0

Steady-state override

Dear TOUGH community,

I'm running some simple simulations and reach steady-state too early.

"

 !!!!!!!!!  FOR 10 CONSECUTIVE TIME STEPS HAVE CONVERGENCE ON <NumNRIterations> = 1
       ==>  WRITE OUT CURRENT DATA, THEN STOP EXECUTION

"

Yet, I do want the simulations to keep going because action is around the corner and I need to export specific times. I could reduce RE1 but it is the only solution? I couldn't find any other input parameter which would help.

For example, could I change the number of consecutive time stepswhich trigger the "stop execution" (from "10" to "100") ? Or is it hardwired in the source code?

 

Thank you

Ludo

2 replies

null
    • Finsterle GeoConsulting
    • Stefan_Finsterle
    • yesterday
    • Reported - view

    Hi Ludo,

    In case you are using iTOUGH2, simply set MOP2(1)=2 and it enforces at least 2 Newton-Raphson iterations, so it never converges in a single iteration and always updates the primary variables. This feature may also be available in TOUGH3 (I'm not sure, though).

    In iTOUGH2, you can also increase the maximum number of time steps converging in a single  iteration using command >>> consecutive: max_iter1. Increasing the fixed number of 10 in whatever version of  TOUGH you are using would also be easy.

    However, my main recommendation would be to accept the insight that nothing is happening in the system for a while. Then just remove all the tiny-spaced printout times causing the problem. In a simple post-processing step, just take the output at the beginning of this inactive period and copy (rather than "exporting") it as many times as you need until the period converging in a single time step is over - after all, you know that the primary variables stay the same until you turn the corner.

    I hope everything is going well down under.

    Cheers,

    Stefan

      • Ludovic_Ricard
      • 11 hrs ago
      • Reported - view

      Hey Stefan,

      Thank you very much for your quick reply.

      All good down under. Hope you had a good time at the symposium and all is well your side too.

      I am using Tough4 and it does have MOP2 implemented (not yet ready to couple with iTough2). The MOP2(1) = 2 worked the trick. The simulation kept going until the next action and it respected all 'TIMES' record.

      I like your idea about the copy of the data rather than letting the system run. However, I might need to think deeper about how to implement that in practice.

      I'm exporting results at times of the DTS observations. Of course I do not know if and when the system would go steady state (although, the DTS observations should give me some decent pointers). If it does, then I could check the final time and compare it with the expected end time and potentially relaunch the simulation with a revised 'TIMES' Keyword and INCON. 

       

      Best

      Ludo

Content aside

  • 11 hrs agoLast active
  • 2Replies
  • 6Views
  • 2 Following