Recent Waste Modelling Studies and Developments

Modelling efforts have attempted to couple geochemical speciation and reaction with transport to assess release from waste materials. Several modelling codes have been developed by researchers for their own use or have become available either commercially or in the public domain. Examples of the latter include but are not limited to the Leaching Expert System (LeachXS – database of leaching data, including the geochemical speciation and transport framework, ORCHESTRA), PHREEQC, GEMS-Selektor, and MINTEQA2.

These codes have been used to model contaminant release during various leaching tests (i.e.

batch, pH-dependence, monolithic or percolation) or under field conditions. It is noted that from these codes GEMS-Selektor is the only true Gibbs free energy minimisation algorithm whereas other popular codes such as LeachXS and PHREEQC involve equilibrium constants.

A series of papers (Batchelor and Wu, 1993; Park and Batchelor, 1999a, 1999b, 2002a, 2002b) describes the development of a code to model reactions and transport from solidified wastes. The Pitzer ionic interaction model was initially used to calculate activity coefficients and the thermodynamic database of the program MINTEQA2 was used and extended to

90 include cement hydration products. CSH variable stoichiometry was described by a set of empirical regression equations (Reardon model, described further in Batchelor and Wu, 1993). The use of the Pitzer ionic interaction model however, led to slow and unreliable convergence and was later replaced by the Davies equation. In addition Reardon’s model was replaced with Berner’s model for the description of the variable stoichiometry of CSH.

Berner’s model exhibited better agreement with experimental data at pH > 11. This model assumes that the variable stoichiometry and variable solubility of CSH can be described as a non-ideal mixture of two independent solid phases. This assumption is applied to three discrete regions defined by different values of the molar ratio of calcium to silicate in CSH.

The solid phases for each region are:

o SiO2 and CaH2SiO4, 0 < Ca/Si ≤ 1.0

o CaH2SiO4 and Ca(OH)2, 1 < Ca/Si ≤ 2.5

o CaH2SiO4 and Ca(OH)2, 2.5 < Ca/Si

Sorption of alkalis (Na, K) was incorporated in the model using a set of regression equations taking into account the water-to-cement ratio and its effect in the surface area of CSH. A linear pH-dependent sorption model was also included to account for sorption of metals through surface complexation according to the reaction:

= SOH + M²⁺ → ≡ SOM⁺ + H⁺

where ≡ SOH is an available sorption site; M²⁺ the metal ions; and ≡ SOM⁺ a sorption site occupied by ion M²⁺. Finally the chemical speciation code was coupled with a transport model based on one dimensional Fickian diffusion. The chemical speciation code calculated the mobile fraction of components which was entered as a variable in the diffusion equations.

The latter was solved using a Crank-Nicolson finite difference method.

Hamil et al (2005) simulated leaching of Pb, Cd, Cr and As from cementitious waste using PHREEQC. The experimental procedure involved preparation of metal-spiked cement mixes and leaching according to a modified TCLP procedure, using 0.1M or 0.6M acetic acid or a municipal solid waste landfill leachate. Model input included mineral phases assumed to be present in the cementitious waste and their calculated quantities, availability of heavy metals,

91 complexation reactions with organic ligands, adsorption on hydrous ferric oxides and silica gel and solid solution formation. In addition, dissolution rates were included for cement minerals while equilibrium with their possible equilibrium-controlling species was assumed for most solid compounds. Model output was the total concentration in solution for each element of interest, its speciation at different time steps, SI values of relevant minerals and solution pH. Modelling results showed that precipitation was the dominant solubility-controlling mechanism for Pb and Cd. In the presence of acetic acid and at low pH values Pb and Cd were present as acetates whereas at high pH values hydroxide species dominated.

Simulations showed that complexes with organic ligands where the dominant heavy metal species when landfill leachate was used as the leachant. The authors also suggested that at high pH values release of As may be controlled by the competition between calcium carbonate and calcium arsenate competition.

Astrup et al (2006) used LeachXS to perform speciation calculations based on data from a pH-dependence leaching test on APC residues and identified possible solubility-controlling minerals for Al, Ba, Ca, Cr, Pb, S, Si, V, and Zn. The plausibility of each mineral was evaluated by calculating element concentrations in equilibrium with specific minerals obtaining solubility-type curves and comparing with the experimental data. Predicted concentrations for most elements based on solubility control from selected minerals present in the thermodynamic database were within one order of magnitude in the pH range 4.5-12.5.

Some data points for Pb, Zn, Cr and V could not be explained solely by solubility control and the authors suggested that processes such as redox, sorption or incorporation in mineral phases may be important at certain pH values.

A similar approach was used by Hyks et al (2007) using ORCHESTRA to identify solubility-controlling phases in washed and raw APC residues varying also the time of the pH-dependence leaching test used. The same controlling phases were suggested for Al, Ca, Mg, Si, S, Mo, Zn, Cd, and Cu regardless of the equilibration period or untreated/washed nature of APC residues, whereas leaching of Ba, Sr was better described by considering a longer equilibration time (kinetic effects). It is also noted that the authors compared the activity coefficients for Ca, S determined with the Davies and Pitzer equation and established that the same solubility-controlling phases are obtained in both cases.

92 Tiruta-Barna (2008) developed a code in PHREEQC to model dynamic leaching tests with continuous renewal of leachant. In the model system, dissolution/precipitation reactions begin at the solid/liquid interface while additional chemical reactions (acid/base, complexation, redox) occur in the aqueous phase. Soluble chemical species diffuse from the pores of the material towards the leachate and the resulting changes in the pore water composition affect in turn mineral solubility. Surface corrosion (dissolution) is also assumed to occur at the solid surface in contact with the leachant. The main model building blocks were a diffusion compartment (porous block) in contact with a leaching compartment, which was assumed to behave according to an open stirred reactor scheme. Exchange of chemical species occurs between the two compartments. Model predictions for two cases involving i) laboratory leaching tests on a cement S/S material and ii) leaching from a pilot scale water storage pool constructed with a S/S material, were within one order or of the same order of magnitude as the experimental data. In the case of the water storage pool scenario, CO2 uptake incorporated in the model produced accurate predictions for the evolution of pH. Predictions for non-reactive elements (Na and Cl) however, were not in good agreement with the experimental values which was attributed to an increase in porosity with time.

More recently Martens et al (2010) modelled various extraction leaching tests and a modified Dutch NEN 7345 diffusion test for stabilized/solidified MSWI bottom ash also using PHREEQC. Their approach involved similar steps with previous studies including selection of appropriate mineral phases, experimental measurement and model input of hydrous ferric oxides and amorphous aluminium mineral for heavy metal adsorption, determination of the potential formation solid solutions (i.e. calcite-cerrusite) and in the case of the diffusion test, determination of the effective diffusion coefficient. The main difference in the study is the use of a different thermodynamic database (CEMDATA07.1) for various cement hydrates such as CSH, AFm phases (monosulphoaluminate, strätlingite), AFt (tricarboaluminate, ettringite), hydrogarnet and hydrotalcite. The model gave good prediction for Ca, Na, K and Pb but underestimated release of Al and Mg. Release of Ca was found to be controlled by hydration products such as portlandite, the jennite-like and tobermorite II-like solid solution strätlingite and calcite. In contrast with the study by Halim et al (2005) leaching of Pb was not only controlled by precipitation/dissolution but sorption by surface complexation and solid solution formation adequately described its amphoteric behaviour.

93 The above studies demonstrate that geochemical and transport modelling is a useful tool for identifying the chemical and transport processes that are involved in the release of contaminants, based on laboratory leaching tests. Moreover, modelling can also be used to model numerous site conditions such as leaching from recycled construction materials, concrete performance and leaching in landfills, where experimental studies are limited by the time required to assess long-term effects (i.e. > 100 years).

De Windt et al, (2007) used a reactive transport code to extrapolate results of batch and dynamic tests to a simplified landfill cell scenario for a Pb-containing waste treated with CEM I and considering a period of 100 years. The researchers used identical chemistry and mineralogy to cubic laboratory specimens. Additional model inputs included the rainfall and effective infiltration rate with the waste assumed fully saturated, while CO2 penetration was partly incorporated. The effect of fracture or micro-crack development in the matrix was evaluated. Model results showed that Pb leaching from damaged S/S matrices was up to 100 times greater (extreme conditions) compared to undamaged monoliths, but was 0.1% of the total Pb content. Considerations in such models are the complex nature of the field scenario in terms of matrix degradation, carbonation (CO2) effects, reactions under unsaturated conditions or drying and wetting cycles as well as the effects of landfill liners.

Van der Sloot et al (2007) followed a combined experimental-modelling tiered approach in order to evaluate the potential for environmental impact of stabilised waste. The authors utilised data from laboratory leaching tests (i.e. pH-stat and monolithic) and lysimeter/field studies coupled with modelling (using LeachXS) to assess the long-term effects of stabilized waste disposed in hazardous waste landfills. pH-stat leaching tests and subsequent modelling provided an insight on the solubility controlling phases (geochemical characterisation).

Modelling of tank leaching data was then used to validate the geochemical characterization derived from the pH-stat test under diffusion-release, taking into account the physical properties of the matrix (i.e. porosity, tortuosity). The next step involved determining the interactions between the stabilized waste (both in monolithic and crushed form) with the layer of soil simulating landfill conditions (release by diffusion and percolation). This integrated, tiered approach was considered a basis for improving the understanding of the factors affecting release in field conditions and for contributing to improved hazardous waste landfill design. The authors attributed differences between the measured and modelled results to the description of processes such as sorption, complexation to organic matter and kinetic

94 effects during leaching. They also noted the complexity of processes such as carbonation that can have a marked effect on leaching characteristics.

As it is observed from the above examples, modelling of cementitious waste systems is a complex process. Laboratory leaching tests coupled with geochemical modelling provides a framework for characterising the material of interest/concern and assessing release under different conditions (i.e. leaching test, field scenarios etc.). This can result in improved management decisions pertaining to either treatment or disposal of wastes or landfill design.

In the case of stabilized/solidified waste using cementitious binders, it can contribute to improved mixed design and assessment of long-term release which is not feasible with laboratory leaching tests.

One of the basic principles for the success of the model is an accurate formulation of the initial problem as well as identification of the possible solubility-controlling phases and their thermodynamic properties. Furthermore, the immobilization mechanisms considered in the model may play a very important role in the accuracy of the results and the eventual fit with experimental data. This requires knowledge of characteristics of the S/S waste such as plausible solubility-controlling solids and their thermodynamic properties and physical properties of the S/S matrix.

2.7 Summary

This review demonstrates that the chemistry of S/S is complex. The effect of metal-containing waste on cement hydration, and thus the effectiveness of the treatment depends on the nature of the waste. Combination of heavy metals may induce hydration poisoning effects, compared to retardation effects when metals are studied in isolation. APC residues represent a problematic waste since it contains not only combinations of heavy metals, but also high concentrations of chlorides. S/S has been previously used for the treatment of APC residues with the majority of the studies using Portland cement types at additions no greater than 60 wt.%. In addition, the focus in these studies has been on the metal immbolisation capacity of S/S APC residues, whereas the effects and fate of chloride, its most abundant problematic constituent, have been less detailed. Finally, geochemical modelling is increasingly used to provide a deeper understanding of S/S immobilisation processes for elements of concern.

95