Preserving CAD

15 August, 2013

Computer-Aided Design (CAD) models are a type of file no archivist really wants to deal with. Not only do they tend to be large, complex and split across several files, they also tend to be poorly supported by any version of any system other than the one used to create them. So the archivist has to put in a lot of effort, and invest heavily, to ensure they can be read in the long term.

But a lot of valuable information is tied up in CAD models, from the designs for cars, ships and aircraft, through the plans for large public buildings, to irreplaceable data from archaeological digs. It is not a problem we can afford to ignore.

Recognising this, the DPC (Digital Preservation Coalition) held a Briefing Day in 2010 on Preserving CAD and commissioned a Technology Watch Report on the subject. A second Briefing Day was held on 26 July 2013 at the Institute of Mechanical Engineers in London to launch the report, and as I was the one to write it, I went along to give the opening talk.

My talk was billed as an overview, so I went through the basics of CAD: what it is used for, what the main 3D modelling techniques are, and what special capabilities have been added to CAD software over the years. I then introduced some important exchange and visualisation standards such as IGES, STEP and JT, before providing some recommendations on how to approach preserving CAD.

I was followed by Sophie Herail from Airbus, who gave an overview of LOTAR International (the name comes from LOng Term Archiving and Retrieval). This ambitious initiative is developing a series of standards for archives of product model data, and involves most of the big names from the aerospace and defence industries. The LOTAR standards explain how to use existing standards like OAIS (ISO 14721) and STEP (ISO 10303) to build an effective industrial archive. This is especially welcome in the case of STEP, as it is extensive – the largest standard ISO has produced, in fact – and bewildering to the beginner.

Next up was Kurt Helfrich of RIBA (Royal Institute of British Architects). He discussed the challenges of archiving the plans for the current redevelopment of London's King's Cross station. I think the most valuable insights Kurt offered were about the whole process of preserving design documentation, rather than the technicalities of CAD:

  • handling intellectual property rights,
  • keeping sensitive data secure,
  • appraisal and selection,
  • chasing up missing originals, and
  • getting producers thinking early about preservation.

In my talk I'd discussed the three main preservation strategies of emulation, migration and normalisation. Hans Ulrich Heidbrink of the SHAMAN Project discussed a fourth approach exemplified by the Multivalent tool: a universal document viewer which allows annotations and edits to be layered on top of the original document. SHAMAN used this approach with product data from Philips, though CAD software being what it is they had to normalise the models to visualisation format JT first. This wasn't the problem it might have been, since Philips don't need to adapt or interpret their designs: they either re-use designs as they are or start from scratch. And a plus point of JT is that one can manufacture from it.

The last three speakers each identified emerging trends. Stuart Jeffrey could give a perspective both from his current work with the Digital Design Studio at the Glasgow School of Art, and his previous work at the (ADS) Archaeology Data Service. He predicted that in the cultural heritage sector

  • archives will need to be more selective about what they preserve;
  • there will be more coordination between archives on what they will accept;
  • workflows will have to improve to bring down preservation costs;
  • those whose heritage it is will get more of a say in how it is preserved.

He also suggested that 3D printouts of CAD models might be a viable preservation strategy in cultural heritage contexts.

Ruggero Lancia (University of Glasgow) felt the big issues in architecture were in the fast-developing areas of BIM (Building Information Models), generative modelling and digital fabrication. He identified a need to train architects to better manage their records – they shouldn't have to rely on specialist archivists all the time. He also demonstrated that while open standards such as IFC (Industry Foundation Classes) for BIM are promising, they aren't ready for serious practical use yet.

Closing the day was Sean Barker (BAE Systems), who noted that:

  • engineers care more about models than the files used to store them;
  • many types of model are used to represent a product, and engineers increasingly have to consider them as a set;
  • the different types of model are becoming more integrated.

On this basis, Sean predicted that engineers will stop using files to store their data, and instead use databases. In time, therefore, archivists will be preserving databases as a matter of course, with file preservation becoming niche.

If there was one message I took from the day, it is that we cannot afford to stand still on CAD preservation. It is already a moving target, and unless CAD vendors, customers and industry regulators take the issues seriously it will be an uphill struggle to keep pace.

The presentations from the Briefing Day are available from the DPC website, as is the Technology Watch Report 'Preserving Computer-Aided Design' [PDF, 930KB].