Why Euan, you almost sound like...

nevermind :P

The advent of new technology - mainly horizontal drilling and 3D and 4D seismic has led to a justifiable expectation of higher recovery factors over the years.

I'm curious as to why these technologies have not turned around the Texas and North Sea declines.

BTW, a question I have been meaning to ask you. Were there ever any material restrictions on drilling in the North Sea?

I'm not going to question your eyesight, but I will question the ability of the HL method to "see" how much oil is in Haradh (or Khurais, etc.).

Similarly, most reservoirs in the lower 48 (the East Texas field being an exception) have rather low recovery rates, mostly because they were produced early. If you and oilmanbob and thousands of others decide to go after a lot of that oil using EOR wo whatnot, won't that give a different production decline profile and a higher URR than predicted by the current HL plot? Now, there's no way Texas or the lower 48 are going to come close to re-peaking, but that is not the only consideration.

I don't disagree with your message. I just question one of your sales pitches.

Note my question up the thread--Why hasn't better technology reversed the Texas and North Sea production declines?

IMO, regardless of whether we are Capitalists, Communists or Britney Spears Worshipers and regardless of whether it is Texas, the North Sea or Saudi Arabia, we generally find the biggest fields first. A production peak does not mean that we stop finding fields and it does not mean that we stop trying to extract the last barrel of recoverable oil out of the reservoir.

It does mean that we can't offset the declines from the old, larger oil fields.

That is my point about using the wildly different post-peak Lower 48 and Russian production profiles. The post-1970 and post-1984 cumulative production numbers were basically what the HL models predicted they would be.

The thing about HL is it tends to give you a URR not surprisingly based on how the fields are developed. Its insensitive to the exact nature of the development except that the growth should reasonably follow the assumed population like pattern. It need not be maximum growth etc etc. It also probably underestimates the amount of oil extracted in the tail end of a fields life when more aggressive methods are used. But since its primary utility is near peak and post peak its faults probably don't effect its "predictive" power.

Its very much a sledgehammer and a simple tool but considering the uncertainties involved its seems to be a pretty good tool. If HL is pointing to peak you need in my opinion some strong evidence that the region is not close to peaking. Its simplicity is both its strength and in a sense its weakness. In general people that object seem to not understand it. Its a simple model of the exploitation of a resource that makes the assumption that future exploitation of the resource will follow previous methodology. I.e the future is like the past and any slowing in the growth rate of production is caused by depletion of the underlying resource.
Thus its a fairly sensible simple model and useful in the case of incomplete data. So far none of the arguments against HL are that strong in general its the bias of the people that dislike the model that reject it.

And finally the URR calculated but HL and by other measures is probably not the same number. This is easy to see for example since people that dismiss HL are assuming a massive investment will be made as a field declines to increase URR and also enhance production rates while HL does not make this assumption. This is why it does not capture the flat peak we find in practice or the so called dogleg. The assumption that the future is the same as the past fail at peak but it becomes strong afterwards.

Anyway I'm not going to debate HL the facts will become clear soon enough and with issues such as Export Land and the potential for the oil industry itself to crumble etc etc
I don't see the projected production rates as relevant now.
So I think HL's and all the other depletion based models are either irrelevant are close to irrelevant now. Politics, war, weather and economics will determine how much oil is produced from now on out and I expect the real amounts will be far less than predicted by depletion modeling. We are going to kill the golden goose.

there has been a paradigm shift in OPEC's view of the market,

I don't think there has been much investigation of this and I believe it may have significant explanatory power.


OPEC went through an early post-nationalization phase which was the equivalent of the kids finally getting the keys to the bank vault. There was little or no market discipline and the attempt was made to maximize immediate returns.


OPEC now appears to have significant market discipline and I suspect this is due to increasing financial sophistication and their own recognition that they have a finite resource and need to manage that finite resource to maximize value extraction. They see themselves as pumping dollars, not oil, and will act as necessary to maintain the value of the income stream.


I suspect that within 6 months we will see OPEC taking more oil off the market in anticipation of a US/UK recession.

The assumption that production needs to be maximized to do a valid HL prediction is just that and assumption with no supporting evidence.

Please explain how HL - or any method - is supposed to correctly divine the amount of oil in the ground if the rate of production - the only data it's given - is determined by economics rather than geology.

HL isn't magic; it follows the same "garbage in/garbage out" rule that every other algorithm does, and production data that is determined by fiat rather than geology is - when looking for geological limits - garbage.

Moreover, RR's theoretical examination of HL pointed out quite clearly that intentionally-constrained production will cause HL to give too-low results. It's pretty easy to see for yourself, though. Suppose a field comes online producing 1 unit of oil per day, and is developed in one of two ways:

1. 10% production increase the first year, 9% the second, 8% the third, and so on.
2. Production intentionally constrained to 1 unit per day.

If you graph that, the first approach will show a nice peak shape, and using data from years 6 through 16 gives a nicely-fitting HL that predicts an URR of 32 (vs. a true URR for that series of 36; not too bad when predicted from a currently-produced value of 24).

By contrast, using data from years 6 through 16 of the intentionally-constrained production gives a nicely-fitting HL (R^2 of 0.95) that predicts an URR of 22...which the other well management scheme has already exceeded.

And it's not like this second well development approach is somehow unreasonably unrealistic - the first approach is producing in excess of 1.5 units/day for the entire data period, so there's no difficulty in achieving 1.0 units/day with the constrained well. Moreover, the unconstrained well has a production shape (rise-peak-fall) that looks very similar to the "theoretical ideal" from Hubbert's 1956 paper, so this seems like input data that HL should be able to handle.

But constrained production makes it fail utterly.

First how HL works has been a bit of a mystery. Or better why it would work. WebHubbleTelescope (WHT) has made some great comments on it. Its a birth death models what the heck does it have to do with oil ?

So I've proposed this is how it works.

Its measuring the life/death cycle of wells. They are born when they are drilled and die when they water out. So this pins the logistic equation to something that follows a life death cycle. Production rates are a good proxy for the life and death of the average well.

Next how the heck does this relate to reservoir size or URR.
The key insight here is that the changes in production become dominated by the rate of death of wells in a mature field and if the rate new wells are created is consistent then you have a heuristic to guess the point at which all well would be dead. The critical factor is that the change in production is now primarily because of the death of wells and new wells are being consistently added even though the rate production is increasing is beginning the slow. This means the field is well developed. You don't have to maximize production just be consistent.

The critical factor is that changes in production from births is now linear and non-linear effects are caused by the death of wells. This is related directly to the size of the reservoir. And easy way to think about it is drilling holes in a damn dead holes occur when the water level drops below the hole. Once you drill enough holes and watch the flow rates its fairly obvious that you can estimate how much water is in the reservoir. Or you could consider drilling holes in a bucket. This is actually close to classic calculus problems. And you can see that the critical factor is that the rate of change is governed by "dry" holes and I need to be consistent with how fast I drill new holes. So I can discount the effect of new drilling.

Now when do these conditions hold ?

1.) Once the rate of increase in production begins to slow.
2.) As long as the drilling campaign is consistent.
3.) When well deaths contribute a significant amount to the change in production.

And when does it not hold ?

1.) Early in the life of the field before its fully developed and well deaths are not important since this is what we are really measuring.

2.) During periods with the drilling campaign changes drastically. This is common near the peak in production and leads to changes in the field development in attempts to maintain and increase production.

3.) Slightly post peak you probably still have a extensive campaign going.

So this means you have two places to apply HL. On the upslope before the peak and on the down slope after the peak. The caveat is the downslope URR may be overstated if advanced technology has increased extraction rates but eventually it will snap back to the original HL plot as wells die faster.

And finally what is this URR ?
Its the reasonably recoverable URR it does not include the long tail 2bpd type production. Since well life/death is under economic control. This is a lot of oil thats not in the HL URR estimate and I'm sure its the cause of a lot of trouble. The HL URR should be considered the URR of semi-easy oil not total URR. The total is at least 10-30% higher but the last 10-30% will be recovered at low production rates with minimal economic gain. The reason its not in HL is that the rate of well death and the nature of the wells changes dramatically when you go to stripper wells. HL is measuring effectively low water cut production during primary/secondary production. Consider how much of the US production came from stripper wells and you will see why you have a disconnect between HL URR and other ways to measure URR. They are not the same variable !!!!

Now the key point is West Texas correctly did the HL analysis. From reading his posts I think he basically followed the same line of reasoning. He knew to correctly reject the dogleg and did the right thing and focused on steady production. Kudos.

If you don't believe in HL then use the following guidelines for your HL plots and lets look at where it fails and why.
I think we have enough data to prove/disprove this physical explanation for HL.

WT its a absolute myth that HL only works when you maximize production. I have no idea how that got started it has no basis. I outlined why I believe your numbers you did it right.
I suspect you may have not understood why your right it took me a while to figure out what HL is measuring. But I'll repeat HL gives URR under the following conditions.

1.) Production has begun to slow because individual wells have reached the end of their productive life.
2.) The drilling campaign is effectively in a steady state i.e well are "born" under a set of constant conditions. This need not be to obtain the maximum production rate. The critical factor is production is dominated by the rate wells are closed in.

HL is a life death model of the drilling campaign and gives URR if the campaign is fairly steady and changes in production are from closing in wells and thus pretty much directly caused buy the underlying URR. If you break production into four periods initial, mature, post peak, and mature decline. HL should be used in the mature phase or mature decline ( new steady drilling state). It is not a good estimator of URR during the initial phase and peak periods. So I think Euans basic assumptions of maximum production as a prerequisite for HL are false and unsubstantiated.

You did it right and correctly discounted the dogleg up phenomena. The dogleg is probably better termed the Ohh S&*t period.

Finally the URR from HL is probably best described is the amount of oil that can be reasonably extracted most advanced extraction method will work to move this forward in time but the total URR is probably correct for commercial extraction.
The bottom up methods tend to give the utopian extraction rates and thus are in my opinion less useful than HL. But the point is the URR from HL is not quite the same as URR from bottom up approaches its a slightly different and my opinion far more realistic URR. A lot of the excess oil that is part of the URR from other methods would be produced in a long tail at low production rates so it distorts the issues surrounding peak and the near term post peak world. We could care less how much oil is being produced 30 or 60 years from now since it will be a very different society at that point.

Not to get too far off tangent - but I grew turnips this year- they grew ok but they tasted lousy. I left them for the deer and even they didnt eat them. However, my 37 tomato plants have produced a cornucopia. And i mean it. And I am SO glad I grew them, not because the sky is falling, but because a)I enjoyed it b)I gave several hundred away to new neighbors, and c) they taste fantastic!

There are three parts to ELP.

Economize, Try to live on half of less of your income.

Localize, Reduce the distance between home (preferably small energy efficient housing) and work to as close to zero as possible.

Produce, Try to work for, become (or invest in) a provider of essential goods and service.

The "P" part does not necessarily mean growing your own food. It does mean that huge numbers of people who survive in the discretionary side of the economy because of the efforts of the food and energy producers are in deep trouble.

As I said up the thread, I could be wrong about Saudi Arabia, but let me know when their average annual crude + condensate production exceeds 9.6 mbpd.

Texas and the Lower 48 as a Model for Saudi Arabia and the World (May, 2006)
http://www.energybulletin.net/16459.html

In summary, based on the HL method and based on our historical models, we believe that Saudi Arabia and the world are now on the verge of irreversible declines in conventional oil production. While there will be massive efforts directed toward unconventional sources of oil, we predict that unconventional sources of oil will only serve to slow and not reverse the decline in total world oil production.

Get back to me when KSA returns to 9.6 mbpd, ok? What's amusing here is that WT and Ace have been consistently closer to the reality of the situation than any other forecasts yet people argue that the more optimistic forecasts must be true because... because... because... why? Because you say so? Like I said, get back to me when the data demonstrates your position but not before.

Decent evidence he is wrong? What evidence? There is none, as Euan himself has admitted in the past. There is simply a decline that people have tried to force fit to OPEC cuts totally ignoring that KSA was falling for months before any OPEC cuts were even announced!!! Ah, but we can ignore that inconvenient fact, can't we? We can ignore that KSA production began falling before prices fell, can't we? Because we want to, because it supports our theological position, because the alternative is unthinkable. Well think again, Asebius.

"The greatest shortcoming of the human race is our inability to understand the exponential function." -- Dr. Albert Bartlett
Into the Grey Zone

"Regarding ELP, I'm not sure all those folks who moved to the country and are now farming turnips on your advice will thank you. Just bought one for the same price as they were sold last year. They didn't go up."

Hmmm...wheat peaked at $7.50 a bushel..up from $3-4...what a couple of years ago...???
Hay is up too...almost double...
(edit) The latest edition of Capital Press(an Ag weekly)- front page- blue berries up to $1.90 a lb. That is up from $.83 a lb 5 years ago. Retail is $3.00 a PINT!!
I never liked turnips anyway and I do like fresh bread and a nice juicy steak with sea salt on the BBQ...
Get ready to bend over as biofuels sucks money out of your wallet...(oops no oil production problems... .-)

DelusionaL - happily sitting drinking wild turkey and looking to add...oh my god!...vegetable/berry production as a way to stradle the transition between the life as we know it (optional purchases of pretty yard plants) and what must come next. Best of luck to you and remember I think it takes 3 years to learn how to grow a consistently predictable crop year in and year out...and as my competitor found out maybe longer...;)(hehe)

But how can we ever realize the waste that we have engineered into the auto-life-style? Based on a finite resource. We "cannot" be that stupid...(?)

That math applies to oil production/human behavior isn't a stretch. It applies to the world around you, planets, plants(my field), etc. etc. So why are we so convinced that we are above being a part of the equation? arrogance?

Oops... the weenies need turning on the propane fire...and when it runs out I will cook with wood...weenies because the deer was full of tape worm cysts, 1-3 per back strap steak...yup just that gross.

Hmm curve fitting is all about a choosing point to fit.
HL is a simple method more rigorous ones exist. But curve fitting is still more art than science. If your serious about finding a better method then HL its fairly easy to do non-linear least square/filtering approaches. I looked at them and the fit provided by HL and felt that they did not provide and advantage considering the quality of the data.

One of my biggest problems with this post is it did not show the actual curves given the various HL fits. We are pretty good at doing visual fitness tests with the mark one eyeball.
I dislike posts that don't show the curve and data together. The HL plots are not that useful. Also their is no reason you can generate a non-linear least squares fit number given the HL to get a error term. The lack of mathematical rigor when people question HL drives me nuts.
I'll just shut up on this subject if the objections to HL had more merit and where better formed we might have a worthwhile discussion instead of dealing with people who don't even understand the fundamentals of curve fitting.

Show me one non-linear least squares fit and then I'll change my tune lets at least start with the basics.

There is certainly some truth in the notion that one could choose their date in an arbitrary fashion and "create" a line pointing to where they would like. Many of the HL analyses of KSA start with 1991 and use all the data to the present. Whether we use C+C or C+C+NGL will shift the line leftward and change the slope somewhat. I have done a number of these plots with C+C data from the EIA and in my opinion a more reliable estimate of URR is found by using the monthly data from 1973, rather than just the annual data. We have the data for C+C, but only annual data for NGL so I created a dataset using the monthly data for C+C and adding the average NGL output (in thousands of barrels per day) for each year to the monthly data for C+C to see what the HL would look like for C+C+NGL. Why would I bother?

The benefit of adding the monthly data is a much narrower 95% confidence interval and a better estimate of URR.

SD=standard deviation
Data from EIA "All Months and Years January 1973-Present
OPEC Countries and Total OPEC" (C+C) on the following page http://www.eia.doe.gov/emeu/international/oilproduction.html and "Selected Countries, Persian Gulf, Total OPEC, and World Total, Years 1970-2006" for NGL (same page as above).

The upper line (yellow) is pretty close to the HL that Euan uses with a URR of 241 GB, this corresponds to the Upper 95% line of the regression. My central estimate gives a URR of 217 GB and the lower line (lower 95%) gives a URR of 196 GB. The data from January 1991 to May 2007 was used for the regression, but all data from Jan 1973 to May 2007 is shown on the graph. R squared was 0.79.

Note that a similar analysis on the annual data from the EIA was performed (data from 1991 to 2006 was used for the regression) and the central estimate for URR was 236 GB. The lower esimate in this case was 160 GB and the upper estimate was 383 GB. R squared was 0.76. This analysis was performed using 16 data points (vs. 195 data points for the monthly plot), typically fewer data points will result in a poorer estimate.

One further comment on Euan's suggestion that KSA was producing below capacity from 1992 to 2002. Although KSA may not have been producing flat out, I do not think they were holding back a lot of capacity since 1991 (altough in the eighties clearly this was the case.) The two points that Euan chooses (1991 and 2005) may have been years that KSA was producing above their sustainable capacity. Gulf War I was the reason in 1991 (loss of Kuwait's and Iraq's output) and the attempt in 2004-5 to bring oil prices back under control by ramping up output to maximum. They gave up on this in the fall of 2006, probably because this level of output was not sustainable. Just my 2 cents.

I've written elsewhere that I suspect that too many people are expecting too much from Hubbert Linearization (particularly with regard to predicting URR) but your assessment of the matter is not, in fact, the only way to choose data points. Another methodology has been recommended in the past and it does not involve a fixed number of data points but rather in taking all those data points from the time that the HL plot stabilizes and going forward. Yes, this gives you a range of URRs but so what? Ask 10 geologists to give you their URR of a field and you will get 11 different answers, though they are all likely to be similar. Likewise with HL, you get a range of values that gives a strong indication of the possible URR.

Now, as I have said, I have strong doubts about HL because that is not the method that Hubbert himself employed. HL is a simplification of Hubbert's equations developed by Professor Deffeyes. If you actually bother to read Hubbert's 1956 paper, you discover that he was working from prior URR estimates, not developing the estimates himself. Thus I personally think that using HL to predict URR goes beyond what was intended with Hubbert's original work. However, even though I disagree with HL as a URR predictor, the fact remains that others have proposed other objective criteria that do not presuppose fixed numbers of data points, such as taking all data points once the linearization stabilizes.

"The greatest shortcoming of the human race is our inability to understand the exponential function." -- Dr. Albert Bartlett
Into the Grey Zone

Excerpt of a post over on the EROEI thread:

I asked the following question a few months ago:

A man is driving a car for two laps on a one mile circular track. He averaged 30 mph on the first one mile lap. How fast does he have to drive on the second lap to average 60 mph on both laps? The answer is infinitely fast, because it is really a time question. A average speed of 60 mph over a distance of two miles takes two minutes, but the first lap consumed two minutes, so it is physically impossible for him to drive fast enough on the second lap to average 60 mph on both laps.

I used this analogy in regard to the Texas oil production peak and decline. From 1972 to 1981, nominal oil prices went up by about 1,000% and the industry responded with the biggest drilling boom in history, which resulted in a decline in production from 3.5 mbpd in 1972 to 2.5 mbpd in 1982.

In other words, how fast did Texas have to drill in order to keep oil production rising? The answer is that it appears to be physically impossible.

We have seen a similar situation in the North Sea, where we have seen a crude oil production decline rate of about 5% per year since 1999 while oil prices have increased at about 18% per year.

IMO, Saudi Arabia--which is currently showing lower crude oil production in response to higher oil prices and increased drilling activity--is at about the same stage of depletion as Texas in the Seventies.

While we can make money finding smaller oil fields, the case histories and mathematical models suggest that we will not be able to increase our aggregate conventional oil production, i.e., the function of oil companies post-peak regions is to slow the rate of decline.

Jeffrey and Ron, in the model I have presented there is 2 Gbs production from yet to find fields produced between 2017 and 2028. And there is 2.2 Gbs production from discovered undeveloped fields from 2015 to 2028. Do you think these are reasonable guestimates?

And would you care to comment on the other elements - Ghawar and Heritage assets - and say where you think I am overestimating.

And whilst as you say, all 1 mmbpd fields in the world are in decline, in Saudi Arabia, Khurais has lain fallow for decades and is forecast to produce at over 1 mmbpd - so there you have a 1 mmbpd field that is not yet started to produce (ignoring half hearted efforts uisng vertical wells in the past).