Suction-Driven Installation of a 20 m-Diameter Circular Steel Cofferdam: A Full-Scale Field Test in Jebudo, Republic of Korea

Abstract

Cofferdams provide dry, stable working conditions for construction in marine environments. However, conventional methods often require significant time and cost for installation and removal, and are prone to leakage. This study proposes a novel method for the rapid and efficient construction of a large-diameter circular cofferdam using suction-driven installation and extraction. As opposed to conventional suction bucket foundations, the upper part of the cofferdam remains exposed above the water surface, and several prefabricated segments are assembled to form a single suction unit. A full-scale field test was conducted in Jebudo, Republic of Korea, using a 20 m-diameter, 13 m-high circular steel cofferdam. The test program included the design and fabrication of a suction cover and an optimized piping system. The key measurements during installation included the suction pressure variation with the penetration depth, leakage at the segmental joints, structural deformations, and inclination. The cofferdam successfully penetrated to a target embedment depth of 5 m at an average rate of 1.83 m/h and was safely removed using reverse suction. Although suction technology has been widely applied to offshore foundations and anchors, this study is the first to demonstrate its feasibility for large cofferdams. These results provide a foundation for future offshore applications of suction-driven cofferdam installations.

1. Introduction

Cofferdams are essential temporary hydraulic structures that enable construction in marine and river environments by providing dry and stable working spaces. They are widely applied in bridge pier construction, offshore platforms, harbor works, and other infrastructures in which safe access to the seabed is required. Cofferdams improve constructability, structural quality, and worker safety by reducing the external water pressure and stabilizing the surrounding soils. Despite their importance, conventional cofferdam methods such as the sheet pile, cellular, and caisson types are associated with significant challenges, including long installation and removal durations, high costs, frequent leakage, and difficulty in maintaining verticality and stability under hydraulic and geotechnical loading [1,2,3,4].

Suction-driven installation has been increasingly considered as a promising alternative to overcome these drawbacks. Suction caisson foundations, which are widely applied in the offshore energy and oil and gas industries, provide rapid and quiet installation by creating a pressure differential between the inside and outside of a closed skirted structure. This method eliminates the need for pile driving or heavy excavation, reduces noise and vibration, and allows for reversible extraction by applying reverse suction. In recent decades, extensive research has addressed suction caisson foundations in sands, clays, and layered soils, focusing on the penetration behavior, bearing capacity, cyclic and combined loading, and long-term stability [5,6,7,8,9,10]. Design procedures have been proposed for both sand and clay conditions. Centrifuge model tests, laboratory studies, and full-scale field trials have supported the development of design guidelines that are currently widely adopted in offshore wind and deepwater anchoring projects [11,12,13,14,15,16].

However, the application of suction principles to cofferdams remains relatively unexplored. As opposed to the suction buckets used for offshore foundations, cofferdams are designed to create a dry construction environment by enclosing a soil volume that must be dewatered following installation. This introduces unique technical challenges because part of the cofferdam structure remains above the waterline, requiring the simultaneous removal of air and water to develop suction, and the segmental joints must remain watertight under both hydrostatic and suction loads. In addition, cofferdam walls and lids experience significant structural stresses during suction penetration, raising concerns regarding lid deformation, wall buckling, and bolt pretension losses. Laboratory experiments, numerical simulations, and small-scale model studies have been used to investigate aspects such as seepage behavior, suction-induced soil failure, connection leakages, and verticality control methods [17,18,19,20,21,22]. However, these studies have been limited in terms of the scale and conditions, and, to date, only a few field demonstrations at diameters up to 5 m have been reported [23,24,25].

However, large-diameter cofferdams present several additional challenges. For diameters exceeding 20 m, the suction lid is subjected to substantial compressive loads, and the segmental walls must resist both the hoop and axial stresses. The soil resistance increases with the penetration depth, as the tip resistance and shaft friction increase, requiring careful control of the suction pressure to avoid soil boiling in sands or plugging in clays. Moreover, achieving uniform penetration without tilting requires precise verticality control systems, which often involve lifting cables or hydraulic cylinders. These technical issues emphasize the need for full-scale field validation before suction-driven cofferdams can be adopted for major marine infrastructure projects [26,27,28,29,30].

In addition to these geotechnical considerations, structural and hydraulic stabilities have been studied in the context of cofferdam design. Numerical simulations have been conducted to analyze seepage discharge and piping safety factors, and model-scale tests have been performed to examine the deformation patterns under suction loading and hydrodynamic effects. Structural analyses have been conducted to evaluate the stiffened lid performance and wall stability, identifying the risks of local buckling and excessive deformation in large-diameter cofferdams. Experimental studies have also highlighted the role of joint performance, showing that leakage or bolt pretension loss can compromise both the penetration efficiency and dewatering performance [31,32,33,34,35,36]. Despite these advances, there remains a lack of comprehensive field-scale validation linking suction pressures, penetration rates, structural deformations, leakage, and overall constructability for cofferdams at realistic offshore project scales.

Recent developments in offshore wind energy, floating platforms, and large-span marine infrastructure have increased the urgency for exploring new cofferdam solutions. At present, offshore construction frequently demands temporary works that can be rapidly installed and removed with minimal noise, vibration, and environmental impact. Suction-assisted cofferdams have the potential to fulfill these requirements by offering fast and reversible installation, segmental constructability, and inherent circular stability. Demonstrating their feasibility for large diameters under real marine conditions is a necessary step toward developing reliable design standards and practical guidelines [36].

Therefore, this study fabricated a 20 m-diameter, 13 m high circular steel cofferdam composed of three prefabricated segments and a suction lid, and conducted a full-scale field installation and removal test in Jebudo, Republic of Korea. This work represents the first full-scale verification of suction-driven cofferdam technology, extending previous small-scale demonstrations ($\le$5 m diameter) to realistic offshore construction conditions.

In contrast to earlier field and laboratory studies that mainly explored small-scale penetration behavior, this study introduces new structural and hydraulic configurations that improve the constructability and performance. Specifically, the system employs lateral hose connections that are optimized for shallow-water operation. The top lid remains exposed above the waterline, and a segmented arched lid with a bolted-flange modular assembly, as well as ethylene-propylene diene monomer (EPDM) gaskets, is used to enhance watertightness and stability under high suction pressures. These innovations simplify on-site handling, reduce the construction time, and improve the sealing performance compared with conventional cofferdam methods.

Following penetration to the target depth, the cofferdam interior was dewatered to confirm its effectiveness in providing a dry construction environment, and the structure was safely extracted using reverse suction. The outcomes demonstrate not only the scalability and engineering efficiency of the proposed system, but also its potential for cost-effective and environmentally sustainable applications in offshore bridges, harbor structures, and renewable energy projects.

2. Steel Cofferdam Installation Process and Fabrication

2.1. Cofferdam Installation Process

Figure 1 depicts the overall construction process and concept of a steel pipe cofferdam. The steel pipe cofferdam method has several advantages: (1) the suction pressure enables rapid and efficient installation; (2) the cofferdam can be divided into multiple circular steel pipe segments to minimize the weight of each component; (3) the circular single-member system provides inherent structural stability; and (4) the cofferdam can be easily dismantled following use by reversing the suction pressure. These unique features render the method highly applicable to offshore and nearshore foundation works, in which stability, constructability, and removability are critical.

The stepwise construction sequence is illustrated in Figure 1. First, the cofferdam is installed through self-weight penetration and suction. The suction-driven installation mechanism of the cofferdam is conceptually similar to the bucket foundation technology, which has been widely applied in offshore wind and oil platforms [5,6,13]. Once the initial penetration is secured, the temporary top lid is removed, and additional circular modules are connected vertically to achieve the required height. This modular assembly minimizes the lifting weight of each segment and facilitates offshore handling. Subsequently, ground stabilization inside the cofferdam is performed through soil improvement or seabed preparation to ensure a stable working platform.

Following stabilization, the water that is trapped inside the cofferdam is pumped out, creating dry or semi-dry working conditions for foundation works. Subsequently, the structural foundation elements, such as piles or footings, are installed within the enclosed space. The vertical structural column is then constructed inside the cofferdam at the designated elevation. Finally, once the permanent foundation and column have been completed, the cofferdam is removed by reversing the suction pressure, which allows for efficient extraction without disturbing the installed foundation. This sequence highlights the dual functionality of suction pressure in both installation and removal, demonstrating the engineering efficiency of the method.

Figure 2 shows the detailed process of the self-weight penetration and connection of the steel pipes. This figure expands upon the modular assembly step introduced in Figure 1 and illustrates the fabrication, transportation, seabed preparation, penetration, and segment connection procedures in greater detail. The process starts with the fabrication of circular steel pipe segments and the top lid at an onshore facility (Figure 2a). These prefabricated units are then transported to a nearby dock or quay and prepared for offshore deployment.

During lifting operations, chokers are used to minimize eccentric or inclined loads that could otherwise induce deformation of the steel pipes (Figure 2b). Radial braces are installed inside the pipe segments to reduce the risk of distortion during transport further. These braces are temporary and are designed to be dismantled once the pipe has penetrated the seabed. Next, the prefabricated elements are loaded onto a barge and transported to the designated offshore construction site. The installation location is confirmed through hydrographic surveying, and divers inspect the seabed to ensure that large cobbles or boulders have been removed. If necessary, seabed leveling is conducted to prepare a stable bearing surface for the cofferdam foundation (Figure 2c).

Once the site has been prepared, the bottom pipe segment is lifted from the barge using a revolving crane equipped with a guide frame and lowered to its target position. The pipe then penetrates the seabed under its own weight, with crane operators carefully adjusting the wire tension to maintain verticality. The inclination is controlled to remain within 0.5°, either by precise crane adjustments or using automated control systems (Figure 2d).

Following the initial penetration, additional pipe segments are connected. The flange surface of the penetrated lower segment is cleaned, and sealing materials are inserted before positioning the upper segment. The upper segment is aligned so that the flange and bolt holes match precisely, aided by horizontal guide keys (Figure 2e). Workers then bolt the upper and lower segments together from a safety platform, working in pairs at diametrically opposite positions to ensure uniform tightening. Pneumatic or hydraulic impact wrenches are used to apply consistent tension across all bolts. Following the initial tightening, torque wrenches are used to achieve the target torque value through multiple passes, ensuring that all bolts carry a uniform load.

The same procedure is applied when connecting the top lid to the uppermost pipe segment (Figure 2f). Once all elements have been fully assembled, the bolt tension is re-checked and a final visual inspection is carried out to confirm that the joints are watertight and that no leakage can occur during subsequent suction penetration.

Figure 3 illustrates the process of suction installation and pile construction within the steel pipe cofferdam. This figure highlights the suction pump system, suction-driven penetration process, internal water removal, ground stabilization, and, ultimately, construction of structural foundations inside the cofferdam. The procedure starts with the arrangement of suction pumps on the barge (Figure 3a). The system is composed of a large pump that can draw seawater and a Venturi ejector equipped with both inlet and outlet connections. Piping is installed around the barge perimeter to allow seawater intake, and flexible hoses such as rubber or PVC (T-line) hoses are connected to the Venturi inlet. To facilitate handling, hoses are divided into 10–15 m segments and joined using quick-release couplings such as camlock fittings. The ends of the hoses are connected to the cofferdam inlet pipes and suction pump, with particular care taken to prevent air inflow at the joints. Depending on the cofferdam diameter and seabed conditions, multiple suction pumps can be operated simultaneously, and when necessary, additional support barges can be mobilized. As the cofferdam extends above the water surface and traps air internally, the suction pump initially fills the cofferdam with water to expel air through air vents in the top lid before suctioning begins.

Once the cofferdam is filled with water, suction pumping begins to discharge water from the interior to the surrounding sea. The pumping rate and suction pressure are carefully controlled to avoid exceeding the suction pressure determined during the penetration design, thereby preventing soil instability or failure at the seabed. Throughout the process, the verticality is continuously monitored to ensure that the cofferdam remains within a tolerance of 0.5°. If deviations occur, the crane wire tension is adjusted, or hydraulic cylinders that are placed between the lifting wire and cofferdam are used to correct the inclination during penetration. When installation is complete, the top lid is removed and placed back onto the barge (Figure 3b).

Following installation, the water inside the cofferdam is drained using submersible pumps and dewatering equipment (Figure 3c). Because the internal water level is lower than the external sea level, inflow through seepage may occur, particularly in soft soils where ground instability and piping failure are concerns. To counter this, a layer of rockfill is placed on the seabed within the cofferdam to reduce the seepage-induced instability and improve the ground bearing capacity (Figure 3d). Additional stabilization measures, such as grouting or ground reinforcement, may also be applied when required. Once stabilization is achieved, the cofferdam interior can be maintained under dry or semi-dry conditions, enabling construction activities to proceed safely.

The final step involves foundation and structural construction within the cofferdam (Figure 3e). This typically includes the installation of sacrificial pipes, the placement of reinforcement cages, concrete casting, seabed foundation work, pile installation, and the construction of superstructures. Throughout the process, structural safety is continuously monitored, and emergency equipment is prepared to repair minor leaks or seepage as they occur. This ensures that the cofferdam remains stable and watertight until the foundation and pile works are completed.

2.2. Configuration of the Circular Steel Cofferdam

Figure 4 shows a schematic of the steel pipe cofferdam. A prototype with a diameter of 20 m was fabricated to apply the circular cofferdam construction technology to offshore field conditions. Considering the specifications of available offshore lifting equipment, the cofferdam was designed and manufactured using single steel plates with a yield strength of 355 MPa. The system consists of an arched top lid and top, a center, and a bottom steel pipe.

Internal stiffeners were installed within the structure to prevent structural deformation during transportation and suction installation. In addition, lifting lugs were installed at eight horizontal locations to enable safe handling with marine cranes. The external surface of the cofferdam was coated with protective paint to prevent corrosion during offshore operation. The top and center steel pipes were fabricated as 3.5 m modules. The bottom steel pipe was designed as a single 6 m unit with a wall thickness of 30 mm.

For assembly, flanges were attached to both ends of the top and center steel pipes as well as to the top of the bottom steel pipe to enable bolted connections with the top lid. To ensure watertightness, 360 bolt holes were drilled horizontally at a spacing of 1° on each flange. Inside each bolted joint, a 36 mm-wide and 10 mm-deep groove was fabricated to accommodate an EPDM gasket (36 mm in diameter), which provided sealing against leakage. For the lifting operations, eight lugs were symmetrically installed on both the top and bottom steel pipes, allowing the cofferdam to be hoisted with wires from eight points.

Before offshore deployment, a trial assembly of the top, center, and bottom steel pipes and top lid was performed onshore to confirm the dimensional tolerances and fabrication accuracy. Six pairs of male–female guide keys were installed vertically inside the structure to assist in alignment during the offshore assembly, ensuring a precise connection between the top, center, and bottom steel pipes. Horizontal working platforms with guardrails were installed at both the top and bottom steel pipe connections to ensure worker safety and ease of assembly. Additional safety hooks were attached to the external surface of the cofferdam to secure workers during the offshore operations.

Suction piping was fabricated and installed along the sidewall of the center steel pipe for the installation of the 20 m-diameter large-scale cofferdam. Owing to the size of the cofferdam, the suction hoses must withstand both a hydrostatic pressure of up to 13 m from the water head and the dynamic energy of the water flow. This makes it difficult to maintain verticality during the initial penetration and demands highly durable hose couplings. A symmetric suction piping system was developed and tested to mitigate these challenges.

The suction piping consisted of 8-inch sidewall pipes, 5-inch Y-shaped distribution pipes, and 33° elbow joints, all fabricated from spunbond-polypropylene–polyethylene (SPP) material. The sidewall piping was connected in pairs at 3 m intervals, forming a curved path with elbow connections. The layout was designed such that the suction hoses could be installed symmetrically on opposite sides of the cofferdam, thereby minimizing the hose length and ensuring balanced forces during suction penetration. The Y-distribution pipes and joints were bolted to the sidewall piping, guiding the inflow and outflow in opposite directions to improve the efficiency and stability. This configuration reduces the risks of uneven penetration and piping failure, which are critical concerns in cofferdams of this scale.

Figure 5 shows the specifications of the top lid. The top lid serves as a cover that seals the upper portion of the cofferdam during suction installation, allowing suction pressure to be applied inside the structure. The lid must fully cover the top opening of the cofferdam and ensure tight contact such that internal water can be discharged to the outside through suction pumps without leakage. Flanges were installed along the perimeter of the lid and at the upper end of the top steel pipe to enable bolted connections. To guarantee watertightness, sealing materials such as water-swelling seals or O-rings were inserted into the grooves between the flanges. The lid must withstand vertical suction pressures of up to 100 kPa with no intermediate supports other than the cofferdam wall.

The lid was designed as a segmented arched structure to achieve structural efficiency. This design was validated by Kim et al. [37], who performed a structural behavior analysis of arched cofferdam lids under vertical pressure. Their results demonstrated that a segmented arched configuration optimizes stability against suction loads. The final design of the top lid consisted of a central lid plate, radial and ring stiffeners, and additional reinforcement members. Specifically, 24 radial stiffeners, four horizontal ring stiffeners, and supplementary ribs were incorporated to ensure both rigidity and safety during offshore operations.

Table 1. Specifications of the 20 m-diameter circular steel cofferdam.

Category	Component	Quantity	Diameter (m)	Height (m)	Thickness (mm)	Weight (MN)
Top lid	Lid	1	20	1.8	30	1.12
	Reinforcement	24	–	–	10	0.58
Coffer dam	Top pipe	1	20	3.5	10	0.80
	Center pipe	1	20	3.5	30	0.80
	Bottom pipe	1	20	6.0	30	1.09

lists the specifications of the 20 m-diameter circular steel cofferdam, including the top lid, reinforcement members, and modular steel pipe components. The top lid, with a diameter of 20 m and a height of 1.8 m, was fabricated from 30 mm-thick steel plates and reinforced with 24 radial stiffeners, each 10 mm thick. The total weight of the lid and reinforcement was approximately 1.70 MN.

The cofferdam body consisted of top, center, and bottom steel pipes. The top steel pipe measured 3.5 m in height with a 10 mm wall thickness, whereas the center steel pipe was also 3.5 m in height but with a 30 mm wall thickness, resulting in equivalent weights of approximately 0.80 MN each. The bottom steel pipe was the largest component, with a height of 6.0 m and a wall thickness of 30 mm, weighing approximately 1.09 MN. The assembled structure had a diameter of 20 m and a combined weight of more than 4.5 MN, ensuring sufficient self-weight for penetration under suction loading.

2.3. Instrumentation

Figure 6 shows the instrumentation of the steel pipe cofferdam. A comprehensive monitoring system was installed to verify the performance of the 20 m-diameter circular cofferdam during suction installation. Sensors were mounted on both the cofferdam structure and the surrounding ground to capture the key hydraulic and structural responses during testing.

To measure the pore water pressures generated by suction pumping, pore water pressure transducers were installed both inside and outside the cofferdam wall (P1 and P2). Additional pore water pressure transducers were embedded at four locations along the internal depth of the cofferdam (P3–P6) to record changes in the pore water pressure during penetration. A water level gauge was positioned to track variations in the sea level during installation (W1).

The structural responses of the cofferdam were monitored using strain gauges. Twelve strain gauges were installed along the steel pipe walls and arranged to measure both the axial (V1–V4) and horizontal (H11–H8) strains, enabling detailed monitoring of deformations caused by suction loading. To evaluate the connection performance between the cofferdam modules, bolt load cells were mounted at four flange positions (BT1–BT4), and a linear variable displacement transducer (LVDT) was used to assess the flange movement and watertightness before and after installation.

The penetration depth and installation speed of the cofferdam were measured using an onshore station that continuously tracked the vertical displacement of the structure. To verify the alignment, a biaxial inclinometer was mounted on the top lid (Inc1), providing real-time monitoring of the verticality during suction penetration.

Twenty-four sensors were attached to the cofferdam across the different measurement categories. These included six pore water pressure transducers (four internal and two external), 12 strain gauges (eight horizontal and four axial), four bolt load cells, an LVDT at the flange connections, one water level gauge, and one biaxial inclinometer. This monitoring system enabled continuous observation of the suction pressure, pore water pressure variation, structural deformation, connection performance, vertical alignment, and installation rate.

All of the instruments were selected and calibrated to ensure reliable measurements under marine field conditions. The pore water pressure transducers had a rated capacity of 5.0 kg/cm², an output sensitivity of 0.6–1.0 mV/V, and nonlinearity within ±1% rated output (RO), suitable for recording suction pressures up to 500 kPa. The water level meter featured a rated range of 30 m with the same accuracy class. The foil strain gauges mounted on the cofferdam wall had a rated capacity of ±5000 × 10⁻⁶ strain, an output of 2.0–2.5 mV/V, and nonlinearity within ±1% RO, whereas the bolt tension meters used at the flange joints had a rated capacity of 20 tonf and comparable accuracy. The biaxial inclinometer measured angular deviations within ±5° at an output range of 1–4 V, enabling verticality control with a resolution of approximately 0.01°. The LVDTs (Model CDP-25) had a 25 mm stroke and nonlinearity within ±0.5% full scale, allowing for precise measurement of the flange displacement.

All sensors were precalibrated using certified reference loads or pressures, and zero offsets were verified immediately before installation. Data acquisition was performed using a DEWESoft DS-Net dynamic data logger (Austria), which is capable of recording up to 10 kS/s per channel across 28–32 channels within an operating range of 0–70 °C. The system simultaneously collected the strain, displacement, temperature, and DC voltage signals at a nominal sampling frequency of 1 Hz for quasi-static monitoring. An IEEE 802.11 an/ac TDMA protocol with MIMO 2 × 2 radio mode in the 5.25–5.85 GHz band was used for wireless transmission, providing stable mid- to long-range communication. The raw signals were filtered using a digital low-pass Butterworth filter and normalized to the rated output of each sensor to remove noise and ensure consistency. These procedures guaranteed accurate, synchronized, and traceable data for evaluating the suction pressure, structural response, and alignment during field installation.

3. Test Bed and Design Considerations

3.1. Site Description

A full-scale field test was conducted offshore of Jebu Island, Republic of Korea, in the vicinity of the Jeongok port (Figure 7a). The test site was located within the construction area of the Jebu–Jeongok marine cable car project, which aimed to connect Jeongok port and Jebu port with a 2.12 km-long cableway. For this project, the marine tower foundations were originally designed as groups of four cast-in-place bored piles, each with a diameter of 2.0 m. A 20 m prototype steel pipe cofferdam was selected for installation at the test bed to validate the applicability of large-diameter circular cofferdams for such offshore foundation works.

Figure 7b provides an overview of the test bed during the field trial. The photograph shows the circular steel cofferdam positioned offshore, supported by a marine crane and barge system. Temporary working platforms and ancillary equipment are visible around the cofferdam, illustrating the actual offshore construction environment. The site is characterized by a significant tidal range, which presents both challenges and opportunities for construction. The tidal water level varies from +4.42 m at high tide to −4.339 m at low tide, yielding a total tidal range of almost 9 m. Within this regime, approximately 3.5 m or greater water depth is available for approximately 5 h per tidal cycle, providing a limited but sufficient working window for marine operations.

3.2. Geotechnical Properties of the Test Bed

Figure 8 shows the geotechnical profile of the test bed. The seabed stratigraphy is dominated by a thick deposit of sedimentary clay, extending to a depth of approximately 30 m below the seabed. This soft clay layer is representative of the marine deposits that are typically encountered in tidal flat environments along the west coast of the Republic of Korea. Beneath the clay, interbedded layers of sedimentary sand and weathered soil are observed, generally between 30 and 40 m in depth, providing locally improved strength compared with the overlying clay. At greater depths, the profile transitions into weathered and soft rock, which forms a geotechnical bearing stratum.

Table 2. Design soil properties at the test-bed site.

Layers	SPT N	Depth (m)	Cohesion (kPa)	Internal Friction Angle (°)	Elastic Modulus (kPa)	Poisson’s Ratio
Clay	$\mathrm{N} \le$ 2	0–4	25.0	0.0	4000	0.40
	$2 <$$\mathrm{N} \le$ 6	4–10	30.0	0.0	4000	0.40
	6 < N	6–10	75.0	0.0	5000	0.39
Sand	20 < N	10–15	–	27.0	20,000	0.35

lists the soil properties at the test-bed site. The soil parameters were determined using a combination of in situ and laboratory tests. Standard penetration tests (SPTs) were conducted to determine the relative density and consistency of the soil layers. Laboratory tests, including unconsolidated undrained triaxial compression and unconfined compression (qu) tests, were conducted to evaluate the undrained shear strength. Additional index tests, such as the Atterberg limits, natural water content, and grain size distribution, supported the soil classification. The elastic modulus and Poisson’s ratio values were estimated from correlations with the SPT N-values and oedometer consolidation tests.

Based on the SPT results, the clay layers were subdivided into three categories. In the uppermost zone, between a 0 and 4 m depth, the soil consisted of very soft clay with SPT N-values of less than or equal to 2, corresponding to an undrained shear strength of approximately 25 kPa and representing the weakest portion of the profile. From a depth of 4–10 m, the clay became slightly stiffer, with SPT N-values in the range of 2–6 and an undrained shear strength of approximately 30 kPa. At depths between 6 and 10 m, the clay exhibited SPT N-values greater than 6, corresponding to medium-stiff clay with an undrained shear strength of approximately 75 kPa and slightly higher elastic modulus values.

Beneath the cohesive deposits, a sand layer was encountered at depths of 10–15 m. This layer showed SPT N-values greater than 20, an internal friction angle of 27°, and an elastic modulus of approximately 20,000 kPa. Compared with the overlying clay, this sand deposit provided a significantly stronger bearing horizon and, thus, represented the geotechnically competent stratum within the soil profile.

3.3. Theoretical Considerations

When considering the stability and installation behavior of a suction-type cofferdam, it is essential to understand the balance between driving and resisting forces during penetration. When the cofferdam is seated on the seabed, penetration occurs once the driving force induced by suction exceeds the total soil resistance acting at the skirt tip and along the sidewalls. The driving force is proportional to both the pressure differential between the foundation’s interior and exterior and the cross-sectional area, which scales with the square of the diameter. In contrast, the total soil resistance increases roughly in proportion to the diameter. Consequently, larger-diameter foundations require a smaller pressure difference to achieve penetration. This scaling relationship explains why large cofferdams and suction buckets can be efficiently installed even in shallow water, provided the seabed soil maintains adequate stability.

For safe installation, two limiting suction pressures must be evaluated: the lower bound suction pressure, representing the minimum suction required to initiate penetration, and the upper bound suction pressure, beyond which soil failure or hydraulic fracturing may occur. The lower bound defines the suction necessary to overcome soil resistance along the skirt tip and walls, while the upper bound denotes the critical pressure at which soil heave or piping could begin beneath the lid. As illustrated in Figure 9, the area below the lower bound line denotes a condition where installation cannot proceed, whereas the region above the upper bound line signifies potential soil instability. The allowable suction during installation must, therefore, remain between these two boundaries. The maximum possible suction pressure generally increases with water depth, since the hydrostatic pressure difference between the inside and outside acts favorably on the lid. Depending on local water level variation, soil strength, and permeability, the maximum suction may coincide with or differ from the upper bound envelope.

In cohesive soils, the tip resistance (q_u) can be calculated using Equation (1) representing the bearing capacity of the soil at the skirt edge while the internal plug compresses during penetration. The shaft resistance (Q_skin) shown in Equation (2) contributes significantly to total resistance. Because bucket-type foundations generally have larger diameters than conventional piles, the sidewall area is proportionally larger. Accordingly, both inner and outer skin friction must be considered when estimating penetration resistance. As the skirt penetrates deeper, this friction increases as the soil adheres to and shears along the steel walls, adding substantially to the total resistance.

$$q_u=c^{\prime }{N}_c+q^{\prime }{N}_q+0.5{\gamma }^{\prime }t{N}_r$$

(1)

$$Q_{skin}=\sum _{i=1}^N{\int }_0^{L_i}\pi D{f}_{s}dz$$

(2)

where c′ represents the cohesion of the soil, q′ corresponds to the effective overburden pressure at depth. The submerged unit weight of the soil is denoted by $\gamma \prime$, and t is the thickness of the steel pipe. N_c, N_q, N_r refer to the bearing capacity coefficients, D represents the diameter of the steel pipe and f_s denotes the unit skin friction acting along the foundation wall.

For the consideration of self-weight penetration, it must first be verified that sufficient embedment occurs under the structure’s own weight before suction pressure is applied. In other words, the initial penetration by self-weight should be deep enough that additional suction can effectively promote further penetration. This self-weight penetration depth can be estimated by assuming zero suction and equating the driving force (F_D) to the total soil resistance (Equation (3)). Achieving this balance ensures that the cofferdam is stably seated prior to the application of suction pressure.

$$F_D=p_s{A}_s+p_t{A}_t+W^{\prime }+Q$$

(3)

where p_s and p_t denote the suction pressures beneath the top lid and at the tip of the cofferdam, respectively; A_s and A_t refer to the inner area of the top lid and the cross-sectional area at the cofferdam tip, respectively; W′ represents the effective weight of the cofferdam; and Q signifies the applied overburden load.

The upper bound suction pressure (p_sc) expressed by Equation (4), can be derived from the concept of confined groundwater uplift stability. When upward suction acts at the base of the foundation, the upper limit is governed by equilibrium among the upward suction pressure, the self-weight of the soil plug, the adhesion along the inner wall, and shear resistance at the soil–steel interface. Exceeding this limit may trigger soil heave or piping beneath the foundation, defining the ultimate safe limit for installation.

$$p_{sc}=({\gamma }^{\prime }z+2c_u)/Fs$$

(4)

where $\gamma \prime$ represents the submerged unit weight of the soil, z is the penetration depth of the cofferdam including upward heave, c_u refers to the undrained cohesion of the clay, and Fs denotes the factor of safety.

When evaluating the stability against piping, it is necessary to consider seepage-induced failure. Piping occurs when the upward seepage force exceeds the submerged unit weight of the soil, reducing the effective stress to zero. The safety factor against this failure mode is typically evaluated using the concept of the limiting hydraulic gradient, with conditions deemed stable when the factor of safety (Fs) exceeds 2.0. In this study, seepage stability inside a circular steel cofferdam was analyzed considering the water level difference between the interior and exterior. A conservatively high permeability coefficient of approximately 10⁻² m/s was assigned to the surface clay and silty sand layers. For a foundation diameter of D = 20 m, with external and internal water levels at H.W.L. El(+1.32 m), and an internal water level of El(−2.74 m), respectively, the resulting differential head of 4.06 m induces seepage from the outside toward the inside. The analysis results show that the piping safety factor increases with embedment depth, displaying marked improvement when the embedment exceeds about 0.2D to 0.4D. For embedment depths greater than 1.25 m, the safety factor satisfies the river design criterion of Fs = 2.0, while depths of 2.5 m or more yield Fs = 4.0, indicating very stable seepage conditions.

4. Field Installation and Measurement Results

Figure 10 presents the sequential process of the field installation and extraction of the 20 m-diameter circular steel cofferdam. The procedure started with lifting the lower cofferdam segment using a revolving crane and lowering it to its designated position (Step 1). The cofferdam penetrated the seabed under its self-weight until no further settlement was observed (Step 2). Subsequently, the guide beam was manually separated (Step 3), and the upper cofferdam segments (Nos. 1 and 2) were stacked on top of the lower unit. Guide keys were employed to ensure the precise alignment of the bolt holes during lowering, and the flanges were fastened using a pneumatic impactor. Final tightening was performed with a torque wrench to maintain uniform bolt tension, and visual inspection confirmed proper flange contact (Step 4).

Following assembly, a top cover plate was placed to seal the cofferdam interior, and the suction pumps were connected (Steps 5 and 6). Because the cofferdam initially contained an air pocket above the waterline, one pump was operated to evacuate air, whereas another simultaneously injected water, enabling the interior to be filled rapidly (Steps 7 and 8). Once filled, the injection pump was stopped, and suction pumping continued to discharge water from the interior. This created a pressure differential between the inside and outside of the cofferdam that induced suction-assisted penetration into the seabed. The magnitude of the suction pressure and corresponding penetration rate were controlled by adjusting the discharge flow rate (Step 9). During this stage, the verticality of the cofferdam was monitored in real time to ensure accurate installation.

Figure 11 shows the variation in the suction pressure with the penetration depth during the installation of the 20 m-diameter circular cofferdam. Following the self-weight penetration, suction pumps were activated to discharge water from the cofferdam interior, creating a pressure differential between the inside and outside. In this study, the suction pressure was defined as the pressure difference between the cofferdam interior and ambient hydrostatic condition, corrected for the potential head difference arising from the pore pressure gauges located above the seawater level at the initial stage of installation.

The measured suction pressure generally increased with the penetration depth, indicating that a greater driving force was required as the cofferdam penetrated deeper into the seabed. This trend reflects the increasing soil resistance mobilized against the structure, consisting of both the tip resistance at the skirt edge and shaft friction along the sidewalls [8]. The gradual increase in suction pressure with depth is theoretically consistent with the expected soil resistance behavior for a cylindrical structure under increasing penetration, where higher suction is needed to overcome the growing passive resistance mobilized by the surrounding soil as the embedded length increases. The stepwise pattern in Figure 11, showing alternating increases and plateaus, represents the pump control process and the transient equilibrium between applied suction and soil inflow. During each hold period, partial dissipation of pore pressure and readjustment of soil stress occur, stabilizing the soil plug and preventing hydraulic fracturing or piping. Minor fluctuations likely correspond to local variations in soil stiffness or permeability, indicating that penetration was governed by soil–structure interaction rather than by abrupt failure.

At the target embedment depth of 5 m, the average suction pressure acting within the cofferdam was approximately 63 kPa. This value is within the theoretical design range, below the expected cavitation and joint sealing limits, confirming that the applied suction was sufficient to maintain continuous penetration while preserving the structural and hydraulic integrity of the system. The average penetration rate recorded during the installation was 1.83 m/h, which exceeded the planned installation rate of 1.0 m/h. A higher penetration rate is advantageous for construction efficiency, provided that excessive soil disturbance or soil failure is avoided. These results demonstrate that the target depth of 5 m could be successfully achieved using suction-assisted installation even under shallow-water conditions, where part of the cofferdam remained above the seawater level.

Figure 12 and Figure 13 present the measured vertical and horizontal strains of the 20 m-diameter circular cofferdam wall during the suction-assisted penetration. Prior to installation, the fabrication and transportation of the cofferdam did not induce noticeable predeformation, and the designed curvature of the steel pipe segments was fully maintained. This confirmed the structural adequacy of the fabricated sections with respect to the design geometry.

The experimental results show that both the vertical and horizontal strains increased approximately linearly with the penetration depth, with positive values indicating tensile strain and negative values indicating compressive strain. Notably, the horizontal (circumferential) strains were generally greater than the vertical strains, reflecting the increasing external loads applied to the cofferdam wall as the suction pressure developed. This trend is consistent with observations from previous smaller-scale tests on 5 m-diameter cofferdams [29]. The maximum vertical compressive strain of the top pipe reached approximately −81.9 μm/m under a suction pressure of 63 kPa, while the bottom pipe experienced a similar compressive strain of −76.4 μm/m. These compressive strains were induced by the negative pressure acting inside the cofferdam, which compressed the structural wall.

The strain patterns observed in Figure 12a,b indicate that vertical compression was relatively uniform across the height of each segment, suggesting that the axial membrane forces developed gradually as the structure penetrated deeper. The similarity of magnitudes between the top and bottom pipes implies that global bending of the wall was limited and that the shell primarily acted in membrane compression rather than flexural distortion. This behavior is theoretically consistent with a thin cylindrical shell subjected to external pressure, where axial strains are secondary to the dominant circumferential action.

In terms of the horizontal strain, the maximum tensile strain observed at the top pipe was approximately +325.7 μm/m, whereas smaller strains were recorded closer to the bottom segment embedded in the seabed. Interestingly, strain gauges positioned 180° opposite to one another recorded similar magnitudes, whereas those at 90° spacing exhibited opposite signs (tension vs. compression). This asymmetry was attributed to the influence of the internal stiffeners and bracing members installed in four orthogonal directions, which redistributed the local stresses. As shown in Figure 13, the circumferential strain distribution forms a characteristic alternating pattern of tensile and compressive zones around the pipe perimeter, representing local ovalization of the circular section under differential suction loads. Such a pattern is a typical shell response, where circumferential tension develops at two opposing sides while compression forms at the perpendicular axes, maintaining equilibrium of the hoop stresses. The smaller horizontal strain observed at the bottom pipe further confirms the confining effect of the surrounding soil, which restricts lateral expansion and suppresses ovalization at deeper levels.

From a theoretical standpoint, the dominance of horizontal over vertical strain indicates that the cofferdam wall primarily resisted suction-induced loads through hoop (circumferential) tension and compression, consistent with thin-shell behavior under external pressure. The limited magnitude of vertical compression and the stable, reversible strain trends demonstrate that the structure deformed elastically throughout the process. Moreover, the correspondence between opposing gauges (H1–H3, H2–H4) validates that the global symmetry of the wall was preserved and that local stress concentrations remained within the elastic range of the steel.

These results collectively demonstrate that although the cofferdam wall experienced measurable deformations during suction installation, the magnitudes were within the expected range and were consistent with the design assumptions. The close agreement between measured strain trends and theoretical shell behavior confirms that the 20 m cofferdam maintained structural stability under suction pressures up to 63 kPa, with no irreversible deformation observed upon installation completion.

Figure 14 shows the changes in the inclination measured during the suction installation process. In the figure, the sign convention for inclination is defined such that positive angles in the X- and Y-directions indicate tilting toward the top left, whereas negative values represent tilting toward the bottom right. Owing to repeated tidal fluctuations following self-penetration, the circular steel cofferdam could not maintain its initial verticality and tilted by −0.49° in the X-direction and −0.88° in the Y-direction during the early stage of penetration. When penetration was complete, the inclination was observed to be −0.56° in the X-direction and −0.80° in the Y-direction.

The observed inclination histories in Figure 14 indicate that the cofferdam experienced minor but progressive tilting primarily during the early penetration stages, which corresponds to the period when self-weight and initial suction first overcame the uneven seabed contact. As penetration advanced and the skirt became more fully embedded, the inclination gradually stabilized, confirming that soil confinement and side friction provided increasing rotational restraint. The small residual inclinations at the final stage (below 1°) fall well within acceptable tolerance limits for large-diameter marine structures and demonstrate effective control of vertical alignment throughout the installation.

From a theoretical standpoint, minor inclination during early suction stages is expected due to asymmetric suction flow paths or small variations in soil stiffness beneath each segment. These factors can induce unbalanced tip resistance and slight angular rotation until uniform penetration is achieved. The subsequent stabilization of the inclination confirms that differential suction pressures were equalized and that the cofferdam reached a symmetric and stable embedment condition. Because the verticality of the circular steel cofferdam is strongly influenced by the inclination during the initial penetration stage, securing vertical alignment at the outset is important.

The circular steel cofferdam is composed of multiple steel pipe segments and a top plate connected by bolted joints, making it essential to ensure the complete sealing of the connections. To verify this sealing, bolt tension gauges were installed at four locations along the cofferdam joint plane, and the bolt forces were measured before and after suction installation. In addition, visual inspections were conducted before installation, after installation, and after dewatering inside the cofferdam to check for leakage. No signs of leakage were observed.

As summarized in

Table 3. Changes in bolt tension force before and after installation.

Installation Process	Bolt Tension (kN)
	BT1	BT2	BT3	BT4
Before	101.55	101.68	107.34	101.24
After	84.78	87.39	94.68	87.79
Difference	16.51%	14.05%	11.79%	13.29%

, the initial bolt pretension, applied using a torque wrench, ranged from 101.24 to 107.34 kN. Following suction installation, the measured values decreased to 84.78–94.68 kN, corresponding to a tension loss of approximately 11.8–16.5%. This reduction was attributed to the stress relaxation between the bolts and nuts induced by the applied suction pressure. However, the loss was not sufficiently significant to impair the watertight performance.

From a theoretical perspective, the measured tension loss is consistent with the combined effects of mechanical seating, gasket compression, and time-dependent relaxation of the joint. When suction pressure is applied, the negative internal pressure slightly increases the separation tendency at the flanges, temporarily redistributing bolt loads and causing a small but recoverable reduction in preload. In addition, microscopic flattening of the contact surfaces and viscoelastic deformation of the EPDM gasket contribute to gradual relaxation over time. The observed range of 10–17% is typical for large-diameter flanged joints under cyclic external pressure, indicating that the joint stiffness design was appropriate.

It is also important to note that the final bolt tensions remained well above the minimum required to maintain gasket compression and sealing capacity under the maximum suction pressure of 63 kPa. The absence of leakage during all inspection stages validates that the bolt pattern, tightening procedure, and gasket selection provided adequate clamping force even after accounting for expected relaxation. These results confirm that the segmented bolted connections maintained structural integrity and hydraulic tightness throughout suction installation.

Furthermore, to evaluate the flange deformation during installation, two strain gauges and two displacement transducers were installed on the flange at diametrically opposite positions (180° apart), and two additional strain gauges were mounted on the adjacent steel pipe. The measurements indicated that the maximum vertical displacement of the flange during suction installation was only 0.004 mm, which was negligible. The maximum strain recorded on the flange was −66.4 μm/m, whereas the maximum strain on the adjacent steel pipe was −87.2 μm/m. All measured values remained within the allowable stress limits, confirming that no structural problems occurred during installation.

Through a full-scale demonstration test, a 20 m-diameter circular steel cofferdam was installed on the seabed, following which the suction top plate was removed and the interior water was pumped out using dewatering pumps to allow for visual inspection of the internal ground conditions. As shown in Figure 15, the cofferdam was successfully penetrated to the target depth of 5 m by suction without inducing soil failure, and ground-sealing equipment was deployed to demonstrate stabilization. The interior of the cofferdam remained effectively sealed from external seawater even during high tides caused by tidal fluctuations, thereby confirming its functionality and performance as a cofferdam.

5. Discussion

The field results from the 20 m-diameter suction-installed cofferdam provide new insights into the feasibility of applying suction technology on an unprecedented scale. Several aspects merit critical discussion in relation to the existing literature.

Compared with previously reported 5 m-diameter field trials [31,35,37], the present 20 m full-scale demonstration introduces multiple design and execution innovations. First, it verifies the feasibility of suction penetration for a partially exposed large-diameter cofferdam, which is a condition that has never been validated in previous submerged models. Second, the newly designed arched top lid and segmental bolted joints with EPDM gaskets were proven to resist high suction loads without leakage or structural compromise. Third, this study provides the first integrated dataset linking the suction pressure, penetration rate, strain distribution, bolt pretension, and verticality control at this scale. Therefore, the novelty lies not only in dimensional scaling but also in demonstrating a new construction system and monitoring methodology that bridges laboratory-scale concepts and real-world marine applications.

Specifically, the measured suction pressures increased steadily with the penetration depth, reaching approximately 63 kPa at the target embedment depth of 5 m. This trend is consistent with suction caisson studies in sand and clay, in which soil resistance mobilization requires progressively higher driving pressures [6,8]. The observed average penetration rate of 1.83 m/h was significantly higher than the planned value of 1.0 m/h, indicating that the applied suction pressures were sufficient to overcome the tip and shaft resistances without triggering soil boiling or piping. These results compare favorably with smaller-scale trials on 5 m cofferdams [31], demonstrating that penetration efficiency can be maintained even at diameters that are four times larger.

The instrumentation showed that the horizontal strains (+325 μm/m) were larger than the vertical strains (−82 μm/m), reflecting circumferential tension in the cofferdam wall under internal suction. The observed magnitudes are within typical allowable strain limits for offshore steel structures (<1000 μm/m for service conditions), confirming the structural adequacy of the design. Previous model-scale tests also reported higher hoop strains than axial strains [25], suggesting that circumferential reinforcement is particularly critical for large-diameter cofferdams.

Bolt pretension losses of 12–16% were recorded during the suction penetration. This reduction aligns with the stress relaxation values reported in flange connections of suction buckets [31,34]. Importantly, no leakage was detected during dewatering, and the flange deformations were negligible (0.004 mm), confirming that the EPDM gasket system maintained adequate sealing under suction and hydrostatic loading. This indicates that segmental construction using bolted joints is viable for cofferdams of this scale, provided that pretension losses are considered in the design.

Inclination changes of less than 1° were observed despite tidal fluctuations. Although such deviations were not significant for constructability, they highlight the importance of controlling the alignment during the early penetration stage. Similar conclusions were drawn by Park et al. [34] in suction pile penetration experiments, in which the verticality correction was most effective when applied immediately after the initial self-weight penetration. The ability to maintain near-vertical alignment in a 20 m cofferdam supports the feasibility of extending this method to offshore bridges or wind turbine foundations, which have strict tolerance requirements.

The results demonstrate that suction-driven cofferdams with large diameters can achieve rapid penetration, stable structural performance, and reliable watertightness. The suction approach offers faster installation, reduced noise and vibration, and reversible removal compared with conventional cofferdam methods, making it well-suited for offshore wind, bridge, and harbor projects where temporary dry working environments are required. Further research should focus on the long-term performance under cyclic wave loading, optimization of the lid stiffener design, and scaling laws for penetration resistance in layered soils.

In addition, the field test was conducted in a large tidal range of approximately 9 m, which is characteristic of the west coast of the Republic of Korea. Such tidal fluctuations temporarily changed the water head acting on the cofferdam and may have slightly affected the suction pressure distribution and apparent penetration rate, particularly during the transitions between low and high tides. The influence of these fluctuations was minimized by performing suction operations within the mid-tide interval when the water depth remained relatively stable for several hours, and by continuously monitoring the verticality and suction pressure in real time.

However, the effect of tidal variation on the suction efficiency and pressure measurement reproducibility could not be quantitatively isolated in this single full-scale test. Future field programs should include controlled monitoring under different tidal conditions to assess these interactions more effectively.

Although the present field test was performed on a clay seabed that is representative of typical tidal flat deposits along the west coast of Korea, the soil type can significantly influence the suction-driven installation process. The rapid dissipation of excess pore pressure may reduce the effective suction pressure in sandy or highly permeable soil, necessitating a higher pumping capacity or modified skirt geometry to prevent premature water inflow and soil boiling [27,28,38].

Conversely, the penetration resistance may increase sharply in very stiff clay or interbedded layers with high strength contrast, leading to nonuniform embedment or excessive skirt deformation. As a result, the current findings reflect the behavior in soft, cohesive soil conditions. Additional studies are necessary to assess how the system performs in dense or loose sand, as well as in dense clay strata. Future research should also address the potential limitations related to the suction loss, permeability contrast, and drainage boundary effects under varying seabed compositions to extend the applicability of the proposed technology to broader offshore environments.

The results of this study have significant implications for offshore construction. The full-scale results demonstrate that suction-driven cofferdams can substantially reduce the installation time and noise/vibration compared with conventional sheet-pile or caisson systems, enabling faster project delivery and improved environmental compliance. The modular segmental design allows for reusable components, thus lowering fabrication and mobilization costs.

In the context of offshore wind foundations and bridge pier construction, the ability to install and remove large cofferdams rapidly without heavy piling operations offers clear logistical advantages. From a design-standardization perspective, the present dataset provides key parameters, such as suction–penetration curves, allowable strain levels, and flange-pretension limits, which can inform future guidelines for suction-driven cofferdam design and operation and facilitate broader adoption in marine infrastructure projects.

6. Conclusions

In this study, a large-scale circular steel cofferdam with a diameter of 20 m and a height of 13 m was fabricated and successfully demonstrated through a full-scale field test in a real marine environment. The primary objective of this study was to verify the feasibility and performance of the suction-driven installation and removal of large-diameter cofferdams, thereby extending suction technology beyond its traditional application to foundations and anchors. Through detailed instrumentation and on-site measurements, this study confirmed that the proposed system can be installed efficiently and maintain structural stability and watertightness under real marine conditions. The creation of a dry internal working environment was also visually confirmed, thereby fulfilling the intended purpose of the cofferdam concept. The main findings are summarized as follows:

A cofferdam with a diameter of 20 m was successfully installed at the target penetration depth of 5 m, even when the top of the structure remained exposed above the water surface. As opposed to the conventional suction bucket method, in which hoses are connected from the top, an alternative configuration with lateral hose connections was adopted for shallow-water conditions, which proved to be more efficient for suction penetration. The suction top plate was subjected to the highest compressive loads during installation. The plate, with a diameter of 20 m and a height of 1.8 m, was fabricated using 24 steel modules, forming a quasi-arching structure. Additional reinforcement was provided using four circular stiffeners and 24 radial stiffeners. The field verification confirmed the structural durability of the system.

During installation, the applied suction pressures were maintained within the upper and lower bounds of the design range. The observed self-weight penetration exceeded the predicted value, which was attributed to the increased structural weight of air caused by tidal fluctuations in the test area. For transport and constructability, the cofferdam was fabricated in three segmented units joined by bolted flanges, with water stops inserted between the joints. The measurements revealed that the horizontal strains were significantly higher than the vertical strains, with the largest values recorded at the top plate connections and along the joints of the longer segments.

Bolt pretension losses of approximately 12–16% occurred during suction penetration. However, the flange deformations remained negligible, and no leakage was observed at the designed strain levels, confirming both watertightness and structural stability. Following installation, the interior water was removed to allow for a visual inspection of the seabed. No evidence of soil piping or seepage was observed in the clay foundation, demonstrating that the cofferdam effectively fulfilled its intended cutoff function.

Overall, this research successfully achieved its objectives by demonstrating, for the first time at full scale, that suction-driven installation can be applied to a 20 m-diameter circular cofferdam to achieve rapid penetration, stable structural behavior, and reliable sealing under marine conditions. In practical terms, the field results provide several engineering implications for design optimization. First, the measured suction-penetration relationship confirmed that the suction pressure should be progressively controlled with the depth to prevent soil boiling or plugging, indicating the importance of real-time monitoring and feedback control systems for large-scale installations. Second, the arched lid geometry and flange reinforcement layout adopted in this study proved to be effective against compressive loads and may serve as a reference for designing suction lids with greater diameters. Third, the lateral hose configuration demonstrated higher installation efficiency under shallow-water conditions and is recommended when the top plate remains above the water surface. Additional pumping capacity or skirt stiffening may be required for deeper or highly permeable soils to maintain uniform penetration and prevent leakage. Collectively, these findings, together with the identified limitations for different soil types and penetration depths, provide a foundation for developing design guidelines and optimization strategies for future suction-driven cofferdam applications.