ANSYS Mechanical

ANSYS Mechanical is a CAE (Computer Aided Engineering) software that simulates structural responses by the finite element method. The simulation analysis includes structural strength, stress, vibration, etc., and it can be combined with other ANSYS software for coupling analysis. In some cases, the fluid flow will cause the displacement of the solid, and the displaced solid will change the flow field of the fluid. Fluid–structure coupling is divided into one-way (1-way) and two-way (2-way) fluid–solid coupling. One-way fluid–solid coupling is to treat an already calculated result as another boundary condition to be calculated. Bidirectional coupling is divided into fully coupled, implicit, and explicit coupling. Full coupling calculates the simultaneous equations of fluid and solid as a matrix, which is often difficult to be solved. Implicit coupling is generally used by iteration until the convergence target is reached. Explicit coupling requires very a small time step and is very time consuming; therefore, it is

not recommended. Figure 2 shows the FSI (Fluid-Structure interaction) module of ANSYS, and the implicit method was chosen in this study to perform the fluid–structure coupling analysis.

**Figure 2.** The system coupling and the two-way FSI simulation in ANSYS workbench.

## *2.3. Monopile OWT*

This paper refers to the 5 MW offshore wind turbine developed by NREL. The radius of the three blades is 61.5 m, and the hub height is 90 m above the water surface. The water depth is assumed to be 20 m. The top of the tower has a diameter of 3.87 m and a thickness of 0.019 m. The bottom diameter is 6 m, and the thickness is 0.027 m. The total length of the tower is 77.6 m. The diameter and the thickness of the pile foundation are 6 m and 0.060 m respectively, and the length of the pile foundation is 30 m. The material of the supporting structures is steel, and the properties are as follows: the Young's modulus is 210 GPa, the shear modulus is 80.8 GPa, and the density is 7850 kg/m3. In this study, the rotor, wind turbine blade, and nacelle are assembled as a lump mass (RNA). Figure 3 shows the 5 MW OWT and the simplified OWT model for the simulation.

**Figure 3.** (**a**) The 5 MW offshore wind turbine (OWT) developed by NREL; (**b**) a simplified OWT model.

To reduce the dynamic response of the OWT, a TLD is installed at the top of the OWT. Figure 4 shows the simulation chart. To perform the simulation, the environmental loads were determined first. Since the turbulent wind field was considered, the TurbSim was used to generate a turbulent wind field, and then we input the obtained data to FAST. Then, the force data generated by FAST was input into ANSYS Transient Structural. The TLD geometry was determined based on the tuned frequency of TLD. We used ANSYS Fluent to simulate the sloshing flow in TLD, and then we used system coupling to complete the fluid–structure coupling simulation.

**Figure 4.** The flow chart of the simuation performed in this study.

One TLD was assumed first, and the simulated responses were compared to those without TLD, and the motion reduction ratio was calculated to evaluate the motion control effect. Different load conditions were assumed, such as harmonic ground motion, real earthquake excitation, coupled wind and wave loads, etc. Once the motion reduction was insignificant, more TLDs (2-TLD and 3-TLD) were then added to evaluate the possible increasing on motion reduction control. For fluid–solid interaction simulation, the SpaceClaim model was imported into Fluent, and the area of the structure was deactivated except for the TLD part. After the meshing was generated, the fluid–solid coupling interface was defined. The VOF calculation and the multi-phase flow were used in Fluent analysis. The turbulence mode was a standard k − model.

#### **3. Results and Discussion**

#### *3.1. Model Validation*

Before presenting the numerical simulations made in this study, the experimental measurements were performed to validate the simulation results obtained in this study. A simple experimental model (shown in Figure 5) was set up. The supported tower was a 3-meter height PVC pipe with a diameter of 12 cm. The RNA (rotor nacelle assemble) was replaced by a lump weight of 1 kg. The cylindrical TLD was set at the top of the RNA. The density, Young's modulus, and Poisson's ratio of the PVC pipe are, respectively, 1532 kg/m3, 3070 MPa, and 0.4. As illustrated in Figure 6, the OWT model was set on a shaking table that can be moved back and forth with an AC motor. The maximum moving distance (r) of the shaking table is ±30 mm, and the highest revolutions of the motor is 2000 r.p.m. The AC motor can reciprocate according to the programmed displacement path and frequency. The motion trajectory of the instrument is based on the sine wave. With a maximum amplitude of ±5 mm reciprocating movement, a circular platform is attached to the square platform on which the OWT was installed. A Keyence laser displacement meter is set at the proper elevation to measure the time history of displacement of the RNA.

**Figure 5.** The dynamic displacement decay history of free vibration of a scaled model OWT.

The free vibration experiment was firstly made to evaluate the damping coe fficient of the system. Figure 5 shows the decay history of the dynamic displacement, and the damping ratio of the experimental set up can be calculated according to the following equation

$$\ln\left(\frac{\chi\_1}{\chi\_j}\right) = \frac{2\pi j\zeta}{\sqrt{1-\zeta^2}}.\tag{12}$$

According to the decay history, the damping ratio ζ = 0.0119 was determined.

**Figure 6.** The simple experimental set up of OWT with a tuned liquid damper (TLD) on top.

3.1.1. Simulated Natural Frequency and Dynamic Response Validation

To confirm the accuracy of experimental measurements, the fundamental frequency of the scaled OWT model was checked. The base platform was excited by a harmonic force, F(t) = A <sup>ω</sup>xsin( <sup>ω</sup>xt), where A is the amplitude of the excitation and ωx is the excitation frequency. ANSYS was used to calculate the corresponding dynamic displacement response. The ANSYS FEM-Modal set up was based on the experimental model, and the natural frequency of the model was calculated. The calculated natural frequency (2.751 Hz) is very close to the experimental measurement (2.75 Hz). Figure 7 clearly shows the agreemen<sup>t</sup> of the simulation and experimental results, and the peak dimensionless displacement occurred when the exciting frequency ( ωx) is equal to the natural frequency ( ωs) of the model, i.e., <sup>ω</sup>x/<sup>ω</sup>s = 1. Figure 8 shows the comparison of the dynamic displacement obtained by ANSYS simulation and those of the experimental measurements and the agreemen<sup>t</sup> is also very good. The dynamic displacements measured in this section were also used as a reference to evaluate the effects of TLD on the response reduction of the OWT.

**Figure 7.** The comparison of ANSYS simulated results and the experimental measurements; X/A: the dimensionless displacement of OWT; <sup>ω</sup>x/<sup>ω</sup>s = 1: excitation frequency (ωx) is equal to the natural frequency of OWT (ωs).

**Figure 8.** The history of the displacement at the rotor nacelle assembly (RNA) of OWT under excitation wih exciting frequency = natural frequency of the OWT.

#### 3.1.2. Fluent TLD Simulation Validation

In the ANYSY fluid–solid coupling, the accuracy of the hydrodynamic force calculation is important, and Fluent may correctly transmit the force of the fluid acting on the tank wall to the structure. Figure 9 illustrates the comparison of the Fluent-simulated force with the experimental measurements of Krabbenhøft (2011) [20], and the agreemen<sup>t</sup> is very good.

**Figure 9.** The comparison of the forces acting on the tank wall; solid line: ANSYS results; dashed line: experimental measurements (Krabbenhøft, 2011) [20]. Tank length = 0.59 m, water depth = 0.02 m, the exciting amplitude = 0.02 m, and the exciting frequency = 2.36 rad/s.

#### 3.1.3. ANSYS Mechanical Model (Fluid–Structure Interaction) Validation

The comparison made in the previous sections validated the accuracies of simulations made by the ANSYS Fluent and ANSYS structure models. In this subsection, we further validated the accuracy of the fluent and structure coupling model. The TLD was added on the top of the scaled OWT model, and an experiment was performed to investigate the motion reduction of the OWT.

The exciting force is a harmonic motion, the displacement amplitude is set as 0.001 m, and the exciting frequency is set to equal to the natural frequency of the OWT. ANSYS-Transient-Structural and ANSYS-Fluent are connected to System Coupling for the data transmission setting and time control. Figure 10 illustrates the TLD mesh model and the fluid–structure coupling interface. Figure 11 shows the schematic of the projects included in the simulation of the interaction between the TLD and OWT.

**Figure 10.** (**a**) The mesh arrangements of fluid in TLD and (**b**) the fluid–structure coupling interface.

**Figure 11.** The schematic projects included in the TLD + OWT interaction simulation.

Figure 12 further compares the dynamic displacements of RNA of OWT with and without TLD, and the motion of RNA of OWT is remarkably reduced when the TLD is installed at the top of the RNA of OWT. Figure 13 also shows that the results obtained by ANSYS simulation and the experimental measurements agree very well. The amplitude of the OWT with the TLD is much smaller than that without the TLD; therefore, we use different vertical axes to clarify the difference between the experiment and simulation results. The response of the OWT with the TLD is much more complicated than that without TLD, and more difference between experiment and simulation can be expected, but the deviation is still below 10%. The accuracy of the ANSYS system coupling was confirmed and can be used to study the motion reduction effect of the TLD on the OWT.

**Figure 12.** Dynamic displacement of RNA of OWT: (**a**) without TLD; (**b**) with TLD.

#### *3.2. TLD on Motion Reduction of OWT*

As shown in Figure 12, the TLD may have a very good motion reduction effect on OWT when it is under a harmonica ground excitation. In the following sections, we will investigate the damping effect of the TLD on the OWT when it is under real environmental load conditions. The design standard IEC 61400-3 defines the load conditions DLC (Design Load Case) for the structural design of offshore wind turbines, including all operating conditions of OWTs, such as startup, normal operation, shutdown, etc. The structural design of the offshore wind turbine often considers the extreme 50-year regression period, while the fatigue analysis is based on the general sea conditions. The environmental loading used in this study refers to the feasibility study of the Taiwan Power Company's offshore wind power generation second phase plan. Table 2 lists the estimated wind and wave conditions and extreme wind and waves at the Zhangbin Industrial Zone project.


**Table 2.** The wind and wave conditions at the Zhangbin Industrial Zone project. DLC: Design Load Case.

#### 3.2.1. Wind Field Simulation

As mentioned in the previous sections, TurbSim was used to simulate the turbulent wind field of the whole domain, and the output binary file was directly used by AeroDyn in FAST. The input parameters of TurnSim are shown in Table 3.


**Table 3.** Input file description (Turbsim) (DLC 1.2).

For the wind field condition of DLC 6.2, the Uref is changed to 54.16 m/s, and the generated file was submitted to AeroDyn to expand into a global wind field. In the AeroDyn, the direction of the wind can be set. In the case of the annual average wind speed of the wind field conditions of the DLC 1.2, the wind turbine directly faces the windward direction, so the wind direction angle is set to 0◦. In the extreme load DLC 6.2, the side winds of the 10-year average extreme wind speed and the 50-year regression period are blown toward the blades, so the wind direction angle is set to 90◦. Figure 13 shows the wind speed in three directions with an annual average wind speed at 90 m elevation. Since the annual average wind speed is blown toward the wind turbines, the wind speed in the x direction is greater than the other two. Figure 14 shows the corresponding 10-min average extreme wind speed in a 50-year regression period. The wind field is designed to have a crosswind e ffect on the wind turbines, so the wind speed in the y direction is the largest. The fatigue load in DLC 1.2 is in normal operation, and the turbine blades rotate normally, but in the DLC 6.2 limit load condition, the wind turbine is parked because of excessive wind, and the blade is turned parallel to the wind to reduce the force on the blades. Meanwhile, in the FAST setting, the blades are parked, and the blade angle is set to 90◦. Therefore, we may expect that RNA has larger force in the x direction, whereas at extreme wind speeds, RNA has larger force in the y direction, since the wind blows from the y direction.

**Figure 13.** Annual wind speed Vave = 9.2 m/s at 90 m elevation.

**Figure 14.** 50-year, 10 min average extreme wind speed Vref = 54.16 (m/s) at 90 m elevation.

#### 3.2.2. Wave Simulation

In FAST, HydroDyn is used to calculate hydrodynamic loads. The regular wave and irregular wave can be selected. In the wave condition of fatigue load, an irregular wave is selected. The PM wave spectrum was selected for the normal sea state, whereas the JONSWAP spectrum is selected for the ultimate load condition, because the JONSWAP spectrum can show the characteristics of extreme sea conditions. Figure 15 shows the wave histories of normal and extreme sea states.

**Figure 15.** (**a**) Normal wave condition, Hs = 1.4 m, Tp = 6.1 s; (**b**) 50-year extreme wave condition, Hs = 8.24 m, Tp = 12.01 s.

#### 3.2.3. Structural Load

The forces exerted on the OWT can be further calculated by FAST once the wind field and wave fields were established. The OWT includes a rotor, a nacelle, a tower, and a pile. The focus of the study is on the tower, which does not consider the pile part and the deformation of the blade and the rotational speed of the blades. Therefore, the rotor, the blade, and the nacelle are combined here (RNA) as a lump mass point.

#### 3.2.4. The Loads on RNA

Figure 16 shows the forces of the RAN acting on the top of the tower in DLC 1.2 and DLC 6.2 conditions, respectively. As shown in the figure, RNA has larger force in the x direction for the average wind speed condition, whereas at extreme wind speeds, RNA has larger force in the y direction, since the wind blows from the y direction.

**Figure 16.** The forces on RNA: (**a**) normal wind condition; (**b**) extreme wind condition.

#### 3.2.5. The Loads on Wind Tower

It can be seen from Figure 6 that the tower with a total length of 77.6 m in FAST is divided into 10 sections, and the length of each section is 7.76 m. The force of the tower is calculated by AeroDyn; Figure 17 depicts the force acting on each tower section at the average wind speed condition, and the force in the x direction is obviously greater than in the other directions. Figure 18 depicts the corresponding forces under extreme wind speed, and the force in the y direction is the largest among all directions.

**Figure 17.** The force on each section of the tower under normal wind conditions (blue line: x direction; red line: y direction).

**Figure 18.** The force of each section of the tower under extreme wind conditions (blue line: x direction; red line: y direction).

#### 3.2.6. The Loads on Pile of the OWT

The length of the supported pile of the OWT is 30 m and it is fixed on the sea bottom, and 20 m and 10 m of it are below and above the sea surface, respectively. HydroDyn is used to calculate the force of the wave on the pile. Figure 19 shows the corresponding forces applied on the pile.

**Figure 19.** Force applied on the pile of the OWT: (**a**) normal wave condition; (**b**) extreme wave condition.

#### 3.2.7. Tuned Frequency of TLD

As reported in Chen and Hunag (2015) [21] and Chen and Yang (2018) [13], the best motion reduction effect of TLD on the structure may occur when the natural frequency of the TLD is tuned to the frequency of the exciting forces. The effect of waves on the tower top displacement is much less than that of wind, so it is useless to tune the frequency of the TLD to the frequency of waves. In addition, because turbulent wind is a random variable, we did a spectrum analysis of the wind and found that no specific peak exists, so it is not feasible to tune the TLD frequency to the frequency of the wind. However, tuning the frequency of the TLD to the frequency of the structure also has a good damping effect (Chen and Huang, [21]; Frandsen, [22]). Therefore, the natural frequency of the TLD used in this study was tuned to the fundamental frequency of the supported tower of the OWT.

#### 3.2.8. The Displacement of Tower Top

The calculated forces in the previous section were applied to the finite element method (FEM) model of ANSYS. The dynamic response of offshore wind turbines, including the deformations and the stresses of piles and towers, were calculated in this section. Figure 20 shows the displacements of the top of the tower under DLC 1.2 and DLC 6.2 conditions.

**Figure 20.** The displacement of the tower top in (**a**): DLC 1.2 condition; (**b**) DLC 6.2 condition.

#### *3.3. TLD Application on Motion Reduction of OWT*

The monopile offshore wind turbine is subject to wind and wave loads, which may cause dynamic responses of the structure. This study tried to install the cylindrical TLD on the top of the offshore wind turbine, as shown in Figure 21, to suppress those responses. The diameter of the TLD is 3.846 m, which is about the same size of the inner diameter of the tower top. The natural frequency of the liquid in the cylindrical tank can be calculated by ω<sup>2</sup> = λgR tanh<sup>λ</sup>hR . The water depth of TLD can be determined when we tune the natural frequency of TLD to be equal to the natural frequency of the structure; then, the water depth of TLD = 0.35 m was calculated.

**Figure 21.** The conceptual sketch of the TLD on the top of the tower.

The mechanical properties of the OWT are shown in Figure 22. The OWT model is drawn by ANSYS SpaceClaim and TLD was also added, which were all input to the Transient Structural. In the Transient Structural, the parameters of the material such as density, Young's modulus, and damping

coefficient, etc. were set. Then, the FAST-calculated wind and wave loads were input. The fluid–solid coupling interface was set as shown in Figure 23.


**Figure 22.** Transient Structural (the parameters setting).

**Figure 23.** Fluid–solid interface.

Then, the SpaceClaim model was imported into Fluent, and the area of the structure was deactivated except the TLD part. After the meshing was generated, the fluid–solid coupling interface was defined. The VOF calculation was started in Fluent to simulate the multi-phase flow, and the solid boundary was set as the wall. The turbulence mode was a standard *k*-<sup>ε</sup> model. In general multi-phase flow, air and water are often set as incompressible fluids to simplify the model, but in the fluid–structure interaction calculation, the boundary of the fluid changes with the solid and the volume of the fluid changes with time, which is very prone to problems in the calculation of Fluent, so it is necessary to set the air as an ideal compressible gas to avoid problems in calculation.

## 3.3.1. Convergence Test

In order to save calculation time, a hexahedral structural mesh is used here. Three difference mesh sizes (1.4 m, 1.0 m, and 0.7 m) were used to do the convergence tests, and the DLC 6.2 conditions were used as the external force conditions. Table 4 lists the comparison of the root square of displacements of RNA of various mesh sizes used, and all the results were about the same; the mesh size = 1.4 m was selected in the later simulation. Table 5 lists the convergence tests for mesh selection for TLD simulation, and the results of 0.09 m and 0.06 m are about the same and therefore, the mesh size = 0.09 m was selected in the later simulation. The time steps = 0.005 s and 0.0025 s were tested, and the results are about the same, and the time step = 0.005 s was used in the later simulations.


**Table 4.** The convergence test for tower FEM mesh selection.


**Table 5.** The convergence test for TLD mesh selection.

#### 3.3.2. TLD on Motion Reduction of OWTs

As shown in Figure 14, the TLD has a significant motion reduction e ffect when OWT is under a harmonic excitation. While the OWTs are mostly under wind and wave loads, in this section, we investigated the motion reduction e ffects of TLD on OWT when it is under DLC 1.2 and DLC 6.2 load conditions. Figure 24 illustrates the time histories of fore-aft and side-to-side displacement of RNA of OWT with and without TLD. Although the results presented in the Figure 25 do not show obvious damping e ffects of TLD on OWT motion reduction, Figure 26 shows the FFT of the displacement responses and indicates that the TLD may reduce response peak intensity by 44% and 24% in fore-aft and side-to-side displacements, respectively. When the environmental load condition DLC 6.2 was applied, the results shown in Figure 27 clearly demonstrate the e ffects of TLD on the motion reduction of OWTs.

**Figure 24.** Displacement of RNA for an OWT under the DLC 1.2 condition: (**a**) fore-aft movement; (**b**) side-to-side movement.

**Figure 25.** Amplitude spectrum of RNA displacement for an OWT under the DLC 1.2 condition: (**a**) fore-aft movement; (**b**) side-to-side movement.

**Figure 26.** Displacement of RNA for an OWT under the DLC 6.2 condition: (**a**) fore-aft movement; (**b**) side-to-side movement.

#### 3.3.3. Multiple TLDs on Motion Reduction of OWTs

In the previous section, one TLD presents mild motion reduction effects on OWT when it is under DLC 1.2 and DLC 6.2 load conditions. As reported in Chen and Yang's study, the best damping effect of TLD on the structure might occur when the natural frequency of TLD is tuned to the natural frequency of the structure. Since the OWT is a slender structure and the natural frequency is small, therefore, the water depth of the liquid in TLD is also small. The damping effect of a single TLD might be limited. Then, we increase the number of TLDs, which all have same natural frequency as the natural frequency of the structure. Figure 27 shows the motion reduction effect of multiple TLDs on OWTs when it is under harmonic ground excitation, and the 3-TLD has the best motion reduction effect, which is nearly 100% better than that of the 1-TLD.

**Figure 27.** Multiple TLDs on OWT motion reduction and harmonic ground excitation.

As mentioned by Jin et al. (2007) and the results obtained in this study, the TLD did have an excellent motion reduction effect on OWT when it is under harmonic ground excitation. Meanwhile, the environmental loads that OWT may experience include wind, waves, and real earthquakes. Then, we further investigated the motion reduction effects of the multiple TLDs on OWT when it is under real earthquake, wind, and wave loads. Figure 28 depicts the comparison of the damping effects of various TLDs on an OWT under DLC 1.2 and DLC 6.2 load conditions. The motion reduction effects of various TLDs are about the same when the OWT is under DLC 1.2 load conditions. Figure 28 even shows larger peak performance for the 3-TLD, whereas the narrower band can be found for the 3-TLD, and the root mean square of the FFT spectrum of three cases (1-TLD, 2-TLD, and 3-TLD) are about the same in the DLC1.2 condition. The response of OWT under DLC 1.2 is virtually small, and the difference among various TLDs is also insignificant. The force of the DLC 6.2 condition is much larger in the y-direction, and thus, the side-to-side displacements were shown in Figure 28 when the OWT was under the DLC 6.2 condition. The much larger displacement occurred when the OWT was under extreme loading condition (DLC 6.2), and the effect of TLD on motion reduction control became more obvious; more TLDs also enhance the reduction effects. The corresponding comparison of OWT under real earthquakes is shown in Figure 29, and 3-TLD also has the best motion reduction effects among others.

**Figure 28.** *Cont*.

**Figure 28.** (**a**) and (**b**): The fore-aft displacement and corresponding amplitude spectrum of OWT with various TLDs under the DLC 1.2 condition; (**c**) and (**d**): The side-by-side displacement and corresponding amplitude spectrum of OWT with various TLDs under the DLC 6.2 condition.

**Figure 29.** The side-to-side displacement of the OWT with various TLDs: (**a**) and (**b**), El-Centro earthquake; (**c**) and (**d**), Chi-chi earthquake.

#### *3.4. Fatigue Analysis*

Figure 30 shows the location where the maximum stress occurs in the absence of TLD according to ANSYS simulation. This position is the intersection of the tower and the pile. Then, we made the fatigue analysis of the stress at the junction of the tower and the supported pile. Figure 31 shows the time series of the maximum stress at the intersection of the tower and the pile.

**Figure 30.** The location of the maximum stress occurred in the OWT.

**Figure 31.** The comparison of the history of the stress at the interaction of the tower and pile of the OWT with various numbers of TLDs.

For a typical offshore structure, the fatigue load history spans a period of 20 years corresponding to about 10<sup>8</sup> wave load cycles (an average wave load period of 6 s). We used the S-N curve to determine the fatigue life at the interaction of tower and pile of the OWT (Ju et al. [23]). In this study, the S-N curve (structural detail class E) of DNVGL-RP-C203 [18] was used to assess the fatigue damage at the interaction of the tower and pile of the OWT. The S-N curve can be expressed as follows

$$
\log N = \log \overline{a} - m \log \Delta \sigma \tag{13}
$$

where *N* = predicted number of cycles to failure for stress range; *m* = the negative inverse slope of the S-N curve; and log *a* = the intercept of log N-axis by the S-N curve. Table 6 lists the details of the parameters of the S-N curve.


**Table 6.** S-N curves for tower and pile (Structural detail class E).

The rainfall counting was used to obtain the stress counting. The Miner cumulative damage theory was used to estimate the degree of damage to the structure. The Miner's rule states that if there are *k* different stress levels (with linear damage hypothesis) and the average number of cycles to failure at the ith stress, *Si*, is *Ni*, then the damage fraction, *C* can be expressed as

$$\sum\_{i}^{k} \frac{n\_i}{N\_i} = \mathbb{C} \tag{14}$$

where *ni* is the number of cycles accumulated at stress *Si* and *C* is the fraction of life consumed by exposure to the cycles at the different stress levels. Usually, fatigue damage occurs when *C* > 1.

Figure 32 shows the histogram of stress rainflow counting of the uncontrolled OWT and controlled OWT with difference numbers of TLDs. Finally, we apply the Miner cumulative damage theory and S-N curve to infer the fatigue damage value of the OWT. Table 7 lists the fatigue life evaluated based on Miner's rule of uncontrolled OWT and OWT with different numbers of TLDs, and an OWT with 3 TLDs may increase the fatigue life for 20 years. The OWT with a single TLD can also increase the fatigue life for 13 years. The 3-TLD can not only reduce the motion response but also increase the fatigue life (37% more) of the OWT.

**Figure 32.** The histogram of stress rainflow counting of uncontrolled OWT and controlled OWT with difference numbers of TLDs.


**Table 7.** Fatigue life of using various TLDs.
