[mazdak]

Hello,

I'm trying to model liquid flow in a piston pump. When the piston pushes the flow out of the chamber, the suction is closed and when the piston reverses its motion, the suction is open. So, I use the following BCs:

first half of the cycle:
inlet (suction is closed) = wall
outlet (discharge is open) = pressure outlet (a very very high pressure)

second half:
inlet (suction is open) = pressure inlet (a very very low pressure)
outlet(discharge is closed) = wall

My questions:

1- When I run the case, at the moment when the discharge is opened, I get very high velocities in the domain like shock waves. This means that since the chamber was filled with a low pressure liquid, for the flow to overcome the very high pressure at the outlet, it needs to have a high velocity. Is this physical?

2- what happens if I treat the liquid as a compressible liquid?

3- What should I do for the following warnings: reverse flow and turbulence intensity set to 10E+05?

Sorry for the lengthy post.

[Hamidzoka]

Hi;
It is impossible to pressurize a liquid just by a moving piston cause liquids are incompressible. What a piston pump (and also other kinds of pumps!) does is increasing kinetic energy of the liquid through a moving piston (or a rotating impeller). This increased kinetic energy may turn into pressure downstream the pump if necessary. An example would be a liquid being pumped to an elevated tank.
So, compressible models will not be helpful since the fluid will not experience any change in its density unlike a gas in a similar domain. This will only adds to complexities of your model.
Fluid velocity is expected to be function of downstream conditions, piston velocity and dimensions as well as chamber geometry. Why you think there must be a shock? what is the velocity levels? What about outlet boundary conditions? How is it set?
Maybe a schematic graph of your domain chan help.

Regards

[FMDenaro]

As first approximation, you could try to understand what happens using the sound velocity of the liquid (higher than gas but a finite value) and start with the one-dimensional theory for omoentropic flows with a moving piston (see the book of Zucrow).

However, high velocity does not automatically implies a shock.

[mazdak]

thanks for your reply.

As you can see, when the piston moves towards right to push the flow out of the chamber, non physical velocities appear. The way in which I'm modeling the
flow is by switching the BCs at the inlet and outlet.

piston moving towards right: inlet = wall; outlet = pressure outlet (10000 psi)
piston moving towards left: inlet = pressure inlet (100 psi); outlet = wall

the motion of the piston is sinusoidal modeled by dynamic mesh layering.

it's interesting that I have modeled the same thing in a 2d geometry almost identical to the 3d one and had no problem. The time step is 0.001 s.

[mazdak]

I have also added a picture of the mesh used.