Moving Models into Reality: An Automated Workflow to Calibrate Forward Stratigraphic Modeling; Application to Hanifa and Arab-D in Central Saudi Arabia

Abstract

AbstractForward stratigraphic modelling (FSM) is an evolving technology for understanding the geology between wells for the purpose of exploration and field development. As opposed to the use of geostatistics, this process-based modeling approach uses physical equations for key controls on deposition, such as initial bathymetry, eustatic sea level change, subsidence rate, wave energy, and other environmental conditions. The output simulation is a 3D cellular model with properties like lithology, porosity, and water depth.FSM results do not honor the interpretation from the drilled wells. This means that the prediction accuracy from the generated model may not be high. The current process of calibrating forward stratigraphic models is time-consuming and tedious. In this paper, we propose an automated workflow to improve the accuracy of a forward stratigraphic model by automating its calibration to facies data from wells.For our case study, we use an initial stratigraphic model of the Hanifa and Arab-D in central Saudi Arabia. The modeling area covers 430 km by 370 km, the cell size is 10 km, and the simulation time step is 100,000 years. In the resulting model, cells are assigned to seven index facies based on their lithology, wave energy, and water depth. Initially, we conducted a sensitivity analysis to identify the environmental parameters with critical influence on the final model. Subsequently, we ran several simulations with varying values for these critical parameters. To measure the match between the different simulation models and the observed well facies, we used facies from 16 wells. The simulation run with the highest match was used as the best forward stratigraphic model.Uncertainty maps, based on superimposing several simulations, were generated to check which areas of the simulation are replicated more often than others, meaning they bear, relatively, the lowest uncertainty. This approach may be used to evaluate the risk of drilling new wells in specific locations, or to provide a measure of the uncertainty for subsequent reservoir simulations. In the future, we will use seismic facies and attributes, in addition to well data, for the model calibration.