Density driven flow in porous media: How accurate are our models? Wolfgang Kinzelbach Institute for Hydromechanics and Water Resources Engineering Swiss.

Slides:



Advertisements
Similar presentations
ICHS 4, San Francisco, California, USA, September 2011 Experimental study of the effects of vent geometry on the dispersion of a buoyant gas in a small.
Advertisements

A mathematical model of steady-state cavitation in Diesel injectors S. Martynov, D. Mason, M. Heikal, S. Sazhin Internal Engine Combustion Group School.
1 A parallel software for a saltwater intrusion problem E. Canot IRISA/CNRS J. Erhel IRISA/INRIA Rennes C. de Dieuleveult IRISA/INRIA Rennes.
A parallel scientific software for heterogeneous hydrogeoloy
Flow in porous media: physical, mathematical and numerical aspects
Groundwater Hydraulics Daene C. McKinney
Principles of Groundwater Flow
Modeling of the Current Distribution in Aluminum Anodization Rohan Akolkar and Uziel Landau Department of Chemical Engineering, CWRU, Cleveland OH
Approximate Analytical/Numerical Solutions to the Groundwater Transport Problem CWR 6536 Stochastic Subsurface Hydrology.
Yhd Soil and Groundwater Hydrology
Introduction to Environmental Engineering Lecture 15 Water Supply and Groundwater.
Fluidyn -FLOWSOL March D numerical simulation of surface flows.
1 Modélisation et simulation appliquées au suivi de pollution des nappes phréatiques Jocelyne Erhel Équipe Sage, INRIA Rennes Mesures, Modélisation et.
Ali Zafarani Subsurface Processes Group University of California, Irvine.
Features of POLLUSOL Flow model Flow model Homogeneous, Isotropic, Heterogeneous and Anisotropic medium Homogeneous, Isotropic, Heterogeneous and Anisotropic.
Numerical Simulation of Dispersion of Density Dependent Transport in Heterogeneous Stochastic Media MSc.Nooshin Bahar Supervisor: Prof. Manfred Koch.
Midterm Review. Calculus Derivative relationships d(sin x)/dx = cos x d(cos x)/dx = -sin x.
Ground-Water Flow and Solute Transport for the PHAST Simulator Ken Kipp and David Parkhurst.
Aspects of Conditional Simulation and estimation of hydraulic conductivity in coastal aquifers" Luit Jan Slooten.
An efficient parallel particle tracker For advection-diffusion simulations In heterogeneous porous media Euro-Par 2007 IRISA - Rennes August 2007.
Free vs. Forced ConvectionFEFLOW ® Exercise Salt intrusion from top Vertical cross section.
REVIEW. What processes are represented in the governing equation that we use to represent solute transport through porous media? Advection, dispersion,
Why do we have storms in atmosphere?. Mid-atmosphere (500 hPa) DJF temperature map What are the features of the mean state on which storms grow?
Geothermal Application 1 Summer School Heat extraction from a sloped sandstone aquifer Vertical cross section of the model domain.
Upscaling, Homogenization and HMM
Upscaling of Foam Mobility Control to Three Dimensions Busheng Li George Hirasaki Clarence Miller Rice University, Houston, TX.
Numerical Methods for Partial Differential Equations CAAM 452 Spring 2005 Lecture 9 Instructor: Tim Warburton.
Direct and iterative sparse linear solvers applied to groundwater flow simulations Matrix Analysis and Applications October 2007.
Diffusion Mass Transfer
Density-Dependent Flows Primary source: User’s Guide to SEAWAT: A Computer Program for Simulation of Three-Dimensional Variable-Density Ground- Water Flow.
BIOPLUME II Introduction to Solution Methods and Model Mechanics.
8. Permeability (Das, chapter 7)
Hydraulic Routing in Rivers
LINTON UNIVERSITY COLLEGE SCHOOL OF CIVIL ENGINEERING
What process is simulated by these moving dots ? a)- Diffusion b)- Dispersion c)- Advection d)- Free convection e)- Something else f)- This is NO groundwater.
BsysE595 Lecture Basic modeling approaches for engineering systems – Summary and Review Shulin Chen January 10, 2013.
Hans Burchard Leibniz Institute for Baltic Sea Research Warnemünde How to make a three-dimensional numerical model that.
Formulation of the Problem of Upscaling of Solute Transport in Highly Heterogeneous Formations A. FIORI 1, I. JANKOVIC 2, G. DAGAN 3 1Dept. of Civil Engineering,
Darcy’s Law and Flow CIVE Darcy allows an estimate of: the velocity or flow rate moving within the aquifer the average time of travel from the head.
Scenarios 1.Tidal influence 2.Extreme storm surge (wave overtopping, max. limit 200 l/s/m, period 2 h) Outlook calibration and validation of 3D model transfer.
Laboratoire Environnement, Géomécanique & Ouvrages Comparison of Theory and Experiment for Solute Transport in Bimodal Heterogeneous Porous Medium Fabrice.
Multiphase Field-scale Modeling: Brine Transport Ann Cook Per Per Hatlevik Jonathan Levine Brice Loose Keegan Roberts Amber Sallerson Katy Schulte Martina.
Summerschool Modelling of Mass and Energy Transport.
Advection-Dispersion Equation (ADE)
Mathematical Background
Groundwater pumping to remediate groundwater pollution March 5, 2002.
Hydraulic Routing in Rivers Reference: HEC-RAS Hydraulic Reference Manual, Version 4.1, Chapters 1 and 2 Reading: HEC-RAS Manual pp. 2-1 to 2-12 Applied.
Tracers for Flow and Mass Transport
Saffman-Taylor Instability of Hele-Shaw Cell
FREE CONVECTION 7.1 Introduction Solar collectors Pipes Ducts Electronic packages Walls and windows 7.2 Features and Parameters of Free Convection (1)
FLOW THROUGH GRANULAR BEDS AND PACKED COLUMN
Darcy’s Law Philip B. Bedient Civil and Environmental Engineering Rice University.
CE 3354 Engineering Hydrology Lecture 21: Groundwater Hydrology Concepts – Part 1 1.
The Islamic University of Gaza Faculty of Engineering Civil Engineering Department EENV 5326 Groundwater Modeling.
(Z&B) Steps in Transport Modeling Calibration step (calibrate flow & transport model) Adjust parameter values Design conceptual model Assess uncertainty.
Distributed Resistances and Fan Models Chapter 4.
 p and  surfaces are parallel =>  =  (p) Given a barotropic and hydrostatic conditions, is geostrophic current. For a barotropic flow, we have and.
Conservation of Tracers (Salt, Temperature) Chapter 4 – Knauss Chapter 5 – Talley et al.
1 Simulation of the Couplex 1 test case and preliminary results of Couplex 2 H. HOTEIT 1,2, Ph. ACKERER 1, R. MOSE 1 1 IMFS STRASBOURG 2 IRISA RENNES 1.
Environmental Engineering Lecture Note Week 10 (Transport Processes) Joonhong Park Yonsei CEE Department CEE3330 Y2013 WEEK3.
A Brief Introduction to Groundwater Modeling
Example application – Finite Volume Discretization Numerical Methods for PDEs Spring 2007 Jim E. Jones.
For a barotropic flow, we have is geostrophic current.
Free vs. Forced Convection
Review 1.
Diffusion Mass Transfer
Philip B. Bedient Civil and Environmental Engineering Rice University
Transport Modeling in Groundwater
Transport Modeling in Groundwater
Philip B. Bedient Civil and Environmental Engineering Rice University
Presentation transcript:

Density driven flow in porous media: How accurate are our models? Wolfgang Kinzelbach Institute for Hydromechanics and Water Resources Engineering Swiss Federal Institute of Technology, Zurich, Switzerland

Contents Examples of density driven flow in aquifers Equations –Formulation –Special features of density driven flows Benchmarks –Analytical and exact solutions –Experimental benchmark: Grid convergence –Experimental benchmark: Fingering problem Upscaling issues Conclusions

Density driven flows in groundwater resources management Sea water intrusion Salt water upconing under freshwater lenses (both on islands and inland) Salt water fingering under playa lakes and saltpans Flow around salt domes (nuclear waste repositories) Brine injection Leachate from waste deposits Even the ordinary tracer experiment...

Saltwater Intrusion Salt water Fresh water

Formation of toe Fresh water Salt water

Saltwater Upconing Salt water Fresh water

Example: Salt Water Upconing on Wei Zhou Island Thesis Li Guomin

Upconing Freshwater Lens

Alternative Extraction Strategies

Salinity backflow from Chotts in Tunisia Oasis

200 km Salt fingers on islands in the Okavango Delta

Islands and salt crusts in Delta

Schematic cross section of an island Transpiration Evaporation Increasing salinity of GW Trona saltcrust Increasing salinity of GW gravity vs. upward flow

Instability on the Islands Critical Wooding Number: k f = m/s, u ET =10 -8 m/s instable stable

Simulation of fingering t=900 d c max =11 mg/lt=2900 d c max =30 mg/lt=6000 d c max =54 mg/lt=8500 d c max =75 mg/l t=12400 d c max =110 mg/l t=16800 d c max =235 mg/l t=25000 d c max =350 mg/l t=32500 d c max =350 mg/l t=46500 d c max =350 mg/l t=66000 d c max =350 mg/l

Flow in the vicinity of a salt dome Recharge Discharge Top of salt dome Salt water - fresh water interface No density difference With density difference

Mass balance total mass Mass balance salt Darcy law Dispersion tensor Constitutive relationships Boundary conditions (many combinations) Basic Equations expressed in mass fraction c and pressure p e.g. Possible simplification: Boussinesq approximation

Features of density driven flow -Non-linearity -Consistency problem of boundary conditions -Rotational flow with closed streamlines -Plus all difficulties known from advective- dispersive transport

Flow in porous media and rotation Darcy-flow in heterogeneous porous media is rotational Example: But we still have: In density flow, rotation is non-trivial: closed streamlines Rotational when not parallel to For constant k/ kfkf

Numerical solution and testing of codes Analytical solutions Exact solutions Inter-code comparison Experimental benchmarks Grid convergence All computations are made with d 3 f, a density flow model using unstructured grids, finite volume discretization, multigrid solver, error estimator, automatic local refinement/coarsening, parallel computing

Idea of exact solution (steady state) Pressure equation Salt mass fraction equation Assume any differentiable functions p(x,z), c(x,z) Assume any domain Assign function values as first kind boundary conditions on boundary of that domain

Plug functions into flow equations Pressure Salt mass fraction Right-hand sides are not zero: They are interpreted as source-sink terms So analytical expressions are exact solution for problem with - these source-sink terms and - first kind boundary conditions with given function values Only good if source-sink terms are small and do not dominate the problem

Analytical expressions for exact solution (steady state) Pressure Salt mass fraction Values in example tuned to make sources/sinks small: =20, =12, h=.14, =1, 0 =1, =1, x =0.1, z =0.02, x s =1, z s =-0.1 In PDE: n=1, g=1, k=10, =1, D m =1, / c=0.1, (c)= 0 + / c c=1+0.1c

Analytical Salt mass fraction Pressure Values between 0 and 1.13 p units Values between 0 and 1 c units Plugged into equations for c and p

Source-sink distributions Total mass Salt mass Red: max. input Turquoise: 0 input Red: max. input Light blue: 0 input Blue: output

Computed (with 4 grid levels) Salt mass fraction Pressure Values between 0 and 1.13 p units Values between 0 and 1 c units

Error Pressure Salt mass fraction Red : computed value too large by % Blue: computed value too small by % Red : computed value too large by % Blue: computed value too small by %

Experimental benchmark 3D transient experiment in box with simple boundary and initial conditions Measurement of concentration distribution in 3D with Nuclear Magnetic Resonance Imaging Measurement of breakthrough curves Drawback: Test of both model equations and mathematics Way out: Construction of a grid convergent solution inspired by the physical experiments

Experimental setup Cube filled with glass beads of diameter 0.7 mm Size of model 20*20*20 cm 3 Injection of dense fluid on bottom center hole Application of base flow via top corner holes In unstable case: Injection from below and rotation All parameters measured except transverse dispersivity, diffusion coefficient

Experimental setup continued 20 cm Salt water stable situation unstable situation

Stable situation: Low concentration contrast

NMR images of diagonal section: stable situation at low concentration contrast Injection End Flushing Equilibration

NMR images of diagonal section: stabel situation at high concentration contrast Injection Equilibration Entraining End

Two modes No density contrast Large density contrast

Experimental breakthrough curves Low contrast High contrast

Comparison of computed and measured breakthrough curves Low contrast High contrast Choice of parameters within intervals given through measurements of those

Comparison of concentrations along diagonal section Low contrast case, end of experiment

Comparison of concentrations along diagonal section High contrast case, end of experiment

Level # grid points , , , ,146,689 Grid convergence: Low contrast case unit x at level 7: 1.56 mm

Grid convergence: Low contrast case

Level# grid points , , , ,146, ,974,593 Grid convergence: High contrast case unit x at level 8: 0.78 mm

Grid convergence: High contrast case

Error of grid-convergent solutions Low contrast High contrast

Unstable case: NMRI data

NMRI of vertical and horizontal section: fingering experiment Vertical section Horizontal section

Fingering: Experiment vs. Computation

Finger velocity: comparison of experiment and computation

Causes for poor perfomance Numerical dispersion smoothes out fingers and eliminates driving force Initial perturbance not known well enough Start of fingers on microlevel, not represented by continuum equations

Influence of heterogeneity on density flow Homogeneous Henry problem Heterogeneous Henry problem

Definition of Henry problem: Homogeneous aquifer Hydraulic conductivity 1E-2 m/s, effective diffusion coefficient 1.886E-05 m 2 /s Boundary conditions: Left fresh water flux given at 6.6E-05 m/s Right hydrostatic salt water, salt mass fraction kg/kg

Solution of Henry problem: Homogeneous aquifer Relative concentration contours between 0 (left) and 1 (right) in steps of 0.1

Heterogeneous Henry problem: Permeability distribution Lognormal distribution, exponential autocorrelation Arithmetic mean 1.68E-9 m 2, Geometric mean 1.02E-9 m 2 Variance of log(k) = 1 Corr. lengths: horizontal 0.05 m, vertical 0.05 m Red: 3.5E-08 m 2, Blue: 2.6E-11 m 2

Heterogeneous Henry Problem: Concentration distribution Relative concentration contours between 0 (left) and 1 (right) in steps of 0.1 Eff. diffusion coefficient as in homogeneous case Question: Are there equivalent effective parameters to mimick main effect of heterogeneity in a homogeneous model?

Heterogeneous Henry Problem Comparison (zero local dispersion) Heterogeneous case Only diffusion Homogeneous case: Permeability equal arithmetic mean of heterogeneous case Only diffusion Homogeneous case: Permeability equal geometric mean of heterogeneous case Only diffusion

Heterogeneous Henry Problem Comparison (zero local dispersion) Heterogeneous case only diffusion Homogeneous case: Permeability equal geometric mean of heterogeneous case only diffusion, zero dispersion Homogeneous case: Permeability equal geometric mean of heterogeneous case, diffusion plus macrodispersion after Gelhar&Axness

Heterogeneous Henry Problem Comparison (with local dispersion) Heterogeneous case Diffusion + local dispersion Homogeneous case: Permeability geometric mean Diffusion + local dispersion Homogeneous case: Permeability geometric mean Diffusion + macrodispersion after Gelhar&Axness

Effective dispersion parameters - Stable situation with flow against direction of density gradient effective longitudinal dispersivity given by Gelhar & Welty, A 11 (with density gradient) < A 11 (without density gradient) - Unstable situation with flow in direction of density gradient at sufficiently large Wooding/Raleigh number, dispersivities grow to infinity due to fingers forming - Horizontal flow towards a fixed concentration leads to boundary layer. Dispersion at upper right boundary and at stagnation point is upstream diffusion: c/c 0 = 1 – exp (-x/ L ) Upstream diffusion

Conclusions Density flow of increasing importance in groundwater field Tests for reliability of codes are available Density flow especially with high contrast numerically demanding: Grid convergence may require millions of nodes Numerical simulation of fingering instabilities still inadequate Heterogeneities can be handled by effective media approach in situation without fingering New numerical methods are in the pipeline