Hubbert’s Method
Last week, I read Hubbert’s Peak, by Kenneth Deffeyes. There he gives a more detailed description of the methods Hubbert used to predict the U.S. oil production peak. That was good to read; I was missing a key piece of the puzzle when I was looking at Hubbert’s work earlier.
Deffeyes’ text is a bit flighty, so I’m still not sure I have it right. I should probably see if there’s a good synopsis on the web. But here goes.
Hubbert used two methods. The first presumed a bell-shaped (albeit the exact form was open) curve for production and required as input the total amount of oil reserves in the ground. The latter was the piece I missed. As such, Hubbert had to hold off on making his prediction until he could make a good guess as to how much undiscovered oil the U.S. had.
The second method involved noting that if you plot cumulative production and cumulative discovery versus the year on the same graph, the curves are parallel. For the U.S. data, there is a lag of about eleven years. In other words, the total amount of oil sucked out of the ground up to year Y equals the total amount of oil discovered by year Y - 11. So once discoveries peak, you can expect production to peak eleven years later.
As such, both methods sound plausible to an amateur like myself, but it looks like they have limited use for future predictions, since they require predicting future oil discoveries – wherein the rub has lain all this time.
Deffeyes uses the first method to predict the global oil peak. To do this, he needs to predict the total amount of undiscovered oil reserves in the world. He uses Colin Campbell’s estimates of 1.8 trillion barrels, and predicts a peak in 2003 (the book was published in 2001).
I have to admit, I’m underwhelmed by the predictive power of Hubbert’s theories. Maybe I’m still missing things, and just don’t get it. But it seems to me like the crux of the issue all along has been “how much oil is left”, and Hubbert’s theories don’t address that at all. The other issue of importance for the peak oil crowd is the shape of the production curve on the post-peak side. Hubbert does predict that relatively precisely, but there I have less confidence in the theory because I’m not sure there’s enough real-world data to back it up. It sounds like Hubbert based his assumption of symmetry about the peak from looking at historical data from individual oil fields. That’s reasonable, but I can see technological or market changes affecting the declining production curve. For example, with better technology we can suck oil fields dry faster; that argues against symmetry.
Deffeyes makes an interesting comment: “I have never known anyone to switch sides as a result of an intellectual analysis of Hubbert’s methods.” Strange thing to say about a scientific “theory”. To me, that’s another way of saying that Hubbert’s methods don’t address the real issues (or that they are outright faulty).
So I’m really skeptical; not that oil will peak someday, but that Hubbert’s methods are particularly useful. But like I said, I might still be missing something.
Deffeyes’ text is a bit flighty, so I’m still not sure I have it right. I should probably see if there’s a good synopsis on the web. But here goes.
Hubbert used two methods. The first presumed a bell-shaped (albeit the exact form was open) curve for production and required as input the total amount of oil reserves in the ground. The latter was the piece I missed. As such, Hubbert had to hold off on making his prediction until he could make a good guess as to how much undiscovered oil the U.S. had.
The second method involved noting that if you plot cumulative production and cumulative discovery versus the year on the same graph, the curves are parallel. For the U.S. data, there is a lag of about eleven years. In other words, the total amount of oil sucked out of the ground up to year Y equals the total amount of oil discovered by year Y - 11. So once discoveries peak, you can expect production to peak eleven years later.
As such, both methods sound plausible to an amateur like myself, but it looks like they have limited use for future predictions, since they require predicting future oil discoveries – wherein the rub has lain all this time.
Deffeyes uses the first method to predict the global oil peak. To do this, he needs to predict the total amount of undiscovered oil reserves in the world. He uses Colin Campbell’s estimates of 1.8 trillion barrels, and predicts a peak in 2003 (the book was published in 2001).
I have to admit, I’m underwhelmed by the predictive power of Hubbert’s theories. Maybe I’m still missing things, and just don’t get it. But it seems to me like the crux of the issue all along has been “how much oil is left”, and Hubbert’s theories don’t address that at all. The other issue of importance for the peak oil crowd is the shape of the production curve on the post-peak side. Hubbert does predict that relatively precisely, but there I have less confidence in the theory because I’m not sure there’s enough real-world data to back it up. It sounds like Hubbert based his assumption of symmetry about the peak from looking at historical data from individual oil fields. That’s reasonable, but I can see technological or market changes affecting the declining production curve. For example, with better technology we can suck oil fields dry faster; that argues against symmetry.
Deffeyes makes an interesting comment: “I have never known anyone to switch sides as a result of an intellectual analysis of Hubbert’s methods.” Strange thing to say about a scientific “theory”. To me, that’s another way of saying that Hubbert’s methods don’t address the real issues (or that they are outright faulty).
So I’m really skeptical; not that oil will peak someday, but that Hubbert’s methods are particularly useful. But like I said, I might still be missing something.
posted by Jim at
10:40 PM
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home