DECISION SUPPORT SYSTEM FOR MAINTENANCE DREDGING IN THE RIVER RHINE 

N. Douben

Institute for Inland Water Management and Waste Water Treatment | RIZA

P.O. Box 9072; 6800 ED Arnhem; The Netherlands

tel: +31 (0)26 3688593

E-mail: n.douben@riza.rws.minvenw.nl 

S.E. Werners

Resource Analysis

P.O. Box 2814; 2601 CV Delft; The Netherlands

tel: +31 (0)15 2191519

E-mail: saskia.werners@resource.nl

Abstract: The river Waal, the largest Rhine branch in the lower delta is the most important shipping route in the Netherlands. It connects the Rotterdam harbour with the German Ruhr area. Due to economic growth and the improved accessibility of eastern Europe shipping on the river Rhine will continue to increase in the future. As a result the fairway dimensions of the river Waal have to be enlarged. At an ‘Agreed Low River Level’, the minimal depth should be increased from 2.50 to 2.80 metres and the minimal width from 150 to 170 metres. The fairway enlargement will be achieved through several river engineering measures of which dredging is the most important. The development of a Decision Support System (DSS) for dredging on the river Rhine (Waal) is now in progress. It will be fully operational by the end of 2002. The concept of the DSS as a sediment tool, is based on the combination of bed level monitoring-data on the one hand with simplified water level and morphological predictions on the other hand. With the DSS, decisions can be made on whether dredging is necessary during critical low-water periods and after floods. Furthermore, the DSS analyses dredging and dumping strategies, maintains a database with dredging information, determines relative economic benefits and visualises the effects of different strategies.

Keywords: decision support system, fairway management, maintenance dredging, morphological predictions, relative benefits, river Rhine, water level predictions

1    INTRODUCTION

The Rhine is the most important river of the Netherlands, entering the country from Germany. A few kilometres downstream of the border the Pannerdensche kop bifurcation divides it into the river Waal and the Pannerdensch canal. The free flowing river Waal is the largest Rhine branch and one of the worlds busiest inland shipping routes. On average, every three minutes a vessel passes Nijmegen. Its a vital link in the shipping lane from the North Seaport of Rotterdam to the industrial Ruhr area in Germany where the worlds biggest inland harbour (Duisburg) is situated (Figure 1). Table 1 lists some general characteristics of the river Waal.

Part of the economy of the Netherlands is based on its strategic position as a distribution country. Due to economic growth and improved accessibility of eastern Europe, waterborne transport will continue to increase in the future. However, especially during low discharges width and depth in parts of the fairway of the river Waal presently limit the loading capacity and negatively influence safety, rapidity and efficiency. Consequently, the Dutch government has commissioned the enlargement of the fairway. With reference to the greed Low River Level’ (OLR), which on average occurs approximately 20 days a year, the minimal depth will be increased from 2.50 to 2.80 metres and the minimal width from 150 to 170 metres. The enlargement is already in progress and will foremost be realised through dredging, supplemented with river engineering measures such as fixed layers, submerged groynes and submerges vanes and groyne extensions. Maintenance dredging is roughly estimated to average 750,000 cubic metres annually.

Fig. 1    The Netherlands and its most important rivers

Table 1    Characteristics of the river Waal

CHARACTERISTICS OF THE RIVER WAAL
Length		100 km
Discharge	Mean	1,600 m3/s
	Lowest	460 m3/s (under ice cover, 02-18-1929) 505 m3/s (no ice cover, 11-04-1947)
	Highest	8,035 m3/s (01-03-1926)
Mean sediment load		280,000 m3/year (sand and gravel) 1,400,000 m3/year (silt)
Regulation width		260 m
Mean dredged amount		750,000 m3/year
Inland navigation	Amount of shipping movements	175,000 vessels/year (1998)
	Transported load	165 million metric tons/year (1998)
	Collective loading capacity	340 million metric tons/year (1998)

Decisions about dredging are complex. Future river discharges and morphological changes remain uncertain and variations in the annual discharge regime of the river Rhine may be expected due to global climate change. In addition, daily maintenance dredging in the busy fairway of the Waal requires an accurate monitoring and decision tool. The directorate Eastern-Netherlands of Rijkswaterstaat decided to commission the development of a Decision Support System (DSS) for maintenance dredging in the fairway of the river Waal. The DSS aims to support decisions towards cost efficient dredging, making full use of river morphology processes. The DSS will be fully operational by the end of 2002 [Douben, 1998].

Although the first DSS approaches date back several decades, their use in river and fairway management is still very limited. Major papers on the matter were published by Andreu et al. [1996], Olson & Courtney [1992] and Reitsma [1996] and Fedra & Jamieson [1996]. A specific DSS for the Trinity river in Texas (USA) is reported in Ford & Killen [1995]. In the Netherlands several (decision) support systems have been built for river management and landscape planning. The Institute for Inland Water Management and Waste Water Treatment | RIZA and WL|Delft Hydraulics have built a DSS for the Dutch Rhinebranches in the frame-work of the Landscaping Project river Rhine [Silva & Kok, 1996]. Sprague and Carlson [1982] described a few starting points, applicable for a DSS for dredging on the river Waal.

2    DEVELOPMENT OF A DSS FOR DREDGING

Once developed, decision support systems have more then once lead to disappointments with the users [Ubbels & Verhallen, 1999]. Therefore the development of the DSS for dredging is based on the principle of ‘Learning by doing’. Strong participation of the future end users characterises the development process. A suitable method for this kind of development is called Rapid Application Development (RAD) [Kerr & Hunter, 1994] (Figure 2).

Fig. 2    Traditional development and Rapid Application Development (RAD)

In the traditional, linear approach to development (left side of Figure 2), users are consulted only during the initial analysis phase. Thereafter design, development, testing, and deployment are executed in series. The end users will only see the DSS once it is deployed, when changes as a result of miscommunication and changing requirements are difficult to implement. In contrast, in the RAD methodology several rounds of analysis, design, construction, testing and deployment follow each other in an iterative cyclic process. Each cycle produces a new version of the DSS, which becomes an aid for analysis and design, by giving users a concrete context for feedback. Users participate in testing, and development becomes an iterative process of refining successive versions of the DSS. This approach facilitates optimal communication resulting in a DSS that fits to the users’ requirements almost exactly. Optimisation of the development process does not only take place by periodic consultation of the end users, but the various intervening DSS-versions are tested by applying them in daily dredging decisions as well. Hence, the DSS-versions are already operational during the development process.

Parallel to the RAD process, so-called ‘Parallel studies’ provide methods for simplified water level and morphological predictions, as well as a module for the computation of relative benefits. These studies prevent the development process from being delayed by time consuming and intensive research.

The first analysis phase of the development process started at the beginning of 1998 and has resulted in a large amount of information about the desired tasks and structure of the DSS [RA & WL|Delft Hydraulics, 1998/1999]. Furthermore the analysis pointed at various dredging activities over the hydrological year. Decisions have to be made about (i) large scale dredging activities (rofile dredging’) after a period of floods and (ii) small scale dredging activities (shallow spots) during the whole year. These two types of decisions require specific (input) data which differ in time and space.

Presently, the third intervening version of the DSS is operationally used by the directorate Eastern-Netherlands of Rijkswaterstaat. The next version will be completed in June 2001. The development process is intended to be finished by the end of 2002.

3    ELEMENTS OF THE DSS FOR DREDGING

The concept of the DSS is based on the combination of bed level monitoring-data with simplified water level and morphological predictions. A schematic plan of the DSS is illustrated in Figure 3. The figure shows the most important functional aspects (modules) of the DSS: (1) data input, (2) data management, (3) calculation & analyses, (4) presentation & interface and (4) assessment & evaluation.

Fig. 3    Schematic plan of the DSS for dredging on the river Waal

Daily data input consists of the discharge, water levels, rainfall and the normative depth. Furthermore bed level data and buoy positions have to be read into the DSS fortnightly. As a result of bed level degradation on the river Waal, a Dredging Reference Level (DRL) has to be calculated annually and fed into the DSS as well.

The calculation & analysis module is fed by data from the input and data management modules. By defining several boundary conditions, the user can compose strategies based on spatial dredging and dumping volumes. These calculations are executed by different Arc View applications. Together with short and/or long term water level and morphological predictions and relative benefits, dredging strategies can be composed. Readjustment of boundary conditions provides the opportunity to set time related priorities and to optimise dredging strategies.

In the assessment & evaluation module either shallow spot or profile dredging can be studied. Dredging strategies can be compared, based on dredging and dumping volumes, costs and benefits, dredging efficiency as well as an indication of the required dredging time. Ba sically the instrument contains two different decision support systems, one for shallow spot dredging and one for profile dredging.

Strategies and their consequences can be visualised in the presentation module, which is also based on Arc View applications (see for an example Figure 4). Through the interface the user can add data, access the dredging data base and compose and compare dredging and dumping strategies.

Fig. 4    Example of a scenario for shallow spot dredging

The DSS is built as a client-server application. It can be used on a PC within an Internet Explorer environment, while the programme itself runs on a server. Communication between the PC and server is provided by an Intra- or Internet connection.

4    PREDICTION METHODS

At present three different prediction methods are under development as part of the parallel studies: (i) water level predictions, (ii) morphological predictions and (iii) estimation of relative benefits. For the water level and morphological predictions, a distinction is made between short (until two weeks) and long term (one to six months) predictions.

4.1    Water level predictions

Long term water level predictions are applied to anticipate for large scale profile dredging, since this type of dredging needs a longer preparation period. The basic assumption for long term water level predictions is the base flow concept (Lyne-Hollick algorithm [Nathan & McMahon, 1990 and Chapman, 1991]) in combination with a recession curve which is simulated by regression equations [van der Made, 1982]. Conversion of the base flow into water levels takes place with a stage-discharge curve.

Short term water level predictions are applied to optimise small scale shallow spot dredging. They will be calculated with an extended version of the Multi Linear Regression (MLR) model based on statistical relations between several upstream water level gauges and rainfall stations [de Ronde, 1982]. The model is extended from a three day forecast towards a fortnight prediction and is normally used for daily water level forecasts.

4.2    Morphological predictions

During the first phase of the parallel study on morphological predictions, simple prediction methods have been composed to give the users some guidelines for future bed level developments and morphological recovery after dredging. The study is based on relevant literature, numerical model computations and monitoring data from dredging tests that have been carried out during the last three years [Klaassen & Sloff, 2000; Schepman, 2000 & Siegfried, 2000]. The morphological behaviour of the river Waal during differing discharge stages has been studied as well. Shallow spots in the fairway are partly caused by standard river-morphology processes (spiral flow and secondary currents). A distinction is made between the forced morphological phenomena (as a result of the river geometry) and free moving bedforms such as dunes. It was also concluded that existing 2D-morphological models are not accurate enough for the specific purpose, their running time is too long (several hours) and the required knowledge level for utilisation is too high for the users of the DSS.

Studies are carried out to make the following morphological predictions operational:

l        Short term predictions for the benefit of shallow spot dredging. These predictions will be based on the fortnight bed level measurements in combination with normative depths and concentrate on free moving bedforms (dunes). As a result of the short time frame, only 1D processes will be taken into account. In this case morphological variations mainly consist of bedform celerity and transformations.

l        Short term recovery of point-bars. The forced morphology of the river Waal is largely determined by point-bars, which are important in relation to the navigable width of the fairway. In relation to morphological recovery after dredging, two different phenomena can be distinguished: (i) upstream filling (sedimentation) of the point-bar and (ii) tilting of the cross-section towards its equilibrium position. Hence, morphological recovery of point-bars will be estimated with a 2D approach.

l        Long term recovery predictions are based on general relaxation characteristics. These predictions give an estimation of the remaining effects from dredging activities in the preceding year. In general, it deals with morphological processes as a result of higher discharges during winter and springtime. The long term recovery predictions are necessary to anticipate on possible (profile) dredging activities.

l        Short term recovery of dunes. After dredging, the bed level of the fairway is relatively flat. Depending on the development of the discharge, free moving bedforms will arise after a certain period of time. Knowledge about these recovery processes enables the optimisation of dredging scenarios. Klaassen & Sloff [2000] describe several prediction methods which possibly can be utilised in the DSS for dredging.

4.3    Estimation of relative benefits

Benefits (also in comparative and relative ways) are an important parameter regarding time related priorities and optimisation of dredging strategies. The calculation method of the first version of a module in which relative benefits (comparative transportation costs) of inland shipping can be calculated, is based on an increasing normative depth as a result of dredging activities, which (theoretically) results in an increasing loading capacity. After dredging and recovery, transportation of a certain amount of load is possible with a lesser amount of vessels which reduces the total transportation costs. Figure 5 shows the relation between the normative depth and the mean load per CEMT Va vessel (1,500 - 3,000 metric tons) on the river Waal. Similar relations can be derived for other vessel-categories (e.g. push tugs, container vessels, coasters, etc.) and contain probability distributions from historic records of passing vessels. The increase of the mean load per vessel is limited until the total loading capacity is reached. At the concomitant normative depth, dredging does not lower the transportation costs for this specific vessel type anymore.

Fig. 5    Mean load per CEMT Va vessel as a throughout the period of morphological function of the normative depth

5    FUTURE DEVELOPMENTS

Long term water level predictions are already implemented in the DSS. In June 2001 the short term water level predictions will be implemented as well.

The presently operational version of the DSS does not contain a module for morphological predictions yet. At the moment, morphological predictions are based on empirical knowledge and bed level measurements. The first versions of the various simple prediction methods will be implemented in the DSS version of June 2001. These methods have to be further tested and developed with the aid of dredging tests, numerical model computations and empirical facts provided by daily users.

An improved version of the existing module for the calculation of relative benefits will also be implemented in the DSS version of June 2001.

After deployment of the fourth intervening version in June 2001, the development team will give special attention to evaluation and user support, to work towards the final version by the end of 2002.

6    CONCLUSIONS AND RECOMMENDATIONS

With the DSS for dredging on the river Waal, decisions can be made on whether dredging is necessary during critical low-water periods and after floods. Daily maintenance of the fairway can be optimised with the aid of such a DSS. The DSS supports dredging decisions, analyses dredging and dumping strategies, maintains databases and computes costs and (relative) benefits of dredging.

Based upon daily experiences of the users, the DSS provides a firm basis for decisions about dredging activities. The DSS has to operationally prove itself during everyday circumstances, which provides the developers useful feedback from the users. Participation of users is inevitable when a complex DSS, such as described in this paper, is to be developed. The RAD method requires a certain amount of flexibility, which can be a problem when large organisations are involved.

The concept of a combination of bed level monitoring-data on the one hand with simplified water level and morphological predictions on the other hand seems to be suitable for the specific problems surrounding daily dredging decisions on the river Waal. 

Acknowledgements

The authors would like to thank all the participants and future users, working at the directorate Eastern-Netherlands of Rijkswaterstaat, which is funding the project. Gratitude is expressed towards the development team, which consists of staff members from the Institute for Inland Water Management and Waste Water Treatment | RIZA, Resource Analysis and WL|Delft Hydraulics as well as the Transport Research Centre for the development of the benefit calculation method. Finally the authors would like to thank ir. G.J. Klaassen, ir. H.A. Zanting, ir. W. Silva, ir. R.H. Smedes and Mrs. E. Knaapen for their constructive comments on the draft version of this paper.

References

Andreu, J., Caprilla, J. & Sanchis, E. (1996), Aquatool, a generalised decision-support system for water resources planning and management, Journal of Hydrology 177, pp. 269 - 291.

Chapman, T.G. (1991), Comment on ‘Evaluation of automated techniques for base flow and recession analysis’. Water Resources Research, Vol. 27, no. 7, pp. 1783-1784.

Douben, N (1998), Inland shipping on the Dutch river Waal; past, present and future, Lecture on World Delta Symposium New Orleans (USA).

Fedra, K. & Jamieson, D.G. (1996), The Waterware Decision Support System for river basin planning: 2. Planning capability, Journal of Hydrology 177, pp. 199 - 211.

Ford, D.T. & Killen, J.R. (1995), PC-based Decision-support system for Trinity river, Texas, Journal of Water Resources Planning and Management.

Kerr, J. & Hunter, R. (1994), Inside RAD, McGraw-Hill Inc.

Klaassen, G.J. & Sloff, C. (2000), Bed level predictions and morphological recovery relations for the benefit of dredging on the river Waal (in Dutch), WL|Delft Hydraulics.

Made, J.W. van der (1982), Quantitative analysis of river discharges (in Dutch), Ministry of Transport, Public Works and Water Management. Section Water management and Hydrodynamics.

Nathan, R.J. & McMahon, T.A. (1990), Evaluation of automated techniques for base flow and recession analysis, Water Resources Research, Vol. 26, no. 7, pp. 1465-1473.

Olson, D.L. & Courtney Jr., J.F. (1992), Decision support systems and expert systems, Macmillan Publishing Company, New York.

RA & WL|Delft Hydraulics (1998), Analysis phase DSS for dredging, results specialistic explorations and assessment of alternatives (in Dutch), Resource Analysis & WL|Delft Hydraulics.

RA & WL|Delft Hydraulics (1999), Functional design 2003; DSS for dredging and fairway management in the river Rhine (in Dutch), Resource Analysis & WL|Delft Hydraulics.

Reitsma, R.F. (1996), Structure and support of water resources management and decision making, Journal of Hydrology 177, pp. 253 - 268.

Ronde, J,G, de (1982), The forecasting and warning system of 慠ijkswaterstaat’ for the river Rhine, Rijkswaterstaat Section Water management and Hydrodynamics.

Schepman, F. (2000), Morphology of the river Waal with and without dredging; Testing of prediction methods with Arc View (in Dutch), Thesis, Delft University of Technology & WL|Delft Hydraulics.

Siegfried, A. (2000), Morphological response of the river Waal to dredging activities, Thesis, University of Karlsruhe & WL|Delft Hydraulics.

Silva, W & Kok, M. (1996), Landscape Planning of the river Rhine in the Netherlands; Towards a balance in river management, RIZA & WL|Delft Hydraulics.

Sprague, R.H. & Carlson Jr., E.D. (1982), Building effective decision support systems, Prentice Hall, Englwood Cliffs.

Ubbels, A. & Verhallen, A. (2000), Suitability of decision support tools for collaborative planning processes in water resources management, RIZA.