Hydrogen dissolution in brine

[ylg]

Dear all,
I would like to estimate the dissolution of hydrogen in brines (3 and up to 5M NaCl @50°C and P from 15 to 150bars).
I could model this based upon ex22 from Phreeqc (V3) manual with phreeqc.dat database. However, it does not work with Pitzer database even when replacing H2(g) by Hdg(g). I wanted to compare the difference induced by the 2 databases. Here is the input file:

DATABASE Phreeqc\database\pitzer_modified.dat
TITLE H2 solubilities at high H2 pressure

SOLUTION 1
GAS_PHASE 1
    -fixed_volume
    H2(g) 0
    H2O(g) 0
REACTION
 H2(g) 1;
 0 27*1
INCREMENTAL_REACTIONS true
REACTION_TEMPERATURE 1
        25.0

SELECTED_OUTPUT 1
    -file                 H2_Water_Pitzer_25.out
    -temperature          true
    -water                true
    -totals               H H(0) H(1)  O 
    -gases                H2(g)  H2O(g)
END

I tried to add H2(g) to the gas section of pizter.dat  (copied from phreeqc.dat)
But in the output file there is no addition of H2
Reaction 1.   

     1.000e+00 moles of the following reaction have been added:

                    Relative
   Reactant            moles

   H2(g)                1.00000

                    Relative
   Element             moles
   H                    2.00000

-----------------------------------Gas phase-----------------------------------

Total pressure:  1.00      atmospheres
    Gas volume:   1.00e+00 liters
  Molar volume:   1.59e+05 liters/mole

                                                        Moles in gas
                                            ----------------------------------
Component                log P           P     Initial       Final       Delta

H2(g)                   -22.05   8.934e-23   1.787e-26   0.000e+00   0.000e+00
H2O(g)                   -1.50   3.142e-02   6.284e-06   6.284e-06   0.000e+00

Does anyone has a clue on how to solve this issue?
Thanks

[dlparkhurst]

Pitzer.dat database has limited redox capabilities. I think you are best to use "Hdg", which is defined as a dissolved aqueous species and a gas component (Hdg(g)). Here is your file adjusted to run with pitzer.dat. Note that Hdg represents H2, two hydrogen atoms.

Code: [Select]

SOLUTION 1
GAS_PHASE 1
    -fixed_volume
    Hdg(g) 0
    H2O(g) 0
REACTION
 Hdg 1;
 0 27*1
INCREMENTAL_REACTIONS true
REACTION_TEMPERATURE 1
        25.0
SELECTED_OUTPUT 1
    -file                 H2_Water_Pitzer_25.out
    -temperature          true
    -water                true
    -totals               Hdg
    -gases                Hdg  H2O(g)
USER_GRAPH 1
    -headings               H2 H2(aq) P(H2(g))
    -axis_titles            "H2 added, moles" "H2(aq), mol/kgw" "Pressure, atm"
    -initial_solutions      false
    -connect_simulations    true
    -plot_concentration_vs  x
  -start
10 GRAPH_X SYS("Hdg")
20 GRAPH_Y TOT("Hdg")
30 GRAPH_SY PR_P("Hdg(g)")
  -end
    -active                 true
END

[ylg]

Thanks David for your very hepful comment.
Using  Hdg enables correct match of experiments. However, I notice a significant difference with phreeqc database when using Hdg(g) or H2(g). The run  with H2(g) and phreeqc database does not match the experiment. There is about twice as much dissolved H(0) than Hdg (see attached graph)

In addition, I noticed that with Hdg(g) there is always 0 for g_Hdg in the export despite some Hdg(g) in the ouptut file pqo (Moles in gas)

[ylg]

Hi David,
First I would like to indicate that the export of g_Hdg is correct. My apologies for this erronous statement.

In addition to my preview reply, I did run the case with 5M NaCl Brine with Hdg using pitzer and phreeqc database as below:

DATABASE C:\Phreeqc\database\pitzer.dat
SOLUTION 1
   units mol/kgw
   Na 5
   Cl 5
GAS_PHASE 1
    -fixed_volume
    Hdg(g) 0
    H2O(g) 0
REACTION
 Hdg 1;
 0 27*1
INCREMENTAL_REACTIONS true
REACTION_TEMPERATURE 1
        25.0

SELECTED_OUTPUT 1
    -file                 Hdg_dp_Brine-5M_Pitzer_25.out
    -temperature          true
    -water                true
    -totals               H H(0) H(1)  O  water Hdg
    -gases                Hdg(g)  H2O(g)
END
 
The results with pitzer database are not consistent with the experiment while those using phreeqc database are in good aggreement (see attached file). This is rather counter intuitive as I though pitzer database was better suited for this case.
Do you have some explanation?

[dlparkhurst]

I think the difference is the activity coefficient for Hdg(aq). The activity coefficient for Hdg(aq) in phreeqc.dat is log10(gamma) = 0.1I (I=ionic strength).

There are no interaction parameters for Hdg in pitzer.dat, so the activity coefficient is 1.0. Based on CO2(aq), I think the parameters lambda and zeta would be used for uncharged species in the Pitzer approach, but you would have to check if you want to pursue it.

[ylg]

Thanks David for the inside
As I am not a specialist of pitzer model, could you advice on some useful reference (introduction and description) and relevant experiment or parameter estimation?
Best regards,

[dlparkhurst]

This reference gives most of the equations, but you can use the reference list to find the original papers. I think you will need to track forward to find more references related to uncharged species.

https://pubs.er.usgs.gov/publication/wri884153

[ylg]

Thanks david for the reference.
Meanwhuile, i look in more details into the phreeqc warning which indicates a non convergence issue which may explain the results obtained with pitzer database.
Using
KNOBS
-logfile true
the logfile contains several warnings such as

Iterations in pitzer_revise_guesses: 1
Ineq had infeasible solution, kode 0, iteration 4
....
Ineq had infeasible solution, kode 0, iteration 19
WARNING: Numerical method failed, switching to numerical derivatives.

Number of infeasible solutions: 9
Number of basis changes: 0

Number of iterations: 113
Number of gamma iterations: 3

This looks like non-convegence issues?
does it means Phreeqc is computing derivatives which do not take into account the solution as changing from 5M to pure water does not change the solution using pitzer database?
I obtain the same results and same warnings with other piter database (from Yucca Mountain project) as soon as introducing Peng-Robinson EOS (with pc, tc, omega for H2).
I'll appreciate your view on this?
Regards,
Yann

[dlparkhurst]

If there is no "ERROR" message, the simulations completed successfully. PHREEQC takes several approaches to finding convergence; the WARNING messages indicate that some of them failed. It will say something like ERROR:" failed on all sets of convergence parameters" if none of the approaches is successful.