Learning Forum

[Murtala Musa]

I am trying to simulate the heave decay of a cylindrical spar hull. The residual seems to stay steady and not drop (see uploaded images below). The solution seems to be converging at every time step, as can be seen highlighted in the console. 
I made some changes to the default convergence criteria by lowering all the residuals threshold criteria from 0.001 to 0.0005. but the residuals behaved the same, ie residual not droping down.
Is this a normal thing, or is there a problem somewhere?
By the way, I use 6dof with smoothing and remashing enabled.

Thanks for your time 

[Rob]

The residuals look fine, that's the usual profile for a transient model. How do the monitors for postiion etc look? Note, you've only run about 0.0227s of real time so there's a good chance the block hasn't moved very far. 

[Murtala Musa]

 

The heave position monitor looks fine; it is oscilating as I expected. Although it does not match precisely with the experimental data that I am trying to validate, do you have any suggestions on how I can increase the accuracy? 
After letting the experiment run for some hours, this is what the residuals look like (see image below).
The heave decay monitor, which I am trying to match with experimental data, is also attached below. For reference, Palm_exp and Palm_Openfoam are the data I am trying to validate using Fluent.

I am thinking about decreasing the mesh size once more to make it fine.

 

[Rob]

If the mesh is too coarse and/or time step too high it'll explain accuracy issues. The other potential problem is checking the initial condition is the same as in the paper. It won't be the first time the paper's definition of a starting point or other piece of critical information was omitted. 

[Murtala Musa]

One more question, please: I don't have Microsoft Visual Studio installed. I am using the Fluent built-in compiler to build and load my UDF instead. Is that a problem? If there is a built-in compiler in Fluent, why is there a need to install the Visual Studio compiler? most of the tutorial I see use visual studio to compile their udf.

[Rob]

The built in compiler is generally preferred, and is CLANG based. Visual Studio was the only option for Windows until some versions back but it was getting difficult to find the free parts and comply with the various licence agreements. We'd been asking for a built in compiler for about 15 years when it arrived! Prior to first asking we (Fluent staff) all used UNIX or LINUX desktops.... 

[Murtala Musa]

The reason I ask is because fluent give me some warning in the console, saying visual studio compiler is not found or broken, something like that. 
so, you are saying I should continue using the built-in compiler and ignore the waring in the console? 

[Rob]

If it's a warning and the solver is reading & using the UDF it's probably safe to ignore. 

[Murtala Musa]

Hello Rob, 
Sorry to bother you once more. In this heave decay analysis, I am using dynamic mesh with remeshing and smoothing in my setup. I just learn about overset mesh which doesnt require smoothing and remeshing. what do you think about that? can I apply it in this case to model heave decay?  

[Rob]

You can use overset for this. Rather than remeshing Fluent will recalculate the data transfer cells as the object moves. You may not save much time overall, but you will avoid getting skew cells: it's a trade off on the two models. If you look in the Tutorials under Fluent's help there's an example.