Introduction

As essential chemical raw materials, dangerous goods play a vital role in the economy1,2. However, road traffic crashes involving HAZMAT transportation can easily lead to severe consequences. Compared with HAZMAT related activities in fixed locations, transportation is far more susceptible to external environmental influences, making the situation more complex and variable. Therefore, conducting safety risk assessments based on road traffic crash scenarios that describe transportation tasks is highly valuable for preventing such crashes3,4.

Existing research on HAZMAT transportation-related road traffic crashes has primarily focused on railway transportation.5,6, road transport7,8 and maritime transport9, and has established a relatively complete research framework from the perspectives of road traffic crash statistical analysis10,11,12, formation mechanism13, influencing factor analysis14,15,16, safety risk assessment17,18,19, route optimization20,21, and emergency response22. However, road traffic crash scenario analysis of HAZMAT transportation on highways remains uncommon. At present, scholars mainly conduct scenario analysis by conducting statistical analysis on accident cases or establishing causal models. Statistical analysis relies on historical data and summarizes accident scenarios based on dimensions such as road characteristics, time, and causes, and proposes corresponding safety measures23,24. Causal models use accident causal theory to describe the causes, processes, and consequences of accidents, thereby explaining the accident mechanism and providing a theoretical basis for accident prevention25,26,27. Common methods for constructing causal models include Bayesian networks28,29, event trees30, fault trees31, fuzzy logic32, and Monte Carlo simulation33. However, these methods have difficulty achieving the desired results when dealing with complex collision processes in the real world34. As an emerging causal modeling method, accident networks have shown higher efficiency in capturing and analyzing complex collision mechanisms35,36. As the main mode of road traffic, highways are characterized by high speeds and large traffic volumes. In addition, special sections such as tunnels and bridges introduce additional risks, making them more dangerous than ordinary roads. Therefore, decomposing traffic tasks in this process helps to analyze collision scenarios more deeply.

Overall, while the risk assessment of dangerous goods transportation has been extensively studied, research on highway HAZMAT transportation crash scenario analysis remains limited. HAZMAT transportation crashes typically result from a combination of multiple risk factors and outcomes, making their comprehensive and detailed analysis highly challenging. To address this gap, this paper proposes a crash scenario analysis method AcciNet-RFs, which incorporates risk factors into crash networks and enables a systematic analysis of highway HAZMAT transportation crash scenarios from a task-based perspective.

Methodology

Overview

First, a road traffic crash scenario model is constructed using the AcciNet-RFs method. This model identifies road traffic crash scenarios and generates the corresponding scenario network diagram by integrating the highway HAZMAT transportation HTA tree diagram, task network, and actor network. Next, through risk level assessment, the identified crash scenarios are predicted and classified, and a crash classification model is developed to achieve the purpose of prevention in advance. The task analysis, identification of risk factors, and determination of their associated weights in the AcciNet-RFs method are based on 523 highway HAZMAT transportation road traffic crashes37. It should be noted that the scope of this study on highway HAZMAT transportation road traffic crashes excludes loading and unloading operations, focusing instead on the transportation process.

Data sources

The data for this study mainly come from three sources:

  • The Chemical accident Information Network operated by the Chemical Registration Center of the Ministry of Emergency Management38, which has been continuously publishing records of chemical road traffic crash related to production, operation, storage and transportation since 2012;

  • The Disaster accident Information Platform of the Ministry of Emergency Management of the People’s Republic of China39, which provides major road traffic crash investigation reports including chemical road traffic crash;

  • The China Dangerous Chemical Logistics Network operated by the Dangerous Chemical Logistics Branch of the China Federation of Logistics and Purchasing40, which publishes information on domestic and international dangerous chemical logistics and transportation road traffic crash.

The statistical period for highway dangerous goods transportation road traffic crash in this study is from January 1, 2018 to December 31, 2021, and a total of 678 road traffic crash cases were collected. After collection, the original data were manually screened for problems such as non-highway road traffic crash, non-hazardous chemical road traffic crash, irrelevant road traffic crash types, duplicate records and missing information. Finally, 523 highway dangerous chemical transportation road traffic crash data that met the research criteria were obtained to ensure the accuracy, completeness and operability of the data set.

Road traffic crash scenario model

The method employed in the road traffic crash scenario model is AcciNet-RFs. Salmon et al.41,42 first proposed AcciNet (the Accident Network) at the Proceedings of the 2020 HFES 64th International Annual Meeting. AcciNet takes the actor network and task network of the system under analysis as its main inputs and uses them to identify crash-contributing factors, the interrelationships among these factors, and the relevant actors. Given the volatile and complex nature of highway HAZMAT transportation crashes, AcciNet is highly suitable for analyzing such events. However, the occurrence of these crashes is essentially driven by the incomplete execution of transportation tasks, with risk factors serving as the root causes of task failures. Therefore, during AcciNet analysis of highway HAZMAT transportation crashes, it is necessary to incorporate explicit identification of risk factors. Building on this, the present study proposes a new method, AcciNet-RFs, for analyzing highway HAZMAT transportation scenarios. This method enhances the original AcciNet approach by adding a risk factor identification process, enabling a one-to-one correspondence among participants, hierarchical tasks, and risk factors. As a result, AcciNet-RFs provides a full-process, all-element crash scenario analysis that allows highway HAZMAT transportation crashes to be examined more comprehensively and effectively.

The AcciNet-RFs workflow adds four new steps to AcciNet and adjusts three existing steps, as shown in Fig. 1. The green shaded area in the figure represents the new steps, and the yellow shaded area represents the adjusted steps.

As shown in Fig. 1, although the AcciNet-RFs workflow was developed based on an analysis of road traffic crash involving hazardous materials transport, the root cause of any road traffic crash lies in the presence of risk factors that lead to incomplete task execution. Therefore, similar to the AcciNet framework proposed by Salmon et al.38,39, AcciNet-RFs is also applicable to road traffic crash handling in other industries. It is particularly suitable for scenarios with complex systems, numerous potential risk factors, and where the traditional “human, machine, environment, and management” classification cannot fully describe the road traffic crash.

HTA build

To draw a highway HAZMAT transportation road traffic crash scenario network diagram, an HTA analysis of highway HAZMAT transportation road traffic crashs, including risk factor identification was first conducted. Finally, an HTA tree diagram was developed based on road traffic crash case analysis, document review, on-site inspection and observation, and expert seminars (Fig. 2). Among them, rectangular nodes represent tasks, and elliptical nodes represent risk factors that may affect the successful completion of tasks. It should be noted that in China, HAZMAT storage devices are generally provided by HAZMAT storage units, and the transportation company is not responsible for this (except for tank trucks). Therefore, the inspection of HAZMAT storage devices mentioned in the HTA tree diagram refers specifically to tank trucks.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Flow chart of AcciNet-RFs operation.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

HTA tree diagram for highway HAZMAT transportation.

The HTA tree diagram of highway HAZMAT transportation is shown in Fig. 2. It comprises 11 first-level tasks and 13 s-level tasks. Based on this structure, 16 common risk factors that may occur across different tasks are identified. Various combinations of these tasks and risk factors constitute different HAZMAT transportation road traffic crash scenarios. It should be noted that the framework presented here reflects the general process of HAZMAT transportation in China; specific transportation cases should be adapted accordingly prior to analysis. In Fig. 2, risk factors are no longer categorized under the traditional dimensions of people, machines, environment, and management, but are instead directly associated with specific tasks. This approach provides a clearer representation of the tasks and corresponding risk factors throughout the entire highway HAZMAT transportation process.

Task network analysis

The highway HAZMAT transportation task network is constructed by obtaining the last level of hierarchical tasks from the highway HAZMAT transportation HTA tree diagram and identifying the interrelationships existing between them based on the above relationship types. For example, in highway HAZMAT transportation, “obtaining transportation tasks” and “designing transportation tasks” are related, because the transportation company only needs to design transportation tasks after acquiring the transportation tasks. The task network represents the HTA tree diagram output as a network, displays vital tasks and their interactions43, and can conveniently display the interactive relationships between tasks in the entire work system. After the highway HAZMAT transportation task network was first constructed, it was gradually refined based on feedback from expert seminars. Finally, it formed the “Highway HAZMAT Transportation” task network, as shown in Fig. 3. In Fig. 3, circular nodes represent tasks, and arrows connecting tasks represent relationships between tasks.

As shown in Fig. 3, the highway HAZMAT transportation task network contains 20 tasks, which are interrelated with other tasks. These include tasks related to transportation task acquisition and obtaining road passes, as well as pre-transportation inspections and tasks related to the transportation process. Other essential tasks include HAZMAT transportation process supervision, vehicle dispatching, etc. Finally, functions for when dangerous situations occur are included. Only when each task is completed can the goal of safe and compliant highway HAZMAT transportation be achieved.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Task Network for expressway hazardous chemical transportation.

Actor network analysis

One of the main inputs to AcciNet-RFs, the actor network, shows information about the different people and organizations that share responsibility for related road traffic crashs and, in particular, shows which tasks in the task network they are undertaking. Therefore, the highway HAZMAT traffic road traffic crash scene network diagram can be accurately established only by constructing an accurate and reasonable actor network. The highway HAZMAT traffic actor network is shown in Fig. 4.

In highway HAZMAT transportation, the main participants include transportation companies, drivers, escorts (when there are two drivers in the exact vehicle, the driver not performing the driving task is the escort), HAZMAT transport vehicles, other vehicles, and relevant government units. Those actors have management, observation, and other responsibilities. When these participants perform their duties conscientiously, they can more effectively complete various tasks in the road transportation of dangerous goods, thereby safely transporting hazardous goods.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Actor network of expressway hazardous chemicals transportation system.

Road traffic crash scenario network diagram

The road traffic crash scenario network diagram is the final presentation of AcciNet-RFs analysis. When creating a road traffic crash scenario network diagram, all identified contributory factors are mapped to the task network and displayed as a shade (light red) on the outermost layer. The actors and risk factors associated with each road traffic crash contributory factor are also added to the task network diagram. In the middle shading, dark shading (red) represents risk factors that lead to inappropriate factors. Tasks that represent “normal performance” in the incident scenario network diagram are shaded lightly (green) to indicate that no failure or inappropriate behavior occurred. Relevant actors associated with each instance of average performance are added to the task network diagram. This annotated task network is called a road traffic crash scenario network diagram.

Road traffic crash classification model

Risk is defined as the product of the probability of a traffic road traffic crash and its corresponding consequences44. The road traffic crash classification model uses weights based on hierarchical tasks and risk factors to calculate the risks of different HAZMAT transportation road traffic crash scenarios described in the Road traffic crash Scenario Model, then classifies them to indicate the relative degree of risk.

For highway HAZMAT transportation road traffic crashs, the degree of risk controllability will directly impact the probability of risk occurrence. For example, road traffic crashs are possible in highway HAZMAT transportation tasks in rainy and snowy weather. However, suppose the potential impact of rainy and snowy weather is predicted in advance while assessing environmental risk levels; the transportation time may be changed, or corresponding measures may be taken. Protective measures to improve risk controllability will most likely reduce the probability of road traffic crashs. Based on the perspective of risk mitigation, this article adds the risk parameter of “risk controllability” to three different dimensions of risk (possibility of occurrence (P), severity of consequences (L) risk controllability (S)), and builds a road traffic crash classification model. Conduct quantitative assessments on highway HAZMAT transportation road traffic crashs to provide a basis for accurate and effective safety control measures.

Risk assessment model

The risk assessment model mainly comprises the formula for solving the severity of consequences, the formula for solving the possibility of occurrence, and the formula for solving the risk controllability.

  1. (1)

    The formula for solving the severity of consequences.

    The severity of the consequences of highway HAZMAT traffic road traffic crashs mainly includes loss of personnel and property and damage to the surrounding environment. Therefore, the severity of the results is primarily related to the quantity of HAZMAT, its hazardous properties, and the sensitivity of the surrounding environment (Table 1). The sensitivity of the surrounding environment is mainly related to the population density and surrounding environmental functions (farmland, drinking water sources, etc.) of the HAZMAT transportation section(Table 2). According to relevant research45,46, combined with the highway HAZMAT transportation characteristics, the HAZMAT consequence risk value is calculated according to Eq. (1), in which the HAZMAT own risk level is calculated according to Eq. (2)47,48.

    $$P = \frac{{m \cdot N \cdot H}}{4}$$
    (1)

    In Eq. (1), P is the HAZMAT consequence risk value; m is the HAZMAT transportation volume classification; N is the HAZMAT’s risk level; H is the environmental sensitivity.

    Table 1 Hazardous chemicals transportation volume classification (m).
    Full size table
    Table 2 Environmental sensitivity grading (H).
    Full size table
    $$\:\:\:N = \frac{{N_{H}^{2} + N_{F}^{2} + N_{R}^{2} + N_{S}^{2} }}{{4(N_{H} + N_{F} + N_{R} + N_{S} )}}$$
    (2)

    In Eq. (2), NH is a health hazard; NF is flammability; NR is chemical reactivity; NS is a particular hazard; the HAZMAT consequence risk value calculated using Eq. (1) and Eq. (2) is taken within [0,4] value.

  2. (2)

    The formula for solving the probability of occurrence.

    Many different combinations of hierarchical tasks and risk factors during highway HAZMAT transportation. Think of these different hierarchical tasks and risk factors as the coordinate axes of the corresponding n-dimensional space coordinate system, with the coordinate origin as the safe base point. The n-dimensional space coordinates are composed of the influence degrees of these hierarchical tasks and risk factors to the space of the origin. Distance represents the possibility of a road traffic crash. The possibility of a road traffic crash can be calculated by Eq. (3).

    $$\:L = \sqrt {(\omega _{1} l_{1} )^{2} + (\omega _{2} l_{2} )^{2} + \cdots + (\omega _{n} l_{n} )^{2} } /\sqrt n \:\:$$
    (3)

    In Eq. (3), l1、l2、…、ln are the evaluation values of the considered risk factors, and the value range is [0–4]; ω1、ω2、…、ωn are the weight values of the corresponding risk factors and the value The range is [0–1]; n is the number of risk factors, and the road traffic crash probability calculated by this formula takes a value within [0,4].

  3. (3)

    The formula for solving the controllability of risk.

    The idea of solving the risk controllability is consistent with the risk possibility, as shown in Eq. (4).

    $$\:S = \sqrt {(\mu _{1} s_{1} )^{2} + (\mu _{2} s_{2} )^{2} + \cdots + (\mu _{n} s_{n} )^{2} } /\sqrt n$$
    (4)

    In Eq. (4), s1、s2、…、sn are the evaluation values of the considered risk factors, and the value range is [0–4]; µ1、µ2、…、µn are the weight values of the corresponding risk factors, and the value The range is [0–1]; n is the number of risk factors, and the road traffic crash probability calculated by this formula takes a value within [0,4].

Index system construction and weight determination

Constructing a safety risk indicator system and adopting scientific assessment methods to evaluate the degree of risk are common in HAZMAT safety risk assessment. This article is based on the highway HAZMAT transportation HTA tree diagram. After expert discussion, the hierarchical tasks and risk factors are classified according to the two dimensions of occurrence possibility and risk controllability, and a highway HAZMAT transportation safety risk index system is established (Fig. 5). These include a total of 3 occurrence possibility indicators and a total of 6 risk controllability indicators. Obtaining transportation tasks is only related to the severity of consequences. The severity of consequences is calculated based on HAZMAT transportation volume, its characteristics, and environmental sensitivity, so it is not reflected in Fig. 5. Classified task reports and investigation road traffic crashs are mainly handled by relevant government departments afterward, so they must be reflected in this part. In the figure, the light shade (yellow) is the occurrence possibility indicator, and the dark shade (blue) is the risk controllability indicator. To eliminate the influence of subjectivity to the greatest extent, we focused on eight significant road traffic crashs in China from 2010 to the present with detailed road traffic crash investigation reports, analyzed the probability of occurrence of each graded task and risk factors, and invited experts to comment (professors of safety engineering at Guizhou University and staff of safety risk assessment institutions). The data reviewed by experts will be used as the final weight value of each risk factor. If different vehicles are involved in the road traffic crash, the focus will only be on whether the HAZMAT transport vehicle’s tasks during the entire transport process were fully completed. The final determination of each hierarchical task and indicator weight is shown in Fig. 5.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Risk index system of expressway hazardous chemical transportation.

Usually, the index system is constructed based on risk factors. In this case, if the first-level index is defined as a consequence severity index, its subordinate index is a consequence severity index. However, this article focuses on the tasks that must be completed while transporting hazardous chemicals on highways, supplemented by risk factors. This makes the sub-indicators of the first-level indicators have both occurrence probability indicators and risk factors. Controllability indicators. At the same time, the same risk factor may appear in different tasks and have different weights. To this end, this paper first classifies the risk factors. Then, it calculates the risk magnitude of each classification task sequentially based on the established three-dimensional risk matrix assessment model. The specific classification of risk factors is shown in Table 3.

Table 3 Risk factor grading.
Full size table

Risk level calculation

Based on using a semi-quantitative analysis method to classify the three parameters of consequence severity, occurrence possibility, and risk controllability, the commonly used “multiplication” calculation is used to calculate the magnitude of road traffic crash risk. As shown in Eq. (5).

$$\:R = f(P,L,S) = P \times L \times S$$
(5)

Combining the severity of consequences, possibility of occurrence, and risk controllability classification, it can be seen that the range of safety risk magnitude is [0, 64]. This paper divides the safety risk magnitude into four levels: unacceptable risk (major), Risk that is acceptable but requires rectification (larger), tolerable risk (normal), and sound risk (low), as shown in Table 4.

Table 4 Risk magnitude classification description.
Full size table

During risk assessment in AcciNet-RFs, the underlying risk factors should be used as the basis and calculated upward step by step due to the multi-level grading tasks involved. To ensure security, it is essential to conduct a risk assessment for each level of classification task. The obtained score can then be used to determine the corresponding risk level. The risk of the final task can be calculated by calculating the previous level of classification tasks based on the index weight. However, since the possibility of consequences is only involved in calculating the final task risk level, only the likelihood of occurrence and risk controllability are calculated when calculating the graded task. Among them, the risk factors of some hierarchical tasks only involve the possibility of occurrence (or risk controllability). Coupled with teaching hierarchical tasks and the risk factors, the calculated risk magnitude value will rarely exceed four. Therefore, if the risk magnitude calculated in this article is a decimal, to avoid road traffic crashs as much as possible, it is rounded up, and the rounded value is the final risk magnitude. In the process of calculating the risk magnitude of a graded mission. If the final risk level exceeds level four, it will be determined as the maximum risk level (level four).

Example applications

Road traffic crash introduction

The case in this article is selected from the tank truck explosion road traffic crash on my country’s highways on June 13, 202049,50 (after this referred to as the 6·13 Wenling tank truck explosion road traffic crash). This road traffic crash resulted in the death of 20 people while 175 were hospitalized. The direct economic loss was 94.77815 million yuan. The 6.13 Wenling tank truck explosion road traffic crash was a major production safety road traffic crash in which a liquefied petroleum gas tank truck drove over a highway ramp at excessive speed, caused rollover, collision, and leakage, and then caused an explosion.

Road traffic crash scenario network diagram drawing

According to the investigation report on the June 13 Wenling tank truck explosion road traffic crash46, it can be seen that in this road traffic crash, the driver, transport vehicle, and tank were fully qualified and in compliance with the specifications, the driving time met the requirements, and the liquefied petroleum gas filling met the criterion. In the end, the road traffic crash investigation team determined that the direct cause of the road traffic crash was that the driver failed to take deceleration measures in time when driving the vehicle from a road with a speed limit of 60 km/h to a curved road with a speed limit of 30 km/h, causing the vehicle to roll over. It was determined that the indirect cause of the road traffic crash was that the rotating anti-collision guardrail was not constructed according to the design and did not meet the requirements of relevant technical standards. The road traffic crash investigation report also pointed out the main reasons for the road traffic crash and the responsibilities of supervision and other aspects.

It can be obtained from the road traffic crash investigation report46 that the road traffic crash promotion factors related to the main participants in the 6·13 Wenling tanker explosion road traffic crash in getting the transportation task and completing the transportation task are as follows:

  1. (1)

    When the driver drove the vehicle from a section with a speed limit of 60 km/h to a curved section with a 30 km/h speed limit, he failed to take deceleration measures in time.

  2. (2)

    The rotating anti-collision guardrail was not constructed according to the design and did not meet the requirements of relevant technical standards.

  3. (3)

    The transportation company and its principal persons in charge failed to implement the main responsibilities of safety production, such as GPS dynamic supervision and truthful uploading of electronic road bills.

  4. (4)

    The transportation authorities have not fulfilled their responsibilities in issuing HAZMAT road transport operating licenses and assessing the dynamic GPS monitoring of HAZMAT transport vehicles.

  5. (5)

    Public security organs fail to perform their duties of managing the traffic order of HAZMAT transport vehicles.

Based on the above road traffic crash promotion factors, the road traffic crash network diagram of the 6.13 Wenling tank truck explosion road traffic crash was drawn, as shown in Fig. 6.

As shown in Fig. 6, the 6.13 Wenling tank truck explosion road traffic crash was caused by road traffic crash-promoting factors such as insufficiently performed tasks, entirely performed tasks, and failed tasks. The incident scenario network diagram shows multiple instances of tasks being insufficiently or inappropriately performed as follows:

  1. (1)

    Designing transportation tasks: Transportation companies fail to fully understand the safety conditions of transportation route infrastructure when designing transportation tasks. Therefore, when the HAZMAT transport vehicle lost control due to speeding, the anti-collision guardrail failed to play its role in safety protection.

  2. (2)

    Assessing road risk levels: Transportation companies fail to consider the safety level of transportation route infrastructure effectively after completing the design of transportation tasks.

  3. (3)

    Obtaining a pass: The transportation company illegally entrusted other units to produce the electronic waybill and failed to perform its supervisory duties, resulting in the false filling of the electronic waybill. At the same time, when the relevant government units reviewed the electronic road bill declaration, they failed to detect any false declarations and issued the pass.

  4. (4)

    Driving the vehicle: The vehicle speeded through a curve that required deceleration, causing the HAZMAT transport vehicle to lose control and rollover.

  5. (5)

    Observe the road conditions: The driver made a mistake in observing the road conditions while driving the vehicle and did not slow down, causing the HAZMAT transport vehicle to roll over.

  6. (6)

    Supervise the driving speed of HAZMAT vehicles: the transportation company failed to implement GPS supervision responsibilities; the escort did not play a role in reminding and stopping.

  7. (7)

    Report danger: This task could not be performed because the driver and escort were dead at the time of the road traffic crash.

  8. (8)

    Respond to emergencies: This task was not performed because the driver and escort were deceased at the time of the road traffic crash.

In addition, the road traffic crash scenario network diagram also shows the risk factors that lead to the occurrence of inappropriate factors. For example, the task of driving a vehicle is not performed effectively because the driver is speeding. The driver’s speeding is due to a lack of crisis awareness and incomplete professional skills when encountering curves.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

Highway HAZMAT transportation road traffic crash scenario network diagram.

Road traffic crash classification assessment

The network diagram of the 6·13 Wenling tank truck explosion road traffic crash scenario shows failed or uncompleted hierarchical tasks throughout the transportation process and the risk factors that led to task failure. Combined with the investigation report on the 6.13 Wenling tank truck explosion road traffic crash, the output table of safety risk factors is drawn as shown in Table 5.

Table 5 Output table of risk factors for 6.13 Wenling tank car explosion road traffic crash.
Full size table

Based on the safety risk factor output table of the 6·13 Wenling tank truck explosion road traffic crash and with experts’ opinions, a safety risk factor classification table of the 6·13 Wenling tank truck explosion road traffic crash was drawn (Table 6).

Table 6 Classification of risk factors for 6.13 Wenling tank car explosion road traffic crash.
Full size table

Table 6 shows the classification of each risk factor in the 6·13 Wenling tank truck explosion road traffic crash. Combined with the road traffic crash classification model in Sect. 2.3, the calculated first-level layered task risk classification score of the 6·13 Wenling tank truck explosion road traffic crash is shown in Table 7.

Table 7 6.13 Wenling tanker explosion road traffic crash level 1 hierarchical task risk classification.
Full size table

According to the risk classification of the first-level layered tasks in Table 7, the probability of the 6·13 Wenling tank truck explosion road traffic crash is calculated to be three; the risk controllability is two.

The HAZMAT transported in the Wenling tanker explosion on June 13 was liquefied petroleum gas totaling 25.36t. According to the characteristics of liquefied petroleum gas, it is determined that the HAZMAT transportation volume classification (m) is level four, the health hazard classification (NH) is level two, the flammability classification (NF) is level four, the reactivity classification (HR) is level four, and the special hazard Sex rating (NS) is one. There are residential buildings, factories, and other people gathering areas around the road traffic crash. Based on the number of casualties in the road traffic crash, the environmental hazard classification was determined to be Level 3. The HAZMAT risk value P calculated according to Eq. (1) and Eq. (2) is 2.6 points, and the corresponding consequence severity level is level three, that is, P = 3.

The safety risk calculated according to Eqs. (4–6) is R = 18 points, and the risk level is serious (level three). consistent with the actual circumstances of the road traffic crash. It proves that the three-dimensional risk matrix evaluation system built in this paper has certain practical application value. Then, the safety risk R calculated according to Eq. (5) is 18 points, and the risk magnitude is serious (level three).

Discussion

Highway hazardous materials road traffic crash scenario analysis helps understand the patterns of road traffic crash occurrence, while road traffic crash classification models support the prediction or retrospective assessment of the risks of tasks at all levels during the transportation process, thereby providing guidance for risk management.

This study used the AcciNet-RFs method to construct a road traffic crash scenario model, which can reveal the relationship between adequate and inadequate task execution, as well as the risk factors that lead to inadequate task execution, thereby enabling a more comprehensive and systematic analysis of road traffic crash causes. However, the AcciNet-RFs method has not yet been widely validated and applied. Future research will collect a large number of highway hazardous materials road traffic crash cases to verify the stability and generalization ability of the model, and continuously improve and refine it based on actual conditions to fully evaluate its effectiveness. The road traffic crash classification model based on AcciNet-RFs focuses on the transportation process and has not yet incorporated emergency response resources along the transportation route, such as fire stations51,52. Therefore, the risk assessment indicator system can be further enriched based on existing research. It should be noted that the original design intention of AcciNet-RFs was to model the interaction between tasks, participants, and risk factors, focusing on road traffic crash mechanism modeling and task process reconstruction, rather than directly predicting road traffic crash probability or outcomes. This is fundamentally different from risk-based models such as Bayesian networks and road traffic crash trees. Therefore, to maintain the simplicity and operability of the method, this paper employs a multiplicative model, commonly used in risk assessment. Future research can further optimize the risk quantification process within the existing framework to achieve a seamless integration of road traffic crash mechanism modeling and risk assessment. The applicability of AcciNet-RFs depends on the completeness of task decomposition and risk factor identification, and there is still a potential risk of failure in extreme cases. For example, when input information is severely missing, expert weights vary significantly, or new risk factors outside the framework emerge, the model may produce biased or even misleading results. Future research can mitigate these risks by expanding the case library, incorporating multi-source validation, and conducting sensitivity analyses. This study primarily employed task identification and internal factors such as “expertise,” “crisis awareness,” and “monitoring failure” to conduct risk scenario analysis, thereby ensuring objectivity. However, the current risk quantification process does not explicitly capture typical driving behaviors such as speeding, drunk driving, and fatigued driving, which may limit the model’s practical relevance in road traffic safety. Future research will expand the applicability of the AcciNet-RFs framework by incorporating these behaviors-speeding in particular, along with drunk driving, fatigued driving, and illegal overtaking-as secondary risk factors. This enhancement will enable a closer alignment between risk factors and HTA task nodes, thereby improving both the explanatory power and practical utility of the model in road traffic safety.

This analysis of highway hazardous materials road traffic crash scenarios is based on the Chinese highway hazardous materials transport task system and road traffic crash cases, which may limit its general applicability to other countries. Nevertheless, this study provides a valuable reference. Furthermore, the AcciNet-RFs method can be applied not only to the creation and risk assessment of highway hazardous materials road traffic crash scenarios but also to other types of transport scenarios, such as rail, aviation, and maritime transport. By adjusting the risk factor identification system and scenario network diagram, this method can adapt to the characteristics and risk conditions of different modes of transportation.

Conclusion

This paper proposes an advanced method, AcciNet-RFs, to construct a highway HAZMAT transportation road traffic crash scenario network diagram for analyzing the complex crash processes in highway HAZMAT transportation. Based on this framework, quantitative risk assessment is applied to classify crashes, thereby facilitating prevention.

To develop the crash scenario network diagram, the HTA method was first employed to identify 11 first-level tasks and 13 s-level tasks involved in highway HAZMAT transportation. On this basis, 16 common risk factors potentially associated with different tasks were identified, forming a comprehensive highway HAZMAT transportation risk factor identification system. From the lowest-level tasks in the HTA tree, a task network comprising 20 hierarchical tasks was established. In addition, six main actors in the transportation process were identified, and an actor network was constructed.

As a demonstration, the 6.13 Wenling tank truck explosion in China was analyzed. A corresponding crash scenario network diagram was generated, and the risk values of hierarchical tasks at each level were calculated using the crash classification model. The resulting scenario network diagram enables the creation of potential crash scenarios, accelerates the risk assessment of highway HAZMAT transportation, and supports the identification of critical safety measures in advance—thus advancing the transition from risk assessment to risk management.