SANDING - AN EXPERT-SYSTEM APPROACH FOR ASSESSMENT AND CONTROL IN WELLS

The sand production problem is plaguing the petroleum industry by its adverse effects on thousands of oil and gas fields throughout the worl d. A tremendous amount of money is spent each year on attempts to pred ict and control sand influx and/or repair wells and equipment damaged by sand. Sand inflow into the well during production leads to casing a brasion and failure, formation damage and distortion, and frequent san d removal and cleaning. The sand control process has a major influence on the type of the well completion design and it influences the guide lines for the completion process. In addition, many wells are currentl y being produced below their potential in order to restrict sand influ x or erosion, and/or as a result of poorly designed or installed sand control methods. Evidently, the sand prediction and control problem is exceedingly complex and suggests the use of heuristics and the approp riateness of the expert systems technology. An automated sand control consultant and expert system has been developed, The system is aimed a t assisting users in predicting sand occurrence during production and in selecting and designing the proper sand-exclusion treatments. The k nowledge base of the system is based on an easy upgrade, easy expand f ormat and involves four primary modules, thus giving the end-user grea ter flexibility to tentatively access and evaluate different scenarios of knowledge processing. Input data can range from ''not known'' form ation characteristics and/or well stimulation requirements, for which the system gives conservative recommendations based on the remaining k nown facts, completion characteristics of the hole, well history, and geological probability; to cases with detailed information available, in which case very elaborate and precise recommendations are prescribe d. This paper describes the knowledge involved in various modules of t he sanding system, as well as future plans and further developments.