Td
T.O. de Niet
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
Fleet sizing under operational uncertainty: A KLM case study at Schiphol
Electric Ground Support Equipment Operations at Airports
Airports are increasingly electrifying Ground Support Equipment (GSE) fleets to reduce local emissions and support decarbonisation targets. Although electric Ground Support Equipment (eGSE) are well suited to many airside operations, electrification changes the fleet-sizing problem. Vehicle availability is no longer determined only by task duration, location, and travel time, but also by battery state, charging duration, charger access, and operational charging rules. These constraints are especially relevant during aircraft turnaround operations, where multiple time-critical ground-handling tasks must be completed within narrow service windows.
This thesis investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of eGSE in airport turnaround processes. A structured review of the GSE and eGSE modelling literature shows that existing studies provide important building blocks for routing, scheduling, energy management, and charging analysis. However, the joint integration of task execution, individual vehicle availability, charging behaviour, infrastructure constraints, and operational uncertainty remains limited. In particular, many models either simplify charging and vehicle-level states, treat fleet size as a fixed input, or do not represent uncertainty. This motivates the development of a simulation-based decision-support approach in which eGSE fleet sizing is evaluated as a dynamic vehicle-availability problem.
A rule-based Discrete Event Simulation (DES) framework is developed to represent daily eGSE operations. The model includes individual service tasks, vehicle-specific states, airport location groups, travel times, battery state of charge, charging sessions, charger capacity, refilling, dumping, depot-return behaviour, and service-specific operating rules. Vehicles are assigned to tasks based on task urgency, time-window feasibility, travel time, battery state, and operational resource constraints. Task-timing uncertainty and travel-time uncertainty are included to evaluate how stochastic operational variability affects service-window performance and fleet-size requirements.
The framework is applied in a case study at KLM Ground Services (KLMGS) at Apron Services (AAS). The validated case-study scope includes three vehicle groups: water vehicles, toilet vehicles, and loaders. These groups represent both depot-based resource-constrained operations and stand-side service operations. The model is verified using synthetic test cases and validated through operational data checks, expert judgement, service-demand validation, energy and charging behaviour, water-demand consistency, and uncertainty-representation checks. Simulation experiments are then used to assess baseline performance, deterministic fleet sizing, task-timing and travel-time uncertainty, combined uncertainty, demand-case robustness, idle forward staging, feasibility-aware task selection, and sensitivity to operational and charging parameters.
The results show that required fleet capacity depends strongly on the service-level interpretation. Under deterministic operating conditions, the smallest fleet sizes that achieve 100% on-time task completion are 12 water vehicles, 11 toilet vehicles, and 23 loaders. Under combined task-timing and travel-time uncertainty, the strict robust thresholds increase to 16 water vehicles, 15 toilet vehicles, and 25 loaders. These are the smallest tested configurations that achieve 100% on-time task completion across all 30 stochastic replications. However, if a very small number of short internal service-window violations is operationally acceptable, lower fleet sizes may also be defensible. Under this pragmatic interpretation, at least 11 water vehicles, 9 toilet vehicles, and 23 loaders achieve ≥99.90% on-time task completion. These results should be interpreted carefully, because the performance metric measures completion within internal service windows and does not directly measure aircraft departure delay.
The experiments also show that operational-control assumptions can materially affect fleet-size outcomes. Idle forward staging, where idle vehicles remain near the aircraft stand instead of returning immediately to the depot, reduces unnecessary deadheading and improves service performance for selected vehicle groups. For toilet vehicles, the deterministic strict requirement decreases from 11 to 9 vehicles. Under combined uncertainty, idle forward staging reduces the strict robust threshold from 16 to 14 vehicles for water vehicles and from 15 to 10 vehicles for toilet vehicles. For loaders, the fleet-size threshold remains unchanged, but driven distance and depot-return movements are substantially reduced. A feasibility-aware task-selection variant further shows that dispatching logic can reduce the number of late tasks under scarce-fleet conditions, although it may increase the lateness severity of tasks that are already infeasible.
The sensitivity analysis indicates that charging is not the main driver of late task completion within the tested configurations. Energy-related parameters mainly affect charging sessions, charger occupancy, and charging-infrastructure utilisation. Task punctuality is more strongly constrained by vehicle availability during demand peaks, travel-time assumptions, service-time assumptions, and operational positioning logic. Charging therefore remains important for infrastructure planning and vehicle availability, but it is not the dominant bottleneck in the tested case-study settings.
Overall, the study shows that explicitly simulating operational requirements, uncertainty, and charging changes the interpretation of eGSE fleet sizing from a static vehicle-count problem into a dynamic availability problem. Required fleet capacity depends not only on the number of tasks, but also on when and where vehicles are needed, how uncertainty clusters demand, how quickly vehicles can recover between tasks, and how charging and supporting infrastructure affect vehicle availability. A simulation-based approach therefore provides a useful decision-support method for assessing eGSE fleet capacity, operational robustness, and charging-related resource use in airport ground handling. The reported fleet sizes should be interpreted as operational fleet-capacity requirements under the tested service-level assumptions. Final implementation decisions should add a technical reserve for maintenance, failures, battery degradation, charger unavailability, and other sources of vehicle downtime. ...
This thesis investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of eGSE in airport turnaround processes. A structured review of the GSE and eGSE modelling literature shows that existing studies provide important building blocks for routing, scheduling, energy management, and charging analysis. However, the joint integration of task execution, individual vehicle availability, charging behaviour, infrastructure constraints, and operational uncertainty remains limited. In particular, many models either simplify charging and vehicle-level states, treat fleet size as a fixed input, or do not represent uncertainty. This motivates the development of a simulation-based decision-support approach in which eGSE fleet sizing is evaluated as a dynamic vehicle-availability problem.
A rule-based Discrete Event Simulation (DES) framework is developed to represent daily eGSE operations. The model includes individual service tasks, vehicle-specific states, airport location groups, travel times, battery state of charge, charging sessions, charger capacity, refilling, dumping, depot-return behaviour, and service-specific operating rules. Vehicles are assigned to tasks based on task urgency, time-window feasibility, travel time, battery state, and operational resource constraints. Task-timing uncertainty and travel-time uncertainty are included to evaluate how stochastic operational variability affects service-window performance and fleet-size requirements.
The framework is applied in a case study at KLM Ground Services (KLMGS) at Apron Services (AAS). The validated case-study scope includes three vehicle groups: water vehicles, toilet vehicles, and loaders. These groups represent both depot-based resource-constrained operations and stand-side service operations. The model is verified using synthetic test cases and validated through operational data checks, expert judgement, service-demand validation, energy and charging behaviour, water-demand consistency, and uncertainty-representation checks. Simulation experiments are then used to assess baseline performance, deterministic fleet sizing, task-timing and travel-time uncertainty, combined uncertainty, demand-case robustness, idle forward staging, feasibility-aware task selection, and sensitivity to operational and charging parameters.
The results show that required fleet capacity depends strongly on the service-level interpretation. Under deterministic operating conditions, the smallest fleet sizes that achieve 100% on-time task completion are 12 water vehicles, 11 toilet vehicles, and 23 loaders. Under combined task-timing and travel-time uncertainty, the strict robust thresholds increase to 16 water vehicles, 15 toilet vehicles, and 25 loaders. These are the smallest tested configurations that achieve 100% on-time task completion across all 30 stochastic replications. However, if a very small number of short internal service-window violations is operationally acceptable, lower fleet sizes may also be defensible. Under this pragmatic interpretation, at least 11 water vehicles, 9 toilet vehicles, and 23 loaders achieve ≥99.90% on-time task completion. These results should be interpreted carefully, because the performance metric measures completion within internal service windows and does not directly measure aircraft departure delay.
The experiments also show that operational-control assumptions can materially affect fleet-size outcomes. Idle forward staging, where idle vehicles remain near the aircraft stand instead of returning immediately to the depot, reduces unnecessary deadheading and improves service performance for selected vehicle groups. For toilet vehicles, the deterministic strict requirement decreases from 11 to 9 vehicles. Under combined uncertainty, idle forward staging reduces the strict robust threshold from 16 to 14 vehicles for water vehicles and from 15 to 10 vehicles for toilet vehicles. For loaders, the fleet-size threshold remains unchanged, but driven distance and depot-return movements are substantially reduced. A feasibility-aware task-selection variant further shows that dispatching logic can reduce the number of late tasks under scarce-fleet conditions, although it may increase the lateness severity of tasks that are already infeasible.
The sensitivity analysis indicates that charging is not the main driver of late task completion within the tested configurations. Energy-related parameters mainly affect charging sessions, charger occupancy, and charging-infrastructure utilisation. Task punctuality is more strongly constrained by vehicle availability during demand peaks, travel-time assumptions, service-time assumptions, and operational positioning logic. Charging therefore remains important for infrastructure planning and vehicle availability, but it is not the dominant bottleneck in the tested case-study settings.
Overall, the study shows that explicitly simulating operational requirements, uncertainty, and charging changes the interpretation of eGSE fleet sizing from a static vehicle-count problem into a dynamic availability problem. Required fleet capacity depends not only on the number of tasks, but also on when and where vehicles are needed, how uncertainty clusters demand, how quickly vehicles can recover between tasks, and how charging and supporting infrastructure affect vehicle availability. A simulation-based approach therefore provides a useful decision-support method for assessing eGSE fleet capacity, operational robustness, and charging-related resource use in airport ground handling. The reported fleet sizes should be interpreted as operational fleet-capacity requirements under the tested service-level assumptions. Final implementation decisions should add a technical reserve for maintenance, failures, battery degradation, charger unavailability, and other sources of vehicle downtime. ...
Airports are increasingly electrifying Ground Support Equipment (GSE) fleets to reduce local emissions and support decarbonisation targets. Although electric Ground Support Equipment (eGSE) are well suited to many airside operations, electrification changes the fleet-sizing problem. Vehicle availability is no longer determined only by task duration, location, and travel time, but also by battery state, charging duration, charger access, and operational charging rules. These constraints are especially relevant during aircraft turnaround operations, where multiple time-critical ground-handling tasks must be completed within narrow service windows.
This thesis investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of eGSE in airport turnaround processes. A structured review of the GSE and eGSE modelling literature shows that existing studies provide important building blocks for routing, scheduling, energy management, and charging analysis. However, the joint integration of task execution, individual vehicle availability, charging behaviour, infrastructure constraints, and operational uncertainty remains limited. In particular, many models either simplify charging and vehicle-level states, treat fleet size as a fixed input, or do not represent uncertainty. This motivates the development of a simulation-based decision-support approach in which eGSE fleet sizing is evaluated as a dynamic vehicle-availability problem.
A rule-based Discrete Event Simulation (DES) framework is developed to represent daily eGSE operations. The model includes individual service tasks, vehicle-specific states, airport location groups, travel times, battery state of charge, charging sessions, charger capacity, refilling, dumping, depot-return behaviour, and service-specific operating rules. Vehicles are assigned to tasks based on task urgency, time-window feasibility, travel time, battery state, and operational resource constraints. Task-timing uncertainty and travel-time uncertainty are included to evaluate how stochastic operational variability affects service-window performance and fleet-size requirements.
The framework is applied in a case study at KLM Ground Services (KLMGS) at Apron Services (AAS). The validated case-study scope includes three vehicle groups: water vehicles, toilet vehicles, and loaders. These groups represent both depot-based resource-constrained operations and stand-side service operations. The model is verified using synthetic test cases and validated through operational data checks, expert judgement, service-demand validation, energy and charging behaviour, water-demand consistency, and uncertainty-representation checks. Simulation experiments are then used to assess baseline performance, deterministic fleet sizing, task-timing and travel-time uncertainty, combined uncertainty, demand-case robustness, idle forward staging, feasibility-aware task selection, and sensitivity to operational and charging parameters.
The results show that required fleet capacity depends strongly on the service-level interpretation. Under deterministic operating conditions, the smallest fleet sizes that achieve 100% on-time task completion are 12 water vehicles, 11 toilet vehicles, and 23 loaders. Under combined task-timing and travel-time uncertainty, the strict robust thresholds increase to 16 water vehicles, 15 toilet vehicles, and 25 loaders. These are the smallest tested configurations that achieve 100% on-time task completion across all 30 stochastic replications. However, if a very small number of short internal service-window violations is operationally acceptable, lower fleet sizes may also be defensible. Under this pragmatic interpretation, at least 11 water vehicles, 9 toilet vehicles, and 23 loaders achieve ≥99.90% on-time task completion. These results should be interpreted carefully, because the performance metric measures completion within internal service windows and does not directly measure aircraft departure delay.
The experiments also show that operational-control assumptions can materially affect fleet-size outcomes. Idle forward staging, where idle vehicles remain near the aircraft stand instead of returning immediately to the depot, reduces unnecessary deadheading and improves service performance for selected vehicle groups. For toilet vehicles, the deterministic strict requirement decreases from 11 to 9 vehicles. Under combined uncertainty, idle forward staging reduces the strict robust threshold from 16 to 14 vehicles for water vehicles and from 15 to 10 vehicles for toilet vehicles. For loaders, the fleet-size threshold remains unchanged, but driven distance and depot-return movements are substantially reduced. A feasibility-aware task-selection variant further shows that dispatching logic can reduce the number of late tasks under scarce-fleet conditions, although it may increase the lateness severity of tasks that are already infeasible.
The sensitivity analysis indicates that charging is not the main driver of late task completion within the tested configurations. Energy-related parameters mainly affect charging sessions, charger occupancy, and charging-infrastructure utilisation. Task punctuality is more strongly constrained by vehicle availability during demand peaks, travel-time assumptions, service-time assumptions, and operational positioning logic. Charging therefore remains important for infrastructure planning and vehicle availability, but it is not the dominant bottleneck in the tested case-study settings.
Overall, the study shows that explicitly simulating operational requirements, uncertainty, and charging changes the interpretation of eGSE fleet sizing from a static vehicle-count problem into a dynamic availability problem. Required fleet capacity depends not only on the number of tasks, but also on when and where vehicles are needed, how uncertainty clusters demand, how quickly vehicles can recover between tasks, and how charging and supporting infrastructure affect vehicle availability. A simulation-based approach therefore provides a useful decision-support method for assessing eGSE fleet capacity, operational robustness, and charging-related resource use in airport ground handling. The reported fleet sizes should be interpreted as operational fleet-capacity requirements under the tested service-level assumptions. Final implementation decisions should add a technical reserve for maintenance, failures, battery degradation, charger unavailability, and other sources of vehicle downtime.
This thesis investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of eGSE in airport turnaround processes. A structured review of the GSE and eGSE modelling literature shows that existing studies provide important building blocks for routing, scheduling, energy management, and charging analysis. However, the joint integration of task execution, individual vehicle availability, charging behaviour, infrastructure constraints, and operational uncertainty remains limited. In particular, many models either simplify charging and vehicle-level states, treat fleet size as a fixed input, or do not represent uncertainty. This motivates the development of a simulation-based decision-support approach in which eGSE fleet sizing is evaluated as a dynamic vehicle-availability problem.
A rule-based Discrete Event Simulation (DES) framework is developed to represent daily eGSE operations. The model includes individual service tasks, vehicle-specific states, airport location groups, travel times, battery state of charge, charging sessions, charger capacity, refilling, dumping, depot-return behaviour, and service-specific operating rules. Vehicles are assigned to tasks based on task urgency, time-window feasibility, travel time, battery state, and operational resource constraints. Task-timing uncertainty and travel-time uncertainty are included to evaluate how stochastic operational variability affects service-window performance and fleet-size requirements.
The framework is applied in a case study at KLM Ground Services (KLMGS) at Apron Services (AAS). The validated case-study scope includes three vehicle groups: water vehicles, toilet vehicles, and loaders. These groups represent both depot-based resource-constrained operations and stand-side service operations. The model is verified using synthetic test cases and validated through operational data checks, expert judgement, service-demand validation, energy and charging behaviour, water-demand consistency, and uncertainty-representation checks. Simulation experiments are then used to assess baseline performance, deterministic fleet sizing, task-timing and travel-time uncertainty, combined uncertainty, demand-case robustness, idle forward staging, feasibility-aware task selection, and sensitivity to operational and charging parameters.
The results show that required fleet capacity depends strongly on the service-level interpretation. Under deterministic operating conditions, the smallest fleet sizes that achieve 100% on-time task completion are 12 water vehicles, 11 toilet vehicles, and 23 loaders. Under combined task-timing and travel-time uncertainty, the strict robust thresholds increase to 16 water vehicles, 15 toilet vehicles, and 25 loaders. These are the smallest tested configurations that achieve 100% on-time task completion across all 30 stochastic replications. However, if a very small number of short internal service-window violations is operationally acceptable, lower fleet sizes may also be defensible. Under this pragmatic interpretation, at least 11 water vehicles, 9 toilet vehicles, and 23 loaders achieve ≥99.90% on-time task completion. These results should be interpreted carefully, because the performance metric measures completion within internal service windows and does not directly measure aircraft departure delay.
The experiments also show that operational-control assumptions can materially affect fleet-size outcomes. Idle forward staging, where idle vehicles remain near the aircraft stand instead of returning immediately to the depot, reduces unnecessary deadheading and improves service performance for selected vehicle groups. For toilet vehicles, the deterministic strict requirement decreases from 11 to 9 vehicles. Under combined uncertainty, idle forward staging reduces the strict robust threshold from 16 to 14 vehicles for water vehicles and from 15 to 10 vehicles for toilet vehicles. For loaders, the fleet-size threshold remains unchanged, but driven distance and depot-return movements are substantially reduced. A feasibility-aware task-selection variant further shows that dispatching logic can reduce the number of late tasks under scarce-fleet conditions, although it may increase the lateness severity of tasks that are already infeasible.
The sensitivity analysis indicates that charging is not the main driver of late task completion within the tested configurations. Energy-related parameters mainly affect charging sessions, charger occupancy, and charging-infrastructure utilisation. Task punctuality is more strongly constrained by vehicle availability during demand peaks, travel-time assumptions, service-time assumptions, and operational positioning logic. Charging therefore remains important for infrastructure planning and vehicle availability, but it is not the dominant bottleneck in the tested case-study settings.
Overall, the study shows that explicitly simulating operational requirements, uncertainty, and charging changes the interpretation of eGSE fleet sizing from a static vehicle-count problem into a dynamic availability problem. Required fleet capacity depends not only on the number of tasks, but also on when and where vehicles are needed, how uncertainty clusters demand, how quickly vehicles can recover between tasks, and how charging and supporting infrastructure affect vehicle availability. A simulation-based approach therefore provides a useful decision-support method for assessing eGSE fleet capacity, operational robustness, and charging-related resource use in airport ground handling. The reported fleet sizes should be interpreted as operational fleet-capacity requirements under the tested service-level assumptions. Final implementation decisions should add a technical reserve for maintenance, failures, battery degradation, charger unavailability, and other sources of vehicle downtime.
Electric Ground Support Equipment Operations at Airports
Fleetsizing under operational uncertainty: A KLM case study at Schiphol
Airports are electrifying Ground Support Equipment (GSE) fleets to reduce local emissions, but electric GSE changes both fleet-capacity and operational-demand planning. This paper investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of electric Ground Support Equipment (eGSE) during airport turnaround operations. A rule-based discrete-event simulation represents individual vehicles, service time windows, airport travel, battery state, finite charging infrastructure, and service-specific resources. The focused case study considers two operationally distinct vehicle types at KLM Ground Services at Amsterdam Airport Schiphol: depot-based toilet vehicles and stand-side loaders. The model is extensively verified and validated for fitness for purpose using flight and task data, airport distances, track-and-trace movement data, measured energy consumption, state-of-charge and charging observations, waste and dumping logic, historical arrival deviations, and expert review. Water vehicles were also implemented and validated in the underlying study but are omitted from the detailed paper results to avoid repetition. Under combined task-timing and travel-time uncertainty, the strict fleet requirements are 15 toilet vehicles and 25 loaders. Pragmatic lower bounds of 9 toilet vehicles and 23 loaders achieve 99.90% and 99.93% mean on-time completion, respectively, but retain residual service-window risk. Operational-control experiments show that idle forward staging reduces the strict stochastic toilet requirement from 15 to 10 vehicles and substantially reduces deadheading. Feasibility-aware task triage reduces the number of late tasks under scarcity, but increases the lateness of tasks that are deprioritised. Charging is not the binding cause of lateness in the tested reference configurations, although it materially affects charging sessions, charger occupancy, peak use, and infrastructure demand. The scientific contribution is therefore a decision-support method that estimates fleet capacity and diagnoses how demand timing, positioning, dispatching, uncertainty, and charging resources shape operational performance.
...
Airports are electrifying Ground Support Equipment (GSE) fleets to reduce local emissions, but electric GSE changes both fleet-capacity and operational-demand planning. This paper investigates how operational requirements, uncertainty, and charging affect the required fleet capacity and operational demand of electric Ground Support Equipment (eGSE) during airport turnaround operations. A rule-based discrete-event simulation represents individual vehicles, service time windows, airport travel, battery state, finite charging infrastructure, and service-specific resources. The focused case study considers two operationally distinct vehicle types at KLM Ground Services at Amsterdam Airport Schiphol: depot-based toilet vehicles and stand-side loaders. The model is extensively verified and validated for fitness for purpose using flight and task data, airport distances, track-and-trace movement data, measured energy consumption, state-of-charge and charging observations, waste and dumping logic, historical arrival deviations, and expert review. Water vehicles were also implemented and validated in the underlying study but are omitted from the detailed paper results to avoid repetition. Under combined task-timing and travel-time uncertainty, the strict fleet requirements are 15 toilet vehicles and 25 loaders. Pragmatic lower bounds of 9 toilet vehicles and 23 loaders achieve 99.90% and 99.93% mean on-time completion, respectively, but retain residual service-window risk. Operational-control experiments show that idle forward staging reduces the strict stochastic toilet requirement from 15 to 10 vehicles and substantially reduces deadheading. Feasibility-aware task triage reduces the number of late tasks under scarcity, but increases the lateness of tasks that are deprioritised. Charging is not the binding cause of lateness in the tested reference configurations, although it materially affects charging sessions, charger occupancy, peak use, and infrastructure demand. The scientific contribution is therefore a decision-support method that estimates fleet capacity and diagnoses how demand timing, positioning, dispatching, uncertainty, and charging resources shape operational performance.