This thesis presents a COLREG-compliant collision avoidance and detection algorithm for the safe navigation of autonomous surface vessels. The method builds upon the Velocity Obstacle algorithm and implements the concept of the ship domain and the ship arena into the framework. The International Regulations for Preventing Collision at Sea – the COLREGs – are analyzed individually and incorporated in the designed collision avoidance algorithm. The COLREGs were used to determine the preferred passing side and optimized motion for every possible encounter situation with an obstacle vessel. The ship domain and arena help to perform proactive alterations when required, improving on the apparent shortcomings of the VO algorithm. This framework leads to a more proactive collision avoidance algorithm that guarantees COLREG compliance and safe navigation in mixed-traffic environments. The robustness and wide adaptability of the developed algorithm are validated in different encounter situations, ranging from vessel-to-vessel batch simulations to complex, multi-vessel scenarios.

During single-blade offshore wind turbine (OWT) installation, wind disturbance results in blade root motion. The sensitivity to this wind disturbance is more significant for larger blades, reducing the allowable installation weather limit. A potential solution is a Hook Mounted Compensator (HMC) between the crane and the load that can provide high-precision compensation across multiple degree of freedom (DOF), and compensate for the influence of wind on the OWT blade. Several Actuation Method (AM)s have been identified that can be used in a HMC. However, these AMs have not been modelled dynamically and their effectiveness for the desired application is unknown. The aim of this study is to analyse and assess the compensation effectiveness of these AMs in a HMC. For this purpose, the AMs are simulated and analysed in various levels of complexity and DOFs. Initially, the system is considered in a 1DOF. A linear representation is formulated, based on which a Proportional-Integral-Derivative (PID) controller for each AM is formulated. This enables simulation of the Single-Blade Installation System (SBIS) in the time domain. Assessment criteria are used to quantify the compensation effectiveness and actuation input of the AMs. To simulate the SBIS for each AM, 3DOF numerical models are developed that describe the operation of the AMs in their respective operational planes. Six actuation methods are included in the assessment. Based on the assessment results, three AMs are combined in a HMC concept. A XY table is used to control blade root position in the x- and y-directions, gyroscopes for the z-direction, and COG shifting with counterweight to prevent gyroscope saturation. The PID controllers based on the linear 1DOF representation are used, operating in parallel, to control the blade root position in all DOF. The blade root motion is reduced by 96% along the blade axis and 86% radially. The blade root motion is sensitive to the mean wind speed, turbulence intensity, and angle of the incoming wind. The HMC shows robustness to variations in tugger line angle and pretension. The static blade pitch angle affects the magnitude and distribution of the wind-induced moment on the blade. For a pitch angle of -180, the moment around the z-axis is minimal, providing optimal HMC performance. The HMC concept can not directly actuate the blade around the z-axis. Instead, the XY table translates the blade at the COG, resulting in increased cylinder stroke and high power consumption for other pitch angles. Two AMs were assessed to actuate the blade around the z-axis, both showing poor performance. Further exploration of AMS that can directly control the blade rotation around the z-axis could lead to improvement of the HMC concept. Additional improvement can be made in optimization of counterweight and gyroscopes sizes, potentially reducing weight and improving system performance. Further suggestions for future work include; exploring the compensating for the now neglected hub motion using the HMC, examining the stability of the HMC once the blade is mated to the hub, and the impact of sensor noise and delay on the HMC’s performance.

Inland waterway transport is a low CO2 emission alternative to road transport. A shift towards more inland waterway transport could also help reduce road congestion and noise pollution. Infrastructure bottlenecking, particularly at locks, is part of the reasons preventing this shift. Congestion is leading to delays. Locks can be physically improved, or the passage of the vessels through the locks can be optimized through scheduling. Recent work introduced a novel switching-max-plus-linear system approach to scheduling vessels passing through networks of waterways and locks, also introducing a novel routing component to the scheduling problem. Switching max-plus-linear systems are a convenient way to model scheduling systems using max-plus-linear algebra. The switching-max-plus-linear model only considered locks with a single chamber that can only process one vessel at a time. Additionally, it only considered four specific waterway network configurations, rather than any arbitrary network configuration. Real locks can have multiple chambers, and they can process multiple vessels at the same time if they are placed according to regulations in the two-dimensional space of the chamber. The scope of this report was then to build upon this switching max-plus-linear model by adding support for arbitrary network configurations, and multi-chamber, multi-vessel locks with proper twodimensional ship placement, to answer the main research question: How can multi-vessel, multichamber locks with ship placement be integrated into the SMPL IWT scheduling model? Three mathematical scheduling models formulated as SMPL systems were introduced, each subsequent model building on the previous one. The first introduced support for arbitrary network configurations. The second introduced support for multi-chamber locks and allowed vessels to pass through lock chambers at the same time, provided that their assigned one-dimensional sizes fit into the assigned one-dimensional capacity of the chamber. The third introduced support for twodimensional ship placement through a Tetris-like placement sequence also modeled as a switching max-plus-linear system. The models were translated to mixed integer linear programming models, and arrival time and arrival time offset objectives were added, so they could be used as the rules by which a scheduler would build and solve offline scheduling optimization models on a case-by-case basis. Auxiliary objectives to promote vessels slowing down, rather than waiting stationary in the waiting areas, were also added. The models were all found to be working as intended and implemented correctly through the use of a number of verification cases and tests. Complexity tests showed that the solution times for all three models already became larger than a practical limit of 15-20 minutes for simple scenarios with 10-15 vessels. A heuristic model that mimics how vessels are assigned to lockages in practice was built. Comparisons to the scheduling models on the verification cases showed negligible differences for the models in single-lock cases, but it indicated that the multi-vessel, multi-chamber models may outperform practice in multi-lock cases. Future research is recommended to focus on online optimization to account for disturbances, distributed optimization to reduce calculation times, and validation of the model’s performance with real data.

Traditionally, the control point (CP) of a dynamically positioned vessel is located around its center of gravity (CoG). However, during offshore operations, other locations, such as the crane tip or gripper position, become more critical due to the load being located there. This thesis proposes shifting the control point from the CoG to an alternative location, specifically the gripper position, to minimize the DP footprint at this location. Minimizing the DP footprint at the gripper location could lead to smaller motion deviations at this critical point. Consequently, this could potentially improve the workability of the vessel, which in turn could lead to higher operational yields. Additionally, the risk of potential damage and unsafe situations offshore is mitigated.

The conventional DP system contains a Kalman Filter, to filter out the first-order motions, a P(I)D controller that calculates the demanded forces to keep the vessel in place, and a thruster allocation algorithm that distributes the demanded forces to all the thrusters in an optimized way.

A new design for a DP system with the control point at the gripper position for the Deepwater Construction Vessel (DCV) Aegir is presented and evaluated. Time domain simulations were performed with the Aegir containing its conventional DP system and with the Aegir containing the newly designed DP system. These time domain simulations were performed using Orcaflex, which contains a model of the Aegir. This model of the vessel is connected to an external Python code, that contains the DP system, including a thruster allocation.


As the new DP system at the gripper location has to cope with coupled equations of motion, it is equipped with a new Multiple-Input-Multiple-Output (MIMO) PD controller, that consists of a decoupling module and separate PD controllers for each DOF. To obtain state estimates for the second-order motions at the gripper location, the state estimates calculated by the Kalman Filter are translated to the gripper location.

The effects on DP performance were assessed by evaluating and comparing the motion responses in the horizontal plane and DP footprint at the gripper location of both models. Also, the thruster behavior and energy consumption of both models are compared. This was done for an incoming wave direction of 135 degrees, a peak period (Tp) of 8 seconds, and significant wave heights (Hs) from 1.0 m to 2.0 m with increments of 0.5 m. The wave spectrum used is a JONSWAP spectrum. As recommended by the classification society, three-hour simulations are performed for the so-called 'base case' sea state with Hs = 1.5m. However, due to the extensive simulation time only one-hour simulations were performed for the cases with Hs = 1.0m and 2.0m.

When comparing the motion responses, one of the most remarkable results is an improved yaw response for the gripper control point model in all sea states that are considered in this study. Further on, the motion responses for sway appeared to be bigger for the gripper control point compared to the center control point in Hs = 1.5m and Hs = 1.0 m. However, the differences in sway are observed to be marginal for Hs = 2.0m.

The results show that the DP footprint has slightly improved in the x-direction for the gripper control point model compared to the center control point model for the base case. The same observation is done for Hs = 2.0m, but the differences found between the models in Hs = 1.0m are marginal. Also, the DP footprint was observed to be slightly larger in the y-direction for Hs = 1.0m and Hs = 1.5m for the gripper control point.

The total thrust outputs as delivered by the DP system during the simulations were converted to power, by using the propeller diagrams (for DP-speed state) for the specific types of thrusters the Aegir is equipped with. From these results, it became clear that the gripper point control model consumes less energy in all tested sea states compared center control point model.

From the results presented in this study, it is concluded that the system itself has potential, but no hard conclusions can be drawn for the system in its current form. Problems were observed with the current thruster allocation algorithm from HES, which need to be explored in more detail and resolved. It is recommended to look into developing a stable working thruster allocation algorithm for both control point models, such that more accurate dynamic simulations can be performed and the system can be assessed under more sea states.

Planning rolling stock maintenance based on prognostic data (Condition Based Maintenance, CBM) is a trend since the ideal timing for maintenance before failure can be planned. With the use of prognostics, a failure can be predicted so Corrective Maintenance (CM) can be avoided. Traditional rolling stock decision-making for Preventive Maintenance (PM) is based on the time and/or mileage since last PM routine at the maintenance depot. A sufficient amount of literature is available that considers rolling stock PM planning optimization methods. Planning rolling stock maintenance is constrained by the required availability for passenger operations, the conditions for PM and the maintenance depot capacity. However, the integration of CBM with the rolling stock PM planning has not been researched. CM also needs to be performed and is considered a disruptive element in the maintenance planning. It is expected that integrating CBM in the rolling stock PM planning is less disruptive than CM. This study investigates how CBM impacts the rolling stock PM planning by formulating a deterministic MILP that uses a rolling horizon framework that minimizes the maintenance costs. An approach that optimizes the rolling stock PM planning while being disrupted by unexpected failures that lead to CM is compared with an approach that integrates CBM that is disrupted by predicted failures that can be planned in advance. The outcome of the model demonstrates that CBM is less disruptive to the maintenance planning than CM, because the time to failure gives the model flexibility to find the ideal moment to perform CBM. Conclusions and recommendations of this study can be used for implementing CBM approaches for rolling stock.

The Dutch railway network is a dense network with 7000 kilometers of railway tracks spread over only 42 000 square kilometers [1, 2]. Additionally, the railways are responsible for 1.4 million passengers every working day [1]. These factors make both the infrastructure and timetable of the Dutch railway network extremely intricate. On top of that, the amount of train trips is expected to grow with 40 % by 2040 [3] and with that the complexity of managing the railway network will increase even more. In the case of conflict in the timetable, action must be taken to resolve this conflict in order for the train operations to continue. In current Dutch practice this task resides with the train dispatchers. However it can be difficult for the dispatcher to oversee the situation when multiple trains get involved and taken actions will have consequences on yet another selection of trains. This increasingly complex task has resulted in a grown interest in so-called Traffic Management Systems (TMS), which are an intelligent systems that use a conflict resolution algorithm to find a solution to a conflict. A TMS can be used to support the train dispatcher...

Ship-to-Ship (STS) cargo transfers can significantly improve the efficiency of offshore installations while reducing their associated costs. However, the added complexity of cargo transfers involving ship-mounted cranes at sea poses significant challenges. To address this, crane manufacturers have introduced Relative Heave Compensation (RHC) systems to assist crane operators. Nevertheless, concerns have emerged among installation vessel companies regarding the relative heave measurement systems, which rely on Motion Reference Unit (MRU) sensors placed on both ships. Given that many supplier vessels lack the required sensors and the communication protocols pose bureaucratic hurdles, this research focuses on proposing novel, feeder-independent relative heave measuring solutions. This work introduces two unexplored sensor units, namely the LiDAR-MRU and RADAR-MRU. In the absence of actual sensor data, a simulation workflow using Simulink and Unreal Engine (UE) is presented to generate synthetic data. After, four distinct measurement solutions are developed, implemented, and evaluated using the simulated data in MATLAB. The implemented solutions estimate relative heave distance and speed with a Mean Absolute Error (MAE) ranging from 0.3-5.2% and 0.8-4.3% of the reference's maximum amplitude, and processing times from 1.8-91.4 [ms]. The results underscore their potential for field implementation. However, future validation is needed with actual sensor data.

Marine vessels execute many missions during their life cycle, each associated with a different required power profile. The required power is to be provided by the power plant, which has a fixed set of equipment such as diesel engines, generator sets, electric machines, and batteries. Typically, the control of vessel's power plants consist of two levels; a primary level with local controllers for the power plant components, and a secondary level that determines the distribution of the required power to the various components in the power plant. The state-of-the-art only considers a multi-level power plant control architecture assuming a fixed power plant layout, but in practice, the fixed power plant layout may not suffice for generating the power demand dictated by a new mission. This may lead to inefficient use of components, risk of overloading components, or the inability to deliver the required power. To handle this issue, equipment modifications such as additions, removals or replacements would probably be necessary, along with modifications in the multi-level control scheme. To enable the seamless operation of the multi-level plant control system after the modifications is essential for guaranteeing safety and reducing the downtime, which could be achieved by a control architecture that allows for modular use of the power plant components.

This thesis presents the design methodology of a mission-oriented modular control system for marine power plants. To this end, first power profiles, power plant layouts and control systems of multiple vessels such as tugboats, offshore support vessels, cargo ships and cruise ships are analyzed. By decomposing the power profile in two components, the propulsion and auxiliary power demand, the correlation between the power profile of a vessel and its mission is derived, and an algorithm that computes the power profile using mission and vessel data is proposed. Furthermore, the correlation between the power profile and the layout of the power plant is also investigated, with emphasis on how changes in the power profile result in power plant automation modifications. A modular secondary control level is then designed to cope with the required power plant automation modifications, by combining the Equivalent Consumption Minimization Strategy (ECMS) with Supervisory Switching Control (SSC). In this thesis we consider battery modifications, following the example of Wärtsilä's ZESPacks. Simulation results are used to show the performance of the proposed switching control methodology, in relation to the stability of the components in the power plant after automation modifications occur.

The main contribution of this thesis is the novel approach for the secondary level power plant control system, introducing modularity to the otherwise assumed fixed layout of the power plant. Furthermore, the proposed algorithm can be used to determine the expected power profile for a new mission, to identify required modifications of the power plant equipment.

Railway networks are to play an increasingly large role in European transportation. This has boosted the urgency of railway innovations, of which the development of decision support systems for conflict resolution is an important aspect. This research contributes to this development by formulating a suitable mathematical approach for railway networks equipped with moving block signalling systems. Two dispatching actions to reschedule trains are applied, namely retiming and reordering. The designed approach is an extension to an existing method, based on graph theory, that is able to reschedule trains in case of conflict. The novel method uses additional node- and arc types in order to ensure moving block suitability. The new node type enables the possibility to create nodes that are related to trains, rather than infrastructure. The new arc type ensures a continuously safe time interval between two trains in the absence of trackside signals. An optimization problem, with the objective of minimizing the maximum propagated delay, is formulated. Hereafter, the performance is evaluated by a case study in the Rotterdam-The Hague corridor. According to the experimental results, the designed model is able to reduce delay propagation up to 50% for the majority of input situations within 10 seconds of computation time. Overall, the designed method shows promising results, but further research will be necessary to make it applicable in practice.

Inland waterways form a natural network infrastructure with the capacity for waterborne transport of people and goods for moving freight from seaports to the hinterland. Recently, Inland Waterway Transport (IWT) has been promoted more extensively by the European Union and various governments as it plays a crucial role in reducing road congestion and CO2 emissions from transport. However, the advantages of IWT are not fully exploited due to inefficiencies in the logistics system, such as long waiting times at locks and sub-optimal navigation on waterways. Currently, no scheduling at infrastructures or routing optimisation of the overall waterway network is happening. The scheduling of vessels through a lock is usually performed on a First In First Out basis, providing an opportunity for improvement. Hence, this thesis aims to design a scheduling strategy for generating an optimal plan for sending inland vessels through a waterway network with minimal delays, yielding a significant positive impact on the modal shift towards IWT. 
A promising approach to scheduling problems is by using Switching Max-Plus-Linear (SMPL) systems. SMPL systems have proven to be effective in various Discrete-Event Systems and transportation networks. Using SMPL models is convenient since non-linear scheduling problems can be described linearly using Max-Plus operators without compromising on the system dynamics. Moreover, as the SMPL systems can be transformed into Mixed-Integer-Linear-Programming (MILP) problems, it is also possible to use fast optimisers for solving the scheduling problems.
This thesis will show how one can describe IWT systems, consisting of; waterways, vessels and locks, as SMPL systems. The optimal schedule for the inland vessels is determined based on multiple input parameters, including waterway network lay-out, the sailing speeds of vessels and arrival deadlines of the vessels. The scheduler will return the individual vessel routing and overall vessel order in the waterway network. This routing and order selection is defined using binary control variables, turning the IWT scheduling problem into a MILP problem, which will allow finding the solution to large scale IWT scheduling problems in a reasonable computation time. Furthermore, this thesis will show how the goal of minimising the cumulative arrival times of all vessels in a network can be achieved. This is done for different types of waterway network cases, for which the results are shown and analysed.

This research work tries to provide a model-based, distributed Fault Diagnosis (FD) framework that will eventually act as an early warning system for online monitoring of the marine propulsion process, in order to avoid future failures and accidents. Furthermore, the introduction of adaptive thresholds in the monitoring modules of the monitoring agents helps minimize false alarms and reduces the conservativeness in decision-making. Finally, this work comes to provide an integrated sensor and process FD framework that can be adopted on-board marine vessels in the future, and narrow down the gap, regarding FD for marine vessels which is closely linked with maritime safety and maintenance decision-making.

The inland waterway once enabled an industrial revolution, yet the emergence of coalescent road networks has seen its true worth be all but disregarded. Despite the amenity of unimodal travel being compelling, growing awareness for sustainability has reignited interest in more fuel efficient modalities for transportation. Autonomous shipping has the potential to increase efficiency, reliability and safety and will arguably play a major role in the evolving transport revolution, returning the transport modality to its former glory.

The main objective of this research is to cater a collision avoidance strategy to the inland waterway through the development of tailored Guidance and Navigation Systems. An approach to local path planning is introduced to handle the challenges of collision avoidance on the inland waterway and a better understanding of the primary role that stereovision sensors could assume in enabling inland autonomy is gained. Achieving this objective requires first a reflection upon existing work through a study into the state-of-the-art. Subsequently, the Guidance and Navigation Systems are developed and implemented on a scale test vessel. Finally experimental testing is conducted for the evaluation of the developed system performance.

This thesis presents the design of a performance prediction simulation model for an ship-to-shore crane. Current STS crane models do not take new trends in container terminal operations into account like automation. These models do not predict the performance accurately. This can lead to logistical and financial consequences. Automation systems that are present in the cranes should be taken into account when modelling. State-of-the-art models do not model the cycle path of the crane accurately, crane components are modeled in sequence while in real life the movements between the crane components are parallel. Moreover what is not included in these models are the accelerations/decelerations of these different components of the STS cranes. A new STS crane model is developed that consists of process blocks and all the lacking points. Data was used from a reference crane to verify and validate the new model. To evaluate the performance of the proposed prediction model, a comparison has been made with existing prediction models.

Formation control of autonomous surface vessels (ASVs) is researched extensively in the last years, since it has promising applications. However very few designed strategies have been validated during real experiments. In this research, two
control methods for distributed leader-follower formation control are proposed: A Nonlinear Model Predictive Control (NMPC) method and an MPC method using Feedback Linearization (FL). These two designed methods are compared with a conventional Proportional-Integral (PI) control method. The performance of the proposed strategies is evaluated through simulations and real experiments using small-scale Roboat units. Both designed control methods outperform the conventional PI control method in simulation and real experiments.