Summary
This paper applies compositional streamline simulation to model a real
field-scale project that is a combination of enhanced condensate recovery and
geologic storage of CO2. These processes are inherently
compositional, and detailed compositional fluid descriptions must be used to
represent the displacement behavior accurately. We demonstrate that
compositional streamline simulation, along with the use of analytical solutions
for condensate displacement, is computationally efficient enough to permit high
resolution of spatial heterogeneity as well as detailed characterization of the
fluid system.
We present a simulation study comparing streamline and finite-difference
results for 2D and 3D examples to demonstrate that the compositional streamline
method is an efficient computational method for modeling CO2 storage
and condensate vaporization in gas reservoirs. Although the streamline method
makes many simplifications regarding the effects of gravity and capillary
crossflow, in heterogeneity-dominated systems such as the condensate system
presented, comparison of finite-difference and streamline results confirms that
these simplifications are reasonable.
Introduction
Concerns about rising concentrations of CO2 in the atmosphere
have led to consideration of CO2 injection into depleted gas
reservoirs as a way to store CO2 that otherwise would be released
into the atmosphere (Oldenburg and Benson 2002). Gas-reservoir settings may be
attractive as potential CO2-storage sites because they are known to
have a seal that can trap buoyant gas. Moreover, depleted gas reservoirs may
contain condensate, a portion of which could be recovered by means of
CO2 injection (Jessen and Orr 2004a), offsetting a portion of the
costs involved in injection. In this paper, the interplay between
CO2 storage and enhanced condensate recovery is examined by use of
compositional streamline simulation.
Gas-cycling schemes for enhanced condensate recovery are inherently
compositional because condensate is moved primarily by transferring components
to the mobile vapor phase. Hence, evaluation of the performance of such
processes requires the use of compositional simulation. Recovery efficiency of
a gas-injection scheme is determined partly by the local displacement
efficiency and partly by fluid flow within the reservoir. Local displacement
efficiency is controlled by the phase behavior of mixtures of the injection gas
with the fluids present in the reservoir, which is, in turn, strongly
influenced by the fluid description used for equation-of-state calculations of
phase behavior. Fluid flow is often controlled by reservoir heterogeneities.
Breakthrough of injected CO2 at production wells will limit the
amount of economically recoverable condensate, influencing the portion of the
cost of CO2 storage that can be offset by gas and condensate
production. Therefore, accurate evaluation of the performance of a gas-cycling
scheme requires both high-resolution representation of heterogeneity in the
reservoir and use of an adequate number of components to describe the phase
behavior of the fluid.
Finite-difference compositional simulation is the conventional way to solve
such problems. This approach involves solving a material balance written for
each component, for each reservoir element (gridblock), in each timestep of the
simulation, requiring at least one flash calculation per gridblock per
timestep. For large models or complex fluid descriptions, this method can be
sufficiently expensive computationally that field-scale calculations are
impractically slow. To reduce computation time, current industry practice is to
simplify the geological model and fluid description. As a result, there is
clearly some loss of accuracy resulting from the less-detailed representation
of phase behavior and reservoir heterogeneities, as well as from the effects of
numerical errors caused by the use of large gridblocks.
An alternative to conventional finite-difference compositional simulation is
compositional streamline simulation (Thiele et al. 1997; Crane et al. 2000;
Jessen and Orr 2002). In this approach, the flow is represented as a series of
1D displacements along streamlines. For more details on streamline simulation,
see the review by King and Datta-Gupta (1998).
In reservoir displacements that are dominated by effects of heterogeneities,
streamline locations change slowly; hence, streamlines can be updated
infrequently. The resulting simulations run much faster than comparable
finite-difference simulations (Thiele et al. 1997; Jessen and Orr 2002)if an
efficient method is available for solving the 1D compositional flow
problem.
© 2007. Society of Petroleum Engineers
View full textPDF
(
3,288 KB
)
History
- Original manuscript received:
11 August 2005
- Meeting paper published:
2 February 2003
- Revised manuscript received:
18 October 2006
- Manuscript approved:
20 March 2007
- Version of record:
20 August 2007
View Cart
