Integrating Disruption Propagation and Buffer Allocation in Smart City Manufacturing Clusters: A Transport-Energy Performance Model for Intermodal Supply Chains

Abstract

This study integrates disruption propagation and inventory buffer allocation into a unified analytical framework and develops a transport-energy performance model (TEPM) for intermodal supply chains operating within smart city manufacturing clusters (SCMCs). The analyzed system consists of a single intermodal logistics node (ILN), a city logistics node (CLN), and multiple manufacturing buildings, forming a multi-layer network with strong interdependencies between transport, logistics, and production processes. Disruptions may occur at different system layers, including the ILN, transport links, the CLN, and manufacturing entities, and propagate downstream through the network (ripple effect), affecting material availability and production continuity. The proposed model employs the multi-layer Bayesian network method (MLBNM) to represent disruption propagation and interdependencies between system components under uncertainty. Inventory buffers at the CLN and manufacturing nodes are incorporated as decision variables and act as key mitigation mechanisms, enhancing system resilience. Buffer allocation also affects transport-energy performance through its interaction with transport activity and production continuity, linking inventory-based resilience with the energy requirements of material transport. The model integrates network structure, disruption propagation, and inventory-based decision-making into a unified framework. A scenario-based analysis evaluates system performance under different disruption conditions and buffer allocation strategies using a representative SCMC configuration. The results for the analyzed scenarios show how buffer placement and supply chain configuration affect resilience and transport-energy performance within the modeled system. The proposed approach provides a framework for jointly analyzing probabilistic disruption propagation, inventory-based mitigation, and transport-energy performance in intermodal SCMC supply chains.

1. Introduction

Smart City Manufacturing Clusters (SCMCs) combine production and logistics within dense urban environments. Their structure is inherently multi-layered: external supply flows enter the system, are consolidated in urban logistics nodes, and are then distributed to multiple manufacturing buildings. This tight integration improves efficiency but also creates strong interdependencies between transport, logistics, and production processes [1,2,3]. These interdependencies become particularly relevant when transport-energy performance is considered. In SCMC systems, transport energy consumption depends on the intensity and organization of material transport operations [4,5]. Inventory and flow-management decisions may indirectly affect this performance by changing transport activity and material availability for production. At the same time, such tightly coupled systems are highly sensitive to disruptions. Disturbances may occur at intermodal entry points, within the city transport network, or inside the cluster itself, and their effects propagate through the whole supply chain. This propagation, often described as the ripple effect, leads to imbalances between supply and demand at downstream nodes [6,7]. Therefore, analyzing disruptions requires a network perspective that captures how local failures affect the entire system [8].

Probabilistic approaches have become increasingly relevant for representing interdependencies in intermodal SCMC supply chains under uncertainty. In particular, multi-layer Bayesian network models enable the explicit linkage between supply chain structure and disruption propagation mechanisms [9,10], supporting the identification of critical nodes and flows. However, analyzing disruption propagation alone is insufficient from an operational perspective, as effective mitigation requires coordinated decision-making. Inventory buffering remains a primary mechanism for mitigating disruptions, with safety stock (SS) providing a baseline level of protection against uncertainty. Inventory buffers at city logistics nodes (CLNs) and production sites absorb variability and reduce the impact of disruptions on material availability [11,12]. Nevertheless, disruption modeling and inventory decisions are generally treated separately, with limited consideration of how buffer allocation should adapt to network structure and disruption dynamics [13]. This limitation is particularly significant in SCMC systems, where structural and operational factors jointly determine system performance. Recent studies have begun integrating disruption modeling with inventory management [14], yet the interaction between buffer allocation and transport-energy performance remains underexplored. Moreover, existing approaches do not fully utilize probabilistic representations to support coordinated decision-making across supply chain layers. As highlighted in recent research on resilient supply chain design, integrating uncertainty, operational decisions, and system efficiency constitutes a key research direction [15].

This study therefore aims to develop a transport-energy performance model (TEPM) for Intermodal Supply Chains in SCMCs by extending the M2 framework presented in the previous study [14]. M3 retains the MLBNM-based stochastic disruption-propagation mechanism of M2 and broadens its operational and inventory scope. Disruptions are differentiated across e-truck transport, CLN transshipment, e-van transport, and unloading at individual SCMBs, while waiting-state variables represent production material temporarily unavailable during disrupted operations. The methodological novelty of the TEPM lies in integrating operationally differentiated disruption propagation, waiting-state material flows, two-level inventory buffering at the CLN and SCMB levels, and transport-energy assessment within a single modeling framework, thereby extending M2 from disruption-aware inventory management to the joint evaluation of material availability, production continuity, and transport-energy performance. Buffer stock (BS) at individual SCMBs complements SS at the CLN, providing inventory protection at both network and manufacturing-building levels. Disruption and inventory states are linked to production continuity and transport-energy performance. Therefore, the following research questions are formulated:

RQ1. How do disruptions propagate across the intermodal SCMC supply chain networks and affect production material availability and manufacturing continuity?

RQ2. How should buffer inventories be allocated and replenished within the network structure to mitigate disruption impacts effectively?

RQ3. How do buffer allocation and replenishment, as well as the intensity of transport operations, influence transport-energy performance of the SCMC supply chains under disruption conditions?

RQ4. How does the integration of disruption propagation and buffer allocation influence the transport-energy performance of intermodal supply chains in SCMCs?

The proposed research framework achieves the study aim. It addresses the formulated research questions by integrating disruption propagation modeling, buffer allocation, and energy-oriented evaluation within a unified SCMC supply chain framework.

This paper is structured as follows. Section 2 reviews previous studies on SCMC supply chains, disruption propagation, Bayesian modeling, and inventory buffering. Section 3 describes the materials and methods used in the study. Section 4 presents the problem definition, notation, and assumptions of the proposed model. Section 5 introduces the analyzed scenarios and the performance evaluation framework for intermodal SCMC supply chains under disruption conditions. Section 6 discusses the results and their managerial implications for buffer allocation and disruption mitigation. Finally, Section 7 outlines the conclusions and directions for future research.

2. Literature Review

2.1. SCMC Supply Chains and Intermodal Logistics Structures

SCMCs constitute a specific form of urban production systems in which manufacturing, logistics, and distribution processes are spatially integrated within dense city environments [1,2]. Earlier studies showed that urban manufacturing can be organized as multi-floor production structures supported by dedicated logistics arrangements rather than conventional industrial layouts separated from the urban fabric [1,3]. This research was later extended through frameworks linking SCMC functioning with city logistics nodes, production network design, and urban freight organization [2,4]. The significance of SCMCs should also be interpreted in the context of adaptive and human-centric manufacturing systems. Industry 5.0 perspectives emphasize that resilient and sustainable production requires integration between physical flows, digital coordination, and decision-support mechanisms [16,17]. This is particularly relevant for SCMCs, where performance depends on coordinating production, logistics, and information flows within spatially constrained environments [18,19]. Accordingly, SCMCs can be understood as interdependent production–logistics systems shaped by structural coupling between transport infrastructure, logistics nodes, and manufacturing buildings [16,18], forming system-dependent networks rather than sets of collocated units [5,20].

Within SCMCs, supply chains are organized as layered intermodal systems. The essence of intermodal transport is efficiently using multiple modes of transportation throughout the supply chain. A key aspect of this approach is minimizing external transport costs while maintaining high delivery reliability. External flows enter through the ILN, which connects to maritime, rail, and road infrastructure and transfers them to CLNs for consolidation, storage, and redistribution before delivery to smart city manufacturing buildings (SCMBs) [2,4,20]. This ILN–CLN–SCMB structure represents a multi-layer architecture in which each layer performs a distinct function while remaining operationally interdependent [2,6]. The intermodal dimension transforms SCMC supply chains into coordinated transport networks combining upstream and intra-urban operations [15,19]. Intermodal freight systems are structurally sensitive to coordination problems because they require synchronization between transport modes, transfer points, and delivery schedules [8,15]. Recent studies indicate that such structures should be analyzed not only in terms of efficiency but also in terms of capacity to absorb and redistribute disturbances across network layers [15,21]. Consequently, SCMC supply chains should be treated as intermodal network systems in which upstream disturbances may directly affect final-stage production performance [8,15]. This network perspective is supported by research on resilience and viability in supply chains, showing that performance depends on node criticality, structural dependencies, and flow configurations rather than linear process logic [7,22,23,24]. Structural efficiency alone is insufficient, as operational continuity depends on interactions among topology, coordination, and recovery capacity [21,25]. In tightly coupled logistics systems, local inefficiencies may generate system-wide effects, particularly when transport, storage, and production functions are interdependent [22,26]. This is especially relevant in SCMCs, where production continuity relies on regular supply flows and coordination at logistics nodes [6,20]. Digitalization further reinforces this system perspective. Data integration, digital monitoring, and AI-supported supply chain management improve visibility and coordination in complex logistics networks [27,28], enabling SCMCs to function as coordinated network systems rather than isolated processes [29,30]. This is critical because operational decisions concerning transport, node utilization, and production cannot be separated from system-wide information availability [16,27].

Overall, SCMC supply chains should be analyzed as intermodal, multi-layer, and system-dependent structures characterized by strong interdependencies between transport organization, logistics nodes, and manufacturing processes [1,4]. However, despite these advances, the literature remains limited in explaining how disruptions propagate through such network structures and how these dynamics affect system-wide operations [6,9].

2.2. Disruption Propagation and Inventory-Based Mitigation in Intermodal SCMC Supply Chains

Disruptions in SCMC supply chains arise from intermodal transport dependencies, coordination failures, and operational variability across logistics nodes. Disturbances originating at transport interfaces or consolidation points propagate through synchronized flows and affect downstream operations, including urban distribution and production processes [6,8]. Multimodal transport structures and territorial integration effects further increase variability and coordination complexity across network layers [31]. Disruption propagation is commonly described as a ripple effect, where local disturbances spread through structural and operational dependencies across the network [9]. In SCMC environments, this mechanism is intensified by the ILN–CLN–B configuration, which creates strong interdependencies between intermodal transport, consolidation processes, and manufacturing activities [10,32]. As a result, local disturbances may evolve into system-wide phenomena affecting multiple layers simultaneously. Probabilistic modeling approaches provide a suitable framework for analyzing disruption propagation under uncertainty. Bayesian network methods represent conditional dependencies between system components and support the analysis of stochastic interactions between disruptions and operational states [33]. The MLBNM extends this approach by incorporating hierarchical and inter-layer relationships, allowing it to represent disruption propagation across transport, logistics, and production layers [32,34]. Empirical studies confirm that disruption dynamics depend on probabilistic interactions between system variables and latent conditions rather than on deterministic relationships alone [35]. In this context, disruption propagation constitutes a key dynamic mechanism shaping SSCR in intermodal SCMC systems [14].

System performance under disruption is closely related to resilience and the ability to maintain material availability for production processes. In SCMCs, production continuity depends on synchronized inbound flows, making the system highly sensitive to supply disturbances [7,22]. Broader system characteristics, including organizational performance and sustainability conditions, also influence disruption impacts [36,37]. These factors reinforce the need for mitigation mechanisms directly related to material flow stability. Inventory is the primary operational mechanism linking disruption propagation and system performance. Reserve capacity in the form of SS or BS absorbs variability in supply flows and limits the amplification of disturbances across network layers [11,12]. In intermodal SCMC systems, buffer allocation must reflect both disruption probability and inter-layer dependencies, as shortages at one layer may propagate and affect downstream operations. Contemporary inventory models under uncertainty highlight the importance of integrating stochastic demand, disruption likelihood, and system dynamics into buffer allocation decisions [38,39]. Sustainable inventory management approaches also indicate that buffer policies must align with environmental and operational objectives, showing that buffer levels influence both system efficiency and environmental impact [40]. Data-driven and machine learning-based approaches support adaptive inventory control by enabling dynamic adjustment of buffer levels in response to changing system conditions [41,42], while advanced analytics improve demand forecasting and decision-making accuracy in uncertain environments [43]. Inventory-based mitigation introduces trade-offs between robustness, cost, and resource utilization. Higher buffer levels improve protection against disruptions but increase holding costs and operational burden [44]. Sustainability-oriented studies indicate that excessive inventory can negatively affect environmental performance due to additional storage and handling requirements [45,46]. At the same time, circular economy perspectives emphasize the need to balance inventory levels with resource efficiency and system sustainability [47]. In this respect, inventory-based mitigation represents a key operational mechanism supporting SSCR by stabilizing material flows under uncertainty [14]. Alternative mitigation measures include switching transport modes, for example, from maritime transport to air transport during the COVID-19 pandemic or relocating the production of materials and semi-finished products closer to the final assembly and finished goods manufacturing sites according to the concept of nearshoring [48].

The effectiveness of inventory-based mitigation depends on the supply chain’s structural configuration. Network design determines how disruptions propagate and how buffers can absorb variability across system layers [49]. Scenario-based and stochastic approaches demonstrate that optimal inventory placement requires simultaneous consideration of network topology and disruption dynamics [50]. Digitalization further enhances inventory-based mitigation by improving system visibility and predictive capabilities. AI-supported supply chain systems enable early detection of disruptions and support dynamic adjustment of inventory policies [27,28]. At the same time, data-driven frameworks facilitate coordination across multiple nodes and operational layers in intermodal SCMC systems [29,51]. Artificial intelligence applications further improve forecasting accuracy and enable adaptive control of inventory systems under uncertainty [52,53]. Still, they also introduce new dependencies and risks that resilience-oriented supply chain models must account for [54,55,56].

The reviewed literature demonstrates that multi-layer probabilistic dependencies govern disruption propagation in intermodal SCMC supply chains, while mitigation is primarily achieved through inventory-based mechanisms. However, existing studies typically analyze disruption propagation and inventory optimization separately, without explicitly linking probabilistic modeling to integrated system-level decision frameworks. This limitation becomes particularly relevant when considering transport-energy performance in intermodal SCMC supply chains. Inventory decisions influence material availability and system resilience and indirectly affect transport energy consumption through their impact on transport activity and flow organization. However, the energy dimension remains insufficiently integrated within SSCR-oriented modeling frameworks. Extending the analytical framework therefore requires incorporating transport-energy performance into disruption- and inventory-oriented analysis of intermodal SCMC supply chains.

2.3. Transport-Energy Performance in Intermodal Supply Chains in Smart City Contexts

Transport-energy performance has become a central dimension of intermodal supply chain design in smart city environments, where logistics systems operate under increasing environmental and regulatory constraints. In SCMC contexts, transport energy consumption is shaped by transport intensity, vehicle utilization, and the organization of material flows. This perspective requires integrating transport energy considerations into supply chain design and operational decision-making [57,58]. Recent research emphasizes the transition towards zero-emission supply chains, in which logistics operations are progressively decarbonized through the electrification of transport and the integration of renewable energy sources [59,60]. In intermodal SCMC systems, this transition primarily affects urban distribution and short-haul transport segments, where e-trucks and e-vans are increasingly deployed. The effectiveness of such systems depends not only on vehicle technologies but also on the availability and spatial distribution of charging infrastructure, which becomes a critical element of network performance and operational feasibility [20,61]. Charging infrastructure constitutes a key component of transport-energy systems in SCMC supply chains. Charging facilities for e-trucks and e-vans are typically integrated into logistics nodes such as ILNs and CLNs, enabling coordination between transport operations and handling processes [62]. Renewable energy sources, particularly photovoltaic systems installed at logistics facilities and accompanied by energy storage facilities, are increasingly used to supply these charging systems [63]. Charging operations can be synchronized with loading and unloading activities, allowing vehicles to be charged during handling processes without introducing additional operational delays [62]. This configuration can improve the energy performance of electric freight transport.

The energy performance of SCMC supply chains is also strongly influenced by logistics nodes, particularly ILNs and CLNs, which are energy-intensive due to storage, handling, and transshipment operations. Integration of renewable energy sources enables partial energy self-sufficiency and reduces dependence on external energy supply. In this context, logistics nodes may operate as energy prosumers, simultaneously consuming and generating energy within the system. Such configurations help stabilize energy demand and improve overall system efficiency [64,65]. These facility-related energy flows provide a broader context for the system but remain outside the energy-accounting boundary of the present model, which is limited to electricity consumption by e-trucks and e-vans.

An important aspect of energy performance relates to the energy losses associated with disruptions in transport and handling processes. Transport disruptions (TD), such as delays, rerouting, and stop-and-go conditions, increase energy consumption due to inefficient vehicle utilization and extended operation times [52,66]. Similarly, handling disruptions at logistics nodes, including waiting times, congestion, and rescheduling of loading and unloading operations, lead to additional energy use related to equipment operation, idle time, and repeated handling activities [1,67]. These effects may increase the energy consumption associated with transport and handling operations, particularly in tightly coupled intermodal SCMC systems. In the present study, we quantify only the transport component of these disruption-related energy effects.

At the operational level, transport-energy performance is closely related to inventory and transport decisions. Buffer allocation can affect transport frequency, vehicle utilization, material availability, and production continuity [14,42]. Changes in buffer levels can alter transport activity and vehicle utilization under disruption conditions [14,68]. These interactions link inventory-based disruption mitigation with the transport activity required to maintain material flows under uncertain conditions. Evaluating this relationship requires an integrated framework that combines disruption propagation, inventory buffering, and transport-energy assessment.

Despite growing interest in sustainable and energy-aware supply chains, existing studies typically address electrification, renewable energy integration, and logistics optimization separately. Limited attention has been given to the combined analysis of disruption propagation, inventory-based mitigation, and transport-energy performance within intermodal SCMC systems. In particular, the effects of buffer allocation and disruption propagation on transport activity, production continuity, and associated transport energy consumption remain underexplored. Addressing this gap requires modeling frameworks that explicitly link transport-energy performance with disruption dynamics and operational decision variables. This approach enables evaluation of how buffer allocation and transport configuration influence transport-energy performance under uncertainty. The proposed model responds to this need by integrating transport-energy assessment into the analysis of intermodal SCMC supply chains, extending disruption-aware and inventory-based approaches.

3. Materials and Methods

Figure 1 presents a large urban area structured around multiple SCMCs, spatially distributed within the large city and forming decentralized production–logistics units. Each SCMC is organized around a CLN, which performs local consolidation, inventory buffering, and distribution functions, ensuring coordination of material flows within the cluster. The conceptual structure reflects selected European metropolitan regions, including Berlin, Amsterdam, and Łódź, providing a spatial reference for the generic SCMC configuration rather than direct empirical parameterization of the model. A common feature of these cities is their urban structure, comprising numerous districts that form separate clusters, linked by a multi-lane ring road which serves as the main access route to the various parts of the city (Figure 1).

External intermodal flows enter the system through an ILN, the primary interface between long-distance transport systems and the urban logistics network. Two upstream configurations are considered: (i) sea–land transport chains with transshipment at a port terminal, and (ii) land-based intermodal connections supported by inland terminals and rail corridors. These configurations enable the integration of different transport modes within a unified supply structure.

From the ILN, the freight is distributed to individual SCMCs via road-based links, forming the urban transport layer. This layer allocates flows across clusters and synchronizes deliveries with local demand and production requirements. The presented structure reflects a hierarchical integration of interurban intermodal corridors and intra-urban cluster-based logistics, where decentralized coordination centers play a key role in organizing efficient, synchronized supply flows across the urban system.

The study adopts a structured, multi-stage research design to ensure methodological transparency and reproducibility. The overall framework combines theoretical and analytical elements with simulation-based evaluation. It is grounded in a literature review that provides the conceptual basis linking SSCR, smart city manufacturing, disruption propagation, and inventory management. The research procedure is organized into the following stages:

• Model development—The MLBNM is utilized to capture causal relationships and represent uncertainty related to disruption propagation within intermodal SCMC supply chains, thereby establishing the probabilistic foundation of the proposed TEPM of Intermodal Supply Chains.
• Model formulation—Building on this foundation, the proposed TEPM of Intermodal Supply Chains is constructed to integrate disruption propagation, BS allocation, and transport-energy performance assessment under uncertainty.
• Simulation-based evaluation—The model is examined using a case-based simulation approach with representative SCMC data to assess its behavior under different disruption scenarios and inventory configurations.
• Analysis and implications—The obtained results are evaluated to derive managerial and theoretical insights, focusing on transport-energy performance, operational resilience, and production continuity. The analysis also identifies potential directions for future research in sustainable urban manufacturing systems.

This structured approach ensures consistency between conceptual assumptions, model development, and simulation-based evaluation, while addressing intermodal supply chain uncertainty, resilience, and transport-energy performance within the proposed TEPM framework.

The modeling framework is based on Material Flow Analysis combined with the MLBNM [10,14,68]. MLBNM is applied to model disruption propagation in intermodal SCMC supply chains and to analyze its impact on inventory levels, including the allocation of SS and BS at CLNs and SMEs of SCMCs. The approach captures probabilistic dependencies between disruption sources, transport operations, logistics nodes, and material availability, enabling the analysis of uncertainty propagation across the network [10,33,68]. Its multi-layer structure reflects interactions between intermodal transport, ILNs, CLNs, and production systems, allowing the assessment of how local disturbances affect system-wide performance. In contrast to deterministic or purely simulation-based approaches, MLBNM provides a stochastic framework that supports dynamic updating of system states under changing conditions [10,32]. This is particularly relevant for evaluating buffer allocation strategies in response to disruptions and their impact on transport-energy performance. Building on this modeling feature, MLBNM enables the integrated analysis of disruption propagation, buffer allocation, and transport-energy performance within intermodal SCMC supply chains.

Figure 2 shows the evolution of the modeling approaches applied to intermodal SCMC supply chains, comparing the deterministic decision-support model (M1) [1], the stochastic inventory management model based on MLBNM (M2) [14], and the proposed TEPM (M3). Model M1 represents a deterministic framework focused on flow coordination between ILN, CLNs, and SCMBs, assuming stable operating conditions and neglecting disruption effects [1]. Model M2 extends this approach by incorporating stochastic disruption propagation using MLBNM and introducing the SS mechanism to support system resilience under uncertainty [14]. For clarity and consistency, the stock maintained at the CLN, previously denoted as BS in earlier versions of the model, is referred to as SS throughout the remainder of this paper. This terminology more accurately reflects inventory’s role in supporting supply chain resilience.

The proposed M3 model retains the MLBNM-based disruption-propagation mechanism and CLN’s SS incorporated in M2, while broadening the model’s operational scope and inventory structure. M3 distinguishes disruptions affecting e-truck transfers from the ILN to the CLN, handling processes at the CLN, e-van transfers from the CLN to individual SCMBs, and unloading at the SCMBs [14]. Waiting-state variables track production material temporarily unavailable during disrupted transport or handling operations. M3 also incorporates BS at individual SCMBs in addition to SS at the CLN, allowing inventory protection to be assessed at both the network and manufacturing-building levels.

These disruption and inventory states relate to production continuity, vehicle transfer activity, vehicle utilization, annual transport energy consumption, and energy consumption per effective building-production day. Thus, M3 integrates operationally differentiated disruption states, waiting-state material flows, two-level inventory buffering, production-continuity assessment, and transport-energy performance within a single operational model.

From an operational perspective, the TEPM follows a sequential decision logic. Disruption states determine whether production material can proceed through the ILN–CLN–SCMB network or waits temporarily. SS at the CLN provides network-level protection against upstream disruptions, whereas BS at individual SCMBs provides local protection against disruptions affecting final delivery and handling. The resulting material availability determines production continuity, while the associated e-truck and e-van transfers determine total transport energy consumption. Transport-energy intensity per effective building-production day links this energy consumption to the production continuity achieved under each analyzed inventory and disruption configuration.

Quantitative assessment of material flows and inventories was performed using Material Flow Analysis [1]. The model uses the following key performance indicators: the number of intelligent reconfigurable trolleys (IRTs) used as intermodal transport units [69], vehicle capacity utilization levels, the number of e-truck and e-van transfers, and inventory levels represented by SS, BS, and overnight stock (OS) [8,14]. These parameters support decision-making regarding inventory control strategies in SCMC supply chains, reflecting different operational policies such as “zero shortages” or “acceptable waiting time for supplies”. For clarity, SS, BS, and OS represent distinct inventory concepts in the TEPM. SS refers to inventory maintained at the CLN for network-level protection, whereas BS refers to inventory maintained at individual SCMBs for local protection against material shortages. OS is not a predefined protective stock level; it represents the remaining production material at the CLN or SCMB after satisfying daily demand.

The proposed M3 model was implemented in MATLAB (version 2025b) and evaluated through scenario-based simulation of SCMC system performance under disruption conditions, varying BS levels, and different transport configurations.

4. Problem Definition, Notation, and Assumptions

4.1. Problem Definition

Intermodal supply chains in urban environments are constrained by limited space, fragmented demand, and the need to coordinate multiple logistic functions within dense city structures. In SCMCs, production, consolidation, and distribution activities operate in proximity, requiring continuous, synchronized flows across interconnected network layers. The system integrates production sites, ILNs, CLNs, and urban demand zones. High infrastructure utilization, restricted access conditions, and limited handling capacity lead to congestion at CLNs and delays in transshipment processes. Figure 3 presents the structure of the considered intermodal supply chain and the associated disruption propagation processes in the context of SCMC operations.

Local disturbances arising at any node may propagate through the network, affecting both upstream supply and downstream distribution. SMEs within SCMCs are particularly sensitive to such disruptions. Limited storage capacity and dependence on frequent deliveries reduce their ability to absorb flow variability. As a result, delays at ILNs or CLNs quickly translate into production interruptions and delivery inefficiencies. This effect is amplified in intermodal systems, where synchronization between transport and handling stages is required. Urban traffic conditions further increase system instability. Congestion, travel time variability, and stop-and-go dynamics reduce transport reliability and increase energy consumption. Inefficient routing, extended vehicle operation times, and repeated handling operations increase energy intensity in freight flows, which conflicts with urban sustainability objectives [14].

A key limitation of intermodal supply chains in SCMCs is the lack of coordinated mechanisms linking disruption propagation with buffer allocation decisions. While centralized structures such as CLNs enable flow consolidation, they also increase system sensitivity to local disturbances, which can propagate across network layers and disrupt material availability and production continuity. At the same time, BSs are often allocated based on static or local criteria, without considering their role in mitigating disruption effects within the overall network structure. This disconnect leads to suboptimal system performance. Insufficient buffering amplifies disruption impacts, whereas excessive buffering increases inventory requirements and operational burden. As a result, the relationship between disruption dynamics, buffer allocation, and transport-energy performance remains insufficiently understood in intermodal SCMC supply chains.

Existing approaches typically analyze disruption effects, buffer strategies, or transport-energy performance separately, without capturing their interdependencies in multi-layer logistics networks. In SCMCs, where production and logistics processes are tightly coupled, this separation limits the ability to assess system resilience and transport-energy performance under disturbance conditions jointly.

This study addresses these limitations by developing a model that integrates disruption propagation and buffer allocation with transport-energy performance assessment in intermodal SCMC supply chains. The proposed approach combines probabilistic modeling based on the MLBNM with network-dependent buffer allocation and a transport-energy assessment framework [14,70]. This integrated perspective enables the analysis of how disruptions propagate across the network, how BSs should be allocated to mitigate their impact, and how these decisions influence transport-energy performance.

4.2. Notation

Table 1. List of indices and parameters.

Indices and Parameters		Unit
Indices
$a_{md}$	availability of m-type production material on day d ($a_{md}=$ 1—available, $a_{md}=$ 0—unavailable)
$d$	day d = 1, 2, 3, … 365
$m$	type of production material, $m=A,B,C,\dots P$
$n$	number of SCMBs in the SCMC, $n=\mathrm{1,2},3,\dots$, N
k	number of SCMCs and CLNs in a large city, $k=\mathrm{1,2},3,\dots$, K
${\delta }_d^{truck}$	compliance of e-truck transit (ILN→CLN) with the schedule on day d (1: undisrupted, 0: disrupted)
${{\delta }_d}^{tranship}$	compliance of the transshipment process in CLN with the schedule on day d (1: undisrupted, 0: disrupted)
${\delta }_d^{van\left(n\right)}$	compliance of e-van transport (CLN→SCMB(n)) with the schedule on day d (1: undisrupted, 0: disrupted)
${\delta }_d^{unload\left(n\right)}$	compliance of the unloading process in SCMB(n) with the schedule on day d (1: undisrupted, 0: disrupted)
$R^{van }$	maximum cargo capacity of e-van	ITR
$R^{truck }$	maximum cargo capacity of e-truck	ITR
Parameters
Independent parameters
$B{S}_{m,d}^{SCMB\left(n\right)}$	volume of BS of m-type of production material at SCMB	ITR
$B{S}^{SCMB\left(n\right)}$	total volume of BS	ITR
BSL	BS level $BSL=B{N}_A{N}_B\dots N_P$, where ${N_A,N}_B\dots N_P$, denote the BS levels assigned to individual production materials. $N_A{N}_B\dots N_P=\{\mathrm{0,1},2\dots \}$	ITR
${SS}_m$	volume of SS of m-type of production material	IRT
$SS$	total volume of SS	ITR
$SSL$	SS level	%
${\eta }_{undisrupt}^{truck}$	energy-consumption coefficient per undisrupted e-truck transfer	kWh
${\eta }_{disrupt}^{truck}$	energy-consumption coefficient per disrupted e-truck transfer	kWh
${\eta }_{undisrupt}^{van}$	energy-consumption coefficient per undisrupted e-van transfer	kWh
${\eta }_{disrupt}^{van}$	energy-consumption coefficient per disrupted e-van transfer	kWh
Dependent parameters
$e_{md}^{van\left(n\right)}$	surplus volume of m-type production material additionally loaded onto e-vans on day d	ITR
$e_{md}^{truck}$	surplus volume of m-type production material additionally loaded onto e-trucks on day d	ITR
$E_Y$	total annual transport energy consumption	kWh
$E_{avg}$	average daily transport energy consumption	kWh/day
$E_{prod\_day}^B$	transport-energy intensity per effective building-production day	kWh/bpd
$N_{undisrupt}^{truck}$	number of undisrupted e-truck transfers during one year	transfers
$N_{disrupt}^{truck}$	number of disrupted e-truck transfers during one year	transfers
$N_{undisrupt}^{van}$	number of undisrupted e-van transfers during one year	transfers
$N_{disrupt}^{van}$	number of disrupted e-van transfers during one year	transfers
$N_Y$	total number of vehicle transfers during one year	transfers
$N_{PROD}^{SCMB\left(n\right)}$	number of production days at SCMB($n$) during one year	days
$Q_{m,d}^{CLN}$	OS of m-type of production material at the CLN after day d	IRT
$Q_{md}^{SCMB\left(n\right)}$	OS of m-type of production material at the SCMB after day d	tonnes, IRT
$T_{md}^{truck}$	volume of m-type of production material affected by disrupted e-trucks transit on day d	IRT
$T_{md}^{tranship}$	volume of m-type of production material affected by disrupted transshipment process in CLN on day d	IRT
$T_{md}^{van\left(n\right)}$	volume of m-type of production material affected by disrupted e-vans transit on day d	IRT
$T_{md}^{unload\left(n\right)}$	volume of m-type of production material affected by disrupted unloading process in SCMB(n) on day d	IRT
$S_{md}^{SCMB\left(n\right)}$	supply of m-type of production material based on the order of SCMB(n) on day d	IRT
$S_{md}^{SCMB\left(n\right),arr}$	amount of m-type production material delivered to SCMB($n$) on day $d$	ITR
$S_{md}^{CLN}$	delivery of m-type of production material to the CLN on day d	IRT
$S_{md}^{CLN,avail}$	amount of m-type production material available for dispatch from the CLN to the SCMBs on day $d$	ITR
$S_{md}^{ILN}$	available quantity of m-type production material at the ILN on day $d$	ITR
${SD}_m^{CLN}$	cumulative period of shortage of m-type of production material at the CLN during one year	days
${SD}_m^{SCMB\left(n\right)}$	cumulative period of shortage of m-type of production material at the SCMB(n) during one year	days
${SV}_m^{CLN}$	shortage of m-type of production material at the CLN on day d	IRT
${SV}_m^{SCMB\left(n\right)}$	shortage of m-type of production material at the SCMB on day d	IRT
$U_Y^{truck}$	annual average e-truck loading capacity utilization rate	-
$U_Y^{van}$	annual average e-van loading capacity utilization rate	-

lists all indices and parameters used in the subsequent stages of the analysis. The key parameters of the SCMC operational model include daily demand for production materials ($D_{md}^{CLN}, D_{md}^{CMFMB\left(n\right)}$) reported by CLNs and SCMBs, shortage of production materials in CLN and SCMBs ($S{V}_m^{CLN}, S{V}_m^{CMFMB\left(n\right)}),$ inventories (SSs and BSs) volumes ($S{S}_m^{CLN}, B{S}_m^{SCMB\left(n\right)}$) kept in CLN and SCMBs, and OS volumes ($Q_{md}^{CLN}, Q_{md}^{CMFMB\left(n\right)}$) kept in CLN and SCMBs.

Compared to the previous study, this analysis extends the notation by introducing disruption indicators (δ) representing equipment and transport failures, waiting-state parameters ($T_{md}^{unloadCLN}$,$T_{md}^{truck}$,$T_{md}^{van}$,$T_{md}^{unload\left(n\right)})$describing production material temporarily unavailable due to disrupted operations, annual transport energy consumption indicators ($E_Y$, $E_{avg}$, $E_{prod\_day}^B$), and vehicle loading capacity utilization indicators ($U_Y^{truck}$, $U_Y^{van}$).

The parameters were classified as independent or dependent variables in accordance with the research assumptions.

4.3. Assumptions

For the TEPM, compared to the previous inventory management model, extensive assumptions were established regarding inventory management under supply chain uncertainty. Building upon the disruption framework established in the previous study [14], this proposed model extends disruption modeling to encompass the entire SCMC intermodal supply chain, from long-distance transport preceding ILN supply to final delivery to SCMB production facilities. Figure 4 presents a new study area within the supply chain network with identified new urban-scale disruption points affecting system operations. Blue indicates disruptions associated with ILN–CLN transport and CLN operations, whereas green indicates disruptions associated with CLN–SCMB transport and SCMB operations.

Five categories of operational disruptions are explicitly modeled using binary disruption indicators (δ):

• Production material availability disruptions at the ILN ($a_{md}$): unavailability of m-type material at the ILN on day d due to upstream supply chain issues.
• E-truck transit disruptions (${\delta }_d^{truck}$): transport failures on the ILN to CLN route.
• Transshipment disruptions (${\delta }_d^{tranship}$): technical failures causing delays in cargo handling at the CLN.
• E-van transit disruptions (${\delta }_{nd}^{van\left(n\right)}$): transport failures on the CLN to SCMB(n) routes.
• Unloading disruptions (${\delta }_d^{unload\left(n\right)}$): equipment failures causing delays in unloading operations at SCMB(n).

When disruptions occur (δ = 0), affected IRTs enter temporary waiting states until operations resume, as detailed in Section 5.1. The probabilities of handling disruptions (breakdowns in handling and/or storage areas) and transport disruptions (accidents, congestion, and vehicle failures), illustrated in Figure 4, were derived using Bayesian network analysis following the approach developed in the previous study [14]. In the present study, these probabilities are used as input parameters for generating disruption scenarios rather than being re-estimated from new empirical data. The TEPM adopts the following assumptions to define the operational configuration of the analyzed scenarios; these assumptions should not be interpreted as universal characteristics of SCMC supply chains:

• Orders from SCMB enterprises for production materials are processed according to two rules: ‘next-day delivery’ (orders collected on day d are delivered on day d + 1) and ‘completion to vehicle capacity’ (material volumes may be increased to optimize vehicle utilization). Supply shortages may occur due to: (a) product unavailability at the ILN ($a_{md} = 0$), (b) transport or equipment disruptions (δ = 0) delaying delivery (c) any combination of the above affecting any material type.
• In the analyzed scenarios, the demand for production materials ordered by SCMB enterprises is assumed to vary according to next-day production plans generated independently for each SCMB. Consistent with the previous model configuration, the maximum possible daily fluctuation in demand for each production input is set within the range of 0–100% [1,14]. The analyzed scenarios further assume that materials are ordered shortly before their planned use, with minimal inventory held on site.
• Each CLN maintains two types of IRT storage: (a) SS area ($S{S}_m^{CLN}$): storage reserved for SS maintained to ensure continuity of supply during supply disruptions (i.e., when $a_{md} = 0$), (b) OS area ($Q_{md}^{CLN}$) for IRTs exceeding immediate demand due to vehicle capacity optimization. CLN facility parameters constrain total storage capacity. In case of transshipment disruptions (${\delta }_d^{tranship} = 0$), affected IRTs temporarily accumulate in waiting states ($T_{md}^{tranship}$) until operations resume.
• Each SCMB has severely limited storage capacity represented in the model by the buffer size parameter (BS). As small and medium enterprises typically occupy multi-story urban buildings, available space prioritizes production over storage. Pre- and post-production storage must be minimized. Additional constraints arise from floor load limits and freight elevator capacity. In case of unloading disruptions (${\delta }_d^{unload\left(n\right)}= 0$), affected IRTs temporarily accumulate in waiting states ($T_{md}^{unload\left(n\right)}$) until operations resume.
• The energy-accounting boundary of the TEPM is limited to electricity consumption associated with e-truck and e-van operations. Transport operations required to replenish SS at the CLN and BS at the SCMBs are included in the energy assessment through the corresponding vehicle transfers. Energy consumption associated with inventory storage, handling equipment, and other facility operations is outside the scope of the present model.
• The energy consumption by vehicles varies depending on road conditions and unforeseen events occurring in road transport. For this analysis, two states of energy consumption in road transport are considered: ‘undisrupted’—normal transport conditions, and ‘disrupted’—conditions in which an event has occurred, such as an accident, vehicle failure, or traffic congestion.
• Vehicles are charged at two locations: ILN and CLN. Dedicated charging stations are provided for e-trucks at ILN and for e-vans at CLN. Their number and capacity ensure that vehicles can be charged without delays during breaks between transfers.

5. Results

5.1. Model Formulation

The model utilizes input data on production material demand and vehicle utilization levels to determine optimal inventory levels and recommend transportation options for IRT loads. The solutions are consistent and deterministic, given the defined operational assumptions and decision structure. Building on the model developed by Wiśnicki et al. [1,14], an extension was introduced to simulate supply chain disruptions. These disruptions represent periods during which goods are unavailable at various supply chain nodes due to operational or stochastic factors. This mechanism reflects the broader body of research on the ripple effect in supply networks [9,23]. The probabilities of disruption consequences at individual supply chain nodes were previously derived using Bayesian network analysis, following the approach in [14]. The trigger–event–consequence causal structure enables modeling cause-and-effect relationships under the uncertainty of the SCMC supply chain [9,27]. In the present TEPM, these probabilities are treated as predefined stochastic input parameters adopted from the previously developed MLBNM framework rather than as probabilities estimated from a new empirical dataset.

Based on these predefined probabilities, the model generates random sets of days when a given product is unavailable ($a_{md} = 0$) or when transport and operational disruptions occur (δ = 0). A single fixed realization of the generated disruption events was used consistently across all analyzed SSL and BSL configurations. This common disruption input, shown in Figure 5, provides a consistent basis for comparing alternative inventory configurations without introducing variation from different stochastic disruption realizations. Product availability matrices and disruption indicators then serve as inputs for subsequent decision-making stages. In this simplified probabilistic model, the consequences of disruptions are analyzed rather than their detailed root causes.

Moreover, the decision-support model presented in [14] is modified or upgraded, as follows:

Delivery of m-type of production material based on the order of CLN:

$$S_{md}^{CLN}={\delta }_d^{truck}\cdot \left(a_{md}\cdot S_{md}^{ILN}+e_{md}^{truck}\right)+T_{md-1}^{truck}$$

(1)

To account for disruptions occurring in transport and handling operations, the proposed model introduces additional state variables describing production material that is temporarily unavailable.

The amount of production material delayed during e-truck transport between the ILN and the CLN on a given day is given by:

$$T_{md}^{truck}=\left(1-{\delta }_d^{truck}\right)\cdot {(a}_{md}\cdot S_{md}^{ILN}+e_{md}^{truck})$$

(2)

The variable represents the cargo delayed on a given day. This delayed cargo is carried over to the following day and is then included in the material delivered to the CLN. Consequently, disruptions occurring on consecutive days affect separate daily vehicle transfers and may generate successive delayed loads.

Production material awaiting transshipment at the CLN is determined as:

$$T_{md}^{tranship} =\left(1 -{\delta }_d^{tranship} \right)\cdot \left(S_{md}^{CLN}+ T_{md-1}^{tranship} \right)$$

(3)

Based on these quantities, the amount of production material available for dispatch from the CLN to the SCMBs is determined (Equation (4))

$$S_{md}^{CLN,avail}={\delta }_d^{tranship}\cdot \left(S_{m,d}^{CLN}+T_{md-1}^{tranship}\right)$$

(4)

Material affected by a transshipment disruption remains in the corresponding waiting state and becomes available only when the transshipment operation resumes. Consistent with the previous version of the model [14], the quantity of production material dispatched from the CLN to each SCMB consists of the material assigned to that SCMB and, when required, additional production material loaded to improve e-van capacity utilization.

The following variables describe the material flow between the CLN and individual SCMBs (Equations (5)–(7)):

Material delayed in transit from CLN to SCMB(n):

$$T_{md}^{van\left(n\right)} =\left(1 -{\delta }_d^{van\left(n\right)}\right)\cdot \left(S_{md}^{SCMB\left(n\right)}+e_{md}^{van \left(n\right)}\right)$$

(5)

Material delivered to SCMB(n):

$$S_{md}^{SCMB\left(n\right),arr} ={\delta }_d^{van\left(n\right)} \cdot \left(S_{md}^{SCMB\left(n\right)}+e_{md}^{van \left(n\right)}\right)+T_{md-1}^{van\left(n\right)}$$

(6)

As in the case of e-truck transport, this represents the load delayed on a given day, which is carried over and delivered to the corresponding SCMB on the following day. Consecutive transport disruptions therefore affect separate daily e-van transfers and may generate successive delayed loads.

IRTs awaiting unloading at SCMB(n):

$$T_{md}^{unload\left(n\right)}=\left(1-{\delta }_d^{unload\left(n\right)}\right)\cdot \left(S_{md}^{SCMB\left(n\right),arr}+T_{md-1}^{unload\left(n\right)}\right)$$

(7)

Material affected by an unloading disruption remains temporarily unavailable for production until the unloading operation resumes. If unloading disruptions occur on consecutive days, previously delayed material remains waiting together with newly arriving material.

Finally, the quantity of production material available for manufacturing processes at each SCMB is determined in two stages. First, the amount of regularly available material is calculated by combining the current-day arrivals, material released from the previous unloading waiting state, and the overnight stock carried over from the preceding day:

$$\begin{array}{c}{S}_{md}^{SCMB\left(n\right),avail}\\ =\left\{\begin{array}{cc}{S}_{md}^{SCMB\left(n\right),arr}+Q_{md-1}^{SCMB\left(n\right)}+T_{md-1}^{unload\left(n\right)}& ,S_{md}^{SCMB\left(n\right),arr}>0 \wedge {\delta }_d^{unload\left(n\right)}=1\\ Q_{md-1}^{SCMB\left(n\right)}+B{S}_{md-1}^{SCMB\left(n\right)}& ,S_{md}^{SCMB\left(n\right),arr}=0\wedge {\delta }_d^{unload\left(n\right)}=1\\ 0& {,\delta }_d^{unload\left(n\right)}=0\end{array}\right.\end{array}$$

(8)

When BS is used to compensate for the absence of regular material supply, only the quantity actually withdrawn from BS is included in the replenishment requirement and added to the following-day order. The CLN then aggregates this replenishment requirement and reflects it in the order it places with the ILN.

The remaining production material after satisfying daily demand constitutes the overnight inventory at the CLN (Equation (9)) and at each SCMB (Equation (10)). At the SCMB, production is carried out only when unloading operations are available, and the quantities of all required production materials are sufficient to satisfy the corresponding daily demand. The overnight inventory at CLN and at SCMB(n) are therefore determined as:

$$Q_{md}^{CLN} =\max \left(0,S_{md}^{CLN,avail}+ Q_{md-1}^{CLN}- \sum _{n=1}^N{S}_{md}^{SCMB\left(n\right),arr}\right)$$

(9)

$$\begin{array}{c}\begin{array}{cc}\hfill Q_{md}^{SCMB\left(n\right)}=& \max (0,Q_{md-1}^{SCMB\left(n\right)}+S_{md}^{SCMB\left(n\right),arr}+T_{md-1}^{unload\left(n\right)}-(D_{md}^{SCMB\left(n\right)}\hfill \\ & -W_{md}^{SCMB\left(n\right)}\left)\right){, if \delta }_d^{unload\left(n\right)}=1\wedge S_{md}^{SCMB\left(n\right),avail}\ge D_{md}^{SCMB\left(n\right)} \forall m\in M\hfill \end{array}\\ \mathrm{or}\\ \begin{array}{cc}\hfill Q_{md}^{SCMB\left(n\right)}=& Q_{md-1}^{SCMB\left(n\right)}+S_{md}^{SCMB\left(n\right),arr}+T_{md-1}^{unload\left(n\right)}, {if \delta }_d^{unload\left(n\right)}=1\wedge \exists m\hfill \\ & \in M{S}_{md}^{SCMB\left(n\right),avail}<D_{md}^{SCMB\left(n\right)}\hfill \end{array}\\ \mathrm{or}\\ Q_{md}^{SCMB\left(n\right)}=Q_{md-1}^{SCMB\left(n\right)}{, if \delta }_d^{unload\left(n\right)}=1\end{array}$$

(10)

where $D_{md}^{SCMB\left(n\right)}$ denotes the daily demand for production material at SCMB(n), and $W_{md}^{SCMB\left(n\right)}$ denotes the quantity of material actually withdrawn from BS. If the required quantities of all production materials are available, daily demand is satisfied, and only the remaining material is retained as overnight stock. If at least one required production material is insufficient, production is not carried out, and the physically available material is carried over to the following day. Material awaiting unloading is maintained separately in the corresponding waiting state and is not included in the overnight stock.

At both inventory locations, protective stock is restored before any remaining surplus is classified as overnight stock. At the CLN, material required to restore the predefined SS level is allocated to SS first, and only the remaining surplus is retained as overnight stock. Analogously, at each SCMB, material required to restore the predefined BS level is allocated to BS first, while any remaining surplus is retained as overnight stock. The quantities withdrawn from SS and BS are included in the replenishment requirement and reflected in the following-day order.

The extended material flow model described above enables the determination of the daily number of fully completed e-truck and e-van transfers throughout the analyzed supply chain.

Thus, the total number of vehicle transfers per year:

$$N_{disrupt}^{truck}=\sum _d{N}_{d,disrupt}^{truck}$$

(11)

$$N_{undisrupt}^{truck}=\sum _d{N}_{d.undisrupt}^{truck}$$

(12)

$$N_{disrupt}^{van}=\sum _{n\in N}\sum _d{N}_{d.disrupt}^{van\left(n\right)}$$

(13)

$$N_{undisrupt}^{van}=\sum _{n\in N}\sum _d{N}_{d.undisrupt}^{van\left(n\right)}$$

(14)

$$N_Y=N_{disrupt}^{truck}+N_{undisrupt}^{truck}+N_{disrupt}^{van}+N_{undisrupt}^{van}$$

(15)

where

• $N_{disrupt}^{van},N_{d.undisrupt,}^{truck} N_{d.disrupt}^{van},N_{d.undisrupt}^{van}:$number of e-truck and e-van transfers carried out on day d.

The annual average vehicle loading capacity utilization rate is calculated as the ratio of the total volume of production material loaded into the vehicles to the total loading capacity of all e-truck or e-van transfers performed during the simulation period.

$$U_Y^{truck}={\displaystyle \frac{\sum _d\left(\sum _m\left(a_{md}\cdot S_{md}^{ILN}+e_{md}^{truck}\right)\right)}{R^{truck}\cdot \left(N_{disrupt}^{truck}+N_{undisrupt}^{truck}\right)}}$$

(16)

$$U_Y^{van}={\displaystyle \frac{\sum _d\sum _n\sum _m\left(S_{md}^{SCMB\left(n\right)}+e_{md}^{van\left(n\right)}\right)}{R^{van}\cdot \left(N_{disrupt}^{van}+N_{undisrupt}^{van}\right)}}$$

(17)

These vehicle flows constitute the basis for the transport-energy assessment of the proposed logistics system. The total transport energy consumption is calculated from the number of vehicle transfers and the corresponding energy-consumption coefficients assigned to road transport-related disruptions.

$$E_Y ={\eta }_{disrupt}^{truck}\cdot N_{disrupt}^{truck}+{\eta }_{undisrupt}^{truck}\cdot N_{undisrupt}^{truck}+{\eta }_{disrupt}^{van}\cdot N_{disrupt}^{van}+{\eta }_{undisrupt}^{van}\cdot N_{undisrupt}^{van}$$

(18)

The average daily energy consumption is calculated as the ratio of the total annual energy consumption to the number of simulation days:

$$E_{avg} ={\displaystyle \frac{E_Y}{365}}$$

(19)

Transport-energy intensity per effective building-production day accounts for production continuity and is calculated as the ratio of total annual transport energy consumption to the total number of effective production days across all SCMBs:

$$E_{prod\_day}^B={\displaystyle \frac{E_Y}{\sum _n{N}_{PROD}^{SCMB\left(n\right)}}}$$

(20)

This indicator does not represent energy efficiency in isolation. It reflects both transport energy consumption and production continuity because its denominator depends on the number of effective building-production days achieved under disruption conditions. A decrease in this indicator may therefore result from lower transport energy consumption, increased production continuity, or both. For this reason, it is interpreted alongside total annual transport energy consumption, the number of e-truck and e-van transfers, and vehicle loading capacity utilization.

5.2. A Case Study

5.2.1. Model Inputs

The TEPM was evaluated through scenario-based simulation using the enhanced test data prepared for the previous version of the model, i.e., the MLBNM-based inventory management model. The model’s assumptions and its individual parameters reflect the characteristic spatial relationships of the Berlin agglomeration, including the locations of ILNs outside the ring road, the locations of CLNs near the largest shopping centers in the various districts, and the distances and journey times of e-trucks and e-vans. The input data used for the TEPM evaluation were expanded as follows:

(1)

The probabilities of urban-scale disruptions, including transport processes: e-truck transit, transshipment in CLN, e-van transit, unloading in SCMB, are shown in Figure 4.

(2)

It is assumed that SSL maintained at the CLN takes five different levels: 0%, 50%, 70%, 90%, and 100%. Where SSL = 100% denotes a configuration of 167 ITR of production material A, 41 ITR of production material B, and 56 ITR of production material C. These quantities are calculated based on the maximum stock shortages identified in the CLN over the course of a year.

(3)

Different BSLs were subsequently analyzed for the SCMBs. The scenarios considered included the following quantities of the three production materials (A, B, C) held in each SCMB: (a) B000, (b) B111, (c) B222, (d) B444. For example, B111 means a BSL configuration of 1 ITR of production material A, 1 ITR of production material B, and 1 ITR of production material C. Scenario B444 was set as the maximum stock level following an expert analysis of the spatial planning of the existing SCMBs and the willingness to sacrifice a maximum of approximately 12 m² of production space in order to maintain the BS.

(4)

The electricity consumption of e-trucks and e-vans was assumed to be 25 kWh per e-truck transfer and 6 kWh per e-van transfer, respectively. These are average yearly energy-consumption rates characteristic of the Berlin metropolitan area. Transport disruptions affecting e-truck and e-van operations were assumed to increase energy consumption by 20%.

All remaining model inputs and assumptions follow those presented in the previous model [14], except for the extensions described in Section 5.1.

5.2.2. Model Evaluation

The TEPM was evaluated using a representative SCMC case study. The integrated MLBNM–Material Flow Analysis framework was applied to assess SCMC supply chain performance under different disruption conditions and alternative inventory configurations. A single fixed realization of the disruption events generated for the five disruption categories was used as the stochastic input for all analyzed inventory configurations. Using the same disruption realization across the SSL and BSL variants provides a consistent basis for comparing their effects without introducing variation from different stochastic disruption inputs. Consequently, the quantitative results should be interpreted as scenario-specific outcomes rather than statistical estimates obtained from repeated stochastic simulations. Figure 5 presents the corresponding disruption matrix that describes (1) production material availability in ILN disruptions, (2) e-truck ILN-CLN transit disruptions, (3) breakdown in CLN, (4) e-van CLN–SCMB transit disruptions, and (5) breakdown in SCMB. The evaluation results are organized into three groups.

The first group characterizes inventory dynamics and production material shortages resulting from disruption propagation. The results are divided into two sub-groups: intermodal-scale disruptions, which include all disruption categories across the SCMC intermodal supply chain, and urban-scale disruptions, which are limited to disruption events occurring within the urban part of the supply chain. In the urban-scale scenario, production materials are assumed to be produced in proximity to the ILN and therefore remain continuously available at this node, representing a nearshoring configuration.

Table 2. Matrix of model outputs related to production material shortages in CLN and SCMBs at different levels of SSL and BSL = B444 caused by urban-scale disruptions.

BSL = B444	SSL
Urban-Scale	0%	50%	70%	90%	100%
max${SV}_A^{CLN}$ [IRT]	16.0	0.0	0.0	0.0	0.0
max${SV}_B^{CLN}$ [IRT]	17.0	0.0	0.0	0.0	0.0
max${SV}_C^{CLN}$ [IRT]	36.0	8.0	0.0	0.0	0.0
${SD}_A^{CLN}$ [days]	5	0	0	0	0
${SD}_B^{CLN}$ [days]	5	0	0	0	0
${SD}_C^{CLN}$ [days]	5	4	0	0	0
max${SV}_A^{SCMB\left(1\right)}$ [IRT]	10.32	10.32	10.32	10.32	10.32
max${SV}_B^{SCMB\left(1\right)}$ [IRT]	12.75	12.75	12.75	12.75	12.75
max${SV}_C^{SCMB\left(1\right)}$ [IRT]	26.10	26.10	26.10	26.10	26.10
max${SV}_A^{SCMB\left(2\right)}$ [IRT]	6.14	6.14	6.14	6.14	6.14
max${SV}_B^{SCMB\left(2\right)}$ [IRT]	10.85	10.85	10.85	10.85	10.85
max${SV}_C^{SCMB\left(2\right)}$ [IRT]	20.80	20.80	20.80	20.80	20.80
max${SV}_A^{SCMB\left(3\right)}$ [IRT]	5.96	5.96	5.96	5.96	5.96
max${SV}_B^{SCMB\left(3\right)}$ [IRT]	6.68	6.68	6.68	6.68	6.68
max${SV}_C^{SCMB\left(3\right)}$ [IRT]	27.00	27.00	27.00	27.00	27.00
max${SV}_A^{SCMB\left(4\right)}$ [IRT]	11.12	11.12	11.12	11.12	11.12
max${SV}_B^{SCMB\left(4\right)}$ [IRT]	10.50	10.50	10.50	10.50	10.50
max${SV}_C^{SCMB\left(4\right)}$ [IRT]	21.30	21.30	21.30	21.30	21.30
${maxSD}_m^{SCMB\left(1\right)}$ [days]	23	21	21	21	21
${maxSD}_m^{SCMB\left(2\right)}$ [days]	32	29	29	29	29
${maxSD}_m^{SCMB\left(3\right)}$ [days]	26	23	23	23	23
${maxSD}_m^{SCMB\left(4\right)}$ [days]	23	21	20	20	20
$N_{PROD}^{SCMB\left(1\right)}$ [days]	342	344	344	344	344
$N_{PROD}^{SCMB\left(2\right)}$ [days]	333	336	336	336	336
$N_{PROD}^{SCMB\left(3\right)}$ [days]	339	342	342	342	342
$N_{PROD}^{SCMB\left(4\right)}$ [days]	342	344	345	345	345

and

Table 3. Matrix of model outputs related to production material shortages in CLN and SCMBs at different levels of SSL and BSL = B444 caused by intermodal-scale disruptions.

BSL = B444	SSL
Intermodal-Scale	0%	50%	70%	90%	100%
max${SV}_A^{CLN}$ [IRT]	156.00	72.00	38.00	8.00	0.00
max${SV}_B^{CLN}$ [IRT]	24.00	11.00	3.00	0.00	0.00
max${SV}_C^{CLN}$ [IRT]	43.00	19.00	2.00	0.00	0.00
${SD}_A^{CLN}$ [days]	155	13	4	1	0
${SD}_B^{CLN}$ [days]	23	2	1	0	0
${SD}_C^{CLN}$ [days]	14	7	2	0	0
max${SV}_A^{SCMB\left(1\right)}$ [IRT]	41.88	10.46	10.32	10.32	10.32
max${SV}_B^{SCMB\left(1\right)}$ [IRT]	40.65	12.88	12.75	12.75	12.75
max${SV}_C^{SCMB\left(1\right)}$ [IRT]	120.65	36.15	26.20	26.10	26.10
max${SV}_A^{SCMB\left(2\right)}$ [IRT]	30.42	21.20	7.02	6.14	6.14
max${SV}_B^{SCMB\left(2\right)}$ [IRT]	50.55	22.03	10.85	10.85	10.85
max${SV}_C^{SCMB\left(2\right)}$ [IRT]	103.35	51.10	20.80	20.80	20.80
max${SV}_A^{SCMB\left(3\right)}$ [IRT]	31.28	10.34	5.96	5.96	5.96
max${SV}_B^{SCMB\left(3\right)}$ [IRT]	49.93	21.05	6.68	6.68	6.68
max${SV}_C^{SCMB\left(3\right)}$ [IRT]	73.15	27.00	27.00	27.00	27.00
max${SV}_A^{SCMB\left(4\right)}$ [IRT]	45.00	22.48	15.94	11.12	11.12
max${SV}_B^{SCMB\left(4\right)}$ [IRT]	48.23	29.08	18.00	10.50	10.50
max${SV}_C^{SCMB\left(4\right)}$ [IRT]	76.80	32.10	21.30	21.30	21.30
${maxSD}_m^{SCMB\left(1\right)}$ [days]	88	23	22	21	21
${maxSD}_m^{SCMB\left(2\right)}$ [days]	86	33	28	29	29
${maxSD}_m^{SCMB\left(3\right)}$ [days]	94	27	24	23	23
${maxSD}_m^{SCMB\left(4\right)}$ [days]	92	29	22	20	20
$N_{PROD}^{SCMB\left(1\right)} [$days]	277	342	343	344	344
$N_{PROD}^{SCMB\left(2\right)}$ [days]	279	332	337	336	336
$N_{PROD}^{SCMB\left(3\right)}$ [days]	271	338	341	342	342
$N_{PROD}^{SCMB\left(4\right)}$ [days]	273	336	343	345	345

and Figure 6, Figure 7, Figure 8 and Figure 9 present the corresponding numerical and graphical results.

The second group compares model outputs related to transport activity and the associated transport-energy performance of the SCMC supply chain.

Table 4. Matrix of model outputs related to transport-energy performance of the SCMC supply chain.

		BSL
		B000	B111	B222	B444
$N_Y$  [transits]	SS = 0%, urban	5031	5025	5026	5032
	SS = 100%, urban	5039	5036	5037	5042
	SS = 0%, intermodal	4994	4938	4984	4999
	SS = 100%, intermodal	5022	5020	5021	5000
$E_Y$  [kWh/year]	SS = 0%, urban	48,169	48,129	48,128	48,195
	SS = 100%, urban	48,210	48,191	48,198	48,249
	SS = 0%, intermodal	47,913	47,608	47,882	47,984
	SS = 100%, intermodal	49,324	49,306	49,337	47,959
$E_{prod\_day}^B$  [kWh/bpd]	SS = 0%, urban	37.311	37.166	36.993	35.542
	SS = 100%, urban	36.801	36.759	36.624	35.296
	SS = 0%, intermodal	64.747	58.994	53.202	43.622
	SS = 100%, intermodal	38.176	38.133	38.010	35.084
$U_Y^{truck}$	SS = 0%, urban	0.977	0.977	0.978	0.977
	SS = 100%, urban	0.977	0.978	0.980	0.978
	SS = 0%, intermodal	0.980	0.979	0.980	0.977
	SS = 100%, intermodal	0.981	0.981	0.979	0.982
$U_Y^{van}$	SS = 0%, urban	0.792	0.793	0.793	0.794
	SS = 100%, urban	0.791	0.792	0.792	0.793
	SS = 0%, intermodal	0.797	0.808	0.799	0.797
	SS = 100%, intermodal	0.797	0.798	0.798	0.800

and Figure 10 report the corresponding numerical and graphical results.

The third group proposes enhanced model parameters to improve transport-energy performance and production continuity in the SCMC supply chain.

Table 5. Enhanced model parameters.

	BS
	100% BSL = B444 (4A4B4C)		Max Demand BSL = B6815 (6A8B15C)
	${\mathit{E}}_{\mathit{p}\mathit{r}\mathit{o}\mathit{d}\_\mathit{d}\mathit{a}\mathit{y}}^{\mathit{B}}$  [kWh/bpd]	${\mathit{N}}_{\mathit{P}\mathit{R}\mathit{O}\mathit{D}}$  [Days]	${\mathit{E}}_{\mathit{p}\mathit{r}\mathit{o}\mathit{d}\_\mathit{d}\mathit{a}\mathit{y}}^{\mathit{B}}$  [kWh/bpd]	${\mathit{N}}_{\mathit{P}\mathit{R}\mathit{O}\mathit{D}}$  [Days]
SSL = 0%, urban	35.542	1356	33.135	1458
SSL = 0%, urban, w/o unloading disruptions	35.055	1375	33.146	1460
SSL = 100%, urban	35.296	1367	33.151	1458
SSL = 100%, urban, w/o unloading disruptions	34.962	1379	33.152	1460
SSL = 0%, intermodal	43.622	1100	38.351	1280
SSL = 0%, intermodal, w/o unloading disruptions	43.511	1107	37.484	1288
SSL = 100%, intermodal	35.084	1367	32.942	1458
SSL = 100%, intermodal, w/o unloading disruptions	34.907	1379	33.152	1460

and Figure 11, Figure 12 and Figure 13 summarise the corresponding transport and energy indicators.

The numerical and graphical results enable an integrated evaluation of disruption propagation, inventory allocation, transport operations, and energy consumption within the analyzed SCMC supply chain. The results show that disruptions directly affect stock dynamics and production continuity within the analyzed SCMC supply chain. As illustrated in Table 2 and Table 3 and Figure 6, Figure 7, Figure 8 and Figure 9, the results obtained for intermodal-scale disruptions differ significantly from those for urban-scale disruptions. Specific observations include:

(1)

The first measure of the intensity of disruptions in the SCMC supply chain is the maximum shortage of production materials in CLN (max${SV}_m^{CLN}$) and the maximum cumulative period of shortage of production materials in CLN ($max{SD}_m^{CLN}$). For the analyzed fixed disruption realization, the largest differences between intermodal-scale and urban-scale disruptions are observed for production material A, with the maximum shortage reaching 156 ITR compared with 16 ITR, and the maximum cumulative shortage period reaching 155 days compared with 5 days, corresponding to differences of 975% and 3100%, respectively. For urban-scale disruptions, shortages of production materials in the CLN fall to zero and do not cause any delivery delays once the SS is 50% full (SSL = 50%). For intermodal-scale disruptions, this effect is only achieved when the SS is 100% full (SSL = 100%). The above relationships demonstrate the effective role of SS maintained in the CLN as a guarantor of the continuity of production material supplies along the ILN–CLN section.

(2)

A second measure of the intensity of disruptions in the SCMC supply chain is the magnitude of the maximum shortages of production materials in SCMBs (max${SV}_m^{SCMB\left(n\right)}$) and the maximum cumulative period of shortage of production materials in SCMBs (${maxSD}_m^{SCMB\left(n\right)}$). For the analyzed fixed disruption realization, comparing intermodal-scale and urban-scale disruptions shows an increase in maximum shortage magnitude of up to 703% (49.93 ITR/6.68 ITR); this applies only to scenarios with SSL ranging from 0% to 70%. A characteristic feature is the almost constant level of maximum shortages in SCMBs when urban-scale disruptions occur, and the same level of these shortages for intermodal-scale disruptions when SSL > 70%. A similar pattern is observed for the maximum cumulative shortage period, with the difference between intermodal-scale and urban-scale disruption scenarios reaching up to 400% (92 days/23 days).

(3)

The above relationships show that, in the analyzed supply chain, maintaining adequate stock at CLN does not ensure the continuity of production material supplies in the final section CLN–SCMB. Even in the scenario with the highest stock levels in SCMB (B444), production material shortages occur. This results from disruptions during e-van transit and breakdowns in the SCMB handling area. Regardless of the SSL at CLN, BS = B444 proved insufficient and resulted in maximum cumulative supply shortages of over 20 days for each production material in every building.

(4)

The maximum cumulative period of shortage at the SCMB ($max{SD}_m^{SCMB\left(n\right)}$) is strictly related to the number of production days at SCMB ($N_{PROD}^{SCMB\left(n\right)}$). In other words, a shortage of production materials caused by disruptions at one or more sections of the SCMC supply chain, even after replenishing the shortages from the BS (B444), results in an inability to produce on the same day.

5.3. Transport-Energy Performance Assessment

Table 4, Figure 10 and Figure 13 present the model outputs related to transport activity and the associated energy performance of the SCMC supply chain, allowing the following observations.

(1)

The first observation is that there are very slight changes in the total number of vehicle transfers ($N_Y$), regardless of the area affected by disruptions, i.e., urban-scale or intermodal-scale disruptions, and the BSL level. These changes do not exceed 1% for all analyzed disruption scenarios and BSL levels ranging from 0% to 100%. The number of transfers carried out by e-trucks and e-vans determines the total annual transport energy consumption ($E_Y$), which also shows minor changes not exceeding 3% for all analysed disruption scenarios and BSL values. In contrast, transport-energy intensity per effective building-production day varies substantially more across the analyzed scenarios. The substantially larger variation in this indicator primarily reflects changes in production continuity rather than a comparable reduction in total transport energy consumption. The limited variation in the number of vehicle transfers across the scenarios indicates that transport activity is driven primarily by demand and vehicle loading capacity utilization rather than by the spatial extent of disruptions. Variations in total annual transport energy consumption also reflect the number of disrupted e-truck and e-van transfers, for which the model applies higher energy-consumption coefficients.

(2)

Transport-energy intensity per effective building-production day takes higher values under intermodal-scale disruption scenarios and decreases as SSL and BSL increase. Although the trends are evident, their magnitude varies across scenarios, with the largest difference reaching 74% for SSL = 0% (64.747 kWh/bpd versus 37.311 kWh/bpd) and 33% under intermodal-scale disruptions (64.747 kWh/bpd versus 43.622 kWh/bpd). These differences primarily reflect changes in production continuity rather than proportional changes in total transport energy consumption. The results indicate that maintaining appropriately sized stocks at the CLN and SCMBs can improve production continuity under disruption conditions, thereby reducing transport-energy intensity per effective building-production day.

(3)

The values of auxiliary parameters such as the utilization rate of e-trucks ($U_Y^{truck}$) and e-vans ($U_Y^{van}$) indicate their direct impact on the total number of vehicle transfers and, consequently, on total transport energy consumption. Even a slight change in these parameters, multiplied by the number of vehicle transfers, may affect transport energy consumption. The data presented indicate greater potential for improving the utilization of e-vans’ cargo space, for which the utilization rate is 19% lower than that of e-trucks.

The observations above indicate the need to adjust BS levels, particularly the BSL maintained at SCMB, to improve production continuity and transport-energy performance. In addition, the study considered new scenarios in which unloading disruptions, i.e., relating to various types of breakdowns in the handling area of SCMB, were eliminated. Table 5 and Figure 11, Figure 12 and Figure 13, which present transport and energy indicators, allow for the following observations and conclusions:

(1)

Setting the BSL in SCMB to a level corresponding to the maximum daily order, i.e., BSL = B6815 (29 IRT units in the 6A, 8B, 15C configuration), reduces the transport-energy intensity per effective building-production day between 5% and 16%, depending on the scenario analyzed. The lowest value of $E_{prod\_day}^B$ was achieved for the SSL = 100% under the influence of intermodal-scale disruptions (32,942 kWh/bpd). This reduction primarily reflects the increase in effective production days achieved with the higher BSL and should therefore not be interpreted as an equivalent reduction in total transport energy consumption. This stock should correspond to at least one order generated in the JIT system, which is a quantity greater than the previously assumed configuration of 12 IRT units (B444), resulting from the willingness to allocate space at SCMB’s disposal for this purpose.

(2)

The reduction in transport-energy intensity per effective building-production day is primarily associated with the increase in the number of effective production days, ($N_{PROD}^{SCMB\left(n\right)}$), which, following an increase in the stock level (BSL = B6815), rises by between 6% and 16%. The new stock levels, combined with the elimination of unloading disruptions in the SCMB, enable the maximum possible number of production days across the four SCMBs (1460 days). This confirms the identified critical stage in the SCMC supply chain: handling operations in the manufacturing building. Under the analyzed scenarios, disruptions at this stage caused production interruptions; therefore, eliminating them was necessary to achieve complete production continuity.

(3)

The enhanced model parameters indicate a relationship between the utilization rates of e-trucks and e-vans and transport-energy performance. The highest vehicle utilization rates are achieved for the increased stock level BSL = B6815, which is also associated with the lowest transport-energy intensity per effective building-production day.

6. Discussion

The results obtained using the proposed TEPM provide the following responses to the four research questions.

(1)

Ad RQ1. The proposed TEPM shows that disruption propagation in intermodal SCMC supply chains follows a multi-layer network mechanism in which upstream disturbances progressively affect downstream logistics and production operations. Product availability at the ILN, road transport disruptions along the ILN–CLN and CLN–SCMB sections, and breakdowns during transshipment and unloading operations generate cumulative delays that propagate through successive logistics layers. For the analyzed scenarios, intermodal-scale disruptions produce a greater negative impact than urban-scale disruptions. Under the fixed disruption realization used in the analysis, the maximum cumulative period of production material shortage increases from 23 to 92 days per year, an increase of up to 400%. From a managerial perspective, nearshoring production facilities that manufacture the required raw materials may reduce exposure to upstream disruptions; however, nearshoring itself was not modeled as a decision variable in the present study.

(2)

Ad RQ2. The results indicate that effective buffer allocation should be determined by both network topology and disruption exposure rather than by uniform inventory policies. The proposed framework demonstrates that CLNs and SCMBs perform complementary buffering functions within the SCMC supply chain. A larger-capacity SS at the CLN mitigates upstream disruptions affecting multiple downstream manufacturing buildings. In contrast, a smaller-capacity BS at the SCMB primarily protects individual manufacturing processes against disruptions in the CLN–SCMB section and breakdowns in the SCMB handling area. For the analyzed SCMC configuration and fixed disruption realization, maintaining the SS at the CLN at 100% and the BS at the SCMB at a level corresponding to the maximum daily order substantially improved production continuity. Complete production continuity, corresponding to the maximum possible 1460 building-production days across the four SCMBs, was achieved when these inventory levels were combined with the elimination of unloading disruptions at the SCMB. Maintaining the SCMB BS at this level presents a practical trade-off, as it requires allocating more storage space than manufacturers typically anticipate within SCMBs.

(3)

Ad RQ3. The proposed model shows that buffer allocation directly influences transport-energy performance through its interaction with transport operations, material availability, and disruption mitigation. Increasing buffer capacity primarily improves production continuity, while its effect on the total number of vehicle transfers and annual transport energy consumption remains limited. This relationship reveals an important resilience–energy trade-off. Higher inventory buffers can improve production continuity under disruption conditions without substantially changing total transport energy consumption. However, additional energy requirements associated with increased storage and handling remain outside the transport-energy accounting boundary of the present TEPM. Transport-energy intensity per effective building-production day shows that transport-energy performance is strongly influenced by production continuity under disruption conditions. Its variation therefore reflects the combined effect of transport energy consumption and the number of effective production days, rather than changes in transport energy consumption alone. The results further show that total annual transport energy consumption varies only slightly across the analyzed scenarios, because the number of vehicle transfers is determined primarily by production material demand and vehicle loading capacity utilization. Buffer allocation therefore affects transport-energy performance mainly by maintaining production continuity rather than substantially reducing transport activity or total annual transport energy consumption.

(4)

Ad RQ4. The proposed TEPM demonstrates that integrating disruption propagation modeling with network-dependent buffer allocation provides a more comprehensive basis for operational decision-making than treating these elements independently. By combining probabilistic disruption modeling with inventory allocation within a unified analytical framework, the model evaluates production continuity, material availability, system resilience, and transport-energy performance simultaneously. The results demonstrate that changes in the transport-energy indicator expressed per building-production day are primarily associated with changes in production continuity. For the analyzed scenarios, maintaining high stock levels at the CLN and SCMBs, as indicated in the answer to RQ2, together with eliminating unloading disruptions at the SCMBs, supports production continuity and is associated with lower transport-energy intensity per effective building-production day under the analyzed disruption conditions.

The responses to the research questions show that the proposed TEPM integrates disruption propagation, probabilistic buffer allocation, and transport-energy performance evaluation within a single analytical framework. This provides the basis for the theoretical and managerial implications discussed below.

This study extends the M2 framework [14] by linking its MLBNM-based stochastic disruption-propagation mechanism with operationally differentiated disruption states, waiting-state material flows, and inventory buffering at the CLN and SCMB levels. M3 retains M2’s disruption-propagation mechanism rather than introducing a new probabilistic method. SS at the CLN provides network-level buffering, while BS at individual SCMBs provides local protection against material shortages affecting manufacturing operations. These elements link disruption propagation and inventory states to material availability, production continuity, and transport-energy performance across interconnected logistics and production layers.

The proposed TEPM supports operational decision-making for planning and managing intermodal SCMC supply chains under disruption conditions. Rather than relying on fixed SS policies, decision-makers can determine the appropriate location and capacity of BSs based on disruption probabilities, network topology, and operational characteristics, consistent with recent advances in adaptive inventory management and resilient supply chain planning [40,41]. The model is particularly applicable to operators of ILNs and CLNs, manufacturing enterprises, and urban logistics planners coordinating material flows across interconnected logistics and production networks. It enables the identification of disruption-sensitive logistics nodes and evaluation of alternative buffer allocation strategies and their effects on production continuity and energy consumption before implementation [42,49].

From a practical perspective, inventory policies should differ by the location and source of disruption risk. SS at the CLN should primarily protect the network against upstream and intermodal disruptions affecting multiple SCMBs, whereas BS at individual SCMBs should mitigate local delivery and handling disruptions. Buffer levels should therefore be adjusted to disruption exposure rather than uniformly increased across the network. Disruption-response measures should prioritize critical downstream handling operations, particularly unloading at SCMBs, since failures at this stage directly affect production continuity. Transport-energy performance should be managed jointly with inventory decisions by monitoring vehicle utilization, transport frequency, and production continuity; higher inventory levels should be adopted only when their resilience benefits justify the associated storage requirements and transport activity. This trade-off should be interpreted within the assumptions and configuration of the analyzed TEPM scenarios. Its magnitude and direction depend on the adopted disruption patterns, demand and buffer levels, network structure, vehicle utilization, and transport-energy parameters; different operational configurations may therefore produce different relationships between inventory-based resilience and transport-energy performance.

For the analyzed scenarios, the results indicate that disruption propagation, buffer allocation, and transport-energy performance should be analyzed as interdependent components of SCMC supply chains rather than as separate operational problems. Their integration within a unified analytical framework strengthens both theoretical understanding and practical decision-making for resilient and energy-efficient intermodal logistics systems. The discussion presented above forms the basis for the conclusions and future research directions outlined in the next section.

7. Conclusions

The principal contribution of the TEPM is demonstrating interdependencies between disruption propagation, inventory buffer allocation, production continuity, and transport-energy performance. Higher inventory protection reduced material shortages and supported production continuity, while the effects on transport activity and total transport energy consumption remained comparatively limited. For the analyzed scenarios, the total number of vehicle transfers and annual transport energy consumption varied by less than 1% and 3%, respectively, while transport-energy intensity per effective building-production day changed by up to 74%, indicating that this performance indicator is strongly influenced by production continuity under disruption conditions.

From a practical perspective, the results support differentiated inventory planning across the SCMC network. It has been demonstrated that SS at the CLN can provide network-level protection against upstream and intermodal disruptions. In contrast, BS at individual SCMBs can mitigate local delivery and handling disruptions. The TEPM enables decision-makers to compare alternative buffer configurations and identify disruption-sensitive locations before implementation, while considering their implications for production continuity and transport-energy performance.

Several limitations of the proposed TEPM should be recognized. The model considers a single ILN supplying multiple CLNs and SCMBs and assumes scenario-based demand patterns and continuous year-round production. Disruption probabilities and selected operational parameters were adopted from the previously developed framework and representative SCMC scenarios rather than estimated from a large-scale empirical dataset. The energy assessment is limited to electricity consumption associated with e-truck and e-van operations. It excludes energy use related to inventory storage, handling equipment, charging infrastructure, and other facility operations. Economic costs, greenhouse gas emissions, and adaptive real-time operational control are also outside the model boundary. Applying it to supply chains with different demand patterns, storage constraints, or vehicle energy-consumption characteristics therefore requires recalibrating the relevant parameters. Although developed for SCMC supply chains, the TEPM can be adapted to other urban logistics systems characterized by multi-stage material flows, intermediate logistics nodes, disruption-prone transport and handling operations, and inventory buffering, provided that the network structure and operational parameters are adjusted accordingly. Finally, the evaluation is based on a single fixed realization of disruption events applied consistently across all inventory configurations and on representative SCMC scenarios rather than extensive empirical observations. Accordingly, the present case study should be regarded as a scenario-based evaluation of the model’s internal behavior and decision logic rather than as empirical validation of the TEPM. The reported quantitative results are specific to the assumed network configuration, disruption conditions, demand patterns, inventory settings, and transport parameters and should not be generalized directly to other SCMC or intermodal supply chain configurations. Consequently, the present findings demonstrate the analytical capabilities and internal consistency of the proposed modeling framework under the examined conditions, but do not establish its empirical validity or external generalisability. Validation against observations from multiple real-world intermodal logistics networks and heterogeneous operating environments is therefore required before drawing broader empirical conclusions about the model’s predictive performance and practical applicability.

Future research should extend the proposed TEPM to support adaptive assessment of disruption propagation and inventory buffer allocation under changing operational and transport energy conditions. Incorporating real-time operational and transport energy data through IoT-based monitoring, digital twins, and AI-supported disruption forecasting would enable continuous updating of disruption probabilities and more accurate assessment of their impacts on energy consumption and system resilience. Further research should investigate the applicability of the TEPM in heterogeneous intermodal supply chain networks operating under diverse disruption scenarios, transport configurations, and the seasonality of energy supply conditions, while extending the model to incorporate renewable energy integration and advanced energy management strategies. Future validation should include empirical observations from large-scale intermodal logistics networks and a comparison with alternative inventory and disruption-mitigation strategies.