Search

Obtaining high-resolution maps of precipitation data can provide key insights to stakeholders to assess a sustainable access to water resources at urban scale. Mapping a nonstationary, sparse process such as precipitation at very high spatial resolution requires the interpolation of global datasets at the location where ground stations are available with statistical models able to capture complex non-Gaussian global space-time dependence structures. In this work, we propose a new approach based on capturing the spatial dependence of a latent Gaussian process via a locally deformed Stochastic Partial Differential Equation (SPDE) with a buffer allowing for a different spatial structure across land and sea. The finite volume approximation of the SPDE, coupled with Integrated Nested Laplace Approximation ensures feasible Bayesian inference for tens of millions of observations. The simulation studies showcase the improved predictability of the proposed approach against stationary and no-buffer alternatives. The proposed approach is then used to yield high resolution simulations of daily precipitation across the United States.

Anomalous transport in advection-dominated convergent flow tracer tests can occurs due to small-scale heterogeneities in aquifer hydraulic properties. These result in fluctuations of the groundwater velocity field and complex connectivity patterns between injection and extraction wells. While detailed characterization of heterogeneity is often not possible in practice, a proper understanding of what fundamental physical mechanisms can give rise to macroscopic behaviors that are measurable is essential for proper upscaling of solute transport processes. We analyze here how heavy-tailed breakthrough curves can arise in radially convergent flow to a well. The permeability fields are three-dimensional multi-Gaussian fields with varying statistical geometry and degrees of heterogeneity. We consider transport of conservative tracers from multiple injection locations by varying distance and angle from the extraction well. Anomalous power law tailing in breakthrough curves is attributed to a variety of features including the initial vertical stratification of the solute that arises due to a flux-weighted injection, the injection distance to the well relative to the depth of the aquifer, and the statistics of the heterogeneity field as defined by the correlation length and variance of the permeability. When certain conditions cooccur for a given injection, such as strong connectivity contrasts between aquifer layers, injection distances comparable to the horizontal heterogeneity integral scales, and large global variances, breakthrough curves tend to scale as a PL with unit slope at late time. These findings offer new insights to understand what physical processes must be understood to develop and choose appropriate upscaling approaches that might reproduce such anomalous transport in heterogeneous advection-dominated systems.

Assessing health risk in hydrological systems is an interdisciplinary field. It relies on the expertise in the fields of hydrology and public health and needs powerful translation concepts to provide decision support and policy making. Reliable health risk estimates need to account for the uncertainties and variabilities present in hydrological, physiological, and human behavioral parameters. Despite significant theoretical advancements in stochastic hydrology, there is still a dire need to further propagate these concepts to practical problems and to society in general. Following a recent line of work, we use fault trees to address the task of probabilistic risk analysis and to support related decision and management problems. Fault trees allow us to decompose the assessment of health risk into individual manageable modules, thus tackling a complex system by a structural divide and conquer approach. The complexity within each module can be chosen individually according to data availability, parsimony, relative importance, and stage of analysis. Three differences are highlighted in this paper when compared to previous works: (1) The fault tree proposed here accounts for the uncertainty in both hydrological and health components, (2) system failure within the fault tree is defined in terms of risk being above a threshold value, whereas previous studies that used fault trees used auxiliary events such as exceedance of critical concentration levels, and (3) we introduce a new form of stochastic fault tree that allows us to weaken the assumption of independent subsystems that is required by a classical fault tree approach. We illustrate our concept in a simple groundwater‐related setting

Construction of underground structures (e.g., subway lines, railways and highway tunnels) is inherently hazardous, posing risks to both workers and machinery at the site and to surrounding buildings. The presence of groundwater may increase these risks. We develop a general probabilistic risk assessment (PRA) framework to quantify risks driven by groundwater to the safety of underground constructions. The proposed approach is fully compatible with standard PRA practices, employing well-developed risk analysis tools based on the fault tree analysis method. The novelty and computational challenges of the proposed approach stem from the reliance on a combination of approaches including extracting information from databases, solving stochastic differential equations, or relying on expert judgment to compute probabilities of basic events. The general framework is presented in a case study and used to estimate and minimize risks at a construction site of an underground station for a new subway line in the Barcelona metropolitan area. &nbsp

Aquifer remediation is a challenging problem with environmental, social, and economic implications. As a general rule, pumping proceeds until the concentration of the target substance within the pumped water lies below a prespecified value. In this paper we estimate the a priori potential failure of the endpoint of remediation due to a rebound of concentrations driven by back diffusion. In many cases, it has been observed that once pumping ceases, a rebound in the concentration at the well takes place. For this reason, administrative approaches are rather conservative, and pumping is forced to last much longer than initially expected. While a number of physical and chemical processes might account for the presence of rebounding, we focus here on diffusion from low water mobility into high mobility zones. In this work we look specifically at the concentration rebound when pumping is discontinued while accounting for multiple mass transfer processes occurring at different time scales and parametric uncertainty. We aim to develop a risk‐based optimal operation methodology that is capable of estimating the endpoint of remediation based on aquifer parameters characterizing the heterogeneous medium as well as pumping rate and initial size of the polluted area.

Turbulence causes rapid mixing of solutes and fine particles between open channel flow and coarse-grained streambeds. Turbulent mixing is known to control hyporheic exchange fluxes and the distribution of vertical mixing rates in the streambed, but it is unclear how turbulent mixing ultimately influences mass transport at the reach scale. We used a particle-tracking model to simulate local- and reach-scale solute transport for a stream with coarse-grained sediments. Simulations were first used to determine profiles of vertical mixing rates that best described solute concentration profiles measured within a coarse granular bed in flume experiments. These vertical mixing profiles were then used to simulate a pulse solute injection to show the effects of turbulent hyporheic exchange on reach-scale solute transport. Experimentally measured concentrations were best described by simulations with a non-monotonic mixing profile, with highest mixing at the sediment-water interface and exponential decay into the bed. Reach-scale simulations show that this enhanced interfacial mixing couples in-stream and hyporheic solute transport. Coupling produces an interval of exponential decay in breakthrough curves and delays the onset of power-law tailing. High streamwise velocities in the hyporheic zone reduced mass recovery in the water column and caused breakthrough curves to exhibit steeper power-law slopes than predictions from mobile-immobile modeling theory. These results demonstrate that transport models must consider the spatial variability of streamwise velocity and vertical mixing for both the stream and the hyporheic zone, and new analytical theory is needed to describe reach-scale transport when high streamwise velocities are present in the hyporheic zone.

The reconstruction of smooth density fields from scattered data points is a procedure that has multiple applications in a variety of disciplines, including Lagrangian (particle-based) models of solute transport in fluids. In random walk particle tracking (RWPT) simulations, particle density is directly linked to solute concentrations, which is normally the main variable of interest, not just for visualization and post-processing of the results, but also for the computation of non-linear processes, such as chemical reactions. Previous works have shown the superiority of kernel density estimation (KDE) over other methods such as binning, in terms of its ability to accurately estimate the "true" particle density relying on a limited amount of information. Here, we develop a grid-projected KDE methodology to determine particle densities by applying kernel smoothing on a pilot binning; this may be seen as a "hybrid" approach between binning and KDE. The kernel bandwidth is optimized locally. Through simple implementation examples, we elucidate several appealing aspects of the proposed approach, including its computational efficiency and the possibility to account for typical boundary conditions, which would otherwise be cumbersome in conventional KDE.

International audience; Medium and flow heterogeneities lead to global transport patternsthat cannot be captured by large scale transport models that arebased on Fickian transport mechanisms. Early and late solute and particle arrivals, and non-linear evolution of dispersion arenon-Fickian manifestations of spatial heterogeneity on large scale transport. Such behaviors can be modeled by approaches thatare based on time and/or space non-local conservation equations. A key question is the identification and quantification of the heterogeneity controls on global non-Fickian transport patterns.Here, we focus on the roles of broad heterogeneity distribution, andheterogeneity and flow correlation on global non-Fickian transport.We identify the signatures of disorder and correlation dominatedlarge scale transport in the distributions of solute arrival times,solute dispersion and spatial solute distributions. Starting fromwell-defined small scale stochastic flow and transport descriptions,we quantify large scale transport by ensemble averaging over the Lagrangianparticle dynamics, which leads to a global transport modelthat is given by a coupled continuous time random walk (CTRW) characterizedby correlated particle velocities. We formalize this correlated CTRW in terms of an evolution equation for the global particle distribution, anddiscuss the impact of the correlation of subsequent particle velocitieson the transport behavior, as well as the limits of perfect and nocorrelation.

International audience; From numerical simulations of pore-scale flow in porous media, we demonstrate the existence of an intermittent-like behaviour of Lagrangian velocities similar to the one observed in turbulent flows. This phenomenon, characterized by non-Gaussian distributions of Lagrangian velocity increments and long-range correlation of Lagrangian accelerations, is at the origin at the breakdown of the classical upscaled models. For transport in porous media this is manifested by anomalous scaling of the temporal evolution of the characteristic dispersion length, called anomalous dispersion. Long range correlation is related to the existence of stagnation zones and localized high velocity channels. While for turbulence, intermittency of Lagrangian velocities can be represented by multifractal random walk, for porous media we show that the dynamical picture is different and that this process is well captured by a correlated continuous time random walk.

International audience; Contaminant transport in heterogeneous aquifers occurs mostly in the networks of intersecting channels. Here we investigate solute transport in channels with a periodically varying aperture, when the flow is still laminar but sufficiently fast for inertial effects to be non-negligible. The flow field is computed for a two-dimensional setup using a finite element analysis, while transport is modeled using a random walk particle tracking method. Recirculation zones are observed when the aspect ratio of the unit cell and the relative aperture fluctuations are sufficiently large; under non-Stokes flow conditions, the flow in non-reversible, which is clearly noticeable by the horizontal asymmetry in the recirculation zones. After characterizing the size and position of the recirculation zones as a function of the geometry and Reynolds number, we investigate the corresponding behavior of the longitudinal effective diffusion coefficient. We characterize its dependence on the molecular diffusion coefficient, the Péclet number, the Reynolds number, and the geometry. The proposed relation is a generalization of the well-known Taylor-Aris relationship wich is only valid for a channel of constant aperture at sufficiently low Reynolds number. Inertial effects impact the exponent of the Péclet number in this relationship; the exponent is controlled by the relative amplitude of aperture fluctuations. For the range of parameters investigated, the measured dispersion coefficient always exceeds that corresponding to the parallel plate geometry under Stokes conditions; in other words, boundary fluctuations always result in increased dispersion. The transient approach to the asymptotic regime is also studied and characterized quantitatively. We show that the measured characteristic time to attain asymptotic conditions is controlled by two competing effects: (i) the trapping of particles in the near-immobile zone and, (ii) the enhanced mixing in the central zone where most of the flow takes place (mainstream), due to its thinning.

Póster presentado en la American Geophysical Union Fall Meeting, celebrada en San Francisco del 3 al 7 de diciembre de 2012., From numerical simulations of pore-scale flow in porous media, we demonstrate the existence of an intermittent-like behaviour of Lagrangian velocities similar to the one observed in turbulent flows. This phenomenon, characterized by non-Gaussian distributions of Lagrangian velocity increments and long-range correlation of Lagrangian accelerations, is at the origin at the breakdown of the classical upscaled models. For transport in porous media this is manifested by anomalous scaling of the temporal evolution of the characteristic dispersion length, called anomalous dispersion. Long range correlation is related to the existence of stagnation zones and localized high velocity channels. While for turbulence, intermittency of Lagrangian velocities can be represented by multifractal random walk, for porous media we show that the dynamical picture is different and that this process is well captured by a correlated continuous time random walk.

Ponencia presentada en la XIX International Conference on Computational Methods in Water Resources (CMWR 2012), celebrad del 17 al 21 de junio de 2012 en la Universidad de Illinois., Local full mixing is often a key assumption that is needed in dispersion and reactive transport models. Complete mixing implies a loss of memory of Lagrangian velocities alon g particle trajectories. Thus, reaching this state is the main condition for Fickian dispersion. The existence of a locally fully mixed scale, which we call here mixing scale, is also required for defining the concept of local concentration and thus for us ing the mass action law for chemical reactions at local scale. The existence of this scale of homogenization can be assumed by invoking diffusive mixing. In this case, the mixing scale is identified to the characteristic diffusion scale. Yet, in heterogene ous velocity fields, where shear may act at all scales, the definition of the scale where elements are fully mixed is not obvious. E xperimental observations of anomalous dispersion and chemical kinetics in heterogeneous porous media suggest that incomplet e mixing at different scales may be at the root of these non classical behaviors. In order to investigate this we define a l ocal mixing scale as the largest volume which can be considered as fully mixed and propose a method for estimating it from particle separation statistics (Le Borgne et al., Phys Rev. E, 2011). In homogeneous media, local mixing is ensured by diffusion and the mixing scale grows like the square root of time. In heterogeneous velocity fields, the growth of hom ogenized areas is limited by the shear action of the velocity field. Hence, for strong heterogeneities, we find that the mixing scale grows subdiffusively, implying the persistence of incomplete mixing over long time scales. As the latter implies a significant correlation of Lagran gian velocities, we show that this subdiffusive scaling of mixing is related to a superdiffusive scaling of plume dispersion. Using different examples, we discuss how the analysis of the dynamics of local mixing can be used for understanding global dispersion and chemical kinetics laws.

International audience; Contaminant transport in heterogeneous aquifers occurs mostly in the networks of intersecting channels. The sinusoidalchannel geometry is relevant to transport moderately-disordered porous media, as well as (to some extent)to flow and transport in fractures with varying apertures. Here we investigate the spreading of a finite amount ofsolute entering such a channel of periodically-varying aperture.In channels of uniform apertures (parallel plate), when solute buoyancy is negligible, the advection and diffusionprocesses eventually lead to the well-known asymptotic Taylor-Aris dispersion regime. After this asymptoticregime has been reached, the solute progresses along the fracture at the average fluid velocity, according to aone-dimensional longitudinal advection-diffusion process. The corresponding diffusive term features an apparentdispersion coefficient instead of the molecular diffusion coefficient. In many real applications the relevant channelsdo not have constant aperture. Prior works have shown that deviation from the parallel plate geometry can significantlyalter the behavior, leading to relative increases or even decreases in the apparent dispersion. These studiesassume small Reynolds number and thus that flow is governed by the Stokes equation. While this is very often areasonable assumption, the Reynolds number can sometimes be of order unity or larger such that inertial effectsare no longer negligible. Increased inertial effects lead to the presence of recirculation zones, which represent immobileregions that can have a significant impact on solute transport and in particular on the asymptotic dispersion.We address flow and geometry configurations for which inertial effects affect flow and transport. In these conditions,flow can no longer be predicted by analytical solutions. We compute the stationary velocity fields based onNavier-Stokes equations, using a numerical scheme based on a finite element analysis. The transport problem isthen solved numerically by Lagrangian particle random walk simulations based on the Langevin equation. Dependingon the geometry parameters (ie the aspect ratio of a cell and the relative amplitude of the aperture fluctuations)and on the Reynolds number, the size and the position of the recirculation zones vary. This phenomenon is not onlyresponsible for longer residence times of the solute in these zones but also for higher velocities in the middle ofthe channel. At short times, this leads to unusual spreading patterns and oscillations in the velocity and apparentdispersion evolutions over time. At longer time, solute trapping is clearly visible in the recirculation zones. Wecharacterize the asymptotic dispersion coefficient as a function of the geometry parameters, the Péclet numberand of the Reynolds number. In our parameters range, a higher Reynolds number value systematically leads toan increase in the asymptotic dispersion coefficient. In tracer tests, higher apparent dispersion coefficient are ofteninterpreted from the parallel plate model. This study shows that, at sufficiently high Reynolds number values,the channel uniform aperture inferred in that manner would be larger than the real mean channel aperture, thediscrepancy arising from the effect of aperture variations.

International audience; The effect of spatial concentration fluctuations on the reaction of two solutes, A þ B* C, is considered. In the absence of fluctuations, the concentration of solutes decays as Adet ¼ Bdet t 1. Contrary to this, experimental and numerical studies suggest that concentrations decay significantly slower. Existing theory suggests a t d/4 scaling in the asymptotic regime (d is the dimensionality of the problem). Here we study the effect of fluctuations using the classical diffusion-reaction equation with random initial conditions. Initial concentrations of the reactants are treated as correlated random fields.We use the method of moment equations to solve the resulting stochastic diffusion-reaction equation and obtain a solution for the average concentrations that deviates from t 1 to t d/4 behavior at characteristic transition time t . We also derive analytical expressions for t as a function of Damköhler number and the coefficient of variation of the initial concentration.

Ponencia presentada en la XIX International Conference on Computational Methods in Water Resources (CMWR 2012), celebrad del 17 al 21 de junio de 2012 en la Universidad de Illinois., Natural flow fields in porous media display a complex spatio - temporal organization due to heterogeneous geological struc tures at different scales. This multiscale disorder implies anomalous dispersion, mixing and reaction kinetics (Berkowitz et al. RG 2006, Tartakovsky PRE 2010). In this context, classical continuum models based on Fickian mixing may misrepresent reactive t ransport. Using two dimensional pore scale SPH numerical simulations of flow and transport, we demonstrate the non Gaussian nature and the long range temporal correlation of the Lagrangian velocity field. The main origin of these properties is the existenc e of very low velocity regions where solute particles can remain trapped for a long time. Another source of strong correlation is the channeling of flow in localized high velocity regions. Thus, this result questions the applicability of classical Langevin approaches f or modeling mixing and reaction kinetics. In order to define an effective upscaled model, we adopt a upscaled model that takes into account the statistical properties of the pore scale Lagrangian velocity field. Analyzing the pore scale statistical properties of the flow, we show the spatial Markovian, and temporal non Markovian, nature of the Lagrangian velocity field. Therefore, an upscaled model can be defined as a correlated Continuous Time Random Walk (Le Borgne et al. PRL 2008) in two dimension. This account for both non Gaussian velocity distribution and long range temporal correlation property. The key feature of this model is the definition of a transition probability density for Lagrangian velocities across a characteristic correlation distance. We quantify this transition probability density from pore scale simulations and use it in the effective random walk model. In this framework, we discuss the ability of this effective model to represent correctly dispersion, mixing and reaction kinetics.

EGU General Assembly 2011, 3-8 Abril 2011, Viena (Austria), Proper characterization of the mixing processes occurring in the subsurface is crucial for describing a large variety of geochemical problems, including cave formation, calcite precipitation and dolomization, chemical speciation and microbial reactions among others. Several measures of mixing can be found in the literature. When transport can be fully defined in terms of conservative components, i.e. when for example all reactions are locally instantaneous and the governing transport equation of all solute species is based on an advection-dispersion equation (ADE), De Simoni et al. [2005] showed that the vector of reaction rates per unit volume of fluid is proportional to a measure of mixing that involves the gradients of concentrations of conservative components and local dispersion. In real field settings, the gradients of concentrations involved in the definition of mixing are highly influenced by the natural variability of the aquifer properties. The erratic velocity fields typically obtained in natural aquifers produce highly distorted solute plumes that enhance mixing. The relationship between physical heterogeneity and mixing is still quite unknown. A promising method for understanding mixing is tracer visualization at high spatial and temporal resolutions. The methodology consists in visually inspect the behavior of an optical tracer injected into a porous medium. Full picture of the solute distribution is obtained by analyzing the relationship between pixel intensity and tracer concentration in time and space. The problem of these techniques is that the contour and geometry of the porous media, light fluctuation, and brightness variations due to non uniform light among others produce noisy estimates of concentrations that cannot be used to directly estimate mixing. We present a methodology to visualize local values of mixing from noisy images of optical tracers based on an iterative nonparametric local regression technique. The methodology is used to provide a full depiction of the local mixing processes occurring in a heterogeneously packed sand box aquifer, consisting in sharp interfaces of two sand materials with contrasting hydraulic properties.

17 páginas, 11 figuras. Número de artículo W02508., We study the influence of buoyancy and spatial heterogeneity on the spreading of the saturation front of a displacing fluid during injection into a porous medium saturated with another, immiscible fluid. To do so we use a stochastic modeling framework. We derive an effective large-scale flow equation for the saturation of the displacing fluid that is characterized by six nonlocal flux terms, four that resemble dispersive type terms and two that have the appearance of advection terms. From the effective large-scale flow equation we derive measures for the spreading of the saturation front. A series of full two-phase numerical solutions are conducted to complement the analytical developments. We find that the interplay between density and heterogeneity leads to an enhancement of the front spreading on one hand and to a renormalization of the evolution of the mean front position compared with an equivalent homogeneous medium. The quantification of these phenomena plays an important role in several applications, including, for example, carbon sequestration and enhanced oil recovery.

International audience; [1] In this technical note we study mixing in a two‐dimensional linear shear flow. We derive analytical expressions for the concentration field for an arbitrary initial condition in an unbounded two‐dimensional shear flow. We focus on the solution for a point initial condition and study the evolution of (1) the second centered moments as a measure for the plume dispersion, (2) the dilution index as a measure of the mixing state, and (3) the scalar dissipation rate as a measure for the rate of mixing. It has previously been shown that the solute spreading grows with the cube of time and thus is hyperdispersive. Herein we demonstrate that the dilution index increases quadratically with time in contrast to a homogeneous medium, for which it increases linearly. Similarly, the scalar dissipation rate decays as t−3, while for a homogeneous medium it decreases more slowly as t−2. Mixing is much stronger than in a homogeneous medium, and therefore we term the observed behavior hypermixing.

Multicomponent reactive transport in aquifers is a highly complex process, owing to a combination of variability in the processes involved and the inherent heterogeneity of nature. To date, the most common approach is to model reactive transport by incorporating reaction terms into advection-dispersion equations (ADEs). Over the last several years, a large body of literature has emerged criticizing the validity of the ADE for transport in real media, and alternative models have been presented. One such approach is that of multirate mass transfer (MRMT). In this work, we propose a model that introduces reactive terms into the MRMT governing equations for conservative species. This model conceptualizes the medium as a multiple continuum of one mobile region and multiple immobile regions, which are related by kinetic mass transfer processes. Reactants in both the mobile and immobile regions are assumed to always be in chemical equilibrium. However, the combination of local dispersion in the mobile region and the various mass transfer rates induce a global chemical nonequilibrium. Assuming this model properly accounts for transport of reactive species, we derive explicit expressions for the reaction rates in the mobile and immobile regions, and we study the impact of mass transfer on reactive transport. Within this framework, we observe that the resulting reaction rates can be very different from those that arise in a system governed by an ADE-type equation.

The difficulty in developing models for waves in turbulent flows is a key problem in the analysis of the complexity of turbulence. We present a method to find and filter perturbations that are generated by the flow of deterministic waves from the power spectrum in the atmospheric boundary layer (ABL). The perturbation model proposed assumes that the amplitude and frequency of such waves decay with time exponentially. For illustrative purposes, we apply the technique to three time series of wind velocities obtained with a sonic anemometer. This analytical procedure allows us to filter waves of the proposed structure with a 99% significance level in the power spectrum. We have applied the same method to 540 such wind series, all painting similar results. We then compare the fractal dimension of the original series to those from which the waves have been removed. We find that the fractal dimension of the filtered waves is slightly less than that of the original series. Finally, we consider the fractal dimension of the studied series as a function of the length-scales and dissipation rate of kinetic energy per unit mass. Our results suggest an increase of fractal dimension with both length-scale and dissipation rate of kinetic energy.

Many modern low-energy ventilation schemes, such as displacement or natural ventilation, require the use of temperature stratification in a space. The ability to extract air at elevated temperatures is necessary for energy-efficiency and free cooling. The adoption of these energy-efficient ventilation systems still requires that they provide an acceptable level of indoor air quality and comfort. It is widely believed, although not comprehensively studied and validated, that such low energy ventilation systems will be more effective at removing contaminants. In this work, the focus is on the transport of passive and particulate contaminants in a displacement-ventilated space. By representing heat sources as ideal sources of buoyancy,simple analytical models have been developed that allow us to compare the average efficiency of contaminant removal between traditional and modern low-energy systems. These models are then extended and solved numerically to study and compare the spatial distribution of contaminants within the space of interest. In order to validate the models, small and large scale laboratory experiments were conducted. The results of the models and the experiments are ultimately extended to understand the fate of particulate contaminants, which suffer the additional effect of gravitational settling.

Natural and forced displacement ventilation offer promising strategies for cooling buildings in an energy efficient manner. However, there is the issue of whether or not they can provide adequate indoor air quality. Here, we investigate this in two studies - one for passive and the other for particulate contaminants. The complex internal stratification of buoyancy, typical of displacement systems, affects the transport of contaminants in a non trivial manner. While on average traditional and modern low energy ventilation systems are comparable regarding efficiency of contaminant removal, we show that the detailed vertical distribution of contaminants tells quite a different story regarding personal exposure. Further, most previous studies on natural displacement ventilation focus only on steady states, while most true buildings have continuously changing heat loads. Here, two projects on transients in natural displacement ventilation are presented. The first studies the influence of sudden changes in heat loads, while the second focuses on heat sources that vary in a periodic fashion. Many benefits of natural ventilation are observed, including that it is self controlling in a very favorable manner and that a well designed space is capable of handling even very large changes in heat load. This thesis presents a separate investigation of contaminants in coastal aquifers. Due to buoyancy effects, saltwater from the ocean often intrudes into aquifers, which gives rise to a nontrivial flow pattern that is very different from that typical of inland aquifers. Using a simple mathematical model, we investigate the influence of these flow patterns on the transport of conservative contaminants in a coastal aquifer. We find that depending on the specific characteristics of the aquifer that the saltwater intrusion can play a significant role on transport. It forces contaminant towards the upper seaward boundary, thus causing elevated contaminant discharge into the surf zone at beach areas

The development of an underground station below the water table requires rigorous and careful planning. As such an in depth knowledge of the station construction designs and the related hydraulic behavior are required in order to design the most appropriated drainage system. Moreover, it is important to ensure the correct drainage of the excavation site to avoid liquefaction and to minimize water seepage. As an example of underground construction we consider the tunnel for the new line in the Barcelona metropolitan area (Línia 9) through the town of Prat de Llobregat in the Llobregat River’s delta, which is currently under development. Most of the tunnel sections are constructed with Tunnel Boring Machines (TBM) and the stations and the ventilation shafts are constructed using “Cut & Cover” technology between sheet pile walls. A typical station on the L9 line lies partially between the shallow and the main aquifer. Specifically, it is located in the middle layer which in made up of silts and clays. In order to design the most appropriated drainage system we present a methodology whereby we generalize the hydraulic problem. This method enables rapid and efficient estimation of fluxes as well as a general understanding of the problem in a simplified manner.

Field data for surface storage zone in Lluta River

Introduction to special section on Modeling highly heterogeneous aquifers : Lessons learned in the last 30 years from the MADE experiments and others INTRODUCTION

This thesis deals with the development of tools and analysis to characterize and predict artificial recharge and radial convergent solute transport processes in heterogeneous media. The goal is to provide new insights to understand how heterogeneity, which is the main natural source of uncertainty in decision-making processes related with groundwater applications, can be controlled and its effects predicted for practical purposes in these topics. For hydrogeological applications, accurate modeling of phenomena is needed, but it is uncertain. Uncertainty is derived from the spatio-temporal random distribution of hydrodynamic (physical, chemical and biological) variables affecting groundwater processes, which is translated into random distribution of modeling parameters and equations. Such randomness is of two types: epistemic, when it can be reduced increasing the sample frequency of an experiment; aleatory, when it cannot be reduced when more information is analyzed. Sometimes hydrodynamic processes occur at so small scales that they become impossible to characterize with traditional methods, and from a practical perspective, this is analogous to deal with aleatoric model parameters. However, if some constitutive relationship (either empirically, theoretically or physically based) can be built between processes across different scales, then small-scale processes can be reproduced by equivalent large-scale model parameters. Uncertainty becomes amenable to be treated as epistemic randomness, and large-scale characterization techniques can be used to improve the description, interpretation or prediction of these processes. This thesis deals with these topics. The manuscript is composed by two main parts (the first on artificial recharge and the second on solute transport), each of them divided into three chapters. In chapter 1 of each part, a tool is developed to obtain quantitative information to model a selected variable at coarse grid resolutions. In the case of artificial recharge, satellite images are used to model the spatial variability of the infiltration capacity on top soils with a metric-scale detail. In the case of solute transport, a new method to estimate density from particle distribution is shown. In chapters 2, it is explored what processes occurring at the fine scales can affect the interpretation of artificial recharge and solute transport processes at larger scales. In the first part, a combined method that joins satellite images and field data along with a simple clogging model is used to display the equally-possible spatio-temporal mapping of the infiltration capacity of topsoil during artificial pond flooding activities. In the second part, numerical three-dimensional models are used to simulate transport in heterogeneous media under convergent radial flow to a well at fine scale. It is shown that an appropriate model framework can reproduce similar observations on contaminant temporal distribution at controlling section similar to those obtained in the field tracer tests. It is also provided a physical explanation to describe the so-called anomalous late-time behavior on breakthrough curves which is sometimes observed in the reality at larger scales. In the chapters 3, models are used to define the uncertainty around operating parameters in the optic of prediction and management on artificial recharge and solute transport. In the first case, a probability framework is built to define the engineering risk of management of artificial recharge ponds due to random variability of the initial distribution of infiltration, which controls several important clogging factors based on theoretical approaches. In the case of solute transport, it is discussed how equivalent parameters based on mass-transfer models can be related with the geometrical distribution of hydraulic parameters in anisotropic formation, when convergent flow tracer tests are used.