The performance of PHREEQC databases for modelling the solubility of CO2, N2, CO2-N2 mixtures, calcite and barite at elevated temperatures, pressures and electrolyte concentrations

Modelling of geochemical reactions in the deeper subsurface can provide valuable insight into the state and dynamics of a system where processes are complicated to observe. In addition, it allows prediction of the outcome of interventions, such as geothermal exploitation or CO 2 injection. The degree to which the modelling portrays reality depends in part on the ability of the thermodynamic data to describe solubility of gases and solids accurately. For most modelling software, the thermodynamic data is typically compiled in databases, which have been tested to some extent. However, the capabilities of the databases to accurately calculate solubility over a wide range of conditions are most often not well defined. This study tests the performance of 13 PHREEQC databases, benchmarking them against an empirical dataset with 3147 measurements of the solubility of CO 2(g) , N 2(g) , CO 2(g) - N 2(g) mixtures, calcite, and barite at variable temperature, pressure, and electrolyte concentration. Our work shows that the databases employing the Peng-Robinson equation of state and molar volume changes of reaction are generally capable of correctly describing experimental values of solubility at higher pressure as intended. Databases relying on the Pitzer approach for description of activity coefficients often provide reasonably accurate calculations of solubility at an extended range of electrolyte concentrations at variable temperature and pressure. For many databases, including those based on Pitzer equations, calculations substantially deviate from measured values at conditions that are unexpected based on general expectations to database performance. This is particularly true for simulations of calcite solubility. Identifying these deviations requires tests such as the one given here. Our approach allows selection of databases based on the objective criterion that calculations should match empirical data at or near the modelled conditions as closely as possible. We exemplify the use of our results for database selection for the simulation of calcite scaling in a model geothermal well. To enable users and developers of databases to conduct tests of performance themselves, the supplementary information contains experimental data, input files and Python code for calculations.