Tom Shelley finds out what it takes to overcome complex design problems to build world beating products.
An automotive company makes a materials breakthrough and transforms itself from a craft to a mass production business, while still making sports cars with world beating performance; an aero engine maker embraces a new software technology to manage increasingly complex design problems; while offshore and military suppliers tackle constantly changing requirements in exceptionally unforgiving environments.
What all these applications have in common is the problem of managing increasing complexity in a competitive environment, which requires a combination of innovative technologies, careful management and advanced software to help manage processes, keep track of everything and ensure everyone appropriate has access to the same data.
A classic example of this combination was to be found at McLaren Automotive's recent, high-profile press conference to launch the new MP4-12C road car, Ron Dennis, the executive chairman of McLaren Automotive explained his motivation for entering the competitive arena of car manufacturing. "Since 1966, 106 F1 teams have come and gone – only Ferrari and ourselves are still in the pit lane." Having made the decision, the two keys to giving the new production cars their edge were to combine technical innovation with a process that made use of the company's F1 technology. This is complicated enough, one might have thought, but to adapt it to volume manufacturing adds a range of additional complications.
The most striking technical innovation in the MP4-12C is its use of a tubular carbon fibre 'MonoCell', which protects the occupants and provides the car with most of its structural strength. The rest is provided by an aluminium frame carrying the 3.8litre M838T engine (which is bolted on the back) and a sacrificial aluminium structure to provide crash protection (which is bolted on the front).
The structure is made by resin transfer moulding, with a cycle time of four hours, two hours of which are spent being cured under pressure in a 400 tonne press. It is finished by CNC milling to a dimensional tolerance of less than 0.5mm. When asked how the hollow sections were produced under pressure, Claudio Santoni, function group manager – body structures declined to reveal details. He did admit, however, that it took four years and 94 trial mouldings to get the process right and, when asked about the possibility of using CAD and FEA to get the design right first time, he responded with a wry smile that this only applied to products that were pretty similar to ones that had been made before.
As regards the task of achieving relative mass production – the plan is to manufacture 1,000 cars per year in a completely new factory – operations director Alan Foster said: "My challenge has been to combine traditional Formula One craftsmanship with more mass production car building techniques." This meant instituting formal methods of supply chain management and component traceability, modelling the production process using Dassault's Delmia suite and using 3DCS for dimensional analysis. Catia V5 was the CAD package used and, according to head of engineering Neil Patterson, PDM was undertaken using a combination of its in-house 'Maxim' ERP system as master for Bill of Material and configuration management in combination with Enovia 3DComm, which is updated as an ongoing process.
The car was virtually driven for 300,000 miles on McLaren's simulator to tweak the design, and some 20 physical prototypes have been built and road tested in various parts of the globe at ambient temperatures from -50ºC to +50ºC. Test mileage is said to be around a million miles, including prototypes being driven 24hours at a time by relays of drives and there have been around 100 crash tests.
McLaren is not alone in pursuing perfection in this way, however. The same almost obsessive attention to technical innovation, management of the design and production process is to be found at the equivalent market leader in aerospace engines, Rolls-Royce.
The complexity of the engine-building process is put in context by Jan Larsson, EMEA marketing manager for CAD and PLM supplier Siemens PLM. He points out that, while a typical jet engine in 1960 had 3,000 parts, by the 1990s, this had risen to 20,000, while the lifespan of such an engine is typically around 30 years and data on it needs to be accessed for around 50.
Given that Rolls-Royce has what Larsson describes as "terabytes of data" to cope with, it is clear that a product of this complexity requires highly sophisticated information management.
The company uses commercial products NX for CAD and Teamcenter for PLM from Siemens PLM, but, not content with the capabilities of commercially available software, also makes extensive use of DRed (Design Rationale editor), a bespoke programme developed by the University of Cambridge Engineering Design Centre.
DRed originally came out of two EPSRC 'Grand Challenge' projects: KIM – Knowledge and Information Management and HIPARSYS – High Performance and Robust Systems. One of the findings of KIM, according to Dr Peter Heisig at the EDC was that the number one knowledge capture need seen by engineering companies was to capture rationale, ie 'Why did we do it this way and not the other way'. By comparison, design descriptions and changes only came in at second, third and fourth respectively.
The original capability of DRed was to set out complex design problems where lots of functions interact and show interactions between different design approaches as well as pro and con arguments for different possible solutions over multiple pages. Lightbulb symbols represent ideas and plain text blocks facts. Green text is true and red text marks it as false. Lines through text indicate that an argument has been rejected, but it is still retained in case it needs to be revisited.
The product has recently been enhanced by the addition of "Blocks" that can be used to show the relationship between different parts and functions both inside and outside the machine being studied to create Functional Analysis Diagrams. White blocks represent external parts, while grey blocks represent internal parts. Red-coloured relationships are harmful, such as those that create windage or excess heat, while green ones are beneficial.
This process has been used extensively by Rolls-Royce in the development of a gearbox in the Trent 1000 engine for the Boeing 787 'Dreamliner'. Apart from helping get to the right answers quicker, at the end of each study, a day is saved in not having to write up documentation, since the system records everything considered, including ideas that have been rejected. The EDC has applied to it a number of other engineering design problems, such as optimising hydraulic motors. Because the methodology is completely generic, it has the potential to be applied to management problems as well as engineering.
Just as complicated as aero engines, if not more so, are some of the pieces of equipment for offshore work, where products are very often one-offs and may be designed and developed by small companies without access to sophisticated big ticket computer systems. The environment is exceptionally hostile.
David Blood, a design engineer with Tech Safe Systems, which makes complicated launch and recovery systems and split-purpose control cabins, relies on Autodesk Inventor. Designs include hydraulics, electrics, electronics and fibre-optic connections to ROVs (Remotely Operated Vehicles. When asked how he ensured that everything worked, he responded: "We do lots of software simulation, mostly using Autodesk Inventor. Part of the simulation involves finite element analysis, for which we used to use ANSYS but now we do it all in the Inventor environment."
Even so, designs are not always right first time. Says Blood: "There are always minor problems, typically arising from manufacturing tolerances and slight oversights." Management controls and data control includes full drawing control and release and export of Bills of Materials directly from Inventor to the company's 123 Insight MRP/ERP/CRM system.