Saturday, January 29, 2011

An Export Land Model Analysis for the USA-Part2


...it becomes difficult to imagine that global oil production could gently waft down from lofty heights in a graceful smooth and continuous curve spanning decades. Rather, the picture that presents itself is one of stepwise declines happening in more and more places, and eventually encompassing the entire planet.

Dmitry Orlov, Peak Oil is History

In Part 1, I presented my best estimate Export Land Model analysis for the USA, based on my analysis of trends in petroleum production and consumption for the USA and its top ten import sources.  In the article I laid out my explicit assumptions.

Here, in Part 2, I examine some different possible future scenarios, based on alternative pessimistic and optimistic assumptions about the trends in petroleum production and consumption for the USA and its top ten import sources.

Warning 
If you are the type of person whose hates the feeling of cognitive dissonance, you may not enjoy this article.  I explore several different possible future scenarios here (and still more in part 3), and, when you get through them all, you might feel more confused than ever about what might happen to the USA's consumption in the future.  If you are looking for: “And that's the way it is,” you may find this article troubling—but I encourage you to work through it nevertheless.

I presented in part 1 my best estimate of what I believe to be most likely trend of future USA consumption—but I readily admit that I might be wrong (in either direction!), and, I am willing change my view if I find compelling evidence to do so.    Part of having the flexibility of mind to do so comes from thinking through alternative scenarios, such as presented here.   Try it and you will understand the issues better, I think.

I don't think that any of the following assumptions are, by themselves, ridiculous, because there is some support either in the production data itself, or, for geopolitical reasons.  In my opinion, however, it is unlikely that these would all happen together.  Nevertheless, I believe that exploring the plausible extremes of pessimism and optimism help to set outer boundaries on what we can expect going forward.

Pessimistic Export Land Model Estimate—why you should prepare for hard times
Figure 4 below shows the Export Land Model analysis, analogous to that shown in Figure 3 in Part 1, but with a few key assumptions shifted towards the pessimistic side:

1) USA—no change in assumptions.

2) Canada only sends its net exports to the USA. Canada's production is still assumed to increase, as represented by the modified Hubbert equation fit to the 1980-2009 production data. This could happen if, for example, because of global oil shortages, Eastern Canada was no longer able to import petroleum from Europe, Africa and the Middle East.  

3) Mexico—no change in assumptions (i.e., still no net exports by 2015). 
 
4) Venezuela—no change in assumptions (i.e., still no net exports by 2023). 

5) Saudi Arabia—no change in assumptions (i.e., still no net exports by 2024). 

6) Nigeria's production follows the trend represented by the Hubbert equation fit to its more recent 2002-2009 production data, and, Nigeria continues to send 53% of its net exports to the USA (i.e., now no net exports by 2018).  This could happen if, for example, civil unrest and militant attacks increase, and/or, foreign investments don't occur to increase or maintain production.


7) Russia's production continues to follow the trend represented by the Hubbert equation fit to the 1999-2009 production data, but Russia stops its linear increase in the percentage of exports to the USA.  That is, exports remain fixed at 2009 levels of 8%.  I still assume that Russian domestic consumption stays flat (i.e., still no net exports by 2022, but not increasing amounts of exports to the USA up to that point). 

8) Algeria—no change in assumptions (i.e., still no net exports by 2030). 

9) Angola's production increases at “only” 24 %/yr and its ultimately recoverable petroleum equals only 11.5 bbs, as represented by the unconstrained Hubbert equation best fit to the 2001-2009 production data. I still assume that Angola continues to send at least 24% of its net exports to the USA (i.e., this still results in no net exports by 2025, but in lesser quantities). 

10) Iraq's production continues to follow the trend represented by the Hubbert equation fit to the 2003-2009 production data (this was scenario 1 in my original trend analysis for Iraq; i.e., now, no net exports by 2030).  This would be consistent with Iraqi's recent history of periodic recoveries and war, and/or with a lack of foreign investments.  I still assumed that Iraq's domestic consumption follows the upward trend represented by the Hubbert equation fit to 2003-2009 consumption data, and, that Iraq continues to send at least 26% of its net exports to the USA.

11) Brazil—no change in assumptions.



A large part of the steep decline in Figure 4 at 2010 is due to my pessimistic assumption made that Canada starts to send only it net exportable petroleum to the USA.  That amounts to a minus 0.4-0.5 bbs/yr difference, as compared to my best estimate model shown in Figure 3.  As I have already explained in Canada—Petroleum Superpower or Super-slave? this would be difficult for Canada to do.  But, if over the next decade or so Venezuala, Saudi Arabia, Russia, Angola, Algeria etc..., all have no net exports, then where would Canada import its oil from in order to make up for the extra quantity, over its net exports, that it presently exports to the USA?  At some point in time in this scenario something like this would happen—but probably it would play out on a more gradual timeframe than depicted in the figure.

According to this "pessimistic scenario," the annual decline in production from 2011-2021 is in the 4-5 %/yr range, instead of 3-4 %/yr, as compared to the best estimate model.  Consequently, there is a 29% decline from 2005 to 2013, and another 23% decline from 2013 to 2021.  That's a 52% drop over 16 years

This is still not as severe a decline as what Russia experienced in the early 1990s, but substantially more severe than my best estimate shown in Figure 3.  As before, there are no prospects for an upswing through the 2020s and beyond—just a shallower rate of decline.  According to this scenario, by 2025, the only two significant import sources to the USA would be from Canada and Brazil.  

Optimistic Export Land Model Estimate—you should still prepare for hard times
Figure 5 below shows the Export Land Model analysis, analogous to that shown in Figure 3 in Part 1, but with a few key assumptions shifted towards the optimistic side:

1) USA—the rate of decline in domestic production in the USA stabilizes and remains roughly constant, as represented by the sum of the Hubbert equation and modified Hubbert equation fits to the different components of production (the stabilization is mainly due to increases in NGL production, see Figures 7 and 9 in Trends in USA Petroleum Production and Consumption.

2) Canada—no change in assumptions (i.e., the same continued trend of increased exports).

3) Mexico—production follows the longer term trend represented by the modified Hubbert equation fit to the 1987-2009 data (this was depicted as the dashed green line in Figure 4, here).  This may happen, for example, if alternative production sources (e.g., KMZ and others) are developed in time to mitigate the decline in the Cantarell field (i.e., now net exports continue past 2030). 
 
4) Venezuela—no change in assumptions (i.e., still no net exports by 2023). 

5) Saudi Arabia—production is assumed to stay constant at 3.847 bbs/yr (the average value of production from 2003 to 2009)—at least until 2030 (this was depicted as the dashed green line in Figure 3, here).  Domestic consumption still assumed to continue its exponential increase, and Saudi Arabia is still assumed to send 18% of its net exports to the USA.

6) Nigeria—no change in assumptions (i.e., still no net exports by 2039). 

7) Russia—no change in assumptions (i.e, still no net exports by 2023). 

8) Algeria—no change in assumptions (i.e., still no net exports by 2030). 

9) Angola—no change in assumptions (i.e., still no net exports by 2025). 

10) Iraq's production is assumed to explode upwards as represented by the Hubbert equation fit to 2003-2009 production data and assuming the production rate increases 17.5 %/yr and the ultimately recoverable petroleum (Qº) of 115 bbs (this was scenario 2 in my original trend analysis for Iraq).  This could happen if Iraq remains politically stable, the foreign investments come through, and, the projected quantity of oil really is in the ground (i.e., Qº = 115 bbs). I still assume that Iraq's domestic consumption follows the upward trend represented by the Hubbert equation fit to 2003-2009 consumption data, and, that Iraq continues to send at least 26% of its net exports to the USA.

11) Brazil—no change in assumptions (i.e., the same continued trend of increased exports).



The predicted consumption curve actually tracks pretty closely to the Hubbert equation best fit to the USA’s 1984-2009 consumption data (originally presented in Figure 10 here), a portion of which  I depict in the figure (dash red line). 

The predicted consumption curve might be characterized as a type of "rosy peak oil scenario," such as described in Orlov’s article, with a gentle wafting down in production, and, US consumption following suit.  For this scenario, however, the “wafting” would not last for decades, but rather, about one decade. 

According to the "optimistic scenario," the annual decline in production from 2010-2021 is predicted to be quite modest at less than 1%/yr.  That is less than the rate of decline in the past few years (i.e., about 9% since 2007). Production actually picks up slightly from its 2009 levels, and consequently, the total decline in the consumption rate is only 8% from 2005 to 2013, followed by and additional 5% decline from 2013 to 2021.  That's only a 13% drop over 11 years. 

The annual rate of decline in consumption does not steepen by more than -1%/yr until 2028.  Thereafter, throughout the 2030s, the consumption rate declines at a steeper rate of about 2-3 %/yr.  That steepening decline occurs for a few different reasons.  One reason is that assumed exponential increase in domestic consumption eventually takes out all of Saudi Arabia's production.  That is, despite an assumed continued production rate of 3.847 bbs/yr, Saudi Arabia's net exports still goes to zero by 2043.  Additional reasons are that Mexico's and Iraq's production start to go into steep decline through the 2030s, and hence, their net exports decrease even more rapidly.  

By 2040, USA consumption is predicted to be 64% of its peak consumption in 2005—a 36% drop over 25 years.

In summary, the USA consumption decline is delayed a decade and then the consumption decline from 2030 to 2040 is slightly milder than the predicted decline from 2010 to 2030 as depicted for my best estimate scenario in Figure 3.   The difference in the consumption decline curves in Figure 3 compared to Figure 5 is mainly due to the optimistic production assumption for Iraq.  Iraq's assumed enormous increase in production helps mitigate the declining exports from several other countries that are headed towards zero net exports by 2030 (e.g., Venezuela, Russia, Angola, Algeria).

What is most striking about the "optimistic scenario” presented in Figure 6 is that the USA’s consumption is still predicted to decline.  This is despite the optimistic assumptions that: the long-term decline in US production stabilizes, Mexico pulls out of its death-spiral of production decline, Saudi Arabian production does not decline at all, and, Iraq undergoes one of the most dramatic increases in production the world has ever seen.  We are talking here about Iraq's production rate from 2020 to 2030 eclipsing that of Saudi Arabia's assumed continued high rate of production.  Add to that my original assumptions of continued increases in production and exports from Canada and Brazil.  Despite all that “rosy” thinking, there are still not enough net exports from the top ten to put the USA’s consumption above its peak in 2005.  Just stop and think about that for minute.

Now, consider this: what would it take for there to actually be a growth in consumption—that is, a return to the consumption rate of 2005 and higher?

A "Sunshine and Lollipops" Export Land Model Estimate—a return to "normal" growth
To get the USA's consumption back to and above its all time 2005 high, I assume that USA, Mexican and Saudi Arabian production has stabilized, and, that there is continued growth in production (and exports to the USA) from Canada, Iraq and Brazil, as described above for the Optimistic Export Land Model Estimate.

Additionally, I assume that the trends of declining production from Venezuela, Russia, Angola, Algeria and Nigeria all stop and stabilize. Specifically, I have assumed that production, and exports to the USA, from Venezuela, Russia, Angola, Algeria and Nigeria all stay at their respective 2009 levels throughout 2010-2040. 

Figure 6 shows what this fantastic scenario would look like:

According to this "super optimistic" scenario, the peak in the USA’s consumption rate occurs in 2024 at 7.94 bbs/yr, which is about 4.5% larger than the peak in 2005.  Not surprisingly, the predicted peak in consumption in 2024 coincidences with the peak in Iraq's exports.  After that, Iraqi exports go down, mainly because Iraq's production goes into decline.  By 2036, the USA's consumption rate is back down to 2009 levels (about 6.9 bbs/yr). 

Based on this, I characterize this "super optimistic" scenario as predicting enough oil for a slight uptick in consumption over the first half and then a slight downtrend for the second half of a 25 year period.

There is one big problem with this scenario—it is a fiction at the moment.   I see no evidence in the production data to support the assumptions that the predicted production declines for Saudi Arabia, Venezuela, Russia, Angola, Algeria and Nigeria would all stop and stabilize. 

There is some hope, however, that the decline curves for these countries will not be a steep as represented by Hubbert equation fits to these data. 

As shown in my analysis of the US production data, the decline side of the production curve did not follow a simple logistic curve very well (see e.g., Figure 1 here).  Rather, the decline-side of the production curve, as modeled using the modified Hubbert equation, was more consistent with a steady increase in the fraction of recoverable oil (i.e., Q¥ ).  That increase corresponds to an about 0.53 percent increase per year (in the vernacular of the modified Hubbert equation, fcq = 1.0053).   This is probably reflecting improvements in the technologies of recovery or refining over the long period of the USA’s decline curve. 

It is possible that the decline side of the production curves for one or more of Saudi Arabia, Venezuela, Russia, Angola, Algeria, Nigeria, Brazil or Iraq will be similar to the USA’s.   For example, if hydraulic fracturing technology is brought to these countries, it might mitigate the steepness of the decline-side of the production curve. 

My trouble is that all of these countries are either in the early stages of decline or still on the growth-side of the production curve, and therefore I can’t apply the modified Hubbert equation to fit the decline side of the production curve. 

But, even if these countries all follow more of a USA-style decline side production curve, that still would not mean constant production for 30 year, as I assumed for the exports portrayed in Figure 6.  Rather, the production rate would decline, but at a more gradual rate than modeled using the traditional Hubbert equation.  This, in turn, should slow the rate of decline in exports as compared to Figure 3. 

For example, according to this USA-style production decline scenario, net exports for Saudi Arabia would probably hit zero somewhere between 2023 (as predicted assuming a declining production curve that follows the Hubbert equation—the solid green line in Figure 3 here) and 2044 (as predicted assuming a constant production rate—the dashed green line in Figure 3 here).  As this pertains to USA consumption, I picture in my mind something half-way between Figure 6 and Figure 3. 

Nevertheless, I think that Figure 6 is still useful for illustrating just what it would take to have a modest increase in consumption, similar to what we saw over the 25 year period from 1982 to 2007.  It seems unlikely to me that the days of “normal” growth are coming back, however.

Optimist-lite
And what if Iraq is not able to boost its production as hoped for and predicted by some analysts and the Iraqi's?  What if instead, Iraq is only able to maintain its present production at about 1 bbs/yr, but, all the other optimistic assumptions represented in Figure 5 still apply?  

Welcome to optimist-lite—really Iraq-lite, to be more specific.

The prediction for an “optimistic-lite scenario,” shown in Figure 7 below, is that the USA's predicted consumption curve is apparoxiamtely inverted, as compared to that shown in Figure 5.  That is, as shown in Figure 7, from about 2016 to 2025, the consumption rate declines at about 2-3 %/yr, and then, from 2025 on, the consumption rate decline is shallower at less than 1 % /yr.  

By 2040, the USA's consumption rate is predicted to end up at 62% of its peak in 2005—a 38% decrease over 25 years.  That’s about the same total decline as predicted under the "optimistic scenario" shown in Figure 5—it’s just that the decline pathway takes on a more near-term pessimistic shape.

These various optimistic outlooks all rely heavily upon the hoped-for bounty of massively increased oil production from Iraq and/or continued stable production from a number of the USA's top 10 import sources.  If the Iraq bounty is realized, then it probably will allow the USA to stumble along with a modest annual declining rate of consumption for another decade or so before succumbing to the about same decline as predicted in my best estimate presented in Figure 3 of Part 1.  

In Part 3, I will explore some implicit assumptions that are buried in this Export Land Model Analysis: fungibility, what happens to the exporting countries, and ERoEI. 

Thursday, January 27, 2011

An Export Land Model Analysis for the USA-Part 1

As Matt Simmons pointed out several years ago, the critical problem with post-peak exporting regions is that we would have two exponential functions (declining production and generally increasing consumption) working against net exports. From the point of view of importers, it is quite likely that we are facing a crash in oil supplies. In my opinion, what I have described as the “Iron Triangle” is doing everything possible to keep this message from reaching consumers.

In an essay posted on The Oil Drum blog in January 2006, I warned of an impending net oil export crisis, and I used what I called the Export Land Model (ELM) to illustrate the detrimental effect on net oil exports of declining production and increasing consumption.


It has taken me four months to get here, but I am finally in a position to use my analysis of trends in petroleum production and consumption for the USA, and its top ten import sources, to do an “export land model” analysis for the USA.  Actually, I have been doing a running analysis as I finished each country, but now its time to step back and look at the data from a broader perspective. 

But first, some background.

Overview of the Export Land Model
As you can see from the quote at the top of this article, the concept of the “Export Land Model” as coined by Jeffery Brown, was inspired by the late Matt Simmons and is based on a simple but important idea.  The net amount of oil available for export from an oil producing country (i.e., an “Export Land”) should equal that country’s total production, minus its domestic consumption.  It follows therefore, that if you can accurately predict the future trends in an export land’s oil production and consumption then you can also predict what that country’s future net oil exports will look like. 

This is particularly important in the period after peak oil, when oil production begins to decline from the exporting countries, but the domestic consumption rate may still be increasing.  The Export Land Model shows that the rate of decline in oil exports will be even faster than the decline in production if there is also increasing domestic consumption.  For example, if the production rate was exponentially decreasing, and, the domestic consumption rate was exponentially increasing, then the export rate, represented by the difference of production minus consumption, would decline at a double exponential rate.  In this kind of scenario, net exports hit zero long before production does.   This has very serious implications for oil importing countries, like the USA.  If the USA cannot produce or import the amount of oil it has in the past, then its consumption must go down. 

Export Land Model analysis done by Brown and Foucher
In the original article from 2006, Hubbert Linearization Analysis of the Top Three Net Oil Exporters, Brown noted that two of the top three oil producers, Norway and Russia were producing below their peak production levels and suggested that the third, Saudi Arabia, was on the verge of declining production:

As predicted by Hubbert Linearization, two of the three top net oil exporters are producing below their peak production level.   The third country, Saudi Arabia, is probably on the verge of a permanent and irreversible decline.   Both Russia and Saudi Arabia are probably going to show significant increases in consumption going forward.  It would seem from this case that these factors could interact this year produce to an unprecedented--and probably permanent--net oil export crisis.

The prediction of production data was referring to an earlier 2005 posting by Stuart Staniford, titled When Does Hubbert Linearization Work?, which presented production data for Saudi Arabia, Romania, USA, Iran, and Norway.

In a subsequent article (also described in a 2007 article by Sam Foucher, Declining net oil exports--a temporary decline or a long term trend?, published at The Oil Drum), A Quantitative Assessment of Future Net Oil Exports by the Top Five Net Oil Exporters Brown, and his colleague "Khebab," did a more detailed analysis of the top five oil exporting countries (Saudi Arabia, Russia, Norway, Iran, UAE), again based on the using the linearized version of the Hubbert equation to estimate future production, and, “a Monte Carlo analysis based on the observed growth rates over the last 10 years” to estimate future consumption.   The main conclusion from the study was that future net exports from these top five exporters were all likely to decline in the future.  Their average rate of the decline in exports was estimated to be -6.2%/yr, with net exports hitting zero in about 2031.

My reasons for doing a Export Land Model Analysis of the USA
There is not single a person that I have explained the idea of the Export Land Model to that does not understand its implications for the USA.  But usually, the follow-up questions are something like: “well just how important is this for the USA, and, how does this apply to USA’s actual import sources?”  For instance, of the above-mentioned top five exporters, only one, Saudi Arabia is a top five exporter to the USA (see e.g., Where in the world does the USA import its oil from?).  

I could only offer a vague response, concerning the fungibility of oil.  That is, if there are export declines in the top five producers, then this will probably track through the entire global market, including the specific exporters that are important to the USA. 

There have also been criticisms in the use of the linearized version of  Hubbert’s logistic equation to model oil production, some of which I have discussed in a previous series (see e.g., Refining the Peak Oil Rosy Scenario Part 4: Testing Hubbert’s linear and nonlinear logistic models) and which the interested reader can easily find discussed in detail at other websites, such as The Oil Drum

So, I embarked on what turns out to have been a long project to assess for myself how the Export Land Model applies to the USA.  Long project, because I had to analyze the production and consumption data for 11 countries—the USA and its top ten import sources, and, assess the relative amounts of petroleum that get exported to the USA.

Features of my approach
There are several things about my approach that differ from what Brown and colleagues did:
1) After having high hopes for using the linear logistic model, I abandoned it in favor of a non-linear least squares (NLLS) analysis of the logistic model.  NLLS can be readily implemented within an Excel spread sheet, and, I found it to give more reasonable results, when tested on simulated data (see e.g., part 4, part 5 and part 6 of my earlier series).

2) I developed and tested a modified version of the nonlinear logistic equation (referred to here and in past articles as the modified Hubbert Equation), to handle situations where the production data seriously deviates from a simple logistic curve. (see e.g., Refining the Peak Oil Rosy Scenario Part 7: An improved logistic model)

3) I analyzed both production and consumption data using the Hubbert equation (or the modified Hubbert Equation), for the reasons discussed in Refining the Peak Oil Rosy Scenario Part 4: Testing Hubbert’s linear and nonlinear logistic models).

4) Like Brown et al., I primarily relied on petroleum production as reported in a massive spread sheet, the BP statistical review.  Petroleum production in the BP statistical review “includes crude oil, shale oil, oil sands and natural gas liquids.”  However, in a few cases where a more detailed analysis was needed (e.g., some individual components USA production), or, the data was not present in the BP statistical review (e.g., Angolan or Iraqi consumption; Brazil ethanol production), I have turned to data provided by the Energy Information Agency (EIA). 

5) My analysis considers not only the net exports a country produces, but also the percentage of net exports that specifically gets exported to the USA.  In some cases, like Mexico, this is not a big deal, because virtually all exports go to the USA.  In the case of Canada, even more than its net exports goes to the USA, forcing Canada to be both a major petroleum exporter and importer (due to an agreement under NAFTA, see Canada—Petroleum Superpower or Super-slave?).  Along similar lines, up until recently, Brazil also exported to the USA, but was a net importer.  In most cases, however, only a fraction of a country's net exports goes to the USA, and estimating the size of this fraction is about as important as the petroleum production and consumption analysis itself.

Why the top ten?
I have analyzed the petroleum production and consumption trends for the USA and its top ten import sources for 2009: Canada;  Mexico; Venezula (direct and indirect via the USVI); Saudi Arabia; Nigeria; Russia; Algeria; Angola; Iraq; and Brazil (and no, those north sea oil producers, the UK and Norway are not among the top ten exporters to the USA). 

Previously, in Where in the world does the USA import its oil from?, I pointed out that the top ten export sources, when added to the USA’s domestic production, consistently accounted for 86 (85.8±2.5) percent of the USA’s consumption over the last 36 years:



This suggested to me a simple prediction formula for estimating USA consumption:

            Predicted consumption = (production USA + Imports top ten) / 0.86

That is, if one takes the sum of USA production plus imports from the USA's top ten export sources and divide this sum by 0.86, the result should give a good estimate of USA consumption. 

In Figure 2 shows exactly that: the black open circles equals the USA's reported production plus the top ten's reported USA exports, all divided by 0.86. 

Also shown in Figure 2 is the USA's reported production and consumption rates for 1949-2009 (from the BP statistical review) and the reported imports to the USA from these top ten export sources for 1973-2009 (from the EIA publication, U.S. Total Crude Oil and Products Imports). All rates are in units of billions of barrels per year (bbs/yr)

The agreement between the reported consumption and predicted consumption looked remarkably good to me.  For instance, linear regression analysis of reported USA consumption (the red circles) versus predicted consumption, shown in Figure 2 (the black open circles), gives an r2 equal to 0.90 (see inset Figure 1).  That is, 90% of the variation in USA consumption over the period 1973-2009 is explained by this simple prediction formula. 

These findings lead me to believe that, given an accurate estimate of the USA's future production plus future imports from the top ten export sources, I could make a pretty good prediction of what the USA's consumption will be going forward.

Export Land Model analysis for the USA—My best estimate
Well, without further adieu, Figure 3 shows all the data in Figure 2, plus my best estimated predictions for USA production and the import trends for the top ten import sources going forward for 2010 to 2030.  The solid red line shows my prediction for USA consumption for 2010-2030 based on the above formula.

There is a steep decline in consumption predicted for the period from 2010-2022.  An inflection point, at 2023, reflects the prediction that net export from some major suppliers, Mexico, Venezuela, Saudi Arabia, have dropped to zero by 2023 and the USA is now importing its oil mainly from Canada, Iraq and Brazil.  Over the next twenty years, net exports from several of the top ten import sources are predicted to go to zero for: Mexico (2014), Venezuela (2022), Saudi Arabia and Russia (both 2023), Angola (2024), and Algeria (2030).  A small amount of exports still comes from Nigeria until 2039, when its exports are also predicted to go to zero.

The predicted decline in USA consumption is part of a trend that started in 2005, when the US reached its peak in petroleum consumption at 7.6 bbs/yr.  By 2009, consumption had dropped from the peak by about 10%.  According to the prediction curve, by 2013, consumption will be about 80% of the 2005 peak, and by 2021, about 60% of the 2005 peak. 

The predicted petroleum consumption rate declines at a rate of -3 to -4 %/yr from 2011 until 2023, and then tapers off to less than -1%/yr from 2025 on.  That is a decline rate that the USA has experienced before, and therefore is not necessarily the "end of the world," but probably the end of the world as Americans have known it recently.  For instance, from 1978 to 1982, the consumption rate declined from a peak of 6.88 to 5.58 bbs/yr.  That was a -19% decrease over five years, or about -3.75%/yr.  Of course, by 1984, consumption was on its way back up—but I'm not expecting that to happen this time.   Rather, this time, after experiencing a -20% decline from 2005 to 2013, there would be another -20% decline from 2013 to 2021, and then no prospects for an upswing through the 2020s and beyond.   

The predicted 40% decline in consumption from 2005 to 2021 is not unprecedented for a global superpower to experience either.  From 1990 to 1996, Russia's consumption went from 1.87 bbs/yr to 0.98 bbs/yr and consumption has stayed flat thereafter (see e.g., Figure 2 here).  That's a -48% decline in only 7 years—a whooping decline rate of -6.8 %/yr!  But once again it was not the end of the world for Russia, but it did end of Russia's (or the Soviet Union's) place as a global superpower.  The USA's petroleum consumption from 2010 to 2030 might well follow a similar trend as Russia's did from 1990 to 2010, but be a little bit more spread out. 

No way—your wrong man!

And that's the way it is
–Walter Cronkite. CBS news

Well no...that's not necessarily the way it is, and even Walter knew that. 

There are a lot of explicit (and some implicit) assumptions wrapped up into Figure 3.  In my previous series of articles on Trends in Petroleum Production and Consumption, I have given my analysis rational for each of the eleven countries shown in Figure 3 and the interested reader is directed to these articles for more details. 

Although Figure 3 represents my best estimates, I would be amazed if the Land Export Model that I have presented here turned out to be exactly correct.  For example, here are some of the major assumptions I have made in arriving at the predicted consumption curve shown in Figure 3:

1) USA production continues its downward trend, as represented by the modified Hubbert equation fit to the total production data from 1980-2009.

2) Canada continues to export more than its net exports to the USA, as per its obligation under NAFTA, and Canada's production continues to increase, as represented by the modified Hubbert equation fit to the 1980-2009 production data.

3) Mexico's production continues to follow its short-term trend, as represented by the Hubbert equation fit to the 1999-2009 production data (i.e., no net exports by 2015).  

4) Venezuela's production continues to follow its production trend from 1985-2009, as represented by the Hubbert equation fit, and, Venezuela continues to send 75% of its net exports to the USA (i.e., no net exports by 2023). 

5) Saudi Arabia's production continues to follows the trend represented by the Hubbert equation fit to 1985-2009 production data; domestic consumption continues its exponential increase, and Saudi Arabia continues to send 18% of its net exports to the USA (i.e., no net exports by 2024). 

6) Nigeria's production continues to follows the trend represented by the Hubbert equation fit to 1983-2009 production data, and, Nigeria continues to send 53% of its net exports to the USA.

7) Russia's production continues to follows the trend represented by the Hubbert equation fit to 1999-2009 production data, Russia continues the linear trend of sending an 0.6%/yr increasing proportion of its net exports to the USA, and, Russian domestic consumption stays flat (i.e., no net exports by 2023). 

8) Algeria's production continues to follows the trend represented by the Hubbert equation fit to 1995-2009 production data, and, Algeria continues to send 30% of its net exports to the USA (i.e., no net exports by 2030). 

9) Angola's production increases at 30%/yr, its ultimately recoverable petroleum equals 13.5 bbs, and, Angola continues to send at least 24% of its net exports to the USA (i.e., no net exports by 2025). 

10) Iraq's production remains at 1 bbs/yr (this was scenario 3 in my original trend analysis for Iraq), Iraq's domestic consumption follows the upward trend represented by the Hubbert equation fit to 2003-2009 consumption data, and, Iraq continues to send at least 26% of its net exports to the USA.

11) Brazil's production continues to follows the trend represented by the Hubbert equation fit to 1980-2009 production data for the sum of crude oil, LPG and RG, plus, the modified Hubbert equation fit to the ethanol production data for 1995-2010, and, Brazil continues to send 12% of its net production to the USA.

Here's one more prediction: some readers will find some of these assumptions to be overly pessimistic, and other readers will find some of these assumption overly optimistic (maybe even the same assumptions!).  And, you maybe right! 

In part two, I will explore some of these pessimistic and optimistic scenarios in further detail—I hope that you will join me, and, feed back is always welcomed.

Saturday, January 22, 2011

Trends in Brazilian Petroleum Production and Consumption

As noted in Where in the world does the USA import its oil from?, based on the data reported in U.S. Total Crude Oil and Products Imports (“Total Imports”),  I estimated that Brazil was the tenth largest petroleum exporter to the USA in 2009.


The historic data in “Total Imports” suggest that Brazil’s importance as a petroleum supplier to the USA has always been relatively modest:  there were very small exports to the USA until the 1980s then up to 1-2% of the USA’s total imports in the late 80s, then back below 1% to almost nothing in the mid 90s, then a steady increase in the 2000s until, in 2005, Brazil once again supplied over 1%, and now 2.6% of the USA’s total exports in 2009 (0.309 mbd). 

Brazil’s Oil production history and potential
Perhaps it should be no surprise that Brazilian exports to the USA for the last 30 years have always been fairly modest—throughout this period Brazil has been a net importer of oil.  According to the EIA’s Country Analysis Brief Brazil is the 10th largest energy consumer in the world and the 3rd largest consumer in the Western Hemisphere, behind the United States and Canada.  In 2006, 49% of Brazil’s energy consumption was from the consumption of oil, followed by hydroelectric power (36%) and natural gas (7%). 

The offshore Campos and Santos Basins account for the bulk of Brazil’s proven oil reserves of about 13 bbs, and, for the steady increase in oil production in the 2000s.  The EIA estimates that Brazils increasing crude oil production will make it a net oil exporter in 2009. 

Similar to Mexico, Venezuela, Nigeria and Angola, most of Brazi’s oil production is controlled by a state-owned entity.  Petrobras been ramping up production through the 2000s, and since 2008, has deployed two floating production, storage, and offloading (FPSO) platforms having a total production capacity of 0.36 mbd.  In the 2000s the Brazilian government has allowed a number of privately owned companies, usually in co-operation with Petrobras, to start projects that should provide a total production capacity of about 0.35 mbd: Shell’s Bijupira-Salema project (0.05 mbd), BC-10 project (0.1 mbd); Chevron’s Frade project (0.1 mbd); StatoilHydro’s Peregrino field (0.1 mbd).

Then there are the sub-salt fields.  The Tupi field, discovered in 2006, is alone estimated to have 5-8 bbs of recoverable reserves oil and natural gas. Oil and gas recovery from Tupi, and similar sized sub-salt fields, will be very challenging however, as these reserves occur in a subsalt zone that average 18,000 feet below the ocean surface.  This will require the construction of multiple FPSO platforms and advanced technology to recover the oil.  Nevertheless, the potential is there—estimates of the recoverable oil and natural gas reserves in the entire sub-salt fields are +50 bbs of oil equivalents.   This would increase Brazil’s proven petroleum reserves from 13 bbs to over 60 bbs.

Brazils plans for the coming decade to double its petroleum production are very ambitious indeed:

State-run Petrobras oil giant unveiled the spending plans as Chief Executive Officer Sergio Gabrielli set out the company's strategy to build capacity in the run-up to 2020, when Brazil will have doubled its production to 5.4 million barrels a day from 2.7 million barrels a day at present. 


The EIA’s country analysis brief, however, sounds a more cautious note: 

In large part due to this sizable slate of recent expansions, EIA expects that Brazil’s total oil production could reach 2.81 million bbl/d in 2010. This forecast takes into account the above-mentioned projects and an estimate for decline rates at Brazil’s older, mature fields. This could make Brazil one of the largest sources of new, non-OPEC oil supply growth. However, recent experience has shown that forecasts of non-OPEC supply growth have generally been over-optimistic, so there is considerable downside risk to this forecast. Such risks include larger decline rates at mature fields and delays to project schedules.

Even past estimates by Petrobras have projected more modest increases in crude oil production:

Brazil's state oil company, Petrobras (PBR.N: Quote, Profile, Research, Stock Buzz), projects its crude output in Brazil should reach 2.37 million barrels per day in 2011, up a steep 27 percent from this year's planned 1.86 million bpd, and then rise to 2.81 million bpd by 2015.

             
Brazil’s “other liquid”—ethanol
No discussion of Brazil’s present and potential liquid petroleum production would be complete without considering ethanol production from sugarcane.  As noted in the EIA country analysis brief, Brazil is the second largest producer of ethanol, and the world’s largest exporter. 

Brazil has aggressively moved towards the use of flex-fuel vehicles that can run on 100 percent ethanol or an ethanol-gasoline mixture.  All gasoline in Brazil contains 20-25 percent ethanol, and, about 90% of all new cars sold in Brazil are flex-fuel vehicles.  Despite this strong move toward using ethanol as a major liquid transport fuel, Brazil’s growing domestic ethanol production exceed domestic consumption and it seems likely that ethanol exports will increase in the future. Presently, as the largest ethanol exporter in the world, Brazil holds over 90 percent of the global export market.

Brazil’s present ethanol production is the result of a decades-old plan:

Brazil's military dictatorship launched the national ethanol program in 1975, when about 90 percent of its fuel consumption depended on foreign oil. The government offered subsidies to sugar cane growers and forced service stations in every town of at least 1,500 people to install ethanol pumps. By the early 1980s, almost all new cars sold in Brazil ran on 100 percent ethanol.


At present, Brazil’s ethanol exports to the USA are modest at 0.013 mbd in 2008, according to the EIA country analysis brief.  This is due at least in part to US-imposed import tariffs designed to protect US farmers, producing corn-based ethanol, from the likely decline in prices if Brazil could freely export to the USA without a duty fee and a 54 cent/G tariff being imposed (see e.g., Sugar cane ethanol's not-so-sweet future).

An end-around the duty fee and tariff has been exploited by Caribbean countries that import ethanol from Brazil, slightly process the ethanol (from hydrated to anhydrous) and then re-exporting the ethanol to the USA for profit.  Under the Caribbean Basin Initiative (CBI), the 2.5% duty and 54 cent/G is waived.  But this still doesn’t avoid the US import quota of 7% of U.S. ethanol consumption (see Ethanol fuel in Brazil).
The tariff was set to expire as the end of 2010, but Congress extended it

A little-noticed provision of the law extends ethanol tax credits ($.45 per gallon, plus a bonus for small producers) and tariffs on ethanol imports ($.54 per gallon), previously set to expire at the end of 2010. Should the rest of us also celebrate? I think not.


However, even if Brazil’s ethanol exports (directly or indirectly through the Caribbean) are limited—the domestic ethanol use helps make more crude oil available for export, as would more ethanol exports to Japan, China or Europe would potentially make more crude oil available for export to the USA. 

The ERoEI of sugar cane based ethanol is much higher than corn based ethanol
We have all heard by now about the poor energy return on energy invested (ERoEI) for ethanol produced from corn, which at best, has a ratio of slightly about 1:1.  This does not appear to be the case for ethanol produced from sugar cane, however:

Let’s turn next to trade and comparative advantage. In the case of ethanol, comparative advantage belongs, hands down, to sugarcane-based ethanol from Brazil. The net energy yield from sugarcane-based Brazilian ethanol is about 8:1, compared to barely more than, or perhaps less than, 1:1 for the U.S. corn-based product. 


The sustainability of Brazilian ethanol appears for the most part to be as advertised. The indications are that the EROEI is at least 8.3, which would actually make it better than for gasoline.

Sugar cane based ethanol production has the potential to be expanded about 7x
The Brazilian government has set a zoning limit on the amount of agricultural land that can be used for sugarcane cultivation:

Nowadays, sugarcane cultivation uses less than 1% of the Brazilian lands (7.8 million hectares). Projections of the Ministry of Agriculture indicate that if Brazilian production doubles till 2017, at most 1.7% of the lands will be used.

The group of restrictions regarding the environment, economy, society, climate risks, and soil conditions, set by ZAE Cana, guides the expansion of Sugarcane in 7.5% of the Brazilian lands (64.7 million hectares). According to the new criteria, 92.5% of the national territory is not suitable for sugarcane plantation.


If presently 1% of the land is being used and up to 7.5% could be used, then assuming a constant yield of sugarcane, ethanol production could potentially be increased about 7.5 times over present projection.

Brazil’s export trend to the USA
As Brazil is not a net petroleum exporter, I cannot do has I have in the past for other countries and look at Brazilian exports to the USA as a percentage of its total net exports.  Instead, I plotted the petroleum exports to the USA since the early 90s, as reported in “Total Exports” versus Brazil’s total petroleum production (reported by the EIA):
Since a low percentage of 0.5%, Brazil’s exports to the USA have steadily increased to 12% in 2009.  Linear regression analysis of the data from 1997-2009 gives an r2 = 0.96, slope 0.88%/yr.  If the linear trend continuous then by 2030 Brazil would be exporting 30% of its gross production to the USA.  I think that is unrealistic, but the trend certainly doesn’t suggest any downturn in exports to the USA.  However I think that strong competition for Brazil’s oil from China, India, Europe and others will stop the upward trend. 

As a compromise between applying this linear trend versus or a 5-year aver age, for the export analysis to follow, I have assumed that the percentage of Brazil’s exports to the USA will continue at its 2009 percentage (12%) going forward.   

Non-linear least squares (NLLS) analysis of total petroleum production
For most of my past analysis of other countries I have used of the petroleum production data for 1965-2009 reported in the BP statistical review.   However, for Brazil the BP data seriously underestimates Brazil’s total production because it does not consider that “other liquid,” ethanol.  To account for this, I have used the 1980-2009 data presented in the EIA’s Brazil Energy Profile. 

Figure 1 below compares the total petroleum production reported by BP and the EIA and the respective NLLS best fits to these data using the Hubbert Equation.  Also shown are some of Petrobras’ stated future production rate expectations.

It is apparent that the EIA total production numbers are higher than the BP numbers (which report crude oil and NGL production), and, that the Hubbert Equation fits are quite different as well.  The Hubbert Equation fits the BP production data fairly well (light blue solid line: best parameters: “a” = 0.082; Q∞ = 59.46; Qo = 0.36) and the prediction curves lines up pretty well with Petrobras’ earlier crude oil production estimates for 2011 and 2015. 

The fit  “blows up” when I attempted to fit the Hubbert equation to the EIA total production data (which reports crude oil, LPG, refinery gains and “other liquids”) as indicated by the exponentially shaped best-fit curve with an absurdly high Q∞ (dark blue solid line: best parameters: “a” = 0.0613; Q∞ = 106; Qo = 2.52). 

The EIA glossary defines Refinery Processing Gain (RG) as “[t]he amount by which the total volume of refinery output is greater than the total volume of refinery input for a given period of time. The processing gain arises when crude oil and other hydrocarbons are processed into petroleum products that are, on average, less dense than the input.”  Generally this is a small number amounting to 2-6% of the total petroluem production. 

The EIA glossary defines Liquefied Petroleum Gases (LPG) as “hydrogen-based gases derived from crude oil refining or natural gas fractionation. They include ethane, ethylene, propane, propylene, normal butane, butylene, isobutane, and isobutylene. For convenience of transportation, these gases are liquefied through pressurization.”  It is apparent that LPG is substantially the same as what the BP statistical review calls NGLs (the liquid content of natural gas where this is recovered separately).   I can tell this is the case, because the sum of the EIA’s crude oil plus LPG is nearly identical to the total production number reported in the BP statistical review.  

There is no entry in the glossary  for “other liquids,” but, in the glossary of the Dec 2010 International Petroleum Monthly (IPM), this was defined as “Biodiesel, ethanol, liquids produced from coal and oil shale, non-oil inputs to methyl tertiary butyl ether (MTBE), Orimulsion, and other hydrocarbons.”  I think that for Brazil, “other liquids” mainly refers to ethanol. 

The Hubbert equation best fit actually comes pretty close to the goal of 5.4 mbd announced by Petrobras’ CEO Sergio Gabrielli, so I can’t say that this is a ridiculously high prediction curve.  However, the yearly increase in the production rate would have to be a lot larger than 6.1%/yr, as implied by the best fit value of the rate constant “a,” if a more reasonable value for Q∞ is assumed. 

This is illustrated by the dashed line shown in Figure 1, which presents the Hubbert equation fit to only the 2005-2010 data with the constraint that production hits 1.97 bbs/yr (5.4 mbd) in 2020, and that Q∞ equals 60 bbs.  Now, the best fit value “a” equals 0.134, implying that a year-over-year increase of 13.3% in the production rate would be needed to get to this doubling in production compared to present rates.  Given the complexities of extracting oils from thousands of feet below the ocean, however, this seems like an unrealistic goal.

Modified Hubbert Equation fit to Brazil’s “other liquid”
I turned to my modified Hubbert Equation (good old equation 9, from part 7 of my earlier series of articles) in an effort to tease some more realistic predictions from the EIA production data.  The EIA does separately report the component parts of total production: crude oil, LPG, RG and “other liquid” production.  Perusal of these data suggested that it is the “other liquids” component of production that has growing exponentially recently.   This is illustrated in Figure 2 below, which shows “other liquids” (i.e., ethanol) production reported by the EIA.



It is apparent that after reaching a peak in production in about 1990, ethanol production declined for a few years and then just took off like a rocket for the later half of the 1990s and throughout the 2000s. 

Figure 2 shows the best fit of the Hubbert equation to the 1980-1994 range of data (solid purple line: best fit parameters “a” = 0.24; Q∞ = 1.34; Qo = 0.12) suggesting a 24%/yr increase in ethanol production through this period—that’s pretty impressive, but, it is starting from a very small production rate in 1980. 

Figure 2 further shows the best fit of the modified Hubbert Equation to the 1995-2010 range of data with the values of “a” Q∞ and Qo fixed to their best fit values from the Hubbert fit to the 1980-1994 range, but fcq (yearly fractional change in Q) allowed to vary (fca fixed to 1).  The best fit value for fcq equals 1.065.  The Srss for the best fit value with fca (fractional change in “a”) varied (fcq fixed to 1) for 1995-2010 plus the best fit of the Hubbert equation for 1980-1994 was significantly better than the Hubbert equation fit to the full range of data (F-test p<0.0001).  The Srss for the best fit value with fca (fractional change in “a”) varied (fcq fixed to 1) was an order-of-magnitude larger than the Srss obtained with fca varied, and, varying both fca and fcq did not result in a significantly better fit (F-test p<.1).

The fcq value obtained from the best fit to the modified Hubbert equation is consistent with there being a yearly 6.5% increase in the Q∞, which in this case, is the total recoverable ethanol. 

Putting it all together—predicting Brazil’s petroleum production
I also fit the Hubbert equation to the sum of Brazil’s other petroleum sources as reported by the EIA, that is, the sum of crude oil, PGL, and RG.  Figure 3 shows this data (open circles) and the best fit to the data (long dashes: best fit parameters: “a” = 0.0816; Q∞ = 57.92; Qo = 1.31).


The best parameters are real close to the best parameters obtain from the fit the data presented in the BP statistical review (shown in Figure 1 and discussed above), but not identical, because the EIA data refinery gains and the BP data doesn’t include this.  Like the fit to the BP data, however, the production rate is predicted to top out at ~1.2 bbs/yr in 2025 and then decline thereafter.  The Q∞ value of 58 bbs is plausible if the sub-salt fields are added to the known reserves.  If this trend held it would make Brazil a major petroleum producer in the 2020s.  For instance, if this production trend and Saudi Arabia’s production were to continue as predicted then Brazil’s predicted production, for these non-ethanol sources, in 2025 would be just slightly below Saudi Arabia’ predicted production (1.4 bbs/yr). 

Also reproduced in Figure 3, for reference, is the production data for that “other liquid”—ethanol and the Hubbert equation and modifed Hubbert equation best fits (previously presented in Figure 2). 

It apparent, that at present, ethanol production (0.2 bbs/yr) is about four times smaller than the total production from Crude, LPG and RG.  However if these trends continued, by about 2038, ethanol production would actually exceed these other petroleum sources.   Is that even possible?  

Well, yes I think that it is.  Recall that the Brazil’s sugarcane zoning implies that about 7.5 times more land could be used than at present.  Potentially therefore, the present rate of ethanol production of 0.2 bbs/yr could be expanded to 1.5 bbs/yr.  That would be more than enough to exceed the production rates expected from the other petroleum sources.  And, by the mid-2030s I suspect that Brazil will have very strong financial incentive to do just that, if they can.

Finally, Figure 3 shows the sum of the production rates from Crude, LPG and RG and ethanol production.  The continued upward trend in production after 2025 declines somewhat, but still increases.  Of course, this is due the predicted increase in ethanol production.  Even if ethanol production were to increase at half the rate as projected in Figure 3, that would still be enough to flatten out Brazils total production as production rate from the conventional sources go into decline after 2025. 

Overall therefore, prospects look pretty good that Brazil has the potential to become a major net petroleum exporter for next few decades at least—provided that they don’t use it all themselves that is.

Non-linear least squares (NLLS) analysis of total petroleum consumption
Figure 4 shows the consumption data for 1965-2009 reported in the BP statistical review (open circles) and the production data for 1980-2009 reported in the EIA.  Like the production data, the number from these two sources are close but not identical to each other.  I can’t see what the reason for the difference is. 



Also shown are the best fits two these two data sets using the Hubbert equation.  The best fit parameters to the EIA data (solid line: “a” = 0.065; Q∞ = 54.1; Qo = 5.61) are very similar to the BP best fit parameters (dashed line: “a” = 0.060; Q∞ = 54.6; Qo = 3.16), except for Qo (because one data set starts at 1965 and the other starts at 1980).  The EIA and BP best fit curve predict a peak in consumption in about 2013 and 2014, respectively.  The annual rates of increase in production are estimated to be 6.5%/yr and 6.0%/yr, respectively.  The estimated total petroleum consumed at ~54 bbs is just slightly under the estimates of total petroleum produced (58 and 59 bbs, for the EIA and BP data respectively).

Therefore for the export land model analysis to follow it would not matter much which of these curves were used, but since I have used the EIA data for production, I will stick with the EIA data for consumption.

Predicting future trends in Brazilian petroleum exports
Figure 5 shows the EIA production and consumption data from 1980-2010, the best fits obtained using the Hubbert equation analysis of the consumption data and the sum of the Hubbert equation and modified Hubbert equation fits to the production data.



The predicted export curve (solid green line) is calculated based on the difference between the production and the consumption curves shown in the figure.  Additionally, I show the estimated “measured” export data from 1980-2009 (i.e., the BP statistical review reported production minus the EIA reported consumption).  Because the consumption rate has exceeded production the measured net exports are actually negative until 2007, goes negative in 2008 and then slightly positive in 2009.  The prediction curve suggests ever upward exports as domestic consumption peaks and declines in the early 2010s but production continues to increase, spurred on by increases in both convention petroleum and ethanol production until 2025, and continued increases in ethanol production at least through 2040.

Impact on USA
Figure 6 reproduces the USA production and consumption data and predicted trends, plus the data from the previously discussed top-nine petroleum exporters to the USA measured and predicted future exports, as presented in my previous articles in this series. 

The lime green line shows the addition of Brazils exportable petroleum to the USA, which equals the predicted total production (i.e., the solid blue line shown in Figure 4) multiplied by 0.12, corresponding to the percentage of Brazilian total production sent to the USA in 2009.

The lime green circles show my estimate of the last past few years of Brazils measured exports to the USA based on the best fit to the EIA production data and the percentages of USA exports, derived from the EIA data (using the linear regression analysis slope and intercept of the 1994-2010 data shown in the above graph, “Percent of Brazil's total production (EIA) exported to USA”).


Based on the predicted export trend, Brazil’s exports to the USA by 2015 are predicted to equal about 0.15 bbs/yr and 0.17 bbs/yr by 2020. That is 131% and 155%, respectively, of Brazils’s estimated exports to the USA in 2009 (0.111 bbs/yr). This upward trend continues through 2030 and beyond. If these trends are accurate, by 2025 Brazil will be third, only behind Canada and Iraq, as the largest important export source to the USA.

Well, at last I have completed my analysis of production and consumption trends for the USA and its top ten import sources .  It is time now to put it all together and apply this to an Export-land model analysis for the USA, and ponder the implications, which is what I will be doing for the next few posts.