Summary
This paper presents the results of a data-driven, field-modeling (DDFM)
evaluation applied to a high-temperature reservoir in Australia for the purpose
of determining the significance of chemistry, reservoir, well, and
hydraulic-fracture characteristics on well production. The DDFM approach has
identified key production drivers for a gas-well field in Australia. This
information has been useful in explaining hydraulic-fracture well production
and providing guidelines for future fracture stimulation success.
A DDFM process is used to develop a model for 32 wells completed in a
complex, 250 to 350°F gas reservoir in Australia. This type modeling technique
uses data from the field, including chemical formulation, geology, reservoir,
well, completion, hydraulic-fracture stimulation, and production results. The
data is integrated into a common format and resolution, and then a visual and
statistical evaluation is performed. Relevant correlations and useful trends
are noted. Next, an effort is made to develop a predictive model that can be
used to provide an overall explanation as to what parameters drive production
in the well field. Or, in effect, derive a high-level understanding about the
effect of the fracturing process on the reservoir. This is accomplished by the
use of data modeling/optimization technologies, including artificial neural
network (ANN) and genetic algorithms. The resulting ANN model can then be used
to evaluate the production associated with various hydraulic-fracturing
scenarios and/or characteristics. Validation of conclusions and/or resolution
of difficult interpretation issues are done by detailed evaluation and modeling
of key wells.
Hydraulic-fracture stimulation scenario evaluations performed by the ANN
model have yielded some expected as well as unexpected results. As expected,
reservoir characteristics, such as pay thickness, porosity, and water
saturation have a dominant effect on well production. What was unexpected is
the significance of well operations and stimulation fluid chemistry on well
production. The practice of killing the well after stimulation and using
inappropriate perforation techniques can reduce gas production by as much as
one-half, while the use of a high-temperature gel breaker in combination with a
reduction in base-gel polymer load can provide a 67% increase in
production.
Introduction
This paper analyzes reservoir, well, and completion information from the
Toolachee and Daralingie formations in the Moomba and Big Lake fields. The work
focuses primarily on the performance of fracture-stimulated wells.
The specific objectives were as follows:
- Attempt to predict post-fracture production from available preproduction
data.
- Identify key parameters that influence fracture effectiveness and
production performance.
- Determine which completion/stimulation methods and designs have been
successful in the past, and use the information to improve future designs.
- Use the three previously listed understandings to improve fracture
candidate selection.
In turn, a more general objective is to perform a pilot trial to determine
the applicability and benefits of applying ANN analysis as a tool for
understanding and improving well performance as part of a 1999 integrated
reservoir study.
For well-performance comparison purposes, a 12-month, normalized, cumulative
production volume (NCPV) for each well is determined. Because wells from all
stages of the field life are considered, this normalization removes the effect
of declining reservoir pressures or changes in surface pressures caused by
installation of compression, etc. The 12-month NCPV is determined for a
"datum" reservoir pressure of 3,600 psi and a flowing tubing head
pressure of 500 psi by multiplying the first 12-month actual cumulative gas
production volume by (3,6002 to 5002)/(reservoir
pressure2 – ftp2).
© 2009. Society of Petroleum Engineers
View full textPDF
(
1,181 KB
)
History
- Original manuscript received:
3 January 2007
- Meeting paper published:
28 February 2007
- Revised manuscript received:
14 May 2008
- Manuscript approved:
19 June 2008
- Published online:
2 March 2009
- Version of record:
26 February 2009
View Cart
