Yuguang Chen & Louis J.
Durlofsky talk with ScienceWatch.com and answer a
few questions about this month's Fast Moving Front in the
field of Mathematics.
Article: A coupled local-global upscaling approach
for simulating flow in highly heterogeneous
formations Authors:
Chen, Y;Durlofsky,
LJ;Gerritsen, M;Wen, XH
Journal: ADV WATER RESOUR
26 (10): 1041-1060 OCT 2003
Addresses: Stanford Univ, Dept Petr Engn, Green Earth
Sci Bldg, Stanford, CA 94305 USA.
Stanford Univ, Dept Petr Engn, Stanford, CA 94305
USA.
Chevron Texaco Explorat & Prod Technol Co, San
Ramon, CA 94583 USA.
Why do you think your paper is highly
cited?
Upscaling is a numerical technique which is routinely applied in simulating
flow in subsurface formations. Upscaling is used to coarsen highly detailed
geological descriptions to scales that are suitable for flow simulation,
while maintaining the impact of important fine-scale flow features.
This paper introduced a new class of upscaling methods—local-global
upscaling, in which global coarse and local fine-scale flows are solved
together. Traditional upscaling methods fall into the category of local
methods, which are efficient but lack accuracy due to the need for assumed
local boundary conditions. On the other hand, existing global methods can
provide better accuracy. But this type of method requires solving global
fine-scale flow, which can be computationally intensive. The local-global
upscaling method combines the advantages of both local and global methods.
It can effectively capture the global effects, but avoids solving the
global fine-scale problem.
The impact of local boundary conditions is a well-known issue in upscaling
and in multiscale modeling in general. Local-global treatments provide a
new means to efficiently address this issue, which may explain why other
researchers have cited this paper.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
Coauthor
Louis J. Durlofsky
Yes, the paper describes a new methodology to compute coarse-grid
properties for flow simulation models. The results are more accurate than
those from existing methods with comparable efficiency. The procedure also
addresses the issue of global flow dependency of upscaled properties.
Would you summarize the significance of your paper
in layman's terms?
The simulation of fluid movement in subsurface formations, such as aquifers
and petroleum reservoirs, involves the computational solution of the
governing flow equations. The input model often contains millions of grid
blocks (discrete data volumes) which are used to represent the subsurface
geology more realistically. However, current flow simulators cannot handle
these models efficiently, meaning that the computational time for flow
simulation is prohibitively long.
Upscaling is applied to reduce the number of grid blocks used in the
simulation model, while at the same time preserving the flow response of
the original high-resolution model as closely as possible. In the petroleum
industry, for example, upscaling is routinely applied in oil reservoir
evaluations. The subsequent flow simulations are then used to make
decisions regarding oil and gas production.
Our paper presents a new upscaling methodology to generate the upscaled
properties. Flow results using this technique can be much closer to those
of the target high-resolution model than results using standard methods.
How did you become involved in this research and
were any particular problems encountered along the way?
In recent years, techniques for the more realistic description of
subsurface formations were developed. The resulting geological models can
involve complex connectivity of rock properties (permeability) and present
high contrasts in permeability values. For these cases, standard local
upscaling methods may give large errors. This motivated us to develop more
accurate procedures. The local-global upscaling method developed in this
paper is well suited for these highly heterogeneous reservoir models.
During the development of this method, we found that the use of upscaled
transmissibility (the numerical analog of permeability) provided superior
accuracy to the use of coarse-scale permeability. The switch from the use
of permeability to transmissibility provided improved accuracy and
robustness.
Where do you see your research leading in the
future?
This general methodology is finding applications in this and related
fields. Consistent with the ideas behind local-global upscaling, global
coarse solutions can be used together with local fine-scale calculations
for a variety of problems. Some examples include the treatment of more
general flows driven by source/sink terms (e.g., wells), upscaling of
multiphase flow parameters, modeling flow systems involving more complex
physics, capturing full-tensor effects in rock properties, and the
development of more accurate and robust multiscale simulation techniques.
Do you foresee any social or political implications
for your research?
Our research is focused on computational methodologies. We expect new
techniques that apply these ideas to be used in practical applications such
as petroleum reservoir simulation.
Yuguang Chen
Lead Research Scientist
Reservoir Simulation Research
Chevron Energy Technology Company
San Ramon, CA, USA
Louis J. Durlofsky
Professor and Chair of Energy Resources Engineering
Department of Energy Resources Engineering
Stanford University
Stanford, CA, USA Web