I think that you'll not really know what happens unless you model the
whole thing.

You might be able to tweak the whole thing for best performance in
software then. There are likely too many things you'd have to change
to do a manual optimization with the number of elements you're talking
about, but at least you'll know what the pattern is.

I did this for a 2m/432 combo where I hung the 432 4 element off the
nose of the 2m 5 element, but, admittedly, this was a MUCH easier
situation, as the 432 beam by itself had 20dB F/B. The pattern was
changed by proximity to the 2m antenna, but it was an improvement.

I adjusted the spacing of the 432 beam for the best F/B and the 2m beam
ignored the 432 entirely.

73,
Dan
N3OX
www.n3ox.net