Agent-Based Simulation of Digital Interoperability Thresholds in Fragmented Air Cargo Systems: Evidence from a Developing Country

Abstract

Background: This study investigates how varying levels of digital interoperability affect coordination and performance in Indonesia’s decentralized air cargo system, reflecting the inefficiencies typical of fragmented digital infrastructures in developing economies. Methods: An Agent-Based Model (ABM) was developed to simulate interactions among shippers, freight forwarders, airlines, ground handlers, and customs agents along the CGK–SIN/HKG export corridor. Six simulation scenarios combined varying levels of digital adoption, operational friction, and behavioral adaptivity to capture emergent coordination patterns and threshold dynamics. Results: The simulation identified a distinct interoperability threshold at approximately 60%, beyond which performance improvements became non-linear. Once this threshold was surpassed, clearance times decreased by more than 40%, and capacity utilization exceeded 85%, particularly when adaptive decision rules were implemented among agents. Conclusions: Digital transformation in fragmented logistics systems requires both technological connectivity and behavioral adaptivity. The proposed hybrid framework—integrating Autonomous Supply Chains (ASC), Graph-Based Digital Twins (GBDT), and interoperability thresholds—provides a simulation-based decision-support tool to determine when digitalization yields system-wide benefits. The study contributes theoretically by linking behavioral adaptivity and digital interoperability within a unified modeling approach, and practically by offering a quantitative benchmark for policymakers and practitioners seeking to develop efficient and resilient logistics ecosystems.

1. Introduction

The IT-enabled transformation of air cargo logistics has become a critical imperative, particularly in emerging economies that continue to face inherent inefficiencies and limited coordination [1,2]. Although global platforms have demonstrated the advantages of integration, numerous air cargo systems, especially those in developing nations, remain fragmented, with poor interoperability, outdated processes, and behaviorally constrained collaboration practices [3,4,5]. Such environments hinder the responsiveness and agility of logistics networks, making them highly vulnerable to disruption and congestion [6,7].

Despite the proliferation of digital transformation initiatives across the air cargo sector, many of these technology-centered initiatives lack adequate consideration of the interconnection between digital adoption and behavioral flexibility [8,9]. Additionally, this study employs the interoperability threshold concept within the agent-based model to demonstrate how coordination benefits emerge only after a critical level of digital connectivity is reached [10,11,12]. Simulation-based studies are limited in modeling the dynamic transformation of coordination behaviors within digitally dispersed environments [13,14], highlighting a methodological gap in how such systems are represented and analyzed.

Indonesia’s logistical complexity and fragmented digital landscape are emblematic of many emerging economies. As an archipelagic nation characterized by high logistical diversity and uneven levels of digital maturity, its air cargo sector reflects a typical developing-country ecosystem in which coordination frictions persist despite initiatives such as the National Logistics Ecosystem (NLE) and the Indonesia National Single Window (INSW). Empirical evidence from Indonesia’s manufacturing sector further demonstrates that digital supply chain adoption significantly enhances performance [15]. Indonesia’s decentralized logistics network—with varying digital maturity across nodes—thus offers a natural laboratory for examining how interoperability and behavioral adaptivity interact under conditions of institutional heterogeneity common to emerging economies.

This study addresses these gaps by developing an agent-based simulation model that encapsulates Indonesia’s air cargo export system. Through the incorporation of behavioral logic, variations in digital readiness, and interactions across six stakeholder simulation scenarios, the model examines the effects of fragmentation, adaptivity, and interoperability thresholds on coordination outcomes. Drawing on theories of complex adaptive systems [16] and learning-based decision-making in multi-agent environments [17,18], this study contributes both theoretically, by linking behavioral adaptivity with digital interoperability in a unified framework, and practically, by offering a template for developing resilient, digitally supported logistics ecosystems.

2. Research Methods and Design

2.1. Research Gap and Positions

Digital transformation in air-cargo logistics has attracted increasing scholarly attention, and existing studies continue to separate macro-level policy analyses from micro-level technological applications. Previous works often assume homogeneous technology adoption across actors, overlooking the behavioral heterogeneity and institutional fragmentation that shape logistics networks in developing economies [1,2,19]. In practice, shippers, freight forwarders, airlines, ground handlers, and customs authorities operate with uneven digital readiness, producing asynchronous decision-making and limited coordination across the supply chain [10,20,21]. Consequently, empirical and simulation-based models rarely capture the realities of fragmented interoperability, where behavioral and structural diversity determine performance outcomes.

The existing literature on the digitalization of air-cargo systems can be broadly grouped into two strands: macro-level studies emphasizing governance or regulation, and micro-level research focused on single technologies such as blockchain or electronic airway bills [12,22,23]. As summarized in

Table 1. Selected Prior Studies on Air Cargo Digitalization.

No.	Source	Focus	Method	Context	Research Gap
1	[20]	Air cargo network	Spatiotemporal Analysis	Global	Micro-level operational dynamics
2	[25]	Security standardization in cargo	Conceptual	Global	Focused on security, not operational integration
3	[12]	Electronic air waybills (e-AWB)	Case/tech-specific	Global	Single-technology view; assumes uniform adoption
4	[23]	Blockchain for tracking and tracing	Applied/tech	Supply chains	Limited to traceability; neglects multi-actor coordination
5	[26]	Geographically limited	Case study	China’s Belt and Road	Threshold analysis and behavioral adaptivity
6	[22]	IATA ONE Record	Case study	Latvia	Lacks simulation-based validation and cross-context generalizability
7	[27]	Terminal scheduling optimization	Data-driven	European airports	Ignores multi-actor behavioral interactions
8	[14]	Agent-based modeling in transport	Simulation	Europe	Applied to passengers, not air cargo logistics

, most contributions remain descriptive or case-specific and lack behavioral or systemic integration. Simulation studies, although growing, primarily address passenger or infrastructure settings rather than multi-actor cargo coordination [14,24]. Collectively, the literature reveals three persistent gaps: (1) limited investigation of fragmented digital infrastructures typical of developing-country air-cargo systems, (2) insufficient integration of behavioral adaptivity and interoperability thresholds within digital-transformation frameworks, and (3) the absence of empirically calibrated agent-based simulations replicating coordination frictions found in practice [6,8,9].

To address these gaps, this study develops an Agent-Based Model (ABM) that represents heterogeneous agents and their adaptive decision-making under varying degrees of digital adoption. The model is grounded in a hybrid conceptual framework integrating three theoretical structures, Autonomous Supply Chains (ASC), Graph-Based Digital Twins (GBDT), and Digital Interoperability Thresholds, to operationalize the complex phenomenon of coordination within digitally divided logistics systems. The novelty of this framework lies in its ability to move beyond a purely technological-determinist perspective by explicitly linking connectivity (GBDT), behavioral adaptivity (ASC), and systemic emergence (thresholds) [6,13,28].

In this integration, the ASC provides the behavioral foundation where agents learn and adapt through local decision rules [13,16]; the GBDT supplies the digital and structural layer that maps information flows and interdependencies among agents [26,27]; and the interoperability threshold concept captures the critical point at which these interactions produce non-linear performance improvements [28,29]. Together, these components form a multi-layered socio-technical model explaining how micro-level adaptivity and digital connectivity interact to create system-wide coordination gains in fragmented logistics networks [6,19,24].

Each framework carries inherent limitations: the ASC overlooks governance and institutional control mechanisms, the GBDT requires high data intensity and computational resources, and the interoperability-threshold approach assumes relative temporal stability. Recognizing these limitations justifies their combination, as each framework compensates for the conceptual blind spots of the others. As depicted in Figure 1, the conceptual model illustrates this dynamic relationship. The upper layers represent autonomous decision-making among shippers, freight forwarders, airlines, ground handlers, and customs agents, while the structural layer visualizes information flows and digital interconnections through the GBDT. The lower level captures digital-readiness variance, structural fragmentation, and interoperability thresholds as frictional forces that determine whether the network evolves toward a self-optimizing and resilient configuration or remains in a fragmented, asynchronous state [6,19].

2.2. Model Design

Building on the conceptual framework described in Section 2.1, the agent-based simulation was developed to represent interactions among heterogeneous actors along the Soekarno–Hatta (CGK) export air-cargo corridor. The model captures how autonomous agents, each operating with bounded rationality and exhibiting different levels of digital capability, interact within the system. These interactions collectively generate coordination outcomes at the system level. This operationalization enables the observation of emergent patterns of delay, utilization, and information exchange under varying levels of digital interoperability.

The simulation was implemented using NetLogo 6.3.0, a widely adopted platform for agent-based modeling in logistics and transportation research. NetLogo was selected because it effectively represents micro-level behavioral rules and visualizes macro-level emergent dynamics in real time, which aligns with the study’s objective to examine coordination emergence under fragmented digital environments. The software’s modular structure allows flexible rule scripting, stochastic experimentation, and scenario-based testing, enabling systematic comparison between adaptive and non-adaptive agent behaviors [28].

Five principal agent types were modeled: shippers, freight forwarders, airlines, ground handlers, and customs officers. Each agent possesses key attributes such as a digital-readiness index, communication delay, learning rate, and capacity limit, which follow decision rules derived from empirical observation and expert validation [4,30]. For instance, forwarder agents choose airline partners based on historical on-time performance and trust indices, while customs agents prioritize clearance queues according to document completeness. Adaptive agents modify their partner selections through reinforcement learning, whereas non-adaptive agents rely on fixed routines. This design operationalizes the Autonomous Supply Chain (ASC) logic by allowing decentralized, self-organizing decision-making to emerge without central control [6,24].

The inter-organizational relationships among agents are represented through a Graph-Based Digital Twin (GBDT), which mirrors both the physical flow of shipments and the digital exchange of information. Nodes denote organizational entities, and edges represent document transmissions or shipment hand-offs. The graph evolves dynamically as agents create, verify, or amend data, permitting analysis of ripple effects from local disruptions. Model calibration employed empirical parameters derived from the CGK corridor: an average communication delay of ≈12.22 h per shipment, redundant documentation averaging 4.2 submissions, and airline-capacity utilization of ≈55.08%, with peaks reaching ≈85%. These values were obtained through process mapping, document audits, and validation interviews with freight-forwarding, airline, and customs personnel, and they typify the coordination frictions of developing-country logistics systems.

To examine how performance changes under different connectivity and adaptivity conditions, six simulation scenarios were designed, as summarized in

Table 2. Selected Simulation Scenarios and Their Key Attributes.

Scenario ID	Scenario Name	Digital Adoption	Operational Friction	Agent Adaptivity	Primary Research Question
S1	Baseline	Low (<20%)	Information incompleteness, capacity constraints	OFF	What are the performance characteristics of the current fragmented system?
S2	Technological Adoption Only	Medium (≈60%)	Information incompleteness	OFF	Does technology adoption alone improve coordination without behavioral changes?
S3	Process Reengineering Only	Medium (≈60%)	Capacity constraint	OFF	Can process changes alone overcome system fragmentation?
S4	Stakeholder Collaboration Only	Medium (≈60%)	Information incompleteness	ON	How much improvement can adaptive behaviors generate without full digitalization?
S5	Partial Integration	High (>80%)	Capacity constraints	ON	What synergies emerge when technology and adaptivity combine under real-world constraints?
S6	Full Integration without Behavioral Adaptivity	Very high (95%)	Minimal friction	OFF	What happens when technological saturation occurs without adaptive learning?

. Each scenario varies the degree of digital interoperability, the presence or absence of adaptive learning, and the operational frictions (e.g., customs delay, capacity constraints, information incompleteness). The scenarios represent sequential stages of digital transformation rather than a full-factorial experiment, moving from low-connectivity fragmentation to near-full integration. The model records three key performance indicators, including average clearance time, communication delay, and capacity utilization, which are aggregated over 200 replications to ensure statistical stability. This experimental design provides a transparent link between the theoretical constructs of ASC, GBDT, and interoperability thresholds, and their empirical manifestation in the air-cargo logistics system.

The simulation integrates ASC-driven agent behavior, GBDT-based structural mapping, and interoperability thresholds to capture micro–macro coordination dynamics, as shown in Figure 1. ASC governs micro-level learning and decision behaviors, GBDT structures the digital and physical interactions among agents, and interoperability thresholds quantify emergent performance shifts at the macro level. Their combined dynamics determine whether the air-cargo system evolves into a self-optimizing network or remains trapped in asynchronous fragmentation. Together, the six scenarios simulate the transition from fragmentation to full digital integration, enabling the assessment of adaptivity and interoperability on system performance.

2.3. Calibration

Model calibration was undertaken to ensure that the simulation realistically represented the operational conditions of the Soekarno–Hatta (CGK)–SIN/HKG export air-cargo corridor. Empirical evidence and process-mapping results were used to set baseline values for key performance indicators including communication delay, clearance time, and capacity utilization. These values were cross-checked against industry benchmarks and stakeholder feedback to reproduce actual coordination patterns under partially digitized ecosystems [8,29]. The purpose of calibration was not statistical fitting, but to establish structural and behavioral fidelity that mirrors real operational dynamics.

Calibration followed a two-stage procedure that combined empirical estimation and expert validation. In the first stage, numerical parameters were derived from workflow documentation, process-mining outputs, and internal operational reports obtained from freight forwarders, airlines, and customs offices at CGK. The average communication delay was 12.22 h per shipment, redundant documentation averaged 4.2 submissions, and airline-capacity utilization was 55.08%, peaking at 85% during congestion. In the second stage, these baseline parameters were iteratively adjusted through consultations with three domain experts representing forwarding, airline, and customs sectors to ensure that the model’s emergent patterns were consistent with observed field conditions [24].

Agent behavior was parameterized using both empirical data and theoretical principles. Adaptive agents applied the Autonomous Supply Chain (ASC) logic, learning from prior outcomes to update partner-selection trust weights and optimize coordination [13]. Non-adaptive agents followed static procedural rules to emulate conventional siloed operations. The Graph-Based Digital Twin (GBDT) structure supported calibration by aligning network connectivity, message latency, and transaction density with the communication frequencies observed in the CGK process map [31]. Through this dual calibration, behavioral and structural validity were established simultaneously.

Calibration continued until the simulated performance indicators converged within ±10% of empirical averages. Sensitivity tests were then conducted by varying digital-readiness indices and learning rates within ±20% of their calibrated values. The stable output confirmed that model results were robust and not overly sensitive to small parameter perturbations. This approach follows common practices in logistics simulation studies, where calibration emphasizes conceptual and behavioral coherence rather than exact statistical replication [6,28].

Figure 2 presents the existing air cargo export process in Indonesia, derived from process mapping of the CGK corridor. The model depicts interactions among shippers, freight forwarders, airlines, ground-handling agents, and customs authorities, highlighting inter-organizational dependencies and bottlenecks such as redundant documentation, manual verification, and asynchronous communication. Quantitative analysis showed an average of 4.2 redundant document submissions, a 12.22 h average communication delay, clearance-time variability between 24 and 72 h, and average capacity utilization of 55.08% with peaks up to 85%. These empirically observed frictions were encoded as constrained parameters within the simulation, forming the baseline of Scenario 1 (Baseline) and influencing document-processing algorithms across all subsequent scenarios.

The BPMN model thus served as the operational foundation for defining agent-interaction rules and system constraints. By embedding real-world process frictions into the simulation, the model captured the heterogeneity of digital readiness, behavioral adaptivity, and coordination efficiency across logistics actors. This grounding ensured that analyses of interoperability thresholds and adaptive behaviors were firmly anchored in the tangible operational realities of Indonesia’s air-cargo system, enhancing both validity and explanatory depth.

2.4. Validation

Model validation was conducted through a structured, multi-step methodology combining code verification, expert validation, sensitivity testing, and quantitative benchmarking. The aim was to ensure that the calibrated simulation not only operated faithfully according to its conceptual and BPMN design but also reproduced the coordination dynamics and operational patterns observed in Indonesia’s air-cargo export system. This process followed established validation practices for agent-based logistics models [6,17,28].

(1)

Face Validation

The initial phase involved iterative verification and debugging within the NetLogo environment. Parameter sweeps were conducted to test extreme conditions and trace agent decision paths, confirming that the model executed its programmed logic consistently with the conceptual framework and BPMN diagram (Figure 2). To establish behavioral credibility, face validation was carried out by three domain experts: a senior freight forwarder with over 15 years of experience, an airline cargo operations manager, and a customs-clearance practitioner. Following Bhatt and Zaveri’s (2002) standardized procedure, they reviewed the model’s workflow and behavioral rules [17]. All experts confirmed that the simulation accurately represented real-world operational issues, including asynchronous communication and redundant documentation commonly observed within the CGK export corridor.

(2)

Sensitivity Analysis

A sensitivity analysis was then applied to test the robustness of results by varying critical parameters (±20% around baseline values), including digital-adoption rates, document-error probabilities, and capacity thresholds. Despite considerable changes in absolute performance, the relative scenario rankings remained stable. Across all tests, the interoperability threshold consistently emerged at approximately a 60% digital-adoption rate, validating the model’s stability and its capability to capture non-linear coordination dynamics under fragmented digital conditions.

The threshold value of approximately 60% was identified through a structured sensitivity analysis performed across a digital-interoperability range of 0.30 to 0.90. A series of 200 replications per increment (Δ = 0.05) were executed, and the resulting clearance-time and capacity-utilization metrics were averaged to generate a smooth S-curve of system performance. To quantify the inflection point, a second-order polynomial and a piecewise spline-fitting procedure were applied to the aggregated data. The derivative of the fitted curve (Δ performance/Δ interoperability) exhibited a distinct global maximum at 0.59 ± 0.05, indicating the point of the most rapid efficiency gain. This inflection corresponds to the emergence of synchronized adaptive coordination among agents—an operational manifestation of the interoperability threshold. Beyond this level, performance improvements exhibited diminishing returns, confirming that digital investments yield non-linear benefits only once the connectivity–adaptivity coupling exceeds this critical point.

(3)

Quantitative Validation

Quantitative validation was conducted by comparing the simulation outputs with empirical benchmarks derived from operational data collected from the CGK export corridor. This process aimed to verify the model’s ability to replicate real-world coordination dynamics and operational delays observed within Indonesia’s air-cargo ecosystem. The results indicated that the combined export clearance process required an average of 46.18 h, reflecting the cumulative duration of documentation, inspection, and release stages, while communication delay averaged 12.22 h per shipment, primarily due to asynchronous exchanges between freight forwarders, ground handlers, and customs authorities. Meanwhile, the average airline slot utilization of 55.08% revealed a moderate level of efficiency across the corridor and pointed to persistent coordination gaps among key stakeholders. Together, these benchmarks served as reference points for assessing the model’s empirical validity and its capacity to reproduce real-world operational frictions and inter-agent dependencies characteristic of fragmented logistics systems.

Accuracy was evaluated using the Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) to quantify the deviation between simulated and observed performance metrics, as shown in the following equations:

• Mean Absolute Percentage Error (MAPE)

$$MAPE={\displaystyle \frac{1}{n}}\sum _{i=1}^n\left|{\displaystyle \frac{Y_{actual,i}-Y_{sim,i}}{Y_{actual,i}}}\right|\times 100\%$$

• Root Mean Square Error (RMSE):

$$RMSE=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^n({Y_{actual,i}-Y_{sim,i})}^2}{n}}}$$

The resulting values (

Table 3. Quantitative Validation Results Against Empirical Benchmarks.

Scenario	Scenario Name	MAPE Clearance (%)	RMSE Clearance (Hours)	MAPE Capacity (%)	RMSE Capacity (%)	Digital Interoperability Score
S1	Baseline	60.88	28.12	5.48	3.02	0.35
S2	Tech Adoption Only	14.11	6.52	20.55	11.32	0.55
S3	Process Reengineering Only	33.82	15.62	7.66	4.22	0.40
S4	Stakeholder Collaboration Only	3.43	1.58	41.43	22.82	0.45
S5	Partial Integration	16.85	7.78	55.05	30.32	0.65
S6	Full Integration without Behavioral Adaptivity	36.83	17.02	2.56	1.39	0.85

) demonstrate strong congruence between simulated and real-world data. For example, Scenario 4 (Stakeholder Collaboration Only) achieved the highest validation accuracy, with a clearance-time MAPE of 3.43% and an RMSE of 1.58 h, confirming that adaptive behavior alone can significantly improve coordination even without full technological integration. Conversely, Scenarios 1 and 6 reflected the persistent underutilization of capacity (≈55%) characteristic of low-interoperability systems, while intermediate scenarios (S2–S5) exhibited progressive improvement consistent with the emergence of a digital-interoperability threshold near 0.45–0.65.

As shown in Figure 3, the RMSE values remain low and stable, particularly for communication delay at approximately 11 h. This result closely aligns with the empirical benchmark of 12.22 h. The MAPE trend confirms that Scenario 4 aligns best with real clearance-time dynamics, whereas Scenarios 5–6 demonstrate diminishing improvements beyond the interoperability threshold. These quantitative outcomes provide empirical proof that the model captures both behavioral adaptivity and digital-connectivity effects accurately.

(4)

Structural Validation and Visual Comparison

Beyond numerical validation, the model’s structural validity was further examined through a comparative visualization using the Graph-Based Digital Twin (GBDT) framework. This approach contrasted Scenario 1 (Fragmented System) and Scenario 4 (Digital Adaptive System) to demonstrate how coordination structures evolve under different digital and behavioral configurations. As illustrated in Figure 4, the fragmented baseline (Scenario 1) exhibits a highly centralized coordination pattern dominated by freight forwarders, ground handlers, and customs authorities. Such a concentration of control reflects a system prone to communication asymmetries, overreliance on specific actors, and bottlenecks in information transfer—typical of low-digital-readiness environments in emerging logistics markets [6,24].

In contrast, the Digital Adaptive System (Scenario 4) displays a markedly decentralized and interconnected structure. Adaptive agents within this scenario facilitate real-time information exchange and dynamic decision-making, producing a flatter coordination topology. The visualized flows of communication are more evenly distributed among agents, reducing delays and improving system responsiveness. This transformation highlights the model’s ability to replicate realistic transitions from hierarchical and siloed operations toward collaborative, self-organizing digital ecosystems, a phenomenon also observed in empirical studies of digitally enabled logistics networks [19].

Complementing this qualitative visualization,

Table 4. Structural Metrics Across Comparative Scenarios.

Structural Metric	S1: Fragmented Baseline	S4: Digital Adaptive System	Improvement
Network Density	0.32	0.78	+144%
Average Path Length	4.2	1.8	−57%
Communication Redundancy	22%	68%	+209%
Bottleneck Nodes	8	2	−75%
Adaptive Connections	0%	85%	+85%
Agent Adaptivity	OFF	ON	-

compares the network-level structural evolution between Scenario 1 (Fragmented Baseline) and Scenario 4 (Digital Adaptive System). The metrics demonstrate how digital interoperability and behavioral adaptivity jointly reshape the coordination architecture. Scenario 4 represents the first configuration incorporating adaptive agent behavior, enabling network self-organization and synchronization of digital information flows.

Table 4 shows that network density rises from 0.32 to 0.78 (+144%), signifying stronger connectivity and interaction frequency among agents. The average path length decreases from 4.2 to 1.8 (−57%), reflecting faster communication cycles and reduced coordination distance. Additionally, the number of bottleneck nodes representing dependency points decreased by 75%, while adaptive connections increased by 85%, indicating higher agent autonomy and more diffused decision-making across the network. Together, these metrics quantitatively reinforce the visual evidence from Figure 4, confirming that digital adaptivity not only enhances efficiency but also reshapes the coordination architecture of the system into a more resilient and distributed network.

Collectively, the visual and quantitative findings validate that behavioral adaptivity, when integrated with digital interoperability, generates a systemic reconfiguration that is both structurally and functionally transformative. The results emphasize that the benefits of digitalization in fragmented logistics systems extend beyond operational efficiency and include architectural transformation, reshaping the underlying topology of inter-organizational interactions into a more resilient, self-organizing, and learning-oriented network structure.

3. Results

This section presents simulation outcomes from six scenarios (S1–S6) designed to evaluate how varying degrees of digital adoption and behavioral adaptivity influence coordination efficiency in Indonesia’s fragmented air-cargo export system. Three primary performance indicators were examined, including clearance time, communication delay, and capacity utilization, to capture both operational and systemic effects of digital interoperability. The results reveal how these dimensions interact to produce non-linear performance improvements and threshold effects characteristic of complex adaptive logistics systems [6,24].

3.1. Scenario-Based Outcome Patterns

Table 5. Simulation Results for Key Performance Indicators Across Scenarios.

Scenario	Scenario Name	Description	Avg. Clearance Time (Hours)	Avg. Communication Delay (Hours)	Avg. Capacity Utilization (%)
S1	Baseline	Low digital adoption, high friction, no adaptivity (baseline)	73.85	23.41	54.7
S2	Technological Adoption Only	Threshold digitalization (≈60%), high friction, with adaptivity	52.10	18.95	68.3
S3	Process Reengineering Only	Threshold digitalization (≈60%), capacity constraint, no adaptivity	61.92	22.88	59.1
S4	Stakeholder Collaboration Only	Fully digital, capacity constraint, with adaptivity	44.25	14.37	89.5
S5	Partial Integration	Fully digital, no friction, with adaptivity (optimal benchmark)	38.50	10.02	97.8
S6	Full Integration without Behavioral Adaptivity	Very high (95%) digital adoption, minimal friction, no adaptivity—technological saturation without learning	70.12	21.05	56.2

summarizes the comparative performance outcomes across the six simulated scenarios, each representing different combinations of digitalization and behavioral adaptivity. The baseline (S1) depicts the existing fragmented state of Indonesia’s air-cargo corridor, showing prolonged clearance times (73.85 h), significant communication delay (23.41 h), and underutilized capacity (54.7%). These results reflect operational inefficiencies typical of developing-country logistics systems, where manual documentation, asynchronous data sharing, and institutional fragmentation create cumulative coordination frictions [6].

As digital adoption increases, coordination performance improves non-linearly. Scenario 2 (Technological Adoption Only) achieves a 29.4% reduction in clearance time relative to S1, but the effect becomes substantial only once digital interoperability surpasses roughly 60%. This finding indicates that technological upgrades alone are insufficient to induce systemic change unless supported by behavioral readiness and mutual trust among agents [19,32].

Further variations across Scenarios 3 to 6 highlight how behavioral adaptivity amplifies the impact of digitalization in air-cargo coordination. Scenario 3 (Process Reengineering Only) yields only modest improvements, confirming that procedural redesign without adaptive coordination cannot fully overcome structural fragmentation.

In contrast, Scenario 4 (Stakeholder Collaboration Only) demonstrates the strongest empirical alignment with real CGK operational data, reducing clearance times to 44.25 h and communication delay to 14.37 h. These results suggest that collaborative and adaptive agent behavior is crucial for achieving synchronized decision-making, particularly when digital maturity remains uneven across stakeholders.

Scenario 5 (Partial Integration) delivers near-optimal system performance, with clearance times of 38.50 h and capacity utilization reaching 97.8%, indicating a balance between digital readiness and adaptive learning.

However, Scenario 6 (Full Integration without Behavioral Adaptivity) still shows residual inefficiencies even under friction-free conditions. This outcome is intentional, as Scenario 6 simulates a fully automated yet non-adaptive system. Although digital connectivity is maximized, the absence of learning mechanisms prevents agents from reorganizing interactions dynamically, resulting in stagnation similar to the fragmented baseline. The contrast between S5 and S6 therefore demonstrates that technological saturation alone does not guarantee coordination efficiency.

Collectively, these findings underscore that adaptive collaboration, not merely technological saturation, is the primary determinant of resilience and efficiency in digitally fragmented logistics ecosystems [6,11,24].

As illustrated in Figure 5, the transition from S1 to S6 reveals distinct improvement trajectories across all performance indicators, yet with differing sensitivities to digital and behavioral factors. Clearance time and communication delay display steep reductions between S3 and S4, marking the onset of adaptive coordination effects. Meanwhile, capacity utilization increases more gradually, suggesting that physical resource efficiency responds more slowly than informational or behavioral adaptation. These visual patterns reinforce the numerical findings discussed above, confirming that the integration of digital connectivity with adaptive behavior yields the most substantial coordination gains and that purely technological solutions cannot replicate these system-wide improvements.

3.2. Cross-Scenario Comparative Insights and Threshold Dynamic

The comparative analysis across all scenarios demonstrates that coordination efficiency follows a non-linear progression rather than a steady incremental trend. As illustrated in Figure 6, performance remains nearly stagnant within the low-digitalization range (S1–S3), where interoperability levels are below 60%. Once this critical point is exceeded, the system experiences a rapid increase in efficiency, confirming the presence of a digital-interoperability threshold. This inflection point indicates that digital tools alone do not yield significant transformation unless paired with adaptive behavioral mechanisms that enable stakeholders to exploit new connections dynamically. The behavior of this system resembles tipping-point phenomena in innovation diffusion theory [32] and the self-organizing transitions observed in complex adaptive logistics networks [6]. Below the threshold, communication delay and capacity inefficiencies persist due to fragmented trust and information asymmetry, whereas beyond it, the emergence of adaptive agents fosters synchronized coordination and real-time learning across the network.

Behavioral adaptivity thus functions as a force multiplier for digital investment. Under comparable levels of digital readiness (~60%), the adaptive configuration (S4) achieves a 13.2% higher clearance-time improvement compared with its non-adaptive counterpart (S3). This adaptive amplification effect underscores the importance of dynamic behavioral responses, including trust-based partner switching, self-optimization, and continuous performance feedback, in maximizing the outcomes of digital transformation. The resulting S-curve pattern demonstrates that air-cargo digitalization behaves as a phase-transition phenomenon, wherein moderate connectivity produces only marginal gains, but once adaptive learning matures, system performance accelerates exponentially before plateauing. These results align with findings from Baena et al. (2024) and Shadkam and Irannezhad (2025), which identified similar non-linear adoption and coordination effects in agent-based simulations of logistics and transport systems [19,24]. Collectively, the cross-scenario comparison confirms that digital interoperability and behavioral adaptivity are mutually reinforcing levers of logistics performance, providing a socio-technical explanation for the resilience and transformation capacity of developing-country air-cargo ecosystems.

As shown in Figure 6, coordination performance increases following an S-curve pattern as digital interoperability improves. At lower connectivity levels (S1–S3), the system remains trapped in a marginal-gains region, where investments in digitalization yield limited returns. Once the critical threshold (~60–65%) is surpassed, performance rises sharply, driven by adaptive learning and behavioral coordination among agents. The “adaptivity effect,” illustrated by the 13.2% improvement between S3 and S4, marks the onset of exponential gains as agents dynamically reorganize workflows and information exchange. Beyond an interoperability score of 0.75, the curve flattens into a diminishing-returns region, indicating that further digitalization provides smaller incremental benefits once behavioral intelligence stabilizes. This visual trend corroborates the simulation’s quantitative results and supports the theoretical argument that digital interoperability and adaptive behavior co-evolve non-linearly to generate systemic efficiency gains [6,19,24,32].

3.3. Quantitative Validation and Synthesis of Coordination Dynamics

The simulation’s quantitative outcomes were cross-validated against empirical benchmarks derived from operational data from the CGK export corridor, focusing on three key indicators, including clearance time, communication delay, and capacity utilization. As summarized in the validation results (Section 2.4), Scenario 4 (Stakeholder Collaboration Only) achieved the strongest alignment with real-world data, recording a clearance-time MAPE of 3.43% and RMSE of 1.58 h, closely approximating the empirical mean of 46.18 h. This high level of consistency confirms that adaptive behavior alone can significantly enhance coordination, even in the absence of complete digital integration. In contrast, Scenario 5 (Partial Integration) achieved marginally better operational performance but diverged more substantially from empirical baselines, underscoring its role as an aspirational rather than present-state benchmark. These findings affirm the external validity of the simulation and demonstrate that the model realistically replicates the coordination constraints and adaptive learning patterns observed in Indonesia’s air-cargo operations.

Beyond numerical accuracy, the simulation reveals a set of systemic coordination dynamics that explain how behavioral and technological interactions co-evolve. Efficiency gains emerge only after interoperability surpasses approximately 60%, confirming the existence of a digital threshold consistent with Rogers’ (2003) diffusion theory [32]. Adaptive agents consistently outperform static configurations by dynamically adjusting to evolving coordination bottlenecks, validating the behavioral assumptions embedded in the model. Furthermore, the synergy between digitalization and adaptivity produces exponential performance improvements, as visualized in Figure 5 and Figure 6. The clearance-time gains follow an S-curve trajectory, remaining slow below the threshold, rising steeply around 60–65%, and plateauing beyond 80%. This pattern indicates that air-cargo digitalization behaves as a phase-transition process rather than a linear progression [6,19,24]. Collectively, these findings substantiate that digital interoperability and adaptive behavior are mutually reinforcing mechanisms that transform fragmented logistics systems into coordinated, resilient networks.

4. Discussion

This section interprets the results of the simulation in relation to the conceptual frameworks of Autonomous Supply Chains (ASC), Graph-Based Digital Twins (GBDT), and digital interoperability thresholds. The aim is to explain how behavioral adaptivity and technological connectivity interact to shape coordination dynamics within fragmented air-cargo systems. The discussion connects the findings to established theories in digital logistics and complex adaptive systems while offering practical insights for logistics stakeholders in developing economies such as Indonesia.

4.1. Interpretation of Key Findings

The simulation findings confirm that digitalization in fragmented logistics systems enhances coordination only when accompanied by adaptive behavioral mechanisms. This is reflected in the non-linear relationship between digital interoperability and performance, where efficiency gains remain minimal below a 60% threshold and increase sharply once adaptive learning emerges. The results validate the proposition that the benefits of digital transformation are not solely technological, but instead arise from behavioral intelligence and collaborative adaptation among supply chain actors.

In Indonesia’s air-cargo corridor, which involves multiple semi-autonomous stakeholders with varying levels of digital readiness, behavioral adaptivity allows agents to self-organize, reconfigure information flows, and mitigate inefficiencies such as redundant documentation and asynchronous communication. The empirical alignment between simulation outputs and real-world performance benchmarks supports the conclusion that improving coordination efficiency requires the co-evolution of digital and behavioral capabilities, reinforcing the view that logistics digitalization represents a socio-technical transition rather than a linear automation process [6,32].

The non-linear S-curve behavior observed across scenarios also provides insight into how transformation unfolds in practice. When interoperability remains below the critical threshold, coordination continues to rely on manual interventions, perpetuating information bottlenecks and idle capacity. Once digital connectivity and adaptive feedback surpass the tipping point, systemic improvements emerge as agents begin to share trust-based data and align decision logic. This behavior exemplifies phase-transition dynamics in complex adaptive systems, where small increases in connectivity yield disproportionately large gains in system performance. Consequently, the findings advance understanding of how adaptive behavior amplifies the value of digital investments within fragmented transport ecosystems.

4.2. Theoretical Implications

This study contributes to theory by integrating Autonomous Supply Chains (ASC), Graph-Based Digital Twins (GBDT), and interoperability thresholds into a unified agent-based framework that explains the transformation of coordination structures. The ASC perspective captures decentralized decision-making and self-organizing behaviors among logistics agents [13,33], while the GBDT architecture models digital connectivity and information flows that enable real-time coordination [11,14]. Together, these frameworks form a multi-layered socio-technical model in which behavioral autonomy and digital transparency interact dynamically to produce emergent resilience [9,19,24]. This integration extends earlier ABM studies that focused predominantly on technology adoption [24] or static optimization [27], positioning digital logistics as an adaptive network shaped by learning, feedback, and inter-agent trust [19].

The identification of a measurable interoperability threshold (~0.6) advances digital-maturity and innovation-diffusion theory by operationalizing the moment at which localized behavioral changes scale into system-wide transformation. The transition from fragmented to synergistic coordination mirrors Rogers’ (2003) diffusion curve but introduces a quantitative link to operational outcomes such as clearance time and capacity utilization [32]. This finding bridges the conceptual gap between behavioral adaptation and system-level efficiency, offering a theoretical mechanism for digital resilience in logistics networks. Moreover, the model provides empirical support for the assertion that system robustness does not emerge from centralized optimization, but rather from distributed intelligence among semi-autonomous agents—aligning digital logistics research with broader discourse on complex adaptive supply chains [6].

4.3. Practical Implications

From a managerial and policy perspective, the results emphasize that achieving digital transformation in logistics requires synchronizing technological infrastructure with human adaptivity. In the Indonesian context, the feasibility of implementing a digitally integrated freight community platform should prioritize not only technical interoperability but also behavioral readiness across logistics stakeholders [22,34]. Although initiatives such as the Indonesia National Single Window (INSW) and early-stage Air Cargo Community Systems (ACCS) have been introduced, integration remains limited across key actors. For instance, while Soekarno-Hatta International Airport has initiated the development of an air cargo digital platform under the National Logistics Ecosystem (NLE) agenda, progress remains in preparatory or pilot stages, with limited stakeholder adoption. Likewise, although INSW enables partial electronic document exchange, integration with freight forwarders and ground-handling agents remains insufficient. Policies such as the 2020–2024 NLE roadmap emphasize inter-agency coordination, yet implementation continues to vary across logistics corridors.

Policymakers can facilitate progress by designing shared digital governance frameworks that incentivize data exchange and cross-organizational learning. Training programs based on simulation or digital twin environments can help stakeholders experience coordination benefits prior to full implementation, thereby reducing resistance to change. Additionally, institutionalizing adaptive practices, including decentralized decision rights and performance-based feedback mechanisms, can accelerate the transition of logistics organizations toward autonomous, data-driven, and resilient ecosystems. The Indonesian air-cargo corridor may thus serve as a reference model for other emerging logistics systems, demonstrating that sustainable efficiency gains arise not from technology alone, but from the continuous interaction among digital integration, human learning, and system-level feedback.

5. Conclusions and Future Research Directions

This study developed and validated an agent-based model (ABM) to explore how digital interoperability and behavioral adaptivity interact in fragmented air-cargo logistics systems. By integrating Autonomous Supply Chains (ASC), Graph-Based Digital Twins (GBDT), and digital interoperability thresholds, the research provides a new theoretical lens for understanding digital transformation as a socio-technical co-evolution process rather than a purely technological upgrade. The simulation results confirm the presence of a non-linear threshold effect, where coordination efficiency remains stagnant at low levels of digitalization but rises sharply once interoperability exceeds approximately 60%. At this point, agents begin to self-organize and optimize information flows. These findings bridge empirical operational data with theoretical constructs, advancing both the methodological and conceptual foundations of digital logistics.

From a managerial perspective, the results demonstrate that technological adoption must be synchronized with behavioral readiness. For logistics practitioners, particularly within developing contexts such as Indonesia, success depends on enabling data-sharing ecosystems and fostering adaptive collaboration across freight forwarders, airlines, ground handlers, and customs authorities. Initiatives such as IATA ONE Record, e-AWB, and the National Single Window can significantly enhance efficiency if complemented by training and governance mechanisms that promote trust and continuous learning. For policymakers, the findings suggest that investment in digital infrastructure should be accompanied by capacity-building programs and institutional incentives that stimulate inter-organizational interoperability and behavioral innovation.

Theoretically, this study contributes by operationalizing interoperability thresholds within an ABM framework, quantifying the tipping point between fragmented and synergistic coordination. Methodologically, it demonstrates how combining ABM with GBDT can capture emergent system behavior and validate complex coordination dynamics using empirical data. Nevertheless, several limitations constrain the extrapolation of these findings. The model simplifies certain behavioral variables and is calibrated on a single corridor (the CGK export network), which reflects Indonesia’s specific institutional and infrastructural contexts. Its generalization to other air-cargo corridors may vary depending on governance structures, regulatory environments, and digital maturity levels. Moreover, agent behavior in the simulation considers only adaptive learning, excluding strategic or interest-based interactions that may influence coordination in real markets.

Future research should further examine the role of communicable modeling frameworks in promoting the adoption of simulation-based decision support across real-world logistics ecosystems. Potential extensions include calibrating multimodal and multi-corridor networks, integrating policy-driven interventions (e.g., data standardization incentives), and embedding game-theoretic or machine learning-based adaptive behaviors to simulate long-term strategic dynamics. These refinements would enhance the model’s external validity and applicability to broader digital transformation agendas in global air cargo.