BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:A Computational Model for Freezing and Thawing in Soil
DTSTART;VALUE=DATE-TIME:20180517T165900Z
DTEND;VALUE=DATE-TIME:20180517T170100Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-740@events.interpore.org
DESCRIPTION:Speakers: Rafid al Khoury ()\nA thermo-hydro-mechanical (THM)
finite element model is developed to simulate freezing and thawing in soil
. The governing equations are based on averaging theory and include conser
vation of mass\, momentum and energy. The constitutive models constitute t
he equation of state (EOS) for water\, Clausius-Clapeyron relationship for
cryogenic suction\, and empirical relationship for the melting point depr
ession and unfrozen water content. The model is capable of simulating all
important phenomena occurring during soil freezing\, including: freezing i
nduced heaving\, convective-conductive heat transfer\, water flow to the f
reezing region\, and porosity change due to cryogenic suction and solid de
formation.\n\nhttps://events.interpore.org/event/2/contributions/740/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/740/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Understanding Wicking in Textile by Multiscale Imaging and Modelin
g
DTSTART;VALUE=DATE-TIME:20180517T160500Z
DTEND;VALUE=DATE-TIME:20180517T162000Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-783@events.interpore.org
DESCRIPTION:Speakers: Robert Fischer (Empa Swiss Federal Laboratories for
Materials Science and Technology)\nTextiles are porous media found in a wi
de variety of configurations\, resulting from weaving\, knitting or croche
ting yarns into networks. Textiles consist of multiple scales of fiber\, y
arn\, fabric and multilayered systems\, each showing their own porosity sy
stem and complexity. Wicking\, or the spontaneous liquid capillary uptake
in the resulting multiscale pore system\, needs a multiscale approach. We
present a multiscale framework to predict the wicking behavior incorporati
ng pore network modeling and continuum transport modeling approaches\, us
ing high-resolution lab X-ray tomography for identifying the pore configur
ations. For validation\, we use synchrotron X-ray fast-tomography of water
uptake at yarn scale and neutron projection at textile scale.\nAt yarn sc
ale\, we first obtain the yarn configuration by highly resolved submicron
X-Ray computed tomography. Then\, we extract\, from this geometry\, the in
formation required to develop a pore network capturing all the appropriate
complexity\, namely long undulating pores\, with a loose system of throat
s and contacts. The configuration of yarns is particularly challenging whe
n analyzed towards building a pore network. Then we develop a pore network
model that allows simulating capillary uptake in the yarn system. Given t
he long aspect ratio of these pores\, we track the developing of liquid fi
lm building up along the yarn and filling up of the yarn pore space.\nWe t
hen isolate a representative element of the textile. Knowing the intra-yar
n structure and transport properties\, we examine the inter-yarn wicking p
roperties by imaging fabric structures at lower spatial resolution of seve
ral microns but larger field of view. A mesoscale pore network is construc
ted on top of the microscale yarn pore network. This modeling approach giv
es us the three-dimensional permeability tensor of a given fabric element\
, i.e. a certain knitting stitch or woven pattern.\nFabrics are repetitive
patterns of such structural elements. We use the derived mesoscale transp
ort properties to model the water transport on fabric scale. A Darcy’s t
ype continuum approach allows us to predict the wicking behavior of differ
ent fabric patterns\, also considering gravity. Previous neutron projectio
n experiments and finite-element modeling studies showed good agreement of
the multi-porous modelling with the observed wicking behavior (Parada et
al. 2017a\,b).\nUnderstanding the capillary uptake and redistribution of l
iquid water in textile not only can improve comfort in clothing but also p
rotect firefighters under extreme conditions or find application in medici
ne. \n\nReferences\n1. Parada M\, Vontobel P\, Rossi RM\, Derome D\, Carme
liet J. (2017a) Dynamic wicking process in textiles\, Transport in Porous
Media\, 119:611–632.\n2. Parada M\, Zhou X\, Derome D\, Rossi RM\, Carme
liet J. (2017b) Modelling wicking in textiles using dual porosity approac
h\, accepted for publication in Textile Research Journal.\n\nhttps://event
s.interpore.org/event/2/contributions/783/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/783/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Diffusion and dispersion with heterogeneous reaction in homogeneou
s porous media: The macroscale models revisited
DTSTART;VALUE=DATE-TIME:20180517T170800Z
DTEND;VALUE=DATE-TIME:20180517T171000Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-784@events.interpore.org
DESCRIPTION:Speakers: D. Lasseux (CNRS - I2M)\nMass transport combined wit
h heterogeneous reaction in homogeneous porous media is a common process e
ncountered in chemical engineering that is of major concern for many appli
cations ranging from packed bed reactors to porous electrodes. In these sy
stems\, reactants are transported by diffusion (and eventually by advectio
n) inside the pores where chemical reactions take place at the solid-fluid
interfaces. Modelling the macroscopic behavior of these mechanisms is of
prime importance and has been the subject of numerous studies [1\, 2\, 3\,
4]. However\, in almost all the analyses reported in the literature\, the
Kinetic number\, Ki\, referred to as the ratio between the characteristic
time associated to diffusion and the characteristic time associated to re
action at the pore-scale\, is considered to be exceedingly small compared
to unity. Many industrial processes are indeed operating in this range of
Ki\, but this constraint is however not always fulfilled. Under these circ
umstances\, the purpose of the present work is focused on the development
of macroscopic models in a range of Ki ≤1\, relaxing the above mentioned
restriction. \nThe study is focused on single-phase transport of a single
chemical species undergoing a first-order heterogeneous reaction in rigid
and homogeneous porous media. In addition\, the advection problem is assu
med to be decoupled from the transport/reaction mechanisms. Macroscopic mo
dels are derived\, with and without advection\, using the volume averaging
method and the associated closure problems are provided to compute the ef
fective diffusion (or dispersion) and reaction-rate coefficients. In order
to elucidate the impact of the Kinetic number on the coefficients involve
d in the upscaled equations\, a Maclaurin expansion in Ki is carried out\,
yielding models for which the corrections at the successive orders in Ki
and the necessary closure problem to compute them are clearly highlighted
and numerically solved in periodic unit cells. Validations of the macrosco
pic models are carried out from comparisons with direct numerical simulati
ons and a discussion is provided on the impact of the corrections. In part
icular\, it is shown that the impact of the Kinetic number is significant
on the effective reaction-rate coefficient as well as on the convective ma
croscopic term in the average transport equation when the Péclet number i
s non zero but that Ki has a completely negligible contribution to the eff
ective diffusion (or dispersion) tensor.\nKeywords: Diffusion\, Heterogene
ous reaction\, Upscaling\nReferences\n[1] Whitaker S.\, The method of volu
me averaging. Kluwer Academic Publishers\, Dordrecht\, the Netherlands (19
99).\n[2] Ryan\, D.\, Carbonell\, R.G.\, and Whitaker\, S. 1980. Effective
diffusivities for catalyst pellets under reactive conditions. Chemical En
gineering Science\, 35\, 10-16.\n[3] Ochoa-Tapia\, J.A.\, Stroeve\, P.\, a
nd Whitaker\, S. 1994. Diffusive transport in two-phase media: Spatially p
eriodic models and Maxwell's theory for isotropic and anisotropic systems.
Chemical Engineering Science\, 49\, 709-726.\n[4] Le\, T. D.\, Lasseux\,
D.\, Nguyen\, X. P.\, Vignoles\, G.\, Mano\, N.\, and Kuhn\, A. 2017. Mult
i-scale modeling of diffusion and electrochemical reactions in porous micr
o-electrodes. Chemical Engineering Science\, 173\, 153-167.\n\nhttps://eve
nts.interpore.org/event/2/contributions/784/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/784/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Examination of Capillary Pressure-Saturation-Interfacial Area Rela
tion under Dynamic Conditions using Volume-Of-Fluid (VOF) Method
DTSTART;VALUE=DATE-TIME:20180517T171100Z
DTEND;VALUE=DATE-TIME:20180517T171300Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-780@events.interpore.org
DESCRIPTION:Speakers: Santosh Konangi (University of Cincinnati)\nConventi
onal two-phase flow equations (Richards Equation) for porous media at the
macroscale require capillary pressure (Pc) and relative permeability (Kr)
measured as a function of saturation (S). However\, these equations lack
solid theoretical foundation\, and there is still a considerable gap betwe
en the theory and experiments. In typical experiments\, an “average”
macroscopic capillary pressure is measured as the difference between the p
ressures of the non-wetting-phase reservoir at the inlet and wetting-phase
reservoir at the outlet of a porous medium. Traditionally\, this macrosc
opic phase pressure difference is assumed to be equal to the pore-scale ca
pillary pressure arising due to the curvature of the menisci at the fluid-
fluid invasion front. Many theoretical and experimental studies have show
n that the macroscopic definition is valid only at equilibrium (i.e. stati
c conditions) and if the phases are connected (Ferrari et al.\, 2013). Un
der non-equilibrium (dynamic) conditions\, when the fluids are moving\, th
e dynamic capillary pressure measured in experiments is a combination of c
apillary pressure at the invasion front and the pressure head caused by vi
scous effects (Lovoll et al.\, 2011). \nThe Pc-S relationship is non-uniq
ue\, and is flow process dependent\; different Pc-S curves are defined for
drainage and imbibition experiments\, resulting in “hysteresis” (Joek
ar-Niasar and Hassanizadeh\, 2012). Gray and Hassanizadeh (1991\, 1993) d
eveloped a theoretical framework for unsaturated capillary flows which pro
posed that the inclusion of specific fluid-fluid interfacial area will exp
licitly define the state of the system\, resulting in a unique relation be
tween capillary pressure\, saturation and interfacial area (Pc–Sw–awn)
. In this study\, we investigate the capillary pressure–saturation rela
tion under equilibrium and non-equilibrium conditions using pore-scale dir
ect numerical simulations (DNS). Direct numerical simulations (DNS) allow
for high resolution description of the geometry and time evolution of int
erfaces\, thereby permitting us to investigate the uniqueness of transient
Pc–Sw–awn surfaces. \nThe porous medium is represented by a quasi-tw
o-dimensional flow network of cylindrical obstructions. Hex-dominant comp
utational grids are generated to accurately resolve the inter-cylinder por
e space. The Navier–Stokes (NS) equations are solved in the pore space
on an Eulerian mesh using the open-source finite-volume computational flui
d dynamics (CFD) code\, OpenFOAM. The Volume-of-Fluid (VOF) method is emp
loyed to track the evolution of the fluid–fluid interface\; a static con
tact angle is used to account for wall adhesion. Simulations of drainage
and imbibition are performed for different capillary numbers by controllin
g the flow rate of the non-wetting (polydimenthlysiloxane oil) and wetting
(water) fluids. From these micro-scale simulations\, the pore-scale capi
llary pressure is directly determined at the fluid-fluid invasion front\;
this capillary pressure depends on the pore morphology and interfacial ene
rgy at the fluid-fluid interface\, without accounting for the viscous diss
ipation which is dependent on system size and invasion speed. The pore-sc
ale capillary pressure is upscaled using the fluid-fluid interfacial area
to estimate the macroscopic equilibrium (quasi-static) and non-equilibrium
(dynamic) Pc-Sw curves\; the Pc–S–awn surface is constructed to deter
mine whether the data points from drainage and imbibition processes fall o
n a unique surface under transient conditions.\n\nhttps://events.interpore
.org/event/2/contributions/780/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/780/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Unsupervised Machine Learning Based on Tensor Factorization
DTSTART;VALUE=DATE-TIME:20180517T164100Z
DTEND;VALUE=DATE-TIME:20180517T165600Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-775@events.interpore.org
DESCRIPTION:Speakers: Velimir Vesselinov (Los Alamos National Laboratory)\
nIn general\, unsupervised machine learning (ML) methods are powerful tool
s for data analyses to extract essential features hidden in data. The inte
gration of large datasets\, powerful computational capabilities\, and affo
rdable data storage has resulted in the widespread use of ML in science\,
technology\, and industry. Here we present applications of ML to character
ize (1) reactive transport data observed at groundwater contamination site
s\, and (2) model simulations representing fast irreversible bimolecular r
eactions. Our ML method is based on Tensor Factorization techniques and is
applied to reveal the temporal and spatial features in the analyzed data\
n\nhttps://events.interpore.org/event/2/contributions/775/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/775/
END:VEVENT
BEGIN:VEVENT
SUMMARY:A posteriori error estimates\, stopping criteria\, and adaptivity
for a two phase flow with exchange between phases as a nonlinear complemen
tarity problem in porous media
DTSTART;VALUE=DATE-TIME:20180517T154700Z
DTEND;VALUE=DATE-TIME:20180517T160200Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-776@events.interpore.org
DESCRIPTION:Speakers: Jad Dabaghi (Inria Paris)\nIn this work we develop a
n a posteriori-steered algorithm for a two phase compositional flow with e
xchange of components between the phases in porous media. The discretizati
on of our model is based on a backward Euler scheme in time and a finite v
olume scheme in space. The phase transition is treated introducing a formu
lation based on Henry's law. The resulting nonlinear system is solved via
an inexact semi-smooth Newton method. The key ingredient for the a posteri
ori analysis are the discretization\, linearization\, and algebraic flux r
econstructions allowing to devise estimators for each error component. The
se enable to formulate criteria for stopping the iterative algebraic solve
r and the iterative linearization solver whenever the corresponding error
components do not affect significantly the overall error. Numerical exper
iments are performed using the semi-smooth Newton-min algorithm as well as
the Fischer--Burmeister algorithm and the GMRES iterative linear solver t
o show the efficiency of the method.\n\nhttps://events.interpore.org/event
/2/contributions/776/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/776/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Multiscale Computation of Pore-Scale Fluid Dynamics
DTSTART;VALUE=DATE-TIME:20180517T162300Z
DTEND;VALUE=DATE-TIME:20180517T163800Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-771@events.interpore.org
DESCRIPTION:Speakers: Yashar Mehmani (Stanford University)\nUnderstanding
the dynamics of fluid flow and transport in porous media is important in s
everal subsurface applications including geologic CO2 storage\, hydrocarbo
n recovery\, geothermal energy\, and groundwater hydrology. In order to co
ntrol and optimize said dynamics\, it is imperative that these processes b
e considered at the pore (or micro) scale. Pore-scale models provide a use
ful means of approaching such problems. However\, current direct numerical
simulation (DNS) methods can be prohibitively expensive\, even though the
y produce the highest fidelity predictions. On the other hand\, certain su
rrogate models (e.g.\, pore-network models) are considerably less expensiv
e and can be used to approximate the fluid flow physics in porous media. H
owever\, current surrogate models often lack the ability to control/shrink
their prediction errors. In other words\, the concept of “convergence t
o a solution” is absent. In this work\, we present a new computational f
ramework for simulating fluid flow dynamics at the pore scale. We demonstr
ate that through a combination of multiscale\, multiresolution\, and domai
n decomposition concepts the Navier-Stokes equations can be solved very ef
ficiently on porous materials with structures of arbitrary complexity. Mor
eover\, the framework provides the ability to converge\, in a step-by-step
fashion\, to the full DNS solution through successive iterations. This re
nders the approach flexible for a variety of applications in/outside geosc
iences\, which pose different tolerances for error.\n\nhttps://events.inte
rpore.org/event/2/contributions/771/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/771/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Risk Assessment of Carbon Sequestration into A Naturally Fractured
Aquifer at Kevin Dome\, Montana
DTSTART;VALUE=DATE-TIME:20180517T170500Z
DTEND;VALUE=DATE-TIME:20180517T170700Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-785@events.interpore.org
DESCRIPTION:Speakers: Bill Carey (Los Alamos National Laboratory)\nAs actu
al CO2 injection is unlikely to take place at Kevin Dome\, Montana\, the B
ig Sky Carbon Sequestration Partnership has turned to maximizing the value
of existing data acquired at the site. We present the risk assessment wor
k done using the National Risk Assessment Partnership (NRAP) to Kevin Dome
\, Montana. Geologic CO2 sequestration in saline aquifers poses certain ri
sks including CO2/brine leakage through wells or non-sealing faults into g
round water or land surface. These risks are difficult to quantify due to
data availability and uncertainty. One solution is running large numbers o
f numerical simulations on the primary CO2 injection reservoir\, shallow r
eservoirs/aquifers\, faults\, and wells to assess leakage risks and uncert
ainties. However\, a full-physics simulation is usually too computationall
y expensive. NRAP integrated assessment model (NRAP-IAM) uses reduced orde
r models (ROMs) developed from numerical reservoir simulations of a primar
y CO2 injection reservoir to address this issue. A powerful stochastic fra
mework allows NRAP-IAM to explore complex interactions among many uncertai
n variables and evaluate the likely performance of potential sequestration
sites. In this study\, we investigate the sensitivity of a variety of unc
ertain parameters to CO2/brine leakage through (1) legacy wellbore and (2)
fault pathways. We found major uncertain parameters to which the potentia
l CO2 leakage through legacy wellbore is sensitive including values of fra
cture permeability\, end-point CO2 relative permeability\, capillary press
ure\, and permeability of confining rocks. CO2 and brine leakage through f
ault pathways is sensitive to fracture permeability\, length of the faults
\, and fault displacement.\n\nhttps://events.interpore.org/event/2/contrib
utions/785/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/785/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Fast Forecast of Future Reservoir Performance Using Deep Learning
DTSTART;VALUE=DATE-TIME:20180517T170200Z
DTEND;VALUE=DATE-TIME:20180517T170400Z
DTSTAMP;VALUE=DATE-TIME:20210918T155658Z
UID:indico-contribution-181-782@events.interpore.org
DESCRIPTION:Speakers: Hoonyoung Jeong (University Of Texas At Austin)\nFut
ure reservoir performance under reservoir uncertainty has been estimated c
onventionally by posterior model simulations followed by history matching
of prior models to observed data. In the history matching step\, however\,
more than hundreds of simulation runs may be required to calibrate prior
model parameters such as facies\, permeability\, and porosity. To address
the computational challenge\, we propose a fast deep-learning-based approa
ch that avoids expensive multiphase flow simulations needed in conventiona
l history matching approaches for future reservoir performance forecast. O
ur deep-learning-based approach consists of four procedures: generation of
training data\, data space reparameterization\, training\, and forecastin
g. First\, prior models are simulated to generate training data set. The d
imension of the training data set is reduced using principal component ana
lysis (PCA) for computational efficiency. The low dimensional transformed
training data set is used to train a deep neural network (DNN) model. Once
observed data is available\, future reservoir performance is forecasted u
sing the trained DNN models. By doing so\, we can avoid the expensive his
tory matching step and forecast the future reservoir performance in observ
able data space instead of reservoir model parameter space. For a real-fie
ld-data-based reservoir model case\, the average relative error of our lea
rning-based approach in forecasting future reservoir performance is 4.14%\
, which indicates that our deep-learning-based approach can provide fast a
nd accurate prediction during the field operation.\n\nhttps://events.inter
pore.org/event/2/contributions/782/
LOCATION:New Orleans
URL:https://events.interpore.org/event/2/contributions/782/
END:VEVENT
END:VCALENDAR