Tunnel Boring Machine Excavation Research Articles

Automated position control of tunnel boring machine during excavation using deep reinforcement learning

Position control of tunnel boring machines (TBM) is critical to ensure precise construction and meets design specifications. However, the high dependence on human intervention largely affects the efficiency and reliability of the operations due to slow response, inconsistent judgment and high risk of errors. Targeting to automate the position control of the TBM to enable it to the planned route in a more efficient and reliable manner, this research proposes a deep reinforcement learning (DRL) method considering spatial-temporal dynamics to perform automated real-time TBM operations. The DRL algorithm is embedded with a time-series deep learning model to simulate the working environment with dynamic time-space variations. The whole process is coupled with the designed reward and evaluation metrics for model enhancement. To showcase the viability and applicability of the suggested approach, a project from Singapore is utilized as an illustrative example. The outcomes indicate that the suggested method exhibits robust capabilities in forecasting and controlling TBM positions, with an R2 of 0.86 and 0.92 in forecasting the horizontal and vertical deviations of the TBM tail three steps ahead and an overall improvement of 50.7 %. The proposed method is capable of providing an optimal strategy to intelligently forecast and control the TBM deviation. The novelty of this method comes from the feasible integration of time series prediction and DRL to simulate the spatial-temporal dynamics along with TBM excavation and obtain a position control model for optimal tunneling strategy, contributing to the effective facilitation of TBM tunneling to improve efficiency and reliability.

TBM excavation of São Paulo Metro Line 6 South in heterogeneous ground in urban area

AbstractThe São Paulo Metro is expanding its network in the São Paulo Metropolitan Area inhabited by 21 million people. Metro Line 6 connects the north of the city to the existing metro network further South. The construction of Line 6 currently represents the largest infrastructure project in Latin America, combining the excavation of tunnel boring machine (TBM) and conventional tunnels, cut‐and‐cover stations and a number of deep emergency exit shafts. This paper presents the main features of the TBM tunnel excavated in the South stretch through the Tertiary soils of the São Paulo Basin and through the challenging Embu Complex, which consists of a mixture of fractured rock, saprolite and residual soils. The alignment runs under the dense urban grid of São Paulo under high‐rise buildings and other infrastructure, such as existing Metro lines.

Probabilistic assessment of rockburst risk in TBM-excavated tunnels with multi-source data fusion

In recent years, tunnel boring machines (TBMs) have become extensively utilized in underground engineering projects. However, as a prevalent type of dynamic geological hazard, rockburst poses a serious threat to the safety of personnel, equipment, and property. To ensure the safe advancement of TBMs, we integrate the data of microseismic monitoring, TBM excavation, and surrounding rock and establish multiple probabilistic assessment models of rockburst risk using probabilistic ensemble learning and quantum particle swarm optimization. In order to identify the most optimal model for geological engineers, a comprehensive evaluation system is devised based on metrics like accuracy, Cohen’s kappa, and F1-score. This system provides a thorough assessment of the models’ global and local generalization performance. The results indicate that the classification and regression tree-extremely randomized trees (CART-ERT) hybrid model achieves the highest score, demonstrating superior generalization performance with an accuracy of 93.06% and a Cohen’s kappa of 0.9074. Moreover, the F1-scores for no rockburst, low rockburst, moderate rockburst, and strong rockburst are 0.9412, 0.9444, 0.9189, and 0.9189, respectively. Based on the operational framework of the CART-ERT hybrid model, a user-friendly graphical user interface (GUI) system is developed. This significantly enhances the practicality and deployability of the model. Through application in a TBM diversion tunnel project in Xinjiang, the on-site early-warning accuracy of rockburst with the GUI system reaches 90%. Notably, the GUI system maintains high early-warning sensitivity and reliability for high-intensity rockburst such as moderate and strong ones. Lastly, to enhance the model’s interpretability, a variable importance measure (VIM) analysis on 14 features extracted from three types of heterogeneous data is conducted to assess their contributions to the early warning of rockburst. We discover that the cumulative energy change rate of microseismic events exhibits the highest importance score, indicating its crucial role in rockburst early warning.

126 Size distribution and characterization of particulate matter generated during tunnel boring machine excavation

126 Size distribution and characterization of particulate matter generated during tunnel boring machine excavation

Machine learning-based prediction model for disc cutter life in TBM excavation through hard rock formations

Replacing damaged or worn disc cutters is costly and time-consuming, which can significantly reduce the utilization and advance rate of tunnel boring machines (TBMs). Accurate prediction of disc cutter life is crucial for optimizing TBM operation in tunneling projects. This study introduces a machine-learning-based prediction model designed to forecast disc cutter wear, incorporating the analysis of each cutter’s travel length and the intervals between cutterhead interventions (CHI). The principle underlying the developed machine learning approach involves usage of multiple learning algorithms, ensemble learning methods, and a sophisticated analysis of these factors to identify patterns and relationships essential for effectively predicting wear rates. Employing CHI report data from the Daegok–Sosa tunneling project’s hard rock excavation, the model evaluates 15 wear-influencing factors, providing precise wear predictions. This study proposes the ensemble machine learning (ML) methods, namely Random Forest (RF) and Extreme Gradient Boosting (XGB), for accurate wear rate predictions and discern influential factors. The proposed model demonstrated exceptional prediction accuracy, as evidenced by a root mean square error of 0.049 mm for a single-ring excavation length. Notably, this model innovatively accounts for variable wear rates of cutters based on the individual cutter travel lengths as well as other geological and operational parameters. Furthermore, the predicted cutter consumption rate showed reasonable correspondence when compared to the actual CHI records. The proposed model is expected to improve existing disc cutter life prediction methods and reduce the cost and time for replacing damaged or worn disc cutters.

Model-based offline reinforcement learning framework for optimizing tunnel boring machine operation

Research on automation and intelligent operation of tunnel boring machine (TBM) is receiving more and more attention, benefiting from the increasing construction data. However, most studies on TBM operations optimization were trained by the labels of human drivers’ decisions, which were subjective and stochastic. As a result, the control parameters suggested by these models could hardly surpass the performance of a human driver, even the possibility of subjective incorrect decisions. Considering that the geomechanical feedback to TBM under drivers’ actions is objective, in this paper, a transformer-based model called the geological response for tunnel boring machine (GRTBM), is proposed to learn the relationship between operation-adjust and TBM monitoring changes. Additionally, with the model-based offline reinforcement learning, this paper provided a novel approach to optimizing the TBM excavation operations. The decision processes, recorded in the Yin-song TBM project for a waterway tunnel in Jilin Province of China, were used for the validation of the model. By adopting an implicit perception of geological conditions in the GRTBM model, the suggested method achieved the desired state within a single action, greatly outperformed the practical adjustments where 500 s were taken, revealing the fact that the proposed model has the potential to surpass the capability of human beings.

Force and wear prediction model of disc cutter under compression shear failure mode based on dense core–cavity expansion theory (U-713)

Calculating disc cutter wear and predicting wear life are long-standing issues affecting the efficiency and cost of tunnel boring machines (TBMs). During the TBM excavation process, the wear of the disc cutter is a comprehensive outcome of the interaction between the tool and rock. Studying the mechanisms of tool rock breaking and wear can assist in developing a more precise model for predicting cutter wear. Therefore, based on the rock-breaking and wear mechanisms of a cutter, this study developed a model for predicting the cutter wear rate and life. The model is based on the theory of dense core–cavity expansion combined with the motion state of a cutter. First, the TBM excavation process and rock-breaking mechanism of a cutter were analyzed in detail. A cutting force model under the compression shear failure mode based on the dense core–cavity expansion theory was established. Based on the force model, a prediction model for the wear rate and life of the disc cutter was established, considering the working characteristics and wear mechanism. Furthermore, the impact of penetration on the prediction model was investigated. In addition, to demonstrate the practicality of the method, this study conducted a comprehensive examination using engineering examples. The results showed that the average difference in the cutter average wear rate using the developed model was less than 3% compared with the actual engineering result. In addition, the average difference in the wear life rate was less than 5%. Finally, the developed prediction method was compared with other models. The wear rate fitting accuracy was improved by 4.91–10.85% and the wear life fitting accuracy was improved by 5.11–12.19% by the developed method. This improvement confirms the reasonableness and reliability of the method. This method can estimate the wear rate and lifespan of a cutter more accurately and reliably than other methods, providing a practical tool to TBM engineers to make informed decisions and optimize tool replacement programs.

Soft ground micro TBM jack speed and torque prediction using machine learning models through operator data and micro TBM-log data synchronization

Tunnel Boring Machines (TBMs) are pivotal in underground projects like subways, highways, and water supply tunnels. Predicting and monitoring jack speed and torque is crucial for optimizing TBM excavation efficiency. Conventionally, skilled operators manually adjust numerous tunnelling parameters to regulate the machine's progress. In contrast, machine learning (ML) algorithms offer a promising avenue where computers learn from operator actions to establish parameter relationships autonomously. This study introduces an innovative approach to enhancing operator monitoring and TBM data comprehension. A robust correlation between TBM operator behaviour and TBM logged data is established by leveraging an Optuna-assisted ML methodology—the research light on the intricate dynamics influencing TBM advance rate parameters. Operational data is collected from micro slurry tunnel boring machine (MSTBM) umbrella support excavations. The proposed framework harnesses Optuna, an advanced hyperparameter optimization platform, to dynamically refine jack speed and torque settings. Through meticulous analysis of the interplay between TBM operator decisions and real-time logged data, the AI model discerns patterns, empowering informed decision-making. Using Optuna, a range of models, including random forest (RF), K-nearest neighbours (kNN), decision tree (DT), XGBoost, Support Vector Machine (SVM), and Artificial Neural Network (ANN) were automatically compared and tuned. The best model's (RF) performance is evaluated through a correlation coefficient (R2) of 96%, mean squared error (MSE) of 119.7, and mean absolute error (MAE) of 4.42 for jack speed decision making while 83% of R2, MSE of 0.62, and MAE of 0.42 for the torque decision making. This intelligent model can assist the TBM operator in making decisions about TBM control.

Disc-cutter induced rock breakage mechanism for TBM excavation in rock masses with different joint shear strengths

When tunnel boring machines (TBMs) excavate through jointed rock masses, the cutting efficiency is strongly affected by the shear strength of joints, the mechanism of which, however, remains poorly understood. In this study, a series of disc-cutter indentation tests were conducted on granite rock mass specimens with different joint shear strengths. During the indentation, the cracking process was recorded by a digital image correlation (DIC) system. The deformation and strength of specimens, cracking behavior, rock breakage mode and cutting efficiency were quantitatively investigated. In addition, to investigate the combined effects of joint shear strength, orientation and spacing on the rock breakage mechanism, numerical rock mass models were established based on a particle flow code PFC2D. Experimental results reveal that the cracking of primary and secondary cracks changes from the mixed shear-tensile to tensile mode in the initial stage, while the joint shear strength does not affect the cracking mode in the subsequent propagation process. The rock breakage mode is classified to an internal block breakage mode, a cross-joint breakage mode and a cutters-dependent breakage mode. The cross-joint breakage mode is optimal for improving the cutting efficiency. Numerical simulation results reveal that the increase in the joint shear strength changes the internal block breakage mode to cross-joint breakage mode for rock masses of particular ranges of joint orientation and spacing. These findings provide basis for improving the TBM cutting efficiency through jointed rock masses.

Prediction of the tunnelling advance speed of a super-large-diameter shield machine based on a KF-CNN-BiGRU hybrid neural network

The tunnelling advance speed (TAS) of a super-large-diameter tunnel boring machine (TBM) significantly affects project progress and safety. Effective prediction of the TAS of a TBM is crucial for early warning and evaluation of its excavation performance. This paper proposes a model based on convolutional neural network (CNN) and bidirectional gated recurrent unit (BiGRU) to predict TAS. This model can more effectively extract and retain many features; additionally, it considers the temporal relationships among the operating parameters of the TBM. Additionally, a Kalman filter (KF) is incorporated into the model to effectively remove input parameter noise, leading to a more accurate prediction of TBM excavation speed. A case study indicated that the pressure of the excavation chamber and the pitch angle of the shield tunnelling machine were correlated with the advance speed. The research results show that the proposed model can accurately and effectively predict TAS.

Energy-driven TBM health status estimation with a hybrid deep learning approach

This study provides a hybrid deep learning approach that enables the researchers to estimate the Tunnel Boring Machine (TBM) health status during excavation process using energy consumption data. By analyzing the energy forms and paths in the tunneling process, the key energy factors' consumption performance behavior affecting TBM safety is identified. The Ensemble Empirical Mode Decomposition (EEMD) method and a Convolution neural network-Long short term memory-Depth Neural Network (CNN-LSTM-DNN, CLDNN) model are integrated for Health Performance Parameter (HPP) prediction. A classification method is constructed by the Generative Adversarial Imputation Networks (GAIN) model for health status estimation. The SHapley Additive exPlanations (SHAP) method is performed to quantify the relationships between energy consumption and system health status. A tunneling case in Singapore is used to test the effectiveness of the developed methodology. It is found that (1) the proposed method provides TBM cutter wear and cooling oil prediction with a R-square (R2) of 0.8591 and 0.8727, respectively, and (2) classification results indicate that the estimation F1-score of 0.9863 can be achieved based on health status evaluation indicator (HSEI). The novelty of the proposed approach lies in its capabilities of (1) fully extracting and utilizing the energy consumption information in every regard to predict TBM health performance; (2) providing a good one-step-ahead health status estimation for the TBM excavation process. Therefore, this study proposes a feasible direction for the application of energy consumption data in TBM health status estimation.

Prediction of geological composition using recurrent neural networks and shield tunnel boring machine data

ABSTRACT Tunnel Boring Machines (TBMs) are large-scale excavation tools used commonly in transportation tunnel construction. While tunnelling, TBMs generate data at large scales, often at levels difficult to parse using traditional statistical techniques. Utilising this data can be highly beneficial to obtain a better understanding of TBM excavation conditions, and such understanding can be applied to machine and technique selection. This paper presents a novel method for sequential estimation of geological composition using advanced machine learning algorithms and the data collected from TBMs. In this approach, we use Recurrent Neural Networks and Long Short-Term Memory (RNN-LSTM) models as a hybrid machine learning algorithm for processing sequential and time-series data. The results from an excavation case study demonstrate that the proposed method is an effective approach for sequential estimation of geological composition as encountered by TBM during operation. It is worth noting that TBM data captures the signature of the ground, and the developed model in this study was successful in predicting the ground geological composition even without using the borehole data.

Analysis of rock cuttability based on excavation parameters of TBM

Rock cuttability has great influence on the rock excavation efficiency of TBM (tunnel boring machine). In order to evaluate rock cuttability in real time, quickly, accurately and efficiently during TBM excavating, the relevant excavation parameters of Zagros, Kerman and Bazideraz tunnels were first collected. Then, the regression analyses between excavation parameters and rock cuttability were carried out. The two-dimensional regression analyses studied the relationship between operating parameters (thrust F and rotation speed RPM) and the characterization parameters (torque T and penetration rate PR). The three-dimensional regression analyses were utilized to create the PR and specific energy SE models based on operating parameters. The result shows that the established three-dimensional regression models have good prediction performance, and its performance is superior to two-dimensional models. Moreover, the prediction model of uniaxial compressive strength UCS and the classification model of rock cuttability were founded based on SE. The rock cuttability is divided into three levels, namely, easy (level 1), medium (level 2), and poor (level 3), in which the corresponding SE ranges are 0 to 6, 6 to 10 and exceeds 10 kWh·m−3, respectively. Finally, the intelligent algorithms, combined with excavation parameters, were introduced to establish UCS prediction model and rock cuttability classification model, and the good prediction performance was achieved. The above studies can provide necessary references and ideas for real-time, rapid, accurate and effective evaluation of rock cuttability based on TBM excavation parameters, and has certain guiding significance for engineering application.

Key Technology of TBM Excavation in Soft and Broken Surrounding Rock

Tunnel Boring Machine (TBM), as the most advanced construction equipment for rock tunnels, has been widely used in railway tunnel engineering, water conservancy tunnel engineering, urban rail engineering, coal mine roadway engineering, and other fields. However, the disadvantage of TBM lies in its poor adaptability to complex geological conditions, resulting in a significant reduction in tunneling efficiency and even major accidents due to improper TBM construction measures, especially in weak and fractured strata, which pose a great challenge to TBM construction. Therefore, it is necessary to study the key technologies of TBM construction in weak and fractured strata in order to improve the efficiency of TBM construction. Based on the background of the TBM construction of the Gaoligongshan Tunnel on the Southwest Darui Railway, this paper provides a reference for safe and efficient TBM excavation in weak and fractured areas through research on equipment, excavation parameter control, and construction measures.

Time series clustering-enabled geological condition perception in tunnel boring machine excavation

For automatic geology perception, this paper develops a DTW-Kmedoids-enabled time series clustering framework to gain a clear insight into the geological condition ahead of the tunnel boring machine (TBM) head. The core idea of DTW-Kmedoids is to incorporate dynamic time warping (DTW) as a distance metric into a popular clustering algorithm Kmedoids, aiming to aggregate time series data reflecting similar conditions of surrounding geology into the same cluster. It has been verified in a sequence dataset recording 377 operational rings from a TBM tunneling project in Singapore. Results indicate that the proposed DTW-Kmedoids algorithm outperforms Kmedoids and softDTW-Kmedoids in partitioning the whole dataset into four meaningful clusters on behalf of four geological conditions. Moreover, it shows merits in great robustness to handle varying-length temporal sequences and even fragmentary data instead of the full observations. Furthermore, an online clustering mechanism is well designed to adaptively update the cluster center, which can accomplish the dynamic recognition of the forward geological circumstances ring by ring. In short, the practical value behind the hybrid time series clustering is to automatically and quickly perceive lithology categories along the designed tunnel trail with no requirement of data labeling, possibly allowing for further inference of geological risk and guidance in safe excavation.

Solving problem of encountering a huge cavern in a TBM drive for water diversion tunnel project in Northern Thailand

AbstractA huge cavern in a zone of thrusted and sheared clastic sedimentary rocks was encountered during drill and blast (D&B) and tunnel boring machine (TBM) excavation of water transfer tunnel in Thailand. The cavern floor made of thick and loose pile of fallen rock blocks hampered the advance of the TBM through it. Because the size of the cavern was large (50 m × 40 m) and the extent of the thickness of the pile of rock blocks could not be precisely determined, the adopted ramification was to solidify a zone of loose rock fragments around the tunnel alignment. This measure is to allow a safe TBM excavation passing the cavern and ensure the long‐term function of the segmented concrete‐lined tunnel to convey water without problem of loss from subsidence damage. The scheme of the solidification consisted of staged grouting that began with polyurethane (PU) foam grouting to form a bottom barrier of the intended cement grout zone. It was followed large void filling by flowable concrete and then cement injection for the remaining smaller voids. The work that took 8 months to complete started with the excavation of a bypass adit to the front of TBM cutterhead as an access to the cavern and the installation of dewatering and ventilation systems to the cavern so that the grouting work could be made.

TBM performance prediction using LSTM-based hybrid neural network model: Case study of Baimang River tunnel project in Shenzhen, China

Accurately predicting tunnel boring machine (TBM) performance is beneficial for excavation efficiency enhancement and risk mitigation of TBM tunneling. In this paper, we develop a long short-term memory (LSTM) based hybrid intelligent model to predict two key TBM performance parameters (advance rate and cutterhead torque). The model combines the LSTM, BN, Dropout and Dense layers to process the raw data and improve the fitting quality. The features, including the ground formation properties, tunnel route curvature, tunnel location and TBM operational parameters, are divided into historical/real-time time-varying parameters, time-invariant parameters and historical/real-time output prediction data. The effectiveness of the proposed model is verified based on a large monitoring database of the Baimang River Tunnel Project in Shenzhen, south China. We then discuss the influence of the prediction mode, neural network structure and time division interval length of historical data on the prediction accuracy. The significance evaluation of input features shows that the historical output prediction has the largest influence on the prediction accuracy, and the influence of ground properties is secondary. It is also found that the correlations between input features and the output prediction are coincident with their interrelationships with the ground properties and ease of TBM excavation. Finally, it is found that the prediction results are most affected by the total propulsion force followed by the rotation speed of the cutterhead. The established model can provide useful guidance for construction personnel to roughly grasp the possible TBM status from the prediction results when adjusting the operational parameters.

Development of a PLSR-BRT Model for Predicting the Performance of Tunnel Boring Machines

Scientific prediction of the penetration rate (PR) of tunnel boring machines (TBMs) is of great significance to the selection of tunnel construction methods, the prediction of construction progress, and the evaluation of benefits. In view of the high nonlinearity, fuzziness, and complexity of the TBM excavation process, this paper attempts to present a new algorithm that integrates partial least squares regression (PLSR) with boosted regression trees (BRT) for predicting the PR of TBMs. For this purpose, 150 datasets were obtained from the Lanzhou water conveyance tunnel project in China. Six impact factors were selected as the input layers of the proposed model by single-factor correlation analysis. To develop the PLSR-BRT model, two principal components of the influencing parameters were extracted via the PLSR method, and the BRT algorithm was employed to establish the predictive model. In addition, other models such as PLSR back propagation neural network (BPNN), BRT, PLSR, support vector regression (SVR), and artificial neural network (ANN) were adapted for use in this problem to verify the predictive accuracy of the proposed model. The results demonstrated that the PLSR-BRT model achieved the best predictive performance compared with the other models, with coefficient of determination value (R2) and root mean square error (RMSE) of 0.96 and 1.78, respectively, for the testing set. The PLSR-BRT model can maximally avoid both overfitting and inadequate fitting. In view of engineering applications, the complexity and redundancy of the field dataset can be solved via reducing the dimensionality of input parameters, and the training samples could be expanded by a boosted method so that the discontinuously acquired geological parameters are unable to restrain the model performance. This method serves as an effective approach for TBM PR prediction and has excellent potential for a variety of scientific and engineering applications.

Analysis of TBM parameters during tunnelling in adverse conditions – The case study of the Frejus Safety Tunnel, Western Alps

AbstractOne of the main hazards during tunnel boring machine (TBM) excavation in hard rock is related to the occurrence of unforeseen adverse conditions, leading to unfavourable rock mass behaviour during tunnel boring. In this respect, the continuous analyses of TBM excavation parameters may represent an interesting tool for a continuous monitoring of geotechnical conditions at the tunnel face and for early detection of changing ground conditions in the rock mass. This work provides a back‐analysis of TBM data recorded during the excavation of the Frejus Safety Tunnel, bored across the French‐Italian border in the Western Alps. It is a 13 km‐long tunnel excavated by a 9.4 m diameter, single‐shield TBM through strongly anisotropic calcareous schist, with cover up to 1800 m. Main problems faced during TBM excavation were related to the local occurrence of buckling‐related convergence and consequent asymmetrical loading on the precast segment rings. Correlating TBM performance parameters to the encountered geological and geomechanical conditions provides the possibility to identify which parameters are most affected by geomechanical properties of the rock mass at the tunnel face and by rock mass behaviour. Conversely, the analysis indicates which parameters can be adopted as monitoring tool of ground conditions during TBM advance. In addition, this work highlights how buckling phenomena can be distinguished, in terms of TBM operation and performance, from other adverse conditions as, e.g., blocky ground or squeezing.

The application of the guideline for the geotechnical design of underground structures with TBM excavation in Austria

AbstractIn 2013, the Austrian Society for Geomechanics published the Guideline for the Geotechnical Design of Underground Structures with TBM Excavation. The Guideline contains general procedures for the geotechnical design and construction. In the design phase, the first steps are evaluating different ground conditions by classifying different ground types, different ground behaviours, their categorization into ground behaviour types, and the determination of construction measures derived from the ground behaviour. On this basis the expected system behaviour in each system sector of the tunnel boring machine (TBM) is predicted and described in detail to enable comparison in the construction phase. In the construction phase, geotechnical relevant ground parameters as well as observations regarding the actual system behaviour are recorded and compared with the assumptions of the design. The geotechnical design is updated continuously based on the encountered ground conditions and geotechnical observations. The improved quality of the geotechnical model allows an optimization of the construction while meeting all safety and environmental requirements. In the first part of the article, the general procedures of the guideline are described and parameters to describe the system behaviour in each system sector adequately are listed. In the second part of the article, the implementation of the guideline during design and construction at the Semmering Basetunnel in Austria is described.