top of page
Luiz Fernando Bez - Confirmed Speaker - MRPM 2026

Luiz Fernando Bez

Using Approximate Bayesian Computation (ABC) to analyze uncertainties in relaxometry experiments and their consequences in permeability estimation.

Nuclear Magnetic Resonance (NMR) techniques enable fast and non-destructive characterization of porous media, ideal for laboratory analysis and well logging in the field. Relaxometry is a technique that stands out among the various NMR tools, as it provides accurate measurements of the medium’s porosity, as well as an estimation of  pore size distributions within it. However, these observations are subject to noise levels and uncertainties inherent to the process. We propose a combination of Approximate Bayesian Computation (ABC) with reduced-order T2 distribution models as a framework for statistical analysis and uncertainty propagation. The T2 time distribution is decomposed as families of gaussian pore size distributions with uncertainty bounds associated with their medium pore size, standard deviation, and relative contribution for the relaxometry experiment. We show that these uncertainties can be propagated using traditional petrophysical correlations, such as the Kenyon model, thus providing upper and lower bounds to permeability estimations.

 

Acknowledgement/Funding: This study was financed, in part, by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brasil (CAPES), Brasil. Process Number 001

1) BEZ, L. et al. Large scale voxel-based fem formulation for NMR relaxation in porous media. Transport in Porous Media, v. 151, p. 2405–2430, 08 2024.

2) Toni, T., Welch, D., Strelkowa, N., Ipsen, A., Stumpf, M.: Toni t, welch d, strelkowa n, ipsen a, stumpf mp. Approximate bayesian computation scheme for parameter inference and model selection in dynamical systems. Journal of the Royal Society, Interface / the Royal Society 6, 187–202 (2008).

bottom of page