Conceptual Design Exploration in Architecture Using Parametric Generative Computing: A Case Study

This paper documents design strategies using Grasshopper and Rhino 3D as an instructional tool for conceptual design. It discusses the underlying concepts of generative design and includes examples using Grasshopper with Rhino 3D for both massing and for basic structural layouts. It also discusses the necessary skill set, beyond that associated with the operation of the underlying CAD applications, required for students to utilize these applications. It then proposes a framework for incorporating generative design into CAD courses utilizing a 2-D to 3-D sequence of instructional activities. Part 1: Introduction The digital revolution and its associated discourse is increasingly influencing all of the design fields, particularly architecture . In his book Constructing Complexity, William Mitchell referenced to shift to digital design in architecture stating that “buildings were once materialized drawings, but now, increasingly, they are materialized digital information – design with the help of computer-aided design systems, fabricated by means of digitally controlled machinery, put together on-site with the assistance of digital layout and positioning devices, and generally inseparable from flows of information through global computer networks.” However, design exploration is an integral aspect of the design process in any discipline. Traditionally sketching has functioned as a primary conceptual design tool due to its indeterminacy and ambiguity. Goel [3] suggested that the ambiguity in sketching promoted cognitive shifts from one proposed conceptual idea to other alternative concepts, a process he referred to as lateral transformation. Won [4] proposed that during the drawing process designers demonstrate a “seeing behavior” in which they will concentrate on the figural properties of a sketch. He stated that as a result the designer may “see the image as something else” and added that the shift of ‘seeing’ to ‘seeing as’ stimulates imaging. Similarly, Suwa and Tversky [5] proposed that as designers inspect sketches “they see unanticipated relations and features that suggest ways to refine and revise ideas.” As design practices have been restructured around Computer Aided Design (CAD), processes to integrate digital technologies in conceptual design have been ongoing. The success of these applications has been limited. CAD has been perceived as a medium intended for production that is difficult to use in the early stages of the design process where the priority is creativity rather than precision. [6, 7, 8] The precision inherent in CAD results in a lack of ambiguity making commercially available computer aided design applications largely ineffective as a medium for design exploration. Most commercial CAD applications have provided some form sketch emulation mode [9] or have provided display options that generate digital representations with hand-drawn characteristics. The former were largely perceived as ineffective (Figure 1). The latter were typically perceived P ge 22368.2 as yet another alternative display which, while providing an effective representation tool, has not been widely adopted as a design interface (Figure 2). Figure 1. AutoCAD Sketch command example Figure 2. AutoCAD 3-D Wireframe display (above), with Sketch emulation (below) As an alternative, applications have been developed that attempt to operate in a manner that emulated traditional manually-based graphic techniques. These applications can include tools to support 2-D sketching mechanisms. Autodesk’s Architectural Studio interface was based on a trace-paper overlay mechanism in which designers could use drawing tools that created linework modeled after traditional markers and pencils. It could also merge sketches into 3-D models, thus bridging the gap between 2-D and 3-D graphics. However, its limited adoption has been attributed to the lack of wide-scale adoption of pen-based input devices . Many other sketch based 2-D to 3-D have been proposed or developed by researchers over the past decades. These include seminal applications such as Sutherlands SketchPad, a constraint-based drawing environment developed in the 1960’s, and STRAIT, a program developed in the 1970’s that interpreted sketch geometry as straight lines . Recent developments in the interface between sketching and digital design include “Digital Clay,” “SmartPaper,” and displacement modeling. [14] However, it may be argued that attempts to re-create the traditional sketch-based design processed with a computer may be an inappropriate strategy for conceptual design in a digital environment. Further, it may be argued that such a strategy is an extension of the early adoption of CAD, in which many of the commercially available applications were developed and used as electronic versions of manual drafting practices. Other than a change in the medium and devices used, no change in the actual processes associated with producing the work actually occurred. P ge 22368.3 This has resulted in the criticism that CAD has failed to meet the expectations of its users and that it’s true potential has gone unrealized. Therefore a more effective strategy may lie in reconceptualizing conceptual design by utilizing processes that embrace and exploit computation. Generative design approaches have emerged from the search for strategies to facilitate the exploration of alternative solutions in design, using computers as variance-producing engines to navigate large solution spaces and to achieve unexpected but viable solutions. [15] Kolaravec used the term “digital morphogenesis” to refer to design processes in which digital media is not used for representation but as a generative tool for the derivation of form and its transformation. [16] He stated that “the predictable relationships between design and representation are abandoned in favor of computationally-generated complexities” and that “models of design capable of consistent and continual dynamic transformation are replacing the static norms of conventional processes.’ For Kolaravec, generative computing, or, as he referred to it, “digital morphogenesis,” is a “radical departure from centuries-old traditions and norms in architectural design” – the emphasis shifts from form-making to form-finding.” In generative design, algorithmic procedures are often used to produce arrays of alternative solutions based on predefined goals and constraints, which the designer then evaluates to select the most appropriate or interesting. [17] This position is reiterated by Oxman , who stated that “the generative model is the design of, and interaction with, complex mechanisms that deal with the emergence of forms deriving from generative rules, relations and principles.” However, she also argued that designer interactivity is a key component. She stated that “Interaction has a major priority in this model” and added that “in order to employ generative techniques in design, there is a need for an interactive module that provides control and choices for the designer to guide the selection of desired solutions.” Chase [18] made a clear distinction between CAD and generative design. He argued that CAD applications do not have the exploratory potential of generative computing. He further argued that “traditional CAD software can aid students in understanding their designs, and develop their knowledge and skills in areas such as geometry, but they act only as aids; the user must directly input and manipulate forms”, and added that “the power of generative design tools is that these can guide a novice down an exploratory path.” The use of generative design technologies have been expanding architectural education and among design professionals as well. Generative design is a parametric computer modeling technology that is typically operated using an alternative interface for a Computer Aided Design application. Examples of these are Generative Components, which works with Microstation, and Grasshopper, which works with Rhino 3D. Part 2. Developing an introduction to Generative Computing using Grasshopper Generative design processes are characterized by the following: 1. A design schema that provides criteria requirements 2. A means of creating variations 3. A means of selecting desirable outcomes P ge 22368.4 Based on these characteristics, generative design environments provide significant advantages for conceptual design as the emphasis is on exploration of alternatives. However, one of the most significant advantages is that generative design environments are dynamic and interactive, providing real-time visual feedback as the geometric and dimensional variations are manipulated. A generative computing application that is rapidly expanding in use is Grasshopper, which runs with Rhino 3D. This expanded can be attributed to two factors. First, the extensive modeling capabilities of Rhino 3D, particularly in terms of nurbs (non-uniform rational b-spline) curve and surface modeling, has lead to its widespread adoption among architectural educators and professionals. The command structure has many parallels with applications AutoCAD as well as 3D Studio, thus reducing the learning curve for students already familiar with this application. Secondly, the graphical interface of Grasshopper provides an explicit representation of the geometric relationships and sequences used to generate the digital model. This explicit representation is linked to the Rhino 3D viewports. This enables designers to receive immediate visual feedback as these relationships are manipulated by user-defined mathematical and geometric parameters. Additionally, Grasshopper utilizes a simple strategy for storing variations of the results of these manipulations. A process, driven by clicking on a single icon, is used to store an option. When combined with layers, this process, referred to as “baking”, can store any number of variations on discrete layers, so the options can be saved for future d