96
Figure 5. The exponential decline curve fits reasonably well with the production of the Norway giant Eldfisk, where
pressure depletion has caused major reservoir compaction and subsidence problems. The Nigerian Jones Creek,
which has been severely disturbed by wars, rebel attacks and sabotage, provides a good fit with the model if dis-
turbed data points are disregarded. In both these cases the production curve has been fluctuating around an approxi-
mately exponential decline curve.
Average decline rate
The Uppsala giant field database includes 331 giant oil fields with a combined estimated URR of over
1130 Gb, using estimates adopted by Robelius (2007). 214 fields are land-based (about 65% of the total),
while 117 are offshore installations (about 35%). To calculate the decline rate of giants that were in de-
cline as of the end of 2005, we considered only the 261 fields classified as post-plateau and in decline. Of
these, 170 were land-based and 91 offshore. IEA (2008) gives an average depletion factor, defined as
cumulative production divided by initial 2P reserves, of 48% for their super-giants and giants. Höök et al.
(2009) found that most giant fields leave the plateau phase and reach the onset of decline when around
40% of the URR has been produced, and combined with IEA's average depletion factor, it is not surpris-
ing that the majority of the fields are categorized as in decline.
Because the number of fields is so large, our approach provided reasonable statistics and reasonable
mean, median and production weighted values for the giant oil fields as a group. The production weighted
values were created by weighting the decline rate against the peak or plateau production level for each
field, thus giving greater importance to fields with high production. The production weighted decline is
lower than the mean value, because fields with high production levels often tend to be larger and decline
slower than the rest. More details can be found in Höök et al. (2009).
The statistical uncertainty is difficult to estimate, since production data contains political influences,
differences in definitions, reporting practice and many other parameters, making conventional statistical
error estimate hard to apply. A histogram showing the distribution of the decline rates for all the post
plateau fields considered in this analysis is shown in Figure 6.
A traditional statistical analysis based on the assumption that production data measures approximately
the same thing, results in standard deviations of around 5% and may be seen as a rough attempt to put a
number on the inaccuracy (Höök et al., 2009). In comparison, neither IEA (2008) nor CERA (2007) pro-
vides any uncertainty estimates and hence it is hard to judge the statistical variations in their results. This
study makes no attempt to provide detailed analysis of the uncertainty; rather, it only concludes that the
results are accompanied with significant uncertainty. Two significant digits will be used here, to make
comparisons with CERA (2007) and IEA (2008) easier, despite the fact that the results of Höök et al.
(2009) indicate that only one digit should be utilized because of the significant uncertainties in many of
the underlying reserves estimates and production figures.