[Josh]

Hi folks,



I'm simulating a rectangular cylinder half exposed to air and half to water with a periodic (sinusoidal) rolling motion for the purpose of crudely mimicking a ship hull rolling back and forth. It rolls clockwise about the axis facing the picture to an amplitude of 0.05 rad, then counterclockwise to the same amplitude, and so on.








My main goal here is to extract the cylinder moment about the Z-axis so I can calculate added mass and roll damping coefficients. I have successfully ran the simulation with a cylinder fully submerged in water and obtained moment values comparable to published Fluent values. See below for an example case.





However, as soon as I introduce an Eulerian multiphase free surface (i.e., 50% submerged in water, 50% exposed to air - see top picture), my moment values become consistently negative. See below.





Since moment calculation report is dependent upon continuum reference density and the default reference density is for air (1 kg/m^3), I thought I would change the reference density to the volume averaged value (roughly 500 kg/m^3 for a 50/50 air/water mix). This yielded better moment value behaviour. See below.




The problem with changing the reference density is that the entire solution is affected. Suddenly, the higher reference density was causing a pressure gradient, which caused the water to push all the air out of the domain. Even if I replace the pressure outlet boundaries with no-slip walls, the gradient still exists and skews the solution. The Help documentation also indicates that having a reference density higher than that of the lightest fluid is bad practice.

I've been troubleshooting this case for weeks. I've tried so many parameterization tests my head is spinning. Any ideas on how to obtain the proper moment behaviour?

Full disclosure: I used to work for CD-adapco in Engineering Services.

[ping]

since moment is just the local pressures times the local areas and thus reference heights will balance out

so i cant see any reason for the moment to be affected by the reference alt and dens

these need to be set to sensible and recommended values based on your air to get the model running well

[Josh]

Hi ping,

Thanks for the reply. I received this message about a week ago from tech. support:

I was looking through the help documentation and found a page that is relevant to this case. It's at:
Modeling Physics > Modeling Flow and Energy > General Navier-Stokes > Modeling Gravity



The working pressure for the simulation is determined in part by the reference density, so the body forces on the boat are going to scale with the reference density as well. I think with the reference density as air you transposed the moment far down into the negative since the body forces were so out of scale. I haven't heard back from our marine specialist to corroborate this yet, but at the moment it appears that changing the reference density is the way to go.

In short, the reference density affects the solution because it changes the working pressure to the piezometric pressure.

Any other ideas?

[ping]

those settings are irrelevant in this case and this proves it:

i took the boat in waves tutorial and replaced its volume mesh with a static volume the same size then added an overset region with a plank like you image at the water surface which has forced rotation motion

then made the 5th order wave flat with no velocity and rotated the plank at 20rpm then reversed it to -20rpm then reversed it again

the monitored moment does what we expect flipping from positive to negative so i dont know what you are doing incorrectly

[Josh]

Hey ping,

I've done the boat in waves tutorial before. The only difference between my simulation and that in the tutorial is that they used the DFBI solver. I'm using a simple rotation motion with zero-velocity flat wave.

Regardless, I'll try your method and post back.