Keywords: expansive soils, knowledge base, artificial intelligence, swelling, foundations, geotechnical, knowledge-based system.

In the past, computer aids for geotechnical engineering have been restricted to formal numerical analysis of well-defined problems or to subsidiary aids such as computer-aided drafting or text processing. More recently, some new types of computer aids have evolved, knowledge-based expert systems, which grew originally from research in artificial intelligence. These expert systems are computer programs, which originally were intended to mimic the performance of a human expert in a limited problem domain.

Many expert systems have been implemented to solve problems that are some form of diagnostic reasoning, ranging from medical diagnosis to troubleshooting generators, locomotives, and telephone networks[1,2]. Much of the knowledge in expert systems is heuristic in nature (consists of rules of thumb that are often, but not always, true).

Geotechnical engineering inevitably involves the use of engineering judgement and dealing rationally with considerable uncertainty. Because of the inherent heterogeneity of most soil deposits and complicated constitutive relations that apply to geologic materials, the primary task confronting the geotechnical engineer typically involves defining and formulating problems that are at once tractable and realistic. Knowledge-based systems can be useful in situations which involve expert judgements for effective problem solving, as is so often the case in geotechnical engineering.

Many researchers have applied knowledge-based systems to the field of geotechnical engineering[3,4]. Due to the heuristic nature of the domain of foundations in expansive soils, the uncertainties associated with the problem of foundations in expansive clays and the need for an expert opinion justify the application of knowledge-based systems. Prior to arriving at a final foundation design in an expansive clay site, the following steps must be taken by the geotechnical engineer:

1. Identification & classification of potential expansiveness of the clay deposit.
2. Quantification of potential or probable vertical movement.
3. Evaluation of design alternatives to determine the suitable foundation design.

An expert system shell, CONFOUND[5] is thought to suit the needs for the development of the knowledge-based system in the knowledge domain of foundations in expansive soils. Based on the above methodology, the process for selection of suitable foundation type in expansive soils has been divided into three main parts `Classification', `Heave Prediction' and 'Evaluation of Design Alternatives'. These three parts comprise the knowledge-based system.

This paper describes how the knowledge based system associated with foundations in expansive soils (FESAID) was developed using the CONFOUND shell. The system was developed to aid in the following processes:

1. Swell potential classification using either of three methods. Method described in Reference[6], BRE method[7] and Dakshanamurthy and Raman's method[8]. Each method has different suitabilities and confidence levels.
2. Heave prediction ranges for the clay deposit, as described in reference[6], considering its swell potential classification, moisture variation and the applied structural load.
3. Foundation selection with guidelines for the alternative treatment procedure such as moisture control or ground treatment procedure. On-line help can be reached during the consultation session to provide further explanations where needed.

1

B.G. Buchanan, G.L. Sutherland, and E.A. Feigenbaum, "Heuristic DENDRAL: A Program for Generating Explanatory Hypotheses in Organic Chemistry", In Machine Intelligence 4, Editors B. Meltzer & D. Michie, Edinburgh University Press, 209-254, 1969.

2

L.A. Rossman and J.T. Decker, "A Rule-based System for Evaluating Final Covers for Hazardous Waste Landfills", Expert Systems for Civil Engineers: Knowledge Representation, Editor R.H. Allen, ASCE, 1992.

3

J. Oliphant and D.I. Blockley, "Knowledge-Based System: Advisor on the Selection of Earth Retaining Structures", Computers & Structures, 40, 173-183, 1991. doi:10.1016/0045-7949(91)90470-7

4

A. Giolas and D.G. Toll, "A Knowledge-Based System for Correlations Using Object Hierarchies", Computers and Geotechnics, 25, 123-140, 1999. doi:10.1016/S0266-352X(99)00023-3

5

D.G. Toll and R.J. Barr, "A Decision Support System for a Geotechnical Application", Computers and Geotechnics, September 2000. doi:10.1016/S0266-352X(01)00014-3

6

M.E.M. Hamadto, "Expansive Soil Behaviour & the Development of a Knowledge-Based System Associated with Foundations in Expansive Soils", PhD Thesis, Heriot-Watt University, Edinburgh, June, 2000.

7

Building Research Establishment, "Low-rise Buildings on Shrinkable Clay Soils", Digests 240, 241 & 242, HMSO, UK, 1980.

8

V. Dakshanamurthy and V. Raman, "A Simple Method of Identifying an Expansive Soil", Japanese Society of Soil Mechanics & Foundation Engineering, 1, 1, 97-104, 1973.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £78 +P&P)