 |
|
|
|
 |
 |
||||||||||
 |
|
||||||||||
 |
|
September 2008 |
|||||||||
|
|
|||||||||||
| Everyone talks about the advances we can expect in future vehicles and powertrains, and the shorter development times, but the real crux of the matter is bringing all the simulation and analysis together with the design software – and getting it really to work. Without that, it’s hard for cars to be radically different to what they are today. The current generation of engineers has this preconception that they design the part before passing it to some expert who does the specialist analysis: looking at the fatigue, thermodynamics, manufacturability, crashworthiness, the list is long. The model is pulled around by different departments who then kick it back to the designer for changes and the process has to start again. All the while, the manufacturing people are waiting to start work on the jigs, tools and fixtures. Getting the design engineer, like it or not, to do more of the analysis work at the beginning in the CAD stage is pretty damn important. Brad Heers, in charge of Dassault Systèmes’ Simulia software, says: “Many software developers have tried to make simulation more accessible to designers. Some have succeeded, although not even close to what is possible. Simulation is still often an afterthought in the minds of the CAD user.
The vast majority of simulation that makes a difference to the design is still performed by experts.” Why is that? Heers has a couple of theories. One is that to use the tools requires an understanding of things like nodes, elements, mesh densities and other abstracts. These are difficult for the average designer who may only want to know whether their design is stiff or strong enough. “There is also a false notion that non-experts don’t need powerful underlying simulation technology,” says Heers. “The opposite is true. Tools for non-experts have to be just as powerful since they have less tolerance for working around issues where the technology is not robust enough. Very stable programs mean more of the settings and decisions can be hidden from the non-expert while still providing reliable results.” And Heers reckons that, because many interfaces are not a natural extension of the CAD environment, it also makes it harder for them to catch on with designers. There’s a strong argument for firms to do all their engineering in one software environment, something that the big players like Dassault Systèmes, the producer of Catia, are obviously keen on. Its United FEA approach uses a single model for noise and vibration, durability and crashworthiness. “Often these three groups use different simulation tools, and the resulting duplication of effort can be costly,” says Heers.
Since taking over Fluent and most recently Ansoft, Ansys’ simulation portfolio has grown to cover structural, fluids, thermal and electronic design. The firm is developing simulation that allows you to view finite element analysis (FEA) and computational fluid dynamics (CFD) simulation results on the same geometry. “We believe it will be increasingly important to do real multi-physics analysis of engineering designs,” says Ansys automotive manager Werner Seibert. “That’s the way the real world is. It will reduce the stack-up effect of errors, handed from one analysis domain to another.”
Seibert says this is essential if companies are serious about reducing the number of physical prototypes. “Often the simulation process chain comprises inadequately linked steps of different pieces of software,” he says. “The interfaces are still operational, but they are in fact slowing down the overall efficiency of the engineering process.”
That’s something that Maplesoft, a relative newcomer to the car industry, is also addressing. The firm’s approach – which Toyota is introducing – combines technical calculations, advanced mathematical analysis and design documentation. It recently launched Maplesim, which allows you to connect standard physical components such as gears, electric motors, joints, springs and dampers to create a system-level simulation model quicker and more reliably. “Errors are detected early and potential problem areas are identified automatically,” says Maplesoft’s R&D chief Laurent Bernardin. “Models need to be based on the underlying physical laws to guarantee accuracy, but also need sophisticated simplification techniques to make these models practical – in terms of simulation time and testability.”
Some engineers argue that using the tool that gets the best results is still more important. The magic formula, says Fraunhofer Institute’s simulation engineering expert Klaus Wolf, is “code-coupling”. The institute’s MpCCI open-interface allows engineers to combine FEA structural codes with CFD codes to look at issues like thermal coupling and cooling for engines and exhausts.
“Instead of mapping median thermal values from a CFD simulation as loads onto a structural stress analysis, the usual one-way, one-step coupling approach, MpCCI allows a two-way, n-step iterative coupling between two running codes,” says Wolf. “The results are much more accurate and realistic, especially if you’re expecting local effects in the models.” But a multi-physics approach to the simulation means a multi-disciplinary approach to the engineering. It’s not enough just to have tools that the designer can use, the different design and analysis departments need to cooperate and be aware of the complexity of the other side. “The first step is to start strong interactive working between a company’s departments,” says Wolf. “As long as structural people see CFD effects as just a static load or the CFD guys only see the structure as a wall around their flow field, multidisciplinary working won’t get good results.” True, but with more development often dispersed across several time zones, it is not always going to be possible. PLM software, rather than empathy, might be a better way of holding the mess together. Siemens PLM is working on ways to provide tighter integration between CAD and CAE with the aim of making it easier for the design engineer to perform first-pass analyses before handing the design over to a simulation specialist for the detailed work. “Knowledge and application of the full capabilities of software is a hurdle for any package,” says Eric Sterling, the firm’s marketing chief. “We solve this by designing the interface to suit the different roles.”
Sterling sees software modularity and the drive to service-oriented architecture as key parts of CAE’s evolution in the next five years. “The web will also help increase collaboration along the supply chain, using things like wikis,” he says. “The effect for the car industry will be better design decisions.” It seems that we’re approaching a point where the CAE is advancing more quickly than we can learn to exploit it fully. In the short term, the degree to which different packages can interact will determine how closely different engineering disciplines work together. But, ultimately, we’re going to need a new generation of engineers who not only have expertise in particular product areas, but who also have a broader grasp of CAE tools. |
Read full interviews with leading CAE experts: Ansys: multi-physics analysis pays off
|
||||||||||
