Design and build
Over the past decade, an exciting and powerful computer-aided design and manufacturing (CAD/CAM) tool has emerged in industries other than architecture and construction. Most new consumer and industrial products, notably in vehicle design, are now digitally designed using solid and surface modelling software and then manufactured using data from those models.Virtual representations become tangible objects by using computer numerically controlled (CNC) machines to automatically form materials into models, prototypes, production tools, or actual components. Maybe it is time that architects took advantage of the opportunities on offer.
These intriguing new digital material-forming technologies are generally classified as either additive or subtractive manufacturing processes, depending on whether they use CNC software to direct the addition (building up layers of a model), or the subtraction (effectively carving away) of material to create a finished model.
Additive material-forming technologies are commercially referred to as rapid prototyping (RP) processes, but they are technically known as solid free-form fabrication processes.
RPprocesses have the capacity to produce complex and freeform components that would otherwise be impossible to create. At present, they are generally used during design to make relatively small representational models and prototype parts, but they are also increasingly used to manufacture functional metal and plastic components.
Three-dimensional subtractive formation utilizes milling machines or routers to carve components or models out of solid blocks of material. These machines are available in a vast array of sizes, from small, inexpensive desktop CNC milling machines useful for model making, to enormous gantry-based machines used in shipbuilding and aircraft manufacture.
It is generally true for all architectural applications that additive and subtractive material forming processes are most useful and cost-effective when applied to complex, intricate or freeform designs, or when they can be used to eliminate fabrication by enabling the substitution of one component for what would otherwise be several different parts.
Rapid prototyping processes incrementally deposit, cure or bind very thin layers of material within the 'build chamber' of a machine. At present, most build chambers are relatively small, but, in industries such as automotive design, larger components are routinely subdivided, built in sections and then bonded together.
In order to use rapid prototyping machines, the solid or surface modelling software used by a designer must have the capacity to export data in the STL format. This file format is named after stereolithography, (SL), which was commercialized in 1986 by 3D Systems in California and is the oldest RP technology. (SL is simply the generic reference for the process which produces a physical, three-dimensional object from a 3D CAD file). STL files convert and abstract surface data into triangular facets. When creating an STL file, a designer can select a resolution appropriate for the intended application. All software used by product and industrial designers should be able to generate STL files, and STL output is increasingly available in architectural modelling programs such as, for example, FormZ.
Rapid prototyping software divides an STL file into very thin horizontal layers. The thickness of these layers is specified by the machine operator and determines the relative accuracy of external surfaces. Thicker layers result in faster formation and thinner layers in greater accuracy.
Manufacturers of RP machines develop and continually update their own proprietary CNC software that automatically generates the instructions and toolpaths required to build a part.
Over the past few years, a new generation of relatively inexpensive rapid prototyping machines has become available. These machines are intended for use in professional offices and are often referred to as 'concept modellers' because they quickly and inexpensively produce models for use in communicating, evaluating and verifying designs. The trade-off made for speed and low cost is dimensional accuracy, but for many architectural applications this relative lack of precision is not nearly as important as in the automotive or aerospace industries, especially in the early design stages.
There are a wide variety of additive technologies, but the dominant commercially available processes are:
Stereolithography machines (3DP). These manufactured by 3D Systems. The build chamber (the area where the model is formed) in a stereolithography machine is a vat of ultraviolet-sensitive liquid polymer, and the process works by incrementally tracing horizontal sections with a laser beam. The laser solidifies the polymer layer by layer from the bottom up until the component is completely built. As can be seen in the photograph, stereolithography machines automatically build delicate structures to support overhanging features. These supports are easily removed by hand, and finished components are usually lightly sanded. The stereolithography process is precise and produces strong, longlasting parts.
Selective laser sintering (SLS) machines. These are manufactured by the DTM Corporation in Austin, Texas, and EOS in Germany. SLS machines are very versatile and can process a wide variety of materials, including polymer powders, sand, and metal powders. Like stereolithography, SLS is a laser-based process, but instead of using laser light, laser heat is used to fuse grains of polymer together (grains of sand or metal are pre-coated with a binder, or, in EOS' machines, a binder is deposited between layers of material). As with 3DP, SLS parts are supported by surrounding powder while being built.
Fused deposition modelling (FDM) machines. These are manufactured in the US by Stratasys and use a computer-controlled extrusion head to incrementally deposit layers of melted thermoplastic. This material solidifies immediately after being deposited, and finished parts are quite strong.
Multi-jet modelling (MJM). Z Corporation's 3D colour printers use a standard ink-jet print head to deposit a water-based liquid binder onto starch (cellulose) or plaster powder. This binder is deposited in an area corresponding to a thin horizontal cross-section of the model.
The build chamber is then lowered by the specified layer thickness, and a roller spreads another layer of powder before the next section is printed.
This process is repeated until the model is finished, at which time it is removed from the build chamber, dried, and infiltrated with wax or resin to increase strength.
Laminated object manufacture (LOM). The build platform in a LOM machine is fed by a roll of paper precoated on the bottom side with an adhesive.A heated roller bonds a new sheet to the layer below, and each layer is cut using a laser in a process very similar to two-dimensional laser cutting. After a part is built, it is removed from a solid block of paper and easily finished like wood. LOM machines have large build chambers and, depending on part geometry, they can be cost-effective to use for bulky model components. If a model component can be CNC milled from wood, however, this approach is often less expensive.
Depending on the geometry, size and intricacy of an architectural model or component prototype, CNC milling can be significantly less expensive to use than rapid prototyping. Materials such as wood, foam, plastic and wax are less expensive to CNC mill than harder materials because they can be cut quickly without tool wear or loss of accuracy.
In order to use subtractive technologies, data in digital models is typically exported in a neutral format such as IGES. These neutral files are then imported into software such as MasterCAM, X-CAM, or SurfCAM, which generates the G-code or toolpaths required to instruct a CNC machine.
CNC mills and routers are generally classified on the basis of the number of axes or directions in which they be manipulated. Threeaxis machines can move in the x, y, and z directions but cannot be manipulated to access undercuts and other complex features. Five-axis machines can accommodate more complicated shapes. Parts can also be subdivided to facilitate subtractive formation. CNC milling can be especially effective for making terrain models and vacuum forming moulds.
Having a desktop rapid prototype device in every architects' office will not be commercially viable for some time, but like design firms in other industries, architects can easily have models made by sending data to independent RP processing businesses. These companies may specialize in a particular technology, or they may offer a variety of different RP systems.
Corporate design departments often own several RP machines, and an architectural firm may find that purchasing its own concept modeller is a viable option. It can also be argued, however, that the need to use a variety of technologies for different applications would make it advisable to outsource requirements.
University RP facilities are also often interested in partnering with businesses. In the UK, De Montfort, Warwick, Nottingham and Leeds universities are all conducting worldclass applied-RP research and also collaborate with private industry.
Dr Kevin Rotheroe is an architect at Freeform, One Stuyvesant Oval Unit 9D, New York, US. E-mail krotheroe@ nyc.rr.com