Using Transparent Intumescent Coatings to Increase the Fire Resistance of Glass and Glass Laminates

Transcription

1 Using Transparent Intumescent Coatings to Increase the Fire Resistance of Glass and Glass Laminates F.A.Veer*, M. van der Voorden*, H. Rijgersberg*, J.Zuidema+ *Building science group, Faculty of architecture, Delft University of Technology +Laboratory of materials science, Delft University of Technology Keywords 1 = Fire resistance 2 = Intumescent coatings Abstract The poor fire resistance of glass is one of the limitations in the use of glass as a structural material. Although recent developments in the field of borosilicate glass have lead to the introduction of glass panels with improved fire resistance the large scale use of borosilicate glass is not very likely in the near future. Intumescent coatings have been developed to improve the fire resistance of steel, aluminium and wood. A transparent intumescent coating should be able to increase the fire resistance of glass structures without affecting the unique transparency of glass. Tests and numerical studies have been conducted on stressed specimens of annealed float glass, tempered glass, glass laminates and advanced types of glass beams to test the effect of the intumescent coating on the fire resistance. Introduction In modern architecture glass plays an increasing role. This includes some example of glass used as a structural element. Current building code have little or no provision for glass used in this manner. One of the problems is the fire resistance of glass structures. Although some data is available on the use of glass in facades with metal structural elements nothing is known about structural glass used for overhead constructions. To investigate the fire resistance of structural glass tests were conducted on beams of annealed float glass, chemically toughened glass, chemically toughened glass laminated with polycarbonate foil, and a special laminate with insulating cavity. Experimental set-up To conduct the tests a special testing rig was made. The beams were subjected to 4 point bending using weights. Sufficient weights were used to achieve a maximum stress in 4 point bending of 24 MPa. This is a reasonable load for an advanced glass laminated composite beam, see Veer et al. [1]. A propane burner, connected to a large gas bottle with a reducing valve providing a constant pressure, was placed at a constant distance from the beam. This ensured a constant flame with a temperature of 650ºC at the surface of the beam. The flame hit the beam between the central loading supports of the 4 point bending apparatus thus ensuring that the most heavily stressed region was subjected to the highest thermal loading. At the back of the specimen, directly behind the flame, the temperature was measured regularly using a thermocouple. This particular test configuration was chosen because no suitable standard test for structural glass exists. The temperature selected was based on the following premises : Enclosures with large overhead structural glass will architecturally be used as entrances or atria. These enclosures will not have a lot of combustible materials inside and will thus usually be open spaces where a fire cannot encounter enough combustible materials to achieve temperatures in the roof of more than 650ºC. The test is thus realistic and even conservative for the likely application A photograph of the test rig in use is shown in figure 1. It was necessary to place the test rig under a ventilation system to remove the strong smell which occurred during the testing of the laminated specimens. The burning and evaporation of the polycarbonate otherwise causing a strong stench that would linger for several hours. 392

2 Figure 2: Laminated beam with insulating cavity Figure 1: Test rig for fire resistance experiments Configurations tested The following beam configurations were tested: A 6 mm thick annealed float glass B 6 mm thick glass chemically toughened to a prestress level of 120 MPa C 3 mm thick continuous chemically toughened glass laminated with 1 mm thick polycarbonate foil in a ply of 3 layers of glass and two of PC D 3 mm thick segmented chemically toughened glass laminated with 1 mm thick polycarbonate foil in a ply of 3 layers of glass and two of PC, the glass being bonded to the PC in an overlapping pattern [1]. E a laminated beam as above with an insulating cavity on the surface, this is illustrated in figure 2. Results for configuration A The results of the tests are summarised in table 1. Interesting is that simple annealed float glass will not fail under the imposed thermal load if no external force is applied. If the glass is loaded and the thermal load is added the annealed float glass will fail quickly. Temperature measurements indicate that the surface temperature at the back at the moment of failure is approximately 250ºC. On exposure to fire the intumescent paint rapidly forms a durable foam insulating the glass from the flame. This is visible in figure 3. The insulating effect of the intumescent coating extends the time to failure. This is caused by insulating the glass from the heat and thus slowing down the temperature rise. The temperature at the back of the specimen at failure was again around 250ºC, but the time to failure was increased from 2.4 minutes to 19 minutes. All specimen had a length of 400 mm and a height of 40 mm. These configurations were tested with and without a protective layer of transparent intumescent The intumescent paint used was Flameguard HCA-TR. The laminates were made using the DELO photobond 4455 photo catalytic acrylic adhesive cured under blue light for 2 minutes. Figure 3: Insulating foam layer formed from intumescent paint Results for configuration B The temperature in stationary condition was not high enough to induce sufficient stress in the 393

3 chemically toughened glass. Temperatures at the back of the specimen of 300ºC were measured. Results for configuration C The continuous laminate did not fail but the PC in the centre of the specimen started to melt and later evaporate. Significant damage occurred but the outer ends of the specimen showed no delamination and the adhesive was strong enough to hold the specimen together, despite the wedging that occurs due the volume increase of the polycarbonate during heating, melting and boiling. The intumescent paint coating slowed down the temperature rise and limited the damage to the specimen. Results for configuration D The segmented laminate failed rapidly by failure of the adhesive layer in the centre. The glass segments in the centre were dislodged after which the specimen buckled, as shown in figure 4. The intumescent paint slowed down the temperature rise but failure occurred in the same way. The polycarbonate was completely intact, failure having occurred due to cohesive failure of the adhesive under the thermal loading. Figure 4: Buckled specimen of segmented laminated glass Results for configuration E The segmented beam with an insulating cavity did not fail but significant thermal damage could be observed in the centre. It should be noted that the damage to the laminate took some 5 minutes to become visible without the intumescent paint and 10 minutes with the intumescent paint. Fires of low temperature and/or short duration should thus not cause any significant damage to structural elements with a similar configuration. Thus it should be possible to combine adequate mechanical safety and fire safety in a single transparent structural element. Numerical simulation of thermo-mechanical loads To investigate the problem further coupled field finite element studies were made using the ANSYS finite element software package. The focus of these studies were the failure of the annealed float glass of configuration A, the heat transport in the continuous laminate of configuration C and the failure of the segmented laminate of configuration D. The analysis of the annealed float glass beam showed that the measured temperature at failure had induced a thermal stress of only 7 MPa which was superimposed on the 24 MPa resulting from the mechanical loading. This is shown in figure 5. This also explains why the chemically toughened beams did not fail. Even at the higher temperature of 300ºC which was obtained in the stationary condition the maximum stresses would have been about 40 MPa, which is one third of the pre-stress which was obtained by the chemical toughening. The heat flow analysis of the continuous laminate shows that the heat is primarily transported by the glass in the length of the beam. This increases the area across which heat is transferred to the polycarbonate. Thus a large area of adhesive and large volume of polycarbonate is affected by the heat. This means that the build-up takes more time. The temperature rises until the polycarbonate Table 1 : Summary of experimental results, average of 3 to 5 tests in each category Configuration Average time to failure Condition of specimens at end of test A without load >30 minutes Intact A 2.4 minutes broken A with intumescent paint 19 minutes broken B >40 minutes Intact B with intumescent paint >40 minutes Intact C >30 minutes PC evaporated over central 10 cm C with intumescent paint >30 minutes PC evaporated over central 7 cm D 1.45 minutes cohesive failure between glass segments and PC D with intumescent paint 4.1 minutes cohesive failure between glass segments and PC E >30 minutes PC in core shows melting over central 7 cm E with intumescent paint >30 minutes PC in core shows melting over central 6 cm 394

4 first starts to melt and later to evaporate. After this the heat is being transported by the glass to the colder ends until a steady state situation has developed. Presumably some heat is transferred between the glass plies after the PC has evaporated by conduction and radiation but this has not been modelled due to lack of reliable data. In this situation the thermal stress is insufficient to cause failure of the chemically toughened glass. In the segmented laminate the heat flow is different, the heat that touches the glass surface has to pass through the adhesive layer to the polycarbonate to go to the remainder of the glass. The segmentation prevents heat being transferred in the length of the glass ply. This causes a very high thermal strain on the adhesive which results in debonding. Figure 6 shows the calculated temperature distribution after 30 seconds for this configuration. The temperature distribution clearly shows that the adhesive behind the segment which is touched by the flame is subject to high temperatures. This explains the debonding of the central segments and the buckling of the beam shown in figure 4. Figure 5 : Stresses in Annealed float glass specimen subjected to mechanical and thermal loading Discussion The results suggest that annealed float glass is a very unsafe structural material with regard to fire safety. Relatively small exposure to heat will result in failure due to inhomogeneous thermal strain. Toughened glass beams are considerably more resistant, as they have far higher strength. Small fires should pose no danger to structural tempered glass members. If the fires are of too long a duration the pre-stress level of thermally tempered and chemically toughened glass will fall but this will require a significant exposure at high temperature. Off course it will be almost impossible to determine how far pre-stress levels have fallen after fire damage. This means that complete replacement of unprotected structural glass members after exposure to fire should be necessary as a safety precaution. Continuous laminates made from toughened glass should also be resistant to fire as long as a large enough region is not exposed to high temperatures. If a mechanical safety is available to prevent the glass plies form rotating and being subjected to buckling the structure should be safe. Severe damage to the polymer interlayer should however be expected requiring complete replacement after the fire. Another problem is that the maximum size of continuous laminates is limited by the sizes of glass that can be provided. Thus this will provide a solution for spans of upto 4m in length. Discontinuous laminates are not very resistant to fires. Their nature will cause quick local debonding of the glass segments near the fire after which the structure will buckle. This can be considerably improved by adding an outer layer forming an insulating cavity. A structure of this type should have significant mechanical and fire safety while retaining adequate transparency. As the limitation in size of these beams will limited to the processing ability in the factory and the ability to transport the element, size of 20 m or more are possible. The use of transparent intumescent coatings will improve the fire safety of glass structures. It should be noted configurations that are not intrinsically safe cannot be absolutely protected by this means. A more likely usage of transparent intumescent coatings is to limit the damage done to laminated members and thus obviate the need for replacement after small fires. The coating can easily be removed and replaced. Figure 6: Temperature distribution in segmented laminate after 30 seconds Conclusions Spanning beams of toughened glass should be adequately fire resistant considering the fire risks in the spaces where they will be used for architectural reasons. Beams of continuous 395

5 laminated glass should also have adequate fire safety providing provision is made to prevent rotation of the glass plies which will cause buckling. In practice these beams will suffer damage to the polymer interlayer which will require replacement of the element after the fire. Large transparent beams of segmented laminated glass can be made to have adequate mechanical and fire safety by using the appropriate construction. These are still susceptible to damage to the polymer interlayer. The use of intumescent paint reduces the heat build-up and thus slows down the development of thermal strain in the glass and on the adhesive layer. This increases the safety for short duration and/or low temperature fires. use of intumescent paint alone will not provide fire safety for a structural member that does not have sufficient build in fire safety. References 1 F.A.Veer et al., proceedings 4th glass processing days, Tampere