How organizations view “time” shapes how they coordinate and utilize operational resources. This study examines how proactive temporal framing, treating future contingencies as requiring present preparation and coordination, is associated with more efficient operational resource utilization. This study uses linguistic future-time reference (FTR) as a language-based proxy for this framing: weak-FTR languages express future events in present-tense forms, making the future feel psychologically closer. Using global airline data from 1994 to 2012, this study finds that airlines in weak-FTR contexts show higher efficiency in resource utilization. This association is stronger when more codeshare partners also come from weak-FTR contexts, although the moderation evidence is context-dependent. The study reframes the trade-off between future orientation and efficiency around operations and identifies temporal alignment as a boundary condition under temporal complexity.
As air traffic operations grow increasingly complex, traditional methods for evaluating air traffic controllers (ATCOs) face challenges of subjectivity and limited standardization. This study investigates the use of large language models (LLMs) to assist instructors by drafting candidate feedback text for instructor-in-the-loop review, rather than autonomously judging ATCO performance. A structured keyword-sentence parallel corpus is developed by extracting key terms from historical evaluation comments using TF-IDF and aligning them with representative sentences based on cosine similarity. A locally deployed LLM is fine-tuned to generate complete, keyword-conditioned feedback sentences from keyword inputs, enabling more standardized and scalable evaluations. To enhance output quality, a hybrid text generation strategy combining Top-k, Top-p, and temperature tuning is applied, balancing fluency, relevance, and diversity. Furthermore, an interactive evaluation platform is built to support drafting, instructor review, and manual editing, ensuring human oversight for atypical or safety-critical cases. In addition, privacy protection is treated as a deployment constraint rather than a primary contribution, and de-identification and controlled access are applied in the data and system pipeline. Experimental results, including comparisons with zero-shot and few-shot GPT-4 API baselines, indicate that the proposed model achieves stronger alignment with standardized ATC feedback conventions while supporting scalable instructor-in-the-loop drafting. This approach offers a practical and extensible solution for intelligent ATCO feedback drafting assistance and contributes to the broader integration of artificial intelligence (AI) into aviation training systems.
Airline baggage policies generally rely on uniform weight allowances combined with excess-baggage fees. From a payload-allocation perspective, this design may obscure the scarcity value of the aircraft payload and generate implicit incidence patterns across passengers. We study a Total Weight Allocation (TWA) benchmark in which passengers receive tradable baggage rights derived from a fixed per-passenger payload allowance net of body weight and trade these rights in a competitive market. When the payload capacity binds, the TWA implements a payload-efficient allocation by equalizing the marginal valuations of discretionary baggage capacity, exactly under quasi-linear preferences. Efficiency is independent of the initial assignment of rights, whereas incidence depends on endowments: conditional on baggage demand and travel needs, lower-body-weight passengers tend to be net sellers and higher-body-weight passengers tend to be net buyers. A non-tradable seating baseline preserves feasibility without changing the tradable baggage margin. The theoretical results extend to convex weight-cost formulations and multidimensional payload constraints. An illustrative numerical exercise based on assumed parameter distributions reports model-implied efficiency gains relative to non-tradable uniform baggage allowances and the transfer patterns implied by tradable rights; it is not an empirical calibration or a route-specific prediction. A per-kg fee equal to the payload shadow price can replicate the benchmark trading allocation while transferring scarcity rents from passengers to the airline, highlighting the efficiency–incidence trade-off in baggage-policy design.
This qualitative study explains how airline cabin crew reconcile a service identity with safety enforcement during in-flight disturbances. Using a grounded theory design, we conducted two in-depth interviews with thirty female crew members at a South Korean full-service carrier. We analyzed transcripts via open, axial, and selective coding until theoretical saturation. Findings show that service-first display rules, weak recognition of safety authority, uneven sanctions, complaint-weighted evaluations, and ambiguous manuals heighten anxiety and role conflict. We theorize a sequential mechanism in which appraisal alters perceived control, and institutional signals channel incidents into two reproducible pathways, namely Path A, responsibility avoidance (responsibility shifting, symbolic performance, feigned ignorance, indifference), and Path B, uncertainty reduction (manual-based performance, systematic record-keeping, peer coordination, reliance on cockpit authority). We formalize this as the Uncertainty Avoidance–Service Safety Conflict (UASSC) model, reframing uncertainty avoidance from a national trait to an enacted micro-mechanism mediated by organizational rules and evaluation systems. Practical levers that reliably shift decisions from Path A to Path B include increasing the safety weight in complaint systems, improving manual clarity and access, making leadership backing visible at the point of action, digitalizing contemporaneous records, and strengthening legal and media deterrence. The model yields testable propositions and a practice-ready template for managing safety enforcement in service-intensive, safety-critical settings.
To investigate the impact of rainy weather on approach operations in the terminal maneuvering area (TMA), a multi-dimensional trajectory data analysis framework is proposed. The Air Traffic Management Airport Performance (ATMAP) algorithm is combined with Meteorological Terminal Aviation Routine Weather Reports (METAR) and Automatic Dependent Surveillance-Broadcast (ADS-B) trajectory data to filter approach trajectories under no-weather and rainy-weather conditions. The density-based Ordering Points To Identify the Clustering Structure (OPTICS) clustering algorithm is employed to identify nine typical approach trajectory patterns under no weather. By integrating a multi-classifier based on Synthetic Minority Over-Sampling Technique (SMOTE) oversampling and the RUSBoost algorithm, the rainy-weather trajectories are matched with the no-weather trajectory patterns. Based on the proposed trajectory centripetal aggregation degree (TCAD) indicator, three deviation levels — low, medium, and high — are defined. Subsequently, from four analytical dimensions — flight time, trajectory efficiency, pilot-controller radio communication pressure, and air traffic control (ATC) strategy — the impacts of rainy weather are systematically quantified using statistical hypothesis testing, effect size estimation, and the introduced trajectory stretch degree (TSD) and the proposed pressure index of pilot-controller radio communication (PIRC). Taking Beijing Capital International Airport as a case study, the results show that different trajectory patterns are affected by rainy weather to significantly different degrees. Further analysis reveals that differences in the chosen deviation modes are the fundamental causes of the varying impact degrees across trajectory patterns in different analytical dimensions, and three typical ATC strategies are identified. This framework can provide quantitative decision-making support for air traffic control units in predicting approach operation situations and optimizing airspace resource allocation under rainy weather conditions.
Flight delays impose substantial costs on airlines, airports, and passengers. While existing studies predominantly focus on predicting average delays, extreme delays cause disproportionate operational disruptions and passenger dissatisfaction. This study introduces a distributional sensitivity analysis framework to identify factors that specifically drive extreme flight delays. Using 2.08 million U.S. domestic flights from 2024, we employ quantile regression models combined with Global Distributional Importance and Tangent-SHAP to decompose factor contributions across the entire delay distribution. Our analysis reveals three distinct effect patterns: tail effects, exhibited by weather conditions and carrier service history, which disproportionately amplify extreme delays; location effects, exhibited by realized delay variables such as departure delay and late aircraft delay, which shift the entire distribution approximately uniformly; and center effects, exhibited by flight duration and distance, which primarily affect typical delays. The classification is supported by an extensive robustness suite covering estimator noise, quadrature design, train–test partitioning, and correlation-preserving perturbations. We identify 60 min of departure delay as a critical threshold (bootstrap 95% confidence interval 55–65 min), beyond which 67 percent of flights experience severe arrival delays exceeding 60 min. Summer months exhibit 2.04 times higher tail risk than fall. The estimated annual cost of delays exceeding 60 min is approximately 1.9 billion USD, with potential savings of approximately 1.0 billion USD through targeted interventions. These findings provide actionable insights for airlines and airports to prioritize interventions targeting extreme delay prevention rather than average delay reduction.
This study develops a four-dimensional framework to evaluate how regulatory divergence between ASC 842 and IFRS 16 distorts financial metrics, operational control assessments, and asset valuations during airline consolidations. It argues that comparability problems do not arise from accounting treatment alone, but from the interaction of operational control doctrine, lease form, and acquisition-date valuation in airline mergers. The study uses a comparative case framework centered on Korean Air-Asiana as the primary post-COVID merger case, with Air France-KLM and United-Continental as supporting reporting and merger benchmarks. Particular attention is given to dry versus wet lease classification, right-of-use asset and lease liability recognition, EBITDA distortion, leverage effects, and merger-related lease remeasurement. The study concludes that lease reform improved transparency but did not eliminate analytical inconsistency, especially in cross-border airline competition and distressed merger settings.
War in Ukraine creates a significant impact on global air transportation system due to the wide area of involved airspaces. On February 24, 2022, Ukrainian air navigation service provider closed national airspace and suspended all navigation services. The ground network of radio navigation aids has been locked. Ukrainian network of radio navigation aids includes 16 Distance Measuring Equipment (DME) and 7 collocated VHF Omnidirectional Range (VOR) stations, which is also a part of the European radio navigation aids network. Service areas of Ukrainian radio navigation aids were provided in airspaces of neighboring countries that allow this equipment to participate in the pair formation process of the on-board positioning system. In the paper, we estimate the configuration of the European navigation aids network and study the impact of closed network on the performance of positioning by pairs of DME/DME and VOR/VOR in airspaces of neighboring countries: Poland, Slovak Republic, Hungary, Romania, and Moldova. Key performance parameters including availability, number of available pairs, positioning accuracy, and system reliability (mean time between failures and probability of failure-free operation) are used in area analysis to identify the impact of closing Ukrainian sub-network of navigational aids. Proposed methodological framework provides a systematic approach for assessing navigation network performance and highlights the implications of network disruptions for aviation safety in European airspace.
Air cargo transportation plays a key role in global trade, especially for time-sensitive and high-value products. Accurate prediction of air cargo demand is essential for informed decision-making on infrastructure planning, capacity management, and resource allocation across the air transport sector. While previous studies have largely advanced air cargo demand forecasting through model development and performance comparison, this study introduces a framework to explain and compare established predictive models based on data characteristic analysis (DCA). By examining key time series characteristics, such as stationarity, seasonality, and complexity, across statistical, machine learning, and deep learning approaches, this research examines how intrinsic properties of demand data are associated with forecasting performance. A rolling horizon design is also employed to evaluate how dynamic changes in data characteristics influence model performance over time. The findings reveal that statistical models are particularly sensitive to the mutability and complexity of air cargo demand data, whereas machine learning and deep learning models demonstrate stronger adaptability under diverse demand data conditions. Overall, this study shifts the emphasis from developing new forecasting models to explaining model suitability and supporting model selection, offering both theoretical insights and practical guidance for stakeholders by highlighting which models are best suited under specific data conditions.
The Climate Adaptation Reporting Power (ARP) established adaptation reporting requirements for critical national infrastructure in the United Kingdom to inform the 5-yearly UK Climate Change Risk Assessment (CCRA). This paper presents the first comprehensive analysis of the reports submitted by ten airports across the four ARP rounds to date (ARP1- ARP4, 2011-2024). Classifying over 900 reported risks, it provides the first complete synthesis of the climate risks and adaptation actions reported by the airports and the first complete mapping of airport climate risks since the second UK CCRA in 2017. Findings demonstrate that the initial mandatory round catalysed engagement: all airports embedded climate risks into corporate risk registers, identified adaptation actions, and nine established monitoring processes. Subsequent voluntary rounds show sustained participation and improved reporting quality, although inconsistencies in timeframes, climate scenarios, and risk classifications hinder comparability. Four recommendations for the ARP fifth round are identified: (1) provide clear templates and guidance identifying what good-quality reports should contain; (2) adopt sector-specific risk classifications to support network-level comparability; (3) broaden reporting to a wider set of UK airports, particularly smaller regional and local airports; (4) consider a mandatory approach to ensure thorough identification of climate risks, adaptation actions and gaps, and interdependencies. The study demonstrates the clear value of regulatory mandates in driving climate adaptation at airports and highlights lessons for other countries considering similar initiatives: the need for strong support and guidance, and early standardisation of inputs such as timeframes, climate scenarios, and risk categories. Adopting comparable measures internationally would strengthen resilience across the global aviation network.
Air traffic management is moving towards higher levels of human–automation teaming and decision support. Brain–computer interfaces (BCIs) can infer controllers’ cognitive workload from electroencephalography (EEG), enabling workload-aware automation. However, EEG-based decoding is highly sensitive to compound distribution shifts across subjects and operational scenarios, which degrades cross-domain representation learning and continuous monitoring. We propose a Dual-Shift Contrastive Alignment Learning Network (DSCAL-Net) for EEG-based cognitive workload decoding in air-traffic control. DSCAL-Net first employs a heterogeneous spatio-temporal encoding module that tightly integrates multi-band local time–frequency dynamics with connectivity-aware global networks, so that local features are shaped by global context and global connectivity is refined by local evidence. A factorized projection module then constructs subject-invariant and scenario-invariant subspaces within a shared latent space, while dual-shift contrastive alignment jointly regularizes subject-invariant, scenario-invariant and dual-domain representations, mitigating compound shifts while preserving class discriminability. Using multi-level workload EEG recorded in high-fidelity aerodrome-control simulations, DSCAL-Net is evaluated against representative baselines under cross-scene, cross-subject and dual-domain protocols, achieving average accuracies of 93.22%, 73.25% and 84.96%, respectively, with consistently improved accuracy and prediction stability. Simulated online monitoring further shows that the workload index derived from DSCAL-Net can smoothly track multi-stage workload evolution, demonstrating its transferability and potential as a computational foundation and proof-of-concept for future neuroadaptive support in air-traffic control.
This study investigates factors influencing passenger acceptance of mandatory sustainable aviation fuel (SAF) surcharges by extending the Theory of Planned Behavior (TPB) with Environmental Knowledge (a cognitive factor), Personal Norms from the Norm Activation Model (a moral factor), and Perceived Fairness. Analyzing 307 Thai air travelers using PLS-SEM, the model explained 55.8% of the variance in acceptance, with Perceived Fairness (β = 0.487) and Personal Norms (β = 0.288) emerging as the primary predictors. Environmental Knowledge and Attitude did not directly influence acceptance but exhibited significant indirect effects, with Personal Norms fully mediating these relationships. Findings suggest that in mandatory contexts, the traditional cognitive route of the TPB may function differently; while knowledge predicts attitude, attitude alone does not ensure acceptance without normative activation. Cognitive and moral frameworks operate sequentially within the model, yet justice-based evaluations demonstrate the strongest link to acceptance. To reduce passenger resistance, airlines and policymakers should prioritize transparent, equitable communication alongside normative framing.
A key challenge in airport ground support equipment (GSE) operations lies in fragmented decision-making and weak coordination across heterogeneous service providers, which arise from decentralized operational structures and limited information exchange. In addition, directly solving a fully integrated multi-type GSE scheduling model is computationally intractable due to large problem scales and complex task interactions. To address these challenges, a novel two-level augmented Lagrangian coordination framework is proposed to decompose the integrated problem into independent subproblems while enhancing global schedule consistency. In the proposed approach, the master problem determines global task start times by optimizing flight delay reduction and schedule robustness, whereas service-type subproblems optimize vehicle routing and local scheduling decisions to minimize travel cost. To enhance computational efficiency, an adaptive large neighborhood search (ALNS) heuristic is embedded to generate high-quality warm-start solutions for the subproblems. Computational experiments based on field data indicate that the proposed framework enhances routing efficiency relative to a greedy local scheduling benchmark, leading to a marked reduction in total vehicle travel cost while preserving schedule robustness, with only a moderate increase in flight delay, thus effectively balancing delay performance, robustness, and operational cost within practical computational time.
The global carbon abatement consensus, particularly on the Carbon Offsetting and Reduction Scheme for International Aviation (CORSIA), has effectively overturned the realistic basis of airlines' revenue management, including capacity pricing and seat allocation. This has created a novel scenario for studying airline operation management under the carbon reduction consensus. In this analysis, we consider an air-transport service supply chain consisting of two airlines and a downstream travel agent under the carbon emission trading system (ETS). The present study examines the impact of carbon emission trading and pricing strategies within two distinct organizational structures: the parent-subsidiary structure and the independently operated structure. The results demonstrate that the organizational structure of airlines does not influence the correlation between operational decisions and the carbon emission allowance allocation scheme, as well as the carbon emissions trading price. Incentivizing airlines to reduce emissions is advantageous for both increasing ETS activity and for airlines when a lenient allowance allocation scheme is adopted for low-emission airlines and a strict allowance allocation scheme is adopted for high-emission airlines. The independently operated structure is more conducive to encouraging airlines to invest abatement technology. With regard to the potential for abatement technologies, there is a possibility of a rebound effect for high-emission airlines under this organizational structure, whereby they may invest in abatement technologies. Furthermore, the ETS has a crowding-out effect on high-emission airlines, as a high carbon emission trading price will result in a reduction in their capacity inputs. Encouraging joint operation at low carbon prices and independent operation of high-emission airline at high carbon prices can effectively reduce carbon emissions in the industry.
Urban Air Mobility (UAM) is advancing rapidly, and its services are inherently linked with ground transport networks, making their impacts on traffic flow and efficiency a critical concern for urban mobility planning and policy decision-making. However, research on the implications of UAM for ground transportation remains limited and contested. Existing studies show divergent findings at the system-wide level and often overlook spatial heterogeneity and localized effects. This study develops an extended SUMO-TraCI framework that (i) incorporates realistic car-based first-/last-mile access and dispatch mechanisms, (ii) enables dual-scale analysis by capturing both system-wide traffic efficiency and localized congestion around vertiports, and (iii) supports multi-scenario testing, providing a broadly applicable framework for UAM evaluation. Based on five scenario experiments varying vertiport numbers (10, 30, 60, 80, and 100) relative to a baseline, UAM demonstrates modest yet positive citywide impacts on traffic flow reducing VKT and TTT modestly under the baseline eligibility setting, while sensitivity tests further reveal a non-linear and saturating benefit pattern around the 60–80 vertiport range as infrastructure density increases. Network-level spatial patterns indicate slightly greater relief on peripheral corridors and more frequent minor increases in central areas, with localized changes more noticeable on high-capacity roads than on local streets. This study introduces a scalable, multi-scenario evaluation framework for understanding and managing the multimodal impacts of UAM, providing actionable insights to guide evidence-based UAM planning and support sustainable city development.
This study presents a hierarchical forecasting framework for air traffic in Brazil, leveraging data provided by the National Civil Aviation Agency (ANAC). The framework integrates hierarchical representations of Brazilian air traffic with base forecasting models, such as ETS and SARIMA, and employs classical, as well as state-of-the-art optimal reconciliation techniques to enhance the accuracy of air traffic demand forecasts. Our results demonstrate that forecast reconciliation strategies can substantially improve the reliability of air traffic demand forecasts, thereby supporting more effective resource allocation, infrastructure planning, and tourism management. Our findings further show that classical reconciliation strategies, specifically bottom-up and top-down approaches – in case of the latter particularly Gross–Sohl Method F (TDGSF) – consistently provide strong performance across hierarchical levels. Rather than favoring more complex methods, the results emphasize the importance of aligning reconciliation strategies with data structure and volatility. Finally, our findings highlight the potential of hierarchical forecasting to capture the complex dynamics of Brazil’s air travel market, leading to better decision-making and strategic planning.
Numerous studies have analyzed airline itinerary choice behavior to understand passenger preferences and the trade-offs they make among key flight service attributes such as total elapsed time, number of connections, and fare. However, limited attention has been given to evaluating the influence of latent attitudinal constructs on itinerary choice decisions. This study examines the influence of latent attitudes and demographic characteristics on passengers' itinerary choice behavior and identifies distinct patterns of preference heterogeneity among air travelers. Two modelling approaches using stated preference data collected from 614 respondents are employed. We begin by estimating a multiple indicators multiple causes (MIMIC) model, which incorporates latent attitudes identified through exploratory and confirmatory factor analysis. The model reveals three latent attitudinal factors, viz “comfort and hygiene conscious”, “in-flight service seeking” and “time and punctuality-oriented”. To capture heterogeneity in preferences, we extend our analysis using a latent class choice model (LCCM), incorporating socio-demographic characteristics and latent variables as class membership covariates. The LCCM results reveal two distinct flyer segments, each exhibiting different sensitivity to airline itinerary attributes. Class 1 accounts for approximately 78% of the sample, while Class 2 comprises the remaining 22%. The first segment (Class 1) displays high preference for reduction in connection time and elapsed time while the second segment (Class 2) reflects greater sensitivity to service quality (legroom and meals). Besides, Class 1 flyers display a strong inclination to nonstop itineraries, while Class 2 flyers show increased reception for connected itineraries. The willingness to pay for connection time and elapsed time reductions are $14 and $16, respectively, for Class 1 air travelers. While flyers falling under Class 2 are willing to pay $10 and $13, respectively, for 1 h reduction in connection time and elapsed time. Moreover, elasticity analysis indicates that Class 1 passengers exhibit a strong aversion to extended connection times on direct itineraries. Additionally, a unit fare increase on direct flights leads to a 2.33% shift toward nonstop options. Females and highly educated individuals are slightly more represented in Class 1, indicating demographic influences on latent class membership. The findings demonstrate variations in flyer preferences especially in the context of latent attitudes of individuals. By accounting for passengers’ willingness to pay for connection time, and total elapsed time specific to different passenger segments, airlines can implement differentiated pricing strategies tailored to passenger sensitivities, thereby optimizing revenue generation while ensuring equity.
This study investigates the development of monotony in air traffic control (ATC) using a multi-modal measurement setup applied in a field study at two operational towers of differing traffic intensity: Tartu (low-traffic) and Tallinn (medium-traffic). The results provide a first step toward establishing a reference baseline for assessing monotony in uneventful low-traffic tower environments using an evidence-based approach. Monotony is conceptualized as a state of reduced physiological and cognitive activation resulting from low task stimulation, repetitive and uneventful operational contexts. We operationalized this concept using eye-tracking indicators, including blink duration, blink frequency, percentage eye closure (PERCLOS), and saccade velocity. Other measures included cardiac indicators derived from electrocardiography (ECG), the Karolinska Sleepiness Scale (KSS), and the Psychomotor Vigilance Task (PVT). These measures were collected from seven air traffic controllers across multiple shifts over an eight-day period.Linear mixed-effects models were fitted to each indicator, revealing consistent time-on-task effects at Tartu across several modalities, including blink duration, PERCLOS, HR (HR), root mean square of successive differences (RMSSD), and standard deviation of normal-to-normal intervals (SDNN). These findings support a pattern of growing parasympathetic activation and reduced arousal associated with monotony. In contrast, Tallinn showed signs of active physiological regulation under higher workload conditions, such as elevated baseline HR and decreasing SDNN, indicative of sustained engagement. The sympathovagal balance (LF/HF) ratio increased over time at both sites, but in the context of stable or increasing RMSSD/SDNN at Tartu, this may reflect compensatory physical activity rather than sympathetic dominance. Monotony indicators differed significant between towers, with up to a 7.0-fold stronger temporal change in PERCLOS and a 1.4-fold difference in SDNN at Tartu compared to Tallinn.The study further discusses methodological feasibility in sterile ATC work environments and highlights how operational differences, such as task rotation and self-directed off-position activities, impact data collection and interpretation. Findings underscore the need for tailored monotony mitigation strategies, particularly in low-traffic towers, and advocate for continued research in operational environments to capture complex real-world phenomena.
This study examines all papers published in the Journal of Air Transport Management (JATM) from 1994 to 2024 to analyze what data researchers use, how they share it, and the implications for impact and reproducibility. We employ an automated methodology to process collected articles. Specifically, we use Google Gemini language model to identify dataset types and data availability statements, and Non-Negative Matrix Factorization (NMF) to article texts to extract latent research topics. We find that while datasets like airline schedules, operating costs, and airport financials dominate the field, most are shared only upon request — or not at all. Further, the fraction of shared data has not grown significantly in the past decades.
Freight forwarders procure air-cargo capacity through a combination of long-term block space agreements (BSAs) and short-term ad hoc purchases. In practice, BSA allotments must be fixed in advance, while daily demand and the availability of ad hoc capacity are uncertain; moreover, ad hoc capacity is typically rate-ordered, with lower-cost options exhausted first. Despite its operational importance, existing models often treat ad hoc capacity as deterministic or fully available and rarely integrate long-term contracting with short-term allocation under such supply uncertainty. This paper examines the weekly capacity-planning problem of a freight forwarder facing these interacting uncertainties and develops a forwarder-centric two-stage stochastic framework that jointly models advance BSA contracting and ex post allocation decisions across multiple capacity sources, explicitly incorporating minimum chargeable weight (MCW) requirements and rate-ordered stochastic ad hoc supply. Analysis of a stylized setting provides insight into how freight rates, penalty costs, supply risk, and MCW requirements jointly shape optimal contracting decisions. Building on these insights, we propose a computationally efficient heuristic that translates demand distributions and stochastic supply scenarios into implementable BSA allotments. An empirical case study on a Thailand-Japan trade lane shows that the proposed heuristic achieves near-optimal performance, reducing average costs by about 35% relative to current practice and closing the gap to the full-information benchmark from 56.4% to as low as 2.4%. The results highlight the importance of explicitly modeling ad hoc capacity uncertainty and show that, once this supply-side uncertainty is accounted for, improvements in demand forecasting yield substantially greater operational benefits than further refinement of supply assumptions.
