Monday, June 6, 2011

The US Crude Production Peak is not Symmetrical

People that worry about the peaking of global oil supplies often use symmetrical curves as simple models for how production will peak and then decline, with logistics and Gaussians being popular choices.  This goes back to M. King Hubbert (and I've done some of this myself).  The United States is the poster child for this kind of analysis, since this region was the first to be developed at scale and production peaked in 1970.

However, it seems increasingly clear that the US production curve is far from symmetrical (perhaps driven by higher prices since the 1970s, and especially in the 2000s).  Using data from the EIA for production and reserves, we can see that the decline side is slower than the growth side for both:

Here I have expressed the reserves on the right scale in millions-of-barrels-per-day-years (ie the number of years that the the current reserves would permit you to produce at a steady rate of one million barrels/day).  Production is on the left scale in mbd.

You can see the noticeable slowing of decline in both production and reserves in the last decade.

If you look at the ratio (reserves/production, often abbreviated to R/P) you can see that it declined initially from very high values, reached a low in the 1980s, but has improved slightly in the last couple of decades:

We have been running with about 10 years worth of reserves for fifty years now.  The general picture seems to be, that although we are scraping the bottom of the barrel - hunting for oil in very deep water in the Gulf of Mexico, fraccing very tight rocks to squeeze oil out by the drop, pumping in CO2 to loosen up the gunk - the bottom of the barrel is a more productive environment than one might have feared.  Thus overall decline rates are low.

To see the asymmetry of the production curve more clearly, here I plot both the before and after peak periods on the same scale:

You can see that production now, forty years after the US peak, is pretty much twice as high as it was forty years before the peak (nor can this be written off as an artifact due to Alaska, which is down to 600kbd these days).

If global oil production behaves like the US, we might expect peak oil to take the form of a long plateau and a slow decline, which would tend to maximize hope for eventually successful adaptation (albeit with quite a bit of pain along the way).

Update: Here is the curve ex-Alaska.

It's still far from symmetric.


Kweksma said...


Doesn't your R/P graph just show that the SEC rules for reserve reporting works like a charm : oil companies are required to report reserves as being 10 times the amount of oil being produced in a year. Apparently everybody does that, and has been doing it for decades.

Glenn said...

You might look at Orlof on this subject. There is a qualitative difference between decline curves in any particular country, which still has access to other countries oil, and the decline curve of world supplies, or the decline curve of the last few countries with oil left.

Jonas said...

Interesting subject!
Is it possible that the net energy is more or less symmetrical?

KLR said...

The last graph is interesting - would like to see that applied to other nations, where possible; some countries have very messy production curves of course.

The US is very exceptional, I'd say - plenty of wealth and proximity to markets = no end of capital directed towards EOR. Don't think you'd say the same about Iran, say, where Ahmadinejad broke ground on a new refinery - something like 8 years ago; it subsequently became something of a bad joke. A roundup of refinery projects from last year had completion dates for this year or next listed - but all the projects seem to be <10% finished: 7 Oil Refineries Moving Ahead – In Iran | Sweetness & Light.

Last week I read that Iran had dealt with its refined product shortages by simply converting over petrochemical plants, but that sounded more than a bit propaganda-ish. To return to the point of this article, I'm betting that the US experience has much to do with free flow of capital and access to markets, neither of which will necessarily apply to many producing nations, including most of OPEC of course.

Jon said...

Is Alaska included in the downslope? Alaska was a very different geological region than the lower 48 and will skew the curve.

Without Alaska I think the curves will be much tighter. If this were a graph of net energy, I think they would be tighter still, or even a faster downslope on the backside.

gar said...

Jonas: My thoughts also. Add to that high prices paying for extensive EOR which helps squeeze the sponge more rapidly. The sponge contains a finite amount of hydrocarbons after all, squeezing it more rapidly leads to steeper decline rates down the road. Which is what can be seen in the North Sea.

Timaeus said...

Interesting Post! A nice way of pointing out empirically what one might expect and hope... that technological advances are allowing us to recover more oil per reservoir than we have historically, and that there they are having a noticeable effect.

KLR said...

Jon - here's a graph of Lower 48, with reserves and discoveries to boot: Boyce Lower 48. Commentary on this graph here: Permalink, from June 23, 2010 DB.

Take out Alaska and you simply lose that 10 year hump Prudhoe provided, nothing too exciting. You see a mild plateau/uptick post price shock. Haven't seen it applied to R/P, but R/P's not the most useful tool, unless you're an alarmist politician. ;) HL gives no end of false alarms, too. Prices and supply/demand are where it's at as far as real prognostication goes.

As examples you could plot R/P for nations like China, where their reserves stayed even flatter than the notorious OPEC countries, for decades on end. No one had any incentive to provide anything better, it was too costly to look into, so call it good. A couple years ago I plotted the same kind of graph Stuart shows here, reaching the conclusion that reserves just aren't that useful, other than as a broad indication of how large a resource is, and even then you need capital to extract it in the first place, per my first post.

Jon said...

I did a quick plot and if you exclude Alaska then the Lower 48 curve is almost perfectly symmetrical until about 20 years after the peak when the down slope stops falling as fast.

Then I removed the deep water Gulf of Mexico (Federal) but left in the off shore (shallow water) state production and the downslope fall was faster than the rise.

It looks like preventing a fast drop is really about bringing new regions and fields into production. The exact same thing is seen in the public north sea field data.

Stuart Staniford said...


I show the curve above ex-Alaska. It's still not symmetric. I can't see any reasonable justification for excluding the deep GoM (obviously if you cherry pick the right subset of things, you can ultimately make it symmetric, but that's a pretty meaningless exercise).

Mason Inman said...

Often people say that Hubbert was wrong about the U.S., because the production curve was not symmetrical. But they forget that he drew boundaries around the area he considered in most of his papers; he usually did not include Alaska or deep offshore production.

Stuart--with all your data, could you put together a curve for just the lower 48 production, for on-shore and shallow off-shore? Jon (comment above) said he did and it was very symmetrical; I tried doing the same once but wasn't sure I was slicing the data correctly.

I know your point here wasn't to prove Hubbert right or wrong, but I thought it would be interesting to see nonetheless.

Jon said...

Hi Stuart, My goal was not to cherry pick. Laherrere has often commented he finds overlapping bells work better at estimating ultimate because they incorporate new regions / technology as they come on.

Once I removed Alaska the match was nearly symmetrical until about 20 years post peak and then started to deviate. Why? Was it new technology boosting old fields? Or was it new fields that had not be tapped before? Subtracting the deep water GOM and having the curves stay pretty well matched makes me think is was mostly GOM. Although there is still a small uptick at the end.

I don't know the answers. Having burned myself with confirmation bias in the past, I like to poke at inconsistancies.

KODE said...

According to Ivanhoe, in his article King Hubbert - Updated(PDF!), Hubbert himself wrote:

"This complete cycle has only the following essential properties: The production rate begins at zero, increases exponentially during the early period of development, and then slows down, passes through one or more principal maxima, and finally declines negative exponentially to zero. There is no necessity that the curve P as a function of t, have a single maximum or that it be symmetrical. In fact, the smaller the region, the more irregular in shape is the curve likely to be. On the other hand, for large areas such as the United States or the world, the annual production curve results from the superposition of the production from thousands of separate fields. In such cases, the irregularities of small areas tend to cancel one another and the composite curve becomes a smooth curve with only a single practical maximum. However, there is no theoretical necessity that this curve by symmetrical. Whether it is or is not will have to be determined by the data themselves." (my emphasis)

Hubbert seems to have drawn symmetrical curves out of expediency; I have read anecdotes by people claiming to have known him that he could get rather irate when people ascribed to him the notion that the curve would necessarily be symmetrical.

Stuart Staniford said...


Well, the small uptick at the end is probably mainly the Bakken and other "resource plays", which are likely to become a bigger deal going forward. Like the deep GoM, they were there all along, but are enabled now by technological developments (fraccing and horizontal drilling in the former case, and the panoply of techniques required to drill deepwater in the latter case).

Stuart Staniford said...


Thanks for the references. My point, of course, was not to critique Hubbert, but rather to observe that it seems that high prices really are supporting techniques that wouldn't be viable at $10/barrel, and those are opening up resources that are causing the US curve to be non symmetrical. Presumably, the same possibilities exist elsewhere in the world also, which strengthens the argument for a slow global decline.

KLR said...

Jon and others - here's a link to my .ods spreadsheet file for US Crude Oil Production. It might prove useful, as I've edited out extraneous text in the headers, calculated YOY, and collected all sorts of other bits of EIA data here.

.ods is Open Office format, if you want I can upload .xls for MS Word etc too.

PADD 3 Fed Offshore doubled 1994-2001 from .8 to 1.6 mb/d, FWIW.

Anonymous said...

The idea that the downslope would be less steep as a result of technology and price seems reasonable to me. I do wonder what the ratio there is, and also whether technological solutions may be subject to a diminishing returns problem over time.

For instance, when new technology (or more accurately, existing technology that becomes economically feasible) is used to increase production, do the production characteristics of those efforts parallel historical numbers? Are rates of increase higher or lower than with traditional technology? Are depletion rates higher or lower? Etc.

Thinking of it as summing overlapping curves, if the newer sources end up depleting more quickly then, eventually, after some period during which the added production slows the decline, the steeper decline rates will combine to actually increase the overall decline rate. At least, without ever increasing amounts of drilling.

There seems to be some evidence that this is the case with natural gas in shale. If the same turns out to be true for these oil sources, then we may look back on this as simply a short term bulge before a more precipitous decline. Hopefully, that will not be the case because we could use all the time we can get.

buck smith said...

The US government is sitting on a number of bakken size reservoirs in the western US. All they have to do is lease them to get the production (and the money)

Steven Kopits said...

The way it's been described to me, there's no point in exploratory drilling for resources beyond ten years, because the discount rate (time value of money) makes such drilling uneconomical. Thus, if you have ten years of reserves, you develop those reserves before going to drill for more.

As for symmetry, I might argue that the notion of peak oil is largely intra-systemic. So, for a given set of technologies applied to a given set of resources, we might expect a pretty good degree of symmetry. However, if the resource base is expanded (a la Alaska thirty years ago) or technology changes (horizontal drilling, hydrofracking in shale oils), then previous assumptions about peak oil may prove incorrect.

TechGuy said...

I don't think your chart includes two very import factors:

1. EROEI. 50 years ago, it took 1 barrel of oil to produce about 100 barrels. Today the ration has slipped to about 8:1

2. Price of oil on the market has considerably change. Currently we are at cusp where higher prices dramatic effect demand. If consumers were about to afford $1000/bbl, than your assumption would be correct that production declines will be slower. However, I believe Price will have a big impact in future production in just a few years.

FWIW: I believe that production over over the long term will follow a horizontal flipped Chi-Square chart. since when oil production started back in 1861, the technology, infrastructure and demand was low. As time progress, the technology, infrastructure and demand all soared. This resulted in a long tail since production was limited. Moving forward to the future, technology has its limits and at some point there will be a fast decline in production. We are still too close to the peak to see this event coming. I belive the tide will go out when two more super giant fields are watered out, causing too much production to be lost to be replaced by the smaller fields.

Steve From Virginia said...

US extraction appears to be equal to that of the WWII period.

Pretty sobering. Unless more oil can be brought to market from new wells faster than the current wells deplete our rate of extraction will drop to Depression- levels by the end of the decade.

With 100 million more people and 200 million more automobiles and ... debt. Someone (thing) is going to do without.