Department of Geology and Geological Engineering, Colorado School
of Mines, Golden, CO
Abstract: Forward and Inverse Stratigraphic Modeling of Fluvial
Systems: A Bridge to the Future of Reservoir
Management
Numerical stratigraphic models are capable of accurately
simulating stratigraphic geometries and facies attributes of
petroleum reservoirs. Correctly constructed stratigraphic models
may be converted into a petrophysical model for reservoir
simulation. Accurate
simulations
which match the stratigraphy of a
reservoir
are generated by models which have: (1) geologically
reasonable algorithms which adequately simulate the stratigraphic
process-response system; (2) a set of correctly acquired
observations; and (3) accurately determined values of model process
parameters. The first condition is accomplished by translating
correct knowledge about the process-response system into computer
code. The second is accomplished through careful measurement of
properties which are comparable to attributes of the simulation.
The third may be accomplished by trial-and-error
simulations
combined with subjective judgment of comparability, but this
approach produces nonunique and nonrobust answers. A better
approach is through simultaneous, multi-parameter stratigraphic
inversion which acquires estimates of process parameter values
directly from the observations.
A new forward model of fluvial stratigraphy was developed using
this philosophy of stratigraphic modeling. The Multi-Scale Fluvial
Model (MFM) consists of nested, deterministic models that simulate
sediment accumulation in fluvial/floodplain environments as a
function of processes operating at variable spatial and temporal
scales. The forward model algorithm reproduces the self-organized
criticality characteristics observed in natural fluvial systems by
incorporating nonlinear dynamics, feedback, buffers, thresholds,
and memory. Sedimentary facies of variable grain-size distributions
accumulate by geomorphic processes, but are variably preserved as a
function of stratigraphic base level. Model output is comparable to
easily collected observations of multiple sedimentologic (e.g.,
grain size, environment) and stratigraphic (e.g., thickness,
geometry, facies distributions) attributes. Model simulations
of
sedimentary facies are convertible to a 3-D petrophysical model to
be used in
reservoir
fluid-flow
simulations
.
In the next phase, model inversion, simulations
should correctly
describe 3-D
reservoir
properties beyond well bores because the
simulation will have matched observations at the well bores.
Inversion produces the best estimates of the forward model
parameters, an estimate of the uncertainty in these parameters, an
estimate of the confidence interval of the best fit forward model,
and suggests sources of errors and shortcoming in the
observations.
AAPG Search and Discovery Article #90928©1999 AAPG Annual Convention, San Antonio, Texas