[Pops]

So I thought we might work on the 10 basic facts as individual threads.

Right now our fact numero uno is:

1. Oil is a finite resource on a human time scale, unimpeded extraction typically follows a bell-shaped curve with a maximum followed by terminal decline.

The first part; oil is finite, I'm pretty sure is not in doubt.

But the second part, apologies to Hubbert, is wrong I think. If not wrong from a 500 year historical perspective, from an up close, limited attention span, human perspective it has proven to be a poor predictor. And even if it is ultimately right, what use is it as a basic fact since there is no such situation as "unimpeded extraction"?

The biggest problem with Hubbert's method is that to guess the midpoint of extraction (and so the peak) one must first guess the total amount of oil that will ever be extracted and only then plot out a nice smooth curve.

Making that prediction of the ultimate resource has been done at least 2 ways, first simply by looking at discovery; since you can't pump what you can't find, the amount discovered surely indicates the amount that can be extracted. I'll talk more about discovery later.

The second way is called the "Hubbert Linearization", a simple formula that divides the amount produced in a period by the cumulative amount produced to that point. A series of such periods are then plotted against cumulative production and where a trendline intersects the x axis is the estimate of total.

Here is a post by Stanifod back in '05 and the chart with Staniford's fit estimating something under 2.4Tbbls in yellow, Dryfesses in pink and (I assume) his guess at how a couple of others would look:




BTW, my math stops somewhere around Pie Are Square so please correct it at any point.

--
Sticking with the linearization method, the first problem I have is the the resulting URR derived is assumed to be a relatively static number - otherwise what good is it? But it isn't, it is derived from past production and production methods change over time due to technology, demand, oil price. Even worse, depending on which particular year you start the trendline gives pretty different estimates of ultimate.

Here is a great chart from Here (oil+gas in tons of oil equivalent) showing a range of ultimates depending on the starting year of the trend:




--
So a little about about backdating reserve growth. First, reserve numbers are always influenced by economic and political considerations so should rightly be suspect. But It is pretty well the received wisdom that discovery peaked back in 1964 or some such. For my math impaired brain, the chart showing reserves having peaked was the clincher - after all you can't drill what you can't find.

Here is a comparison of the amounts of original discoveries, and reserve additions either backdated to the the date the region was originally discovered or the date of the reserve addition.




The idea that discovery has peaked is pretty dependent on backdating reserve growth. That is, ascribing reserve additions in known fields to the date the field was first discovered rather than the date the new technology or price increased the reserve. So if field A initially was assigned a URR of 1 million barrels in 1950 and an improvement in technology in 2000 made it possible to extract 2 MM barrels, the additional "discovery" is credited to 1950 rather than 2000.

Opponents of that view say this gives a false impression of the role of increasing technology and ongoing "discoveries" above, below and around fields know for years.


--
What Hubbert's peak doesn't take into account is:
• Technology that increases the URR
• Scarcity pricing that increases the URR
• Non-homogeneity of the resource that affects flow rates - this is about x-heavy/tar/etc and is relly a separate topic from C+C peak)


Now, as Keith has been saying, even huge changes in URR don't really make for huge changes in PO date. At current extraction rates, a simple logistics curve shows only a 25 year difference in the peak date between a URR of 2 trillion barrels and 4 trillion barrels. Campbell and Laherrere's SciAm paper in '98 guessed 1.8Tb I think, giving a peak date around 2005, even if they were off by 100% that makes a simple Hubbert peak come in at 2030.

There are also some countries, in fact a lot, that show a peak way before 50% of their ultimate production. Texas (who thinks it's a country) was thought to have peaked at almost 60% of URR but with this latest flush from tight oil might wind up moving it's peak back - or maybe way forward, though I doubt that. If someone would like to look that up and find a citation for how many regions appear to have peaked before 50% it would be great.

… Anyway, I don't have the patience to keep saving this to work on later so I'll just throw it up and see what you-all think, tear it up!

Keep it relevant to the point "bell shaped curve" in our Fact Number One.


www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA488967
http://www.econbrowser.com/archives/201 ... nd_pr.html
https://docs.google.com/viewer?url=http ... p12109.pdf
http://www.graphoilogy.com/search/label ... earization
http://www.hubbertpeak.com/fleay/climaxingoil/

[ROCKMAN]

I think it's useful to go back to Hubbert's original work. Over time it has been misused/misstated many times. Above all else he specifically and emphatically stated the model applied ONLY to a defined geologic domain. There's a good reason the DW GOM, North Slope and shales don't fit his model: they weren't part of his model. He specifically states the model is restricted to a specific geologic province. His model was designed to predict only the development and decline of the known geologic trends of the lower 48 states. He states that including any trend outside of his basic assumptions violates the model.

So what has happened? Folks constantly violate the model and construct their own models that try to combine Hubbert's model with different models. Which is OK except when they start referring to their efforts as "Hubert models" because they aren't. But they could be. Take the US shale plays. They represent a unique geologic domain...a domain that wasn't part of Hubbert's original work. The correct approach IMHO is to construct a unique Hubbert curve for the shales as well as the DW GOM, DW Brazil, etc. Of course, won't be very easy since most of these domain are in their early stages of deferment. Again note that Hubbert constructed his model many decades after development of his trends began.

It's little wonder that combining Hubbert's well defined model with other models produces crappy results...they should. Makes as much sense as combining the production curves of the Bakken trend with Ghawar Field and using it to predict the future. And it's worth noting that Hubbert clearly points out the curves won't approach anything close to perfect symmetry. If I recall correctly he specifically states the shape of the post peak curve wasn't predictable. And again I'll point out that the loosely defined "bell shaped" term doesn't mean a symmetric curve as many like to imply. Most of the bell shaped curves of natural data tend to be very skewed.

And lastly I don't recall Hubbert linking his curve to ultimate recovery such that the peak would represent the 50% URR point in time. In fact, I think he has the expectation that post peak production volumes FOR A UNIQUE GEOLOGIC DOMAIN will exceed the pre-peak production. Memories fade with age so please refer me to any specify passage from Hubbert's work I got wrong.

[MD]

So I thought we might work on the 10 basic facts as individual threads.

Right now our fact numero uno is:

1. Oil is a finite resource on a human time scale, unimpeded extraction typically follows a bell-shaped curve with a maximum followed by terminal decline.

"unimpeded" should be removed from the statement. All oil flows are impeded in one way or another. There are so many types it's impossible to get an apples comparison, especially when you consider that each type results in a different mix of refined product and ecological footprint(energy burden deferred!).

[raizcapoeira]

Yes, Hubbert Linearization doesn't work for all liquids (because it employs logistic regression, a data modeling culture, not an algorithmic modeling culture). Why is this even discussed anymore? The level-headed peakists knew that a long time ago, that's why they started using moving away from HL and towards other methods of curve-fitting and bottom-up analysis.

[raizcapoeira]

I think it's useful to go back to Hubbert's original work. Over time it has been misused/misstated many times. Above all else he specifically and emphatically stated the model applied ONLY to a defined geologic domain. There's a good reason the DW GOM, North Slope and shales don't fit his model: they weren't part of his model. He specifically states the model is restricted to a specific geologic province. His model was designed to predict only the development and decline of the known geologic trends of the lower 48 states. He states that including any trend outside of his basic assumptions violates the model.

So what has happened? Folks constantly violate the model and construct their own models that try to combine Hubbert's model with different models. Which is OK except when they start referring to their efforts as "Hubert models" because they aren't. But they could be. Take the US shale plays. They represent a unique geologic domain...a domain that wasn't part of Hubbert's original work. The correct approach IMHO is to construct a unique Hubbert curve for the shales as well as the DW GOM, DW Brazil, etc. Of course, won't be very easy since most of these domain are in their early stages of deferment. Again note that Hubbert constructed his model many decades after development of his trends began.

It's little wonder that combining Hubbert's well defined model with other models produces crappy results...they should. Makes as much sense as combining the production curves of the Bakken trend with Ghawar Field and using it to predict the future. And it's worth noting that Hubbert clearly points out the curves won't approach anything close to perfect symmetry. If I recall correctly he specifically states the shape of the post peak curve wasn't predictable. And again I'll point out that the loosely defined "bell shaped" term doesn't mean a symmetric curve as many like to imply. Most of the bell shaped curves of natural data tend to be very skewed.

And lastly I don't recall Hubbert linking his curve to ultimate recovery such that the peak would represent the 50% URR point in time. In fact, I think he has the expectation that post peak production volumes FOR A UNIQUE GEOLOGIC DOMAIN will exceed the pre-peak production. Memories fade with age so please refer me to any specify passage from Hubbert's work I got wrong.

Yes, his curve was not a bell curve. It was an almost symmetrical logistic distribution curve. Big difference. Hubbert's function wasn't a Gaussian (bell curve).

Excellent point about only plugging in known geologic domain. The Hubbert curve wasn't an absolute. It's just a predictive model employing logistic regression.

[ROCKMAN]

R - "HL doesn't work for all liquids". But it's not suppose to, does it? It shouldn't work when combining different oil producing domains either, should it? Perhaps I missing the point but as I just mentioned elsewhere Hubbert defined his US model to apply ONLY to the known and fairly mature trends of the lower 48 states. Of course it didn't predict gains from the DW GOM or the shales...they weren't part of the model. Some of what's said about Hubbert's work is similar to someone taking a production model for Ghawar Fld, combining it with a model for Mexico's Cantrell Fld, and then use the combined crappy combined model to argue that the Ghawar Fld model was flawed. The US shales as well as the DW GOM fields deserve their own separate models. They are unique geologic domains very different than the one Hubbert based his original work upon. Layering those models onto of Hubbert's original model violates one of his most clearly stated premises for his model: it is restricted to the known geologic domains he derived his statistics from.

Wouldn't it be nice if someone bothered to present the current plot of Hubbert's model that only includes the data it was designed to include?

[raizcapoeira]

R - "HL doesn't work for all liquids". But it's not suppose to, does it? It shouldn't work when combining different oil producing domains either, should it? Perhaps I missing the point but as I just mentioned elsewhere Hubbert defined his US model to apply ONLY to the known and fairly mature trends of the lower 48 states. Of course it didn't predict gains from the DW GOM or the shales...they weren't part of the model. Some of what's said about Hubbert's work is similar to someone taking a production model for Ghawar Fld, combining it with a model for Mexico's Cantrell Fld, and then use the combined crappy combined model to argue that the Ghawar Fld model was flawed. The US shales as well as the DW GOM fields deserve their own separate models. They are unique geologic domains very different than the one Hubbert based his original work upon. Layering those models onto of Hubbert's original model violates one of his most clearly stated premises for his model: it is restricted to the known geologic domains he derived his statistics from.

Wouldn't it be nice if someone bothered to present the current plot of Hubbert's model that only includes the data it was designed to include?

Yes, that was precisely my point -- that it doesn't work, and it's not supposed to. Methinks the future of Predicting all liquids is going to be simulation modeling. Models based on quite disparate assumptions (e.g., physical simulation vs. economic optimal depletion) have produced approximately bell-shaped production profiles, but data do not support assertions that any one model type is most useful for forecasting future oil production. In fact, evidence suggests that existing models have fared poorly in predicting global oil production. The greatest promise for future developments in oil depletion modeling lies in simulation models that combine both physical and economic aspects of oil production.

[Pops]

Actually a little off of our topic about our First Fact but Hubbert spent the first paragraph of his Nuclear Energy And The Fossil Fuels talking about Columbus discovering continents then as time went by others only discovered islands until finally the only thing left to discover were the intricacies of the shorelines. Not only does he predict world peak around 50 years hence (That would be March 8, 2006 if you're counting) he talks about coal and gas and <Gasp!> economics in that paper as well.

If the world should continue to be dependent upon the fossil fuels as its principal source of industrial energy, then we could expect a culmination in the production of coal within about 200 years. On the basis of the present estimates of the ultimate reserves of petroleum and natural gas, it appears that the culmination of world production of these products should occur within about half a century, while the culmination for petroleum and natural gas in both the United States and the state of Texas should occur within the next few decades.

This does not necessarily imply that the United States or other parts of the industrial world will soon become destitute of liquid and gaseous fuels, because these can be produced from other fossil fuels which occur in much greater abundance. But it does pose as a national problem of primary importance, the necessity, both with regard to requirements for domestic purposes and those for national defense, of gradually having to compensate for an increasing disparity between the nation's demands for these fuels and its ability to produce them from naturally occurring accumulations of petroleum and natural gas.

--
But that really isn't the point here, I'm not whupin up on Marion or defending him (even though I think he did a bang up job - we have a thread about his paper over here in fact). What I'm trying to find is the Number One basic fact about peak oil (taking oil as finite of course).

[Pops]

Thanks MD. I agree we should strike out the unimpeded part.

In fact I think we should not try to describe the curve beyond the basics

How about this:

Fossil fuels are a finite resource; their extraction starts and ends at zero with a peak somewhere between.

[kuidaskassikaeb]

Um, if we actually have enough resources to fry the planet before we run out, they are, for all practical purposes, infinite.