Dampening force - CFD piston

jojoFD · January 13, 2022, 12:20

Hello,

I have problems to calculate the flow losses on my CFD study and could use some help.

The case is a piston with holes on the sides, oscillating (sinus curve) inside a closed 8m-long domain pressurized at 200bars (cf simplified picture).

I want to estimate the flow losses (dampening force, drag force) through the holes. So far I calculate (using Fluent) the global force acting on the whole piston and I substract it to the force acting on the rod (as the end of the rod is going out of the domain --> no counteracting force).

The problem is that at the beginning of the simulation, I get a higher force acting on the rod than on the rest of the piston, so my initial drag force is negative (-2000N) which doesn't make any sense to me. And this offset continues for the rest of the simulation (cf drag force curve (blue) / yellow curve : velocity).

Does anyone have any tips for me on how to estimate the flow losses properly, please?
That would be a great help, thanks a lot in advance!

LuckyTran · January 13, 2022, 13:01

You have two periodic functions that are out of phase and are trying to compare the differences in their amplitudes. You just need to go into the right representation. For example, you could calculate the "abs" of the transfer function.

jojoFD · January 13, 2022, 13:05

Thanks for your answer.

The velocity is only plotted to know where I am at in the simulation. It's the 2000N offset at t=0s that puzzles me.

jojoFD · January 13, 2022, 13:13

And also having an "abs" value would rob me of the direction of the force, which is also an important component for me in the case of any occurring suction effect / compression effect. Don't you think?

LuckyTran · January 13, 2022, 13:14

You missed the reason that I put "abs" in quotes. I'm talking about the modulus of the complex amplitude. You can still get the phase from the complex matrix. Anyway you have two things to characterize the damping.

Is t=0 truly the start of your simulation? from initial conditions? That's not going to be the same for a statistically stationary case because you guess the initial condition that isn't the same as the actual solution. That is, you need to let the piston oscillate a few times first.

jojoFD · January 14, 2022, 05:26

Oh right, didn't get that, thanks for clarifying. But I don't see how that would work..

Little clarification on my side, I used the term "damping force" when I am really looking for the drag force. But as the damping is connected to the viscous drag, I supposed I could get one with the other.

Yes, I had that in mind as a potential source of errors, but I ran previous cases with more oscillations and the offset is still here so it was decided that for the drag force I'm trying to measure + the dynamic layering technique used, the first oscillation would be good enough for now.

jojoFD · March 11, 2022, 05:40

*** Update on this problem ***

I ran a new piston motion simulation with a very low velocity. And it appears that the way I calculate the drag force (force on the piston - pressure on the rod * rod surface) give a constant value at a very low velocity as well. And it exactly corresponds to the initial drag value at t=0s I get for a high speed velocity. If I substract this initial value to all the values I get during the whole piston motion at high speed, the drag curve is then perfectly centered.

Another thing is that between the cases at different velocities and mesh, this initial drag value changes as well. So this phenomenon is still a mystery...but looks like a static part to be removed or initialized.

If anyone understands this better than be, I'd be keen to know more

thanks !

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
CFD Design...The CFD Future	John C. Chien	Main CFD Forum	20	November 20, 2015 00:40
Force can not converge	colopolo	CFX	13	October 4, 2011 23:03
ASME CFD Symposium, Atlanta, July 2001	Chris R. Kleijn	Main CFD Forum	0	September 13, 2000 05:48
CFD for fans & blower housings	David Carroll	Main CFD Forum	8	August 24, 2000 18:25
ASME CFD Symposium, Atlanta, July 2001	Chris R. Kleijn	Main CFD Forum	0	August 21, 2000 05:49

January 13, 2022, 13:01		#2
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	You have two periodic functions that are out of phase and are trying to compare the differences in their amplitudes. You just need to go into the right representation. For example, you could calculate the "abs" of the transfer function.

January 13, 2022, 13:05		#3
jojoFD New Member Join Date: Mar 2021 Posts: 18 Rep Power: 5	Thanks for your answer. The velocity is only plotted to know where I am at in the simulation. It's the 2000N offset at t=0s that puzzles me.

January 13, 2022, 13:13		#4
jojoFD New Member Join Date: Mar 2021 Posts: 18 Rep Power: 5	And also having an "abs" value would rob me of the direction of the force, which is also an important component for me in the case of any occurring suction effect / compression effect. Don't you think?

January 13, 2022, 13:14		#5
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	You missed the reason that I put "abs" in quotes. I'm talking about the modulus of the complex amplitude. You can still get the phase from the complex matrix. Anyway you have two things to characterize the damping. Is t=0 truly the start of your simulation? from initial conditions? That's not going to be the same for a statistically stationary case because you guess the initial condition that isn't the same as the actual solution. That is, you need to let the piston oscillate a few times first.

January 14, 2022, 05:26		#6
jojoFD New Member Join Date: Mar 2021 Posts: 18 Rep Power: 5	Oh right, didn't get that, thanks for clarifying. But I don't see how that would work.. Little clarification on my side, I used the term "damping force" when I am really looking for the drag force. But as the damping is connected to the viscous drag, I supposed I could get one with the other. Yes, I had that in mind as a potential source of errors, but I ran previous cases with more oscillations and the offset is still here so it was decided that for the drag force I'm trying to measure + the dynamic layering technique used, the first oscillation would be good enough for now.

March 11, 2022, 05:40		#7
jojoFD New Member Join Date: Mar 2021 Posts: 18 Rep Power: 5	* Update on this problem * I ran a new piston motion simulation with a very low velocity. And it appears that the way I calculate the drag force (force on the piston - pressure on the rod * rod surface) give a constant value at a very low velocity as well. And it exactly corresponds to the initial drag value at t=0s I get for a high speed velocity. If I substract this initial value to all the values I get during the whole piston motion at high speed, the drag curve is then perfectly centered. Another thing is that between the cases at different velocities and mesh, this initial drag value changes as well. So this phenomenon is still a mystery...but looks like a static part to be removed or initialized. If anyone understands this better than be, I'd be keen to know more thanks !