Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

CFD in Astrophysics, ideas for a senior project

  1. Dec 8, 2009 #1
    Hello everyone, I am an looking for a senior project for aerospace engineering. In my Applied Computational Fluid Dynamics (CFD) course it came up that one of the many applications of CFD is astrophysics and the modeling of planet/star formation.

    Our professor suggested it to me as a possible idea for my senior project. I have done some intense research but i still don't not have a clear view of the subject. It is very broad field and a lot of things are beyond my undergraduate capabilities. All the papers and information I found has been extremely exciting but most of it seems beyond my reach and time commitment.

    I was wondering if anyone here had any ideas for me on what problem I should take on for my senior project. I need ideas for something that can be contained in a senior project and not a masters/Ph.d thesis.

    Thank you!
  2. jcsd
  3. Dec 10, 2009 #2


    User Avatar
    Science Advisor

    I think you'd be mad to embark on a projecct without discussing it with someone from the relevant field. I'm sure your professor knows about CFD, but if you want to do a CFD in astrophysics project you really need to talk to an astrophysicist (a current professor), otherwise you'll spend to long on the 'boring' bits trying to isolate the problem you want to solve, and less time on the interesting part; actually solving that problem.

    The thing about astrophysics is that while CFD can be very important, that part of a problem tends to be easy compared with other physics, such as nuclear processes or magnetic fields. That's probably not true in all cases, but that's why you'd need to speak to an expert to find the areas in which the CFD part is the greatest source of uncertainty.
  4. Dec 10, 2009 #3
    You probably should talk to someone that's already has a live code to see what can be done. If someone already has a running code, then there are usually a lot of tweaks that you can do that would be within a senior project.

    If you don't have any other ideas....

    Modifying a CFD code to do original physics is probably well outside a senior project, but one thing that would be useful and would probably fit into a senior project is to take an open source code, get it to run a simple simulation (like a Sedov blast wave) and then package it so it can be installed on a Linux Standard Base compliant machine as an RPM would be extremely useful.

    Something else you can do is to start with a simple simulation (i.e. Sedov blast wave), try to simulate it with a number of different packages, and then write a review with suggestions on improving the ease of use.
  5. Dec 10, 2009 #4
    Also see....


    Something else that would be within the time frame of a senior project is to take some CFD code with a semi-realistic test problem, and then compile the code with different compiler options and different compilers (gfortran/g95/ifort) and then see what produces the best performance. I'd also be interested in seeing if you take a machine and then switch on and off compiler options like SSE3, what happens to performance.

    A lot of my ideas right now are more CS than physics related, and that's because if you have a new code, you are likely to be spending about a month just getting it to run at all. Also the interesting thing about astrophysical CFD code is that you rarely have a situation in which someone understands all of the physics within the code. What people will do is to take results and equations that other people have derived and add it to a CFD code, and try to understand one piece of the puzzle, very, very well, and even that will give you a Ph.D. dissertation.

    Something else that you might consider doing is to try to get something running with GPU technology. GPU technology is something of a game changer and no one really knows how to use it in CFD, so that if you can figure out a way of using GPU's to do even minor things "i.e. Hello World" that would be interesting.
  6. Dec 10, 2009 #5


    User Avatar
    Science Advisor

    Nice link there, it reminded me of the http://www-theorie.physik.unizh.ch/astrosim/code/doku.php" [Broken] which contains a similiar set of information. That might also come in handy.
    Last edited by a moderator: May 4, 2017
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook