Happy to admit when I'm wrong.

Dimensions for the x and z for all cases are the same - and after a second review, I agree - even though the main branches look odd, they are dimensionally accurate.

The only hiccup then would be the small branches, but those are unimportant for our purposes.

Consequently, it seems the only way to "fix" this currently is to be prepared for this condition, or split the model into two different CAESAR files. Unfortunately, the manual split within ISOGEN does not currently appear robust enough to handle this, as it seems to crash. Sometimes it works, sometimes not - perhaps it has something to do with multiple pipe circuits within the system. (Using the term "circuit" to distinguish it from the term "loop," so as not to confuse with expansion loops.)

If/when it comes up, I'll speak up.

Thanks.