Summary
Well-placement decisions made during the early stages of exploration and
development activities have the capability to improve later placement decisions
by providing more information (greater certainty). Therefore, recovery and
efficient use of information may add value beyond the amount of oil recovered.
This study proposes an approach that emphasizes the value of time-dependent
information to achieve better decisions in terms of reduced uncertainty and
increased probable net present value (NPV). Unlike previous approaches,
well-placement optimization is coupled with recursive probabilistic
history-matching steps through the use of the pseudohistory concept. The
pseudohistory is defined as the probable (future) response of the reservoir
that is generated by a probabilistic forecasting model. To test the results of
the proposed approach, an example reservoir was investigated with multiple
realizations, all of which match the same production history. The results of
this study showed that subsequent well-placement decisions can be improved when
probabilistic production profiles obtained from the wells, as they are drilled,
are incorporated in the optimization scheme..
Introduction
Well placement is one of the important decisions made during the exploration
and development phase of projects. Most of the time, the large number of
possibilities, constraints on computational resources, and the size of the
simulation models limit the number of possible scenarios that may be
considered. In these cases, optimization algorithms become extremely valuable
in searching for the optimum development scenario.
Various approaches have been proposed for production optimization.
Bittencourt (1994) optimized the scheduling of a field using the polytope
algorithm. Beckner and Song (1995) applied the traveling salesman framework on
a well-placement problem, using simulated annealing (SA) to find the optimum
locations of the wells. Bittencourt and Horne (1997) hybridized genetic
algorithms (GA) with the polytope algorithm and tabu search and referred to
this hybrid optimization technique as HGA. HGA was observed to improve the
economic forecasts and CPU effort during optimization. Pan and Horne (1998)
used kriging as a proxy to the reservoir simulator to decrease the number of
simulations. Guyaguler et al. (2000) showed that the number of simulations
required to optimize the injector well locations decrease when an HGA was
coupled with a kriging proxy. Yeten et al. (2002) coupled GA with hill-climbing
methods and an artificial neural network (ANN) proxy to optimize the type,
location, and trajectory of nonconventional wells. Guyaguler and Horne (2001)
assessed the uncertainty of the well-placement results using utility theory
together with multiple realizations of the reservoir.
All these approaches considered only the information that was available at
the beginning of the optimization process. Data that would become available as
the reservoir developed in time was not taken into account.
© 2006. Society of Petroleum Engineers
View full textPDF
(
3,746 KB
)
History
- Original manuscript received:
7 June 2004
- Revised manuscript received:
8 June 2005
- Manuscript approved:
16 January 2006
- Version of record:
20 April 2006
View Cart
