Shape Grammar Approaches in Architecture

Introduction

Shape Grammars are amongst the most well-known computational design methods used in architecture, although it is important to realise that the word “computational” does not necessarily refer to the use of a computer. A computational method is one that follows a rigorous formal logic, akin to a mathematical process, to produce an outcome. In the 1970s and 80s, when the first major developments in design computing occurred, the “computation” was typically manual or graphic. That is, written or drawn processes were used as the basis for computational logic, in large part because software and hardware to automate these processes were not available. In more recent years, with rapid advances in computing and visual programming languages, the manual or graphic processes used in Shape Grammars have been increasingly embedded in software. Thus, throughout the present chapter, references to computing are largely about a process, rather than the tools (hardware, software or methods) used to enable it.

Conceptually, a Shape Grammar is a “production system” that consists of two parts: a set of rules about shape transformations and a generation or behaviour engine that determines which rules are applied and when. The transformation rules typically include geometric operations like “rotating”, “enlarging” or “mirroring” a shape. The engine determines the orders in which these geometric operations occur, because a grammar is not random. The output of the production system is a shape that conforms to the rules and behaviours of the larger system, or language, it was derived from.

The origins of Shape Grammar research are traditionally traced to the seminal works of Stiny and Gips (1972) and Stiny and Mitchell (1978). The former of these used grammatical means to analyse the shapes used in painting and sculpture, while the latter applied a similar method to the architecture of Andrea Palladio, effectively creating the first architectural Shape Grammar. Since that time, art, design, architecture and engineering languages have been the most common subjects of Shape Grammars. Shape Grammar research in architecture can also be broadly categorised in terms of the three levels of architectural representation the grammars cover: schematic, two-dimensional and three-dimensional. The first of these, the schematic level, addresses the composition of spatial programs using basic modular shapes, much like a “bubble diagram” in the early stages of the architectural design process (J. H. Lee & Gu, 2018; Stiny & Mitchell, 1978). The second level, two-dimensional representation, deals with architectural dimensions and orientations as well as various types of design elements, like walls, doors or windows. The third level, three-dimensional representation, requires a more complex grammatical approach to the configuration of connections between forms (Cui & Tang, 2014; Flemming, 1990; Koning & Eizenberg, 1981). Because three-dimensional architectural forms require larger and more complex rule sets, two-dimensional Shape Grammars, typically floor plans, have been the most common type (Cagdas, 1996; Eloy & Duarte, 2011).

This chapter presents an overview of Shape Grammar research published between 2008 to 2018 and indexed in three online journal databases: Taylor & Francis online, SAGE Journals and ScienceDirect. These databases cover many of the major international journals in the architecture and design domains and could be regarded as capturing any emerging trends in these fields over the last decade. The first two databases allow us to identify articles using “Shape Grammar” or “Shape Grammars” as “author-specified keywords”, while the last permits a search only in “title, abstract or author-specified keywords”. Nonetheless, most of the collected articles from the last database list “Shape Grammar” or “Shape Grammars” as keywords.

This chapter reviews the general characteristics of the emerging research in the three databases and then provides an in-depth review of three main themes in the research: descriptive analysis, design generation, and generative applications. The first of these three also includes a background to the theme, which introduces many key concepts that are applied in the next chapter (Chapter 3).