Artificial intelligence is unlikely to replace engineers, says Charles Clarke, but knowledge-based tools are already playing an important part in testing and ‘genetically engineering’ complex design processes.
Buying the right software package could, in future, turn you into the requisite type of engineer, it is claimed – which is almost like saying that buying a set of spanners turns you into a qualified mechanic. Artificial intelligence is not an idea that will go down well with experienced engineers, but there’s no doubt that certain software tools can take the tedium out of everyday design work.
Knowledge based engineering (KBE) has been with us since the 1980s and large companies such as Boeing, BAE Systems and NASA, along with car giants Jaguar and Lotus Engineering, have bought into it. But it hasn’t taken off more widely until recently, because in its traditional form it was prohibitively expensive – around £50,000 a seat just for the software was the norm.
More affordable modules for the likes of CATIA (Knowledgeware) and Unigraphics (Knowledge Fusion) are now finally helping to bring this technology to the masses, although it must be an absolute essential to justify both cost and time taken to harness its full potential.
There are are a number of related approaches to KBE.
The traditional approach is Knowledge Automation, which has more in common with computer programming than CAD/CAM, where knowledge is captured with a view to automating the process for the production of new variants.
The second method is Knowledge Advisor, which uses information to guide the designer throughout the process.
Then there is Knowledge Expert Assistant, a relative newcomer and the design equivalent of a spell checker. As you generate or change any part of your design, the consequences of any change are continually checked against the rest of the assembly.
Finally there is the Generative Model approach, which uses the ICAD system from Knowledge Technologies International, now part of Dassault SystÃ¨mes.
Those who have adopted it can thank their engineers who had the vision to recognise its potential – in some cases betting their jobs on the outcome.
But the companies also had a specific requirement for this kind of approach and the resource and commitment to be able to use it to its full potential – the Boeing 777 and the Jaguar S Type, for instance. Early users have all made huge investments in KBE development and are reaping the rewards in terms of time and cost savings on new variants or re-designs.
The ICAD generative model is like the DNA or the genetic fingerprint of the design process. Every detailed facet and nuance of the design and its possible variants are captured and linked to ‘top level’ or functional inputs. This DNA can then begenetically engineered by manipulation of these functional inputs, so ‘more range’, ‘more passengers’ or ‘different engines’ can produce a new airframe design in a fraction of the time it would normally take.
When Boeing wanted to produce a long-range variant of the 777 it required more fuel and hence placed a greater load on the plane. These functional inputs were put into a generative KBE model of the airframe and in less than a week over 2,500 newfuselage parts, stringers, frames and skins were accurately detailed from the generative modeller. In the following weeks Boeing managed to go through a large number of re-designs, using this approach, to achieve an optimum long range design.
The ability to re-design a complete airframe in a matter of weeks saves an incredible amount of time in getting a design onto the market, while the financial saving can run into millions.
Another advantage with ICAD is that because it is mandatory to capture every facet, nuance and variant of a design, the design intent stored in the ICAD Generative Model is so real that once captured it never leaves the company.
However, while AI has obvious benefits, buying a solid modelling package does not make you a designer – the ability to innovate does. So until AI can truly innovate, there’ll always be a need for engineers and designers.