A methodology is presented for determining reaction kinetics from coreflooding: A core is flooded with reactive brine at different compositions with injection rates varied systematically. Each combination is performed until steady state, when effluent concentrations no longer change significantly with time. Lower injection rate gives the brine more time to react. We also propose shut-in tests where brine reacts statically with the core for a defined period and then is flushed out. The residence time and produced brine composition are compared with the flooding experiments. This design allows characterization of the reaction kinetics from a single core. Efficient modeling and matching of the experiments can be performed as the steady-state data are directly comparable to equilibrating the injected brine gradually with time and do not require spatial and temporal modeling of the entire dynamic experiments. Each steady-state data point represents different information that helps constrain parameter selection. The reaction kinetics can predict equilibrium states and time needed to reach equilibrium. Accounting for dispersion increases the complexity by needing to find a spatial distribution of coupled solutions and is recommended as a second step when a first estimate of the kinetics has been obtained. It is still much more efficient than simulating the full dynamic experiment.
Experiments were performed injecting 0.0445 and 0.219 mol/L MgCl2 into Stevns Klint (Denmark) and Kansas (USA) chalks at 100 and 130°C (North Sea reservoir temperature). Injection rates varied from 0.25 to 16 pore volume per day (PV/D), while shut-in tests provided equivalent rates down to 1/28 PV/D. The results showed that Ca2+ ions were produced and Mg2+ ions retained (associated with calcite dissolution and magnesite precipitation, respectively). This occurred in a substitution-like manner, where the gain of Ca was similar to the loss of Mg2+. A simple reaction kinetic model based on this substitution with three independent tuning parameters (rate coefficient, reaction order, and equilibrium constant) was implemented together with advection to analytically calculate steady-state effluent concentrations when injected composition, injection rate, and reaction kinetic parameters were stated. By tuning reaction kinetic parameters, the experimental steady-state data were fitted efficiently. The parameters were determined to be relatively accurate for each core. The roles of reaction parameters, pore velocity, and dispersion were illustrated with sensitivity analyses. The determined reaction kinetics could successfully predict the chemical interaction in reservoir chalk and outcrop chalk containing oil with strongly water-wet or mixed-wet state.
The steady-state method allows computationally efficient matching even with complex reaction kinetics. Using a comprehensive geochemical description in the software PHREEQC, the kinetics of calcite and magnesite mineral reactions were determined by matching the steady-state concentration changes as function of (residence) time. The simulator predicted close to the identical production of Ca as loss of Mg. The geochemical software predicted much higher calcite solubility in MgCl2 than observed at 100 and 130°C for Stevns Klint and Kansas. The methodology supports reactive flow modeling in general, but especially oil-bearing chalk reservoirs, which are chemically sensitive to injected seawater in terms of wettability and rock strength.