It’s been some time I am talking about what I think of automation and “push-a-button” solutions to my newsletter subscribers.
I know that this is a potentially “sensitive” topic for some people in the FEA field…and many people working in the FEA software industry would prefer you not to know about what I am about to share…
But that’s the position I took! I swore I would tell you all the truth about FEA and it includes of course the traps of the FEA software industry you shouldn’t fall in…
Automation and “Push-a-button” are one of those traps…
This time, I decided to record the audio as well :)
(I tried to speak slowly and articulate… but that doesn’t change the fact that I am French… ;-) )
[soundcloud url=”https://api.soundcloud.com/tracks/324215056″ params=”auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false&visual=true” width=”100%” height=”450″ iframe=”true” /]
It is tempting to want automation for your FEA process
It’s always a very tempting to automate something and I admit myself that I like very much automation !
When I have some repetitive tasks that I have to do, if there’s a way to automate those tasks then of course I’d like to find this way in order to do it
The problem with automation is that it usually has to be ultra-specific to what you do in order to work properly
Every task, every process, and especially in simulation, is very specific to the type of model that you have, the type of industry, the type of…
A lot of things are involved in your project and that’s why it’s something complex !
But when you do the same thing over and over for years then you might find a way to automate that by reducing some of the parts that take time in order to create some kind of automated workflow that will help you to solve faster your process
This is a very good thing and I absolutely cannot criticize that
So why am I talking about automation today ?
There are different kind of Automation in FEA
Well… the automation I’m talking about is not exactly the same automation that an engineer may think about when he thinks about automation…
There is good automation and there is bad automation you can say… (and also a mix of both ;-) )
Let me explain what I mean by that!
The word « Automation » may be used as well by other industries such as marketing in order to sell a software product so let me explain what I mean by that
FEA Software are complex tools for engineers with many functions… that can be overwhelming!
The problem with engineering pieces of software is that they can do a lot of things…
They have a lot of features and they are ultimately very complex to use
That’s why it’s not so easy to be an engineer because you have to to know a lot of things and you have to know the tools that correspond to what you want to do inside the software in order to solve the problem
Now… for those who just start to use such kind of software for the first time, this is very overwhelming !
When, for the first time, you see this software with all those functions everywhere and you don’t know where to look at, it’s actually natural to be overwhelmed!
So you start doing a few tutorials, but you soon realize that actually it’s much more complex than you thought and that it requires a lot of repetitive learning, doing a lot of tutorials, a lot of focused efforts in order to get something that works and to understand the real thing
The things is that…
This is also a big problem for people who actually sell the software (Marketers), because they would like to have much more users
And they would like also of course to sell much more licenses of their software!
How marketers “solve” the FEA software complexity with automation…but do they really?
One of the problem for them is that the software is very complex, so it’s an usual barrier for the people to say « oh I have no one in my team who knows how to use this software so I cannot buy it »
So one of the things they have found in order to make those kind of people purchase the software is to tell them « don’t worry everything is automated, you don’t really need to know how to use all the functions, most of the functions can be performed just by pushing a button »
How is such push-a-button automation done?
Well… they basically develop features that automate a process which is problematic for the user.
The problem with that is that the complexity is only hidden behind the automated algorithm and many of the decisions that are critical to the accuracy of your simulation are reduced to a simple choice that the algorithm is doing for you inside a blackbox.
It’s convenient for them, because you don’t get to see HOW the algorithm operates and it may sometimes do some wrong operations that you wouldn’t even be aware off.
The trap of the hidden complexity
My point is that there is 2 types of complexity. There is the complexity due to engineering and the multitude of choices that you have to take… and there is the “hidden” complexity that you don’t see which makes everything look “easy”.
After all, if everything was so easy, why does the algorithm often doesn’t work, get errors or gets it wrong?
My take on that is that if at least you have the options inside your software to take manually the commands, then you do have a certain control over your model and results.
The real problem is when you don’t have those options… and the automated push-a-button solution does everything behind your back… and fails, sometimes without even telling you something got wrong…
Do you see what I mean now?
Marketers use this “automation” principle as some kind of pretext to make its look more easy than it actually is…
But what they are actually doing is just hiding the complexity away from you.
The problem is that you shouldn’t fall into the trap to think that this is the truth!
Push-a-button automation may not be suitable for your specific needs
I mean, it look more easy because you can theoretically do something very complex just by pushing a button
But…
When you get down to the real task and project, a lot of those automated functions may not be suitable for your project!
Because imagine… if you have an automated function that is supposed to work for every project well as I discussed at the beginning this kind of function is not ultra-specific to your specific project or industry, it has to be very general to suit almost every type of projects
Which means that if you have a very special model… there are some chances that this automation won’t be suitable
The problem is that you have been sold already the fact that you can solve every model using this automation just by pushing a button
So that’s why a lot of people don’t realize that this was basically a « mild » lie
The workflow of an FEA beginner who has been sold the push-a-button strategy may look exactly like that…
They just import a model… click
Mesh everything…click
Applied boundary conditions and materials… a few more clicks
Then, they get the results and get something that looks correct and a lot of data
So what’s the problem with that?
The problem of data accuracy
The problem with that is that those data may be totally wrong…
And when you have no way to compare those data with real testing data or if you have no experience in the kind of data you actually want to get from this kind of software…
Then it becomes very difficult to actually tell if those data are correct or not!
So when you have no time, you’re in the project and someone ask you « Hey! Can you do a simulation? »
Then, you know… it’s tempting to say: « oh yeah I can do it »
And then in a few steps you get something done, you click, you get some colors and then you do a report…
That can be done very quickly!
…but you know, this kind of behavior is not the behavior of someone who is responsible!
Related article: The dangers of bad accuracy in FEA
Simulation is ultimately about engineering!
It’s about understanding what is REALLY happening behind the physics that is involved into what you want to simulate!
So… don’t let the FEA software company fool you by telling you that they have everything automated for you
You should understand what is behind those nice words and really get down to the real engineering thing in order to understand and solve correctly and accurately your engineering problem!
YES, engineers need to know some basics about Engineering mechanics, Strength of Materials, Design & material science, and of course FEA simulation…
And if someone tells you that you don’t need all that… don’t believe him…at best this guy is not an engineer (not everyone has this change ;-) )…at worst, he is trying to convince you to buy his software…
Related article: 4 essential steps to solve FEA problems like a pro
——
Ok, that’s all for today! That was a hell of an article ;-) and I sincerely hope that you enjoyed it!
Let me know if it helps you in ANY way, it always makes me happy to know that I am helping engineers to become better and better in FEA simulation.
The world needs REAL engineers, that’s why I am doing all that.
If you like that article, you can also help me by sharing it with your team or with your network on Linkedin, a sincere THANK YOU for helping me to spread good practices of engineering!
Angus Ramsay says
Well done Cyprien. I recently went on the Ansys Aim workshop and was shocked to see design engineers being sold mutil-physics software!!!
Cyprien says
Thank you Angus, yes…exactly!
Muthuzz says
Oh…It is Very Nice Article….Now i clearly knows about the things behinds FEA softwares…
Thanks Cyprien
George says
Excelent. It is a very interesting oppinion.
Murali says
I am in full agreement with you. Complete automation may not possible in FEA simulations. possible to automate some few areas in the entire cycle. In pre-processing stage, meshing can be automated to some extent, same case with boundary conditions. Solution is already automated with batch mode commands, but one have to struggle manually, if there are any solver errors. Post-processing can be done, if one can build the right story board to output the required results.
I have done full automation projects in turbo machinery applications in late 90’s for Ansys and NISA solvers. Few to name, water lifting pump for hydro power plants, LP Turbine balding. But with the current complex CAD models, automation may not possible. It’s possible to do only A to B comparison with so many assumptions. We can not match the results with testing.
Cyprien says
Thank for your message Murali! I am glad that you share my view about automation. I think it isn’t a subject that everyone understands and it sometimes bring confusion into projects that require more experience to be taken care of. You seem to have a great experience automating FEA projects, I would be glad to hear more about the solutions you developed in the 90s. Feel free to send me a direct email ;-)
stephane says
I just get into your youtube channel and your website, i teach in an engineering school, and i want to involve i and my student in FEA. We carry out science projets with salome meca. This is the kind of article i’m going to share with same because i fight to make them understand FEA is not a push button tool.
Thanks a lot.
Stephane.
Cyprien says
Merci beaucoup Stephane pour le commentaire! J’espère que mes videos sur Youtube vont les aider ;-)
Mikhail says
Everything is very correct. At the ANSYS workshop, the marketer always talks about the magic button that is in the newer version. But I use my own button, I called this button “DO” and in each task I write macros for this button in the APDL language (APDL is a built-in language similar to FORTRAN in ANSYS )
Mirek says
Cyprien, great article, I completely agree on FEM automation :) I was using a couple of FEM software in my carrier like radios, Abaqus, lsdyna… and recently it came to me to use Ansys workbench. My first thought was “what a nice tool”, but after I learned the mechanism philosophy that stays behind it, I say it’s just a piece of crap. Ansys and fluent in the current form are good generators of colorful pictures. It’s scary but Ansys people on workshops saying that program control is always better, that it always will pick the right path during the solution. They are just selling their products, they turned off thinking in many of ansys users :(… and what I must say ansys bought lsdyna because ansys got one of the best marketers, the lstc has never had. It would have been better for the simulation world if it is the other way around…
Cyprien says
Can’t agree more with you Mirek! Automating is good when you know and understand what you are doing… but Marketers always make you think that this is the miracle solution to get anything done just by clicking on a button.
taju says
I can fully agree with this. Good example is when Ansys changed their marketing strategy. Some years ago they started selling the software to non expert managers, not FEA engineers.
Saniket Agarkar says
Yes,It’s really helpful.
Bo Feng says
You corrected my concept of FEA, thank you!