This paper describes the development of an expert system for deciding whether to drill underbalanced or overbalanced. The expert system screens rock parameters that favor the design of an effective underbalanced drilling program. Once it is concluded that a particular reservoir presents a good candidate for underbalanced drilling technology (UBD), a screening procedure is then followed to select the optimal UBD technique. The following list of underbalanced techniques is considered in our expert system: liquid drilling (flow drilling), dry air drilling, nitrogen drilling, mist drilling, nitrified mud, aerated mud, and stiff foam drilling. The system screening procedure is put in an interactive advisor system that accounts for subsurface and surface safety, minimum rock formation damage, and wellbore stability. The expert uses a set of known facts and heuristics to measure the reservoir susceptibility for damage caused by clay swelling, fines migration, lost circulation and wellbore collapse.
A knowledge base development tool (Resolver) that consists of various utilities for developing and maintaining the rules, and implementing inference mechanisms was used in developing the expert system for selecting a candidate UBD technique. The principal attractive feature of ReSolver is that it supports fuzzy logic among other confidence modes. Membership functions were defined to assist the expert system in making decisions when the decision variables fall in a "gray area". Membership functions were constructed for variables such as lost circulation, clay swelling, fines migration, hard drilling potentials, cost benefit, gas influx and water influx potentials, fire potential, and stuck pipe potential. When the final outcome consists of a set of drilling fluids rather than a single one, these drilling fluids options were even screened further by the expert system to assure that the UBD fluid density is adequate within the pressure window. If the expert system still recommends more than a single drilling fluid option, a confidence level is given with each option.
Drilling underbalanced offers several benefits compared to conventional drilling techniques. These include, increased penetration rate and bit life, reduced probability of sticking the drillstring downhole, minimized lost circulation, improved formation evaluation, increased well productivity, and the reduction of stimulation treatments. The extent to which it is possible to achieve any of these benefits is generally controlled by the properties of the target reservoir and the overlying formations, and in some instances even by the specific characteristics of the well being drilled. The need for a fuzzy expert system that guides the user to optimally drill a well underbalanced becomes evident when we consider the delicate balance between the numerous coupled factors affecting the success of this operation.