Inferring pressure profiles from neutron data through Bayesian calibration

Peter Moonen (1, 2) , Hannelore Derluyn (1)

1 LFCR - Laboratoire des Fluides Complexes et leurs Réservoirs
2 DMEX - Développement de méthodologies expérimentales

We propose employing Bayesian calibration to match observed neutron data with calculated moisture contents. Hereby the calibration effort aims at minimizing the uncertainty regarding the experimental boundary conditions, after which the model can be interrogated for pressure data. The proposed formulation explicitly accounts for image noise and model bias. The approach is illustrated for the case of evaporative drying in limestone.

Mots clés: neutron radiography; Bayesian calibration; Gaussian Processes

Domaines: Science des matériaux [cond-mat.mtrl-sci]; Mécanique des fluides [physics.class-ph]; Mécanique des matériaux [physics.class-ph]; Génie mécanique [physics.class-ph]; Mécanique des matériaux [physics.class-ph]; Géophysique [physics.geo-ph]

Peter Moonen, Hannelore Derluyn. Inferring pressure profiles from neutron data through Bayesian calibration. 2nd International Conference on Tomography of Materials and Structures ICTMS 2017, Jun 2017, Lund, Sweden. ⟨hal-01815873⟩