MODFLOW horizontal boring

Document Sample
MODFLOW horizontal boring Powered By Docstoc
					Soil Vapor Extraction
    Contamination in the Vadose Zone
•   The vadose (unsaturated) zone acts as a “buffer zone” for protecting the quality
    of the underlying ground water.
•   When contaminated, however, it acts as a source zone for ground water
    pollutants and gaseous emissions.


                                     Residual                               Vadose Zone

    Capillary Fringe                                 Vapors

                             Separate Fluid Phase
     Water Table
                                                Dissolved in Ground Water

       Ground Water Flow

                                                                            Saturated Zone
        Soil Vapor Extraction (SVE)
•   Targets the removals                                        Vapor/ Vacuum
    of VOCs from the                          Pressure    Flow Liquid Pump Treatment
    vadose zone by                            Gauge       Meter Separator    Unit
                            To GW Treatment
•   Shown to be effective
    at removing NAPL,                  Monit. GW         SVE
    aqueous, and sorbed                Well Well         Well
    phases                                                                  airflow
•   Encourages aerobic                                                      paths
•   Proven technology
    with some design
    guidance (rule-of-

              Case Study: JP-4 Spill
                at Hill AFB, Utah
•   A JP-4 fuel spill site at Hill, AFB , Utah was selected to be modeled using

•   This is probably the most comprehensive data set collected from a field
    application of SVE.

•   Several pilot studies, and a full scale soil vapor extraction (SVE) operation
    under a variety of flow conditions were conducted at the site.

•   Data for gas concentrations, contaminant concentrations, mass removals and
    vacuum pre- and during-operation were provided by the researchers.
                      VENT3D Description
•   VENT3D solves the 3-D vapor phase advection-dispersion equation for a mixture of
    compounds. It computes the 4-phase distribution for each compound between vapor
    moving periods, assuming equilibrium partitioning between phases.
•   The model domain is discretized into blocks.

    parameters common to domain
    contaminant conmposition, foc, x, y,  b.
                                                                           Qi,j,k
    parameters unique to cell volume
    permeability, contaminant concentration,
    injected air humidity, Qi,j,k.
    parameters unique to layer                                        x,i
    permeability anisotropy,z, porosity,
    moisture content.
    Boundary Conditions
    - ground surface open or closed to atmosphere.
    - lateral boundaries at amospheric pressure.      layers     1000
    - lateral boundaries: no flux or known flux
    -bottom surface represents water table:no flux.
              VENT3D Flow Algorithm
•    The model solves the 3-D steady-state gas flow equation using finite-

      P2    P 2    P 2 
       k         k         k          2 WRT
                                        
    x  x  y  y  z  z  dx dy  dz  MW
          x            y          z

                                                   k x, y, zPx, y, z
                                v x, y, z 
                                               x, y, z
                                               *  
                                          k  k  
                                                 n 
    where k= soil vapor permeability tensor (L2), P= soil-gas pressure (M/LT2),  = soil-gas viscosity (M/LT),W=
     vapor mass flux source/sink (M/T), R= universal gas constant (ML2/T2 mole oK), T= temperature (oK), MW=
     molecular weight of soil-gas (M/mole), k*= intrinsic permeability (L2),
    = air filled porosity (dimensionless), v= interblock gas flow (L/T).
    VENT3D Transport Algorithm
•   Knowing the 3-dimensional interblock flows, the 3-D advection-dispersion
    equation is solved by finite-difference for each chemical compound.
                         F    DnCn  qCn 
                    t     n

                                          qx qy
                       Dn  Dm I   x yz
                                                         3.3
                                 Dm  Do                     2
   where Mn = total molar concentration of compound n in mixture (mole/L3), Cn =molar concentration of compound
    n (mole/L3), q = vapor discharge vector (L/T), Fn=volumetric molar loss/addition rate of compound n (mole/L3T),
    Dn= the dispersion tensor (L2/T) , Dm= molecular diffusion coefficient (L2/T), Do= free air diffusion coefficient
    (L2/T), I = identity vector,  xyz= vapor dispersivity values (L), qx,qy= vapor flow in x,y,z, direction (L/T), q =
    magnitude of the discharge vector (L/T).
                            Phase Partitioning
•   Equilibrium partitioning is assumed between the different phases for each compound.
    Equilibrium is re-calculated at the end of each time step. The total molar concentration of
    each compound is expressed as a function of the vapor concentration and the sum of the
    molar concentrations in the 4 phases:

              M RT MH O RT Kdn RT H O 
     Mn  Cn   HCV    2
                                            2

                 Pn     n PnV   n PnV MWH2 O 
                                   #c ompounds
                                                   Cn RT
                                                   P  v
                                        n1           n

   where MHC=molar concentration of NAPL phase in soil (mole/L3), Pv= compound vapor pressure (atm), MH2O=
    molar concentration of dissolved phase (mole/L3),n=activity coefficient of compound in water (dimensionless), Kdn=
    distribution coefficient of compound (dimensionless), = soil density (M/L3), MWH2O= molecular weight of water (18
    gm/mole), H2O= soil moisture flag=1 if present, 0 if not present.
                           Site History
•   27,000 gallons (76,500 kg) of JP-4 jet fuel spilled in January 1985.
•   Contaminated area west of spill area = 160 ft * 120 ft *50 ft.
•   A field study of soil venting was performed at the site in 1988-1989.
           Pilot & Full-Scale Tests
               Date                                        Activity

                                      Pilot Tes ts

        January 18, 1988        Extraction from Vent 7 at 62 ft3 /min for 2 hours

         January 19,1988         Extraction from Vent7 at 127 ft3 /min for 2 hours

                                 Extraction from Vent7 at 202 ft3 /min for 2 hours

         January 20,1988         Extraction from Vent7 at 172 ft3 /min for 4 hours

                                 Extraction from Vent7 at 209 ft3 /min for 8 hours

                                 Full-s cale Operation

    December 18-March 11, 1988   Extraction from Vent 7 at 250 ft3 /min with
                                 occasional system shutdown
      March 11-April 2, 1989     Extraction from Vent 10 at 250 ft3 /min

       April 2-April 22, 1989    Extraction from Vent 9,10,11 at 350 ft3 /min.
                                 Performed several flow tests
       May 15-May 26, 1989       Extraction from Vent 5,6,7,8,9,10,11 at 500
                                 ft3 /min, followed by 14 days shutdown
      June 10-August 15, 1989    Extraction from Vent 5,6,7,8,9,10,11 at
                                 918 ft3 /min
     August 15-October 7, 1989   Extraction from Vent 9,11 at 650 ft3 /min.
                                 Heat injection from Vent 10 at 93 ft3 /min
            VENT3D Vertical Layers                  ground surface
                          0 ft
                                       3.75 ft             layer 11
                          3.75 ft
•   The contaminated                                       layer 10
    soil volume is        6.75 ft
                                                            layer 9
    divided into 11                       3ft
                          9.75 ft
    layers, each                          3ft               layer 8
    having different      12 .7 5 ft
    initial contaminant                                     layer 7
                          16 .7 5 ft
    concentrations in                     5ft               layer 6
    soil.                 21 .7 5 ft
                                          5ft               layer 5
                          26 .7 5 ft
                                          5ft               layer 4
                          31 .7 5 ft
                                          5ft              layer 3
                          36 .7 5 ft
                                          5ft              layer 2
                          41 .7 5 ft

                                       8.25 ft   soil      layer 1

                          50 ft
            VENT3D Horizontal Grid
•   Each layer is
    divided into a 16 *
    12 orthogonal
    grid, each grid cell
    is 10 ft wide *10
    ft. long. The grid
    shows the
    subareas to which
    the vertical vent
    area was divided.
    Each subarea is
    presented by a soil
    boring V1,
                             JP-4 Composition
        •   Reported and estimated standard weight fractions for JP-4 components.

We ight Fractions from ORNL GC Analysis
                                                   JP-4 Mass Fractions Estimated from GC
 # Carbon Atoms         We ight Fraction %
        C5-C6                     0                Analysis at Southern Petroleum Lab
        C6-C7                     0                  #Carbon Atoms          mass fraction %
        C7-C8                  0.166                         C5                  0.005
        C8-C9                  0.223                         C6                   0.05
       C9-C10                   0.16                         C7                   0.15
      C10-C11                  0.116                         C8                  0.225
      C11-C12                  0.131                         C9                  0.145
      C12-C13                  0.105                        C10                   0.11
      C13-C14                  0.053                        C11                   0.12
      C14-C15                  0.029                    C12+C13                   0.16
      C15-C16                  0.016                        C14                  0.025
                                                            C15                   0.01
      C16-C17                  0.001
          Determining Site Permeability

•   The horizontal and vertical permeabilities at the site were estimated using
    GASSOLVE, a computer program developed by Falta (1996).

•   GASSOLVE uses analytical solutions to the steady-state gas flow equation for
    different boundary and initial conditions.

•   The permeability is found by fitting the analytical solution to pressure data
    collected from air permeability tests.

•   GASSOLVE adjusts the permeability until it reaches a minimum residual sum
    of squares between the calculated and observed pressures.
                         GASSOLVE Results
      •   Pressure data from four pilot tests were used by GASSOLVE to determine the
          air permeability of the formation.

      •   Results indicated:
           –   horizontal permeability = 40 darcys.
           –   vertical permeability = 1 darcy.

                             Open ground Surface
                            te st1     te st2      te st3               te st4
hl permeability               35         40         42.7                 42.4
vl permeability                1        1.3         1.1                  1.02
Res. Sum Sq.              2.10E-05   4.70E-05    7.10E-05             1.00E-04
                                 Flow Calibration &Validation
•   Five pilot tests were simulated by VENT3D using GASSOLVE estimated
•   Three of the full-scale operation flow tests were simulated by VENT3D.

    Calculated Vacuum (in H2O)




                                                                pilot tests
                                                                flow tests
                                     0   5      10        15          20
                                             Measured Vacuum (in H2O)
                         Vacuum Adjustment

•    VENT3D calculated pressure was adjusted at extraction wells using the
     equation adapted from Anderson and Woessner, 1992.

                             Q g    r 
    P                      2  w
                        P j        ln e 
     w                     i,   k     rw 

 where Pw is the well pressure, P i,j is the calculated pressure, Qw is the pumping/injection rate, k is the
  permeability,  is the fluid viscosity,  is the fluid density, g is the gravitational acceleration, r e is the
  effective well block radius and rw is the well radius.
              Vacuum Adjustment Results
Q (m3/s ec)     Obs erved Vacuum VEN T3D Vacuum   Adjus ted Vacuum
                     (in H2O)       (in H2O)          (in H2O)

                           Pilot Tests
    0. 029            5.4                 4.0           4.1
    0. 059           10.9                 8.1           8.4
    0. 080           16.0                 11.2          11.6
    0. 094           20.0                 13.1          13.6
                            Flow Tests
    0. 029            8.3                 4.4           4.5
    0. 080           22.0                 14.7          15.1
    0. 080           22.5                 14.9          15.2
    0. 074           15.7                 13.3          13.6
    0. 0505          16.0                 11.6          11.8
    0. 048           16.5                 9.9           10.1
                Transport Calibration

•   The total initial contaminant mass was adjusted until the measured mass
    removals from the pilot tests and calculated mass removals from VENT3D
    matched within less than 20% error.

•   The initial total mass was estimated to be as high as 76,000 kg based on the
    spill volume.

•   Five pilot tests were simulated and mass removals were compared.
    Transport Calibration Using Pilot
   Data from pilot tests were not successful in the calibration step due to some
    discrepancy in reported data.
          This EPS image does not contain a screen preview.
          It will print correctly to a PostScript printer.
          File Name : calm   arem.eps
          Title : calmarem.eps
          Creator : MATLAB, The Mathworks, Inc.
          CreationDate : 07/03/98 13:09:51
          Pages : 1
     Transport Calibration Using Full
            Scale Operation
•   Mass removals recorded at the beginning of the full-scale operation were used to
    verify the calibration process.
           This EPS image does not contain a screen preview.
           It will print correctly to a PostScript printer.
           File Name : calibmass.eps
           Title : calibmass.eps
           Creator : MATLAB, The Mathworks, Inc.
           CreationDate : 07/03/98 13:12:30
           Pages : 1
                    Transport Validation
•   Mass removals from simulations representing the whole operation were
    compared with the measured values.
        This EPS image does not contain a screen preview.
        It will print correctly to a PostScript printer.
        File Name : rem7670.eps
        Title : rem7670.eps
        Creator : MATLAB, The Mathworks, Inc.
        CreationDate : 07/03/98 13:17:26
        Pages : 1
       Initial & Final Soil and Gas
    This EPS image does not contain a screen preview.
   It will print correctly to a PostScript printer.
    File Name : resgas.eps
    Title : resgas.eps
    Creator : MATLAB, The Mathworks, Inc.
    CreationDate : 07/03/98 15:35:43
    Pages : 1
                      Average Extraction Rates
    To evaluate whether the detail of the flow history was necessary,the
     25 flow tests conducted during the full-scale study were represented
     by one simulation.

        Q (ft3/s)


                                  different Qs for flow tests


                    Q average



              Average Extraction Rates
•   VENT3D estimated a 2% difference in removals.

      This EPS image does not contain a screen preview.
      It will print correctly to a PostScript printer.
      File Name : aveextr.eps
      Title : aveextr.eps
      Creator : MATLAB, The Mathworks, Inc.
      CreationDate : 03/18/98 16:30:09
      Pages : 1
         Predicting System Behavior
•   VENT3D was used to predict the extent of contamination if SVE was not
    carried out at the site.
         This EPS image does not contain a screen preview.
         It will print correctly to a PostScript printer.
         File Name : remdiff.eps
         Title : remdiff.eps
         Creator : MATLAB, The Mathworks, Inc.
         CreationDate : 06/25/98 14:04:46
         Pages : 1
        Predicting System Behavior
    This EPS image does not contain a screen preview.
   It will print correctly to a PostScript printer.
    File Name : resdiff.eps
    Title : resdiff.eps
    Creator : MATLAB, The Mathworks, Inc.
    CreationDate : 07/03/98 15:50:02
    Pages : 1
•   The design and performance of remediation systems can be greatly improved
    through the use of mathematical models.

•   VENT3D proved successful in representing site characteristics with respect to
    subsurface air flow and for simulating the performance of a vapor extraction
    study conducted at a JP-4 jet fuel spill site at Hill AFB.

•   Air permeability was appropriately estimated using GASSOLVE.

•   VENT3D helped in determining the initial total contaminant mass at the site.
                  Conclusions (cont.)
•   3-D modeling provided a closer match to field measurements than 2-D

•   The loss of accuracy in 2-D modeling was small from a design standpoint and
    was accompanied by considerable savings in computer time.

•   JP-4 could be represented with a mass-equivalent 10-compound mixture, and
    even a single component representing the mixture.

•   The gain in accuracy provided from modeling the multi-component mixture
    also came at the cost of extra computational effort.
                  Conclusions (cont.)
•   The slight improvement in accuracy by using 3-D modeling and detailed multi-
    component representation of the jet fuel does not justify the increased
    computational effort.

•   We therefore propose that for similar applications, one can represent the
    mixture by a smaller number of compounds and use a two-dimensional model
    without considerable loss of accuracy.

•   VENT3D was useful in demonstrating the changes in JP-4 composition during
    SVE .
                   Conclusions (cont.)
•   In this case, a surface seal was predicted to have minimal effect on cumulative
    mass removals.

•   If a model was to be used for design purposes, it would be more convenient to
    be able to use average (constant) flow conditions instead of going through a
    lengthy, complicated process of running a large number of simulations.

•   Data from long-term studies give a better description of site conditions and
    system behavior than data from pilot tests.
•   VENT3D calculated soil gas concentrations did not match well with observed
    concentrations due to the following:

     –   mass distribution from soil cores was not accurate.

     –   VENT3D does not allow for specifying different mass fractions at different locations.

     –   VENT3D does not account for mass transfer limitations.

     –   The site domain was modeled as a homogeneous formation.

Shared By: