A framework for history matching - StreamSim Technologies, Inc.

Questions & Answers
Benefits
“What benefits does Mepo give me compared to
my existing conventional history matching (HM)
workflow?”
“Usually I use one history matched
simulation model as base case for prediction runs.”
The Mepo workflow yields several matches of
acceptable quality. Thus, the uncertainty in your
simulation model is assessed before predictions are
made.
“Manual HM is usually time consuming, and I
often find it difficult to obtain a good match.”
With the application of scalable CPU clusters, Mepo will
reduce the amount of time used for HM, and the total
turnover time to update models is reduced significantly.
Mepo utilises state-of-the-art global optimisation
methods, with the capability to search the whole solution
space looking for ways to achieve a HM. When the
simulations are running, the engineer can analyse output
data, and adjust the workflow by modifying the
optimisation strategy or selecting new parameters
(discrete or continuous) into the HM.
“I judge the quality of the match by appearance, and
don’t use any quantitative measure.”
Mepo uses a customisable objective function to
measure the difference between simulated and
measured history data. Individual weighting schemes
for measurements and/or time periods are available.
Prior information (e.g. correlations between the HM
parameters like permeability and porosity) can be
added as penalty terms to the objective function.
“I modify only one parameter at a time, and choose
the ones which appear to be the most promising.”
A one parameter at a time approach ignores correlation
effects. The Mepo workflow includes global optimisation
and experimental design with the capability to modify
several parameters at a time.
“I have to manually edit and evaluate each
simulation.”
Mepo has an integrated pre- and post-processor, which
automatically checks the quality of the HM, and
generates new input files for simulation. You can interact
with this process at any time.
“How is Mepo different from other history
matching software?”
Optimising HM projects.
Mepo is a flexible framework for HM, an optimisation
environment with an integrated pre- and postprocessor.
The Mepo workflow currently includes
Bayesian analysis, experimental design, local and global
optimisation methods, and allows the inclusion of new
optimisation methods.
The graphical user interface is user friendly, and gives
full control and overview of the simulations. Mepo
provides a structured approach to HM, and produces
comprehensiveness, transparency and reliability of
results.
Global optimisation methods and parallel
processing.
A unique feature of Mepo is the use of global
optimisation and multiple CPUs. An Evolution
Strategy is applied to find several matches within
determined acceptable quality parameters. These
matches are then applied in prediction runs assessing
the model uncertainty. Distribution of these runs to a
number of parallel CPUs significantly reduces the simu-
lation time. Mepo proposes and generates new cycles
of runs based on the previous results. A cycle of runs
typically ranges from 2 – 20, but can consist of any
number of runs.
Both discrete and continuous parameters can be
changed.
A major benefit of Mepo is the ability to change both
discrete and continuous parameters. Typical parameters
that cause difficulties in HM studies are fault locations
and relative permeability curves (discrete), and
permeability and pore volumes (continuous).
Changing parameters and optimisation methods
during study.
An advanced algorithm management integrated in Mepo
allows to steer the optimisation process, e.g. changes in
the optimisation strategy (e.g. method), and to activate
or deactivate parameters as the HM study progresses.