Multimodal reservoir porosity simulation: An application to a tight oil reservoir

At appraisal stage of a reservoir characterization, a key step is the inference of the reservoir static properties, such as porosity. In this study, we present a new nested workflow that optimally integrates 3D acoustic impedance and geophysical log data for the estimation of the spatial distribution of reservoir porosity, which is applied to a tight sandstone oil reservoir located in Quebec, Canada. In this workflow, 3D seismic is the main source of spatial information. First, the statistical petrophysical relationship between acoustic impedance and reservoir porosity is established using collocated geophysical log data. Second, a conventional least-squares post-stack inversion of the impedance is computed on the seismic grid. The fit between well log data and numerically computed traces was found to be inaccurate. This leads to the third step, involving a post-stack stochastic impedance inversion using the same seismic traces not only to improve well and trace fit but also to estimate the uncertainty on the inverted impedances. Finally, a Bayesian simulation algorithm adapted to the estimation of a multi-modal porosity distribution is used to simulate realizations of porosity over the entire seismic grid. Results show that the over-smoothing effect of least-squares inversion has a major impact on resource evaluation, especially by not reproducing the high-valued tail of the porosity distribution. The adapted Bayesian algorithm combined with stochastic impedance inversion thus allows a better reproduction of the porosity distribution and improves estimation of the geophysical and geological uncertainty. •Bayesian sequential simulations were used to infer porosity in an oil reservoir.•BSS algorithm takes into account multi-modal distribution of porosity.•Stochastic seismic inversion improves correlation between well data and seismic.•Using the BSS algorithm with SSI cube reduces the bias on porosity realizations.