Gianfranco Carrara

Knowledge-based computational support for architectural design

Abstract The process of architectural design aims to define a physical form that will achieve certain functional and behavioral objectives in a particular context. It comprises three distinct,but highly interrelated, operations: (1) definition of the desired objectives; (2) production of alternative design solutions; (3) evaluation of the expected performances of the solutions and their comparison to the predefined objectives. Design can be viewed as a process of research for a solution that satisfies stated needs, while at the same time adapting the need to the opportunities and limitations inherent in the emerging solution. Computational techniques were developed to assist each one of the three operations, with varying degrees of success. We propose to intergrate all three operaitons into one whole, by developing a computational model that will facilitate smooth transition from one operation to another. The role in supporting this model will include providing a database of prototypical design objectives and solutions, storing project-specific design goals and solutions, and predicting their expected performances. This paper discusses the rationale and background for developing such a knowledge-base design system, and presents the parameters for implementing it as a computational tool to support architectural design. Examples from a prototype implementation serve to illustrate the discussion.

Computer Supported Design Studio

Abstract The paper presents the ongoing experimentation of a Computer Supported Design Studio (CSDS). CSDS is part of our continuing effort to integrate computers and networks in the design studio. We recognise three corner stones to CSDS: memory, process and collaboration. They offer a framework for the interpretation of the pedagogical aspects of the teaching of architectural design in relation to the innovations produced by information and communication technologies. The theme of the 1998 CSDS is a railway station in Turin, Italy, to be incorporated in a reorganised rail transport system. The choice of this theme emphasises the realistic simulation aspects of the studio, where technical problems need to be interpreted from an architectural point of view.

Multi-modal representation of design knowledge

Abstract Explicit representation of design knowledge is needed if scientific methods are to be applied in design research, and if computers are to be used in the aid of design education and practice. The representation of knowledge in general, and design knowledge in particular, have been the subject matter of computer science, design methods, and computer-aided design research for quite some time. Several models of design knowledge representation have been developed over the last 30 years, addressing specific aspects of the problem. This paper describes a different approach to design knowledge representation that recognizes the multimodal nature of design knowledge. It uses a variety of computational tools to encode different kinds of design knowledge, including the descriptive (objects), the prescriptive (goals) and the operational (methods) kinds. The representation is intended to form a parsimonious, communicable and presentable knowledge-base that can be used as a tool for design research and education as well as for CAAD.

Constraint-bounded design search

Publisher Summary The design process requires continual checking of the consistency of design choices against given sets of goals that have been fulfilled. Such a check is generally performed by comparing abstract representations of design goals with these of the sought real building objects (RBO) resulting from complex intellectual activities closely related to the designers culture and to the environment in which he operates. This chapter presents a possible formalization of such representations concerning the goals and the RBO that are usually considered in the architectural design process by the culture in the environment. The representation of design goals is performed by expressing their objective aspects and by defining their allowable values. The resulting system of requirements defines the set of allowable solutions and infers an abstract representation of the sought building objects that consists of the set of characteristics which are considered relevant to represent the particular kind of RBO in regard to the consistency check with design goals.

A Performance-Based Paradigm of Design

This paper proposes an alternative approach to existing design paradigms (and their CAD implementations) that are based on the traditional, causality-based notion that ‘Form follows Function.’ The proposed approach, which will be referred to as performance-based design, is founded on the argument that the relationship between form and function is contextual rather than causal. Hence, the expected performance of a given design proposal can only be determined by an interpretive evaluation, which considers the form (and other physical attributes) of the proposed design, the functional objectives (goals) that it attempts to achieve, and the circumstances under which the two come together (the context). The paper develops a performance-based design methodology and demonstrates its application in an experimental, knowledge-based CAD system.

Knowledge assistants in the process of architectural design

Abstract The following pages contain a description of the methodological system common to researchers in Computer Aided Design (CAD) in the process of development at the CAD Laboratory in the Faculty of Engineering in Romes “La Sapienza” University. In this context a software system for building and architectural design is being created; the system is based on the concepts of constraints propagation and knowledge-based systems aimed at cooperating with the designer in carrying out his/her work. The objective is a system capable of showing all the implications and aspects of project choices made in relation to the objects and limitations considered. After outlining the methodological approach used, we shall illustrate the structure of this system.