By David Strahan. Published in Geoscientist, and Petroleum Review, April 2007.
When it comes to estimating the scale of oil and gas resources, the United States Geological Survey has a reputation of coming up with some very large numbers. Nobody quite knows why, but its estimates always seem to be at the upper end of the spectrum, and on occasion have turned out to be absurdly over-optimistic. As a result, some observers have dismissed the most recent USGS assessment of the world petroleum endowment as inflated and irrelevant. After studying the fractious oil depletion debate for the last two years, however, I have come to exactly the opposite conclusion: the Survey has provided numbers that are not only relevant but which also strongly and unexpectedly support the case for an early peak in global oil production.
The USGS’s reputation goes back a long way. When the Shell geologist M. King Hubbert published his now legendary paper in April 1956, with its startling prediction that American oil production would peak and start to fall within 15 years, the Survey’s Deputy Chief Geologist Vincent McKelvey was among those who tried to discredit the forecast by promoting an estimate of the US oil resource that was almost three times higher than the biggest number used by Hubbert. This mattered because the central hypothesis of Hubbert’s work was that oil production tends to go into terminal decline at about the midpoint of depletion, when half the oil that will ever be produced from a given region is still underground. So the bigger the resource estimate, the longer the peak would be deferred. As I describe in The Last Oil Shock, this may have been McKelvey’s aim.
The USGS numbers were proved conclusively wrong in 1970 when US oil production peaked and started to fall right on schedule, and by the “Arab oil embargo” three years later, which rubbed in just how right Hubbert had been; despite the most intense political and economic incentives to produce more oil, American output continued to drop. McKelvey was finally sacked in 1978, an event about which the USGS website is understandably coy.
So when at the turn of the century the USGS released a new estimate of global oil resources that seemed to be wildly optimistic, even mildly skeptical observers could be forgiven for suspecting the Survey was up to its old tricks. The World Petroleum Assessment – 2000 provoked howls of disbelief with its estimate that total production and reserves of conventional crude and natural gas liquids could amount to 3,300 billion barrels, substantially higher than all other credible estimates. Yet subsequent events have recalibrated these numbers in such a way as to change their message entirely.
To be precise, the World Petroleum Assessment did not set out to produce a forecast of ‘ultimate’ – the total oil that would ever come out of the earth’s crust – but rather an estimate of the amounts of oil, gas and natural gas liquids “that have the potential to be added to reserves in the next 30 years (1995 to 2025).” To do this the Survey produced two separate estimates: one was based on a geological appraisal of how much oil might be discovered during the forecast period; the other a calculation of how much supply could be expanded through ‘reserve growth’ – the tendency for known fields to yield more than originally expected as time goes on, through some combination of conservative initial estimates and technological improvements. The USGS concluded that reserve growth could add 612 billion barrels of crude oil by 2025, and discovery a further 649 billion barrels. Discovery and reserve growth of natural gas liquids (NGLs) would add another 249 billion barrels. When added to industry data for historical production and known reserves, the total amounted to over 3,300 billion barrels. This figure was all the more extraordinary because the USGS had not even assessed the entire world.
The World Petroleum Assessment was attacked not only for the size of its numbers, but also the relevance of the methodology that produced them. As part of the discovery estimate, for example, the USGS attributed 47 billion barrels to offshore East Greenland, provoking some hilarity. The oil may exist, of course, but how many companies would rush to drill in the area that produced the iceberg that sank the Titanic? As for reserves growth, many analysts objected to the extrapolation of formulae derived from the history of US oilfields to the rest of the world, which they claimed was inappropriate. But these arguments may now be academic in the light of a further study carried out by the USGS in 2005, which assessed how well the forecasts of the World Petroleum Assessment were performing against reality, and which now allows us to draw some useful conclusions from the original numbers.
The USGS self-assessment showed that their reserve growth forecast for crude oil was on track but their discovery forecast was badly adrift. One quarter of the way through the original forecast period, actual reserves growth had delivered about the right proportion of the Survey’s prediction, but real-world oil discoveries were underperforming by almost 60 percent. The USGS discovery forecast implied that an average of almost 22 billion barrels of oil would have to be found every year for thirty years, but in the first 8 years of the forecast period discovery had averaged just 9 billion barrels annually.
The USGS sought to explain away the discovery deficit as the result of a relatively low oil price and low levels of exploratory drilling, implying that discovery could rise in future. But these arguments do not stand scrutiny. The volume of oil discovered annually has been in decline for forty years, while the price of oil has risen fairly strongly over the same period. As Harry Longwell, then Executive Vice President in charge of exploration and production for ExxonMobil, wrote in 2002: “Most of our discoveries were made in a much lower price environment than today….contrary to some widely held beliefs, discovered volumes have not been closely related to price fluctuations”. He went on to remark, “Another trend is much clearer than price. It’s getting harder and harder to find oil and gas”.
As a result, oil discovery is much more likely to keep on falling in line with its historical trend than to start rising again. But even if it were to remain steady at 9 billion barrels a year until 2030 – highly unlikely – the undershoot against the USGS forecast would still be almost 400 billion barrels (22 – 9 x 30 = 390), meaning that the USGS ‘ultimate’ for conventional oil should be reduced to 2,600 billion barrels at the very most, and 2,830 including NGLs. (The USGS did not reassess its original NGL forecast, but since both its oil and gas discovery forecasts are underperforming by almost exactly the same margin, and since NGLs are a by-product of natural gas production, I have deflated the USGS NGL discovery forecast by the same proportion). So what can these recalibrated USGS figures tell us?
Hubbert’s contention that oil production tends to go into terminal decline at about the midpoint of depletion has been demonstrated repeatedly the world over. The midpoint of our recalibrated ‘USGS ultimate’ including NGLs is 1,415 billion barrels. According to figures from IHS Energy, cumulative production to the end of 2006 was 1,100 billion, leaving 315 billion to go. Production of conventional crude and NGLs in 2006 was some 28.4 billion barrels, and if we factor in the IEA’s demand growth forecast of 1.7%/year between now and the peak, the midpoint arrives in January 2017. Even if we inflate all the figures by a reasonable factor to take account of that proportion of the world that the USGS did not assess, the midpoint is only delayed until 2021. That peak date coincides almost exactly with the view of Total chief executive Thierry Desmarest.
This approach may appear crude compared to modern ‘field-by-field’ modeling techniques, but it does have the distinct advantage of including an apparently well calibrated forecast of reserve growth, a factor which peak oil forecasters are sometimes – but often wrongly – accused of ignoring. My method may be rough and ready, but to take issue with its recalibrated ‘USGS ultimate’ of 2,830 billion barrels, you must either ignore the strong reasons for believing that discovery in the real world will greatly undershoot the USGS forecast, or assume that empirical reserve growth will suddenly accelerate and greatly outperform the USGS forecast.
In fact there is some reason to believe that reserve growth may also start to falter soon. Michael Rodgers, head of upstream economics at the respected international consultancy PFC Energy, notes that in deep water Gulf of Mexico the general experience over the last decade has been that companies’ initial reserve estimates have proved to be either accurate or overly optimistic. In other words, subsequent ‘reserve growth’ has either been nonexistent or negative – the fields have produced less than originally expected. As companies get better at reserve estimation, and as the average size of fields discovered continues to fall, Rodgers thinks this may become the norm: “I doubt we will see the reserve growth for fields discovered after 1990 that we had for those found earlier”.
This development may already be starting to affect the oil supply picture globally. According to figures from IHS, although discovery has fallen increasingly short of consumption since the early 1980s, until recently reserve growth has always more than filled the gap. In 2004 and 2005, however, oil discovery and reserve growth combined failed to match consumption for the first time.
So while I accept the USGS reserve growth forecast is on track for the time being, the risk is that this will also prove overoptimistic. This means that both the deflated discovery figure and the reserves growth estimate should be regarded upper limits, and therefore that a conventional liquids ultimate of 2,830 billion barrels is about tops.
The recalibrated USGS numbers suggest a peak by 2017-21, but their resource assessment only covers conventional liquids. Some apparently faith-based forecasters argue, perhaps with echoes of Vincent McKelvey, that unconventional fuels will put off the witching hour for decades to come. But this is to confuse the undoubtedly vast scale of unconventional resources for the speed with which they can be produced. As I show in The Last Oil Shock, of the varieties that are currently practical – Canadian and Venezuelan bitumen, gas-to-liquids, compressed natural gas and biofuels – none is likely to grow quickly enough to fill the hole left by crude. They may defer the overall liquids peak by a few years, but it is conventional oil, which still makes up the overwhelming bulk of the oil supply, that will largely determine the date.
When I interviewed Tom Ahlbrandt, project chief for the USGS World Petroleum Assessment, in early 2005, he professed amazement at the furore that his study had provoked among peak forecasters. “I don’t know why they’re wrestling with us”, he said, “We did the numbers straight up, it was scientifically based, and the numbers we have should cause concern in the first half of this century”. He was more right than perhaps he knew. The re-evaluated USGS numbers now provide powerful corroboration for those who forecast a global oil peak by around the end of the next decade.