Simulation times

[Model1]

Dear,

I would like to couple PhreeqC to an external software.
I was wondering if simulation times can be (significantly) reduced by eliminating the non-required species and reactions from the PhreeqC database, or by using a SELECTED_OUTPUT statement ?

Thanks.

[dlparkhurst]

Phreeqc makes lists of species that are appropriate for the current calculation. I do not think there is too much savings, but perhaps some if you have thousands of species and phases.

SELECTED_OUTPUT does not affect the actual geochemical calculations, but there is time involved in writing a selected output file. The shorter the list of selected-output items, the less processing and writing. My feeling is that you should work first to get working code, and then optimize the slow parts.

Note that IPhreeqc (available at http://wwwbrr.cr.usgs.gov/projects/GWC_coupled/phreeqc/) provides an interface to PHREEQC that can be implemented in C, C++, and Fortran codes. There is also IPhreeqcCOM that will work with Matlab, Excel, and Python (see http://www.phreeqpy.com/).

[Model1]

I have coupled IPhreeqc to Tornado.
As I understand, while using IPhreeqc, there is no reading and writing of input and output files required during data transfer. I was wondering if the Phreeqc database is read at every time step during model simulations or only at the first simulation ?

In my case, using the stand-alone PHREEQC engine, I obtained a high reduction in simulation times when using a reduced database instead of the full Phreeqc.dat database. Hence, this may be very relevant when coupling models...

[dlparkhurst]

IPhreeqc normally only reads the database once with a call to LoadDatabase.,  A second call to LoadDatabase would clear all definitions (SOLUTIONS, EQUILIBRIUM_PHASES, etc), which would make an inefficient coupling.

If you are reading the database at every iteration to run Phreeqc, then the size of the database would make a difference.