I was interested to see today that the possible leakage rate from the oil spill in the Gulf of Mexico may need to be revised up again. Amongst other things, I noted in a post here that the leak was inside the total disaster capacity plan for BP. The rate at the time was 200,000 gallons/day (aka 5000 barrels) vs. a declared disaster capacity of 300,000 gallons/day – both above the original 162,000 estimated ‘worst case’ of course.
There now seems to be sources that would put the true loss rate as much higher – from 1 million gallons/day up to about 4 million (100,000 barrels/day) (e.g. see here, here and many many more places). There is a pattern, after disasters, where people compete to give the highest estimate, with no more demand for the logic than the over-low estimates. There is a good note on this from Reuters here, where it is noted that the highest of these estimates is close to a physical impossibility. Nothing surprising so far in the pattern of reporting. But, it is worth looking at the words used, and comparing to the original Exploration Permit from BP.
From the Reuters note :-
SkyTruth, an environmental group that has looked at satellite pictures of the sheen from April 28, estimates the leaks are closer to 20,000 bpd. Two Texas-based petroleum geologists, who did not want to be identified, said that rate or a bit less was much more likely than the 100,000 bpd flow
Other sources seem to back up this view that 20-30,000 barrels/day for a single well is about the best that can be had.
Now, lets look at the Exploration Plan (there is a copy here). On page 7-1, in the section on worst case scenario determination it notes that the ‘Volume Uncontrolled Blowout (per day)’ is 162,000 gallons – just under 3900 barrels/day. That is less than 20% of what a good single well can produce when operating normally, never mind a blow-out.
Even if we take the 20,000 end of the scale, BPs worst case was less than 20% of the possible well output. It also means that their total disaster capacity for a blow-out, at 300,000 gallons (7100 barrels/day) was considerably less than half of what a single good well can produce. I can find nothing else in the document that helps understand how this worst case figure is derived. You can, at this point, colour me confused.
Circling back to the title of this post, the apparent hubris in estimation here is material. It is already a pretty bold thing to be talking about a worst case scenario. A worst case has very few constraints, and is frankly pretty much meaningless. My idea of worst case might be different to yours, and the phrase does nothing to help us get a mutual understanding.
But, if you are looking to think about about a ‘worst case’, a good place to start is basic physical limits. If a single well can produce 30,000 barrels, when operating normally, then a good starter for an estimate of an uncontrolled blowout might be that. It should certainly give you cause for concern and be worth of note if you pick a number below that. To pick a number far below that should make a reader stop and ask themselves what fabulous powers the modeller has to predict all possible chains of events, and all uncertainties, to arrive at such a low estimate. Top tip – no-one is that smart, and you should be especially suspicious of anyone that thinks they might be. The onus of proof is on them, and 19 times in 20 they will fail the test.
In fact I have an even topper tip. Never let the words ‘worst case’ pass your lips. Talk about bad outcomes, disaster scenarios, stress tests or whatever you will, but don’t delude yourself or others by pretending that you are smart enough to work out the worst it could be. You can’t.