Rail noise sources - impact noise and contact filter

[espen-wold-adsk]

Hi,

I am looking at your code - and in particular some of your unit tests - in an attempt to use it as an independent verification for the results of my own implementation of CNOSSOS. In the area of rail noise sources I came across a few discrepancies that I would love to get your thoughts on:

• It looks like your code applies the contact filter to the total roughness after adding the contribution from impact noise, while the CNOSSOS documentation seems to imply that the impact noise should be added to the total-roughness-after-application-of-contact-filter (equations IV-6 and IV-10). I verified this by modifying the order of operations in my own code, and thus producing values matching the expected output in some of your tests.

• If I am reading your code correct, you are not making use of a 'JoinDensity' parameter, and rather directly applying what I had interpreted as L_R, Impact-single as per equation IV-11 in the CNOSSOS documentation.

I hope you have time to help me understand these discrepancies!

Regards,
Espen

[pierromond]

Thank you very much for your comment ! We will look at it and come back to you :)

[ochiello]

Hi Espen,

Thank you for your very pertinent comments. You're absolutely right on both issues.

• the contact filter should not be applied to the roughness specific to impact noise; this is an error that we will quickly correct!
• not taking 'joint density' into account is the method recommended for the production of noise maps for the French national rail network (SNCF réseau): the "french" equivalent roughness (L_R_ Impact) was determined with this assumption.

Just one question: which document are you referring to when you mention the CNOSSOS documentation ?

Regards,
Olivier

[pierromond]

Done on 8 october and in NoiseModelling 5.0 ! Thank you