Smoke Disappears due to mesh size difference

[IFAB-AJ]

Hello,

I have a simple test file here which I ran with FDS 6.9.1-1751. The bottom side has a mesh size of 6cm and the top side has a mesh size of 3cm. As you can see in the image, the smoke disappears as it enters the area where the mesh size decreases to 3cm. I also attached the fds file here.

Mesh_Test_V2.txt

[gforney]

fds converts soot density to an opacity, smokeview uses to draw smoke, using a formula something like
alpha = 255*(1.0-exp(-KSDX)) where K is conversion factor, S is the soot density and DX is a path length. these alpha opacities are integers ranging from 0 to 255. for your case, the smoke is very thin. for your upper mesh the DX is cut in half, this is enough to cause all alpha's to be zero.

we are thinking about the best way to solve this in the mean time a work around is to coarsen the grid in the X direction, change 100 to 50 for the number of grid cells in the X direction .

[IFAB-AS]

Hello,

we are using the thin smoke to represent the test setup for smoke detection for Rail vehicles according to ARGE guideline part 1 (used in Germany, but also in many other countries used to prove early detection). The target is to find good positions for smoke detection under various conditions before the vehicle is build. We did get the soot yield by modelling the test setup from the latest ARGE (only German text available, this image is translated):

The test for a type A smoke source is successful if:

• ambient temperature is between 20-25 °C, humidity 30-70%
• smoke detector in position 1 is detecting the smoke with a range of 15 - 35 s and detector at position 2 in range 20 - 40 s.
• transmission measurement <95% between 60 s - 120 s
• Increase in temperature at Position 1 after 60 s shall not be > 20 °C

In the real test, a small pan of Methanol is used (169 cm²) together with an aerosol generator. We try to represent this in the FDS simulation with a small Methanol fire (approx. 5 kW) and give this fuel a soot yield in such a way that the criteria above are met.

This fire/soot is placed in the vehicle simulations afterward. As the ceiling structure is important (HVAC, ceiling is not flat, other devices) we used the small grid with 3 cm at the ceiling (we did show convergence for this value in the past for a different vehicle).

If it is possible to show the smoke in SMV, this would be good for searching errors and the "visual experience". Would it be possible to include a factor that scales the term in brackets, like alpha = 255*(factor*(1.0-exp(-KSDX))) or 255*(1.0-exp(-KSDX)* factor) or 255*(1.0-exp(-KSDX/factor))?

[drjfloyd]

A slice file of optical depth or soot density would give you a quantitative measure of the amount of soot present (with smoke 3d in smokeview the fact that you see a small amount of smoke there doesn't tell you if the amount is likely to cause detection).

[mcgratta]

We have made some progress on this issue and we are testing it now. The key point is how the soot density information is conveyed from FDS to Smokeview using 1 bit integers (0-255). We have found that transferring the information directly as a density rather than an "opacity" provides us with a greater range of visibility.