The paper presents an innovative approach to the development of a knowledge-based system aimed at support matrix stimulation design. The basic role played by preliminary context evaluation is enhanced. Context evaluation Involves assessing whether or not the stimulation must be carried out, through technical-economical feasibility analysis and risk assessment. A conceptual model and a system architecture have been specifically defined for this domain. allowing fast and expressive system prototyping. Extensive exemplifications of system reasoning performances are provided. performances are provided. Introduction The large internal demand for matrix stimulation jobs compared to the small number of expert available. the failures caused by lack of skill and inadequate operation control have led to the decision to develop a tool devoted to the broad support of matrix stimulation activity. The target objectives of such a tool are as follows: 1 - outline of detailed procedures for matrix stimulation planning, for selection of chemicals, equipment and service companies and for supplier and service quality control; 2 - data analysis and post-job evaluation; 3 - support for less-experienced users to avoid wasteful and uneconomical matrix stimulations and to design at expert level those which are necessary and cost effective. Knowledge-based systems (KBS) offer a solution to reach those goals. They are well-suited for domains of application where expertise plays a fundamental role In achieving optimal results. Therefore the huge amount of expertise involved in matrix stimulation design makes it the typical kind of domain where this computer technology could be successfully applied. Of the few KBS that have been previously developed in this field, all have highlighted the complexity of the domain. Cram et al. (1986) first Investigated the possibility of using KBS for matrix treatment design. They focused mainly on the selection of optimal chemicals, disregarding the process design. Alegre et al. (1988) dealt with formation damage diagnosis by defining classes of causes, mechanisms and related types of damage. Jia-li Ge et al. (1989) similarly approached damage diagnosis by focusing mainly on handling data uncertainty, using fuzzy logic. In this paper we present MAST (MAtrix Stimulation Treatment), a knowledge-based system developed for improved damage diagnosis and matrix treatment design MAST not only helps with correct damage diagnosis and supports the selection of chemical types and quantities, but it also evaluates economical opportunities and plans the appropriate operations sequence. We recognize that the availability of a domain expert is a crucial point in making a KBS fully reliable. Therefore MAST capabilities have been enhanced by the adoption of a design and evaluation method developed in-house: the key factors increasing the success ratio of matrix stimulations jobs have been identified extensive statistical analysis on more than 650 matrix stimulation jobs. carried out over the past 11 years in 9 countries, allowed us: to understand the reasons for past 11 years in 9 countries, allowed us: to understand the reasons for failures, to obtain a full scale lab (i.e. field) response, to compare lab-test Information with field Indications, and to identify and validate specific techniques aimed at optimizing matrix stimulation design and performance. performance. Currently MAST is the prototype stage and it covers only damage diagnosis. The validation phase is underway and. at the present time, the system has been successfully tested on a present time, the system has been successfully tested on a significant set of field cases. The system is implemented using a commercial knowledge engineering environment, based on an object-oriented paradigm and Common-Lisp programming language. The user interacts with the system using a mouse and programming language. The user interacts with the system using a mouse and a window-based interface. 2 PROBLEM SOLVING IN MATRIX STIMULATION The MAST architecture, shown in fig. 1, reflects the complex line of reasoning used by the expert for optimum matrix stimulation design The expert considers the following groups of Input data: 1 - well objective (i.e. production and/or recovery optimization, economical production rate, etc.); 2 - well data (i.e. bottom-hole flowing pressure. bottom hole temperature, net pay thickness, etc.); 3 - field and history data (i.e. reservoir lithology, mineral contents, static pressure, permeability. porosity, previous treatments. production history); 4 - laboratory results (i.e. acid sensitivity, wettability, formation packing, formation cementation, etc.).

No abstract