Decarbonizing Freight Through Intermodal Transport: An Operations Research Perspective—Part II: Modal Configurations and Sustainability Pathways

Abstract

Building upon the methodological synthesis presented in Part I, this second part of our two-part survey examines how operations research (OR) models have been applied to capture the broader dynamics of intermodal transport (IMT) in pursuit of decarbonization. The analysis integrates chronological, modal, and sustainability-oriented perspectives to reveal how IMT strategies evolve across transportation modes, policy environments, and temporal contexts. We identify how efficiency gains, modal shifts, and low-carbon technologies interact within OR frameworks, and assess their implications for emissions reduction, energy use, and network resilience. By bridging technical modeling approaches with system-level sustainability objectives, this study offers a holistic understanding of the pathways through which OR supports the transition toward low-carbon freight systems and highlights research gaps for future interdisciplinary work.

1. Introduction

In the last decade, operations research (OR) has played an increasingly important role in advancing decarbonization strategies in freight transportation (FT), one of the most carbon-intensive sectors globally, particularly in systems involving multiple transportation modes, namely intermodal (IMT), multimodal (MMT), and synchromodal (SMT) transport. FT accounts for 8–11% of global GHG emissions, and growing economies in Asia, Africa, and Latin America are projected to triple global freight demand by 2050, potentially doubling freight-related emissions and making FT the highest-emitting sector worldwide [1]. Road freight, however, emits over 100 times more CO₂ than maritime transport per unit of freight-distance and accounts for approximately 80% of the global growth in diesel consumption [2]. In response to this challenge, governments worldwide have implemented ambitious policy frameworks to accelerate freight decarbonization. For instance, the European Union (EU) set modal shift goals in its “White Paper on Transport”, aiming to transfer 30% of road freight over 300 km to rail or waterways by 2030, and 50% by 2050 [3]. Similarly, in the United States (US), the Advanced Research Projects Agency-Energy (ARPA-E) has started initiatives to enhance the IMT sector by promoting low-carbon emission modes through technology and data-driven solutions [4]. These policy developments underscore the growing urgency and practical relevance of OR driven approaches to freight decarbonization. Despite this growing research interest, no existing review has, to the best of our knowledge, consolidated these efforts in a comprehensive and structured manner. Early reviews on IMT have concentrated primarily on OR modeling techniques [5,6], simulation frameworks [7], and long-haul optimization [8], largely overlooking sustainability considerations and carbon emissions. Conversely, broader surveys on FT decarbonization have predominantly taken qualitative approaches [9,10], with limited focus on operational modeling. Consequently, a critical research gap exists at the intersection of IMT decarbonization and quantitative OR-based operational frameworks.

To directly address this gap, this survey is presented as a two-part study. Part I of this two-part survey established the methodological foundations for analyzing decarbonization efforts in IMT. A systematic literature search was employed across Web of Science, Scopus, IEEE, and ABI Inform (ProQuest), using structured keyword strings and explicit inclusion criteria to identify studies applying OR methods to intermodal freight decarbonization between 2010 and 2024. It systematically collected, classified, and synthesized studies published between 2010 and 2024 that apply OR techniques to support emission reduction objectives. The review provided a comprehensive taxonomy of OR methods used across key problem domains, including network design, routing, and demand estimation, and identified major methodological trends, research gaps, and emerging opportunities. Detailed mappings of problem classes, modal configurations, objective-function components, decision variables, modeling assumptions, and solution techniques across all reviewed studies are provided in a comprehensive summary table in the Appendix of Part I of this two-part study. Key findings underscored the growing need for multistage stochastic optimization frameworks to better capture uncertainty and operational disruptions, as well as the increasing potential of data-driven and machine learning-based approaches to enhance decision-making, coordination, and resource efficiency.

Part II (the present paper) builds on this foundation by offering a comprehensive, application-oriented synthesis of OR-driven decarbonization studies in FT from 2010 to 2024. Specifically, this paper systematically analyzes how OR techniques support emissions reduction and operational efficiency across MMT, IMT, and SMT systems, while explicitly integrating modal structure, sustainability strategies, decision levels, and emission considerations. By categorizing studies according to modality mix, operational decision context, sustainability objectives, and methodological approach, this review develops a unified framework that connects operational modeling with practical decarbonization strategies in IMT. Furthermore, the chronological assessment captures the rapid growth and methodological evolution of this research stream since 2018, highlighting emerging trends, dominant solution paradigms, and persistent research gaps. Together, these contributions advance existing literature by moving beyond isolated methodological surveys and qualitative assessments toward a structured, operationally grounded understanding of how OR can effectively support sustainable FT systems.

The remainder of this Part II is organized as follows. We provide a chronological analysis of the reviewed studies in Section 2, followed by an integrated modal and sustainable analysis in Section 3. Section 4 examines research gaps and future directions, and Section 5 summarizes our findings. A comprehensive list of acronyms used throughout this paper is provided in Abbreviations.

2. Chronological and Thematic Evolution of the Literature

Since the early 2000s, research on freight decarbonization has evolved from unimodal cost and emission assessments [11] toward comprehensive systems analyses that integrate environmental, economic, and policy dimensions. Figure 1 illustrates this progression. To clarify the temporal evolution of this research stream, the literature is grouped into four dominant phases that approximately span the periods 2009–2014 (Early Interventions), 2015–2017 (Development of New Concepts), 2018–2021 (Advancement of Concepts), and 2022–2024 (Recent Developments). These four chronological phases and their associated time boundaries were identified through thematic clustering and temporal publication patterns, guided by key inflection points in policy developments and methodological evolution. The Early Interventions phase (2009–2014) begins with the seminal work of Kim et al. [12], which established systematic CO₂-cost trade-off analysis, and concludes as deterministic optimization frameworks matured and international climate policy began to crystallize. The Development of New Concepts phase (2015–2017) is anchored in the adoption of the 2015 United Nations Sustainable Development Goals and the Paris Agreement, marking a shift toward policy-driven research emphasizing carbon taxation, subsidies, regulatory interventions, and the integration of stochastic modeling. This short phase represents a transitional period during which policy instruments were systematically embedded into operational models. The Advancement of Concepts phase (2018–2021) reflects the maturation of these frameworks alongside the emergence of synchromodal systems, real-time optimization, and sector-specific applications, with the COVID-19 pandemic further accelerating resilience-oriented and uncertainty-aware modeling. Finally, the Recent Developments phase (2022–2024) captures the post-pandemic surge in publications and methodological diversification, characterized by rapid growth in game-theoretic analysis, SD modeling, and carbon peak prediction studies addressing near-term climate targets. These breakpoints represent approximate thematic demarcations rather than rigid temporal boundaries. While individual studies may introduce concepts ahead of or beyond these time windows, this classification reflects the dominant thematic and methodological shifts observed across the literature. Moreover, although these phases follow a general chronological progression, considerable temporal overlap exists, as foundational concepts introduced in earlier phases continue to be refined, extended, and operationalized in subsequent research.

2.1. Early Interventions (2009–2014)

A foundational study by Kim et al. [12] analyzes cost-CO₂ emission trade-offs between IMT and truck-only networks, revealing a near-linear inverse relationship between freight cost and emissions. Building on this, Kim and Van Wee [13] use Life Cycle Assessment (LCA) to quantify direct and production emissions across modes and processes (e.g., drayage), confirming IMT’s lower carbon footprint. Their follow-up work [14] reinforces these results with additional data (e.g., loading units), whereas Comer et al. [15] extend the modeling to a geospatial hub-and-spoke framework, highlighting the environmental benefits of shifting freight from road to waterways.

Building on IMT network design (ND) advances, several studies integrate emission costs into planning models. Bauer et al. [16] embed GHG emissions into intermodal cost optimization focused on fleet scheduling, while Chang et al. [17] include mode-specific environmental costs to encourage low-emission short sea shipping and ease port congestion. Later, Zhang et al. [18] optimize mode selection by incorporating carbon costs from transport, transfer, and inventory, and Le and Lee [19] extend this globally through vehicle-specific emission modeling across truck, sea, and air modes.

Dry ports have emerged as a key strategy linking seaports to inland terminals by rail to facilitate modal shifts. Early findings report up to a 25% emission reduction from their use [20]. Building on this, Henttu and Hilmola [21] integrate congestion, noise, and CO₂ impacts into ND optimization, while Lättilä et al. [22] assess dry port feasibility under unimodal (road) and intermodal port-hinterland setups. Similarly, Iannone [23] internalize emissions, congestion, and accidents within generalized cost functions (e.g., transit holding cost) in a multi-commodity capacitated dry port model.

Advancing carbon-aware pricing, Chaabane et al. [24,25] apply LCA within carbon trading frameworks. Meanwhile, Pishvaee et al. [26] introduce demand and capacity uncertainties into supply chain (SC) ND, and Holmgren et al. [27] assess how CO₂ taxes, infrastructure investment, and policy shifts influence modal choice. Similarly, Fahimnia et al. [28] evaluate carbon pricing in closed-loop SCs, and Zhang et al. [29] examine its effect on terminal design. Extending these efforts, Hoen et al. [30] analyze how carbon regulations affect mode selection, later expanded by Rezaee et al. [31] to incorporate carbon price and demand uncertainty. At the tactical level, Fahimnia et al. [32] assess carbon tax impacts on financial and emission outcomes, while Zhang et al. [33] analyze their influence on mode choice and infrastructure investment.

Expanding the sustainability scope, Sawadogo et al. [34] identify routes that jointly minimize environmental, social, and economic impacts including noise and accident risk, while Kim et al. [35] optimize facility investments by minimizing social costs under emission limits. Recognizing the need to align stakeholder priorities, Kengpol et al. [36] develop a decision support system minimizing cost, time, risk, and CO₂ emissions. In parallel, Soysal et al. [37] explore cost-emission trade-offs in food SCs considering road design, vehicle type, load, perishability, and backhauls, while Pan et al. [38] integrate mode choice, demand allocation, and routing to analyze cost-CO₂ relationships.

Several studies address emission reduction through seaport expansion and selection. Rodrigues et al. [39] evaluate alternative seaport gateways and ND, Chen et al. [40] examine service ND (SND) and traffic flow management, and Rodrigues et al. [41] analyze container re-routing and congestion impacts. Furthermore, Liotta et al. [42] integrated sourcing and production decisions within IMT by linking emissions to transported volume. Previously, Chaabane et al. [24] incorporated Bill of Materials (BOM) constraints into the IMT sourcing and production framework.

This phase established foundational methodologies for quantifying emission trade-offs in IMT systems, demonstrating the environmental benefits of modal shifts from road to rail and waterways. Key contributions include the integration of carbon costs into ND and pricing frameworks, the adoption of LCA and activity-based emission modeling, and the recognition of dry ports as strategic infrastructure for facilitating low-emission freight corridors. Early studies primarily relied on deterministic optimization approaches emphasizing infrastructure planning and cost internalization, while gradually expanding sustainability assessments beyond carbon emissions to encompass broader social and economic externalities. These early interventions laid the groundwork for subsequent policy-oriented research and multi-objective optimization frameworks.

2.2. Development of New Concepts (2015–2017)

Following the 2015 adoption of the UN 2030 Agenda for Sustainable Development [43], research increasingly emphasized government intervention, particularly tax and subsidy policies, as key levers for freight decarbonization. Duan and Heragu [44] pioneered the analysis of IMT behavior under carbon tax policies, while Wang et al. [45] examined taxation effects within government and company-third-party logistics alliances. In the same year, Bouchery and Fransoo [46] assessed emission and the modal-shift effects of train subsidies, and Zhang et al. [47] extended this to MMT contexts, linking government decisions to operational and emission costs in SND routing.

Subsequent studies increasingly coupled fiscal instruments with ND. Lin et al. [48] balanced investment, transport, and emission costs to guide rail infrastructure selection and freight flow management, whereas Mostert et al. [49] evaluated road tax policy and health externalities on modal splits between road, rail, and waterways. Their follow-up study [50] further assessed subsidies and external-cost while integrating economies of scale and a door-to-door scheme. Haddadsisakht and Ryan [51] incorporated demand and carbon-tax uncertainty, showing that adaptive capacity planning mitigates volatility. Likewise, Zhang et al. [52] integrated infrastructure investment and subsidies to meet CO₂ targets, and Choi et al. [53] demonstrated that road taxes and containerization exert strong influence on modal shifts.

Around 2020, Jiang et al. [54] modelled regional logistics networks that jointly determine park locations, capacities, and railway subsidies under national carbon targets, while Li and Zhang [55] examined government emission policies and pricing strategies to promote modal shifts from road-only to road-rail through coordinated decisions among government, operators, and consigners. Later, Yang et al. [56] developed a joint policy-network design for capacity expansion in road-water systems, Qian et al. [57] optimized regional modal splits for freight, and Gallardo et al. [58] planned multimodal networks through ND and SND decisions that aimed toward net-zero freight. Pedinotti-Castelle et al. [59] and Halim [60] further evaluated modal-shift strategies and underscored the need for sustained public investment in rail infrastructure.

Terminal operations, a critical component of IMT, have gained prominence in OR models. In multi-commodity SND problems, Qu et al. [61] estimate CO₂ emissions using an activity-based function while accounting for terminal handling costs, whereas Rudi et al. [62] consider mode type, energy source, and topography in emission estimation and include transshipment from carrier replacement. Further contributions include Baykasoğlu and Subulan [63], who optimize sustainable load planning, and Ji and Luo [64], who minimize transshipment time and cost to alleviate congestion. Additionally, Zhou et al. [65] address combined mode-vehicle costs, while Sun et al. [66] jointly minimize CO₂ and population exposure risks in road-rail systems.

In addressing uncertainties like network disruptions in IMT decarbonization, stochastic and hybrid methods have become prominent tools. Demir et al. [67] pioneered this approach in SND by incorporating uncertain service times and demand. Building on this, Hrušovskỳ et al. [68] added in-transit inventory costs to capture travel time variability, while Layeb et al. [69] applied a different method to generate more accurate travel time data. Extending this perspective, Sun et al. [70] considered multiple uncertainties (e.g., railway capacity and loading times) in rail-road networks, and Baykasoğlu and Subulan [71] modelled cost, capacity, and transit time uncertainty in fleet planning for IMT systems.

In ND and dry port research, Tran et al. [72] integrated inland connections into global container networks to minimize maritime, hinterland, and emission-related costs. Tsao and Linh [73] extended this by incorporating stakeholder-driven storage pricing. Xu et al. [74] quantified transport and transfer emissions at dry and seaports while modeling port competition and shipper routing, and Liu et al. [75] captured dynamic fuel consumption under varying fleet configurations and cargo volumes in port-hinterland systems. Other studies focussed on enhancing freight efficiency and sustainability: Tsao and Thanh [76] integrated all sustainability dimensions under uncertainty in dry port ND, Wei and Dong [77] linked per-container emissions to local regulations in cross-border logistics, and Li et al. [78] assessed multimodal cost-environment trade-offs in seaport selection.

In port-hinterland IMT, Wang et al. [79] integrated production and transport scheduling under uncertainty for port-centric SCs, while Dai and Yang [80] addressed container ND with waterways and terminals under demand uncertainty. Abu Aisha et al. [81] optimized terminal layouts and hinterland transport to lower costs and emissions, and Yin et al. [82] minimized transportation costs, emissions, and travel time through optimized hinterland divisions. Within IMT SND, empty container repositioning (ECRP) remains a pivotal area. Lam and Gu [83] first integrated ECRP into MMT routing with emission constraints, followed by Zhao et al. [84], who modelled container leasing and routing under demand-supply uncertainty, and Castrellon et al. [85], who assessed ECR strategies with varying container substitution via dry ports.

Various modeling approaches have been developed to estimate GHG and CO₂ emissions in IMT systems, generally classified as energy- or activity-based and distinguished by their precision as microscopic or macroscopic [86]. Bridging these scales, Kirschstein and Meisel [87] proposed mesoscopic road-rail models using operational parameters (e.g., speed, weight) to assess energy demand. Heinold and Meisel [88] extended this to large-scale estimation, capturing shipment, vehicle, and route attributes. On the activity-based side, Pizzol [89] integrated error propagation into LCA for sea-road transport, while Tao and Wu [90] included well-to-wheel emissions from transshipment, highlighting ECRP’s role in hinterland transport. Guo et al. [91] further captured emissions from electric modes across hinterland networks, addressing prior neglect of node-level emission complexities.

This phase marks a conceptual shift toward policy-driven decarbonization, emphasizing government intervention through carbon taxation, rail subsidies, and infrastructure investment as primary mechanisms for promoting modal shifts. The growing adoption of stochastic, hybrid, and simulation-based approaches reflects increasing recognition of real-world operational complexity. The emergence of terminal operations as a distinct research focus highlights the importance of transshipment efficiency in overall network emissions. Additionally, refinements in emission estimation methodologies, spanning micro-, meso-, and macroscopic scales, provide more accurate and operationally relevant tools for quantifying environmental impacts across complex multimodal networks.

2.3. Advancement of Concepts (2018–2021)

In 2018, emission reduction strategies in food SCs increasingly targeted operational measures like postponing packaging to lower freight weight. Harris et al. [92] examined IMT routing and packaging configurations for wine distribution, assessing how route, mode, and packaging affect CO₂ and sulfate emissions. Ma et al. [93] proposed a shipper-oriented scheduling model for cold chains that jointly minimizes cost, quality degradation, and refrigeration emissions. Likewise, Maiyar and Thakkar [94] optimized hub location and IMT decisions in food grain SCs, incorporating capacity and availability constraints across hubs, vehicles, and handling operations.

Around the same time, Demir et al. [95] and Laurent et al. [96] advanced freight planning tools supporting stakeholder decisions on modal shifts using historical and real-time data. The former provided an offline planning tool offering timely environmental and economic insights, while the latter introduced the concept of an intermodal carbon-efficient boundary. Concurrently, SMT gained traction for enhancing intermodal efficiency through dynamic planning and real-time information exchange [97]. Key advances include decision and scheduling frameworks balancing energy use and cost in SC design [98], real-time SMT ND under capacity and disruption constraints [99], online shipment-service matching with flexibility considerations [100], and dynamic routing with real-time synchronization under disruptions [101].

To improve reliability in IMT and MMT networks, Sun [102] modelled uncertainties in network capacity, departure, and loading times within a road-rail system, estimating CO₂ emissions through activity-based methods [70]. Conversely, Wang et al. [103] addressed node-capacity uncertainty in road-inland waterway networks under regional carbon taxation [33]. Extending this to SC reliability, Kabadurmus and Erdogan [104] incorporated carbon cap-and-trade mechanisms capturing production and transport emissions with BOM constraints [24], while Mousavi Ahranjani et al. [105] evaluated strategic and tactical SCND decisions for bioethanol systems under operational disruptions.

Advances in decarbonization and routing planning continue in 2020. While Sun and Lang [106] integrated schedule-based and time-flexible services in MMT routing earlier, considering CO₂ emissions and generalized costs (e.g., transport, inventory), recent studies broaden this scope. Maneengam [107] propose a multi-objective model minimizing transport and emission costs while ensuring on-time delivery, balancing cost, time, and GHG emissions in MMT planning. Wang et al. [108] couple operational and environmental efficiency metrics, whereas Heinold and Meisel [109] introduce order-specific emission limits set by shippers using egalitarian and payload-based allocation within road-rail IMT network.

FT has been examined from various logistics service provider (LSP) perspectives to understand decision-making and operational strategies. Fulzele et al. [110] analyze factors (e.g., damage, delays, and GHG emissions) influencing modal shift decisions. Similarly, Wang et al. [111] study competitive pricing among LSPs to balance mode choice, revenue, and emission reduction, while Zhang et al. [112] address dual uncertainty in MMT under cap-and-trade policy [24] to minimize transport cost, emissions, and time. Tiwari et al. [113] develop a third-party LSP model separating long- and short-haul decisions under carbon taxation, and Wu et al. [114] design a fourth-party LSP platform optimizing bulk freight distribution. Finally, Li and Sun [115] assess how uncertain demand and carbon trading prices affect MMT routing design.

Incorporating travel time into SND formulations affects not only routing and scheduling but also customer satisfaction and overall network efficiency. In this direction, Lu et al. [116] integrate hub location and vehicle routing with travel time in an MMT network combining highways and rail. Similarly, Liang et al. [117] optimize paths by balancing storage and delay penalties, enhancing service quality through timeliness and economic satisfaction indexes. In multi-commodity MMT settings, Qi et al. [118] include departure dates and emissions in door-to-door planning, whereas Xie et al. [119] extend this by adding air transport and carbon costs in cross-border logistics. Sun et al. [120] further advance rail-road routing with time-varying travel parameters and fuzzy time windows, accounting for dynamic emissions and a hub-and-spoke system supported by direct road transport [70]. Beyond operational optimization, Zhang et al. [121] and Heinold et al. [122] explore how eco-label systems encourage environmentally conscious customer behavior in IMT.

This phase demonstrates a pronounced shift toward operational complexity and real-time decision-making, exemplified by the emergence of synchromodal transport systems that leverage dynamic routing and information exchange to enhance network efficiency under uncertainty. Sector-specific applications, particularly in food SCs, illustrate how operational tactics such as packaging postponement and cold-chain optimization contribute to emission reductions beyond modal shifts alone. The growing emphasis on logistics service provider perspectives and competitive dynamics reflects recognition that decarbonization strategies must align with commercial incentives and market structures. Additionally, the integration of travel time considerations into ND highlights the multi-dimensional nature of sustainable freight planning, where environmental objectives must be balanced against service quality and customer satisfaction.

2.4. Recent Developments (2022–2024)

In 2022, several studies on synchromodality focussed on optimizing schedules, user preferences, and resource utilization. First, [123] analyzed carrier preferences for cost, time, and emissions, optimizing total costs (i.e., container, fuel, and carbon tax) alongside travel and waiting times. In another study, [124] accounted for shippers’ heterogeneous priorities through a decision-making framework that balances multiple performance criteria, while Zhang et al. [125] enhanced service flexibility by integrating fixed and adaptive routes with complex scheduling to improve resource utilization. Building on these efforts, Oudani [126] introduced blockchain technology to SMT, coupling it with energy-efficient management and multi-criteria evaluation to generate green and optimal transport strategies.

Advancing low-carbon MMT practices requires aligning stakeholder interests, and, in recent years, particularly 2023, a sharp rise in game theory (GT) applications toward this goal have occured. These models emphasize operational coordination under carbon regulation and stakeholder interaction. Shams et al. [127] compare cap-and-trade, carbon offset, and carbon tax schemes from governmental, economic, and environmental perspectives, showing how policy design shapes freight operators’ responses. Wu and Zhang [128] integrate carbon taxation into dry port planning, capturing government-shipper interactions under capacity limits and empty container flows. Similarly, Rahiminia et al. [129] examine rail operator-shipper pricing dynamics to align profitability with sustainability via triple bottom line objectives. Further, Chen et al. [130] assess government-market coordination through subsidy mechanisms maximizing emission reductions per unit subsidy while considering demurrage and operational emissions.

In the same year, system dynamics (SD) models were widely applied to policymaking analyses of port-hinterland IMT systems. SD captures feedback loops, time delays, and nonlinear interactions shaping long-term effects of carbon taxation and related policies. Within this context, Zhong et al. [131] simulated transport feedbacks to assess how carbon taxes influence CO₂ reduction and economic costs, highlighting the need for region-specific policy design. Nassar et al. [132] evaluated fiscal, regulatory, and infrastructure measures promoting shifts to lower-emission modes, showing their effectiveness declines over time as systems stabilize. Extending this perspective, Guo et al. [133] combined SD with Monte Carlo simulation to assess policy mixes, including cost-based pricing, road expansion, rail improvements, and subsidies, under economic transition uncertainty, revealing that policy interactions diminish overall effectiveness despite strong individual impacts.

Predicting carbon peak times is vital for guiding climate action and energy management, as this turning point marks the start of emission decline essential for sustainability and policy planning. Research in this area centers on identifying peak timings and managing emissions across sectors. In this context, Zuo et al. [134] design an annual control model to manage freight energy consumption, projecting a 2029 peak in the study region, consistent with Yu et al. [135], who link emission peaks to energy intensity. Guo et al. [136] simulate carbon peak timing in container IMT networks, showing that meeting pre-2030 targets requires optimizing transport structure and network connectivity, alongside Ke et al. [137], who use an adaptive genetic algorithm (GA) to adjust freight sharing rates to align emission with transport factors including safety, cost, speed, and flexibility.

Advances in GA-based approaches also emerged in 2023: Zhang and Chen [138] and Yang et al. [139] applied GA-based multi-objective models to enhance container MMT routing, jointly minimizing cost, time, and emissions. The former improving service efficiency and the latter addressing multi-task conditions from the MMT operators’ perspective. Similarly, Li and Wang [140] optimized secondary hub locations, cargo flow, and mode selection in hierarchical hub-and-spoke networks, capturing trade-offs among hub capacity, clearance efficiency, and carbon taxes. Extending these efforts, Shoukat and Xiaoqiang [141] optimized logistics networks connecting dry ports and seaports, refining green logistic designs by distinguishing MMT and IMT operations.

Recent studies highlight the operational complexity of IMT under uncertainty and dynamic conditions. Sun et al. [142] address first- and last-mile operations by integrating soft time windows, uncertain truck speeds, rail capacities, and carbon tax policies into path planning. Li et al. [143] extend this by incorporating uncertain demand, volume, and transshipment times, combining GT with dynamic weighting to balance objectives in real time. For time-sensitive logistics, Liu [144] model cold-chain routing under congestion, weather, and breakdowns, minimizing penalties and emissions while ensuring timely delivery. Similarly, Kurtuluş [145] develop a disruption recovery model for container shipping using speed control and port skipping. While port skipping is effective, its benefits decline when shipping lines and terminals share real-time data and coordinate time windows and handling rates.

The pivotal role of shippers’ choices in shaping sustainable transportation strategies is also highlighted in recent research. Sun et al. [146] integrate river channel upgrades, ship deployment, and routing optimization to expand inland waterway use and reduce trucking dependence under carbon taxation. Wu and Zhang [147] analyze shippers’ operational choices across modes, terminals, and ports, showing how capacity expansion, taxes, and subsidies drive modal shifts and affect logistics costs and emissions. Meanwhile, Guo et al. [148] optimize feeder shipping operations by coordinating route planning, scheduling, and fleet allocation, demonstrating how carbon costs influence MMT network efficiency.

Strategic improvements in bulk cargo distribution increasingly rely on MMT integration that accounts for physical constraints and customer preferences. In this context, Ko et al. [149] analyze biomass SCs under transport cost uncertainty, incorporating railcar leasing, emission, and social costs to assess how limited rail capacity drives road-dominant modal shifts. de Almeida Rodrigues et al. [150] develop a discrete-event simulation model enabling joint shipper-consignee decisions across seven transport strategies involving mode, port type, and cargo consolidation. Extending to bulk freight, Feng et al. [151] optimize mode choice, routing, depot allocation, and containerization by integrating customer preferences and physical constraints (e.g., bridge heights), promoting inland waterway use within MMT networks.

Looking ahead to 2024, studies by Yin et al. [152] and Zhang et al. [153] compare carbon pricing strategies. Yin et al. [152] assess combined policies like rail freight subsidies and carbon trading, highlighting limited modal shift success and weak differentiation among regional pricing mechanisms. Their multi-objective IMT ND model integrates transport costs, travel times, and carbon trading. Zhang et al. [153] propose a multi-objective MMT planning model minimizing transport cost and time under fuzzy demand and time intervals across varied carbon policies.

Additionally, Ghisolfi et al. [154] and Derpich et al. [155] focus on modal and energy transitions. Ghisolfi et al. [154] evaluate how electrification, biofuels, fleet renewal, and modal shifts reduce emissions in Brazil’s freight system, while Derpich et al. [155] adapt hub-and-spoke strategies to regional demand, emphasizing consolidation and hub concentration in MMT ND.

This phase is characterized by methodological diversification and increasing analytical sophistication in addressing freight decarbonization. The growing adoption of game-theoretic and SD models reflects recognition that effective policy design requires explicit representation of stakeholder interactions, feedback mechanisms, and long-term system behavior beyond static optimization. Technological integration, particularly blockchain and real-time data sharing, emerges as a key enabler of synchromodal coordination and operational transparency. The emphasis on carbon peak prediction and scenario-based policy evaluation highlights the urgency of meeting near-term climate targets, while continued advances in genetic algorithms and multi-objective optimization demonstrate ongoing efforts to balance cost, service quality, and emissions. Collectively, these developments signal a transition from conceptual feasibility toward the operational deployment of low-carbon freight systems within complex, uncertain, and multi-stakeholder environments.

3. Integrated Analysis of Literature

This section reviews key trends in the literature on modality mix, logistics decision levels, and emission considerations in OR models. The analysis covers studies from 2010 to 2023, excluding 2024 due to the limited number of publications (only four), which could skew overall trends. The modality and decision-level analyses are conducted independently across the entire literature corpus, with all modes represented at all decision levels and vice versa.

3.1. Modality Analysis

To analyze the distribution of research across transportation modalities, it is essential to understand how different mode combinations are represented in the literature. Figure 2 provides a comprehensive overview, presenting seven distinct transport mode combinations and illustrating the focus and frequency of studies within each.

The combination of road, rail, and water transport is the most studied in IMT research, with extensive work centered on SND. Interest in this mix surged in 2023, driven by global decarbonization goals and pandemic-related disruptions that underscored the need for sustainable and resilient transport systems. Here, the term water-based transport is used as a general descriptor covering inland waterways, short sea shipping, and maritime transport. The road-rail combination is also widely examined, particularly for hinterland connections like dry ports that enable modal shifts at inland terminals. This pairing leverages the complementary strengths of each mode: rail provides efficiency and cost savings over long distances, while road offers flexibility and time efficiency over short hauls. By contrast, air transport is rarely studied (about one study per year) due to high costs, capacity limits, and handling challenges. Some studies also address global SC network design, focusing on travel time, risk, and freight damage considerations.

Next, we present the evolution of IMT and MMT studies in Figure 3. IMT publications have steadily increased over time, reflecting its growing role in enhancing the financial and environmental performance of FT. This sustained growth has been driven by tightening carbon regulations and emission reduction targets, advances in intermodal infrastructure and terminal automation, and the maturation of OR-based optimization frameworks, which together have reduced traditional coordination barriers and strengthened the operational viability of intermodal systems. These drivers have also influenced modeling approaches, shifting early deterministic and static formulations toward stochastic, robust, and dynamic optimization frameworks. However, in 2021, the pandemic temporarily shifted focus toward healthcare logistics and emergency response, causing a brief decline. This disruption further accelerated the adoption of resilience-oriented models that explicitly capture uncertainty, network disruptions, and adaptive capacity planning. By 2022–2023, attention returned to SC optimization, with IMT’s potential to improve efficiency and resilience regaining prominence. The 2023 surge in both IMT and MMT research can stem from post-pandemic recovery, renewed interest in SC resilience, and evolving policy initiatives. The surge in both IMT and MMT research during 2022–2023 reflects renewed emphasis on multimodal integration, operational robustness, and policy-aligned decarbonization strategies, supported by growing adoption of real-time optimization, data-driven forecasting, and advanced computational methods. For this figure, SMT is viewed as an extension of IMT and is included in the IMT dataset.

Planning decisions are typically categorized as strategic, tactical, or operational [156]. Strategic decisions involve long-term planning such as network design, facility location, and infrastructure investment. Tactical decisions address medium-term concerns including fleet sizing, service frequency, and resource allocation. Operational decisions focus on short-term execution such as routing, scheduling, and real-time adjustments. While surveying the literature, we identified a significant volume of studies addressing more than one decision level simultaneously. To capture this integration, we introduce a Mixed category, defined as studies that jointly consider two or more decision levels, including strategic-tactical, tactical-operational, strategic-operational, and fully integrated strategic-tactical-operational formulations. Following this framework, studies are organized over time in Figure 4. The results reveal a clear and growing trend toward integrated decision-making, with research addressing multiple decision levels increasing substantially since 2012 and the Mixed category becoming dominant in recent years, particularly sharp increase in 2023. This shift reflects the growing recognition that decarbonization strategies require coordinated optimization across infrastructure planning, service design, and operational control, rather than isolated treatment of individual planning horizons. It also highlights the increasing complexity of OR models and their emphasis on cross-level interactions, system-wide trade-offs, and coordinated policy and operational planning.

Since the Mixed category represents the largest share of studies in Figure 4, a detailed breakdown is provided in Figure 5. The figure shows that strategic-tactical combinations account for the largest share (54%), followed by tactical-operational (25%) and strategic-operational (15%), while fully integrated models addressing all three levels remain relatively rare (6%). This continued rarity can largely be attributed to several structural barriers, including the substantial computational complexity arising from multi-scale decision coupling, the extensive data requirements needed to ensure consistency across planning horizons, and the methodological challenges associated with uncertainty propagation and model validation. These factors significantly increase model dimensionality and solution difficulty, often rendering fully integrated formulations computationally intractable for large-scale real-world networks and prompting most studies to adopt partial integration strategies that balance modeling fidelity with computational tractability. The dominance of strategic-tactical integration suggests that researchers prioritize linking infrastructure and capacity decisions with service design and resource allocation, as these combinations most directly influence modal shift feasibility and long-term network efficiency. Although still limited, fully integrated multi-level models are emerging as an important research direction, underscoring a broader methodological transition toward holistic, multi-scale decision frameworks capable of capturing the dynamic interactions and trade-offs inherent in real-world freight decarbonization pathways.

Interestingly, research focusing solely on operational planning has declined sharply over the past three years, suggesting a shift toward integrating it with other decision levels. Tactical decisions remain the most studied, followed by strategic ones, reflecting the dominance of deterministic and stochastic ND research. This decline in operational studies, alongside the growing emphasis on strategic and tactical levels, signals a broader move toward long-term planning and resilience.

3.2. Sustainability Analysis

As sustainability becomes central in OR, understanding how carbon emissions are integrated into model formulations is essential. This section reviews key approaches for incorporating emissions, categorized into five main types, each representing a distinct strategy for embedding carbon considerations into OR models.

• Carbon estimation: Using emission estimations or factors that are minimized within OR models.
• Carbon cap: An emissions limit, often regulated by authorities.
• Carbon cost: Assigning a monetary value charged per unit of emissions (e.g., taxes).
• Cap-and-trade: Carbon credits that can be bought or sold in a market regulated by a third party.
• Combination: Any mix of the above methods.

Figure 6 presents the number of studies by classification over time. Carbon cost (43%) and carbon estimation (40%) dominate the literature. Early research focused on hub location and terminal selection driven by carbon cost modeling, whereas recent studies emphasize improving network-wide operating efficiency. Similarly, carbon estimation initially guided infrastructure expansion decisions but now primarily assesses the environmental impacts of modal shift strategies.

Another trend in recent years is the growing focus on improving the precision of emission estimation models. Cap-and-trade mechanisms are mainly applied in SCND decisions, while carbon caps were initially used to evaluate government policies. However, recently, attention has shifted toward combining emission strategies, reflecting the need for stronger, coordinated government actions to decarbonize the sector.

These policy instruments differ fundamentally in implementation dynamics. Carbon caps impose static regulatory constraints, carbon taxes generate volatile price signals, and cap-and-trade schemes create stochastic, market-driven prices. These differences directly shape OR modeling choices: static policies favor deterministic formulations, while dynamic instruments require stochastic or robust optimization. The dominance of carbon cost and emission estimation approaches (83%) thus reflects a preference for modeling tractability over the computational and data challenges associated with market-based policy mechanisms.

It is crucial to note that these strategies have varying degrees of real-world application. For instance, carbon estimation is commonly used by companies to monitor and reduce emissions, often guided byframeworks like the GHG Protocol [157]. Carbon taxes, implemented in 30 jurisdictions as of 2020 [158], encourage emission reductions through financial penalties, while cap-and-trade systems, such as those in California, allow firms to trade carbon credits, incentivizing mitigation [159].

Although widely studied in academia, particularly within the OR community, many approaches are explored through idealized models optimizing cost, emissions, and operational efficiency. Real-world implementation, however, is more complex due to challenges like profit loss, carbon price volatility, regulatory variation, and technological limits. Firms may adopt these strategies voluntarily in regions with strong environmental policies or market incentives, but compliance is typically driven by mandatory regulation.

4. Future Research Directions and Identified Gaps

This section outlines existing research gaps, suggests future directions, and highlights key areas that require further exploration.

Pricing Mechanisms: Revenue management remains underexplored in IMT research, resulting in limited attention to pricing mechanisms for transportation services. Addressing this gap, particularly through stakeholder collaboration, could enhance economies of scale [90,123], while integrating quantity discounts across modes offers another promising direction [129]. Uncertainty in multi-period dynamic investment and pricing models [52] is largely overlooked in our surveyed literature, underscoring a rich area for future study. Examining how strategic decisions (e.g., port location) affect pricing among governments, carriers, and shippers, while accounting for transport and processing capacities could yield more realistic insights [128]. Further, analyzing pricing determinants like demand-supply balance may benefit from AI-driven demand forecasting [160] integrated with behavioral factors.

Alternate Energy Sources and Autonomous Vehicles: The transition to alternative energy sources in IMT necessitates holistic estimation models, like LCA frameworks, to evaluate transport and transshipment emissions [74], accounting for uncertainty in emission rates across modes. This is increasingly relevant with the rise of battery-electric locomotives, hybrid vehicles, and hydrogen-powered systems. Strategic studies are needed to assess required investments versus energy efficiency gains for better financial planning [65]. Research should also address infrastructure development, like charging networks and resource capacity for long-haul electric trucks and trains [98] and evaluate the feasibility of 1 MW megachargers and battery-swapping stations for longer-distance transport.

Integrating autonomous vehicles, namely self-driven trucks and drones, within intermodal logistics can significantly enhance efficiency and sustainability. Drones, or unmanned aerial vehicles (UAVs), extend ground transport, particularly for last-mile urban deliveries. Operating from intermodal hubs, UAVs enable rapid local distribution and seamless mode transitions while also supporting monitoring, inspection, and real-time data collection. However, sustainability and emission impacts of multimodal drone delivery remain underexplored. UAVs could also support cold-chain logistics by combining last-mile de-consolidation with end-haul transport. Future studies should evaluate trade-offs between traditional and autonomous systems, comparing their respective efficiencies and benefits.

Container De/Consolidation and Repositioning: Freight consolidation, deconsolidation, and empty container repositioning remain critical yet underexplored topics in IMT research. Integrating consolidation and delivery decisions can yield more effective solutions for improving system-wide sustainability [116]. Future studies should also examine how consolidation affects smaller shipments and packaging [92], as these factors influence emissions and efficiency. Developing GT models that incorporate containerization technologies in shipping markets presents another promising direction [151].

Container size, also often overlooked in the current literature, directly shapes consolidation and deconsolidation strategies influencing costs, carbon footprints, load optimization, and space utilization. Ignoring this factor limits sustainability insights. For repositioning, accounting for container shortages and network contingencies can strengthen ECR evaluations. Trip-sharing strategies, combining bulk transport with ECR, can offer particular promise by reducing costs for all parties.

5. Conclusions

In this second part of our two-part survey, we analyze how IMT, MMT, and SMT freight systems have evolved toward decarbonization. Building on the methodological foundations of Part I, this study highlights chronological, modal, and sustainability trends illustrating the sector’s growing integration of carbon reduction strategies. Our synthesis shows how operational innovations, policy interventions, and technological advances collectively shape emissions trajectories across transport modes. We also identify persistent challenges such as uneven adoption of low-carbon technologies, weak policy coordination, and data gaps that constrain system-level efficiency. Overall, this second part complements Part I by offering a temporal and practical perspective on OR’s role in advancing low-carbon freight transitions.