Question about PATH_LENGTH

[Dylan B.]

Hello,

This topic following my old discussion about panel radiation. 

Today, I have a question about the paramter "PATH_LENGTH". 

I'm stuying the impact of a panel radiant on one target at the surface of soil. 

In the FDS guide, we learn that "PATH_LENGTH might be the distance from the fire to the target'  but i don't understand what is, physically, this PATH_LENGTH paramter. 

For instance, if we are in this situation (This is the top view): 

If we want the incident heat flux on our pipeline at the surface of the soil , what distance has to be set on PATH_LENGTH, the middle distance of the pipeline (15m) ?  (the red form is our frontline of fire) or perhaps several simulation must be launch at different PATH_LENGTH in order to have the incident heat flux at different point. 

I'm sorry for technical question but I have read the technical reference and i don"t understand the realistic sense of PATH_LENGTH. 

Thank you in advance if you can help me. 

D.B.

[Kevin McGrattan]

Thermal radiation has a range of wavelengths, each with a unique absorption coefficient. But FDS assumes only one absorption coefficient, which is a combination of all the values. To calculate this effective coefficient, FDS must use a particular PATH_LENGTH. That is, the effective coefficient is appropriate for distances comparable to that. If the distances you are interested in are very different than the default 0.1 m, set the PATH_LENGTH.

[Dylan B.]

Thank you for you answer Kevin. 

So the distance of PATH_LENGTH is the range where the average of absorpption coefficient is validate. 

So if we have a target at 5m and another at 20m, it is better to do two simulation because the distance arn't comparable (I imagine a sensitive study for determine the influence). 

[Kevin McGrattan]

PATH_LENGTH is the distance over which the effective absorption coefficient is optimum. Another way to look at is this -- suppose you have a function 

f(x) = A_1 exp(-k_1 x) + A_2 exp(-k_2 x) + A_3 exp(-k_3 x)

that you want to approximate with only a single exponential term

g(x) = A_eff exp(-k_eff x)

How do you choose k_eff ? You have to choose a certain distance and calculate k_eff that is appropriate for that distance. 

But, yes, do a sensitivity study. Sometimes the results are not sensitive to PL, sometimes they are.

[Dylan B.]

Thank you !  

Its clearer to me now.