WATER DEMAND AND SUPPLY ANALYSIS USING A …environ.chemeng.ntua.gr/en/Uploads/Doc/Papers/Water/2001_Water... · WATER DEMAND AND SUPPLY ANALYSIS USING A SPATIAL DECISION SUPPORT

INTRODUCTIONThe need for efficient, equitable, and sustainablewater allocation policies has increased in importancebecause of the growing scarcity and competition forwater across sectors. The management process ofwater resources must prevent and resolve conflictsamong different users. Sectoral approaches to waterresource development and management have beenand are still dominant (Lilburne et al., 1998; Salman

et al., 2001) but there is a need for a shift towards anintegrated or cross-sectoral approach to avoid frag-mented and uncoordinated development of waterresources (Rosegrant et al., 2000; Staudenrausch andFlugel, 2001). Water allocation policies can best beexamined at a hydrological basin level, which linksessential hydrologic, economic and institutional rela-tionships as well as water uses and users and alloca-tion decision making process.

Global Nest: the Int. J. Vol 3, No 3, pp 199-209, 2001Copyright© 2001 GLOBAL NEST

Printed in Greece. All rights reserved

WATER DEMAND AND SUPPLY ANALYSIS USING A SPATIAL DECISION SUPPORT SYSTEM

1 National Technical University of AthensSchool of Chemical Engineering

Heroon Polytechniou 9, Zografou Campus, 157-80, Greece

2 Technological Education Institute (TEI) AthensFaculty of Technological Applications

Ag. Spyridonos and Palikaridi, 122-10, Egaleo, Athens, Greece

3 Dr. Chemical Engineer

* to whom all correspondence should be addressede-mail: [email protected]

ABSTRACTA prototype Spatial Decision Support System for the evaluation of water demand and supply manage-ment schemes is presented. The water basin is topologically mapped to a network of spatial objects rep-resenting the physical entities and their connections. Several GIS functions, which include datainput/update, network derivation from the basin map and network building/modification are incorpo-rated. The tool integrates suitable models for demand site requirements calculation and water alloca-tion. Alternative scenarios can be constructed, trends and interactions of the complex water system canbe analysed, strategies to solve water allocation conflicts can be evaluated and necessary infrastructureinterventions can be planned in advance in order to meet water needs. The tool is demonstratedthrough a case study, involving the current situation and future policies for a typical Greek island.

KEY WORDS: Integrated water resources management, spatial decision support system, water alloca-tion model.

E. MANOLI1

G. ARAMPATZIS1

E. PISSIAS2

D. XENOS3

D. ASSIMACOPOULOS1,*

Received: 01/06/02Accepted: 18/06/02

arampatzis.qxd 16/9/2002 5:52 Page 199

Integrated Water Resources Management(IWRM) has to promote the co-ordinated develop-ment and management of water, land and relatedresources, in order to maximize the resultant eco-nomic and social welfare in an equitable mannerwithout compromising the sustainability of vitalecosystems (Bouwer, 2000; Albert et al., 2001). Oneof the central IWRM challenges is to find the rightmix of management tools (Simonovic, 2000) whichmay vary from situation to situation. Such toolsrange from water resource assessment, demandregulation and regulatory instruments, socialchange instruments, conflict management, econom-ic instruments and information and communicationinstruments. Mathematical models based on ahydrological basin spatial mapping offer the frame-work that allows an integrated analysis of differentwater-related sector elements, such as agriculture,municipal and industrial water supply and tourism. The interdisciplinary nature of water problemsrequires new methods to integrate the technical,economics, environmental, social, and legal aspectsinto a coherent environment (McKinney et al.,1999). Nowadays, there is a need for multi-objec-tive, multipurpose and multi-facility projectapproach to solve water resource allocation prob-lems. In a modeling framework, the objective func-tion is an essential instrument to reflect the host ofrules, principles and constraints in water resourcemanagement. Several objectives (economic, effi-ciency, social well-being, environmental, sustain-ability, etc.) have to be dealt with simultaneously.Spatial decision support systems (SDSS), are aclass of computer systems in which the technolo-gies of both GIS and DSS are applied to aid deci-sion makers with problems that have a spatialdimension. GIS is a general-purpose technologyfor handling geographic data in digital form,offering a spatial representation of waterresource systems, but currently little predictiveand related analytical capacities are available forsolving complex water resource planning andmanagement problems. DSSs are interactive pro-grams, which embed traditional water resourcesimulation and optimisation models, with adapta-tion of new approaches, to support users in semi-structural or ill-structural problem solving.In the present work a prototype SDSS for waterresource management in hydrological basins is pre-sented. A hydrological basin is topologicallymapped to a network of spatial objects. The SDSS

integrates suitable models for demand site require-ments calculation and water allocation on the basisof alternative scenarios. The following sectionsgive an outline of the architecture, the modellingframework and the operational details of the sys-tem. A characteristic case study, involving the cur-rent situation and future policies for a typicalGreek island is presented in order and demon-strates the effectiveness of the proposed approach.

TOOL ARCHITECTUREThe structure of the developed SDSS is presentedin Fig. 1. The central objective in the design of thesystem is to integrate data, models and decisionanalysis processes into a unified software pack-age. The system was implemented within thecomputational environment of Microsoft VisualBasic. The GIS functionality is embedded withobjects of the MapInfo MapX ActiveX compo-nent. Users interact with the system via a GIS map-based user-interface, which provides the function-ality of inputing information and viewing ofresults through appropriate maps, diagrams andtables. A network representation of the hydrolog-ical basin is derived from the core database.Characteristic scenarios can be developed withthe use of a network editing tool and futureassumptions that affect demand, supply andhydrology can be specified. Scenarios are evaluat-ed with the aid of a demand calculation proce-dure and a water allocation model. Scenarios canthen be planned, simulated and evaluated and thedecision-maker can undertake rational actionswith respect to his objectives.

DatabaseThe GIS database is the heart of the spatial andoperational information system as well as thestorage system that allows communication andintermediate storage between models and subse-quent reporting modules. The object model of thedatabase is presented in Fig. 2. The database hasbeen developed around a geographical hierarchy,which is dictated by the very same nature of avail-able information. The hierarchy is implementedthrough a collection of maps (chartographic rep-resentation) and a collection of tables withattribute data and time series (tabular represen-tation), connected through the data-binding pro-tocols supported by the MapX technology.

200 MANOLI et al.


201WATER DEMAND AND SUPPLY ANALYSIS USING A SPATIAL DECISION SUPPORT SYSTEM

Figure 2. Central database object model and attribute data

Figure 1. Architecture of the Spatial Decision Support System


For each area identified as demand or supplyregions, irrigated areas, industrial plants, surfaceand groundwater resources, storage and distribu-tion networks are retrieved from the database.Each entity is fed with appropriate attribute data,which refer to permanent and seasonal popula-tion, agricultural water requirement, waterresource availability, their monthly variation andtheir associated economic cost and money flows.

Supply Requirements CalculationThe estimation of the supply requirement over aspecified time period is based on a hierarchical dis-aggregation of water demand data. The first levelcorresponds to demand sites. Below, specific activi-ty levels are defined. Activity levels in the SDSSinclude permanent and seasonal population for set-tlements and towns and irrigated areas per croptype. Water demand is calculated by multiplying theoverall activity level by a water consumption rate.Activity levels or water consumption rates can beprojected using functions describing the specificcharacteristics of each demand site or activity level.

Water Allocation ModelSeveral methodologies have increasingly beenused over the last decades for the optimal design,planning and operation of water resource sys-tems. The two basic categories of water resourcemodels are simulation and optimisation models(Wardlaw,1999). Mays (1996) carried out a widereview of these models. Some authors (Mannochiand Mecarelli, 1994; Reca et al., 2001) introducedeconomic objective functions in irrigation waterallocation models. However, many of these mod-els are not readily adaptable to the case of allo-cating water on a hydrological basin level.In the present work, water allocation is achievedthrough a simulation model. A network represen-tation of the hydrological basin is derived fromthe database (Fig. 3). Nodes represent the con-nection between these entities. To capture thefeatures of the water systems� function, differenttypes of node are incorporated. These includesprings, wells, boreholes, water treatment plants,demand sites, etc. The links correspond to theman-made or natural water conduits, such aspipelines, canals, river reaches, etc. The frame-work of the network is constructed by connectingthe nodes and links according to their physicallocations in the water resource system.

Each node i can be classified into one of the follow-ing three categories (i) supply node which is char-acterized by a positive monthly supply rate si, (ii)demand node which is characterized by a monthlydemand rate di, and (iii) transhipment node. Foreach link j two characteristic variables are intro-duced: (i) the link capacity cj which represents themaximum monthly flows allowed (unbounded linkscan be defined by assigning a sufficiently largecapacity), and (ii) the link monthly flow rate fj (thedecision variables of the problem).In situations of water shortage, a conflict arises ofhow to distribute the water available at supplynodes, among the demand sites that are connect-ed to them. The model can solve this problemusing two user defined priority rules. First, com-peting demand sites are treated according to theirpriorities. Each demand site is characterized by apriority, ranged from 1 (highest priority) to 10(lowest priority). During a water shortage, higherpriority demand sites are satisfied as fully as pos-sible. These priorities are useful in representing asystem of water rights. On the other hand, supplypriorities can be used when a demand site is con-nected to more than one supply node. These pri-orities are attached to the links and are useful in

202 MANOLI et al.

Figure 3. Network representation of a water resource

system


ranking the choices of a demand site for obtainingwater.The mathematical concept of the model is to findstationary solutions for each time step (month).For each time step the problem is to find the flowon the network (a set of link flows) that minimizesthe water shortage on all demand nodes:

(1)

subject to the following constraints:

� Supply constraints associated with all supplynodes:

(2)

� Demand constraints associated with alldemand nodes:

(3)

� Flow conservation constraints associated withall transhipment nodes:

(4)

� Capacity constraints associated with all links:

(5)

The model is solved by first constructing a reductionto a standard maxflow problem and then using astandard algorithm to solve the maxflow problem.The maxflow model applies to a basic network, i.e. anetwork which has exactly one source node (s) andone sink node (t). A flow in a basic network is a setof nonnegative link flows, satisfying the conditionsthat no link�s flow is greater than the link�s capacity(Eq. 5) and that the total flow into each internalnode is equal to the total flow out of that node (Eq.4). By the above conditions, the total flow out of thesource node is always equal to the total flow into thesink node. This common value is called the value ofthe flow. Given a basic network, the problem is tofind a flow of largest possible value (a flow such asno other flow from s to t has larger value).

The model formulated above (Eqs. 1-5) isreduced to an equivalent maxflow problem usingthe following transformations (Fig. 4):� A dummy source node (s) is added to the net-

work.� A dummy link from s to each supply node is

added to the network. The capacity of each link isset to the supply rate of the corresponding node.

� A dummy sink node (t) is added to the net-work.

� A dummy link from each demand node to t isadded to the network. The capacity of eachlink is set to the demand rate of the corre-sponding node.

It can be easily shown that the maxflow problemto the transformed network is equivalent to theoriginal problem. The maxflow problem is solvedusing the Ford-Fulkerson method, known as theAugmenting-Path Maxflow algorithm (Dolamand Aldous, 1993; Sedgewick, 2002). To handle the priority system, an iterative goalapproach is used inside each time step. The ideais first to come as close as possible to meeting thehighest priority demand sites and then to try tocome as close as possible to meeting the next pri-ority demand sites but ensuring that the highestpriority demand sites do not compromise.

0 ≤ ≤j jf c

all outcoming links all incoming links

0− =∑ ∑j jj j

f f

all incoming links all outcoming links

− ≤∑ ∑j j ij j

f f d

all outcoming links all incoming links

− ≤∑ ∑j j ij j

f f s

all demand nodes all incoming links

minimize

−

∑ ∑i ji j

d f


Figure 4. Network transformations


SDSS OPERATIONAL ASPECTSThe developed SDSS consists of three basic mod-ules allowing for a complete representation ofdemand/supply scenarios. These are:� Water demand analysis and supply require-

ments estimation module;� Network editing module;� Water allocation and water shortage estima-

tion module.All results are presented via fully customisablegraphs and tables, in order to permit a completeevaluation of existing and proposed infrastructurefor meeting demand needs.For a demonstration of the SDSS application, acase study for the island of Syros (Fig. 5) wasundertaken. Syros is located in the centre of theCyclades complex, is the administrative centre ofthe prefecture and covers an area of 84 km2. Dueto the important administrative role of the island,the permanent population has shown a consider-able increase during the last decades. The perma-nent population is 20,220 inhabitants (2001 cen-sus), nearly 70% of which concentrated at thecapital of the island, Ermoupolis. The rapidtourism development that has been experiencedduring the last 10 years has as a result the aban-donment of traditional agricultural and stock-breeding activities.Urban water consumption for the entire island inthe year 2000 was about 900,000 m3 of which 25%was allocated for tourism activities. Irrigationwithdrawals for the same period were estimatedat 1,200,000 m3. Natural water resources are lim-ited with scarcity problems being more acute dur-ing the arid summer period. With low rainfall(approximately 400 mm yr-1) there are limitedoptions for the exploitation of surface waterresources. Therefore, with the exception ofErmoupolis, which relies on desalination, irriga-tion and urban water demand are met through theextensive exploitation of groundwater resources.The depletion of the island�s aquifers and overex-ploitation during the summer period continue topose a threat for economic development andpreservation of future water resources.

Water demand analysis and supply requirementsestimation moduleThe disaggregation of water use sectors derivedfrom the database is presented in Fig. 6. Activitylevel data, month variation, water use rates and

projection functions can be modified for each sce-nario introduced. For the case study undertaken,parameters used are summarized in Table 1.Estimation of irrigation water needs was based ondata from the 2000 agricultural census and con-sumption rates for the most important crops.Figure 7 presents the annual water demand forA. Syros and Ermoupolis agglomerations whileFig. 8 depicts the monthly variation of waterdemand for irrigation and domestic use as it isestimated for permanent and seasonal popula-tion needs, in 2030.

Network modifying moduleThe network derived from the database repre-sents the base scenario or existing conditions. Ifthe user wants to introduce changes into the sys-tem, he has to use the network building and edit-ing module (Fig. 9). The module is based on theuse of graphical tools for introducing nodes anddrawing links, as well as related actions for appro-priate modification and reshaping of the network.In all case, the user can benefit from the graphicstools supported by MapX GIS technology for per-forming the necessary structure and networkinterconnections.Alternative water supply scenarios were devel-oped in order to meet the water needs ofErmoupolis, A. Syros and irrigation purposes, upto the year 2030. The latter involve evaluation of

204 MANOLI et al.

Figure 5. Syros island water resources



Figure 6. Demand parameters and estimation modules

Table 1. Water Demand Estimation Parameters

Permanent Population

Seasonal Population

Irrigation

Growth Rate 1.5% 3% up to 20101.5% for the period

2010 - 2030

0%

Consumption Rate 150 l/d/capita150 l/d/capita

Figure 7. Annual water demand for A. Syros and Ermoupolis


existing infrastructure, dam construction anddetermination of the appropriate time horizon forother interventions such as desalination unit con-struction. The present situation regarding watersupply is presented in Table 2. A number of studies have proposed in the past theconstruction of a dam in the Aetos basin. In the

scenario of the present case study, the dam isexpected to be fully operational by 2005. With acapacity of 400,000 m3 and a maximum annualwithdrawal of 300,000 m3, it should primarily meetthe domestic demand of A. Syros and Ermoupolisand secondarily irrigation demand. The proposedinfrastructure is presented in Figs. 9 and 10.

206 MANOLI et al.

Figure 8. The monthly demand in 2030

Figure 9. Network modifying module


Water allocation and water shortage estimationmoduleThe results from the water allocation module indi-cate that existing network infrastructure and bore-holes can adequately satisfy the population of A.Syros up to the year 2022. For the period 2022 -

2030 small deficits (5,000 m3 in 2030) are evidentduring the peak tourist season (July and August).However, the municipality of Ermoupolis experi-ences severe water shortages and cannot rely onthe existing desalination units in order to meet therapid demand growth (Fig. 11).


Figure 10. Demand site priorities and network configuration for Aetos dam construction

Figure 11. Evaluation of existing infrastructure for Ermoupolis and A. Syros

Table 2. Present water supply status for A. Syros and Ermoupolis

Water resources Total water supply (m3 d-1)

A. Syros Boreholes 360

Ermoupolis Desalination units 3,460


The water shortage for the scenario with theAetos dam construction is presented in Fig. 12.From the year 2010 a water shortage appears inthe irrigation sector since water is allocated forthe needs of Ermoupolis. Urban demand growthresults in a direct shortage from the year 2014.What is important to notice is that the irrigationdeficit is not constant as it would be expectedfrom the estimation of irrigation needs. From theyear 2024 it increases since water is withdrawed inorder to serve the demand of Ermoupolis duringthe summer months.As an additional water supply option, the scenariointroduces a desalination unit for which accordingto Fig. 12, appropriate time for construction isaround the year 2010 The estimation of peakmonth shortages and water availability indicatethat with a capacity of 1,500 m3 d-1, the unit will beable to meet domestic demand in Ermoupolis up

to the year 2025, leaving sufficient water supplyfrom the dam to meet irrigation needs. The unitshould be rebuilt in the year 2025 with a capacityof 2,000 m3 d-1 in order to meet water needs up tothe year 2030.

CONCLUSIONSA prototype spatial decision support system forthe evaluation of water demand and supply man-agement schemes has been outlined. The systemintegrates a spatial database of the study area andits infrastructure, tools to for network editing andspecifying assumptions that affect demand, supplyand hydrology, model to perform demand analy-sis and water allocation and components to man-age and present the information. The tool wastested for a characteristic case study that demon-strated its effectiveness in analysing and support-ing decision making.

208 MANOLI et al.

Figure 12. Water shortage in domestic and irrigation sectors after dam construction

REFERENCESAlbert, X., Mark, O., Babel, M.S., Gupta, A.D. and Fugl, J. (2001), Integrating Resource Management in South

East Asia, Water21 October 2001, 25-30.Bouwer, H. (2000), Integrated Water Management: Emerging Issues and Chalenges, Agricultural Water

Management, 45, 217-228.Dolam, A. and Aldous, J. (1993) Networks and Algorithms: An Introductory Approach, Wiley. Lilburne, L, Watt, J. and Vincent, K. (1998), A Prototype DSS to Evaluate Irrigation Management Plans,

Computers and Electronics in Agriculture, 21, 195-205.Mannochi, F., Mecarelli, P. (1994), Optimization Analysis of Deficit Irrigation Systems, J. Irrig. Drain. Eng., 120

(3), 484-503.Mays, L.W. (1996), Water Resources Handbook, McGraw-Hill, New York.


McKinney, D.C, Cai, X, Rosegrant, M.W., Ringler, C. and Scott C.A. (1999), Modelling Water ResourcesManagement at the Basin Level: Review and Future Directions, International Water ManagementInstitute, SWIM Paper 6.

Reca, J., Roldan, J., Alcaide, M., Lopez, R. and Camacho, E. (2001), Optimisation Model for Water Allocation inDeficit Irrigation Systems I. Description of the Model, Agricultural Water Management, 48, 103-116.

Rosegrant, M.W., Ringler, C., McKinney, D.C., Cai, X, Keller, A. and Donoso, G (2000), Integrated Economic-Hydrologic Water Modeling at the Basin Scale: the Maipo River Basin, Agricultural Economics, 24, 33-46.

Salman, A.Z., Al-Karablieh, E.K. and Fidher, F.M. (2001), An Inter-Seasonal Agricultural Water AllocationSystem (SAWAS), Agricultural Systems, 68, 233-252.

Sedgewick, R (2002), Algorithms in C++ Part 5: Graph Algorithms, Addison Wesley Longman.Simonovic, S.P (2000), Tools for Water Management One View of the Future, IWRA Water International, 25, 76-88.Staudenrausch, H. and Flugel, Q.A. (2001), Development of an Integrated Water Resource Management System

in Southern African Catchments, Phys. Chem. Earth, 26, 561-564.Wardlaw R. (1999), Computer Optimisation for Better Water Allocation, Agricultural Water Management, 40, 65-70.



WATER DEMAND AND SUPPLY ANALYSIS USING A …environ.chemeng.ntua.gr/en/Uploads/Doc/Papers/Water/2001_Water... · WATER DEMAND AND SUPPLY ANALYSIS USING A SPATIAL DECISION SUPPORT

Documents