3.2.1. Risk Identification of TBM Construction Based on the WBS-RBS Method
One of the key factors in conducting TBM construction risk analysis is to determine the influential factors and establish a rational and effective risk index system [
19]. In project management, the work breakdown structure (WBS) is a crucial concept that involves systematically dividing the construction process into a well-defined hierarchical framework. On the other hand, the risk breakdown structure (RBS) is a technique used to progressively categorize potential risks originating from major sources into more specific risk factors until they become negligible [
20]. Hillson et al. [
21] were the first to combine the WBS and RBS methods, constructing a WBS-RBS risk identification coupling matrix for projects, identifying project risk factors, and establishing a project’s risk index system.
In tunnel TBM construction project risk management, according to the site environment and risk management design requirements, the target and limit of risk identification are preliminarily determined. For TBM construction of the tunnel, the goal of risk identification mainly refers to whether safety accidents will occur during TBM construction, including but not limited to equipment failure, casualties, environmental pollution, etc. The boundary of risk identification is more extensive, from the preparation work before TBM construction to the work in TBM excavation construction, etc., for example, in the early stage of TBM construction, the equipment needs to be comprehensively inspected and debuggable to ensure its safety and reliability. It is also necessary to conduct systematic safety education and skill training for personnel to improve their safety awareness and operational skills. In the normal driving process, all parameters should be monitored in real time. In the initial support stage, it is necessary to strengthen the tunnel support in time to avoid a collapse caused by groundwater flooding or rock mass loosening. As for the auxiliary construction stage, such as drainage, ventilation, and transportation, it is also necessary to pay attention to problems such as equipment failure and personnel injury.
Establish the work breakdown structure for TBM construction [
22,
23,
24,
25,
26,
27].
The WBS work breakdown structure method is founded on system engineering theory, enabling the gradual subdivision of a complex and extensive TBM tunnel project. Based on management content, it is divided into a single operational process, further disintegrating the TBM construction project into multiple work units that possess excellent controllability and operability.
According to the Work Breakdown Structure (WBS) principle, the TBM construction process (W) is divided into two levels:
- (1)
Based on the TBM construction process, the first level of the WBS is further decomposed into four stages: TBM construction preparation (W1), TBM tunneling construction (W2), TBM preliminary support (W3), and TBM auxiliary equipment construction (W4). The stage of TBM construction preparation primarily refers to the preparatory phase when the TBM enters the tunnel excavation.
- (2)
According to the operation conditions of each process of TBM construction, the first-level WBS is adjusted and optimized according to the characteristics of the process to form a more detailed second-level WBS, which is specifically the operation unit, and intuitively indicates that the first-level construction stage is composed of various second-level operation units. The obtained decomposition structure for TMB construction work using WBS method is illustrated in
Figure 2.
- 2
Establish the decomposition structure of TBM construction risk sources [
28,
29,
30,
31]
By using the RBS method, the potential risks in the TBM construction process are analyzed layer by layer, and a complete risk decomposition tree is established. In this process, the various risks that may cause accidents are classified and summarized in detail to ensure that the various safety risk factors faced in the TBM construction process can be clearly defined and presented in a clear and identifiable state.
The main risks in terms of geology include collapse, sudden water influx, rock burst, and leakage of harmful gases during excavation or support processes. Therefore, factors contributing to these risks are considered to be the joint effect of the rock integrity coefficient, elastic deformation performance index, rock mass classification, internal development degree of the rock mass, moisture content of the rock mass, structural planes within the rock mass, tunnel depth, and groundwater conditions. Additionally, considering the stability, bearing capacity, and deformation characteristics from a geological perspective involves examining the presence of joints and bedding planes as well as whether they are located in an earthquake-prone zone. From a rock-properties perspective, physical properties such as abrasion resistance, frost resistance, permeability resistance, creep behavior, thermal properties, and strength are taken into account.
The equipment risk factors include the type of TBM selected according to the geological conditions, the designed cutterhead structure, the cutters for different rock types, the design of different types of surrounding rock support, the selection of related auxiliary equipment, etc. The equipment risks with higher occurrence frequency include non-timely ventilation and dust removal, TBM blockage, discontinuous slagging, wear of the cutterhead or cutter, the setting of tunneling parameters inconsistent with the actual situation, etc.
The influence of human-made risks in TBM construction cannot be ignored. It has a high degree of importance, because the interference and influence of human factors on TBM construction exist all the time in the construction process, and human activities exist in every construction link, so the area of human-made risks is full of uncertainty. If the organizational structure of the construction management team is unreasonable, it may lead to poor information flow and low decision-making efficiency, affect construction safety, and even lead to serious human-made risks in TBM construction.
The environmental risks mainly include damage to underground pipelines, surface subsidence, groundwater pollution, and building damage. Surface subsidence is a common risk in the process of underground excavation, which may lead to different degrees of surface subsidence, and then lead to deformation and cracking of building structures. Damage to underground pipelines will lead to slow construction progress, so it is also a problem that cannot be ignored. Construction wastewater generated in the process of excavation, and the discharge of pumped leaking water, may pollute groundwater.
According to the engineering RBS principle, a two-level risk decomposition is conducted for TBM construction risk sources (R).
- (1)
Firstly, the primary Risk Breakdown Structure (RBS) for TBM construction projects and associated environmental risks comprises four categories: geological risk sources (R1), equipment-related risk sources (R2), human-induced risk sources (R3), and surrounding environment risk sources (R4).
- (2)
Secondly, through a comprehensive analysis of TBM construction risks, and considering the decomposition direction of first-level risk sources, various secondary-level risk structures are derived from the initial structure. The resulting decomposed structure for TBM construction risks obtained using the RBS method is depicted in
Figure 3.
- 3
Establishment of TBM construction risk identification coupling matrix
According to the established Work Breakdown Structure (WBS) and Risk Breakdown Structure (RBS), the lower-level units of both structures are interconnected to derive the coupling matrix for identifying construction risks in TBM projects [
25]. In this matrix, “0” signifies that there is no risk generated by the interconnection, while “1” indicates that a risk is generated.
In constructing the coupling matrix, we first ensure that each item (
W) in the WBS and each risk category (
R) in the RBS are clearly identified. Then, by matching one by one, we create a crossing point for each pair of
W ×
R that indicates whether
W creates a risk during construction. For example, hazardous gas leakage is generated by the intersection of “technical preparation
W11” and “tunnel depth
R15, construction management status
R31, security education situation
R33”, reflecting that the technical preparation caused by the tunnel burial depth is too large, the construction management is confused, and the safety education is insufficient, resulting in the occurrence of harmful gas, but non-timely feedback and solutions to the problem. The likelihood of accidents increases greatly. The outcomes of the TBM construction risk identification coupling matrix are presented in
Table 1. Finally, the specific risks are as follows: (1)
W11R15,
W11R31,
W11R33: Hazardous gas leakage; (2)
W12R11,
W12R12,
W12R12,
W21R14,
W35R14: Rock blast; (3)
W12R12,
W12R13,
W21R13,
W35R13: Surging water; (4)
W12R14,
W12R15,
W21R11: Collapse; (5)
W13R22,
W13R23,
W22R21,
W21R22,
W41R15: Discontinuous slagging; (6)
W14R21,
W14R31,
W14R32,
W14R33: TBM jamming; (7)
W21R21,
W21R22,
W21R31,
W21R32, W
21R
33: Support destabilization deformation; (8)
W21R41,
W42R42: Poor ventilation and dust collection; (9)
W23R21,
W23R22,
W24R21,
W24R22: Deviation of digging direction; (10)
W31R11,
W31R23,
W31R24,
W31R25: Cave collapse; (11)
W32R23,
W22R21,
W21R22: Water accumulated in the hole; (12)
W31R31,
W31R32,
W33R13: Lining leakage; (13)
W41R24,
W41R31,
W41R41,
W41R42,
W41R43: Underground pipeline damage; (14)
W34R13,
W34R14,
W21R15,
W44R15: Deformation of surrounding buildings; (15)
W21R42,
W13R21,
W13R22,
W44R42: Stratigraphic ground settlement.
Initially, a comprehensive evaluation is conducted to identify potential risks associated with TBM projects involving the construction of exceptionally long tunnels. This assessment follows a hierarchical structure model based on the fundamental principles of the analytic hierarchy process (AHP). The model comprises three layers: the target layer, criterion layer, and indicator layer, arranged in descending order of significance. The target layer represents TBM construction risk for ultra-long tunnels (
U), while the criterion layer comprises
U = [
U1,
U2,
U3,
U4]. The indicator layer includes geological risk
U1 = [
U11,
U12,
U13,
U14], equipment risk
U2 = [
U21,
U22,
U23,
U24], tunnel risk
U3 = [
U31,
U32,
U33,
U34], and surrounding environment risk
U4 = [
U41,
U42,
U43]. The final system for evaluating TBM construction risks is illustrated in
Figure 4.
3.2.2. Extended Matter-Element Model with Variable Weight
Based on the current research status both domestically and internationally, it is evident that the TBM has witnessed a growing utilization in the construction of lengthy tunnels and underground projects in recent years. However, traditional drilling and blasting methods have posed challenges in adapting their risk assessment techniques to the evolving trend of constructing ultra-long and deeply buried tunnels. Insufficient attention has been given to researching risk assessment methods specifically tailored for TBM construction [
32].
The initial development of the Matter-Element Extension Model was based on a robust mathematical statistical framework. By integrating principles from both Matter-Element Theory and Extension Set Theory, this model systematically establishes classical domains, segment domains, index levels, and matter elements for precise evaluation using empirical data. Furthermore, it effectively determines appropriate weights while calculating the correlation degree between the target matter element and each rating level to derive a comprehensive assessment outcome. Since its inception, scholars have continuously enhanced this model through various research endeavors. For instance, they have addressed issues related to single index values exceeding limits by employing standardization techniques; furthermore, they have improved evaluation accuracy while minimizing information loss by introducing the closeness criterion as an alternative to the maximum membership criterion [
33].
According to the principles of matter-element theory and extension set, the classical domains, segment domains, index levels, and matter elements are established in the extension model for evaluation purposes using collected data. Additionally, their respective weights are determined. Finally, the comprehensive rating is obtained by calculating the correlation degree of each rating level. In order to further enhance the model’s effectiveness, scholars address the issue of single index values exceeding the limit of matter elements through standardization processing and introduce a closeness criterion to improve evaluation accuracy while reducing information loss.
The assessment criteria are classified into j tiers based on national standards or pertinent specifications. The index weights are adjusted in accordance with the variable weight theory, considering the variability of each evaluation index value as opposed to using a traditional fixed-weight approach. Ultimately, the determination of the comprehensive evaluation level involves assessing the grade variable’s closeness and characteristic values for each level of the matter element being evaluated and selecting the highest obtained closeness value. Furthermore, we assess the closeness between consecutive levels by utilizing the distinctive value of the grade variable.
The fundamental unit that defines entities in terms of “thing, characteristic, and quantity” (represented by P, C, and V, respectively) is known as the matter element. It serves as the elementary logical operation unit in the extension model of matter elements. The proposed approach aims to optimize the evaluation model for extending matter elements through adjustments in variable weights.
- (1)
Establish classical domain, section domain, and matter element to be evaluated.
The evaluation level denoted as
, which is the
jth level, consists of
n unique features (
c1,
c2, …,
cn) and their respective value ranges (
v1,
v2, …,
vn), representing the classical domain. The defined boundaries for this value range are
and
.
In the equation,
represents the overall grade of the evaluated entity;
vp1,
vp2, …,
vpn denotes the segment of
that corresponds to
c1,
c2, …,
cn, specifically referred to as the nodal domain.
In the equation, represents the matter element under evaluation; v1, v2, …, vn denotes the measured data of in relation to c1, c2, …, cn, respectively.
- (2)
Normalized processing
The issue of exceeding the limit in the single index value of the matter element to be evaluated can be addressed through normalization processing, thereby normalizing the classical domain matter element
.
Similarly, the evaluation element
needs to be normalized, and the following results are obtained:
- 2.
Determine variable weights of indicators.
Firstly, according to the established prediction index system, the 1~9 scale method proposed by Saaty (
Table 2) is adopted to analyze the relative importance of any two indicators at each level [
34], and the judgment matrix of prediction indices is obtained.
where
is an element in the judgment matrix
, and
satisfies
.
Secondly, Equations (7) and (8) are used to calculate the weight vector
and the maximum eigenvalue
of
[
35]:
Finally, in order to further determine the reliability of the calculation results, Equation (9) is used to verify whether the matrix
meets the consistency requirements:
The random consistency index (
) value can be located in
Table 3, indicating the location where it is presented.
When the CR value is less than 0.1, it indicates that the matrix meets the consistency requirements, and the calculation results are more reliable.
Through the introduction of the variable weight theory, the influence of index value volatility is taken into consideration based on traditional fixed weights, thereby achieving an improved dynamic calculation method for variable weight indices [
36]. The variable weight theory is proposed based on factor space theory. The fundamental statement of the variable weight theory states that if
represents the factor state vector,
represents the factor constant weight vector, and
represents the state variable weight vector, then the variable weight vector
can be expressed as a normalized Hadamard product of
and
:
where
i = 1, 2, …,
n,
is the equilibrium function, usually an exponential function with the natural constant
e as the base.
The analytic hierarchy process is utilized to assess the relative importance of each factor index, considering both subjective and objective evaluation methods. Moreover, the weight vector for state variables is determined by taking into account the equilibrium function and measured index value, thereby incorporating the impact of fluctuations on index weights. As a result, a calculation formula is devised to determine the variable weight of evaluation indices.
where
= 1.0;
;
L represents the rating level of index ;
represents the overall count of ratings within the assessment system;
c determines the threshold used to determine the state of incentives or punishments within the equilibrium function.
After the normalization,
where
;
, is the penalty;
, is the reward.
Therefore, the constructed variable weight state vector can be used to punish or encourage according to the measured values of the index.
- 3.
The value of the closeness function needs to be determined.
Theoretical analysis [
37] has led to the development of an asymmetric formula (
p = 1) for evaluating closeness, which replaces the previous maximum membership criterion.
where
represents the proximity factor;
denotes the measure of separation;
is the weight.
In addition, the proximity of the item under evaluation to every grade is:
In the formula,
where
represents the count of assessment indicators.
- 4.
Rank confirmation.
The evaluated item is categorized as level
.