deleted duplicate

Another masterpiece by Rembrandt as he pencils in the details of reality!

The difference in vision between so called "optimists" and pessimists" with respect to the peak in world oil production is often caused by a view of future technological development in the oil industry.

For some reason that reminds of this painting by that other Rembrandt... can you tell who the optimists and the pessimists are in this painting?
http://intranet.haef.gr:61/dilaismas/024.jpg

Very clear summary, thank you! I will be presenting this conundrum to my energy class in a couple of weeks so look forward to the rest of the series and the inevitable "if I told you what I know I'd have to kill you" comments from reservegrowthrulz2 and his USGS analyst buddies.

The average recovery factor plotted in graph 6 that the Saudis project is not necessarily off-scale if one allows delivery over a very extended timeframe. So the question for TOD experts remains can AramCo really manage reservoir depletion to maintain Saudi production? And the question for IEA/EIA remains who else can build on that shaky base to boost the current global plateau to meet the needs of new customers?

Forgive me if this has been rehashed endlessly, but does boosting both recovery and flow rates specifically from SA secondary fields require proven tech beyond the sophisticated water flood drilling and reservoir visualization that they have mastered for their "oldies but goldies"?

I'll point out again what's been previous noted as a very flawed assumption IMHO that US oil recovery is a good model for global oil recovery. Above all else the character of the companies producing the great majority of the last fraction of oil in the US is a polar opposite of most of the rest of the world. US production is dominated by small independent companies (who produce more oil than all US Big Oil combined) who, one average, produce less than 10 bopd per well from privately owned lands. I don't have the stats for the global comparison but it's easy to assume it's significantly greater than 10 bopd/well. Last time I saw number for Saudi they were doing between 3,000 - 5,000 bopd per well. Additionally most of global production is from govt owned lands.

US stripper production is very labor intensive. The smaller margins typically do not allow a large over head....they don't pay much in the way of salaries. Thus stripper operators tend to exist on the basis of sweat equity. A real life example: south of San Antonio, a husband/wife team produce about 25 bopd total from 20 wells. They pay no salaries: both handle all the operations. Might not sound like much but at current prices they are producing over $700,000/year. Multiply that by thousands of similar operations and you see why more than half of US production is attributable to these companies. More significantly, with re: URR, their field is a gravity drainage reservoir. They've made that same 25 bopd for the last 30 years. And those wells will continue to make that volume for the next 50+ years. Just a rough guess but URR for this field could easily exceed 80%. but there's a big catch: it could take about 100 years to reach that number. This is another reason using US URR can be very misleading: the time factor. This is a major assumptive error IMHO: NOC's are not going to bleed the last X% of recovery from their fields over periods of many decades: the profit margin will likely to be just to thin for them to pay for the over head. And without an independent entrepreneurial class there will be no one to take over for the NOC's.

Re: future US EOR efforts - I see nothing of any significance on the horizon as far as new technology/chemicals that will change this game to any significant degree. Every field that could benefit for EOR is undergoing EOR today and, in the majority of cases, have been for decades. CO2 EOR is often touted as a big potential source for NEW EOR. It isn't IMHO for a simple reason: if an operator has commercial access to CO2 and his field would benefit then it's already being done. And if the field doesn't have a source for the CO2 then it's not very likely it will in the future. As far as future improvements the service/chemical companies have spent 100's of millions of $'s over the decades developing better EOR. At the moment none of these companies are pushing anything that amounts to a step change. And these same companies daily offer the various technologies to the NOC's around the world. If the NOC's aren't taking advantage of these state of the art EOR methods there is a good reason IMHO: it's not worth it for them. That might change in time. But I suspect when/if that time comes the prime motivation will not be to increase exports but to satisfy internal demand. And the major fields, like Ghawar, have had EOR methods applied for decades. Some improvements could develop but again the question is how significant and, more importantly IMHO, what is the time? Increasing Ghawar URR by 15% would significant. But if it takes as long to produce that 15% (very typical of EOR methods)as it did to recover the first 50% or so than what would be the impact? I see this as the same problem the "drill, baby, drill" folks have: increased US drilling will add reserves (and that is a good thing). But the flow rate gain of these reserves won't likely be a game changer in the bigger supply/demand picture IMHO.

I think this is a great topic to revisit since no one seems to understand the main idea. Rembrandt says that "The only institute that has done exensive studies with respect to the growth of recoverable reserves over time is the United States Geological Survey." , but doesn't delve into their own analyses which were originally covered by Attanasi & Root, Verma, and others of the USGS. They essentially gave up on an explanation apart from applying some questionable heuristics.

Much of this reserve growth is purely a statistical phenomenon.

Any time you have a level of uncertainty in some measurement and all you have are indirect means of estimating, conservatively you will always have a phenomena such as reserve growth.

It is very easy to show that with a degree of uncertainty given some constraints (i.e. applying the maximum entropy principle) you will get what is called a hyperbolic reserve growth curve. Laherrere has consistently referred to this empirical observation without ever giving a derivation for it.

I have derived this a while ago in a top-level post on TOD. The basic gist in this situation is that you have an initial uncertainty of how far along searching the volume you are at and you also have uncertainty in how much to try to probe. Whenever you have two uncertainties that act as a ratio of probabilities you end up with what is called a ratio distribution. The ratio distribution for the simplest case of maximum entropy in these uncertainties gets you the hyperbolic distribution, which looks like:

U(x) = U0/(1+L/x)

where x related to the amount of perceived amount of the volume probed, and U0 is the initial estimate of the recoverable amount, with L representing he median value of volume probed. The amount of reserve growth is a conservative indication of where you are at on this curve, placed a few years in the future. It is never reported as the UO since the USA requires conservative estimates, and that is actually a bit of a WAG. Note that the middle east OPEC countries will always report U0 since all we see are step functions in the reported reserve amounts. They don't bother to give the smooth transition, quite opposite to the USA's practices. This is very important to understand the continuous transition observed by the USA versus the step functions reported by SA and others!

The key thing about hyperbolic reserve growth is that you can be anywhere on the curve. Get on it too early where you have the greatest uncertainty and it looks like you have quite a bit of growth. Later on where the uncertainty goes down, you have much less reserve growth. The thing that makes it different than an exponential profile is that it is much more gradual and actually falls into the fat-tail category of statistics. That is why people have a lot of problems with it as well, as it is not amenable to conventional statistical measures. For example, a hyerbolic curve in time doesn't have a variance or even have a concept of expected value, as in average time to reach the asymptote! It does however have a concept of URR because you have to supply some sort of initial estimate, the only issue is how slowly you will climb this curve.

The ideas behind hyperbolic growth come from laws of probability, maximum entropy, and Bayesian reasoning. The only way that it would not apply is if you can actually physically measure something by building up enough constraints. When your guiding constraints have uncertainty in themselves, you will always get hyperbolic growth.

Here is an example of one of Laherrere's hyperbolic curves. A creaming curve only differs from a reserve growth curve in that the number of wildcats takes the place of time (I am likely the only person that has pointed this out):

Note the glitches that occur when either new sources are discovered or new technologies are applied. These become new estimates in the original U0 estimate.

The beauty of this approach is that it will lead directly to the Logistic sigmoid when you apply an accelerating exponential rate to the search function. That is, you replace x with exp(rt) where r is a search rate and t is time, and you get this:
U(t) = U0/(1+C/exp(rt))
which is exactly the Logistic that King Hubbert suggested long ago. I refer to the general form that covers both reserve growth and discovery as the Dispersive Discovery Model.

There is also a way to "linearize" the constant growth reserve growth curves, analogous to Hubbert Linearization
http://mobjectivist.blogspot.com/2008/10/significant-no-hyperbole.html
(I shifted my curve slightly vertically, because it aligns exactly on top of Laherrere's hyperbolic curve)

The next step is to apply uncertainty to the U0 as well. This will get us another curve that will appear similar to the hyperbolic curve, but will actually have even a fatter tail.

BTW, you will not find this explanation in any professional textbook, as afar as I am aware. Most of the professionals still call it an "enigma", as in they don't understand it. Even Laherrere has not bothered to explain it, seeming to keep it at the level of a heuristic. This drives me crazy that no one ever tries to explain this stuff, because understanding is really the basis of making good analysis and policy decisions.
So the sad fact is that you can only find this explanation on TOD and on my blog http://mobjectivist.blogspot.com.

thanks for this post rembrandt. i have a couple not so tiny disagreements with the post.

graph 6 makes no distinction between reservoir drive mechanisms which can cause recovery to vary from 10 to 90 % depending upon the drive mechanism and reservoir management practices and independent of size. so correlation based on ooip is meaningless, possibly worthless and certainly misleading.

saudi arabia's recovery factor of 50% is probably too low. take a reservoir with high porosity and permeability, low viscosity oil, and steep dip and you have gravity separation. water injection/pressure maintenance only assures that the oil viscosity and gas saturation will remain low or zero(think ghawar - arab d). my wag for recovery factor for ghawar -arab d is 65% . many,many examples of the > 65% recovery from gravity separation exist and you only have to look a short distance north, abqaiq, for such an example.

i don't see much need for or benifit from eor for ghawar - arab d.

It appears to me as an outsider that "technology improvement" is a little too glib. I assume that the geology and mechanics of an oil field are pretty well understood. From that would it be reasonable to conclude that it is possible to plot out a path for what will need to be done to increase recovery rates? e.g. while we may not have the capability to get a man to Mars and back today we have a pretty good idea of what technology we will need to overcome the obstacles to get there. Didn't people imagine horizontal drilling before it became practical. If I am right then "technological improvements" 10 years from now should already be in the discourse -if not on the drawing boards.

The alternative view is that technology improvement is the equivalent of punctuated equilibrium. Something dramatic comes to pass that nobody imagined and it radically alters the situation.

If the former is the case then past is prologue and the increased recovery rates and the technology should be pretty firm. If it is the latter then expecting future increases is really more akin to buying a lottery ticket.

Heuristic--pertaining to a trial-and-error method of problem solving used when an algorithmic approach is impractical.

Web Hubble Telescope--I always read your posts because you have something very useful to say. Yet I don't understand them because I am not mathematically savvy. My education on statistics is limited to one graduate level course. From your writings, I am believing that your understanding of statistics is better than the average geologist or economist.

I am posting this rather than just emailing you because I assume that many others have the same problem. I really want to understand what you are saying. You may talk more about my concerns in your blog. If so, refers us to specific articles in your blog. You frequently talk about how frustrated you are when others use heuristics. I want to know more in lay terms about why. What are the limitations of heuristics? What can they tell you? What are the alternatives? What are their strengths and limitations? My assumption is that heuristics identify data patterns without any valid explanation for the cause of those patterns.

The IEA concludes from this chart that:

"Using data from Figure 7.10, it has been calculated that the average recovery factor for Roadifer's 1987 sample of 300 giant oilfields was 33.3% compared to 38.6% for Lahherere's 1996 sample of 200 giant oil fields. This analysis suggests that the average giant oilfield's recovery factor increased by 5.3 percentage points in the space of nine years, or 0.6 percentage points per annum. In the unlikely event that giant oilfield's recovery factors were to continue to increase at 0.6 percentage points per annum, then by the year 2020 the average recovery factor would be some 14.2 percentage points higher than in 1996. The average giant oilfield in 2020 would therefore have an average recovery factor of 52.8%." (IEA, World Energy Outlook, page 100)

The IEA is making a basic error in methodology here, even if the figures quoted are valid. The 0.6% per annum increase in recovery factor is not really that at all: it is an 0.2% per annum decrease in the amount of oil left in the ground when the field is abandoned. As the amount of oil left behind decreases each year, the additional amount of oil represented by a constant percentage of that amount decreases. In other words, if we are to extrapolate anything in a linear fashion, it should be a fraction of what would otherwise be left in the ground.

According to my calculations using this method, the recovery factor by 2020 will be 41.9%, not 52.8%. Or to be more in keeping with the accuracy of the numbers we are dealing with, around 40% instead of around 50%.

That's not to admit that I believe the 1987 data is in any way similar to the 1996 data. I'm just pointing out that even if the comparison is valid, the conclusion from it about increased recovery factor is not.

I think the EOR, technology, and "rate" vs "ultimate recovery" issue is a good deal more complicated than is suggested in this discussion. It is certainly true that in many cases EOR and technoloigical advances only allow us to produce the same amount of oil faster. Rockman, in a post above has a good example of how stripper production works. In some situations very low rates can be maintained profitably over a very long times.

However, I would argue that there are many other cases where increased rate does in fact result in increased recovery. For example, at Prudhoe Bay, much of the remaining oil is in "Zone 1" (roughly the bottom quarter of the formation). Unlike the superb reservoir rocks in the upper 3/4, the oil in Zone 1 is in thin, discontinuous sands with much lower porosity and permiability. The technology of coil tubing horizontal side track wells have made it possible to produce these sands at a reasonable rate. In theory one could produce them at a much slower rate with conventional wells. However, there is the matter of maintaining a high enough rate to keep the the Alaska pipeline ("TAPS") in operation. When the rate falls below some level, TAPS will cease to operate. At that point oil which might have been produced (at a lower rate) will be left in the ground. In this situation, higher rate (due to technology) directly translates to higher ultimate production.

I believe a similar argument would hold for nearly all deepwater production. No matter how much oil is in place the rate needs to be high enough to to support the infrastructure necessary to get the oil ashore. So from a big picture point of view, to understand the rate vs recovery issue, we need to somehow separate those cases where we will ultimately recover the same oil but more slowly, from those cases where either we recover it fast or not at all.

In many cases so-called reserve growth is not a mystery because a field is not a single pool but a series of adjacent pools which a) were not identified at the time of discovery and declaration of OOIP and b) are made econokmic for development because of the infastructure installed for the first pool. Anybody who has watched the development of any one field over time could see that the named field is in fact many or several pools added to the main discovery over years. The cobined set of reservoirs usually retain a single name so that when we look back at production history we see more reserves, both recovered and in place, than were first declared. Various stages of EOR or simply more intensive drilling also raises the value of the recovered volumes. In the current oil world especially in high cost deep water plays we often see quite low levels of recoverable reserves simply because the operator sets his production levels at the most profitable, choosing to leave the rest in place. In other words, the volume produced is a financial not a geological function. When the USGS raises the potential for the Orinoco Belt for example, they are simply adjusting the recovery factor in view of expected oil prices and the application of more intensive drilling or in this caser horizontal wells. So this kind of anlayis si important but more simple than the real world of oilfield economics and geology.

In many cases so-called reserve growth is not a mystery because a field is not a single pool but a series of adjacent pools which a) were not identified at the time of discovery and declaration of OOIP and b) are made econokmic for development because of the infastructure installed for the first pool. Anybody who has watched the development of any one field over time could see that the named field is in fact many or several pools added to the main discovery over years. The cobined set of reservoirs usually retain a single name so that when we look back at production history we see more reserves, both recovered and in place, than were first declared. Various stages of EOR or simply more intensive drilling also raises the value of the recovered volumes. In the current oil world especially in high cost deep water plays we often see quite low levels of recoverable reserves simply because the operator sets his production levels at the most profitable, choosing to leave the rest in place. In other words, the volume produced is a financial not a geological function. When the USGS raises the potential for the Orinoco Belt for example, they are simply adjusting the recovery factor in view of expected oil prices and the application of more intensive drilling or in this caser horizontal wells. So this kind of anlayis si important but more simple than the real world of oilfield economics and geology.

Rembrandt,
You may recognize the following as a sentence from you lead article:

"If reserve growth is however caused mainly by technological development, it is more likely that the trend in past reserve growth will continue."

I recognize that at the point where you say this, you are presenting a argument that further on, you seem to reject. But can you explain a detail of the argument that puzzles me? For me 'technological development' is an ambiguous phrase. In this context is it the development/discovery of a new technology? the deployment of a recently discovered technology to a particular oil field? or something else? Depending on what is meant by technology development, the conclusion 'growth will continue' may or may not be true. Is this ambiguity something that you can explain? Or is it something that you intended to display?

the amount of recoverable reserve increases over time due to changes in technology, economy, insights. But also expected recoverable reserves increase over time due to past underestimates.

"Recoverable reserves" (an economic term, not a geologic/SI term) could just as easily decrease in time due to changes in technology, economy and insights.

Rembrandt, are you still doing your excellent oilwatch monthlies? I haven't seen one for a couple months.

Hi Rembrandt, good work, I have a question. In these graphs of recovery percentage versus number of wells it is not clear how the caption percentage is computed. When you get to it, this not an urgent thing.
cheersJuan