Filter factor in Digital Filtering Method

[stan_lw]

I have a question regarding the filter factor for the Digital Filtering Method. If I correctly understood the theory Section 1.3 (please correct me if I am wrong), N is the filter width and n is a ratio between a length scale L and grid spacing Δx, i.e., n = L/Δx. It is also mentioned that N should be chosen such that N ≥ 2n. However, in Section 2.2, Table 2.2 specifies that the filterFactor entry should be ≥ 4, i.e., N/n ≥ 4 or N ≥ 4n. Any reason why you implement a larger filter factor (N/n ≥ 4) than required by the theory (N/n≥ 2)?

Also, I think there is a typo regarding the definition of n, where n should be L/Δx (unit-less) instead of LΔx (unit of meter squared) in the documentation.

[ajaybh]

Hey,


Let me have a look at this and get back to you asap.

[pmackenz]

The theoretical requirement is N/n>2 but larger values will increase accuracy.  On the other hand, very large values increase the computational effort for generating the inflow conditions.  N/n >=4 is a recommendation based on experience.