Scheda di dettaglio – i prodotti della ricerca
| Dato | Valore |
|---|---|
| Title | Porous media flux sensitivity to pore-scale geostatistics: A bottom-up approach |
| Abstract | Macroscopic properties of flow through porous media can be directly computed by solving the Navier-Stokes equations at the scales related to the actual flow processes, while considering the porous structures in an explicit way. The aim of this paper is to investigate the effects of the pore-scale spatial distribution on seepage velocity through numerical simulations of 3D fluid flow performed by the lattice Boltzmann method. To this end, we generate multiple random Gaussian fields whose spatial correlation follows an assigned semi-variogram function. The Exponential and Gaussian semi-variograms are chosen as extreme-cases of correlation for short distances and statistical properties of the resulting porous media (indicator field) are described using the Matèrn covariance model, with characteristic lengths of spatial autocorrelation (pore size) varying from 2% to 13% of the linear domain. To consider the sensitivity of the modeling results to the geostatistical representativeness of the domain as well as to the adopted resolution, porous media have been generated repetitively with re-initialized random seeds and three different resolutions have been tested for each resulting realization. The main difference among results is observed between the two adopted semi-variograms, indicating that the roughness (short distances autocorrelation) is the property mainly affecting the flux. However, computed seepage velocities show additionally a wide variability (about three orders of magnitude) for each semi-variogram model in relation to the assigned correlation length, corresponding to pore sizes. The spatial resolution affects more the results for short correlation lengths (i.e., small pore sizes), resulting in an increasing underestimation of the seepage velocity with the decreasing correlation length. On the other hand, results show an increasing uncertainty as the correlation length approaches the domain size. |
| Source | Advances in water resources |
| Keywords | pore-scale geostatisticsporous media generationLattice Boltzmann methodPore-scale flow simulationSeepage velocityMatérn covariance model |
| Journal | Advances in water resources |
| Editor | C.M.L. Publications,, Southampton, Regno Unito |
| Year | 2017 |
| Type | Articolo in rivista |
| DOI | 10.1016/j.advwatres.2017.02.002 |
| Authors | Paolo Roberto Di Palma, Nicolas Guyennon, Falk Hesse, Emanuele Romano |
| Text | 366885 2017 10.1016/j.advwatres.2017.02.002 Scopus 2 s2.0 85012115643 ISI Web of Science WOS WOS 000401379400008 pore scale geostatistics porous media generation Lattice Boltzmann method Pore scale flow simulation Seepage velocity Matern covariance model Porous media flux sensitivity to pore scale geostatistics A bottom up approach Paolo Roberto Di Palma, Nicolas Guyennon, Falk Hesse, Emanuele Romano Istituto di Ricerca sulle Acque Consiglio Nazionale delle Ricerche, Istituto di Ricerca sulle Acque Consiglio Nazionale delle Ricerche, Department of Computational Hydrosystems UFZ Helmholtz Centre for Environmental Research Germany, Istituto di Ricerca sulle Acque Consiglio Nazionale delle Ricerche Macroscopic properties of flow through porous media can be directly computed by solving the Navier Stokes equations at the scales related to the actual flow processes, while considering the porous structures in an explicit way. The aim of this paper is to investigate the effects of the pore scale spatial distribution on seepage velocity through numerical simulations of 3D fluid flow performed by the lattice Boltzmann method. To this end, we generate multiple random Gaussian fields whose spatial correlation follows an assigned semi variogram function. The Exponential and Gaussian semi variograms are chosen as extreme cases of correlation for short distances and statistical properties of the resulting porous media indicator field are described using the Matern covariance model, with characteristic lengths of spatial autocorrelation pore size varying from 2% to 13% of the linear domain. To consider the sensitivity of the modeling results to the geostatistical representativeness of the domain as well as to the adopted resolution, porous media have been generated repetitively with re initialized random seeds and three different resolutions have been tested for each resulting realization. The main difference among results is observed between the two adopted semi variograms, indicating that the roughness short distances autocorrelation is the property mainly affecting the flux. However, computed seepage velocities show additionally a wide variability about three orders of magnitude for each semi variogram model in relation to the assigned correlation length, corresponding to pore sizes. The spatial resolution affects more the results for short correlation lengths i.e., small pore sizes , resulting in an increasing underestimation of the seepage velocity with the decreasing correlation length. On the other hand, results show an increasing uncertainty as the correlation length approaches the domain size. Preprint Porous media flux sensitivity to pore scale geostatistics A bottom up approach 2017 Di Palma et al AWR Porous media flux sensitivity to pore scale geostatistics A bottom up approach.pdf Articolo in rivista C.M.L. Publications, 0309 1708 Advances in water resources Advances in water resources Adv. water resour. Advances in water resources. paolorobertodipalma DI PALMA PAOLO ROBERTO emanuele.romano ROMANO EMANUELE nicolasdominique.guyennon GUYENNON NICOLAS DOMINIQUE |