Modeling scatter from anodized surfaces

Anodized surfaces can result in such variable BSDF behavior that there isn't an appropriate "standard" model. It will depend heavily on the wavelength, anodization process, and the underlying surface (material, smooth, blasted, milled, etc.). 

For example, the picture below shows the BSDF from two different anodized surfaces on aluminum substrates, based on measurements at 1550 nm and shows how variable the BSDF can be.

It is not uncommon that you see black anodized parts that are perfectly diffuse at small angles of incidence but then exhibit an additional specular component as you get to high angles of incidence (right-side graph). Modeling such a surface treatment in FRED would need both a scatter model as well as a coating model (where the coating would have an AOI dependence with 0 reflectance at normal incidence but non-zero reflectance at higher angles).

In summary, for an accurate model you'll need to either measure a sample, or find out details about your specific process and then hopefully find some equivalent BSDF data in a published paper.