by Gill Chapman, Senior Lecturer in Computer Aided Design, School of Cultural Studies, Sheffield Hallam University. Intergraph MicroStation for 3D Industrial Design
In the summer of 1992 the CAD hardware and software in the School of Cultural Studies at Sheffield City Polytechnic (now Sheffield Hallam University) was completely replaced. The necessity of choice focused staff minds on their educational priorities for CAD and on the type of equipment which would best enable them to fulfil them. The major CAD users were industrial designers, but a range of CAD activities from engineering design to fine art had to be catered for.
At this time the industrial design section was putting in place a new strategy for IT teaching on the two courses BA (Hons) Product Design and BA(Hons) Design with Applied Technology. It was important that the strategy came first, at least in essentials, and the new equipment came second. This meant that the software and hardware were specified to fit the strategy rather than, as is often the case, the other way round. The strategy is described in detail in my companion case study 'Introduction of 3D CAD into Industrial Design Courses'. This case study looks at the reasons for our choice of Intergraph MicroStation as our main teaching software and describes three years' experience of using it.
When I took up my position four years ago the facilities were actually quite impressive by Polytechnic standards. There were 12 Apollo Workstations running AutoCAD, CGAL (a surface modelling and animation package) and IDEAS GEOMOD. The packaging design section had just invested in 12 Mac LCs running Aldus Pagemaker and Freehand, plus a Mac IICx running PhotoShop. These were heavily used for the large graphics component of the HND in Packaging. There was also a reasonably good word-processing facility provided on PCs by the Computer Services department of the (then) Polytechnic, which was hardly used by design students. The CAD facility was also under-used, due to lack of staff time and expertise.
Unfortunately it became clear as the new strategy began to be implemented that the existing hardware and software were not as good as they at first appeared. CGAL was very difficult to use, was not in general use outside the Polytechnics and would not communicate in any way with any other packages. GEOMOD ran very badly on our Apollos, which were out of date when they were bought and which were also unable to run current versions of standard design software such as AutoCAD 11. Money became available to upgrade the facility and it was decided that it should be replaced. Exhaustive research was begun into what should be purchased, based on our teaching and learning philosophy and our research needs.
The first and most essential step when buying new IT equipment is to sort out exactly what is to be achieved when it arrives and how. If the underlying philosophy is not clear the new purchase is unlikely to be successful, however much money is spent on it. The decisions made by staff at Sheffield Hallam are described fully in my other case study, but it is worth summarising them briefly here.
The first priority was to emphasise three dimensional computing, both because we believed that this will be essential for students working in industry in future and because it has much greater educational benefit for three dimensional designers than does 2D CAD. It was also decided that all the students should be given realistic 3D CAD experience and not just the enthusiastic few. The use of this technology would be encouraged as an educational and creative tool and would be integrated as far as possible into the rest of project-based design learning.
It is often suggested that the second step, after clarifying the underlying philosophy, should be to make a check-list of the various activities to be undertaken and create a software specification to accommodate them. If money was no object this would be good advice, but if resources are limited it may lead to much wasted time spent in creating a wish list that can never be afforded. It is probably better to have some idea of a realistic hardware platform at the beginning of the exercise, and this will also narrow down a bewildering array of software altematives.
Because all students were to be given the chance to learn quite advanced computing it would not be acceptable to end up with less than the existing 12 3D seats, but the available resources would not run to replacing the old Apollos with other workstations. The choice was therefore between Macs and PCs. The Computer Services department, which would be looking after the facility, favoured PCs because they wanted to network them for management and security reasons. Luckily, 3D work was just becoming a viable proposition on 486 PCs and it was decided to look first at PC software. In the event, it was found that we could afford one or two workstations as well for advanced work, should that offer a significant advantage. Computer Services favoured Suns because they had experience with them and felt they offered the best value for money at the time.
Another problem was the nature of 3D CAD within industrial design and the lack of any kind of industry standard . Industrial design spans a spectrum between art and engineering and ideally both ends of this spectrum should be available within one piece of software. It was obvious that this ideal would be hard to find because of the way CAD has grown up within different disciplines, all of which impinge on product design, but none of which have been specifically aimed at it. 2D draughting on computer has been primarily aimed at architecture or engineering. 3D surface modelling has often been added onto architectural draughting packages in a half-hearted and unsatisfactory way. Altematively, 3D solid modelling has been available via the engineering profession, in which case it has also provided a link into CNC or CAM, but it has been very unfriendly to designers. The third route into modelling has been via packages aimed at computer graphics and animation. These surface modellers are very strong on the presentation side but they are not always sufficiently accurate for industrial design, nor do they have any way into CAM. This situation is no doubt one reason why the revolution predicted for product design within industry has been slow to take off .
Despite these difficulties, software had to be decided upon. A check-list of criteria was evolved which included:
As expected all this was not available within one package. The choice three years ago seemed to come down to two software providers, either AutoDesk with AutoCAD 11 and 3D Studio, or Intergraph with MicroStation 4 and ModelView, which had just come out for the PC. As a basic teaching platform, AutoCAD and MicroStation scored fairly equally on the first criteria of detail and accuracy. MicroStation scored much better on no.2, ease of use, and also on no.3 as it was the only software with NURBS then running on a PC. AutoCAD had more industrial credibility at PC level, but Intergraph was known within industry for their workstation software and they were also making progress with MicroStation. For basic rendering, MicroStation was better than AutoCAD, but for more sophisticated work, and particularly for animation, 3D Studio was far superior to ModelView. If we could afford one or two workstations it seemed to be an advantage that Intergraph provided software for these as well as for the PC, so that we could maintain contact with solid modelling within one suite of packages. We looked at Intergraph EMS running on a Sun, and at I/Design which had recently come out and was aimed specifically at industrial designers.Initial Perception of Software Performance
|Ease of use||Good||Poor||Poor||Good|
|Sculptural surfaces||Good||Poor||Poor||Industrial credibility||Adequate||Very good|
|Rendering||Good||Very good||Poor||Very good|
It was eventually decided MicroStation 4 would be the main teaching platform, with four copies of ModelView for advanced presentation and animation of MicroStation models. One copy of EMS was put on a Sun Spark workstation, the idea being that some MicroStation models would be taken into EMS and edited using the solid modelling techniques available in that software. Four copies of 3D Studio were also purchased for the fine art students wishing to do animations, as their need for accurate models is not as great as that of the industrial designers.
There was already a A0 pen plotter in the graphics lab, as well as video output. For colour stills we take photographs uf the screen, which is surprisingly satisfactory. Since the initial outlay, a PowerMac has been purchased, plus PhotoShop and a colour scanner. There is no problem with sending image files between the PCs and the Mac via the network.
The total budget over two years was œ70,000, which incuded for the basic MicroStation facility:
|Price per Item (UK Pounds)||Total Price(UK Pounds)|
|12 x 486 PCs with 16Mg RAM, graphics cards, super-VGA screens.||2500||30000|
|12 x MicroStation 4||100||1200|
|12 x MicroStation 5||350||4200|
|4 x ModelView||350||1400|
|4 Gigabite Server||8000||8000|
MicroStation 4 was put to use with the first Year 1 group in the autumn of 1992. In January 1993 it was replaced by MicroStation 5. As there are two industrial design courses, the new teaching and learning plan has now run six times; twice on our old software, three times on MicroStation 4 and once on MicroStation 5.
All the industrial design students now do a 'Design with Computing' unit in their first year. The teaching methods are described in 'Introduction of 3D CAD into Industrial Design Courses', section 3. There follows here a description of the skills which the students learn in their classes, which will also give a flavour of MicroStation's capabilities. It has evolved somewhat with use, particularly with ~e upgrade to version 5.
Because MicroStation is reasonably intuitive for designers to use, the students are able to progress quite rapidly through basic 3D CAD modelling techniques. Before they start, they have had about six hours computer familiarisation and word-processing instruction. They then plunge straight into 3D. The majority of the students are able to cope with each session in two hours.Session 1 - Introduction
For some groups, classes have stopped here in Year l and a two week project has been set using the commands used so far. After only ten hours work most of the students are able to start using the computer to create their own 3D designs (see Fig 1in 'Introduction of 3D CAD into Industrial Design Courses').
With other groups classes have continued:Session 6 - Complex Model Creation
Rendering facilities are much more sophisticated in MicroStation 5 than they were in version 4. Despite the lack of ray-tracing, they are as nearly as good as ModelView and much easier to learn. First year students usually find out quite a lot for themselves, but advanced rendering techniques are not actively taught until Year 2. Previously using ModelView and now MicroStation 5 students leam:
As mentioned elsewhere , 2D draughting is leamed after modeUing, not before. MicroStation is a very good tool for the production of engineering drawings, at least the equal of AutoCAD in power and facility and considerably friendlier to use. For die-hard AutoCAD users it can be configured to look like their old favourite. After learning 3D, students teach themselves 2D with the manuals, or create drawings from their models.
Because of time constraints, animation has not been generaUy undertaken until Year 3. Intergraph software performs reasonably weU for 'waUc-throughs' or'fly-rounds', which are now available within MicroStation 5 as weU as ModelView. For animating moving parts of a model, which is the main purpose of animation for industrial design students, the situation is not satisfactory (see section 7).The quite impressive results achieved by some students has been in spite of, rather than because of, the software.
Apart from work done on an individual basis by a few students this aspect of our stated aims is not yet in place. Last year one of the third year product designers took a design right through from a CAD model done on EMS to a stereo-lithography one, with the help of two different industrial sponsors. Taking MicroStation designs right through to physical models is the main content of our current IT development.
Inevitably, not everything that was hoped for from MicroStation has come about, but it has nevertheless proved its worth as a low cost modelling package. The most important aspect has also been the most successful, in that a high proportion of first year students have become confident in the use of 3D modelling on computer. They are therefore genuinely able to assess its merits and decide how far they want to extend their use of the tool. Compared with the old software, students now find this learning curve much easier to climb. When the use of CAD as a creative tool is looked at the benefits of the new software are even clearer. Very few of the first groups used the computer creatively, whereas several of the later groups did, even in their first computer project and after only ten hours of fommal teaching.
The major problems encountered have come at later stages when students move on to more advanced work. The B-spline surfaces which reaXy tipped the balance in favour of MicroStation initially tumed out to be full of bugs in version 4 and it took several frustrating months to work out which commands were reliable and how to get round the various problems. In version 5 these commands are much more robust, however one command which would have been useful had it ever worked has been dropped altogether - an admission of failure, presumably. Similarly, the Boolean operations which are new in version 5 are very unpreditcable. A proper solid modelling add-on package is now in the pipeline, which will hopefully work better. It seems to be the case that one can only rely on the commands which have been in the software for at least two releases. I gather that this is not so in the case of AutoDesk software.
Another problem has been that MicroStation does not really work well when run on a network. There have been endless problems which have been hard to diagnose because they could be the software or they could be the network. Many of these seem to relate to the way MicroStation uses temporary memory space on the hard disks. Our Computer Services department is continuing to work on these problems but they are still not entirely resolved.
ModelView has been a disappointment. The interface is confusing and does not relate well to the MicroStation interface. It is extremely clumsy for animating parts of working models, involving constant to-ing and fro-ing between the two pieces of software. It has also refused to run from the network, making the whole process even more fiddly and time-consuming. The situation has improved somewhat with the enhanced rendering and 'fly-through' facilities within MicroStation 5. As soon as we locate a satisfactory add-on package to animate moving parts of the model within MicroStation, ModelView will be dropped with sighs of relief all roundSoftware Scores after Extensive Use
|Ease of use||Very Good||Very good||Very Poor|
|Sculptural surfaces||Poor||Good||Industrial credibility||Adequate||Adequate|
|Rendering||Good||Very good||Very good|
Apart from the excellent work of one student, EMS 2 has also been of limited use so far and is under review. Although simple MicroStation models can be transferred to EMS, models containing B-splines have proved problematic, and B-splines are heavily used by our students to create the shapes they want. It is hoped that EMS 3 and MicroStation 5 will work better together, but this has not been tested so far.
Our latest purchase has been two copies of Pathtrace CAM software which takes in MicroStation models and produces CNC code for the machine tools in our own and the engineering department workshops. This exercise is still at the testing stage. It is hoped that the students will soon have the option of taking their modelling skills either towards advanced presentation work, or towards CADCAM and more scientific applications such as FEA. They will then have spanned the art to engineering spectrum, if not with one piece of software. at least with one data model.