Summary
The portfolio of gas sources to supply a liquefied-natural-gas (LNG) project
may involve many diverse fields, each with its range of uncertainty and degree
of maturity. For project approval, it is necessary to aggregate the
reserves/resources of all these fields into project-level representative
numbers, either deterministic or probabilistic.
Arithmetic addition of all low estimates (1P or P90) and all high estimates
(3P or P10) is known to overstate the range of uncertainty. On the other hand,
independent probabilistic addition tends to produce unrealistically narrow
ranges. The correct answer would be obtained by using correlated addition, but
this requires the estimation of all correlations between field-resource
estimates.
This paper presents a simplified and pragmatic approach, partial
probabilistic addition. A hierarchy of “resource containers” is defined from
individual reservoirs to total project level, and resources are aggregated from
bottom upward using either arithmetic or probabilistic addition—whichever is
more appropriate. This amounts to setting weak correlations to 0 and strong
correlations to 1. Expert opinion is solicited only concerning the strength of
dependencies, rather than being asked to specify elusive correlation
coefficients.
The validity of the approach is investigated by applying partial
probabilistic addition to a synthetic portfolio of fields with the same
log-normal size distribution. The study indicates the feasibility of estimating
a global P90 with a controlled error, whereas for a P10, the error can become
quite large. The method is then discussed using a real LNG case study.
Partial probabilistic addition is a practical method. It is easy to
understand and to explain. It can be summarized in a single table where a color
code indicates how its value was calculated for each cell. These features
facilitate technical interchange and quality control, which are key to reducing
as much as possible the degree of arbitrariness inherent in the modeling of
dependencies.
Introduction
The portfolio of gas sources to supply gas to an LNG project may be very
diverse. It may include associated gas from producing oil fields and
nonassociated gas from fields at different stages of development. Some of these
gas fields are producing, others are delineated but undeveloped, others are
discovered but not delineated, and some are still explora¬tion targets. For
nonoperated fields, there is little or no control on how resource figures are
calculated. To support the final investment decision, the resources of all
these fields need to be consolidated into project-level representative numbers.
These may be deterministic estimates 1P/2P/3P of proved , proved + prob¬able,
and proved + probable + possible volume, respectively, or probabilistic low,
best, high estimates P90/P50/P10. For example, the three numbers 67/100/150 Bcf
mean that there is a 90% chance of having 67 Bcf or more, a 50% chance of
having 100 bcf or more, and a 10% chance of having 150 Bcf or more. The low
value 1P or P90 is of special interest for project approval because it is
compared with a predefined resources threshold. [According to SPE terminology
the word “resources” is used rather than “reserves” when commerciality is not
proved (SPE 2001).]
Producing an estimate of minimum resources that can be used reliably as a
decision tool is no easy task. It involves two aspects: (a) Which numbers to
add? and (b) how to add them? The first question refers to the longtime debate
between deterministic and probabilistic estimates. Deterministic 1Ps used for
booking reserves are by design very conservative. They certify a volume
considered as proved today. However, this figure does not reflect the
project 1P, the volume that can reasonably be expected in the future
after planned fields have been brought on stream and delineation work has been
completed. Such an estimate is better derived by probabilistic means.
Concerning the aggregation of estimates, in the deterministic approach there
is little choice other than summing all 1Ps, all 2Ps, and all 3Ps. For
probabilistic estimates, arithmetic summation of all low values (P90) and all
high values (P10) is known to overstate the range of uncertainty. On the other
hand, independent probabilistic addition tends to produce suspiciously narrow
ranges. The correct procedure is a probabilistic addition taking into account
the dependencies between field-resource estimates. Assessment of these
dependencies is the key issue. Methodologies have been proposed to estimate
correlations between fields in a semiqualitative manner through an analysis of
commonalities between fields (Carter and Morales 1998; Van Elk et al. 2000).
However, the practicality of these approaches is in question when dealing with
a large number of fields and reservoirs. Interesting guidelines for the
aggregation of estimates for public disclosure have also been drafted (SPE
2001; COGEH 2004), but the present paper does not address reporting issues.
A simplified and pragmatic approach, partial probabilistic addition ,
is proposed. Groups are formed containing fields that tend to be either
dependent or independent, and their resources are summed accordingly by
arithmetic or probabilistic addition. This amounts to setting weak correlations
p to 0 and strong correlations to 1. The validity of this approach is
first investigated using a synthetic example. Then the method is applied to
determination of the global P90 of an LNG project.
© 2008. Society of Petroleum Engineers
View full textPDF
(
1,780 KB
)
History
- Original manuscript received:
4 June 2004
- Meeting paper published:
26 September 2004
- Revised manuscript received:
9 May 2007
- Manuscript approved:
17 August 2007
- Version of record:
25 April 2008
View Cart
