Ground water and contaminant flow modelling in Bremen area

Table of Contents

3 Executive Summary

4 Project Description
4.a Introduction
4.b Site Description
4.b.i Hydrologic Setting
4.b.ii Site description – more data

5 Data Assimilation

6 Description of the groundwater model
6a Spatial discretization
6b Temporal discretization
6c Heterogeneity and layering
6d Solver
6e Boundary Conditions

7 Discussion of Flow Model Results

8 Description of scenarios
8.1 Scenarios without cleaning well
8.2 Scenarios with a cleaning well

9 Conclusion

10 Recommendation for further work
10.1 Faults
10.2 Contaminants
10.3 Seasonal data

11 References

12 Table Of Figures

Declaration of Authorship

3 Executive Summary

In the northern german lowlands north west of Bremen and south west of river Weser a factory wants to extract 70 cubic meters of groundwater per hour. They want to pump clean water despite of a landfill as contaminant source south of the well (=upstream of the ground water flow). The well is very close to the western one of two parallel faults within the middle layer, with conductivities between 1 and 1E-9 m/s and is strongly influenced by them. The uppermost layer (= Saale) has the highest conductivity and is therefore most important for the flow regime. A MODFLOW model with 3 layers, given elevations and heads in the river and in the rim of the area was run many times to check if and how the well can can deliver the 70 cubic meters per hour of clean water.

Lowering pumping rate, displacement of the well away from the fault, … is only sufficient for the case of low conductivity (closed) faults. For open (high conductivity) faults these measures are not sufficient and a cleaning well north of the landfill is necessary. By trying lower and lower pumping rates of the cleaning well the minimum extraction rate for the cleaning well was found.

4 Project Description

4.a Introduction

The purpose of the groundwater model for an area south west of river Weser in Bremen is to check if and how a well in the midst of the model area can pump 70 cubic meters per hour for a new factory without getting contaminated water from a landfill site at the southern (lower) border of the area. The well is located at the more western one of two parallel faults in the middle layer running south west to north east (perpendicular to river Weser). The conductivity of these faults can vary between two extremes: 1 m/s for the open water filled fault and 1E-9 m/s for the closed fault filled with very fine sediments. Well location, depth and pumping rate can be changed a little bit. If unavoidable also a cleaning well can be placed to catch conta­mi­na­ted water. With ModelMuse, MODFLOW and other USGS-tools the flow of contaminants must be checked and visualised and a combination of parameters found that enables the factory to get clean groundwater in a suitable way. The task is described in more detail in [4a] and [4b].

4.b Site Description

Bremen is the capital of a northern westerly german state of the same name. It lies on both sides of the river Weser in the northern german flatlands 60 kilometres upstream of its estuary on the North Sea. Bremen has an elevation of about 11 meter a.s.l. Its geographic coordinates are 53° 5' N, 8° 48' E or in UTM projected coordinates the position is 32U 487091 5880729. The UTM coordinates are similar to the Gauss-Krüger coordinates that are given for the origin (= upper left corner, column 0, row 0) of the model area: x=3 478196.26, y= 5887455.54. The model area is 6.2 km (west-east) * 4.4 km (south-north) = 27.28 square km large (62 columns * 44 rows with 100m*100m wide cells) and mainly south west of the river Weser located. The area is located within the following left map [1]. The right one [6] indicates elevations:

Abbildung in dieser Leseprobe nicht enthalten

4.b.i Hydrologic Setting

The ground of the northern german lowlands close to North Sea is structured into at least 3 layers. The 2 uppermost are quarternary. The uppermost is Saale called after a tributary of Elbe river and one of the last glaciations between 130000 and 300000 years before present that left sediments of high conductivity Kx = 1 E-5 m/s.

The middle layer is Elster called after river Elster and an older glaciation between 300000 and 400000 years before present that left sediments of low conductivity kx = 1 E-5 m/s.

The lowest layer we deal with is tertiary i.e. older than quarternary (more than 2.6 mega years before present) and has a medium conductivity of kx = 1 E-4 m/s.

River Weser is flowing to the north-west like the rivers Ochtum (south of Weser) and Lesum (north of Weser) that join the river Weser north west of the area shown in the map. The inner city of Bremen lies on dunes and is the highest point in the area (yellow greenish colors in Fig. 2).

4.b.ii Site description – more data

Bremen has a moderate oceanic climate of classification Cfb (Köppen-Geiger). Winds are mainly westerly from the Atlantic Ocean. Mean annual precipitation is 671 mm (nearly exclusive rain). The annual mean temperature is 9.2 °C. For more detailed meteorological and climatic data see the following table ([2], [3]):

Abbildung in dieser Leseprobe nicht enthalten

Fig. 3 Climate data for Bremen

River Weser has a mean discharge of 337 cubic meters per second.

5 Data Assimilation

Geological and hydrogeological data from around 70000 bore holes in Bremen and Bremer­haven is available from “Geologischer Dienst für Bremen – GDFB” (https://www.gdfb.de/).

For the MODFLOW-model a text file (3_Layers_DATA.txt) as digital elevation model – DEM with geometric z-coordinates in 62*44 cells in each of 3 layers as point data and a shapefile (GW_Boundary_GK3.shp) with hydraulic heads for all 210 cells at the model rim (2*44 + 2*62 -2*1 = 2* (44+62-1) = 2*105 = 210) were given instead of any recharge data or heads or groundwater levels in single observation wells. Analysis of bore hole data was not necessary, because geometry of the layers (see above) and conductivities (highest k=1E-3 in top layer = Saale, lowest K=1E-5 in middle layer = Elster, intermediary k=1E-4 in lowest layer = tertiary, special values of 1 and 1E-9 for fault zone where Elster layer is less than 4m thick) were explicitly given. The x- and y-coordinates of the river were given as shapefile “Weser_AOI_DHDN3_z3.shp”. The river bed conductance K=1E-5 m/s, thickness M=3 m, cell length L=100m and width W=100m was explicitly given, so that a constant C=k*L*W/M = 0.1m^2/3sec = 0.0333 m^2/sec as Direct-Value for the RIV-module was determined.

The point elevation file “3_Layers_DATA.txt” should have been prepared in a way, that no discrepancies like elevation (lowest layer) > elevation (middle layer) i.e. negative thickness of middle layer are found. But 3 points were overlooked and I adjusted them to elevation (lowest layer) = elevation (middle layer) – 1m before loading it in ModelMuse with File-Import-Points as a single object.

Abbildung in dieser Leseprobe nicht enthalten

Fig. 4 Three Elevations in 3_LAYER_DATA.txt that need correction

6 Description of the groundwater model

Groundwater modeling is described in [15] and in many publications of USGS. Some of them can be found in the References section of this paper.

6a Spatial discretization

In ModelMuse a model with 62 columns (column 1 = west border, column 62 = east border), 44 rows (row 1 = north border, row 44 = south border) both of 100m width, 3 layers and origin at x=3478196.26 and y=5887455.54 (easting and northing in meter in Gauss-Krüger zone 3 or DHDN3 around longitude of 3*3°=9°E) was created. This area represents 6.2 km * 4.4 km = 27.28 square km north west of the city of Bremen. The geometric z-values of Model_Top, Up­per_Aquifer_Bottom = Saale_Bottom, Middle_Aquifer_Bottom = Elster_Bot­tom and Lower_Aqui­fer_Bottom = Tertiary_Bottom were imported as point text file “3_Layers_DA­TA.txt”. The layers were renamed according to the description in chapter 4.b. The first layer (upper­most layer) was renamed to Saale. The second (middle) layer is Elster and the third (lowest) layer Tertiary. The horizontal spreading of river Weser was im­por­ted as shapefile “Weser_AOI_DHDN3_z3.shp”. The hydraulic / piezometric heads in all cells at the very rim of the model area were imported as (the z-values in) shapefile “GW_Boundary_GK3.shp” for the GHB-Module.

All cells are active. The layers have no further discretization (Vertical discretization = 1 in MODFLOW Layer Groups) and are of type Confined. The Confined-type was selected as described and justified in [4b]. The related storitivity S uses the default value of 1E-5.

6b Temporal discretization

Under Model-MODFLOW-Time “Steady State” was selected (1 stress period from -1 to 0), i.e. no time discretization.

6c Heterogeneity and layering

The layers were marked as rectangular objects each with 2 z-values equal to layer borders and got horizontal conductivities Kx of 1E-3 (Saale), 1E-5 (Elster if thicker than 4m or 1 or 1E-9 in other case which represents the fault), 1E-4 (Tertiary):

Abbildung in dieser Leseprobe nicht enthalten

Fig. 5 Kx-values of the 3 layers

For the conductivity Kx within the faults two global Variables HK_FaultHi=1 and HK_FaultLo = 1E-9 are used. The formula “If(...)” does not need the Abs()-Function for the difference of Layer-z-values, because for all points was checked that Tertiary has a lower elevation than Elster and discrepancies (only 3 points) were corrected before importing the text-file. Vertical conductivity is Ky=0.1 * Kx as default.

Visualizing Kx for the middle layer (= Elster) makes the 2 faults visible (2 parallel red lines from lower left to upper right). They are perpendicular to the river Weser that flows from lower right to upper left corner. A well at x=3480869 (east) and y=5885587 (meters north of equator) i.e. in the cell at column 27 and row 19 is found at the western rim of the western fault. A landfill site (= contamination source) is found at x=348437 and y=5883148 i.e. in the cell at column 23 and row 44 at the southern rim between the 2 faults:

Abbildung in dieser Leseprobe nicht enthalten

Fig. 6 Visualization of kx – middle layer showing fault

6d Solver

The default solver PCG (Preconditioned Conjugate Gradient package) with default para­me­ters was used.

[...]