Results of the simulation of the evolution of fracture networks show that fracture networks with homogeneous spatial fracture distribution yield fracture development resembling worm-holes in porous media.
The application of percolation clusters as initial fracture networks allows to develop networks that resemble the topology of real caves. However, flow and dissolution tend to homogenize fracture cross sections along the developed fracture.
Simulating heterogeneity in dissolution rates produces a more heterogeneous distribution of fracture cross sections geometry. An unexpected result is that heterogeneity in dissolution rates control the fractures that are enlarged by dissolution. Fractures with lower reaction rate are enlarged before fractures with higher reaction rate because water can travel longer into the fracture network without getting saturated and then connecting the injection points with the outlet. The fractures that control the development of the geometry are at the end of the dissolution front, the maximum distance measured from the injection point at which water still can dissolve and enlarge fractures.
All simulations showed that fracture aperture after dissolution tends to a bi-modal dis-tribution. The second mode represents the enlarged fractures and it could be an exponential or a log-normal distribution similar to the one of real caves shown in figure 6.1. What has been surveyed in caves could be the second peak. The first peak is impossible to survey be-cause fractures be-caused by stresses on the rock are too small to explore but might contribute significantly to flow.
Results show that adding more initial information to the simulations, such as reaction rate heterogeneity, yields more realistic networks. However, the initial problem is precisely the lack of information about the system, either stratigraphical data, fracture network geometry or flow rates. Thus, the simulation of fracture enlargement can produce meaningful results only if some reliable information in known for the system.
The results presented in this chapter are only a first attempt to investigate the statistical and spatial distribution of conduit diameters in a karstic network. Further work is still needed to characterize in a systematic manner the networks and cross sections obtained by simulating mineral dissolution. The role of the geological history of a site with the evolution of the boundary conditions and accounting for a reasonable model of the heterogeneity of the sediments is still to be investigated.
Conclusions
7.1 Main results
In this work, we applied the concepts of reactive transport and calcite dissolution kinetics to develop simulations tools to explore the development of caves in the study area. The main questions were: why large karst systems developed in the study area? and, what can we learn from these caves to understand other caves?
The following is a summary of the results discussed in the context of the main questions of the thesis. Also, possible future developments are proposed for theoretical speleogenesis research and interpretation of data available from the study area.
Why large karst systems developed in the eastern coast of the Yucatan Peninsula?
In chapter 3, the particular characteristics of the study area were presented in detail and are enumerated here:
1. The karst aquifer in the area of Tulum has a high primary porosity and permeability resulting from its geological history which is described as eogenetic.
2. High primary and porosity are related to Pleistocene sediments deposited in prograding coastline reef-rimmed environment. However, the horizontal and vertical extent of Pleistocene sediments is not known in detail. It is possible that most of the karst system is located in the porous and friable limestones.
3. The low relief topography of the Yucatan platform combined with sea level fluctuations due to glaciations have caused displacement of the coastline in the scale of kilometers.
A moving coastline implies displacement of flow outlet of the karst system, displacement of the mixing zone between the freshwater and saline layers in the aquifer and variability in the groundwater flow direction in the conduit network. The combined effect of these processes may contribute to the apparent lack of order or preferential development of direction of the karst network.
4. It was observed that the largest karst network of the study area is located near an area where sea level drops to 300 meters in a relatively short horizontal distance. Under lower sea level conditions, this relief provides stronger hydraulic gradient compared to neighboring zones with gentler carbonate platform slopes.
The main novel result of this thesis is the conceptual model of karst conduit development by tidal pumping presented in chapter 5. It was shown that exchange of reactive water between a karst feature and the porous rock matrix around it, can produce dissolution relevant to the development of karst networks. Tidal pumping is stronger closer to the coast, but the coastline has been moving over the whole study area due to sea level changes.
To further investigate the theory of karst development by tidal pumping, it would be worth obtaining and analyzing a set of rock samples from the field. If a dissolution front is developing due to tidal pumping, differences of porosity should be observed between a sample near the cave wall surface and samples deeper on the wall. Drilling underwater is a complicated task, but there are also dry caves in the study area that developed in underwater conditions. Core drillings could also be done in other coastal karst networks where the conditions for tidal pumping are met.
Are existing conceptual models of speleogenesis adequate to describe karst networks on the study area?
The study area is different to other regions because limestone is quite permeable at the pore scale. Thus, groundwater flow occurs in karst conduits and if hydraulic conditions are met, water exchange can occur between the karst conduit and the porous wall around it. This exchange, known as tidal pumping, dissolves limestone not only at the conduit wall surface but also at the interior of the wall.
Based on these observations from the study area a conceptual framework and a simulation tool were produced to simulate dissolution by tidal pumping. Simulation results show that the action of tidal pumping combined with rock properties and groundwater chemistry from the study can lead to significant conduit enlargement rates. The simulation results contribute to explain the development of large karst cavities found in the study area.
What can we learn from these caves to understand other caves?
This question aims to respond to the initial problem of information gaps on cave geometry.
Exploring and surveying karst networks is a challenging task if not impossible. A simulation tool was developed to explore the evolution of fracture networks driven by mineral dissolution.
Results of the simulations are networks with topological and geometrical properties that can be compared to the properties of real networks.
The work done during the time frame of the thesis generated more questions than answers.
We observed that the combined action of groundwater flow and dissolution can produce a statistical distribution of fracture apertures (conduit cross section size) that shares some characteristics with the one observed in real karst networks. However, flow and dissolution are not sufficient to generate the spatial distribution of fracture aperture if we assume a simple setup. Instead, it is necessary to supply more information to the model such as heterogeneity in dissolution rates to be able to generate networks having a spatial distribution of aperture closer to real networks. However, this is a paradoxical situation because the purpose of simulating speleogenesis was to bridge gaps on karst network data. If reliable structural geology, stratigraphical and mineralogy data is available, a reliable simulation of speleogenesis can be done.