©Copyright JASSS

  JASSS logo ----

Yvonne Haffner and Stefan Gramel (2001)

Modelling Strategies for Water Supply Companies to Deal with Nitrate Pollution

Journal of Artificial Societies and Social Simulation vol. 4, no. 3,

To cite articles published in the Journal of Artificial Societies and Social Simulation, please reference the above information and include paragraph numbers if necessary

Received: 29-Jun-01      Published: 301-Jun-01

* Abstract

The computer based model presented in this paper regards strategies for water supply companies to deal with nitrate pollution of groundwater aquifers. In Germany, as well as in many other European regions, nitrate pollution is one of the most important problems for water protection and water supply. The simulation of an existing water supply company shows a high level of conformance between simulation results and economic data of the company. The simulation of scenarios with high nitrate pollution shows important differences between the strategies of using deeper aquifers, of technical treatment of raw water, and of co-operation with the agriculture regarding costs and environmental sustainability. Also these results reflect fairly well the situation in Germany.

Keywords: Nitrate, water supply, object-oriented simulation, environmental sustainability

* Situation of German Water Supply


At present, water supply in Germany suffers mainly from quality problems but also in some regions from quantity problems. In several regions of the country, raw water quality is not satisfactory and is even deteriorating (SRU 1998), due to diffuse emissions mainly from agriculture (e.g. nitrate, pesticides). Elsewhere, there is regional or local water scarcity, as water resources are over-exploited. Regarding both problems, water resource management must be oriented towards greater environmental sustainability.

An important issue in this context is the restructuring of the German water supply system, towards a greater participation of the private sector. The push for privatizating water supply has many reasons: The public sector is seeking opportunities to sell off assets in order to increase its revenues, if only temporarily1 ; water supply is thought to become more efficient, as management expertise of private companies can be utilized;  cost reductions due scale economies in private businesses are anticipated, and, as a result, an enhanced competitiveness of the German water industry (e.g. BMWI 2001Spelthahn 1994). Expectations about the effects of privatization on environmental sustainability are less concrete and more ambiguous (e.g. UBA 2000; Gramel & Haffner 2001; Entelmann et. al. 2000).

If sustainability of water resource management is the objective, the question poses itself whether the water industry (and especially water supply companies) is capable of pursuing it, given its current organization, and whether the decisionmakers in this industry are faced with the appropriate incentives. Sustainable management (and especially investment decisions) requires accounting for environmental objectives while at the same time not neglecting socio-economic ones.

It is not easy to define generally applicable criteria for the environmental sustainability of water supply and resource management. Local and regional factors play a crucial role and determine whether, in a specific water supply situation, the constraints are more of a quantitative or of a qualitative nature.

The model we present in this article focuses on a typical water quality problem: elevated nitrate concentrations in groundwater caused by agricultural emissions. Nitrate in groundwater is one of the most pressing environmental problems in Europe (European Environmental Agency 2001), and in Germany as well.

Figure 1 shows the prevalence of this problem.  In Germany, in the year 1995  around 11% of groundwater samples had nitrate concentrations above the legal  limit valueof 50 mg/l 2 ; and about a quarter has considerably elevated values (>25 mg/l, LAWA 1995). This data is based on the results of the monitoring networks of the federal states.

Figure 1

Figure 1. Distribution of nitrate concentrations in groundwater samples (LAWA 1995)

The presented model considers three strategies to address the nitrate pollution of raw water:

The user has to chose one of the three strategies (deeper aquifers, nitrate treatment, co-operation with the agriculture) in attribute "strategy". By this a adequate object is defined in the reposistory. This object can be adapted through the user by the change of preselected values.

Additionally, the user has to define which nitrate concentration is the sign for the start of a measurement. If the strategy ‘co-operation with the agriculture’ is selected the time difference between the start of the co-operation and the effect on the groundwater6 has to be respected (see chapter Water resource). In this case the nitrate concentration of the groundwater has to be (without water treatment) below the limit value of 50 mg/l!

The model was implemented using the Versatile Simulation Environment for the Internet (VSEit), a software framework, described in another article in this issue of JASSS (Brassel 2001). If you like to see the simulation run in your browser, visit the model's home page at www.vseit.de/de/vseit/haffner/water1/index.htm and follow the instructions given there.

* Overview of the model

Figure 2 shows the classes of objects that are required to model the strategies described above. The classes and their relationships as an UML class diagram. A class is represented as a rectangle devided in three sections. The top section contains the class name, the middle section a list of attributes describing the state of an object, and a bottom section containing the methods an object may perform.

Figure 2

Figure 2. An overview of WATER 1

Two main aspects of the water supply company are modeled: First, the water withdrawal and water delivery to the housholds. Second, decisions about investment, reinvestment, and loan taking.
For water withdrawal there is the class water ressource with specific phyiscal attributes and methods. Different kinds of plants are needed for withdrawal, treatment, storage, and distribution. The class CustomerTypes fixes the water demand. To keep the plants working the company has to reinvest and there for the company has to take out loans. The bank grants loans with specific credit terms. The company has to do investments if a measure against the nitrate pollution has to carry out. Because the strategy 'Co-operative agreements with agriculture' is a specific measure cooperation is a single class.

* Description of single classes

Water Supply Company and Plants

In the following , we use italic font for the names of methods and single apostrophes for the names of objects as tey appear in the model.
The technical infrastructure of the water supply company consists of individual plants which are instances of class 'Plant'. The plants can be devoted to different activities: raw water extraction, conventional raw water treatment, storage and distribution (this is the classification typically used by engineers).  Accordingly, plants in the model are called 'well',' treatment', 'water storage system', 'distribution network' and 'other'. There are also designated deeper wells, devoted to withdrawal from deeper aquifers, and a nitrate treatment plant.  In the model, these are called 'deeper wells' and 'nitrate treatment'.
The model user can see this setup  in the repository, as shown in figure 3. The figure shows different plants, some connected to the water supply company by lines, some without connection lines. The former are plants which are owned by the water supply company;  the latter can be purchased during the simulation run, whereupon they will receive a line connecting them to the company.

Figure 3

Figure 3. Water supply company with plants

The different colour saturation shows the actual book value of the plants. Book values can be expressed as a percentage of purchase cost. If a plant is newly constructed, its book value is identical with its purchase costs. In this case, the proportional book value is 100% and therefore the colour of the plant is saturated; in case of a totally depreciated plant with proportional book value of 0%, the colour is transparent.

Figure 4 shows the attributes of the water supply company. There are three different types of attributes:

Attributes that can be edited by the user are underlay white, non editable are underlay grey.

Figure 4

Figure 4. Attributes of the water supply company

The water supply company has to fulfill different tasks, or assignments, during a simulation run:

These assignments are implemented in individual methods, roughly as follows:

  1. The company obtains information about the nitrate concentration of the groundwater in the method observeNitrate. First a method in the object 'water resource' will be called up which calculates the actual nitrate concentration and returns this value. When the nitrate limit reaches the critical value, the company begins to carry out the selected measure.

    If the strategy of withdrawing from deeper aquifers ('deeper aquifers') is chosen, the old wells will be replaced by deeper wells. The old wells will be deleted from the list of plants and the deeper wells will be added.

    DeeperWells dw = (DeeperWells)sim.repository.getClient("deeper wells");
    Well w = (Well)sim.repository.getClient("wells");

    If the strategy 'nitrate treatment' is chosen, a nitrate treatment plant will be added to the list of plants.

    NitrateTreatmentUnit nt = (NitrateTreatmentUnit)sim.repository.getClient("nitrate treatment");
    entity().add(HASINFRASTRUCTURE, nt);

    Because a nitrate treatment does not replace the existing water treatment plant, the latter will not be deleted. The use of a nitrate treatment plant requires additional raw material. Therefore the variable costs per cubic meter increases. The variable costs have to be changed.

    entity().set(VARIABLE_COSTS, entity().get(VARIABLE_COSTS) + nt.entity().get(NitrateTreatment.OPERATING_COSTS));

    If the strategy ‘co-operative agreements’ is selected, a contract between water supply company and farmers will be introduced. The costs of this action are accounted for in another method.
    The effect of this measure on nitrate concentrations in groundwater is delayed because of the time it takes for the emissions to seep through the unsaturated zone. Accordingly, in the object 'water resource', a method is called up that calculates and returns this delay.  Also, an action is created which is activated after the delay time and passes the new nitrate emission on the water resource. The resulting new value can be used to calculate the nitrate concentration in the groundwater.

    this.scheduleIn(water.transportDelay(), false,
      new Action("reduceNitrateRecharge", Action.HIGH_PRIORITY)
        protected void execute()
          sim.output("Started reducing nitrate recharge."):
  2. With the method registerWaterDelivery the company calculates the amount of water it delivers in two steps: First, the water consumption of the last period is recorded;  secondly, the water loss is added to customer demand. The water loss depends on the condition of the distribution network. In the model, it is assumed that the worse the network’s condition (depending on book value), the more leaky it is. This can lead to a higher water delivery even if real customer demand is stable or decreasing. Depending on the proportional book value the water loss is between 0% and 25% of the water transported through the network.

  3. The total costs of the water supply company consists of depreciation sum, labour costs, variable costs, interests for credits, other costs, and if necessary costs for co-operative agreements with agriculture. The total costs are calculated with method calculateCosts.

  4. During every time step, the company decides upon its reinvestment with the method reinvest. In the current model version, the company pursues the target to ensure a specific level of security and reliability of operation. This level is specified by setting the attribute 'security book value'. The higher this value, the more reliably the plant works.5
    If the actual book value of a plant is smaller than its security book value it calculates the required reinvestment in method calculateReinvest.

    double differenz = a.entity().get(a.PROPBOOKVALUE) – a.entity().get(a.SECURITYBOOKVALUE);
    if (differenz < 0)
      reinvestment += a.calculateReinvestment(10.0);

    The reinvestment sum for all plants is given to the bank which produces a credit.

    if (reinvest > 0.0)
      Bank bank = (Bank)sim.repository.getClient("bank");
  5. The water price for the next period will be fixed in method calculateWaterPrice.

Co-operation with agriculture

This object will be generated if the model user chooses the strategy “co-operative agreements”. There are three types of cost, incurred at different stages of the cooperation agreement.  First, the water supply company has to incur one-off implementation costs for specifying the terms of the contract, for surveys of soil conditions, and negotiation with the decisionmakers. Secondly, during the early stage of co-operation, the costs for consultation are higher than in later periods (Gramel et al. 2001, Gramel & Urban 2001). These attributes can be edited by the model user, together with some others, as shown in figure 5.

Figure 5

Figure 5. Co-operation

Water resource

In the current model version, it is assumed that there is no quantity problem in water supply: there is enough groundwater to satisfy the demand for drinking water. Nitrate in groundwater caused by agriculture is the problem to be solved.
Therefore the object ‘water resource’ has to calculate the nitrate concentration in the groundwater depending on some physical attributes. This is done with the following formula which depends on a differential equation describing the mixing process of the recharge water with the water of the aquifer:

c(t) = (c0-crecharge) * e – [G * t / (H * nf)] + crecharge    (Rohmann et al. 1985)

c recharge: nitrate concentration of the grounwater recharge [mg NO3/l]
G: quantity of groundwater recharge [l/m2*a]
H: thickness of the aquifer [m]
nf: effective porosity [-]
c0: nitrate concentration at beginning [mg/l]
t: time
c(t): nitrate concentration at time t [mg/l]

In the repository, the user can view the actual nitrate concentration in the groundwater as shown in figure 6.

Figure 6

Figure 6. Water resource

After the co-operative agreement with agriculture is concluded, the water supply company calls up the method transportDelay to calculate the delay in the change of nitrate concentrations in the groundwater, caused by the travel of the emissions through the unsaturated zone. It uses the following formula (which is not an exact one but based on expert experience):

Dt = nFK * d/GWnew   (HMU 1996)

nFK: net field capacity [l/m2]
d: thickness of unsaturated zone [m]
GWnew: quantity of groundwater recharge [l/m2*a]
Dt: transport delay through unsaturated zone [years]

Figure 7 and 8 show the nitrate concentration with and without co-operative agreement:

Figure 7

Figure 7. Nitrate concentration without co-operative agreement

nitrate concentration of the groundwater recharge: 80 mg/l
quantity of groundwater recharge: 200 l/(m2*a)
thickness of the aquifer: 25 m
effective porosity: 0.35 mg/l
nitrate concentration at the beginning: 10 mg/l

Figure 8

Figure 8. Nitrate concentration with co-operative agreement

same values as used in the context of figure 7;

Co-operative agreements with agriculture are concluded when the nitrate concentration reaches 40mg/l. The reduction of nitrate emission amounts to 25% - i.e. to 20mg/l.

Bank and credits

To carry out reinvestments, the water supply company has to take out loans. The model user can specify credit period and interest rate. In the current model version, only one kind of credit is possible. It would be useful to have different kinds of credits;  but then, the water supply company would have to have rules to decide which one it would like to take.
In the beginning the model user can define existing credits and initialise them with four attributes. The credits which are generated during a simulation run are visible in the repository with their remaining pay-back time and annual interest burden, as shown in figure 9. Once they are repaid, they will be deleted.

Figure 9

Figure 9. Bank and credits

Customer types

In future model versions, there will be different customer types with different demand and behaviour. The current model version does not yet make full use of this object. 'Customer types' is just needed for the water demand and is therefore already available.
In the method demandLastYear 'customer types' calculate their water demand for the last period. So far, the demand is the same in every period.

* Simulation results

The model version described in this article is mainly used for the evaluation of the structure of the model. Therefore, we are interested in those simulation cases which allow comparison with empirical data. An existing water supply company has been chosen to work out this simulation. The data of this case study are integrated into Figures 10 and 11.

The level of company debt is not an input into the model because the decision to take out additional loans is determined by an algorithm of the model. That is why the comparison of simulation results with the data of the case study is a criterion for the quality of the model.
Figure 10 shows the simulation results: After 20 years the interest burden reaches a quite stable level of 600 000 to 700 000 DM. The existing water supply system has an interest burden of about 650 000 DM.

Figure 10

Figure 10. Interest loads

Also, the total costs and the proceeds of the simulation show a quite similar level to the real values (total costs: 5.4 and proceeds: 5.2 million DM).

Figure 1

Figure 11. Costs and proceeds

Already this version of the model allows scenarios which go beyond the comparison with the reference water supply system. For example, the nitrate concentration and costs of different nitrate-strategies can be simulated.

As still described at the end of  the chapter 'Situation of German Water Supply'  the user of the model selects the strategy and has to define which nitrate concentration is the sign for the start of a measurement.

The VSEit framework offers a special window showing the status and of key steps of the simulation.  Figure 12 shows this window for the example of a co-operation strategy7 . Prior to the start of the simulation, the simulation has to be initialized (simulation step “zero"). The start value for the beginning of the co-operation is 35 mg/l. After 26 years this nitrate concentration has been exceeded and the co-operation is started. After a delay of 10 years the reduction of the nitrate in the groundwater begins.

Figure 12

Figure 12. Simulation steps of a co-operation model

Figure 13 shows the adequate nitrate concentration in the groundwater. If the strategies “Deeper wells” or “Nitrate treatment” are chosen, the nitrate concentration of this aquifer exceeds the limit value of 50 mg/l whereas in the case of a co-operation the nitrate concentration does not exceed the limit value.

Figure 13

Figure 13. Groundwater nitrate concentration depending on the strategy

Figure 14 shows the annual total costs8 under the conditions of the three strategies (co-operation with the agriculture, deeper aquifers, nitrate treatment).
The measurements of the strategies “Deeper aquifers” and “Water treatment” begin at a level of 45 mg/l nitrate9 - and adequately the costs.

Figure 14

Figure 14. Annual total costs under the conditions of different strategies

The use of deeper aquifers is the strategy with the lowest costs. Sometimes this strategy is not possible as the hydrogeological situation does not allow water withdrawal from deeper aquifers.
Co-operations with the agriculture are a little bit cheaper than a additional nitrate treatment to reduce the nitrate concentration. After the period of the first very intensive agricultural consultation10 the costs of the co-operation drop once again. On the other hand, co-operations have to be started earlier (in this case: 15 years) than the other strategies. Additionally, the effect of a water treatment is sure whereas agricultural measurements have risks concerning the impact on the groundwater.

The financial comparison of these two strategies also depends much on the interest rate. The costs of water treatment consist mainly on investment costs whereas the investment costs of co-operations are very low. Accordingly treatment plants lead to more credits so that the water treatment is relatively cheaper if the interest rate (and consequently the burden of interest of the credits) is low.
Due to these factors, it is not easy to compare the efficiency of the two strategies.

These results reflect fairly well the present situation in Germany: Together with the mere mixture of raw waters and the purchase of water from other water supply companies the use of deeper (and less polluted) aquifers is the most widespread strategy to deal with high nitrate concentrations of groundwater. Co-operations with the agriculture are much less frequent;  also water treatment to reduce nitrate concentration does not exist often (e.g. Rohmann 1985).

* Conclusions

The current version of WATER represents a first self-contained model of the problem domain. Its main purpose is the evaluation of the principal model structure and behavior. As shown above, the economical data computed by the model do match the empirical data quite well. Moreover, the outcome of several measures for dealing with nitrate pollution seems to produce plausible scenarios.

The behavior of the water supply company has to be supplemented in the next model version. The company has to be equipped with additional target variables like profit maximization, price stability, and so on. To solve the trade-offs between those targets special conflict solving rules have to be implemented. Also, several types of water consumers will be introduced differing in their water demand depending on type and size of houshold, awareness of environmental problems, etc.

These extensions of the model should be relativly easy to implement with VSEit continuing the stepwise development of the current model. The individual modelling steps so far have been:

  1. Implementation of the water supply company with its plants and internal bookkeeping
  2. Introduction of the water consumer and a water resource
  3. Implementation of bank and credits in order to allow for the reinvestment in plants and for financing measures to reduce nitrate pollution.

Experience showed that these extensions could be implemented by persons without programming background. A basic introduction into the Java programming language was sufficient, since special knowledge about, say, the design of user interfaces or file handling is not required to utilize the VSEit framework. All in all, VSEit allows the modeler to concentrate on model development and, at the same time, provides sufficient flexibility to adapt to the modelers ideas.

* Notes

1 This has to be seen in the context that German municipalities are often heavily indebted (e.g. Nassmacher 1999).
2 50 mg/l NO3 = 11.2 mg/l NO3-N
3 Water-bearing stratum
4 If the layer between the higher nitrate polluted aquifer and the deeper aquifer is permeable the pollution of the deeper aquiver is also possible – accelerated by the water withdrawal from the deeper level.
5In fact the process of reinvestment is more complex. E.g. a water treatment plant consists of various parts with different useful life expectancies.
6Especially because of the unsaturated zone and the mixture of the percolating water with the groundwater
7Nitrate concentration of the groundwater at the beginning: 10 mg/l, nitrate concentration of the recharge without co-operation: 80 mg/l, nitrate concentration of the recharge with co-operation: 20 mg/l , groundwater recharge: 150 mm/a, thickness of the aquifer: 25m, effective porosity 0.35, net field capacity of the soil: 150 l/m², thickness of the unsaturated zone: 10m
8The curves are quite smooth as the annual total costs e.g. does not show the investment costs of one year but the distributed costs for the credit during several years.
9Which is exceeded in the year 41
10In the context of this simulation 10 years

* References

BMWI, Bundesministerium fuer Wirtschaft und Technologie Eds (2001) Optionen, Chancen und Rahmenbedingungen einer Marktöffnung  für eine nachhaltige Wasserversorgung, Berlin,

BRASSEL K H (2001) Flexilbe Modelling with VSEit, the Versatile Simulation Environment for the Internet. Journal of Artificial Societies and Social Simulation, 4(3), http://jasss.soc.surrey.ac.uk/4/3/10.html.

ENTELMANN  I, Diersen G and Wichmann K (2000) Potentiale für eine Nachhaltige Wasserwirtschaft durch Privatisierungsprozesse? gwf – Wasser/Abwasser, 3/2000. pp 157-163.

EUROPEAN ENVIRONMENTAL AGENCY Eds (2001) Environmental signals 2001, Copenhagen, http://reports..eea.eu.int/signals-2001/index_html

GRAMEL S and Haffner Y (2001) Privatisierung und Ressourcenschutz - eine Problemskizze. Umweltmagazin, 18th August 2001 (forthcomming).

GRAMEL S, Neumüller J, Peter M and Urban W (2001)  Kooperationen zum Grundwasserschutz zwischen Landwirtschaft und Wasserversorgung in Hessen - Wirkungen und Kosten. Landnutzung und Landentwicklung, 5 or 6/2001 (forthcomming).

GRAMEL S and Urban W  (2001) "Co-operative agreements between agriculture and water supply". In Environmental Agency UK (Eds), Protecting Groundwater,  conference on 4-5 October 2001, Birmingham, http://www.environment-agency.gov.uk/pgw

HMU Hessisches Ministerium für Umwelt, Energie, Jugend, Familie und Gesundheit and Kuratorium fuer Technik und Bauwesen in der Landwirtschaft e.V. Eds (1996) Landwirtschaft in Wasserschutzgebieten, Wiesbaden, not published.

LAWA Länderarbeitsgemeinschaft Wasser Eds (1995) Bericht zur Grundwasserbeschaffenheit Nitrat, Stuttgart.

NASSMACHER H (1999) Kommunalpolitik in Deutschland, Opladen.

ROHMANN U and Sontheimer H (1985) Nitrat im Grundwasser, Karlsruhe.

SPELTHAHN S (1994) Privatisierung natürlicher Monopole – Theorie und internationale Praxis am Beispiel Wasser und Abwasser, Wiesbaden.

SRU, Rat von Sachverständigen für Umweltfragen Eds (1998) Flächendeckend wirksamer Grundwasserschutz – Sondergutachten, Wiesbaden, http://www.umweltrat.de/gutach98.htm

UBA, Umweltbundesamt Eds (2000) Liberalisierung der deutschen Wasserversorgung – Auswirkungen auf den Gesundheits- und Umweltschutz, Skizzierung eines Ordnungsrahmens für eine wettbewerbliche Wasserwirtschaft, Texte 2/00 des Umweltbundesamtes, Berlin, http://www..umweltbundesamt.de/wasser/themen/index.htm


ButtonReturn to Contents of this issue

© Copyright Journal of Artificial Societies and Social Simulation, 2001