[esp-r] return of the CFD NaN

[Andrew Cowie]

Hi all,

Once again i am getting the dreaded NaNs from the CFD.  So far i have traced them back to the variable EDf; they are first occuring in LISOLV3 when called from CALCED.  As far as i can tell the problem is that SU is getting ridiculously huge (e+35 after the first iteration, then e+36, then e+37) until Fortran freaks out and calls it Inf, which turns C in the E-W sweep in LISOLV3 to Inf, which is somehow instigating the NaNs in PHI (havn't worked out exactly how yet).  I havn't traced SU back through CALCED yet, I just thought I'd check to see if anyone has any idea why this might be happening before i spend another day up until 3am code crawling.

I am working with the model that my previous thread was about, with the slab blockage as an ersatz diffuser.

Thanks in anticipation,
Andy Cowie