PSI - Issue 27

Aldias Bahatmaka et al. / Procedia Structural Integrity 27 (2020) 6–13 Bahatmaka et al./ Structural Integrity Procedia 00 (2019) 000 – 000

9

4

a time-marching approach and obtain a time series of wind pressures and forces. RANS can simulate time-averaged flow and unsteady flows in which large vortices and generated periodically in the wakes of simulations. In a simple statement, LES predicts wind loads on both claddings and structural frames. Then, RANS should be limited to estimate time-averaged wind forces on simulations. Tominaga et al., (2008) discussed for LES computation and the velocity fluctuation corresponding to the experiment at the inflow boundary. Fig. 3. presents the inflow profile obtained by computations at the driven region in LES.

Fig. 3. Inflow profile by computation in LES (Tominaga et al., 2008).

From the above figure, the computation in the driver section closely estimated the turbulence property of the inflow condition in the experiment and the uniform time-averaged velocities in the lateral direction. On the other hand, prediction by RANS calculation has been conducted (Tamura et al., 2008).

Fig. 4. Mean wind pressure coefficient on low-rise buildings predicted by the RANS model at the center section (Kawamoto et al., 1998). From Fig. 4. shows the time-averaged pressure coefficients on the center section of the low-rise building. The standard k – ε model overestimates the pressure coefficients on the frontal surface becau se the turbulence energy is overproduced at the impinging flow and the approaching flow does not separate from the leading edge of the roof. The approaching flows predicted by the modified k – ε models, the k – − model. 3.2. Evaluation of wind loads on ship analysis Results for the LNG carrier are presented in Fig. 5. The figure presents a comparison between wind force coefficients determined in the wind tunnel and by numerical analyses using three computational grids . The comparison of numerical and experimental results shows reasonable agreement, which is evident mostly for yaw moments. Although the first numerical results (CFDtetra) under predicted the wind tunnel measurements (EXP) by about 50% for longitudinal forces, the consistency has remained good.