Learning Forum

[blackbaddl]

Hi everyone, 

I encountered a problem while postprocessing on HPC System. When switching from steady-state to transient solving I do not get the turb-kinetic energy any more. I am not shure wether this is a technical issue or I did not get the concept completely yet. 

I downloaded the last time step .dat file and tried it by hand. The same, no turb kinetic energy. What is the problem ? 

These are my solver settings:

 

/define models solver pressure yes 

/define operating-conditions gravity no

/define models energy yes no no no no 

/define models viscous kw-sst yes

/define models viscous curvature-correction yes

/define materials change-create air , yes incompressible-ideal-gas , no yes sutherland , , , , no no no

;

;

;---Solver Settings---

;Pressure Velocity Coupling Scheme [20] ?

;; 20 for SIMPLE

;; 21 for SIMPLEC

;; 22 for PISO

;; 24 for Coupled

/solve set p-v-coupling 21

;

;Convective discretization scheme for Pressure [10]

;; 10 for Standard

;; 14 for PRESTO!

;; 11 for Linear

;; 12 for Second Order

;; 13 for Body Force Weighted

/solve set discretization-scheme pressure 14

;

;Convective discretization scheme for Momentum [0]

;; 0 for First Order Upwind

;; 1 for Second Order Upwind

;; 2 for Power Law

;; 4 for QUICK

;; 6 for Third-Order MUSCL

/solve set discretization-scheme mom 6

;

;Convective discretization scheme for Energy [0]

;; 0 for First Order Upwind

;; 1 for Second Order Upwind

;; 2 for Power Law

;; 4 for QUICK

;; 6 for Third-Order MUSCL

/solve set discretization-scheme temperature 6

;

;Convective discretization scheme for Turbulence Kinetic Energy [0]

;; 0 for First Order Upwind

;; 1 for Second Order Upwind

;; 2 for Power Law

;; 4 for QUICK

;; 6 for Third-Order MUSCL

/solve set discretization-scheme k 6

;

;Convective discretization scheme for Turbulence Dissipation Rate [0]

;; 0 for First Order Upwind

;; 1 for Second Order Upwind

;; 2 for Power Law

;; 4 for QUICK

;; 6 for Third-Order MUSCL

/solve set discretization-scheme omega 6

;

;---set under-relaxation factors when divergence occurs---

/solve set under-relaxation temperature 0.6

/solve set under-relaxation turb-viscosity 0.6

;/solve set under-relaxation omega 0.6

;/solve set under-relaxation pressure 0.6

;/solve set uner-relaxation mom 0.4

;/solve set under-relaxation density 0.8

;/solve set under-relaxation body-force 0.7

;

/solve set high-order-term-relaxation enable? yes

;

/solve initialize compute-defaults all

/solve initialize hyb-initialization

;/solve iterate number-of-iterate

/solve set timestep 0.01

/solve dual-time-iterate 150 300

;

 

By the way the solution converged by itself and the residuals of  K and omega existed of course

 

Thank you in advance

[DrAmine]

Are u post processing in Fluent,? What is missing? You are able to solve TKE and you its residual.

[blackbaddl]

Yes I do postprocessing in fluent. Basically everything affecting turbulance is missing in postprocessing. I was able to solve the case perfectly and because of the residuals I assumed that turbulance values should have been callculated.

The are the residuals of the last timestep I downloaded from the HPC to look into that by hand.

 

  iter  continuity  x-velocity  y-velocity  z-velocity      energy           k       omega     time/iter

 12759  1.2295e-03  2.2280e-06  1.6802e-06  2.1652e-06  9.5707e-17  2.5318e-05  4.3953e-05  0:07:09  234

 12760  1.1969e-03  2.1917e-06  1.6538e-06  2.1311e-06  9.5311e-17  2.4826e-05  4.3069e-05  0:07:05  233

 12761  1.1655e-03  2.1563e-06  1.6278e-06  2.0977e-06  9.5589e-17  2.4347e-05  4.2149e-05  0:07:01  232

 12762  1.1354e-03  2.1216e-06  1.6023e-06  2.0649e-06  9.5669e-17  2.3879e-05  4.1312e-05  0:06:57  231

 12763  1.1060e-03  2.0877e-06  1.5775e-06  2.0329e-06  9.5496e-17  2.3423e-05  4.0452e-05  0:06:54  230

 12764  1.0772e-03  2.0543e-06  1.5530e-06  2.0013e-06  9.5454e-17  2.2977e-05  3.9660e-05  0:06:51  229

 12765  1.0499e-03  2.0216e-06  1.5290e-06  1.9703e-06  9.5616e-17  2.2542e-05  3.8836e-05  0:06:49  228

 12766  1.0232e-03  1.9897e-06  1.5054e-06  1.9400e-06  9.5554e-17  2.2117e-05  3.8086e-05  0:06:46  227

!12767 solution is converged

 12767  9.9754e-04  1.9584e-06  1.4823e-06  1.9103e-06  9.5626e-17  2.1703e-05  3.7303e-05  0:06:44  226

Flow time = 1.5s, time step = 150

 

So the goal is to do a basic postprocessing on HPC System because each of these time steps has 35 GB of data.

[blackbaddl]

This is completely missing. All I changed is solving transient and not steady state. All other settings are the same

[DrAmine]

That is weird can you start a dummy case just to check if you face same issue

[blackbaddl]

Sorry for the very late reply, 

the HPC system was down for quite a while due to maintance and I have not been at the project for some time because of being sick.

 

However, yes there is the same issue in a dummy case.... It is very weird. We decided to stay with steady state simulations because there is not a real added value in transient solutions for our project at this moment. But this might be worth looking into in more detail some time in the future.....

 

I faced another big big problem with Fluent on HPC. I am not able to mesh. This a lot worse for us at the moment as we are trying to implement an efficent workflow "backwards". (Just the order of implementation and development) 
But I will post another question with details on it. Would be great if somebody is looking into that.

 

Thank you