Beneath the planned alignment of a new highway in Central Java, Indonesia, engineers face a problem that reads like a riddle from the frontiers of civil engineering: how do you carve two colossal tunnels, each spanning 20.3 meters and standing 13.8 meters tall, directly beneath existing streets and residential housing, through soft volcanic soil that can barely hold its own shape? Each tube, designed for three lanes of traffic, will stretch 385 to 405 meters long with a cross-sectional area of 224 square meters, excavated under a maximum overburden of just 32 meters. These dimensions exceed anything previously built in Indonesia, and the ground above the tunnels is not empty terrain but a living urban fabric of roads and homes where even modest settlement could crack foundations and disrupt daily life. The answer, according to a new study, lies in a carefully engineered marriage of intensive geotechnical investigation and a sweeping analysis of tunneling practices drawn from around the world.
Super large-span tunnels, generally defined as those with spans between 18 and 24 meters and cross-sectional areas of 200 to 300 square meters, have become essential infrastructure for highways, high-speed railways, and subway stations across the globe. The Liantang Tunnel in China, part of the Shenzhen Eastern Transit Expressway, reaches excavation dimensions of 30 by 18.4 meters with a maximum area of 428 square meters. Los Angeles hosts a crossover cavern 17.7 meters wide and 11 meters high in the Regional Connector Transit Corridor. Teixonera Station in Spain and Los Leones Station in Chile measure 20.3 by 13.5 meters and 18.9 by 11.8 meters respectively, while the Stepney Green Caverns in the UK span 17 meters wide and 14 meters high. Australia contributes the 24-meter Eastern Distributor in Sydney and the famous Y-junction caverns at Shafston, Kedron, and Lutwyche, the last reaching a remarkable 27-meter span. Each of these projects confronted unique geology, and each offers lessons for the Indonesian tunnels now on the drawing board.
The construction method at the heart of such projects is the sequential excavation method, or SEM, which divides the enormous tunnel cross-section into smaller segments excavated in sequence over short round lengths. This fragmentation makes an otherwise unmanageable excavation controllable, allowing support to be installed almost immediately after each segment is opened. Three SEM variants dominate worldwide practice: the top-heading and bench or full-face method (THB), the central or cross diaphragm method (CD), and the side-drift method (SD). Selection traditionally depends on the excavation span and the ground competence factor, defined as the ratio of ground compressive strength to in-situ stress. China has developed its own variations, including the three-bench method for high-quality rock and the three-bench seven-step excavation method, specifically devised for super large-span high-speed rail tunnels in poor-quality ground such as loess. In soft ground, where the stability of a freshly excavated face may last only hours, immediate support and continuous convergence monitoring become matters of survival rather than preference.
The geotechnical investigation behind the Indonesian project was correspondingly exhaustive. Between January and June 2022, engineers conducted 40 geotechnical drillings spaced 25 to 30 meters apart, totaling 3012 meters of depth, alongside 30 seismic refraction lines stretching 3450 meters and 14 electrical resistivity lines covering 5000 meters. Fourteen drill holes hosted seismic downhole tests reaching 40 meters deep, and 32 pressuremeter tests at depths of 10 to 42 meters determined the deformation modulus of the ground. Cone penetration tests with pore pressure measurement reached 25 to 30 meters at three locations, standard penetration tests were run at 2-meter intervals in every hole, and two trial groutings were carried out near the tunnel portals. Laboratory testing added physical and mechanical property evaluations, consolidation tests, swelling pressure tests, X-ray diffraction, X-ray fluorescence, and scanning electron microscopy. The shear wave velocity of the clayey silt was found to be 200 meters per second up to 30 meters depth, rising to 332 meters per second at 42 meters, with competent bedrock estimated at 760 meters per second some 150 to 160 meters below the surface.
What the surveys revealed was ground that demands respect. At tunnel depth, the tunnels will pass through volcanic tuff material weathered into clay and silty sand, predominantly clayey silt with stiff to very stiff consistency, characterized by standard penetration values of 11 to 22. The undrained shear strength ranges from 68 to 130 kilopascals and the compressive strength from 140 to 260 kilopascals, yielding a ground competence factor of just 0.02 to 0.04, a figure that signals soft ground and complex stress-strain conditions with likely unstable tunnel face behavior. The groundwater table sits below the tunnel invert, an advantage, but the ground permeability of roughly 10 to the power of negative 7 meters per second proved too low for grout injection to penetrate effectively. On-site trial grouting demonstrated that injected material could not be controlled along the length of a grouting pipe, ruling out conventional pipe roof support. Laboratory analysis, meanwhile, confirmed the clay exhibits no swelling behavior: X-ray diffraction found none of the clay minerals commonly linked to swelling, such as montmorillonite, smectite, or bentonite, and measured swelling pressures ranged only from 0.5 to 6.4 kilopascals.
To place these findings in context, the researchers assembled a database of 75 tunnel construction methods with spans of 15 to 27 meters from 13 countries, including 41 highway tunnels, 6 high-speed rail tunnels, 26 railway tunnels, and 2 hydropower tunnels, sourced from journal articles, conference proceedings, and technical reports. When these projects were plotted against established SEM selection charts, which rely on the ground competence factor and excavation span, a striking pattern emerged: global practice does not strictly follow the empirical guidelines. For super large-span tunnels with spans of 18 to 24 meters and competence factors below 10, both the multi-graph approach and the SEM selection chart recommend the side-drift method, and the Indonesian tunnels fall squarely in this region. Yet contractors worldwide have frequently adopted the CD and THB methods in the SD region, and vice versa, apparently preferring to compensate with robust ground support measures rather than adhere rigidly to classified construction methods. The majority of tunnels with competence factors below 1 and depth-to-span ratios below 2 were built using the CD and SD methods, suggesting genuine flexibility in method selection.
The global database also illuminated typical ground support for tunnels of comparable span and strength. Shotcrete thicknesses of 30 to 35 centimeters, steel arch spacing of 0.4 to 0.6 meters, and rockbolt lengths of 4 to 6 meters are commonly used, with rockbolts typically 25 to 32 millimeters in diameter. Forepoling, where employed, uses lengths of 8 to 12 meters and diameters of 63 to 89 millimeters, but the researchers judged it unsuitable here: the ground is both too weak and too fine-grained for forepoles to be effective, creating a significant risk of collapse beneath and between the forepoles. Final lining thicknesses of 60 to 70 centimeters are common in Class V ground, compared with 40 to 50 centimeters in Class II to IV conditions. Drawing on these benchmarks, the study identified the central diaphragm method as the most suitable choice for the Indonesian twin tunnels, being less complex and requiring fewer excavation segments than the side-drift method, which should reduce both construction costs and overall project duration.
The proposed design incorporates several innovations for Indonesian tunneling practice. Pre-confinement will consist of 53 sub-horizontal jet grouting columns, each 60 centimeters in diameter and 18 meters long, inclined at 6 degrees with 0.5-meter circumferential spacing and 12-meter longitudinal spacing, producing a 6-meter overlap between successive columns that forms a protective umbrella ahead of the face. Fiberglass dowels in the form of GFRP pipes will reinforce the core face, preventing the large, uncontrolled face extrusion that can destabilize unsupported soft ground. The initial lining combines 4-bar butterfly lattice girders at 1-meter spacing, a double-layer wire mesh system, and 30-centimeter fiber-reinforced shotcrete, supplemented by 6-meter underpinning micropiles welded to the girders for elephant foot-lock reinforcement. Lattice girders were chosen over steel arches because the risk of shotcrete shadowing, meaning voids and inadequate bonding, is much higher for steel arches, and standard steel sections would require heavy bolting and welding that slow assembly and raise safety risks. The excavation sequence adopts large 10-meter segments with a short 12-meter lagged distance between neighboring drifts, and ring closure through invert completion within one to two tunnel diameters of the furthest face, supporting an expected advance rate of 1 meter per day.
To verify the design, the researchers built a three-dimensional finite element model measuring 90 meters deep, 200 meters wide, and 42 meters along the excavation direction, using 268,731 elements with higher-order tetrahedral elements concentrated near the tunnel boundary where stress and displacement gradients are steepest. The ground was represented by the hardening soil model, known for effectively simulating stiffness and strength in sands, clays, and silts. The results were reassuring on every measure. Roof settlement and horizontal convergence remained around 0.8 to 1.4 centimeters after both tunnels were fully excavated, far below the allowable limits of 6 to 12 centimeters specified by Chinese transport ministry standards and up to 10 centimeters under Japanese Society of Civil Engineers criteria for three-lane highway tunnels in soft ground. The model predicted displacements consistent with monitoring data from the global database, including roof settlements of 1 to 1.5 centimeters and tunnel strains of roughly 0.1 percent. Surface settlement measured 0.5 centimeters after the first tunnel was completed, rising to 0.8 centimeters after the second, remaining below the 1-centimeter allowable limit with differential settlement of 1 in 500 for category 1 structures under national and international standards. Induced major principal stresses of 1.19 to 1.26 megapascals, against an in-situ stress of 640 kilopascals, indicated that the jet grouting columns artificially create a localized arch effect near the excavation perimeter, allowing the ground to redistribute stress more effectively and reducing the risk of face instability. Structural analysis confirmed that all internal forces in the composite lining segments stayed within the capacity envelope, with compression dominating after full excavation. Although the twin tunnels remain in the planning phase and field validation must await construction, the study’s methodology of integrating geotechnical investigation with global practice offers a template that extends well beyond Indonesia, as the researchers continue collecting data on tunnel span, ground strength, construction methods, and ground supports with the aim of developing universal guidelines for tunnel construction.
Subject of Research: Sequential excavation method design for super large-span soft-ground tunnels
Article Title: Designing sequential excavation method for super large-span tunnels: Integration of geotechnical investigation and global tunneling practices
Article References: Prassetyo, S. H., Lim, E., Basarah, Y. I., Kartiko, R. D., Hendriyawan, Lim, A., Aditya, F., Fahrudinsyah, M. H., Tuwan, G. A., Mulyadi, S. S., Wattimena, R. K., Winarsa, A. J. D., Danubroto, R. W., & Wisanggeni, T. (2026). Designing sequential excavation method for super large-span tunnels: Integration of geotechnical investigation and global tunneling practices. Results in Engineering, 32, Article 113669. https://doi.org/10.1016/j.rineng.2026.113669
Image Credits: AI Generated
DOI: 10.1016/j.rineng.2026.113669
Keywords: tunnel engineering, sequential excavation method, geotechnical investigation, soft ground tunneling, central diaphragm method, jet grouting, numerical modeling, surface settlement, Indonesia, infrastructure, Designing, sequential
