Velocity increase around fixed Floating body when mass is increased
Hi all,
I came up with this already in a prior post. However, I couldn’t fix it, so far. So here comes my problem. I am trying to benchmark and validate a falling floating body through fluid, similar to the CaseFloatingSphereVal2D. I only did the following changes in the Def of the case.
- Increased the FtPause 5s
- Generated the floating object in the middle of the domain intead at the top
- Increased the relative weight to 2.6.
I observed that an artificial increase of the velocities at the boundaries of the floating body, even if it is fixed by the FtPause, which I cannot explain so far. If I reduce the weight back to 1.3 everything is fine. This also happened in case of the DEM "module". Shifting helped a bit but did not solve the problem.
Please note that I took extra care that the floating is filled with particles, as suggested previously. I also tested all options to modify the weight of the floating as in the CaseTemplate, as well as various positions. I attached video on my gdrive:
https://drive.google.com/open?id=0B3ajAgMnxT6kdlV2TFhwRXh5aEk
How can I avoid these velocities ??? Can anybody explain that to me ???
Any help on this is appreciated.
H
I came up with this already in a prior post. However, I couldn’t fix it, so far. So here comes my problem. I am trying to benchmark and validate a falling floating body through fluid, similar to the CaseFloatingSphereVal2D. I only did the following changes in the Def of the case.
- Increased the FtPause 5s
- Generated the floating object in the middle of the domain intead at the top
- Increased the relative weight to 2.6.
I observed that an artificial increase of the velocities at the boundaries of the floating body, even if it is fixed by the FtPause, which I cannot explain so far. If I reduce the weight back to 1.3 everything is fine. This also happened in case of the DEM "module". Shifting helped a bit but did not solve the problem.
Please note that I took extra care that the floating is filled with particles, as suggested previously. I also tested all options to modify the weight of the floating as in the CaseTemplate, as well as various positions. I attached video on my gdrive:
https://drive.google.com/open?id=0B3ajAgMnxT6kdlV2TFhwRXh5aEk
How can I avoid these velocities ??? Can anybody explain that to me ???
Any help on this is appreciated.
H
Comments
These seem to be inevitable issues of using dynamic boundary conditions with simple SPH. It gets very technical (you can check 10.1016/j.jcp.2016.12.004), but basically the error we have by the boundaries (and we treat floatings as boundaries in some aspects) is because we have NO way of assuring that we are first order consistent under a scattered particle distribution. That is an issue because density is at best first order and we automatically get a density error, a discontinuity, at the interface (that gap everyone notices around the boundaries). What you seem to have is the evolution of that discontinuity after some time, fluid particles get denser by accumulating error by the body and they start to flow down by gravity. Delta-SPH should alleviate some of those problems but it only works for some time.
Don't wait for 5 seconds. Don't expect anything ultra-better until we can implement some serious higher-order schemes or reconstructions.
All the best
I should point out that by setting a proper value of alpha (viscosity) the problem will be resolved.
Thank you so much for your input. That explains a lot. Best H