ABSTRACT. The Strait of Hormuz carries roughly one fifth of global crude oil trade, yet how geopolitical shocks at the Strait propagate beyond directly exposed tanker routes remains poorly understood. This paper provides novel systematic evidence on the diffusion of Hormuz geopolitical risk through the tanker freight market. We define diffusion as the transmission of a localized geopolitical shock from directly exposed market segments to indirectly exposed segments, conditional on the absence of a common fundamental driver, and examine three dimensions, cross route within a vessel type, cross vessel between VLCC and Aframax markets, and cross tenor along the FFA term structure. The analysis covers six Hormuz related episodes spanning 2019 to 2026, with two Red Sea episodes (2023 to 2024) as placebo. We use daily Baltic Exchange TCE rates for four VLCC routes (TD3C, TD2, TD15, TD22) and two Aframax routes (TD8, TD14), together with TD3C FFA prices at four tenors (BOM, M1, M2, M3). Identification combines a market model event study with the Cowan (1992) generalized sign test, preferred over parametric alternatives because the high cross-sectional correlation among VLCC routes (ρ ̅= 0.746) attenuates parametric power to near zero. Mechanism inference uses subsample VARs estimated on normal period trading days. Episodes are partitioned ex ante by institutional severity, with low severity shocks (E2, E3) involving operational disruptions and high severity shocks (E6, E7) involving sanctions on VLCC port access. Three findings emerge. First, low severity Hormuz shocks generate systematic cross route diffusion within the VLCC segment, with exposed routes outperforming unexposed routes by a Cowan z statistic of 3.378 at CAR [0,5] (p < 0.001) and unexposed routes responding with a one-to-two-day lag. Second, this cross-route pattern disappears in high severity episodes (z = 0.875) and is replaced by extreme cross vessel diffusion, where exposed Aframax TD8 exceeds TD14 by Δ = 1.008 in the JWLA-033 episode. Third, the FFA term structure encodes shock persistence, with the M1-M3 slope steeply positive (0.184) in the 2019 operational shock and near zero (−0.007) in the 2025 sanctions escalation. Slope changes Granger cause future spot returns at the one-day horizon while the reverse does not hold, showing that the FFA curve leads the spot market. These results are consistent with a partially segmented markets framework in which two frictions, partial segmentation between vessel type markets and slow-moving capacity within each market, jointly determine the dimension of diffusion. Low severity shocks activate fleet reallocation across VLCC routes, while high severity shocks that eliminate VLCC port access force vessel substitution toward Aframax. A normal period VAR confirms that VLCC-Aframax coupling weakened sharply between the 2019 to 2022 and 2024 to 2026 subsamples (Granger F falls from 20.39 to 2.19). The paper shows that the dimension of geopolitical risk diffusion depends on the institutional character of the shock rather than its magnitude alone, and that the FFA term structure functions as a leading indicator of spot adjustment in this non storable freight market.

ABSTRACT. FuelEU Maritime introduces banking and borrowing of compliance balance as flexibility mechanisms to support compliance with greenhouse gas (GHG) intensity targets in maritime transport. However, their impact on shipping companies’ strategic fleet decisions has received limited attention in the literature. This paper develops a mixed-integer linear programming model for fleet renewal and deployment under FuelEU Maritime. The model incorporates compliance balance, banking, borrowing, penalties, ship purchases, and retrofit decisions. A case study of a 4,000-TEU container liner service on a trans-European route is conducted. The results show that the timing of fleet transition is primarily determined by the penalty level rather than by the availability of banking and borrowing. In all scenarios, diesel vessels are optimally retrofitted to bio-methanol-powered vessels, while borrowing is mainly used immediately before the retrofit decision to anticipate future compliance surpluses. The analysis also shows that large surpluses accumulated through banking before 2050 can provide a transition buffer for compliance after 2050, while also potentially delaying progress toward the long-term GHG intensity targets of FuelEU Maritime.

ABSTRACT. This study examines the applicability of a freight market forecasting method based on maritime big data to Panamax bulk carriers. The target variable is the grain freight rate for vessels of 65,000-100,000 DWT operating between the U.S. Gulf and Japan. Statistical data, including market indicators, commodity prices, and freight rates, were combined with AIS-based vessel movement data obtained from satellite and terrestrial sources. Seasonal changes in trade routes were also considered during data extraction. After missing values were imputed and the movement data were smoothed using a 3-day moving average, a multilayer neural network was developed for forecasting. The model was evaluated using out-of-sample data to assess its applicability beyond the training period. The results show that the model combining statistical data and vessel movement data achieved the highest accuracy in terms of both MDA and MSE. In contrast, models using only limited input data showed substantially lower accuracy. These findings indicate that integrating statistical data with vessel movement data improves forecasting accuracy in the Panamax bulk carrier market.

ABSTRACT. Maritime transport plays a central role in international trade, and ports serve as key interfaces linking global shipping networks with inland logistics systems. Trailers transporting maritime containers are therefore essential for connecting container terminals with clients in their hinterlands. However, few empirical studies have examined container-trailer mobility using large-scale trajectory data across multiple ports. This study analyses ETC2.0 data to investigate the mobility patterns of container trailers entering a container terminal at a port in the Tohoku region of northern Japan. Trailer trajectories derived from ETC2.0 data are used to identify stop locations and to construct tours that represent logistics activity cycles. Based on the constructed tours, key mobility indicators describing stop behaviour, travel distance, and travel time are analysed and compared across different ports in the region. The distributions of these indicators are further examined to reveal differences in operational characteristics among the ports. The results reveal systematic differences in trailer mobility patterns across ports and contribute to a better understanding of inland transport patterns of maritime containers and interactions between ports and their hinterlands.

ABSTRACT. Indonesia’s port sector is under growing pressure to become more efficient while also meeting environmental sustainability goals. This study explores the use of Systematic Layout Planning (SLP) to improve the spatial layout of New Priok Container Terminal 1 (NPCT1) in Jakarta, with the aim of reducing CO₂ emissions and improving terminal performance. By redesigning the layout and analysing cargo flow, the study compares the existing and proposed arrangements using average prime mover travel distance and estimated annual emissions as evaluation indicators. The results show that the average travel distance decreased by 15.75%, from 2.54 km to 2.14 km per trip. This operational improvement was accompanied by a reduction in estimated annual CO₂ emissions from 20.99 million kg to 18.94 million kg, equivalent to 9.97%. These findings suggest that layout optimisation can improve internal traffic flow, reduce unnecessary movement, and contribute to greener terminal operations. Overall, the study demonstrates the practical value of SLP as a planning approach for improving sustainability in container terminals.

ABSTRACT. Urban freight in megacity regions faces increasing complexity driven by labour shortages, rapid growth of e-commerce, and carbon neutrality requirements. This paper presents Tokyo FreightSim (TFS), an agent based urban freight simulation model for the Tokyo Metropolitan Area implementing five sequential modules of facility location choice, business to business (B2B) freight demand, business to consumer (B2C) parcel demand, delivery tour formation, and network assignment, and reports its validation against observed freight data. This paper further assesses input data availability required to build TFS for the purpose of deploying a similar model in international context. The result reveals that the establishment-level survey of Tokyo Metropolitan Freight Survey (TMFS) and the household-level survey of Personal Parcel Receipt Survey are essential for demand generation. Especially B2C parcel demand modelling requires integration of multiple datasets including weight conversion and item classification reconciliation because neither TMFS nor household respondents can fully characterise individual deliveries, which underscores the importance of conducting dedicated household-level parcel surveys alongside establishment-side freight surveys.

ABSTRACT. With the rapid expansion of the agricultural e-commerce market, short food supply chain models such as community group buying and online direct sales have increasingly emerged. Taking Meizao cherries as the research object, this study employs a discrete choice experiment and applies mixed logit and latent class logit models to measure consumer preferences for purchasing cherries through e-commerce-based short supply chains. The findings suggest that: 1) consumers’ preference for product attributes ranks, in descending order, as reputation, green certification, delivery time, place of origin, interaction mode, and price; 2) consumers with higher age and income levels show a stronger preference for next-day delivery, while those with higher awareness of short supply chains exhibit a stronger preference for high-reputation cherries; 3) based on preference heterogeneity, consumers can be classified into four segments: quality-oriented, efficiency-priority, environmentally conscious, and rational comprehensive types. This study provides practical implications for e-commerce platforms to optimize operational strategies for short-chain food products and offers policy insights for governments to promote the development of short food supply chains.

ABSTRACT. E-commerce has grown in the past decade, leading to changes within the logistics chain in urban areas. In Europe and the United States, there has been an annual increase of 15-20%, while in Japan, e-commerce market share in retail was around 6-8% during the start of the COVID-19 pandemic. This growth has led to an increase in urban freight activity. Fulfillment centers handle a large volume of shipments to be delivered to last-mile facilities or households. E-tailers are then motivated to locate these fulfillment centers close to the demand they serve to realize fast deliveries. However, high-density urban areas face constraints on land availability for logistics activities while also experiencing negative externalities associated with freight traffic. Despite its relevance in policy and research, there are limited studies to understand this relationship. E-commerce-driven demand has only recently been integrated into agent-based freight simulators, which traditionally consider business-to-business (B2B) commodity flows between firms and establishments. The growth of e-commerce has made it necessary to include business-to-consumer (B2C) commodity flows between households and microhubs. However, this interaction between e-commerce-driven demand and the spatial distribution of fulfillment centers has not been fully investigated. This study develops an integrated model of the e-commerce logistics chain, from supplier to fulfillment center to household, with an agent-based urban freight simulator based on the framework of SimMobility Freight (Sakai, et al, 2020). Using a synthetic region calibrated to the Tokyo Metropolitan Area, the impacts of fulfillment center relocation are evaluated. The results indicate that fulfillment centers in densely populated areas play a key role in e-commerce-driven freight traffic, since their absence leads to longer fulfillment center-to-microhub trips. These findings highlight the importance of proximity of fulfillment centers to the demand

ABSTRACT. Urban freight regulation and industrial spatial planning have long been hindered by decision blind spots due to the lack of precise freight industry profiles. Existing trajectory-based research fails to identify industrial functions from massive truck trajectories, restricting data-driven optimization of freight systems and urban economic activities. This study uses trajectory data from over 100,000 heavy-duty trucks to propose an industry identification method integrating trajectory semantics. We apply BERT to extract POI/AOI industrial semantic features, construct a multi-dimensional feature system, and develop a classification model combining BERT and XGBoost. The model achieves 87.42% overall accuracy and a macro-average F1 score of 0.8815, significantly outperforming other models. POI uniqueness and trip sequence features are crucial, especially the continuity of freight activities reflected in trip sequences. OD flow analysis reveals that Chongqing’s freight activities are highly concentrated in core areas and decline outward, consistent with local industrial distribution. This research provides an effective framework for extracting industrial semantics from freight trajectories and supports urban logistics optimization and management.

ABSTRACT. Demand-responsive transit (DRT) refers to a class of transport services whose routes can be flexibly adjusted according to passenger demand. A more advanced form of DRT is the flexible bus service, which allows real-time booking and dynamic route updates. Although such systems have attracted growing attention, existing studies mainly focus on the routing problem, with limited attention to how service plans and prices can be designed to influence the demand side. This paper proposes an incentive-based pricing framework for real-time flexible bus services, where passengers are offered multiple arrival deadline options and the operator uses prices to incentivize users to choose the option that is most beneficial to future system performance. The problem is formulated as an event-based discrete Markov decision process. A closed-form optimal pricing rule is derived, and a compact state aggregation scheme together with a simulation-based neural-network value function approximation is developed to estimate the potential value of each option. Numerical experiments on a scaled-down Sioux Falls network show that the proposed approach increases total reward while reducing average passenger delay and vehicle operating cost relative to benchmark policies. The results demonstrate the potential of pricing as an active demand-management tool in dynamic flexible transit operations.

ABSTRACT. Efficient school mobility is a critical component of suburban transport systems, particularly in developing countries where railway networks play a major role in daily commuting. While previous studies have examined suburban railway performance, limited research investigates the satisfaction of school commuters within developing-country railway systems, particularly in the Sri Lankan context. This study addresses this gap by analysing the determinants of satisfaction among school railway commuters in the Gampaha District, with specific attention to operational reliability, referring to consistent train punctuality and travel time stability. A structured survey of 397 school commuters was conducted. Reliability analysis was applied to confirm the internal consistency of measurement items, while the Relative Importance Index was used to prioritise key service attributes. Spearman correlation and multiple regression analysis were then employed to identify the most influential factors affecting commuter satisfaction. The findings reveal that punctuality and travel time efficiency are the dominant predictors of satisfaction. Passenger comfort and safety also demonstrate statistically significant but comparatively weaker effects. Demographic variables show no significant influence on satisfaction levels, indicating that reliability issues affect commuters consistently across user groups. Based on these findings, the study proposes AI-supported timetable optimisation as a potential approach to improve scheduling reliability and service stability. The research contributes practical evidence to support reliability-focused planning of suburban railway systems for time-sensitive educational mobility.

ABSTRACT. This paper proposes a bi-level optimization model to address interactions between a ridesharing service provider and two groups of commuters (i.e., car-owners and non-car-owners) who choose mode and departure time to pass a shared bottleneck. Building upon Vickrey's bottleneck model, the lower level addresses commuters’ choices among being solo drivers, ridesharing drivers, ridesharing passengers, and public transit riders under arbitrary pricing strategies of the platform. The ridesharing platform aims at maximizing its own profit by seeking the optimal price for ridesharing passengers and optimal wage for ridesharing drivers. This paper presents analytical monotonicity that captures how exogenous variables (e.g., commuter population size, and car ownership) impact the ridesharing platform's pricing strategies, profit, and market share. The analytical results suggest that commuters' mode choice and the platform's pricing strategies change significantly in the number of total commuters and the proportion of car-owners.The analytical results provide suggestions for pricing decisions of ride-sharing platform under different commuter structures.

ABSTRACT. Cities undergoing rapid urban growth and population ageing face increasing pressure to maintain equitable access to essential services for older adults. This study proposes a micro-scale evaluation framework based on accessibility, provision, and quality (APQ) to assess community-level medical and public transport conditions. Accessibility is measured by equivalent walking distance (EWD), which adjusts route length using segment-level impedance related to slope, surface condition, and barrier-free continuity. Provision and quality are used to capture facility capacity and service performance. After indicator standardisation, the three dimensions are integrated into a community-level evaluation system. Using Shenzhen as a case study, the results show clear cross-community differences and substantial within-community heterogeneity in both medical and public transport conditions. The findings suggest that approaches based only on proximity or facility counts may overlook important differences in service accessibility and performance.

ABSTRACT. Safe and accessible public transport infrastructure is an essential component of sustainable mobility systems, yet pedestrian safety at bus halts remains insufficiently addressed in many developing countries. This study evaluates the infrastructure conditions and pedestrian safety characteristics of bus halts located along the A9 highway corridor between Kandy and Jaffna in Sri Lanka. A total of 295 bus halts (79 urban and 216 rural) were analyzed using geospatial observations derived from Google Maps and Google Street View. The assessment considered several infrastructure elements including pedestrian crossing availability, the relative placement of crossings with respect to bus halts, passenger shelters, signage, and bus bay markings. Descriptive statistical analysis, Chi-square testing, K-means clustering, and a multinomial logit (MNL) model were applied to identify patterns in infrastructure provision and factors influencing pedestrian crossing placement. The results indicate that only 32% of bus halts provide designated pedestrian crossings, while approximately 70% lack passenger shelters, highlighting significant deficiencies in basic public transport infrastructure. Cluster analysis classified bus halts into three groups—high-risk, moderate-risk, and well-equipped facilities, with the majority of high-risk stops located in rural areas. The MNL model further shows that urban locations significantly increase the probability of pedestrian crossings being placed in front of bus halts (p = 0.007), whereas no significant relationship was observed for crossings located behind the halt. These findings reveal substantial inconsistencies in bus halt design and emphasize the need for standardized infrastructure guidelines, improved pedestrian crossing placement, and targeted safety improvements to enhance pedestrian safety and public transport accessibility along major transport corridors.

ABSTRACT. Capacitated Vehicle Routing Problem (CVRP) becomes complex following growth the number of delivery locations. However, most studies using Adaptive Large Neighborhood Search (ALNS) only report average solution quality and operator usage statistics, without explaining why certain operator combinations fail on specific instances. Until now, there has been no research that examines the characteristics of cases before ALNS optimization to predict which operators will be effective. This study to research how CVRP instance characteristics influence the performance of destroy and repair operators. Through experiments, the study analyzes the performance of commonly used operators across several benchmark CVRP instances. The analysis examines correlations between various destroy–repair operator pairs on instances with different characteristics, such as clustered versus dispersed node distributions, homogeneous versus heterogeneous demand patterns, and diverse depot configurations, represented through instance features. Furthermore, the prediction results are integrated into a rule-based ALNS framework and the findings indicate that operator effectiveness is correlated with specific instance features. Certain operator combinations consistently achieve high capacity utilization on clustered instances but fail to balance routes in dispersed distributions. Experiments are conducted on common some benchmarks of CVRLIb to predict the most influential operators. The results show a improvement success rate in operator prediction, and validation demonstrates that the rule-based ALNS achieves convergence on average 1.71 times faster than standard ALNS. These findings enable the development of operator selection rules based on instance characteristics.

ABSTRACT. This paper addresses the Heterogeneous Fleet Vehicle Routing Problem with Time Windows (HFVRPTW), a critical problem in last-mile delivery. Adaptive Large Neighborhood Search (ALNS) is a popular approach for such problems but often suffers from local optima entrapment. To overcome this limitation, we propose a novel ALNS framework incorporating temporary split delivery operators. Traditional operators maintain feasibility throughout the search. Our approach takes a different route. It temporarily generates split-delivery solutions to explore broader spaces, while ultimately returning a feasible non-split solution. The proposed algorithm is evaluated on Solomon-based heterogeneous fleet instances. Experimental results show that our algorithm achieves superior performance, outperforming the best-known solutions (BKS) on 73.2% of instances with an average gap of -3.08%, and surpassing existing literature results on 64.3% of instances with an average gap of -2.64%. Ablation studies confirm that the temporary split operators account for 2.28% of the 2.64% total improvement.

ABSTRACT. The joint order batching and picker routing problem (JOBPRP) is a combinatorial optimization problem arising in warehouse order-picking operations, where customer orders must be grouped into batches and the corresponding picking routes must be determined to minimize the total travel distance. This paper investigates the JOBPRP in a multi-block warehouse environment and develops an integrated framework, namely ALNS-ACA-CA. In the proposed framework, an association-based clustering approach (ACA) is first employed to generate high-quality initial batching solutions by improving the spatial compactness of orders. Then, the combined algorithm (CA) is adopted to evaluate the picking route length of each batch in a multi-block warehouse. Based on this routing evaluation, an adaptive large neighborhood search (ALNS) procedure with multiple destroy and repair operators is designed to iteratively improve the batching solution. Computational experiments on a representative instance show that the proposed framework achieves the shortest total picking distance among the compared algorithm combinations. The results also indicate that the association-based batching strategy can effectively reduce unnecessary travel and that batching and routing decisions are strongly coupled in the JOBPRP.

ABSTRACT. The single container loading problem (CLP) is a three-dimensional packing problem which has to fill the container with a set of boxes. The objective is to maximize the space utilization of the container. In this work, multi-population biased random-key genetic algorithm (BRKGA) and a tree-search-based method (lookahead mechanism) are combined. To evaluate the efficiency of box arrangement strategies including layer and block when dealing with differently heterogenous box instances, original BRKGA algorithm is used to test on 1500 classical instances. With the pattern above, we test our hybrid algorithm adopting layer strategy on instances BR1-7 and observe that the performance of the algorithm outperforms than other competitive algorithms but a little disadvantage in some instances compared with the state-of-the-are beam search algorithm.

ABSTRACT. Aircraft maintenance scheduling has become increasingly difficult as airline operations grow more dynamic and resource-constrained. Traditional manual scheduling approaches often struggle to coordinate maintenance tasks under overlapping time windows, resource conflicts, task priorities, and operational uncertainty, resulting in reduced efficiency and limited adaptability. This paper presents the design and implementation of an intelligent aircraft maintenance scheduling system that transforms maintenance planning from a manually coordinated process into a dynamically optimized decision process. The proposed system integrates constraint-based planning with multi-objective optimization to generate feasible and efficient maintenance schedules while balancing operational performance and safety requirements. Hard constraints, including crew availability and maintenance deadlines, are strictly enforced, while soft constraints are incorporated to improve schedule quality and reduce operational conflicts. To support practical deployment, the system adopts a modular architecture consisting of task management, optimization, and visualization components, enabling real-time schedule adjustment and intuitive decision support. Simulation experiments demonstrate that the proposed approach improves maintenance completion rates and resource utilization under complex operational conditions. The study suggests that intelligent scheduling can move aircraft maintenance from reactive coordination toward proactive and adaptive operational management.

ABSTRACT. The closure of the Cambodia-Thailand border in July 2025 severely disrupted logistics between the two countries. In 2024, land transport had carried about 70% of total 6.07 million tons and $4.8 billion in trade with Thailand, forcing a reconfiguration of Cambodia’s logistics network.

This study analyzes how logistics with Thailand shifted to alternative routes in response to the border closure, based on data from the Sihanoukville Autonomous Port (PAS) and the General Department of Customs and Excise (GDCE) of Cambodia. The results revealed a sharp increase in cargo volume handled at Sihanoukville Port. Furthermore, due to connectivity enhancement in the region driven by the Greater Mekong Subregion (GMS) initiative, the use of the Thailand-Laos-Cambodia route and even the Thailand-Laos-Vietnam-Cambodia route were also observed. This has highlighted the critical need to establish redundant transport routes.

However, the shortest Thailand–Laos–Cambodia route crosses the Thailand–Laos border (Chong Mek/Vang Tao) and the Cambodia–Laos border (Trapeang Kreal/Nongnokkheane). The former is not designated as a cross-border point under the GMS-CBTA. As a result, Laos has levied taxes on transit cargo from Thailand bound for Cambodia, revealing the institutional barriers to seamless regional connectivity.

ABSTRACT. This study conducts a scenario analysis of rail container transport between China and Europe using a global logistics intermodal network simulation (GLINS) model that accounts for differences in cargo types and transit cargo characteristics. The model used in this study is a modified version of the conventional GLINS model, with changes made to its network structure and computational process. The simulation model we developed accurately replicates the distribution of cargo volumes across the global maritime and long-distance land transport networks in Eurasia. Using the simulation model we developed, we conduct a scenario analysis of the subsidy system implemented for the China Railway Express—a rail transport service between China and Europe—and the effects of infrastructure investments in Kazakhstan. The model’s results suggest how reductions in transport costs and travel times influence the uptake of transcontinental transport services.

ABSTRACT. Thesis Abstract

Rise of reusable launch vehicles and availability of orbital refuelling infrastructure can enable ultra-fast transport of high-value, time-critical freight through suborbital space between any two locations on Earth in under two hours (Walton & Goehlich, 2022). The ability to ship goods from Frankfurt to Melbourne in a fraction of current time could radically transform the way we move goods. A key advantage is reaching microgravity beyond Karman line (100km) for reduced air resistance enabling 35 times higher velocity of 7.8km/s compared to conventional freighter jets travelling at 800km/h. Under poor weather conditions such as tornadoes, floodings and wildfires that disable freight movement through traditional modes of transport, the suborbital logistics corridors can enhance synchromodality and deliver 10x reduction in travel time. For high-value and time- sensitive freight segments like vaccines, critical spare parts, military assets, or emergency aid the speed could be a game changer. However, a successful setup of these operations require well-connected multimodal logistics networks that can link suborbital launch sites with existing gateways and hubs (Santoro et al., 2014). This research aims to investigate if such suborbital transport lanes are feasible and whether they can be integrated into global multimodal logistics networks to maximize operational efficiency.

While the technological feasibility of suborbital spaceflight is rapidly maturing, there is a gap in the literature and practice on how this new mode of transport can be connected to ground-based logistics networks for cargo transfer of door-to-door shipments. Key issues include ground handling operations, linehaul at terminals, alignment with first- and last-mile delivery modes and routes, and operational synchronization with existing infrastructure. The central research question is whether its economically and practically feasible to introduce high-speed transport corridors via suborbital space and how can these transport nodes be effectively integrated into existing multimodal logistics networks to improve end-to-end delivery performance for high-priority and time critical shipments?

In terms of key objectives, the research aims to: - Develop a comprehensive multimodal logistics network that combines existing logistics infrastructure (air-, rail- , seaports) with suborbital transport corridors; - Identify optimal network design and routing strategies by analysing high-volume cargo lanes and hubs; - Distinguish which industry segments (semiconductors, automotive etc.) and freight types (e.g. dangerous goods) are best positioned to benefit from suborbital transport links; - Simulate operational scenarios to assess how the integration of suborbital transport corridors affect delivery performance and test network resilience using discrete-event simulation.

The study focuses on constructing a hybrid logistics network model including suborbital transport nodes as ultra-fast cargo lanes, connected to traditional air, land, and sea terminals. A Mixed-Integer Linear Programming model is planned to be used to solve routing and network design optimization problem under varying cost, time, and capacity constraints. Discrete-event simulation tools like AnyLogic and SimPy are considered to be used to model terminal operations, delays and queueing, ground handling and cargo transfer processes, measuring performance under different environmental conditions and disruption scenarios.

To summarize, the research aims to deliver a framework for integration of suborbital cargo nodes into pre- existing logistics systems, including optimized routing strategies, infrastructure design recommendations, and logistics performance benchmarks. The findings aim to guide both policymakers and logistics service providers in preparing for a future where space is not only a destination, but a high-speed transport corridor.

References: [1] Santoro, F., Bellomo, A., Bolle, A., & Vittori, R. (2014). The Italian spacegate: Study and innovative approaches to future generation transportation based on high altitude flight. Acta Astronautica, 101, 98–110. http://dx.doi.org/10.1016/j.actaastro.2014.03.020 [2] Walton, R. O., & Goehlich, R. A. (2022). SPACE CARGO: Ultra-fast Delivery on Earth –Potential of Using Suborbital Space Vehicles for the Transportation of Cargo. International Journal of Aviation, Aeronautics, and Aerospace, 9(1). https://doi.org/10.15394/ijaaa.2022.1671

ABSTRACT. Cross-border transport of the medical products runs into the stochastic clearance delays at the border checkpoints, and the multi-stage inspections generate delay distributions whose right tails outrun the cold-chain tolerance window. When the clearance delay exceeds the thermal viability limit of the shipment, the domestic routing models become infeasible. The bottleneck sits at the border, not on the domestic legs of the route. We develop a compliance-aware framework around a deterministic finite automaton (DFA) for the clearance sequence, coupled with a two-stage stochastic mixed-integer linear program that carries a CVaR 0.95 tail-risk penalty. The border delays are calibrated through a five-stage WCO Time Release Study decomposition. We introduce the Border Cold-Chain Transfer Pod (BCTP), a buffer that is installed at the checkpoint and that pauses the cold-chain degradation clock. The pod investment dominates the domestic transport as the risk-control lever. The framework is instantiated on the Heihe-Blagoveshchensk corridor, where the RhD-negative blood prevalence is forty times higher on the Russian side. Experiments across three border regimes reveal a sharp summer-to-winter UAV switching pattern that is driven by the battery derating, with an EVPI of $4,770 per day under the disruption regime.

ABSTRACT. Against the background of global project-based development, the Project-Driven Supply Chain (PDSC) serves as a critical carrier for integrating global resources and supporting the achievement of project objectives. PDSC is characterized by temporality, diverse stakeholders, and complex supply chain network structure, and its risks can easily spread across network levels, resulting in a ripple effect. While existing studies have explored the ripple effect, and find that they can significantly impair network resilience and project delivery, these researches are mostly in a two-level network context and lack quantitative calculation of inter-project coordination. Thus, their findings cannot be directly applied in the context of PDSC, which has cross-level risk transmissions and inter-project coordination. Therefore, this paper focuses on the cross-level ripple effect of PDSC risks, exploring its impact on network resilience and moderating role of inter-project coordination. Based on synergy theory, this paper first introduces the Copula function to quantify the coordination relationships among stakeholders at the project level. Second, a cross-level Bayesian network model is constructed reflecting the multi-level network characteristics of PDSC. Layers of the raw material supplier, manufacturer, core supplier, project, and strategic objective are incorporated to analyze the transmission paths of risks within the network. Finally, the effectiveness of the model is verified through model comparison, resilience assessment, and ripple effect analysis. Key risk nodes are identified, and the moderating effect of inter-project coordination on risk transmission is revealed. The findings show that the negative impact of node disruption on the probability of achieving strategic objectives in the cross-level model is significantly higher than the non-cross-level model, verifying the existence of the risk ripple effect. The identified key risk nodes are predominantly core stakeholders with close cross-level connections and high coordination intensity, whose disruption triggers rapid chain reactions. Risks propagate layer by layer along PDSC paths, and inter-project coordination exerts a two-way moderating effect, that is high coordination accelerates risk diffusion, while a reasonable coordination mechanism helps mitigate negative impacts and enhance network resilience. Compared with traditional two-level models, the proposed cross-level Bayesian network model has significant advantages in resilience assessment accuracy and scenario adaptability. Theoretically, this study enriches the research of PDSC risk management by focusing on the risk ripple effect on network resilience and the key role of inter-project coordination in mitigating it. Practically, it provides PDSC stakeholders with practical guidance to identify risk nodes, optimize coordination mechanisms, and evaluate network resilience. The findings can help improve the stability and risk resistance of supply chains in the context of global project-based development.

ABSTRACT. This study examines inventory strategy changes in the Japanese manufacturing sector during the post-pandemic period. Using firm-level panel data on Japanese listed manufacturing firms from fiscal year 2015 to 2024, we document a significant and persistent increase in inventory days following the COVID-19 pandemic. Fixed-effects models reveal substantial within-firm inventory growth, suggesting strategic adjustment rather than compositional shifts across firms. Difference-in-differences analysis reveals heterogeneous effects: firms with higher overseas business exposure, as measured by pre-pandemic overseas sales ratio, show significantly stronger increases in inventory levels. Event study estimates show no evidence against parallel pre-trends and reveal that the heterogeneous effect of overseas exposure peaked in FY2021, remained elevated through FY2022-2023, and converged by FY2024. Pre-trend analysis documents a 2.1% annual inventory trend prior to COVID-19; after controlling for this pre-existing trend, the net post-pandemic level shift remains 5.2%. Inventory composition analysis shows that the increase is most pronounced in raw materials, particularly in the electronics and electrical equipment industry, consistent with upstream safety stock accumulation. The relationship between inventory levels and firm value (Tobin's Q) does not show significant change in the post-pandemic period. Overall, this study provides firm-level empirical evidence consistent with a persistent shift from just-in-time toward just-in-case inventory strategies following disruptions in international supply chains.

ABSTRACT. Designing a reliable humanitarian relief logistics network (HRLN) faces severe challenges caused by the extreme unpredictability of natural disasters regarding their timing, location, and magnitude. After a disaster occurs, the simultaneous transportation of relief supplies and evacuees often severely congests the limited logistics infrastructure, which may also suffer from disruptions. In order to mitigate these surface transport bottlenecks, this study introduces a truck-drone collaborative system for the efficient distribution of emergency and basic supplies. This paradigm shift drives the integration of crucial decisions in HRLN design: pre-disaster emergency facility location, relief inventory pre-positioning, and drone fleet deployment, alongside post-disaster human evacuation planning and coordinated truck-drone transportation. We mathematically frame this integrated issue as a two-stage distributionally robust optimization (DRO) model, dividing the timeline into pre- and post-disaster phases. An ambiguity set relying on the type-1 Wasserstein distance and historical observations is constructed to properly characterize the probability distributions of evacuee volumes and available road capacities. The proposed model is then transformed into a computationally tractable equivalent suitable for Benders decomposition algorithms. A practical case study based on the Chinese Yushu earthquake is examined, and multiple managerial implications for integrated truck-drone HRLN operations are proposed.

ABSTRACT. Cambodia is experiencing remarkable economic growth, and port development by both the public sector and private capital is highly active. By having the government indicate future nationwide cargo demand and investment volumes to facilitate appropriate adjustments, it is possible to prevent haphazard development and ensure orderly port growth. To this end, the authors collaborated with the Cambodian Ministry of Public Works and Transport to forecast nationwide container demand and estimate container generation and concentration volumes across six divided zones in Cambodia. Based on these findings, the required scale of terminal development for each zone was estimated. Additionally, the study estimated changes in container flow resulting from the development of the Funan Techo Canal. This paper introduces these estimation methodologies and their results. It also summarizes key points of discussion encountered during the estimation process, identifies the limitations of current container cargo demand estimation methods, and aims to contribute to the improvement of future demand estimation methodologies.

ABSTRACT. With the development of global economy and social activities, the production of organic waste continues to rise, characterized by diverse sources and complex composition, posing significant challenges to environmental pollution control and resource recycling management. Concurrently, the construction industry, particularly concrete production, is among the most resource-intensive sectors, contributing substantially to global environmental challenges. Concrete, the most widely used construction material ,has a substantial environmental impact due to its high carbon emissions and extensive resource consumption. Driven by global sustainable development goals and circular economy principles, the convergence of concrete's environmental impact and the increasing burden of organic waste management presents a unique opportunity for sustainable innovation. Incorporating biochar derived from organic waste into concrete production offers a dual environmental benefit: reducing the carbon footprint of concrete while diverting waste from landfills. This study has aimed to construct a multi-period and multi-echelon reverse supply chain network by integrating various nodes and introducing the aforementioned innovative process, so as to reduce costs and carbon emissions, and enhance resource utilization.A multi-objective optimization model based on mixed-integer linear programming (MILP) is proposed to formulate the problem. Then, we resort to the Gurobi optimization solver. The finding results indicate that, in contrast to the conventional concrete supply chain without organic waste integration, the proposed multi-period, multi-echelon reverse supply chain network, utilizing coffee grounds as a representative organic waste resource, achieves enhanced performance in both economic and environmental dimensions.This research provides theoretical support for synergistically addressing waste management and low-carbon construction, with potential for incorporating more complex factors to enhance the model's practical value in future studies.

ABSTRACT. This study examines how logistics platforms in Japan coordinate unintegrated logistics resources within Japan’s logistics structure amid labor shortages and rising demand for more sophisticated logistics services. In mature B2B logistics sectors, core flows remain largely organized by incumbent logistics firms and 3PL providers. In segments dominated by small and medium-sized enterprises, however, warehousing capacity, trucking capacity, local delivery capacity, and cross-firm surplus resources remain widely dispersed. From a transaction cost perspective, the paper argues that logistics platforms create value not by replacing the existing system, but by aggregating and coordinating unintegrated resources through matching, standardization, visibility, human adjustment, and data accumulation. A comparative analysis of six Japanese cases examines how platforms coordinate unintegrated resources, which coordination functions platform routines can standardize and digitalize, and how platforms complement rather than replace incumbent firms. The findings indicate that platform mechanisms support tracking, OCR, document matching, and preliminary decision support, whereas exception handling, responsibility allocation, quality assurance, and relationship management still depend on human intervention. The study concludes that logistics platforms in Japan function primarily as complementary coordination mechanisms whose longer-term significance lies in data accumulation and the emergence of partly 4PL-like coordination.

ABSTRACT. This study develops a PI simulator incorporating spatio-temporal freight information, applied to the greater Tokyo metropolitan area. Scenario analysis across hub count, placement distribution, freight volume, handling time, and delivery deadlines reveals that geographically dispersed hub placement minimizes total cost, diminishing returns become prominent as hub count increases up to 25 hubs, and reducing handling time is the most impactful operational lever.

ABSTRACT. Returnable transport items (RTIs) are vital for supply chain efficiency, yet systems often suffer from spatiotemporal imbalances and insufficient closed-loop reprocessing coordination. This study investigates a multi-period closed-loop supply chain for shared transport containers in the Physical Internet (PI) context, and proposes an integrated production-inventory-routing problem (PI-PIRP) model, where production is endogenized as centralized container cleaning. To solve this NP-hard problem, we propose a multi-stage heuristic algorithm that serves as a PI-enabled coordination mechanism, synchronizing reprocessing, inventory collaboration, and transportation integration through functionally coupled decision modules. Numerical experiments across multiple scales verify the model's logical consistency and evaluate operational trade-offs. Sensitivity analyses quantify the critical triggering effect of safety stock thresholds on procurement costs, the logistical efficiencies derived from vehicle capacity expansion, and the substitution effect between cleaning throughput and external procurement. This study provides valuable insights for the collaborative management of returnable transport items in interconnected logistics webs.

ABSTRACT. High-density Puzzle-Based Storage Systems (PBSS) face severe routing complexities. Traditional exact models (NIPA) suffer from the dimensionality curse, while state-relaxed models (NIPF) experience catastrophic state amnesia. To break this bottleneck, we propose the Interference-Aware Rolling Horizon Batch (IA-RHH) framework. Spatially, an IA-A* engine with dynamic bounding-box penalties steers searches away from congestion. Temporally, a batch-based decoupling strategy with an innovative Soft-Failure mechanism enforces precise physical state inheritance, transforming intractable deadlocks into asynchronous rescheduling opportunities. Extensive simulations reveal a critical performance inflection point. Starting from the 8*8 scale, IA-RHH shows decisive superiority. At extreme 10*10 congestion, it secures higher fulfillment rates and drastically slashes solving time compared to the NIPA baseline, providing a computationally efficient and resilient algorithmic foundation for next-generation autonomous warehousing.

ABSTRACT. Industrial production planning is a core decision-making problem in supply chains and is typically formulated as a large-scale mixed-integer linear programming model. In multi-plant, multi-period, and multi-level production environments, practical instances may contain tens of millions of decision variables and constraints due to detailed modeling of production, procurement, transportation, delivery, capacity limits, and bill-of-material dependencies. Although the Relax-and-Fix heuristic is widely adopted in industrial practice, its linear relaxation phase and repeated subproblem solving remain major computational bottlenecks for problems of this scale. To address this challenge, this paper proposes a graph-neural-network-assisted learning-to-optimize framework for large-scale industrial production planning. The proposed framework integrates a lightweight aggregated graph neural network with the Relax-and-Fix heuristic to identify decision variables that are likely to remain inactive in the final solution and fix them to zero before optimization. To enable scalable learning, we construct a heterogeneous graph that captures item-level interactions among production, inventory, delivery, replacement production, and capacity while aggregating plant and time dimensions. The resulting compact graph representation preserves the hierarchical structure of the bill of materials and is further enriched by static attributes and historical optimization data, allowing the model to learn recurring structural patterns in rolling-horizon planning. Numerical experiments on real-world industrial datasets show that the proposed framework can substantially reduce the optimization dimension and solver effort while maintaining high solution quality. In particular, the method consistently decreases linear-programming solving time under different pruning thresholds, with up to about forty percent reduction and less than one percent loss in demand fulfillment. Additional sensitivity analyses on pruning thresholds and loss-function configurations further verify the robustness and flexibility of the framework. Overall, the proposed approach provides a scalable and practical solution for large-scale industrial production planning and offers a general paradigm for graph-neural-network-assisted dimensionality reduction in mathematical optimization.

ABSTRACT. The rapid growth of e-commerce has driven firms to redesign distribution networks to better balance cost efficiency and operational flexibility. This paper studies the Category-Assignment Multi-Echelon Distribution Network Design Problem (CA-ME-DNDP) that integrates facility specialization and flexible regional distribution center (RDC) activation in a multi-period setting. In the proposed framework, each central distribution center (CDC) is dedicated to a specific product category to capture heterogeneous storage and handling requirements, while RDCs can be selectively activated to enhance responsiveness. We first formulate an arc-based model that jointly determines category assignment, facility activation, and multi-period transportation decisions. By exploiting the underlying network structure, we further propose a path-based formulation that yields a more compact representation. To solve large-scale instances, we design a tailored selective-intensification adaptive large neighborhood search (SI-ALNS), which enables efficient generation of high-quality solutions. Extensive computational experiments show that selective RDC activation outperforms mandatory full activation, reducing total cost by 7.46\% on average. The proposed multi-echelon structure also significantly improves performance over pure direct delivery, achieving a 9.99\% cost reduction. Sensitivity analyses further reveal that network configuration is governed by the interplay between transportation economics and spatial scale, which endogenously determines the transition between centralized and decentralized distribution structures.

ABSTRACT. High-value imported fresh produce is increasingly distributed through dual-track cold chains that combine air freight for the early premium market with sea freight for the later market. This structure creates intertemporal demand cannibalization and capacity rationing challenges. We develop a two-period Stackelberg game with strategic consumers and endogenous retailer preservation effort to examine pricing, preservation investment, and fulfillment allocation under sufficient supply, priority fulfillment, and deferred fulfillment. Results show that freshness decay and preservation efficiency have asymmetric effects: a higher freshness discount compresses system-wide profits, whereas higher preservation efficiency increases value under sufficient supply. Under binding capacity, however, preservation efficiency may become a trap by intensifying strategic waiting and reducing total profit. The optimal fulfillment regime shifts from deferred to priority fulfillment once capacity exceeds the critical threshold.

ABSTRACT. In a context where the speed and responsiveness of warehouse systems are becoming increasingly important, rack-Climbing Storage and Retrieval (CR/SR) systems are emerging as a promising solution for high-density automated logistics. These systems utilise a fleet of rack-climbing robots capable of combining horizontal and vertical mobility to perform retrieval and storage operations in warehouses for bin storage, where human workers carry out the picking activity to fulfil customer orders. This article develops an analytical model based on queuing theory, which considers the behaviour of both picking stations and the rack-climbing robot fleet, in order to investigate how the layout affects performance, waiting times, the average mission lead-time, and the utilisation rates of both robots and picking stations; the model then analyses the stability of the system as the number of rack-climbing robots employed varies. A series of warehouse configurations is analysed as both storage capacity and the aspect ratio of the warehouse vary, in order to isolate the impact of the layout, indicating that geometry plays a central role in determining the efficiency of the system. The proposed approach provides a useful model for comparing design alternatives in the early stages of development.