PSI - Issue 64

Margherita Autiero et al. / Procedia Structural Integrity 64 (2024) 1798–1805 Author name / Structural Integrity Procedia 00 (2019) 000–000

1802

5

geometry of the fire model into different cells as the load level of structure, i.e., where the combustion material (UDC) is present. Therefore, the final model of the ARSW was created in CFAST, as shown in Fig 2a, where a first large compartment that represents all structure volume was modelled. Within this one, one transversal cross section was modelled, by modeling the 5 racks shoulder and 1 steel compartment representing the bottom steel frame. For the central three shoulders, the first six load levels which were directly affected by the presence of a fire, were modelled as a single compartment/cell but communicated among them hanks to horizontal openings (ceiling/floor vent). The result was a model with seven compartments for each shoulder: six single compartments related to the first six levels where, with several delays, the fire develops and the seventh compartment grouping the remaining load levels where no fire was defined. The lateral shoulders were not modelled as a compartment because they were considered cold. Figure 2b shows the temperature distributions recorded by the CFAST thermocouples placed at the first seven levels near the horizontal elements i.e., the hot zone. By analyzing temperature results, it is possible to appreciate, how the time delay imposed for the HRR curves was found also in the natural fire curves. All the curves reach a maximum temperature equal to about 1120 °C.

θ [°C]

1200

2nd level 3rd level 4th level 5th level 6th level 7th level

1000

800

1st level 3th level 4th level 5th level 2nd level 6th level 7th level

600

400

200

first ignition

0

0

15

30

45

60

time [min]

(a) (c) Fig. 2. (a) CFAST model (b) temperature distributions recorded by the thermocouples placed at the first 7 levels near the horizontal elements (c) deformed shape at 5.8 minutes in the Diamond environment. 4. Thermo-mechanical modelling A typological frame, in the CA direction, of the ARSW structure shown previously, was modelled in SAFIR software by using beam elements. The structure was analyzed in 3D space by blocking out-of-plane displacements thanks to restraints. At the structural base, fixed restraints were considered while the braces, horizontal beams of the shoulders, and diagonal and vertical elements of the truss were modelled as pinned beams. The exceptional load combination was considered, and dynamic analyses were carried out by considering the nonlinear geometry and material nonlinearity, and for a more correct modelling of Class 4 steel elements through beam elements, a modified constitutive low was considered to consider in beam-based numerical models, local instabilities that can occur in thin sections (Franssen et al. 2014). Since the literature assesses that the most indicated simplified fire model for a member outside the fire area is the localized model (O. Vassart et al., 2016), the thermo-mechanical analysis was carried out by considering different fire models for different structural elements. For all the uprights the LOCAFI model was used while for all the other structural elements, the results of the proposed multi zone model were used. Fig. 2c shows the deformed shapes of the structure at 336 seconds (i.e., 5.6 minutes), which is the last step of the structural analysis provided by SAFIR. At this time the critical element could be the transversal beam of the central shoulder at the 2 nd load level, which is a coupled C150x50x15mm and 2mm thickness, which shows a buckling shape. Fig. 3a shows the temperature distribution within the cross-section, obtained from the thermal analysis, where it is possible to appreciate a little temperature difference among the web, the flanges, and the lips. Fig. 3 shows the temperature distribution along the first level of the upright, obtained with LOCAFI, in comparison, with the one obtained from the thermal analysis carried out with the CFAST natural fire curve. By analyzing the results, it is possible to see that LOCAFI provides different temperatures along the vertical development of the column, unlike by using the zone model where it is possible to obtain only one uniform temperature, on the other hand, it is possible to see that CFAST could be seen as a mean value of the LOCAFI one. To investigate the achievement of the capacity of the beam, the compression (b)