PSI - Issue 64

Flavio Stochino et al. / Procedia Structural Integrity 64 (2024) 1782–1789 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

1786

5

latter one. The chain has been modelled with 40 beam elements, while the concrete central hinge and the ends are represented by 144 plates elements. The chords and diagonals are modelled with 101 beams elements and have 1510 degrees of freedom. The constraint condition imposed by the column on the arch is modeled using two trolleys equipped with two springs that have translational stiffness along the horizontal axis. The stiffness of the springs is calculated considering the elastic theory of slender beam: = 3 3 (1) where E is the modulus of elasticity, J is the moment of inertia, and L is the column length. This formula ensures that the mechanical interactions between the arch and its supports are accurately represented, allowing for a detailed analysis of the arch's response under thermal stress. The thermal load has been applied using the temperature-time curves obtained through the CFD analysis. The chain was divided into 40 nodes, with different temperature-time tables assigned every 8 nodes. For the lower part of the arch, which is approximately 21 meters long, the three different temperature-time tables were placed at nodes spaced every 7 meters. The calculation step used for this analysis does not affect the results; hence, the same step frequency as that used in the nonlinear structural analysis was employed, which is one step per second. This approach facilitates a transient nonlinear structural analysis that accounts for temporal variations in temperature and the nonlinear material properties of the model, thereby allowing for an assessment of the arch's mechanical behavior under varying thermal conditions UNI EN 1992-1-2.

Fig. 4. Arch thermomechanical FE model.

5. Results The simulation sequence is depicted through a series of images analyzed to understand the fire's behavior during the simulated period. At 60 seconds (Figure 4), the fire was confined only to the initial ignition zone. Ten minutes into the simulation the fire appeared to remain in the ignition area, but it's important to note that the flames indicate regions with a heat release rate exceeding 40 kW/m³, suggesting that combustion reactions might be occurring in other areas, albeit to a lesser extent. By 1500 sec the flames had spread throughout the warehouse with significant heat release, and temperatures were high enough to trigger the removal of openings; windows and doors changed color from blue to white, indicating activation, except for windows on the north wall's left side, which did not reach the activation temperature and thus remained closed for the duration of the simulation. The absence of coverage in the simulation meant that these openings did not affect air recirculation; three windows stayed closed likely due to the absence of combustible materials within a 3.5 m radius. Further into the simulation, at 3000 seconds, a notable reduction in internal materials is observed as temperatures begin to decline. By 5000 seconds, the fire is nearing self-extinction, indicating the simulation's progression towards an end without external intervention.