What is the difference between fluid transfer to solid force and stress when solving FSI?

Official adapters and tutorials
1

I want to know what is the difference between fluid transfer to solid force and stress when solving FSI? I found in the Turek-Hron FSI3 example that early examples were transmitting forces, but now the examples are transmitting stress. And in the acceleration scheme, both stress and displacement are used as acceleration factors, but in previous examples, only displacement was used. I found it difficult to converge the transmission force during testing, is that the case?

2

I want to know what is the difference between fluid transfer to solid force and stress when solving FSI?

The main difference is that force depends on the area, while stress does not. With a force, you would apply a conservative mapping, while with a stress, you would apply a consistent mapping.

Exchanging forces is a bit trickier, as you always need to have the correct area of each face. Exchanging stresses is, therefore, easier, and potentially less prone to accuracy errors.

I found in the Turek-Hron FSI3 example that early examples were transmitting forces, but now the examples are transmitting stress.

Each simulation code has different restrictions. We found that coupling stresses was easier to integrate into deal.II, that’s why we use it now. Previous examples were probably coupling CalculiX, where we had already implemented forces coupling.

And in the acceleration scheme, both stress and displacement are used as acceleration factors, but in previous examples, only displacement was used.

This is not an easy question, and we recently added some documentation regarding primary and secondary data in coupling schemes. Data that is used to compute the acceleration factors is called primary data, and what data can be primary data depends on the coupling scheme and on who is the second participant.

We now use a parallel-implicit scheme (which means we can use both fields in the acceleration), while the previous example probably used serial-implicit.

As a rule of thumb, using more information to compute the factors helps, at least as long as that information is not too noisy.

1 Like
3

Hi, @Makis

Is it possible for us to get Turek-Hron FSI3 example of couping OF and CalculiX?

4

@Ulrich has already posted a setup of the FSI2 setup with OpenFOAM and CalculiX (of course, older versions): Turek-Hron FSI 2 Benchmark

There is also an open PR in the tutorials: Turek & Hron FSI3 with CalculiX and OpenFOAM by JoSchrdr · Pull Request #304 · precice/tutorials · GitHub

1 Like