Untitled - Technische Universiteit Eindhoven

108 7. Permeability reduction in porous materials by in situ formed silica gelFig. 7.5: Scanning electron microscope image of Bentheim sandstone.image of the surface obtained at a low magnification is given in Figure 7.5. The imageclearly shows the coarse structure of the grains which are held together by siliceous cement.Though the SEM-EDX analysis is semi-quantitative at best a little amount of claywas found in local spots on the sample surface. The clay mainly consisted of silica andaluminum without other cations, and is therefore likely to be kaolinite. Contamination ofthe non-swelling clay due to cations introduced by the buffered solutions in the treatedrock samples may enhance the swelling behavior. The XRD results only showed quartz,which indicates that the clay found in the SEM analysis was only present in small quantities.Despite the absence of swelling clays the grains appear to slip when a load is appliedon the rock sample. The presence of water significantly reduces the elastic modulus ofthe rock and leads to an enhanced degree of VE relaxation with respect to a dry rock.Beam bending of treated sandstoneThe experimental results obtained for two series of treated sandstone are discussed inthis section. The rock samples in series A were treated with acid-catalyzed gel, whereasthe rock samples in series B were treated with base-catalyzed gel. The TMOS concentrationwas varied in both series, but no additional sodium-chloride was added to thegelling solutions. The properties of the treated rock samples are listed in Table 7.3. Thetotal reaction and aging time was 72 hours for all rock samples. Glycerol from fresh andtightly sealed bottles was used in the bath and for the exchange with the original poreliquid. Exposure to air was avoided as much as possible for the glycerol in the exchangebath. Therefore it is assumed that the glycerol in the samples is free of water, so thatthe viscosity at 26 ◦ C is ≃ 0.870 Pa s [138]. The bulk modulus K L is 4.80 GPa [138].The normalized load curves for series A and B are shown in Figure 7.6 and Figure 7.7,respectively. The total measurement time for each curve is about one hour. Two experimentsin series A were extended to more than 10 hours. All load curves show a monotonicdecrease within the times of measurement and none of the load curves reaches a plateau.The overall decrease in magnitude of the normalized load varies but appears to increase