Beth, the cost factor is really the main issue. I remember, in the third quarter of the last century of the last millenium, buying the large number crunching computers and analysis software for a spacecraft plant. Of course, simulation of all kinds was done early and often. Money was not much of an issue. On the other hand, the software cost as much as the hardware (almost exactly). And the hardware was much more expensive then.
Making it easier to use is also a big advantage. Even so, the software does not solve the problem for you. It makes it easier to explore more design options and to avoid more prototypes. I have a neighbor, a PhD in Mechanical Engineerin, who was published using one of the major CAE tools. When I asked him about his experience is was somewhat less than enthuastic about it. He did see value in the tool, but the tool does not solve your problem. It still takes time and effort. The tools are just that. They are far superior than not using a tool, but there are still groups that I run into that use their own software for their own model. I see that as a long term risk, but they still do it.
I am intrigued by the cloud model for this. Adapting the algorithms to a cloud environment is a lot of work, but of course you are selling it to a large number of customers. One outcome of that is that the techniques are being used for products like bicycles. In the past, you might have thought that overkill. Not anymore.
Thanks for your input, Naperlou. It is amazing how the cost of the hardware, and hence the software, has come down in price so significantly essentially putting this kind of capability in the hands of the smaller companies like specialty bike manufacturers or sporting goods makers. I completely agree with your comments that the tool can't solve the problem. I think we made it very clear in the piece that you have to understand the engineering problem and the science in order to put these tools to use and get accurate results. Definitely a learned skill set, no doubt.
Great article, Beth. Putting CFD and FEA in the hands of non-specialists is a worthy goal. When I took a couple of classes in FEA in 1976, seasoned engineers would ask me, "What's that?" By now -- 36 years later -- it seems to me we should have been able to find ways to make it accessible to people who have engineering backgrounds, but don't have FEA expertise. Same goes for CFD.
Enjoyed your exploration of the subject Beth. I too see a growing desire among ANSYS customers to drive simulation further up stream in the design process and make simulation tools more accessible to non-specialists. Making the tools easier to use is something that many of us in the industry are focused on, but as Keith points out, it doesn't address all of the challenges. I see a number of best practices emerging in our customer base that both small and large companies can benefit from:
Documenting simulation best practices for specific points in the design process (and for specific products)
Automating simulation workflow – either with off the shelf tools, or through customization to include these best practices
Instituting internal certification and training programs specific to these best practices
Capturing and reusing design knowledge from past designs
@Gregfallon: Thanks so much for those great suggestions. All really good stuff. It definitely puts into perspective that even if the tools are becoming easier to manuveur, this is still hard stuff and new policies, training, and workflows are essential for managing and coordinating simulation initiatives as they move from isolated one-off studies to a more concerted, enterprise effort.
I recently heard a story from a test Engineer. When he failed a design in the lab, the design Engineers told him that his test setup was wrong. He challenged them on that, and they responded, "It has to be wrong, the design passed simulation."
Simulation is a great tool, but its users must understand that it is an approximation. The real world is far more complex, and often takes a design outside what the simulation can handle. The best the simulation can do is detect some of the flaws early. It will never replace thorough testing. It is just another way of improving the odds that the design will pass.
Finally, let's be clear on this "cloud computing" thing. "The Cloud" has been used as a catch phrase to cover a wide variety of Internet enabled processes. In this case, they are talking about a company providing high-performance computing services. It is really no different than what was done 20 years ago, except that a clever web-based interface means that you can submit your design directly to the computer through the Internet, instead of sending a file to a "Customer Support Engineer" who would translate it and run it through the computer when your reserved slot became available.
A convenient step forward, no doubt. But hardly the radical paradigm shift that it is being hyped as.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.