PSI - Issue 59

B. Ganendra et al. / Procedia Structural Integrity 59 (2024) 238–245 Ganendra et al. / Structural Integrity Procedia 00 (2019) 000 – 000

242

5

Displacement-control method was applied on two BC to replicate the displacement in the midspan of the beam under static loading with two other BC was set as support points as described in Figure 3. The steel properties are also provided from tensile test conducted prior to the experiment and are shown in Table 3.

BC-2, BC-3 U 1 =0 U 2 = -150

BC-1 U 1 = 0 U 2 = 0 U 3 = 0

BC-4 U 1 =0 U 2 = 0

Fig. 3. Boundary condition setup in ABAQUS.

Table 3. Steel properties for FE modelling. Yield Stress ( ) Tensile Strength ( ) Young's Modulus ( ) Poisson Ratio (-) MPa

Elongation Ratio (-)

1.88 x 10 5

190

315

0.36

0.51

4. Results and Discussion 4.1. Flexural Strength

Non-linear static analysis is used in current study to analyse the flexural strength of the steel cylindrical shells. All four models were meshed using four-node element (S4R) within four mesh sizes, i.e., 20, 10, 5, 2.5. These different mesh sizes are utilized to determine at which mesh size the closest prediction of the ultimate load can be obtained. The comparison of ultimate load of each model obtained from the numerical simulation and experiment are shown in Figure 4. The graphs shown in Figure 4 are presented in terms of ratio between mesh size over the thickness of the shell on the x-axis and load carrying capacity of the numerical simulation ( ) over the load carrying capacity from the experiment ( ) on the y-axis. All four figures showed that the value of the ratio of ( ) ( ) always nearly 1, indicating that the numerical simulation are able to predict the ultimate load of the cylindrical shell models close to the actual result. Deviations are still can be seen on each graph. This phenomenon revealed the fact that the prediction of the ultimate load is sensitive to the mesh size used in the numerical simulation, even though at certain point the effect of mesh size would be less significant. Similar pattern is occurred in all graphs shown in Figure 4 where the deviations are much smaller and the graph line became more stable when the value of ratio between mesh size / thickness reached 5. In addition, the most accurate prediction of the ultimate load of each model are obtained in different mesh sizes, whereas the most accurate ultimate load prediction of model DT75 and DT125 is obtained when the value of ratio between mesh size / thickness is 10 and the most accurate ultimate load prediction of model DT75 and DT125 is obtained when the value of ratio between mesh size / thickness is 5.