[JLoenartz]

Dear colleges,


I came across some challenges, when I tried to simulate a pressure surge inside a closed domain. Attached is an image, that should give an idea, what I am trying to simulate and I will use this image to explain the problem.

In a transient calculation using rhoPimpleFoam, I want to simulate the flow into a sealed off domain, after a valve is abruptly opened. While the valve is closed, we have ambient pressure and temperature inside of the fluid-domain. When the valve is opened, air flows into the fluid-domain from a pressure vessel (pressure of 300bar and 60°C). The opening process of the valve is to be considered instantly. There is no outlet and the simulation is supposed to run, until there is an equiblibrium inside the fluid-domain.



The pressure-vessel is not modelled explicitely, instead I want to put the inlet on a cross-section of the pipe, as shown in the attached image.

My first questions are concering the boundary conditions. A velocity-inlet is of course out of the question, because at this point, the celocity at the inlet is unknown.



My idea was, to use a combination of the following boundary conditions on the inlet:


p:

    inlet
    {
        type    totalPressure;
        p0      uniform 300e+5;
        value   $internalField;
    }

T:

    inlet
    {
        type        totalTemperature;
        T0          uniform 333;
        gamma       1.4;
    }

and U:

    inlet
    {
        type    pressureInletOutletVelocity;
        value   $internalField;
    }

In theory, these bc's should account for the decrease of pressure and temperature, when the air accelerates while flowing into the pipe.


But I obtained some really questionable results from this set of boundary conditions. The flow reached velocities of more than double the speed of sound in the pipe, which should not be possible and temperatures fell to 40K. And after a short while the calculation crasehd, because of negative initial temperatures.



My next thought was, to approximate the flow conditions by using codedFixedValue for the boundary conditions, but this should prove very complicated and feels like guessing. So I thought I might ask in the forum, if maybe someone has encountered a problem like this before and has a reliable solution or setup ready.

Thank you very much in advance,
Jonas