Proceedings of the 6th International Conference on the Application

of Physical Modelling in Coastal and Port Engineering and Science

(Coastlab16)

Ottawa, Canada, May 10-13, 2016

Copyright ©: Creative Commons CC BY-NC-ND 4.0

1

PHYSCAL SCALE AND 3D-NUMERICAL MODEL SIMULATIONS OF

TIDAL FLOW HYDRODYNAMICS AT THE PORT OF ZEEBRUGGE

WAEL HASSAN1, MARC WILLEMS2, JORIS VANLEDE2

1 Ghent University / Flanders Hydraulics Research, Belgium, wnmhassan@gmail.com

2 Flanders Hydraulics Research, Belgium, marc.willems@mow.vlaanderen.be, joris.vanlede@mow.vlaanderen.be

ABSTRACT

The combination of various hydraulic models (both physical and numerical) and available field data is a unique research

platform to setup an intercomparison between different results of these models. This kind of research is rather scarce due to

lack of data or only one single model in most research projects.

In a large integrated research project for the port of Zeebrugge, Flanders Hydraulics Research had the opportunity to

combine the results of two models in order to perform a detailed analysis of the tidal current in front of a coastal port and to

find the best alternative layout for this port. It was also a good occasion to evaluate and to improve the various modeling

techniques.

The model intercomparison showed corresponding results and also differences in the current velocity. Based on the

analysis it can be recommended to consider not only the output values of a model but to calculate the relative outcome of

different runs to interpret the impact of modified layouts of the port.

KEYWORDS: physical model, numerical model, hydrodynamics, tidal current.

1 INTRODUCTION

This study is part of an ongoing integrated study to optimize the nautical accessibility of the port of Zeebrugge (Belgium).

The hydrodynamic part of this large research project aims to reduce existing tidal cross-currents (> 2 m/s, or > 4 kn in Figure

1 (Vlaamse Overheid Afdeling Kust, 2011)) to a safe limit (1 m/s) at the port entrance and to find the best solution out of

proposed design scenarios. A detailed analysis of spring tidal flow dynamics around the port has been carried out using 2

models: (i) a large physical scale model and (ii) a 3D numerical model using Telemac3D.

This paper focuses on the tidal flow hydrodynamics and presents the main results of the model calibrations with existing

on-site data and model intercomparison. Three different design scenarios are used to compare the physical and numerical

model results.

2

Figure 1. On-site measurement of the maximum flood current at the port. (© Vlaamse Hydrografie)

2 THE PHYSICAL AND NUMERICAL MODELS

2.1 Physical model

A distorted physical scale model of the port configuration was constructed (Willems et al., 2011) at Flanders Hydraulics

Research (FHR), with a fixed bed, an area of 2000 m² (16 km by 10 km in prototype) and a 1:300 horizontal and 1:100 vertical

scale. The size of the physical model and provisional boundary conditions of water levels, velocities and discharges were

calculated by using a Delft3D numerical model especially calibrated for the Zeebrugge area. By optimizing the time varying

boundary conditions and the roughness of the breakwaters and the nearshore zone, the physical model was successfully

calibrated (Willems et al., 2014) to obtain a very good agreement for the tidal current (both magnitude and direction) in the

access channel to the port. Also the eddy formation in the outer port was modeled correctly.

The physical model simulates a mean spring tidal cycle (12.5 hours in prototype, 25 min in model) and repeats this tidal

cycle continuously. The model needs one tidal cycle to start up; model measurements can be performed from the second tidal

cycle.

3

Figure 2. 3D visualisation of the physical model (colour = bathymetry of the model).

Figure 3. General view of the physical model (55 m x 35 m).

Figure 4. Detailed view of the port in the physical model (13 m x 13 m).

Different measuring techniques are used to measure water level variations and flow velocities. The water level in the

model is measured by analog remote ultrasonic sensors. Flow velocities are measured in two different ways:

The time varying magnitude and direction of the tidal current at specific locations are registered by six 4 quadrant

electromagnetic liquid velocity meters at a certain level in the water column (point measurement).

The time varying surface velocities and current patterns are calculated in specific areas by the Particle Tracking

Velocimetry technique (PTV). Two video cameras register 50000 floating hollow white polypropylene particles

(diameter 15 mm). In the Streams software these particles are identified in the captured images and all movements

are tracked to calculate the velocity vector of all particles (Nokes, 2012). The PTV software generates Lagrangian

flow statistics and is able to interpolate these velocities onto a regular grid to obtain Eulerian velocity information.

Finally this technique results in velocity vector maps of the area under investigation over the tidal cycle (e.g. Figure

8, left plot).

In the physical model the velocity was measured up to 3000 m in front of the port.

2.2 Numerical model

The numerical Telemac3D model is used to simulate the spring tidal cycle around the port at a larger domain than the

physical scale model. The domain of the Telemac model spans a circle with a radius of 70 km around the port of Zeebrugge.

The bathymetry close to and inside the port is identical to the physical model (Figure 5).

4

Figure 5. Bathymetry of the Telemac3D model (IMDC, 2013).

In the Telemac model the 3D shallow water equations are solved by the finite element method in combination with the

advection-diffusion equation for tracers like salinity and temperature. All parameters are discretized in a triangular grid

(IMDC, 2013). After choosing the right parameters for the time step, vertical layers, boundary conditions, viscosity and HLES,

the model was calibrated by optimizing the Manning coefficient.

The Telemac3D model simulated a neap tide – springtide – cycle (fourteen days) with the springtide at the end of this

cycle. The on-site data used for this calibration (Vlaamse Hydrografie, 2016) is different from the data used for the calibration

of the physical model.

3 SCENARIOS

Three different port layouts were selected for the comparison (Figure 6).

scenario T0 present situation with 2 breakwaters (reference layout for model calibration, Figure 6a);

scenario T17 layout with a modified bathymetry in front of the port entrance (sea bed level lowered10 m

below the actual access channel, in the curved rectangle in Figure 6b);

scenario T20_2 layout with a modified seabed in front of the entrance (= T17) and an additional western

breakwater (Figure 6c).

The reference layout T0 was used to calibrate both models based on prototype measurements. As the available prototype

measurements were different for both models, the calibration results were slightly different.

The two modified layouts T17 and T20_2 were designed to reduce the cross current in the access channel in front of the

port entrance (Hassan et al., 2014). It was verified and confirmed that the additional breakwater in T20_2 did not introduce

model effects due to the distance between the breakwater and the model boundaries.

5

Figure 6. Layout of the 3 scenario’s (upper: T0; left: T17; right: T20_2).

4 RESULTS AND DISCUSSION

4.1 Water level

The water levels in the 2 models and in prototype are shown in Figure 7. The presented prototype curve is the mean

springtide. In general there is a very good agreement between the models and prototype, taking into account the differences

in the setup and the calibration of both models. However, there is a small difference between 6h and 3h before highwater but

this is acceptable because it is within the confidence band and at the end of the ebb phase when the tidal current is relatively

small. In the rest of the tidal cycle, and especially the last 2 hours before highwater when the level raises quickly and the tidal

flood current is strong, the water levels are almost identical.

Figure 7. Comparison between the water level of the models and prototype (scenario T0).

6

4.2 Velocity

In this section the velocity of the maximum flood current in the tidal cycle (50 minutes before highwater) is discussed.

Figure 8 shows the velocity of the maximum flood current in scenario T0 as predicted by both models. The values of the color

maps are the same in both figures. Although both figures look very different at first sight, the flow fields are comparable and

the flow velocity in the access channel is very similar. The difference in cross current in the access channel is 0.20-0.25 m/s

(Figure 9) and the velocity gradient along the channel is almost the same. These observations are important to assess the

nautical accessibility of the port.

Figure 8. Velocity of the maximum flood current in scenario T0 (left: PhysMod – right: Telemac).

Figure 9. Maximum velocity of the cross current in the access channel (scenario T0).

Figure 10 shows the predicted velocity of the maximum flood current in scenario T17 by both models. The velocity

difference with scenario T0 (lower plots in Figure 10) shows that the lowering of the bathymetry in front of the entrance has

the same effect in both models.

The difference in cross current in the access channel is approximately 0.20 m/s (Figure 11, left), and the velocity gradient

along the channel is almost the same. The reduction of the cross current (Figure 11, right) is identical in both models. This

means that both models show the same impact of this scenario on the existing tidal current. To overcome the small differences

between the calibrated models to assess the nautical accessibility of the port for this scenario, the simulated impact should be

applied velocities measured in the field.

7

Figure 10. Velocity of the maximum flood current in scenario T17 and velocity difference with scenario T0

(left: PhysMod – right: Telemac).

Figure 11. Maximum velocity (left: scenario T17) and reduction (right: (T17-T0)/T0*100) of the cross current in the access

channel.

Figure 12 shows the velocity of the maximum flood current in scenario T20_2 (upper plots) and the velocity difference

with scenario T0 (lower plots) in both models. It is clear that the new breakwater results in the same increase in velocity in

front of the breakwater head and in the same reduction in velocity near the port entrance in both models. On the other hand,

the increase of the velocity extends over a different distance. The increased velocity just crosses the access channel in the

physical model (left plots) but it drops just before the access channel in the Telemac3D model (right plots). As a consequence

there is a difference in cross current in the access channel at 700-2000 m in front of the port entrance (Figure 13). This

difference has significant consequences for the evaluation of the nautical access as the safety limit (1 m/s) is violated in the

physical model but achieved in the Telemac3D model. Finally the cross current at 200-300 m in front of the port entrance is

approximately 70% smaller in both models.

8

This conclusion shows that model results have to be analyzed carefully, and some uncertainty about the magnitude and

the location of the maximum velocity of the tidal current has to be considered.

Figure 12. Velocity of the maximum flood current in scenario T20_2 and velocity difference with scenario T0

(left: PhysMod – right: Telemac).

Figure 13. Maximum velocity (left: scenario T20_2) and reduction (right: (T20_2-T0)/T0*100)

of the cross current in the access channel.

9

5 CONCLUSIONS

The tidal current in front of the port of Zeebrugge has been investigated in a physical scale model as well as in a numerical

model. Both models were successfully calibrated based on different on-site data. A detailed intercomparison has revealed a

good agreement between the models but also some differences in the results that have to be considered when analyzing the

impact of a modified port layout.

Based on this unique combination of available field data and models, it is proven that velocities calculated in different

models can be somewhat different and uncertainty about the current velocities should be taken into account. In this case it is

better to analyze the relative impact of a modified scenario by comparing the altered velocities with the calibrated ones. This

relative impact can then be used on field data to obtain the best possible results for scenario analysis.

REFERENCES

Hassan W., Willems M., Troch P., 2014. ‘A detailed hydrodynamic study on tidal flow at the port of Zeebrugge’, Coastlab’14 Europe Congress, book of proceedings, 2014, Varna–Bulgaria.

IMDC, 2013. Havenuitbreiding Zeebrugge-Telemac model - Calibration and validation. (I/RA/11401/13.072/JUD).

Nokes R., 2012. ‘Streams version 2.00. System Theory and Design’, University of Canterbury, Christchurch, New Zealand.

Vlaamse Hydrografie, 2016. Meetnet Vlaamse Banken – Welcome [Homepage of www.meetnetvlaamsebanken.be], [Online]. Available:

http://www.meetnetvlaamsebanken.be/Default.aspx?Page=&L=en [20/01/2016].

Vlaamse Overheid Afdeling Kust, 2011. ‘Stroomatlas: haven van Zeebrugge 2011: stroommetingen in de Pas van het Zand en de haven van Zeebrugge’, Afdeling Kust, Oostende, België.

Willems M., Van Dingenen B., Delgado R., Verwaest T., Mostaert F., 2011. Nautische toegankelijkheid haven Zeebrugge: technisch

ontwerp schaalmodel. WL Rapporten model 780_03. Waterbouwkundig Laboratorium, Antwerpen. (in Dutch).

Willems M., Hassan W., Heyvaert G., 2014. ‘Calibration of the large physical model of the port of Zeebrugge’, 3rd IAHR Europe Congress, book of proceedings, 2014, Porto–Portugal.