The textural and chemical results discussed in the previous section indicate that the sand and fluid undergo significant changes during the course of each experiment. These changes, namely the growth of secondary mineral phases, are likely to affect the flow of water through the sand and result in a corresponding drop in permeability. In fact, the fluid chemistry suggests that the system's chemical state is closely linked to the observed permeability reduction. These observations suggest that the increases in solubility affect the rate and degree of precipitation and permeability reduction. From the data presented so far, it is clear that dissolution of primary quartz and labradorite results in the supersaturation of at least one additional mineral phase. However, the mere presence of a secondary mineral precipitate does not mean that all of the observed permeability change is caused by this authigenic phase. Other possibilities are that permeability reduction was induced by fines created during compaction or by compaction related to the minor creep observed during the four days of each experiment.
Several additional experiments have been conducted to constrain the mechanism by which permeability reduction occurs. As a first step, we tested whether the permeability changes are a result of mineral precipitation or of some other mechanical or flow-induced mechanism. This was achieved by conducting a quartz-only experiment. Quartz does not possess the chemical components necessary for clay mineral formation, so any permeability reduction that is observed cannot be a result of precipitation. Figure 9 shows the result of this quartz-only experiment. Unlike the quartz/labradorite experiments conducted under similar conditions, no discernible permeability change is observed. This indicates that all the permeability reduction observed in the reported experiments is a result of dissolution/precipitation reactions. If the permeability changes had been caused by the minor axial creep observed in all experiments or by clogging of pore throats by fine particulates, then a similar degree of permeability reduction should have been observed in the quartz-only experiment. Figure 9 shows this distinct difference between the quartz and quartz/labradorite experiments. The quartz/labradorite experiment shows a rapid decrease in permeability at early times with a leveling off at about 40% of the initial value, whereas the quartz-only experiment does not undergo any reduction of permeability over the four day duration of the experiment. Figure 9 also shows the results of a labradorite-only experiment. In this case, permeability reduces much more than in the analogous quartz/labradorite experiment. The reason for this behavior will be discussed below.
Another experiment was conducted to determine the effect of fluid flow on permeability reduction. In this experiment, a quartz/labradorite sand was subjected to a predominantly no-flow condition, with flow only being initiated briefly once or twice per day, so that permeability could be measured. The permeability data from this experiment is compared to the analogous continuous flow experiment on figure 10. It is clear that there is little difference in permeability evolution between the two experiments, indicating that the permeability change is not related to flow, and that the permeability reduction process is reaction-controlled rather than transport-controlled.
To this point, the data and observations clearly suggest that permeability reduction is caused by precipitation of a secondary mineral phase, probably a clay mineral such as smectite. The next question to answer is: Why does the rate of permeability reduction decrease with time? The most obvious answer is that because there is a fixed volume of pore fluid being circulated, the rate of precipitation slows with time due to chemical equilibration of the fluid with the starting mineral assemblage. If the solution becomes saturated with respect to labradorite, then dissolution will stop and no more Ca, Na or Al will be liberated for production of a secondary mineral species. System saturation with respect to labradorite occurs when (Q/K) = 1, where Q is the ion activity product (IAP) of the solution and K is the solubility constant for labradorite. The IAP is controlled by the activities of each dissolved species raised to some power which is determined by the composition of the dissolving labradorite. At early times, the solution is very undersaturated, so dissolution occurs at a fast rate. As the solution approaches equilibrium, dissolution slows, also causing authigenic phases to form at a slower rate.
To test this hypothesis, another quartz-labradorite experiment was conducted in which pore fluid was extracted in the middle of the experiment and replaced with fresh de-ionized water. If the proposed model is correct and permeability reduction rate is dependent upon the system saturation state, then the rate should increase once the system is replenished with new fluid. The reasoning behind this argument is that dilution of the fluid will once again cause the solution to become undersaturated with respect to quartz and labradorite. An undersaturated system leads to re-intensified dissolution of labradorite, with the components necessary for smectite formation (i.e. Ca, Al) being introduced to the fluid at a faster rate. Simply stated, once the system is flushed with de-ionized water, the rate of permeability reduction should increase to what it was near the beginning of the experiment. Figure 11 shows that there is clearly a significant change in the rate once the new fluid has been introduced to the system. Since a number of studies suggest that precipitation of smectite is rapid, relative to other silicate minerals [Hajash and Bloom, 1991; Zuddas and Michard, 1993], an increase in dissolution rate may be directly linked to greater precipitation and faster permeability reduction. There is a transient of permeability in the several cycles following the flush, which is believed to be caused by re-dissolution of previously formed smectite. This process quickly ceases upon saturation of the fluid with respect to smectite.
In this experiment, pore fluid was sampled periodically during the course of the experiment. As expected, Si rapidly increases in the early stages and then levels off after a few hours (Figure 12a). At the time of the pore fluid flush, approximately a day and a half into the the experiment, there is 90 ppm Si in the fluid. Ca and Al both increase rapidly during the first few hours, after which they begin to decrease slowly (Figures 12 b,c). This decrease is a result of mass removal from the fluid via clay mineral precipitation. Na does not show any peak and subsequent decrease (Figure 12 d), which is consistent with the EDX analysis of the precipitate, which found no Na present. The rapid increase of Si to concentrations that are much greater than all other elements may in part be attributed to the presence of quartz. However, fluid from the labradorite-only experiment also shows high Si concentrations, suggesting that the precipitating smectite has a greater Al/Si ratio than does the dissolving feldspar. Furthermore, the precipitation of any other Si-bearing phases must be negligible. Once the concentration of Si builds up in solution and dissolution of labradorite slows down (as described earlier), the chemical evolution of the system begins to reflect the precipitation process. Precipitation is evidenced by the concurrent decrease in Ca and Al several hours into the experiment. It should be noted that after their initial increase, the decrease of Al and Ca are exponential in form, similar to the permeability evolution curves.
Following the pore fluid flush, all elements markedly decrease except for Al, which increases by 60% over the course of several hours. This increase may be related to re-dissolution of previously formed precipitate as evidenced by the short- lived increase in permeability following the flush. The absence of any increase in Ca, Na or Si following the flush is probably a result of several factors. Before attempting to explain these results, some assumption must be made concerning the exact composition of the precipitating phase. According to the textural observations from this study, the dominant mineral precipitating is a Ca-rich smectite. Geochemical modeling using EQ3/6 [see companion paper by Aharonov et al.,1997b] indicates that the smectite would be Ca-beidellite of approximate composition Ca0.165Al2.33Si3.67O10(OH)2. Although a beidellite of such composition is rich in Si, re-dissolution would not cause a noticeable increase because of the abundance of Si in solution immediately preceding the flush. Stated another way, the mass of Si removed during the flush is much larger than the Si re-introduced to solution during redissolution of smectite. For Al, the equilibrium concentration is much lower, so that redissolution is more significant. In the case of Ca, it is the low molar abundance in beidellite which is likely responsible for the absence of any increase following the flush. Although equilibrium values for Ca are low, there is only one Ca atom liberated for every eight Al atoms during dissolution of beidellite, again causing any increase to be swamped by the reintroduction of new fluid. Predictably, Na shows no increase following the flush due to its absence in the authigenic beidellite.
The preceding test clearly shows that the shape of the permeability curves is closely linked to the saturation state of the fluid. However, the rate at which equilibrium is approached will depend on the rate constants for dissolution of labradorite and quartz. Furthermore, if the dissolution rate of quartz is much greater than for labradorite, then dissolved Si derived from quartz will control how rapidly the solution will attain labradorite equilibrium and therefore how much feldspar can be dissolved. If the effective dissolution rate of quartz is slower than that of labradorite, the system will behave as if no quartz is present with all Si being derived from labradorite. If such is the case, then the effect of quartz in the experiments is negligible.
Two tests may be used to determine the relative rates of dissolution for quartz and labradorite. The first is analytical and uses the chemical evolution data from the pore fluid flush experiment to predict expected chemical evolution at different relative rates of dissolution. In the companion paper, Aharonov et al. [1997b, fig. 6] have determined that the fluid chemistry indicates a quartz dissolution rate which is 60 times greater than that of labradorite. To experimentally confirm this result, an experiment using a labradorite-only sand was conducted (Figure 9). The total permeability drop for the labradorite-only experiment is an order of magnitude greater than for the quartz/labradorite experiment and the time-constant in the decay curve significantly longer. This result qualitatively indicates that quartz dissolves significantly faster than does labradorite. This increase in time constant is a manifestation that all Si is being derived from labradorite. Such a change in the time constant implies that the fluid takes longer to reach equilibrium, thus allowing more time for active dissolution of feldspar to occur. As a result, there is more Ca and Al introduced to solution which can eventually be used for the formation of smectite. This enhancement of Ca and Al-transfer to the fluid eventually results in more precipitation and a greater degree of permeability reduction.
These experimental results have thus isolated the processes responsible for the precipitation-sealing phenomenon observed. In the companion paper, Aharonov et al. [1997b] have used these constraints in constructing a theoretical model of the precipitation-sealing process. The curves superimposed on figure 6 are the resulting permeability evolution curves for different temperatures and stresses and are described by an equation of the following form:
K/Ko = [1-µ(1-exp(1.5e-5t))]^2 (1)
This expression essentially describes an equilibrating system precipitating one authigenic mineral phase. The exponent of 2 represents a possible exponent by which permeability and porosity may be related as precipitation occurs. The time constant in (1) is an average value and varies little between experiments, and can be thought of as representing the smectite precipitation rate. The constant µ is dependent on temperature and stress and essentially describes the point at which equilibrium is reached. A detailed description and derivation of this expression are given in Aharonov et al. [1997b].
The experimental system used in the present study is one in which temperature and mass are held constant (closed system) during any given run, with reaction only occurring while the fluid is out of equilibrium. In reality, geological systems are usually open, with precipitation often being caused by temperature induced solubility changes [i.e. Person et al., 1996]. Work on sedimentary basins around the world suggests that precipitation often occurs at a critical temperature [Hunt, 1990]. To test this idea and assess how it may fit into the chemical model proposed in this paper, an experiment was conducted in which temperature was stepped down incrementally (Figure 13), with the possible expectation that sudden drops in permeability will occur with each temperature change. These permeability changes might arise due to temperature induced supersaturation of clay phases or other minerals that are thermodynamically stable. In fact, figure 13 shows that no such drop in permeability is observed with decreasing temperature. The small offsets in permeability that are observed at each temperature change are simply a manifestation of uncertainty with temperature control and the subsequent mismatch when making the required viscosity correction in the permeability calculation. The effect of dropping temperature is to bring the solution closer to equilibrium with respect to feldspar, thereby slowing down the dissolution/precipitation process, but not causing a sudden drop in permeability. The decrease in dissolution rate is manifested by the distinct changes in dK/dt with each temperature step. At each temperature step, there is likely some temperature induced precipitation of silica or clay minerals, but this process is probably secondary to the slowing of dissolution and precipitation. In the final stage of the experiment, temperature is again increased to 200°C, upon which a resurgence of permeability reduction is observed. This second phase of intense permeability reduction is a result of suddenly undersaturating the fluid due to increased temperature. As in the pore fluid flush experiment, this temperature induced undersaturation is accompanied by enhanced dissolution and precipitation. It should also be noted that as in the flush experiment, the increase in temperature also causes a temporary increase in permeability before the enhanced reduction of permeability (inset, Figure 13). This increase in permeability occurs for the same reason; the fluid is undersaturated with respect to feldspar, quartz, and smectite, causing previously formed smectite to dissolve.
In this section it has been proposed that permeability reduction is caused by precipitation of a secondary mineral species, most likely Ca-beidellite. The rate at which precipitation occurs is determined by the equilibrium state of the primary mineralogy with respect to the reacting fluid, with Si being the dominant element controlling equilibrium. Several experiments have been conducted to test this idea. The results of all the experiments were successful in that they strongly support the proposed hypothesis. Permeability reduction has been modeled as a result of one precipitating phase, but in reality there are probably minor amounts of other thermodynamically stable minerals which also contribute to permeability reduction. For example, in a similar study done by Scholz et al. [1995], minor amounts of a fibrous mineral, probably a zeolite, were observed. Since these minerals were not observed in SEM in this study, they probably affect permeability to a much lesser extent and have little effect on the chemistry of the system.
Go to next section
Go back to table of contents