iMOD User Manual version 4.4 (html)

11.10Tutorial 10: Multi-Node Well- and HFB Package

This tutorial gives an introduction to the Multi-Node Well Package (MNW, see Section 12.30) by using it in an unconfined quasi 3-D transient model. It also compares the MNW package with the conventional WEL package. We also add to this tutorial the HFB package, developed to block the horizontal flow from a particular direction.

Outline This is what you will do:

Required Data For this tutorial you need the following iMOD Data Files/folders:

Getting Started

Load the Modelling Project in 3-D

We will load the modelling project and generate a 3-D image of our model.

We want to display the well in 3-D together with the modellayers, so we need to configure the Z-settings of the well via IPF Configure, let’s do that.

We will now load the upper- and lower elevations per model layer, we use the Special Open option from the Project Manager. This option allows you to quickly read a selection of IDF files in a particular order from the current model project to the iMOD Manager. In that way, it is easy to port the files in the right order to the Profile Tool and/or 3-D Tool.


Figure 11.137: Example of the Special Open window.

You’ll notice that prior to the 3D tool the 3D IDF Settings dialog appears. In this dialog the appearance of the IDF-files can be configured. For example, an IDF can be represented by planes (quads between mids of gridcells giving a smooth surface) and/or cubes (representing the grid cells as flat surfaces, like Lego-blocks). To visualize aquitards as solids we will combine each bottom of an aquifer with the top of the aquifer lying underneath it.

To show the well we need to instruct iMOD to ignore the associated txt file temporarily and use the Z and Sec.Z-Crd as set previously. Therefore do the following:

Now we see our well.

The image might look like this:


Figure 11.138: 3-D image of our model.

The 3D-image allows you to observe that the well is penetrating all model layers; in fact the well screen is for a length of 1.0 meter in model layer 1, model layer 2 is fully penetrated and layer 3 contains 5 meters (half of the thickness of that aquifer) of the well screen. The well extracts from all three model layers, proportional to the respective length of the well screen in each layer; this will be computed by iMOD when the model definition is translated to the MF2005 WEL package. The total strength of the well is 10,000.0 m\(^3\)/d from December 1\(^{\rm st}\) 2016 up to December 1\(^{\rm st}\) 2040. Starting from December 1\(^{\rm st}\) 2040 the well is turned off (0.0 m\(^3\)/d). This is specified in the WEL.TXT file associated to the WEL.IPF. Below is the content of that file.

DATE , -9999.0
Q , -9999.0
20401201, 0.0

Run the Model

Because we want to be able to simulate layers falling dry we apply a model with unconfined model layers. In that way the areas that fall dry are no part of the simulation until these model cells are re-wetted again. When unconfined model layers are applied in iMOD, iMOD includes the wetting option of MF2005 automatically.


Figure 11.139: Example of the Layer Types window: assigning layer type ’Convertible (HNEW-BOT)’ to all layers.

It is important to know that MF2005 includes the option to simulate model cells becoming dry when the hydraulic head of that cell drops below the bottom of that model cell. To ensure that dry cells can become part of the simulation again, iMOD includes the wetdry-option in the LPF-package automatically: it is not needed to specify extra input for this option. iMOD defines the wetdry-option to all active model cells that are part of an unconfined model layer. Whenever the head underneath the dry cell (\(h_n\)) is higher than 0.1 meter above the bottom of the model layer (BOT), it becomes wet again. Using this option, is more stable than using all four adjacent model cells as well. In the iteration, the head at that cell is than initially set to by following equation:

\begin{equation} h = {\rm BOT} + {\rm WETFCT} \left ( h_n - {\rm BOT} \right ), \end{equation}

whereby WETFCT=0.1. These are programmed internally in iMOD as they give the most robust approach. However, whenever it is still needed to modify this, the (advanced) model user can modify the exported MF2005-files outside iMOD themselves.

From the PRJ-file iMOD has read the transient characteristics of your model; it starts at 1\(^{\rm st}\) of December 2017 00:00:00 and ends at the same date. More input was not yet given to the model, but we can extent the simulation period of the model by simply defining another end date, let’s do that.


Figure 11.140: Example of the iMOD Define Simulation Configuration window.

After step Item 39. the Define Simulation Configuration window should look like in Figure 11.140.

The model will generate results on a monthly time step which is based on the definition of the input data, we can inspect this.

iMOD will now first create the necessary MODFLOW2005 files; as the model is tiny this will be will finished quickly. Immediately thereafter the simulation starts. You’ll see that the model starts in a separate DOS-command window and it will echo the simulation progress. It is a transient simulation of 396 stress periods, it probably will take something like 20 seconds of runtime (e.g. on a computer with a 2.6 GHz processor).

Note: {installfolder} refers to the full path of the directory you installed iMOD in (e.g. D:\iMOD).

Inspect the result of simulation

Let’s inspect the hydraulic head of the first model layer and the well rates and generate time series.

iMOD will load all selected result files into the iMOD Manager and displays the result on the graphical canvas. Use your experience learned from the previous Tutorials to display the computed heads (HEAD) and the extraction rate (BDGWEL) as time series as shown in the following figure. Be aware that it can take a few seconds, as iMOD needs to open 2049 files.


Figure 11.141: Time Series of computed hydraulic heads and abstraction rates at the location of the well using the WEL package: heads in layer 1 (blue line), layer 2 (turquoise line) and layer 3 (cyan line), abstraction rates [m3/day] in layer 1 (red line), layer 2 (green line) and layer 3 (yellow line).

Note: The extraction in model layer 1 is inactive as soon as the layer becomes dry. Also observe that the layer is re-wetted and the extraction of layer 1 is re-activated again as a result of deactivation of the extractions in the other model layers.

The total extraction for the entire duration of the model is less than was assigned to the model. So, instead of taking out 87.8E\(^6\) m\(^3\)/d, the amount of extracted water was 86.8E\(^6\) m\(^3\)/d. One of the advantages of the MNW package is that the total extracted amount remains intact once a model layer falls dry. Another improvement is that the extraction rate declines gradually as a model layers tends to dry, instead of abrupt as with the WEL package. The other layers will get an increased extraction for those case. Let’s observe that in the coming part of this tutorial.

Creating the Multi-Node Well (MNW) input

This will clean the entire Project Manager first before loading in the selected PRJ file.

Now we have selected our MNW-well and the values for the different attributes are presented in the table. We can see that that screens of the well starts at 96.0 m+MSL and ends at 70.0 m+MSL. This is similar to our previous well modelled by the conventional WEL package. You can also see that the methodology of computing well loss is given by the keyword THIEM and the appropriate parameter relevant to that is the RADUIS (\(r_w=0.25\) m), see section Section 12.30 for more detailed information about the MNW package.


Figure 11.142: Attribute values for the MNW-well.

MNW computes a hydraulic head in the cell \(h_n\) such that it equals the computed hydraulic head at the well minus a head loss term (e.g. the Thiem equation, see (Konikow2009)). Here we neglect head loss due to skin and local turbulence effects for that particular cell, so:

\begin{equation} h_{\rm WELL}-h_n = \frac {Q_n}{2 \pi T}\rm {ln}\frac {r_0}{r_w}, \end{equation}

where \(Q_n\) is the well rate (m\(^3\)/d), \(T\) is transmissivity of the aquifer (m\(^2\)/d) at the well, \(r_0\) is the effective radius of a finite-difference cell (m), this is assumed for isotropic conditions as \(r_0=0.14\sqrt {\Delta x^2+\Delta y^2}\); and \(r_w\) is the actual radius of the well.

Note: Because \(r_0\) is typically much larger than \(r_w\), the head in a pumping well will be lower than the model-computed head. The head in the pumping well is not equal to the hydraulic head saved by the model.

Okay, let’s run the model with the MNW package.

Again, iMOD will first create the necessary MODFLOW2005 files and start the simulation immediately. Similar to the model using the WEL package, this model including the MNW package will also probably take no more than something like 20 seconds to run.

Compare the result of WEL and MNW simulation

Let’s inspect the hydraulic head of the first model layer and the computed distribution of extraction rates and generate time series.

iMOD will load all selected result files into the iMOD Manager and displays the result on the graphical canvas.

pictures/tutorial12/mnw2_qrates1.png pictures/tutorial12/mnw2_qrates2.png

Figure 11.143: Time Series of computed extraction rates using the MNW package in layer 1 (red), layer 2 (orange) and layer 3 (violet); total time series (above) and zoomed in from 2040 onwards (below).

As expected, you might observe that the total extraction rate varies and that the extraction rate for the first model layer slowly decrease to zero. At the same time the extraction of the deeper aquifers, increases to sum up to 10,000 m\(^3\)/d. If we look at the zoomed in image (bottom) for the period after 2040, where we turned off the well, we observe that the well rates vary although there is no external rate specified.

So, what is happening?

Well, one of the features of the MNW package is the capability of simulating intra borehole flow, actually water can move from one aquifer - through the borehole - to another aquifer. Due to the stopping of the pumping, the deeper aquifers recover quicker from the computed draw down than the unconfined aquifer, mainly due to the low storage coefficient. This causes an overpressure from the deep aquifers to the shallow one and generates a groundwater flow that migrates directly through the borehole into the first aquifer.


Figure 11.144: Time Series of computed hydraulic heads at the location of the abstraction well: in layer 1 using the WEL package (red line), and heads in layers 1 to 3 using the MNW package (blue, turquoise and cyan lines respectively).

The figure of timeseries of computer hydraulic heads at the location of the well clearly shows that there is an overpressure that causes this intra borehole flow. Moreover, when comparing the hydraulic heads in layer 1 cells at the well location, in the model with the MNW package the cell remains wet for a longer period of time compared to the model with the WEL package. This is caused by the MNW package decreasing the extraction amount gradually and therefore decreases the draw down rate of the ground water head. Due to the early mentioned intra borehole flow, the heads in the model with an MNW package recover more quickly than in the model with the WEL package.

So the MNW package can really add some extra features concerning the behaviour of a well in your model. Speaking of more detail, it seems that there is a horizontal barrier (sheet pile wall) blocking the flow to our well. Let’s see how to incorporate this with iMOD into our model.

Enhancing the model with a Horizontal Flow Barrier (HFB) input

Let’s create our sheet pile wall.

The following line could be on your screen.


Figure 11.145: Outline of our sheet pile.

Let’s save the sheet pile wall.

Now we have to add this sheet pile to our modelling project.


Figure 11.146: Example of the iMOD Project Manager window.

So, we’re ready to run this model.

Also here again iMOD will first create the necessary MODFLOW2005 files and starts the simulation. Similar to the previous 2 models, this model including the MNW + HFB package will also probably take no more than something like 20 seconds to run.

Compare the results of the MNW and HFB simulation

Let’s inspect the hydraulic head of the first model layer and generate time series.

iMOD will load all selected result files into the iMOD Manager and displays the result on the graphical canvas.
During the simulation iMOD translates the manually drawn sheet pile wall - which we saved earlier as SHEET_PILE.GEN - to a continuous (kinked) line coinciding exactly with the lateral cell faces it intersects; when utilizing the HFB package the specified resistance is assigned to these cell faces. It is always a good idea to examine the result of such a translation, e.g. to check whether the discretization has resulted in a sheet pile wall that is fully continuous and thus behaving like a true barrier. Let’s open that file.

You should see, more-or-less, the following image. In white is the actual position of the sheet pile in the model. Due to the chosen grid size, it is a little bit shifted and crenelated due to the rectangular simulation network.


Figure 11.147: Display of the possible outcome of our HFB model.

You could try to experiment with more complex shapes for the HFB and/or modify the resistance of the sheet pile.