Summary
Fluid-characterization data used to build representative reservoir fluid
models for reservoir and production engineering are generally scarce and very
costly because of the expense of fluid sampling and analysis and operational
related problems of well intervention. The problem turns more dramatic when the
field to be studied has been producing for some time and its bottomhole
pressure has reached the bubblepoint. In those cases, initial reservoir fluid
data is more difficult to be used because some portion of the hydrocarbon’s
heaviest or lightest components, whether oil or gas is produced, may have been
left in the reservoir and initial PVT data are no longer valid or very
difficult to match to current fluid behavior.
One of the most common practices of obtaining fluid data is by taking well
bottomhole samples, or by obtaining surface oil and gas mixture samples, and
analyzing them in the lab using recombined methods. Because of the costs
involved, including loss of time and production, operating companies do not
obtain fluid data with the required frequency. Accordingly, engineers are left
with the only option of using nonupdated or insufficient data that will affect
the outcome of their simulation studies.
This work describes a rapid and low-cost methodology to obtain an updated
compositional model for produced fluids using gas chromatography, oil
characterization, and GOR from analysis of the separator fluid discharge. The
required samples are easily collected with low cost to the operating
company.
Then, a case history is shown wherein compositional fluid models obtained
with the proposed method are used in well, surface-network, and
process-facility simulation models, with excellent results. A comparison of
field data against the simulation model results is also shown.
Introduction
One of the most frequent problems faced by production and reservoir
engineers when dealing with simulation studies is the availability of a
reliable characterization of the produced fluids (hydrocarbons, water, and so
on) because correct fluid modeling represents one of the most important factors
in the development of a successful reservoir, well, network, or
process-simulation study, in terms of value of the conclusions and
recommendations that could result from them. Frequently, the necessary data to
build a production simulation model are scarce or very difficult to obtain.
Commonly, fluid data information sources are initial PVT reports and some field
data which are, generally, non-representative of the currently produced fluids.
This situation becomes more complicated when fluids are produced from saturated
reservoirs, where the heaviest and/or the lightest fractions of the
hydrocarbons have been segregated and left in the formation (Rojas 2005) and
the produced fluids become more different from early production stage fluids.
In other words, reservoir conditions change as the fluids are produced, and
correct representation of fluid behavior in simulation models using initial
data becomes harder to match.
The usual alternative to obtaining the necessary fluid data consists of
taking field fluid samples (for example, bottomhole fluid samples) and their
corresponding laboratory analysis. This type of information, which should be
taken regularly because of its changing nature, involves the use of resources
that, normally, companies are not willing to spend. Consequently, most fluid
data are, in most cases, estimated using only the engineer’s field
experience.
This paper describes a practical methodology to obtain updated recombined
compositional fluid models of a well or group of wells producing form the same
reservoir based on an iterative process. This process uses a specialized flash
simulation software, and field data with low-cost acquisition. Data consists of
gas chromatography (as molar flow), oil characterization (molecular weight,
density, and viscosity), and GOR taken at separator discharge.
This method was used successfully in an integrated well-network-process
simulation study, allowing engineers to represent the fluid corresponding to 10
different fields involved in the analysis and, therefore, helping them to
obtain a good representation of the production system that was modeled,
detecting opportune areas of optimization, and making recommendations to
increase production.
© 2007. Society of Petroleum Engineers
View full textPDF
(
1,187 KB
)
History
- Original manuscript received:
30 July 2007
- Meeting paper published:
14 November 2007
- Manuscript approved:
17 August 2007
- Version of record:
20 December 2007
View Cart
