[Ben0000]

I'm trying to simulate the effect of wind on building, specifically the the pressure distribution on the surface of the building. I've had success with a constant wind velocity with respect to time, attached is a screenshot of the results, but when I try to make the wind velocity change with time I'm running into some issues. The problem is that I'm getting a pressure gradient across the entire domain (see other attachment). This gradient is then being applied onto the building surface, and essentially overwhelms any other pressure that occurs not as a result of the gradient. This gradient also changes significantly between different time steps.

At first I thought this was nonphysical behaviour, but looking at the Navier-Stokes Equations I can see that a non-zero du/dt will lead to some constant dp/dx, i.e. a pressure gradient. I'm using the Von Karman wind turbulence model to generate the time-varying wind velocity, and I've checked this against real-world data to confirm that the acceleration (du/dt) has a realistic magnitude.

My question is how would I go about modelling the time-varying wind around the building without the pressure gradient, or if the pressure gradient is necessary, without the pressure gradient applied onto the building? Currently, it seems that the entire domain changes velocity simultaneously, would it be possible to model some sort of gust generator that generates these varying wind speeds upwind of the building, and they propagate through the domain and around the building?

Below are my current boundary conditions for p and U (freeWall refers to the left, right and top boundaries of the domain).

p:

dimensions      [0 2 -2 0 0 0 0];

internalField   uniform 0;

boundaryField
{
    inlet
    {
        type            zeroGradient;
    }
    outlet
    {
        type            fixedValue;
        value            uniform 0;
    }
    freeWall
    {
        type            slip;
    }
    floor
    {
        type            zeroGradient;
    }
    buildingWall
    {
        type            zeroGradient;
    }
}

U:

dimensions      [0 1 -1 0 0 0 0];

internalField    uniform (45 0 0);

boundaryField
{
    inlet
    {
        type            timeVaryingMappedFixedValue;
        setAverage        off;
    }
    outlet
    {
        type            zeroGradient;
    }
    freeWall
    {
        type            slip;
    }
    floor
    {
        type            noSlip;
    }
    buildingWall
    {
        type            noSlip;
    }
}

[joshwilliams]

A flow is driven by pressure gradients. For example, if you have incompressible flow in a pipe with pressure = 0 Pa at the outlet and 0 Pa at the inlet, the fluid within the pipe will be stationary. It is only a difference/gradient in pressure between the inlet and outlet that creates flow. So, there is no way to simulate your problem 'without a pressure gradient' if I understand you correctly.