Field experimental study on stratum classification of marine engineering based on cone penetration test

This study focuses on the intertidal wind farm in Rudong, Jiangsu Province, and aims to enhance the accuracy of marine stratum classification using cone penetration test (CPT) technology. In situ CPT experiments were conducted with a 20 T double-bridge penetration tester to obtain parameters such as cone tip resistance and sleeve friction, and the raw data were optimized through wavelet denoising. The Robertson soil classification method was then applied to calculate the soil behavior type index and achieve continuous, high-resolution identification of seabed strata. The results show that the processed CPT data effectively identify thin layers, interlayers, and interbedded strata in complex marine sediments, revealing clear vertical variation patterns in sediment properties. Overall, the CPT-based stratigraphic classification approach demonstrates clear conceptual logic, operational simplicity, and strong repeatability, providing reliable technical support for foundation investigation, pile design, and engineering evaluation of offshore wind farms.

1 Introduction

During the crucial period of global energy structure transition towards clean and low-carbon, off-shore wind power, with its significant advantages such as abundant resources, high power generation efficiency, and low environmental impact, has become a strategic choice for countries to address climate change and ensure energy security. According to the International Energy Agency (IEA), the global installed capacity of off-shore wind power has grown at an average annual rate of over 20% over the past decade. It is projected that by 2030, off-shore wind power will meet approximately 10% of the global electricity demand (IEA, 2019). However, the marine environment has unique characteristics such as high salt fog, strong winds and waves, and complex geological conditions, which make the construction of marine wind farm much more difficult than those on land (Randolph and Gourvenec, 2011). The stability and safety of the foundation engineering directly determine the lifespan, operational efficiency and economic benefits of the entire wind power project, and have become one of the core bottlenecks restricting the large-scale development of marine wind farm (Byrne and Houlsby, 2003).

The foundation for marine wind farm, as the key structure connecting the wind turbine and the seabed stratum, needs to bear multiple complex loads such as the weight of the turbine, wind load, wave force, tidal current force, and seismic action. The design and construction quality of this foundation rely on the precise understanding of the physical and mechanical properties of the seabed stratum (Kerstin and Peter, 2009). Traditional geological exploration methods such as drilling for sample collection and standard penetration tests have limitations such as low exploration efficiency, high costs, significant sample disturbance, and poor data continuity (Lunne et al., 2009), making them unable to meet the requirements of marine wind power foundation engineering for “high precision, high resolution, and high efficient” of exploration data. In this context, the Cone Penetration Test (CPT) technology, with its unique advantages, has gradually become the core technical means in the field of marine wind power geological exploration (Robertson and Cabal, 2010).

The static probing technology involves slowly pushing a probe with sensors into the ground at a constant speed. It can simultaneously collect parameters such as the resistance at the cone tip, the frictional resistance on the side walls, and the pore water pressure. This method can directly reflect key information about the stratification characteristics, bearing capacity, compressibility, and liquefaction potential of the ground (Robertson, 1991). Compared with traditional surveying methods, the CPT technology has several significant advantages: high surveying efficiency (the time for a single borehole survey is only 1/3-1/5 of that for drilling and sampling), strong data continuity (it can achieve a centimeter-level resolution for stratigraphic profile drawing), minimal sample disturbance (in situ test avoids the damage to the stratum structure during the sampling process), good cost-effectiveness (no need for mud circulation and reduced usage of consumables), and wide adaptability (it can be applied in various strata such as soft soil, sand layers, and complex marine environments) (Lunne et al., 2009; Robertson and Cabal, 2010). Therefore, conducting a systematic study on the application of static probing technology in the foundation investigation of off-shore wind power project holds significant theoretical value and practical significance for optimizing the foundation design, reducing engineering risks, and promoting the high-quality development of the off-shore wind power industry (Kerstin and Peter, 2009; Jardine et al., 2006). Although static probing technology has been widely used in onshore engineering, its applicability in complex marine geological conditions, the optimization methods for data processing, and the accuracy of stratigraphic classification still require systematic research. Especially in the intertidal zone and coastal areas of the ocean, the strata often exhibit obvious non-uniformity, interlayers, and gradual changes, and traditional survey methods are difficult to achieve high-resolution continuous identification. Therefore, this paper takes the tidal flat wind farm in Ruyang, Jiangsu Province as an example. Through on-site CPT tests, combined with wavelet denoising and Robertson soil classification methods, it systematically explores the key technical issues of CPT in the identification of marine strata, aiming to provide an efficient and reliable geological investigation and evaluation method for marine wind power foundation project. Recent advances in soil constitutive modeling further demonstrate that the mechanical response of marine and coastal sediments is strongly influenced by particle-scale thermodynamic processes, cementation, and microstructural evolution. Granular thermodynamic frameworks have been applied to describe the coupled thermal-hydraulic-mechanical behaviors of multi-phase geomaterials (Bai et al., 2025), the constitutive characteristics of coarse-grained soils considering particle breakage (Bai et al., 2025), and the dissociation and strength evolution of hydrate-bearing sediments under temperature-driven and multi-field interactions (Bai et al., 2023). Moreover, experimental studies on bentonite buffer materials under Fe(II)-rich reducing environments have highlighted the significant impacts of chemical conditions on swelling performance and microstructural changes, providing valuable insights into the behavior of fine-grained soils in complex geochemical settings (Zhang et al., 2025). These studies collectively reinforce the importance of coupling CPT interpretations with an improved understanding of soil thermodynamic behavior and micro-mechanisms to achieve more reliable stratigraphic classification and parameter evaluation in marine wind farm foundation investigations. The correlation between soil engineering and mechanical parameters is evaluated by in situ test results such as shown in Table 1. The summary is listed in the following table.

Table 1

Table 1. The correlation between soil engineering and mechanical parameters and in situtests such as CPT (Mayne, 2007; Mayne et al., 2010; Kulhawy and Mayne, 1990; Hegazy and Mayne, 1995).

2 Overview of the test site area

As an example of the in situ CPT test for the 50 MW capacity expansion project of the Jiangsu Longyuan Ruyang off-shore (Intertidal Zone) Wind Farm Demonstration Project, it is located on the intertidal zone along the coast of Jiangsu Province. Figure 1 is situated to the east of the Yungou Port Ring Port Area and outside the channel of Liu Bu Gate, in the sea area. It is located to the northeast of the 150 MW Intertidal Zone Phase I Demonstration Project of Jiangsu Province. The terrain of the site is the coastal intertidal zone to the subtidal zone, with the ground elevation ranging from −0.5 ∼ −5.6 m (based on the 1985 National Elevation Benchmark). The average high tide level is 2.2 m, and the average low tide level is −2.41 m. This project plans to install 20 Goldwind 2.5 MW wind turbines, with a total installed capacity of 50 MW.

Figure 1

Figure 1. Schematic diagram of the location of the intertidal wind farm and test site.

3 Test methods

Figure 2 was used for in situ test. A 15 cm² double-bridge probe was employed. The flow chart is shown in Figure 2 was automatically recorded. The cone tip calibration coefficient was 2.4 kPa, and the side wall calibration coefficient was 0.024 kPa. A LMC-310 microcomputer recorder enables the portable microcomputer to analyze and process the test data. Before the test, the probe is calibrated, and after the test, it is rechecked. The linear error meets the design requirements, and the test results are reliable.

Figure 2

Figure 2. Physical and schematic diagrams of the 20T double-axle probe static penetration instrument.

Generally speaking, the typical CPT test curve has some distinct characteristics for different strata. For instance, the cone tip resistance q_c is high in sandy soil and low in clayey soil, and the friction ratio R_f = f_s/q_c is low in sandy soil and high in clayey soil, etc. The specific soil properties and the description of the test curves are shown in Table 2. According to the cone penetration test data to classify the soil layers (Wu et al., 2024), the following steps should be followed:

1. Divide the resistance-depth curve of the cone penetration test probe into segments. The basis for the segmentation is to conduct a comprehensive segmentation based on the magnitudes of various resistances and the shapes of the curves. If the curve segment with lower resistance, higher frictional resistance, greater excess pore water pressure, and smaller curve variation represents the soil layer, it is mostly clay; while the curve segment with high resistance, low frictional resistance, very small excess pore water pressure, and a jagged, sharply changing shape is sandy soil.

2. The interface depths of each soil layer are accurately determined based on concepts such as critical depth. During the uniform penetration process of the cone penetration test from the surface, the cone head resistance gradually increases (except for the influence of the hard shell layer), and reaches a relatively constant value at a certain depth (the critical depth). The critical depth and the first relatively constant value segment of the curve represent the first layer. When the probe continues to penetrate to the vicinity of the second layer, the resistance of the probe will be affected by the combined influence of the upper and lower soil layers and change (either increasing or decreasing). Generally, the depth in the middle of the curve change segment is the layer depth. The second layer also has a relatively constant value segment, and this pattern continues downward.

3. After the above two steps, the probe resistance parameters of each layer of soil should be calculated and averaged respectively. The average values can be used to determine the names of soil layers and soil layer (category) names. The method for determining these names can be based on various empirical graphs. The soil layer profile of the site can also be obtained using the multi-hole static penetration curve.

Table 2

Table 2. Characteristics of the relationship between CPT test curves and soil properties.

4 Test results

The wave forms of the lateral friction force and the cone tip resistance curve in the original test data fluctuate greatly, with long jagged-like abrupt changes, damage, irregularity, and local peaks that are interlaced and spaced closely, with low human readability. Therefore, when classifying soil properties, the original results of the in situ test are first optimized using the wavelet denoising method. The basic principle of this method is to use a set of functions to represent or approximate a signal or function. During in situ test, it is inevitable to be affected by the external environment, or due to human operation and systematic deviations, resulting in “noise” errors in the test data. To objectively reflect the true test data, it is very necessary to separate the test results from the noise, and this is also the basic basis for stratigraphic classification. Using wavelets for signal denoising and reconstruction is a currently widely applied method. It can observe the signal from coarse to fine, obtaining the overview of the signal on a large scale and the details on a small scale. It is accomplished through three steps: ①Select a wavelet and perform N-level wavelet decomposition on the signal; ②For each layer of the coefficients obtained from the decomposition, select a threshold and apply soft thresholding to the detail coefficients; ③Use the processed coefficients to reconstruct through wavelet.

Using the Matlab platform, the wavelet basis and the denoising scale (Zhang, 2012) were selected. The spacing was set to 0.25, 0.5 or 1.0 m. The original test data and the data after setting the interval for denoising (Liu and Wu, 2004) are shown in Figures 3, 4 as (a), (b), and (c). From the original data, it can be seen that there are many burrs in the graph. This is because the test results are often interfered by various uncertain factors, such as high-frequency noise from the acquisition equipment and systematic deviations in the low-frequency part. The denoised data eliminates the relatively blurry fluctuation phenomena in the original data, retains the overall regularity and detailed signals of the original data, having the typical variation characteristics of CPT test data. It can better identify thin layers, interlayers and interbedded layers in the strata. It can be seen that it has removed most of the noise in the signal, with almost no distortion, and has higher recognition accuracy. The larger the interval scale, the fewer curve inflection points and the smoother the curve. However, the interval cannot be too large, otherwise it will lead to the distortion of the effective data. An interval of 0.5 m or 1.0 m is relatively reasonable, which can grasp the overall stratum pattern and increase the determination efficiency.

Figure 3

Figure 3. CPT test indicators at Hyjt0001 measurement point along the test depth curve and its denoised curve (f_s is the blue line in the figure, MPa and q_c are the red lines in the figure, MPa): (a) Original curve. (b) 0.25 m interval fitting. (c) 0.5 m interval fitting.

Figure 4

Figure 4. CPT test indicators at Hyjt0002 measurement point along the test depth curve and its interpolation curve (f_s is the blue line in the figure, MPa and q_c are the red lines in the figure, MPa): (a) Original curve. (b) 0.5 m interval fitting. (c) 1.0 m interval fitting.

5 Classify soil layers based on the CPT test indicators

5.1 CPT test index calculation I_c

The CPT method for classifying soil types based on the classification index of stratum soil properties does not provide precise names for soil particles based on the particle size distribution of the soil. Instead, it offers an engineering classification based on the mechanical properties of the soil. In European and American countries, the CPT soil classification technology is often named as “soil behavior type” (SBT), and the stratum state is reflected by the soil class index I_c (Lunne and Kleven, 1984). Robertson, (2009) defined the soil class index as shown in Formula 1 as follows:

$I_{\mathrm{c}}={\left({\left(3.47-\log Q_c\right)}^2+{\left(\log F_r+1.22\right)}^2\right)}^{0.5}(1)$

$Q_c=\left(\frac{q_c-{\sigma }_{v0}}{P_{a2}}\right){\left(\frac{P_a}{{{\sigma }_{v0}}^{\prime }}\right)}^n(2)$

$F_r=\left[f_s / \left(q_c-{\sigma }_{v0}\right)\right]\times 100\%(3)$

Among them, q_c represents the tip resistance, f_s represents the side resistance of the cone, σ′_v0 represents the effective overburden pressure, σ_v0 represents the total overburden pressure, n is the stress index, as shown in Formula 3 is a reference pressure with the same unit as the effective overburden pressure σ′_v0 (when the unit of σ′_v0 is kPa, P_a = 100 kPa), and P_a2 is a reference pressure with the same unit as q_c and σ_v0 (when the units of q_c and σ_v0 are MPa, P_a2 = 0.1 MPa). Robertson classified the soil properties based on different soil index I_c and soil characteristics. When I_c > 2.6, a simplified linear stress relationship can be used to establish the soil property classification table (n = 1). When I_c ≤ 2.6, according to the statistical law of the test scatter points, n is 0.5. Therefore, when I_c ≤ 2.6, as shown in Formula 1 is calculated using n = 0.5, and the I_c value is recalculated based on the new Qc value. If the newly calculated I_c is still less than 2.6, the calculation classification should be performed using a stress index of n = 0.5. If the iteration result is greater than 2.6, a stress index of n = 0.75 should be selected for the calculation classification.

Using Robertson’s calculation formula for the soil index I_c, the variation curves as shown in Table 3 along the depth at the two measurement points were calculated. At the same time, the area was divided into six regions based on the boundaries. The I_c boundaries were 1.31, 2.05, 2.6, 2.95 and 3.6. The Robertson soil classification diagram is shown in Figure 5. Among them, when I_c is greater than 3.6, it represents the VI zone, which is for organic soil and peat soil; when I_c is between 2.95 and 3.6, it represents the V zone, which is for clayey soil to silt-clay soil range; when I_c is between 2.6 and 2.95, it represents the IV zone, which is for silt-mixed soil type, from silt-clay soil to clayey-silt soil range; when I_c is between 2.05 and 2.6, it represents the III zone, which is for sandy-mixed soil type, from sandy-silt soil to silt-sandy soil range; when I_c is between 1.31 and 2.05, it represents the II zone, which is for silt-sand to clean sand range; when I_c is less than 1.31, it represents the I zone, which is for dense sandy soil to gravelly sand range.

Table 3

Table 3. Robertson soil classification indicators based on I_c(Robertson, 1991).

Figure 5

Figure 5. Based on Robertson’s CPT soil classification chart (Robertson and Robertson, 2006).

5.2 Based on the I_c classification, the seabed strata are divided

Based on the I_c–depth profile at Hyjt0001 (Figure 6), the soil behavior varies continuously with depth and can be divided into several intervals with distinct engineering implications. Although the I_c curve represents a continuous transition of soil properties, grouping the depth profile into segments helps highlight the dominant behavior within each interval.

Figure 6

Figure 6. Soil property index I_c curve along depth at Hyjt0001 measurement point.

From 0 to 5.0 m, I_c values fall within Zone II, indicating sandy silt to silty sand. The curve position near the upper boundary reflects more sand-like behavior, whereas values approaching the lower boundary indicate increased silt content. Between 5.0 and 8.3 m, I_c ranges within Zone III, suggesting a sand–silt mixture with gradual internal changes. In contrast, the 8.3–10.3 m interval shows a clear shift to Zone V, corresponding to silty clay to clay, which forms a distinct weak layer.

From 10.3 to 19.0 m, the soils revert to Zone III characteristics, but the I_c curve shows a subtle upper–lower trend: the 10.3–16.0 m segment is more sand-dominated, while 16.0–19.0 m displays increased silt content. The 19.0–23.0 m interval transitions into Zone IV, representing clayey silt to silty clay. Below this depth, from 23.0 to 40.1 m, I_c values mostly fall in Zone V, though thin layers of silt (35.0–37.0 m) and sandy clay (37.0–40.1 m) interrupt the overall clayey profile. The deepest section (40.1–50.1 m) again shows Zone III behavior, dominated by sandy silt with minor variations.

These observations demonstrate that I_c-based classification effectively captures the internal layering and subtle transitions that are easily overlooked in traditional empirical interpretations. The method provides a structured and consistent framework for identifying weak layers, assessing soil variability, and supporting engineering decisions related to foundation design.

To further demonstrate the advantages of the I_c-based stratigraphic classification over traditional manual interpretation, a comparative analysis was conducted using measurement point Hyjt0001. In conventional q_c–f_s curve–based manual delineation, the interval from 5.0 to 10.0 m is commonly classified as a single “silty soil layer” or a “silt–silty clay transitional layer.” However, the I_c profile (Figure 6) reveals clear internal differences in engineering behavior that are not captured by manual judgment. The segment from 5.0 to 8.3 m has I_c values mainly within 2.05–2.60 (Zone III), indicating a sand–silt mixture, whereas the 8.3–10.3 m interval exhibits Ic values rising to 2.95–3.60 (Zone V), corresponding to silty clay to clay. Under manual interpretation, this distinct weak clayey interlayer is typically merged into a broader unit, failing to reflect its true engineering properties.

This discrepancy is of considerable engineering significance. Treating the entire 5.0–10.0 m interval as a single layer using manual methods may lead to inaccurate estimates of pile side resistance, consolidation behavior, and weak-layer thickness, thereby affecting decisions on pile type and required embedment depth. In contrast, the I_c-based method can clearly identify the abrupt change in soil behavior and accurately highlight the 8.3–10.3 m weak interlayer, providing more reliable input for pile foundation design and reducing potential engineering risks.

Similarly, within the 23.0–40.1 m depth range, manual interpretation typically categorizes the entire section as “clayey soil,” whereas the Ic profile identifies a thin silt layer at 35.0–37.0 m and a sandy clay interlayer at 37.0–40.1 m. These subtle variations are often overlooked through manual experience-based judgment but have direct implications for evaluating consolidation characteristics, settlement behavior, and pile toe performance. Therefore, the I_c-based approach not only improves stratigraphic resolution but also effectively reduces subjective misinterpretation arising from operator experience variability.

Having understood the regularity of the gradual change in the soil properties of the strata, that is, from the distribution curve in Figure 6, it can be determined that: from 0 to 5.0 m, it is sandy silt to pure sand soil; from 5.0 to 8.3 m, it is silt-sandy soil to sandy silt type; from 8.3 to 10.3 m, it is clayey soil; from 10.3 to 19.0 m, it is sandy silt to sandy silt type; among which from 10.3 to 16.0 m is close to the upper boundary of area 5, which can be defined as sandy silt; from 16.0 to 19.0 m is close to the lower boundary of area 5, which can be defined as silt; from 19.0 to 23.0 m, it is clayey sandy soil to sandy clay type; from 23.0 to 40.1 m, it is sandy clay to clayey soil type range; among which there is a thin layer of silt in the interval of 35.0–37.0 m, and there is an interlayer of sandy clay in the interval of 37.0–40.1 m. From 40.1 to 50.1 m, it is sandy silt layer, except for the thin layer of silt in about 42.0 m, the distribution of sandy silt type soil is relatively uniform, which is basically consistent with the classification test results and is more efficient. Compared with the empirical judgment method, the engineering concept is also clearer.

If the lateral friction force and the cone tip resistance at the CPT-Hyjt0001 measurement point are denoised and fitted with a precision of every 0.25 m, and then the soil property index is calculated, and compared with the curve of 0.5 m precision, as shown in Figure 7, the distribution curve of the soil property index and the corresponding soil layer naming range.

Figure 7

Figure 7. Soil property index I_c at the Hyjt0001 measurement point along the depth curve (with 0.25 m accuracy and 0.5 m accuracy).

At a 0.25 m accuracy, the curves exhibit a more jagged distribution, making it easier to make detailed judgments. However, when the curves overlap and intersect, the judgment becomes somewhat complicated. Compared to the 0.5 m accuracy curves, the regularity is basically the same. At different depths, the range where the curves fall is almost without deviation. Moreover, the 0.5 m accuracy curves can eliminate overlapping curves for judgment. Thus, in the foundation bearing layer or the selection of mechanical parameters for important strata, high-precision curves can be used for simultaneous judgment to grasp the overall pattern.

Taking the CPT-Hyjt0002 measurement point with 0.5 m accuracy and 1.0 m accuracy as an example, as shown in Figure 8, for non-essential projects, judging by 1.0 m accuracy can also meet the requirements, and it can also meet the requirements for identifying the continuous distribution characteristics of the strata. According to the different color ranges of the curve varying with depth, the strata can be classified into soil types and soil properties based on the corresponding names. The discrimination method is clear and straightforward, which can greatly reduce human errors.

Figure 8

Figure 8. Soil property index I_c at the Hyjt0002 measurement point along the depth curve (with 0.5 m accuracy and 1.0 m accuracy).

As shown in Figures 9, 10, for the sake of illustration, the test values with a 0.5 m accuracy were calculated for the CPT-Hyjt0003 and Hyjt0012 measurement points within the site area. Hyjt0003 shows that from 0 to 4.0 m, it is either pure sand or silt-like soil, and the curve is closer to sandy mixed soil. Therefore, it can be basically determined as silt-like soil. While Hyjt0012 shows that from approximately 0–7.0 m, it is silt-like soil. Based on the static exploration holes laid out on the site and the measured data, it can also reveal the undulating characteristics of the site’s strata. The curve forms of the two measurement points are different. Hyjt0003 reflects that there are weak interlayers in the shallow part of the stratum, and there are two weak soil layers at 4.0–7.0 m, and soft clay from 9.0 to 13.0 m. While Hyjt0012 reflects that above 8.0 m, it is all silt-like soil with better soil properties, and soft clay from 8.0 to 11.0 m. The stratum distribution of this site is more uniform than that of Hyjt0003. This is also an important feature of the complex stratum distribution in marine sites, that is, the stratum deposition is less uniform than on land. Some stratum distribution ranges, lateral and longitudinal characteristics vary greatly.

Figure 9

Figure 9. Interpolation of 0.5 m accuracy at Hyjt0003 measurement point and distribution of soil property index I_c.

Figure 10

Figure 10. Interpolation of 0.5 m accuracy at Hyjt0012 measurement point and distribution of soil property index I_c.

The deeper strata, Hyjt0003, are located below 36.0 m, and the measurement point Hyjt0012 is situated below 33.0 m. Both show fine sandy soil with good properties and uniform distribution. The cone tip resistance q_c of the static exploration boreholes ranges from 8.46 to 13.66 MPa, with an average value of 11.46 MPa. The burial is moderate, the thickness is large, and combined with the standard penetration count, the density is classified as dense soil. Based on the compressibility, it is classified as medium compressibility soil. The engineering characteristics are good, and it can be used as the bearing layer at the end of the pile foundation.

In order to further highlight the stratigraphic structure characteristics and spatial differences of different measuring points in the study area, the CPT analysis results of Hyjt0001, Hyjt0002, Hyjt0003 and Hyjt0012 were comprehensively compared. In general, the shallow strata (0–5 m) in the study area are generally composed of sandy silt-silt mixed layers, reflecting the common characteristics of high-energy sedimentary environment in the intertidal zone. However, since below 5 m, the stratigraphic structure of each measuring point began to show significant differences: Hyjt0001 had a significant weak interlayer of cohesive soil at 8–10 m, while Hyjt0003 had two weak thin layers at 4–7 m. In contrast, the 0–8 m of Hyjt0012 is a more continuous silty sand layer, and the formation uniformity is obviously better than that of other measuring points. In deep strata (below 30 m), continuous sandy silt layers are generally developed at each measuring point, but there are still differences in thickness and uniformity. For example, both Hyjt0003 and Hyjt0012 show a thick layer of dense silt-fine sand, but the same depth section of Hyjt0001 still contains local thin layer of silt or sandy clay interlayer. This fluctuation and interlayer development characteristics show that the study area is significantly affected by the sedimentary dynamics of the intertidal zone and the transport of the ancient water system, and the lateral continuity of the strata is limited.

6 Conclusion

The actual strata are extremely complex in terms of their sedimentary history and formation environment. They are generally non-uniform and gradual in nature, often with one type of soil mixed with another. This is a common geological law of the Quaternary sedimentary strata. On the basis of understanding the basic geological engineering concepts, reasonable and economically appropriate stratigraphic classification methods should be adopted. Through the discussion and application research in this study, the following brief conclusions can be drawn.

1. It is quite necessary to optimize the original test data, which can eliminate noise errors, smooth the test curve, enhance the human readability, and grasp the overall pattern of the test curve.

2. The chart classification method of the geotechnical index of strata can obtain the continuous indicators of stratum characteristics along the depth, thereby enabling us to understand the progressive change phenomenon of the distribution characteristics of stratum geotechnical properties along the depth. At the same time, by integrating the test data from multiple boreholes in the site, we can grasp the distribution law and undulation situation of stratum geotechnical properties, which is of great significance for foundation engineering, especially for pile foundation engineering.

3. The stratigraphic classification method based on the CPT soil property index and its application examples reveal that this method has a clearer and more definite judgment concept. Especially for complex strata with interbedded and intercalated layers, it reduces human operational errors and misjudgments. It is a rapid, retestable, high-precision, and quantitative analysis-based judgment method, and is worthy of promotion and application.

4. Future research may focus on integrating I_c-based stratigraphic characterization with three-dimensional geotechnical modeling, refining stress-index selection for different marine geological conditions, and expanding the method to CPTU-derived parameters or machine-learning-assisted classification. Such developments can further enhance the predictive capability and engineering applicability of CPT-based stratigraphic interpretation in offshore wind farm investigations.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions

HX: Writing – original draft. TL: Writing – review and editing. YG: Writing – review and editing. ZC: Writing – review and editing. ZM: Writing – review and editing.

Funding

The authors declare that no financial support was received for the research and/or publication of this article.

Conflict of interest

Authors HX, YG, and ZC were employed by Northwest Engineering Corporation Limited.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The authors declare that no Generative AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

Bai, B., Zhou, R., Yang, G., Zou, W., and Yuan, W. (2023). The constitutive behavior and dissociation effect of hydrate-bearing sediment within a granular thermodynamic framework. Ocean. Eng. 268, 113408. doi:10.1016/j.oceaneng.2022.113408

CrossRef Full Text | Google Scholar

Bai, B., Wu, H., Zhou, R., Wu, N., and Zhang, B. (2025a). A granular thermodynamic framework-based coupled multiphase-substance flow model considering temperature driving effect. J. Rock Mech. Geotech. Eng. 17, 5816–5828. doi:10.1016/j.jrmge.2024.11.017

CrossRef Full Text | Google Scholar

Bai, B., Zhang, B., Chen, H., and Chen, P. (2025b). A novel thermodynamic constitutive model of coarse-grained soils considering the particle breakage. Transp. Geotech. 50, 101462. doi:10.1016/j.trgeo.2024.101462

CrossRef Full Text | Google Scholar

Byrne, B. W., and Houlsby, G. T. (2003). Foundations for offshore wind turbines. Phil. Trans. R. Soc. A 361, 2909–2930. doi:10.1098/rsta.2003.1286

PubMed Abstract | CrossRef Full Text | Google Scholar

Hegazy, Y. A., and Mayne, P. W. (1995). Statistical correlations between Vs and CPT data for different soil types. Proc. Int. Symposium Cone Penetration Test. 2, 173–178.

Google Scholar

International Energy Agency (IEA) (2019). Off-shore wind outlook:world energy outlook special report.

Google Scholar

Jardine, R. J., Standing, J. R., and Chow, F. C. (2006). Some observations of the effects of time on the capacity of piles driven in sand. Geotechnique 56 (4), 227–244. doi:10.1680/geot.2006.56.4.227

CrossRef Full Text | Google Scholar

Kerstin, L., and Peter, H. (2009). Design of monopile foundations for offshore wind energy converters. Contemp. Top. Deep Found., 512–519. doi:10.1061/41021(335)64

CrossRef Full Text | Google Scholar

Kulhawy, F. H., and Mayne, P. W. (1990). Manual on Estimating Soil Properties for Foundation Design. Palo Alto, CA: Electric Power Research Institute (EPRI), 306p. EPRI Report EL-6800.

Google Scholar

Liu, S. Y., and Wu, Y. K. (2004). On the state -of- art and development of CPT in China. Chin. J. Geotech. Eng. 26 (4), 553–556. doi:10.3321/j.issn:1000-4548.2004.04.025

CrossRef Full Text | Google Scholar

Lunne, T., and Kleven, A. (1984). Role of CPT in north sea foundation engineering. Norw Geotech. Inst. PublN139, 14P. 21 (3), 76–107. doi:10.1016/0148-9062(84)91773-X

CrossRef Full Text | Google Scholar

Lunne, T., Robertson, P. K., and Powell, J. J. M. (2009). Cone penetration testing in geotechnical practice. Soil Mech. Found. Eng. 46 (6), 237. doi:10.1007/s11204-010-9072-x

CrossRef Full Text | Google Scholar

Mayne, P. W. (2007). “Cone penetration testing-A synthesis of highway practice,” in NCHRP synthesis 368. Washington, DC: Transportation Research Board (TRB), 117p.

Google Scholar

Mayne, P. W., Peuchen, J., and Bouwmeester, D. (2010). Estimation of soil unit weight from CPTs. Proc. 2nd Int. Symposium Cone Penetration Test. (CPT’10, Huntingt. Beach, CA) 2, 169–176.

Google Scholar

Randolph, M. F., and Gourvenec, S. M. (2011). Off-shore geotechnical engineering. Boca Raton, FL: CRC Press. doi:10.1201/9781315272474

CrossRef Full Text | Google Scholar

Robertson, P. K. (1991). Soil classification using the cone penetration test: reply. Can. Geotech. J. 28 (1), 176–178. doi:10.1139/t91-024

CrossRef Full Text | Google Scholar

Robertson, P. K. (2009). Interpretation of cone penetration tests — a unified approach. Can. Geotech. J. 46, 1337–1355. doi:10.1139/T09-065

CrossRef Full Text | Google Scholar

Robertson, P. K., and Cabal, K. L. (2010). Guide to cone penetration testing for geotechnical engineering. Signal Hill, CA: Gregg Drilling and Testing Inc, 138.

Google Scholar

Robertson, P. K., and Robertson, K. L. (2006). Guide to cone penetration testing and its application to geotechnical engineering. California. Signal Hill, CA: Gregg Drilling and Testing Inc.

Google Scholar

Senneset, K., Janbu, N., and Svan, G. (1982). “Strength and deformation parameters from cone penetration tests,” in Proceedings of the European symposium on penetration testing, 863–870.

Google Scholar

Wu, Z. Q., Li, T. R., Li, Q., Duan, W., Zhan, Z. N., and Cal, G. J. (2024). Research progress on the application of cavity expansion theory in in-situ testing. Mech. Eng. 46 (4), 689–702. doi:10.6052/1000-0879-23-221

CrossRef Full Text | Google Scholar

Zhang, D. (2012). Engineering application of MATLAB wavelet analysis. Beijing: China Machine Press.

Google Scholar

Zhang, M., Tian, L., Xie, J., Hu, D., Zhou, F., Zhang, Y., et al. (2025). Experimental study on the swelling performance of GMZ24 bentonite buffer material under Fe(II) reducing environment. Nu. Eng. Des. 442, 114280. doi:10.1016/j.nucengdes.2025.114280

CrossRef Full Text | Google Scholar