search

Xiangyu-Hu / SPHinXsys

SPHinXsys provides C++ APIs for engineering simulation and optimization. It aims at complex systems driven by fluid, structure, multi-body dynamics and beyond. The multi-physics library is based on a unique and unified computational framework by which strong coupling has been achieved for all involved physics.
https://www.sphinxsys.org/
Apache License 2.0
258 stars 192 forks source link

A phenomenon in Eulerian framework #307

Closed ZhentongWang closed 5 months ago

ZhentongWang commented 10 months ago

when I try the case of backward-facing step case with Re=400, I found out a phenomenon but I think this phenomenon also occur in other cases like flow around a cylinder case. At beginning , I don't introduce the numerical dissipation from the Riemann problem, https://github.com/Xiangyu-Hu/SPHinXsys/blob/319520e02f631847f85160c0e8b55648c395eafa/tests/user_examples/ExtraSources/extra_sources_share/common_weakly_compressible_eulerian_classes.h#L93-L98 here, I set parameter in the limiter as 0, then we can see this not using Riemann solver. and the result of the velocity contour in this case is limiter0_t_0 3 and the acostic timestep is (https://github.com/Xiangyu-Hu/SPHinXsys/blob/319520e02f631847f85160c0e8b55648c395eafa/tests/user_examples/ExtraSources/extra_sources_share/common_weakly_compressible_eulerian_classes.cpp#L16-L19)

I have tried two approaches to solve this, first one is that increase the parameter in the limiter, and I try parameters 5 and 10 in the limiter limiter5_t_0 3 and limiter10_t_0 3 as shown, there is a pattern when the parameter is low, this is also occurring in the flow around cylinder case , but we thought that as sound propagation before, but probably it is not. and with the increase of the parameter in the limiter, the pattern is dissapearing gradually.

Second one is that I decrease the acostic timestep (parameter in the limiter is also 0) , I use the equation like this 0.5(0.6 / Dimensions smoothing_lengthmin / (reduced_value + TinyReal);) and the result is like limiter0_t_0 15 the pattern is not showing in this situation.

Compared this two approaches, I think we could decrease the acoustic timestep a little bit cuz we could get higher accurate result in the low resolution. what do you think of this ? professor Hu

DrChiZhang commented 10 months ago

High-order time integration may also be helpful to remove the noise.

ZhentongWang commented 10 months ago

High-order time integration may also be helpful to remove the noise. thanks for the advice. I will check that later

Xiangyu-Hu commented 10 months ago

I am wondering whether "set parameters to zero" means no or maximum numerical dissipation?

ZhentongWang commented 10 months ago

I am wondering whether "set parameters to zero" means no or maximum numerical dissipation?

means no numerical dissipation from the Riemann solver

Xiangyu-Hu commented 10 months ago

I am wondering whether "set parameters to zero" means no or maximum numerical dissipation?

means no numerical dissipation from the Riemann solver

Does it means no riemann solver.

ZhentongWang commented 9 months ago

I am wondering whether "set parameters to zero" means no or maximum numerical dissipation?

means no numerical dissipation from the Riemann solver

Does it means no riemann solver.

Yes. just using average value as middle state value

Xiangyu-Hu commented 5 months ago

Not sure the issue still there.