Invariant point calculation failes

[Frank Bok]

Hi folks,

 

I've tried to model some invariant point in salt solutions and sometimes the workbench doesn't converge at the invariant point but swaps back and forth between the two mineral phases. Then it ends with messages like:

 

N-R didn't converge after 400 its., maximum residual = 7.65e-11, Xi = 0.4929
Cutting step size to find phase assemblage
Swapping Thenardite in for Mirabilite
Step 87, Xi = .4929 (2 iterations)
-- Can't converge, abandoning path.
-- Xi step is too small

I have attached a .rea example file, the output and a graphic. Maybe you can give me a hint?!

 

Best regards,

Frank

Invariant_point.rea

React_output.txt

[Brian Farrell]

Hi Frank,

 

If you're using GWB10, you have some control over the parameters used in the Pitzer calculations. Using the pitz_precon command (accessible from Config - Iteration) you can change the maximum number of passes through the Pitzer preconditioning loop before the Newton-Raphson iteration begins. This value defaults to 10, but decreasing it can sometimes help the program to converge. If you change the value to 5, your model will run to completion, though it still has some weird oscillatory behavior.

 

You can read more about the new pitz_precon command, as well as pitz_relax and pitz_dgamma in the GWB Reference Manual.

 

Hope this helps get you started,

 

Brian Farrell

Geochemist

Aqueous Solutions LLC

[Frank Bok]

Hi Brian,

 

thank you for the information. Beside of the oscillation I get the invariant point.

 

Best regards,

Frank