YD
Y.M. Dwarkasing
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
2 records found
1
Load Hosting Capacity of Urban Traction Networks
Case study of the Calandlijn metro network in Rotterdam
Master thesis
(2026)
-
Y.M. Dwarkasing, P.T.M. Vaessen, Dennis van der Born, A. Lekić, I. Diab, J. Verschoor
Dutch urban distribution grids face growing congestion, with many areas unable to accommodate new connections until reinforcement is completed years from now. Light-rail traction power systems, dimensioned to power trams and metros, represent an underexplored asset on the same grid: their substations are sized for peak vehicle loads, leaving capacity that could in principle host additional external loads such as fast chargers or local industry. This thesis develops a method to quantify and allocate that hosting capacity. Train traffic makes the demand on each substation vary unpredictably from second to second; the network’s thermal limits depend not on instantaneous load but on how long an overload persists; and a load added at one substation shifts power flows across the entire line. A worst-case assessment returns conservatively small numbers; a single-snapshot average ignores the events that actually constrain operation. Treating the substations collectively, as a single aggregate capacity, hides the spatial trade-offs that determine where loads can actually be sited. The method addresses this in three stages. A simulation of the network generates a large pool of realistic operating scenarios. The pool is then reduced to a representative subset using a new clustering approach that preserves the scenarios most relevant to feasibility. The reduced set feeds a multi-objective optimisation, formulated with one capacity objective per substation, that exposes the spatial trade-offs between total hosting capacity and how evenly the load can be distributed. Probabilistic operating limits replace deterministic worst-case bounds where the underlying engineering standards admit them, so that brief admissible transients do not subordinate the result. Of the two metaheuristics evaluated, swarm optimisation handles the search space most effectively at moderate confidence levels, though its advantage narrows as the reliability requirement tightens. Applied to the Calandlijn of the Rotterdam metro, operated by Rotterdamse Elektrische Tram N.V., the method identifies a hosting capacity of 2 to 5 MW across the line’s twelve substations, with the achievable value depending on how evenly the load is distributed. A more important finding emerges from the analysis: the binding limit is not uniform across the network. On some feeders it is the contractual cap on utility intake rather than the traction infrastructure itself, while on others the smaller substations run close to their thermal limit. The distinction matters for the operator, since the two cases call for different remedies: contract renegotiation in one, physical reinforcement in the other.
...
Dutch urban distribution grids face growing congestion, with many areas unable to accommodate new connections until reinforcement is completed years from now. Light-rail traction power systems, dimensioned to power trams and metros, represent an underexplored asset on the same grid: their substations are sized for peak vehicle loads, leaving capacity that could in principle host additional external loads such as fast chargers or local industry. This thesis develops a method to quantify and allocate that hosting capacity. Train traffic makes the demand on each substation vary unpredictably from second to second; the network’s thermal limits depend not on instantaneous load but on how long an overload persists; and a load added at one substation shifts power flows across the entire line. A worst-case assessment returns conservatively small numbers; a single-snapshot average ignores the events that actually constrain operation. Treating the substations collectively, as a single aggregate capacity, hides the spatial trade-offs that determine where loads can actually be sited. The method addresses this in three stages. A simulation of the network generates a large pool of realistic operating scenarios. The pool is then reduced to a representative subset using a new clustering approach that preserves the scenarios most relevant to feasibility. The reduced set feeds a multi-objective optimisation, formulated with one capacity objective per substation, that exposes the spatial trade-offs between total hosting capacity and how evenly the load can be distributed. Probabilistic operating limits replace deterministic worst-case bounds where the underlying engineering standards admit them, so that brief admissible transients do not subordinate the result. Of the two metaheuristics evaluated, swarm optimisation handles the search space most effectively at moderate confidence levels, though its advantage narrows as the reliability requirement tightens. Applied to the Calandlijn of the Rotterdam metro, operated by Rotterdamse Elektrische Tram N.V., the method identifies a hosting capacity of 2 to 5 MW across the line’s twelve substations, with the achievable value depending on how evenly the load is distributed. A more important finding emerges from the analysis: the binding limit is not uniform across the network. On some feeders it is the contractual cap on utility intake rather than the traction infrastructure itself, while on others the smaller substations run close to their thermal limit. The distinction matters for the operator, since the two cases call for different remedies: contract renegotiation in one, physical reinforcement in the other.
This report describes the design and implementation of a sensor-based proximity sensing system implemented to assist the camera-based system. The goal of the sensor-based system is to provide a potential alternative to the camera-based system, and also to compensate for any of the downsides of this system, such as relatively worse performance in low visibility conditions, gauging distances accurately and implementing a collision warning mechanism. This will be done through the use of ultrasonic sensors. The sensor considerations and sensor
network will be described in depth. Furthermore, analysis of numerous signal processing techniques will be performed. These techniques will differentiate objects based on their respective echoes, which allows for accurate mapping onto the user interface, once the locations of the objects have been obtained through 3D trilateration. The prototype implementation along with the system integration with the camera-based system will be discussed once all of the previous work is done. Afterwards, the results of the project will be discussed and the work will be concluded with some future recommendations. ...
network will be described in depth. Furthermore, analysis of numerous signal processing techniques will be performed. These techniques will differentiate objects based on their respective echoes, which allows for accurate mapping onto the user interface, once the locations of the objects have been obtained through 3D trilateration. The prototype implementation along with the system integration with the camera-based system will be discussed once all of the previous work is done. Afterwards, the results of the project will be discussed and the work will be concluded with some future recommendations. ...
This report describes the design and implementation of a sensor-based proximity sensing system implemented to assist the camera-based system. The goal of the sensor-based system is to provide a potential alternative to the camera-based system, and also to compensate for any of the downsides of this system, such as relatively worse performance in low visibility conditions, gauging distances accurately and implementing a collision warning mechanism. This will be done through the use of ultrasonic sensors. The sensor considerations and sensor
network will be described in depth. Furthermore, analysis of numerous signal processing techniques will be performed. These techniques will differentiate objects based on their respective echoes, which allows for accurate mapping onto the user interface, once the locations of the objects have been obtained through 3D trilateration. The prototype implementation along with the system integration with the camera-based system will be discussed once all of the previous work is done. Afterwards, the results of the project will be discussed and the work will be concluded with some future recommendations.
network will be described in depth. Furthermore, analysis of numerous signal processing techniques will be performed. These techniques will differentiate objects based on their respective echoes, which allows for accurate mapping onto the user interface, once the locations of the objects have been obtained through 3D trilateration. The prototype implementation along with the system integration with the camera-based system will be discussed once all of the previous work is done. Afterwards, the results of the project will be discussed and the work will be concluded with some future recommendations.