NITA Fractional Step - Residuals & Convergence
 

Hello :)

This days I found out how NITA Fractional Step speeds up about 10x the CPU-time for simulations around a cylinder. With this feature, I dared to run RANS k-w transition SST with SAS.

I understand that NITA performs inner iterations. And this can be "controlled" with Controls->Advanced->Expert. This is well explained in Fluent's manual.

But... (there is always a but :rolleyes:) I can not fully understant what is happening inside each time-step. I activated verbosity=1. And I get the following in 1 time-step:


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

x-momentum equation:
tol. 2.6483e-09
0 1.8538e+00
1 9.7318e-03

y-momentum equation:
tol. 1.2665e-10
0 3.2576e+00
1 1.5402e-02

x-momentum equation:
tol. 2.6483e-09
0 2.3432e-02
1 1.4507e-03
2 2.8526e-04
3 6.6125e-05

y-momentum equation:
tol. 1.2664e-10
0 2.6327e-02
1 2.2418e-03
2 4.9086e-04
3 1.2797e-04

pressure correction equation:
tol. 9.3473e-16
0 1.8695e-03
1 9.2260e-04
2 4.2316e-04
3 2.1374e-04
4 1.1580e-04

pressure correction equation:
tol. 8.9469e-17
0 1.7894e-04
1 1.1491e-04
2 6.7756e-05
3 4.1333e-05
4 2.6478e-05
5 1.6730e-05
6 1.0049e-05
7 6.3437e-06
8 3.9266e-06
9 2.3948e-06
10 1.4192e-06
11 7.1497e-07
12 5.0770e-07
13 3.5760e-07
14 2.3073e-07
15 1.8530e-07

k equation:
tol. 6.5366e-12
0 1.5345e-01
1 1.3704e-03

omega equation:
tol. 1.8028e-05
0 1.2873e+02
1 9.0926e-01

k equation:
tol. 6.5366e-12
0 3.1488e-03
1 1.6761e-04
2 2.6553e-05
3 5.0554e-06

omega equation:
tol. 1.8028e-05
0 3.3582e+00
1 8.1995e-02
2 9.8305e-03
9018 1.8695e-03 2.9803e-04 5.2372e-04 8.0712e-03 1.7060e-03 0:00:01 1
9018 1.7894e-04 3.7672e-06 4.2326e-06 1.6562e-04 4.4504e-05
step flow-time vel_out flow-time cl cd
9018 1.5686e+01 5.4330e-01 1.5686e+01 -1.5954e+00 1.7933e+00
Flow time = 15.68559s, time step = 9018


Here the expert configuration is by default. And I want to understand how this values affect the inner iterations, so, what are those numbers for: tol., 0, 1, 2... etc. Where they appear from?

What I would like to do is manipulate the configuration to get scaled residuals of 1E-5, which is the number I use with iterative method.

Thank you guys! :)

Is there someone with experience with NITA solution methods???
I need to understand this NITA Residuals, and compare them to ITA Residuals