A knowledge-based system, named LOGIX, has been developed for the evaluation of petrophysical data. The objectives of LOGIX are to provide less-experienced petrophysicists with easy access to petrophysical expert knowledge and to provide petrophysical experts with a means of considering all data simultaneously in a consistent manner when developing new petrophysical models; an important feature of the system is the ability provided to the petrophysical engineer to incorporate his professional interpretational skills in the knowledge base. In LOGIX, typical petrophysical heuristic expertise ("rules-of-thumb") and reasoning processes are used to identify reservoir zones, to determine lithology and type of pore fluids, and to calculate porosity and hydrocarbon saturation. Moreover, the system highlights alternative solutions and inconsistencies in the data set. The set of petrophysical rules, present in a knowledge base, can easily be modified by an experienced petrophysicist. LOGIX'S reasoning mechanism consists of a hypothesis-and-test procedure in which some 1600 petrophysical rules are combined to test the likelihood of a large number of hypothesised rock and porefill models. The likelihood is indicated by a confidence factor. The most-likely model is given by the rock and porefill model with the highest confidence factor. An important feature of the system is its capability to explain the reasoning process. LOGIX has been tested on well data from siliciclastic and carbonate reservoirs. Examples of knowledge-based interpretations are presented in this paper. LOGIX 'S interpretation concurred well with core measurements, geological descriptions of cuttings and side-wall samples, and well-test data. LOGIX has been installed in some of Shell's major operating companies, where it is currently undergoing extensive testing under operational conditions, with a view to installing it worldwide as a tool complementary to existing deterministic and statistical petrophysical interpretation systems.

This content is only available via PDF.
You can access this article if you purchase or spend a download.