1. Introduction
Hazardous chemical warehouses and storage tanks are locations for the storage and maintenance of hazardous chemicals such as chemical raw materials, chemical drugs, chemical reagents, pesticides, etc. Because of the large quantity, variety and high risk of hazardous chemicals, the potential storage hazards can exceed those of the production, transportation and use of hazardous chemicals. Human factors play an important role in the occurrence of hazardous chemical accidents in China [
1,
2]. Therefore, it is necessary to analyze and identify human factors in hazardous chemical storage accidents. In addition, this paper continued the research in reference [
3], combining fault tree analysis (FTA) with human factors analysis and classification system (HFACS) model to analyze multiple hazardous chemical storage accidents. Based on this approach, we further explored the relationship between human factors in hazardous chemical storage accidents using the chi-square test and odds ratio analysis. According to the above methods, this paper realized the quantitative analysis of human factors.
It should be noted that the human factors studied in this paper not only refer to individual behaviors that directly lead to an accident, but also include other organizational factors such as organizational supervision and resource management, because individual people are not isolated and act as members of an organization. As a result, individual behavior is affected by other people, technology and the organization, and these factors restrict and influence each other. Therefore, the study of human factors should consider individual factors as well as organizational factors related to human behaviors.
Currently, certain human factors and proposed relevant models are available, including the software–hardware–environment–liveware (SHEL) model [
4], Swiss cheese model [
5] and the HFACS model [
6]. Among these, the HFACS model has been widely recognized and adopted in many industries. Dekker noted that the HFACS model is the most powerful tool for human factors analysis of various accidents [
7]. In the field of aviation, Shappell analyzed the data from 1020 aviation accidents in the United States and found that the majority of accidents were caused by the aircrew and environment, and the number of accidents related to supervision and organizational reasons was significantly reduced [
8]. Daramola used the HFACS model to analyze aviation accidents in Nigeria and concluded that the most common causes of accidents were skill-based errors, the physical environment and inadequate supervision. Supervision violation to crew resource management to decision errors was considered the most likely path to accidents [
9]. Michal et al. used the accident analysis method combining HFACS with a systems-theory accident model and processes (STAMP) to analyze the Überlingen air accident and confirmed the feasibility of the STAMP-HFACS analysis method [
10]. Rashid et al. proposed the Human Factors Analysis and Classification System-Maintenance Extension (HFACS-ME) model for helicopter maintenance accidents and statistically analyzed 58 helicopter maintenance accidents to study the survival rate of helicopter maintenance accidents and the distribution of accident severity [
11].
The application of HFACS in accident research also includes coal mining, maritime, medical, railway, chemical and other industries. For example, Patterson and Shappell used the HFACS-Mining Industry (HFACS-MI) model to analyze 508 coal mine accidents in Queensland and concluded that skill-based errors are the most common unsafe behavior, with no significant difference between different types of mines [
12]. Chauvin et al. analyzed the human factors and organizational factors of ship collision accidents in Britain and Canada using the improved HFACS model [
13]. The analysis showed that most collision accidents were caused by decision errors. Baysari et al. analyzed railway accidents in Australia using the HFACS and Technique for the Retrospective and Predictive Analysis of Cognitive Errors (TRACEr) methods and suggested the effectiveness of the two methods [
14]. However, each tool seems to ignore certain important factors related to the occurrence of errors. Cohen et al. used HFACS-Healthcare to identify systemic vulnerabilities during surgery [
15]. Hale et al. used the HFACS model to analyze 26 fatal building accidents and found deficiencies in planning and risk assessment, hardware design, purchase and installation, and contracting strategies [
16]. In the chemical industry, Gong and Fan analyzed the “11·13” explosion accident at the PETROCHINA Jilin petrochemical biphenyl factory using HFACS and classified the human factors that led to the accident, confirming the usefulness and feasibility of the HFACS for accident analysis in the chemical industry [
17]. Zhou et al. improved the items of HFACS and used the improved HFACS to analyze the “8·12” Tianjin Binhai New Area explosion accident [
18]. That research showed that the interaction between different levels of human factors in the Ruihai company led to the accident, and the accident investigation report displayed limitations in the identification of human factors and guidance for similar accident prevention. A review of the main relevant information of the HFACS is shown in
Table 1.
The HFACS model addresses the defects of the Reason model and gives a specific definition of the loopholes in each layer of the Reason model, which is more conducive to the study of the classification and mechanism of human factors. However, there are many reasons that exist for hazardous chemical storage accidents [
2,
19,
20], and qualitative analysis of the causes of hazardous chemical storage accidents alone is not sufficient to ensure the effectiveness of the analysis results. Therefore, it is necessary to use other methods to quantitatively analyze the cause of an accident. For this reason, this paper collected information from 42 hazardous chemical storage accidents. The collected accident data were summarized, including the date of accident, enterprises, type of accident and number of deaths. The main sources of accident data include the national and local emergency management departments at all levels, the websites of local governments, the official website of the China Chemical Safety Association, and the chemical registration center of the Ministry of Emergency Management of the People’s Republic of China. Then, according to the characteristics of hazardous chemical storage accidents, the HFACS model was modified to make it more suitable for the analysis of hazardous chemical storage accidents. Second, the improved HFACS model was applied to accident analysis. In this way, the causes and high-frequency human factors of accidents under the HFACS framework were obtained. Finally, a chi-square test and odds ratio analysis were used to test the significance and relevance between the four levels of factors under the improved HFACS framework.
4. Using χ2/OR to Analyze the Relevance of Human Factors in Hazardous Chemical Storage Accidents
To conduct a quantitative study of human factors in hazardous chemical storage accidents, the relevance among the factors in 42 accident cases was further studied based on the HFACS analysis. In this paper, a chi-square test and odds ratio analysis method were used to test the significance and relevance of four level factors in the improved HFACS framework, so as to realize the key step of human factor quantitative analysis.
Taking the calculation of the relevance between Communication and Coordination and Decision Errors in the improved HFACS model as an example, we calculated the
χ2 value and
OR value. Original hypothesis (H0): there is no significant causal relationship between Communication and Coordination and Decision Errors. Alternative hypothesis (H1): there is a significant causal relationship between Communication and Coordination and Decision Errors.
Table 5 shows the statistical results of the frequency of occurrence in accidents. The
χ2 value and
OR value were calculated using Equations (1) and (2).
By calculation, χ2 = 4.582 and OR = 4.000 > 1, and by df = 1, we can obtain p = 0.032 < 0.05. Therefore, we rejected H0 and accepted H1. This means that there was a significant causal relationship between Communication and Coordination and Decision Errors. In addition, the OR value greater than 1 indicated that the occurrence of Communication and Coordination can increase the possibility of Decision Errors.
We used the above method to analyze the relevance among the four levels of factors in the improved HFACS model. We screened the causal relationships between different levels of factors that satisfy
p < 0.05 and
OR > 1, and eliminated the causal relationships of human factors that did not meet the conditions, e.g., Resource Management and Planned Inappropriate Operations, Organizational Climate and Failure to Correct Problem, Inadequate Supervision and Technological Environment, Failure to Correct Problem and Physical Environment, Personal Readiness and Perceptual Errors, etc. Thus, we obtained the results shown in
Table 6. According to the inspection results, the causal relationship diagram of human factors in hazardous chemical storage accidents was obtained, as shown in
Figure 2.
5. Results Analysis
According to the four levels of the improved HFACS framework (including Organizational Influences, Unsafe Supervision, Preconditions for Unsafe Acts, Unsafe Acts) and the results of the χ2 test and OR analysis, the results of quantitative calculation were analyzed as follows.
5.1. Defect of Organizational Influences
Organizational influences include Resource Management, Organizational Climate and Organizational Process. From
Table 5, inadequate Resource Management has the greatest impact on Inadequate Supervision. In other words, if resource management of a hazardous chemical storage enterprise is poor, the possibility of insufficient supervision will increase to 20.7 times (
OR = 20.667). Resource management loopholes are primarily reflected in human resources, equipment and facilities resources, funds, and other aspects. The specific performance issues are unreasonable personnel allocation, lack of a qualification examination system for special operation personnel, poor quality of safety management personnel, insufficient equipment and facilities, or quality defects. The poor organizational climate also leads to the occurrence of supervisory violations and inadequate supervision. A poor organizational climate will increase the possibility of Inadequate Supervision to 4.5 times (
OR = 4.455) and the possibility of Supervisory Violations to 7.8 times (
OR = 7.800). Poor organizational climate includes insufficient safety investment, insufficient risk management policies, “focusing on efficiency, ignoring safety”, and poor safety culture.
In addition, Organizational Process loopholes have a significant impact on Planned Inappropriate Operations and Supervisory Violations in Unsafe Supervision. Organizational process loopholes will increase the probability of Planned Inappropriate Operations to 6.2 times (OR = 6.231) and the probability of Supervisory Violations to 4.8 times (OR = 4.848). Organizational process loopholes are mainly reflected in systems, procedures, production supervision and other aspects. Examples of this include where the enterprise has not formulated a specific safety management system or the system is incomplete, the regulatory system has loopholes, the organization and management of the site are disordered, and relevant operation instructions are lacking.
5.2. Unsafe Supervision
Unsafe Supervision includes Inadequate Supervision, Planned Inappropriate Operations, Failure to Correct Problem and Supervisory Violations. It can be observed from
Table 5 that Inadequate Supervision, Planned Inappropriate Operations, and Failure to Correct Problem in the Unsafe Supervision level have a significant impact on Technological Environment, Physical/Mental Limitations, Communication and Coordination, and Personal Readiness in the Preconditions for Unsafe Acts level. In particular, Inadequate Supervision can directly cause the emergence of three unsafe factors in the next level. If a hazardous chemical enterprise suffers from inadequate supervision, it will increase the possibility of worker physical or mental limitations to 5.1 times (
OR = 5.143), the possibility of poor communication and coordination to 8.8 times (
OR = 8.800), and the possibility of insufficient personal readiness to 14.3 times (
OR = 14.286). Inadequate supervision refers primarily to situations in which the manager fails to offer sufficient training and guidance on hazardous chemical knowledge to the employees in their daily work. It results in employees lack of clarity relative to the physical and chemical properties and dangers of related hazardous chemicals or lack of supervision in fire operation sites. This situation further leads to poor information exchange and communication between the upper and lower levels of employees.
Planned Inappropriate Operations has a significant impact on the Technological Environment and Communication and Coordination in the Preconditions for Unsafe Acts level. Planned inappropriate operations will increase the possibility of a poor technological environment to 4.7 times (OR = 4.722) and poor communication and coordination to 4.3 times (OR = 4.333). Planned inappropriate operations is manifested as improper collocation between team members or authorization of unqualified team members for work, resulting in insufficient cooperation and communication among team members. In addition, planned inappropriate operations also refers to the improper allocation of resources, and thus it might lead to differences in equipment allocation among different teams, resulting in the risk of a poor technological environment. Failure to correct the problem will increase the probability of technological environmental problems to 4.3 times (OR = 4.275). Failure to correct the problem refers to the failure of the supervisor to find problems or correct the problems in time in the hazardous chemical storage process, resulting in the continuous existence of hazards. Poor technological environment refers to equipment and facility failures, lack of protective devices, lack of electronic monitoring facilities, unreasonable control design, etc. Therefore, it is easy to increase the risk of a poor technological environment if problems are not found or not solved in time, or hazards are not investigated adequately.
5.3. Preconditions for Unsafe Acts
Preconditions for Unsafe Acts includes the seven aspects of Physical Environment, Technological Environment, Adverse Mental States, Adverse Physiological States, Physical/Mental Limitations, Communication and Coordination, Personal Readiness, among others. However, for the 42 hazardous chemical storage accidents collected, only Technological Environment, Physical/Mental Limitations, Communication and Coordination, and Personal Readiness have a significant impact on Unsafe Acts. Among these, the lack of personal readiness is one of the main reasons for unsafe acts, especially for skill-based errors and violations. The lack of employee personal readiness can increase the probability of skill-based errors to four times (OR = 4.000) and increase the probability of violations to 13 times (OR = 13.000). Personal readiness refers to a lack of knowledge and skills for the related hazardous chemicals or a lack of physical strength and energy of the front-line workers before work. The main manifestations are insufficient knowledge about hazardous chemicals, a lack of mastery of skills required by the position, failure to wear personal protective equipment (PPE), insufficient rest, etc. Therefore, the lack of personal readiness will inevitably increase the possibility of skill-based errors and violations.
Physical/Mental Limitations and Communication and Coordination have a significant impact on the occurrence of Decision Errors. Physical or mental limitations can increase the probability of decision errors to 4.1 times (OR = 4.083). Communication and coordination can increase the probability of decision errors to 4 times (OR = 4.000). Physical/mental limitation refers to a lack of experience and the ability of employees to function in complex situations. Communication and coordination refers to insufficient cooperation among team members and lack of information exchange between superiors and subordinates. If these two factors are defective, it will inevitably lead to decision errors under different situations. Decision errors mainly refer to emergency judgment errors, emergency response errors, improper selection, problem handling errors, etc. In addition, the technological environment also has a significant impact on skill-based errors (OR = 9.000). A poor technological environment may lead workers to be unfamiliar with important equipment and ignore operational details, resulting in skill-based errors.
5.4. Unsafe Acts
Unsafe Acts includes Skill-based Errors, Decision Errors, Perceptual Errors and Violations. According to the statistics of unsafe acts in 42 hazardous chemicals storage accidents, violations by front-line workers account for the largest proportion, reaching 85.714%, followed by decision errors and skill-based errors, accounting for 52.381% and 47.619% respectively, and finally perceptual errors, accounting for 7.143%. Violations mainly refer to the violation of the existing rules and various safety operating procedures and risky operations of front-line workers.
Decision errors refer to the errors caused by improper employee estimation of the situation, including three main types of errors in emergency situations: process errors, selection errors and problem-solving errors. Skill-based error refers to mistakes in skill-related behaviors of employees, mainly including poor operation technology, blind operation blind, improper use of PPE, etc. Perceptual errors are manifested by individual cognition and actual conditions such as visual errors, information understanding errors, wrong directions, etc.
According to the above analysis, the main factors leading to unsafe acts of employees are personal readiness, communication and coordination, and technological environment, whereas unsafe acts are primarily reflected in violations and decision errors. Therefore, managers should enhance training on professional knowledge and skills for front-line operators, improve the professional development of employees, and assure good job preparation to reduce the possibility of unsafe behaviors.