Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

CFD

  • 18-05-2013 09:41AM
    #1
    like a ninja
    Registered Users, Registered Users 2 Posts: 23 ✭


    Hi everyone, I am thinking of doing a FYP involving using cfd to analyse wind effects on a crane structure. I am using pro engineer to model the crane and am not too sure what detail i should be putting into the model, eg motors as simple cylinders. I am trying to keep the memory (MB) it uses as lowest possible also if this helps.Part of the model has a lattice structure and i am not sure if this could be difficult to mesh in cfd.

    I have never using CFD (ANSYS CFX is used my college) but if anyone has any tips or tricks and advice in what the best way to approach it.

    Thanks


Comments

  • #2
    hypersonic
    Registered Users, Registered Users 2 Posts: 320 ✭✭hypersonic


    I like your style, that is a bruiser of a problem!
    if I was doing it I would stripe the geo right down, think circle, squares and triangles.
    meshing won't be a problem if the geo is simplified. the tricky part will be getting a good quality mesh with a low cell count.


  • #3
    5uspect
    Moderators, Arts Moderators Posts: 10,527 Mod ✭✭✭✭5uspect


    I don't think a simulation with the fidelity of individual bolts and motors is within the scope of an FYP, it's not even within the capabilities of CFD itself.

    Reynolds Averaged Navier Stokes (RANS) simulations are fairly cheap computationally but rely heavily on turbulence models which simply don't work well beyond a few proven cases. Large Eddie Simulation (LES) is a better approach, and it becoming more mainstream due to today's more powerful multicore CPUs. Direct Numerical Simulation (DNS) is the ultimate in CFD, but is limited to low Reynolds Numbers and supercomputers.

    The best approach is to start off simple and build up the complexity of the model. You're probably going to need to use finite element for the structure and finite volume for the flow. Ansys' Workbench makes this a lot easier nowadays.

    So start with a simple rigid structure, a tall pylon, and add a steady wind load to it. Look at and understand the physics. Then add the top of the crane, you you have an unsymmetrical T, but still a primitive geometry. Look at how wind direction affects the structure.

    Then change the flow to something more realistic, look up wind data, and figure out a periodic time dependent load representing the historic data. The main concern is probably rare but strong gusts.

    Then go to a coupled fluid-structure interaction. Here you'll probably need to couple your finite element for the structure and finite volume. Again the workbench in Ansys should make this easy enough.

    Nest I would make a small model of a section of the crane's complex lattice. Since it's pretty much repeated everywhere you can just model one section. See how it responds to loading compared to your previous box section and use the data from that study to update your previous studies. You could even try LES for this bit.

    You should really look at some of the Ansys training material before you start. CFD is a tricky beast to master and can give some pretty useless results if you're not careful.


  • #4
    fear_gaoithe
    Registered Users, Registered Users 2 Posts: 5 ✭fear_gaoithe


    Hi, I am preparing to study CFD next semester, however I need to get a grasp of some key concepts/prerequisites before I take the course such as the derivation of the Navier Stokes equations (they weren't covered in my undergrad engineering course...). Is anyone in the Dublin region around to give some grinds on Navier Stokes during the summer and/or early autumn? Email me here if you can help me out!


  • #5
    5uspect
    Moderators, Arts Moderators Posts: 10,527 Mod ✭✭✭✭5uspect


    I'd start here:
    http://www.core.org.cn/OcwWeb/Mechanical-Engineering/2-25Fall-2005/LectureNotes/index.htm


  • #6
    Darren1o1
    Registered Users, Registered Users 2 Posts: 744 ✭✭✭Darren1o1


    For the basics try an introduction to computational fluid dynamics the finite volume method, H. Versteeg, W. Malalasekera. It is pretty good. I would also be looking at Ansys. Prep my modelling and meshing your structure or similar. Meshing will be a huge part coming from someone who wrote his masters on CFD.


  • Advertisement
  • #7
    chris85
    Registered Users, Registered Users 2 Posts: 4,502 ✭✭✭chris85


    As mentioned before Versteeg is good, I have another book to recommend but cant remember on hand.

    As mentioned meshing will be crucial and many CFD problems are solved using incorrect meshes which leads to useless results. Also important to read up on solution methods in CFD such as the solvers used and which one to use for a particular problem and its implications for meshing.

    Where ever possible simplify the problem to reduce computational effort required. Also you need to have some sort of experimental data, either out there already or you do experiment yourself, in order to verify your CFD model.


  • #8
    Johnny Serpico
    Registered Users, Registered Users 2 Posts: 4 ✭Johnny Serpico


    Get that Versteeg book, I used it all the time. My project was covered in references to it.

    I did a similiar FYP to this involving wind tunnels.

    Keep the structure to the most basic geometry, the unsymetrical T mentioned earlier. No one is interested in a highly complicated shape, it will just lead to questions about modelling and accuracy & relevance that are out of scope.

    The mesh is one of the most important parts, spend a lot of time on it.

    Remember you have to be able to justify everything in your project so know your theory and know why you made every decision in Fluent/Ansys etc...


Advertisement