Sunday, December 26, 2010

Trends in Angolan Petroleum Production and Consumption

As I noted in Where in the world does the USA import its oil from?, based on data from U.S. Total Crude Oil and Products Imports (“Total Imports”) I estimated that Angola is number eight of the largest petroleum providers to the USA in 2009.

Like Algeria, Angola has supplied substantial amounts of oil to the USA for at least the last 35 years. According to data in “Total Imports,” at least since 1985, Algeria has been among the top 10 of total petroleum importers to the USA.

Angola, after winning its independence from Portugal in 1975 fell into a 27-year civil war with the main two warring faction being the Soviet Union/Cuban-supported MPLA and the USA/South Africa-supported UNITA.

A third but important faction is the FLEC “Front for the Liberation of the Enclave of Cabinda.” Cabinda is important because that is the province where the majority of Angola’s petroleum comes from via off-shore oil fields. Cabinda province is isolated from the rest of Angola by the Congo. Although the civil war in Angola is officially over, the sub-factions FLEC are still an active militant force destabilizing Cabinda.

After a formal cease fire in 2002 Angola started to rebuild, largely supported by its revenue generated by its oil exports.  According to the EIA country analysis brief for Angola, oil accounts for over 90 percent of export revenue and over 80 percent of GDP. China has played a major role in the rebuilding of Angola’s infrastructure, no doubt in the hopes of gaining increased access to Angola’s nature resources, including its oil. (see e.g., Angola: China's African foothold).  Indeed, Angola may have just become China’s top import source in 2010, surpassing Iran and Saudi Arabia (see e.g., Angola becomes China's largest crude oil source).  It remains to be seen if this relationship between China and Angola continues to grow or wane.  For instance, there are signs that the relationship has cooled as indicated by Angolans protesting and attacking Chinese-imported workers, the non-payment of Chinese businesses and decisions by the Angolan state oil company, Sonangol, to block ownership interests by the Chinese-state owned oil company CNOOC.  (Hostility Toward Workers Cools Angola-China Relationship).  Its not clear to me if this is just posturing by Angola in order to get a better deal from China, or, a true change in attitude towards China since 2002.

In 2007 Angola became a member of OPEC and therefore, in principle, subject to OPEC-defined quotas on production.  The change in Angola’s estimated oil reserves after joining OPEC in 2007 gives a nice example of how OPEC members get around the quota system.  Here are yearly estimates of oil reserves from Angola (in billions of barrels), since 1990 as reported in the BP statistical review
I can see the jump from 2001 (6.5 bbs) to 2002 (8.9 bbs) as being due to the end of the civil war.  But the increase from 2006 (9 bbs) to 2007 (13.5 bbs) may include Angola’s joining OPEC and Angola wanting to give itself a higher ceiling under the quota system.

Nevertheless, the EIA country analysis brief points out that there are several up and coming off-shore projects in development and these projects, once online, should add substantially to Angola’s production capacity by about the mid 2010s:

Despite limitations imposed by OPEC, companies operating in Angola are on track to significantly ramp up their offshore developments in the short- and medium-term. Major projects summarized below represent tie-ins to support existing developments and also new developments that are underway. Industry analysts have estimated that Angolan production capacity could peak between 2.5 and 3 million bb/d by 2015 based on existing discoveries

Additionally, the EIA reports expected increased production in several existing fields which would push the production capacity even higher.  So, that jump in reserves in 2007 could be reflective of these new sources of oil—its not clear to me if all of these are figured into the 13.5 bbs estimated reserve or not.   Nevertheless at least to take advantage of this new capacity under the OPEC quota system, I expect to see Angola make new announcements of higher oil reserves in the coming years. 

According to the EIA country analysis brief, Angola’s exports in H1 2009 were split between about one-third going to the Americas (mainly the USA and then Canada) one-third going to Asia (mainly China, India and Taiwan) and the remainder going to Europe and South Africa: 
The above figure depicts the USA as the biggest single country importer, taking about 31% of Algerian net exports.  However, the EIA’s own data shows that Angolan annual exports to the USA have been on a down-trend since the early 90s, as shown from data in “Total Exports” and the EIA's Angola energy profile:
The downward trend since 1992, from about 70% in 1993 to about 25% in 2009, follows a fairly stronger linear trend (r2 = 0.86) with a slope of -2.6 percent per year.  If this trend were to continue then by 2019, Angola would export no petroleum to the USA!  I think that is unrealistic, but on the other hand, I don’t see anything that would cause the return of the percents of exports to the USA back to where it was in the 1990s.  China, India and probably others will be standing by to purchase any oil that the USA or Europe doesn’t take. 

As a compromise between applying this linear trend versus just taking a 5-year average, for the export analysis to follow, I have assumed that the percentage of Algeria’s exports to the USA will continue at its 2009 percentage (24.5%) going forward.   

Non-linear least squares (NLLS) analysis of total petroleum production
Figure 1 shows total petroleum production for 1965-2009 as reported in the BP statistical review.
After a fairly flat to declining low rates of production in the mid-70s to early 80s, probably due to being in the heights of the civil war, production picked up again from 1983 to 2001, and then shifted into high gear in 2002, with the end of the civil war. 

As I would expect by now with this sort of data, the Hubbert equation fit to the full data range blows up and looks essentially like an exponential fit (“a” = 0.0815; Qo = 0.22; Q∞ = 7124).  A Hubbert equation fit to the sub-period 1982-2009 also blows up and looks about the same (not shown; “a” = 0.0822; Qo = 0.85; Q∞ = 42082).  Extending this prediction curve out to about 2030 would put Angola’s production rate on par with Saudi Arabia, which I don’t think is realistic given what is presently known about the size of Angola’s reserves.

As shown in Figure 2, fitting the Hubbert equation to the sub-period 1982-2001 did not blow up (long dashed line; “a” = 0.158; Qo = 0.43; Q∞ = 7.23) and fitting the sub-period 2001-2009 also did not blow up (short dashed line; “a” = 0.240; Qo = 1.05; Q∞ = 11.47).  However, the sharp downturn in the fit to the 2001-2009 period, in my opinion, does not seem realistic given that there are several new projects about to come online in the mid 2010s, and, that will take the production up to about 3000 mbd (equivalent to about 1.1 bbs/yr).
Attempts to fit my modified Hubbert equation 9 to the 2002-2009 data range, using the best fit parameters from the Hubbert equation fit to the 1982-2001 range as the fixed values for “a” Qo and Q∞, did not give very satisfactory fits (not shown).  The fit where only fca was varied was poor; the fit where fcq was varied was a pretty close fit to the data but the best value of fcq of 1.14 rockets the production rate up to even higher values than the Hubbert fit. 

To produce a more reasonable model of production than an ever increasing exponential curve, I repeated the Hubbert equation fit to the full data range but with Q∞ fixed to the sum of annual production from 1965 to 2009, 8.3 bbs, plus the reported proven reserve of 13.5 bbs (i.e., Q∞ =21.8 bbs). It was apparent that using the sum of production to date plus 1x proven reserve value as the value of Q∞ doesn’t give a very good fit to the existing production data, as the best fit using this Q∞ predicts a peak production rate that is  less than the present production rate. 

In fact, as shown in Figure ,1 even setting Q∞ equal to 2x or 3x the reserve (i.e., 35.2 and 48 bbs, respectively) still doesn’t given a very good fit to the recent existing production data.  For instance as the best fit using the Q∞ at past production plus 2x the reserve value put the production rate lower than the EIA’s expectations regarding the upcoming projects and the total production capacity of about 1 bbs/yr .  The past production plus 3x the reserve value put production rates above 1 bbs/yr, but not until the mid 2020s which is a decade later than the EIA prediction.  Finally past production plus 4x the reserve value (Q∞ = 61.5 bbs) puts production at about 1 bbs/yr by 2016 which is pretty close to the EIA’s expectations regarding the timing of the upcoming projects and the added production capacity.

Therefore, for the export analysis to follow, I have used the best fit of the full time range to the Hubbert equation with Q∞ fixed to 61.5 bbls.  More on this assumption at the end of the article.

Non-linear least squares (NLLS) analysis of total petroleum consumption
The BP statistical review does not split out the data from Angolan petroleum consumption, so instead I used the consumption data from 1980-2009 as reported in the EIA’s Angola Energy Profile.  

Figure 3 shows the Angola’s consumption data from 1980 to 2009 as reported by the EIA. 

Note that the vertical scale is an order-of-magnitude lower than the scales shown in Figures 1 and 2—that is, like Algeria, historically, Angola has not used much of the petroleum it produces.  Of course, the quarter-century long civil war would explain this.

Around the late 90s there are some signs of increased consumption and then a substantial (about 50%) increase in 2001.   Thereafter consumption continues at this fast rate.

The Hubbert equation to the full data range (solid line in Figure 3) is not very good, so I performed another fit to just the data from 1996-2009 (long dashes best fit parameters: “a” = 0.0837; Qo = 0.11; Q∞ = 11.51).  This gives a much better fit, and it is this later fit that I used for my export analysis. 

Predicting future trends in Angolan petroleum exports
Figure 4 shows the production and consumption data from 1965-2009 and 1980-2009, respectively, the best fits obtained using the Hubbert equation analysis of the 1996-2009 time period of consumption data and the Hubbert equation fit to the 1965-2009 period of production data with Q∞ fixed at 61.5 bbls:
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 is so low these measured export rates are nearly identical to the production rates.

If these predicted production and consumption trends continue, it suggests that Angola will continue to be a net exporter for several decades to come, due to combination of production not peaking until the late 2020s and domestic consumption increasing from a very low present value.  Eventually, by the late 2020s, production peaks and the yearly increase in consumption (8.4%/yr, based on the best fit value of “a”) starts to substantially cut into exports, but still, these curves predict that there would be net positive exports until at least the 2050s.

Impact on USA
Figure 5 reproduces the USA production and consumption data and predicted trends, plus the data from the previously discussed top-seven petroleum exporters to the USA measured and predicted future exports, as presented in my previous article, Trends in Algerian Petroleum Production and Consumption.
The indigo-blue line shows the addition of Angola’s exportable petroleum to the USA, which equals the predicted total exports (i.e., the solid green line shown in Figure 3) multiplied by 0.245, corresponding to the percentage of Algeria total exports sent to the USA in 2009. The indigo-blue circles show my estimate of Angola’s measured exports to the USA based on the BP statistical review and EIA production and consumption data and the percentages of USA exports, derived from the EIA data (shown in the above graph, “Percent of Angolan net Exports to USA”).


Based on the predicted export trend, Angola’s exports to the USA by 2015 are predicted to equal about 0.22 bbs/yr and 0.27 bbs/yr by 2020. That is 143% and 179%, respectively, of Algeria’s estimated exports to the USA in 2009 (0.153 bbs/yr). This upward trend continues until 2027-2030 when the exports peak at 0.32 bbs/yr and then declines as production goes past its peak and domestic consumption in Angola become relatively more significant.

If these trends are accurate, then by 2018, Angola will be second only behind Canada as the most important export source to the USA as Venezuela, Saudi Arabia, Nigeria, Russia and Algeria exports go into decline.

There is more speculation built into these export predictions than I would like, and I could imagine this being on the optimistic side. For instance if the reverse of 13.5 bbs turn out to be an accurate estimates of the present field and new fields about to come on line, then the production curve will follow more closely the curve corresponding to Q∞ = 21.8 bbs. That would leave far less exports, although Angola would still be an important exporter at least out to the early 2030s. Or, if the 15-year trend for steadily declining percentage of exports to the USA continues, then even if Angola had petroleum to export, the amounts going to the USA could become insignificant by the late 2010s. Although I would not discount either of these scenarios, I would like to see more evidence to support them before changing my prediction. 



Wednesday, December 29, 2010
I am not happy with the analysis I did of Angola's petroleum production. I describe below my second attempt at getting a more realistic prediction curve.

I was not happy with my previous analysis because I had assumed a Q∞ value that is 4x the estimated reserves, plus past production, to get the dQ/dt value up to level and time frame that the EIA had predicted based on the upcoming projects (e.g., at least ~1 bbs/yr by the mid 2010s) plus the added production capacity. However, there is no basis for my assuming a Q∞ of 61.5 bbs to get to that dQ/dt. Also, in retrospect I realized that it was incorrect to add past production to Q∞ for the purposes of this analysis because I am only trying to cover a time span from 2001 and on. So, Q∞ should have been fixed equal 13.5 bbs.

Also after re-considering this analysis, I realized that I could take a different approach to get the production rate up to the EIA predicted level and time frame: fix Q∞ equal to 13.5 bbs and fix "a" to the best fit value from fitting the sub-period 2001-2009 (short dashed line in Figure 2; “a” = 0.240). In other words assume that production continues at the same rate as it has been in recent years, and, assume that total recoverable oil equals 13.5 bbs.

Figure 6 shows the result of such a fit using the Hubbert equation (alternating two short and one long dashes), again using only the data from the sub-period 2001-2009. Also for reference, I put two "Xs" to mark the spots that would correspond to a production rates of 2.5 mbd (0.91 bbs/yr) and 3 mbd (1.095 bbs/yr) in 2012 which is the production range estimated by the EIA.

Since "a" and Q∞ are fixed, the NLLS analysis amounts to adjusting Qo to get the best fit to the data (best fit Qo = 0.68). That is, I have esstentially fixed the steepness and area of the curve and used NLLS to adjust the lateral position of the curve in the plot of dQ/dt vs time to best fit the 2001-2009.

Given these assumptions, the production rate would top out at about 0.8 bbs/yr.  That falls short of the range predicted by the EIA. That is, even if the production rate continues to increase at its present rate, for the given estimated reserves of 13.5 bbs, we can get to about 1 bbs/yr, which is mid-range of the EIA estimate.

After further reflection, I realized that I could use the "constraints" feature of the SOLVER function in EXCEL to help estimate what the rate of increase in production rate (i.e., what the value of the rate constant "a" would have to be to give a dQ/dt value of 1 in 2012. That is, in addition to constraining the value of Q∞ equal to 13.5 bbs, SOLVER allows me to constrain the value of dQ/dt equal to 1 specifically in 2012. Then I can perform the NLLS analysis with "a" and Qo allowed to vary to get the best fit to the production data for 2001-2009.

The results of this best fit are also shown in Figure 6 (solide line: "a" = 0.297; Qo 0.46; Q∞ fixed to 13.5; dQ/dt fix to 1 in 2012).

As I expected, the value of "a" has to increase to an even higher value ("a"=0.30) than the value of "a" for the best fit for the 2001-2009 period ("a"=0.24).  So, we are talking about a year-over-year increase in the production rate of about 30% in order to get to the mid-range predicted by the EIA by the mid 2010s. I suppose that that this is possible, since there an annual growth rate in production of 24% over the 2001-2009 period. But it would have to be one of the steepest petroleum production growth rates that the world has every seen.

Of course, as is inherent for any model based on the Hubbert equation, the decline side of the production curve is predicted to be just as steep as the growth side of the curve. That means that after peaking in 2015 production will drop steeply (i.e., 30%/yr) and as illustrated in Figure 6, by 2016, production will be back down to the same level as in 2009, and by 2021, down to about the level it was at in 2001.

So, in the absence of a further increase in reserves, Angola would have a decade of massive production and then just as rapid a return back to the same levels of production at the end of the civil war—would Angola's leaders really do that? Sure they would, there would be massive some profits to re-building the country's infra-structure, and, some profits for them to gain as well.

Figure 7 is a revised version of Figure 4, using this revised production curve (solid line) shown in Figure 6.
Once again, because the consumption rate is so low, the predicted export curve (solid green line) is nearly identical to the production rate at least until the mid 2020s when production decreases, and increased domestic consumption, causes net exports go to zero by 2025.

I think that this more realistic production curve gives a more realistic prediction of Angola's exports as it pertains to the USA.  This is shown in Figure 8, which is a revised version of Figure 5, above.

The curve still predicts that Angola will be an important petroleum exporter to the USA for the next decade. Based on the predicted export trend, Angola’s exports to the USA by 2012 peaks out at about 0.24 bbs/yr, which is higher than that predicted for Algeria or Russia but less than Nigeria. By 2015, exports are about 0.19 bbs/yr which is still higher than in 2009 (0.15 bbs/yr), but by 2020, down to about 0.06 or about 37% of exports in 2009, and then zero by 2025.

Thursday, December 23, 2010

Trends in Algerian Petroleum Production and Consumption

As I discussed in Where in the world does the USA import its oil from? Algeria is number seven on the list of largest petroleum provider to the USA in 2009 based on data from U.S. Total Crude Oil and Products Imports (“Total Imports”). 


Algeria has long been an important supplier to the USA.  For instance, according to data in “Total Imports,” at least since 1975, Algeria has been usually among the top 10 of total petroleum importers to the USA.

According to the EIA’s country analysis brief for Algeria,  the State-owned entity, Sonatrach, controls the oil and natural gas sectors in Algeria, and foreign companies have to enter into partnership agreements with Sonatrach to develop Algeria's oil fields.  The EIA country analysis brief indicates that Algeria’s plan is to maintain crude oil production capacity at about its current level, and points hopefully to Sonatrach’s announcement of the discovery of 16 oil fields since the beginning of 2009 as a sign that this will help replace declines in older fields. 

However, also in 2009 we have Algeria’s Oil Minister announcing the plan to go look for oil and gas in shale and compact formations “to ensure a safe and durable supply of hydrocarbons and petrochemical products to our traditional partners in Europe, Asia and North America.”  (from: Algeria Looking for Shale Gas Deposits, Study Nuclear Power, Minister Says)

To me, this looks like a sign that conventional oil is at or near peak and that now the less attractive sources such a shale oil, requiring substantial foreign investments, is what will be in store for Algeria’s future production. 

According to the EIA country analysis brief, Algeria’s exports are split between about one-third going to North America (the USA and Canada) and two-thirds going to QECD countries like France, Germany, Italy, UK, Japan and South Korea: 

The USA was the biggest single country importer, taking about 27% of Algerian net exports in 2009.

The EIA's Algeria energy profile and “Total Imports” indicates that Algeria’s exports to the USA have trended upwards since at least the mid 90s:




The upward trend, however, is not very strong (linear regression r2 = 0.5).  Therefore for the export analysis to follow, I have assumed that the percentage of Algeria’s exports to the USA will continue at its past 5-year average of 30.5±4 percent.   

Non-linear least squares (NLLS) analysis of total petroleum production
Figure 1 shows total petroleum production for 1965-2009 as reported in the BP statistical review.



After a gradual increase in the production rate through the 1970s and 80s, production peaked in the early 90s and then took off in the late 90s until about 2004, where production plateaued for about four years and then declined in 2009.  The EIA might want to write-off that decline in 2009 to OPEC quota restrictions, but that would not explain why there was a plateau in production from 2004 to 2008 when world oil demand was high and prices were rising.   Again, to me this looks like a sign of a production peak.

Figure 1 shows the best fit of the Hubbert equation to the entire data set (solid line best fit parameters: “a” = 0.0326; Qo = 9.33; Q∞ = 97.51).   This didn't look like a very good fit to the data as a whole, and so I tried two more Hubbert equation fits to two different smaller time segments: 1965-1994 (long dashed line, best fit parameters “a” = 0.04790; Qo = 7.44; Q∞ = 39.83) and 1995-2009 (short dashed line, best fit parameters “a” = 0.105; Qo = 9.33; Q∞ = 27.24).  As suggested by the difference in value of the rate constant "a" for these two periods, there was a more than a doubling in the production rate from the 1965-1994 period to the 1995-2009 period.

I also tried a fit using my modified Hubbert equation 9 to the 1995-2009 time span,  in which parameters “a” Qo and Q∞ were fixed to the best fit parameters for the 1965-1994 time span, and one or both of fca or fcq were varied. 

Figure 2 shows the best fit for the case where fca was allowed to vary and fcq was fixed equal to 1:


The best fit value for fca was 1.043 (solid green line).  Allowing only fcq to vary gave a poorer fit, and allowing both fca and fcq to vary did not result in a significantly better fit (F-test, p>.1).  This suggests that the behavior of Algerian production since 1995 is most consistent with an about 4.3 percent per year increase in the production rate with no increase in Q∞.  That is, it looks like Algeria has just been pumping the oil that it has at an increasing rate, without expanding its reserves. 

The best fit using modified equation 9, however, did not give a better fit than the Hubbert equation fit to the 1995-2009 time period so I will stick with the Hubbert equation fit for the further analysis to predict future exports.  I note however, that as illustrated in Fig. 2, these two fits for the 1995-2009 period (solid green and short dashes) are nearly identical, so it would not have mattered too much which one I had chosen to use for my export analysis. 

Non-linear least squares (NLLS) analysis of total petroleum consumption
Figure 3 shows the Algerian consumption data from 1965 to 2009 as reported in the BP statistical review:
 



The first thing to note about this plot is that the vertical scale here is about an order-of-magnitude lower than the scales shown in Figures 1 and 2—that is, Algeria doesn’t use much of the petroleum it produces.  However, after a plateau in the mid 90s and then down turn in the late 90s, consumption took off in strong upswing through the 2000s. 

That upward swing in the last decade causes the Hubbert equation to the full data range (solid line in Figure 3) to be quite a poor fit.  The Hubbert equation fit to the smaller period from 1965-1996 (long dashes best fit parameters: “a” = 0.130; Qo = 0.067; Q∞ = 2.34) was much better, however. 

To better model the consumption trend from 1997 to 2009, and beyond, I again turned to my modified Hubbert equation 9, setting “a” Qo and Q∞ to the best fit parameters for the 1965-1996 time span and allowing one or both of fca or fcq to vary.   



Figure 4 shows the fit results for the case where fcq equal to 1.04 and fca fixed equal to 1.  The overall best fit to the 1965-2009 time span using the Hubbert best fit parameters for 1965-1996 plus the modified best fit for 1997-2009 tme span (sum of  rss = 0.0007) was significantly better (F-test p<0.0001) than the fit obtained using the Hubbert equation for the full 1965-2009 time span (rss = 0.0117).  Allowing fca to vary and fixing fcq=1, did not result in a better fit, and varying both fca and fcq did not result in a significantly better fit (F-test, p>.1).

The value of fcq equal to 1.04 is consistent with the Q∞ for consumption increasing at a yearly rate of 4%/yr for the period 1997-2009.  I extended this trend out to 2030 in Figure 4 to illustrate that, although domestic consumption has been fairly low in the past, it will start to more substantially eat into production in the future if the consumption trend continues. 

Predicting future trends in Algerian petroleum exports
Figure 5 shows the production and consumption data from 1965-2009 and the best fits obtained using the modified Hubbert equation analysis of the 1997-2009 time period of consumption data and the Hubbert equation fit to the 1995-2009 period of 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 1997-2009 (i.e., the BP statistical review reported production rate minus the reported consumption rate). 

The predicted production and consumption trends suggests that Algeria’s net exports would end in 2029, due to combination of declining production and increasing domestic consumption.

Impact on USA
Figure 6 reproduces the USA production and consumption trends, plus the previously discussed top-six petroleum exporters to the USA measured and predicted future exports as presented in the previous article, Trends in Russian Petroleum Production and Consumption.



Based on the predicted export trend, Algeria’s exports to the USA by 2015 is predicted to equal about 0.151 bbs/yr and only 0.102 bbs/yr by 2020.  That is 91% and 62%, respectively, of Algeria’s estimated exports to the USA in 2009 (0.162 bbs/yr).  If the trends continue, somewhat similar to Nigeria, Algeria will continue to be an increasingly important exporter to the USA in the 2010s, overtaking Mexico by 2012 and Russia by 2018 in terms of its exports to the USA, before dropping to very low levels of exports by the mid 2020s, and none by 2029.

Algeria adds another small sliver of petroleum to the USA's total import sources, at least until 2029.

Saturday, December 18, 2010

Where in the world does the USA import its oil from?

This article is a follow-up to address a question I posed at the end of the article, Trends in Nigerian Petroleum Production and Consumption.

After considering the USA’s petroleum imports from its top five source countries, I pondered the question: where did the additional ~1-2 bbs/yr petroleum come from for the years 2004-2009 to account for the discrepancy between the USA’s consumption and the USA’s domestic production, plus the imports from its top 5 importers? 

I can now re-phase this question, “from the top 6 importers,” because after considering number six, Russia (see Fig 4 also reproduced below), there is still an about 1.5 bbs/yr discrepancy.


Even after considering the top six petroleum importers and domestic production, I still can not account for about 1.5 bbs/yr worth of petroleum imports into the USA especially in those peak consumption years around 2005-06. 

Where did that petroleum come from?

Since posing this question, I have found a source of information that helps shed some light on where in the world the USA imports its oil from.  The EIA has published a spread sheet entitled, U.S. Total Crude Oil and Products Imports (hereinafter, “Total Imports”). 

 

“Total Imports” is a massive spreadsheet that presents average US imports (in units of 1000s of barrels per day) of Crude Oil plus Petroleum Products, for the years 1973 to 2009 from about 30 to 90 difference source countries depending on the year being considered.  

It is just what I needed to answer my question.

 

Top Suppliers of Petroleum Imports to the USA in 2009

One thing I would like to correct, from the start, is a mistaken assumption that I had made about Fig. 5.4 of the Annual Energy Review for 2009 (and its associated Table 5.4).  Here it is one more time:


My mistake was to assume that the EIA was actually showing the top nine countries in their order of importance.  I should have been made more suspicious by the title, “Selected Countries.” 


According to the data presented in “Total Imports,” however, the top 10 countries exporting petroleum to the USA are presented below.



Notice that I have added to Venezuela, the separate entry in “Total Imports,” of imports for the US Virgin Islands (VI).  As already discussed in Trends in Venezuelan Petroleum Production and Consumption, the VI imports substantially all of its oil from Venezuela, and then passes the oil and petroleum products on to the USA.  This would actually put Venezuela + VI ahead of Mexico as the second largest petroleum exporter to the USA in 2009.

 

I find it curious that Fig. 5.4 of the Annual Energy Review excludes Algeria and Angola but rather includes the UK.  Why would the EIA choose to not feature the imports from Algeria and Angola, which together account for about 8 percent of the total exports, and instead, show imports from the UK, which provided only about 2 percent?  According to “Total Imports,” the UK is only the twelfth largest supplier, behind Columbia at 2.4 percent.   Actually, thirteenth, if I were to consider VI separately (VI would be tied with Columbia at 2.4%). 

 

Maybe the EIA’s goal was to portray the UK as still an important supplier, or, to suggest (by omitting Algeria and Angola) that petroleum from Iraq or Brazil are relatively more important than they actually are. 


The values in parenthesis above the bars in Figure 1 show each country’s the exports to the USA as reported in “Total Imports,” after conversion of 1000s of barrels per day into units of bbs/yr.  The sum of the imports from these ten countries (or 11, if you wanted to count VI separately) equals about 78% of the USA’s imports for 2009 or about 3.3 bbs/yr.  Added to my USA production for 2009 (2.6 bbs/yr; from the solid blue data point at 2009 in Fig. 4, from Trends in Russian Petroleum Production and Consumption, reproduced above) gives about 5.9 bbs/yr or about 80 percent of my estimate of USA’s consumption for 2009 (7.4 bbs/yr; from the solid red line in Fig. 4 above), or, about 87 percent of the reported consumption (6.8 bbs/yr; from the red data point in Fig. 4 for 2009).

To account for a larger portion of US imports, I considered all the countries in “Total Imports,” that contributed at least about 1 percent to the total US imports for 2009.  I ended up with sixteen countries (Norway contributed only 0.92 percent, but I accepted this as about 1 percent) shown in Figure 2 below:



The sum of the imports from these sixteen countries equals about 88% of the USA’s imports for 2009,  about 3.7 bbs/yr.  Added to USA estimated production for 2009 (2.6 bbs/yr) gives about 6.3 bbs/yr or about 85 percent of my estimated consumption for 2009 (7.4 bbs/yr), or, about 93 percent of the reported consumption (6.8 bbs/yr).

To account for the remaining 12% of the USA’s imports, I would have to “scrape lower into the barrel” of importers so to speak.  According to “Total Imports,” there were 82 countries that made a "non-zero" contribution to the USA’s imports in 2009, which for the purposes of the spread sheet was at least 1000 barrels/day. 

The sum from all 82 countries equaled 4.3 bbl/yr which added to US production of 2.6 bbl/yr gives 6.9 bbl/yr which is just slightly higher than the reported consumption of 6.8 bbl/yr.  Perhaps the discrepancy is a rounding error, or, because the 6.8 bbl/yr value is from the BP Statistical Review, while the 6.9 bbl/yr is from the EIA’s “Total Imports.” 

Thirty-three countries each contributed from 0.1 to 0.9 percent of USA’s total imports.  These countries, 17 through 50, in total contributed about 11 percent of the USA's imports.  The top five among this group are Equatorial Guinea (0.76%); Trinidad & Tobago (0.72%); Libya (0.68%); Azerbaijan (0.64%) and Chad (0.59%).  The remaining thirty-one countries (e.g., countries like Japan, Qatar, China, Egypt, Malaysia etc...) each contributed less than 0.1 percent of USA's imports, and in total contributed about 1% of the USA's imports.

Historic Petroleum Imports to the USA from the Top Importers of 2009
Figure 3 shows how well these top 10, or top 16, import source countries account for USA’s imports in the past.  Figure 3 is a variation on Figure 4, which still shows production and consumption data from the USA and my best fits, but replaces the export data derived from BP statistical review, and my NLLS analysis of this data, with the import data presented in “Total Imports.” 



Also shown is the additional contribution from the remaining countries 17-81 (with VI’s exports combined with Venezuelan).  

Figure 4 casts the data shown in Figure 3 in percentage terms.  That is, for each year, USA reported production plus the imports from the top 10 importers for 2009 (i.e., the countries shown in Fig.2), the top 16 importer for 2009 (i.e., the countries shown in Fig. 3), and all of the importers (i.e. at least 1000 b/d) for the year in question are shown as a percentage relative to reported consumption for the year in question.


For the entire 36 year period, USA production plus the top 10 importer from 2009, on average, accounted for 85.8±2.5 percent of USA consumption.  USA production plus the top 16 importers from 2009, on average accounted 90.3± 3.3 percent of USA consumption.  Considering only the 28 year time span from 1981 on, to exclude the dip centered at 1978, did not substantially change these averages (86.4±2.2 and 91.8±1.1, respectively).

That dip centered at 1978 is due the fact that, back then, there were significant sources of imports from a number countries that are not among the top 10 or top 16 providers in 2009, as illustrated in Figure 5 below:


Figure 5 considers all of the countries in 1978 that contributed at least about 1% to the USA’s total imports in that year.  As before, I combined the contributions from Venezuela and VI (8% and 5.3% respectively).  I might well have also added to Venezuela the petroleum from other Caribbean countries like Netherland Antilles, Bahamas and Puerto Rico (Trinidad & Tobago produces much of its own oil) but I couldn’t verify that, like VI, they were importing from Venezuela and then re-exporting to the USA.  

USA imports from Iran ended in 1979 after the Iranian-hostage taking incident and imports from Libya ended in 1982 after the USA declared an import/export ban.  As indicated above, since about 2005, after then ban was lifted, Libya has been making a come back as an important import source.  Substantial imports from Indonesia continued for another decade, but by the mid-1990s was less than 1%. and continued declining  Petroleum from the UAE dipped below 1% after 1980. 

Using the top 10 importers to predict USA consumption
Not withstanding the discrepancy in late 1970s, the sum of USA production plus the imports from the top 10 importers for 2009 accounts for a fairly constant 86 percent of USA consumption for a plus-three decade period of time.  This suggests that these data would be useful for predicting the USA’s consumption going forward.

For instance, consider Figure 6 below, which presents the same USA production and consumption data as shown in Figure 3:
Additionally, I have shown the sum of USA production plus the imports from the top 10 import sources (dashed green line) and this sum divided by 0.86 (solid green).  With the exception of the late 1970s, the solid green line tracks the reported USA consumption data pretty well.

This analysis gives me some assurance that by just considering and predicting USA production and imports from the top 10 importers, I can make a reasonably accurate prediction of USA consumption going forward. 

Now, I just need to finish the analysis for the remaining top 10: Algeria, Angola, Iraq and Brazil!

Next stop...Algeria. 

Saturday, December 11, 2010

Trends in Russian Petroleum Production and Consumption

Russia is number six on the list of largest petroleum provider to the USA according to Annual Energy Review for 2009 (see e.g., Fig. 5.4 and Table 5.4).  According to the EIA’s country analysis brief for Russia,  most of Russia’s oil presently comes from Western Siberia, although untapped oil from Eastern Siberia, the Caspian Sea, and Sahkalin (in the Far East ) may play a larger role in the future.

Relatively little of Russia’s oil exports finds its way outside of Europe and Asia—only about 6% were exported to all of the western hemisphere: 


Unlike Saudi Arabia or Nigeria, which have had sent a fairly steady percentage of their net exports to the USA, Russian exports to the USA have come from essentially nothing to eight percent over the last decade. 

For instance, the EIA Russian energy profile indicates that Russia’s exports to the USA have gone from 24 in 1998 to 465 mbd in 2008.  Additionally the EIA’s published spread sheet entitled, U.S. Total Crude Oil and Products Imports (hereinafter, “Total Imports”)  indicates that in 2009, USA exports from Russia further increased to 563 mbd.  As a percentage of Russia net exports, this puts exports to the USA alone at 8%. So, the above chart from the EIA is somewhat out-of-date.

Therefore, although the USA is still a relatively minor consumer of Russian petroleum exports as compared to a number of European countries, there is a strong trend for Russia to export a greater proportion of its petroleum to the USA.  For instance, a linear regression analysis of the data presented in the above figure has an r2 of 0.91 (slope 0.602; intercept -1201), suggesting a near linear increase in the percentage of exports to the USA equal to about 0.6%/yr.

For the analysis to follow, I have assumed that Russia’s exports to the USA will continue to increase according to this linear trend (0.6%/yr) going forward.
   
Non-linear least squares (NLLS) analysis of total petroleum production
Figure 1 shows total petroleum production for 1985-2009 as reported in the BP statistical review (it appears that prior to 1985, production was grouped under, “Other Europe & Eurasia”).

After a local peak in production in late 1989, there was a large, about 50 percent, decline in production in the early 90s, no doubt reflecting the collapse of the Soviet Union, a plateau through the mid 90s, then an up-turn in production in the late 90s and indications of a plateau since about 2006. 

Figure 1 also shows the best fit of the Hubbert equation to the 1999-2009 data (solid line best fit parameters: “a” = 0.17; Qo = 14.4; Q∞ = 85.6). I did not attempt a fit using my modified equation 9 to such a short time span of data. 

The best fit to the 1999-2009 data looks a lot like the best fit to Mexico’s production data from 1999-2009 and, if this trend holds, it suggests a rapid decline production (e.g., half of peak production in 2009 by about 2018).  Perhaps this trend will break with new production in eastern Siberia or other locations, but I see no trend in the data to support this.

In the export analysis to follow, I have using the NLLS analysis of the Hubbert equation best fit to the 1999-2009 data.
 

Non-linear least squares (NLLS) analysis of total petroleum consumption
Figure 2 shows the best fit of the Hubbert equation to Russia’s petroleum consumption data (as reported in the BP statistical review.  As illustrated in Figure 2, Russia’s domestic consumption since 1998 has been fairly flat and about half of its peak consumption in 1989. 

The red line in Figure 2 shows the best-fit of the Hubbert equation to the full data set from 1998-2009 (“a” = 0.036; Qo = 39.6; Q∞ = 110).  The best fit suggests that domestic consumption during this period may peak in 2014 although the rate of change in consumption is fairly flat.

Predicting future trends in Russian petroleum exports
Figure 3 shows the best fits obtained using the Hubbert equation analysis of the 1998-2009 time span of consumption data and the 1999-2009 time span of production data as well the 2002-2009 time span of 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 “measured” export data from 1991-2009 time period (i.e., the BP statistical review reported production minus the EIA reported consumption). 

The predicted production and consumption trends suggests that Russia’s net exports would end in 2023, due to declining production and steady consumption at about 1 bbs/yr.

Impact on USA
Figure 4 reproduces the USA production and consumption trends, plus Canadian, Mexican, Venezuelan, Saudi Arabian and Nigerian petroleum measured and predicted future exports presented from the previous article, Trends in Nigerian Petroleum Production and Consumption.
 
The dark red line shows the addition of Russia’s exportable petroleum to the USA, which equals the predicted total exports (i.e., the solid green line shown in Figure 3) multiplied by the percentage of exports calculated from the slope and intercept obtain from the linear regression analysis done on the data in the graph, “Percent of Russian net exports to USA.” presented above.  The dark red circles show my estimate of Russia’s measured exports to the USA.


To make a better estimate of measured exports to the USA, I have refined my calculation as compared to previous articles in this series.  Namely, rather than apply the last year or several year average percentage of Russia’s exports to the USA to the past data, I have applied the actual year-by-year percentages to the USA, as calculated from the first graph in this article (extracted from the EIA energy profile, and “Total Imports”).  Thus the estimated measure exports to the USA equals the total measured production minus total consumption, the green circles in Figure 3, times this variable percentage.  I only depict measured values from 1998 and on because prior to 1998, there were substantially no exports to the USA from Russia.  

I also went back and performed the same type of correction to the measured estimated exports to the USA for Nigeria and Saudi Arabia —however the effect of this correction is barely visible in Figure 4.

Based on the predicted export trend, Russia’s exports to the USA by 2015 is predicted to equal about 0.178 bbs/yr and only 0.066 bbs/yr by 2020.  That is 85% and 31%, respectively, of Russia’s estimated exports to the USA in 2009 (0.21 bbs/yr).  Remember, these predictions are based on my assumption that Russian exports will continue to increase at 0.6%/yr (i.e., about 11% and 14% of Russian net exports in 2015 and 2020 respectively).  However, this does not mitigate the overall downward trend in Russian production, and hence, net exportable petroleum.  Nevertheless, for a few years, Russia’s importance as a supplier of petroleum to the USA will increase.  For instance, if the export trends for Mexico and Russia continue, by 2012, Russia should overtake Mexico in importance as a supplier to the USA. 


For 2010, I predict that Canada, Mexico, Venezuela, Saudi Arabia, Nigeria and Russia’s exports, together with USA’s domestic production, will provide about 68% of the USA’s predicted consumption.  By 2015, USA domestic production and imports from these current top six foreign providers will only amount about 56 percent of the USA’s predicted consumption.  Once again, by 2020 the USA is down to getting exports essentially from Canada and a small amount from Nigeria


Moreover, if the trend for a production decline does continue, I would expect that it will be most cost-effective for Russia to supply Europe through the pipeline networks connected to Russia, and to drop or reduce the percentage of net exports to its more marginal customers such as USA and China.


That small sliver of additional Russia oil added to the summation curve shown in Figure 4, tells me that I have reached a stage in this analysis where the further consideration of the USA's imports from other countries will be adding ever-smaller amounts to the summation curve.  I will continue to explore this for a few more countries, maybe up to the top ten of exporters to the USA, but I expect that the additions of petroleum exports for countries seven through ten will probably only bring the total up to about 75-80% of the USA’s predicted consumption for 2010.  The remaining 25-20% would have to be filled by more minor suppliers.


But, perhaps, one of countries seven through ten will show some signs of expanded exports going forward?