case study: i t t d c t l d n i l m d li integrated conceptual and

Case Study:I t t d C t l d N i l M d liIntegrated Conceptual and Numerical Modeling

Chemwest Industrial Site

ObjectivesDesign a conceptual model, consisting of

Geologic StructureHydrogeologic PropertiesHydrogeologic PropertiesBoundary conditions and wells

Simulate Groundwater Flow with MODFLOW and FEFLOWCompare simulated results with field dataCompare simulated results with field dataDemonstrate the value in having a grid and simulator-independent conceptual model that is integrated with the numerical model

2

Background - SiteChemWest Laboratories Ltd. manufactured inorganic and organic industrial chemicalsBenzene was found in the groundwater 25 years agoBenzene was found in the groundwater 25 years agoOriginated from an underground storage tank (UST) several hundred feet up-gradient of the well.A pump and treat recovery program was initiated to hydraulically contain the

benzene plume and to ultimately capture it. After more than ten years of continuous groundwater pumping substantially, the extent of the benzene plume has diminished substantiallyR li l d h l f i i ll i f h iRecent sampling revealed that only four monitoring wells in one area of the site contained benzene concentrations above the prescribed groundwater criteria. ChemWest proposed to the U.S. EPA that the pump and treat operations be terminated and allow natural biodegradation to eliminate the remainingterminated and allow natural biodegradation to eliminate the remaining benzene. SWS was asked to develop a numerical groundwater model to assess the feasibility of this plan

3

feasibility of this plan.

Steps for Conceptual Modeling1. Collect and Visualize Data2. Design Geologic Structure

D fi Fl P ti3. Define Flow Properties4. Define Hydrologic Boundaries

4

1. Collect and Visualize DataGIS Data:

Locations of water bodies, roads, plant and property boundariesPumping and Observation wellsPumping and Observation wells Borehole log data

Site Plan: DXF fileWater table (initial heads)Water table (initial heads)Cross-section interpretations

Combine to define geologic surfaces

5

2D Views

Locations of:•Norman RiverNorman River•Deer Creek•Bass Lake•Salmon Pond

6

Cross sectionsAA’AA’

BB’

AA’, Vert Exag. 10

BB’ V t E 10

7

BB’, Vert Exag. 10

Fence Diagrams

8

Generate Surfaces from XS and Log Data

Top Sand

Top BedrockGround Surface

9

2. Design Geologic StructureProvide the model boundary (polygon)Provide Geologic Surfaces as InputD fi H i R lDefine Horizon RulesInterpret 3D Geologic bodies

10

Convert Surfaces to Horizons

11

Generate 3D Structural Model

12

3. Define Flow PropertiesSelect the geometry (polygons, structural zones)Define the method (constant, shp attribute, 2D or 3D Grid)P id th l / tt ib t (C d ti it St ti it I iti lProvide the values/attributes (Conductivity, Storativity, Initial Heads, etc..)

13

Property Zones

1

4

23

4

Upper Formation: Constant Kx = 0.0006 m/s Lower Formation:

Using Polygons:1. 0.0009 m/s2 0 00025 /

14

2. 0.00025 m/s3. 0.0003 m/s4. 0.0012 m/s

4. Define Hydrologic BoundariesSelect the geometryDefine the method (constant,

h tt ib t 2D 3D G id)shp attribute, 2D or 3D Grid)Provide the values/attributes

Boundary Types Used:Constant head (west edge)River2 Ponds (modeled as Lakes)Recharge

15

Constant HeadNorman River modeled as a constant head:

602 m; above the dam

587 m; below the dam587 m; below the dam

16

RiverStage:

598 m stage at the eastern edge587 m at the discharge point to the western river boundary587 m at the discharge point to the western river boundary

River bottom is 1.5 m below the surface of the river.Conductance estimated to 10 m2/day

598 m

587 m

17

PondsThe 2 ponds as Type 3 Lake boundariesStage: 599.0 mB tt 597 5Bottom: 597.5 mLeakance: 100 m2/day,

18

RechargeUniform across top of model domain254 mm/yr

19

Steps for Numerical Modeling - MODFLOW

1. Define Uniform Grid2. Refine Grid3. Populate Grid with Conceptual Model Data4. Review, make cell based edits5 Translate to MODFLOW Packages5. Translate to MODFLOW Packages6. Run Simulation7. Analyze Results

V lid t ith C t l M d l d A ti8. Validate with Conceptual Model and Assumptions

20

MODFLOW Step 1: Generate MODFLOW GridInitial uniform grid of 100 m x 100 m,

=>Results in 40 columns, 85 rowsUpper formation is refined to create 2 layersUpper formation is refined to create 2 layersDeformed grid layers follow conceptual model horizons

Plan view of the grid Cross-sectional view of grid

21

MODFLOW Step 1: Generate MODFLOW Grid

22

Step 2: Refine MODFLOW GridThe grid is refined by a factor or 2 around area of interestR lti id i 119 R * 56 C lResulting grid is 119 Rows * 56 Col(Total 6664 cells)

23

Step 3: Populate Grid with Conceptual Model DataDefine Inactive CellsPopulate Cells with Properties and Boundary Conditions

Cell representation of

24

Kx values in 2D, XS, and 3D View Boundary conditions

Step 4: Review and Edit Cell Attributes

25

Step 5: Translate to MODFLOW Packages

26

Step 6: Run SimulationMODFLOW-2005with PCG Solver

27

Step 7: Analyze Results

Heads ContoursXS View: Row 75

Head Contours, ,Plan View: Layer2

28

Step 7: Analyze Results

Reasonable Calibration to Heads

Mass Balance <0.1%

29

Step 8: Compare Results with Field Data

Overlay heads (contours, colormap), pathlines, compare to p , p

conceptual model

30

Steps for Numerical Modeling - FEFLOW

1. Define Superelement Mesh; Add-Ins (Points, lines, polygons)

2. Generate Mesh3. Populate Mesh with Conceptual Model Data4 Review make element-based edits4. Review, make element based edits5. Run Simulation6. Analyze Results

V lid t ith C t l M d l d A ti7. Validate with Conceptual Model and Assumptions

31

FEFLOW Step 1: Define Superelement MeshProvide Line and PointAdd-Ins:

Calibration PointsCalibration PointsConstant Head BoundaryRiver and Lakes

32

Step 2: Generate MeshUsing Triangle Mesh generatorEst. 6500 elementsR fi d liRefine around linesand edge (flow boundaries)

33

Step 2: Generate Mesh

34

Step 3: Populate Mesh with Conceptual Model DataProvide run settings (date, type)Select the Options for representing boundaries

35

Step 4: Review Cell Attributes, and Adjustments

36

Step 5&6: Run Simulation and Analyze Results

37

Step 6: Analyze Results

38

Step 7: Compare Results with Field Data

Overlay heads (contours, colormap), pathlines, compare to

t l d lconceptual model

39

SummaryThe conceptual model provides the ability to:

understand the site conditions before running a simulationchoose between MODFLOW or FEFLOWchoose between MODFLOW or FEFLOWProvides ability to select different grid types

Grid and mesh is easily generated and populated from conceptual objectsobjectsThe MODFLOW and FEFLOW Simulation show reasonable likenessIntegrated Conceptual and Numerical Modeling allows for:

Effi i d i f l i l d l d iEfficient design of multiple models to reduce uncertaintyManage multiple modeling scenariosComparing simulated results with observed field data to improve model credibilitycredibility

40