PRT texit is negative (unexpected)

[p-ortega]

Hi all,

I updated a model (DISV) from MF6 v5 + PRT (nightly build) to the official 6.6.2 release.

When running PRT using the CBB and HDS files from a previous GWF model, I’ve noticed:

• It fails with 'texit is negative (unexpected)'. The stress period at which it fails depends on the EXIT_SOLVE_TOLERANCE with lower values causing the failure to occur later in the simulation.

Any help on what to look for would be greatly appreciated.

Pablo

[p-ortega]

A really low EXIT_SOLVE_TOLERANCE solved it

[javgs-bd]

Just yesterday we run a PRT model updating from 6.6.1 to 6.6.2 and got results which are notably different. We are in the process of trying to replicate this behaviour in a minimal model to report it, although we don't know if we will success.
I wonder which fundamental change in the updated PRT might be triggering this.

[wpbonelli]

@p-ortega configurable EXIT_SOLVE_TOLERANCE was introduced in 6.6.0, there is some info here. We debated what this should default to — lower is safer but slower. I think the original hardcoded value was 1e-7 where the current default is 1e-5.

What value did you find necessary? And do you see comparable results from the particle tracking model with 6.6.2?

@javgs-bd if your model is also DISV, EXIT_SOLVE_TOLERANCE could come into play. But a more likely culprit is #2281. Some logic determining vertical vs lateral motion was pretty badly broken on DISV grids in 6.6.0-1 (I am the guilty party).

[p-ortega]

@wpbonelli lemme ask but probably yes.

[p-ortega]

Should I reopen this?

[wpbonelli]

@p-ortega awesome, I opened an issue to track this.

[p-ortega]

@wpbonelli Can I have an email to send the files?

[wpbonelli]

wbonelli [at] ucar [dot] edu. thanks for sending.