Summary, Conclusions and Recommendations

Under fully controlled laboratory conditions, with constant chemical groundwater quality, one type of collector (quartz grains), a range of Darcy flow velocities, a range of grain sizes, and in columns of limited height, the E. coli concentration decreases logarithmically with the transport distance. Also, there is reasonable agreement between the measured filter coefficients from breakthrough curves and the filter coefficients calculated using the single collector contact efficiency, and assuming a constant sticking efficiency for the E. coli - quartz system. Because the E. coli - quartz grain system behaves according to the classical CFT, as defined in chapter 1 of this thesis, the theory provides a valuable framework for predicting the transport of E. coli in porous media under saturated conditions. This is one of the main outcomes of the work carried out by Matthess et al. (1991a, 1991b) on an extensive set of column experiments. However, based on the research presented in this thesis, this outcome needs to be modified for the following reasons: The conclusions of Matthess et al. (1991a, 1991b), and many other workers, are based on effluent concentrations from column experiments only. As was concluded by Bradford et al (2002 , 2003 ), Tufenkji and Elimelech (2004a, 2005), Li et al. (2004) , Tong and Johnson (2007), and by our work ( Chapter 8 ), such type of experiment does not give information on the transport distance dependent retention processes taking place in the column. Therefore, more types of experiments are required. In literature, column extrusion experiments have been mentioned a number of times, and we added a new type of experiment, the so called flow reversal experiment, in which the occurrence of non-adhesive retention in column experiments is directly assessed; We have demonstrated in columns of spherical glass beads with various diameters, that, when the beads were large, or column heights were too limited, dual mode deposition, as a result of attachment heterogeneity among members of the E. coli population, was completely masked, because in such cases just one type of (fast) deposition dominated the retained profiles, resulting in log-linear retained bacteria masses. Therefore, the extrapolation of experiments, where masking of dual mode deposition has taken place, to predict removal of E. coli at a larger scale will almost certainly lead to overestimation Of E. coli removal; The CFT does not include straining. In columns of ultra-pure angular quartz sand, from eluted concentrations, column extrusion experiments, flow reversal experiments, and model fitting of eluted concentrations, we found that straining could be as much as 25-30% of the total bacteria mass applied to the column, while the maximum attached fraction was 30-35%. Since straining occurred in all grain size fractions we used, we are convinced that a good deal of literature on colloid retention processes, using coulmns of angular collectors, and neglecting straining or non- adhesive retention, have come to erroneous conclusions with regard to the nature of the processes causing retention in their experiments. In these cases, as was mentioned above, the extrapolation of such results to predict removal of E. coli at a larger scale will almost certainly lead to overestimation of E. coli removal; The CFT does not include geochemical heterogeneity as a result of the presence of various minerals in the sediment. Since these minerals can have different surface charges as a result of pore water chemistry and/or lattice imperfections, and since E. coli is also (negatively) charged, mineralogical Variations in the sediment can have an influence on attachment. We found that in columns of limited height (2 cm), due to the presence of positively charged goethite or calcite, the sticking efficiency, as determined from eluted E. coli concentration curves increased from 0.2 to 0.9. Also, the sticking efficiency of PRDl, the model virus that we used, calculated in the same way as described above, increased 100%, from 0.05 to 0.1, when goethite coated quartz grains were added to the sediment. Characterization of the surface of the collector is therefore of utmost importance, especially if various mineral assemblages are considered or for comparison and interpretation of sticking efficiency values in literature; In cases of dual porosity aquifers, like the shallow unconsolidated alluvial aquifer below Sana’a, where we proved that certain parts of the aquifer were inaccessible for microorganisms, the importance of CFT was limited. Here, decay rate factor, pore water flow velocity, and pore volume available for fast flow in the aquifer were the most important factors determining the transport of E. coli . Relatively small volumes of fast flowing polluted water in aquifers can give rise to considerable microbiological contamination, while hydrochemical patterns may be hardly affected. More in general, we have demonstrated the added value of using microbial tracers for the conceptualization of contaminant transport processes in aquifers.