An object-oriented module for geochemical and reactive transport modeling

Resumo

Accurate prediction of contaminant migration requires reactive transport modeling. The increasing complexity and the traditional procedure-oriented type of programming hinder codes reuse and transportability. The objective of this thesis is to present a Fortran90 module using object oriented concepts to simulate hydrobiogeochemical processes (CHEPROO, CHEmical PRocesses Object-Oriented). CHEPROO consists of a general structure with two classes. The Nodal Chemistry class represents local chemistry and contains geochemical state variables. It provides functions related to basic operations (evaporation, mixing, etc) and can easily grow on this direction (extreme dry conditions, biochemical state variables, etc). The Chemical System class includes kinetic and thermodynamic models to describe reactions between and within phases. As such, it can grow in the direction of increasingly complex chemical systems (solid solutions, microorganisms as individual phases, etc). These two classes are overlaid by CHEPROO, a general structure designed for interaction with other codes. CHEPROO can be used as a geochemical tool for modeling complex processes such as biodegradation or evaporation at high salinities. Yet, many CHEPROO functions are devoted to coupling chemical processes to other phenomena (e.g. flow, transport, mechanical, etc). Different approaches to reactive transport can be easily implemented into existing conservative transport code with a minimal number of changes. Modeling concentrated solutions demands the use of ion-interaction models such as Pitzer equations, involving a large number of operations. Implementation of these models in large reactive transport simulations is computationally demanding. CPU time depends on the efficiency of 1) Pitzer algorithm itself, and 2) the speciation algorithm. Pitzer equations are implemented in CHEPROO using a compact matrix scheme. Speciation is based on a Newton-Raphson method using analytic derivatives of Pitzer equations. I demonstrate that the code is robust, in that it converges in a broad range of cases, and efficient, in the sense that CPU time compares favorably to other codes. CHEPROO is applied to two real problems. The first one involves modeling the formation of eriochalcite (CuCl2·2H2O), a highly soluble salt, rarely found in Nature. Significant amounts of efflorescent eriochalcite have been described on coastal mine tailings at Chanaral (Chile). They represent an environmental problem. Modelling is necessary for eventual remediation and was done by including heat transfer, vapour and water flow, as well as chemical reactions (secondary salt precipitation; silicate and sulphide dissolution and Fe(II) oxidation, cation exchange between major ions). Atmospheric conditions were imposed at the top boundary. Results for different scenarios were compared to observed mineralogy and pore water chemistry. The best scenario involves sea-water or a mixing between sea/less-saline water in the groundwater composition. I find that strong competition of other chloride salts (i.e. halite (NaCl) and silvite (KCl)) may inhibit precipitation of eriochalcite. Therefore, the Cl/Na ratio is a key parameter. Cation-exchange between Na and other major ions such as K, Ca, Mg and Cu is proposed to account for sufficiently high Cl/Na ratios. Moreover, exchange may be a more important source of Cu than the slow oxidation of chalcopyrite (CuFeS2). The second example involves remediation alternatives for a 137Cs contaminated soil. Cs sorbs strongly to clay aggregates where water flux is negligible, whereas the mobile portion of the soil (macropores) retains little water and cesium. Remediation alternatives involve infiltration of sea-water enriched with KCl, to promote mobilization of Cs through exchange with K. They are tested using a multicontinuum reactive transport model performed with CHEPROO and other codes, for verification. I find that flushing is a viable alternative.