Ionic composition inversion of water samples

[Mr.white]

Suppose I inject water (Sample A) into a 1000 m deep well, and then pump it back to the surface (Sample B). Can I infer the composition of Sample A based on Sample B? The process of pumping the water to the surface only considers changes in temperature and pressure, without taking water-rock interactions into account, while gas exsolution and mineral precipitation occur during the pumping process.
SOLUTION 1
    temp   25
    pH       8
    pe        4
    redox     pe
    units     ppm
    density   1
    Alkalinity 801 as HCO3
    Ca        12
    Cl        11177
    K         2407.2
    Mg        7
    Na        5616.8
    S(6)      17
    -water    1 # kg

[dlparkhurst]

No, I think you need the composition of sample A to determine the extent of mineral reactions.

[Mr.white]

Sorry for not making it clear. After Sample A reaches the bottom of the well, it reacts with the surrounding rock to form Sample C. Sample C is then pumped up to the wellhead, becoming Sample B. My goal is to use Sample B to infer back to Sample C.

[dlparkhurst]

You can start by using inverse modeling between sampl A and sample B.

You can then try to consider what reactions may have occurred while water moved up the well bore. The most likely reactions in the well bore are degassing and precipitation of calcite (or another scale mineral). You may be able to use forward modeling to use estimated partial pressures or equilibrium conditions to estimate composition C.

[Mr.white]

I would like to use transport modeling to simulate the diffusion of water samples. First, simulate the process of Water Sample A descending from the wellhead to the bottom of the well (transport). Then, simulate water-rock interaction to produce Water Sample B. Next, simulate the transport of Water Sample B from the bottom back to the wellhead to obtain a new water sample, and finally compare it with Sample C. Is this approach feasible?

[dlparkhurst]

Diffusion is a relatively slow process. From porewater work I used to do, a round number for the propagation of a signal at the top of a sediment column downward is about 10 cm per year for pure diffusion. I think, processes other than diffusion would overwhelm diffusion in an open borehole with a temperature gradient.

You can model transport and reactions with PHREEQC, but you should probably start by considering simple batch calculations. If you use inverse modeling to try to identify the pertinent reactions, you can then consider forward modeling equilibrium or kinetic calculations based on those reactions.