h1

TY  - THES
AU  - Liang, Rui
TI  - A universal mathematical model for porosity prediction of fluvial sediments
PB  - Rheinisch-Westfälische Technische Hochschule Aachen
VL  - Dissertation
CY  - Aachen
M1  - RWTH-2025-04980
SP  - 1 Online-Ressource : Illustrationen
PY  - 2025
N1  - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University
N1  - Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2025
AB  - The porosity of riverbed is a key structural property arising from the packs of fluvial sediments in varied sizes and shapes, which is defined as the ratio of pore volume to total volume. It is significant to nearly every investigation related to riverbed. For instance, morphologically, porosity determines the sediment concentration in the river bed and hence the rate of bed level changes. Ecologically, porosity governs the interstitial space of the hyporheic zone for aquatic habitats. Geologically, porosity dominates the exploitable reserve of oil, gas, and groundwater stored in the voids of fluvial deposits. Despite its important role, information regarding the spatial variations in porosity is rarely available in riverbed. Instead, porosity is often simply assumed to be spatially constant, which could cause a systematic error in morphological, ecological, and geological studies. The reason for this is partly due to the costly and arduous effort for in-situ measurements on porosity. As an alternative, mathematical porosity predictors turn out to be an effective way to estimate porosity based on porosity-controlling factors, such as grain size, grain shape and packing state. However, so far, no such a model can provide satisfactory results in terms of universality, accuracy, and efficiency. Regression-based models, while simple to use, is often insufficient when utilized in regions outside the original dataset. On the other hand, existing analytical models despite their general usefulness, are complex to compute and have been found to systematically underestimate porosity due to their intrinsic assumptions.In this thesis, the objective was to develop a novel mathematical porosity predictor that is general, accurate, and simple to apply. As a first step, the grain size effect on porosity was explored by assuming sediment shape as spherical. Unlike traditional analytical models that are typically derived from the analysis of binary mixtures of spheres, and then extended into complex models for arbitrary spherical packings, this study reverses such process by conceptualizing arbitrary spherical packings into a binary spherical mixture. This was achieved based on a newly proposed binary-unit concept, which states that any multi-sized (or continuous) spherical mixture can be transformed into an equivalent binary-unit mixture of spheres through the link of identical grain size statistics of mean, standard deviation and skewness. The obtained binary mixture is actually the most elementary spherical packing unit that can equivalently represent the diversity of intraparticle interactions in the original spherical mixtures, i.e., the mixing and unmixing effects. With this concept, the model, namely the binary-unit conceptual (BUC) packing model, can be readily implemented to estimate the porosity of complex spherical packings solely by leveraging models capable of predicting the porosity of a binary spherical packing. The Westman-equation model is recommended for this purpose. Validation against 85 digital riverbeds of spheres generated through a validated non-smooth granular dynamics (NSGD) algorithm suggested that the BUC packing model is able to provide very accurate porosity predictions, producing a root-mean-square error (RMSE) of 0.01. Next, the non-spherical grain shape effect was integrated into the BUC packing model in order to fully resolve the porosity of fluvial sediments. Initially, an ideal regular shape was employed to simplify the complex grain shapes of fluvial sediments. 241 of sediment particles were scanned in high quality, and then compared to four candidate regular shapes: cuboid, elliptic disk, truncated octahedron, and ellipsoid. And it was found that the ellipsoid renders the best shape similarity to fluvial sediments, allowing it as a reasonable surrogate. Following the concept of equivalent packing diameter, a non-spherical (ellipsoid) sediment mixture can then be converted into a spherical packing with an equivalent size effect on porosity that can be well handled by the BUC model, alongside an initial porosity capturing the isolated non-spherical shape effect at a specific packing stage. The three theoretical transformations, i.e., from sediment to ellipsoid packing, from ellipsoid to spherical packing, and from spherical to binary-unit spherical packing, form the foundation of the integrated BUC (IBUC) packing model. As a result, the IBUC packing model requires only two inputs: the grain size distribution (GSD) of the transformed spherical packing, and the initial porosity. It demonstrated that the GSD of the spherical packing can be well approximated with the measured GSD of the original sediment packing. For practical purposes, the use of a measured mean initial porosity has been proposed as a general representation for a local site being investigated. Despite this simplification, the IBUC packing model still achieved accurate porosity predictions with RMSE of 0.03, when validated against 138 porosity measurement data across four diverse riverbeds: the Rhine, Bès, Galabre, and Kuqa. Overall, the generality, simplicity, and prediction performance of the IBUC packing model positions itself as a state-of-the-art tool for investigating the spatial variability in riverbed porosity. In addition, as a key component of the IBUC model, the binary-unit concept is expected to go beyond porosity estimation, as intraparticle interactions impact a range of other factors. The potential applications involve estimation of permeability in sediment mixtures, determination of the cut-off grain size for morphological alterations, and even prediction of the incipience of sediment transport.
LB  - PUB:(DE-HGF)11
DO  - DOI:10.18154/RWTH-2025-04980
UR  - https://publications.rwth-aachen.de/record/1012294
ER  -