Accurately diagnosing and treating a water control problem can often be costly and difficult. Although numerous water control products and treatments are available, only a small percentage achieve the desired results because the problem was misdiagnosed, the incorrect treatment design was chosen, or operational problems occurred.

A knowledge-based expert system is now available that can help identify possible problems and provide a complete treatment design. Since the knowledge required in this case may involve some uncertainty, a set of numerical probabilities must be used to represent this information. In addition to the rule-based reasoning1  commonly used by many expert systems, this system includes a unique matrix technique that can use available reservoir, production, and injection information to diagnose problems based on probability.

After identifying all potential problems, the system can recommend the most effective treatment for each type of problem, provide an optimum treatment schedule, and introduce the best placement technique based on available data. This type of expertise can be best represented by production rules. The system also features an extensive explanation facility to help the user understand why each recommendation was made.

This PC-based system can operate in a Microsoft® Windows™ environment. Unlike previously designed systems2 requiring an AI shell, C language was used to develop this system.

This paper will address the various design considerations involved in developing this knowledge-based system. Specifically, it will discuss the following topics:

  • the knowledge acquisition process

  • knowledge representation and system implementation, including

    • problem identification/reasoning (using the matrix technique)

    • fluid selection/placement techniques (using the rule-based technique)

  • the operation of the user interface

You can access this article if you purchase or spend a download.