3D or not 3D?

CAD systems come in essentially three flavours. The one you choose depends on what you are designing. Far and away the most common across both the mechanical and the construction industries are 2D systems such as Autocad or Microstation, in which the fundamental input is a line. Join some lines together and you get a drawing. Most 2D systems today also have a parametric capability where a drawing can become a template from which a family of similar drawings can be easily generated. And that's about it for 2D – totally flexible and essentially stupid. If you want to create a beautiful 3D picture of your project you use a special visualisation program, which can often do animation and other clever things.

The second class

The second class of CAD system works by creating a patchwork of surfaces that are manipulated and glued together to form the required external shape. This defines a volume. These systems came from the design of cars and aircraft and are now widely used in all types of mechanical design. They are also coming into favour for prestige architectural projects, eg, Frank Gehry used Catia, an aircraft design system, for the Guggenheim Museum in Bilbao. The building was far too large to be a single Catia project and if you look closely you can see the joins between the various sub-projects quite clearly.

The third class of CAD system is the assembly modeller. This takes pre-defined elements from a catalogue and assembles them in space to form the finished design, just as one might use Lego blocks. This is a very powerful technique, which has the great benefit that it guarantees the integrity of the final project. It works best when the number of element types is small relative to the size of the project, but of course it does not cope at all well with the fancy curved shapes that are fashionable in architecture today.

I started my career in CAD at Applied Research of Cambridge with two assembly modeller projects: the Harness project for the then DHSS; and the Oxsys project for the Oxford Regional Health Authority.

Both systems were based on prefabricated building components, which in the case of Oxsys went to a very fine level of fitted-out detail. The economic driver in both cases was a huge increase in government funds for health building and the need to spend wisely and rapidly. The government funding went away again before very much had been built, but we at ARC were confident that the CAD system itself, called BDS, could be commercialised.

We were wrong. Having to pre-define the building elements and then locate them in space turned out to be too unnatural a working method for the average construction project. After a lot of fruitless sales effort we turned our attention to a 2D architectural drafting package called GDS, which was very successful.

My second experience with assembly modelling was in CAD for process plant design. Here, the technique fitted perfectly. A process plant consists of a very large number of simple manufactured objects such as pumps and valves, connected by a lot of geometrically simple stuff (pipes, elbows, flanges) while the rest of the plant (structural steel, access stairs and ladders, hangers and supports) is also simple to model parametrically.

PDMS

Importantly, the main benefit of assembly modelling – design integrity – was highly valued by the industry. This was particularly true in off-shore projects where the cost of remedial work on site is prohibitive. Cadcentre already had a well regarded plant assembly modeller called PDMS, whose main defect was its antique and cumbersome user interface. Moving PDMS onto the new Silicon Graphics 3D colour-shaded workstations solved that problem and started a new era of true "Legoland" design for process plants. No more need to build a scale model of a plant just to check its integrity. Not surprisingly PDMS soon became the world market leader in CAD systems for plant design and changed the industry in the process.

Could the success that we achieved with PDMS in process contracting be applied to other parts of the construction industry? That depends on the extent to which construction elements can be pre-loaded into a catalogue. Industrial buildings such as factories and warehouses would seem to be likely candidates, as would the simpler type of airport. But such an approach does require a strong "back office" to select, prepare and maintain the data for the project team to use. It is up to the industry practitioners to decide whether the economic case for true 3D design can be made outside the process plant arena. At least we now know what the parameters are.