Re: Aqueous species oscillations within transport calculation

["David L Parkhurst" <dlpark@xxxxxxxx>]

> 1. Equilibrium controlled dissolution
> To a first approximation, the Phreeqc calculation shows similar values
to the experimental concentrations of the effluent water collected the
the end of the laboratory core.  However, the aqueous species
concentrations oscillate rapidly over short timescales,  despite the
amounts of minerals remaining constant.  We are puzzled by this and
wonder if you can help with the explanation of this problem: Is there an
unstable mathematical answer to this oscillation, or have we overlooked
a problem with the chemistry?

I don't think it is the result of the numerical methods. I think the
oscillations are caused by changes in mineral assemblages within cells of
the column. As a mineral disappears from one cell, the chemical composition
of the water in the cell will change to a new equilibrium composition,
probably as a step function. This chemical change propogates down the
column. With 32 cells and so many minerals, it is hard to keep track of all
the changes. I would probably set the problem up in PHAST so I could view
the mineral and chemical fronts more easily with ModelViewer.

With the equilibrium assumption and so many minerals, it is possible that
the chemical system is instantly dissolving all of one mineral to make
another because PHREEQC is always calculating the most stable set of
phases. If this is happening, it may be necessary to select the allowed
phases more carefully or use the kinetic formulation.


> Here is some more explanation of the underlying experiment:
A highly alkaline fluid (Solution 0) is advected slowly (3e-12 m3s-2)
through a column containing Boom Clay ( a clay composed of 7 different
(primary) minerals in equilibrium with its pore water and initially at
pH ~8; Solutions 1-32).  The mass of water in each cell is kept small to
mimic the experimental data.  The minerals are allowed to equilibrate
thermodynamically at each shift, and likely secondary minerals are
allowed to precipitate out.

I would scale everything to a kilogram of water; you would have to do this
with PHAST. Your transport is 1 mm per 1.3 days, with a total run time of
about 20 years. Is this correct?

I have included the input file (Ec64a3)  and an excel file (Ec64a3)
containing all the selected output data plotted up.  You can see from
the Excel file that the Na, K, Si, Ca concentrations oscillate rapidly,
while the primary and secondary minerals remain constant over the same
interval.  I used the Llnl.dat database throughout.

> 2. Kinetically controlled mineral dissolution
I have tried the same calculation with kinetically controlled mineral
dissolution.  In this case, Phreeqc does not converge, despite
alteration of the KNOBS parameters.  Have you any suggestions for the
cause of this?  I have attached an input file (k17) which shows this
calculation attempt.

Again, I would use 1 kilogram of water and the corresponding number of
moles of minerals. The kinetics tolerance is in moles, and by using < 1 ml
of water, the corresponding tolerance would be on the order of 1e-12. I
think there will be numerical problems with small numbers.

Enclosed is a beta version of PHREEQC that uses a different ODE integrator
that may be better for the set of kinetic reactions that you have. It has
not been tested thoroughly, so you should do some testing that it is
working properly. I did a little testing with your setup and it at least
runs, but I did not check to see if it made any sense.

This version of the program only uses the kinetic tolerance from the first
reactant (illite in your case). Start with a tolerance of 1e-6 or 1e-5 and
see if the results are plausible. The runs will still be long.

You may need to be careful with redox. Your initial conditions for solution
0 put the iron as ferric; equilibrated solution 50 has ferrous iron. Just
checking if this is what you want. Pyrite may be difficult to include as a
kinetic reactant. Concentrations of iron or sulfide could be very small (<
1e-12) and PHREEQC sometimes has problems with concentrations this small,
especially if kinetic reactions are trying to add mole transfers in this
range. The SR can change widely with tiny increments of reaction.

David

(See attached file: phreeqc.exe)

David


David Parkhurst (dlpark@xxxxxxxx)
U.S. Geological Survey
Box 25046, MS 413
Denver Federal Center
Denver, CO 80225

Project web page: https://wwwbrr.cr.usgs.gov/projects/GWC_coupled

Attachment: phreeqc.exe
Description: Binary data