JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES / AUGUST 2001 / 109 EXPERT SYSTEM FOR ENVIRONMENTAL QUALITY EVALUATION By Rabee M. Reffat 1 and Edward L. Harkness 2 ABSTRACT: This paper presents the development of an expert system to evaluate the effects on environmental quality of proposed modifications to an office building following a postoccupancy evaluation. The model pre- sented in this paper has been designated Expert System for Environmental Quality Evaluation (ESEQE). The ESEQE model consists of 200 rules covering 65 performance criteria of environmental quality. These perfor- mance criteria cover lighting comfort, acoustic comfort, thermal comfort, and indoor air quality. The performance criteria and the method of evaluation were extracted from a series of interviews with experts in the field of environmental quality in the built environment. The structure and development cycles of the ESEQE model are described. A demonstration of using ESEQE to evaluate the environmental quality in office buildings is presented. The potential benefits of using ESEQE during the design stages are also discussed. INTRODUCTION Buildings should provide high quality environments to sup- port the activities of their occupants. This case study is an endeavour to develop a computer-based model with which to evaluate the environmental quality of buildings postoccupancy or during the design process. Designers could benefit from previous experience of ex- perts’ knowledge built into a knowledge base in an expert system model. A detailed and careful postoccupancy evalua- tion could facilitate the fine-tuning of a recently completed building and could help others learn from the experience (Lushington and Kusack 1990). The Expert System for Envi- ronmental Quality Evaluation (ESEQE) model can assist in documenting successes and failures in a building’s perfor- mance, and it may also assist in trouble-shooting during the commissioning/shakedown period immediately after occupa- tion, thereby identifying unforeseen problems in building use. This study asserts that there is a need for designers to assess environmental quality and that environmental quality evalua- tion (EQE) would enable designers to improve a design or completed building to meet the expectations of clients and users. Development of an expert system model for EQE could provide designers with a more efficient evaluation of office buildings. ENVIRONMENTAL QUALITY CATEGORIES Environmental quality in office buildings includes provision of lighting comfort, acoustic comfort, thermal comfort, and acceptable indoor air quality (IAQ) for the occupants of build- ings (McMullan 1983; Davis 1986; Manning 1987). The eval- uation of environmental quality in offices may be broken down into relevant performance criteria. Many criteria may be rel- evant to achieve the desired comfortable level. A series of interviews was conducted with experts to extract these performance criteria. Two kinds of experts were selected under certain benchmarks to be interviewed. These bench- marks include experience of at least 15 years in the field of specialization, substantial practical contribution to a related field, and comprehensive understanding of the interdisciplinary relationships with the related fields. A combination of theory 1 Lect., Dept. of Arch. and Des. Sci., Univ. of Sydney, Sydney, NSW, Australia. E-mail: rabee@arch.usyd.edu.au 2 Visiting Lect., Facu. of Built Environment, Univ. of New South Wales, NSW, Australia. E-mail: tedh@fbe.unsw.edu.au Note. Discussion open until January 1, 2002. Separate discussions should be submitted for the individual papers in this symposium. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was sub- mitted for review and possible publication on September 6, 2000; revised April 12, 2001. This paper is part of the Journal of Performance of Constructed Facilities, Vol. 15, No. 3, August, 2001. ASCE, ISSN 0887-3828/01/0003-0109–0114/$8.00 + $.50 per page. Paper No. 21975. and practice was considered while selecting potential experts. The experts interviewed totaled 50. There were 29 profession- als and 21 academics. A comprehensive overview of related literature was carried out in which initial ranges of comfort conditions for each performance criterion in the environmental quality categories were established. A structured questionnaire was then developed to collect the necessary knowledge and data. After the completion of the interviews, 65 performance cri- teria for evaluating environmental quality were extracted as shown in Reffat and Harkness (2001, Tables 1–4). A method of weighting and integrating the performance criteria obtained from the experts was developed (Reffat and Harkness 2000). This paper presents the model developed for this integration of environmental quality using expert systems techniques. OVERVIEW OF EXPERT SYSTEMS Expert systems are computer techniques that can be used to model the expertise of humans in a specific domain. Expertise is that knowledge that has been acquired through experience over a period of time in a specific domain and is heuristic in nature. Expert systems provide a technique to model the rea- soning processes of experts and use their knowledge to solve specific problems. Such systems can be used by nonexperts to improve problem-solving capabilities. Expert systems can also be used by experts as knowledgeable assistants. Expert sys- tems are used to propagate scarce knowledge resources for improved consistency of results. Such systems could function better than a single human expert in making value judgements in a specific area of expertise (Turban 1992). Expert systems are best known as self-contained entities that exist quite separately from other computer-aided design sys- tems. Important is the notion of embedding explicit knowledge of the kind that is encoded in expert systems within more general computer design tools (Krishnamoorthy 1996). The task of the expert system developed in this paper is to com- municate, access knowledge, make inferences, arrive at con- clusions, and explain its conclusions. STRUCTURE OF EXPERT SYSTEMS Expert systems are composed of two major parts: (1) the development environment; and (2) the consultation (run time) environment, as illustrated in Fig. 1. The developmental en- vironment is used to build the components and to introduce knowledge into the system. The consultation environment is used to obtain expert knowledge and advice. The development environment contains components that facilitate the creation of expert systems: knowledge acquisition facility, inference en- gine, and knowledge base. The knowledge acquisition facility provides a way to store experts’ qualitative knowledge to ad- dress a given problem. The inference engine controls the rea-