Computational Design

Definition

Computational design refers to the use of algorithms, parametric modeling, and computational processes to generate architectural form, moving beyond digital drafting (CAD) into algorithmic form-finding. Software such as CATIA, Digital Project, Maya, and Grasshopper enable architects to model complex geometries, optimize performance, and automate production. For Hadid and Schumacher, computational design enabled parametricism's fluid, non-standard forms; for others, it raises questions about authorship, the role of the architect, and the relationship between code and creativity.

Carraher (via Stan Allen) periodizes the field in three phases of engagement beginning c. 1990 — the years of the World Wide Web's proposal (1989) and release (1992), the first digital cellular call (1990), and 2G: (1) digitizing the drafting process — "computer-aided tools were used widely, but there was not an overriding ethos as to how they could be utilized for anything other than digitizing the drafting process"; (2) formal speculation — early adopters exploring computational design (and parametric design especially, which "facilitates establishing relationships between elements, introducing a fluidity into the design process"), animation-derived "literal fluidity," the blob, and Gehry Technologies adopting Catia and Maya from the automotive, aeronautical, and shipbuilding industries; (3) the current strategic/operative phase — the designer "transforms agency by engaging with the design and construction processes at all stages," introducing "feedback into the formal development system that takes into account material, manufacturing, economic, performative, and social inputs." The phase model carries Allen's critique of phase two: software "present their architectures within a perspectival frame and an orthographic coordinate system," producing "exquisite objects… rather simple, cave-like interior space and undefined relationships to external conditions." Carraher's own definition of the medium: "Computational design is, at its most fundamental, defined by a series of relationships that can adapt to a range of forces" — designers "uniquely qualified to synthesize and respond to… competing forces by the very nature of the tools they use" (see spatial-agency, heterogeneous-space).

Anderson & Lovell-Anderson (2018) add the interior-scale, agent-based strand: "Agent-based visual scripting and computation is used as a medium for experimentation, decision-making, and problem solving to formulate dynamic surfaces and topographies." Where Carraher's phases track architecture's engagement with computation, their chapter relocates it: "a shift in systems of knowledge, thinking, and problem solving within interior architecture – a paradigmatic fluidity." The machinery is systemic rather than stylistic — an algorithm that "intervene[s] in a system and segment[s] a complex design query into several simplified sub-problems, and later reconstructed to create a solution" — and its form-generation is self-generating (autogenic-design) rather than continuously parametric: micro-level "nodes or agents" extracted from a data source produce macro-level patterns (emergence), simulated through "simple shape grammars in visual scripting" (shape grammars), and made material through "both additive and subtractive means." Computational design here is not a style (parametricism) but the discipline's new epistemology: "designs and their meanings [can] be viewed as the results of computations carried out according to rules of composition and correlative rules of description" (March & Stiny, via the chapter).

Key Thinkers

• Zaha Hadid — early adopter of Digital Project/CATIA to realize Suprematist-inspired forms.
• Patrik Schumacher — parametricism as the style enabled by computational design.
• Rem Koolhaas — critical perspective on computation's effects on architectural labor and authorship.

Related Concepts

• Digital Architecture — the broader field; computational design is its algorithmic subset.
• Parametricism — the stylistic movement enabled by computational processes.
• Abstraction — Hadid's abstractionist vision predated and was realized through computation.
• Digital Fabrication — the downstream half: computational models translated into machine cutting paths, templates, and fabrication sequences.
• Spatial Agency — Carraher's claim for the medium: relationships "that can adapt to a range of forces" as the computational form of agency.

Source Support

Sources in the wiki that discuss this concept:

• source-notes/seabrook-2009-the-abstractionist
• Fabricating interiority (Manack 2018) — the computational-to-material pipeline in interior architecture: "a sequence of design operations began by testing space plan and volume limits against the material dimensional constraints" (North Presbyterian Church); digital modeling that "slices a singular solid shape to create the splines and profiles, which ultimately become the cutting paths and templates used to extract the wood ribs from nominal panels" (Crystal Bridges Museum Store); "standard triangular tile shapes… determined computationally by subdividing a standard, rectilinear sheet of the ceiling cladding."
• Swimming upstream (Satterfield & Swackhamer 2018) — computation as performance translator and borrowed infrastructure: the VarVac Wall's "Grasshopper script that translates desired acoustical performance into a pattern of lines" (the script positions the variable mold's cables, so acoustic mapping, not drawing, generates the form); Gehry Technologies "borrowed parametric software and digital fabrication tools from other industries and applied them to building design and delivery," collaborating with aerospace and automotive engineers; and the ontological claim — "Software has grown into more than merely a tool of representation. It is increasingly paired with digital fabrication tools, a marriage of input and output that become generators of real physical form." See production-stream, mass-customization.
• Technologies: The spatial agency of digital praxis (Carraher 2018) — the phase model and the agency claim: Allen's three phases (drafting digitization → blobs and borrowed Catia/Maya → strategic/operative feedback with "material, manufacturing, economic, performative, and social inputs"); "Computational design is, at its most fundamental, defined by a series of relationships that can adapt to a range of forces"; parametric design's unfulfilled social promise — designers must either "parameterize" qualitative aspects of space or the tools must "become more accessible to the populations in need of design without the mediator of a designer."
• Transforming Interior Volumes (Anderson & Lovell-Anderson 2018) — the interior-scale, agent-based strand: "Agent-based visual scripting and computation… a medium for experimentation, decision-making, and problem solving to formulate dynamic surfaces and topographies"; the algorithm "to intervene in a system and segment a complex design query into several simplified sub-problems"; "Systemic (and rule-based) computation and visual-spatial scripting… where a series of nodes or agents at the micro-level are extracted from a data source"; and the March & Stiny designs-as-computations thesis. See autogenic-design, emergence, shape grammars, surface-mathematics.
• On technological limits (Olsen 2018) — the limit-interrogation strand and its history: the 1990s propagation of "computational design and production methods" as the profession's second great transformation; the technologies of production defined as "the software and hardware used in the design and construction processes"; the novelty-then-expertise curve ("As technologies become embedded within the culture of the discipline, innovations require a deeper level of technological expertise"); the structural branch — finite element modeling "parallels CAM software that emerged from the development of supercomputers in the 1950s… initially harvested by the aerospace and military industries" — and its built use in digital form-finding (Voussoir Cloud's hanging chain; Wiscombe's ANSYS-tested Cantilever; Roberge's gravity-loaded physics simulations). The strand's critical edge: "Computers are excellent tools for optimization and efficiency, and they are also superb at generating excess" — parametric modeling's "endless number of results, begging the question, is more really more?" See technological-limits, form-finding, parametricism.
• Notes on Digital Nesting (Goulthorpe 2002/2006) — the strand's founding document and its earliest critique: John Frazer's An Evolutionary Architecture (1995) as the template for everything above — "one should code not the form but rather precise instructions for the formative process" (quoted p. 137), genetic code "used to breed further populations in a cyclical manner," the computer as "an electronic muse" compressing "evolutionary space and time" (quoted p. 137) — and Goulthorpe's Bachelardian audit of it: the process is "nakedly scientific-rational in its prescriptive manner," suggests "natural selection" but "never offers sufficient parameters or process of selection" (pp. 134–135), and its outputs lack "any… cultural qualification other than their technical feasibility" (p. 134). The source dates computational design's process-over-form doctrine to 1995 and asks what it still owes Bachelard: is the "thinking of the digital itself" (p. 133) yet accounted for? See evolutionary-architecture, emergence.
• source-notes/vanucci-2018-inside-out-outside-in
• source-notes/voordouw-2018-topology-and-interiority

Open Questions

• Does computational design democratize form-making or concentrate power in software and expertise?
• What is the relationship between computational design and the deskilling of architectural labor?
• Is AI-assisted design a continuation of or break from computational design?