Bayesian Networks to Reduce Uncertainty Using Well Data in 3D Model of the Jeanne d’Arc Basin

Summary Basin models have many uncertain input parameters. Hence, instead of single solution simulation results, the results should be calculated and communicated in a probabilistic manner to optimize drilling decisions. We use data from basin models in Bayesian networks to create an integrated workflow that provides: 1. A structured way to vary model parameters, 2. A quantitative way to calibrate the model to observed data without manual adjustments, and 3. A way to update the understanding of model parameters when new data becomes available, without re-running simulations. An example of prolific Jeanne d’Arc basin is used to illustrate how the workflow facilitates constraining the source rock quality, thermal history, and migration pathways. We find that the Kimmeridgian source rock has terrigenous input in the Eastern part of the basin, a high heat flow corresponding to the Jurassic rift, and that the trans-basinal faults are impermeable in models that reproduce field measurements.