ABSTRACT

Most geologists still use the time-honored Spontaneous Potential (SP) baselining for sand-shale discrimination. While relatively fast and easy, it is subjective and nonreproducible. With modern computers, lithology identification from logs can be fast, accurate, and totally reproducible.

In the Gulf of Mexico, porosities under 20 percent are typically considered uneconomic. Neutron values over 50 pu and sonic delta travel times over 110 microseconds per foot, rare elsewhere, are common. These log responses may create problems when trying to identify lithology from logs. Dicriminant analysis and clustering have been discussed in recent literature as methods to identify lithology from logs. Both techniques have failed to be widely used because they are expert intensive and nonreproducible in their calculation. A classification procedure can generate reproducible results. Such a method is the basis for the Geocolumn (FOOTNOTE *) program. A classification procedure requires defining all the possible subdivisions of the group to be classified. In the case of lithology, 182 facies have been identified based upon log responses of eight different wireline logs and assembled into a computer data base. Typically values of density, neutron, sonic delta transit time, gamma ray, photoelectric factor, thorium, uranium, and potassium can be used.

Thousands of two-dimensional crossplots make up such a lithology database. Each crossplot contains the range of expected log responses for the two logs and one lithofacies being considered. The facies typically are defined as ellipses on the crossplots. The placement, size, orientation, shape, and degree of overlap with other facies must all be considered for each crossplot. Overlap is necessary to prevent undefined or unidentified zones, but overlap can introduce instability when data fall in the overlap.

Geological variations such as sediment source and burial history require different databases for different geologic basins. East Texas Cotton Valley sands have low porosity and values over 12 percent are rare. Lithologies common in one basin are rare in other: sideritic sandstone and feldspathic or pyritic shale are examples.

Because the difficult step of data base construction is completed before operation of the Geocolumn program, the program produces objective, repeatable calculations and lends itself to being run at the wellsite. In fact, it is now available at the wellsite. A wellsite lithology derived from logs allows a synergism with mudlogs that provides information never before available.