Learning Forum

[cviveknair]

I am facing an issue when restarting a transient simulation: My continuity residual does not converge even though it was converging in a few iterations in the previous time-steps. Additionally, rerunning the simulation from an earlier time point ( from periodically saved case and data file) results in the same issue even though in the original run the residuals where converging in a few iterations.

I am running fluent in a HPC and use fluent scripts files (.inp) to pass TUI commands; I have used the same procedure to successfully run previous cases (parametric study). I have faced this issue only once before, unrelated to the ongoing study, and simply re-ran the simulation from the beginning, however, I would like to know if this is a bug or something that can be diagnosed/fixed/corrected on my end?

Any help would be appreciated. Thank you.

[Rob]

It shouldn't make a difference. Are you using any adaption registers?

[cviveknair]

Hello,
Sorry for the late reply.
I am not using any adaptation registers. When I rerun it from the very beginning It is behaving properly again.

[Rob]

I've not seen that. What scripts are you using to restart?

[cviveknair]

BEGINNING SCRIPT FILE AS A .INP:
/file/read-case NFP_2C_del_20deg_12M_updated.cas.h5
;/mesh/scale 0.01 0.01 0.01
/define/models/viscous/laminar yes
/define/models/unsteady-2nd-order yes
;Boundary Conditions
;(specify individual zones If needed. Best to set it up properly in pointwise or fluent GUI)....
;/define/boundary-conditions/zone-type inlet velocity-inlet
;/define/boundary-condtions/zone-type ....
/define/boundary-conditions/velocity-inlet inlet no no yes yes no 0.3 no 0
/report/reference-values/compute/velocity-inlet inlet
;Solver Scheme Selection
;24 is coupled ,20 is SIMPLE ,21 is SIMPLEC , 22 is PISO
/solve/set/p-v-coupling 22
;set p-v-controls if PISO (22) skewness correction iterations, neighbor corrections iterations, skewness-neighbor coupling
/solve/set/p-v-controls 2 2 yes
;PISO SIMPLE(C) URF
/solve/set/under-relaxation/mom 0.7
/solve/set/under-relaxation/pressure 0.3
;set p-v-controls if Coupled (24) CFL relaxation, mom under-relaxation, pres under-relaxation
;/solve/set/p-v-controls 10 0.5 0.5
;6 is 3rd order, 4 is QUICK, 1 is 2nd order
/solve/set/discretization-scheme/mom 1
;Convergence criteria
/solve/monitors/residual/convergence-criteria 1e-4 1e-5 1e-5 1e-5
;Solver Initialization
;/solve/initialize/compute-defaults/velocity-inlet inlet
;/solve/initialize/initialize-flow
/solve/initialize/set-hyb-initialization/general-settings200.50.5absolute no no yes
/solve/initialize/hyb-initialization
;Timestep size (CFL based)
;/solve/set/transient-controls/cfl-based-timestepping yes 0.5 1e-05 5 1e-06 1e-03 0.1 1.5 1
;Timestep (Fixed)
/solve/set/transient-controls/fixed-user-specified yes
/solve/set/transient-controls/time-step-size 1e-4
;Force Monitors
;/solve/monitors/force/drag-coefficient yes wall () yes yes "cd-history-FPTS_tval_laminar_Eoff-4M_validate_200it_5e-5_5K" no no 1 0 0
;/solve/monitors/force/lift-coefficient yes wall () yes yes "cl-history-FPTS_tval_laminar_Eoff-4M_validate_200it_5e-5_5K" no no 0 1 0
/file/cff-files no
;Solve
(define k 10)
(do ((i 0 (+ i 1))) ((> i 5))
(ti-menu-load-string (format #f "/solve/dual-time-iterate 10000 200"))
(ti-menu-load-string (format #f "/file/write-case-data NFP_2C_del_20deg_Re250_1e-4_~aK.cas" k))
(set! k (+ k 10))
)

exit
yes

2nd RESTART FILE:

/file/read-case NFP_2C_del_20deg_Re250_1e-4_40K.cas
;Convergence criteria
/solve/monitors/residual/convergence-criteria 1e-4 1e-5 1e-5 1e-5
;Timestep size (CFL based)
;/solve/set/transient-controls/cfl-based-timestepping yes 0.5 1e-05 5 1e-06 1e-03 0.1 1.5 1
;Timestep (Fixed)
/solve/set/transient-controls/fixed-user-specified yes
/solve/set/transient-controls/time-step-size 1e-4
/file/cff-files no
;Solve
(define k 40)
(do ((i 0 (+ i 1))) ((> i 1))
(ti-menu-load-string (format #f "/file/read-data NFP_2C_del_20deg_Re250_1e-4_~aK.dat" k))
(ti-menu-load-string (format #f "/solve/dual-time-iterate 5000 200"))
(set! k (+ k 5))
(ti-menu-load-string (format #f "/file/write-data NFP_2C_del_20deg_Re250_1e-4_~aK.cas" k))
)

exit
yes

I have used the same script for previous cases and have not faced any issues.

[Rob]

In the more recent versions of Fluent the time step is saved in the data file, so if you set time step etc as in P2 above anything you added will be overwritten. Try using:

/file rcd NFP_2C_del_20deg_Re250_1e-4_aK.cas
/solve/dual-time-iterate 5000 200
/file wcd NFP_2C_del_20deg_Re250_1e-4_aK_%t

Move all monitors to another folder before kicking off the run. As an aside, if you need 200 iterations per time step drop the timestep. Rough optimum for least iterations per second is 10-15 iterations per time step.

[cviveknair]

Hi Rob.
I missed your last update. I think the issue was with the continuity residual resetting at the beginning of the new run an as such would never go down because its already so small. All my monitors would pick up the physical values correctly. The 200 time steps was just overkill on my part. the process would just converge in 5-10 iterations.
To fix the problem I simply run a single large time step (twice the intended timestep size) at restart and then go back running the rest of the simulation. This "fixed" the problem.

[Rob]

Good to hear it.