[Next] [Previous] [Up] [Top]
PHREEQC is a general geochemical program and is applicable to many hydrogeochemical environments. However, several limitations need to be considered.
PHREEQC uses ion-association and Debye Hückel expressions to account for the non-ideality of aqueous solutions. This type of aqueous model is adequate at low ionic strength but may break down at higher ionic strengths (in the range of seawater and above). An attempt has been made to extend the range of applicability of the aqueous model through the use of an ionic-strength term in the Debye Hückel expressions. These terms have been fit for the major ions using chloride mean-salt activity-coefficient data (Truesdell and Jones, 1974). Thus, in sodium chloride dominated systems, the model may be reliable to higher ionic strengths. For high ionic strength waters, the specific interaction approach to thermodynamic properties of aqueous solutions should be used (for example, Pitzer, 1979, Harvie and Weare, 1980, Harvie and others, 1984, Plummer and others, 1988).
The other limitation of the aqueous model is lack of internal consistency in the data in the database. Most of the log K's and enthalpies of reaction have been taken from various literature sources. No systematic attempt has been made to determine the aqueous model that was used to develop the log K's or whether the aqueous model defined by the current database file is consistent with the original experimental data. The database files provided with the program should be considered to be preliminary. Careful selection of aqueous species and thermodynamic data is left to the users of the program.
The ion-exchange model assumes that the thermodynamic activity of an exchange species is equal to its equivalent fraction. Other formulations use other definitions of activity, mole fraction for example, or additional activity coefficients to convert equivalent fraction to activity (Appelo, 1994). No attempt has been made to include other or more complicated exchange models. In many field studies, ion-exchange modeling requires experimental data on material from the study site for appropriate model application.
PHREEQC incorporates the Dzombak and Morel (1990) diffuse double-layer and a non-electrostatic surface-complexation model (Davis and Kent, 1990). Other models, including isotherms and triple- and quadruple-layer models have not been included in PHREEQC.
Davis and Kent (1990) reviewed surface-complexation modeling and note theoretical problems with the standard state for sorbed species. Other uncertainties occur in determining the number of sites, the surface area, the composition of sorbed species, and the appropriate log K's. In many field studies, surface-complexation modeling requires experimental data on material from the study site for appropriate model application.
The capability of PHREEQC to calculate the composition of the diffuse layer (-diffuse_layer option SURFACE keyword data block) is ad hoc and should be used only as a preliminary sensitivity analysis.
PHREEQC tries to identify input errors, but it is not capable of detecting some physical impossibilities in the chemical system that is modeled. For example, PHREEQC allows a solution to be charge balanced by addition or removal of an element. If this element has no charged species or if charge imbalance remains even after the concentration of the element has been reduced to zero, then the numerical method will appear to have failed to converge. Other physical impossibilities that have been encountered are (1) when a base is added to attain a fixed pH, but in fact an acid is needed (or vice versa) and (2) when noncarbonate alkalinity exceeds the total alkalinity given on input.
At present, the numerical method has proved to be relatively robust. Known convergence problems--cases when the numerical method fails to find a solution to the non-linear algebraic equations--have occurred only when physically impossible equilibria have been posed and when trying to find the stable phase assemblage among a large number (approximately 25) minerals, each with a large number of moles (5 moles or more). It is suspected that the latter case is caused by loss of numerical precision in working with sparingly soluble minerals (that is, small aqueous concentrations) in systems with large total concentrations (on the order of 100 moles). Occasionally it has been necessary to use the scaling features of the KNOBS keyword. The scaling features appear to be necessary when total dissolved concentrations fall below approximately 10-15 molal.
Inclusion of uncertainties in the process of identifying inverse models is a major advance. However, the numerical method has shown some lack of robustness due to the way the solver handles small numbers. The option to change the tolerance used by the solver is an attempt to remedy this problem. In addition, the inability to include isotopic information in the modeling process is a serious limitation.
- Aqueous Model
- Ion Exchange
- Surface Complexation
- Convergence Problems
- Inverse Modeling
User's Guide to PHREEQC - 07 MAY 96
[Next] [Previous] [Up] [Top]