Monolithic Leaching Test

Modelling the monolithic leaching test is a complex task that involves both chemical reactions as well as contaminant transport. A number of parameters and/or variables are taken into consideration namely:

 Sample Geometry;

 Material Properties;

 Element Availability Concentrations;

 DHA and DOC concentrations; and

 Mineral Set.

a) Sample Geometry

One of the challenges in modelling the monolithic leaching test is the fact that in a real-life scenario release from the monolith is three-dimensional with all the surfaces of the specimen exposed to the leachant. LeachXS represents the real-life scenario by creating a grid comprising 24 cells. The allocation of the cells is as follows:

 Cell 0 represents the aggressive leaching solution (leachant);

 Cells 1-21 represent the solid phase layer;

 Cell 22 is the solution used for refreshing the aqueous phase (leachant); and

166

 Cell 23 is a “waste bucket” which is used only in the case of advection.

Assuming that the specimen to be tested is a cube, LeachXS creates the model grid by dividing the solid phase (specimen subjected to the leaching test) into 20 hollow cubes called cells. The geometry of each cell is derived as shown in Figure 6.3. The outer cell (cell 1) in contact with the leachant has a surface area of A1 = 6×L² and a volume of V1 = A1×dx. The next cell (2) has a surface area of A2 = 6×(L-2dx)² and a volume of V2=A2×dx. The geometry of the subsequent 18 cells is derived in the same manner.

Figure 6.3 Idealisation of three-dimensional release in LeachXS

The geometry and dimensions of the specimen can be defined by the user. The distance dx is also a variable in LeachXS called ‘Factor to increase size of successive grid cells’. This can be adjusted by the user in order to vary the geometry of the model cells. Transport is then calculated based on diffusion and concentration gradients between the compartments of the idealised structure in Figure 6.3. The specimens in this study were 50mm cubes. The distance dx was set between 1.25 and 1.28 based on data provided by ECN.

b) Material Properties

Material properties that can be adjusted in the model include bulk density, available porosity, tortuosity and pH. Bulk density values were calculated automatically by LeachXS based on the experimental (observed) data entered in the database. Values for material pH and porosity

167 were obtained from the experimental data and were entered manually using the LeachXS GUI. In particular, pH values were obtained from the ANC test at no acid (0 meq/g) addition.

It is noted that this pH value refers to the natural pH of the material to be tested (i.e. S/S APC residues) and not the pH of the eluate at fractions of the leaching test.

Tortuosity was not determined experimentally for the samples tested in this study. Tortuosity was determined by varying its value in the model and assessing its effect on species with low retardation coefficients such as soluble salts (i.e. Na and Cl). The tortuosity value that resulted in the best fit with leaching data for such elements was selected for subsequent modelling runs.

c) Element Availability, DHA and DOC Concentrations

Input of element availabilities, DHA and DOC concentrations has been described as part of the pH-dependence run. The majority of availability concentrations, as well as DHA and DOC concentrations were obtained from the surrogate material in the LeachXS database and previous modelling studies on stabilised waste conducted by ECN. These parameters are summarised in the following sections for each mix modelled.

d) Mineral Set

The mineral set is one of the key variables in the model. The initial mineral set was based on the Chemical Speciation Finder and pH-dependence runs for the surrogate material. The mineral set was adjusted after each modelling run until a good fit with the measured values was obtained. The mineral set was adjusted based on previous studies in the literature as well as other probable minerals according to LeachXS SI calculations.

In general, due to the lack of data for tortuosity and element availability concentrations the modelling framework for the monolithic leaching test consisted of the following steps:

1. Adjustment of tortuosity values until a good fit is obtained for release of soluble salts;

2. Adjustment of mineral set and/or availabilities until a good fit is obtained for pH and the release of major elements such as Ca, Al, SO42- etc.

3. Adjustment of mineral set and/or availabilities until a good fit is obtained for trace elements of concern such as Pb and Zn.

168

As the w/s ratio is the same for each of the mixes investigated, the modelling study evaluated the effect of binder type and content on the leaching characteristics of the S/S APC residues.

Water content was not investigated in the modelling study due to the fact that varying the w/s ratio (for the same binder addition) did not result in statistically significant results in terms of monolithic leaching as it will be explained in the Discussion chapter.

Modelling results for each element were compared to the observed values by calculating the residuals, i.e. difference between observed and predicted value. Different statistical methods exist for determining the goodness of fit. In the present study due to the low number of data points (8 values for the tank test and 9 fractions for the pH-dependence test), the sum of the squared residuals from all test fractions was used to determine goodness of fit for each element as well as pH. In summary, the problem definition with the best fit for each element and pH was determined according to the following equation:



mX, is problem definition with the best goodness of fit for the elements of concern as well as pH;

z, is the total number of fractions of the leaching test modelled. z = 8 for the monolithic leaching test and z = 9 for the pH dependence test;

169 yiobs, is the measured value (element concentration or pH) determined experimentally for

leaching fraction i.

yipred, is the model predicted value (element concentration or pH) for leaching fraction i.

The overall goodness of fit obtained from the different mineral sets was then assessed based on the goodness of fit of individual elements and pH as shown in Equation 6.2. The sums of the squared residuals for each element and pH were added to obtain a single number for each problem definition. In addition, weights were applied to the different elements in order to avoid skewed results due to large residual values for elements with high leached concentrations (e.g. Ca).



 



 



X min m

M

1

(6.2)

The MX value was then used as an indicator of goodness of fit, and the problem definitions (set of parameters) from the different modelling runs were compared based on these values as well as existing literature to improve the problem definition for subsequent runs. The same methodology for assessing goodness of fit was used for both the monolithic and the pH-dependence test. The comparison between measured and predicted values for the monolithic leaching test was performed on the concentrations (mg/l) for each leaching fraction whereas for the pH-dependence test it was based on the measured release (mg/kg) for each batch extraction.

In addition, the effect of varying the model input parameters was evaluated using statistical techniques such as paired t-tests and analysis of variance (ANOVA). The commercial statistical software package STATA (v. 9.2) was used for all statistical analyses.

170 6.6 Results

This section presents the results from the modelling study for both the monolithic leaching test and the alkaline/acid leaching procedure. The section for the monolithic leaching test is further subdivided based on the different elements of concern. It is noted that although many modelling runs were conducted in order to assess the significance of individual model parameters, the results presented below pertain to the set of parameters that yielded the best fit for the majority of the elements.