Modeling of flow and solutional processes within networks of interconnected conduits in limestone aquifers indicates that enlargement occurs very selectively during the early stages of karst aquifer development under laminar flow. If initial flow paths are uniform in size, almost all enlargement occurs along a single set of connected conduits that lie along a direct path between recharge and discharge locations and are aligned along the hydraulic gradient. With a sufficiently large variation in initial aperture widths, enlargement occurs along the flow path offering the least resistance to flow, but since flow rates in laminar flow are proportional to the fourth power of diameter but only linearly proportional to hydraulic gradient, the preferentially enlarged set of fractures may follow an indirect path. Results disfavor earlier suggestions that nonselective cave patterns result from artesian flows (at least under laminar flow conditions) and that all passages should be competitive until the onset of turbulent flow.

A simulation model developed to explore patterns of fracture enlargement within incipient limestone karst aquifers has been extended to turbulent flow. In contrast to the highly selective passage enlargement that occurs early in cave network development under laminar flow, the transition to turbulent flow results in more general passage enlargement, leading to maze networks when initial fractures are large and hydraulic gradients are high. These results support previously published hypotheses for the development of maze patterns, including formation within structural settings that have created initially large fractures or within flow systems periodically inundated by flooding. Maze development is also favored under turbulent flow when passages are entirely water filled, and where the groundwater flow system is long-lived. By contrast, branched patterns are most common when passages become free-surface subterranean streams, because depression of the piezometric surface along main passages, downcutting along main passages, and possible infilling with sediment of side passages limit the sharing of discharge among interconnected fractures or bedding planes that promote maze development.

Early theoretical investigations into initiation of limestone cave development suggested that water entering small fractures approaches saturation within a short distance and thus appeared to be unable to produce much further enlargement of these fractures. The answer to this problem seems to be White's (1977) "kinetic trigger" idea, emphasizing the precipitous drop in dissolution rates as equilibrium is approached. However, minimum conditions of flow rate, passage geometry, and initial CO2 concentration may exist, below which the waters will so nearly achieve saturation that caves will be unable to develop. Modeling results indicate that under geologically reasonable PCO2 and head gradient values, the minimum initial aperture widths in which caves will develop is of the order of 100/•m. Initial apertures wider than about 500 •m will form passages very easily, even under low head gradients. There will thus be strong differential selection among initial flow paths with regard to cavern passage development. IntroductionOne of the earliest theoretical efforts to study initiation of conduits within karst aquifers was made by Weyl [1958], who coupled calcite dissolution kinetics with laminar flow hydrodynamics. He assumed that diffusion of reaction products was the rate-limiting kinetic step, and calculated a "penetration length" for water entering initial fractures. This described the length that water can flow before reaching a given high level of saturation and which seemed very short compared to the observed lengths of natural cavernous conduits. Weyl's analysis showed that penetration lengths under what he considered natural conditions of fractures would be on the scale of millimeters to meters. Howard [1964] suggested that the first stages of initiation could be enhanced by local acid production, by either bacterially assisted oxidation of organic material [Murray and Love, 1929; Imbt, 1950] or sulfide minerals. Based on theoretical treatment and results of solution experiments, Howard [1964], and Howard and Howard [1967] also concluded that the solution process was not very accurately represented by evl s [1958] model of one limiting component. Curl's [1965] work stressed the slowness of CO2 hydration as a potential limiting step, offering longer penetration lengths. Other suggested mechanisms addressing the problem of short penetration lengths include convergence of flow, addition of oeresh solutions, and enhanced solutional ability due to mix-•ng of waters of differing composition, or "Mischungskorrosion" [B6gli, 1964; Howard, 1966' RunnelIs, 1969; Dreybrodt, 1981a]. Atkinson [1968] speculated that diurnal rock movements as a result of Earth tides or temperature fluctuations (as detected and measured by Davis [1968] in a tNow California cave) could pump undersaturated fluids in small fractures, thereby increasing early dissolution rates.The drop in dissolution rates that accompanies close approach to equilibrium has since been recognized to be an important part of the conduit development process in what White...

No abstract

An approach that couples flow rate measurement with mass balance calculations has been used to evaluate the geochemical evolution and kinetic behavior of Laurel Creek, West Virginia as it flows from a siliceous, noncarbonate caprock onto the Greenbrier Group Limestones, into and through the Laurel Creek‐Cross Road Cave System, and out of the spring at the lower end of the system. At a discharge of 1.0 m3 s−1, the stream was calculated to be removing calcite from the aquifer by dissolution at a rate of 4.1 mg 1−1 hr−1, for a total removal of 46 kg CaCO3/hr. Laurel Creek remained undersaturated with respect to calcite throughout the entire reach studied. Comparison of calculated saturation states with laboratory based kinetic mechanism information provided by the work of Plummer et al. (1978) suggests that the elementary reaction that describes attack by carbonic acid dominates calcite dissolution within the main conduit portion of the flow system.