![]() |
|||
![]()
|
![]() |
![]() |
![]() |
2.8. CONTOURING ASSISTANTSines et al. Contouring Assistant began as a research project at Landmark/Zycor in 1990 (Sines 96). Despite hardships and trial and error, the system was successfully fielded as a product in December 1994. Geoscientists build contour maps of geologic surfaces. These are often referred to as "horizons." Similar maps can be produced for thicknesses and other rock attributes to help scientists understand hydrocrabon deposition and reservoir environments. In the past, much of this work has been performed manually. Geologists have hand-drawn contours after posting values from seismic and well data on a map. According to Sines et al., this operation requires both dense and accurate data in order to provide useful results. The manual operation is very time consuming and places great burdens on the person performing the work. Because of this, computer-based tools have been launched in order to alleviate the burden. Increased speed, ease of update, repeatability, and objectivity can be achieved. However, a computer approach is susceptible to both data volume and data quality. In contrast to the computer, a person is able to apply his expertise and intuition to compensate for lack of data. When ZYCOR launched its surface modeling toolkit called Z-MapPlus, they became aware of the fact that its use was complex. In some aspects, it in fact increased manual labor. This gave rise to the development of a support tool, an intelligent front-end that later became known as the Contouring Assistant. Once built and interacting with the modeling software, Contouring Assistant builds a map that would take hundreds of button clicks and a lot of insight using a more conventional modeling package. The Contouring Assistant knowledge base embodies years of gridding and mapping experience. As an expert system, it incorporates many of the standard features. The project was initiated along the same lines as a standard expert system effort at the time. What makes Contouring Assistant different from the other systems presented thus far is the fact that it successfully applied an object-oriented technique and blended it with the use of standard production rules. The documentation by Sines et al. provides a sincere description of the process from the proof of concept to the commercial system, including descriptions of both setbacks and progress. Classes and objects were used for organizing static knowledge. Class hierarchies represent the Contouring Assistant's understanding of the problem specification. This includes data and default files, area of interest, data and fault spatial characteristics, smoothing, and mapping preferences for the contours. Hierarchies were also built to accommodate objects containing solutions, i.e., gridding algorithms, grid and data operations, and key control parameters. Rules are accommodated within this framework. They create recommended surface modeling procedures based on the input specification. These procedures are constructed as dynamic object hierarchies. They are translated into the syntax compatible with the associated gridding and contouring software. The inference strategy is basically of the backward type. An important retrospective observation that Sines et al. made in their final publication (Sines 96) is the fact that expert systems development suffers from the same pains and challenges as ordinary software development. It appears that the Contouring Assistant project suffered more from these types of problems rather than particulars of the expert system concept itself. The need for good project management with clear-cut goals is especially emphasized. It shines through from this experience that expertise-oriented software projects like the Contouring Assistant are more susceptible to management weaknesses than ordinary software work. Contoruing Assistant was developed with Nexpert Object and runs on a SUN platform. By 1996, ZYCOR had sold more than 125 copies of the system.
|
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |