Content area
Abstract
The present study aims to improve the building envelope design process by developing a systematic approach for synthesis and evaluation of alternatives at the preliminary design stage. Knowledge-based system techniques are utilized to automate the information processing and decision making problems in design. It is possible to develop a knowledge base consisting of building Code requirements, performance standards, material properties, construction types and design heuristics. Such a knowledge base can provide the information necessary to establish the performance requirements of the design context and to generate design alternatives. A hierarchy of building envelope components and materials with their semantic relationships and functional attributes is developed using a schema (frame) based knowledge representation technique. Generation of feasible design alternatives is viewed as a constraint-based search problem in which the envelope components individually and collectively must satisfy the performance requirements of the design context. Problem decomposition and multiple levels of generate-test paradigms are used in the alternative generation process. Ranking and selection of alternatives can then be accomplished by specifying the preferences on construction types and the priorities on performance attributes.
The above methodology has been implemented in a software prototype known as Building Envelope Analysis and Design System (BEADS). The prototype knowledge base includes information from building code, a performance standard, design manuals, practising architects, material properties and cost data handbooks. A design query language (DQL) has been developed to facilitate rapid design revisions, re-evaluation of alternatives and to maintain the knowledge base. DQL provides designer freedom and allows the maintenance of knowledge base without having to familiarize with the development environment. BEADS currently addresses only a few important performance requirements but demonstrates the advantages of knowledge-based system approach. Two sample design problems have been solved to validate the knowledge base and to identify future extensions.
It is shown that knowledge-based system approach to building envelope design can significantly improve the abilities of the designer to consider a large number of alternatives and many performance attributes simultaneously at the preliminary design stage. The knowledge base can also serve as a vehicle for technology transfer in the building industry. The knowledge representation scheme and inference mechanisms developed for the BEADS prototype can be adapted to provide an efficient methodology to automate other formulation type design problems. (Abstract shortened by UMI.)