Stochastic simulations of discrete karstic networks

in: 2020 RING Meeting, ASGA

Abstract

Various approaches exist to model underground flows in karstic settings. Among them, some recent software programs allow running flow simulations on explicit conduit networks (e.g., SWMM, MODFLOW-CFP, Epanet, GROUNDWATER). This opens new avenues to more accurately evaluate the impact of various karstic network architectures on underground flows. This paper presents the key points of a new method aiming to stochastically simulate 3-dimensional discrete karstic networks, necessary to perform such flow impact assessment. This method is designed to directly take hard data into account (known entries and exits of the networks) and to use geological information as conditioning data. The connectivity of the network is imposed thanks to a ratio between the number of intersections and extremities. This should allow us to simulate various types of networks (e.g., vadose, phreatic). The first step of this method is to define a probability grid in the studied zone and draw a chosen number of points inside it. The probability associated to each voxel depends directly on the conditioning data. The number of connections of each point is then computed, respecting the intersections/extremities ratio given in input. Lastly, the connections between the nodes are automatically generated. We present different examples of application of this method, still in its early development stage, and comment the obtained results.

Download / Links

BibTeX Reference

@inproceedings{FRANTZ_RM2020,
 abstract = { Various approaches exist to model underground flows in karstic settings. Among them, some recent software programs allow running flow simulations on explicit conduit networks (e.g., SWMM, MODFLOW-CFP, Epanet, GROUNDWATER). This opens new avenues to more accurately evaluate the impact of various karstic network architectures on underground flows. 
This paper presents the key points of a new method aiming to stochastically simulate 3-dimensional discrete karstic networks, necessary to perform such flow impact assessment. This method is designed to directly take hard data into account (known entries and exits of the networks) and to use geological information as conditioning data. The connectivity of the network is imposed thanks to a ratio between the number of intersections and extremities. This should allow us to simulate various types of networks (e.g., vadose, phreatic).
The first step of this method is to define a probability grid in the studied zone and draw a chosen number of points inside it. The probability associated to each voxel depends directly on the conditioning data. The number of connections of each point is then computed, respecting the intersections/extremities ratio given in input. Lastly, the connections between the nodes are automatically generated.
We present different examples of application of this method, still in its early development stage, and comment the obtained results. },
 author = { Frantz, Yves AND Collon, Pauline },
 booktitle = { 2020 RING Meeting },
 publisher = { ASGA },
 title = { Stochastic simulations of discrete karstic networks },
 year = { 2020 }
}