Summary: | We develop a linear programming inversion scheme to estimate lithology from well log
data. The inversion described severely limits the dependence on core retrieval, and
makes accurate mapping of new parameters possible. There are no limitations on the
number of input well logs or output lithologic fractions. Because the method uses an
L-I norm error measure, it should be more ·robust than otherwise similar techniques.
Other advantages of the technique include the straightforward imposition of inequality
constraints, and strictly positive volume fraction results. As specific examples we apply
the method in a variety of geologic situations, and compare our results to those achieved
via coring. We also test the new method against conventional least squares inversion
algorithms. While both methods perform well, the linear programming inversion yields
improved results when formation petrophysical properties are not known accurately.
|