Optimization/Inversion—a
Technique to Efficiently and Effectively Calibrate Petroleum
System
Models
Hermanrud, Christian1, Jay E. Leonard2, Thomas A. Schutter2, Margaret A. Lessenger2, Marianne S. Karplus2, Christopher N. Wold2, Veit J. Matt2 (1) Statoil ASA, Trondheim, Norway (2) Platte River Associates, Inc, Boulder, CO
Petroleum
system
model calibration has traditionally been
constrained to trial-and-error or linear search methods. With complicated
models, these approaches prove time consuming and may not identify the best
results. Optimization/inversion is an advanced technique widely used in science
and engineering to efficiently and effectively
calibrate a model to measured data. With the large numbers of wells and
associated geophysical log data available today, computerized optimization has
the potential to become a valuable tool for petroleum
system
model calibration.
Statoil and Platte River Associates, Inc. have
developed an optimizer that utilizes an adaptive simulated annealing (ASA)
inversion algorithm. ASA samples the parameter space efficiently and increases
the chance of finding the error surface’s global minima. The optimizer allows
simultaneous, fast, automated calibration of a large number of models to
measured data such as bottom hole temperatures, maturity (%Ro), pressure,
porosity, and permeability. The algorithm identifies the optimized values of user-specified
model parameters such that the model outputs most closely match the measured
data. The user can select a set of model parameters including lithologic, thermal, stratigraphic
(i.e., eroded section), and diagenesis parameters.
The optimizer provides an error response surface as well as the optimized model
parameter values. Furthermore, the optimizer sensitivity results enable the
user to determine if additional data are required and the relative importance
of various parameters to producing accurate models. The optimization/inversion
process streamlines calibration, improves modelling
results, and enables sophisticated determination of default parameters for
basin modelling studies.