CB Center of buoyancy CG Center of gravity FOWT Floating offshore wind turbine MSL Mean sea level RAO Response amplitude operator RNA Rotor nacelle assembly STL Spring tensioned leg TLP Tension leg platform UOU University of Ulsan

Recently, some floating offshore wind turbines (FOWT) have been developed and deployed in deep sea, while a large number of offshore wind turbines with fixed foundations have been installed in water depths less than 50m water deep supporting 3~5MW rotor nacelle assembly (RNA). The installation of two fixed offshore wind turbines at 44m in water depth was made in the Beatrice Wind Farm Demonstrator Project (Wikipedia, 2013).

Several researches on FOWT have been made. Bulder et al. (2002) analyzed a tri-floater platform wind turbine; Lee (2005) studied a 1.5-MW wind turbine; Wayman et al. (2006), Sclavounos et al. (2007), Wayman (2006), Jonkman et al. (2009), Jonkman (2010), Jensen et al. (2011) and Wang and Sweetman (2012) analyzed various tension leg platform (TLP), spar, semisubmersible and barge substructures of FOWT. Also a few floating wind turbine model tests have been performed in wind and waves. Hydro Oil ＆ Energy performed a scale 1/47 model test of a 5MW spar-buoy floating wind turbine at Marintek’s Ocean Basin Laboratory in Trondheim, Norway (Skaare et al. 2007). Principal Power Inc. carried out a scale 1/67 model test of a semisubmersible platform, WindFloat (Roddier et al. 2010). In this model test, a disk was used instead of three blades to obtain aerodynamic thrust forces. WindSea of Norway was tested at Force Technology on a 1/64 scale of tri-wind turbine semisubmersible platform (Windsea, 2013). Reynolds scale was used in wind tunnel and Froude scale in basin, respectively. Model tests of OC3-Hywind were carried out at the wide tank, the University of Ulsan (UOU) on a 1/128 scale and the scale model was moored by 3 catenary mooring lines (Shin, 2011) and moored by a spring-tensioned-leg (STL) (Kim, 2011).

To produce electricity with higher efficiency at lower costs in deep sea, however, it is necessary to consider building wind farms with plenty of FOWTs, not a single FOWT. Then, FOWTs should require not only smaller foot prints to prevent mutual interferences among them, but also installation cost lower than those of existing FOWTs.

In this paper, a new substructure of FOWT satisfying both smaller foot prints and lower installation costs in 50m of water is suggested to support a 5-MW RNA, Jonkman et al. (2009). Its characteristics are as follows: • An inverted conical cylinder type with heavy ballast weight at its bottom to ensure that center of gravity (CG) is lower than center of buoyancy (CB) • A tensioned mooring line with a spring to secure small foot print and low dynamic tensions with a quick installation.

Model tests with a 1/128 scale ratio were carried out in the Ocean Engineering Wide Tank of UOU to predict the characteristics of motions of the FOWT platform in wind and waves. Comparisons are made between the inverted conical cylinder in 50m of water and the OC3-Hywind moored by a STL at 320m deep.

Based on the OC3-Hywind moored by a STL in 320m of water (Kim, 2011), the scale 1/128 model of 5MW wind turbine with an inverted conical cylinder platform was designed for shallow water applications as shown in Fig. 1. It is a three blade horizontal axis reference wind turbine with 90m in hub height. The platform in Fig. 2 consists of two parts. Its upper part is a cylinder to connect to the tower base and its lower one is an inverted conical cylinder with a large ballast plate on the bottom. The inverted conical cylinder was selected for achieving high CB in shallow water and the heavy ballast plate at the bottom was designed for low CG. The large diameter of ballast bottom plate increases yaw inertia of platform. In this platform, CG locates at 23.6m and CB locates at 16.9m below mean sea level (MSL). The model was moored by a STL. Also a spring case for limited extension of spring was installed inside the platform.

Fig. 2 illuminates its detailed drawing with a spring case inside. Principal particulars of the floating offshore wind turbine are given in Table 1.

The model tests were performed in the Ocean Engineering Wide Tank, UOU (L × B × D × D_w = 30m × 20m × 3.0m × 2.5m). An artificial bottom plate (L × B = 18m × 2.0m) was manufactured to satisfy the shallow water boundary condition with 0.391m in water depth. The model was installed at 15m downstream of the wave generator and a wave probe was placed to measure the wave elevation as shown in Fig. 3.

Four passive makers were mounted on the tower of the model in Fig.1 to measure motions in six degrees of freedom by eight VICON cameras (Fig. 3). Before the model test, wind speed was measured by 12 anemometers at the position where the model would be installed. The wind generator produces mean wind speeds up to 10m/s. Its dimension is 2.0m in height and 3.5m in width. Test data was recorded in 100 seconds for regular waves and in 256 seconds for irregular waves.

The principal dimensions are shown in Fig. 4. Four triangle plates, which are made of light plastic material, were attached to the platform for reducing surge and pitch motions of the platform. These plates were used to get large drag forces at lower part of platform.

The model test was carried out in regular and irregular waves without/with wind and rotating rotor to obtain the response amplitude operator (RAO). Wave generator produces 13 regular waves and 4 irregular waves as shown in Tables 2 and 3. An electricity motor is used for driving the rotor. In this test, wind speed and rotor speed are based on the rate wind speed of the NREL 5-MW wind turbine. The 5MW wind turbine rotor operates with rotor speed 12.1rpm at 11.4m/s mean wind speed. Using Froude scaling, rotor speed and mean wind speed of model test are determined as hereunder:

• Rotor speed of scale model = 136.9rpm • Mean wind speed of scale model = 1.007m/s • Two load cases are defined as follow: • LC1: Regular waves, no wind, parked rotor • LC2: Regular waves, mean wind speed and rotating rotor

ISSC wave spectrum was applied to 4 irregular waves as in Table 3. Based on these waves, 2 load cases were defined as follows: LC3: Irregular waves, no wind, parked rotor LC4: Irregular waves, mean wind speed, rotating rotor

Where, H_s is significant wave height and T_p is peak-spectral wave period.

These irregular waves were produced during 256 seconds (about 48minutes at full scale). Comparisons between the theoretical wave spectra and the measured spectra are shown in Fig. 5. There is a good agreement between them.

LC1

Fig. 6 shows the RAOs obtained from model tests of the OC3-Hywind moored by a long STL in 320m of water and the inverted conical cylinder moored by a short STL in 50m of water, respectively. The comparison is made to see how large the motion of inverted conical cylinder is. In surge and heave RAOs, both FOWTs show similar responses. The pitch RAO of the inverted conical cylinder is smaller than the one of the OC3-Hywind below 0.5rad/s. Both models have natural frequencies around 0.26rad/s in surge and pitch. The heave natural frequency of the inverted conical cylinder is not clearly shown in RAO because of large heave damping from the ballast weight bottom plate. Yaw restoring moment is very small because of one STL mooring line. In real sites, both FOWTs need to employ a torque-balanced tensioned-leg and/or a yaw controlling device to keep the turbine operating in upwind direction.

LC2

LC2 includes both wind and wave conditions. The scaling law for aerodynamic loads of FOWT has not been established and still be in dispute for application in basin model test. Froude scaling in this model test was applied to produce wind speed and the rotor-thrust might be underestimated. Both thrust and torque can be representative of those in full scale with both wind speeds and blade pitch angles controlled properly (Martin, 2011).

Due to wind and a rotating rotor, the platform drifts to a new equilibrium position and oscillates around the new position. As can be seen in Fig. 7, the inverted conical cylinder is drifted around 30mm (3.84m at full scale) in surge and inclined about 4° in pitch and then the wind turbine system oscillates around the new position by wave excitations.

Fig. 8 presents the RAOs of two models in regular waves, mean wind speed and rotating rotor. Responses of the inverted conical cylinder moored by a short STL at 50m deep are smaller than those of the OC3-Hywind moored by a long STL at 320m deep, except the peak in surge. Also, all modes of the inverted conical cylinder show peak values around 0.25rad/s. The new proposed inverted conical cylinder with appropriate appendages moored by a short STL shows decent performances in operation conditions, compared with the OC3-Hywind moored by a long STL at 320m deep.

Fig. 9 shows comparison results between RAOs in LC1 and LC2. From that comparison, RAOs in LC1 and LC2 are similar in most of frequencies. But, RAOs in LC2 are higher than those in LC1 near natural frequencies. This phenomenon is caused by that damping near natural frequency are not adequately applied in model test. To verify this phenomenon, another model test with higher scale ratio should be performed.

The behavior of offshore structures in irregular waves may be described in terms of significant amplitude of motion responses at specified sea states. In order to predict the significant amplitude of motion responses of the floating offshore wind Turbine, model tests were carried out in irregular waves (sea states 5~8). As shown in Fig. 10, from the captured motion response, the motion spectra were obtained and their significant heights were calculated.

There is a small static drift motion in irregular waves because of the second-order drift force. The maximum static drift motion is shown in sea state 8 as shown in Table 4 and Fig. 11.

LC3

Fig. 12 shows the significant motion height of the inverted conical cylinder moored by a short STL in 50m deep and the OC3-Hywind moored by a long STL at 320m deep in only irregular waves. In sea states 5 and 6, the significant heights of both models show a small difference in all modes. In sea state 7, the significant heights of the inverted conical cylinder moored by a STL in 50m deep are smaller in both surge and pitch, and a little larger in heave, while, in sea state 8, larger in pitch and smaller in both surge and heave than those of the OC3-Hywind. The inverted conical cylinder was moored by a short STL due to the shallow water depth. In larger excursions, the mooring system of conical cylinder cannot extend because the spring is locked by the case shown in Fig. 2. Therefore the inverted conical cylinder moored by a short STL has a small surge response and a large pitch response, compared with the OC3-Hywind moored by a long STL. And Responses in LC3 looks too big compared with responses in LC1. However, these big responses are only shown in very severe wave conditions, especially sea state 8 (H_s = 15.24m, T_p = 17s).

LC4

Fig. 13 shows the significant motion height of the inverted conical cylinder moored by a short STL in 50m deep and the OC3-Hywind moored by a long STL at 320m deep in irregular waves with a rotating rotor under uniform rated wind. Surge and pitch response of an inverted conical type FOWT are better, but not in heave response. The higher surge response of the OC3-Hywind is caused by a long STL. In sea state 5~7, an inverted conical type FOWT’s heave response are a little higher than OC3-Hywind.

A novel concept of FOWTs, the inverted conical cylinder moored by a STL at 50m deep, was proposed for shallow water applications. In order to estimate motion characteristics, scale model tests in regular, irregular waves and wind were performed in the Ocean Engineering Wide Tank of the UOU.

Having the small responses and the small footprint, the inverted conical cylinder platform was designed with the ballast weight bottom plate for both low CG and large yaw inertia, the inverted conical cylinder for high CB, the STL with a spring case for both small dynamic tensions and the shift in natural frequency, and four triangle plates for small surge and pitch responses.

An inverted conical cylinder type FOWT shows decent motions in most of load cases compared with the OC3-Hywind spar model moored by a STL at 320m deep. Natural frequencies of an inverted conical cylinder type FOWT are around 0.25rad/s in surge and pitch and around 0.19rad/s in heave. In LC1 (regular waves only), the RAOs of this model in natural frequency are 4.0m/m in surge, 0.7m/m in heave and 1.8deg/m in pitch. In LC2 (regular waves with wind and a rotating rotor), the effective RAOs of this model in natural frequency are 6.2m/m in surge, 0.9m/m in heave and 2.8deg/m in pitch. When compared LC1 with LC2, the surge and pitch responses in LC2 are bigger than the ones of LC1 non-universally. The reason why this non universal phenomenon is shown is that the inverted conical cylinder type FOWT has insufficient restoring moment in roll and pitch direction to countervail thrust force by wind and a rotating rotor. And from the results of both LC3 and LC4, the responses of an inverted conical type FOWT are small except for very severe wave condition, sea state 8.

The inverted conical cylinder drifts in a new equilibrium position due to wind and/or a second order wave effect and oscillates around the new position. The thrust in basin model test was scaled in Froude number and may not be comparable to the one in full scale. In near future works, Reynolds number effects on the Froude scale FOWT should be clarified.

In both real sites and model basins, a torque-balanced laying construction for STL is needed to secure large yaw restoring moments as well as yaw controlling devices.