Evaporate a solution using PHREEQC

[Tom]

Q. How can I simulate evaporation of a solution in PHREEQC? I am looking to investigate the minerals which are likely to precipitate.

A. Evaporation can be simulated using the REACTION keyword:

In the example below, 1 kg of a hypothetical solution (100 %) is evaporated.

----------------
SOLUTION 1
pH 7 charge #adjust pH in accordance with charge balance
units ppm
S(6) 1700
Mg 50
Ca 200
Fe 10
Si 10
water 1.0 #1 kg of water

REACTION 1
H2O -1.0 # -1.0 is the relative moles of the reaction.
55.51 moles in 10 steps #1 kg of water is 55.51 moles. 5.551 moles H2O removed per step (=100 ml H2O per step)
END

--------------

From the output you can assess the SIs of phases at each step.

[Charlie]

Adding this block will help assess the reaction progress:

SELECTED_OUTPUT
-reset                false
-step         true
-water               true

[dlparkhurst]

For high evaporation factors, PHREEQC can fail to find the amount of water left in solution, and then the numerical method fails. Another approach that can be useful is to fix the activity of water with a pseudo phase as in the following script.

SOLUTION
   Na   1
   Cl   1
END
PHASES
water
    H2O = H2O
    log_k     0
END
USE solution 1
EQUILIBRIUM_PHASES 1
    water     -0.1 10
END

[liamabrahamsen]

That's an interesting approach. What effect does introducing the pseudo water phase have? Doesn't it simply replace the water that was removed in the evaporation step?

[dlparkhurst]

You could have the same effect using the H2O(g) phase, but the target saturation index would be a bit harder to set up. With H2O(g), the target saturation index is equal to the log of the partial pressure of water that is desired. Whereas, the "water" phase simply removes (or adds) water until the log activity of water is equal to the target saturation index.

[MichaelZ20]

Should we consider reliable the results of modeling strong evaporation when the ionic strength exceeds 6-10?

[dlparkhurst]

The caveats related to high ionic strength always apply. Pitzer.dat would be best, perhaps followed by sit.dat, followed by the other ion association models, in order of reliability.

The ionic composition of the solution makes a difference. Pitzer parameters are fit over a range of concentration, and some elements may have been fit only to a lower ionic strength. Ion association models often rely on activity coefficients fit from chloride and sodium salts, so they will work best in a NaCl medium.