Pros and Cons of Simplified Simulation: A Panel Discussion

Panelists for the discussion on simplified simulation: (Left to right) Sunny Manivannan, VP of Business Development at Rescale; Dennis Nagy, the principle at BeyondCAE; and Jeff Waters, an independent CAE expert and industry watcher.

The conversation began months ago when I posted a motion for discussion in the LinkedIn group called “New Trends in CAE Simulation.” I asked, “Do you agree or disagree? It’s dangerous to simplify FEA and make it accessible for the masses.”

There’s clearly a push among computer-aided design (CAE) software vendors to package their simulation programs for a broader audience. These programs have historically been the domain of Ph.D.-level experts. Some critics see the move to reinvent them for designers and engineers as “dumbing down” a complex process. Others believe greater accessibility to simulation through simpler interfaces would lead to better designs.

in LinkedIn discussion forum, John Chawner, president and cofounder of Pointwise, noted, “Should we simplify it? Ah, here’s the magic. Does simplify mean ‘make it idiot proof?’ Then by all means, let’s go for it. Does simplify mean ‘cripple it so that it’s not useful for anything?’ Then forget it.”

Carlo Poloni, president of Esteco, chimed in, “I believe … the engineer who builds the simulation (FEA) model and the one who uses it might be not the same person and the type of access to information/execution can be kept different. Under this perspective, there is no need (and probably it would be a danger) to simplify too much the specific FEA tool …”

Joe Walsh, a veteran of the CAD community and founder and CEO of intrinSIM, added, “One key aspect is not to ‘simplify’ or ‘dumb down’ the simulation application but to ‘smarten’ them up to encapsulate expert knowledge and range of applicability. One good examples of this is the focused applet or vertical applications … with pre-configured scenarios.”

So I invited three participants of this online conversation — Dennis Nagy, the principle at BeyondCAE; Sunny Manivannan, VP of Business Development at Rescale; and Jeff Waters, an independent CAE expert and industry watcher — to discuss the same motion in an Oxford-style debate.

For a moment, the so-called debate seemed to be in jeopardy as all three panelists came out against the motion. But the agreement among them notwithstanding, Nagy, Manivannan, and Waters engaged in a lively conversation about the pros and cons of simplified simulation with different perspectives.

Manivannan distinguished user-experience improvements and simplifying the physics setups as two different types of simplification approaches. “There’s a lot of idiosyncrasies that are very tool-dependent,” he reflected. “For example, using ANSYS and using MSC Nastran are two completely different experiences. And the skill set required to run these models successfully under each code is very different … That’s another way to simplify — coming to a more standard process on how you set up physics.”

Nagy said, “It’ll be a long time before people who have no engineering training can do simulation, but even that will come, though at a much later stage. So the dividing line now is between simulation experts in a company and product development and design engineers. So making it accessible to that broader community is the key here … We have to figure out how to do it, and not say that it’s dangerous and try to stop it.”

Waters said, “If you take a look at SolidWorks Simulation or Autodesk simulation tools, they’re about as easy as you can make a general purpose tool … But that’s not going to be enough to spread into the general population.” Waters has a more ambitious definition of accessibility to the mass. He cited Twitter as his model for ease of use and mass appeal.

Waters also posed a question to the panel: “What kind of analysis should the masses be doing? Do we expect the masses to be handling … something like flow through a room? Or, when people say the masses shouldn’t be engaging in CAE, are they referring to high-end physics?”

Manivannan revealed he’d like to see even high-school students using simulation. “Get them started with simple simulation that helps explain what’s going on around them,” he said. Nagy added, “The issue of facilitating it — what that costs, and who’s going to pay for it — is one of the problems.”

Nagy observed, “Those of us from education community, end-user community, or vendor community will sometimes point the finger at the other, to say, ‘Well, it’s your responsibility [to simplify simulation] … It’s in everybody’s interest. You somehow have to get that cycle moving.”

I’d like to hear your thoughts too. Would you have supported the motion? Or spoke out against it? What, do you feel, is the right definition for “masses” qualified to use simulation? How easy should the software be? How do you put in safeguards to prevent unintentional errors in a simplified simulation software? Please share them as comments below here.

The full panel discussion is available in the podcast below:

If you enjoyed this post, make sure you subscribe to my RSS feed!

16 Responses to Pros and Cons of Simplified Simulation: A Panel Discussion

  • Kevin De Smet says:

    Yes – it’s dangerous. But even so, it should be democratized. Because if you don’t include simulation in your development process then you’re not really designing in 3D, you’re just drafting in 3D. The physics need to be considered, and should be analyzed. And that’s what simulation is all about.

    This is not new, vendors have been pushing simulation for the mainstream for well over a decade but not many mainstream companies actually use it, I think at least two reasons are: 1) The tools are still too difficult to use 2) The tools are still too expensive

  • I think the term “simplified simulations” needs an explanation or clarification. Which simualtuions could be called simplified? Some CAD engineers would term all simulations as complicated.
    Anyway, the way I understand the term, I think simplified simulations has its value in an initial phase of a development project where the CAD designer can use the build in simulations module in his CAD package to do initial simulations on a proposed solution. When the solutions appraches final design, the simplified simulations will not be sufficient any more, and a more detailed and thorough simulation is needed. This should be handled by the simulation expert with a tool that can include all the important details. The simplified analysis should be restricted to parts. Assemblies are ususally a lot more complicated with bearings, gears, contact surfaces and other important details (Not typically FE geometry). That should be left to the expert with an expert Software.

  • Paris Altidis says:

    Here are my thoughts from back in the early days of the simplified FE simulation:

    http://www.designnews.com/author.asp?section_id=1386&doc_id=223972

    Still hold true no matter how much more sophisticated and expanded the simulation tools in the CAD packages have become.
    Both Analysts and Design engineers should be Engineers First, Analysts Second…And another thing: Analysts and Design engineers are both the weakest link in this debate no matter how simplified or complex tools are being used.

    Nowadays, you can practice your structural engineering skills on your smartphone with schnittkraftmeister (http://www.ifb.tugraz.at/schnittkraftmeister/).
    Is this a way to gain CAE expertise ?? Absolutely not, but the fundamentals always need to be within reach and refreshed from time to time.

  • Kenneth says:

    Kevin, Bernt, Paris, thanks for sharing your thoughts!

    Kevin, quite true. Some large enterprises with sufficient budgets can handle the cost of the software, but the difficulty — the complexity of the software’s interface and learning curve — is a hurdle for most potential users.

    Bernt, you’re right. “Simplified simulation” could mean a software that’s easy to use because its interface is designed to cater to nonspecialists. It could also mean an simulation job that’s simplified because a lot of details about the design is still unknown in the early phase and many critical decisions have not yet been made. When we recorded the panel discussion, we were thinking of simplification in the former sense.

    Paris, thanks for sharing your article! I think the FEA experts will thank you for your suggesting that potential users should take them to lunch to get tips from them. :-)

  • Chuck Ritter says:

    I’ve been involved in CAE for 40 years this month, starting when service bureaus and card punches roamed the Earth. Things have changed just a bit ;). A few thoughts follow.

    You should not accept capabilities lists unquestioningly. Every potential user, whatever their skill set, should ask – Who will support me when I need help? The “Cadd-ins” for analysis simulation have pretty good feature sets. Can they support them like ANSYS, Simulia or Nastran? Will I need that support? How can I tell when I have a good enough answer?

    From Paris’ article – “The answer depends on three things: First, how much time design engineers are willing to spend in understanding the physics/failure mechanisms/weak links of their designs prior to embarking on creating FE models; Next, how honestly they recognize their own limits; and finally, how closely they work with the professional analysts within their companies.” Smaller companies may not have skill gradations and of necessity their engineers may be jacks-of-all-trades. They are often those most in need of analysis democratization yet with the least time to devote to figuring out what is worthwhile to do directly.

    I am not a fan of student engineers spending much time learning to use a specific program. That is time better spent on closed form basics. Sure, bash off a few Cadd-in FEAs to compare to closed form, but invest the bulk of the time in understanding the closed form procedures.

  • Dick Russell says:

    This is an interesting discussion that’s been going on – at least that I’m aware of – for over two decades! It’s been the “holy grail” for CAE vendors – accessing a potential user community much larger than the simulation specialists., and helping simulation better support/impact design decisions. There has definitely been incremental progress in “design oriented simulation”, and many runs have been taken at it from CAD integration, modeling automation tools, simplified roll-based user interfaces, lower priced entry packages, and even through expert systems. And yet the promise has yet to be fulfilled. Though the tools can still be improved, I belive that a large part of the solution need to revolve around engineering process. There is no silver bullet tool solution. I dug out an article I had written back in 1995 for the now defunct CAE Magazine in their regular Manager’s Guide feature titled “Making the Case for Simulation Driven Design” It’s amazing how little things have actually changed since then. Still lots of promise, but still big challanges.

  • Paris Altidis says:

    >> I am not a fan of student engineers spending much time learning to use a specific program. That is time better spent on closed form basics.

    To avoid being surrounded by professional mouse clickers, I’ve been telling my students in FEA classes about sanity checks and all: “If your hand calcs don’t pass the “SMELL” test, your FEA will most definitely “STINK” !!! One or two (closely supervised) humble pies later and the “stench’ went away..

    >> How can I tell when I have a good enough answer?

    True/False >>
    a) All FEA answers are correct ;
    b) All FEA answers are wrong ;
    c) Some are useful .

    All TRUE !!!

  • Kenneth says:

    I happen to agree with those who feel simulation users should first develop sound engineering knowledge and experience so they can intuitively spot simulation results that don’t look right. Sometimes I worry that increased reliance on software-driven simulation might cause a gradual erosion in these valuable intuitions, which can only come from observations of physical tests and real-world simulations.

  • Engineers do not go to school to learn how to use a simulation tool. They go to school to be able to apply mathematical principles to be harness nature to build better products for people. The simulation tool is an enabler, like a pencil and paper, calculator, spreadsheet, etc,.

    If you consider the job of an engineer, it is to make a decision that has a positive influence on the product and the companies direction. Being able to get to that decision point faster and with certainty (accuracy), and more often raises the engineer’s value to the company. If a software solution can improve the number of times an engineer can make a decision that makes a positive impact, then the tool is going in the right direction.

    A few years back, I was an Application Engineer for MSC Software where I would teach a lot of engineers classes in CAE. Before we would even fire up the computers and software, the first items I would tell the students is:

    Don’t get enamored in the tools. They are only enables of a solution. I’ve seen many engineers, new to simulation, build very overly-complex models for the sake of building a model. Don’t fall into the trap. Write the report before you open begin to build any analysis model. This will help you in determining why you are building the model and where you should focus your efforts (in terms of clean load paths, detailed stresses, etc,.). This will always save you time in the long run.

    In keeping with the above, if the simulation tool, can provide certainty, allow the engineer to arrive at decision with certainty quickly, and allow the engineer to communicate this decision to the rest of the product team, then by all means a simpler tools is a better tool.

    At the end of the day, engineering is a form of knowledge that is passed down from engineer to engineer. When I taught classes, it was never about the software solution (ie. how to push buttons in the tool), but more about how to utilize the tool to make a person a better engineer. Therefore, I would provide methods that would allow an engineer to efficiently engage with the tool to solve a problem in a smart way. Therefore, no matter how smart or simplified the software solution becomes, it will always come down to the engineer (end-user) to ensure that they make clear decisions on how to positively impact their products. If they can utilize a simulation solution to get there more efficiently… great!

  • Richard Williams says:

    First of all, I am no expert. I do not even rate myself as a good beginner. However, I have been taught that Embedded FEA, CFD and any other analysis add on’s and even stand alone ones, should be used only to help guide you in building a great prototype that will be ready for physical testing in the real world. A complete study of a real physical test will help shape the right prototype as a final product, that will satisfy the requirements you are designing for. Those that would accept whatever an FEA analysis would show as indisputable proof are really beginners in this field. I personally believe that someday we might be able to rely on analysis programs much more but whether that would ever be at 100% I do not think it will ever be reached. But I have seen a lot happen since I was born well before the TV age.

  • Once simple tools were created for 3D modelling, CNC machining, and printing, the rest of this discussion became quite moot. Could the chair you just designed withstand 300 pounds? What would the failure probably look like. Stress points? Would temperature and humidity change that outcome? If I put the nichrome filaments too close to the frame will the filaments sag so much as to short out my toaster or will air convection help? Will that combustion chamber withstand a 30 second burn with equal amounts of sulfur and kerosene as fuel, using nozzle #6 and a 400psi upstream pressure? The student or the engineer can either build a computer model or build the real thing. Ultimately he/she would like to do both. But remember all the model can do is save work. The initial conditions are ultimately quite immutable. The reality is tested. And the more you work with the tools, the more you understand them… but also the more the tools should understand you, the more real, easy, and extensive the tools should become. Every gamer knows the limits of his or her game physics. And ultimately, no matter what the current experience level of the user, and despite extreme caterwauling and impressive obfuscation by the high end sales people, the future is now, raw, exposed, learning, and growing. I loved it when Algor delivered to virtually every college a free copy of their software or when BobCad created a massive shift in 3D CNC work. But it has never been a question of price point or ease of use or secrecy or fear of bad models, it is always a question of which is cheaper, the model, or the real thing. I have built a house, over 25 sets of bookshelves in my life, and thousands of electrical circuits. A model is nice. And some models were created. But 5″ cracks in oak can suddenly appear out of nowhere even under light loads. And houses defy CAD, simple engineering, or computer modelling. For example, not putting braces between studs can lead to the most wonderful rat cavern you have ever seen. With CNC mills less than $5000 and 3D printers less than $1500, the clever FEA people better start asking whether they have a future… at all!

  • Rob Schorry says:

    Simulation without technical knowledge is like the blind man trying to figure out an elephant – nothing goes well.

    A simple model (cantlever beam or thin shell) can be useful, but not if the user does not undertstand small deflection theory, strength of materials or the limits of linear elastic theory.

    Education first, computer tools second! Yes?

  • Kenneth says:

    Tony, Richard, Edward, Rob: All very good points!

    The panelists spent a lot of time debating who “the masses” are and what “simplification” means. I think we came away with the notion that the masses we’re talking about are people with some basic engineering knowledge. Without that foundation (either from education or from job training), it’s nearly impossible to even know where to put the stresses and loads and how to interpret the pressure distributions in the results.

    Personally, I think simplified simulation software tools are quite safe to use so long as users understand that (1) they’re meant as design guidance, not as absolute answers to questions; and (2) their findings must be reinforced with physical tests in critical phases.

  • Doug Rush says:

    Wow, there is so much truly horrendous software in the world today! I don’t want to mention any names, but I will give you a little hint: PTC. Although I will concede that Mechanica is pretty slick when it works. I think in the last 20 years software in general may have actually gotten worse instead of better. Then again, some software was so bad it could only get better as it should have been scrapped before it was released.

    I believe you can have both simplicity and functionality. Start with simplified commands and give the user the option to dig deeper.

    The key to good software is a) intelligent interface design, b) quality feedback (a message that simply tells you a function failed is completely worthless), and c) a very well designed and implemented help function. In the case of FEA, the help function needs to be extensive. And absolutely NO internet based help!! NO NO NO! Or you can hire enough support people that a user will get instant help 95% of the time.

    Unfortunately I haven’t used Solidworks for some years, but the last time I used it their help was actually helpful. What a novel idea! That alone put Solidworks light years ahead of their so called “competition”.

    The concept for creating good software is really very simple: You put 10 idiots in front of a computer and tell them to use the software. Then you actually listen to and fix what these users complain about. In the case of FEA you probably want to add 5 expert users to this group.

    As you can see I have a very bad attitude regarding software vendors. Because the industry has earned it! For example we flushed a couple hundred thousand down the toilet with MSC. The sad part is that after 30+ years of experience it appears developers haven’t learned much (with a few exceptions). And in general software companies are on the same moral level as genocidal African governments.

  • Kenneth says:

    Doug: Describing software companies as morally the same as “genocidal African governments” is uncalled for. Frankly it doesn’t even make sense. I’m not aware of any software company that has advocated the eradication of an ethnic tribe or a certain race. But you’re entitled to your own opinion on that.

    I agree with your points (1) and (2) about the key to good software. But “no internet-based help”? Do you mean 24-7 phone line should be the only option for troubleshooting? I think it depends on the type of help one needs. With 3D modeling and simulation software, a 5-minute video clip can be infinitely more useful than talking on the phone because the operations required to fix the problem is not easily described in plain words. And I’ve noticed among the younger generation of users the preference to seek help online before they call–It’s just how they prefer to work.

    On your concept of creating good software, I don’t think it works if you observe how “10 idiots” use a piece of software. You have to observe how 10 reasonably intelligent users with varying level of expertise use the beta code.

    Thanks for sharing your thoughts!

  • Pingback: This Week in CFD | Another Fine Mesh

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>