Experimental Investigation of Local Scour Around A New Pile-Group Foundation for Offshore Wind Turbines in Bi-Directional Current

The local scour around a new pile-group foundation of offshore wind turbine subjected to a bi-directional current was physically modeled with a bi-directional flow flume. In a series of experiments, the flow velocity and topography of the seabed were measured based on a system composed of plane positioning equipment and an ADV. Experimental results indicate that the development of the scour hole was fast at the beginning, but then the scour rate decreased until reaching equilibrium. Erosion would occur around each pile of the foundation. In most cases, the scour pits were connected in pairs and the outside widths of the scour holes were larger than the inner widths. The maximum scour depth occurred at the side pile of the foundation for each test. In addition, a preliminary investigation shows that the larger the flow velocity, the larger the scour hole dimensions but the shorter equilibrium time. The field maximum scour depth around the foundation was obtained based on the physical experiments with the geometric length scales of 1:27.0, 1:42.5 and 1:68.0, and it agrees with the scour depth estimated by the HEC-18 equation.


Introduction
The increase of developments in offshore wind farm planning and construction in recent years has stimulated the urgent need to optimize design and reduce costs in the field of offshore wind technology (Matutano et al., 2013;Petersen et al., 2015;Liu et al., 2015aLiu et al., , 2015b. With the long-term exposure of a wind turbine to a marine environment, its foundation may be subjected to tidal currents, which cause local scour around the foundation (De Vos et al., 2011). Wind turbine foundations are sensitive to scour, which can reduce their stiffness and ultimate capacity and alter their dynamic response Prendergast et al., 2015).
The turbine monopile foundations are optimized for severe conditions if they are reliable and cost-effective (Den Boon et al., 2004;Bilgili et al., 2011). However, the pilegroup foundation has unique advantages, which has excellent structural integrity and good adaptability for soft soil. A new high-rise pile group foundation has already been em-ployed in the East China Sea area for the Shanghai Donghai-Bridge wind turbine, which is the first Asian offshore wind farm (Lin and Zhou, 2010;. The foundation is composed of eight inclined open-ended steel piles and a fixed circular concrete cap placed upon its upper surface (Fig. 1). The slope of inclined piles is 6:1, and the length above the bed surface is 15 m. The top elevation of the pile cap is 6.5 m, and the diameter is 14 m. This foundation has strong vertical, horizontal and flexural bearing capacity and is suitable for an offshore wind farm (Lang et al., 2015). However, because of the special constructs, the flow pattern around the foundation and its mechanism of local scour are complicated.
Local scour around monopile and pile-group foundations has been investigated extensively for many years (Sumer et al., 1992;Fredsøe, 1997, 1998;Melville and Coleman, 2000;Ezzeldin et al., 2006;Hosseini and Amini, 2015). Despite the scale effects, experimental study has still been widely used to investigate the mechanisms of local scour. For instance, Jones and Sheppard (2000) conducted a series of experiments and made predictions on the local scour depth for the composite structure in the steady current. Ataie- Ashtiani and Beheshti (2006) carried out experiments on clear-water local scour at pile groups and derived a correction factor to predict the maximum local scour depth. Amini et al. (2012) presented the results of an experiment on clear-water scour at pile groups in steady flows and proposed a new method to predict the local scour depth. Mostafa and Agamy (2011) conducted an experimental study of scour around a single pile and on different configurations of pile groups exposed to waves and currents and determined that the scour depth for pile groups was generally larger than that for the case of a single pile.
In comparison with the number of studies on the scour in mono-directional steady currents, only a few studies on the local scour in bi-directional currents have been carried out (Nakagawa and Suzuki, 1976;Han and Chen, 2004). Moreover, several researchers have investigated the scour depth around traditional ocean structures, however, there is no published literature on the local scour around the new high-rise pile group foundation as shown in Fig. 1. The scour around the new foundation in bi-directional currents needs to be further investigated.
This study investigated the maximum scour depth and morphology of the scour hole around the new pile-group foundation under bi-directional current conditions via a series of experiments. Some preliminary conclusions for the effect of the flow velocity on the maximum scour depth and equilibrium time were drawn. The prototype maximum scour depth and equilibrium time were obtained by using the method of series models. The scour depth was also compared with the estimation by the HEC-18 equation. The paper is organized as follows: the experimental setup and procedure are described in Section 2. In Section 3, the development and morphology of scour holes and the maximum scour depth are given and analyzed. Finally, conclusions are presented in Section 4.

Experimental setup
The experiments are carried out in a 35 m long, 7 m wide, and 1.6 m deep flume (Fig. 2), which is located at the State Key Laboratory Hydraulic Engineering Simulation and Safety at Tianjin University. The flume is composed of an inlet section, transition section, sand section, another transition section and an outlet section. The sand section starts at 13 m from the flume inlet, and the scale model is installed in the middle of the sand section, located 17.5 m downstream of the flume inlet. The flow is driven by the axial flow pumps at the entrance of the flume. The circulating pipelines are installed on both sides of the flume, and two boards for flow-adjusting located at the transition sections are used to reduce the turbulence of flow. Flow velocity and scour depth are measured using acoustic Doppler velocimetry (ADV).
When using ADV for underwater topographic surveys, the instrument should be placed vertically in water to ensure an accurate measurement. A plane positioning equipment with the dimensions of 4.4 m×4.3 m is used in the experiment (Fig. 3). Two guide rails are placed on both sides of the horizontal framework. A vertical frame device with the dimensions of 4.4 m×0.8 m is also placed on the hori-  The bi-directional current is generated by the flow system, and the flow velocity v is described as: where U is the maximum velocity, T is the period, and t is the time. The water depth remains unchanged during the experiment.

Test conditions and procedure
The Shanghai Donghai-Bridge offshore wind farm is located in the East China Sea, where the semidiurnal tide and obvious characteristics of reciprocating current occur. The maximum flow velocity is 1.73 m/s and water depth is 10 m.
The very fine sand that satisfies the similarity would cause flocculation of sediment and affect the experimental d 50 = 0.10 mm θ c results. In this study, the prototype sand with a median particle size (particle size distribution presented in Fig. 4) and a specific gravity of 2.65 was employed. The method of series models was used to eliminate the errors of dissimilarity (see Section 3.3). Three test series were performed with different scale models (Fig. 1b). According to the prototype scale of the foundation, the dimension of the laboratory site and the parameters of the pumps, the geometric length scales of 1:27.0, 1:42.5 and 1:68.0 were adopted in the experiments. The hydrodynamics were scaled based on the Froude similarity criterion ( Table 1). The experimental parameters are given in Table 2, where the skin friction Shields parameters are computed according to Soulsby (1997). The critical Shields parameter is calculated by the following expression (Soulsby, 1997): (2) where is the non-dimensional grain size, in which is the kinematic viscosity of water and is the ratio between the densities of sand and water. Based on Eq.
(2), the critical Shields parameter for is   JI Chao et al. China Ocean Eng., 2018, Vol. 32, No. 6, P. 737-745 approximately 0.077. The sand section, which was 9 m long and 7 m wide, was located at the center of the flume. The sand bed was leveled after the test model was installed, and the flume was filled with water carefully so as not to disturb the bed surface. Eight monitoring points were positioned around the model foundation to reflect changes in the bed elevation at the foundation and to determine equilibrium scour (Fig. 2). The distance between a monitoring point and the outer edge of a corresponding pile was 5.0 cm in Test 3, and the distance was 10.0 cm in the other tests. The monitoring points were named ① to ⑧ for the convenience of explanation. During the experiment, the monitoring points were measured every two hours, and the duration curves of the scour depth were obtained. Following Han and Chen (2004) and Yu et al. (2016), scour was considered to reach equilibrium when the scour depth stabilized, that is, the scour depths of three consecutive measurements in each monitoring point were approximately equal in this study. The 20-cm-long monitoring profiles No. 1 and No. 3 were placed in the direction perpendicular to the flow, with monitoring Points ① and ⑤ as the center (Fig. 2). Then,Profiles No. 2 and No. 4 were positioned in the direction parallel to water flow, with monitoring Points ③ and ⑦ as the center. Twenty-one points were distributed evenly on each monitoring profile and adjacent points were 1 cm apart. These points were measured every two hours in the tests to develop the monitoring profiles.
After achieving equilibrium, the scour depth of the area around the foundation was measured to obtain the morphology of the scour hole. The size of the measurement area was determined based on the extent of the scour. The measurement area was filled with the measuring points that were at a minimum of 1 cm apart. Then, a smooth topographic map was obtained by interpolation.

Development of the scour hole
According to the model of eight characteristic points that was previously mentioned, measurements were made every two hours to obtain a time series of the scour depth around the foundation. For Test 1 (Fig. 5a), the scour depth increased with time during the scour process. Under the bidirectional flow, the scour depth increased faster in 15 h, and then the depth increased gradually until it reached its equilibrium state at 26 h. The scour depth fluctuated due to the motion of the reciprocating flow. When the flow direction changed from the positive to the reverse, the sediment could not move. When the reverse flow velocity gradually increased to the sediment incipient velocity, the flow would carry sediment into the scour hole to be deposited. Though the oscillation with time was observed, the scour depth overall increased and tended to be stable.
The scour depth of the characteristic Point ① is deeper than that of Point ⑤. The two-way axial-flow pump located on the side of Point ① may cause the current to be tangled, and thus, the flow velocities on both sides were not exactly the same. The varieties of the maximum scour depth of the monitoring point and its location with time for Test 1 are listed in Table 3. Most of time the maximum scour depth occurred at Point ①. The scour depths of Points ③ and ⑦ were slightly smaller than that of Point ① (see Fig. 5a). It is mainly because that Point ① was in the upstream or downstream of the pile, while Points ③ and ⑦ were on the side of the corresponding piles. The deep scour pit was not fully developed to the locations of Points ③ and ⑦.
For Test 2 (Fig. 5b), the equilibrium time of scour around the foundation was approximately 24 h. At the beginning, the deposition occurred at the characteristic Points ①, ④, ⑥ and ⑧. After t= 6 h, the scour started, especially at both sides of the foundation Points ③ and ⑦. For Test 3 (Fig. 5c), the scour depth increased during the time of 4 h to 8 h, and the equilibrium time of scour was approximately 20 h. The scour depth around the foundation fluctuated at a certain period of time because of the reciprocating current. However, the overall trends of the scour depth increased. The development of the scour hole was obvious in the early stage of the experiment but then leveled off. The maximum scour depth levels occurred at both sides of the foundation, such as Points ③ and ⑦. The scour depths were relatively small at Points ②, ④, ⑥ and ⑧.
A time series of the scour depth data for the sections around the foundation were created. For Test 1 (Fig. 6) No. 2 and No. 4 were located along the flow direction. The scour pit patterns after equilibrium were slightly asymmetric for Sections No. 2 and No. 4. The maximum scour depths were approximately 10-11 cm. The evolutions of two sections morphology with time were similar. In the beginning, the scour hole developed rapidly until t=14 h, at which the growth rate of the scour hole slowed until reaching stability.
For Test 2 (Fig. 7), the scour pit patterns in Sections No. 1 and No. 3 were inconsistent. At the equilibrium state, the maximum scour depths along Sections No. 1 and No. 3 were approximately 0.5 cm and 2.5 cm, respectively. The scour holes on Sections No. 2 and No. 4 showed slight differences in morphology, where the scour hole for Section No. 2 was substantially more asymmetric than that of Section No. 4. The maximum scour depths were approximately 2.0-2.5 cm.
For Test 3 (Fig. 8), the scope of the scour was narrow along Section No. 1, and the maximum scour depth was approximately 4 cm. The scope of scour along Section No. 3 was wider than that of Section No. 1 with the maximum scour depth of 1.3 cm. The equilibrium scour pits in Sections No. 2 and No. 4 were basically identical. The maximum scour depths of Sections No. 2 and No. 4 were 3.4 cm and 4.2 cm, respectively. Based on the evolution of the scour pit pattern with time, the scour around the foundation reaches the equilibrium state within the first 10 h. The development of the scour hole was relatively slow for the 10-24 h period.

Morphology and depth of the scour hole around the foundation θ c
In Tests 1-4, the movement of the sand particles was considered to occur because the sand ripples were observed, though the skin friction Shields parameter is slightly  Fig. 6. Scour depth of the monitoring section around the foundation for Test 1.
JI Chao et al. China Ocean Eng., 2018, Vol. 32, No. 6, P. 737-745 θ cr θ c < θ cr smaller than critical Shields parameter in Test 3 (see Section 2.2). According to the study of Camenen and Larson (2005), the sediment transport is often observed when due to the uncertainties of the prediction of critical Shields parameter. The scour pits around each pile were connected in pairs and the erosion in the center of the eight piles was relatively small (Fig. 9). Table 4 summarizes the size of scour hole for each test according to the definition of scour pits in Fig. 9a. As we can see, the maximum scour depth occurred at the side pile of the foundation (i.e. No. 3 or No. 7) for each test. This phenomenon is caused by two reasons. One is the sheltering effect. The scour holes Nos. 1, 2, 4, 5, 6 and 8 are affected by sheltering. The reduction of the velocity and horse-shoe vortex strength leads to the decrease in the scour depths. By comparison, the sour holes No For Tests 1-4, the widths of the scour holes outside the foundation are usually larger than the inner widths ( ), however, this conclusion is not applicable for Test 5 (Fig. 10). In Test 5, it seems that only a few sediment particles around the pile foundation have moved and the sand ripples were not obvious because of the relatively small flow velocity. The sheltering effects are indeterminate and the morphology of bed surface is not completely consistent with the results of the other tests.
In this study, Tests 4 and 5 with different maximum flow velocities were performed to preliminarily investigate the effect of the flow velocity on the scour depth and equilibrium time (Fig. 11). The maximum scour depths were induced by the currents with the velocities of 16.5 cm/s, 26.5  cm/s and 36.5 cm/s were 2.5 cm, 7.0 cm and 8.0 cm, respectively, and the corresponding equilibrium times were 26 h, 24 h and 22 h. It was clear that the scour depth increased with an increase of the flow velocity, however, the equilibrium time decreased with the increased velocity. Moreover, the results preliminarily show that the relationship between the maximum scour depth and flow velocity is not linear, as the scour depth increases at a slower rate as the flow velocity increases.

Prototype maximum scour depth and equilibrium time
The method of series models (Allen, 1947;Sha, 1963) has been used to eliminate the errors caused by the movement dissimilarity of model and prototype sands in the movable-bed model test. The prototype sand can be used in the model scale experiment, and the prototype quantities (e.g., the maximum scour depth and equilibrium time) are calculated as: where represents the prototype quantity, is a coefficient, denotes the model scale, and represents the measurements in the model experiment. In Eq. (3), only and are the unknown quantities. Hence, Eq. (3) can be solved by two different sets of experimental results. The prototype quantities could be obtained when , yielding a maximum scour depth of 5.14 m and an equilibrium time of 67.6 h (Fig. 12).

Comparison with empirical formula for the scour depth
In this subsection, the prototype maximum scour depth obtained from physical experiments was compared with the HEC-18 equation (Richardson and Davis, 2001), which was recommended by Federal Highway Administration. In the empirical formula, the maximum scour depth around pile groups is calculated by is the pile group height factor, is the water depth, is the correction factor for the pile shape, is the correction factor for the bed condition, is the correction  g v a * pg factor for armoring by the bed material size, is the gravitational acceleration, is the flow velocity, and is the effective width of an equivalent full depth pier and is computed as follows: where is the sum of non-overlapping projected widths of the piles, is the coefficient for the pile spacing, is the coefficient for the number of aligned rows.
Based on the field condition, the maximum scour depth estimated by HEC-18 equation is 5.55 m. However, some researches indicate that the maximum scour depth under the mono-directional current is usually larger than that in the case of bi-directional flow (Wang et al., 1989;Lu et al., 2005Lu et al., , 2011. Following the investigations of Han and Chen (2004) and Han (2006), a correction factor of 0.95 was employed in the present study. The finally estimated scour depth is approximately 5.27 m, which is consistent with the conclusion of this research.

Conclusions
The bi-directional flow flume has been established to conduct a series of experiments to investigate the local scour around the new pile-group foundation of offshore wind farms at the Shanghai Donghai-Bridge under the effect of bi-directional flow. Plane positioning equipment and ADV were used to measure the flow velocity and topography. The geometric length scales of 1:27.0, 1:42.5 and 1:68.0 were selected, which had equilibrium times of 26 h, 24 h and 20 h, respectively, under bi-directional flow. After reaching equilibrium, the scour pits would occur around each pile and they were connected in pairs in most cases. Three scaled tests had the maximum scour depths of 14.0 cm, 7.0 cm and 5.0 cm, respectively. The maximum scour depth occurred at the side pile of the foundation (i.e. No. 3 or No. 7) for each test and it was caused by the sheltering effect and the shed vortices based on the special construct of the foundation. For the 1:42.5 scale, three different flow velocities of 16.5 cm/s, 26.5 cm/s and 36.5 cm/s have been conducted to investigate the effect of velocity on the scour depth and equilibrium time. The results showed that the scour depth increased with an increase of the flow velocity, however, the equilibrium time decreased with the increased velocity.
Based on the series model method, the maximum scour depth around the foundation is approximately 5.14 m with a   median size of 0.10 mm of very fine sand seabed, water depth of 10 m and maximum flow velocity of 1.73 m/s. It shows good agreement with the scour depth estimated by HEC-18 equation and is capable of giving the design guideline to provide the scour protection for the pile-group foundations of offshore wind farms.