Nowadays many technological and environmental applications have fostered the interest of groundwater literature towards the stochastic modeling of flow and transport. A growing attention has been recently devoted to the attempt of quantifying the effect of non-Gaussian frequency distributions exhibited by many hydro-geological properties, such as the hydraulic conductivity, on the subsurface transport of contaminants. The motivation behind this approach, against the classic deterministic one, is the wide spatial variability featured by those properties in natural formations, which prevents reducing transport to the simpler Gaussian case. As part of the theoretical framework described, this master thesis focuses on the study of conservative transport in mildly heterogeneous hydraulic conductivity fields under mean uniform steady flow conditions. The crucial goals are to validate, by means of numerical simulations, the analytical expressions for Fickian macrodispersion analogues developed in the context of the Generalized sub-Gaussian (GSG) model by Riva, Guadagnini and Neuman for infinite two-dimensional domains and extend them to three-dimensional fields. Numerical results are post-processed with the Monte Carlo method, adopting a Sequential Gaussian Simulator (SGSIM) to generate conductivity fields conditioned to prescribed ensemble means and covariance functions, later rendering them sub-Gaussian by means of statistical subordination. The flow equation is solved numerically for large regular grids and transport is simulated via Random-Walk Particle Tracking (RWPT) applied to pure advection. This work shows first synthetic results on two-dimensional domains at increasing degree of heterogeneity, extending later to slightly heterogeneous three-dimensional fields. These results yield good agreements between analytical and numerical estimates for mild heterogeneity, the matching worsening for markedly heterogeneous fields. Yet the analytical model reveals a successful outcome, at least for assessing the correct orders of magnitude, if heterogeneity keeps moderate. Analytical estimates can thus reasonably replace numerical ones in several cases, depending on the level of accuracy required, enabling to reduce the computational times and opening the path to future studies with reacting flows in strongly heterogeneous fields.
Lo scenario tecnologico e le problematiche ambientali hanno oggigiorno volto l’interesse della ricerca nel campo dell’ingegneria idraulica alla modellazione stocastica dei processi di flusso e trasporto nel sottosuolo. In particolare, si dedica sempre più attenzione allo studio del trasporto sotterraneo di contaminanti in campi di conducibilità idraulica non Gaussiani. L’adozione di modelli stocastici trova ampia giustificazione nell’estrema variabilità spaziale che i parametri idrogeologici mostrano in natura nei mezzi porosi eterogenei, rendendo impossibile ricondurre i fenomeni di trasporto al comune caso Gaussiano. Questa tesi si propone, nel contesto sopra descritto, di studiare il trasporto conservativo di soluto in campi di conducibilità idraulica moderatamente eterogenei, subordinato a condizioni di flusso mediamente uniformi e stazionarie. L’obbiettivo primario è quello di convalidare numericamente le espressioni analitiche di macrodispersione sviluppate mediante il modello Generalizzato sub-Gaussiano (GSG) da Riva, Guadagnini e Neuman per domini bidimensionali infiniti ed estenderle al caso tridimensionale. I risultati numerici sono elaborati attraverso il metodo Monte Carlo. La generazione dei campi avviene con un Simulatore Gaussiano Sequenziale (SGSIM) che condiziona a valori medi e funzioni di covarianza prefissati. I campi di conducibilità idraulica sono quindi resi sub-Gaussiani attraverso una subordinazione statistica. L’equazione di flusso è risolta con metodi numerici e il trasporto simulato con algoritmi di Random-Walk Particle Tracking (RWPT) applicati a casi puramente avvettivi. Questo testo si incentra inizialmente sull’analisi di casi bidimensionali di trasporto in campi progressivamente più eterogenei, focalizzandosi infine su un dominio tridimensionale leggermente eterogeneo. Tali simulazioni consentono di confrontare i risultati numerici e analitici circa l’evoluzione temporale dei coefficienti di macrodispersione, che rivelano un buon grado di concordanza, specialmente per campi poco eterogenei, e una discrepanza crescente all’aumentare dell’eterogeneità. É altresì osservabile che, a patto di limitare la trattazione al caso di domini moderatamente eterogenei, il modello analitico produce stime valide (almeno in termini di ordini di grandezza), trovando così applicazione in svariate situazioni a seconda dell’accuratezza richiesta. Ciò garantisce non solo di ridurre drasticamente i tempi computazionali, bensì apre la strada a future estensioni a casi di trasporto reattivo di contaminanti in campi fortemente eterogenei.
Analytical expressions for macrodispersion in three-dimensional randomly heterogeneous non-gaussian conductivity fields
CERESA, LAURA
2017/2018
Abstract
Nowadays many technological and environmental applications have fostered the interest of groundwater literature towards the stochastic modeling of flow and transport. A growing attention has been recently devoted to the attempt of quantifying the effect of non-Gaussian frequency distributions exhibited by many hydro-geological properties, such as the hydraulic conductivity, on the subsurface transport of contaminants. The motivation behind this approach, against the classic deterministic one, is the wide spatial variability featured by those properties in natural formations, which prevents reducing transport to the simpler Gaussian case. As part of the theoretical framework described, this master thesis focuses on the study of conservative transport in mildly heterogeneous hydraulic conductivity fields under mean uniform steady flow conditions. The crucial goals are to validate, by means of numerical simulations, the analytical expressions for Fickian macrodispersion analogues developed in the context of the Generalized sub-Gaussian (GSG) model by Riva, Guadagnini and Neuman for infinite two-dimensional domains and extend them to three-dimensional fields. Numerical results are post-processed with the Monte Carlo method, adopting a Sequential Gaussian Simulator (SGSIM) to generate conductivity fields conditioned to prescribed ensemble means and covariance functions, later rendering them sub-Gaussian by means of statistical subordination. The flow equation is solved numerically for large regular grids and transport is simulated via Random-Walk Particle Tracking (RWPT) applied to pure advection. This work shows first synthetic results on two-dimensional domains at increasing degree of heterogeneity, extending later to slightly heterogeneous three-dimensional fields. These results yield good agreements between analytical and numerical estimates for mild heterogeneity, the matching worsening for markedly heterogeneous fields. Yet the analytical model reveals a successful outcome, at least for assessing the correct orders of magnitude, if heterogeneity keeps moderate. Analytical estimates can thus reasonably replace numerical ones in several cases, depending on the level of accuracy required, enabling to reduce the computational times and opening the path to future studies with reacting flows in strongly heterogeneous fields.File | Dimensione | Formato | |
---|---|---|---|
master_thesis_Ceresa_Laura.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
24.54 MB
Formato
Adobe PDF
|
24.54 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/142480