[idrama]

Hey folks,

my questions concerns the inlet BC for velocity and pressure. In CFD books, it is alway explained that by known inflow velcoity the inlet pressure is set to zero gradient. However, when I know the velocity than I know via 0.5*rho*v^2=p the inlet pressure. Can I uses this BC also or do I have to apply zeroGradient always?

Cheers

[PGodon]

>>> my questions concerns the inlet BC for velocity and pressure. In CFD books, it is alway explained that by known inflow velcoity the inlet pressure is set to zero gradient. However, when I know the velocity than I know via 0.5*rho*v^2=p the inlet pressure. Can I uses this BC also or do I have to apply zeroGradient always?

A zero gradient means that there is not pressure force at the
boundary. If you don't put a pressure gradient =0 condition at the
boundary and give the pressure there, you may develop a pressure gradient
which in turns will act as a force and increase (or decrease) the velocity
at the boundary, and this might all be inconsistent with the imposed
velocity at the boundary. So usually a zero pressure gradient is in this
case better.

Another thing you need to consider is from the point of view of imposing
mathematically the correct number of BCs.
You are solving a number of equations, and I am not sure what they
are. Possibly equations for the velocities (one in each dimension) and
an equation for the density and possibly one for the pressure/energy.
For each equation you need to know what is the degree of the equation,
I mean are the space derivatives only of first order? or do you have
second order space derivatives? (which means diffusive terms, as for
the inclusion of heat diffusion, viscosity, etc...). For each equation of
1st order you need only ONE BC for each dimension, namely you really
need only to have a conditions given at one end/boundary of the
domain only. For equations of the second order you have to give one boundary
condition at each end/boundary (so TWO BCs for each dimension). If you give too
few or too many BCs you may have numerical instabilities, though these
might not show up because of (artificial) viscosity or because you are
using a low order accuracy solver which damps these instabilities. But
mathematically the number of BCs has to be given that way. To avoid
all instability you need to actually impose these BCs not on v, P and rho
but on the characteristics of the flow (Riemann invariants). But that's
another story all together.

[idrama]

Thanks, PGodon!

For my understanding, it means: When I would have set a velocity inflow BC, say 10 m/s, and at the same time I have set a pressure condition p=1/2*v^2=0.5*10^2=50 then a addtional force of 50 would act on the velocity, i.e. the velocity would increase.

Is my narration correct?

Cheers in advance

[PGodon]

"50" is the pressure, not the pressure gradient...
In the equation for the velocity (or momentum - I am not sure what are the set of equations you are solving here) you will have a term with the gradient of the pressure which acts as a force. If P=50 everywhere near the boundary (inside and outside) then the pressure gradient is zero because the pressure is constant (inside and outside) the boundary - the pressure gradient THROUGH the boundary is to be zero if you don't want any force there. Don't mixed up the pressure and the pressure gradient... I know it might seem confusing as pressure divided by area is force too... but that's not what I was refering too. If the pressure on both sides of the boundary are the same (pressure gradient =0) then these pressure forces cancel each other. If the pressure is not the same on both sides of the boundary (pressure gradient different than 0) then there will be a net resulting force - that's what you want to avoid when setting gradP=0.

[idrama]

Thanks PGodon!

Oh man, it was the entire time in the front of me and I haven't spotted it. Your answer is totally plausible!

Cheers