Safety decision analysis of collapse accident based on “ accident tree – analytic hierarchy process

: To reduce the hazards of collapse accidents in the construction process and to ensure that the lives, health, and property of construction workers are protected, this study used the theory of safety system engineering to identify the hazards of collapse accidents, analyze the hazards, predict the consequences, evaluate the systemic risks, and evaluate the e ﬀ ects and improve them. At the same time, the risk factors of collapse were evaluated qualitatively and quantitatively by using the analysis methods of fault tree analysis (FTA) and the analytic hierarchy process (AHP). Finally, according to the evaluation results, the main factors causing collapse accidents were found; this provided a reliable and practical basis for the prevention of collapse accidents. Then, according to the decisive factors, corresponding measures were taken in advance to achieve the aim of preventing and controlling collapse accidents. The results show that equipment maintenance, material inspection, and construction site safety management play an important role in preventing collapse accidents.


Introduction
In recent years, with the development of the global economy and the progress of science and technology, the construction industry has developed remarkably, and the scale of all types of engineering construction increases year by year; at the same time, the level of urban construction is accelerating. However, with the development of new materials and new technology, the traditional production management model for construction safety is currently facing great challenges. Although the traditional "empirical" safety management method is constantly developing, the overall level is still relatively backward, especially in the management of construction projects. Most managers still adopt a more extensive management model, and the construction quality and construction safety are not very strict [1]. Wang [2], Wu [3], and Shan [4] analyzed the deficiencies in the construction safety management system and put forward relevant suggestions for construction safety management, providing references for the relevant practitioners. Zhan [5], Zhu [6], and Tam et al. [7] also put forward similar views. Many scholars have attached great importance to building safety and have emphasized the important role of innovative building safety management modes in preventing safety accidents. Zhang et al. [8] argue that sensor technology, an effective method of information collection, identification, and processing, provides a new means of promoting building safety management. Lee et al. [9] introduced the development of a safety management real-time locating system (RTLS) based on radio frequency identification (RFID) and conducted a case study to verify the localization performance of the RTLS. The results showed that the RTLS had great potential for the improvement of construction site safety management. The traditional safety management model can qualitatively, but not quantitatively, determine whether a site is safe or not. It can only reveal some superficially dominant causes of accidents [10] and cannot find a clear target value by which to prevent accidents. Therefore, improving the efficacy of construction safety management can ensure the safety and stability of the construction process to the greatest extent and can play a highly significant role in preventing frequent collapse accidents in the construction process.
A building collapse accident is a dangerous accident that often occurs in the process of building construction, and it is one which may occur at any stage of the construction process. Once the collapse accident occurs, it will not only cause huge economic losses but also cause heavy casualties and seriously affect the stability and development of the society. Marzaleh and Peyravi [11] pointed out that the number of building collapses was still increasing worldwide in the twenty-first century, especially in developing countries. By analyzing the 2017 Plasco Building collapse accident, they emphasized the severity of the consequences of building collapses and discussed the lessons learned. Shao et al. [12] and Cheng et al. [13] used data mining methods to analyze the causes of building collapse accidents. In addition, other scholars established the construction accident causation system (CACS) model analysis method [14] and the new Austrian tunnel analysis method (NATM) [15]. Peng et al. [16] studied the steel-reinforced concrete (SRC) structure collapse event of a new factory building in Taiwan and found that the failure mechanism of the reinforced concrete structure during construction was caused by system buckling. Mathebula and Smallwood [17] found that quality problems, design defects, and lack of safety inspections were important causes of the collapse of religious buildings. Many existing studies have analyzed the causes and different aspects of building collapse and have pointed out the direction that construction units should take to prevent and control the occurrence of building safety accidents. At the same time, by reading the literature, it is not difficult to discover that among all construction safety accidents, falling from scaffolding is one of the most common. The occurrence of scaffold and formwork collapse has been increasing in recent years. It is necessary to clarify the causes of scaffold and formwork collapse accidents and to take reasonable and effective measures to prevent them and ensure construction safety. Through systematic barrier management, there is great potential for accident prevention in the construction industry [18]. Dogan et al. [19] first classified the suspended scaffold and then determined and analyzed the causes of collapse accidents. Park and Kim [20] used analytic hierarchy process (AHP) technology to rank the importance of formwork collapse factors. Li and Wan [21] analyzed the direct causes of support collapse from the technical point of view, considering the forms of support failure and different load effects in order to formulate preventive measures. Jahangiri et al. [22] used the method of combining an adaptive neural network fuzzy inference system (ANFIS) with a safety checklist to identify risk factors and predict the risk of scaffold collapse on construction sites. Abas et al. [23] used the average index and the relative importance index (RII) to analyze the main causes of scaffolding accidents on construction sites. The study found that the main causes of scaffolding accidents were neglect of safety rules, improper inspection, and improper foundation stability. Rebelo et al. [24] developed the scaffold use risk assessment model (SURAM) to assess the risk level of different construction stages and different types of work. The result was based on a set of historical accident data. The developed SURAM seemed to help in the prediction of high-risk construction activities and, consequently, the prevention of accidents. A large number of existing studies have recognized and fully discussed the causes of the two most common collapse accidents and have provided a variety of countermeasures that construction enterprises may employ to guide the direction of their construction.
Nowadays, it is becoming more and more important to clarify the causes of building collapse accidents and to ensure that the construction is safe. There are many construction units with a good safety reputation that have adopted a zero-accident vision [25]. Many scholars have studied construction safety by using the fault tree method. Chai et al. [26] used the fault tree analysis (FTA) method to analyze the causes of surface collapse accidents in subway construction; the analysis provided a theoretical basis for surface construction accident prevention and safety management on construction units. Bakeli and Alaoui Hafidi [27], Zermane et al. [28], and Purohit et al. [29] relied on FTAa type of risk assessment method that helps in identifying and analyzing hazards to minimize the probability of accidents. Andriani et al. [30] used similar methods to identify common safety hazards on the construction sites of integrated agricultural enterprises. Chang and Zhao [31] established a construction safety accident tree model, which was combined with the Cobb-Douglas function, and used a nonlinear programming method to derive solutions. The results showed that investment in daily safety management, safety education, and construction health measures could effectively reduce the probability of accidents. Ji et al. [32] studied the main factors affecting the hoisting safety of prefabricated buildings and established a safety accident tree. The important hazard sources affecting hoisting safety were found, and targeted control measures and suggestions were put forward to improve the hoisting environment of the construction site. Geng [33] and Zhou and Feng [34], using the FTA method, systematically analyzed high-altitude falling accidents. By combining the steps of the fault tree construction, the fault tree was simplified using Boolean algebra. Finally, the minimum cut set and the minimum path set were used for quantitative analysis, and targeted safety prevention measures were proposed. Han [35] established an accident map of high-altitude falls by using the FTA method and analyzed the process and risk degree of falling accidents in high-altitude operations; this work significantly ensured the life safety of high-altitude operators. However, using the FTA method alone can only vaguely determine the degree of influence of each basic event, which cannot be quantitatively and accurately analyzed. There are some limitations. The AHP comprises a combination of qualitative and quantitative analysis; it is a systematic, hierarchical analysis method and is one of the mathematical tools of system analysis. Zhang [36], Chen et al. [37], and Hyun et al. [38] put forward safety measures to prevent accidents by using the principle of the AHP to order the importance of accidents.
Scholars have used various methods to evaluate building collapse accidents; however, with regard to preventing building collapse accidents more effectively, the research on the use of a combination of the fault tree and the AHP is limited. In this article, the FTA method is used to determine the key factors affecting the collapse accident index layer. Based on the qualitative analysis of the fault tree, the quantitative analysis method of the AHP is used to quantitatively analyze the collapse accident in the construction process in order to further improve the accuracy and effectiveness of the analysis and to formulate the corresponding preventive measures.

An overall analysis of collapse accidents
In a construction project, a collapse accident is mainly understood as an accident caused by the collapse of structures, buildings, and stacked items, resulting in casualties; the reasons for the collapse are environmental or manmade. The collapse accident is mainly divided into the collapse of the formwork and scaffold, the collapse of the foundation pit trench, and other accidents [39]. Due to the particularity of building construction, the collapse conditions of the two main types are quite different. This study will analyze them separately. Through communication with construction managers and workers who have rich experience in collapse accidents, it was concluded that the stages of the formwork and scaffold collapse accidents are different and that the state of the people and the things in the process of foundation pit collapse is different. In order to better analyze the factors of collapse accidents, formwork and scaffolding were divided according to the construction stage, and the foundation pit collapse was analyzed according to the classification of the people and objects.

Analysis of formwork and scaffold collapse
Through the analysis of the safety accident notifications in the past 10 years, it can be seen that the formwork and scaffold collapse accidents generally occurred in the formwork support stage, the use stage, and the demolding and dismantling stage. 1) Formwork support stage: the causes of the collapse at this stage are no construction plan or no formwork support according to the construction plan, unlicensed personnel for formwork support, no approval and disclosure, and excessive personnel or local concentrated accumulation of materials in the process of formwork support. 2) Use stage: the collapse at this stage is mainly caused by unqualified materials, structural errors, or improper formwork technology. The structural problems generally include insufficient bearing capacity due to the large three-dimensional size of the frame, insufficient scissors, no sweeping rod, insufficient support points, excessive spacing between vertical rods, and unstable connection points. 3) Demolding and dismantling stage: the collapse at this stage is mainly due to concrete that has not yet reached the design strength, demolding and dismantling sequence errors, and excessive local stress or excessive load on the frame.

Analysis of foundation pit earthwork collapse 2.2.1 Unsafe human behaviors
"Three violations" of behavior include illegal command, illegal operation, and violation of safety discipline. Violations include the following: a safe slope is not set up according to the regulations; the requirements for slope support are not adhered to; mechanical operators are not operated according to regulations; the foundation pit groove slope or pile hole surrounding it does not have heavy objects stacked according to the regulations; management confusion; failure to take safety measures; and lack of personnel to monitor the slope and the surrounding environment.

Unsafe conditions of objects
The unsafe conditions include poor soil quality, soft soil caused by the penetration of groundwater or surface water, no effective drainage measures, no derailment fences, no warning signs during the demolition, and environmental deterioration after the rainy season and winter thawing.

Construction of accident tree model
Based on the accident, the fault tree gradually figures out the accident's causes. Then, the logical relationship between the various causes is represented by an arborescence composed of logical gate symbols. The qualitative analysis of the fault tree refers to the calculation of the structural importance according to the minimal cut sets and minimum path sets of the accident tree. Thus, the influence of the events is obtained, and the key causes are found and used to control the accidents.

Accident tree of formwork and scaffold collapse
According to the analysis of the causes of many collapse accidents, it can be seen that the collapse of the formwork and scaffold [40] may occur at the time of erection, use, and demolition. The basic events mainly affecting the collapse include 24 events, such as not setting up according to the construction plan, excessive load, and unqualified materials. The meaning of each symbol for the formwork and scaffold collapse accident is shown in Table 1. An FTA model was established for building collapse accidents, as shown in Figure 1.
To determine the basic events that cause collapse accidents, the fault tree was decomposed by the Boolean algebra method to determine the minimum cut set of the fault tree: From the above formula, it can be concluded that there are 24 minimum cut sets of the formwork and scaffold collapse accident tree; that is, there are 24 basic events that cause the collapse of the formwork and scaffold. The formula also shows that there are 24 possible ways for the system to cause collapse accidents: After determining the minimum cut set of the fault tree, the structural importance analysis, which is the other part of the qualitative analysis of the fault tree, was Collapse when supporting X 9 Insufficient bearing capacity of wooden supports M 2 Collapse in use X 10 The three-dimensional dimension of the scaffold is too large M 3 Collapse in dismantling X 11 Inadequate diagonal bridging M 4 False construction X 12 Inadequate bottom horizontal tube M 5 Heavy load X 13 No plate M 6 Unqualified material X 14 Insufficient braced point M 7 Structural error X 15 Overlong hanging height of the top pole M 8 Inferior process X 16 False pouring order M 9 False order of dismantling X 17 Pump line vibration shaking the support frame M 10 Weak connection point X 18 Pole joint in the same horizontal plane X 1 Weak concrete X 19 No installment in the horizontal connection rod X 2 Overloading due to excessive constructors X 20 Insufficient pre-tightening torque of the fastener X 3 Construction by unlicensed personnel X 21 The pole is not vertical X 4 Lax regulation X 22 Adopting lap joint in the vertical pole X 5 Disapproval and disclosure X 23 A double fastener is not used for the highest node of a vertical bar X 6 Overloading by centralized stacking of materials X 24 Excessive spacing between vertical poles X 7 Adopting inferior steel pipes and fasteners conducted. This analysis involves the calculation of the contribution of each basic event to the top event when the probability of occurrence of each basic event is the same; thus, it belongs to the qualitative importance analysis. In the minimal cut set of the fault tree, the cut set of only one basic event is defined as a first-order minimal cut set; the cut set of two basic events is defined as a secondorder minimal cut set; etc. Each minimum cut set represents a possibility of the occurrence of the top event and thus provides a basis for accident investigation and accident prevention. According to the minimum cut set of the fault tree calculated above, the basic event importance of the firstorder minimum cut set is the largest, and the others can be calculated according to the approximate discriminant: where ( ) I X ϕ i is the importance coefficient of the basic event structure X i ; K j is the jth minimal cut set; and n j is the number of basic events in the minimal cut set K j of the basic events X i .
The following is the structural importance order of each basic event after calculation: The structural importance of each of the basic events X 7 to X 24 is equal and is the highest because each of their minimum cut sets has only one basic event. As their structural importance is the highest, it means that they have the greatest impact on top events, which is also where there are vulnerabilities in the security system. The corresponding measures derived from these aspects can effectively control the occurrence of accidents and improve the safety of the system.

Foundation pit earthwork collapse
The meaning of each symbol for the fault tree of the foundation pit soil collapse is shown in Table 2.
Similarly, using the Boolean algebra method to simplify the fault tree (Figure 2), a total of 40 minimum cut sets of the foundation pit earthwork collapse fault tree could be obtained, indicating that there are 40 possible ways for the system to cause accidents.
After calculation, the order of importance of each basic event structure was By analyzing the importance of the structure, the order of structural importance of the various basic events was revealed. The structural importance of not monitoring the slope and surrounding environment and not supporting the slope according to the regulations makes them the most important basic events; they have the greatest impact on the occurrence of the top event and are more likely to lead to the occurrence of the top event. To improve the safety and reliability of the foundation pit trench earthwork system, all measures must be taken to monitor the slope and the surrounding environment before construction, and the slope support must be carried out in strict accordance with the regulations to prevent the occurrence of collapse accidents to the greatest extent.

Construction of the AHP model 4.1 Quantitative analysis of the AHP
The AHP is a qualitative and quantitative decision-making method of multi-objective analysis proposed by Seaty [41]. Unsafe conditions of objects X 6 The slope is not set according to the rules M 3 Three violations X 7 Poor soil M 4 Disordered management X 8 Penetration of surface or groundwater X 1 No slope is set X 9 No effective drainage measures X 2 False operation X 10 No derailment fences and warning signs X 3 Heavy objects are stacked in pits or pile holes X 11 Environmental deterioration after winter and rainy seasons Figure 2: Accident tree of the foundation pit earthwork collapse.
The main principle is as follows. First, the relevant factors are classified into different levels, including the target layer, the criterion layer, and the indicator layer. Afterward, the relative importance of each layer is determined. Finally, the relative importance of each measure in the indicator layer is calculated, which means that the measures are ordered. The model of hierarchical analysis consists of a target layer U, several criteria layers U i , and an indicator layer U ij . Based on the accident tree analysis mentioned above, the target layer is to prevent collapse accidents during construction. The criteria layer is obtained by neutralizing the intermediate event description of the accident tree. For example, unqualified materials and structural errors are attributed to the lack of regular maintenance and inspection of equipment. Similarly, the indicator layer is to neutralize the basic event description of the accident tree. For instance, the qualification rate of the materials is related to the inferior steel pipe fasteners, unqualified plates, and insufficient bearing capacity of timber struts.
Based on the above fault tree and analysis, a comprehensive index system of the influencing factors of collapse accidents can be constructed, as shown in Figure 3; the analytic hierarchy model is shown in Table 3.

Constructing a judgment matrix, calculating the weight of each index, and checking consistency
To construct the judgment matrix, seven experts were invited to independently assign values to each indicator. Then, a discussion was held to determine the average; finally, the judgment matrix was obtained. The following is an example of the U − U i judgment matrix (Table 4).  The U i − U ij judgment matrix after the calculation is as follows: The results show that the maintenance and inspection of equipment have the highest weight among the various indexes, followed by supervision and management. Safety management measures should be formulated with these two as the focus, followed by safety education and protection measures and environmental status inspection and monitoring.

Analysis and countermeasures
There are huge safety hazards during construction, and collapse is one of the most dangerous accidents. Therefore, measures must be taken to prevent and control the dangers. Based on the accident tree-AHP analysis, collapse accidents are mainly affected by two causes. The first is caused by objects because the equipment materials have not been maintained and inspected as required. The second is caused by human beings because there is no clear supervision and management responsibility system.
In this case, based on the accident tree-AHP analysis, the present study analyzed the basic events of the accident tree and the criteria layer and indicator layer of the analytic hierarchy. Then, the following security decisions were proposed to prevent collapse accidents.

Cause by objectsregular maintenance and inspection of equipment
The following measures can be taken to prevent collapses caused by equipment on the construction site. First, the   equipment should be maintained and inspected, and the machines with problems or overloaded operations should be strictly prohibited. The mechanical equipment must be checked before use, and the examination must be recorded. In addition, the construction equipment should be used according to the specifications, with the prohibition of invalid mechanical equipment. Moreover, the effectiveness and completion of tower crane safety devices should be guaranteed, based on regular inspection, repair, and maintenance. The construction elevator safety devices should be effective and complete to ensure that there is no overloaded use. In brief, all the safety devices should be tested every day before utilization.

Cause by human beingsestablishment of supervision and management responsibility system
A safety management organization should be established; it should consist of a project manager, site manager, project deputy chief engineer, safety director, full-time security officer, and subcontractor. The project manager is the main person responsible for safe production. The safety director takes charge of the supervision and management of the supervision unit and is answerable to the owner. The main contractor is in charge of the project, and the subcontractor is answerable to the main contractor. In addition, the main contractor takes responsibility for the subcontractors designated by the owner with regard to the pile foundation, earthwork, ejector anchor slope protection, integrated wiring, fire alarm equipment, installation, high-voltage power supply, and elevators. In a word, everyone must guarantee safe production. Responsibility and management systems are required for safe production. The security duty of all personnel should be clarified to set up the management system for different inspections. A safe disclosure system is needed. The engineering and technical personnel should ensure the safety of the operation team before construction [42], which must be implemented by the on-site project department, with a signature for verification. In addition, the personnel is required to go into the site and conduct the supervision, management, inspection, and guidance.

Conclusions
(i) Building hazard sources can be identified, and the minimum cut set of the fault tree can be determined by constructing a fault tree model of the formwork, scaffold, and foundation trench collapse accidents. Once the basic event of each minimum cut set occurs, it will lead to the top accident, which can provide the basis for the construction unit to prevent the occurrence of construction safety accidents.
(ii) The degree of the influence of each basic event on the system risk can be determined through the analysis and sorting of the minimum cut set structure importance. The event was found that had the greatest impact on the top event, that is, the minimum cut set with the smallest number of basic events; this was the critical control point in the system. Strengthening the control of such events before construction can effectively ensure the safety and reliability of the system.
(iii) The risk assessment of the system was carried out by using the AHP to determine the optimal strategy for preventing collapse accidents. The results showed that the maintenance of equipment and material inspection, qualified use, and site construction supervision and management have an important impact on the prevention of collapse, and attention should be paid to this by construction units.
(iv) The risk assessment analysis method, combining FTA and the AHP, provides a new way of thinking for the construction industry-related personnel to carry out safe construction work and to ensure the life and health of the practitioners.
Funding information: The authors state that no funding is involved.