CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Total time for simulation (http://www.cfd-online.com/Forums/main/92177-total-time-simulation.html)

Meerkat September 5, 2011 06:48

Total time for simulation
 
Hello,

I am running a hydrodyamics simulation that is using a 3rd order TVD Runge-Kutta time stepping method along with appropriate 3rd order reconstruction routines to calculate primitive variables at cell boundaries for use in a variety of Riemann solvers to calculate the fluxes. However, my question is quite simple:

After each completed loop of the three steps in the RK algorithm, is the total amount of time elapsed increased by a single time step, three time steps or a third of the time step?

I have tried to compare my results against test problems in the literature and while I can reproduce the solution it is usually at the wrong time. I am just not sure how to control the time step in the calculation of the overall time elapsed. Everything else is working fine!:p

Thoughts anyone?

cfdnewbie September 5, 2011 07:17

Hello Meerkat, this link might help you:
http://en.wikipedia.org/wiki/List_of...d-order_method

note that the step (timestep) here is h, and the coefficients are the b's....So in the three step method, your bs are 1/6, 2/3, 1/6 (note they add up to 1!). This means that if you have applied your method correctly, the solution is advanced by h (in space or time).

Might I suggest that you turn to the explicit Euler: du/dt = (u(n+1)-u(n))/delta t... that's the easiest and most simple, I'm quite sure there won't be stability issues with an O(3) method....

So pick the Euler, and try a simple testcase, i.e. a density bubble transported by a constant flow velocity.... thereby, you know the physical speed of the problem and can debug your code!

cheers!


All times are GMT -4. The time now is 01:08.