Passive blast protection of buildings by ductile steel-based envelopes

. The growth of industrialization and the rapid expansion of densely populated urban areas have increased the risk caused by technological and man-made hazards. In particular, the accidental or intentional detonation of high explosives (e.g., improvised explosive devices IEDs), can cause damages to buildings and infrastructures and severe harm people. These risks can be minimized by a better planning, design, and construction, e.g., avoiding brittle construction materials, especially within the building envelope, and adopting more robust construction techniques. The study presented in the paper investigates the capacity of light steel-based wall panels to resist the effects of external explosions. The experimental results showed the ultimate capacity of the wall panels is strongly dependant on the initial design conditions and panel-to-structure fastening solution. A numerical model has been also calibrated using Etabs program.


Introduction
The explosions produced in industrial/chemical facilities, such as hazardous material storage facilities, are events with low probability but high destructive potential.When occur in a densely populated area, such explosions can lead to extensive damage of nearby buildings or infrastructures, resulting in high economic costs and human injuries and fatalities.The explosions may be the result of an accident but may be also intentional.A tragic example from the first group is the accidental detonation of ammonium nitrate in a warehouse located in the port of Beirut on August 4, 2020, which led to a huge explosion, causing the death of more than 200 persons, and injuries to more than 6500 others.Referring to the latter category, despite the worldwide fall in the impact of terrorism, it remains a significant and serious problem in many countries [1].
As the pressure released by the external explosions decreases exponentially with the distance to the target, increasing the stand-off distance is one of the most effective ways to mitigate the effects against structures.In addition, the local resistance of the facade elements must be increased to prevent or limit the spread of damage.Even though these issues are well known and understood, there are relatively few provisions in the current European practice [2], [3].Additional provisions may be found in [4], [5], but their use requires extensive knowledge.To meet this demand, design guidelines and examples of application for steel and composite structures subjected to explosions were recently published within the FAILNOMORE project, entitled "Mitigation of the risk of progressive collapse in steel and composite building frames" [6].
The destructive effect of the high velocity debris projected by the explosion should also require attention, as they can cause severe harm to people [5].Therefore, the reduction of risk can be done by using ductile systems and low fragmentation materials [7], [8], [9].The study presented in the paper investigates the capacity of light steel-based building facades to resist the effects of near field blasts.The wall panels were mounted on two-story fullscale building [10], and assembled in two configurations, i.e., standard wall layers (inner metal liner, thermal insulation, outer weather cladding) and inner metal liner only [11].The results presented in the paper are limited to the second configuration.Numerical models were developed using Etabs software and calibrated against test data.To reduce the computational complexity, the parametric study was done considering an individual wall panel.

Full-scale blast test 2.1 Description of test set-up
The wall system is made from MBS KS100/600 liner trays with 0.88 mm thickness, and S320 GD steel (see [12] for technical delivery conditions).Liner trays are placed horizontally and fixed to each side using six 5.5 mm S-MD55Z self-drilling screws, see Fig. 1a.The technical conditions for fasteners are given in [13].The panels were designed for a characteristic wind pressure qb = 0.5 kN/m 2 .
To simulate the real site conditions and interaction with support structure, the liner trays were mounted on a steel frame structure with two spans of 4.5 m each, two bays of 3.0 m each, and two stories of 2.5 m each (Fig. 1b).The columns are made of HEB 260 profiles, and the transversal beams are made of IPE 270 profiles.The secondary beams are made of IPE 200 profiles (secondary beam between columns) and IPE 180 profiles (current secondary beam).The base connections of the columns are rigid.More details can be found in [10].

Instrumentation and loading scenario
The instrumentation included pressure sensors (PS), acceleration sensors (AS) and highspeed cameras (HCam), see Fig. 2. A LiDAR system was also used to capture the initial and deformed shape of the façade.Pressure sensors 1 and 2 (PS1, PS2) were mounted at 1.5 m height from the ground (centre of the third panel from the ground), and 10 cm from the wall face.PS1 was oriented with the face to the wall, while PS2 was placed with the back to the wall.These two sensors were used to measure the pressures experienced by the panel.
PS3 was mounted at 10 cm from the left end of the wall face, and 1.5 m height from the ground.It was oriented with the back to the wall face and used to measure the pressure developed at the panel end.PS4 was positioned at the back of the charge, at same distance from the charge as sensors PS1 and PS2.This sensor was used to measure the incident pressure in the field.Acceleration sensors (AS) were attached to the centre of the third panel, but on the backside, opposite to the pressure sensors PS1 and PS2.High-speed cameras were positioned at safe distance from the charge (at approximately 30 m distance), one on lateral side of the structure (HCam1) and one at an angle of 15 from the side of the structure (HCam2).The blast loading scenario used cylindrical charges positioned at 1.5 m distance from the midspan of the left-hand side wall, and 1.5 m height from the ground (centre of the third panel from the ground), see Fig. 2. The charge weights were gradually increased until complete failure of the wall panels using the following loading sequence: -T1: two cartridges, total weight of 0.286 kg eTNT; -T2: four cartridges, a total weight of 0.572kg eTNT; -T3: six cartridges, total weight of 0.858 kg eTNT; -T4: eight cartridges, total weight of 1.144 kg eTNT.where eTNT is the TNT equivalence.
Note that eTNT of the explosive, which is defined as the ratio of the explosive's weight to the weight of TNT's charge weight which will produce the same amplitude of a blast parameter at the same radial distance from the charge, was equal to unity.
To classify the explosion, i.e., near-field or far-field, the scaled distance, Z, was calculated using the Hopkinson-Cranz scaling law ( [14], [15]), see equation (1): where R is the stand-off distance in m, and W is the charge weight in kg of eTNT.As seen in Table 1, the scaled distance Z varies from 2.28 m/kg 1/3 for T1 to 1.43 m/kg 1/3 for T4.The larger charges are therefore close to the near field range, i.e., when the scaled distance is smaller than 1 m/kg 1/3 ([16], [17], [18]).In the near-field explosions, the afterburning and detonation products affect the shock wave [16].The sensor measuring the incident pressure in test T4 was damaged during testing, and only partial data were recorded.As a result, a second test was performed for same charge weight (i.e., 1.144 kg) but against a concrete wall, test denoted as R4.The test was conducted with identical equipment and blast parameters (charge height above the ground and distance from the wall, position of pressure sensors), see Table 1.

Experimental results
Fig. 3a shows images with the failure of the wall panels after test T4, with large deformations and distortions and failure of end connection in the 3 rd panel.The complex deformed shape of the wall was also captured using LiDAR system, see Fig. 3b.
The instantaneous displacements of the panels were measured using accelerometers.Thus, using integral calculus, the velocity function was first calculated from the acceleration function, then the displacement function was obtained from the velocity (Fig. 4a).Also, due to the particularities of the calculus, only the peak values are of interest.The maximum residual deflections were manually measured after each test using a simple steel tape ruler (Fig. 4b).

Blast pressure
The maximum blast pressure experienced by the most loaded panel, i.e., 3 rd panel from the ground, was measured with a group of two sensors, P1 and P2, while sensor P4 measured the corresponding incident pressure (see Fig. 2 for position of sensors).The recorded data were also compared with data obtained using formula available in the literature.Thus, the incident overpressure, Pso, was calculated analytically using relationship proposed in [19], see equation ( 2 where Z is the scaled distance, in m/kg 1/3 , and Pso is the overpressure, in bar.The corresponding reflected overpressures, Pr, were calculated using reflected pressure coefficients cr defined using eq.( 3), for a surface perpendicular to the shock front, i.e., angle of incidence  = 0 o ([5], [20]): The results presented in Table 2 compare the analytical and experimental pressures for charges T1 to T4, and additional R4.It may be seen that the maximum pressure recorded in each test is very close to the analytical one.
Fig. 5 presents the reflected pressure history for test T4 in comparison with the one obtained on the concrete wall, R4.As may be seen, the peak pressures are very similar, even the histories in time are different, most probably due to shape irregularities in the profiled liner trays.Note that the experimental incident pressure in T4 was not available, but only the reflected pressure.Therefore, while for the incident pressure the reference experimental value is 3.19 bar (recorded in R4) for the experimental reflected pressure, the reference value is the average of the two pressures measured in T4 and R4, i.e., 11.88 bar.The images taken during T4 with the high speed camera (see Fig. 6a) show the expansion of detonation products from the centre of detonation and also the large transient distortions of the panels, with complete separations of the overlapped webs of adjacent panels.Fig. 6b shows a snapshot with the shock waves, with the clear indication of the three wavefronts, namely the incident wave (1), the reflected wave from the ground (2), and the reflected wave from the wall (3).The contour pressures were obtained using the BOS (Background Oriented Schlieren) technique.

Numerical modelling and simulations 4.1 Calibration of the numerical model
The numerical modelling was done using Etabs software [21], which is widely used for the analysis and design of multi-story buildings but has also advanced features for nonlinear time-history analysis.The numerical model includes the wall panels and the supporting steel frame structure, see Fig. 7a.The consideration of the main steel frame structure allows for more realistic modelling of panel boundary conditions, mainly the in-plane stiffness of the perimeter steel frame, which directly affects the development of catenary action in panels [11].The modelling of steel frame structure (beams, columns, and braces) was done using frame objects and material properties from [10].The wall panels were arranged as single span elements and modelled using layered nonlinear shell elements, see Fig. 7b.The characteristics of the steel material from the panels were taken from the tensile tests on materials [11].The end fasteners that connect the panels to the supporting columns were modelled using nonlinear link elements.The response of link elements was modelled using multilinear plastic force-displacement curves calibrated against test data [11], see Fig. 7c.The overlapped webs of adjacent panels were also connected using nonlinear link elements.The ultimate resistance of the connectors was corrected by using a Dynamic Increase Factor (DIF) of 1.1 to consider the effect of the high deformation rate [6].It should be noted that although the explosion-induced deformation rates are particularly high, the ultimate resistance is less influenced by the deformation rate than the yield strength [5].The observations made during the experimental program indicated that, starting with test T3, the overlapped webs of adjacent panels begin to separate due to the pull-over failure of the seeming fasteners (see Fig. 8).As may be seen from the snapshots taken with high-speed camera HCam2, for the largest blast pressure developed in T4, the webs of the 3 rd panel completely separate from the adjacent panels, see Fig. 6a.Based on this observation, the numerical model was simplified by considering only the 3 rd panel (Fig. 9), which was the most exposed to blast wave.This also reduces the computational complexity of the blast analysis.Note that panel end connections, mesh size and material properties were not changed.The blast pressure was modelled using test data obtained in the experimental program, see Table 2.The pressure-time history considered only the positive phase and neglected the negative phase (Fig. 5), which is usually less important than the positive phase.The pressures applied along the panel span varied according to the angle of incidence, .Considering the very short duration of the blast, a time step of 0.0001 sec was used in the analysis.The calibration of the numerical model was done using the results obtained in test T3 and in test T4 after T3.T1 and T2 were disregarded, as the panel response is quasielastic (T1) or has small residual deformations (T2).
As seen in Fig. 10, for test T3, the maximum residual out of plane deflection of the panel obtained in the numerical analysis is very close to the experimental value (see Fig. 4b, right).For the subsequent test T4, the numerical results also indicate the failure of the panel, with excessive out of plane deflections and analysis interrupted due to the failure of the end links.Fig. 11 shows the deformed shape of the panels, experimental and numerical, after tests T3 and T4.Some differences may be seen in the deflected shapes, but this has little influence on the ultimate capacity of the panel, which is mainly attributed to the catenary response of the wide flange, and little contribution from the two webs.

Conclusions
The accidental or intentional detonation of high explosives (e.g., improvised explosive devices IEDs) are extreme loading events that can cause damages to buildings and infrastructures and severe harm to people.The growth of industrialization and the rapid expansion of densely populated urban areas has increased these risks.
When explosions occur within a short distance of the building's facades, there risk of significant damage or complete failure of the walls increase, thus threatening the life of the occupants and the stability of the structure.Increasing the stand-off distance is an effective measure of reducing the level of damage.If the stand-off distance is small, the classical design of the light steel-based wall panels to resist out of plane pressures becomes uneconomical, as it is based only on their flexural resistance.Increasing the capacity of the end connection can lead to a significant increase in ultimate load capacity, by the development of catenary forces in the panels.The increased deformation capacity allows the explosion-induced energy to dissipate through plastic deformation and ultimately reduce the risks associated with such accidental actions.The numerical simulations showed good agreement with the test data, even using single panels instead of multiple panel assemblies.Further studies are in progress, including complex numerical modelling, different position of the blast charge and pressure intensity.

Fig. 1 .
Fig. 1.View of the experimental setup with detailed view of liner trays (a) and assembly view (b).

Fig. 2 .
Fig. 2. Plan view of the blast test set-up, with pressure sensors, acceleration sensors and high-speed cameras (not to scale).

Fig. 3 .Fig. 4 .
Fig. 3. Failure of the panels after test T4: a) overall view with large deformations and distortions in the lower panels (left) and failure of end connection in the 3 rd panel (right); b) deformed shape captured with the LIDAR system.

Fig. 7 .
Fig. 7. Full numerical model (a), detailed view with panels (b), and force-displacement curves characterising the link elements (c)

Table 1 .
Description of the blast charges.

Table 2 .
Experimental vs analytical incident and reflected overpressures.