Error message filling XP Sanitary output file

The error message:

===> WARNING !! NO CONVERGENCE IN SUBROUTINE CIRCLE.. ID = 1
ALPHA OR PSI = 8.9900000E-01 THETA1 = 5.7455417E+00

fills most of my Sanitary output file (171 MB!) so that not even Word can open it. I'm already running Tolerance=10^-6, and #Iterations = 75, and the simulation takes 9 hours for 22 years*12 pollutants. Behavior is the same for 10^-7 tol/100 iter. The continuity errors are fine (<2%) for both water and pollutants. How can the source of the error be detected and resolved, or the error messages suppressed? fyi, the last couple lines of CIRCLE subroutine are what generates the error. The condition of Q/Qfull= 0.89 is of course unavoidable during continuous simulation, however. The error messages prevent viewing any of the water quality summary output, as they are placed in the .out file after the input echo but before the useful output.

Thank you,

Gene Bosley
Parsons, Brinckerhoff, Quade & Douglas
Bosley@PBWORLD.COM

ULTRAEDIT and other powerful editors will allow you to open such large files as well as replace text. Maybe a simple program to replace such text with a space or line feed could also be handy.

It is possible that only a section of the model is generating the problem so a run of just parts of the model may isolate the location. Once isolated you may want to then make changes to model geometry to reduce the number of occurrences of the error messages.

Anthony Kuch
XP Software Inc.
anthony@xpsoftware.com

Gene,

The small tolerance and large number of iterations that you are using are probably excessive for Transport.

"I'm already running Tolerance=10^-6, and #Iterations = 75"

The circle non-convergence message is common if: (1) you are using a large time step or (2) you have a lot of instances occuring of the flow in the conduit dropping to zero during continuous simulation. Transport is trying to implicitly solve for the flow at both ends of the conduit and failing. You can fix problem (2) by putting in a small (0.001 cfs should do) constant flow into your model at the upstream nodes of your network. This will prevent Transport from dry out during the continuous simulation and not add a lot of extraneous flow to your model simulation.

Regards,
Bob Dickinson
CDM Inc.
DickinsonRE@CDM.COM