Tunnel and Underground SpaceISSN:1225-1275(Print) 2287-1748(Online)한국암반공학회

Borehole deviation refers to the degree of curvature that occurs during actual drilling relative to the intended drilling direction. Since borehole deviation affects the accuracy of deep geological repository design for high-level radioactive waste, precise measurement and proper management are required. This article introduces the types and principles of instruments used for borehole deviation measurement both domestically and internationally, presents case studies from Finland and Sweden, and discusses the causes of borehole deviation and methods for its reassessment. In addition, borehole deviation results obtained from boreholes drilled at depths of approximately 750 m in various rock types during site study for a domestic high-level radioactive waste repository are presented. Most deep boreholes exhibited deviations of around 1°, with relatively larger deviations observed in rock types with stratification, foliation, or schistosity. However, in gneissic rock boreholes where drilling equipment was set on poor ground conditions and multiple weak zones were encountered during drilling, deviations exceeding 7° occurred. The excessive deviation was attributed to the influence of large boulders in the ground where drilling equipment was installed and to frequent weak and fractured zones encountered during drilling. Based on domestic deep drilling experience and international case studies, potential strategies for mitigating and improving borehole deviation include multi-stage drilling, installation of protective casing, placement of concrete pads for drill rig positioning, use of horizontal and vertical measurement devices for drill rig alignment at the start or during drilling, maintaining appropriate drilling rates, and employing drill collars or stabilizers during drilling.

A co-simulation model integrating the multibody dynamics of a continuous excavation-type tunnel boring machine (TBM) with its hydraulic system was constructed. The proposed model enables the evaluation of the thrust jack behavior and control system performance within a virtual simulation environment. The coupled model employs a power-exchange feedback structure, in which pressure variations in the thrust jacks are reflected in the dynamic system, and the resulting velocity responses are fed back as loads to the hydraulic system. Through this approach, the load distribution, pressure response, and Center of Thrust (COT) control behavior of the thrust system were realized. Furthermore, the conventional group-controlled hydraulic circuit was improved by introducing an independent control architecture, in which each thrust jack is individually equipped with a dedicated pressure sensor. Due to the inclined surface of the helical segments, differences in the contact distance among the thrust jacks were observed, resulting in corresponding variations in the time-dependent pressure responses. By controlling the jack pressures according to the specified COT target angles, the effectiveness of the proposed COT control method was verified. The implemented multibody–hydraulic co-simulation model can be effectively utilized in the design phase of TBM hydraulic and control systems to validate the load distribution, hydraulic response, and control algorithm of the thrust module in advance.

This study defines autonomous loading systems and unmanned loading equipment in mining and reviews development trends and applications by major global manufacturers. Representative unmanned loaders and excavators are compared, and their automation maturity is interpreted using the GMG Mining Automation Maturity Model. Most deployed systems rely on teleoperation supported by LiDAR, camera, and GPS sensing for real-time localization, obstacle detection and avoidance, and safety functions, with Wi-Fi based communication for control and data transmission. Current field systems are largely Level 1 to 2, while only a few prototypes approach Level 3. Korean mine cases confirm improved continuous operation, higher productivity, better safety, and reduced costs.

Footage captured by on-site CCTV was labeled to construct a terrain-object training dataset, and the data were partitioned by time to evaluate the time-series nature of terrain changes. Specifically, the first two months were used for training and validation, and the subsequent two months were reserved for testing. Using a deep learning–based instance segmentation model, the experiments achieved stable detection of terrain objects with an mAP@0.5 of 90.6%, and further confirmed the feasibility of detecting future terrain changes from past image frames. These results demonstrate the potential of image-only approaches to recognize and monitor terrain changes at earthwork sites and lay the groundwork for extensions to long-term time-series forecasting and risk-aware alerting, positioning the method as a core enabling technology for safety monitoring in earthwork operations.

This study presents a systematic framework for defining inspection items and classifying damage types to facilitate the automation of tower crane visual inspections using drone-based imaging and artificial intelligence (AI)-based image analysis. Ten major structural components of tower cranes (foundation, mast, bracing, access platform, telescopic cage, turntable, operator’s cabin, counter jib, main jib, and hook block) were analyzed by integrating legal inspection standards, technical guidelines, and accident case studies to derive detailed inspection items. In addition, seven major categories of visual damage—cracks and fractures, corrosion and surface deterioration, deformation, fastening defects, wire rope and sheave damage, hydraulic and electrical failures, and safety device defects—were systematically classified. Quantitative criteria for each category were established to standardize AI training datasets and support the development of automated damage detection models. The proposed framework provides a technical foundation for constructing smart inspection platforms and advancing automation technologies for the safety assessment of construction equipment.

In this study, a conceptual design was carried out for implementing underground smart farms by utilizing the environmental characteristics of underground idle spaces. To achieve this, recent research trends on underground smart farms were analyzed to identify monitoring parameters necessary for optimizing future underground smart farm operations. Based on the identified monitoring parameters, an environmental monitoring system was designed, fabricated, and installed in the idle space of a room-and-pillar mine. Analysis of the initial data revealed that temperature, atmospheric pressure, carbon dioxide concentration, and carbon monoxide concentration satisfied the operating conditions for the three cultivation methods considered in this study. However, additional measures are deemed necessary for humidity and dust control. Based on these results, a modular underground smart farm unit was designed for application in the test space. A smart farm monitoring framework was also proposed to derive optimal growing conditions in conjunction with the underground environmental monitoring system established in this study. Future work will focus on selecting cultivation methods suitable for underground environments and deriving optimal growing conditions through the operation of actual underground smart farm units, based on long-term monitoring results from the system constructed in this study.

The Open TBM (Tunnel Boring Machine) is a mechanical excavation equipment primarily used in sound rock formations, mainly applied to mountain tunnels and waterway tunnels, but its utilization in urban construction projects has also been increasing recently. This study presents a method for developing and applying a directional control model for a main-beam type Open TBM to a simulator as part of the Open TBM simulator development. The directional control model for the Open TBM mathematically describes the horizontal and vertical rotation angles of the cutterhead using the stroke length of the cylinders installed on the gripper carrier. It also mathematically describes the forward stroke length of the thrust cylinder connecting the main beam and the gripper. To utilize the formulated directional control model in the Open TBM simulator, a calculation method for the cutterhead's rotation and movement based on control panel inputs was presented.

This study designed an underground mining system for a graphite deposit and developed a simulation model to evaluate the mining development process. The proposed model incorporates the orebody generated through 3D modeling and simulates three major subsystems: the underground haulage system, the underground crushing system, and the shipment system. A graphite deposit in Korea was selected as the study area, and a site-specific simulation model was constructed accordingly. Using the developed model, optimal equipment operation strategies suitable for the underground mining conditions of the study area were derived. The model implemented through 3D modeling provides a comprehensive representation of the underground mining system. The results of this study demonstrate the potential contribution of the proposed simulation model to improving operational efficiency in actual field applications.

As the backfill design for the Deep Geological Repository (DGR) shifts from conventional compacted block/pellet systems to granular bentonite in Finland, the world’s first country to receive an operating license for high-level radioactive waste, comprehensive data of these granular materials are required. Accordingly, this study presents a comprehensive analysis of the basic physical, chemical, hydro-mechanical properties of two materials, GraFi and GBM, which are being considered as granular backfill respectively in the Finnish and Swiss DGR programs. A total of 14 characteristic evaluations were performed, including particle size distribution, chemical composition and clay activity. The results confirmed that both materials are of the same sodium-type bentonite; however, they exhibited distinct differences originating from the source and manufacturing processes. While GraFi showed relatively higher values for all indicators representing clay activity compared to GBM, it was confirmed that GBM was produced with a higher dry density in terms of manufacturing properties. Additionally, the van Genuchten model parameters for the Soil-Water Characteristic Curve (SWCC) were derived for an installation dry density of 1.4 g/cm³. These fundamental property data will serve as input parameters for future hydro-mechanical models and as core basic data for the development of Korean granular backfill.

This study investigates the feasibility of predicting damage levels around underground openings under deep in-situ stress conditions by combining acoustic emission (AE) analysis with classification learning methods under biaxial loading. Biaxial compression tests were conducted on Berea and Idaho sandstone block specimens containing a central circular cavity, while applying different horizontal stress levels. Representative time- and frequency-domain AE parameters, including counts, energy, amplitude, and initiation frequency, were analyzed with respect to the axial loading history. Despite an approximately twofold difference in uniaxial compressive strength between the two sandstones, the evolution patterns of AE parameters were generally similar, and the influence of horizontal stress level on AE responses was found to be limited. This behavior is interpreted as the effect of lateral confinement, which suppresses microcrack opening and propagation and thereby reduces the sensitivity of AE responses to strength and boundary stress conditions under biaxial loading. A dataset comprising 24 input features—15 AE parameters, 8 mechanical properties, and 1 horizontal stress variable—and four damage levels defined by the ratio of axial load to peak load was then constructed for classification learning. Among 23 classification algorithms tested, the bagged trees algorithm achieved the highest prediction accuracy of 93.9%, with more than 80% classification performance for each damage level. These results demonstrate that the combination of AE features and classification learning methods can quantitatively determine damage levels around circular openings under biaxial compression, and they provide a useful basis for AE-based damage assessment systems for deep underground openings.

This study evaluated the dynamic Mode I fracture toughness of Yeongju granite and Hwangdeung granite using a Split Hopkinson pressure bar (SHPB) system. Using the Short rod (SR) method, Yeongju granite exhibited fracture toughness values approximately 1.08~1.16 times higher than those of Hwangdeung granite. For both rock types, fracture toughness increased with loading rate and followed a power-law relationship. Regression analysis further showed that the loading rate sensitivity of Yeongju granite was approximately 1.1 times greater than that of Hwangdeung granite. The anisotropy factor decreased from static to dynamic loading, indicating that material anisotropy has a relatively small influence on fracture toughness under dynamic loading. Moreover, the crack opening displacement (COD) and crack opening velocity (COV) time histories were consistent with high speed camera images of crack propagation, confirming the reproducibility of crack opening and fracture behavior in the dynamic SR test.