The challenge of time for CAD-archiving

How do we synchronize the 10-20 year lifespan of cars, 40 years lifespan of ships, the maybe 70 year lifespan of aircraft designs, and 100 year lifespan of nuclear reactors, with the limited lifespan of computers and normal design software?

How can we read digital models, many years after the original application that the model was made with ceased to exist? Historically, new versions of CAD systems are released at least every six months, and are replaced or completely rewritten every few years.

The LOTAR project has been established to find the answers.

A Business problem

We are not merely speaking of a pure technical problem. What we see is a real business problem. A complex product may take up to five years to get into production, and be in production for many years. However, design data does not cease to be used when production finishes. It continues to be used in servicing the product, in manufacturing spares, and in modifying the product - not just upgrading it, but responding to changes in regulations or finding new suppliers.

The original CAD system will become obsolete quite early in the life of a major product, especially in the aerospace industry. Not only will it be obsolete, but even if a version is available, the number of people who know how to use it will also get fewer and fewer. Hence the CAD system must be replaced at several points in the product lifecycle. Even if we are able to drop the product after 20 years (and risk losing customers), we would still be legally liable while it continues in service, and usually for several years longer, and would need to provide design data in case of accident.

The LOTAR Project

This gives us the needed background to understand the LOTAR project, which, through the European body for the development of global Aerospace standards (ASD-STAN), aims to develop a standard for long term retention of digital aircraft data.

Project members in LOTAR have included BAE SYSTEMS, Airbus, EADS, Dassault Aviation, SNECMA and MTU, and is supported by the ProSTEP association. The European industry is working closely with its US counterpart, the AIA to promote a global consensus, and with the PDES Inc. Long Term Data Retention project.

From drawing to modelling

Early CAD systems were viewed as an electronic version of the work that had been done by pencil for centuries. The main advantage in the electronic version was that when changing one line all others would be redrawn automatically.

While there has been much development in CAD since the 60’s, most has been about improved tools for creating and checking shapes. In the last few years this has changed. More complex functionality has been added, in which the tool itself generates extra information, such as geometric tolerances.

Model in must be model out

Consequently, if we use the model as the master source of the product design, we need to keep it. However, since the model is always interpreted through software, we need to show that software correctly interprets the model. That is, if we take a model, and put it into an archive, then, when we take it out again, we must show that what went in is what comes out again.

A tecnique for verification

A verification solution is needed. Cloud of points is a technique of putting points over the surface of a model, and using these as the verification mechanism. That is, we can use the cloud of points to automatically check that what went into the archive is the same as what comes out. Further, rather than putting points at random over the surface, points can be put at critical points, either to show that the surface is accurately reproduced, or that edges have not moved significantly.

The focus of LOTAR

We have now defined the data retention problem, and seen a possible verification solution. At the simplest level, one could keep the computer and the operating system and the software. There has been work on this for document retention, but is out of scope for the LOTAR project, since the sorts of modeling tools used in the aerospace industry not only require complex software, but also skilled users.

At a higher level, there is a problem of retaining data, for which the OAIS archiving standard provides not only a reference model, but a template. The next level is to retain the model, that is, to ensure the information remains well structured and accurate when read into a new modeling tool.

This is the focus of the LOTAR project. However, the model does not tell the whole story, and needs to be supplemented by the knowledge of how to interpret the model - an area where much work has started in recent years.