Article

Discussion: Machine learning to inform tunnelling operations: recent advances and future trends

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Therefore, evaluating the settlement hazards induced by tunneling is recommended since they have high significance for the long-and short-term performance of the tunnel. The effects of tunneling on the ground surface have been studied by several researchers [2][3][4]. However, their conclusions have been marred by uncertainty since it is unclear how the various elements affect the tunneling process. ...
... The NNW model showed good performance with 90% and 88% accuracy for models (1) and (2), respectively, as shown in Figures 11 and 12. Furthermore, there is no difference in the performance of the two prediction methods. However, the ANN models outperformed the MLR model in representing the interaction between factors, making the ANN models more valuable as a prediction tool, which corresponded well with the findings of Kaczmarek [43], Mohammadi et al. [17], and Sheil et al. [4]. In comparison, Chen et al. [18] predict the maximum surface settlement caused by EPB shield tunneling using three methods of artificial neural networks, the backpropagation neural network, the radial basis function neural network, and the general regression neural network (GRNN). ...
Article
Full-text available
Nowadays, the need for subway tunnels has increased considerably with urbanization and population growth in order to facilitate movements. In urban areas, subway tunnels are excavated in shallow depths under densely populated areas and soft ground. Its associated hazards include poor ground conditions and surface settlement induced by tunneling. Various sophisticated variables influence the settlement of the ground surface caused by tunneling. The shield machine's operational parameters are critical due to the complexity of shield-soil interactions, tunnel geometry, and local geological parameters. Since all elements appear to have some effect on tunneling-induced settlement, none stand out as particularly significant; it might be challenging to identify the most important ones. This paper presents a new model of an artificial neural network (ANN) based on the partial dependency approach (PDA) to optimize the lack of explainability of ANN models and evaluate the sensitivity of the model response to tunneling parameters for the prediction of ground surface and subsurface settlement. For this purpose, 239 and 104 points for monitoring surface and subsurface settlement, respectively, were obtained from line Y, the west bond of Crossrail tunnels in London. The parameters of the ground surface, the trough, and the tunnel boring machine (TBM) were used to categorize the 12 potential input parameters that could impact the maximum settlement induced by tunneling. An ANN model and a standard statistical model of multiple linear regression (MLR) were also used to show the capabilities of the ANN model based on PDA in displaying the parameter's interaction impact. Performance indicators such as the correlation coefficient (R2), root mean square error (RMSE), and t-test were generated to measure the prediction performance of the described models. According to the results, geotechnical engineers in general practice should attend closely to index properties to reduce the geotechnical risks related to tunneling-induced ground settlement. The results revealed that the interaction of two parameters that have different effects on the target parameter could change the overall impact of the entire model. Remarkably, the interaction between tunneling parameters was observed more precisely in the subsurface zone than in the surface zone. The comparison results also indicated that the proposed PDA-ANN model is more reliable than the ANN and MLR models in presenting the parameter interaction impact. It can be further applied to establish multivariate models that consider multiple parameters in a single model, better capturing the correlation among different parameters, leading to more realistic demand and reliable ground settlement assessments. This study will benefit underground excavation projects; the experts could make recommendations on the criteria for settlement control and controlling the tunneling parameters based on predicted results. Doi: 10.28991/CEJ-2022-08-11-05 Full Text: PDF
... This hampers accurate prediction of jacking force requirements. An exclusive focus on the retrospective analysis of pipe-jacking drives as a means of improving our understanding of the relationship between skin friction and stoppage duration may have limitations, for reasons summarised by Sheil et al. (2020). The calibration of load cells used in conjunction with jacks in the launch shaft and the steering cylinders (and at intermediate jacking points, where present) is not routinely checked on working sites. ...
Conference Paper
Full-text available
In long pipe-jacking drives used for installing utility pipelines, maximum jacking load requirements are usually governed by skin friction at the pipe-soil interface. In addition, field experience has shown that transient peaks in skin friction arise upon recommencement of jacking after stoppages; these stoppage durations can be short (due to the addition of a pipe to the string) or long (due to weekend stoppages or breakdowns) and constitute a risk for pipe-jacking contractors. In this paper, the problem is replicated in the laboratory using direct shear interface tests using a concrete specimen in one half of the apparatus and sand/bentonite mixtures in the other. Once critical state conditions were reached in these tests, stoppages of various durations (from 30 mins up to 2 weeks) were incorporated and the increase in shear stress upon recommencement of shearing was noted. From the experiments, there appears to be a threshold stoppage duration beyond which the skin friction increase appears to plateau, suggestive of a time-limited process within the bentonite. These skin friction data are shown to provide an upper bound to corresponding stoppage data from pipe-jacking drives in sandy ground conditions.
... Therefore, the topic considered in the paper will be of interest to tunnelling contractors aiming to reduce risks associated with excessive jacking forces. It is encouraging to see a laboratory modelling approach used by the authors, given that the calibration of load cells and other sensors are not routinely checked on working pipe-jacking sites, with implications for the reliability of drive data (Sheil et al., 2020). ...
Article
Full-text available
Jacking forces which exceed expectations constitute a risk for tunnelling contractors. One scenario in which high forces may arise is when jacking of lubricated pipes is temporarily halted, which was considered by Li et al. using a programme of direct shear testing. While recognising the importance of the topic to the profession, the purpose of this Discussion piece is to highlight some of the limitations of the study.
Article
Full-text available
Forecasting geological cross-section ahead of tunnel face is an essential ingredient for tunnel design and construction. Geological analysis and drilling have been the most traditional approach for predicting tunnel ahead geological conditions. However, this practice is often subjective, and geological information retrieved from previous tunnel excavation in the same project has not been used quantitatively. In this study, a data-driven framework is proposed to sequentially develop geological cross-sections along planned tunnel trajectory conditioning on site-specific data and prior geological knowledge. The proposed framework dynamically and continuously incorporates geological information revealed from tunnel excavation as additional site-specific data, which provide first-hand direct geological information from the immediate past tunnel sections and can actually serve as the most relevant prior geological knowledge for forward prediction. All the prior geological knowledge is compiled as a site-specific training image database. When the actual geological cross-sections are revealed from tunnel excavation, the training image database is also updated for the next loop of tunnel ahead geological prediction. The proposed method is illustrated using data obtained from a real tunnelling project in Australia. Results indicate that the proposed method continuously provides accurate prediction of geological cross-sections along planned tunnel trajectory with quantified stratigraphic uncertainty.
Article
Machine Learning (ML) and Building Information Modelling (BIM) are two topics that are part of a revolutionizing transformation in the construction industry – commonly referred to as digitalization. Being part of the research for artificial intelligence (AI), most of today's ML applications deal with computational processes that try to make sense of data. Automatic rockmass behaviour classification based on tunnel boring machine (TBM) data or tunnel construction site surveillance via closed‐circuit television (CCTV) analysis is an example for applications of ML in tunnelling. BIM describes a new type of planning, including model‐based collaboration and information exchange, which requires well‐organized storage and handling of data – a precondition and valuable source for any automated analysis method like ML. While other sectors of the construction industry have implemented BIM systems successfully, the development in underground engineering is currently at its beginning with multiple actors working towards common standards for semantics, data exchange formats, etc. This article seeks to combine the two fields by giving an overview of the two topics and then points out four potential fields of applications: semantic enrichment and labelling, automation of technical processes, knowledge derivation and online data analysis.
Conference Paper
This paper presents an implementation of Random Forest (RF), a supervised learning algorithm, to classify the encountered geologic conditions using continuous earth pressure balance machine (EPBM) operation data. This study was performed on a data set from State Route 99 (SR99) tunnel construction in Seattle, WA. Hyperparameter tuning was performed to investigate the effects of RF hyperparameters on the classification performance as well as to determine the best hyperparameter configuration. The role of the features in the classification model was investigated by evaluating the feature importance measures. This study demonstrates that, with straightforward hyperparameter tuning, RF could deliver good classification performance and could infer the geologic transition through the classification probabilities. This study indicates that although several EPBM features had relatively larger “weights” for the classification, it was the interactions among the features that contain the geologic information.
Article
The use of supporting fluids to stabilise excavations is a common technique adopted in the construction industry. Rapid detection of incipient collapse for deep excavations and timely decision making are crucial to ensure safety during construction. This paper explores a hybrid framework for forecasting the collapse of fluid−supported circular excavations by combining physics-based and data-driven modelling. Finite element limit analysis is first used to develop a numerical database of stability numbers for both unsupported and fluid−supported circular excavations. The parameters considered in the modelling include excavation geometry, soil strength profile and support fluid properties. A data-driven algorithm is used to ‘learn’ the numerical results to develop a fast ‘surrogate’ amenable for integration within real−time monitoring systems. By way of example, the proposed forecasting strategy is retrospectively applied to a recent field monitoring case history where the observational method is used to update the input parameters of the data-driven surrogate.
Article
Full-text available
This paper provides a review of the history and development of pipe jacking and microtunneling methods with extensive referencing to the published literature. The application of such methods in comparison with other trenchless technologies is discussed and the various planning, design and construction aspects are introduced. The emphasis of the paper is to trace the academic research and field monitoring results covering critical aspects of design and construction with a particular emphasis on jacking force estimation and the effect of lubrication on jacking forces.
Article
Microtunnelling is an increasingly popular means of locating utilities below ground. The ability to predict the total jacking force requirements during a drive is highly desirable for anomaly detection, to ensure the available thrust is not exceeded, and to prevent damage to the pipe string and/or launch shaft. However, prediction of the total jacking force is complicated by site geology, the use of a lubricated overcut, work stoppages, tunnel boring machine driving style and pipe misalignment. This paper introduces a probabilistic observational approach for forecasting jacking forces during microtunnelling. Gaussian process regression is adopted for this purpose which allows forecasts to be performed within a probabilistic framework. The proposed approach is applied to two recent UK microtunnelling monitoring projects and the forecasts are appraised through comparisons to predictions determined using design methods currently applied in industry. The results show that the proposed framework provides excellent forecasts of the monitored field data and highlights a significant opportunity to complement existing prescriptive design methods with probabilistic forecasting techniques.
Article
The proliferation of data collected by modern tunnel boring machines (TBMs) presents a substantial opportunity for the application of machine learning (ML) to support the decision-making process on site with timely and meaningful information. The observational method is now well-established in geotechnical engineering and has a proven potential to save time and money relative to conventional design. ML advances the traditional observational method by employing data analysis and pattern recognition techniques, predicated on the assumption of the presence of enough data to describe the modelled system’s physics. This paper presents a comprehensive review of recent advances and applications of ML to inform tunnelling construction operations with a view to increasing their potential for uptake by industry practitioners. This review has identified four main applications of machine learning to inform tunnelling, namely TBM performance prediction, tunnelling-induced settlement prediction, geological forecasting and cutterhead design optimisation. The paper concludes by summarising research trends and suggesting directions for future research for ML in the tunnelling space.
Article
In recent years, there has been an increased resort to microtunnelling/pipe-jacking as a means of constructing underground conduits (for water, sewage, gas, and other utilities) to avoid on-street disruption in urban areas. In this paper, technical details of two 1 200 mm internal diameter microtunnels in silty sand totalling over 550 m in length are discussed. While average skin friction values are extremely low for both drives suggesting effective lubrication practice, differences in normalised bentonite volumes appear to be responsible for differences in skin friction. Full or near full buoyancy of the pipeline has been demonstrated for the majority of the drive. The frictional stress increase after a stoppage is shown to depend on stoppage duration but also on the normalised lubriation volume. Interpretation of data in the manner presented in the paper is an important means of assimilating experience of microtunnelling in different ground conditions.
Article
Glacial till (or boulder clay) is the most widespread sediment on the island of Ireland. The behaviour of these tills, especially Dublin Boulder Clay, is now better understood as a result of ground investigations and associated testing for major construction projects, particularly in the east of the country over the past decade. Despite an increase in tunnelling activity in the country over the same time period, there is very little documented evidence on the settlements induced by tunnelling operations in glacial till. In this paper, transverse surface settlement data from two glacial till sites are presented. Four profiles are presented for the ‘soft ground’ TBM-bored section of the Dublin Port Tunnel. Nine profiles are reported for pipe-jacked microtunnels constructed for a sewerage scheme in Mullingar in the Midlands; one of which was measured at the top of a railway embankment under which the pipeline passes. The measured settlements have been interpreted using a standard Gaussian error function, and trough width parameters show dependence on the fraction of the till, fine or coarse, that governs its behaviour. In addition, conservative design estimates of maximum trough settlement and volume loss have been provided, and the impact of boulders is discussed. This paper provides the first empirical guidance for predicting surface settlements above tunnels in Irish soils.
Article
Friction forces usually constitute the main component of jacking loads. As a result of their increase with jacking length, it is these forces which limit the drive length. Therefore, it seems important to be able to quantify them accurately. The field monitorings, carried out as a part of the French National Project ‘Microtunnels’, have shown the effects of parameters such as lubrication, stoppages, deviation and overbreak on friction force values. Frictional stress, deduced from field monitoring, is compared with empirical results extracted from the literature and with the results of most frequently used calculation models.