Can someone make it clear to me? What pressures do I have to transfer from CFD to CSD for a fluid - structure interaction problem? Is it the static pressure, the dynamic pressure or what?

dynamic would involve the square of the velocity of the flow.

I know that. My question is if I have to transfer the static pressure or the total pressure (static + dynamic). Thanks

You transer the surface stress - which is taken from the stress tensor involving the pressure p and the viscous stresses.

Hi Mihai,

the velocity is zero at the wall if you are computing a viscous flow, e.i. pstat=ptot at the wall. Hence you should transfer pstat (or ptot - as you like).

Hope that is of any help.

Good luck, Alan

The velocity may be zero but the shear stress is not. The surface traction is the correct quantity to transfer, and this includes the pressure.

Total pressure (This is the effect of a moving fluid on a solid stationary boundary)

Ag wrote: "The velocity may be zero but the shear stress is not. The surface traction is the correct quantity to transfer, and this includes the pressure."

I would tend to agree with 'ag' that it is necessary to transfer the 'forces' between liquid & solid sub-systems. Treat the fluid & solid separately & use force as the linking mechanism.

Pressure, velocity etc are contained within the fluid subsystem. Use them to derive their effect on the solid surface. Then focus yourself on the solid sub-system subjected to the external forces imposed by the fluid.

diaw...

From all the answers I conclude that the Ag's answer is the most pertinent. I do need to transfer the surface forces which include static pressure and viscuous forces. Velocity is zero on surface so dynamic pressure is zero (we can not talk about dynamic pressure in this case. the term is available only in free stream). Still we do have viscuous forces. So I need to find a way to substract viscuous forces at surface from FLUENT.

Ag wrote: "The velocity may be zero but the shear stress is not. The surface traction is the correct quantity to transfer, and this includes the pressure."

I would tend to agree with 'ag' that it is necessary to transfer the 'forces' between liquid & solid sub-systems. Treat the fluid & solid separately & use force as the linking mechanism.

Pressure, velocity etc are contained within the fluid subsystem. Use them to derive their effect on the solid surface. Then focus yourself on the solid sub-system subjected to the external forces imposed by the fluid.

diaw...

You can compute all stresses on the surface using the velocity evaluated at the baricentre of the element. Imagine that you have some grid discretized with tetrahedra, then you'll have 3 (or 2) nodes touching the surface, and these nodes will have null velocities, while there will be 1 (or 2) nodes left with the velocity computed in the fluid flow near the wall. Therefore, you can use the area of the face of these elements (face that touches the surface only) and its normals to compute the forces.

(ps. I think this method to compute stresses on surfaces is not so accurate but it works...)

Regards

Renato.

I agree.

>>Ag wrote: "The velocity may be zero but the shear stress is not. The surface traction is the correct quantity to transfer, and this includes the pressure."

I would tend to agree with 'ag' that it is necessary to transfer the 'forces' between liquid & solid sub-systems. Treat the fluid & solid separately & use force as the linking mechanism.

Pressure, velocity etc are contained within the fluid subsystem. Use them to derive their effect on the solid surface. Then focus yourself on the solid sub-system subjected to the external forces imposed by the fluid.

diaw...

I couldn't agree any more with diaw

If you don't have the shear stress available accurately, you may as well forget about it. The significance of shear stress versus static pressure is small for most applications. You can start out by only considering the normal stress, i.e. static pressure, make it work first, and then worry about shear stress later on. This is assuming that you are looking at wing or panel flutter, or something along that line, where pressure is really the dominant factor.

The question about which pressure to use has been answered correctly, but I would like to add a simple reminder. The only pressure that really matters is the only "real" pressure which is static pressure. Dynamic and total pressures are concepts, not real thermodynamic and fluid dynamic (primitive) quantities. For example, the total pressure is defined as the pressure that you would get in an "imagined" idealized isentropic stagnation to zero velocity. Neither dynamic nor total pressure can directly be measured in a flow, they have to be converted to static pressure (e.g. by stagnation in front of a pitot tube) or obtained from velocity and density. The static pressure and shear stress are the quantities that enter the structural equations.

From an engineering point of view, what is the effect of a shear stress (that might be 1 or 10 Pascals) compared to the total pressure (101325 Pa or more) This a good cause for numerical divergence?

Ahmed wrote: From an engineering point of view, what is the effect of a shear stress (that might be 1 or 10 Pascals) compared to the total pressure (101325 Pa or more) This a good cause for numerical divergence?

Be careful here. You are referring to Total Pressure in ABSOLUTE terms ie. P + Pref(=101325Pa), this is why it appears to be dominant. If you set the model pressures to be RELATIVE, then Pref is deducted. The shear-stress & pressure terms will now be more comparable. It could very well be that the stress terms now appear to be dominant.

It very much depends on the flow regime & fluid you are considering.

To compute surface shear stresses is no big deal. Calculate the surface velocity gradient (dv/dy) ~ delta_v/delta_y for first cell next to wall. delta_v = cell-centred velocity - 0 for no slip, no suction/blowing wall. delta_y = height of cell centre above wall. Plug in the viscosity & 'bingo' you have wall shear stress. Actually, many packages allow you to extract this boundary information directly.

A lot will depend on what flow-regime you are working with - low-speed incompressible flow, or high-speed inviscid flow (boundary-layer).

diaw...

diaw wrote: Calculate the surface velocity gradient (dv/dy) ~ delta_v/delta_y for first cell next to wall. delta_v = cell-centred velocity - 0 for no slip, no suction/blowing wall. delta_y = height of cell centre above wall. Plug in the viscosity & 'bingo' you have wall shear stress. Actually, many packages allow you to extract this boundary information directly.

Should read: (du/dy) ~ delta_u/delta_y for first cell next to wall. delta_u = cell-centred velocity - 0 for no slip...

Apologies... finger trouble...

diaw...

Diaw The answer to your comments can be developed in a scientific fashion, to start with I invite you to the following, first do you agree that The Navier Stokes equation describes the behaviour of fluid dynamics? the answer is yes, ok then what is the meaning of that "P" that appear in that equation and solved by all CFD programmes that I have come across during the last few years of my professional life? (Continuum Mechanics by Mace and Mace is an excellent reference about the Navier Stokes equations, also a very nice discussion about P can be found in the excellent books by Schlichting or Panton) 2- Now let us assume you have spent few years doing experimental work and was using a pressure transducer to measure the pressure (No hot wire), so what pressure were you reading? Stagnation? total? or what?

Ahmed wrote: Diaw The answer to your comments can be developed in a scientific fashion, to start with I invite you to the following, first do you agree that The Navier Stokes equation describes the behaviour of fluid dynamics? the answer is yes, ok then what is the meaning of that "P" that appear in that equation and solved by all CFD programmes that I have come across during the last few years of my professional life? (Continuum Mechanics by Mace and Mace is an excellent reference about the Navier Stokes equations, also a very nice discussion about P can be found in the excellent books by Schlichting or Panton) 2- Now let us assume you have spent few years doing experimental work and was using a pressure transducer to measure the pressure (No hot wire), so what pressure were you reading? Stagnation? total? or what?

-------- Greetings Ahmed,

Let's begin opening the 'can of worms'...

Firstly, in terms of P - for now, let's refer to it as direct normal pressure on the face of a fluid element. (There are a number of different ideas on this which we can all explore together). The issue to be careful of when simulating is to not let the relative 'numerical magnitude' of the 'absolute pressure' swamp the other stress terms in the calculation. We therefore elect to set a relative reference pressure of zero (pressure is after all a 'potential' - so we can reset our simulation reference to any number we choose for the duration of the simulation & correct it later by adding back the reference). This is what I was refering to in earlier posts.

--------------- Now to your second question: "do you agree that The Navier Stokes equation describes the behaviour of fluid dynamics? "

To be honest - not entirely!!!

1. There are singularities inherent in the N-S equations, which may, or may not be real in nature. My current research program is exploring these in more detail.

2. In the derivation of the N-S equations, we use a merry mix-up of reference frames - continuity=> fixed reference-frame; Momentum & energy=> 'moving' reference frame, corrected via 'convection terms' to a fixed (Eulerian frame). How does this 'moving reference frame' move? What are its characteristics?

--------- To your third question: "Now let us assume you have spent few years doing experimental work and was using a pressure transducer to measure the pressure (No hot wire), so what pressure were you reading? Stagnation? total? or what?"

I would imagine more of 'or what' than anything else. The dynamics, or otherwise of a pressure membrane may present their own 'disturbance' issues. It also depends on the orientation & placement of the pressure-sensor (?damper?).

Back to you...

diaw...

There are two well established views to define what is what we call pressure, and they are 1- Microscopic view Matter is composed of molecules that are continously moving and in their movements they hit themselves and the walls of a boundary. Momentum exchange between these molecules and the boundary is what we call pressure (Kinetic theory of gases, Statistical Thermodynamics) 2- Macroscopic view Pressure is a property of matter that satisfies its equation of state, even if we do not know the exact mathematical form of this equation of state. (Water is such a matter and that is why we have the steam tables) The simple equation of state of an ideal gas or its modifications to a real gas are good examples (Continuum Mechanics,Classical Thermodynamics) No matter which view you like, the pressure is measured in absolute terms. In the early development of engineering practice, people used to talk about gauge pressure (the quantity that can be measured by attaching a guage to a wall) and this practice continue to be today as it was yesterday. CFD,on the other hand, is a relatively new comer (When I started my post graduate studies it was an academic talk, no FLUENT, No StarCD etc...). When developers of commercial CFD programmes started to prepare their programmes for wide commercial use they have to adjust to their clients needs and uses and one of these adjustments is how to introduce pressure values, simply said, would they require the pressure to be introduced using the absolute values and limit the use of their programmes to a limited number of researchers or use the common units and open their programmes to as many clients as possible. As you understand it, the use of gauge values won the battle and today all programmes have the reference value of 101325 Pa (or 14.7 psi) embeded in them though the user can change this value if he/she wishes to Zero but in this latter case he/She has to introduce the pressure values in absolute units. My friend as you can see the subject will need pages and pages to give it full account, the following two references will give you more details and hope they will clarify your understanding about the term pressure 1- Atomic Physics by Nobel winner Max Born has an excelllent account about the Kinetic Theory 2- Thermodynamics Foundations and Applications by E.P. Gyftopoulos and Gian Paolo Beretta has a profound explanations of Classical Thermodynamics Wish you look at them before you post your reply. Wish you luck with your research and let us know about the results