[DoctorDoom]

So the next question is, what magnitude of change in reserves growth/discoveries would give you a plateau at some comfortable time out into the future, say for example, 2050 or so?

I phoned up Rosy Scenario and here were her inputs to the model:

800 Gb yet-to-find, 800 Gb reserve expansion, discovery climbs back to 20 Gb / year (3x the current rate), double the Venezuelan heavy oil deposits, and finally (and most significantly) reduce the rate of demand growth to 1.5%. She also told me that in 2020 we will be visited by aliens who will give us the secret to extracting dark energy.

All of this is barely able to push the peak out to 2045-2050. That is not comfortably far in the future. It's within the lifetime of our kids, and things still get pretty ugly in the decade following the peak. I want to see a solution that will go 200 years or more, that's my definition of comfortable. It's long enough to have some hope of developing fusion.

Code: Select all

      Production     USA       Reserves   Conv
Year  Conv  Hvy    mbd w% ef%  Conv  Hvy  R/P
2005    80    0     20 25   0  1156  200   39
2006    81    0     20 24   1  1167  200   39
2007    82    0     20 24   3  1177  200   39
2008    83    0     20 24   5  1187  200   39
2009    84    0     19 23   7  1196  200   39
2010    86    0     19 23   9  1204  200   38
2011    87    0     19 22  10  1213  200   38
2012    88    0     19 22  12  1221  200   38
2013    90    0     19 21  13  1228  200   37
2014    91    0     19 21  14  1235  200   37
2015    92    0     19 21  16  1241  200   36
2016    93    0     19 21  17  1247  199   36
2017    94    1     19 20  18  1253  198   36
2018    95    2     19 20  19  1258  197   36
2019    95    2     19 20  20  1263  196   36
2020    97    3     19 19  21  1268  194   35
2021    97    4     19 19  24  1273  192   35
2022    98    4     19 18  26  1277  189   35
2023    99    5     18 18  28  1281  186   35
2024   100    6     18 17  30  1284  183   35
2025   101    6     18 17  33  1287  180   34
2026   103    6     18 16  35  1290  177   34
2027   105    6     17 15  36  1291  173   33
2028   106    6     17 15  38  1293  170   33
2029   108    6     17 15  40  1293  167   32
2030   110    6     16 14  42  1293  163   32
2035   119    6     14 11  50  1283  147   29
2040   128    6     14 10  49  1257  131   26
2045   136    6     14 10  49  1173  114   23
2050   136    6     14 10  49   925   98   18
2055   116    6     14 11  49   699   81   16
2060    99    6     14 13  49   506   65   14
2065    70    6     14 19  49   359   48   14


[Soft_Landing]

Just found this thread, excellent post, excellent thread...

Doctor Doom wrote:

So the thrust of my argument is this: decline rate is the inverse of R/P, mature fields easily get down to 15 year R/Ps (e.g. Alaska, North Sea), therefore the world should also be able to get to 15 R/P, and hence the eventual world decline rate will be able to get to 7%. If you work backward from 7% and the existing reserves, it puts the peak out at least 15 years from now (unless growth increases). Any sooner and you get a more rounded plateau of longer duration, and the 7% rate is not reached for another decade or so after the plateau period.

Don't forget that the fastest decline rate achieved by world production *must* be the maximum of the weighted average of maximums for local productions all over the world. So, even though the decline rate in Alaska might reach 7%, the decline rate in other parts of the world would not necessarily be 7% at that time. The *world* consists of oil fields in various stages of production, some at decline rates of 7%, some at decline rates much lower (e.g., Venezuela).

If this isn't yet clear, consider just the Alaskan fields. If overall decline is 7%, then surely, some parts of Alaska are at greater decline than 7%, and some parts are at lower decline than 7%. The reason that decline can reach 7% in Alaska is because various fields in Alaska all came on around the same time frame, so the weighted average of depletion can get to a high level.

In contrast, this situation cannot be extrapolated to world production and decline. The variation in terms of age of fields (and complexity etc.) throughout the world is much higher than variation just in Alaska. Therefore, the weighted average of depletion rates throughout the world would never be expected to match that of Alaska.

This is one of the virtues of using the 'normal' curve or gausian curve to predict production rates. It reduces the temptation to extrapolate (incorrectly) from local trends to the much larger 'system as a whole'.

There's a theorem in statistics known as the "central limit theorem" which provides good reasons why the shape of world production should end up looking something like a 'normal' curve. Of course, politicking in the 70's messed that up. That normal curve, however, can still give us guidelines as to how fast world depletion can be on the tail side of the curve.

The point is that the occurance of 7% depletion in one mature zone cannot be extrapolated to the possibility of 7% depletion for world. You should expect that the 'world as a whole' should have a lower maximum depletion rate than that of it's parts.

Doctor Doom wrote:

After all, by definition 1/2 of the P50 estimates should prove out, right? What I'd rather do is take the weighted average of world-wide P50 estimates, divided by 2, and subtract the P90 numbers. I'm pretty sure the net number for probable reserve growth would be greater than my "plug" figure of 100.

I'm a little confused, I'm not sure what you're trying to achieve here.

A P50 estimate is the estimate of the amount of oil that is recoverable from a certain field with a 50% chance of being too high, and a 50% chance of being too low (and to be mathematically precise, a 0% chance of being correct!).

A P90 estimate is the estimate of the amount of oil that is recoverable from a certain field with a 90% chance of being too high, and a 10% chance of being too low.

P50 and P90 numbers can't be added or subtracted to obtain any meaningful result...

Say you thought a basketballer had 50% chance of scoring 20 points in a game. Your P50 for her is 20. Then, you decide there is only a 10% chance of this basketballer scoring less than 7 points. Your P90 score for her is 7. Now the difference between 20 and 7 is 13.... but 13 doesn't really mean anything...

I have been very much inspired by all this and will certainly have a go at constructing my own sim, to see what comes out.

[pup55]

Doc, thanks for your efforts on this.
Gotta run one more test for us though...

Try to test the Heinberg "Powerdown" scenario, with some kind of middling estmates of reserve growth and/or discoveries and try to use "serious conservation" to extend the plateau. Define "serious" any way you want.

If you want, I think your model is capable of testing "serious American conservation", that would answer the proverbial question "how much longer could we delay the peak, not worrying about the rest of the world, if US energy consumption could be tamed down to some responsible level".

This leaves aside (temporarily) the question of whether or not this is actually implementable.

[Aaron]

Doom - Do you receive email at your address listed here?

[DoctorDoom]

Don't forget that the fastest decline rate achieved by world production *must* be the maximum of the weighted average of maximums for local productions all over the world. So, even though the decline rate in Alaska might reach 7%, the decline rate in other parts of the world would not necessarily be 7% at that time. The *world* consists of oil fields in various stages of production, some at decline rates of 7%, some at decline rates much lower (e.g., Venezuela).

True today while there is a mix of young and old fields. However, eventually one would expect to reach the point where most/all fields are mature, at which point the world R/P ratio should start to approach the R/P typical of mature fields.

There's a theorem in statistics known as the "central limit theorem" which provides good reasons why the shape of world production should end up looking something like a 'normal' curve. Of course, politicking in the 70's messed that up. That normal curve, however, can still give us guidelines as to how fast world depletion can be on the tail side of the curve.

I'll wait for your sim.

A P50 estimate is the estimate of the amount of oil that is recoverable from a certain field with a 50% chance of being too high, and a 50% chance of being too low (and to be mathematically precise, a 0% chance of being correct!). A P90 estimate is the estimate of the amount of oil that is recoverable from a certain field with a 90% chance of being too high, and a 10% chance of being too low. P50 and P90 numbers can't be added or subtracted to obtain any meaningful result.

Let me at least try. That statement was made in response to people who challenged the notion of reserve expansion in my model by saying that it was very unlikely that new technology would enhance recovery. I agree, actually, but some level of reserve expansion is still possible as some of the P50 estimates come true. I will just make up an example to illustrate. Suppose you have a field with reserves of 10 Gb (P90) and 15 Gb (P50). There is some chance that you will someday be able to get 5 Gb more from this field. As time passes, your estimates should sharpen up, eventually the P90, P50, and P10 numbers should all narrow in on the same number (zero!). Let's suppose it's some years down the track, you produce 5 Gb and now this process has you at 6 Gb (P90), 7 Gb (P50), 8 Gb (P10). Your estimates sharpen up and some of the "extra" oil becomes oil you can reasonably count on. There is, of course, always the chance that your initial estimate was too high, and that over time you'll come to realize there wasn't as much as you thought. By definition, though, this is only going to happen 10% of the time; there is a much greater likelihood that you'll be able to up your estimates over time. Subtracting the P90 from the P50 number is, admittedly, crude. I still think it would be interesting to do that. By definition, half of the world's fields should eventually prove out the P50 numbers or better, while the other half won't. Thus, as a crude approximation, it would be interesting to know what (P90 - P50)/2 is, don't you agree?

[DoctorDoom]

If you want, I think your model is capable of testing "serious American conservation", that would answer the proverbial question "how much longer could we delay the peak, not worrying about the rest of the world, if US energy consumption could be tamed down to some responsible level".

I'll probably leave that until I build a more sophisticated model - this one's literally about 100 lines of code. You can kinda see that if you look at the US numbers in the above examples. Those are all based on very serious conservation, to wit:

1. Cars go from 20 mpg average to 40 mpg average, said conversion takes 15 years (this is approximately the time it takes to turn over the US auto fleet at today's rate of 17 million new cars / year). Gasoline is 45% of US oil imports, so this cuts demand to 78% of what it would otherwise have been.
2. Further conservation (unspecified) brings US demand down to 50% of what it would otherwise have been, in line with Germany, a country that has the same per-capita GNP as the US. This is achieved at a rate of 3% improvement per year. Certainly there would have to be further improvements in vehicles to achieve this, less travel by air, less use by agriculture, etc.

Given all that, you can then look at the three columns of numbers for the US in each of the above scenarios. The first column shows Mb/day, the second shows %-of-world, and the third shows where we are on the efficiency effort (starts at 0 and goes to 50).

As you can see, it's helpful but doesn't solve the problem - the US %-of-world can get down to around 20%, after which it starts rising again. (This assumes US "before-savings" demand is still rising at 1.5% per year the entire time.)

The thing I did not do is adjust the world number to reflect the lower US numbers. I just used a plug of 2% growth for the entire world, which includes the US. It would be better to model the US separately from the ROW and add the results back together.

In fact, it would be better still to divide the world producers into major regions, and the world consumers into major regions, and use different numbers for each both for depletion and consumption. That way we could factor in, for example, higher demand growth in India and China. That more sophisticated model will take a lot of coding and combing through source numbers. I'll try to do that next week if I have time.

[notacornucopian]

After reading this thread, I realized my use of the P90, P50, etc terminology has been incorrect. I had believed that the "P" numbers represented a % probabilty of new discovery. So in reality, the P5 number the USGS uses is the one that most closely follows the historic trend, correct ?

Also, to make this even more complicated, how will the " new technology " that we assume will be used to increase reserve growth affect extraction rates and overall production costs ? In other words, even if the oil is there and extractable, will physical constraints prevent the crude from being extracted fast enough to meet the demand ?

[pup55]

thanks, Doc.

I guess the point I was trying to get at is that this model evidently is not too sensitive to reserve growth and/or exploration if it continues at its current rate, and the "powerful determinants" of the plateau are oil-in-ground, and consumption and/or consumption growth rate. It does not matter too much whether India uses it or the US to affect the plateau.

You can nibble around the edges with conservation and/or technology, but it will not make too much difference either way, based on the current knowledge. By your judgement, it does not really matter if the plateau occurs in 2010 or 2050.

But, a lot can happen in 40 years. You would be surprised. Maybe a more serious problem is a shortage of people with any actual engineering talent.

[DoctorDoom]

After reading this thread, I realized my use of the P90, P50, etc terminology has been incorrect. I had believed that the "P" numbers represented a % probabilty of new discovery. So in reality, the P5 number the USGS uses is the one that most closely follows the historic trend, correct ?

Also, to make this even more complicated, how will the " new technology " that we assume will be used to increase reserve growth affect extraction rates and overall production costs ? In other words, even if the oil is there and extractable, will physical constraints prevent the crude from being extracted fast enough to meet the demand ?

My understanding is that the P90 number is the number that you are 90% sure you can extract, the P50 number is a number that has only a 50% chance of being recovered, etc. Once again I'm not assuming we need new technology to support some level of reserve growth, I'm relying (literally) on chance. If your baseline estimates are all P90 numbers (I don't actually know what BP uses), then just by sheer dumb luck we will have some reserve growth because some of the P50 estimates will come true. Of course, if BP's numbers are already P50 numbers, then, like, yikes!

[notacornucopian]

DD - I think I actually did have it right after all - I was thinking in terms of discovery rather than extractable amounts from reserves.

Maybe Ender is right - if the BP figures are bogus for remaining reserves, then perhaps the 2008 to 2010 peak year is a good estimate after all. Campbell did say that if the true figures were made public, we would have an economic catastrophe tomorrow. What happens if you plugged in 130 GB for Saudi Arabia instead of the 260 GB figure ( I am assuming that is what BP used ) ?

[Soft_Landing]

Congratulations on your moderatorship, Doctor Doom. :D

Also, thanks again for putting up a model. I can't believe I didn't think of having a go at that...

About a week ago, I wrote suggesting that even though the crisis in the 70's messed up the expected outcome from the gaussian curve shift, that this technique might still be useful.

I haven't programmed for years, but I built a little excel sheet that take production and total endowment inputs, and then attempts to extrapolate forward using the gaussian curve technique. The model still has some problems that I'm not entirely sure how to fix (e.g., doesn't account for spare capacity), but I was thinking you (and others) might enjoy having a look at it.

Two things.

I haven't a clue how to post it to this page... Is there some way I can upload an Excel Spreadsheet?

Second, it would be helpful to have numbers (data) for prior to 1993. I've just been using the BP Report data. Do you (or anyone) know where I can get production data over a longer frame? Since 1930's would be nice!

[Aaron]

http://www.peakoil.com/modules.php?name=Forums&file=faq#24

[smiley]

Second, it would be helpful to have numbers (data) for prior to 1993. I've just been using the BP Report data. Do you (or anyone) know where I can get production data over a longer frame? Since 1930's would be nice!

Go to the bp site

http://www.bp.com/subsection.do?categor ... Id=2015020

there you can download all the historical data you want in a nice Excel format.

I already tried fitting the production curve to Gaussians, but the results are inconclusive. The statistical error is simply too large. It does give you a nice idea how the R/P ratios etc should behave.

[Soft_Landing]

I just found the production data I was looking for, and have plugged it into my model. The plot of the various predictions is below...



I was unsure how to treat surplus production capacity in this model. However, when I plotted "surplus capacity + production", the curve appeared to be on the same gaussian since 1991. See on the graph how the top of the black 'error bars' on top of the blue production curve make a gentle curve consistent with the yellow curve later.

The yellow curve was generated using an Ultimate of 2300 Gb. I thought that was middle of the road. In that case, peak occurs during 2010.

If you decrease the ultimate to 2000 Gb, peak occurs late in 2006.

If you increase the ultimate to 2700 Gb, peak occurs 2016/2017.

[Ender]

The yellow curve was generated using an Ultimate of 2300 Mb. I thought that was middle of the road. In that case, peak occurs during 2010.

If you decrease the ultimate to 2000 Mb, peak occurs late in 2006.

If you increase the ultimate to 2700 Mb, peak occurs 2016/2017.

Looks good, but surely you mean Gb and thus are out by three orders of magnitude.