Architectural design concepts incorporating glass beams, panels, or generally load-carrying elements and stiffeners for buildings, claddings, windows, and partitions are largely considered in modern high-rise constructions. A multitude of aspects, including motivations related to transparency, aesthetics, illumination, and energy conservation, progressively increased the use and interest for such a still rather innovative constructional material. However, compared to other traditional materials for buildings, standard glass is typically characterized by brittle behaviour and limited tensile resistance. The intrinsic properties of glass, moreover, together with typically limited thickness-to-size ratios for glazing elements, or the mutual interaction of glass components with adjacent constructional elements as a part of full assemblies they belong (i.e., fixing systems, sealants, etc.), as well as the combination of mechanical and thermal phenomena, make glass structures highly vulnerable. Special safety design rules are hence required, especially under extreme loading conditions. In this review paper, a state of the art on structural glass systems exposed to fire is presented. Careful consideration is paid for actual design methods and general regulations, as well as for existing research outcomes—both at the material and assembly levels—giving evidence of current challenges, issues, and developments.
Glass is largely used in buildings as a construction material, to replace and/or interact with traditional structural elements composed of steel, aluminum, timber, and concrete. Major applications of glass in buildings are related to a multitude of aspects, including aesthetics, lightening, transparency, and insulation motivations (see, e.g., Figures
(a)-(b) Typical structural glass applications in buildings and (c)-(d) examples of recent fire event scenarios.
Generally, glass is known to behave as a brittle material with relatively high compressive resistance and limited tensile strength, hence shattering into many dangerous shards [
In this regard, several research studies have been dedicated over the last years to the development and/or assessment of specific design regulations and novel design concepts for structural glass systems, including extended experimental and finite element (FE) numerical investigations related to connections, composite assemblies, and hybrid systems [
Special care has been spent also for the analysis and design of glazing systems under extreme loads, such as explosive events [
Especially in the case of fire accidents, special enhanced safety levels should in fact be ensured, to allow evacuation of buildings (Figures
Multiple aspects are however combined in the overall fire performance of a given structural glass system, such as the typical brittle behaviour of material, the high sensitivity of its mechanical properties to temperature, the high sensitivity of fire performance to geometrical features, glass type, and the mutual interactions between all the system components (i.e., the structural glass assembly, including supports and building components).
As a specific issue of glass systems, in addition, their fire performance cannot be analytically derived but requires fire testing estimations. Advanced FE modelling, in this regard, could represent a valid alternative to time- and cost-consuming experiments. However, major issues for the FE analysis of structural glass elements under fire derive from the current lack of standardized guidelines and general rules able to offer reliable results [
In this paper, a review of experimental research on structural glass systems under fire is proposed. Section
Compared to traditional materials in use of constructions, one of the major factors affecting the design and verification of structural glass elements is represented by its intrinsic features. Even without fault of the designer, a given structural glass element can in fact break unexpectedly, during its service life [
The ULS resistance verification is intended to fulfill the structural safety of a structural glass element [
The SLS verification is aimed at the limitation of deflections. The reference limit values for such deformations mainly depend on the specific applications or support conditions. As in the case of specifications for ULS design, different SLS limit deformation values can be found in standards. A further design condition that should be properly verified (see, e.g., the CNR guidelines [
Fire loading represents, for glass systems as well as for constructions in general, an extreme loading configuration. As such, specific provisions should be taken into account to ensure appropriate performance levels.
Given a glazing system exposed to fire, in accordance with existing standards (see, e.g., the EN 13501-2 regulations [ integrity (classification “E”): glass prevents flames, smoke, and hot gases from passing through. The fire remains contained; limiting radiation (“EW”): glass restricts the amount of heat passing through it to the side which is to be protected; thermal insulation (“EI”): the average temperature of glass on the protected side remains below 140°C; hence, the risk of self-combustion of exposed materials (due to either radiation or convection) can be minimized, and buildings can be evacuated safely and calmly.
The above FR reference criteria can only be determined on the basis of fire experiments, and typical FR rating classes are associated to 30, 60, or 120 minutes of performance. Relevant standards in use in Europe are, for example, the EN 1363-1 document [
In addition to EU provisions, the American Underwriters Laboratory standard [
Compared to other extreme loading conditions which may occur over the lifetime of a given structural glazing system, the main issue of FR glass systems arises from glass response to temperature variations. Conventional glass for application in buildings (Section
Most of glass solutions in existing or novel buildings are realized by using soda lime silica (SLS) glass. Special and limited in number applications only, when a certified level of fire resistance and heat resistance is required, are indeed realized with borosilicate (BS) glass, since offering better performance to temperature changes. BS glass applications in buildings are indeed limited, compared to huge SLS glass use for load-carrying elements, windows, etc. Table
Chemical and physical properties of SLS and BS glass types (at room temperature), in accordance with [
Chemical properties | Physical properties | |||||||
---|---|---|---|---|---|---|---|---|
SLS | BS | SLS | BS | |||||
Silica sand | SiO2 | 69–74 | 70–87 | Density |
|
2500 | 2200–2500 | Mechanical |
Lime (calcium oxide) | CaO | 5–14 | — | Young’s modulus (MOE) |
|
70 | 60–70 | |
Soda | Na2O | 10–16 | 0–8 | Poisson’s ratio |
|
0.23 | 0.2 | |
Boron oxide | B2O3 | — | 7–15 | Tensile resistance |
|
45–120 | 45–120 | |
Potassium oxide | K2O | — | 0–8 | Compressive resistance |
|
1000 | 1000 | |
|
||||||||
Magnesia | MgO | 0–6 | — | Coefficient of thermal expansion |
|
9 | 3.1–6.0 (class 1 to 3) | Thermal |
Alumina | Al2O3 | 0–3 | 0–8 | Specific heat capacity |
|
720 | 800 | |
Others | — | 0–5 | 0–8 | Thermal conductivity |
|
1 | 1 | |
Emissivity (corrected value) |
|
0.837 | 0.837 | |||||
Transition temperature |
|
530 | 530 | |||||
Maximum thermal stress |
|
40–250 | 40–250 |
aDepending on the treatment of glass; btheoretical value, in accordance with [
As far as the transition temperature
Crack occurrence and propagation, however, may even prematurely occur due to possible thermal stresses, hence requiring a typical multidisciplinary approach for such a constructional material. Thermal shocks, that is, cracking due to temperature gradient between heated and unheated glass regions, as well as due to relatively low thermal expansion coefficient of glass, are in fact typically expected to occur when the temperature gradient lies in the order of 40°C for AN glass, up to 100°C for HS glass, and 200–250°C for FT glass [
Further design issues and complexities arise as far as glass systems do not consist of single panes but are assembled in composite laminated sections and/or insulated glass units, as conventionally in use for buildings.
Laminated glass (LG) represents, in general terms, the combination of two or more glass plies together with foils consisting in a certain interlayer type. LG has been first developed for automotive applications, since early 1900, to avoid injuries in case of accidents, and only in the last decades, LG has been largely used in civil engineering applications for structural purposes. As a general rule of the LG concept, the resisting cross section is expected to respond as a composite system to external loads, hence having enhanced mechanical performances than single glass panes, both in the elastic stage and in the postcracked phase. From a mechanical point of view, the first implicit advantage of LG structural applications is that multiple glass layers can be bonded together; hence, the required level of resistance, stiffness, and redundancy can be obtained by using conventional glass thicknesses available on the market. In addition, thanks to the presence of bonding films, LG represents since decades the conventional safety glass solution in buildings, since able to hold together glass shards in case of failure, hence reducing possible risks for people (Figures
Laminated glass: (a)-(b) examples of fractured LG panels and (c) variation of shear modulus for common LG interlayers (PVB and SG degradation with temperature [
Bonding films typically consist of polyvinyl butiral (PVB) films, ionoplast foils (i.e., SentryGlas® (SG)), and ethylene-vinyl acetate (EVA) compounds. As a common aspect of such possible interlayers, besides their different constitutive laws, these films are generally characterized by viscous behaviour; hence, they are generally sensitive to temperature and load-time application, as also emphasized by several research efforts [
In terms of structural design of LG systems under ordinary loads, various methods are available to account for the effects of interlayer degradation over time and temperature increase (see [
Multiple glass panes (monolithic and LG sections) can then be assembled together to act as insulated glass units, both double (i.e., single gas cavity interposed between glass panels) or triple (i.e., double cavity). There, design calculations should take into account the so-called
FR glazing represents a relatively recent solution, known to provide excellent protection for lives and property in the event of fire, and hence may be used as a barrier for fire separation or compartmentation (for a specified duration), enabling occupants to assemble in a relatively safe compartment and acting as a part of an integrated “fire safety strategy” for the full building it belongs. As a crucial aspect of such solutions, FR glass systems require extreme attention in installation detailing. In addition, all the FR components, such as the glazing seals, beads, fixings, and frame, must be compatible and work together to achieve the required performance [
FR glazing, due also to relatively high costs, actually finds limited applications in buildings, especially where protected escapes must be ensured in the case of fire accidents. In accordance with a study carried out by Yang et al. [
In accordance with design guidelines such as [
FR glass: (a) examples of application in a public building, (b) wired glass, (c) double LG with intumescent layer, and (d) working principle of a triple LG with intumescent interlayers (schematic cross section).
Generally speaking, for structural design purposes and mechanical calculations, a given FR system can be conventionally treated as a standard, non-FR glass element (Sections
The performance of glass under high temperatures under heating and fire loading attracted the attention of several experimental research studies, since 1950s, due to the huge use of glazing panels in windows and fenestrations. Most of those investigations are related to thermal shock effects in SLS glass, as well as to its thermal characterization in general, including variations of modulus of elasticity (MOE) and resistance with high temperatures, while only limited experimental studies are currently available for composite glass systems and assemblies under fire or combined fire and mechanical loads (Section
As a conventional nominal value for glass transition temperature, design standards suggest a conventional value
Rouxel and Sangleboeuf [
Due to the intrinsic properties of SLS glass, it is in fact known that, as far as the service temperature increases up to
The elastic properties of standard glass at elevated temperatures have been extensively assessed by Rouxel [
MOE variation in SLS glass and other glass types, as a function of temperature, as reported in [
Earlier experiments were also carried out both on SLS and BS glass components by Kerper and Scuderi [
Close correlation can be observed with MOE variations in standard AN glass specimens, as derived from different literature sources (see Figure
MOE variation as a function of temperature, for SLS annealed and BS glass types.
Worth of interest for structural design purposes is that Kerper and Scuderi [
Following [
Thermal characterization of SLS glass. (a) Variation of MOE and resistance, under thermal shock [
Later on, Xie et al. [
Worth of notice that, as far as different literature references are examined (see, e.g., [
Moving from the material to the assembly level, such a need of further experimental assessment and investigation can be further perceived.
Experiments related to the thermal breakage of specific glazing systems under fire loading have been in fact carried out only recently, that is, for double glazing units [
Fracture of point-fixed glass panels, in accordance with [
In this regard, Chen et al. [
In the case of LG systems, for example, the thermal performance of interlayers of common use should be properly taken into account. In this regard, Debuyser et al. [
Thermal properties of glass, as a function of temperature, as derived from several literature references. (a) Specific heat capacity and (b) thermal conductivity.
Although the relatively large number of experimental studies focused on the thermal performance of glass as a constructional material, limited literature efforts are still available on the fire performance of full glass systems and assemblies (see a selection in Table
Summary of selected experimental research studies on structural glass systems under fire.
Reference and year of publication | Test typology/setup | Specimen size/loading | Glass type | Additional FR tools | |
---|---|---|---|---|---|
Walls, facades, and enclosures | Mejicovsky (2007) [ |
Frame supported, double LG (special setup for heat transmission) | F/T | SLS (FT) | — |
Machalická et al. (2016) [ |
Frame supported, double LG | F/T | FR | Gel-filling layer | |
Manzello et al. (2007) [ |
Frame supported, double LG + monolithic (furnace) | F/T | FR | Gel-filling layer | |
Yang et al. (2011) [ |
Frame supported, monolithic | F/T | FR | — | |
|
|||||
Window retrofit | Koudijs and Csoke (2013) [ |
Double glazing unit | F/T | SLS (AN, HS) | Low-E coating |
Misawa et al. (2013) [ |
Double glazing unit | F/T | SLS (AN) | Low-E coating + refractory film | |
|
|||||
Floors and overheads | Siebert and Maniatis (2008) [ |
LG, frame supported | F/n.a. | n.a. | n.a. |
Davis (2013) [ |
LG, frame supported | F/M | SLS (FT) bonded to FR glass | Liquid laminating film | |
|
|||||
Beams | Veer et al. (2001) [ |
4 point-bending; monolithic, triple LG, insulated + segmented beams (glass flame, bespoke setup) | S/M | SLS (AN) | Intumescent protective coating |
Bokel et al. (2003) [ |
4 point-bending; triple LG (glass flame, bespoke setup) | S/M | SLS (AN), FR glass | Epoxy interlayers | |
Louter and Nussbaumer(2016) [ |
4 point-bending; triple LG | F/M | SLS (AN, HS, FT) | — |
S = small scale; F = full scale; M = mechanical loading; T = only thermal load; n.a. = not available.
Glazing enclosures and walls attracted the attention of researchers especially during the last years, to assess the fire performance of novel FR solutions in place of standard glass. In doing so, the actual boundary and loading configurations were properly taken into account in defining the test setup and methods, so as to reproduce the testing conditions of full-scale specimens as a part of full buildings and complex systems.
Glass enclosures designed for an extension of Washington Dulles International Airport (automated train system for passengers) were tested under fire conditions in 2007, as reported by Mejicovsky [
Glass enclosures under fire. (a) Evidence of partial delamination, as reported in [
Glass elements for overheads and walls were designed in the form of LG sections, composed of two 10 mm thick FT glass panes bonded by a 1.52 mm thick PVB interlayer. Glazing joints were then realized by means of silicone rubber setting blocks and structural silicone sealant joints (Dow Corning 995™ type), while the glazing channel and edge trims consisted of minimum 3 mm thick stainless steel. The fire test was stopped after 35 minutes of exposure, with temperatures in glass over 250°C (up to 400°C in the last 5 tests of the experiment). Postexamination of the glazing system revealed no cracking or dislodgement of components, but localized melting and off-gassing of the PVB interlayer was observed, in the form of small bubbles and/or partial delamination (see details of Figure
A full-scale standard fire test according to EN 1363-1 and EN 1364-1 was reported by Machalická et al. [
Both double LG panels composed of special tempered glass, gel-filled type (SAFTI SuperLite II-XL™ type, with 19.05 mm the total thickness), and monolithic glass panels (6.35 mm the thickness) were assembled together to obtain the glazing wall tested in [
Failure mechanism in FR glass systems, as observed by (a) Manzello et al. [
This is not the case of experimental studies carried out by Yang et al. [
Several types of protective films able to improve the fire performance of existing and novel standard windows are available on the market. Although these coatings do not affect the room temperature elastic stiffness and resistance of a given glass pane to retrofit, the same films can be beneficial in delaying high-temperature effects, hence resulting in increased FR performance.
Koudijs and Csoke [
Experiments on coated glazing windows, as reported in (a) [
Misawa et al. [
Although the mentioned research studies generally proved the potential and efficiency of such special coatings to enhance the FR of glazing systems, several aspects should be still assessed. Wu et al. [
Siebert and Maniatis [
Fire tests on glazing floor, as reported in [
Similar fire experiments are summarized also in [
Fire tests on glazing floor, as reported in [
Aiming to ensure appropriate structural integrity as well as overall performances in the case of fire accidents, the typical glass panel (3 × 1.1 m the maximum size) consisted of a LG section, three 10 mm thick, FT glass layers, bonded together to a 23 mm Pilkington Pyrostop™, by using a special liquid composite bonding (Koediguard™ type). Steel-framing elements were also preliminary treated with intumescent coatings, mineral wool, and fire check boards. Additional expansion joints were finally included in the setup, in order to prevent buckling in the supporting frame members. The fire experiment was carried out in accordance with EN regulations, assigning to the flooring system a standard fire curve and a simultaneous mechanical load, being representative of crowd (5 kN/m2, distributed uniform pressure or 4.5 kN point load (50 × 50 mm foot print), resp.).
The fire performance exceeded 60 minutes of integrity (test stopped after 68 minutes), with maximum recorded temperatures of 67.6°C and no evidence of debonding or failure. Compared to fire insulation requirements (EN 1363-1), where temperature rise should not exceed 140°C (Section
Limited experimental literature background can be found for structural glass elements under fire loading, due to the relatively recent demand of vulnerability assessment and protection for these systems.
Veer et al. reported in [ 6 mm thick, AN glass; 6 mm thick, chemically toughened glass (120 MPa the initial stress); 3 mm thick, chemically toughened glass, laminated with 1 mm polycarbonate (PC) foil (3 glass layers + 2 PC films); 3 mm thick, segmented, chemically toughened glass, laminated with 1 mm PC foil (3 glass layers + 2 PC films). Compared to C, glass layers were bonded to PC foils in an overlapping pattern; An LG beam (as in the case of C and D type specimens), including insulating cavities on the external sides.
Summary of fire experiments carried out in [
All the (a)–(e) configurations were tested both without and after the application of intumescent coating (FlameGuard HCA-TR™ paint type).
A conventional four-point bending test setup was used for these beams. Experiments with no external mechanical loading as well as with additional weights (with up to 24 MPa the corresponding midspan bending stress) were carried out on monolithic AN specimens (type A) (Table
Summary of bending test results reported in [
Specimen | Paint | Failure time (min) | Failure mode |
|
Notes |
---|---|---|---|---|---|
A | No | >30 | I | 250 | — |
No |
2.4 | BR | 250 | — | |
Yes | 19 | BR | 250 | Beneficial effect of intumescent paint | |
|
|||||
B | No | >40 | I | 300 | — |
Yes | >40 | I | 300 | — | |
|
|||||
C | No | >30 | PC evaporation | n.a. | Central 10 cm evaporated; no delamination at the beam ends |
Yes | >30 | PC evaporation | n.a. | Central 7 cm evaporated; no delamination at the beam ends; beneficial effect of intumescent paint | |
|
|||||
D | No | 1.45 | C | n.a. | Premature failure of the adhesive layer between glass segments, with subsequent delamination, dislodgement of segments, and buckling |
Yes | 4.1 | C | n.a. | Beneficial effect of intumescent paint, but similar failure mechanism | |
|
|||||
E | No | >30 | PC melting | n.a. | Significant thermal damage; central 7 cm PC melted (visible after 5 min) |
Yes | >39 | PC melting | n.a. | Beneficial effect of intumescent paint, even with significant thermal damage; central 7 cm PC melted (visible after 10 min) |
The overall experimental investigation gave evidence of some important aspects, as, for example, the potential safety level of structural glass beams under fire.
However, critical aspects were also emphasized for the same specimens, like, for example, in the case of segmented beams (type D, see Figure
Bokel et al. [
Louter and Nussbaumer [
A four-point bending test setup was considered, with end supports protected from fire exposure, and the fire loading was assigned together with a simultaneous, constant mechanical load taking the form of 115 kg at the midspan section. Given the limited stress effect due to the assigned mechanical load (with maximum tensile stresses in the order of 5 MPa at beams’ midspan), the specimens proved to offer a rather stable behaviour under fire, for >40, >45, and >50 minutes in the case of AN, HS, and FT beams, respectively, up to collapse (Figure
LG glass beams tested by Louter and Nussbaumer [
In this paper, a state of art on structural glass systems under fire loading was presented, with careful consideration for current design methods and issues as well as experimental research efforts. Besides the continuously increasing use of glass in buildings as a constructional material able to interact with and/or replace materials of traditional use, the actual behaviour of structural glass assemblies, in general, currently requires further investigations, as well as the application of specific fail-safe design rules. This is the case of glazing systems under ordinary loads, but especially of extreme loading conditions, as, for example, fire accidents.
As shown, the intrinsic features of glass and its interaction with other components (i.e., framing systems, boundary details, etc.) make glazing systems highly vulnerable to temperature variations, as well as combined effects of thermal and mechanical loads, hence requiring multidisciplinary approaches in their design. In doing so, appropriate structural safety levels should be in fact ensured in combination with multiple aspects, such as transparency, aesthetics, and lightening requirements.
At the material level, in particular, a wide set of experimental research can be found in the literature, aiming to assess major effects of high temperatures on MOE, tensile resistance, and thermal cracking of standard glass. Most of these experimental outcomes are in rather close agreement as far as the MOE variation with temperature is considered. But when different literature sources are accounted, however, test results can also give evidence of high scatter in the observed trends, as, for example, in the case of glass thermal resistance (Section
As far as the attention moves from the material to the system and assembly levels, a relatively wide set of experimental investigations can also be found in the literature, with careful consideration for the fire performance of various typologies of glass systems inclusive of a multitude of boundary configurations, fire exposure patterns, and glass types (standard and/or FR glass). As a common aspect of such experimental investigations (Section
Finally, literature efforts have been spent in the last years also to assess the potential and efficacy of special coatings and films for the retrofitting and protection of existing glass windows and systems in general. In accordance with earlier observations, such solutions generally gave evidence of major benefits for uncoated glass specimens, but careful consideration should be still spent to properly optimize their potential.
The author declares that there are no conflicts of interest.
This research study has been carried out within the “Structural Task” activities of the ongoing EU COST Action TU1403 “Adaptive Facades Network” (