Search the Community
Showing results for tags 'surface complexation'.
Found 2 results
Hello, I am using GWB X1t to simulate fluoride transport in a soil column. First, the column is contaminated with fluoride for a certain period of time and then the column was flushed with deionized water. In the simulation, first, I want to simulate the transport process during sorption, where fluoride sorption was described by surface complexation. For surface complexation, I used the general composite approach (I added a new basis species (X) with 1g/mol molecular weight, defined a new element and a mineral with the basis species). For the surface dataset, I defined the reactions in terms of a new surface species (>XOH). For the mineral,if I use the exact density (e.g., 3591 g/l, that I used for my column with 39% porosity) and a site density of 2.6e-5moles/g, the program ends with the message that the porosity is too small. Since the porosity at each grid block is calculated with each run, I decreased the mineral site density (e.g., 1g/l or 0.1 g/l) and accordingly increased the site density. Then the program doesn't converge. However, if I keep both site density and mineral density small, the program runs but the results are not satisfactory (fluoride appears in the outlet just after 1 pore volume because the sorption is small). I appreciate any suggestion on how to keep the desired site density? P.S.: My PHREEQC simulation gives good results with the same site density, mineral conc. and surface reactions. Padhi
Greg Miller posted a topic in The Geochemist's WorkbenchI want to model the sorption of boron on surfaces using the constant capacitance (CC) model. I am trying to use a semi-mechanistic, mainly empirical, model advanced by Sabine Goldberg that uses measured physical and chemical properties of solids to calculate Kint-boron, and K+ and K- of the surface for CC. This is a generalized composite model that assigns these properties to the geomedia as a whole bulk system. Therein lays the first problem, defining a sorbing mineral phase and the surface reactions that go with it. I am very interested in the groups’ thoughts on how to best go about this. My first thought was to just alter FeOH.sdat to reflect the right Ks, surface areas, etc. and then use it as a dummy variable. Then React would be used to simulate boron adsorption on porous media using a pretty complete water quality data set. All other reactions will be turned off. The problem lies in the generalized composite model source of Kint, K+ and K-; it’s a bulk sample parameter, not mineral specific. All of the porous media would need to be Fe(OH)3 (ppt) to use a modified FeOH.sdat. My intuition is that this approach would fall apart when I went to determine bulk concentrations of boron on the solids because of molar considerations and surface site density being defined in moles/mole-mineral. The system is a granular aquifer ~80% quartz, ~20% feldspars so my second approach has been to define the sorbing mineral as SiO2 to avoid the molar volume/mass issues and use a dummy water on the SiO2 as the reacting surface. So, the surface is SiO2-OH2, and sites SiO2-OH- and SiO2-OH3+ are formed by protonation/deprotonation, and boron on the surface is SiO2-HH3BO4-. The borate ion is not considered because the pH of the system is far below the 9.2 pKa. Using a defined mineral works in my very limited understanding of the features of GWB. Using dummy water as a fake metal oxide surface makes the charges and stoichiometry work out right (I think!) and shouldn’t blow up the math anywhere. My thinking is I will avoid the molar problems described above. Is there another approach in GWB that might work better? I have to use the CC model. I can torture PHREEQC to do this with an as yet untested CC add-in, I prefer to use GWB because all my water chemistry data is in .gss, and if this goes well, I might add in some other rxns. There are a fair number of samples to be evaluated that all will need their own .sdat file. Maybe there is a better way? See any big holes in the Qtz + dummy water surface idea? I greatly appreciate anybody’s time to think about this. I am thinking someone has done something close to this before. Greg Miller