taking advantage of messes

A paper came out this week in Environmental Science and Technology, probably my favorite technical journal, suggesting a “risk-managed route” for the Keystone XL pipeline.  The original route made enough stakeholders in Nebraska nervous about their groundwater supplies that the governor requested President Obama to deny the Presidential Permit, and so it was done.  Keystone has a new proposed route, which will avoid much of the Nebraska Sandhills, but still go through some ecologically sensitive areas.  The authors of this paper propose a route that avoids surface water crossings in the canyons of the northern part of the Ogallala aquifer, and instead intentionally crosses spray-irrigated, row-cropped land underlain by nitrate-contaminated groundwater.  The route eventually connects with the existing Keystone 1 pipeline north of the Platte River, rather than the proposed connection in Steele City, NE.

Now, I’m sure that Keystone has its own agenda for why it chose the route it did, and there are parts of this “risk-managed route” that are left unmentioned in the paper (it’s only 5 pages, so it can’t cover everything).  That said, the authors’ proposed route has an interesting idea as far as the irrigated cropland goes.  They assume that pipeline spills are inevitable (over the past 10 years, they cite statistics of 0.8 spills per 1000 miles of pipeline, averaging 364 barrels of oil), and that dilbit will partition into the part that volatilizes (evaporates away) and the part that floats on the water surface (light components called “light non-aqueous phase liquids” or LNAPLs).  The existing irrigation infrastructure is already set up to deal with LNAPL spills in the groundwater: the irrigation wells can extract the water+hydrocarbons, and the irrigation sprayers can enhance volatilization of the components pumped out of the ground.  Interesting thoughts, and true.

The big red flag in my mind is the actual composition of the dilbit (diluted bitumen) to be carried in the pipeline.  Normal crude oil is largely lighter than water, and does in fact float.  But bitumen is more like tar, and it’s diluted with lighter hydrocarbons so it can even flow in pipelines.  I don’t know if bitumen composition is a trade secret or something (I couldn’t find much information), but it’s possible that it would sink under the water table (making it a dense non-aqueous phase liquid, or DNAPL) rather than float.  In fact, the Enbridge spill in Michigan suggests that dilbit will, in fact, sink, though in that case, the dilbit got mixed in with sand and sediment that caused it to sink into the riverbed along 40 miles of the Kalamazoo River.  There’s not much mixing like that in groundwater — things are pretty static.

But think about this: LNAPLs are easier to remediate than DNAPLs, based on the physics of water and non-aqueous phase liquids.  I found a statistic that the average spill cost is ~$2,000/barrel for normal crude oil, but the spill in Michigan is already past $29,000/barrel.  Multiply that by 364 barrels/average spill, and you’re talking a $10.6 million starting point, on average.  That’s an expensive project.

I guess my take-home message is, I like the thinking outside of the box in the paper, but if dilbit is, in fact, DNAPL rather than LNAPL, we need to do a lot more due diligence and risk mitigation before we approve anything.

The biggest oil spill you’ve never heard of

In 2010, there was an oil spill of nearly 1 million gallons on the Kalamazoo River in Michigan.  You may not have heard about it.  The spill came from a leaking pipeline chock-full of diluted bitumen (“dilbit”), a particularly heavy mixture of tar sands oils (plus some lighter-weight stuff).  The company who owned the pipeline, Enbridge, had a leak prevention system in place, and just 10 days before the accident, told federal regulators that it could remotely detect and shut down a rupture in 8 minutes flat.  Unfortunately, it took 17 hours to confirm the spill – operators thought that a bubble was blocking the flow and restarted the flow multiple times to unblock it…only to discover a huge leak was the actual problem.

Apparently remote detection of leaking oil pipelines is pretty difficult, technically speaking, because the flow rate changes so frequently within the lines.  It’s hard to know if some oil is missing if you don’t know how much should be there in the first place.  And there have been more than 100 significant spills per year for the past 20 years.  So it kind of makes sense that environmentalists and farmers are nervous about the proposed Keystone XL pipeline – politics aside, it does sound pretty risky to put that much oil on top of the Ogallala aquifer, water supply to a vast swath of the Midwestern US.

The 2010 Kalamazoo spill brought needed scrutiny to Enbridge, against which a record $3.7 million fine was levied, and will hopefully prompt the company to take safety a little more seriously.  But the spilled “dilbit” is still wreaking havoc on ~40 miles of the Kalamazoo River: being heavier than water, it sank beneath the riverbed and will cost ~$500 million to clean up.  Yikes.  Granted, an aquifer isn’t the same as a river – probably a spill would be pretty localized due to low flow rates – but again, is it worth the effort?  Is it so much to ask that these pipelines not leak?

The latest technology proposed by Keystone XL would be able to detect leaks larger than 1.5% of flow…which is nearly 500,000 gallons per day for the proposed flow rates.  Not that convincing, to be honest.  Oh, and did I mention that Keystone XL will be pumping dilbit, too?  That’s a high-risk proposition.  I think, given these facts, it’s very reasonable for American regulators to be hesitant about the pipeline, and hopefully we can hold the Keystone folks to a high standard while sensibly choosing routes around sensitive areas of the Ogallala aquifer.  Because if we veto this outright, the tar sands oil will travel by pipeline to the Pacific coast and be refined in China — not our problem, per se, but also not good for the environment.

Kansas and Texas: the next Yemen

I wrote earlier about the groundwater overdraft in Yemen, which was largely due to irrigated farming (specifically for qat, a mild stimulant).  I sort of brushed off the idea that the Ogallala aquifer was drying up.  But, researchers recently used satellite data to show that yes, it is drying up, and at an alarming rate.  In the southern portions, the aquifer may be unable to support irrigation as soon as 30 years from now.  The northern portions seem to be holding steady in water level, due to rainfall and infiltration from lakes, so only Kansas and Texas would be affected.  The same study showed that groundwater tables fluctuate widely in California’s Central Valley, although a clear decline is not evident.  These are major farm belts in the US, and the loss of irrigation water would be, quite simply, devastating for those areas.  We would do well to take heed and adopt proper policy measures to sustain these critical but not unlimited water resources.