A couple of weeks ago, I had the pleasure of conversing with Desktop Engineering magazine’s senior editor Kenneth Wong for a podcast recording. He had a simple challenge for me: For a structural engineer who is just beginning to work with fluid dynamics, outline the points important to CFD flow simulation. Additionally, he asked me to explain how to avoid pitfalls when setting up the simulation and what to look for when analyzing the results.
My first thought was that, well, there are great classes, training and free YouTube videos available. Give me a couple of hours and I can turn a structural-expert-but-CFD-newbie into a CFD user. Kenneth understood all this, but his biggest challenge was yet to come. He asked me quite seriously, “And can you get an engineer on the right track in a couple of minutes?”
*** Mission Impossible soundtrack playing inside my head **** Sure! Let’s do it!
Part 1 of the mission: Select an example that is relevant to a structural engineer. We decided to discuss a valve design and the pressure that is exerted by the flow, making sure that the pressure is not high enough to cause structural damage.
Part 2 of the mission: Identify tips and tricks to help engineers avoid typical pitfalls. We decided to focus first on the simulation setup. (Analysis results will be discussed in a future podcast.)
1) Structural engineers have experience studying the structure of the valve. To look at fluid behavior, they need to look at “negative” space of the device — the flow inside the valve and pipe system.
2) Mesh requirements for a fluid simulation are very different than that for a structural simulation. Let’s make sure we tell the CFD newbie how to correctly resolve key aspects of the flow features — near-wall regions (where the flow velocity drops to zero) and regions where the flow curls and swirls (near the valve).
3) Since 99 percent of industrial flows are turbulent, you must understand how to best set up the turbulence model. As there is no single model that will work for all industrial flow, ANSYS software provides many models. Online forums, classes and simulation software providers like ANSYS provide tips and best practices to help engineers master this hurdle.
4) Set up boundary conditions. In this case, there is no need to simulate each meter of piping, so we focus on the valve itself. For this part, the boundary conditions are key: If they are not set properly, the GIGO (garbage in, garbage out) principle kicks in. So we need to be very particular about how to set up inflow and outflow boundary conditions. On the thermal side, we show that special models exists to treat conduction of heat through the pipe and valve material without actually having to simulate the actual process using a structural simulation.
You can listen to the full podcast discussing the simulation setup and let me know if you have experienced any challenges in doing this type of simulation.
And if you want even more information about CFD, check out some free online classes hosted on YouTube.
You can also see how an entire workflow can be executed here in this video.
Have happy and successful simulations!
Gilles
Very nice illustration Gills!
Thank you very Hossam!
Great job Gilles…
I think a lot of mechanical engineers know they should be getting into multiphysics but are not quite sure how to set it up and worried about doing it wrong. They end up just sticking to what they they know and never really simulate reality as well as they could. The same could be said for CFD engineers, Emag engineers or anyone who is comfortable with their expertise in one area, but concerned about getting the wrong answer if they branch out.
I think it is important that you didn’t just tell them the main areas to think about, but you reminded them that they can access help from a variety of sources including directly from ANSYS… Once they get a little help with their class of problems, they will be able to move forward with multiphysics and true virtual protoyping…
Next step… Parametric studies, design exploration, optimization and more!
As usual Simon, you are 100% correct.
Dear Gilles,
Nice post. yet here is another challenge! I want to setup a thermal analysis for a continuous production process (joining two solid composites by local heating). The analysis is completely heat transfer nothing else. However, I do want to transport the mass along with its heat content while keeping the mesh fixed (Eulerian mesh). I used the mass transport option in thermal shells but i have a problem. My velocity is restricted due to “peclet number” and i cannot model the actual process velocity. I have to make my mesh size extremely small (millions and millions of elements) to attain this velocity and keep the peclet number less than 1. My computer gets hanged in the meshing. Can you give me a quick tip how to resolve this issue.
I already asked the local distributor in Netherlands (infinite) but they have no solution.
Kind regards
Muhammad Niazi
Dear Muhammad,
Excellent and very challenging question! I do not have an answer for the moment and I am talking with our Technical Services team to find the best solution. To avoid exchanging ideas and suggestions using the blog comments, I already followed up with you directly via e-mail. Once we have the final answer, we will post-it on the blog.
Best regards,
Gilles
Dear Muhammad,
I shared more information with you but I wanted to update everyone
A colleague at ANSYS sent excellent information I wanted to share. He points out that more information is needed (and Technical Support is working on it directly with you), but even so his answer contains great information.
“I think, not knowing the specific application, I would be hesitant to prescribe a solution and get this dialog misdirected. In my mind, I could be wrong, but there are several applications to consider:
a) Composite lay up and curing
b) Thermal bonding of polymer substrates
c) Thermal bonding in non-wovens
d) Multi-layer tissue papers
ANSYS CFD can analyze all of these applications (and up and down stream of them) in great details. But, it is difficult to ascertain if any of these apply to your situation. Furthermore, if silicone adhesive is involved, you might have additional considerations about chemical bonds. In any case, we will make ourselves available to help you out whichever situation you want to analyze. Given the Peclet number restriction, it appears that focus is on local details. There may be a way to identify different scales of details in some other way than using multi-million cells and small timesteps. Both major CFD tools from ANSYS have capabilities to address your basic need. But, talking in terms of application will improve the speed of resolution to your needs. If you want you can contact us directly. Otherwise, describe the specific situation with a bit more specifics. For example, you seem to have a modeling strategy with fine mesh and fine time step and you seem to have been stuck in meshing. Not knowing the geometry details, suggesting any resolution is difficult. On the other hand, as I mentioned earlier, an alternate modeling strategy may result in a better solution approach.
“
i like Video. thanks you
You are welcome. I ma happy that you liked the video.
Wow. Great job. It would be great if instead of webinars ansys also released a series of small podcasts like this one on some basic topics targeted towards first time users of some models or cases.
Pingback: ANSYS' Top 5 Engineering Technology News Articles - Week of Feb. 25th
Dear Gilles,
Thank you for your great video. This video shows that ANSYS has the capality of doing parametric studies. I alreay have a parametric design table of a fan made by solidworks. Once I import the geometry to Workbench, those geometric paramters won’t exist anymore and its gonna be a fixed geometry not a parametric one. My challenge is to find a way to import the solid work design table to workbench still as parameteric geometry. I really appreciate it if you let me know whether or not Ansys has this capability as well…
Thank you,
Sahar