[Andrew Norfolk]

Hi everyone.

I am currently carrying out a mesh refinement study for a CHT problem and thought I could speed things up by interpolating the solution from a less refined mesh onto a new denser mesh. This means in theory that the initial conditions should be very close to the final solution, especially if mesh convergence has been achieved.

I've been doing this is by clicking on the initial values tab after loading the solution option in workbench, and manually clicking "specify initial values" and selecting my previous results file. I then choose to interpolate from the domain "AIR" in the results file (the source) to domain "AIR" in the NEW MESH (the target). Likewise for the domain "FIN".

When the solution starts the residuals immediately jump back to high values, imbalances to 100% and the solution takes as many iterations to converge as if I had never interpolated at all. I can only assume that I must be doing something wrong as I was expecting the solution to converge much faster.....

+--------------------------------------------------------------------+
| |
| Interpolation of Initial Values |
| |
+--------------------------------------------------------------------+




+--------------------------------------------------------------------+
| |
| ANSYS(R) CFX(R) Interpolator |
| |
| Release 17.0 |
| Build 17.0 2015.11.30-15.46-135354 |
| Mon Nov 30 15:53:59 GMTST 2015 |
| |
| Executable Attributes |
| |
| single-64bit-int32-archfort-optimised-noprof-lcomp |
| |
| (C) 2015 ANSYS, Inc. |
| |
| All rights reserved. Unauthorized use, distribution or duplication |
| is prohibited. This product is subject to U.S. laws governing |
| export and re-export. For full Legal Notice, see documentation. |
+--------------------------------------------------------------------+




+--------------------------------------------------------------------+
| Job Information at Start of Run |
+--------------------------------------------------------------------+

Run mode: serial run

Host computer: PSLCAD003 (PID:676)

Job started: Wed Jul 20 14:52:19 2016


+--------------------------------------------------------------------+
| Memory Allocated for Run (Actual usage may be less) |
+--------------------------------------------------------------------+

| Real | Integer | Character | Logical | Double
----------+------------+------------+-----------+----------+----------
Mwords | 294.37 | 278.38 | 0.20 | 0.01 | 0.30
Mbytes | 1122.95 | 1061.94 | 0.19 | 0.04 | 2.29
----------+------------+------------+-----------+----------+----------


+--------------------------------------------------------------------+
| Host Memory Information (Mbytes) |
+--------------------------------------------------------------------+
| Host | System | Allocated | % |
+-------------------------+----------------+----------------+--------+
| PSLCAD003 | 65403.57 | 2187.41 | 3.34 |
+-------------------------+----------------+----------------+--------+

================================================== ====================
Interpolating Onto Domain "AIR"
================================================== ====================

Total Number of Nodes in the Target Domain = 8246689
Bounding Box Volume of the Target Mesh = 4.00000E-02


Checking user defined source locations from "Interpolation Mapping 1":
Target mesh is different from domain "AIR".

Searching for Candidate Source Domains:
User defined source domain "AIR"
in "Interpolation Mapping 1":
Number of Mapped Nodes = 8246689 (100.0%)
Bounding Box Overlap Volume = 4.00000E-02 (100.0%)

Start interpolation of variables:
Source Domain Name Mapped Nodes
AIR 8246689 (100.0%)

+--------------------------------------------------------------------+
| Variable Range Information |
+--------------------------------------------------------------------+
+--------------------------------------------------------------------+
| Variable Name | min | max |
+--------------------------------------------------------------------+
| Isothermal Compressibility | 9.87E-06 | 9.87E-06 |
| Thermal Conductivity | 2.61E-02 | 2.61E-02 |
| Courant Number | 1.87E+02 | 2.12E+05 |
| Density | 9.45E-01 | 1.20E+00 |
| Density Derivative wrt Pressure at Constant| 9.32E-06 | 1.19E-05 |
| Static Enthalpy | -5.04E+03 | 7.58E+04 |
| Static Entropy | -1.71E+01 | 2.27E+02 |
| Pressure | -1.20E+01 | 3.76E+01 |
| Specific Heat Capacity at Constant Pressure| 1.00E+03 | 1.00E+03 |
| First Blending Function for BSL and SST mod| 2.16E-07 | 1.00E+00 |
| Second Blending Function for SST model | 3.84E-02 | 1.00E+00 |
| Turbulence Eddy Dissipation | 1.11E-01 | 2.11E+03 |
| Turbulence Eddy Frequency | 5.09E+01 | 1.00E+06 |
| Temperature | 2.93E+02 | 3.74E+02 |
| Turbulence Kinetic Energy | 5.64E-04 | 1.94E+00 |
| Velocity | 6.86E-03 | 7.13E+00 |
| Dynamic Viscosity | 1.83E-05 | 1.83E-05 |
| Eddy Viscosity | 8.35E-10 | 1.31E-03 |
| Wall Scale | -3.54E-06 | 7.37E-04 |
+--------------------------------------------------------------------+

================================================== ====================
Interpolating Onto Domain "FIN"
================================================== ====================

Total Number of Nodes in the Target Domain = 922535
Bounding Box Volume of the Target Mesh = 9.76320E-05


Checking user defined source locations from "Interpolation Mapping 2":
Target mesh is the same as domain "FIN".

Start direct copying of variables from domain "FIN".

+--------------------------------------------------------------------+
| Variable Range Information |
+--------------------------------------------------------------------+
+--------------------------------------------------------------------+
| Variable Name | min | max |
+--------------------------------------------------------------------+
| Thermal Conductivity | 2.37E+02 | 2.37E+02 |
| Density | 2.70E+03 | 2.70E+03 |
| Static Enthalpy | 6.34E+04 | 8.27E+04 |
| Static Entropy | 1.91E+02 | 2.42E+02 |
| Real Partition Number | 1.00E+00 | 8.00E+00 |
| Specific Heat Capacity at Constant Pressure| 9.03E+02 | 9.03E+02 |
| Specific Heat Capacity at Constant Volume | 9.03E+02 | 9.03E+02 |
| Specific Volume | 3.70E-04 | 3.70E-04 |
| Temperature | 3.68E+02 | 3.90E+02 |
+--------------------------------------------------------------------+

================================================== ====================
Copy GGI Data Onto Domain Interface "Default Fluid Solid Interface"
================================================== ====================

Checking all source domain interfaces from the source file:
Target GGI is different from domain interface "Default Fluid Solid Interface".

Skip GGI data from the source file.

+--------------------------------------------------------------------+
| CPU Requirements of Interpolation |
+--------------------------------------------------------------------+

Domain Name Tree Setup Interpolation
(secs. %total) (secs. %total)
----------------------------------------------------------------------
AIR 2.57E+01 45.4 % 3.01E+01 53.4 %
FIN 1.08E-01 0.2 % 2.71E-01 0.5 %
-------- ------- -------- ------
Domain Summary 2.58E+01 45.6 % 3.04E+01 53.8 %

Miscellaneous 3.00E-01 0.5 %
--------
Total 5.65E+01

+--------------------------------------------------------------------+
| Job Information at End of Run |
+--------------------------------------------------------------------+

[evcelica]

It sounds like you are doing it correctly. I usually just use the old results file as initial values, and don't specify anything else. You will see a jump in your residuals, I'm surprised about the imbalance though, are you using a very large timestep? You could always put in some monitor points to see what the values are, and if they start at what you expect. or run 1 iteration with a small time step and look at the results.

[mortazavi]

This is my question too. any help please?

[Andrew Norfolk]

My time step was very large as the this enabled me to previously converge the solution in a much shorter time. I am going to try running the interpolated solution using a smaller value and see if it makes a difference. Will post results.