Help with deconstructing the EIA 2037 peak curve?

Sparaxis

I'm sure you're familiar with the unfortunate projection that EIA put out last year :

This is the basis of the 2037 peak that is so often referenced as the US "official" view of the peak.

I find it laughable to posit a global depletion curve based on R/P=10.

(But I find it interesting that even with their extra trillion barrels, a "normal" depletion curve peaks in 2016)

But equally unfortunate, the director of our national lab uses this same graph in his presentations urging more funding and R&D into carbon-neutral alternatives such as cellulitic ethanol. I've engaged him in a chat about the unsuitability of the EIA projection (he's a physicist and not a geologist), but I wanted to add some more facts to my arsenal. Basically, I'm wondering what percent of the URR (3,003 bill bbls) is represented under the curve by the peak in 2037? Is it way beyond 50%? 60% or more? Could any of you whiz-bang modelers with the historical production series be able to pick up from 1999 (looks like the last historical year they have used) and run with their assumptions? There's other evidence I can use with this to show that other oil fields, countries, and even other resources (e.g UK coal) decline at about midpoint of depletion.

Thanks for any help!

EnergySpin · Fission Joined: Jun 25, 2005 Posts: 2381

Hello Sparaxis,
Check the updated Verhulst model thread, we covered pitfalls of estimation using nonlinear regression, and the technical caveats of prediction when all the parameters are unknown. You will find various formulas that will help you answer that question there.
Basically:
- EIA are f..king around with nonlinear regression tools that they barely understand. The same can be said about all the people predicting the peak.
- there are ways to use the data series to come up with any number you want, and in fact we have been able to reproduce the predictions from the various groups by interfering with the stopping rules of the Levenberg Marquadt, Gradient descent estimators
- if you check the ASPO conference proceedings in Lisbon this year you will find that the issue was addressed (check their web site there is a link there)
- the time of the peak can be estimated more robustly than the URR (at least with an error of +/- 4 years
Hubbert himself fitted the data by hand by the way, and he was wrong on the URR of NA.
- it has been found (based on the US data) that the 99% curves are more reliable than the "mean" curves due to the bias of the parameter estimation procedure. That bias is probably a function of the dataset btw.
- Unless people decide to play with Stochastic Differential Equation, mixture modeling approaches (btw there was such a paper in ASPO) we will not achieve better accuracy
- EIA is wrong, Peak is here
_________________
"Nuclear power has long been to the Left what embryonic-stem-cell research is to the Right--irredeemably wrong and a signifier of moral weakness."Esquire Magazine,12/05
The genetic code is commaless and so are my posts.

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

Here some results for the first model:

cumulative production in 1999: 907.36 Gb
cumulative production between 2000 and 2016 (2% growth): 538.89 Gb
cumulative production between 2017 and infinity (2% depletion): Pmax* (1 - alpha) / alpha= 36.97 * (1 - 0.02) / 0.02= 1,811 Gb

URR= 907.36 + 538.89 + 1,811= 3,258 Gb

which is not far from the USGS URR. If we impose URR= 3,000 Gb, the PO date would have to be between 2012 and 2013.

PO production value= P(2016)= 36.97 Gb
PO cumulative production/URR= 44%

Production midpoint= 2022
_________________
______________________________________
http://GraphOilogy.blogspot.com

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

For the second model:

cumulative production in 1999: 907 Gb
cumulative production between 2000 and 2037 (2% growth): 1,511 Gb
cumulative production between 2037 and infinity (P=R/10): 582 Gb

P(2037)= 56 Gb

PO cumulative production/URR= 81%

Production midpoint= 2018

I assumed URR= 3,000 Gb which means that R(2037)= 3,000 - 907 - 1,511= 582 Gb. Therefore, P(2038)= 582 / 10= 58.2 Gb which implies that the peak is in fact in 2038 (with this particular URR).

By construction, the URR has to be 3,000 Gb.
_________________
______________________________________
http://GraphOilogy.blogspot.com

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

Clearly, the first model is the most "credible" with a near symmetric curve and a midpoint near the PO date. The second model is really unrealistic with half of the URR consumed 19 years before peak production!
_________________
______________________________________
http://GraphOilogy.blogspot.com

Sparaxis · Posted: Wed Jul 13, 2005 12:11 am Post subject:

Thank you!!! Both for the references to the various Lisbon conference papers and for running the numbers. 81% URR consumed at peak on the R/P=10 decline curve is astonishing.

But not much more astonishing than Guy Caruso's (EIA director) presentation at the Asian Oil and Gas conference in Malaysia a few weeks ago. I would have been so embarrassed for him to have heard his speech. In it, not only did he repeat all this, but he added in all the Canadian tar sands and Venezuelan bitumens and treated them as conventional petroleum, added them to the USGS 2000 numbers, and created a new peak in 2065 (2% growth) or 2094 (1% growth). His production curve for tar sands alone had maximum output in 2078 at over 112 mmb/d!

Thanks guys.

EnergySpin · Fission Joined: Jun 25, 2005 Posts: 2381

LadyRuby · Posted: Wed Jul 13, 2005 7:14 am Post subject:

It's so ridiculous! How do you like the nice cliff at the end? Like we're going to use oil like it's water, not worrying about conservation, etc. then suddenly drop off the face of a cliff.

This actually is a good argument for drastic measures now to survive the free fall!!

Doly · Expert Joined: Dec 03, 2004 Posts: 4035

You could also say that simply eyeballing the curve it doesn't look credible for the following reasons:

1) The area under the curve post peak is visibly smaller than the area pre-peak (make a rough count of squares if this is not obvious enough by simple eyeballing).
2) The peak is a very pointy peak. This is quite unlikely, as it would mean production somehow ramping up at full speed before falling dramatically. I don't think the physics of oil drilling allow for that.

pup55 · Expert Joined: May 26, 2004 Posts: 3695

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

If we assume a EIA like scenario:
- 2% annual demand growth
- 2% annual depletion rate
I assumed different PO date between 2000 and 2040 and I derived afterward the corresponding URR, PO maximum production rate and % of the URR before peak. Therefore, I generate a subset of annual production scenarios matching the EIA requirements. I obtain the following figure:

The EIA scenario corresponding to PO in 2016 is given by the red dot. If we take the last URR from the 2005 BP ernergy review (2,235 Gb), we have already passed PO since 2003. Assuming an URR equals to high case USGS estimate (3,896 Gb) we have a peak in 2025 which is what the CERA report is claiming. The ASPO URR estimate is about 1,823 Gb which would have put the PO date far before 2000 (again assuming a 2% depletion). It means that the ASPO estimate requires a far greater depletion rate (probably around 5%) in order to put the PO date beyond 2004-2005.
_________________
______________________________________
http://GraphOilogy.blogspot.com

pstarr · Posted: Wed Jul 13, 2005 1:29 pm Post subject:

What is R/P=10 and what is it's significance? Why is it a ridiculous value?

thanks, pete
_________________
director

ree rah rip ram. sunofabitch godamn. hidey didey christ almighty. rah rah crap wav

khebab · Moderator Joined: Sep 27, 2004 Posts: 935 Location: Canada

MD · Posted: Wed Jul 13, 2005 3:04 pm Post subject: