TechArchive

This article was originally written in the period 1995-2000

Only in the past few years have solid modelling systems been used in earnest. Has the conversatism of engineers now been truly overcome?

The very first CADCAM systems started by mimicking conventional engineering design – 2D draughting. Lines, arcs and other geometrical entities were mathematically defined in a 2D coordinate system, which could be manipulated and edited easily.

The addition of a third coordinate within the definition was a simple step forward. However, these initial 3D systems continued to use draughting geometry – lines, arcs and splines – producing drawings akin to a physical model made from pieces of wire.

While a wire frame model enabled engineers to picture and resolve 3D issues, such as complex assembly questions, there were also inherent problems.

A 3D wire frame model required more information to be generated and manipulated than a 2D drawing. Models and their subsequent derived drawings therefore took longer to create than a 2D drawing. Equally, a wire frame model is not a full definition of a real world object, which limits its use. Calculating mass properties or an intersection curve, or defining a styled surface was not easy. Such constraints led to surface modelling, where the mathematics of a surface or skin could be defined and manipulated.

Surface modelling enabled complex shapes, such as a car body panel, to be defined in a way impossible before. Such models cold subsequently be used for shaded image generation or machine tool path generation. However, adding surfaces to a wire frame model added yet more data, often making it more cumbersome for general engineering applications.

Solid modelling overcame these problems by combining all the geometry into a single model that fully defined a real world 3D object. A solid model is created by taking standard primitive shapes – cone, cylinder, torus, sphere and so on – together with swept 2D geometry, and joining, subtracting or intersecting the resultant objects together. A cylinder subtracted from a cube produces a hole, for instance.

Like most advances, though, the development of practical solid modelling software has not been straightforward. The mathematics involved is extremely complicated and it has taken many years to arrive at software that enables real world designs to be created.

Initially, some developers simplified the mathematics by faceting a model – breaking everything into lines, While this showed the benefits of solid modelling, the loss of accuracy was often not acceptable to users. Equally, solid models produce large amounts of data and it is only the recent advance in workstation power and PC technology that has provided hardware capable of supporting it cost-effectively.

All design work is more or less iterative, so models need to be easily changed. Initial solid modellers required the lengthy process of re-running the complete modelling sequence to undertake a change. Thus it was that the parametric, feature-based solid modeller was developed.

Here the rules that define the model are captured, so that a change is undertaken simply by changing a dimension or parameter and regenerating the model from the rules. Features are regularly used geometries, such as a keyway or countersunk hole. These are defined as parametric solids which can be added to the model in a single command.

Developments are continuing in solid model manipulation and the best modellers enable the geometric rules to be changed easily, providing a flexibility in design modelling that is a considerable step forward from the earliest CADCAM systems and the manual processes upon which these were based.

Nearly all products are made from an assembly of components. Building an assembly model by bringing individual models into a single part, as provided by early modelling systems, has proved impractical, given the large amounts of data involved and the work required to reflect a component change.

Leading solid modelling systems now dynamically access parts in an assembly, presenting the structure as a graphical tree. This enables the complete product structure to be defined clearly, variants and all, as well as allowing easy manipulation of the assembly.

In the best systems, groups of users can work concurrently on a single assembly, seeing the results of each other’s work as it happens. Thus models with tens of thousands of components – such as jet engines with all their bolts, washers and pipes – can be created and manipulated with ease.

Early adopters of solid modelling often came up against the limitations of the early modelling systems and would advise prospective users to proceed with caution. Quite simply, a solid modelling system is of little use if it cannot model the products you make, at component and assembly level. There are still software packages available today of which this is true.

On the other hand, the best modelling systems bring real benefits worthy of exploration. A solid model provides a far more complete description of the design intent and can be just as quick to create (including the associated documentation) as a 2D drawing. While it does not necessarily remove the need for a drawing, it can be used far more effectively in all the activities that use design data.

Rapid prototyping can reduce build and test times dramatically. Computer-based analysis of models enables fast design optimisation. Tooling and manufacturing design becomes easier and NC toolpaths can be applied directly to the model. With fully integrated systems, changes can be undertaken easily and rapidly and reflected automatically through all the downstream applications.

All this leads to a more engineering process, enabling higher quality, lower cost products to be produced in competitive time scales, as well as allowing time for new technologies to be introduced or new legislative requirements to be met. It is for these reasons that this technique will soon be the natural starting point for most engineering design.

• Desktop Engineering
• Tel: 01993 883555
• Fax: 01993 883201