Amid the increasing frequency of extreme climate events worldwide, urban transportation systems face dual pressures from structural vulnerabilities and climate-related risks, highlighting the urgent need to enhance climate adaptability and comprehensive resilience. Firstly, this paper systematically reviews the primary impacts of climate change on urban transport network structures and residents’ travel behaviors, revealing issues such as reduced network accessibility and temporal-spatial mismatches in travel demand caused by meteorological hazards. Then, from the perspective of physical and social resilience, the study analyzes the evolutionary characteristics of infrastructure disturbance resistance and recovery capabilities, traveler risk-response behaviors, and governance system adaptation mechanisms. Finally, the paper summarizes sustainable mobility policies guided by low-carbon objectives and transport resilience policy frameworks centered on disturbance absorption and recovery, explores synergistic governance pathways formed through their integration, reviews recent research progress on urban transport networks in infrastructure optimization, coordinated allocation of transport capacity resources, and transport-energy system coupling, and proposes an integrated optimization approach for constructing a multi-modal synergistic, low-carbon resilience-integrated transport system. The findings indicate that enhancing climate adaptability in future urban transportation systems should focus on multi-modal network optimization and technology empowerment, strengthen social equity and policy coordination, and promote a governance shift from infrastructure resilience to system resilience and from single-objective to multi-objective coordination. It is necessary to deeply integrate the coupled development of transport and energy systems, establish a full-cycle resilience assessment framework covering resistance, absorption, recovery, and adaptation to fill the current evaluation gap that focuses on post-climate-disaster impacts while neglecting long-term evolutionary processes. Additionally, integrated modeling of physical and social resilience, combining network topology, travel behavior, and governance mechanism analysis, should be refined to provide customized infrastructure optimization solutions and policy support for cities with diverse climate characteristics.

To systematically summarize the main themes, methods, and findings in current research on the resilience of comprehensive transportation systems, identify existing gaps, and provide insights for future research, this study conducts a detailed review of relevant literature on the resilience of comprehensive transportation systems. More than one hundred articles are retrieved from major Chinese and international academic databases. The review synthesizes existing work from several perspectives, including research objectives and scopes, resilience evaluation indicators, integrated transportation network modeling approaches, disruption and recovery scenario design, and strategies for enhancing system resilience. Current research can be broadly categorized into two spatial scales, intercity and intra-urban, with primary objectives focusing on resilience assessment, identification of critical components, and resilience enhancement. Regarding resilience evaluation indicators, existing studies predominantly adopt two categories: network-analysis-based indicators and traveler-oriented indicators, alongside some integrated indicators and metrics tailored to multimodal transportation. For modeling integrated transportation networks, the literature is reviewed from the perspectives of multilayer network structures, disruption and recovery scenario configurations, and network dynamics. In terms of resilience enhancement, current strategies typically fall into three groups: pre-disaster prevention, post-disaster response, and optimized recovery planning. The results demonstrate that future research should place greater emphasis on demand-side factors and incorporate multimodal transportation characteristics more thoroughly. Disruption scenarios and passenger behavioral responses require more accurate modeling, with fuller consideration of slow modes of travel. Dynamic analyses should pay closer attention to realism and the influence of travel information guidance. Additionally, the complexity of integrated transportation systems and the heterogeneity across different transport modes warrant further investigation.

This study investigates the time-varying vulnerability of high-speed rail (HSR) networks under cascading failures initiated by unexpected train halts that disrupt network links. Based on the actual passenger timetable of December 15, 2023, the paper simulates the complete propagation process of cascading train failures under diverse attack strategies (random and intentional) and intensities. First, a cascading failure model is developed that captures the operational characteristics of HSR networks and analytically dissects the failure propagation mechanism. Subsequently, network performance metrics are designed from both structural and functional perspectives, incorporating temporal dynamics inherent in cascading failures, to establish integrated static and dynamic vulnerability assessment frameworks. Finally, using real-world Chinese HSR timetable data, a topological network is constructed and the interdependencies among trains, lines, and stations are formalized. Through cascading failure simulations, the static characteristics of stations and lines alongside the diurnal operational patterns of trains are analyzed, thereby revealing the temporal evolution of network vulnerability indicators. The results demonstrate that under low-intensity attacks within identical time windows, intentional targeting disrupts significantly more trains than random failures. However, as attack intensity escalates, the differential impact between the two strategies diminishes. Under intentional attacks, trains traversing critical links during morning peak hours exhibit the greatest vulnerability and impose the most severe degradation on daily network performance. Therefore, prioritizing protective measures for morning peak corridors and implementing rapid recovery interventions constitute essential strategies for safeguarding network performance and enhancing resilience.

To address the vulnerability of urban rail transit networks during operational disruptions, this study proposes a systematic evaluation method that integrates network structure, passenger flow dynamics, and passenger behavioral responses. First, a CPT-NL model of travel choice behavior is developed based on Cumulative Prospect Theory (CPT) and the Nested Logit (NL) model. This model captures passengers’ bounded rationality and risk aversion under emergency conditions. Second, a vulnerability analysis framework is designed for urban rail networks within multimodal transportation systems, establishing quantitative vulnerability metrics that reflect the combined impacts of operational disruptions and passenger flow dynamics. Finally, using Shenzhen’s rail transit network as a case study, peak-hour passenger flow data are employed to systematically simulate disruptions at sections and transfer stations, quantifying network performance losses and passenger travel impacts. The results indicate that failures at critical sections or transfer stations reduce network connectivity and trigger large-scale passenger congestion and significant travel delays. During section disruptions, the proportion of affected passengers reaches up to 14.78%, with network vulnerability peaking at 12.02%. During transfer station disruptions, the proportion of affected passengers peaks at 11.73%, and network vulnerability reaches 5.53%. The number of delayed passengers significantly exceeds the number who abandon rail travel due to disruptions, indicating that most passengers opt for detours within the rail network or multimodal travel alternatives. Operational management should strengthen maintenance and emergency preparedness for high-vulnerability nodes, while optimizing multimodal transportation scheduling to improve the resilience and operational stability of urban rail transit systems.

To address the challenges of resilience assessment and post-failure recovery in railway passenger networks following large-scale station failures, this study proposes a railway passenger network resilience assessment and optimal recovery method based on a service efficiency indicator. First, from the perspectives of network service capacity and actual operational characteristics, a railway passenger service network model is constructed. A service efficiency indicator considering the number of operating trains between stations is proposed, and a network service resilience assessment model is established using the resilience curve and the service efficiency indicator. Second, an optimal resilience recovery model is formulated with the objective of maximizing recovery resilience, and an improved Particle Swarm Optimization (PSO) algorithm is developed to determine the optimal recovery strategy under large-scale station failures. Finally, the Chengdu-Chongqing railway passenger network is used as a case study to evaluate network resilience and compare the resilience recovery performance of different strategies under three disturbance scenarios. The results indicate that the service efficiency indicator effectively assesses the impact of station failures on network performance and accurately identifies critical stations. The optimal recovery strategy derived from the resilience recovery model consistently outperforms both random and degree recovery strategies across all disturbance scenarios. Under deliberate attack scenarios, the recovery resilience of the network achieved by the optimal strategy improves by 69.90% and 4.81% compared with the random and degree recovery strategies, respectively.

To address the limitations of existing research in adequately characterizing road-rail topological coupling and the mechanisms of cross-system load transfer and transmission time lags during node failures, this study constructs a topology model for coal multimodal transport networks that incorporates road-rail synergistic structures. A cascading failure evolution mechanism is proposed, explicitly accounting for capacity constraints, cross-system transshipment, and transmission time lags. Addressing functional maintenance and recovery needs throughout the disturbance lifecycle, a three-dimensional resilience assessment index system, comprising resistance, absorption, and recovery capabilities, is established to quantify functional attenuation and restoration under varying disturbance intensities and scopes. Furthermore, two load redistribution strategies based on network structure and capacity distribution are designed to compare resilience under scenarios of key node failure, cascading congestion, and cross-mode substitution. To characterize the impact of differentiated disturbances such as intentional attacks and natural disasters, a dynamic disturbance strategy based on grey information is introduced to simulate attack selection and resilience evolution under varying levels of network knowledge. Finally, a typical Chinese coal transport corridor is used as a case study to simulate the scenario. Simulation results demonstrate that the proposed model effectively captures key node failure characteristics and cascading evolution dynamics. The findings reveal that load redistribution strategies and disturbance modes significantly influence system resilience, providing a theoretical basis for enhancing network resilience, optimizing road-rail capacity allocation, and improving emergency response.

To address the problem that isolated analysis of conventional bus and subway networks fails to capture actual passenger transfer behaviors and inter-network coupling effects, this study constructs a weighted composite network model of conventional bus and subway systems using the Space L modeling method. The model integrates actual spatial distances between nodes, establishing a coupling radius of 660 m via Geographic Information Systems (GIS) and the Amap Application Programming Interface (API), while excluding stations where the actual walking path is excessive despite meeting straight-line distance criteria. Line carrying capacity and passenger travel time are assigned as edge weights to balance network supply capacity with passenger travel costs. Network topological properties are analyzed using Matlab and Gephi, focusing on core metrics such as average degree, average path length, clustering coefficient, and betweenness centrality. Furthermore, two novel attack strategies are proposed: the Node Centrality Importance Attack Strategy (CIAS), based on multi-attribute decision theory, closeness centrality, and betweenness centrality; and the Degree Attack Strategy of Coupling Nodes (DASCN), which accounts for the substitution effects of coupling nodes. An attack efficiency evaluation function is introduced to compare these novel strategies against traditional ones, using Harbin’s main urban area as a case study. Performance is evaluated via global efficiency, the relative size of the largest connected subgraph, and the network connectivity rate. The results demonstrate that the composite network shows significant improvements in average degree, clustering coefficient, network efficiency, and connectivity rate, with decreased average path length and betweenness centrality, indicating enhanced resilience. Under identical node removal scales, the CIAS exerts a greater negative impact on network performance than traditional strategies, whereas the DASOCN has a comparatively smaller impact on global efficiency, the largest connected subgraph, and connectivity rates.

To address operational disruption risks in urban rail transit networks arising from station failures, this study proposes a systematic resilience optimization methodology based on a weighted coupled map lattice (CML) model integrated with an improved simulated annealing algorithm. First, by comprehensively incorporating station degree, inter-station passenger flows, and boarding/alighting volumes, a weighted CML model is constructed to accurately simulate cascading failure processes following station disruptions. Structural and service performance metrics derived from this simulation are then employed to quantify system resilience. Subsequently, two optimization models are established: A critical station identification model aimed at minimizing cumulative resilience loss, and a recovery sequence optimization model for failed stations designed to maximize restoration efficiency. To enhance algorithmic robustness and search efficiency, an improved simulated annealing algorithm featuring dynamic cooling rates and hybrid perturbation strategies is developed. Finally, the Wuhan Metro network serves as a case study for empirical validation. Results demonstrate that, compared to traditional single-indicator-based methods (such as those relying solely on station degree, inter-station flow, or boarding/alighting flow), the critical stations identified by the proposed model (primarily high-flow non-transfer stations) induce 7% to 13% greater network performance degradation upon failure. Moreover, the recovery sequence optimization model and algorithm yield a 3% to 7% improvement in restoration efficiency, validating the effectiveness of the proposed approach in enhancing network robustness and recovery capacity.

Existing methods for identifying critical communities and key stations in transportation networks often lack unified evaluation criteria and a systematic analytical framework. To address this, this study proposes a hierarchical identification framework based on the Leiden algorithm. First, an improved modularity function that integrates both passenger flow and topological features is constructed, systematically incorporating station passenger volumes and network topological characteristics into the Leiden algorithm’s community detection process. Second, based on the community detection results, a multi-dimensional evaluation system is developed that simultaneously considers functional and topological attributes to quantitatively assess the importance of critical communities and key stations. Finally, the applicability of the proposed method is validated using real operational data from Beijing’s integrated bus-metro network, by comparing variations in community structures and critical node distributions across different day types. Results indicate that compared with non-working days and holidays, the number of communities on weekdays decreases by 9.05% and 8.59%, respectively, while the number of large-scale communities (with more than 150 nodes) increases by 16.67% and 40%. Overall, community importance is predominantly driven by functional attributes, but weekday net-works exhibit a more balanced interplay between functional and topological significance. Furthermore, bus stations consistently represent a higher proportion of critical nodes due to their superior spatial coverage and service flexibility. This study provides theoretical insights and methodological support for the structural optimization and differentiated operational planning of multimodal transportation networks.

To accurately characterize and evaluate the performance of flight ground handling operations, this study proposes a method for constructing and analyzing multi-level Stochastic Petri Net (SPN) models of ground handling processes. Both the execution of handling tasks and the connections between handling nodes are modeled as transitions in the SPN, capturing the complex series-parallel relationships across the entire process. A time-performance equivalence simplification method is applied to reduce analytical complexity, after which an isomorphic Markov chain is used to evaluate performance, yielding place occupancy (busy) rates and transition utilizations for the multi-level SPN model. An input-output place-occupancy quadrant diagram visually highlights high- and low-efficiency areas in the process, enabling precise classification of pre-connection and post-connection states. By combining this with transition results and analyzing the cumulative change in steady-state probabilities under dynamic variations of mean transition rates, critical handling nodes and connections are identified, including four critical nodes and six key inter-node connections. Results indicate that critical part often occur at loosely connected nodes, such as the post-catering and in-flight supply connections to cabin door closure. Even when individual task durations are short, prolonged preceding or subsequent states can reduce overall handling efficiency. Furthermore, when the mean occurrence rate of a critical node or connection exceeds 0.2, corresponding to a duration reduced to 5 minutes, the overall process duration stabilizes. This study provides theoretical support and practical guidance for improving turnaround time prediction accuracy and enhancing ramp operation efficiency under collaborative airport decision-making.

To address the problem of approach flight delays and reduced operational efficiency caused by severe weather, a quantitative assessment framework for operational resilience in arrival operations is developed. A Greedy Algorithm (GA) embedded within a Receding Horizon Control (RHC) framework is proposed to optimize the sequencing of weather-affected approach flights, enabling rapid improvement of key performance indicators during the resilience recovery phase and enhancing overall operational resilience. First, based on the dynamic evolution pattern of arrival flight performance, comprising four stages: stable, disrupted, recovery, and new stable, delay duration, on-time performance, and landing throughput under severe weather conditions are selected as fundamental indicators. Incorporating actual operational characteristics, a comprehensive resilience metric is constructed that integrates system redundancy and average recovery performance compensation. Second, an airborne holding cost function is formulated based on aircraft type and delay duration, balancing economic efficiency and fairness principles. Third, the RHC+GA approach decomposes the dynamic optimization problem through a rolling time-domain mechanism. Aiming to minimize flight airborne holding costs, the greedy algorithm yields locally optimal solutions and thereby generates an improved flight sequence under adverse weather conditions. Finally, using a thunderstorm scenario at Tianjin Binhai International Airport for verification. Results show that under the actual scheduling sequence, the operational resilience of arriving flights is 1.529. After reordering with the RHC+GA method, resilience increases to 2.791, an 82.5% improvement, while the average delay per aircraft is reduced by 6 minutes, significantly accelerating operational performance recovery.

To address the issues of low extraction accuracy of track areas and lagged risk response during railway maintenance windows, this paper proposes a safety detection algorithm based on hierarchical semantic fusion. First, to overcome challenges such as the slender geometry of distant tracks, background interference, and high computational redundancy in traditional segmentation networks, a hierarchical semantic fusion-based track area extraction algorithm framework is constructed. By employing a hierarchical feature encoder and a lightweight semantic fusion decoder, the algorithm significantly reduces parameter counts and computational complexity. This enables high-precision, real-time segmentation of track areas under limited hardware resources, providing a reliable spatial foundation for subsequent risk assessment. Second, to adapt to diverse operational requirements and risk control needs during maintenance windows, a hierarchical safety zoning method based on lateral distance measurement is designed, categorizing are-as into a core operation zone (layer A), an auxiliary operation zone (layer B), and a non-operation zone (layer C). Combined with a cross-stage feature-coupled real-time identification algorithm for boundary violations, this approach integrates regional prior information with multi-scale target features to perform multi-task analysis, including worker detection, uniform identification, and helmet detection. This facilitates the precise determination of safety equipment compliance and unauthorized boundary incursions. Finally, the effectiveness of the algorithm is validated on the self-constructed RailScapes railway maintenance window dataset. The results indicate that the proposed algorithm effectively addresses the limitations of traditional video surveillance, such as inadequate coverage of blind spots and high manual dependency. By implementing a tiered early-warning mechanism, the algorithm achieves real-time proactive alerts for violations, providing a high-precision, low-power, and highly resilient technical foundation for active safety protection during railway maintenance.

To examine the resilience characteristics of urban taxi systems under daily disturbances, this study employs a dynamic threshold method to identify disturbance periods and establishes a three-dimensional resilience assessment framework encompassing resistance, recovery, and adaptability. Nine evaluation indicators tailored to industry operational features, such as empty-vehicle response efficiency, idle mileage redundancy, and spatiotemporal matching balance, are incorporated. Indicator weights are determined using the entropy weight method, and a comprehensive resilience index is constructed. Drawing on three weeks of taxi order data from Zhuhai, an empirical analysis is conducted to explore resilience differences across multiple dimensions, including day type (weekdays vs. non-weekdays) and temporal characteristics (morning peak, evening peak, night peak). Results indicate that the overall system resilience is jointly determined by period composition and demand structure. Weekdays exhibit substantially higher comprehensive resilience than non-weekdays, largely due to the presence of the highly organized morning peak period. Specifically, recovery and adaptability indicators on weekdays exceed those on non-weekdays by 54.75% and 110.18%, respectively. The morning peak exhibits the strongest resilience on weekdays, while the night peak is the most vulnerable. Further analysis reveals that under the same period composition (i.e., only evening and night peaks), weekday resilience performance is weaker than that on non-workdays, highlighting the critical influence of demand patterns on resilience levels. Resistance is primarily driven by empty-vehicle response efficiency and spatiotemporal matching balance; recovery is significantly affected by changes in the order completion rate; and adaptability is closely related to service efficiency per unit distance and cross-regional dispatching efficiency. The proposed assessment framework and empirical findings provide valuable theoretical and practical guidance for enhancing resilience and improving daily operational management of urban taxi systems.

To investigate the environmental comfort of public areas in subway stations, this study analyzes the variation laws of environmental parameters based on on-site field measurements. Accounting for the variability in passengers’ clothing insulation and activity levels, an improved Predicted Mean Vote (PMV) calculation model is proposed. Additionally, the Predicted Percentage of Dissatisfied (PPD) model is refined to incorporate the characteristics of sudden ambient temperature changes. Based on these improved PMV-PPD models, the environmental comfort of the station is evaluated, and the sensitivity of PMV and PPD to environmental parameters is examined using sensitivity analysis. Finally, applying reliability theory, a reliability assessment model for station environmental comfort is established by modeling passenger metabolic rate and clothing thermal resistance as random variables. The results indicate that for fixed monitoring points, PMV and PPD values fluctuate around stable baselines, and the improved PMV-PPD model demonstrates greater applicability compared to the traditional model. With constant metabolic rate and clothing thermal resistance, the sensitivity of PMV to air temperature initially decreases and then increases as air temperature rises. Specifically, when the metabolic rate is 58.15 and clothing thermal resistance is 0.35, 0.5, and 0.7, as temperature changes, the minimum sensitivity of PMV to air temperature is consistently 0, while the maximum sensitivities are 0.278, 0.243, and 0.204, respectively. Furthermore, at a metabolic rate of 58.15, the system failure probability gradually decreases as air temperature increases; conversely, at metabolic rates of 116.3 and 139.56, the failure probability increases with rising air temperature.

Frequent lane-changing and acceleration-deceleration behaviors in expressway merge areas restrict overall traffic efficiency and compromise driving safety. The geometric design of these areas is critical to traffic operations; specifically, the length of the acceleration lane is a vital design element that directly impacts efficiency and safety. To address issues of low efficiency and frequent accidents in merge areas, this study investigates acceleration lane length. Based on field-measured data, the Kolmogorov-Smirnov (K-S) test is employed to analyze headway distributions across different segments of the outermost mainline lane within the merge area, and the maximum likelihood method is applied to estimate distribution parameters. Subsequently, the acceleration lane length is calculated by integrating gap acceptance theory with the non-sequential merging behavior of ramp vehicles. Finally, VISSIM is utilized to simulate and compare traffic operations of the merge area before and after the optimization of the acceleration lane length. The results indicate that headway distributions exhibit heterogeneity across different segments of the outermost lane. The optimized acceleration lane accommodates the waiting time required for most ramp vehicles to find a merging gap, thereby enhancing safety of vehicle merging. Under level of service 3 conditions on the mainline, ramp merging volume increased by 7.15%-15.11%, while traffic conflicts decreased by 8.68%-22.16%. Under level of service 4 conditions on the mainline, merging volume increased by 5.47%-14.22%, and conflicts decreased by 2.08%-14.74%. Consequently, the optimized acceleration lane improves both traffic efficiency and safety of the merge area, providing a novel approach for calculating acceleration lane length.

The integration of flexible bus services introduces uncertainty into network topology and edge weights, rendering redundancy analysis methods based on fixed topologies inapplicable. To address this, this study investigates modeling techniques and redundancy evaluation methods for bus networks incorporating flexible services. First, to tackle the topological uncertainty of flexible lines, a random network model is constructed using constraint programming to enumerate all feasible path sets. By introducing interval service probabilities to aggregate these paths and defining mapping relationships among the physical, service, and demand layers, a multilayer “physical-service-demand” network representation is established. Second, an infrastructure-based route diversity index considering supply and demand weighting is developed. This index integrates factors such as the overlap reduction of efficient routes on the physical road network, infrastructure redundancy in the physical layer measured by resistance distance, travel demand distribution, and bus service frequency. Finally, an empirical analysis is conducted using randomly generated bus network schemes with varying degrees of flexibility, based on the bus network data from Jiading District, Shanghai. The results indicate that at the network dimension, the introduction of flexible stops may cause the overall redundancy index of the bus network to fluctuate or decline, primarily due to reduced service frequency and increased path lengths caused by route detours. At the Origin-Destination (OD) dimension, although the number of OD pairs experiencing reduced redundancy due to the addition of flexible stops exceeds those experiencing an increase, significant improvements are observed for specific OD pairs. Notably, the average magnitude of the increase in route diversity is 2.36 times that of the decrease, indicating that the positive effects are more substantial in magnitude. Consequently, when planning and designing flexible bus networks, flexible stop configurations should be determined through simulation and optimization to prioritize redundancy improvements for high-demand OD pairs.

To address the difficulty of jointly optimizing safety and timeliness in route planning for large-scale emergency supply transportation under extreme events, this study investigates railway–highway intermodal route planning. First, a multi-objective mathematical programming model is developed that aims to simultaneously maximize safety probability and minimize transportation time, while comprehensively incorporating flow balance constraints, road hierarchy constraints, and disaster-induced route disruption constraints. Second, a hybrid LCA-NSGA-Ⅱ algorithm is proposed, in which the Label Correcting Algorithm (LCA) is embedded into the Non-dominated Sorting Genetic Algorithm Ⅱ (NSGA-Ⅱ) framework to generate high-quality initial populations. In addition, a Dual-Node Partially Mapped Crossover (DNPMX) operator and a Segment Deletion and Repair (SDR) operator are designed, and an adaptive crossover and mutation mechanism is introduced to enhance search efficiency. Finally, the effectiveness of the proposed model is validated through multiple experiments on a railway-highway multimodal network comprising 104 nodes and 496 arcs, and the performance of the LCA-NSGA-Ⅱ algorithm is compared with that of the Gurobi solver and the conventional NSGA-Ⅱ algorithm. The results indicate that the LCA-NSGA-Ⅱ algorithm achieves an 87.98% improvement in solution efficiency compared to the Gurobi solver. Compared to the traditional NSGA-Ⅱ algorithm, it achieves 76.1% and 1.15% improvements in inverted generational distance (IGD) and hypervolume (HV), respectively. The model effectively balances time and safety objectives and generates distinct routes for different types of supplies between the same origin-destination pair. Sensitivity analysis of disaster intensity levels shows that as the disaster impact weakens, transportation time is reduced by up to 3.33%, while safety probability increases by 4.70%, demonstrating the model’s adaptability to dynamic disaster scenarios. The proposed model and algorithm provide effective decision support for emergency supply transportation scheduling under extreme events, thereby enhancing the reliability and responsiveness of emergency logistics systems.

To address the intensifying competition between road and rail freight transportation and the increasingly severe loss of rail freight volume, this study investigates the coordinated optimization of off-site warehouse location selection and freight flow allocation. First, it identifies that under existing port logistics models, private enterprises, characterized by fragmented, small-scale, and weak operations, lack the incentive and capacity for rail block train shipments. Second, by comparing traditional port transportation routes with those integrated with off-site warehouses, this study highlights how the off-site warehouse model alleviates inventory pressure and reduces costs for small and medium-sized enterprises with limited demand and financial resources. Third, focusing on the coordinated optimization of off-site warehouse location and freight flow allocation, mixed-integer programming model based on service networks is developed. This model comprehensively considers key factors, such as transportation and construction costs, to optimize off-site warehouse layout and enhance road-rail intermodal efficiency. Finally, an empirical study using the port clearance process of non-ferrous ores at Tianjin Port as a case study validates the model’s effectiveness. Research findings indicate that the off-site warehouse model significantly reduces total transportation costs and increases rail turnover compared to traditional models. Specifically, the single-warehouse configuration reduces total operating costs by 16.63% with a 45.18% rail turnover share, while the multi-warehouse configuration reduces costs by 30.81% and increases the rail turnover share to 95.41%.