The Practice and Development of T-Bar Penetrometer Tests in Offshore Engineering Investigation: A Comprehensive Review

: In recent years, the development of marine hydrocarbon resources has led to an increased demand for research on the marine soil bearing capacity and cyclic loading effect in marine engineering design. Because of the difﬁculties and high costs involved in obtaining high-quality soil samples from offshore sites, in situ testing techniques have become the preferred method of determining design parameters in offshore geotechnical engineering projects. This paper provides a review of the current state of marine penetrometer deployment technology used in offshore engineering investigations and presents a summary of the T-bar penetrometer test for measuring marine soft clay. The existing literature research on penetration mechanisms, numerical simulations, laboratory experiments, and ﬁeld tests of the T-bar penetrometer in the ﬁeld of marine geotechnical engineering are analyzed. Finally, the potential difﬁculties, challenges, and prospects of the T-bar penetrometer tests are discussed.


Introduction
The development and expansion of coastal cities worldwide have promoted the continuous development of marine geotechnical engineering techniques and theories [1][2][3]. Marine engineering traditionally concentrated on near and shallow seas, including coastal protection, land enclosure, and port engineering. Nowadays, it has extended to distant and deep seas, which includes cross-sea bridges, offshore oil and gas drilling platforms, offshore wind energy, tidal energy power generation, submarine tunnels, submarine mineral extraction, and land enclosure [4,5]. The increasing scale and difficulty of marine engineering put forward higher requirements on the sensitivity, safety, and durability of marine engineering, especially on the evaluation of the performance of marine engineering under dynamic and cyclic loading effects and related design parameters [6][7][8].
The design and construction of marine engineering, including the main structure and appurtenant facilities such as submarine cables and pipelines, require a large amount of survey-accurate data [9,10]. Soft soils with a high water content, large porosity, low shear strength, high sensitivity, and low surface strength are common conditions encountered in marine engineering projects [11][12][13][14][15]. These characteristics make it highly difficult and costly to obtain low-disturbed subsea soil samples through traditional drilling methods and conduct laboratory geotechnical tests during marine engineering investigation [9,13,16]. Therefore, in situ testing is becoming one of the main methods of determining design parameters. The vane shear test, as a commonly used in situ testing method in soft soils on land, is more prone to be disturbed in subsea super-soft soils, which significantly affects the reliability of the test; on the other hand, in marine engineering surveys, the piezocone penetration test (CPTU) is adopted more often [12].
The piezocone penetration test (CPTU) is a popular in situ testing technique that emerged internationally in the 1980s. It has a solid theoretical foundation, comprehensive functions, precise parameters, high accuracy, and good stability [17,18]. These characteristics make the CPTU suitable for civil engineering design applications worldwide. The CPTU can accurately classify soil layers and identify soil types. Moreover, it can estimate various mechanical and deformation properties of soil, such as undrained shear strength, overconsolidation ratio, sensitivity, compression modulus, Young's modulus, initial shear modulus, consolidation coefficient, and permeability coefficient. Therefore, the CPTU is suitable for testing soft soil in a convenient, economical, and reliable way [16]. Recently, with the processive development of new directions and demands in geotechnical engineering, especially marine geotechnical engineering, many international research institutions (e.g., University of British Columbia, Georgia Institute of Technology, Delft University, and Southeast University) and professional in situ test instrument development companies (e.g., Hogentogler, Vertek, ConeTec, and Furgo Corporation) have developed new sensors for CPTU. These new CPTU sensors promote the CPTU test method to develop in the direction of multi-functionality [10,[19][20][21][22][23][24], which further broadens the application field of the CPTU. Figure 1 shows a schematic diagram of the internal structure of a common CPTU. However, there are also some disadvantages to the CPTU method. For example, when applying the conventional cone penetration test (CPT) in the subsea soft soil, Randolph found that the accuracy of the measured data decreases with the increase in the seawater depth. Further, in 1994, Randolph proposed a novel full-flow theory based on the fullflow penetrometers (including T-bar, ball, and plate) to test soft soils, the apparatus of which is shown in Figure 2 [25]. The full-flow penetrometers are now commonly used in offshore site investigations where the extremely soft soils encountered necessitate an increased measurement resolution and accuracy, and the flow mechanism of soil around the probe can be modeled as plane strain and axisymmetric in the case of the T-bar and ball, respectively. This is suitable for the structures supported on these deposits, including anchors and caissons for offshore oil platforms; nearshore piers, wind turbines, and liquid natural gas platforms; and piles in soft sediments onshore [26][27][28][29][30][31]. As a new in situ testing technology for testing marine soft soil, the T-bar penetrometer has the advantages of high accuracy, good reliability, a large amount of data collection, a convenient testing process, and low cost. However, this testing technique is mainly adopted in onshore engineering, and less application has been performed in offshore engineering, and it especially lacks practice in marine engineering [21]. Therefore, the study of the mechanism of the T-bar penetrometer and its application in marine engineering is of great theoretical significance and engineering value to the development of in situ testing technology in the field of marine engineering in the world. These previous studies on the failure mechanism and penetration resistance of full-flow penetrometer penetration behaviors mainly include analytical solutions, numerical simulations, laboratory experiments, and field tests [31]. In this paper, the development of marine penetrometer deployment technology is reviewed, and the penetration mechanism, numerical simulations, laboratory experiments, and field tests of the T-bar penetrometer in the field of marine geotechnical engineering are analyzed and summarized. Meanwhile, the potential difficulties, challenges, and prospects of the T-bar penetrometer tests are also discussed.

Development of Marine Penetrometer Deployment Technology and T-Bar Penetrometer
The CPT devices used in offshore engineering first appeared in the 1960s. After nearly 60 years of development, various CPT devices were developed for marine environments, including small jack-up platforms, cased platforms, seabed, and downhole modes, and submersible chambers, especially the earliest development and widest applications of CTP in the Netherlands [32]. The history of the development of foreign offshore CPT devices, summarized by Lunne (2010), is shown in Table 1 [32]. When adopting the CPT in marine engineering, the CPT equipment should be pushed into the sea bottom to test the soil on the seabed. There are two ways of pushing a cone penetrometer into the sea bottom; the first one is the seabed mode, which pushes the CPT or a predetermined penetration into the sea floor until it encounters a resistance. The second one is the down-hole mode, which drills a borehole and pushes the penetrometer into the soil at the bottom of the borehole, as shown in Figure 3 [21]. The seabed mode systems are mainly used in deepwater areas, and the penetration tests are driven by wheeling to generate a continuous thrust. Differently, in the downhole technique, the test equipment free falls along the drill pipe to the seabed, where it is hydrostatically pressed into the soft subsea soil layer by the negative pressure created by the seawater. This system can perform CPT and vane shear tests, as well as the sampling after the test. The advanced deepwater CPT tests in the world are listed in Table 2. Here, the development and application of the CPT technique are summarized. In 1965, the Netherlands and France successively used small jack-up platforms to carry out offshore CPT, but due to technical constraints, the maximum depth of touch was around 5 m.
In 1966, the Dutch company Fugro independently developed the "Seaball" and "Wison I" rope-type CPT equipment.
From 1972 to 1974, the Dutch company Fugro further ameliorated the "Seaball" offshore CPT by adding a subsea disc to provide a support reaction force, which was named "Seacalf". At the same time, Pfeiffer upgraded the "Wison I" rope-in-well CPT and developed the "Wison" MK II rope-in-well CPT with a total working water depth of 400 m.
In 1973, the Institute of Oceanography of the Chinese Academy of Sciences developed a seabed-mode underwater CPT, which reached a depth of 7 m, with a maximum test bearing capacity of 33 t/m 2 and a test water depth of 50 m [33].
In 1977, the Netherlands developed a cased platform CPT designed for a working depth of 50 m and a maximum penetration force of 20 t. In the same year, Norway cooperated with the UK and developed the "Strigraz" subsea sinking drilling CPT, with a working depth of approximately 30 m and a maximum penetration depth of 20 m.
In 1982, a new type of CPT machine was developed in Canada and successfully applied to test the shallow continental shelf. This equipment is advantageous for measuring both the tip resistance and pore water pressure. The first was the use of a cable-free system, using a digitally encoded ultrasonic system to transmit data from the probe directly to a recording device at the surface; secondly, the penetration depth was significantly increased by injecting mud behind the cone, reducing the friction between the rod and the soil, allowing it to penetrate 71 m into the soil with a reaction force of only 5 t.
Since the 1970s, applied research in the United States on offshore CPT has progressed relatively rapidly, with the development of a CPT system for marine exploration that has been successfully tested in the sediments of the San Diego Trench at water depths of over 1200 m [34].
By combining the MJ-II top-pressure penetrometer and CPT platform, the China Shipbuilding Industry Institute of The Engineering Investigation and Design Co., Ltd. developed a CPT platform in 2003 [33]. In 2005, the Institute of Engineering and Technology of Jilin University developed the "shallow seabed mode CPT system" with a maximum work depth of 55 m and a maximum test depth of 15 m [35].
From 2001 to 2005, the Marine Geological Survey of Guangzhou, China, relying on the national project "In situ static and dynamic penetration test technology of submarine soil", developed an improved technology of offshore CPT equipment with the hydraulic propulsion system in the tube, with a work water depth of 100 m and a penetration depth of 120 m [36]. The above-developed equipment has a shallow test water depth and a small penetration depth and is less utilized in practical engineering, and the theoretical research lacks sufficient test data.
Numerous piezocone penetration tests demonstrated that as the depth increases, the accuracy of the test results obtained in soft soil from the conventional CPTU declines. In the case of the piezocone, this correction can be significant, particularly offshore, where the ambient pressures can be high. According to Chung and Randolph's (2004) research, the correction required for the piezocone varied between 20 and 40% in the estuarine clay layer [37]. As shown by R. Kelly (2014), the cone correction for tests at the National Field Testing Facility (NFTF), Ballina approaches 40% at depth [38]. According to the studies, the main reasons for this include the following [12]:

•
The high pore water pressure during penetration testing in a high-pressure environment on the seabed and the low strength of the soft soils on the seabed compared to the soft soils on the seabed has a greater impact on the sensitivity of the penetration resistance during testing.

•
There is a significant correction for unequal area effect and uncertainty in the overburden stress correction on penetration resistance. Randolph et al. (1998) developed a full-flow penetrometer in 1994 to solve the above problem by increasing the coverage area of the penetrometer based on the fullflow theory [39]. The undrained shear strength of soft soils is determined by the frictional resistance generated by a viscous fluid-like soil flow around the penetrometer as it passes through the soft soil, as shown in Figure 3. The newly proposed full-flow penetrometer has a more rigorous theoretical solution between the penetration resistance measured by the penetrometer and the strength of the soft soil, and it can be applied using cyclic penetration tests to estimate the remodeling strength of the soft soil and further estimate its sensitivity. The T-bar, ball, and plate penetrometers are shown in Figure 4 [40]. The T-bar penetrometer was first used in marine engineering practice in 1998, and it is currently utilized in a variety of nations and areas for marine surveys. The correction applied to the full-flow penetrometer resistance is much smaller due to the near equal pressure above and below the probe. The correction required for the piezocone varied between 20 and 40% at depth, while the piezoball and T-bar correction remained below 10% [37,38]. Currently, the international standard NORSOK G-001 (Standards Norway 2004) incorporates a T-bar penetrometer. The standard defines a 40 mm diameter, 250 mm long, and 10,000 mm 2 projected area steel T-bar penetrometer (10 times the size of a standard CPT). The cross-sectional area of the connecting rods should not exceed 15% of the projected area of the T-bar penetrometer, nor should the diameter exceed the diameter of the T-bar penetrometer. Additionally, the surface of the T-bar penetrometer should be lightly sandblasted.
Naturally, the T-bar penetrometer is more susceptible to bending moments, which has an impact on the induced load cell in the system. These may lead to spurious changes in the load cell measurements, as it is difficult to achieve the results when the load cells are completely independent of bending effects. However, the T-bar penetrometer can be used as a pipeline component model to provide direct information for the design of pipelines and risers. For super soft clay sites in deep water, it can be very useful to obtain T-bar or ball penetrometer test data in addition to CPTU.

Analytical Solution of the Resistance Factor
The undrained shear strength of soft soils (s u ) is one of the most crucial indicators for analyzing the strength properties and stability of the soils, as well as one of the most important parameters for marine soft soil engineering. Haneng et al. (2014) demonstrated that the shape of the T-bar penetrometer resembles a cylindrical pipe, and that the undrained shear strength derived from the penetration test was more accurate for use as a strength parameter in the design of subsea pipelines and risers in practical marine engineering [34,41]. The results of the model tests and field tests by Weemees et al. (2006) and Yafrate et al. (2007) showed that the L/D ratio (length-diameter ratio) of the T-bar penetrometer should be 4-10 (i.e., the projected area of the T-bar penetrometer is 6.4-15 times the area of the standard cone), which does not affect the dimensional resistance to penetration, and recommended that the L/D ratio of the T-bar penetrometer should generally be greater than 4 [42,43]. Low  that the T-bar penetrometer can estimate the remodeled soil s u and Sensitivity S t , and proposed a series of equations to estimate the relevant design parameters for soft soils [45]. Lunne et al. (2010) analyzed the test results from multiple subsea soft soil sites around the world and discovered that the strength measured using full-flow penetrometer was influenced by strength anisotropy, and that the relationship between the resistance factors of the soft soil remodeling and in situ soils varied primarily with S t , independent of the properties such as the mechanical strength of the soft soil [32].
The resistance factor N T-bar of the T-bar penetrometer can be estimated by using the plasticity theoretical solution of the full-flow theory and the modified penetration resistance q T-bar obtained from the test to estimate s u as follows: In this case, considering the pore pressure and overburden stress for the penetration resistance of the T-bar penetrometer, the penetration resistance can be corrected by the following equation: where q m is the measured penetration resistance, u 0 is the static pore water pressure, and α is the static area ratio; A s is the cross-sectional area of the connection axis; and A p is the projected area of the horizontal axial direction of the tactile penetrometer. Randolph made some small modifications to the above equation, which did not result in significant changes to the calculations and eliminated the necessity for the accurate measurement of u 2 for the T-bar penetrometer [6]. Martin and Randolph et al. (2006) investigated N T-bar based on the plasticity theory and derived their upper limit solution for the friction coefficient α on the surface of the penetrometer and their linear fit [6,46].
This method yields a range of N T-bar from 9.14 (α = 0) to 11.86 (α = 1), and it is generally accepted that a light sandblasting of the penetrometer surface would result in an α of 0.4, with Low suggesting the N T-bar value of 10.5. Randolph et al. give the variation of the resistance factor with the friction coefficient α for the T-bar, ball, and cone penetrometers, as shown in Figure 5 [6].  found that the actual measured N T-bar values were larger when comparing the experimental solution to the theoretical analytical solution for the N T-bar values measured in the vane shear tests and laboratory tests [6]. Randolph analyzed this phenomenon and suggested that the main factors influencing the difference between the theoretical and experimental solutions for N T-bar were sensitivity, strain softening, and strain rate. Randolph and Hloulsby et al. (1984) and Martin and Randolph et al. (2006) derived analytical solutions for the T-bar resistance factors using the ideal elastoplastic intrinsic model based on the Tresca and Mises yield criteria, respectively, according to the limit analysis method [46,47], and Figure 6 shows the assumed stress characteristic lines for the T-bar and ball full-flow penetrometer stresses in the lower limit method, respectively [48].  For the T-bar, the resistance factor N T-bar was based on the theoretical lower and upper bound solutions for a cylindrical plane strain object moving through a rigid plastic medium [47]. Figure 7 demonstrates the upper bound mechanism for both the cylinder and sphere [49]. Figure 8 illustrates the resistance factor N T-bar values for the strain rate correlation and strain softening effects of the T-bar penetrometer [41].

Numerical Simulation
Zhou and Randolph (2009) concluded that natural soft soils are not ideal plastic soils, that relatively high strains are generated in soft soils during penetration, and that strain softening occurs as the soft soil flows through the surface of the T-bar penetrometer [50]. Therefore, in practice, the effect of both the strain rate correlation and strain softening on the penetration resistance factor should be considered in the full-flow penetrometer, and it is necessary to make corrections to the penetration resistance factor.
The association between the shear strain rate and resistance factor was investigated using a variety of numerical methods, with Einav and Randolph (2005) using a combination of an upper limit solution and a strain path method (UBSPM) [49], Klar and Pinkert (2010) using the steady state finite difference method (SSFD) [51], and Zhou and Randolph (2009) using large deformation finite element analysis (LDFE) [50]. All three methods employ a similar logarithmic law to describe the association between the shear strain rate and resistance factor. Einav and Randolph (2005) combined the upper limit solution of the plasticity theory with the strain path method by using the full-flow mechanism and considering strain softening to derive a resistance factor related to the shear strain rate γ and the total plastic shear strain value ξ [49].
where N u,ref is the value of the resistance factor at the strain rate γ ref = 1%/h. The damage factor δ describes the relationship between the shear strength value and the ideal state strength during the gradual softening of soft soils.
where s us and sui are the softened strength and initial strength, respectively; δ rem is the damage factor value for soft soils at full remodeling and is the ratio of the shear strength of fully remodeled soils to the shear strength of de-primed soils, which is the reciprocal of the sensitivity S t ; and ξ 95 is the cumulative plastic shear strain corresponding to the remodeled soils when the remodeling damage reaches 95%. Zhou and Randolph (2009) provide a correction equation for the resistance factor based on the LDFE method considering the effect of strain softening concerning the strain rate dependence [25,50].
where µ is the rate of increase in strength corresponding to a ten-fold increase in strain rate, typically 0.1. By comparing the findings of these three methods, it can be seen that the LDFE analysis yields the most comparable resistance factor values due to its ability to represent the softened shear zone corresponding to the cyclic variation of the cyclic penetration resistance.
The penetration mechanism of the T-bar penetrometer was analyzed by Zhou and Randolph (2009b, 2011) using a large deformation finite element analysis software, as depicted in Figures 9-13 [50,52]. The effects of the penetration rate and strain softening on the resistance factor of the T-bar penetrometer and the range of action effects were analyzed, respectively.    Based on the ABAQUS software, Yang (2018) analyzed the stress change of the soil around the full-flow penetrometer before and after its penetration (Figures 14 and 15), and pointed out that the penetrometer based on the full-flow mechanism would still produce large compressive stress at the lower part of the penetration direction, while a small gap was formed at the connection between the penetrometer and the rod due to the incomplete flow of soil, resulting in small stress [8].   In addition, Randolph and Andersen (2006) used the PLAXIS software to conduct finite element simulations of the penetration test of a T-bar full-flow penetrometer without considering the effect of the penetration rate [53]. The effect of the penetration rate, strength anisotropy, and strain softening on the resistance factor of a T-bar full-flow penetrometer was investigated using the ABAQUS finite element software by Fan et al. (2009) [54].

Laboratory Model Test
In situ testing is a technique used for testing geotechnical properties in the natural environment that has the characteristics of no sampling, simplicity, and speed, and is an effective method for accurately obtaining soil property parameters. Therefore, it is particularly crucial to study the theory and technology of in situ testing based on the field for the investigation of marine soft soil engineering [21].
The in situ testing techniques for evaluating the in situ condition of soils require the accumulation of a large amount of test data, as well as the application of corresponding theoretical derivations and empirical corrections to obtain reliable methods for evaluating soil properties. In situ tests are more time-consuming and cannot determine the variation of a single soil parameter.

1.
Development of the calibration laboratory model tank test In the 1960s, laboratory calibration tanks were first introduced as a common method of laboratory calibration testing. Common calibration canisters include the CRB calibration canister system, the flexible single wall system, and the clay soil calibration canister system. A summary of information and detailed statistics on current calibration tank systems is provided in Table 3 [55]. Ghionna and Jamiolkowski (1991) carried out a laboratory calibration test of an in situ penetration device using the rigid wall test pit method [56]. In this method, a test pit of the required size is dug into the ground, and a test soil sample is inserted for the CPT penetration test. This test was less effective due to the test pit boundary effect.
The flexible wall calibration tank system was developed in Australia in the 1960s to investigate in situ standard penetration tests through laboratory calibration tank tests. Holden (1976) Figure 16 [57,58]. The design theory of this calibration tank system was more mature, and many later improvements based on this type of tank appeared, collectively referred to as CRB calibration tanks. In addition, there is a calibration tank of the University of Oklahoma Calibration, as shown in Figure 17. The CRB flexible double-walled calibration tank has the disadvantage of being complex to operate and to produce soil samples; thus, Villet and Mitchell (1981) designed a flexible single-walled calibration tank that is simple to equip and easy to operate [59]. In 2010, a new flexible single-walled calibration tank device was designed for unsaturated soils at the University of New South Wales, Australia (shown in Figure 16b) [60]. By connecting the soil sample hoop plate inside the single wall to the outer steel wall, this design simplifies both the preparation of soil samples and the control of several boundary conditions.

Boundary state control for calibration tank tests
The flexible wall calibration tank calibration test can be controlled separately for horizontal and vertical boundary conditions, including stress control and strain control. Holden (1976) defined four different boundary control states (shown in Table 4 and Figure 18) [57]. The four states correspond to, respectively, state BC1: horizontal vertical with stress control; state BC2: horizontal vertical with strain control; state BC3: horizontal direction stress control, vertical direction strain control; and state BC4: horizontal direction strain control, vertical direction stress control. Figure 18 depicts a flexible double-wall calibration tank that uses water pressure for stress control in the horizontal direction of the soil sample. Huang (2005) defines a new type of boundary state, called the BC5 state. A schematic diagram of the boundary state and a diagram of the calibration tank setup are illustrated in Figure 19 [62]. In the diagram, σ v is the axial stress, and σ h1~σhn are the transverse stresses, i.e., the force states in the horizontal direction at different locations of the soil sample, which can be controlled.

Factors influencing the calibration tank test
The calibration tank penetration test is an in situ simulation of the in situ site conducted laboratory. Due to the constraints of the soil sample size and the stress state, it does not fully replicate the in situ site conditions, and the data are somewhat different from the in situ test. According to the studies, the influence of the laboratory calibration tank tests is mainly due to boundary conditions and dimensional effects.

•
Boundary effects Parkin and Lunne (1982) investigated the effect of the boundary conditions on the calibrated tank tests at the BC1 and BC3 boundary states [63]. By conducting calibrated tank tests with two different sizes of cones, the effect of the varying boundary conditions and different diameter ratios on the measured cone tip resistance was investigated. The effects of the boundary conditions and size on the resistance of the cone tip is shown in Figure 20. The diameter ratios were investigated by Salgado and Mitchell (1997) using the theory of penetration resistance and controlling the corresponding boundary conditions, soil sample density, and stress state [64]. The study indicates that the measured cone tip resistance is less than the in situ measured value for the BC1 and BC4 boundary conditions. This is due to the attenuated nature of the gain in peritectic pressure compared to the in situ boundary states BC1 and BC4.

•
Size effect Calibrated tank tests on the effect of diameter ratios were carried out by related scholars to derive the effect of diameter ratios on sandy soils [64]. Schnaid and Houlsby (1991) investigated the effect of soil properties on the cone tip resistance by conducting calibration tank tests utilizing three different sizes of cone pressuremeter (corresponding to diameter ratios of 38, 27, and 22) [65]. For dense and medium dense sand, the measured speciated ultimate stresses increased dramatically with the increasing diameter ratios; however, for loose sand, there was no significant effect on the size of the calibration tank.
The theoretical analyses of the causes of dimensional effects on calibration tank tests were focused on two statements. The first statement is by Salgado et al. (1997), who analyzed and explained the dimensional effects of the calibrated tank test using the theory of pore wall expansion [64]. As the penetration pressure gradually grew, the isotropic pressure withstood also gradually increased, and a plastic zone appeared around the pore with a non-linear elastic zone and a linear elastic zone at the periphery (as shown in Figure 19). The second statement, by Wesley (2002), proposes that the change in the vertical stress with cone penetration in the CPT may lead to a decrease in the cone tip resistance as the size of the calibrated tank can decrease [66].

Cyclic Penetration Tests
The T-bar penetrometer tests also measure the resistance during extraction as well as during penetration. The penetration and extraction processes are repeated over a range of depths, and the resistance values are recorded. As the cycle testing continues, the soft soil becomes progressively more disturbed, and its strength gradually diminishes and stabilizes at a certain value after approximately 10 cycles. At this point, the soft soil is considered completely remodeled, and the experiment is completed. Einav and Randolph (2005) employed the cyclic penetration test method to obtain the strength decay curve for the cyclic penetration test data obtained from the T-bar penetrometer, as shown in Figure 21 [49]. The T-bar penetrometer cyclic penetration test [49].
As the cycle progresses, the s u of the soft soil gradually decreases; as reflected in the result, the penetration resistance gradually drops and stabilizes. A graph of the decay of the penetration resistance with the increasing number of penetrations can be made as depicted in Figure 22; the calculated curve and the measured data are compared and analyzed, and on this basis, the parameters δ rem and ξ 95 are used to calibrate [49]. After one lifting cycle, the undrained shear strength of the soft soil produces a decrease of approximately 10%, and generally, after a process of 10 cycles, the soft soil reaches a fully remodeled state, and the undrained shear strength is stabilized. In the derivation of the cyclic resistance, the penetration resistance is normalized by the initial resistance for each cycle. As the soil begins to soften during the first penetration, the final normalized resistance is typically greater than the true value of δ rem . Figure 22. Normalized penetration resistance decays as the cyclic penetration test proceeds [49].
The findings of the cyclic tests can be utilized to calculate the in situ undrained shear strength and the remolded undrained shear strength of soft soils under the initial and final penetration, as well as to assess the sensitivity of the soft soil. Chung et al. (2006) studied shear strength under both drained and undrained loading paths, and the concerning penetration rate noted that the drainage conditions at which the penetration rate affects the shear deformation have a substantial effect on the penetration test result [37]. House et al. (2006) and Randolph (2004) studied the exact relationship between the two loading paths through variable rate penetration tests, as shown in Figure 23 [18,67]. The tests were carried out at a discontinuous logarithmic decrease in the penetration rate, with an initial rate of 20 mm/s. At the beginning of the penetration test, the penetration rate is high, and the soft soil is in undrained conditions. As the penetration rate is gradually reduced, the penetration resistance decreases at the beginning as the cohesive effect diminishes, and as the penetration rate is reduced significantly, the penetration resistance increases significantly with the consolidation effect, until the surrounding soil reaches a state of complete consolidation. During this process, the penetration rate varies in the range of 2-3 orders of magnitude. The data satisfy Equation (8).

Variable Rate Penetration Experiments
The normalized penetration rate V in the equation is defined as follows: (9) where v is the penetration rate, d is the diameter of the full-flow penetrometer, and c v is the vertical coefficient of consolidation for soft soils. House proposes that the consolidation coefficient of the soil can be calculated using the normalized variable rate penetration test described above (House et al., 2001) [67]. As the normalized penetration rate reaches a final value (2)(3)(4) in 50% of the time required for a pore pressure dissipation test in the CPTU, this method can be used to minimize the test duration compared to the conventional CPTU method.
In the case of the T-bar, the flow mechanism of the soil around the penetrometer can be simulated as plane strain during penetration, and the T-bar is assumed to be an infinitely long cylinder. The simple geometries of the T-bar mean that the existing theoretical frameworks can account for the strain rate and strain softening. The numerical models of the soil flow during the T-bar penetration provide insight into the soil behavior and interpretation; however, experience with using theoretical solutions for the T-bar penetrometer remains limited. Therefore, it is necessary to combine the numerical models with specific experimental data.

Application of the T-Bar Penetrometer for Marine Soft Soil Engineering
In recent years, the development of marine engineering has accelerated, placing higher demands on the bearing capacity of soils for the design of marine geotechnical engineering, especially the effects of cyclic loading effects, and assessing the physical and mechanical properties of marine sediments has become a major challenge for the development of marine engineering. For marine soft soil engineering, in situ testing techniques using a full-flow penetrometer can effectively provide the required design parameters. Dejong et al. (2011) concluded that during the penetration process of a full-flow penetrometer, the measured penetration resistance is the result of compressive and shearing forces acting on the penetrometer as the soil is forced to flow around the probe as it is advanced [68]. In comparison to a standard cone penetrometer, the influence of the overburden stress correction on the penetration resistance correction can be significantly reduced, hence, effectively reducing the influence of the correction process on the test reliability.

Evaluation of the Undrained Shear Strength of Soft Marine Soils
Some application tests worldwide were conducted by the University of Western Australia and Southeast University and summarized the empirical values of the T-bar penetrometer resistance factor N T-bar obtained in these tests, as shown in Table 5, where data a are from Lunne [69], data b are from Randolph [18], and data c are from the authors of this paper. The authors of this paper agree with Lunne and Randolph that the N factors derived empirically from s u were found to be in the range of 7.2-14.3 for the T-bar [18,69]. To support the determination of the empirical N T-bar values in the Wenzhou, China test site, the vane shear tests (VSTs) were set at a distance of 1 m from each T-bar penetration test by the authors. A site-specific resistance factor N T-bar can be estimated from a linear-fitted slope of s u versus q net (Figure 24) for the Wenzhou test site based on the VSTs measured strengths, as the reference strengths increase linearly with depth. The calculated average for the resistance factor N T-bar of the spot determined by the VSTs data is 12.0, which is consistent with the study conducted by Lunne [69]. Table 5. Empirical values of resistance factors N T-bar obtained by the University of Western Australia and Southeast University (in this paper) at different sites.

Location N T-bar-DSS , Average
N T-bar-DSS , Rang

N T-bar-FVT , Average
Burswood, Australia 11.9 b -10.9 b Onsoy, Norway 11.9 a , 12.   Tand et al. (1995) stated that the undrained shear strength of soft soils s u can be derived from the ratio of the penetration resistance and the resistance factor through Equation (10) [70,71] as follows: (10) where N T-bar and N kt are the resistance factors for the T-bar and CPTU, respectively. The numerical solution based on the numerical analysis can be used to estimate the undrained shear strength of soft soils based on the penetration resistance, but the numerical solution cannot take into account all the factors affecting the variation of the penetration resistance under actual conditions and does not have a high degree of reliability when applied in practice. Therefore, it is necessary to correct the theoretical numerical solution to obtain the empirical equation by employing the experimental solution obtained from site-specific tests on multiple sites. Figure 25 summarizes Jorat's results for the three penetration resistance factors N for the T-bar, ball, and CPTU at the Onsoy site in Norway, based on the undrained shear strength results of the field vane shear test [70]. The relationship between N and depth, s u , and q net is also summarized. Figure 25. Variation of T-bar, ball, and CPTU resistance factors N with depth, undrained shear strength, and net penetration resistance [70]. Figure 26 illustrates the probability distribution of the resistance factors N for the three penetrometers T-bar, ball, and CPTU in the test, respectively. Based on this, Jorat concluded that the distribution of the resistance factors for the T-bar and ball penetrometers is more concentrated than the distribution of the cone penetrometers in super soft soils, and that the test results are more reliable.

Sensitivity Evaluation of Marine Soft Soils
The sensitivity of marine soft soils S t can be calculated from the in situ undrained shear strength s u and the fully remodeled undrained shear strength s ur measured by the full-flow cycle penetration test of the full-flow penetrometer using Equation (11). (11) where q in is the penetration resistance at the initial penetration, and q rem is the penetration resistance at the end of the complete remodeling cycle. In addition, soil sensitivity is one of the factors influencing the resistance factor N T-Bar , with the value of N T-bar decreasing as the sensitivity of the soil increases. Dejong et al. (2010) summarized the empirical equation for the resistance factor N T-bar with the sensitivity S t from the experimental result [16].
Yafrate et al. (2007) obtained a similar relationship between the remodeling resistance factor N T-bar,rem and the sensitivity of the soil [43]. Figure 27 demonstrates the relationship between the resistance factor N T-bar , the sensitivity S t of the vane shear test, and the extraction factor of the T-bar penetration test. The curve of Equation (6) in Figure 27a is the fitted curve corresponding to Equation (12). Despite the large dispersion in the data, Equation (6) still fits well for soft soil sensitivities of less than 20 (Figure 27a), but there is a large error in Equation (12) for soft soils with high sensitivities.  [43]. Yafrate et al. (2007) described the relationship between the soft soil sensitivity and extraction factor as follows: Combining Equations (12) and (13), the resistance factor N T-bar of the penetrometer increases with the increase of q in /q ext , and has the following relationship: Equation (14) corresponds to the curve of Equation (9) in Figure 27b, which fits well for the resistance factor N T-bar versus q in /q ext .
The authors of this paper modified Equation (13) and developed into Equation (15) from eight samples at two test sites (Wenzhou and Fuzhou) as follows: Equation (15) has R 2 values of 0.95, with a minor increase in the prediction confidence compared with Equation (13) (Figure 28). The results of soil remolding by the full-flow penetrometers were examined through cyclic testing and field vane shear tests at all test sites. Recently, improved full-flow penetrometers were developed and pore pressure sensors were adjusted to obtain parameters in addition to penetration resistance, thereby improving the ability of full-flow penetrometers to estimate geotechnical parameters other than undrained shear strength [72]. This could provide an advantage of the piezo T-bar over the piezocone in the estimation of the in situ consolidation coefficient from the dissipation tests; the dissipation time of the piezo T-bar is usually quicker than that of the piezocone. The authors of this paper developed a set of piezo T-bar penetrometers (with one pore pressure filter at the center or one at the edge) with support from Southeast University (Figure 29). The piezo T-bar penetrometer, featuring pore pressure measurements, is a relatively novel device similar to the piezoball used in dissipation tests, which has the potential to estimate the consolidation parameters for profiling fine-grained soil.

Discussion and Conclusions
As mentioned previously, unlike the piezocone, the full-flow mechanisms of the T-bar or piezoball mean that the resistance values are largely unaffected by high overburden and hydrostatic pressures. Additionally, the larger projected area means a greater resolution in soft soils. The other major benefit of full-flow devices is that they allow for the assessment of the undrained shear strength (both intact and remolded) and the consolidation characteristics in a single test. As a result, some suggestions on future developments of the penetrometers (cone, T-bar, and ball) and associated equipment are recommended to maximize their potential in the characterization of marine soft clays. The main focus has been on lightly overconsolidated clays, with strengths of less than 100 kPa. The selection of appropriate in situ tools (such as the piezocone or T-bar penetrometer) for a site investigation is contingent upon the project requirements, anticipated soil conditions, and specific geotechnical challenges. Table 6 provides a comprehensive overview of the geotechnical issues relevant to marine field developments, including the corresponding soil parameters that can be inferred through in situ testing and their associated reliability. Table 6 serves as a reference for determining the appropriate application of different in situ tests. Notably, the T-bar and ball penetrometers are grouped in the same category due to their closely correlated measured resistances.
In geological formations where accurate stratigraphy and material characterization are essential, the piezocone is recommended as the primary investigative tool due to its extensive usage in deducing material types based on the piezocone parameters. However, when estimating the undrained shear strength, especially in relatively soft materials, the T-bar (or ball) penetrometer should be considered as a complementary instrument. This is because the T-bar (or ball) penetrometer exhibits higher a potential reliability than the piezocone, particularly when the T-bar and ball resistances (q T-bar and q ball ) are correlated with the undrained shear strength values. Moreover, the inferred undrained shear strength derived from the T-bar (or ball) penetration resistance appears to provide a robust predictive basis for assessing the bearing capacity of foundation elements, as illustrated by Watson (1999) [73].
The estimation of the undrained shear strength through a piezocone penetration test exhibits a reduced reliability when applied to backfilled materials compared to the original seabed soil. This discrepancy arises due to the expectedly low cone resistance and pore pressure observed in such materials. To characterize soft clay at shallow depths, the T-bar (or ball) penetrometer test can provide reasonably accurate estimations of the undrained shear strength if executed with meticulous care. Therefore, it is recommended to prioritize the T-bar (or ball) penetrometer as the primary tool, with the vane test serving as a supplementary measuring tool to enhance the reliability of the estimated undrained shear strength. Additionally, the T-bar and ball penetration tests should be performed on box core samples to assess the shear strength profile in the uppermost 0.5 m of the seabed. Note: 1 : high reliability; 2 : high-moderate reliability; 3 : moderate reliability; 4 : moderate-low reliability; 5 : low reliability; γ: soil unit weight; u: in situ pore pressure; OCR: overconsolidalion ratio; K 0 : coefficient of earth pressure at rest; s u : undrained shear strength; s ur : remolded undrained shear strength; S t : sensitivity; c', ϕ': effective stress shear strength parameters; E, G: Young's and shear modulus; M: constrained modulus; G max : small strain shear modulus; k: coefficient of permeability; and c h : coefficient of consolidation.
The consolidation characteristics are usually assessed in situ using piezocone dissipation test data. Meanwhile, the full-flow penetrometers fitted with pore water pressure sensors were developed for determining the consolidation parameters. Recently, piezoball dissipation tests were reported, and theoretical interpretation methods of piezoball were developed that were also found to estimate the coefficient of consolidation to a similar degree of accuracy as the piezocone dissipation tests [12,74]. In addition, the available data suggest that the dissipation around the ball is quicker than the dissipation around the cone. Offshore, where vessel costs are high, this testing efficiency can lead to significant cost savings. The T-bar may be viewed as a model pipeline element, and thus provides direct information for pipeline and riser design, which shows the potential of T-bar penetrometers.

Summary and Outlook
This study reviewed the penetration mechanism and development of the T-bar penetrometer, and the state of its applications in marine soft soil engineering. Then, the numerical simulations and various laboratory experiments of the T-bar penetrometer and in situ testing methods were summarized. After that, the principles for parameter evaluation by applying the T-bar penetrometer in marine soft soil engineering investigations were also described. Due to the complexity of the topic in the present work, there are still many limitations and challenges with the T-bar penetrometer testing technology. The main aspects that need to be investigated are summed up as follows: 1.
An analysis of the theoretical solution of the T-bar penetrometer data. In practical applications, the interpretation of the T-bar penetrometer test data to predict the undrained shear strength and sensitivity of soft soils mainly relies on empirical formulas. However, the evaluated parameters of soft soils through the empirical relationship method lack reliability due to the absence of a large amount of reliable test data. This problem can be solved by gaining a deeper understanding of the mechanism of the T-bar full-flow test evaluation system and the derivation of a more accurate theoretical analytical solution.

2.
The numerical simulation of the T-bar full-flow penetrometer. Recently, the numerical simulation of the T-bar penetrometer is proposed by researchers, which considers the effects of the strain rate, strain softening, and strength anisotropy. However, it is difficult to restore the soil material and the penetration process in the simulation process nowadays, and further development of numerical techniques is needed. 3.
The laboratory model experiment of the T-bar penetrometer. Most of the present research on laboratory model experiments is focused on the traditional CPTU testing of sandy soils, while the research on the penetration mechanism of the T-bar full-flow penetrometer of soft soil is still insufficient. 4.
The T-bar penetrometer field experimental research. The good performance of the T-bar penetrometer technology depends on the large number of accurate field test data, which are used for repeated verification and calibration. In recent years, the T-bar penetrometer technology has mainly been used in Europe and the United States, and the test results are usually available for these areas. However, in many Asian countries, such as China, the research and application of the T-bar penetrometer technology is still in its infancy. Therefore, a large number of field tests still need to be conducted to verify the applicability of the T-bar penetrometer in soft coastal soils in Asian countries.