Integration of geological knowledge in karstic network simulation with L-systems

Pauline Mourlanette and Pierre Anquez and Guillaume Rongier and Pauline Collon. ( 2016 )
in: 2016 RING Meeting, ASGA

Abstract

Well data provide essential information to better characterize karst networks and reduce the uncertainties related to these potential reservoirs. Better estimating these uncertainties calls for simulations of the possible karst geometries conditioned to the data. This paper introduces a method based on a formal grammar, the Lindenmayer system, to simulate both branchwork and anastomotic patterns of karst networks, including topology control and hard data conditioning. In a first step, the object-based method simulates a karst network with a branchwork pattern using L-systems while attraction vectors ensure hard data conditioning. A second step creates loops by reconnecting some conduits to others. During the entire simulation, the karst growth is influenced by input parameters relating to the karst topology. The resulting networks, currently simulated in a 2D local space, show a complete range of patterns going from pure branchwork to pure anastomotic ones. The domain limits and conditioning of hard data are always respected, but the control of karst topology, as proportion of bifurcations, is still at an early stage. Unwanted intersections also still appear. Integration of repulsive and soft data thus appear as short-term perspectives of improvement.

Download / Links

BibTeX Reference

@INPROCEEDINGS{,
    author = { Mourlanette, Pauline and Anquez, Pierre and Rongier, Guillaume and Collon, Pauline },
     title = { Integration of geological knowledge in karstic network simulation with L-systems },
 booktitle = { 2016 RING Meeting },
      year = { 2016 },
 publisher = { ASGA },
  abstract = { Well data provide essential information to better characterize karst networks and reduce the
uncertainties related to these potential reservoirs. Better estimating these uncertainties calls for
simulations of the possible karst geometries conditioned to the data. This paper introduces a
method based on a formal grammar, the Lindenmayer system, to simulate both branchwork and
anastomotic patterns of karst networks, including topology control and hard data conditioning. In
a first step, the object-based method simulates a karst network with a branchwork pattern using
L-systems while attraction vectors ensure hard data conditioning. A second step creates loops by
reconnecting some conduits to others. During the entire simulation, the karst growth is influenced
by input parameters relating to the karst topology. The resulting networks, currently simulated in a
2D local space, show a complete range of patterns going from pure branchwork to pure anastomotic
ones. The domain limits and conditioning of hard data are always respected, but the control of
karst topology, as proportion of bifurcations, is still at an early stage. Unwanted intersections
also still appear. Integration of repulsive and soft data thus appear as short-term perspectives of
improvement. }
}