A focal point of this study is that large drops in permeability may occur in granular media due to the growth of authigenic minerals such as smectite. The experiments presented here show that these drops in permeability may be achieved without any concurrent decrease in porosity. Although the pore volume does not change within any one experiment, the nature and shapes of the pores do. Fibrous smectite coatings such as those shown in figure 8 are rough surfaces possessing a large amount of void space between each fiber. Although this void space represents part of the porosity, it is not effective porosity which contributes to fluid flow. As more smectite grows, the effective porosity decreases and permeability is progressively reduced. A number of studies have been done on natural sandstones, showing that secondary mineral growth can increase surface roughness and cause permeability reduction with very little concurrent porosity reduction [i.e. Thompson, 1991; Aharonov and Rothman, 1996]. Assuming that smectite morphology is identical across all the conditions tested, then the magnitude and rate of permeability reduction depends only on the total volume and rate at which smectite is produced.

The permeability evolution curves shown in figure 6 clearly show that the magnitude of permeability reduction is dependent on both temperature and deviatoric stress. The temperature dependence on permeability evolution is relatively clear. Higher temperatures result in greater mineral solubilities, thus retarding attainment of fluid equilibrium and prolonging the active dissolution and precipitation which cause the observed changes in permeability. However, the permeability dependence on deviatoric stress is more difficult to understand. The Al and Si chemical data (Figure 7, Table 1) clearly demonstrate that stress is exerting some influence on the chemistry of the system, with higher stresses causing an increase in mineral solubility. This difference in solubility can be explained by a pressure solution-type mechanism, which causes rapid transfer of mineral mass to the pore fluid due to the formation of chemical potential gradients. Such stress-induced solubility changes have already been studied for pure quartz systems [de Boer, 1977; Elias and Hajash, 1992]. Elias and Hajash [1992], conducting pressure solution experiments using unconsolidated St. Peter's sandstone, observed an increase in Si solubility up to an effective stress of 69 MPa. Greater stresses were not tested since Hertzian grain fracture is believed to occur at 69 MPa [Zhang et al., 1990], thus complicating interpretation of the data (see below). In our experiments, the increased solubility caused by stress translate into a change in permeability behavior. However, it should be noted that the permeability dependence on deviatoric stress is not a result of pressure solution in the classical sense. In the past, studies on pressure solution have mainly focused on quartz aggregates, where dissolution and precipitation of quartz are accompanied by compaction caused by transfer of Si. This long term deformation of the aggregate combined with local precipitation of quartz results in permeability and porosity reduction. In this study, the stress dependent permeability reduction observed is due to a different process which operates on a shorter time scale. The applied deviatoric stresses in the experiments cause a change in mineral solubility which in turn affects the time required to reach equilibrium. At low stresses, equilibrium Si values are low, allowing the fluid to equilibrate faster than for high stresses. As a result, a shorter time is available for active dissolution and precipitation of smectite to occur.

Studies have shown that at low stresses, compaction of sands occurs mainly by grain reorganization with very little grain crushing taking place [i.e. Zoback and Byerlee, 1976; Zhang et al., 1990]. During this fracture-free stage, greater external stresses translate into higher contact stresses. This stress intensification at the microscopic scale is then manifested by greater Si concentrations in solution. However, once grain crushing begins, any increase in effective stress is accompanied by an increase in total contact area, with the stress no longer rising since it is limited by the crushing strength-therefore there is no further change in fluid chemistry. This concentration plateau is evident from our data and is shown in figure 7b, although the flat portion of the concentration curve begins at a maximum effective stress of 100 MPa (sd = 50 MPa) rather than at 69 MPa. This discrepancy is likely a result of the different material used and the angularity of grains compared to that used in Zhang et al. [1990].

Although pressure solution is likely playing an important role in permeability evolution by changing the system chemistry, deviatoric stress probably also results in some mechanical effects and associated pore topology changes which also contribute to changes in permeability behavior. Variable stress causes differing amounts of compaction to occur, giving rise to different grain size distributions between experiments. The effect of grain size distribution variability is two-fold. First, higher stresses would skew the grain size distribution profile toward the smaller end. This decrease in grain size means that more surface area is available for nucleation and growth of secondary phases. Secondly, high stresses result in a reduced porosity which can be directly related to a decrease in permeability (Figure 5), implying that the pore throats used for fluid transport would be much narrower. The Hagen-Poiseuille law states that a given mass of mineral precipitated along the wall of a pipe (or cylindical pore throat) will have the greatest relative effect on flow when the pipe is narrow. This is because flow rate through a pipe is proportional to the radius of the pipe to the fourth power (r4). In a narrow pipe, r4 is reduced by a proportionately greater amount per unit mass of precipitate formed. So, for a given mass of secondary mineral precipitate, permeability reduction will be greatest when the starting permeability is small.

It is obvious from figure 6, that the function used to simultaneously fit the entire data suite does not fit all the data perfectly. The discrepancy is likely a result of several factors. First, the function used is theoretically based [Aharonov et al., 1997b] and assumes that smectite is the only precipitating phase. This is probably a valid approximation, but according to geochemical modeling using the code EQ3/6 [Wolery et al, 1984], other phases are also thermodynamically stable. However, slow precipitation kinetics may have restricted their formation, rendering them unobservable using the SEM. Minor precipitation of other minerals could cause slight changes in permeability evolution which are not accounted for by the theoretical model. Secondly, although each experiment is assembled in the same manner, the sand grains will assume slightly different initial configurations. Any differences in axial compaction related to these packing variations will cause the pore network to change in subtle ways. These topologic discrepancies can have an observable effect on precipitation induced permeability reduction. In equation (1), the exponent of 2, which relates permeability to porosity, has been assumed to be independent of temperature and stress. In reality, this exponent would be affected by variable compaction since it is a function of pore shape and pore connectivity.

The theoretical expression (1) closely matches the experimental data from 125°C to 275°C. However, the model does not accurately describe the permeability results of the experiment conducted at 25°C. In the 25°C experiment, permeability is observed to increase in the early stages and then level off to a constant permeability. The interpretation of this result is that the kinetics of dissolution and precipitation are extremely slow at low temperature. Initially, the rise in permeability could be caused by flushing of fines from the main pore throats. This process may be important at all temperatures studied, but at high temperature the process is masked by the more predominant dissolution/precipitation reactions. The objective of this study was not to simulate the exact physical and chemical conditions encountered in nature, but rather to understand and quantify how diagenetic reactions affect the permeability of granular media temporally, and to determine how temperature and stress affect these reactions. The experiments in this study were conducted in a closed system with the specimen at a constant temperature, which in natural systems is seldom the case. Most natural systems usually allow a net removal or addition of mass via fluid flow. Furthermore, fluid flow is often driven by compaction or by various heat sources which cause the fluid to traverse isotherms. Changing temperature results in a continuous shift of fluid-rock equilibrium and subsequent dissolution or precipitation. In this study, intense dissolution and precipitation occur only while the fluid is out of equilibrium, which is only the case for the first several days. In a true flow-through experiment in which fresh fluid were continuously injected-these processes would continue and much greater permeability reduction would occur over much longer times.

Go to next section

Go back to table of contents