Abstract
The prediction of the occurrence and the production conditions under which sand influx will occur into the wellbore, has always had great impact on well completion design. Several criteria have been used with limited success, most of which are based on complex mathematical algorithms and which were developed for well defined areas.
A model has been developed to assist in identifying sand production problems utilising drilling, logging, core analysis and well test data. The model emulates how human experts in well completion design solve such problems in a predictive mode. Information was gathered and coded in sets of production rules that represent the basic knowledge and reasoning used by the experts. Inferences from conventional data are used instead of complex mathematical algorithms, even though provision is made to include mathematical algorithms in order to compliment the reasoning process of the model. The computerised model is implemented using Artificial Intelligence Technology. The result is a prototype of a knowledge based system for sand production prediction. The process of developing such a system, its applications and further development is described in this paper.