The unfortunate ammonium nitrate explosion from August 2020 in Beirut, Lebanon, reminds us all of the importance of curtain wall façades’ resistance to extreme loadings. A mid-rise building situated approximately 1km from the explosion was severely damaged (Figure 1).
The façade system of this 7 story high building, about 10 years old, integrated a combination of skylights, unitized and stick frame systems using various DOWSIL™ Structural Glazing Silicones. The façade glass panels were limited in size (about 2.5m width and 0.8m height) and structurally glazed on 4 sides using a 12mm by 6mm joint of DOWSIL™ 993 Structural Glazing Sealant. The glass panes were insulating glass units (IGU) of 8mm toughened and a 55.1mm laminated glass on the inside sealed using DOWSIL™ 3362 Insulating Glass Sealant. On-site inspection after the event showed that no silicone joints were compromised cohesively. The broken glass was still retained around its perimeter by the silicone. The hardness of the material was still within expected range (Figure 2).
A number of design parameters (the use of toughened glass and a limited lamination thickness for the IGU, leading to quick breakage of the glass, small pane dimensions and pane aspect ratio), were probably favorable and limited the loading on the structural joint, but still the performance of this relatively small joint, not designed to resist such types of loading, might surprise.
Figure 3 shows a silicone bonded glassfin that was under installation on a new build project as the Beirut explosion occurred. The glass was broken yet the images illustrate how the PVB foil still retains the broken glass. The wavy shape of the fin demonstrates the extreme plastic deformation and the membrane behavior of the PVB. The silicone bonding at top and bottom kept the deformed glassfin in place. The U-shape and favorable load orientation (occurring perpendicular to the plane of the glassfin rather than in plane) helped reducing the amplitude of the loading experienced by the joint. Once more, this joint demonstrated an excellent performance despite not being designed specifically dimensioned for blast resistance.
Although both cases were not specifically designed to be blast resistant and some parameters were favorable to the joint resistance, the performance of silicone bonded façades (combined with laminated glass) in blast events can ensure improved performance compared to other design options, thanks to the increased retention capacity of the glass membrane post rupture. When the structural joint connection is properly designed to account for parameters such as glass thickness or laminate type, structurally glazed units contribute to limit damage during a blast event. A structural silicone connection has been shown to reduce the ejection of glass panes and maintain their attachment to the frame after glass breakage even in extreme blast loading cases.
What happens during a blast event?
Historically, the first blast-resistant façades would use channel laminated glazed systems with 30-40mm of glass edge covered by sealant gunned in between the metal channel and the glass. This detailing was, however, perceived as expensive and inaesthetic. In 1997, one of the first field tests of a structurally glazed element of 3m x 3m was performed (Figure 4, Hautekeer 2001). The total glazed area was divided in 4 panels which were bonded on the 4 sides using a DOWSIL™ 993 Structural Glazing Sealant joint of only 26mm x 6.4mm. The glass panes were insulating glass units, with 12mm glass laminated with 5 plies of PVB. The charge comprised of a moderate 12kg TNT weight placed quite close at 6.5m from the glass, 800mm above ground level which resulted in the pressure wave hitting the glass after 6msec at a speed close to 500m/s. After testing, it was found that the laminated glass was cracked and deformed in line with the frame system. When breaking, the laminated glass tends to lose its flexural rigidity and pane deformation can be such that the plate escapes from its mechanical fixations. The silicone joint however retained the glass in position and did not show any sign of damage or adhesion loss. The tested silicone joints were much smaller compared to the conventional channel designs used up until this point and this opened the path to the use of structurally glazed details that can withstand bomb blast pressure.
Field testing is expensive and time-consuming due to the high number of tests needed to understand the influence of the different parameters. Therefore, it is important to understand the blast event as well as the silicone performance to be able to accurately model and predict behavior through advanced calculations and simulations.
A complex loading system
The first few milliseconds of a blast event consist of a positive pressure hitting the building (Figure 5, Dewey 2010). The positive phase is determined by its impulse, which is the total load exerted by the explosion whereby not only the overpressure, but the distance from the explosion also plays a role. The pressure of the explosion quickly decreases (as a function of the cube of the distance). During this phase, the glass deflects towards the interior of the building. This is considered to be the most critical phase as glass should not break and enter the building and, therefore, the positive phase is typically the main concern in the design phase. As the load gets reflected by the building and the surrounding construction, its impulse becomes negative.
The glass itself will at first continue to deflect inwards, due to inertia, then it will start deflecting outwards following the negative phase with some delay. This phase shift between plate bending and suction force reduces the glass pane deflection, as the external force is now opposite to the displacement. The glass bending towards the outside of the building is now in phase with the suction force, leading to enlarged deformation at the second flexion peak. This mechanism explains why in most bomb blast testing, the glass detached from the frame is found in front of the façade as glass pane deformation and stress in the joint are the largest during the second deflection peak. The negative phase should not be underestimated, as it has a long duration and will also impact a structure which has been fragilized by the positive wave. Because there is little energy loss associated to glass deformation, most of the energy associated to the glass deformation is recovered and the interaction between the inward and outward movement leads eventually to complex higher modes of deformation and an oscillation of the glass pane similar to a spring system.
The silicone sealant, essential for retaining the glazing, is first compressed by the glass bearing against the frame or mullions, and then subjected to a complex triaxial state of superimposed bending, shear, and tensile stresses as the glazing is crazed and deformed. Once the pressure wave cracks the glass plies of the laminated glass system, the laminate behaves as a membrane with the interlayer ensuring the integrity through plastic deformation. The pane’s flexibility and deflection increase greatly, which will, to the contrary of windloading, mainly load the joint in shear. Besides shear, additional forces impact the silicone joint, such as in and out of plane forces and moments, leading altogether to a complex loading behavior. The deformation of the silicone after the glass breakage is therefore the most demanding part of the loading and the focus of design.
The speed of deformation or strain rate of the silicone is another important element influencing the silicone’s response. The realistic load speed for the structural edge sealant of a façade element under bomb blast can be obtained by analysis of test results. For instance, (Kranzer 2005) reports pressure and displacement versus time histories measured on laminated glass panes (two panes of 3 mm thick float glass laminated with 1.52 mm thick PVB film) exposed to blasts generated by high explosive field charges or pressurized air releases in shock tubes. The glass area loaded by the blasts was 1.0 x 0.8 m² and the blast impulses were designed to take the laminated glass to the point when the glass pane just crazes (referred to as the Break Safely / No Hazard level). Under these conditions, center pane displacements of around 15 mm and maximum center pane velocities of 4.9 to 7.5 m/s were measured using a non-contact, laser-optical displacement measurement technique. Using these experimental blast testing data, typical deflection angles of 4° to 40° and typical line loads in the range of 60- 160 N/mm were estimated for a laminated glass of 1.0 x 0.8 m² size. Using a range of 5 to 20 milliseconds for the positive pressure phase and the typical displacement ratios given above (h/b ratios of 0.03 to 0.2), maximum center pane velocities for successful blast tests, i.e., without the pane dislodging, are estimated to fall in the 4 to 30 m/s range. Considering the post-breakage ductility of the laminated glass at high strain rates, an estimate for maximum movement rates of the sealant ranged between 1 and 15 m/s. Similarly (Mueller 2006) calculates a maximum glass elongation of 160mm after 17ms for an element span of b=800mm. Converting this to glass rotation α, an approximate maximum load speed of the sealant at the glass edge of 0.5 to 0.8m/s is obtained, assuming a 20mm sealant bite. The strain ratio of the sealant itself is, therefore, lower than the speed of the glass during the explosion and typically depends on the geometric behavior of the system. However, it is significantly higher than the strain rate observed for normal or extreme wind events such as typhoons and hurricanes which makes specific silicone material characterization at high speed necessary.
Silicone behavior at high strain rate
Since the deformation process occurs within a few milliseconds, giving rise to stresses in the silicone sealant close to its performance limits, it is of importance when selecting products in the design of blast-mitigating glazing systems to understand the response of silicone sealants to high strain rates and high stress loads.
As a hyperelastic material, the typical strength and elongation of a silicone sealant depend on load speed. The performance of conventional structural silicone materials under normal (wind) loading is typically evaluated at quasi-static strain rates of 5mm/min (~8E-5 m/s) as specified in ETAG002 (EOTA 2012) and ASTM C1135-19 (ASTM 2019). As seen above, typical designs indicate that speeds of 2.5 m/s in tensile and 1.1 m/s in shear should be applicable for most of the bomb blast resistant structural glazing facades. However, it is recommended that a check be made for the relevance of these values for each specific case by the designer of the façade. Otherwise, further investigation could be necessary.
Extensive blast testing at high speed (Yarosh 2010) was carried out on silicones used for structural bonding. The aim of the tests was to investigate the material properties at the increased load velocities occurring in shock waves. High speed tests were conducted at displacement rates of 1.1m/s and 2.5m/s and shear tests at 1.1m/s. These speeds consider the relative displacement between the glass and the frame, in both directions. This means a load application velocity 240 to 300 times higher than the maximum standardized tensile and shear test speeds and 30 000 higher than the standard tensile speed (5 mm/min). A variety of specimen sizes were evaluated. The length of the H-bar joints was 18mm and the depth and thickness were in the range of 6-16mm. The tests were conducted on single sided joints.
The silicone sealants studied all toughened and appeared to stiffen with increasing movement rates. The toughening of the sealant results from a simultaneous increase in maximum strength and strain, which translates into substantially increased fracture energy, which corresponds to the area under the stress-strain curve. For a well-balanced, blast mitigating window design, the increased toughness results in greater blast capacity, assuming the silicone sealant represents the weakest link in the performance chain of the design. In tension, both tensile strengths and corresponding strains increase by a factor of about 2 to 2.5 for all three sealants with an increase in movement rate from 50 mm/min to 5.0 m/s. In the shear experiment, a corresponding increase in shear strength by a factor of 3.5 and strain by a factor of 2.1 is observed. These increased tensile strength and elongation indicate a strain dependent relationship of physical properties. The flexibility of the silicone under these conditions will allow the displacements between glass and frame without breaking cohesively. In addition, the product will be dimensioned based on an increased breaking strength. The values observed for DOWSIL™ 993 Structural Glazing Sealant are reported in Table 1.
Another parameter of importance in blast is the tear resistance. This test evaluates the tensile strength on H bar joints of 12mmx12mmx50mm with a preliminary applied incision of 5mm. The test is normally performed at the standard speed of 5mm/min (EOTA 2012). At this velocity, peak tear strength is around 127N ± 13N for DOWSIL™ 993 Structural Glazing Sealant. When performing this test at a tear velocity of 1m/s, the tensile tear strength increases to 306± 2N. A similar increase in tear elongation can be noted. With ~2.4X increase in tear force and ~2.1X increase in tear elongation, the total tear energy at 1m/s is ~ 5X of the tear energy at 5mm/min (Figure 6). The same increase of performance is observed for shear mode (Figure 7).
These observations can be explained through the unique composition of the silicone sealant. The viscoelastic nature of amorphous rubbers (such as silicones above glass transition and crystallization temperatures), derives from the mobility of the polymer chain on the atomic scale (rotations between molecular units) and on the macroscopic scale (straightening of the chain between cross-links). The strain rate sensitivity reflects the timescale required for these polymer chain re-orientations to take place. At low strain rates the polymer chains have sufficient time to re-orient themselves and the storage modulus of the rubber is low. At high strain rates, the deformation of the polymer chains is restricted to bending and stretching of the chemical bonds, and the storage modulus of the rubber can increase by up to three orders of magnitude. As a silicone bond is loaded, micro cracks will eventually start to form and propagate. The propagation around the fillers in the polymer matrix takes time but if the joint is loaded at high speed, the chains can elongate before the crack has propagated and hence reach higher elongation and strength at failure than when loaded at conventional speeds.
Furthermore, silicone sealants are characterized by an excellent retention of their physical properties along a very broad temperature range. This characteristic is linked to the very low glass transition temperature (Tg) of the central polymer backbone of silicone elastomer, that can reach –120 °C. This low Tg is determined by the inherent flexibility of the polymer molecules and is typical and unique to silicone. Viscoelasticity and, in particular, the Boltzman Superposition and Time/Temperature Equivalence principles for polymeric material predict that such low glass transition temperature would also result, under constant temperature, to a very stable property profile under very short impact load conditions. The theory is predicting the particularly good performance of the silicone sealant when submitted to a very sudden load conditions typical for bomb blast, hurricane or earthquake. The critical point for a structural sealant is to remain elastic and to keep its strength and movement capability. Any glassy behavior of the silicone sealant could lead to cohesive failure of the sealant. With a polymeric glass transition temp of –120 °C versus 20°C for organic elastomer, the silicone is the material of choice for such applications.
Joint dimensioning methods
Some of the deformation conditions that the silicone sealant experiences can be approximately derived. For a 4-sided bonded glass pane, once the glass is broken, the interlayer behaves as a membrane. Assuming the frame is infinitely rigid, the stress will be maximum in the middle of the long edge of the pane and minimum in the corner leading to a deformation as illustrated in Figure 8. Therefore in a simplistic (quasi-static) approach, the maximum force acting on the sealant at the center of the long edge of the glass pane can be estimated by replacing the three-dimensional membrane with a two-dimensional cross section, represented by a flexible “rope” fixed between two points under constant load. In this model, the load is acting perpendicular on the rope at any point along its length. Simple static considerations allow derivation of the line load and its perpendicular components at the fixation points and at the center of the membrane, as follows
With b the span (m), h the deformation of the membrane in the middle (m), p the pressure load (kPa), H0 the horizontal line load on fixation (N/mm), V0 the vertical line load on fixation (N/mm), Qm the tension line load in the middle of the membrane (N/mm), Q0 the resulting line load on the fixation and a the directional angle of resulting load Q0.
Based on calculated line load values and an estimated sealant shear strength at high speed load as described previously, a bite can be easily determined. No safety factors are considered, because bomb blast impact always destroys the element. The calculation model requests an assumption for glass deformation h. This depends on glass thickness, interlayers, glass type (e.g. annealed versus tempered) and potential fracture pattern of glass type. The proposed calculation method is a worst case scenario due to its assumptions (e.g. frame rigidity). In reality, systems can be optimized so that the energy of the blast can be absorbed by plastic deformation of the frame elements or the effect of the interlayer.
Blast events are complex loading phenomena and representative answers for silicone joints can only be obtained by simulating the full system (glass, interlayer, fixation, frame structure). For instance, modeling of laminated glass and its behavior after breakage is not obvious and requires the knowledge of parameters difficult to estimate such as the percentage of delamination of the glass shards adhering to the interlayer. Considering the complexity of the problem and the influence of the various parameters, the above simple analytical calculation methods should be used with caution in the early design phases of a project, to provide an approximate estimate of the required joint dimension to sustain a certain blast load. Finite Element Analysis can be used as an alternative design method. This well-known simulation method allows evaluating strain and stress levels in silicone joints in complex systems or under complex loading conditions. Extrapolating such methods developed for conventional loading speeds of 5mm/min to high-speed complex loadings such as blast events is, however, not straightforward. In a first step, adapted hyperelastic material models based on sealant testing at high-speed rates are needed. Second, the failure mechanism should be carefully studied to ensure which parameter accurately represents experimental validation under blast. Parameters of importance under conventional loading might not be relevant under blast loading anymore. Hyperelastic material models for DOWSIL™ Silicone bonding sealants are available from Dow to be used in your models.
It is a common belief that high strength and modulus are the most desired parameters positively influencing the strength of a bonded connection. Modulus values provided by sealant manufacturers are typically derived from stress-strain graphs through linearization of the curve. These values will be used for approximate analytical calculations as illustrated above. When the real behavior is modelled using FEA and integrating the full hyperelastic material behavior of the sealant in the simulation, it appears that sealants showing a significantly larger linear modulus eventually develop very similar strain levels as sealants with a lower linear modulus. This can be explained by the fact that the bulk moduli are used in FEA and these values are much more similar for different sealants due to the incompressible nature of silicone elastomers. The deflection of glass panes bonded with both types of sealants is very similar confirming the equivalence in the end application of these materials. The strain developed in a lower strength sealant with a higher elongation capacity will also be lower. High moduli sealants typically have a lower sealant deformation capacity as they have been formulated with a higher filler content which can limit elongation at break. As described previously, deformation levels reached by sealants under blast loadings are extreme and a good movement accommodation capacity is key to successful design.
Reducing modulus can actually be beneficial for the overall performance of silicone sealants under blast events, as long as the tensile strength is maintained and hence the elongation at break. This is illustrated in Figure 9 showing two identical joints of 50mmx8mm loaded under the same blast conditions. The Von-Mises stress developed in the high modulus sealant under maximum deformation reaches values that are several orders of magnitude larger than those in the sealant with half modulus.
Besides the influence of the intrinsic material properties, it is possible to optimize the response of the joint by adapting its geometry. It is known that the aspect ratio and hence the thickness of a joint influences its capacity to accommodate deformation (Descamps 2017, 2018). To illustrate this, Figure 9 compares a sealant with 50mm bite and thickness of 15mm (aspect ratio 3) to the previous case (50mm x 8mm, aspect ratio 6), both having the same modulus. The maximum value of Von Mises stress is nearly halved (~5MPa versus 10MPa) for the 15mm thick joint.
It is interesting to note that the joint bite is not the only criteria to consider when dimensioning the joint but that joint thickness also plays a critical role on the capability of the joint to sustain the blast load.
Silicone is the preferred bonding solution when dealing with extreme loading such as those observed in blast events, thanks to its optimal combination of strength and movement capability properties. Appropriate geometrical joint dimensioning guidelines can help to further optimize its performance. However, the complexity of blast loading phenomena requires the use of advanced dimensioning methods to obtain representative results when analyzing the behavior of silicone joints. Reaching results which are representative of reality can only be obtained by simulating the full system (glass, interlayer, fixation, frame structure). This implies a close collaboration between the system supplier, the sealant, interlayer and glass manufacturers, as well as the façade consultants. Project review and analysis should be initiated as early as possible in the project to maximize the benefits of the use of the silicone.
Dow is continuously developing its expertise of blast loading. Please contact your Dow Technical Specialist to further discuss your project requirements.
This article was originally published in IGS Magazine’s Summer 2021 Issue: Read the full Magazine here for more thought-leadership from those spearheading the industry
ASTM C1135-19, Standard Test Method for Determining Tensile Adhesion Properties of Structural Sealants (2019)
Descamps P, Hayez V, Chabih M, Next generation calculation method for structural silicone joint dimensioning , Glass Struct. Eng. DOI 10.1007/s40940-017-0044-7 (2017)
Descamps P, Dimensioning silicone joints used in bomb blast resistant facade systems, in Proceedings of Challenging Glass 6 (2018)
Dewey J. M. ‘The shape of the blast wave: studies of the Friedlander equation’, 21st International Symposium on Military Aspects of Blast and Shock, Israel (2010)
EOTA, ETAG002 Guideline for European technical approval for structural sealant glazing kits (2012)
Hautekeer JP, Monga F, Giesecke A, Brien BO: The use of silicone sealants in protective glazing applications. Glass processing days conference proceedings pp 298-302 (2001)
Kranzer, C., Gürke, G., and Mayrhofer, C., “Testing of Bomb Resistant Glazing Systems Experimental Investigation of the Time Dependent Deflection of Blast Loaded 7.5 mm Laminated Glass,” In Glass Processing Days Proceedings, J. Vitkala (Ed.), Tamglass Oy, Tampere, Finland, pp. 497-503; available online at: http://findarticles.com/ . (2005)
Mueller R, Wagner M, Berechnunng explosionshemmender Fenster- une Fassadenkonstruktionen, IBDRM (2006)
Yarosh K, Wolf A, Sitte S “Evaluation of silicone sealants at high movement rates relevant to bomb mitigating window and curtainwall design.” In Durability of Building and Construction Sealants and Adhesives: 3rd Volume. ASTM International, 2010.
Valérie Hayez is Global Façade Engineering & Architectural Design Engineer for High-Performance Building Solutions at Dow based in Belgium. She provides technical service to the design community, including façade system manufacturers, architects and engineers and communicates industry needs to Dow‘s Research and Development Community. Valérie has developed a broad expertise in façade engineering, including structural performance, fire safety, thermal or acoustic insulation and is active at standardization level. She holds an MSc and a PhD in Applied Sciences (electronics and optics) from the University of Brussels.
Jon Kimberlain currently provides technical expertise and application support for Dow Silicones as a Senior Scientist. With Dow Corning and Dow, Jon has published 25+ research papers on the use of silicone sealants in high performance buildings presented at venues such as Façade Tectonics, ASTM, and GPD Finland. Currently a founding board member of Architectural Glass and Metal Certification Council, he has also been active in GANA and NGA.
Sigurd Sitte is Senior Technical Service & Development Scientist for High Performance Building Solutions at Dow in Germany. He has over 25 years providing expertise to architects, consultants and glass industry professionals on silicones applied in curtain wall and glass applications as well as solar. Sigurd also represents Dow in several research projects with test institutes and universities and coaches global projects with glass and facade applications.