An Example of Model Production

Table 1 identifies the application of model production. This is useful for visualisation, communication and conflict resolution. The application of architectural models is not new and Rapid Prototyping is a proven technology in this respect. For wide spread application in the industry, across architecture and engineering, the transition of information from designer and CAD to Prototyping machine must be simple. This initial investigation to identified some of the issues with using a model from a construction oriented CAD package to produce a physical model using a Rapid Prototyping process.

Virtually all Rapid Prototyping machines accept a data format called STL. This data format describes the object to be built using a single tessellated surface where the inside and outside faces of the surface is defined. What is ‘inside’ is deemed to be solid. This data model is ‘sliced’ into layers. The machine interprets which part of each layer is solid and which is not. The information controls the sequential reconstruction of the object in the physical world, a layer at a time, ground up. There is a continual cycle of operations, applying a new layer of material across the build platform and selectively turning the ‘internal’ parts of that layer solid using an activation process. In the case of the Selective Laser

Figure 2. On the Left, the CAD Image of the Object and on the Right, a Photograph of the Finished Item.

Sintering, the solid parts of each layer of powdered material are melted, subsequently cooling and solidifying.

Where solid 3D CAD software is employed, conversion of digital model information to STL format suitable for Rapid Prototyping machines can be relatively trivial. There are two issues that are important for model production of building components. Although solid modelling is becoming more widely used, a great deal of design information is detailed in 2D CAD systems or are rendered 3D images using surface modelling. It is generally difficult, and for complicated systems practically impossible, to convert these models into good[10] STL files. The machine must be told what is and isn’t solid and hence must have objects with perfectly joined edges, clearly defining volume. For example, it is clear to us that a pool ball is solid, but STL data would describe it as an infinitely thin surface that has a certain volume inside. If there is a hole in that surface, you can’t define what is inside and what is out and hence, what is solid. Problems like this will cause a ‘build’ to fail. Surface modellers and 2D CAD software are not designed to describe volume and so can’t be used to generate an accurate description of a surface that will instruct a machine to make a physical object.

The second issue for construction is that assemblies of components can be problematic. Construction system designs generally describe an assembly of individual components; steel frame systems are a good example. There are construction orientated CAD systems that use solid modelling, Tekla Structures is one example (Tekla 2005). This system models every component down to the nuts, bolts and washers and hence a model of a whole system, is described as an assembly of solid objects. If a model of the complete system is required, the data must be converted to one surface describing a single object. The issue is how to combine multiple objects in the assembly so that the desired model is described as one homogenous unit.

The left hand picture in Figure 2 details the CAD image of an example steel section modelled using Tekla Structures. This software models each component as a solid object but it does not have an integrated STL conversion function. To generate the STL file, the data was loaded into the solid modeller, Rhino 3D (Rhinoceros 2005), using a file format common to both (DXF). Using the automated functions within Rhino, the STL file generated was made up of many surfaces. Each surface defining the individual objects in the design. A model of the entire assembly was required. Using Magics, an STL file editing software package (Materialise 2006), each object surface was combined using an operation which merges one object with another, resulting in a single surface. The dimensions of the object were scaled to suit the size of the Rapid Prototyping machine and then built in nylon using a Selective Laser Sintering process. The final part is depicted on the right hand side of Figure 2.

At present, if a model of a building, component or subassembly is required, it is likely that separate operations will be required to encode the design into solid modelling software in order to produce a STL file that is capable of instructing a machine to build the physical object. If an assembly of solid objects are used to describe the item to be built, additional operations will be required to generate a single object in STL. As software develops, it is likely that the latter will be achievable through prescribed software functions. If solid modelling becomes the standard way of describing building information in the future, the former issue will become obsolete.


There are many examples in construction where new technologies are moving away from the use of traditional methods of building procurement to those that can be designed and fabricated digitally. It is likely that this trend will grow. To enable the full benefits of construction automation, radically new process are required. The freeform approach is well placed to contribute. By engaging with industry, potential applications have been identified. These range from near market solutions to some that will take years, even decades to be exploited commercially.

Generally freeform methods can find ready application in geometrically complex building components or the integration of systems. Specialist applications seemed to be most probable locations for early development. One hurdle to overcome is that of information handling. Solid modelling as a real tool for construction is in it’s infancy when compared to ship building or aerospace engineering. An additional factor is that to be successful, the digital building model information needs to be readily used by all members of the supply chain. This will mean greater use of CAD/CAM manufacturing technologies by suppliers. The modelling example, although fairly trivial, does demonstrate that there is a little way to go before the use and exchange of construction information with Rapid Prototyping type processes becomes trivial. Such issues will be exacerbated when generating full scale structures.