problem with the time step jump in an LES (transient) study


I am running an LES study for a periodic pipe unit which has helical grooves around its periphery.

Although, I know the results have not converged so far, I did wanted to ask about a jump in time step/flowtime that I have observed in the report file that I create.

One of my wall temperatures reports have the following resutls with

time-steps ----wall temperature ------flow time
1185               300.01324                   0.00473999999999
2439               300.33120                  0.009755999999999

Below and above these two numbers is a 1 time step increase (.....1183, 1184 and 2349 , 2340, 2341 ...... etc), except in the above shown result.

I have the following commands for saving of time steps and others.

/file/auto-save/save-data-file-every time-step
/file/auto-save/data-frequency 2500
/file/auto-save/case-frequency each-time

/solve/set/cfl-based-adaptive-time-stepping yes no 1 1 0.000004 100 1 , , , ,
/solve/set/max-iterations-per-time-step 50
/solve/set/time-step 0.000004
/solve/set/number-of-time-steps 250000

My guess is this jump is because of the cfl-based-adaptive-time-stepping yes command that I have. And as a consequence if I try to plot the wall temperature vs time step(or) flow-time, I get a discontinious graphs.

Is that the case why I have this jump in the time step? Or is there an another reason?

Also, adaptive time stepping for cfl seems a nice feature, so should it always be used?. Or more importantly, when should it be not used?


  • RobRob UKForum Coordinator
    edited May 13

    If you're using adaptve time stepping you'll need to report data against flow time and not time step. 

    I tend not to use it as the jump in time step can trigger instabilities in the solver if you have rapid transients or changes in time scale due to shedding etc. 

  • atulsingh92atulsingh92 Member
    edited May 13

    If you are mentioning about creating a report with flow-time, than that is exactly how I have obtained the above talked result.


    /solve/report-files/add t_wall file-name t-wall.out itr-index 1 name t_wall report-defs t_wall , frequency-of flow-time flow-frequency 0.000004 print yes

    The third colum that I show above, is for flow time, with a time step of 4 micro seconds. But you can see the cfl update being 0.004 to 0.009 which is not of 4 micro seconds.

  • RobRob UKForum Coordinator
    edited May 14

    That's the flow time in the two lines of data, not the time step. 

Sign In or Register to comment.