Efficiency of risk management for tunnel security of megaprojects construction in China based on system dynamics

ABSTRACT

There have been repeated security incidents in megaproject tunnels, construction security issues are particularly serious, and the management of security risks is receiving increasing attention. This study was conducted to improve the efficiency of tunnel security risk management, the security risk management process was adopted as an entry point, based on the Observe-Orient-Decision-Action (OODA) model, the influencing factors index system for the efficiency of security risk management in megaproject tunnels from four dimensions: observe, orient, decision and action. The AHP-EWM method is applied to assign index weights to model the system dynamics of security risk management efficiency. The Vensim software was used to simulate and analyse the effect of different indices on the efficiency of security risk management under different conditions, which provides a reference for improving the efficiency of security risk management. The results of the study show that building a comprehensive and uninterrupted risk information awareness network, real-time adaptation of risk control strategies based on the complexity of risks, and shortening risk management delays can effectively increase the efficiency of dynamic security risk management. This provides a theoretical and practical reference for effectively improving the efficiency of risk management for tunnel security in megaproject construction.

KEYWORDS:

• management efficiency
• tunnel security risk
• megaprojects
• system dynamics
• AHP-EWM

1. Introduction

China has the largest number of tunnels, the fastest development rate, and the most complex geological and structural forms in the world (Lu and Cai 2019). While continuously accelerating the modernization of the transportation system, the development trend of large scale, complex geology and high difficulty of tunnel construction has led to an increasingly serious security situation (Jaafari 2003). The length, depth and cross-sectional size of tunnels are increasing, and more and more tunnels are built at high altitude, strong sand, high temperature and cold environment or high stress, strong karst area, resulting in complex and variable geological conditions, which are difficult to predict accurately (Xun, Zhang, and Yuan 2022). Adverse geology such as water and mud surges, large deformations in soft rock, rock bursts, gas and high ground temperatures lead to variable and uncertain security risks in tunnel construction. As a result, the environmental conditions faced by workers during tunnel construction are complex (Ma, Bin Luo, and Chen 2013). Given the complexity of tunnelling, there is a relatively high probability of security accidents during the construction phase of tunnelling, resulting in greater economic and human losses and a large negative impact on society.

A summary of the causes of tunnel security accidents shows that for tunnel construction security risks come from two main sources. Firstly, the complexity of the geological environment faced by the tunnel during the actual construction process (Chatzimichailidou and Dokas 2016; Reza et al. 2019). Secondly, human or subjective causes, mainly due to the increasing size of the tunnels and the management deficiencies of the construction units, which do not pay attention to security management in the actual construction management process (Liu et al. 2022). Although many scientific and reasonable security contingency plans have been proposed, they are mainly designed to deal with unexpected events in tunnel construction, and do not provide real-time monitoring and prevention of daily security risks, so the usual tunnel construction security risk management link is relatively weak (Liu et al. 2018; Sina et al. 2022). As a result, the increasingly challenging tunnel security situation and management weaknesses place greater demands on the efficiency of risk management.

However, tunnel construction is affected by multiple factors, both in the megaproject itself and in the external environment, and the efficiency of security risk management is fraught with uncertainties. A comprehensive, systematic and in-depth analysis and study of the efficiency of security risk management is an important practical guide to improve the level of security risk management in megaproject tunnels. Tunnel security risk management is a dynamic evolutionary process. System dynamics (SD), which starts from the microstructure of the system and simulates and analyses the dynamic behaviour of the system, provides methodological support for studying dynamic, complex multivariate nonlinear systems. Therefore, this study constructs an indexing system of influence factors for the efficiency of tunnel security risk management. Based on SD, an efficiency model for tunnel security risk management model is established, which provides a quantitative and scientific decision-making basis for studying how to improve the efficiency of tunnel security risk management for megaprojects.

The remainder of the paper is organized as follows. Section 2 reviews the relevant literature. Section 3 identifies the factors that affect the efficiency of security risk management. Section 4 constructs the SD model for the efficiency of tunnel security risk management. Section 5 performs empirical simulations and analysis. Section 6 provides a discussion based on the model results. Finally, the concluding remarks are provided in Section 7.

2. Literature review

2.1. Tunnel security risk identification

Security risk identification is a prerequisite for risk management (Cha and Shin 2012). On the one hand, risk identification can be judged by perception and historical experience. On the other hand, it can also be analysed, summarized and organized by various objective data and records of risk incidents, as well as necessary expert visits, in order to find out various obvious and potential risks and their loss patterns (Konstantinos and Konstantinos 2013). Nezarat, Sereshki, and Ataei (2015) summarized eight categories of tunnel construction security risks through geological studies and interviews with experienced technicians. Zhou et al. (2021) proposed a framework for automatic identification and classification of security risks in the metro design phase based on Building Information Modeling (BIM), which facilitates the digital transfer of security risk design information throughout the construction process. Lu et al. (2020) developed a multi-agent based Security Computing Experimental System (SCES) to provide comprehensive risk factor identification for shield tunnel projects. Forcael et al. (2018) identified the main risk factors that threaten the achievement of the tunnel construction project objectives and consider them to be inherent to the tunnel construction project. Chen et al. (2020) extracted incident chains from tunnel security incidents and designed network-level performance metrics and node-level critical security risk identification metrics. Zhang et al. (2016) integrated Building Information Modeling (BIM) and expert systems to propose a new tunnel security risk identification method to address the shortcomings of traditional identification methods. Ye et al. (2022) propose a digital twin-based multi-information intelligent warning and security management platform where multi-source security risk information of tunnel construction can be automatically collected and transmitted. Dong et al. (2018) developed a Knowledge-dynamics Integrated Map (KIM) that visualizes the identified tunnel construction security risks to form a knowledge flow for security risk management.

2.2. Tunnel security risk assessment

Security risk assessment is a quantitative measurement of the likely level of impact or loss from a risk event. Qiu et al. (2020) proposed a comprehensive assessment method for gas tunnel construction security risk based on fuzzy Bayesian network, which determines the results of risk assessment in terms of both risk probability and risk hazard. Li and Qu (2016) designed a two-segment search and constraint algorithm (BSSBA) to obtain the optimal solution for tunnel security facility selection. Han et al. (2018) used Monte Carlo simulation techniques to quantitatively estimate the potential level of security risk for circumferential cracks in pre-existing shield tunnels. Ntzeremes, Kirytopoulos, and Filiou (2020) proposed an evacuation simulation model to improve the efficiency of quantitative risk assessment, which can help security analysts to make more informed decisions about the security level of the tunnel. Hu et al. (2021) developed a tunnel construction security risk assessment model for LSDDTPs based on the material element expansion method and validated it with the East Lake Deep Buried Tunnel project in Wuhan, China as a case study. Guo and Zhang (2021) utilized D-S evidence theory to fuse information and determine tunnel construction security risk probabilities, and used evidence updating methods to capture the dynamic characteristics of risk states. Xun, Zhang, and Yuan (2022) proposed a security risk assessment method based on Building Information Modeling (BIM), Intuitionistic Fuzzy Set (IFS) theory and Dempster-Shafer (D-S) evidence theory, which can perform dynamic security risk assessment for cross-harbour tunnel construction projects under uncertainty, ambiguity, and conflict environments.

2.3. Tunnel security risk decision-making

Security risk decision is the choice of more than two courses of action under the action of multiple uncertainties, and because of the existence of uncertainties, the result of the implementation of the course of action is not predetermined the value of profit and loss (Eirik, Willy, and Ruben 2012). When security risk incidents occur, they require rapid response and decision making by tunnel construction and management personnel (Kinateder et al. 2014; Reza, Sina, and Zoheir 2019). Uncertainty is one of the sources of security risk in tunnelling projects. Uncertainty information not only affects the accuracy of security risk evaluation results, but also affects the reliability of the decision-making scheme (Xia et al. 2018). In order to solve the problem of cognitive uncertainty caused by insufficient use of information in decision making, Guo et al. (2022) proposed a continuous-time Bayesian network (CTBN) technology-based decision support method for subway tunnel construction security risks. Cai, Hu, and Ma (2021) argued that the information fusion approach based on extension theory (ET) performs an effective tool for uncertain information reasoning to address the ambiguity and uncertainty of security risk decisions. Zhang and Chen (2022) used TOPSIS to determine the ranking of various alternative tunnel security risk decision options to deal with multi-criteria decision problems under uncertainty.

2.4. Tunnel security risk control

Security risk control refers to the measures and methods adopted by risk managers to reduce the damage caused by security risk events (Konstantinos et al. 2010). Zhang et al. (2013) extended the security control process to the entire lifecycle of the model application that is prone to security risk events, including continuous control during construction and post-incident control. Bjelland et al. (2021) considered that the system theory approach is suitable to deal with the complexity of tunnel systems and facilitate the control of tunnel construction security risks. Li and Yuan (2016) developed a security risk control framework for shield tunnel construction to address and manage security risks in a systematic way. Lei et al. (2020) analysed the idea of security risk control for shield tunnel construction under existing railroads and formed a complete set of security risk control processes. Fang, Zhang, and Wong (2011) proposed a risk management approach to achieve control of tunnel construction security risks through process control of ground settlement and nearby existing structures. Li et al. (2017) established a sudden water surge mud risk evaluation of unsure measures model to effectively control the tunnel construction security risks caused by sudden water surge mud. Pan et al. (2019) quantitatively analysed the coupling degree between the factors in the tunnel construction security risk system based on the coupling degree theory in physics, so as to achieve the control of the whole security risk system.

3. Influencing factors of efficiency of security risk management

The efficiency of tunnel security risk management for megaprojects is the extent to which construction and management personnel can achieve the desired goals of risk prevention and control by performing risk management tasks during the tunnel construction process (Liu et al. 2021). However, in order to enhance the efficiency of risk management, based on the basic structure of observation-orient-decision-action of the OODA cycle theory, as shown in Figure 1, and combined with risk management-related approaches and relevant literature, an index system of influencing factors for the efficiency of security risk management in megaproject tunnels was constructed, as shown in Table 1.

Figure 1. OODA model.

Table 1. Influencing factors of efficiency of tunnel security risk management.

Primary indicators	Secondary indicators
Observe: Risk perception	A1: Ability to identify security risk
	A2: Security risk warning and communication level
	A3: Risk perception attitude and knowledge state
	A4: Ability to collect security risk information
Orient: Collaborative research and evaluation	B1: Ability to assess security risk posture
	B2: Cross-organizational collaborative security risk research and evaluation ability
	B3: Cross-departmental collaborative security risk research and evaluation ability
	B4: Cross-level collaborative security risk research and evaluation ability
Decision: Risk decision making	C1: Ability to assess security risk management programs
	C2: Degree of cross-organizational involvement in risk decision making
	C3: Degree of information and intelligent decision-making for security risks
	C4: The level of knowledge and competence for security risk decision making
Action: Risk control	D1: Degree of clarity of responsibility for security risks
	D2: Reasonableness of security risk control program selection
	D3: Ability to implement security risk control programs
	D4: Reasonable staffing for security risk control management

4. SD model of efficiency of tunnel security risk management

4.1. SD method and steps

The study of system dynamics focuses on the behavioural patterns and characteristics of the system. Based on the internal information feedback control of the system, computer techniques are used to seek the dynamic correlations between the system structure and behaviour patterns with the help of model simulation analysis. The system dynamics model consists mainly composed of system variables, including state variables, rate variables, auxiliary variables, and constants. The causal relations and feedback mechanisms between the variables are determined by constructing equations. The construction of the SD model is an important component to complete the analysis of the security risk management in the tunnel of the megaproject. The main specific steps are shown in Figure 2, including (1) clarify the purpose of modeling. (2) define the system boundary. (3) determine the feedback relationship. (4) design the equations of variables. (5) run the model simulation.

Figure 2. SD analysis steps for tunnel security risk management.

4.2. Causal relationship of influencing factors

Tunneling is a dynamic and open system in the security risk management process with multiple input and output variables. There are dynamical features such as mutual transmission, derivation, and complexity among the variables. When the value of the security risk management variables is not sufficient to meet the base requirements, new risks may be created, which may be passed on to other risk factors or reduce the level of tunnel security risk management. For example, the geology of the tunnel was surveyed using advance geological predictions. The quality of the geological surveys varies considerably between different skill levels of staff. If their technical level is not rigorously managed, it can easily lead to inaccurate tunnel geology, making it difficult to reveal the unknown geological risk susceptibilities. As a result, the security level of tunnel construction will be reduced and security risk accidents will be induced.

According to the causal circuit and accident causation theory, the factors affecting the efficiency of security risk management in the megaproject tunnel are considered comprehensively, and a causal loop diagram of efficiency of security risk management is constructed based on the causal relationship between each factor, as shown in Figure 3. The complexity of security risk is negative feedback, which has a negative impact on the efficiency metrics of security risk management. The higher the complexity of security risk, the worse the efficiency of security risk management. Moreover, all the other feedbacks in Figure 3 are positive feedbacks, which play a positive role in the efficiency of security risk management.

Figure 3. Causal loop diagram of efficiency of security risk management.

Note: ”+” represents positive feedback and “-” represents negative feedback

In the causal loop diagram of the efficiency of tunneling security risk management, the efficiency of security risk management is used as the horizontal variable. The efficiency of tunneling security risk management is generated by the joint action and interaction coupling of multiple factors. Therefore, a causal loop diagram of system dynamics is used to graphically describe the complex dynamical correlations of a dynamic security risk management system. With the help of Vensim software, the feedback loop for the efficiency of security risk management is analysed. ”+” indicates positive feedback and “-” indicates negative feedback. Based on the SD theory, the influencing factors in the OODA cycle, including risk perception, collaborative research and evaluation, risk decision making, and risk control, have a positive feedback loop that modulates the efficiency on efficiency of security risk management.

4.3. SD flow mapping

The tunnel security risk management system consists of a combination of four subsystems such as risk perception, collaborative research and evaluation, risk decision making, and risk control in the OODA loop. Based on the causal loop graph of efficiency of security risk management, the SD flow graph of efficiency of security risk management is plotted with the help of Vensim software, as shown in Figure 4.

Figure 4. SD flow graph of efficiency of security risk management.

4.4. Construction of SD equation

Based on the influencing factors and system dynamics flow diagram of tunnel security risk management efficiency, the weights of the factors were selected as the variable coefficients to establish SD equations for the system dynamics model of security risk management efficiency, and the SD equations of the main variables are shown in Table 2. The weights of the four subsystems of risk perception, collaborative research and evaluation, risk decision making, and risk control are W_A, W_B, W_C, and W_D. The factor weights of the risk perception subsystem are W_A1, W_A2, W_A3, and W_A4. The factor weights for the collaborative research and evaluation subsystems are W_B1, W_B2, W_B3, and W_B4. The factor weights of the risk decision making subsystems are W_C1, W_C2, W_C3, and W_C4. The factor weights of the risk control subsystems are W_D1, W_D2, W_D3, and W_D4.

Table 2. SD equations of security risk management efficiency.

Variables	Types	Expressions
Efficiency of security risk management	State	Efficiency of security risk management = INTEG (Amount of change in security risk management efficiency)
Complexity of security risk	State	Complexity of security risk = INTEG (Amount of change in complexity of security risk)
Risk perception	State	Risk perception = INTEG (Growth rate of risk perception - Reduction rate of risk perception)
Collaborative research and evaluation	State	Collaborative research and evaluation = INTEG (Growth rate of collaborative research and evaluation - Reduction rate of collaborative research and evaluation)
Risk decision making	State	Risk decision making = INTEG (Growth rate of risk decision making - Reduction rate of risk decision making)
Risk control	State	Risk control = INTEG (Growth rate of risk control - Reduction rate of risk control)
Amount of change in efficiency of security risk management	Rate	Amount of change in efficiency of security risk management = Risk perception × WA + Collaborative research and evaluation ×WB + Risk decision making ×WC + Risk control ×WD
Amount of change in complexity of security risk	Rate	Amount of change in complexity of security risk = Complexity of security risk × WE - Security risk management efficiency × WF
Growth rate of risk perception	Rate	Growth rate of risk perception = A1 × WA1 + A2 × WA2 + A3× WA3 + A4× WA4 +
Reduction rate of risk perception	Rate	Reduction rate of risk perception = Complexity of security risk × WA
Growth rate of collaborative research and evaluation	Rate	Growth rate of collaborative research and evaluation = B1 × WB1 + B2 × WB2 + B3× WB3 + B4× WB4
Reduction rate of collaborative research and evaluation	Rate	Reduction rate of collaborative research and evaluation = Complexity of security risk × WB
Growth rate of Risk decision making	Rate	Growth rate of risk decision making = C1 × WC1 + C2 × WC2 + C3× WC3 + C4× WC4
Reduction rate of Risk decision making	Rate	Reduction rate of risk decision making = Complexity of security risk × WC
Growth rate of risk control	Rate	Growth rate of risk control= D1 × WD1 + D2 × WD2 + D3× WD3 + D4× WD4
Reduction rate of risk control	Rate	Reduction rate of risk control = Complexity of security risk × WD

5. Empirical simulation and analysis

5.1. Analysis of indicator weights

The SD method is used to analyse the efficiency of tunnel security risk management, which requires the assignment of indicator weights. In order to reduce the limitations of the single assignment method, the hierarchical entropy weight method (hierarchical analysis method – entropy weight method) is invoked in the process of determining the weights. A combination of subjective and objective evaluations is used to modify the weights, which makes the weights of each indicator more reliable.

5.1.1. Analytic hierarchy process method

The AHP (Analytic hierarchy process) method, first proposed by American operations researcher Saaty in the 1970s, is a method of qualitative and quantitative analysis for decision making, which has the advantages of being simple, systematic and flexible. The AHP method is commonly used to solve complex multi-objective decision problems, and is also widely used to determine the weights of evaluation metrics. The specific calculational steps of the AHP method are as follows.

(1) Construct the judgment matrix. Through expert consultation and the application of 9 scaling method, a judgment matrix U is constructed based on a two-by-two comparison of the relative importance of indicators in the same tier relative to the indicators in the upper tier, where U_ij indicates the comparison result of the ith indicator relative to the jth indicator.

(1) $\mathit{U}=\left[\begin{array}{cccc}{U}_{11}& U_{12}& \cdots & U_{1\mathit{n}}\\ U_{21}& U_{22}& \cdots & U_{2\mathit{n}}\\ ⋮& ⋮& \mathit{\hspace{0.17em}}& ⋮\\ U_{\mathit{n}1}& U_{\mathit{n}2}& \cdots & U_{\mathit{nn}}\end{array}\right]$(1)

(2) The weights of each indicator ω_i were calculated. the weights of the indicators were determined by using the geometric mean method.

(2) ${\omega }_i=({\prod }_{j=1}^n{U}_{\mathit{ij}}{)}^{1/n}/{\sum }_{i=1}^n{({\prod }_{j=1}^n{U}_{\mathit{ij}})}^{1/\mathit{n}}$(2)

(3) Calculate the maximum characteristic root of matrix U.

(3) ${\lambda }_{max}=\sum _{\mathit{i}=1}^n\frac{{(\mathit{U\omega })}_i}{\mathit{n}{\omega }_i}$(3)

(4) Calculate consistency metrics.

(4) $\mathit{CI}=\frac{{\mathit{\lambda }}_{max}-\mathit{n}}{\mathit{n}-1}$(4)

(5) Determine the consistency of the matrix.

(5) $\mathit{CR}=\frac{\mathit{CI}}{\mathit{RI}}$(5)

The specific values of RI are shown in Table 3, and when CR < 0.1, the matrix consistency test is considered to be passed, while the opposite needs to be corrected.

Table 3. The RI reference value.

Matrix orders	1	2	3	4	5	6	7	8	9
RI	0.00	0.00	0.58	0.90	1.12	1.24	1.32	1.41	1.46
Matrix orders	10	11	12	13	14	15	16	17	18
RI	1.49	1.52	1.54	1.56	1.58	1.59	1.5943	1.6064	1.6133

5.1.2. Entropy weight method

Entropy theory reflects the degree of information disorder, and the more information a certain indicator carries, the more influence it has on decision making. In general, the larger the dispersion of evaluation metrics, the larger the amount of information and weight. On the contrary, the smaller the dispersion of the evaluation index, the smaller the amount of information and weight. Due to the different magnitudes of each evaluation metric, the data needs to be standardized in order to achieve comparability between the data. The specific calculation procedure of the entropic weight method is as follows.

(1) Normalize the column vectors of the judgment matrix.

(6) $b_{\mathit{ij}}=\frac{{\mathit{a}}_{\mathit{ij}}}{\sum _{\mathit{k}=1}^n{a}_{\mathit{kj}}}$(6)

(2) Calculate the index entropy value.

(7) $e_j=-\frac{1}{ln\mathit{n}}\sum _{\mathit{i}=1}^n{b}_{\mathit{ij}}ln{b}_{\mathit{ij}}$(7)

(3) Calculate the information entropy redundancy.

(8) $d_j=1-e_j$(8)

(4) Calculate the entropy weight of each indicator.

(9) $u_j=\frac{d_j}{\sum _{\mathit{j}=1}^n{d}_j}$(9)

(5) Calculate the modified indicator weights.

(10) $q_j=\frac{u_j{\omega }_j}{\sum _{\mathit{j}=1}^n{u}_j{\omega }_j}$(10)

5.1.3. Indicator weights based on AHP-EWM

Based on the entropic hierarchical analysis method, the subjective index weights are determined by the hierarchical analysis method and then the objective index weights are modified by the entropic weight method.

In this paper, a number of experts, including senior engineers of megaprojects and professors of engineering security risk management, were invited to set up an expert scoring group. After consultation and discussion, the judgment matrix for the indicators in this paper was finally determined based on the scale of the explanation matrix, as shown in Table 4.

Table 4. Weighted decision matrix.

Indicator	A1	A2	A4	A5	B1	B2	B3	B4	C1	C2	C3	C4	D1	D2	D3	D4
A1	1	1	2	1/3	2	1/3	1/2	2	3	1/2	1	2	3	1	2	5
A2	5	1	3	2	5	2	5	7	9	2	3	4	7	5	6	9
A4	1/2	1/3	1	1/3	3	1/2	3	5	2	1	1/2	1	2	1/2	1	3
A5	3	1/2	3	1	5	1	1	3	7	3	4	5	5	3	4	7
B1	1/2	1/5	1/3	1/5	1	2	3	4	2	1/3	1/2	1	2	1/2	1	1/2
B2	3	1/2	2	1	1/2	1	2	3	5	2	3	4	5	3	4	7
B3	2	1/5	1/3	1	1/3	1/2	1	2	4	1	2	3	4	2	3	6
B4	1/2	1/7	1/5	1/3	1/4	1/3	1/2	1	2	1/2	1/3	1	2	1/2	1	3
C1	1/3	1/9	1/2	1/7	1/2	1/5	1/4	1/2	1	1/5	1/3	1/2	1	3	1/2	2
C2	2	1/2	1	1/3	3	1/2	1	2	5	1	2	4	5	2	3	7
C3	1	1/3	2	1/4	2	1/3	1/2	3	3	1/2	1	1/2	2	1	2	4
C4	1/2	1/4	1	1/5	1	1/4	1/3	1	2	1/4	1/2	1	2	1/2	1	3
D1	1/3	1/7	1/2	1/5	1/2	1/5	1/4	1/2	1	1/5	1/2	1/2	1	1/3	1/2	2
D2	1	1/5	2	1/3	2	1/3	1/2	2	1/3	1/2	1	2	3	1	2	3
D3	1/2	1/6	1	1/4	1	1/4	1/3	1	2	1/3	1/2	1	2	1/2	1	2
D4	1/5	1/9	1/3	1/7	2	1/7	1/6	1/3	1/2	1/7	1/4	1/3	1/2	1/3	1/2	1

After the computation of the AHP steps, the weights ${\omega }_i$ of each indicator can be obtained. ${\omega }_i$ = {0.063, 0.182, 0.058, 0.131, 0.051, 0.109, 0.071, 0.029, 0.025, 0.083, 0.051, 0.031, 0.019, 0.049, 0.030, 0.016}. CI = 0.142, CR = 0.088 < 0.1, therefore, the consistency test of the judgment matrix passes. The weights u_j for each indicator can be obtained after computing the steps of the EWM. u_j = {0.080, 0.052, 0.053, 0.072, 0.064, 0.070, 0.097, 0.054, 0.056, 0.075, 0.071, 0.006, 0.036, 0.069, 0.053, 0.038}.

Thus, the indicator weights q_i based on AHP-EWM can be calculated. q_i = {0.076, 0.144, 0.047, 0.143, 0.050, 0.116, 0.105, 0.024, 0.021, 0.095, 0.055, 0.029, 0.011, 0.050, 0.024, 0.010}. The corresponding weights of each factor for risk perception, collaborative research and evaluation, risk decision making, and risk control are (W_A, W_B, W_C, W_D) = (0.410, 0.295, 0.200, 0.095). The weight values of the secondary indicators for each subsystem are given in Table 5. Finally, the combined weight of security risk complexity and security risk management efficiency is 0.167 and 0.833 after expert analysis.

Table 5. Results of indicator combination weights.

Primary indicator	Secondary indicator	AHP weights	Entropy weights	Combined weights
A (0.603)	A1	0.144	0.312	0.186
	A2	0.420	0.202	0.351
	A3	0.134	0.206	0.114
	A4	0.302	0.280	0.349
B (0.118)	B1	0.197	0.225	0.170
	B2	0.418	0.246	0.393
	B3	0.274	0.340	0.357
	B4	0.111	0.189	0.080
C (0.147)	C1	0.132	0.214	0.107
	C2	0.435	0.286	0.474
	C3	0.268	0.271	0.276
	C4	0.165	0.229	0.143
D (0.132)	D1	0.1698	0.184	0.112
	D2	0.4226	0.352	0.533
	D3	0.2642	0.270	0.256
	D4	0.1434	0.194	0.099

The factor weights of risk perception subsystem are (W_A1, W_A2, W_A3, W_A4) = (0.186,0.351,0.114,0.349), which suggests that security risk warning and communication level, ability to collect security risk information are the main factors influencing risk perception. The factor weights of collaborative research and evaluation subsystem are (W_B1, W_B2, W_B3, W_B4) = (0.170, 0.393, 0.357, 0.080), which suggests that cross-organizational collaborative security risk research and evaluation ability, cross-departmental collaborative security risk research and evaluation ability are the main factors influencing collaborative research and evaluation. The factor weights of risk decision making subsystem are (W_C1, W_C2, W_C3, W_C4) = (0.107, 0.474, 0.276, 0.112), which suggests that degree of cross-organizational involvement in risk decision making, degree of information and intelligent decision-making for security risks are the main factors influencing risk decision making. The factor weights of risk control subsystem are (W_D1, W_D2, W_D3, W_D4) = (0.112, 0.533, 0.256, 0.099), which suggests that reasonableness of security risk control program selection, ability to implement security risk control programs are the main factors influencing risk control.

5.2. Simulation experimental design

For the purpose of exploring the impact of different metrics on the efficiency of security risk management in megaproject tunneling under different conditions. The experimental design of SD simulations for the efficiency of management of security risk in megaproject tunneling is shown in Table 6. Three simulation experiments are set up in this study. The main purpose of experiment a is to investigate the variation of different metrics and observe the overall variation in the complexity and efficiency of security risk management. The main purpose of experiment b is to observe the impact of risk perception on the efficiency of security risk management. The main purpose of experiment c is to analyse the impact of time delay on the efficiency of security risk management.

Table 6. Experimental design of SD simulation.

Experimental group	Experimental purpose	Parameter Setting
Experiment a	①Changes of different indicators in the process of management of security risks.	The initial value of the complexity of security risk and efficiency of security risk management are set to 1, the initial values of risk perception, collaborative research and evaluation, risk decision making, risk control are set to 0, and the initial values of other constants are set to 1.
	②Observe the overall change of the complexity of security risk and efficiency of security risk management.
Experiment b	Observe the impact of risk perception on the efficiency of security risk management.	Increase the initial value of risk perception and leave the other variables unchanged.
Experiment c	Analyze the impact of time delays on the efficiency of security risk management.	The initial values of growth rate of risk perception, growth rate of collaborative research and evaluation, growth rate of risk decision making, growth rate of risk control are set as step functions: STEP({height}, {stime}), stime is set to 1, 2 and 3 respectively.

5.3. Model simulation experiments

Three representative model simulation experiments were conducted with simulation time T = 8 and simulation step size of 0.5 in hours. The simulations were performed by using the Vensim software. Vensim is a visual modeling tool that allows you to conceptualize, document, simulate, analyse, and optimize models of dynamic systems. Vensim provides a simple and flexible way to build simulation models from causal loops or stock and flow graphs.

5.3.1. Experiment a

The simulation of the Experiment a (1) model is shown in Figure 5, where risk perception rises faster than collaborative research and evaluation, risk decision making, risk control. This indicates that risk perception has the greatest impact on the efficiency of security risk management, followed by collaborative research and evaluation and risk control with the least impact. The simulation of the Experiment a (2) model is shown in Figure 6, complexity of security risk and efficiency of security risk management show a clear antagonistic trend. Security risk management activities have continued to advance over time. The efficiency of security risk management continues to grow, and the complexity of security risks is gradually being degraded and controlled.

Figure 5. Simulation diagram of experiment a (1) model.

Figure 6. Simulation diagram of experiment a (2) model.

5.3.2. Experiment b

The simulation of the model of Experiment b is shown in Figure 7. As the initial value of risk perception increases, the upward trend of security risk management efficiency gradually increases, and correspondingly, the downward trend of security risk complexity significantly increases. Therefore, it can be seen that risk perception is a prerequisite for the implementation of risk management in the process of security risk management in megaproject tunnels. When the level of risk perception is high, the efficiency of security risk management increases rapidly as the level of risk management is improved.

Figure 7. Simulation diagram of experiment b model.

5.3.3. Experiment c

The simulation of the model of Experiment c is shown in Figure 8, where the simulation results with delays of 1, 2 and 3 h are compared and analysed. The results show that the larger the delay time, the slower and worse the rise in the efficiency of security risk management, and the faster and more complex the decrease in the complexity of security risk management. This indicates that when risk perception, collaborative research and evaluation, risk decision making, and risk control are inactive, security risk complexity is less affected by risk management behaviour and becomes progressively more complex.

Figure 8. Simulation diagram of experiment c model.

6. Discussion

6.1. Strategies to improve the efficiency of security risk management

In the process of managing of security risks in megaproject tunnels, the complexity of security risks should be controlled in real time, and the perception, collaborative research, decision making and control of security risks should be comprehensively improved by construction managers and agencies. Real-time collection of construction site security risk information and enhanced communication and exchange between construction personnel and management regarding security risks. Form a real-time, comprehensive and uninterrupted security risk information perception network so that the acquisition, transmission, processing and sharing of security risk information form an efficient circuit. Based on the risk information, the participants and departments are organized to conduct collaborative research and assessment of security risks. Participants worked together to use information technology to make decisions about security risks and to select a reasonable security risk control program to implement risk control.

6.2. Risk perception is key influencing factor

Risk perception is a prerequisite and key element in the management of security risks, according to the weighting of the influencing factors in the management of security risks in megaproject tunnels. Risk perception, as a cognitive factor for tunneling construction and management agents, plays a leading role in the management behaviour of security risks. The core component of risk perception is the perception and awareness of external risks by individuals or groups, which emphasizes the experience gained by influencing the subjective judgment and feelings of individuals. Megaproject tunnel builders work with a large number of security risks, and the higher their level of risk perception, the more aware they will be of the possibility of being injured on the job. Therefore, the level of security risk warning and communication, and the ability to collect security risk information need to be focused on improving to adopt a higher level of security behaviour to meet security needs.

6.3. Reduce delays in management of security risks

Reducing delays in the managing of security risks is an effective measure to improve management efficiency. Participants can maintain close cooperation in security risk perception, research and evaluation, decision making and control. When security risks are found to be hidden, strategies are adopted to control them in a timely manner. The response of project construction and management personnel to security risks should be improved to reduce the delay in security risk management, which can improve the efficiency of security risk management. Moreover, the collaboration of multiple entities and sectors involved in megaproject tunneling can overcome the subjective and objective factors that cause time delays and effectively shorten the dynamic delay time of security risk.

The limitation of this study is that the research method is not applied to the actual project to be tested, but rather the law of influence of its indicator factors on the efficiency of tunnel security risk management is explored by means of simulation tests. In future research, the authors will focus on applying the research results to real megaprojects. This will enhance the objectivity of the indicator data, propose more detailed measures to improve the efficiency of tunnel security risk management, and increase the practicality of the model approach.

7. Conclusion

The analysis of the efficiency of security risk management in megaproject tunneling is an analysis of evolution of complex risk management processes. Based on the OODA model, this study analyzes that affect the efficiency of security risk management in megaproject tunnels in four aspects: observe, orient, decision and action. The SD method is used to construct a system dynamics model of the efficiency of security risk management in megaproject tunneling. Moreover, the influence factors on the efficiency of security risk management are assigned weights based on the AHP-EWM method, and the complex functional relationships of the system dynamics model are dynamically simulated. The conclusions are as follows.

1. The corresponding weights of each factor for risk perception, collaborative research and evaluation, risk decision making, and risk control are 0.410, 0.295, 0.200, 0.095. They are important influencing factors for the efficiency of managing security risks in megaproject tunnels. Risk perception is not only an important prerequisite for the implementation of dynamic security risk management activities, but also the most critical influencing factor.

2. Security risk warning and communication level (0.351), ability to collect security risk information (0.349) are the main factors influencing risk perception. Cross-organizational (0.393) and cross-departmental (0.357) collaborative security risk research and evaluation ability are the main factors influencing collaborative research and evaluation. Degree of cross-organizational (0.474) and degree of intelligent decision making (0.276) are the main factors influencing risk decision making. Reasonableness of security risk control program selection (0.533) and ability to implement security risk control programs (0.256) are the main factors influencing risk control.

3. Risk perception, collaborative research and evaluation, risk decision making, risk control can reduce the complexity of security incidents in megaproject tunnels. Real-time adjustment of risk control strategies based on the complexity of security risks can improve the efficiency of dynamic security risk management.

4. If managing security risks is delayed, it is difficult to degrade the complexity of the risks in a timely manner, which can reduce the efficiency of dynamic security risk management. Therefore, reducing the delay can effectively improve the efficiency of security risk management.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by Fundamental Research Funds for the Central Universities: [Grant Number 2022YJS050]; State Key Laboratory for Track Technology of High-speed Railway: [Grant Number 2021YJ111], China Academy of Railway Sciences.

Notes on contributors

Kai Liu

Kai Liu is a PhD student in the School of Economics and Management at Beijing Jiaotong University. He received his Master of Engineering degree from the School of Civil Engineering and Architecture of Shandong University of Science and Technology in Qingdao, Shandong Province, China. His research focuses on project governance and project management. He can be reached at [email protected].

Yuming Liu

Yuming Liu is a professor in the School of Economics and Management at Beijing Jiaotong University. He received his Ph.D. in Management from Beijing Jiaotong University and his M.S. in Engineering from Hunan University. Liu has published several articles in journals, including Energy Conversion and Management and Energy and Buildings. He is a reviewer for Energy and Buildings, Journal of Cleaner Production, Energy Efficiency, and Building and Environment. He can be reached at [email protected].

Yuanyuan Kou

Yuanyuan Kou is a Ph.D. scholar in the College of Economics and Management, Shandong University of Science and Technology. She completed her Master’s program of Management Science and Engineering from Shandong University of Science and Technology. Her areas of interest include project management, engineering technical management. She can be reached at [email protected].

Xiaoxu Yang

Xiaoxu Yang is a PhD student in the School of Economics and Management at Beijing Jiaotong University. He completed his master’s degree in civil engineering at the School of Civil Engineering and Construction of Beijing Jiaotong University. His research interests are project management and tunnel construction management. He can be reached at [email protected].

Guangzhong Hu

Guangzhong Hu is a master’s student at the School of Economics and Management, Beijing Jiaotong University. He completed his undergraduate program at Harbin Engineering University. His research interests are project management and project governance. He can be reached at [email protected].