[billy7590]

Hi,
I am trying to verify the results of a research paper regarding simulation of toxic gas release. I am using Ansys Fluent Solver.
Basically, first I ran a steady state analysis with wind blowing in a certain direction and certain velocity. Then I added a leakage source with mass flow boundary condition and ran a transient analysis for a gas release for 3 minutes duration (keeping the steady state results as initial condition).

The result that I am looking for is mass fraction of gas at different locations in solution domain. At a height of 1.5 m from ground, the research paper plots its results for a fraction of 2.0e-5. My contour completely matches with the paper, except near the release point.

In the paper, gas dispersion starts at the release point whereas in my case, gas releases disperses ahead of the release point in the forward direction with no dispersion along the plane of leakage source. So my results say that gas is not dispersing in the backward direction whereas in the paper, it disperses even behind the leakage source.

In the attachment, the contours on the left are from the paper. I am simulation the case for wind speed of 0.75 m/s (contour No. 1). My results for the same wind speed are on the right.

What could I be doing wrong? I have modeled everything as recorded in the paper. Please help.

[FMDenaro]

I think that no one can say what is the reason of the difference without the details of the two simulations... Is the grid the same? Is the setting of the BC.s the same? Is the setting of the transient integration the same? Is the formulation the same?

I can only suggest to do also the comparison between the velocity fields

[billy7590]

Originally Posted by FMDenaro

I think that no one can say what is the reason of the difference without the details of the two simulations... Is the grid the same? Is the setting of the BC.s the same? Is the setting of the transient integration the same? Is the formulation the same?

I can only suggest to do also the comparison between the velocity fields

I have tried to keep things identical as much as possible. Grid is also very much similar. So are the boundary conditions and transient integration. Velocity fields are also matching completely. That is why I can't think of any reason that could be causing this difference.

[FMDenaro]

The type and accuracy order of the discretization can also be a relevant difference.

[billy7590]

Originally Posted by FMDenaro

The type and accuracy order of the discretization can also be a relevant difference.

The paper that I am verifying does not give any information regarding discretization. Could you please shed some light on what type of discretization and accuracy selections will give better results ...??