Calibration of rock physics models using borehole acoustic logs and seismic data is a fundamental step in the interpretation of seismic inversion products into petrophysical properties. Sonic logs are affected by noise, environmental, and smoothing effects. However, rock physics models typically assume homogeneous spatial properties. Therefore, to obtain a reliable estimate of petrophysical and fluid properties from borehole measurements, the logs need to be noise-free and their volume of investigation needs to be consistent with the assumptions implicit in the rock physics model. To circumvent these problems, we introduce a new two-step workflow that reconciles acoustic logs and rock physics models. First, we “deconvolve” the spatial sensitivity function inherent to borehole logs to obtain layer-by-layer elastic properties with corresponding uncertainties. Then we estimate layer petrophysical properties via joint Bayesian inversion with Random Walk Markov chain Monte Carlo sampling. The proposed workflow reduces the computational time up to 1,000 times when compared to traditional Bayesian inversion. Rock physics models are calibrated efficiently and accurately even in the presence of noise. Field examples verify that the estimated petrophysical properties are consistent with core data.
Presentation Date: Monday, October 12, 2020
Session Start Time: 1:50 PM
Presentation Time: 1:50 PM
Location: Poster Station 4
Presentation Type: Poster
