Difference between revisions of "Model setup"

From Limestone
Jump to: navigation, search
m (Model setup for fracture flow and transport in limestone aquifers - steps)
m (Mesh generation)
 
(15 intermediate revisions by the same user not shown)
Line 10: Line 10:
 
{| class="body1"
 
{| class="body1"
 
|
 
|
== Model setup for fracture flow and transport in limestone aquifers - steps ==
+
== Steps to setup a model for flow and transport in fractured limestone aquifers ==
This chapter gives an overview of the recommended steps to setup of a model for the simulation of flow and transport in a fractured limestone aquifer.
+
This chapter gives an overview of the recommended steps to setup a model for the simulation of flow and transport in a fractured limestone aquifer.
 
The following list shows the typical steps to setup a model for contaminant transport in a fractured limestone aquifer.
 
The following list shows the typical steps to setup a model for contaminant transport in a fractured limestone aquifer.
 
Expand an item (button on the right) to get more information about it.
 
Expand an item (button on the right) to get more information about it.
Line 18: Line 18:
 
=== Formulation of the modeling objectives ===
 
=== Formulation of the modeling objectives ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Before setting up a model, it is important to define the modeling objectives.
+
Before setting up a model, it is important to define the specific modeling objective(s).
 
The choice of the model concept, model extent, modeling scale and included details should be closely linked to these objectives.
 
The choice of the model concept, model extent, modeling scale and included details should be closely linked to these objectives.
One should aim at including the most relevant features in the model, while keeping it as simple as possible.
+
One should aim at including the most relevant features in the model, while keeping it as simple as possible to address the modeling objective(s).
  
 
The following list gives some examples for modeling objectives:
 
The following list gives some examples for modeling objectives:
* Analyze the distribution and potential spreading of a contaminant in an aquifer
+
* Delineate the capture zone of a well
* Improve the understanding and predictability of contaminant transport
+
* Improve risk assessment, e.g. for a contaminant threatening a drinking water well, in order to assess future actions
 +
* Analyze the distribution and potential spreading of a contaminant plume in an aquifer
 +
* Test hypotheses related to the origin of a contaminant
 +
* Interpret observed / measured field data
 
* Analyze the influence of transient hydraulic conditions (annual variations, pumping in the area) on plume propagation
 
* Analyze the influence of transient hydraulic conditions (annual variations, pumping in the area) on plume propagation
* Develop and optimize a remediation strategy for the source zone
+
* Develop and optimize a remediation strategy for the source zone of a contaminated site
* Optimize a remediation strategy for the contaminant plume
+
* Optimize a remediation strategy for a contaminant plume
* Improve risk assessment, e.g. for a drinking water well, in order to assess future actions
+
</div>
* Delineate the capture zone of a well
+
</div>
 +
 
 +
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
=== Conceptualization and setup of a conceptual model including geology and hydrogeology ===
 +
<div class="mw-collapsible-content">
 +
[[File:SE-NO_profil.jpg |thumb|550px|Example geologic profile of a Geoscene3D model.]]
 +
==== Geologic modeling ====
 +
Borehole data, outcrops, geophysical measurements etc. can give valuable information about the [[ Geology and properties of limestone | geology ]] at a site.
 +
Bits of geologic knowledge can be connected to establish a geologic model, that shows different geologic layers and relevant geologic features (like inclusions or lenses).  
 +
Typically, these layers are characterized by different hydrogeologic properties.
 +
The geometry can be stored as a CAD interpolation of the surfaces, that delineate the geologic layers.
 +
Tools like GeoScene3D can be very useful to manage borehole data and to create interpolation surfaces.
 +
The surfaces can be imported into the numerical model later on.
 +
<br clear=all>
 
</div>
 
</div>
 
</div>
 
</div>
Line 53: Line 69:
 
Then, a simple model can be applied, which can be improved as soon as new data is measured.
 
Then, a simple model can be applied, which can be improved as soon as new data is measured.
  
Modeling was an integral part in the limestone project.  
+
Modeling was an integral part in the limestone project and already done at an early stage of the project.  
Initially, a rough model based on first measurements and available data was setup and used for the planning of further field work and measurements.
+
Initially, a rough model based on first measurements and available data was setup.
These measurements were later on used to improve the model.
+
It was used for the planning of further field work and measurements.
 +
It also helps to identify further data requirements and to plan measurements that support the modeling.
 +
These additional measurements and data can then be incorporated in the model.
 
</div>
 
</div>
 
</div>
 
</div>
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
=== Conceptualization and setup of a conceptual model including geology and hydrogeology ===
 
<div class="mw-collapsible-content">
 
[[File:SE-NO_profil.jpg |thumb|550px|Example geologic profile of a Geoscene3D model.]]
 
==== Geologic modeling ====
 
Borehole data, outcrops, geophysical measurements etc. can give valuable information about the [[ Geology and properties of limestone | geology ]] at a site.
 
Bits of geologic knowledge can be connected to establish a geologic model, that shows different geologic layers and relevant geologic features (like inclusions or lenses).
 
Typically, these layers are characterized by different hydrogeologic properties.
 
The geometry can be stored as a CAD interpolation of the surfaces, that delineate the geologic layers.
 
Tools like GeoScene3D can be very useful to manage borehole data and to create interpolation surfaces.
 
The surfaces can be imported into the numerical model later on.
 
<br clear=all>
 
</div>
 
</div>
 
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
 
 
=== Data acquisition - measurements to obtain relevant model parameters (see list of parameters for each model) ===
 
=== Data acquisition - measurements to obtain relevant model parameters (see list of parameters for each model) ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Line 85: Line 87:
 
=== Implementation of parameters for selected units in the model domain (homogeneous/heterogeneous) ===
 
=== Implementation of parameters for selected units in the model domain (homogeneous/heterogeneous) ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Based on available data, parameter distributions in the model can be defined.
+
Based on available data and the chosen model, parameter distributions have to be defined.
For a flow simulation, the hydraulic conductivity (or permeability) and the porosity have to be specified.
+
For a steady-state flow simulation, the hydraulic conductivity (or permeability) is required.
 +
For transport, parameters like porosity, (effective) diffusion coefficients, dispersivities and sorption parameters are required.
 
</div>
 
</div>
 
</div>
 
</div>
Line 93: Line 96:
 
=== Choice of boundary conditions and sources/sinks ===
 
=== Choice of boundary conditions and sources/sinks ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Boundary conditions have to be chosen according to known values, like a constant head or a known inflow.
+
Boundary conditions should be chosen according to known or delineated boundaries, such as geologic boundaries, known hydraulic head isolines, known flowlines (can be used as no-flow condition perpendicular to them, if the flow field is stable).
 
The most common boundary conditions are
 
The most common boundary conditions are
* Dirichlet conditions (or first-type boundary conditions), where the primary variable is fixed to a value (e.g. fixed head or fixed concentration)
+
* Dirichlet conditions (or first-type boundary conditions), where the primary variable is fixed to a value (e.g. fixed hydraulic head or fixed concentration).
* Neumann conditions (or second-type boundary conditions), where the gradient of the primary variable is specified (e.g. flux across the boundary, no-flow)
+
* Neumann conditions (or second-type boundary conditions), where the gradient of the primary variable is specified (e.g. flux across the boundary, often no-flow boundaries)
* Cauchy conditions (or third-type boundary conditions), which sets a condition to the primary variable and it's derivative (used f.e. for the flux through a river bed).
+
* Cauchy conditions (or third-type boundary conditions), which sets a condition to the primary variable and its derivative (used f.e. for the infiltration flux through a river bed).
The boundary conditions should be chosen according to known boundaries, such as geologic boundaries, known head isolines, known flowlines (these can be used as no-flow condition perpendicular to them).
+
The proper choice of boundary conditions is a very important step and determines the calculated results.
Sources and sinks are employed to include e.g. withdrawal/injection at wells or groundwater recharge due to precipitation.
+
 
 +
Sources and sinks can be added in the model domain to include e.g. withdrawal/injection at wells or groundwater recharge due to precipitation.
 
</div>
 
</div>
 
</div>
 
</div>
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
 
=== For transient models: definition of initial conditions ===
 
=== For transient models: definition of initial conditions ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
text
+
Steady-state models do not require the definition of initial conditions.
 +
Transient problems, however, require to specify the initial value of the primary variables (e.g. concentrations at each location of the model domain.
 
</div>
 
</div>
 
</div>
 
</div>
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
 
=== Mesh generation ===
 
=== Mesh generation ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
 
[[File:MeshExample.png |thumb|right|500px|Example a mesh used for discrete-fracture simulations of the Akacievej tracer tests. The mesh is refined at the horizontal fractures and at the wells.]]
 
[[File:MeshExample.png |thumb|right|500px|Example a mesh used for discrete-fracture simulations of the Akacievej tracer tests. The mesh is refined at the horizontal fractures and at the wells.]]
When working with complex models it is useful to start with a coarse mesh to test the model and the setup with limited time effort.
+
When working with complex models it is useful to start with a coarse mesh to test the model and the setup with limited computational efforts.
For the final simulations, a finer grid can be employed.
+
When the model is properly setup, a finer grid can be employed.
Modern grid generators allow a mesh refinement at specific parts of the mesh.
+
 
Especially at heterogeneities and fractures, at wells and at concentration fronts, the mesh should be sufficiently fine to resolve the local gradients (e.g. of concentration or heads) appropriately.
+
Modern grid generators allow a local mesh refinement at specific locations in the model domain.
This can be tested by a grid refinement study, where the mesh is refined and the results are compared.
+
Locations, where strong gradients of the primary variables (head, concentration) occur, should be resolved at a high resolution, while for the parts of the model domain with only small changes, a coarser mesh resolution may be sufficient.
When the solution does not change with a further grid refinement, the grid resolution is sufficient.
+
Especially at heterogeneities and fractures, at wells and at concentration fronts, the mesh should be sufficiently fine to resolve the local gradients (e.g. of concentration or head gradients) appropriately.
 +
 
 +
 
 +
If the applied mesh is sufficiently fine can be tested by a grid refinement study, where the mesh is gradually refined and the model results are compared.
 +
When the results do not significantly change with a further grid refinement, the grid resolution is sufficient.
 
<br clear=all>
 
<br clear=all>
 
</div>
 
</div>
Line 125: Line 136:
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
 
=== Simulation ===
 
=== Simulation ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Line 135: Line 147:
 
=== Critical evaluation of the modeling results ===
 
=== Critical evaluation of the modeling results ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
Modeling results should be always critically evaluated and tested.
+
Modeling results should always be critically evaluated and the model should be properly tested.
It is, for example, helpful, to have a check the mass balance of a model.
+
It is, for example, helpful, to check the mass balance of a model, i.e. to balance all inflows and outflows and the storage in the model domain.
It is also important to visually inspect the results, by . f.e. visualizing the hydraulic heads and concentrations in the domain and in special areas of interest.
+
It is also important to visually inspect the results, f.e. by visualizing the hydraulic heads and concentrations in the domain and in special areas of interest.
Then, it can be checked, if the results look as expected, if the boundary conditions are fulfilled and if there are any disturbances like oscillations in the model domain.
+
Then, it should be checked, if the results look reasonable and as expected, if the boundary conditions are fulfilled and if there are any disturbances like oscillations in the model domain.
 
Oscillations can be an indication for a too coarse mesh.
 
Oscillations can be an indication for a too coarse mesh.
 +
It is also a good idea to test for grid convergence, i.e. to test if the model results change, when the numerical grid is refined.
 
</div>
 
</div>
 
</div>
 
</div>
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
 
=== Model calibration and validation ===
 
=== Model calibration and validation ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
text
+
more information will be included soon
 
</div>
 
</div>
 
</div>
 
</div>
  
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 
<div class="collapsiblebar mw-collapsible mw-collapsed">
 +
 
=== Model reporting ===
 
=== Model reporting ===
 
<div class="mw-collapsible-content">
 
<div class="mw-collapsible-content">
text
+
more information will be included soon
 
</div>
 
</div>
 
</div>
 
</div>
  
== Example: Setup of models for a contaminated site with a fractured limestone aquifer (Akacievej, Hedehusene) ==
+
== Example: Setup of a COMSOL Multiphysics model for a contaminated site with a fractured limestone aquifer (Akacievej, Hedehusene) ==
The setup of a discrete-fracture model in 2D in COMSOL Multiphysics is described in the following document:
+
The setup of a discrete-fracture model in 2D in COMSOL Multiphysics using Coefficient Form PDEs is described in the following document:
 
* [[:File:WeakFormulation_toolbox.pdf |Instructions for setting up a discrete-fracture model in COMSOL Multiphysics (PDF) - DRAFT]]
 
* [[:File:WeakFormulation_toolbox.pdf |Instructions for setting up a discrete-fracture model in COMSOL Multiphysics (PDF) - DRAFT]]
 +
[https://www.comsol.com/ COMSOL] provides also the possibility to include fracture flow in the Darcy's Law interface, while fracture transport can be included in the Transport of Diluted Species in Porous Media interface.
 +
We refer to the product documentation for more details.
 +
[https://www.mikepoweredbydhi.com/products/feflow FEFlow] provides a tutorial, that describes the steps of setting up a flow and transport model and how to include discrete fractures.
  
The typical workflow for modeling a contaminated site will be demonstrated using an example field site close to Copenhagen.
+
DRAFT: The typical workflow for modeling a contaminated site will be demonstrated using an example field site close to Copenhagen.
 
 
 
[[ Example: Akacievej | Example: Setup of models for a field site (Akacievej, Hedehusene) ]]
 
[[ Example: Akacievej | Example: Setup of models for a field site (Akacievej, Hedehusene) ]]
  

Latest revision as of 12:07, 3 July 2019

Highlights
  • Setup of a model for a fractured limestone aquifer
  • Modeling objectives
  • Typical steps of setting up a model for limestone aquifers

Steps to setup a model for flow and transport in fractured limestone aquifers

This chapter gives an overview of the recommended steps to setup a model for the simulation of flow and transport in a fractured limestone aquifer. The following list shows the typical steps to setup a model for contaminant transport in a fractured limestone aquifer. Expand an item (button on the right) to get more information about it.

Formulation of the modeling objectives

Before setting up a model, it is important to define the specific modeling objective(s). The choice of the model concept, model extent, modeling scale and included details should be closely linked to these objectives. One should aim at including the most relevant features in the model, while keeping it as simple as possible to address the modeling objective(s).

The following list gives some examples for modeling objectives:

  • Delineate the capture zone of a well
  • Improve risk assessment, e.g. for a contaminant threatening a drinking water well, in order to assess future actions
  • Analyze the distribution and potential spreading of a contaminant plume in an aquifer
  • Test hypotheses related to the origin of a contaminant
  • Interpret observed / measured field data
  • Analyze the influence of transient hydraulic conditions (annual variations, pumping in the area) on plume propagation
  • Develop and optimize a remediation strategy for the source zone of a contaminated site
  • Optimize a remediation strategy for a contaminant plume

Conceptualization and setup of a conceptual model including geology and hydrogeology

Example geologic profile of a Geoscene3D model.

Geologic modeling

Borehole data, outcrops, geophysical measurements etc. can give valuable information about the geology at a site. Bits of geologic knowledge can be connected to establish a geologic model, that shows different geologic layers and relevant geologic features (like inclusions or lenses). Typically, these layers are characterized by different hydrogeologic properties. The geometry can be stored as a CAD interpolation of the surfaces, that delineate the geologic layers. Tools like GeoScene3D can be very useful to manage borehole data and to create interpolation surfaces. The surfaces can be imported into the numerical model later on.

Definition of the model scope

The scale also depends on the modeling objectives. The model scope will, for example, be different for the prediction of the spreading of an entire plume (large scale) and for the planning of source zone remedial actions (local scale). An important aspect when choosing the model extent is to have the boundaries sufficiently far away from the most influential features in the area, such as pumping wells. The model extent has to be big enough, so that the boundaries do not influence the results in the area of interest. It should be chosen based on physically meaningful boundaries (f.e. known head isolines or no-flow boundaries).

The modeling of source zone remediation requires a different scale than the risk assessment of a contaminant plume for water works. The following scales can be distinguished:

  • well or borehole scale
  • source zone scale
  • intermediate scale for e.g. pumping tests
  • plume scale

Based on available data and modeling objectives, the model complexity has to be chosen. A very complex and detailed model is not appropriate if only little field data is available. Then, a simple model can be applied, which can be improved as soon as new data is measured.

Modeling was an integral part in the limestone project and already done at an early stage of the project. Initially, a rough model based on first measurements and available data was setup. It was used for the planning of further field work and measurements. It also helps to identify further data requirements and to plan measurements that support the modeling. These additional measurements and data can then be incorporated in the model.

Data acquisition - measurements to obtain relevant model parameters (see list of parameters for each model)

Implementation of parameters for selected units in the model domain (homogeneous/heterogeneous)

Based on available data and the chosen model, parameter distributions have to be defined. For a steady-state flow simulation, the hydraulic conductivity (or permeability) is required. For transport, parameters like porosity, (effective) diffusion coefficients, dispersivities and sorption parameters are required.

Choice of boundary conditions and sources/sinks

Boundary conditions should be chosen according to known or delineated boundaries, such as geologic boundaries, known hydraulic head isolines, known flowlines (can be used as no-flow condition perpendicular to them, if the flow field is stable). The most common boundary conditions are

  • Dirichlet conditions (or first-type boundary conditions), where the primary variable is fixed to a value (e.g. fixed hydraulic head or fixed concentration).
  • Neumann conditions (or second-type boundary conditions), where the gradient of the primary variable is specified (e.g. flux across the boundary, often no-flow boundaries)
  • Cauchy conditions (or third-type boundary conditions), which sets a condition to the primary variable and its derivative (used f.e. for the infiltration flux through a river bed).

The proper choice of boundary conditions is a very important step and determines the calculated results.

Sources and sinks can be added in the model domain to include e.g. withdrawal/injection at wells or groundwater recharge due to precipitation.

For transient models: definition of initial conditions

Steady-state models do not require the definition of initial conditions. Transient problems, however, require to specify the initial value of the primary variables (e.g. concentrations at each location of the model domain.

Mesh generation

Example a mesh used for discrete-fracture simulations of the Akacievej tracer tests. The mesh is refined at the horizontal fractures and at the wells.

When working with complex models it is useful to start with a coarse mesh to test the model and the setup with limited computational efforts. When the model is properly setup, a finer grid can be employed.

Modern grid generators allow a local mesh refinement at specific locations in the model domain. Locations, where strong gradients of the primary variables (head, concentration) occur, should be resolved at a high resolution, while for the parts of the model domain with only small changes, a coarser mesh resolution may be sufficient. Especially at heterogeneities and fractures, at wells and at concentration fronts, the mesh should be sufficiently fine to resolve the local gradients (e.g. of concentration or head gradients) appropriately.


If the applied mesh is sufficiently fine can be tested by a grid refinement study, where the mesh is gradually refined and the model results are compared. When the results do not significantly change with a further grid refinement, the grid resolution is sufficient.

Simulation

After setting up the geometry, boundary conditions, initial conditions, material parameters and simulation parameters (simulation time, solver settings), the actual simulation can be run. It is always a good idea to start with a test run, f.e. with a coarse mesh and a short duration, to test if everything is set as desired.

Critical evaluation of the modeling results

Modeling results should always be critically evaluated and the model should be properly tested. It is, for example, helpful, to check the mass balance of a model, i.e. to balance all inflows and outflows and the storage in the model domain. It is also important to visually inspect the results, f.e. by visualizing the hydraulic heads and concentrations in the domain and in special areas of interest. Then, it should be checked, if the results look reasonable and as expected, if the boundary conditions are fulfilled and if there are any disturbances like oscillations in the model domain. Oscillations can be an indication for a too coarse mesh. It is also a good idea to test for grid convergence, i.e. to test if the model results change, when the numerical grid is refined.

Model calibration and validation

more information will be included soon

Model reporting

more information will be included soon

Example: Setup of a COMSOL Multiphysics model for a contaminated site with a fractured limestone aquifer (Akacievej, Hedehusene)

The setup of a discrete-fracture model in 2D in COMSOL Multiphysics using Coefficient Form PDEs is described in the following document:

COMSOL provides also the possibility to include fracture flow in the Darcy's Law interface, while fracture transport can be included in the Transport of Diluted Species in Porous Media interface. We refer to the product documentation for more details. FEFlow provides a tutorial, that describes the steps of setting up a flow and transport model and how to include discrete fractures.

DRAFT: The typical workflow for modeling a contaminated site will be demonstrated using an example field site close to Copenhagen. Example: Setup of models for a field site (Akacievej, Hedehusene)


Return to Content