Docstoc

Deterministic and stochastic modeling of groundwater flow and

Document Sample
Deterministic and stochastic modeling of groundwater flow and Powered By Docstoc
					Deterministic and stochastic modeling of groundwater
flow and solute transport in the heavily-stressed Bangkok
coastal aquifer, Thailand, and investigation of optimal
management strategies for possible aquifer restoration


        Manfred KOCH1 and Phatsaratsak ARLAI1, 2


1Department   of Geotechnology and Engineering Hydrology,
University of Kassel, Germany


2now at: Department of Water Resources, Nakhon Pathom Rajabhat
University, Nakhon Pathom, Thailand
Contents of Presentation

1.   Introduction
2.   Literature Review
3.   Hydrogeological Overview of Study Area
4.   Statistical and stochastic Approaches to assess reasonable calibrated
     Parameters in a complex Multi-Aquifer System
5.   Modeling Flow and Transport for sustainable Yield Estimation of
     Groundwater Resources in the Bangkok Aquifer System
6.   Numerical Modeling as a Tool to investigate the Feasibility of artificial
     Recharge to prevent possible Saltwater Intrusion into the Bangkok coastal
     Aquifer System
7.   Numerical Investigation of Density-driven Flow Effect in transient State to
     the Seawater and vertical Saltwater Intrusion of the Bangkok-Multilayered
     Aquifer System
8.   Integrating an Groundwater Management Optimization Module and a
     variable Density Flow and Transport Model to investigate sustainable
     Restoration Schemes for the Bangkok Aquifer
9.   Summary
1. Introduction
      problem: Increase of groundwater pumping in the last decades




    History of groundwater withdrawal            Hydraulic heads in 1999




       Land subsidence in 1999 (~1m)    Saline plume migration in 1999
1. Introduction
1.1 Objectives
a. Discriminate unequivocal sources of saltwater contamination in the Bangkok
   aquifer system
b. Future possible sustainable management strategies for the present and future
   safeguard of the groundwater resource will be simulated and then determined.

1.2 Scope of Study
a. Setting up 3D groundwater flow model for Bangkok aquifer using MODFLOW-96
   (Harbaugh and McDonald, 1996)
b. Examine the density effect in Bangkok aquifer system
    • Constant density: MODFLOW-96&MT3DMS embedded in PMWIN (Chiang
        and Kinzelbach, 2001;2005)
    • Variable density: SEAWAT-2000 (Langevin et. al., 2003)
c. Apply models to data in layer 3, 4 and 5 to simulate
    • Present day situation of seawater intrusion or/and saltwater from vertical
        leakage sources in these coastal aquifers and
    • Future saline plumes under several possible management strategies to
        alleviate a future deterioration of the groundwater resources
2. Literature Review
2.1 Saltwater intrusion investigations in the multiple-aquifers
system underneath Bangkok and adjacent provinces
a. AIT: the saltwater contamination in PD and NL originate from vertical leakage of
   adjoining aquifers, connate water and the seawater intrusion.
b. Kogkusai Kogyo (1995): the contaminations intrude from the Bangkok clay and the
   seawater intrusion.


2.2 Density effects on the solute transport
a. Hydrodynamic dispersion α 1/density contrast
b. Significant density-affected vertical plume movement can occur with relatively low
   concentrations
c. With density contrasts of ~ 0.3%, the variable density may have an effect on the
   plume migration
3. Hydrogeological Overview of Study Area




  3D – Geological Map                       Profile of Bangkok Aquifers System

         Recharge (Sanford and Buapeng, 1996):

         a. Insignificant amount of recharge occurs at the
            center of this plain

         b. Main recharge exists at the basin flank of aquifer
4. Statistical and stochastic Approaches to assess reasonable
calibrated Parameters in a complex Multi-Aquifer System
  4.1 Model Implementation


 Model domain and Finite Different grid of layer 5
 Model Area:        = 208*256 km2           = 53248 km2   Flow model
 Model Grid:        One layer =52*55        = 2860        •Dirichlet´s BC. at the basin flanks, 1st
                    9 layers = 25740                      layer, and gulf of Thailand




                                                                        3D of FD of BK aquifers
       Conceptual Model                    Modeled layer 5
4. Assess reasonable calibrated Parameters
4.2 Effective approaches to assess the reliable parameters
 4.2.1 Conventional (forward) trial&error approach:
 The head in 1999 is selected to be the target of the steady state calibration

a. Quantitative assessment




Kogkosai Kogyo, 1995                  AIT, 1997                    The current model
4. Assess reasonable calibrated Parameters
4.2 Effective approaches to assess the reliable parameters
 4.2.1 Conventional (forward) trial&error approach:
   b. Qualitative assessment




                                                                   RMS/head losses is 4%,
                                                                   3%, 4% for layer 3, 4
                                                                   and 5, respectively (5%,
                                                                   Anderson and
                                                                   Woessner, 1992)




   (a)                    (b)              (c)
Observed-(a solid line) and computed (dashed line) heads in 1999
  4. Assess reasonable calibrated Parameters

  4.2 Effective approaches to assess the reliable parameters
  4.2.2 Statistical regression (inverse) approach
                                                                  Yobbi (2000)
                                                                        0.02
MODFLOW&UCODE (Poeter and Hill,
1998) to compute rcs and cc, scrutinizes
uncertainties/ non-uniqueness problem in the
calibration.

                                                Relative scale sensitivity of T-16 subzones


                                                                        0.02




                                               Relative scale sensitivity of Vk-14 subzones
          Subzones of transmissivity
           in layer 3, Phra Padeang
4. Assess reasonable calibrated Parameters

4.2 Effective approaches to assess the reliable parameters
 4.2.2 Statistical regression (inverse) approach


The matrix of cc unveils the non-uniqueness problem. (cc < 0.95, Hill,1998)




  Correlation coefficient matrix of transmissivity (left panel) and vertical leakance
  (right panel) in layer 3, 4 and 5.
  4. Assess reasonable calibrated Parameters
  4.2 Effective approaches to assess the reliable parameters
   4.2.3 Stochastic modeling using MC-simulations and validation of stochastic theory:
    (a) validate the pure stochastic formula                       1*σY2 and 2*σY2




Simulated variograms of head in layer 3 (left panel), 4 (middle panel) and 5 (right panel) with
variances and correlation lengths as specified in above table.
                                 Correlation length at 95% fits
                                 better than at 66% of sill to
                                 Gelhar´s formular.
4. Assess reasonable calibrated Parameters
4.2 Effective approaches to assess the reliable parameters in the
Bangkok aquifers model
4.2.3 Stochastic modeling using MC-simulations and validation of stochastic theory
“                   ”:
(b) investigate the calibration parameters which are responsible for the systematic
misfit error
a. 180 realizations of lnT are generated and run for the Monte Carlo simulations.
b. 180 Monte Carlo simulations with randomly disturbed pumping rates of       varying
   magnitudes (30 - 80 % of the reference value) are performed.
c     of stochastic pumping <       of the stochastic transmissivity field.




              (a)                  (b)                   (c)
    A realization of MC- simulation of T in layer 3 (a), 4 (b) and 5 (c)
5. Modeling Flow and Transport for sustainable Yield Estimation
of Groundwater Resources in the Bangkok Aquifer System
 5.1 Transient calibration: Dirichlet solute BC. at Gulf of Thailand and the Bangkok clay




   Monthly observed (red) and computed (blue) heads from 2000 to 2002 at wells PD 93
   (left panel), NL2 (middle panel), NB61 (right panel) in layer 3, 4 and 5




       Observed (red) and computed (blue) saline concentration (mg/l) in 1995
       in layer 3 (left panel), 4 (middle panel) and 5 (right panel).
5. Flow and Transport modeling for Sustainable Yield Estimation
5.2 Sustainable yield estimation

Sustainable yield concept

                                                                                  1stscheme: Projected pump
                                                                                  for the next 30 yrs based on
                                      Sustainable yield                           the average pumping trend
                                                                                  from 1983 to 2002
                                2 pump schemes for layer 6 to 9
                                                                                  2ndscheme: Projected pump
                               Trial&error pumps in layer 3 to 5                  for the next 30 year relied
                                                                                  on the double average
                                                                                  pumping trend from 1983 to
                                      H3,4,5>0.75*HDec,2002                       2002
                 NO
                                             and/or
                                      Conc3,4,5 < 250 mg/l
                                         YES
     S.Y. for   1stscheme:   Decrease pump 1.2%, 1.2% and 1.9% /yr for layer 3, 4 and 5
5. Flow and Transport modeling for Sustainable Yield Estimation

  5.2 Sustainable yield estimation

 Unmet water demand (umd) is the residual water demand for which the sustainable yield
 cannot serve the expected future water demand and must be supplemented by surface water.




               (a)                           (b)                           (c)


     Unmet water demand (umd) in 2012 (a), 2022 (b) and 2032 (c) in various provinces
  5. Flow and Transport modeling for Sustainable Yield Estimation
  5.2 Sustainable yield estimation




Saline plume (mg/l) distribution in layer 3 (left), 4 (middle) and 5 (right panel) in 2032.
  5. Flow and Transport modeling for Sustainable Yield Estimation
  5.2 Sustainable yield estimation
   Reasons why the S.Y. can not recover the water quality:




The velocity vectors and H in layer 3 (a), 4(b) and 5 (c)




                                                            34% of inflow   Bangkok clay
                                                            6% of inflow    Gulf of Thailand

     Saltwater intrusion profile in 1990 and 1995
6. Investigate Feasibility Schemes to prevent Saltwater Intrusion
6.1 Investigation of the origins of the saline sources


                                                         Observed chloride (mg/l)
                                                         fingerprints for several
                                                         profiles in 1990




                                                               Blue: c> 4000 mg/l
                                                               Blue
                                                               Green: mixing seawater-
                                                               Green
                                                               and vertical saltwater
                                                               intrusion
                                                               Pale green: c>1000 mg/l,
                                                                    green
                                                               from upper layers to the
                                                               layer 3
                                                               Red: c>1000 mg/l, from
                                                               Red
                                                               upper layers to the layer 5

 Map of delineating four types of contamination zones inside
 Bangkok aquifers system
6. Investigate Feasibility Schemes to prevent Saltwater Intrusion
  6.2 Numerical study of the possibility of aquifer restoration

                                         The best non-constructive scheme
                                         1.   10th scheme: shut off pump in layer 5 to 9
                                              (2012 to 2032)
                                         2.   19th scheme: decrease pumping to 60% in
                                              shallow saltwater intrusion zone and shut
                                              off completely the pumps in deep
                                              saltwater intrusion zone (2012 to 2032)

                                         The best integration of non- and
                                            constructive scheme
                                         1.   31th scheme: combination of recharge-
                                              and clean-up wells (2012 to 2032)
                                         2.   32th scheme: combination of recharge and
                                              clean-up wells (2012 to 2032)
6. Investigate Feasibility Schemes to prevent Saltwater Intrusion
    6.2 Numerical study of the possibility of aquifer restoration
   Optimal design of possible aquifer restoration schemes




          WOS (a)                       19th scheme (b)                 31th scheme (c )
    Qualitative effects of the different schemes; WOS (a), the best „non-constructive“
    scheme (b) and best integration of „non- and constructive“ scheme (c)
7. Numerical Investigation of Density-driven Flow Effects in
transient State to the Seawater and vertical Saltwater
Intrusion of the Bangkok Multilayered Aquifers System
   7.1 Comparison of constant density-and variable density
   model of the calibrated parameters
 Constant density model-MODFLOW-96&MT3DMS and variable density model-
 SEAWAT-2000 are simulated for the long term (2003 to 2032) transient behavior of
 saline transport.



                                                            Massive pumping and high
                                                            transmissivities V is high:
                                                                 Dα V
                                                                 D α 1/Density effect
                                                                  (Koch and Zhang, 1992)

 Saline distribution of constant- and variable density in
  2032 at modeled column 21 or UTM-X = 662000 m.
7. Investigation of Density-driven Flow Effects in transient State
    7.2 Sensitivity analysis of hydrodynamic dispersion
      Sensitivity analysis of A (dispersivity)


                                                         Five A cases;
                                                         With pump
                                                         •   Outside major pumping
                                                             zone ---> Plumes of
                                                             variable density model sink
                                                             deeper than one of constant
                                                             density model
                                                         •   Greater A ---> Plumes sink
                                                             deeper in both constant-
                                                             and variable density model
                            Simulated saline conc. of    Without pump
                            constant- (solid line) and   •   Massive uplift >>> density
                            variable density (dashed         effect
                            line) in Dec, 2032) with
                            different       values  of
                            dispersivities.
7. Investigation of Density-driven Flow Effect in transient State
   7.3 Sensitivity analysis of hydraulic anisotropy

    Sensitivity analysis of hydraulic anisotropy Kx/Kz

                                                                       Four Kz cases;
                                                                       With pump
                                                                       •   Outside      major
                                                                           pumping zone --->
                                                                           Plumes of variable
                                                                           density model sink
                                                                           deeper than one of
                                                                           constant    density
                                                                           model
                                                                       Without pump
                                                                       •   Massive uplift >>>
                                                                           density effect


   The simulated saline conc. of constant- (solid line) and variable density (dashed line) in
   Dec, 2032) with different values of Kz.
7. Investigation of Density-driven Flow Effect in transient State
 7.4 Effect of density-dependent flow and transport on the
 trial&error aquifer restoration management schemes

 Effect of variable density flow and transport on the effectiveness of the
 trial&error aquifer restoration management schemes




                                           Snapshots of simulated saline plumes
                                           (kg/m3) for Dec, 2032, using the constant-
                                           (a, c, e) and variable density model (b, d,
                                           f): WOS (upper-left panel), 19th scheme
                                           (upper-right panel) and 31th scheme (lower-
                                           left panel)
8. Integrating an Groundwater Management Optimization Module
and a variable Density Flow and Transport Model to investigate
sustainable Restoration Schemes for the Bangkok Aquifers
  8.1 Optimal sustainable management


                                      Optimal sustainable management



  1st scheme: Sustainable yield           4th    scheme:      Optimize     the       5th,6th,7th scheme: Optimize water
                                          integration of non- and constructive       trade off concept from the scheme
  2nd scheme: Trial&error scheme 19
                                          scheme--obtained       from      the       4, 2, 3
  3rd scheme: Trial&error scheme 10       trial&error scheme 31


                                      Re-simulate 7 schemes with SEAWAT-2000

                                       Best:H-, Pollution recovery, invested costs

                                                      Best scheme
8. Groundwater Management Optimization
  8.2 Results
Results for scheme 4

Applying GWM module embedded within MODFLOW-2000 to optimize the trial&error
integrated non- and constructive scheme (31th scheme from chapter 6)
Objective function and constraints     Comparison of head recovery and monetary costs




                                                             51% savings




  Trial&error well scheme in layer 3      Optimal well scheme in layer 3 (a),
  (a), 4(b) and 5(c).                              4(b) and 5(c).
8. Groundwater Management Optimization
  8.2 Results
Results for scheme 5, 6 and 7

 Embedding GWM within MODFLOW-2000 optimizes the water trade off concept
 for three new schemes
Objective function and constraints of 5,6,7
                                                                    “Water Trade off Concept” = to
                                                                    examine the least-cost effective means
                                                                    through raise the water levels along
                                                                    the front of the seawater intrusion up
                                                                    to zero meter (MSL) either by shutting
                                                                    off the discharge wells or increasing
                                                                    freshwater injection close to the
                                                                    shoreline through in-lieu water
                                                                    supply, in order better restrain
                                                                    seawater intrusion or to reduce the
                                                                    polluted area of the “without scheme”.




  An example of in lieu delivered water supply cell- (white cell), recharge well(blue cell)
  candidates and withdrawal cells (orange cell) in layer 3 (a), 4(b) and 5 (c) of the scheme 5 to 7.
8. Groundwater Management Optimization
   8.2 Results
Quantitative analysis of variable-density effects of all schemes

                                                                   The 6rd scheme is
                                                                    the best scheme




       Summary of % averaged reduction of vertical saline plume pollution
       area, -seawater intrusion area and head recovery in layer 3 to 8
       relative to the wos scheme
8. Groundwater Management Optimization
8.3 Discussion
Choosing the best scheme

                                          The 3rd scheme is
                                           the best scheme




        Comparison of unmet water demand, averaged reduction of
        saline pollution area and total costs of each scheme
9. Summary

• Conventional-, new statistical- and stochastic method are used to evaluate the set of
   calibrated parameters and show that the set of calibrated parameters are realiable
• The sustainable yield concept can satisfy only the hydraulic constraints, but does not
   significantly improve the groundwater quality.
• The saline leakage from the Bangkok clay appears to be the major source of
   contamination, while the horizontal seawater intrusion is the minor pollution source.
• With the co-existing pollution sources, the groundwater management scheme needs
   complex „policy“- or „non-constructive“ measure as well as integrated „non-and
   constructive“ measures to remediate the saline pollution within the Bangkok aquifers
   system. Two best „non-constructive“, and integrations of „non- and constructive“
   measurement are found through trial&error and constant density modeling.
• Through the sensitivity analyses of the hydrodynamic dispersion and the aquifer
   anisotropy, it is found that the variable density of the contamination saline plumes
   does not appear to have a significant feedback effect on the hydraulic flow itself in
   the Bangkok multilayered aquifers.
• From the 7 schemes in Chapter 8 it is discovered that the scheme 3 is the best among
   these 7 schemes with respect to the groundwater hydraulic, quality, invested cost and
   impact to existing groundwater users.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:21
posted:10/12/2011
language:English
pages:32