Lock passage can potentially make up a third of a vessel’s travel time for inland waterway transport and therefore has a large impact on inland waterway and multimodal transport networks.
To reduce the overall environmental and economical impact of transport, there is an increasing interest in inland waterway transport relative to road and rail transport. To keep up with this shift towards inland waterway transport, models are created for the optimal arrangement of multimodal transport and transport on inland waterway networks. Hence, lock operations need to be carefully modelled.
The modelling of lock passage is often based on historical data and generalised for all the locks on the network. However, there are uncertainties in historical data, especially with the prospect of an increasing fleet size and more extreme seasonal changes. Additionally, locks can vary in functionality (e.g. recreational or professional use), operability and strategy (e.g. filling and emptying), structural design (e.g. capacity of lock chamber), and environmental conditions (e.g. water level differences).
These variations impact the passage time of phases in the locking cycle. An overgeneralised validation for a simulation model can result in inaccurate simulations for a specific lock.
Therefore, it is desired to use more lock specific data, which can be achieved by applying vessel specific data in the form of AIS (Automatic Identification Systems). AIS, live location data of professional vessels, is used in waterway traffic management, but is also collected for research purposes. The data is easily accessible for a desired period of time and area.
Literature studies show that analyses and validations of simulation models is a common practice with the combination of GPS based data. Some studies use AIS data around locks, only to find the total passage time of the vessel. A lock passage can be divided in more phases that impact the total passage and waiting time of a vessel, examples are the entering and exiting of the vessel and the
operating time of the lock itself.
The objective of this study is to find a generic method to derive validation parameters for simulation models of locks. Data derived from this study can further be used in the optimisation of lock passages in simulation models. In this way, any desired lock and for any circumstances (e.g. seasonal changes or periods of maintenance) a validation can be performed.
A method was created to translate AIS data to information that is relevant for a lock. This method enabled us to analyse the total passage time and the phases of a lock passage that impact this passage time. Based on the AIS data over a certain time period and considering the geometry of the lock, trajectories of passing vessels were derived. These movements were combined into lock cycles
and therefore the corresponding validation parameters could be identified. One example of these validation parameters was the lock operating time, which is estimated based on the principle that all vessels in the lock chamber stopped moving.
The lock specific data can be used to compare the data with the performance of a simulation model of a lock and period in time. This applies for models that only use the total passage time of a vessel, which in some situations might be sufficient. This also applies to models that consider more detailed lock passage. Since the method is based on positional data, the boundaries of the lock sections can be adjusted to the definitions used in the model.
The method is applied on a cases study of the Volkerak and Kreekrak locks. Firstly, the AIS data translation is performed for a lock to create data sets of the lock passages. Also, data of the water level at the lock is linked to each locking cycle. Secondly, this data is compared to a lock simulation in the same situation. The arrival rates, arrival speed, vessel dimensions and water level difference
are used as input to create a base simulation for various configurations. The lock simulation module of OpenTNSim is used because this is an open-source transport network simulation model developed at TU Delft.
The simulations are compared to the collected data, for a total of 18 segments on the trajectory passing the lock. The segments correspond to the phases of a locking cycle. The average passage time of each segment is compared between the vessels of the simulated and the collected data. Because of the large sample sizes, these comparisons created a view on the overall performance of the simulation for each segment.
The information found in the base simulations was further used to calibrate the OpenTNSim package. The vessel speed was adjusted on the approach and leaving of the lock chamber to match the trajectory of the arriving vessels. Also, the filling time of the the lock chamber was optimised for the simulation.
The Volkerak had a total of 24,446 vessels passing over the time period, this is comparable to the 24,570 vessels that were counted by Rijkswaterstaat in the same period. Three cases were selected to compare the simulation model, one vessel passing without waiting, one vessel passing with waiting and two vessels passing. These resulted in sample sizes of 943, 89 and 561 lock cycles respectively. The case study shows that, with a small sample size of 89 locking cycles, there can be large variability in passage times.
The largest deviations of the simulations relative to the collected data were found in the passage time of the segments between the waiting area and the lock chamber. Another deviation was found in the lock operating time for the case of two vessels passing.
The method used can give a large and useful data set of vessels passing a certain lock. The lock data can be used for statistical analyses. An example includes the number of vessels per locking cycle or the entering time or speed. Also, each locking cycle can be assessed and visualised individually or for a desired time span.
When the data is used for the validation of a simulation model, a large data set is needed to give reliable results. After all there can be large outliers and the performance of the lock is dependent on human interaction, the lock operator and the vessel’s captain. A combination of the AIS based data with other additional data can expand the method.
In conclusion, a highly suitable method was found that enables to use AIS data for the validation of a simulation model of a lock. The method is sufficient for most applications of a lock simulation model, despite being limited to just the movement of the vessels. However, when an accurate simulation of a lock phase like the closing of the gate is desired, other data sources might be more suitable. While this research has a focus on the comparison of the collected data to a lock simulation, the data can also be used for a statistical analysis of the lock when looking at fleet composition, arrival rates and stopping distances.

Nature-based initiatives have emerged as potential solutions to problems caused by saltwater intrusion in deltas found globally, but their successful implementation is enabled partly by a multi-stakeholder approach. The latter involves managing several parallel stakeholder objectives, which usually requires quantitative knowledge to understand possible collisions of interests. On that account, the present work developed a systematic approach (referred to as the `comparison tool') to quantitatively compare the objectives of multiple stakeholders interplaying in a delta. As a first approach, the comparison tool is intended to support decision-making to deal with potential conflicts between freshwater supply and port logistics interests. In particular, the present juncture in the Rhine-Meuse Delta in the Netherlands was used to investigate potential trade-offs generated by the nature-based shallowing (or river bed heightening) of the Rotterdam Waterways. The comparison tool is founded on the objective-based assessment of Building with Nature (BwN) solutions. The effects of a BwN solution are assessed separately for each functional requirement and then are related in a combined assessment. The impact assessment for the shallowing of the Rotterdam Waterways required two numerical modelling studies. First, effects on the hydrodynamics and salt transport in a partially-mixed estuary were modelled with the Operationeel Stromingsmodel Rotterdam (OSR), developed by the Port of Rotterdam Authority. Secondly, changes in meso-scale traffic flows over the port network were modelled with the OpenTNSim developed by TU Delft. Afterwards, the effects on freshwater supply and port performance (capacity and efficiency) were quantified separately and then compared. This research delivered a systematic procedure and demonstrated how a combined assessment could be performed in the context of nature-based solutions to mitigate salt intrusion. The most important outcome entails quantitative trade-offs between port efficiency and freshwater supply over a range of bed level increase from 0.0 m to 3.9 m. In general, results showed that the improvement towards the objective of port logistics always goes to the detriment of the freshwater supply objective while increasing bed level. Also, this study found that a collision of interests between the two types of end-users might worsen for a bed level increase over 2 m. Additional results showed that shallowing could be associated with benefits for freshwater supply through a decrease in the duration of water shortages due to a retreat in salt intrusion. The latter holds for specific environmental conditions of low river discharge and mild wind. Also, this study concluded that shallowing could negatively affect port efficiency due to heavier vessel traffic and more burdensome tidal window restrictions, which can result in an exponential growth of vessels average waiting times if the bed level increases over 2 m. Since the freshwater and port sub-systems were simplified, uncertainty in the results was unavoidably accepted. Despite these simplifications, this research demonstrated the main principles in implementing the comparison tool and lays the groundwork for more comprehensive models. In addition, several recommendations for policy-making are proposed, setting a basis for later discussions between freshwater supply and port-related stakeholders in the Rhine-Meuse Delta.

In the port of Ostend a discharge sluice is going to be constructed as part of an enforced dike ring. The proposed location of this discharge sluice is close to an already existing marina, the Royal Yachtclub Oostende (RYCO), and hindrance is expected regarding the outflow of the sluice in marina direction. When designing the sluice, an optimal balance has to be found between acceptable flow velocities in the downstream area and the capacity of the discharge sluice. It is therefore very important to be able to determine resulting flow patterns.

Formulas and rules of thumb found in literature are not sufficient to determine the resulting flow pattern of this system due to the complex geometry, including a pile row for flow velocity reduction. Other studies have shown that numerical models could simulate flow patterns of discharge sluices with much detail. However, a lot of detail in the results also requires much computational time. An example of a detailed numerical software program that is able to simulate the complete three-dimensional flow field is COMSOL Multiphysics 5.6 (COMSOL). Although it provides the most detail, simulating the flow in the entire area of interest (including the RYCO) for a complete tidal cycle in COMSOL would take too much computational time.

The objective of the present study is therefore to investigate the possibilities of determining the flow pattern downstream of a discharge sluice using a numerical method that requires less computational time but has sufficient accuracy to determine the potential impact of a discharge sluice on nautical activities.

In the present study, two options are considered to determine the flow field downstream of a discharge sluice. Method COMSOL-D3D is a coupled numerical method of a COMSOL and a coarser Delft3D-FLOW 4 (D3D) model. The other option, method D3D, uses only a D3D model and the sluice outflow is schematized by means of the general discharge relation.

As validation of the results is not possible due to a lack of measurement data, the methods are applied to a simplified case. The flow pattern resulting from each method is compared to the results obtained with a so-called baseline method. This method consists of modelling the entire domain with only a detailed numerical model, COMSOL. This is possible since, for validation purpose, the domain of the simplified case is relatively small and only stationary conditions are considered.

In conclusion, there is a lot of potential in the use of both methods in predicting the flow pattern downstream of a discharge sluice. They produce for the simplified case flow patterns similar to those obtained with the detailed method. Moreover, both methods require relatively little computational time compared to a full 3D simulation, method D3D requires the least amount. However, there are a number of conditions for the application of both methods.

The methods cannot be applied in the direct vicinity of the discharge sluice where the flow is highly three-dimensional. If one is interested in the flow in the first meters after the outflow opening or around the pile row, for example for designing the bottom protection, the two considered options are not sufficiently accurate. The flow in this area is too complex to simulate in a D3D model. In this case it is recommended to model the situation completely in COMSOL or a model similar to COMSOL. Furthermore, method D3D can only be applied if the sluice system is simple enough to correctly determine the discharge coefficient analytically/empirically and to simulate the effect of the pile row with a simplification in D3D. It is possible to accurately determine the effect of the pile row on the flow in this study with a schematized porous plate in D3D. Further research must show whether this applies to all types of pile rows.

For method COMSOL-D3D it is important that a correct coupling is made between both models. Here it is important to gradually impose the flow rates in the D3D model. Furthermore, the coupling should be made before the predicted point at which the jets starts deflecting towards the side but downstream of the area at which three-dimensional flows caused by the pile row are present.

It is important to note that due to a lack of validation data there is an uncertainty in the results following from the model approaches. Further research and the use of validation data must show how accurate the results of the considered methods are.

In this research method D3D is applied to the Ostend case. It becomes clear that flow rates exceed predetermined limits for safe operation in the marina. This applies to the entire marina and a large part of the time that the discharge sluice is discharging in marina direction. Measures will therefore have to be taken to prevent this. It is recommended to use method COMSOL-D3D to investigate the optimization between flow velocities in the marina and the discharge capacity. This is due to the fact that the design of the discharge sluice is expected to become much more complex and as a result the discharge coefficient is no longer easy to determine using formulas from literature.