Geocellular rendering of cave surveys in paleokarst reservoir models

Infilled and collapsed cave systems are an important component of many paleokarst reservoirs. Incorporating these features into industrial reservoir models commonly relies on geostatistical modelling methods that often fail to capture key aspects of connectivity, geometry and volume of the paleokarst features realistically. The present work investigates the implementation of realistic cave geometries in geocellular models using survey data from an active karst cave as a starting point. The proposed method utilizes cave survey data to generate a dense equally spaced point-cloud representing the cave system. The point-clouds are used for geometric modelling and subsequent geocellular discretization of the karst system. The volumetric and geometric accuracy of this novel reservoir modelling method is compared to that from two established methods by benchmarking against the cave survey data. Additionally, the interlinkage between grid cell resolution, applied filter cut-off and geocellular rendering are evaluated. This study demonstrates that our proposed novel methodology can provide an excellent geometric and volumetric geocellular rendering of karst systems using cave survey data as input. Employing a combination of cave network maps and forward modelling of collapse and infill may enable model rendering of these features that more closely echoes processes controlling cave and karst breccia formation and geometric characteristics. In turn, this could offer better constraints to forecast paleokarst reservoirs architecture and properties.