Sie sind auf Seite 1von 5

www.nafems.

org

Education & Training

The Expanding Use of Virtual Pr ototyping to Develop Superior Pr oducts


David Reid, of Blue Ridge Numerics, shares some of his views on the controversial subject of the use of simulation software by users other than specialist analysts.

i mulation software is an integral part of the product development process. Over the past decade it has become increasingly vital to many companies.

Relying on their traditional products has not been kind to their balance sheet. Whos going to buy expensive film if you can do it all digitally, virtually for free? Skoda reversal of reputation On the flip side we see examples of companies who have overcome terrible reputations to become leading lights in their field. Skoda have turned themselves around from being the butt of every schoolboy joke to a serious contender that produces cars that not only have a reputation for reliability but also for making brave, bold & effective styling statements. Of course much of the credit goes to fact that they were acquired by VW, and have since capitalised on shared platforms and components. But their ability to deliver top quality products at impressively low prices is quite amazing.

So what are the factors that cause companies to choose to use virtual prototypes over the traditional build and test method? How have some companies managed to successfully implement these tools, showing significant competitive advantages, while others shun their use or struggle to implement them?

Commercial Pressures & Competitive Advantages


A companys reputation and success is based on the quality and performance of their products. Certainly many companies manage to eke out an existence by relying on the momentum created in bygone days, but consumers and customers are fickle and are unwilling to put up with a sub-standard product even if it carries an attractive brand label. Polaroid loss of massive market share One classic example of such complacency is Polaroid. At its height, Polaroid totally dominated the market for home cameras. The concept of instant photographs was very compelling for many consumers. Their business model, of providing cheap cameras and making the profits on the film, was extremely profitable and has been successfully imitated by others companies. Today Polaroid are a shell of their former selves. This is certainly not down to their failure to invest in research and development, but merely their failure to bring any significantly valuable products to market. They even have a museum-type room, where they store all the ideas that never became products. The blame can be laid squarely at the lack of confidence that they had in their concepts. They spent much longer iterating concepts than their competitors, who then stripped their market with the introduction of digital cameras. Their R&D in digital technology was actually very advanced at a similar level to their Japanese competitors, but they failed to deliver any products to market.

Performance is key
Think of the reasons why you choose one product over another. Think of the last few products that you bought. Why did you buy them and not others? Almost always the answer comes back to performance. This performance needs to be measurable and the quickest and often best way to do this is using simulation software. This has obvious benefits over traditional prototyping and testing methods. At conceptual stages it is almost impossible to build and test every idea. Often the development at this key stage is driven by guesswork and experience. The hallmark of really effective innovation is the quantum leap. But these jumps are often achieved using new, unorthodox, unproven, even risky, ideas. Such ideas are often vetoed unless the originator can demonstrate the value of the idea. Innovation by people not software I have often heard people say in the past that This software will make you innovative. This simply isnt true. People are innovative but their innovation is often stifled by Doubting Thomass who want hard evidence and do not trust the intuition of the originator. Page 7

July 2004

Education & Training


Software validates and verifies If that person with a good idea has a way of representing the idea and proving the performance and value of that idea it is far more likely to be implemented. Once the concept is proven it becomes far less risky and seems more orthodox and acceptable. Whats more, the idea often requires iteration and development to tune a rough concept into an optimised product.

www.nafems.org

Success Factors and Limitations


Simulation software is not new though, so why has it taken so long to catch on? Finite element technology is a prominent example. It has been harnessed to successfully tackle structural, thermal and fluid problems and yet many people are unnecessarily scared of it. It is often viewed as a black art that is cloaked in complicated and fragile mathematical hype. Often it is regarded as the domain of highly trained, skilled and educated analysts (typically holding a Ph.D.) who seem to go out of their way to baffle listeners with complex jargon. This results in competent and ambitious engineers shying away from such development tools. This is a shame, since these are precisely the innovative type of people who would benefit most from being able to evaluate, prove and develop their ideas. While many companies have invested in CAE technology, they often make the mistake of housing it all in a separate analysis department and hiring specialised full-time analysts to drive the software. By separating this function from the core design department they are creating a communication barrier and a detrimental sequential process. The analysts tend to regard the design engineers as their intellectual inferiors; mere CAD jockeys, who bring them awful geometry that nearly always needs to be rebuilt in their CAE system. The designers tend to regard the analysts as geeky, slowworking boffins who get bogged down in irrelevant minutiae. They have a tendency to discover problems without presenting any solutions. They are a bottleneck in the process that have to, unfortunately, be included at least once in the attempt to develop a product. The net effect of this environment is that the analysts are left out of the design process as much as possible. Problems get discovered late in the design process when much of the detailed work has already been done. Change becomes painful and the minimum number of iterations is used to get the product to perform at a level that is acceptable. Hardly a recipe for success, where change is discouraged and penalised. Innovation is closely linked to the culture in a company, and can only thrive where people are encouraged to take ownership and explore their ideas. Much has been changing in the CAE arena over the past decade and the willingness by the market to accept new Page 8

products is far higher now than ever before. The emergence of CAE products for the design engineer is now becoming big business, whereas ten years ago it was regarded by many as unfeasible, unrealistic and even dangerous. Knowing how difficult it was to use the existing products, analysts and vendors of such products discouraged the spread of CAE technology to a wider audience. Their fear was based on the dangers of giving such powerful tools to users who lacked the necessary training, skills and experience. So what has changed to make it more accessible? Significant advances have been made in the realms of: Intuitive Graphical User Interfaces Robust underlying technology Automation of tedious manual processes Integration with other software tools Consideration and empathy for the target user Recognition of sensible limits Intuitive graphical user interface design When observing software that is widespread and frequently used, it is apparent that the user does not have to search around to perform the common actions. Users should be able to proceed down their chosen avenue with minimum effort and frustration. Less frequent functions should be available but only if asked for. Importantly it must be recognised that different users have different requirements. Experienced users do not want or need help. Less experienced users need appropriate guidance to take them through an unfamiliar process. While it may seem that these two requirements oppose each other it is possible to design user interfaces that fulfil both by using: Icons with intuitive meaning that are presented in a logical fashion. Menus that drop down to show the most common or relevant choices with the less common choices available at a deeper level. Experienced users can work quickly without being blocked by cumbersome procedures. New users can be lead by means of hints or wizards. Wise use of default settings allows most users to ignore standard and common actions, but unorthodox cases are catered for, by exposing the other choices for selection.

July 2004

www.nafems.org

Education & Training


Figure 1: Structural Load Through a Hole

Robust underlying technology Unfortunately there are many simulation software products that contain technology that is unsuitable for the majority of users. These products may contain complex algorithms, but are highly dependant on the skill and experience of the user. Because of this they are only usable by an elite few. If an inexperienced user generates a result without taking due care and attention, it is easy for them to get answers which are inaccurate. For this reason many people have had their fingers burnt by previous encounters with simulation software. They have either: Found it difficult to get to grips with, or Blindly trusted a result and made an important decision on that figure, only to discover that the accuracy of their simulation was compromised. This is a shame since several vendors have been striving to provide products that combine accuracy with accessibility. To do this, the latest and best technology must be used to ensure that the software automatically seeks out the best answer and does a good job of proving to the user that the results are reliable.

user to specify a complex varying load that varied as the function of a local coordinate system. Nowadays users simply click on the hole and specify a load vector, which figures out the correct distribution and applies it automatically, giving the user graphical feedback so that it can be verified as correct. Another more sophisticated procedure is seen in the realm of CFD. Often the only piece of geometry required is the volume of fluid contained by the design. The user is free to assign boundary conditions such as flow rate, pressure and temperature. But what about the surfaces of the model that will act as walls? In the past it was necessary for the user to specify a particular boundary condition (velocity = 0) to correctly simulate the effects of these walls. While this may be perfectly obvious to a CFD specialist, it is the kind of task that seems unnecessary to a new or infrequent user. Why do I have to put a boundary condition on every wall? Why cant it just assume that no BC implies a wall? Figure 2 shows a half model of the fluid inside a poppet valve assembly. The inlet (purple) has an assigned normal velocity and the outlet (yellow) is set to ambient pressure. The symmetry plane (blue) is used to reduce the computation. The remaining 10 surfaces (red) are therefore wall surfaces. Auto-wall functionality now assumes that any surface without a flow related BC is a wall and therefore drastically reduces the tedious setup work required. In this simple example it wouldnt have been too difficult, but models with thousands of surfaces would be very laborious to deal with. The velocity gradient moving away from the wall into the flow can be quite sharp and it is necessary to be able to capture this effect accurately. Consensus shows that a good way to achieve this is to pack some prismatic elements adjacent to the wall, where the extrude direction of the elements is normal to the boundary. Again this is something that has previously been the domain of the user, often involving an

Figure 2: Fluid Inside a Poppet Valve Assembly

Several products make good efforts to provide evidence of the likely accuracy of the result. For such a function to be useful it needs to be readily available, easy to understand and open to interrogation by the user when more information is required. This ability to interrogate the quality of a result is vital. All users of simulation software should carry a healthy dose of scepticism that requires them to verify the validity of their results before showing them to anybody else. Automation of tedious manual processes In the early days of CAE, it was accepted that much of the users time would be taken up with the laborious generation of decks or combing through reams of printed results to extract point data. Much of this has changed and now allows for easier access to the software. A simple example is the common application of a structural load through a hole. Often this load needs to simulate the presence of a cylinder or pin through the hole. This means that the load needs to be applied normal to the surface of the hole and varying by a cosine function across 180 degrees of the hole. For many years it was up to the

Figure 3: Velocity Results of a Simple Mesh

July 2004

Page 9

Education & Training


Figure 4: Velocity Results Using Mesh Enhancement

www.nafems.org

element of judgement to specify how thick this layer should be and how many element thickness it should contain. Figure 3 shows the velocity results of a model that has a simple mesh. Using mesh enhancement causes a significant difference in the results shown in Figure 4. The maximum velocity increases by 19% and the velocity gradient near the wall is captured far more accurately. A magnified portion of the models is shown in Figures 5 and 6. This automatic mesh enhancement process pushes the mesh away from the wall and inserts some prismatic elements that help to accurately capture the velocity gradient. By setting this to happen as default with some sensible parameters it makes the whole process much easier, simpler and quicker. Obviously there needs to be a manual override that allows the user to turn off or adjust such a feature, but it eliminates a great deal of the judgement from the process. Integration with other software tools Engineers who use digital design tools are not looking for a single all powerful system that does everything, simply because it does not exist. For a long time CAE software has struggled with the need to incorporate complex, real world geometry. Some vendors have tried to provide tools to allow the user to build their own geometry, but these often fall short of being useful, since it is not their core competence. Other methods involve accepting common geometry transfer formats like IGES. These typically suffer from the digital equivalent of Chinese Whispers and often results in surfaces becoming split apart or loose tolerances creeping in and corrupting the geometry.

In the authors opinion, by far and away the best method to use is to build direct interfaces to common CAD systems or the kernels that they are built on. The pinnacle of such interface capabilities is to have a fully integrated system that not only allows geometry to be transferred from the CAD tool to the CAE tool, but for the CAE tool to close the loop and be able to communicate back to the CAD tool. So the smart strategy is not to try and do everything yourself. It is better to link to other software products that are widely used and accepted. For CAE this means linking to all the big CAD vendors, few of whom have focussed on the CAE market, despite the impressive growth that it has shown over the past few years. Does it suit the people who use it? Many full time analysts will spend a significant amount of their time creating precise meshes and grids for their analyses. This means that the pre-processing of their model still takes up a significant proportion of the overall process time. While this might be important to them, it is impractical to expect CAD users to emulate them. These users want to get in with a CAD model and get out with an answer, spending the minimum amount of time setting up their model and extracting a useful answer. The software should use language that is clear and terminology that is familiar to them rather than confusing and scaring them. The key test here is the repeatability of the system. If the same problem is given to two users who have access to the same simulation software, will they get the same answer? If those tools require the users to exercise significant amounts of judgement and require significant skill, experience and training then the answers can be wildly different. Who do you believe? Who is right? It is clear that one of them is wrong, so are either of them right?

Figure 5: Velocity Results of a Simple Mesh: Magnified Portion

Figure 6: Velocity Results Using mesh enhancement: Magnified Portion

Page 10

July 2004

www.nafems.org

Education & Training


But care must be taken when selecting a simulation software tool. It must fit the purpose it is intended for, the environment that it will be used in and, most importantly, the people who will use it.

Conclusion
The development of many products contains some mechanical engineering requirements. To improve and optimise the performance of a product as it is designed in a timely and cost effective manner it is essential to use the right tools. To simulate the performance of a product is often easier, quicker and cheaper than using a virtual prototype. The software market providing such tools has, in the last decade, gone through a revolution. Once it was dominated by expensive products that were difficult to use and connected poorly with other engineering software products. Since then we have seen the emergence of cheaper, more accessible tools that integrate well with incumbent CAD products and allow engineers to push the limits of their designs and explore the effectiveness of their ideas. Product designers can now evaluate their intuition and play with their models giving them a far greater realm of discovery and innovation. Simulation software products have become so accessible that anyone capable of making a 3D solid CAD model should be capable of running structural, thermal and fluid analyses.

Contact
David Reid Blue Ridge Numerics E david.reid@cfdesign.com

Editors Comment
This article represents one vendors view on this controversial subject. Articles and comments supporting different viewpoints would be welcomed by the editor for potential inclusion in future issues. E-mail your thoughts to david@nafems.org

Opportunities for Growth in Analysis with Cape


You may be the department head of an "in-house" analysis department, or the owner of a specialist simulation company. Cape is a major force in analysis consultancy in the UK, and looking to strengthen this position through partnerships in both technical and geographical areas. Our business is centred on structural analysis consultancy, and we are seeking to expand into other technical specialities (CFD, EMS, Multi-physics, impact, etc) as well as enhancing our presence in regional areas. Partnership could take any number of forms, and we are flexible as to how this could be formulated; however technology exchange and business growth would be a common theme. If you feel that a partnership with Cape would benefit your business then contact us (care of NAFEMS), with an outline of your business, to set up an informal meeting and exchange of ideas. Naturally everything will remain confidential.

Opportunities for Growth in Analysis, c/o NAFEMS


Whitworth Building Scottish Enterprise Technology Park East Kilbride UK G75 0QD

e-mail growth@nafems.org
(N.B. NAFEMS are providing a response service only, and have no direct invlovement or affiliation with the companies concerned with this activity)

July 2004

Page 11

Das könnte Ihnen auch gefallen