conventional or oil sands crude in Canada is arguably just now reaching a take-off point… the implication is that we could see a growing scale of shipment of WCSB [Western Canadian Sedimentary Basin] crude by rail in the next one to two years.” In both its 2011 and 2013 environmental reviews, the State Department has noted that rail capacity can be expanded in relatively short time spans – taking at most a year to expand existing facilities.

Nearly two years since State’s 2011 prediction, there has been little evidence of a North Dakota trajectory for tar sands to the Gulf by rail. The two measures available to evaluate this are 1) how much Canadian oil is moving across the U.S. border by rail, and 2) how much Canadian crude is being processed in Gulf Coast refineries (which are also served by a number of pipelines from the Midwest).  

In 2011, less than 5,000 bpd of Canadian crude moved into the United States by rail while 150,000 bpd of Canadian crude was processed in Gulf Coast refineries. While rail data for crude oil across the Canadian border isn’t available for 2012, there is no evidence of significant volumes of tar sands being shipped by rail to Gulf Coast refineries. Gulf Coast refineries processed less than 100,000 bpd of Canadian crude in 2012. In December of that year, the month in which there were discounts in excess of $60 per barrel for tar sands compared to international benchmarks, only 50,000 bpd of Canadian crude was processed in the Gulf. 

State’s new market analysis for rail does not explain why a rail boom has happened in North Dakota and has failed to do so in the Alberta tar sands. In fact, State’s environmental assessment uses many of the same assumptions for rail transport that it did in 2011, without either explaining or acknowledging the failures of those assumptions to accurately predict the lack of rail growth for tar sands producers thus far.

Transportation of tar sands by rail from Alberta to Texas is too costly to support expansion

The reason why rail isn’t a feasible alternative to Keystone XL is that it is simply too expensive to support tar sands expansion. State’s conclusions to the contrary are due to their substantially underestimating the cost of rail transport.  In 2011, State assumed that rail to the Gulf would cost producers $9 to $12.50 per barrel. Now they estimate that it will cost them about $15.50 a barrel.

 In reality, the only tar sands producers which are successfully getting crude from Alberta to the Gulf via rail and barge are doing so at a cost of over $30 per barrel. The State Department’s rail prices are estimates – and the fact that producers are currently paying twice as much to move their product to the Gulf suggests State is significantly underestimate the cost of rail from Alberta. 

Figure 2. State Department rail cost estimates compared to actual costs


Pipeline cost $/bbl (EnSys 2011 forecast)

Rail cost $/bbl (EnSys 2011 forecast)

Rail cost $/bbl (State 2013 draft SEIS)

Actual Rail cost $/bbl (2013)

To Gulf Coast from Edmonton/Hardesty


$9 - $12



The reason that this is important is because the high breakeven price for new tar sands projects cannot bear the high cost of rail. New tar sands mines require oil prices of up to $95 per barrel simply to break even. The projects are uneconomic after tacking on an addition $30 in transport costs. However, there is currently a steep discount for tar sands crude which has brought tar sands prices down to between $50 and $60 per barrel. These discounts are expected to persist and deepen over the next few years.

Production Costs.png

Figure 3. The discount between Cold Lake tar sands and Mexican Maya crude (prices in $ / bbl)

Companies with new tar sands production projects have three options. First, they can sell their tar sands in Alberta at less than $60 per barrel – a $40+ a barrel discount and significantly below the breakeven price of most new projects.  Second, they can opt to spend $25 to $30 per barrel to ship their crude via rail to the Gulf Coast in order to sell it at the price of comparable heavy crudes (right now around $90 a barrel). This nets them $65 to $70 a barrel, which is still below the break- even price for many tar sands projects. Third, they can cancel or postpone their project. The only option here that doesn’t lose money for new producers is the third one.

New tar sands projects are likely to continue to face tight profit margins  

State’s conclusion that rail is an economically feasible option is also based on a fundamental flaw in its analysis of the long term profitability of tar sands production. While State acknowledged that many new tar sands projects are economically challenged, it assumed that oil prices would increase through 2035 and concluded that if production costs stay constant, new tar sands projects would be able to bear slightly higher transport costs. State expresses this argument in the graph below in which it adds its low estimate of the cost difference between rail and pipeline to NEB’s 2011 breakeven price and assumes that these new production costs will stay constant through 2040:  

State Production Costs.png

Source: State Department draft SEIS for Keystone XL, 1.4-53.

There is fundamental flaw in this assessment. It would be generally poor advice in any industry to assume that production costs will remain constant – and that is particularly true with the tar sands industry. Tar sands production is dependent on the cost of labor, material and energy, and these costs have been rapidly increasing and are likely to continue to do so if industry pursues its plan to triple production by 2030. In fact, it appears that the 2011 breakeven prices that are State used are already significantly lower than those faced by tar sands producers today.    

Fig. 4. Increasing Costs of Tar Sands Production


NEB 2011 (Baseline for State’s 2013 draft SEIS)

Alberta 2011 (ERCB)

Alberta 2012 (ERCB)

New In Situ

$51 - $61

$47 - $57

$50 - $78

New Mining (no upgrading)

$66 - $76

$63 - $81

$70 - $91

New Mining w/ upgrading

$86 - $96

$88 - $102


 Tar sands production prices have been rapidly increasing and are likely to continue to do so. The upper bound of tar sands breakeven prices appear to have increased by about $15 a barrel across all types of projects from 2011 to 2012.

Getting the production and transport costs right is critical to understand the impact that Keystone XL will have on tar sands production and climate because relatively small changes have a significant impact on production. Even in its flawed estimates, State forecast if rail increased transport costs by $2 per barrel, it would lead to a reduction of 84,000 bpd of tar sands production, while a $7.50 per barrel increase in cost will lead to a reduction of 315,000 bpd of tar sands production.  In reality, State’s underestimates current rail costs by about $15 a barrel and current production costs by up to $20 a barrel (while assuming they will not continue to increase in the future).

If every $2.50 in higher costs reduces tar sands production by 100,000 bpd, underestimating the costs of production and rail transport by $35 per barrel ignores the very substantial impact that a rejection of Keystone XL would have on tar sands production and climate emissions.

Keystone XL is “a key supply chain link” and a driver of tar sands expansion and climate change

Industry analysts generally recognize this - and this is why they have come to fundamentally different conclusions that the State Department did in its most recent draft SEIS. Banks, financial analysts and industry experts underestand that the event of a rejection of Keystone XL, new tar sands production projects will likely be canceled or postponed. In a recent short term market analysis, RBC Dominion Securities Inc., called Keystone XL “a key supply chain link” and estimated that in the event of a rejection of Keystone XL, tar sands production growth would be reduced by 450,000 bpd by 2017, as production projects are deferred.  CIBC forecasts that by 2030, market constraints will reduce tar sands production by as much as 2.4 million bpd by 2030. Goldman Sachs, TD Economics, Standard and Poor, Wood MacKenzie and others have published financial analysis indicating that importance of Keystone XL to tar sands pricing and production.

"If Keystone XL doesn't happen or gets delayed a full year plus, you're talking about projects having to be put on the shelf," Phil Skolnick, analyst at Canaccord Genuity, March 5, 2013

Infrastructure is needed for tar sands expansion, and it is clear to most observers that the permit decision for Keystone XL plays a critical role in the future of tar sands production and the greenhouse gas emissions associated with it. The first step in addressing climate change is to stop making the problem worse – and that means rejecting the Keystone XL tar sands pipeline and the higher carbon emissions associated with it.

Share | | |


Michael BerndtsonMar 6 2013 01:03 PM

Even though I think mining and processing Alberta tar sands is totally stupid - here are some reasons rail transport is not such a stupid idea with respect to Keystone XL:

1) The flow rate (or transport rate) could be limited with rail. Keystone XL is being design for 900,000 bbls per day. Rail transport would have to be much less since a typical tank car holds around 700 barrels (1 barrel = 42 gallons) and there is only so much track, locomotives and steel. Obama should force O&G's hand on this buy saying something like..."hey since your're cool with rail we don't need the pipeline - let's stick with rail - and I'll recommend to Sec. Kerry we call the whole pipeline deal off."

2) A spill from derailment is typically limited to several tank cars or somewhere in the range of 500 to 5,000 barrels. A pipeline spill can easily exceed that within a short period of time.

3) Capital cost would be much less - however, yes operations costs are lower. However, a pipeline is less costly O&M wise based on volume of product transported. So the argument for a pipeline is...the more oil the better - let's pump more fluid and make more money.

4) The volume of oil sands for train transport possibly could be lower then pipeline transport since less diluent would be necessary. Tar sands needs to be cut almost 50/50 with diluent to flow through a pipe successfully. Undiluted bitumen may have to be plopped into a tank car like jello at the hopper - but still. There would be less need to cut the tar sands with diluent pumped up from Illinois to Alberta and other parts of North America.

5) A rail tank car spill could be cleaned up easier than diluted bitumen via pipeline. Assuming the rail car bitumen is higher in viscosity any spill via derailment would mostly sit at the surface or near surface soil. Diluted bitumen if spilled from a pipeline buried 5 to 15 feet bgs would greatly impact the groundwater, almost immediately.

6) If Obama ever becomes environmentally conscious, we could shut the whole thing down with far less sunk capital: rail transport v. pipeline. In other words, there wouldn't be a big old pipeline costing in the billions sitting underground - not in operations. Rail lines are already there - or could be expanded for better north/south transportation of people. The capital spent on KXL could build thousands of windmills along the great plains.

John ReaganMar 6 2013 01:04 PM

You made the case for building pipelines to oil sources within the US borders quite nicely thanks. We have areas of the US that have trouble getting product to the refineries already. When California develops their new find it will be worse. Why are we busy creating jobs in Canada and focusing on their oil when we can't properly resolve our own internal issues? Well...aside from Speaker Boehner and half of Congress investing heavily in Canadian tar sands that is.

Frans KosterMar 6 2013 01:21 PM

By end of this year, 80,000 b/d railed heavy Canadian crude to Delaware City refinery. Also, some portion of Valero's 70,000 b/d rail terminal in California will be dedicated to Canadian crude. Meanwhile, Mercuria to source up to 210,000 b/d of crude - possibly all of it from Canada - via a rail project to the US Gulf Coast. So that's 360,000 b/d potential railed Canadian, and there's other projects I haven't mentioned or that are in the works. Also, Canadian producers seem to be OK with the current price landscape - not thrilled, but OK. The main casualty seems to be upgraders to light, sweet, as we've got plenty of that now down here. I am afraid you are underestimating the significance of rail, which has felxibility not inherent to pipelines, has already existing infrastructure, and has staying power as we discover more previously unexploited sources of oil. Moreover, rail costs have been inflated, and as crude suppliers become more experienced rail's premium to pipe will decrease. In ND, producers are actually turning to rail to such an extent that Enbridge has trouble filling space on its regional pipelines.

Anthony SwiftMar 7 2013 12:54 PM

Hi Frans –

As an example of the feasibility of tar sands by rail to the Gulf, the Delaware City refinery is a bit of a non sequitur for two reasons – 1) rail to the East Coast is cheaper than to the Gulf and more importantly 2) the East Coast has very limited heavy crude processing capability (about 120,000 bpd, most of which is in the Delaware City refinery. Railing 830,000 bpd of Canadian tar sands to the East Coast isn’t an alternative to KXL, and the decision to rail Canadian heavy to Delaware City would have likely happened regardless of the decision on KXL.

The 210,000 bpd Houston terminal is for oil from West Texas, Cushing, Bakken and Western Canada (including the Canada’s part of the Bakken). Tar sands is a) going to have to compete with all of these sources, b) is at a locational disadvantage, c) requires special cars and unloading facilities, and d) fits few barrels per car than light crude. Based on these factors and rapidly expanding domestic crude in the Permian Basin and Bakken, it’s likely that very little of the terminal’s capacity will be tar sands (and certainly more likely that none will be tar sands than all will be).

And I don’t disagree that rail is more flexible – and will continue to be a major transport option for the country’s tight oil boom. However, there are reasons that tar sands producers haven’t turned to it the way that tight oil producers have, despite similar incentives to do so.

Scott – it sounds like we agree on the main conclusion. But on the details:

1) Southern Pacific estimates its costs at $31 per barrel. We’ll look into whether this is net or gross. But, either way, a model where you move tar sands by rail down and diluent up only works if the vast majority of tar sands is moved by pipeline. If producers must move millions of bpd of tar sands by rail, there will not be demand for millions of bpd of diluent.

2) Accounting for Southern Pacific’s $31 per barrel costs is reportedly based on a united train contract (ten per month). Prices may come down in a higher volume scenario (which introduces other issues), but it is highly unlikely that they will come down as far as State projects in their scenario (which is why nobody in the market has suggested building the projects that the State Department proposes).

3) North Dakota producers were also looking at the same probabilities that Keystone XL (and other pipelines) would be coming on as transport options, and yet they turned to rail in the interim. Tar sands producers have faced even higher discounts than ND producers over the same time period and haven’t. There’s a reason – it’s not as feasible for them based on their costs and constraints.

Paul MagnusMar 7 2013 02:31 PM

You would hope that some media main stream like CNN would do the maths for all....

Michael BerndtsonMar 7 2013 02:37 PM

Anthony, from your reply above you say: "If producers must move millions of bpd of tar sands by rail, there will not be demand for millions of bpd of diluent."

By any chance is this a product searching for a market? Or more specifically, will the massive amount of diluent needed to move bitumen from Alberta come from natural gas liquids and other market-marginal hydrocarbons produced from US shale oil and gas fields? Either way - would rail shipping minimally diluted bitumen (like enough so it pours into a rail tank car) down to refineries east/west/south for mixing upon crude distillation eliminate much of the energy/cost associated with fluid handling?

One of my long time complaints about policy making and project development is how poorly feasibility studies are performed. They typically are done by engineers and consultants in a perfunctory style so the ideas and wishes of the paying client flow nicely through the technical, economic and societal screening steps. For example, if Transcanada wants to build a pipeline from Alberta to Houston, the alternative of - "build a pipeline from Alberta to Houston" almost magically becomes the preferred option - based on the right assumptions and weighting factors. And of course this may be true for any KXL competing forces as well.

I found through experience that the best way to conduct a feasibility study on a project as sensitive as Keystone XL is to first develop and negotiate, with interested parties, the caveats and assumptions - before starting the FS. It saves time, money and reduces embarrassment as the process goes from development to delivery.

Adam BedardMar 7 2013 04:23 PM

Michael: The diluent needed to ship bitumen to the US isn't that massive. By 2017, Canada will need to import about 340,000 b/d of diluent (they will produce about 140,000 b/d of their own diluent). They prefer high API (~70API). Eagle Ford production in 2017 is forecast to average 1.25 (or more) million b/d of which condensate and super light oil can comprise 50%. Combined with the deluge of NGLs in the US, there will be plenty of diluent supply for blending. The problem becomes that there is too much condensate, and so the diluent market is finite, particularly since it essentially gets recycled back up to Canada after it has been through the refinery.
Keystone XL makes sense economically, operationally, and politically on almost all levels. One option that hasn't been mentioned is railing crude out of Albert to a dock, and then barging it from there. This may be viable economically compared to railing it all the way to the Gulf.

Michael BerndtsonMar 7 2013 05:40 PM

Adam, my only experience with tar sands is getting stuck in traffic on an almost daily basis in Northwest Indiana a couple years ago - as new vessels and reactors for bitumen processing where being moved from staging to the Whiting refinery. The by-pass bridge was shutdown so I had to take local roads to my project site - a completely unrelated matter.

Here's my questions and thoughts about Alberta tar sands - assuming it will continue on into the future. Apparently 80 percent of Alberta tar sand is too deep to surface mine. Eventually most or all of bitumen will be from in situ mining techniques. Will the physical/chemical properties change much, effecting transportation or shipping? The reason I ask is that I read a company is piloting in situ electric resisting heating (but not really electric resistance heating pursuant to nomenclature issues) somewhere in the tar sands of Alberta. I'm assuming this technology is a potential alternative to steam thermal. Anyway, it seems to me the hydrocarbon fluid would be much lighter and less viscous from in situ techniques then say the bitumen processed from surface mining - due to the heavy hydrocarbons being left in the ground.

Andy SkuceMar 9 2013 06:57 PM

You made some good points here. I agree with your overall conclusion, that rail is not a viable option for large scale bitumen transport, but I can't help thinking that you have overstated your case.

1) The operating rail transport project that you mention is a very small start-up case. Surely costs will fall for larger scale projects and, in this case once efficiencies are learned and upfront costs amortized..
2) Most of the projects developed in the future will be lower cost per barrel in-situ projects rather than mining. Also, the product most likely to be shipped by rail is raw bitumen rather than upgraded synthetic oil, which is more likely to be pumped to market. Break even prices for future projects liable to be shipped by rail are therefore likely to be closer to $70 than $100.
3) Cost control is certainly a big deal in oil sands projects and has been a serious problem in the past. However, firstly, general inflation is already factored into the product price forecasts, which are in constant 2011 dollars. Also, companies are surely learning from their past mistakes, so they will be better at managing future costs, one might expect. Technology should improve also and also drive down costs. On the other hand, there will be rising costs as companies move to the second geological tier of project with inferior subsurface qualities. Also, natural gas prices may well rise from current record low prices, which will increase operating costs for the in-situ projects in particular.

DavidMar 11 2013 01:20 AM

A majority of Canadians support the West-East pipeline according to an Abacus Data poll released last week. You can find the details here

Irvin DawidMar 12 2013 06:24 PM

Any comments as to whether it's safer or more energy efficient to ship by pipeline rather than rail?
Also - I hope you saw Monday's WSJ article, "U.S. Refiners Turn to Rail to Tap Canadian Oil"

Comments are closed for this post.


Switchboard is the staff blog of the Natural Resources Defense Council, the nation’s most effective environmental group. For more about our work, including in-depth policy documents, action alerts and ways you can contribute, visit

Feeds: Anthony Swift’s blog

Feeds: Stay Plugged In