F. Schulte
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>
29 records found
1
Energy-Aware Steel Slab Logistics
Simulation-Based Evaluation of Thermal Buffer Allocation
Reheating slabs before hot rolling accounts for roughly 80% of the energy consumed by the hot-rolling process, so preserving slab heat during yard logistics is a direct energy lever. At the integrated steel plant used as a case study, insulated thermal buffers (hot boxes) can substantially slow the cooling of slabs. These hot boxes are currently reserved for fracture-sensitive steel grades and, on average, operate at about half capacity. On paper, this presents an opportunity for other slabs to use the spare buffer capacity. The question of whether managing that space with a more sophisticated allocation strategy is unanswered.
The literature provides ample tools for slab logistics modelling and optimisation, but no methodology for establishing whether, and under what operating conditions, a more sophisticated allocation decision improves thermal and logistics performance. This paper attempts to construct that methodology.
A stochastic, thermally coupled discrete-event simulation of a slab yard is developed, driven by arrival, dwell-time, and grade-mix distributions derived from two years of production data, and validated for relative comparisons between strategies. Four storage allocation strategies are developed: a baseline reflecting current practice, a dwell-time heuristic, a rolling-horizon strategy, and a simulated annealing metaheuristic. The performance of these strategies is evaluated under identical random arrival streams through various experiments and paired statistical tests.
Under baseline load conditions, no strategy improves the current practice. Simulated annealing remains indistinguishable after 50 replications, and the heuristic performs worse over all four performance indicators. The developed allocation strategies yield benefits only when two conditions hold: discretionary hot box slack is available, and the placement objective function is non-separable. Under baseline load conditions, neither condition holds. Capacity, not allocation logic, drives performance in this case.
The two-condition criterion and the methodology used to establish it are the value of this research and are what transfer to any slab yard with limited buffer capacity operating under uncertainty. ...
The literature provides ample tools for slab logistics modelling and optimisation, but no methodology for establishing whether, and under what operating conditions, a more sophisticated allocation decision improves thermal and logistics performance. This paper attempts to construct that methodology.
A stochastic, thermally coupled discrete-event simulation of a slab yard is developed, driven by arrival, dwell-time, and grade-mix distributions derived from two years of production data, and validated for relative comparisons between strategies. Four storage allocation strategies are developed: a baseline reflecting current practice, a dwell-time heuristic, a rolling-horizon strategy, and a simulated annealing metaheuristic. The performance of these strategies is evaluated under identical random arrival streams through various experiments and paired statistical tests.
Under baseline load conditions, no strategy improves the current practice. Simulated annealing remains indistinguishable after 50 replications, and the heuristic performs worse over all four performance indicators. The developed allocation strategies yield benefits only when two conditions hold: discretionary hot box slack is available, and the placement objective function is non-separable. Under baseline load conditions, neither condition holds. Capacity, not allocation logic, drives performance in this case.
The two-condition criterion and the methodology used to establish it are the value of this research and are what transfer to any slab yard with limited buffer capacity operating under uncertainty. ...
Reheating slabs before hot rolling accounts for roughly 80% of the energy consumed by the hot-rolling process, so preserving slab heat during yard logistics is a direct energy lever. At the integrated steel plant used as a case study, insulated thermal buffers (hot boxes) can substantially slow the cooling of slabs. These hot boxes are currently reserved for fracture-sensitive steel grades and, on average, operate at about half capacity. On paper, this presents an opportunity for other slabs to use the spare buffer capacity. The question of whether managing that space with a more sophisticated allocation strategy is unanswered.
The literature provides ample tools for slab logistics modelling and optimisation, but no methodology for establishing whether, and under what operating conditions, a more sophisticated allocation decision improves thermal and logistics performance. This paper attempts to construct that methodology.
A stochastic, thermally coupled discrete-event simulation of a slab yard is developed, driven by arrival, dwell-time, and grade-mix distributions derived from two years of production data, and validated for relative comparisons between strategies. Four storage allocation strategies are developed: a baseline reflecting current practice, a dwell-time heuristic, a rolling-horizon strategy, and a simulated annealing metaheuristic. The performance of these strategies is evaluated under identical random arrival streams through various experiments and paired statistical tests.
Under baseline load conditions, no strategy improves the current practice. Simulated annealing remains indistinguishable after 50 replications, and the heuristic performs worse over all four performance indicators. The developed allocation strategies yield benefits only when two conditions hold: discretionary hot box slack is available, and the placement objective function is non-separable. Under baseline load conditions, neither condition holds. Capacity, not allocation logic, drives performance in this case.
The two-condition criterion and the methodology used to establish it are the value of this research and are what transfer to any slab yard with limited buffer capacity operating under uncertainty.
The literature provides ample tools for slab logistics modelling and optimisation, but no methodology for establishing whether, and under what operating conditions, a more sophisticated allocation decision improves thermal and logistics performance. This paper attempts to construct that methodology.
A stochastic, thermally coupled discrete-event simulation of a slab yard is developed, driven by arrival, dwell-time, and grade-mix distributions derived from two years of production data, and validated for relative comparisons between strategies. Four storage allocation strategies are developed: a baseline reflecting current practice, a dwell-time heuristic, a rolling-horizon strategy, and a simulated annealing metaheuristic. The performance of these strategies is evaluated under identical random arrival streams through various experiments and paired statistical tests.
Under baseline load conditions, no strategy improves the current practice. Simulated annealing remains indistinguishable after 50 replications, and the heuristic performs worse over all four performance indicators. The developed allocation strategies yield benefits only when two conditions hold: discretionary hot box slack is available, and the placement objective function is non-separable. Under baseline load conditions, neither condition holds. Capacity, not allocation logic, drives performance in this case.
The two-condition criterion and the methodology used to establish it are the value of this research and are what transfer to any slab yard with limited buffer capacity operating under uncertainty.
Assessing Packaging Systems in Offshore Logistics
A comparative framework for evaluating GHG emissions from single-use and reusable packaging
Packaging plays a necessary supporting role in offshore logistics, but it also contributes to material use, waste generation, and greenhouse-gas (GHG) emissions. Within the logistics operations of Heerema Marine Contractors (HMC), the environmental impact of packaging had not previously been quantified in a structured way, and the benefit of replacing the current single-use system with reusable alternatives was therefore unclear. This thesis develops and applies a comparative decision-support framework to assess packaging-related GHG emissions in offshore contractor logistics and to evaluate whether reusable packaging can reduce that impact. The study focuses specifically on packaging production, transportation, and end-of-life treatment, and is limited to GHG emissions expressed as CO₂eq.
The framework was applied to historical package-flow data extracted from the HMC ERP system for the period 2012–2025. Because packaging characteristics are not recorded systematically in the ERP, packaging configurations were reconstructed using package-assignment rules based on unit of measure, package weight, density assumptions, and transport-support logic. The resulting model combines package interpretation, transport reconstruction, end-of-life modelling, and yard-based validation to estimate the emissions of both the current single-use packaging system and a proposed reusable alternative.
Validation showed that ERP package labels do not always reflect physical packaging practice and that some informal reuse already occurs in the current system, meaning that the baseline should be interpreted as a conservative approximation of a predominantly single-use system.
The results show that the current packaging system is dominated by wooden transport items, particularly pallets and dunnage. Across the full study period, the estimated total mass of single-use packaging was 1546 mt, with corresponding emissions of 1141 mt CO₂eq under cut-off accounting and 783 mt CO₂eq under system-expansion accounting. For the in-house subset used in the reusable comparison, the baseline reusable scenario yielded only a marginal net benefit of 4.3 mt CO₂eq under cut-off accounting, while under system expansion it performed worse than the single-use baseline by 101.5 mt CO₂eq. Sensitivity analysis further showed that the comparative outcome is influenced more strongly by reverse-logistics performance than by reusable transport-item lifetime.
The thesis therefore concludes that packaging-related GHG emissions can be assessed systematically in a data-constrained offshore environment, but that reusable packaging should not be regarded as an inherently superior solution for HMC. Any potential benefit is limited, highly context-dependent, and sensitive to both accounting assumptions and operational conditions. ...
The framework was applied to historical package-flow data extracted from the HMC ERP system for the period 2012–2025. Because packaging characteristics are not recorded systematically in the ERP, packaging configurations were reconstructed using package-assignment rules based on unit of measure, package weight, density assumptions, and transport-support logic. The resulting model combines package interpretation, transport reconstruction, end-of-life modelling, and yard-based validation to estimate the emissions of both the current single-use packaging system and a proposed reusable alternative.
Validation showed that ERP package labels do not always reflect physical packaging practice and that some informal reuse already occurs in the current system, meaning that the baseline should be interpreted as a conservative approximation of a predominantly single-use system.
The results show that the current packaging system is dominated by wooden transport items, particularly pallets and dunnage. Across the full study period, the estimated total mass of single-use packaging was 1546 mt, with corresponding emissions of 1141 mt CO₂eq under cut-off accounting and 783 mt CO₂eq under system-expansion accounting. For the in-house subset used in the reusable comparison, the baseline reusable scenario yielded only a marginal net benefit of 4.3 mt CO₂eq under cut-off accounting, while under system expansion it performed worse than the single-use baseline by 101.5 mt CO₂eq. Sensitivity analysis further showed that the comparative outcome is influenced more strongly by reverse-logistics performance than by reusable transport-item lifetime.
The thesis therefore concludes that packaging-related GHG emissions can be assessed systematically in a data-constrained offshore environment, but that reusable packaging should not be regarded as an inherently superior solution for HMC. Any potential benefit is limited, highly context-dependent, and sensitive to both accounting assumptions and operational conditions. ...
Packaging plays a necessary supporting role in offshore logistics, but it also contributes to material use, waste generation, and greenhouse-gas (GHG) emissions. Within the logistics operations of Heerema Marine Contractors (HMC), the environmental impact of packaging had not previously been quantified in a structured way, and the benefit of replacing the current single-use system with reusable alternatives was therefore unclear. This thesis develops and applies a comparative decision-support framework to assess packaging-related GHG emissions in offshore contractor logistics and to evaluate whether reusable packaging can reduce that impact. The study focuses specifically on packaging production, transportation, and end-of-life treatment, and is limited to GHG emissions expressed as CO₂eq.
The framework was applied to historical package-flow data extracted from the HMC ERP system for the period 2012–2025. Because packaging characteristics are not recorded systematically in the ERP, packaging configurations were reconstructed using package-assignment rules based on unit of measure, package weight, density assumptions, and transport-support logic. The resulting model combines package interpretation, transport reconstruction, end-of-life modelling, and yard-based validation to estimate the emissions of both the current single-use packaging system and a proposed reusable alternative.
Validation showed that ERP package labels do not always reflect physical packaging practice and that some informal reuse already occurs in the current system, meaning that the baseline should be interpreted as a conservative approximation of a predominantly single-use system.
The results show that the current packaging system is dominated by wooden transport items, particularly pallets and dunnage. Across the full study period, the estimated total mass of single-use packaging was 1546 mt, with corresponding emissions of 1141 mt CO₂eq under cut-off accounting and 783 mt CO₂eq under system-expansion accounting. For the in-house subset used in the reusable comparison, the baseline reusable scenario yielded only a marginal net benefit of 4.3 mt CO₂eq under cut-off accounting, while under system expansion it performed worse than the single-use baseline by 101.5 mt CO₂eq. Sensitivity analysis further showed that the comparative outcome is influenced more strongly by reverse-logistics performance than by reusable transport-item lifetime.
The thesis therefore concludes that packaging-related GHG emissions can be assessed systematically in a data-constrained offshore environment, but that reusable packaging should not be regarded as an inherently superior solution for HMC. Any potential benefit is limited, highly context-dependent, and sensitive to both accounting assumptions and operational conditions.
The framework was applied to historical package-flow data extracted from the HMC ERP system for the period 2012–2025. Because packaging characteristics are not recorded systematically in the ERP, packaging configurations were reconstructed using package-assignment rules based on unit of measure, package weight, density assumptions, and transport-support logic. The resulting model combines package interpretation, transport reconstruction, end-of-life modelling, and yard-based validation to estimate the emissions of both the current single-use packaging system and a proposed reusable alternative.
Validation showed that ERP package labels do not always reflect physical packaging practice and that some informal reuse already occurs in the current system, meaning that the baseline should be interpreted as a conservative approximation of a predominantly single-use system.
The results show that the current packaging system is dominated by wooden transport items, particularly pallets and dunnage. Across the full study period, the estimated total mass of single-use packaging was 1546 mt, with corresponding emissions of 1141 mt CO₂eq under cut-off accounting and 783 mt CO₂eq under system-expansion accounting. For the in-house subset used in the reusable comparison, the baseline reusable scenario yielded only a marginal net benefit of 4.3 mt CO₂eq under cut-off accounting, while under system expansion it performed worse than the single-use baseline by 101.5 mt CO₂eq. Sensitivity analysis further showed that the comparative outcome is influenced more strongly by reverse-logistics performance than by reusable transport-item lifetime.
The thesis therefore concludes that packaging-related GHG emissions can be assessed systematically in a data-constrained offshore environment, but that reusable packaging should not be regarded as an inherently superior solution for HMC. Any potential benefit is limited, highly context-dependent, and sensitive to both accounting assumptions and operational conditions.
Reliable container-tracking depends on the quality of estimated time-of-arrival (ETA) data, yet existing logistics platforms offer little guidance on how trustworthy those timestamps really are. This thesis proposes a fit-for-use data-quality (DQ) framework for Digital Container Shipping Association (DCSA)-compliant event logs that flags ETA records likely to deviate from actual time of arrival (ATA) by more than one calendar day.
Event logs from $\sim$90\,k transport legs were preprocessed into records capturing origin-destination pair, carrier, publisher type, and timing information. Four supervised models, namely Linear Regression (LR), Random Forest, XGBoost, and a Neural Network, were trained to predict leg duration. A prediction that placed ATA \(>1\) day from the published ETA labeled that record \textit{low-quality}. Model outputs were evaluated with a precision-oriented \(\mathrm{F}_{\beta}\)-score, where a false alarm is 50 times more costly than a missed detection (\(\beta \approx 0.141\)).
The simplest model prevailed: standard LR achieved the highest overall \(\mathrm{F}_{0.141}\)-score (68.5 \%), balancing few false positives with robust recall, while more-complex tree-based and neural models produced excessive false alarms. When the analysis was narrowed to early-stage ETAs published by carriers (arguably the least reliable yet most operationally valuable subset) LR’s score rose to 72.0 \%. These findings highlight that careful feature engineering and data curation outweigh algorithmic complexity for this task.
The study delivers the first systematic, event-data-only method to quantify DQ in container tracking, enabling near-real-time plausibility checks without AIS feeds. Limitations include a three-month observation window and absence of exogenous factors such as weather or port congestion. Future work should extend the temporal scope, integrate AIS-derived and environmental features, and explore meta-learning techniques to adapt to disruptions. It could also use process-mining to uncover anomalous event sequences to take a different approach in dataquality assessment within container-eventlogs.
By demonstrating that a transparent LR baseline can reliably surface dubious ETAs, the thesis provides a practical blueprint for logistics platforms seeking to bolster trust in their tracking data and to prioritise corrective action where it matters most. ...
Event logs from $\sim$90\,k transport legs were preprocessed into records capturing origin-destination pair, carrier, publisher type, and timing information. Four supervised models, namely Linear Regression (LR), Random Forest, XGBoost, and a Neural Network, were trained to predict leg duration. A prediction that placed ATA \(>1\) day from the published ETA labeled that record \textit{low-quality}. Model outputs were evaluated with a precision-oriented \(\mathrm{F}_{\beta}\)-score, where a false alarm is 50 times more costly than a missed detection (\(\beta \approx 0.141\)).
The simplest model prevailed: standard LR achieved the highest overall \(\mathrm{F}_{0.141}\)-score (68.5 \%), balancing few false positives with robust recall, while more-complex tree-based and neural models produced excessive false alarms. When the analysis was narrowed to early-stage ETAs published by carriers (arguably the least reliable yet most operationally valuable subset) LR’s score rose to 72.0 \%. These findings highlight that careful feature engineering and data curation outweigh algorithmic complexity for this task.
The study delivers the first systematic, event-data-only method to quantify DQ in container tracking, enabling near-real-time plausibility checks without AIS feeds. Limitations include a three-month observation window and absence of exogenous factors such as weather or port congestion. Future work should extend the temporal scope, integrate AIS-derived and environmental features, and explore meta-learning techniques to adapt to disruptions. It could also use process-mining to uncover anomalous event sequences to take a different approach in dataquality assessment within container-eventlogs.
By demonstrating that a transparent LR baseline can reliably surface dubious ETAs, the thesis provides a practical blueprint for logistics platforms seeking to bolster trust in their tracking data and to prioritise corrective action where it matters most. ...
Reliable container-tracking depends on the quality of estimated time-of-arrival (ETA) data, yet existing logistics platforms offer little guidance on how trustworthy those timestamps really are. This thesis proposes a fit-for-use data-quality (DQ) framework for Digital Container Shipping Association (DCSA)-compliant event logs that flags ETA records likely to deviate from actual time of arrival (ATA) by more than one calendar day.
Event logs from $\sim$90\,k transport legs were preprocessed into records capturing origin-destination pair, carrier, publisher type, and timing information. Four supervised models, namely Linear Regression (LR), Random Forest, XGBoost, and a Neural Network, were trained to predict leg duration. A prediction that placed ATA \(>1\) day from the published ETA labeled that record \textit{low-quality}. Model outputs were evaluated with a precision-oriented \(\mathrm{F}_{\beta}\)-score, where a false alarm is 50 times more costly than a missed detection (\(\beta \approx 0.141\)).
The simplest model prevailed: standard LR achieved the highest overall \(\mathrm{F}_{0.141}\)-score (68.5 \%), balancing few false positives with robust recall, while more-complex tree-based and neural models produced excessive false alarms. When the analysis was narrowed to early-stage ETAs published by carriers (arguably the least reliable yet most operationally valuable subset) LR’s score rose to 72.0 \%. These findings highlight that careful feature engineering and data curation outweigh algorithmic complexity for this task.
The study delivers the first systematic, event-data-only method to quantify DQ in container tracking, enabling near-real-time plausibility checks without AIS feeds. Limitations include a three-month observation window and absence of exogenous factors such as weather or port congestion. Future work should extend the temporal scope, integrate AIS-derived and environmental features, and explore meta-learning techniques to adapt to disruptions. It could also use process-mining to uncover anomalous event sequences to take a different approach in dataquality assessment within container-eventlogs.
By demonstrating that a transparent LR baseline can reliably surface dubious ETAs, the thesis provides a practical blueprint for logistics platforms seeking to bolster trust in their tracking data and to prioritise corrective action where it matters most.
Event logs from $\sim$90\,k transport legs were preprocessed into records capturing origin-destination pair, carrier, publisher type, and timing information. Four supervised models, namely Linear Regression (LR), Random Forest, XGBoost, and a Neural Network, were trained to predict leg duration. A prediction that placed ATA \(>1\) day from the published ETA labeled that record \textit{low-quality}. Model outputs were evaluated with a precision-oriented \(\mathrm{F}_{\beta}\)-score, where a false alarm is 50 times more costly than a missed detection (\(\beta \approx 0.141\)).
The simplest model prevailed: standard LR achieved the highest overall \(\mathrm{F}_{0.141}\)-score (68.5 \%), balancing few false positives with robust recall, while more-complex tree-based and neural models produced excessive false alarms. When the analysis was narrowed to early-stage ETAs published by carriers (arguably the least reliable yet most operationally valuable subset) LR’s score rose to 72.0 \%. These findings highlight that careful feature engineering and data curation outweigh algorithmic complexity for this task.
The study delivers the first systematic, event-data-only method to quantify DQ in container tracking, enabling near-real-time plausibility checks without AIS feeds. Limitations include a three-month observation window and absence of exogenous factors such as weather or port congestion. Future work should extend the temporal scope, integrate AIS-derived and environmental features, and explore meta-learning techniques to adapt to disruptions. It could also use process-mining to uncover anomalous event sequences to take a different approach in dataquality assessment within container-eventlogs.
By demonstrating that a transparent LR baseline can reliably surface dubious ETAs, the thesis provides a practical blueprint for logistics platforms seeking to bolster trust in their tracking data and to prioritise corrective action where it matters most.
Increasing Drought Resilience of the Port-Hinterland System
A Case Study on the Rhine River
Master thesis
(2025)
-
F.C. van Steekelenburg, M. Saeednia, N. Pourmohammadzia, F. Schulte, A.J. van Binsbergen
Future demand for container transport is expected to increase, placing pressure on the port-hinterland system of the Rhine river. To meet European climate goals and relieve the congested road freight system, inland waterway transport (IWT) is promoted as a sustainable alternative. However, IWT is vulnerable to low water levels as a result of climate change, as evidenced by significant drought periods in 2018. This research investigates how the strategic establishment of transshipment hubs can enhance the climate resilience of the port-hinterland system of the Rhine river under varying river discharge scenarios. The focus is on three critical bottlenecks in terms of available draught: Druten, Duisburg, and Kaub. This study quantifies the impact of new hubs on the total cost and CO2-equivalent emissions by developing a mathematical location routing model, which is used for a scenario-based analysis with varying river discharge levels and rail capacity levels. Results show that hub establishment significantly reduces both total transport costs and CO2-equivalent emissions during low discharge scenarios, without requiring expansion of existing terminals. The establishment of hubs depends on location and multimodal accessibility, with rail connections only becoming economically viable when the remaining hub-destination distance exceeds certain distance thresholds. The findings provide actionable insights for policymakers and transport authorities on strategic decisions of infrastructure investments to improve the climate-resilience of freight transport along the Rhine corridor to minimize total costs, reduce CO2-equivalent emissions, and relieve congested road networks.
...
Future demand for container transport is expected to increase, placing pressure on the port-hinterland system of the Rhine river. To meet European climate goals and relieve the congested road freight system, inland waterway transport (IWT) is promoted as a sustainable alternative. However, IWT is vulnerable to low water levels as a result of climate change, as evidenced by significant drought periods in 2018. This research investigates how the strategic establishment of transshipment hubs can enhance the climate resilience of the port-hinterland system of the Rhine river under varying river discharge scenarios. The focus is on three critical bottlenecks in terms of available draught: Druten, Duisburg, and Kaub. This study quantifies the impact of new hubs on the total cost and CO2-equivalent emissions by developing a mathematical location routing model, which is used for a scenario-based analysis with varying river discharge levels and rail capacity levels. Results show that hub establishment significantly reduces both total transport costs and CO2-equivalent emissions during low discharge scenarios, without requiring expansion of existing terminals. The establishment of hubs depends on location and multimodal accessibility, with rail connections only becoming economically viable when the remaining hub-destination distance exceeds certain distance thresholds. The findings provide actionable insights for policymakers and transport authorities on strategic decisions of infrastructure investments to improve the climate-resilience of freight transport along the Rhine corridor to minimize total costs, reduce CO2-equivalent emissions, and relieve congested road networks.
Predicting mooring line forces of large-scale vessels using machine learning
A case study at the Port of Rotterdam
Master thesis
(2025)
-
N.K.S. Wilking, J.J. Burgers, Vasso Reppa, A.A. Roubos, Alex van Deyzen, F. Schulte
The increasing size of cargo vessels poses significant challenges for ports in ensuring safe mooring, as larger ships result in higher mooring forces.
Most existing port infrastructure and mooring equipment were designed for smaller ships, making accurate estimation of mooring line forces increasingly critical.
Metamodels, machine learning models trained on numerically simulated data, offer a promising alternative to traditional, computationally expensive simulation-based methods by enabling rapid predictions with a useful level of accuracy.
This study proposes a metamodeling approach for the numerical Dynamic Mooring Analysis (DMA) to predict mooring line forces from input parameters that describe environmental conditions, mooring systems, and ship characteristics.
The methodology is demonstrated in a case study involving a 333-meter container vessel moored at a berth in the Port of Rotterdam.
A total of 11,520 scenarios were simulated using the DMA model aNySIM and used to train and test two candidate metamodels: Linear Regression (LR) and Multilayer Perceptron (MLP).
After evaluating both models on predictive accuracy, efficiency in terms of prediction speed and development effort, and interpretability, the MLP was selected as the preferred DMA metamodel.
It achieved high predictive performance, with an RMSE of 10 kN and an R2 of 0.996, while offering prediction times measured in microseconds. This is more than seven orders of magnitude faster than the numerical DMA, thereby enabling large-batch predictions.
The metamodel revealed that pretension is clearly the most dominant feature for predicting the mean mooring line force, followed by MBL. For the maximum mooring line force, the most influential features were identified as pretension, windvelocity, and wind direction. ...
Most existing port infrastructure and mooring equipment were designed for smaller ships, making accurate estimation of mooring line forces increasingly critical.
Metamodels, machine learning models trained on numerically simulated data, offer a promising alternative to traditional, computationally expensive simulation-based methods by enabling rapid predictions with a useful level of accuracy.
This study proposes a metamodeling approach for the numerical Dynamic Mooring Analysis (DMA) to predict mooring line forces from input parameters that describe environmental conditions, mooring systems, and ship characteristics.
The methodology is demonstrated in a case study involving a 333-meter container vessel moored at a berth in the Port of Rotterdam.
A total of 11,520 scenarios were simulated using the DMA model aNySIM and used to train and test two candidate metamodels: Linear Regression (LR) and Multilayer Perceptron (MLP).
After evaluating both models on predictive accuracy, efficiency in terms of prediction speed and development effort, and interpretability, the MLP was selected as the preferred DMA metamodel.
It achieved high predictive performance, with an RMSE of 10 kN and an R2 of 0.996, while offering prediction times measured in microseconds. This is more than seven orders of magnitude faster than the numerical DMA, thereby enabling large-batch predictions.
The metamodel revealed that pretension is clearly the most dominant feature for predicting the mean mooring line force, followed by MBL. For the maximum mooring line force, the most influential features were identified as pretension, windvelocity, and wind direction. ...
The increasing size of cargo vessels poses significant challenges for ports in ensuring safe mooring, as larger ships result in higher mooring forces.
Most existing port infrastructure and mooring equipment were designed for smaller ships, making accurate estimation of mooring line forces increasingly critical.
Metamodels, machine learning models trained on numerically simulated data, offer a promising alternative to traditional, computationally expensive simulation-based methods by enabling rapid predictions with a useful level of accuracy.
This study proposes a metamodeling approach for the numerical Dynamic Mooring Analysis (DMA) to predict mooring line forces from input parameters that describe environmental conditions, mooring systems, and ship characteristics.
The methodology is demonstrated in a case study involving a 333-meter container vessel moored at a berth in the Port of Rotterdam.
A total of 11,520 scenarios were simulated using the DMA model aNySIM and used to train and test two candidate metamodels: Linear Regression (LR) and Multilayer Perceptron (MLP).
After evaluating both models on predictive accuracy, efficiency in terms of prediction speed and development effort, and interpretability, the MLP was selected as the preferred DMA metamodel.
It achieved high predictive performance, with an RMSE of 10 kN and an R2 of 0.996, while offering prediction times measured in microseconds. This is more than seven orders of magnitude faster than the numerical DMA, thereby enabling large-batch predictions.
The metamodel revealed that pretension is clearly the most dominant feature for predicting the mean mooring line force, followed by MBL. For the maximum mooring line force, the most influential features were identified as pretension, windvelocity, and wind direction.
Most existing port infrastructure and mooring equipment were designed for smaller ships, making accurate estimation of mooring line forces increasingly critical.
Metamodels, machine learning models trained on numerically simulated data, offer a promising alternative to traditional, computationally expensive simulation-based methods by enabling rapid predictions with a useful level of accuracy.
This study proposes a metamodeling approach for the numerical Dynamic Mooring Analysis (DMA) to predict mooring line forces from input parameters that describe environmental conditions, mooring systems, and ship characteristics.
The methodology is demonstrated in a case study involving a 333-meter container vessel moored at a berth in the Port of Rotterdam.
A total of 11,520 scenarios were simulated using the DMA model aNySIM and used to train and test two candidate metamodels: Linear Regression (LR) and Multilayer Perceptron (MLP).
After evaluating both models on predictive accuracy, efficiency in terms of prediction speed and development effort, and interpretability, the MLP was selected as the preferred DMA metamodel.
It achieved high predictive performance, with an RMSE of 10 kN and an R2 of 0.996, while offering prediction times measured in microseconds. This is more than seven orders of magnitude faster than the numerical DMA, thereby enabling large-batch predictions.
The metamodel revealed that pretension is clearly the most dominant feature for predicting the mean mooring line force, followed by MBL. For the maximum mooring line force, the most influential features were identified as pretension, windvelocity, and wind direction.
Over the past decade, consumer shopping habits have increasingly shifted towards online grocery purchases, creating a growing demand for efficient and scalable warehouse operations. As order volumes and complexity rise, the optimization of warehouse processes has become essential to meet requirements while controlling operational costs. This study addresses the Joint Order Batching Picker Routing Problem (JOBPRP) by developing an exact optimization approach and examining the interdependencies among key warehouse processes, such as batching, routing, and product allocation. Using a case study of Crisp B.V., an online grocery retailer, the proposed algorithm was implemented to optimize configurations across multiple temperature-controlled zones with varying operational characteristics, such as differing pick densities and operational constraints. The research shows that all warehouse processes are interconnected and the best performing configuration is depending on the operational characteristics of the warehouse. The optimization approach achieved a 39.14% reduction in weekly travel distance compared to Crisp’s current benchmark, highlighting its potential to significantly enhance travel distances. These findings highlight the significant impact of integrating order batching and picker routing on warehouse efficiency. The study not only demonstrates the critical correlation between warehouse processes but also provides actionable insights for optimizing order picking in high-density, large-scale warehouses
...
Over the past decade, consumer shopping habits have increasingly shifted towards online grocery purchases, creating a growing demand for efficient and scalable warehouse operations. As order volumes and complexity rise, the optimization of warehouse processes has become essential to meet requirements while controlling operational costs. This study addresses the Joint Order Batching Picker Routing Problem (JOBPRP) by developing an exact optimization approach and examining the interdependencies among key warehouse processes, such as batching, routing, and product allocation. Using a case study of Crisp B.V., an online grocery retailer, the proposed algorithm was implemented to optimize configurations across multiple temperature-controlled zones with varying operational characteristics, such as differing pick densities and operational constraints. The research shows that all warehouse processes are interconnected and the best performing configuration is depending on the operational characteristics of the warehouse. The optimization approach achieved a 39.14% reduction in weekly travel distance compared to Crisp’s current benchmark, highlighting its potential to significantly enhance travel distances. These findings highlight the significant impact of integrating order batching and picker routing on warehouse efficiency. The study not only demonstrates the critical correlation between warehouse processes but also provides actionable insights for optimizing order picking in high-density, large-scale warehouses
In this thesis, an exploration into the potential use of path-planning algorithms in modeling cyclist behavior is made, and a novel model utilizing such algorithms, incorporating the commonly found overtaking behavior on bike paths, is developed and assessed.
The investigation started with a literature review on the existing behavioral interpretations and findings of bicycle riding, where the cyclists' interaction with the environment and other cyclists is found to be at different task levels of the riding process. Based on the findings, further analysis into how existing techniques replicate the different layers of behavior is made and assessed. During the assessment, this research determined that path-planning algorithms could best be used to replicate the physical steering behavior of cyclists. An investigation into the inner workings of path-planning algorithms is also done, resulting in the final four-layer conceptual framework based on the two-layer operational framework of bicycle riding, adapting the process into mental (perception, goal orientation) and physical (path planning, movement) layers.
With the adapted modeling framework, a model is developed, verified, and assessed with face-validation against real-world trajectory data. The development and verification step provides insights into the inner workings of the model, showcasing how the four layers of the framework are realized, and how the changing of used parameters would affect the intermediate output between the model layers. The face validation consists of two scenarios: a physical steering and pedaling-focused scenario of chicanes, which is a series of bottlenecks, and an overtaking scenario that focuses on the mental process of overtaking decisions. The results showcased that the developed model can create plausible steering and pedaling behaviors in the chicane scenario. However, the model showed lower accuracy and consistency in predicting the mental overtaking maneuvers.
The assessment result of the developed model showcased the strength of path-planning algorithms in augmenting the existing model with the physical steering capability of the cyclists. The limited accuracy in the overtaking scenario highlights the importance of capturing the mental process of bicycle riding. Future work could further refine the mental layers of the framework, specifically the goal orientation process. The adapted modeling framework also provides a new direction and foundation for further work on the path-planning additions and improvement of bicycle behavioral modeling. ...
The investigation started with a literature review on the existing behavioral interpretations and findings of bicycle riding, where the cyclists' interaction with the environment and other cyclists is found to be at different task levels of the riding process. Based on the findings, further analysis into how existing techniques replicate the different layers of behavior is made and assessed. During the assessment, this research determined that path-planning algorithms could best be used to replicate the physical steering behavior of cyclists. An investigation into the inner workings of path-planning algorithms is also done, resulting in the final four-layer conceptual framework based on the two-layer operational framework of bicycle riding, adapting the process into mental (perception, goal orientation) and physical (path planning, movement) layers.
With the adapted modeling framework, a model is developed, verified, and assessed with face-validation against real-world trajectory data. The development and verification step provides insights into the inner workings of the model, showcasing how the four layers of the framework are realized, and how the changing of used parameters would affect the intermediate output between the model layers. The face validation consists of two scenarios: a physical steering and pedaling-focused scenario of chicanes, which is a series of bottlenecks, and an overtaking scenario that focuses on the mental process of overtaking decisions. The results showcased that the developed model can create plausible steering and pedaling behaviors in the chicane scenario. However, the model showed lower accuracy and consistency in predicting the mental overtaking maneuvers.
The assessment result of the developed model showcased the strength of path-planning algorithms in augmenting the existing model with the physical steering capability of the cyclists. The limited accuracy in the overtaking scenario highlights the importance of capturing the mental process of bicycle riding. Future work could further refine the mental layers of the framework, specifically the goal orientation process. The adapted modeling framework also provides a new direction and foundation for further work on the path-planning additions and improvement of bicycle behavioral modeling. ...
In this thesis, an exploration into the potential use of path-planning algorithms in modeling cyclist behavior is made, and a novel model utilizing such algorithms, incorporating the commonly found overtaking behavior on bike paths, is developed and assessed.
The investigation started with a literature review on the existing behavioral interpretations and findings of bicycle riding, where the cyclists' interaction with the environment and other cyclists is found to be at different task levels of the riding process. Based on the findings, further analysis into how existing techniques replicate the different layers of behavior is made and assessed. During the assessment, this research determined that path-planning algorithms could best be used to replicate the physical steering behavior of cyclists. An investigation into the inner workings of path-planning algorithms is also done, resulting in the final four-layer conceptual framework based on the two-layer operational framework of bicycle riding, adapting the process into mental (perception, goal orientation) and physical (path planning, movement) layers.
With the adapted modeling framework, a model is developed, verified, and assessed with face-validation against real-world trajectory data. The development and verification step provides insights into the inner workings of the model, showcasing how the four layers of the framework are realized, and how the changing of used parameters would affect the intermediate output between the model layers. The face validation consists of two scenarios: a physical steering and pedaling-focused scenario of chicanes, which is a series of bottlenecks, and an overtaking scenario that focuses on the mental process of overtaking decisions. The results showcased that the developed model can create plausible steering and pedaling behaviors in the chicane scenario. However, the model showed lower accuracy and consistency in predicting the mental overtaking maneuvers.
The assessment result of the developed model showcased the strength of path-planning algorithms in augmenting the existing model with the physical steering capability of the cyclists. The limited accuracy in the overtaking scenario highlights the importance of capturing the mental process of bicycle riding. Future work could further refine the mental layers of the framework, specifically the goal orientation process. The adapted modeling framework also provides a new direction and foundation for further work on the path-planning additions and improvement of bicycle behavioral modeling.
The investigation started with a literature review on the existing behavioral interpretations and findings of bicycle riding, where the cyclists' interaction with the environment and other cyclists is found to be at different task levels of the riding process. Based on the findings, further analysis into how existing techniques replicate the different layers of behavior is made and assessed. During the assessment, this research determined that path-planning algorithms could best be used to replicate the physical steering behavior of cyclists. An investigation into the inner workings of path-planning algorithms is also done, resulting in the final four-layer conceptual framework based on the two-layer operational framework of bicycle riding, adapting the process into mental (perception, goal orientation) and physical (path planning, movement) layers.
With the adapted modeling framework, a model is developed, verified, and assessed with face-validation against real-world trajectory data. The development and verification step provides insights into the inner workings of the model, showcasing how the four layers of the framework are realized, and how the changing of used parameters would affect the intermediate output between the model layers. The face validation consists of two scenarios: a physical steering and pedaling-focused scenario of chicanes, which is a series of bottlenecks, and an overtaking scenario that focuses on the mental process of overtaking decisions. The results showcased that the developed model can create plausible steering and pedaling behaviors in the chicane scenario. However, the model showed lower accuracy and consistency in predicting the mental overtaking maneuvers.
The assessment result of the developed model showcased the strength of path-planning algorithms in augmenting the existing model with the physical steering capability of the cyclists. The limited accuracy in the overtaking scenario highlights the importance of capturing the mental process of bicycle riding. Future work could further refine the mental layers of the framework, specifically the goal orientation process. The adapted modeling framework also provides a new direction and foundation for further work on the path-planning additions and improvement of bicycle behavioral modeling.
Metro networks face operational challenges due to increasing ridership and system growth, particularly in managing delay propagation. Epidemiology models have recently been an interesting method in transportation research for studying delays. This study, therefore, aims to see if the Susceptible-infectious-susceptible (SIS) model is suitable to help model delay propagation in a metro network through its ability to reproduce the vulnerability of metro stations for specific instances. Using data from the Washington Metro Network, delay propagation instances were grouped, and the model was trained and tested using a differential evolution algorithm. The results indicate that the vulnerability values as calculated from the data do not follow the expected trend. Also, the model can predict the vulnerability values for the first group more accurately than the second group. However, limitations such as underestimation and overestimation of station vulnerabilities, and sensitivity to training data and parameters were observed. These challenges stemmed from the dynamics between specific parameters, the mismatch in the order of magnitude of model components, and the lack of additional factors.
...
Metro networks face operational challenges due to increasing ridership and system growth, particularly in managing delay propagation. Epidemiology models have recently been an interesting method in transportation research for studying delays. This study, therefore, aims to see if the Susceptible-infectious-susceptible (SIS) model is suitable to help model delay propagation in a metro network through its ability to reproduce the vulnerability of metro stations for specific instances. Using data from the Washington Metro Network, delay propagation instances were grouped, and the model was trained and tested using a differential evolution algorithm. The results indicate that the vulnerability values as calculated from the data do not follow the expected trend. Also, the model can predict the vulnerability values for the first group more accurately than the second group. However, limitations such as underestimation and overestimation of station vulnerabilities, and sensitivity to training data and parameters were observed. These challenges stemmed from the dynamics between specific parameters, the mismatch in the order of magnitude of model components, and the lack of additional factors.
Strategies for Identifying Outlier Parcels in Urban Deliveries
An Explorative analysis
This study presents two novel strategies for identifying outlier parcels in urban deliveries, focusing on cost and environmental impact. With the growth of Business-to-Customer (B2C) e-commerce, the logistics industry faces increased pressure to improve last-mile delivery (LMD) efficiency. Outlier parcels, characterized by higher costs and emissions, pose significant challenges to logistics service providers (LSPs). Using the Marginal Cost Method and the COFRET method, this study develops cost-based and emission-based identification strategies to help LSPs address inefficiencies. The elbow point method is introduced to set objective thresholds for identifying outliers. Results from simulations in the South Holland region reveal significant differences in outlier parcel distributions across various carriers. Additionally, a sensitivity analysis assesses the impact of carbon pricing on parcel identification. The findings provide actionable insights for LSPs to optimize delivery operations and achieve sustainability goals. Future work will explore the application of these methods in different regions and consider alternative delivery methods for handling outlier parcels.
...
This study presents two novel strategies for identifying outlier parcels in urban deliveries, focusing on cost and environmental impact. With the growth of Business-to-Customer (B2C) e-commerce, the logistics industry faces increased pressure to improve last-mile delivery (LMD) efficiency. Outlier parcels, characterized by higher costs and emissions, pose significant challenges to logistics service providers (LSPs). Using the Marginal Cost Method and the COFRET method, this study develops cost-based and emission-based identification strategies to help LSPs address inefficiencies. The elbow point method is introduced to set objective thresholds for identifying outliers. Results from simulations in the South Holland region reveal significant differences in outlier parcel distributions across various carriers. Additionally, a sensitivity analysis assesses the impact of carbon pricing on parcel identification. The findings provide actionable insights for LSPs to optimize delivery operations and achieve sustainability goals. Future work will explore the application of these methods in different regions and consider alternative delivery methods for handling outlier parcels.
Impact of different energy types of military vehicles on the supply chain
A MILP model for an optimal military Vehicle Energy Supply Chain
The Dutch Ministry of Defence (NLMoD) has stated the goal of reducing dependency on fossil fuels by at least 20% by the year 2030 and at least 70% by the year 2050 compared to the year 2010. Research in the NLMoD explores the possibility of changing diesel-fuelled vehicles and weapon platforms to alternative forms of energy such as electric or sustainable fuels. In these projects, the focus is on (part of) the vehicle or energy source itself, but the impact on the Military Supply Chain (MSC) is missing.
This research has developed a Mixed Integer Linear Programming model that can be used to gain insight into the impact of the energy type of tactical vehicles and weapon platforms on the MSC and therefore is able to see what energy type has the lowest impact on that MSC. The impact on the MSC is measured by minimizing the refuel time, number of supply trips, and CO2 equivalent emissions. The model can provide insight into what the minimal requirements of potential energy carriers and conversion devices should be in order to have a similar or better impact on the current diesel MSC. The model is based on the current supply chain of the NLMoD and is expanded with the use of APUs for vehicles, energy generation at Nodes, the use of small supply trucks as energy buffers, compatible supply material, and longer self-sufficient times. Combinations of these are looked at in different policies.
Results show the trend that energy types with lower CO2 equivalent emissions have higher refuel time and number of supply trips. An exception to this is HVO and HVO-electric series hybrid, which also have the least impact on the MSC. Energy types such as hydrogen and electric require huge improvements in energy density, fill speed, and FTW efficiency to come close to the results of current diesel. ...
This research has developed a Mixed Integer Linear Programming model that can be used to gain insight into the impact of the energy type of tactical vehicles and weapon platforms on the MSC and therefore is able to see what energy type has the lowest impact on that MSC. The impact on the MSC is measured by minimizing the refuel time, number of supply trips, and CO2 equivalent emissions. The model can provide insight into what the minimal requirements of potential energy carriers and conversion devices should be in order to have a similar or better impact on the current diesel MSC. The model is based on the current supply chain of the NLMoD and is expanded with the use of APUs for vehicles, energy generation at Nodes, the use of small supply trucks as energy buffers, compatible supply material, and longer self-sufficient times. Combinations of these are looked at in different policies.
Results show the trend that energy types with lower CO2 equivalent emissions have higher refuel time and number of supply trips. An exception to this is HVO and HVO-electric series hybrid, which also have the least impact on the MSC. Energy types such as hydrogen and electric require huge improvements in energy density, fill speed, and FTW efficiency to come close to the results of current diesel. ...
The Dutch Ministry of Defence (NLMoD) has stated the goal of reducing dependency on fossil fuels by at least 20% by the year 2030 and at least 70% by the year 2050 compared to the year 2010. Research in the NLMoD explores the possibility of changing diesel-fuelled vehicles and weapon platforms to alternative forms of energy such as electric or sustainable fuels. In these projects, the focus is on (part of) the vehicle or energy source itself, but the impact on the Military Supply Chain (MSC) is missing.
This research has developed a Mixed Integer Linear Programming model that can be used to gain insight into the impact of the energy type of tactical vehicles and weapon platforms on the MSC and therefore is able to see what energy type has the lowest impact on that MSC. The impact on the MSC is measured by minimizing the refuel time, number of supply trips, and CO2 equivalent emissions. The model can provide insight into what the minimal requirements of potential energy carriers and conversion devices should be in order to have a similar or better impact on the current diesel MSC. The model is based on the current supply chain of the NLMoD and is expanded with the use of APUs for vehicles, energy generation at Nodes, the use of small supply trucks as energy buffers, compatible supply material, and longer self-sufficient times. Combinations of these are looked at in different policies.
Results show the trend that energy types with lower CO2 equivalent emissions have higher refuel time and number of supply trips. An exception to this is HVO and HVO-electric series hybrid, which also have the least impact on the MSC. Energy types such as hydrogen and electric require huge improvements in energy density, fill speed, and FTW efficiency to come close to the results of current diesel.
This research has developed a Mixed Integer Linear Programming model that can be used to gain insight into the impact of the energy type of tactical vehicles and weapon platforms on the MSC and therefore is able to see what energy type has the lowest impact on that MSC. The impact on the MSC is measured by minimizing the refuel time, number of supply trips, and CO2 equivalent emissions. The model can provide insight into what the minimal requirements of potential energy carriers and conversion devices should be in order to have a similar or better impact on the current diesel MSC. The model is based on the current supply chain of the NLMoD and is expanded with the use of APUs for vehicles, energy generation at Nodes, the use of small supply trucks as energy buffers, compatible supply material, and longer self-sufficient times. Combinations of these are looked at in different policies.
Results show the trend that energy types with lower CO2 equivalent emissions have higher refuel time and number of supply trips. An exception to this is HVO and HVO-electric series hybrid, which also have the least impact on the MSC. Energy types such as hydrogen and electric require huge improvements in energy density, fill speed, and FTW efficiency to come close to the results of current diesel.
Redesign of the Car Distribution Process: a Dutch case study
A Holistic Approach in a Capacitated Vehicle Routing Problem to Reduce Direct CO2 Emissions in a Truck-Based Car Distribution Process
Purpose - With the short term need to reduce direct CO2 emissions of trucks in distribution processes, this paper aims to provide an easy to implement solution approach for distribution processes of new cars from holistic perspective. Scarce emphasizes is provided on short-term alternatives in truck-based distribution processes and approaches lacks applicable for large scale problems including split delivery function.
Design/methodology/approach - The distribution process and model methods are analyzed using a literature study and interviews with experts, resulting in the development of a solution approach. Combined with an extensive field research, a solution approach enables the performance evaluation of the current state, and the policy implications. Future designs are used to validate the solution approach by calculating performance differences in multiple relevant evaluation domains.
Findings - The analysis of the current state has identified critical bottlenecks, leading to the development of two promising policies. The application of a new and validated prioritization strategy and permitting more stops per truck has successfully yielded a significant reduction in CO2 emissions. The performance of the solution approach demonstrates high precision on a small scale and yields results comparable to actual practices on a larger scale, suggesting the approach's effectiveness and potential for future application.
Research limitations/implications - This research provides a new solution approach for evaluating direct CO2 emissions of model different designs of distribution processes. Despite its narrow scope, the transportation sector has a significant environmental footprint, and offers the potential for substantial reductions in emissions. From modeling perspective, further research is suggested in integrating split delivery function without using dummy variables.
Originality/value - This paper contributes by identifying critical gaps in the understanding and implementation of system-wide efficient car distribution processes from distribution hubs to car dealers. It not only addresses potential improvements, but also proved efficiency gains of the system with a new solution approach, using a new combination of a state-of-the-art meta-heuristic and a proven split delivery method applicable for large-scale problems. ...
Design/methodology/approach - The distribution process and model methods are analyzed using a literature study and interviews with experts, resulting in the development of a solution approach. Combined with an extensive field research, a solution approach enables the performance evaluation of the current state, and the policy implications. Future designs are used to validate the solution approach by calculating performance differences in multiple relevant evaluation domains.
Findings - The analysis of the current state has identified critical bottlenecks, leading to the development of two promising policies. The application of a new and validated prioritization strategy and permitting more stops per truck has successfully yielded a significant reduction in CO2 emissions. The performance of the solution approach demonstrates high precision on a small scale and yields results comparable to actual practices on a larger scale, suggesting the approach's effectiveness and potential for future application.
Research limitations/implications - This research provides a new solution approach for evaluating direct CO2 emissions of model different designs of distribution processes. Despite its narrow scope, the transportation sector has a significant environmental footprint, and offers the potential for substantial reductions in emissions. From modeling perspective, further research is suggested in integrating split delivery function without using dummy variables.
Originality/value - This paper contributes by identifying critical gaps in the understanding and implementation of system-wide efficient car distribution processes from distribution hubs to car dealers. It not only addresses potential improvements, but also proved efficiency gains of the system with a new solution approach, using a new combination of a state-of-the-art meta-heuristic and a proven split delivery method applicable for large-scale problems. ...
Purpose - With the short term need to reduce direct CO2 emissions of trucks in distribution processes, this paper aims to provide an easy to implement solution approach for distribution processes of new cars from holistic perspective. Scarce emphasizes is provided on short-term alternatives in truck-based distribution processes and approaches lacks applicable for large scale problems including split delivery function.
Design/methodology/approach - The distribution process and model methods are analyzed using a literature study and interviews with experts, resulting in the development of a solution approach. Combined with an extensive field research, a solution approach enables the performance evaluation of the current state, and the policy implications. Future designs are used to validate the solution approach by calculating performance differences in multiple relevant evaluation domains.
Findings - The analysis of the current state has identified critical bottlenecks, leading to the development of two promising policies. The application of a new and validated prioritization strategy and permitting more stops per truck has successfully yielded a significant reduction in CO2 emissions. The performance of the solution approach demonstrates high precision on a small scale and yields results comparable to actual practices on a larger scale, suggesting the approach's effectiveness and potential for future application.
Research limitations/implications - This research provides a new solution approach for evaluating direct CO2 emissions of model different designs of distribution processes. Despite its narrow scope, the transportation sector has a significant environmental footprint, and offers the potential for substantial reductions in emissions. From modeling perspective, further research is suggested in integrating split delivery function without using dummy variables.
Originality/value - This paper contributes by identifying critical gaps in the understanding and implementation of system-wide efficient car distribution processes from distribution hubs to car dealers. It not only addresses potential improvements, but also proved efficiency gains of the system with a new solution approach, using a new combination of a state-of-the-art meta-heuristic and a proven split delivery method applicable for large-scale problems.
Design/methodology/approach - The distribution process and model methods are analyzed using a literature study and interviews with experts, resulting in the development of a solution approach. Combined with an extensive field research, a solution approach enables the performance evaluation of the current state, and the policy implications. Future designs are used to validate the solution approach by calculating performance differences in multiple relevant evaluation domains.
Findings - The analysis of the current state has identified critical bottlenecks, leading to the development of two promising policies. The application of a new and validated prioritization strategy and permitting more stops per truck has successfully yielded a significant reduction in CO2 emissions. The performance of the solution approach demonstrates high precision on a small scale and yields results comparable to actual practices on a larger scale, suggesting the approach's effectiveness and potential for future application.
Research limitations/implications - This research provides a new solution approach for evaluating direct CO2 emissions of model different designs of distribution processes. Despite its narrow scope, the transportation sector has a significant environmental footprint, and offers the potential for substantial reductions in emissions. From modeling perspective, further research is suggested in integrating split delivery function without using dummy variables.
Originality/value - This paper contributes by identifying critical gaps in the understanding and implementation of system-wide efficient car distribution processes from distribution hubs to car dealers. It not only addresses potential improvements, but also proved efficiency gains of the system with a new solution approach, using a new combination of a state-of-the-art meta-heuristic and a proven split delivery method applicable for large-scale problems.
Despite being a long-cherished EU ambition and a crucial key to achieving climate goals, there is still no European High-Speed Rail (HSR) network, with the few completed projects often facing disappointing demand resulting in unprofitability. In order to gain insights to profitable network design, this study develops a new formulation to the “Transport Network Design & Frequency Setting Problem” (TNDFSP), as current literature lacks one that can optimally solve the problem for instances of this size while also accounting for demand elasticity. Our model works with elastic and dynamic HSR demand calculated from optimal fare settings and observed travel mode characteristics from competing alternatives. The optimal solution is largely insensitive to fare changes, and in some measures outperform current state-of-the-art. Our model considers the 111 most populous European cities, along with all their origin-destination (OD) pairs, and finds the most profitable network design within a reasonable solution time frame. The results show HSR can be very profitable in Europe, but only when concentrated around a selected group of the largest cities in the western part of the continent. Despite being profitable and contributing significantly to set Green Deal goals, the viability of our network heavily depends on the willingness of several countries to invest significantly in infrastructure construction.
...
...
Despite being a long-cherished EU ambition and a crucial key to achieving climate goals, there is still no European High-Speed Rail (HSR) network, with the few completed projects often facing disappointing demand resulting in unprofitability. In order to gain insights to profitable network design, this study develops a new formulation to the “Transport Network Design & Frequency Setting Problem” (TNDFSP), as current literature lacks one that can optimally solve the problem for instances of this size while also accounting for demand elasticity. Our model works with elastic and dynamic HSR demand calculated from optimal fare settings and observed travel mode characteristics from competing alternatives. The optimal solution is largely insensitive to fare changes, and in some measures outperform current state-of-the-art. Our model considers the 111 most populous European cities, along with all their origin-destination (OD) pairs, and finds the most profitable network design within a reasonable solution time frame. The results show HSR can be very profitable in Europe, but only when concentrated around a selected group of the largest cities in the western part of the continent. Despite being profitable and contributing significantly to set Green Deal goals, the viability of our network heavily depends on the willingness of several countries to invest significantly in infrastructure construction.
The distribution process of multi-dose vaccines
A robust optimization approach to supply uncertainty
The distribution chain of two-dose vaccines by an air carrier (KLM cargo) with its practical features is modelled to study the influence of supply uncertainty. First the goal is to find an efficient solution approach for the model which gives good quality solutions in reasonable computation time. Secondly with the developed solution method the supply uncertainty is taken into account using robust optimization. Solving the model using an exact solution method leads to a too large increase of computation time with increase of the model size for the purpose of robust analysis. Nevertheless, given the high reliability of the model the results of this commercial model are used as a reference. To circumvent the high computation time, two alternative solution methods are developed and implemented. Respectively the genetic algorithm and the rolling horizon method. A basic implementation of the genetic algorithm does not provide adequate results and gets trapped in a local optimum. An analysis shows that measures are needed to increase the flexibility and stimulate the algorithm to find so called “transaction-less” events. To achieve this, a toolbox is developed. The toolbox consists of analysis tools (measurement of convergence rate, sparsity and a diversity measurement) and tools intended to improve the convergence and accuracy of the solution. These latter are an adapted mutation operator that stimulates the number of transaction-less events (sparsity) and an approach for directed mutation. Each measure by itself has a positive effect on the initial convergence rate, but the algorithm still gets trapped at a somewhat improved local optimum at a level of approximately 10% above the optimum solution. A strong improvement is found by combining these measures of the toolbox resulting in solutions that approach the optimal solution within less than 5% for a single destination at a very high convergence rate. The best found combination of measures are implemented to solve a multi-destination problem. The results prove to be not as good as the result for the single destination: the gap to the optimal solution is roughly 10% and the convergence rate is somewhat slower. Probably this is due to the fact that the boundary conditions impose a reduction of flexibility for the multi-destination setting. On the other hand this finding might provide a base for worthwhile future work. Since this is solver is not (yet) suitable to be used in the robustness analysis. Three different implementations of the rolling horizon method were made: (i) the straight forward (myopic) approach, (ii) extending the time window with relaxed periods and (iii) a shifting rolling horizon. The straight forward approach is significantly improved by using the relaxed and shifting methods...
...
The distribution chain of two-dose vaccines by an air carrier (KLM cargo) with its practical features is modelled to study the influence of supply uncertainty. First the goal is to find an efficient solution approach for the model which gives good quality solutions in reasonable computation time. Secondly with the developed solution method the supply uncertainty is taken into account using robust optimization. Solving the model using an exact solution method leads to a too large increase of computation time with increase of the model size for the purpose of robust analysis. Nevertheless, given the high reliability of the model the results of this commercial model are used as a reference. To circumvent the high computation time, two alternative solution methods are developed and implemented. Respectively the genetic algorithm and the rolling horizon method. A basic implementation of the genetic algorithm does not provide adequate results and gets trapped in a local optimum. An analysis shows that measures are needed to increase the flexibility and stimulate the algorithm to find so called “transaction-less” events. To achieve this, a toolbox is developed. The toolbox consists of analysis tools (measurement of convergence rate, sparsity and a diversity measurement) and tools intended to improve the convergence and accuracy of the solution. These latter are an adapted mutation operator that stimulates the number of transaction-less events (sparsity) and an approach for directed mutation. Each measure by itself has a positive effect on the initial convergence rate, but the algorithm still gets trapped at a somewhat improved local optimum at a level of approximately 10% above the optimum solution. A strong improvement is found by combining these measures of the toolbox resulting in solutions that approach the optimal solution within less than 5% for a single destination at a very high convergence rate. The best found combination of measures are implemented to solve a multi-destination problem. The results prove to be not as good as the result for the single destination: the gap to the optimal solution is roughly 10% and the convergence rate is somewhat slower. Probably this is due to the fact that the boundary conditions impose a reduction of flexibility for the multi-destination setting. On the other hand this finding might provide a base for worthwhile future work. Since this is solver is not (yet) suitable to be used in the robustness analysis. Three different implementations of the rolling horizon method were made: (i) the straight forward (myopic) approach, (ii) extending the time window with relaxed periods and (iii) a shifting rolling horizon. The straight forward approach is significantly improved by using the relaxed and shifting methods...
In recent years, shared mobility systems have had a growing presence in cities all over the world. This is understandable given its numerous advantages such as the reduced need for personal vehicle ownership, reduced traffic congestion and emissions, increased parking efficiency, and cost savings for users. Overall, shared mobility systems offer the potential to revolutionize transportation, providing individuals with more options and helping to create more sustainable, livable cities. For shared mobility systems to fully deliver their benefits, vehicle availability must be maintained at the right place and time. If the vehicle distribution is not optimal, it may lead to overcrowding and shortages which in turn will discourage usage and lead to reduced revenues for the operator. Therefore, ensuring proper balancing of supply and demand is crucial for the success of the shared mobility service. One way to balance supply and demand is through physically rebalancing vehicles within the service area. In this study, a simulation-based optimization model is created and used to determine the optimal rebalancing operations while quantifying system improvement. A case study is conducted using real data from the Dutch moped sharing provider Felyx to examine the impact of performing rebalancing operations in Eindhoven throughout May ’22. The results demonstrate a potential increase in profit of up to 2.06%. By performing the recommended rebalancing actions several times a week in each city where the operator is active, a significant amount of extra profit can be made. This additional profit will even rise as the usage of shared mobility rises in general.
...
In recent years, shared mobility systems have had a growing presence in cities all over the world. This is understandable given its numerous advantages such as the reduced need for personal vehicle ownership, reduced traffic congestion and emissions, increased parking efficiency, and cost savings for users. Overall, shared mobility systems offer the potential to revolutionize transportation, providing individuals with more options and helping to create more sustainable, livable cities. For shared mobility systems to fully deliver their benefits, vehicle availability must be maintained at the right place and time. If the vehicle distribution is not optimal, it may lead to overcrowding and shortages which in turn will discourage usage and lead to reduced revenues for the operator. Therefore, ensuring proper balancing of supply and demand is crucial for the success of the shared mobility service. One way to balance supply and demand is through physically rebalancing vehicles within the service area. In this study, a simulation-based optimization model is created and used to determine the optimal rebalancing operations while quantifying system improvement. A case study is conducted using real data from the Dutch moped sharing provider Felyx to examine the impact of performing rebalancing operations in Eindhoven throughout May ’22. The results demonstrate a potential increase in profit of up to 2.06%. By performing the recommended rebalancing actions several times a week in each city where the operator is active, a significant amount of extra profit can be made. This additional profit will even rise as the usage of shared mobility rises in general.
Repositioning in shared mobility systems
Combining model predictive control and approximate dynamic programming
The global market for personal mobility has transformed over the last decade. Traditional taxi services have to compete with the emergence of ride-hailing services such as Uber and Lyft. Rapid developments in available algorithms and real-time inter-connectivity of travellers and vehicles offer mobility-on-demand (MOD) services new possibilities to maximise the efficiency of the ride-hailing process. It is well established that rebalancing can enable the true potential of a ride-hailing fleet. In this context, rebalancing is defined as the redistribution of idle vehicles over the service area of a ride-hailing operator. In the search for the optimal rebalancing strategy, model predictive control (MPC) and approximate dynamic programming (ADP) methodologies are active fields of research. However, the computational burden of MPC limits the length of the predictive horizon in large MOD applications. This thesis proposes a novel algorithm that combines ADP and MPC. More specifically, we integrated a value function into the MPC framework as a terminal cost. We shorten the horizon of MPC and propose to use the value function as a long-term planner. For the terminal cost, we continue research into piece-wise linear value function approximation. We implement a novel multi-period approximation of the value function, where we use the maximum repositioning length as the horizon. In our case study, the value-based repositioning strategy offers comparable service quality to conventional MPC at 5.2% of the computational burden. The hybrid ADP-MPC algorithm offers flexible horizon partitions between the multi-period value function and MPC. It addresses the shortcomings of both ADP and MPC algorithms in repositioning problems. At peak performance, it offers an increase of 4.5% in service rate and a decrease of 5.2% in the average waiting time over conventional MPC, at roughly half the computational burden. Keywords: Ride-hailing, Mobility on demand, Model predictive control, Approximate Dynamic programming
...
The global market for personal mobility has transformed over the last decade. Traditional taxi services have to compete with the emergence of ride-hailing services such as Uber and Lyft. Rapid developments in available algorithms and real-time inter-connectivity of travellers and vehicles offer mobility-on-demand (MOD) services new possibilities to maximise the efficiency of the ride-hailing process. It is well established that rebalancing can enable the true potential of a ride-hailing fleet. In this context, rebalancing is defined as the redistribution of idle vehicles over the service area of a ride-hailing operator. In the search for the optimal rebalancing strategy, model predictive control (MPC) and approximate dynamic programming (ADP) methodologies are active fields of research. However, the computational burden of MPC limits the length of the predictive horizon in large MOD applications. This thesis proposes a novel algorithm that combines ADP and MPC. More specifically, we integrated a value function into the MPC framework as a terminal cost. We shorten the horizon of MPC and propose to use the value function as a long-term planner. For the terminal cost, we continue research into piece-wise linear value function approximation. We implement a novel multi-period approximation of the value function, where we use the maximum repositioning length as the horizon. In our case study, the value-based repositioning strategy offers comparable service quality to conventional MPC at 5.2% of the computational burden. The hybrid ADP-MPC algorithm offers flexible horizon partitions between the multi-period value function and MPC. It addresses the shortcomings of both ADP and MPC algorithms in repositioning problems. At peak performance, it offers an increase of 4.5% in service rate and a decrease of 5.2% in the average waiting time over conventional MPC, at roughly half the computational burden. Keywords: Ride-hailing, Mobility on demand, Model predictive control, Approximate Dynamic programming
Learning-based path planning for automatic guided vehicles in container terminals
A case study at TBA Group
This thesis has provided insight into how machine learning can be beneficial to path planning in container terminals. Path planning algorithms can be used in environments with automated vehicles. A well known algorithm is the A* path planning algorithm, which is the fastest optimal path planning algorithm under satisfied conditions. However, the behaviour of a container terminal is unknown beforehand, costs can change over iterations. Therefore, Liu et al. [Liu et al., 2019] and Keselman et al. [Keselman et al., 2018] show the advantage of combining A* with Machine Learning. This way, the exploring part of the ML algorithm is combined with the fast andmore precise properties of the A* PP algorithm. This thesis has proposed the machine learning algorithm Vehicle Aware Reinforcement Learning Path Planning Algorithm VARLPPA. This algorithm uses Monte Carlo Control method. This is a model free approach, which has been shown in both experiments to find more efficient solutions in exceptional situations.
...
This thesis has provided insight into how machine learning can be beneficial to path planning in container terminals. Path planning algorithms can be used in environments with automated vehicles. A well known algorithm is the A* path planning algorithm, which is the fastest optimal path planning algorithm under satisfied conditions. However, the behaviour of a container terminal is unknown beforehand, costs can change over iterations. Therefore, Liu et al. [Liu et al., 2019] and Keselman et al. [Keselman et al., 2018] show the advantage of combining A* with Machine Learning. This way, the exploring part of the ML algorithm is combined with the fast andmore precise properties of the A* PP algorithm. This thesis has proposed the machine learning algorithm Vehicle Aware Reinforcement Learning Path Planning Algorithm VARLPPA. This algorithm uses Monte Carlo Control method. This is a model free approach, which has been shown in both experiments to find more efficient solutions in exceptional situations.
Short-Term Forecast of Demand for Train Station-Based Round-Trip Bikesharing
A Case Study of OV-fiets in The Netherlands
Around the world, authorities try to increase the attractiveness of multimodal public transport (PT)-related trips to reduce car usage. To achieve this, a seamless combination between the different modes is necessary. The Dutch train station operator NS tries to enhance the combination of the bike and train by providing a train station-based round-trip bikesharing (SBRT) scheme located at train stations throughout the country. This scheme allows users to rent a bike to connect the train station and their destination. The round-trip characteristic SBRT makes it unique in comparison to widely applied one-way bikesharing schemes. While on the latter a wide range of research exists, little research has been conducted on round-trip bikesharing, especially when being integrated into an existing public transport scheme. This paper aims to fill this gap by identifying potential temporal and weather-related determinants for SBRT-rentals of the Dutch SBRT-system OV-fiets using multiple linear regression (MLR). The results are compared with findings on one-way bikesharing schemes. The results are then used as an input to forecast short-term demand. To identify a best performing forecasting method, the statistical methods MLR and Prophet are compared with the neural-network based method Long Short-Term Memory (LSTM).
It is found that for hourly rentals in an SBRT-system, the highest explanatory power achieved with the number of train travelers leaving the corresponding train station, followed by temporal and weather-related determinants. Further, the magnitude of the correlation between the determinants and the hourly demand differs across the stations in the system. For forecasting, the performance of the methods differs across the stations and forecasted periods due to the stations' distinct characteristics. But, especially in times of uncertainty, LSTM is likely to outperform the others due to it's capability of adapting to short-term changes in the demand. ...
It is found that for hourly rentals in an SBRT-system, the highest explanatory power achieved with the number of train travelers leaving the corresponding train station, followed by temporal and weather-related determinants. Further, the magnitude of the correlation between the determinants and the hourly demand differs across the stations in the system. For forecasting, the performance of the methods differs across the stations and forecasted periods due to the stations' distinct characteristics. But, especially in times of uncertainty, LSTM is likely to outperform the others due to it's capability of adapting to short-term changes in the demand. ...
Around the world, authorities try to increase the attractiveness of multimodal public transport (PT)-related trips to reduce car usage. To achieve this, a seamless combination between the different modes is necessary. The Dutch train station operator NS tries to enhance the combination of the bike and train by providing a train station-based round-trip bikesharing (SBRT) scheme located at train stations throughout the country. This scheme allows users to rent a bike to connect the train station and their destination. The round-trip characteristic SBRT makes it unique in comparison to widely applied one-way bikesharing schemes. While on the latter a wide range of research exists, little research has been conducted on round-trip bikesharing, especially when being integrated into an existing public transport scheme. This paper aims to fill this gap by identifying potential temporal and weather-related determinants for SBRT-rentals of the Dutch SBRT-system OV-fiets using multiple linear regression (MLR). The results are compared with findings on one-way bikesharing schemes. The results are then used as an input to forecast short-term demand. To identify a best performing forecasting method, the statistical methods MLR and Prophet are compared with the neural-network based method Long Short-Term Memory (LSTM).
It is found that for hourly rentals in an SBRT-system, the highest explanatory power achieved with the number of train travelers leaving the corresponding train station, followed by temporal and weather-related determinants. Further, the magnitude of the correlation between the determinants and the hourly demand differs across the stations in the system. For forecasting, the performance of the methods differs across the stations and forecasted periods due to the stations' distinct characteristics. But, especially in times of uncertainty, LSTM is likely to outperform the others due to it's capability of adapting to short-term changes in the demand.
It is found that for hourly rentals in an SBRT-system, the highest explanatory power achieved with the number of train travelers leaving the corresponding train station, followed by temporal and weather-related determinants. Further, the magnitude of the correlation between the determinants and the hourly demand differs across the stations in the system. For forecasting, the performance of the methods differs across the stations and forecasted periods due to the stations' distinct characteristics. But, especially in times of uncertainty, LSTM is likely to outperform the others due to it's capability of adapting to short-term changes in the demand.
When traffic demand is high and the intersection lacks throughput, oversaturated traffic phenomena occur, such as the oversaturation of a turning bay or intersection link. When such unsafe situations emerge, control strategies can be applied to handle or avoid these spillback effects. In this paper, a spillback component is designed that detects and controls spillback in a real-time controlled traffic system (RTCS) that optimizes multiple vehicle types in its optimization, with a high vehicle priority for the light-rail vehicle (LRV). A spillback control strategy was developed that assigns additional priority to the congested link when spillback is detected. For assessment, a simulation study is conducted to gain insight into the effects of applying spillback control strategies on the traffic and network performance at an isolated intersection and in a corridor. The results showed that the intersection performance decreased when a spillback control strategy was applied in the optimization, but the network performance increased during the peak period. Furthermore, a spillback penalty factor effectively decreases spillback durations while it works well in combination with higher prioritized public transit. A too-high priority negatively impacts the intersection and network performance. However, in some cases, a (high) spillback priority is necessary to provide enough priority so that a growing unsafe situation is avoided. Therefore, it is crucial to determine the intended use and applicability when providing spillback priority. The spillback priority is one of the few priorities that can be configured in the utilized RTCS; it is recommended to research the impacts when varying priorities are used.
...
When traffic demand is high and the intersection lacks throughput, oversaturated traffic phenomena occur, such as the oversaturation of a turning bay or intersection link. When such unsafe situations emerge, control strategies can be applied to handle or avoid these spillback effects. In this paper, a spillback component is designed that detects and controls spillback in a real-time controlled traffic system (RTCS) that optimizes multiple vehicle types in its optimization, with a high vehicle priority for the light-rail vehicle (LRV). A spillback control strategy was developed that assigns additional priority to the congested link when spillback is detected. For assessment, a simulation study is conducted to gain insight into the effects of applying spillback control strategies on the traffic and network performance at an isolated intersection and in a corridor. The results showed that the intersection performance decreased when a spillback control strategy was applied in the optimization, but the network performance increased during the peak period. Furthermore, a spillback penalty factor effectively decreases spillback durations while it works well in combination with higher prioritized public transit. A too-high priority negatively impacts the intersection and network performance. However, in some cases, a (high) spillback priority is necessary to provide enough priority so that a growing unsafe situation is avoided. Therefore, it is crucial to determine the intended use and applicability when providing spillback priority. The spillback priority is one of the few priorities that can be configured in the utilized RTCS; it is recommended to research the impacts when varying priorities are used.
A real-time energy management system for a grid-connected solar park using an electrolyser in the Netherlands
Optimizing to maximize the revenue
This paper describes a real-time energy management system developed for a solar park in the Netherlands using an alkaline electrolyser. The optimization problem is split into a two-step optimization, taking into account the specifications of the electrolyser and allowing the electrolyser to respond to changes in the imbalance market. The first optimization step determines the state of the electrolyser one day in advance. The second optimization step determines the electrolyser power, using the state of the electrolyser as an input. Simulations using data from 2020, 2021 and 2022 show that the use of the electrolyser is limited to a number of days in the year with a lot of solar generation, causing the day-ahead prices to be low. Different scenarios have been tested to get insight into how the use of the electrolyser is influenced by these changes. The type of electrolyser, being able to put the electrolyser on standby and allowing the grid to be used for the electrolyser hardly affected the results. At last, different hydrogen prices are compared. The higher hydrogen prices lead to more use of the electrolyser. This real-time EMS contributes to the ability of an alkaline electrolyser to respond to the sudden changes in the grid and by doing this making it possible to use alkaline electrolysers for balancing the grid and contributes to the use of green hydrogen in the industry for a competitive price.
...
This paper describes a real-time energy management system developed for a solar park in the Netherlands using an alkaline electrolyser. The optimization problem is split into a two-step optimization, taking into account the specifications of the electrolyser and allowing the electrolyser to respond to changes in the imbalance market. The first optimization step determines the state of the electrolyser one day in advance. The second optimization step determines the electrolyser power, using the state of the electrolyser as an input. Simulations using data from 2020, 2021 and 2022 show that the use of the electrolyser is limited to a number of days in the year with a lot of solar generation, causing the day-ahead prices to be low. Different scenarios have been tested to get insight into how the use of the electrolyser is influenced by these changes. The type of electrolyser, being able to put the electrolyser on standby and allowing the grid to be used for the electrolyser hardly affected the results. At last, different hydrogen prices are compared. The higher hydrogen prices lead to more use of the electrolyser. This real-time EMS contributes to the ability of an alkaline electrolyser to respond to the sudden changes in the grid and by doing this making it possible to use alkaline electrolysers for balancing the grid and contributes to the use of green hydrogen in the industry for a competitive price.