3D CAD-Is there Another Way?

These are some thoughts on an alternative which, for some applications, maybe a better approach than using a general purpose 3D modelling system. This alternative maybe of interest for you if:

You don’t use your 3D CAD system or you only use a very small subset of the facilities
You've found 3D CAD systems too difficult or cumbersome
It would take too long to define 3D geometry relative to the benefits you already achieve with 2D CAD

Problems with 3D CAD

3D is a much more specialist area than 2D CAD and no single 3D modelling system is suited to all applications. For example, a system designed to model complex doubly-curved 3D surfaces will not be able to model petro-chemical plants. However, this has not stopped some CAD vendors from trying to promote their 3D systems as being suitable for all applications. In our opinion this has never been the case and, we suspect, it may never be so.

One of the most striking conclusions drawn from a visit to a recent Solid Modelling exhibition, is how similar most mid-range 3D modelling systems now are. Mid-range modellers include Pro/Desktop, SolidWorks, SolidEdge, IronCAD and Inventor. As these systems run on Microsoft Windows, the user interfaces now conform to that standard. This is beneficial in that any user who is familiar with one Windows application will find a 3D CAD system that uses the same standard somewhat familiar and it will be, therefore, quicker to learn. Also most mid-range 3D CAD systems are based on one or other of the two common modelling kernels ACIS and ParaSolids. This could be a benefit if it makes the transfer of 3D geometry between systems easier and less error prone. Unfortunately there is still a long way to go before we can claim that the data transfer problems have been solved! (NB OpenHSF is a more recent development which is a significant step forward compared with IGES, VRML, etc.)

But why do all these 3D modelling systems have to work in the same way? Are there not other alternative approaches that can be used to create and manipulate 3D models? We believe there are, but we would not begin to suggest that the alternative outlined below is suitable for all, or even very many applications.

3D systems have become easier to use than they were 20 years ago when 3D wire frame modellers were the norm. But it is still at least an order of magnitude more difficult to work in 3D than it is in 2D. Almost the only exception are pre-prepared and well rehearsed sales demonstrations of 3D CAD systems. It’s always amusing when these demonstrations fall apart after a simple, but realistic, geometry edit is unexpectedly requested by an innocent prospective customer!

Not only are 3D systems more difficult to use, the software is also much more difficult to develop. The maths involved is extremely hard to comprehend and designing programs to handle the large volumes of calculations efficiently is very tricky. From the programmers’ point of view it is often much easier to have the user fully define everything in 3D space in a way that suits the program rather than in a way that corresponds to the user’s view of the real life objects he's trying to represent. Faced with these difficulties programmers often seem to forget that a good system should help the user rather than overloading him with unnecessary work.

General purpose mid-range 3D modelling systems may now be cheap to buy (unless you need a number of the optional extras which can cause the price to rise alarmingly). But all general purpose 3D modellers are expensive to use. They are difficult and time consuming to learn and to use on anything more than trivial geometry. They also need regular practice. This due to the nature of, and the fundamental difficulties inherent in manipulating 3D geometry. The Holy Grail with 3D CAD vendors is now ease of use, but nothing we've yet seen provides any convincing proof of any significant advances. Defining 3D models parametrically only adds to the problems. As a general rule anything in 3D will take at least 4 times longer than the equivalent 2D construction, and all too often users give up after wasting hours unproductively.

3D is a fascinating area and the results can be very satisfying and is sometimes the only factor driving engineers to use 3D rather than 2D. But 3D can be a dangerous waste of time and involve a lot of pain for little real gain. This is especially so in electrical applications where the benefits of 3D are normally much less than in many mechanical engineering applications.

An Alternative Approach

In most applications 3D is only necessary for some aspects of the engineering process. Many design problems can be analysed and largely solved in 2D. For example, a warehouse layout can be largely designed using a plan view and perhaps one or two front and side views of some typical racks showing the stacking of pallets.

With some applications it is not even necessary to work to scale. For example, piping schematic drawings showing the interconnectivity of pipes and equipment enable much design work to be completed long before it is necessary to contemplate the complexities of the scale and size of the various items in 3D or even 2D.

As it is much more difficult to work in 3D than in 2D, wouldn’t it be nice if you could work in 2D and then switch to 3D only when it was necessary? This isn’t possible with applications where 3D is required from the outset, for example, engineering components that involve complex doubly curved surfaces. But for applications where a lot of design work can be safely and productively completed in 2D, why bother with the complexities of 3D until absolutely necessary?

This isn’t a new idea. For example, Cambridge Interactive Systems Ltd. developed a system in the late 1970’s which became known as Medusa. This system is mainly used in 2D but a 3D modeller is also available, if required. We had some significant successes customising Medusa to suit a number of applications including piping and warehouse racking which naturally fell into distinct 2D and 3D tasks. Unfortunately from the mid-1980’s Medusa suffered a number of changes of ownership, and the unique strength of the 3D system now seems to have been largely forgotten or misunderstood.

Special Purpose CAD Systems

An alternative to the general-purpose, jack-of-all-trades approach are 3D CAD systems that are developed for specific applications. Such systems tend to have a much more restricted set of features than general-purpose 3D CAD systems. However, as they are tailored to a specific application they can be significantly easier to use and more efficient.

Now that humble PCs include powerful 3D graphics cards, it is economically feasible to consider custom built 3D modelling systems for certain applications. A custom solution may cost the equivalent of 10 copies of one of the popular off-the-shelf mid-range 3D modelling. Therefore, if more than 10 copies are required the cost may even be less. Moreover the tailored solution should be more efficient for the particular application and the operational cost savings can be considerable.

A system we developed to help design electrical wiring harnesses serves to illustrate this alternative approach.

Example Application

HarnWare is a wiring harness design system that ADE Ltd. develops and supports for Raychem (now part of Tyco Electronics). There are now over 1,000 HarnWare users in more than 300 companies in 30 countries. HarnWare can suggest a wiring harness design sequence and help with many aspects of the design process including:

Part selection to ensure compatibility with mating parts and service conditions
Calculation of wire routes and lengths
Parts listing
Assembly time estimates
Determination of wire gauge and optimum lay of mixed diameter cables

All of these tasks and many more can be effectively undertaken with a non-scaled, 2D drawing of the wiring harness. It is much easier for the user to construct a drawing using fixed sized shapes to represent each part than it would be using scaled 2D or 3D geometry. HarnWare uses the Microsoft Visio 2D drawing system, which is low-cost and ideally suited to this type of work.

However, while non-scaled 2D drawings are perfectly adequate for most aspects of the design process, there comes a point when an accurate to-scale 3D model is useful to check the fit of mating parts and dimensional integrity. With a to-scale 3D model the user can instantly see if the design ‘looks right’. Also an accurate to-scale representation is required if a lay-up board is needed to help assemble the wiring harness.

We, therefore, developed HarnVis which is a 3D modelling system for visualising harness designs generated by HarnWare. HarnVis automatically generates to-scale 3D models which provide “virtual prototypes” of harnesses designed using HarnWare. The user can see what a harness will look like with lengths, diameters and parts shown to-scale. This reduces the potential for errors.

Sometimes alternative Raychem products suggested by HarnWare have different overall dimensions which can affect the fit within the equipment. Overall sizes can easily be checked in HarnVis by selecting one or more parts. HarnVis draws the minimum enclosing orthogonal volume over the selected parts and the dimensions can be queried. By clicking a part in HarnVis, the key part dimensions can be displayed.

Raychem moulded parts are shrunk to fit onto mating adaptors and cables. Therefore, these sizes are infinitely variable, which makes the parts difficult to define using a general purpose 3D modeller. HarnVis was specifically designed to handle these complexities.

The models are intelligent in that the user can access such data as part numbers, materials, finishes, adhesives, etc. simply by clicking a part. As is normal with 3D modelling systems, HarnVis 3D models can be viewed from any angle and centred on any part. Parts are rendered in colour with lighting effects, which yield very realistic representations of the Raychem product materials. These views are ideal for quotation drawings, presentations, exhibitions, etc. They have a more immediate impact than 2D schematic drawings, especially for those who are not trained in reading engineering drawings.

The 3D modelling methods used in HarnVis are finely tuned for this particular type of geometry. These methods are more restrictive than those used in general purpose 3D CAD systems. But with HarnVis the ease of use is dramatically better. The user works in the simple and fast 2D Visio/HarnWare environment and 3D models can be produced fully automatically in only a few seconds whenever the user wishes. The 3D complexities do not get in the way or slow the user down as it is an optional output, rather than a working environment imposed upon him in the manner of most 3D CAD systems. 3D geometry is inherently difficult to work with, especially in a parametric form and, therefore, ease of use is still a serious constraint on the use of general-purpose 3D CAD systems.

Click the above thumbnail images to download the free OpenHSF viewer and view two sample models which were automatically generated by HarnVis from 2D HarnWare designs. The viewer install file is 1.6Mbytes but the two sample OpenHSF files are less than 50 Kbytes each. Therefore, while it may take sometime to view the first OpenHSF file, other files can then be downloaded and viewed in less time than many simple Web pages. More than 13,000 faces are used to define the geometry of this simple harness and larger harnesses can easily require 100,000’s of faces in order to represent the curved surfaces to a sufficiently smooth and aesthetically pleasing standard.