Study Of Window Glass Breaking In Compartment Fires - Report Example

Chapter 4: Results Analysis and Discussion 4.1 Introduction to the chapter This chapter presents the experimental test results obtained from breaking of Window Glass in compartment fires. Representative test results are displayed graphically and later evaluated in this context. Tabulated results for the entire experimental study are then documented with their equivalent numerical analysis. Following this, the report discusses the time to cracking and the conditions of fire at the time. This then followed by a discussion of the relation between the occurrence of glass cracking and compartment gas temperature, and finally the effects of ventilation in a reduced scale compartment. Summary of results Table 1 below summarises the results collected from the window glass breaking tests in a compartment. There were a total of twelve tests using two different ventilation vent sizes i.e. 150mm and 50mm, and three different fuel positions i.e. back, front and middle positions within the fire compartment. In tests number Run 3 (centre & 150mm) and Run 12 (Front, 50mm) compartment fire, there were no cracking observed. Each of other ten tests resulted in single and multiple glass-cracking patterns. For each of the tests conducted in this experiment, the fuel position, vent size, glass breakage temperatures and the corresponding breakage time are listed in table 1 below. The lowest average gas temperatures was recorded in three experiments in Run 1 (150mm, centre), Run 6 (150mm Front), and Run 7 (150mm, Front), i.e. gas temperature 2950C, 298.50C and 290.50C, respectively. The two lowest time-scales recorded were in test runs Run 11 (50mm centre) and Run 10 (50mm, back) which are 16 and 21 min respectively. The average glass breaking time for the ten successful tests was 22.15minutes. All tests recorded cracking gas temperatures in the region nearest to the glass window are higher than the mean gas temperature of the smoke layer, except Run 10. Table 1: The relation between the occurrence of glass cracking and compartment gas temperature 4.1 Glass Breaking patterns The figures in table 1 above shows the breakage patterns recorded from the ten window cracking tests where cracking occurred. In all the tests, the glass cracks initiated from the edge of the glass window and propagated rapidly, such that the entire cracking was finished in less than a one second. In all the tests apart from tests1 and 11, the figures show that there were multiple cracking patterns, with cracks spreading throughout the entire sample and some later joining together. It seems that when the fuel is located at the back of the compartment, cracking pattern tend to follow the shape of flame “V” shape, runs 4, 9, and 10. As this is not seen in any published literature, further study is needed to confirm it. In all scenarios the window glass initially cracks at points of stress concentration, on the edge of glass. This is characteristically at locations where there exist imperfections, such as notches that occur after cutting the glass, or locations where the glass has undergone stresses, for instance on points where glazing points bind the glass in place. This was observed in Run9 where the glazing sealant (putty) was attached to the glass window for a whole week. For instance when the glass starts to fail, it starts to develop a bifurcating fracture pattern followed by multiple cracks propagating from the edge of the glass. Also, it was noted that back fuel position runs showed different Crack patterns forming ‘V’ shape specifying and following the flames. 4.2 Glass cracking relative to heat produced by fire in different locations In terms of fire location, glass cracking in configuration with fire at the middle of the compartment and with either 150mm or 50mm vent occurred at almost the same time (Runs 1 and 2 see Figure 16 below). It was noted several times earlier that fire locations appears insignificant to glass cracking events and this is still true because although indirect, the heat and gases produced by fire regardless of location are also responsible for glass cracking. As shown in Figure 14 gas temperatures recorded at times 1,440 (or 24 minutes) and 1,380 (or 23 minutes) in sequence, were from hot gases produced by fire. Similarly, a 150mm2 PMMA Slab fire source can easily raise the temperature of a compartment area that is roughly 0.08645m3. Note that the type of fuel selected for this small-scale experiment as stated in Section 3.5.3 can provide steady heat of combustion and fire growth similar to those of a furnished compartment thus rapid hot gases production can be expected. Figure 1: Thermocouple arrangement relative glass window and fire location By analysis, the experimental result indicate that glass cracking in relation to heat produced by fire in different locations has no direct effect. For instance, as shown in the thermocouple arrangement below, TC01 to TC04 is at the back of the compartment and very near to the glass window. Logically, if the heat from burning fire located near TC01 (Chan 1) and directly under or attached to the wall TC02 (Chan 2), TC03 (Chan 3), and TC04 (Chan 4) can actually advance the glass cracking and may also lead to complete fall out of glass window which never been recorded with in this experimental study. This is may be because of flame impinging on the glass surface directly. The gas temperatures recorded by TC01 (or Chan 1) is significantly lower than the rest TC02, 3 and 4 (or Chan 2, 3 and 4) which indicates the lower part of the fire box has much lower temperature compared with the upper part. This complies with the general understanding of the existence of a hot layer and a cold layer in a compartment fire. When calculating the average temperature, only the hot layer measurements were considered. As TC1 is judged to be in the cold layer, gas temperatures taken by TC 1 were not included in the average temperature calculation in all temperature data collected throughout. Figure 2: Configurations with fire just under the glass window but with different vent size In configuration where fire is at the other side of the compartment and away from the glass window as shown in Figure 18, the cracking time for both 150mm and 50mm vents are almost the same. Figure 3: Configuration where fire source is away from the glass window 4.1 Time to cracking and the conditions of fire at the time Experimental results analysis suggests that glass cracking occur in all test configurations. However, glass cracking-time occurs after 20-25 minutes as shown in figure 13 below. Figure 4: Glass cracking in relation to vent size and fire location As shown in Figure 16 and indicated by dotted cracking point line, glass cracking in both 150mm vent and 50mm (reduced-scale vent) occurs at same range of time1440 seconds or 24 minutes and 1380 seconds or 23 minutes respectively. However, it should be noted that in Run1 and Run2 where the fire is in the middle of the compartment, the 50mm vent size advanced glass cracking occurrence by about 60 sec. or 1 minute. Note that glass cracking in Run1 and Run2 occurred at 1440 seconds at 294.5 oC and 1,380 seconds at 356 oC respectively. These temperatures were recorded by taking temperature scans of TC3&4 or the thermocouple nearest to the glass. 4.2 Glass cracking relative to gas temperature and ventilation size Figure 5: Vent effects on upper layer temperature and glass cracking As mentioned earlier, the glass cracking time-scales shown in Figure 14 are not associated with fuel or fire location. Note that if these cracking events resulted from direct exposure of heat from the fire or flames, where the fire is located near or under the glass window, the outcomes will be significantly different. In fact, there is a strong possibility that the glass cracking in configurations where fire is located at the front of the compartment occurred much later than the case when the fire is at the back by approximately 300 seconds. On the other hand, changing the vent size resulted in high gas temperatures recordings hence glass cracking time advanced notably (see figure 13 above Runs 1 and 2). Figure 6: Runs 5 and 6, higher gas temperature effect on glass cracking events A The relationship between glass cracking events and rising gas temperature (vent size) is also evident in Run8 and Run12 as shown in Figure 15. Runs 1, 2 with Runs 5 and 9 were chosen to demonstrate gas temperature distributions recorded at glass region by TC3 & TC4. As shown in Figure 16 (Runs 1 and 2), their cracking time is almost similar. In Run 1, the gas temperature recorded at cracking time (1, 440 seconds or 24 minutes) is 295oC. In Run 2, the gas temperature recorded at (1380 seconds or 23 minutes) is 356oC. Where in Run 5 gas temperature of 349oC was recorded at cracking time (1380 seconds or 23 minutes) and in Run 9 gas temperature at cracking time (1260 seconds or 21 minutes) is 391oC. These temperature readings indicate similar variation but smaller temperature difference at 42oC compared to 61oC of Run 1 and 2, which suggest the cracking time also depends on how quick gas temperature rises. Notes: thermal inertia and how quick the temperature of glass can be raised by the hot gas. Figure 7: Higher gas temperature effect on glass cracking events B The above test result suggests that hot gases inside the compartment play an important role in glass breakage or cracking. Note that hot gases accumulate in the upper layer where the glass window is located. Therefore, regardless of fire location and vent size, the glass window during the test was subjected to high gases temperature and subsequent cracking. Note that TC4 or the thermocouple nearest to the ceiling recorded much higher temperature (345.5 oC) compared to TC3 that is directly under it. Flames were observed in the TC4 region. This is considered to be the reason why temperature measured by TC4 is higher. Glass is a poor conductor and they often breaks during a fire mainly because of thermal stress. Its poor conduction ability causes a big temperature difference between the inside surface and outside surface when the glass pane is exposed to hot gases convection. The thermal expansion of the glass window places its covered edge in tension until it cracks. Moreover, once cracking occurs the glass bifurcates or divided in two parts and crack rapidly spread across the window. As the same fuel configurations were used in all tests, heats produced in each experiment are similar. The upper layer temperature is not significantly affected by changing fire location in the manner we did in the experiments. Therefore, the fire location did not affect the cracking time (this must be in the suitable section vent effects) The effect of vent size in glass cracking is clearly demonstrated by the early cracking time due to rapid temperature rise in the upper layer of the compartment. As shown in Figure 19, the glass cracking time in configuration with a smaller vent is always ahead and with noticeable high temperature nearly reaching 600oC. Since smaller vents advances glass cracking time then the bigger the vent, the longer glass window last. In terms of upper layer temperature contribution to glass cracking, analysis of temperatures of both 150mm x 255mm and 50mm x 255mm vent suggest temperature readings in TC10 is higher when ventilation is restricted as shown in Table 4.2.4-A. The temperature level in TC10, which was in the opposite side of the compartment, is around 300oC at time 1,380 when the glass cracking started. Note that this value is lower compared to configurations with a larger vent indicating slower heat release rate and hot gases accumulation in the upper layer due to insufficiency of oxygen supply. Figure 8: Vent effects on upper layer temperature – Front Further examination of the relationship between glass cracking, temperature, and restricted vent, TCs located in the level of glass window reported higher temperature at 316oC near the glass with 50mm vent compared to 296.5oC of 150mm vent. Varying the size of vent also showed a notable increasing in compartment temperature, the smaller the vent size, the hotter it gets and that influenced the glass clacking time too. 4.3 Glass cracking relative to the MLR Fuel mass loss rate is one of the most vital aspects of post flashover in a compartment fire modelling. It determines the fire heat release rate and hence the temperatures recorded in the compartment. Mas Loss rate (g/sec) has been empirically used to determine the burning rate of fuel. Ideally mass loss rate is proportional to the Heat Release rate (HRR). The slope of the mass history gives the mass loss rate during the entire fire development rate. In this experiment the mass loss rate was computed based on eth formula below: Where; MLR—mass loss rate, M0- Initial Fuel mass Mi- Residual mass at time t t- time in sec the results were as shown in table 2 below Table 2: mass loss rate at given combustion time Mass loss Rate (g/s) Time(s) Weight(g) 150 back Weight(g) 150 front Weight(g) 50 back Weight(g) 50 centre Weight(g) 50mm front 30 0.400 0.040 0.0333 0.007 0.007 60 1.400 0.047 0.0200 0.020 0.020 90 2.400 0.042 0.0244 0.024 0.024 120 3.200 0.035 0.0333 0.025 0.025 150 3.800 0.029 0.0307 0.051 0.024 180 4.400 0.026 0.0233 0.046 0.023 210 4.800 0.023 0.0219 0.046 0.027 240 5.000 0.023 0.0200 0.041 0.024 270 5.400 0.021 0.0193 0.038 0.023 300 5.600 0.020 0.0180 0.035 0.021 330 5.800 0.019 0.0200 0.032 0.020 360 6.400 0.019 0.0189 0.031 0.020 390 6.800 0.018 0.0185 0.030 0.019 420 7.200 0.019 0.0186 0.029 0.019 450 7.800 0.018 0.0182 0.028 0.019 480 8.200 0.018 0.0183 0.027 0.019 510 8.800 0.017 0.0188 0.027 0.019 540 9.600 0.019 0.0193 0.027 0.019 570 10.400 0.019 0.0196 0.027 0.020 600 11.200 0.020 0.0217 0.027 0.020 630 12.000 0.021 0.0219 0.027 0.020 660 13.000 0.022 0.0224 0.027 0.021 690 14.000 0.026 0.0229 0.027 0.021 720 15.200 0.027 0.0236 0.028 0.022 750 15.600 0.029 0.0232 0.027 0.022 780 18.000 0.031 0.0241 0.029 0.024 810 19.200 0.034 0.0259 0.030 0.025 840 21.000 0.037 0.0271 0.031 0.026 870 23.000 0.039 0.0267 0.032 0.027 900 24.800 0.042 0.0251 0.034 0.028 930 26.800 0.046 0.0265 0.036 0.030 960 29.200 0.050 0.0283 0.038 0.031 990 31.600 0.054 0.0287 0.040 0.033 1020 34.200 0.057 0.0304 0.042 0.034 1050 37.600 0.062 0.0328 0.043 0.037 1080 41.000 0.066 0.0341 0.045 0.037 1110 44.600 0.071 0.0346 0.047 0.039 1140 49.200 0.076 0.0368 0.051 0.043 1170 53.600 0.080 0.0388 0.053 0.045 1200 59.600 0.086 0.0420 0.055 0.049 1230 66.200 0.090 0.0447 0.059 0.053 1260 73.200 0.095 0.0487 0.062 0.057 1290 81.000 0.101 0.0518 0.067 0.062 1320 88.200 0.105 0.0570 0.070 0.066 1350 97.200 0.110 0.0600 0.074 0.071 1380 105.400 0.117 0.0610 0.077 0.083 1410 113.600 0.121 0.0698 0.083 0.088 1440 122.800 0.126 0.0740 0.088 0.091 1470 132.800 0.131 0.0814 0.100 0.096 1500 142.600 0.137 0.0883 0.104 0.101 1530 152.000 0.143 0.1012 0.109 0.105 1560 162.600 0.148 0.1060 0.112 0.110 1590 172.800 0.157 0.1104 0.117 0.114 1620 184.200 0.164 0.1154 0.122 0.119 1650 194.600 0.171 0.1196 0.126 0.123 1680 206.000 0.176 0.1243 0.130 0.128 1710 218.800 0.179 0.1296 0.135 0.133 1740 231.600 0.180 0.1290 0.140 0.138 Figure 9: Mass Loss Rate of Fuel by Time: Figure 21 shows that all the tests exhibited virtually similar mass loss rate from the initial time, as time increased, the fluctuation in the mass loss rate decreased up to 600seconds, thereafter mass loss rate increased steadily. Tests of door vent 150mm and front fuel position demonstrated the greatest fluctuation. The results suggest the fires used in the experiments were fuel bed controlled fires. Figure 10: MLR graph of Vent size= 50mm, and fuel position- Back, middle and front All the three fuel locations gave similar magnitude and path of mass with respect to time as shown in Figure 23. This showed that the different locations of fuel considered in the experiments do not have significant effect on the fire sizes. All the three tests started off with by gradual mass loss rate up to about 900sec. This later increased steadily up to final fuel combustion after about 2000sec. 4.4 Compartment peak gas temperatures The average peak temperatures in the compartment can be summarized as shown in table 2 below. Ra, Rb and Rc Shows the various, vent size fuel position Table 3: Average Peak temperatures for the tests Average peak temperatures (OC) Ra Average peak temperatures (OC) Rb Average peak temperatures (OC)Rc Run 1 380.00 380.00 380.00 Run 2 483.83 531.17 490 Run 3 313.17 362.66 271.33 Run 5 678 528 493 Run 6 447 544.5 657.67 Run 7 406.5 488 591.83 Run 8 378.67 395.67 548.75 Run 10 630.67 708 523 Run 12 573 529.625 434.625 The peak heat release rate is majorly determined by the size of the fuel and the position of fuel. From the graph above, it is evident that the highest peak temperatures are recorded in Run 10 (vent width size=50m, fuel position=at the back), while the lowest peak temperatures occurred at Run 3 Run 3 (Vent=150mm, fuel=centre) at 313.17 OC. The size of fuel plays a key role in defining the rate of peak heat release. Small fuel size releases lower peak release rate. The peak temperature is also determined by fuel size. Due to forced ventilation, there is remarkable reduction of smoke in the fire compartment in any particular volumetric flow rate, with increased flow rates causing decreased smoke emission. 4.2.1 Fire engineering justifications and analysis. The breaking of window glass is significant in fire engineering field in that the created wall openings created offer new inlets for fresh air as well an exit for the hot fire gases. This increased ventilation alters the burning configuration and the equivalence ratio and consequently the quantities and composition of the freeing combustion flue gases. Also the gases are main cause of fatalities during fire incidences as a result of fumes and smoke. 4.2.2 Conclusion of study There have been a number of prominent topics throughout the dissertation; and the findings in each of the topic area have been significantly dealt with in order to allow several conclusions to be drawn from this paper. The key finding from both qualitative and qualitative analysis of the recoded results from the experiment, permit broad conclusions to be done. This paper reviewed the previous work carried out on the behaviour of glass in compartment fire. 4.3.1 Practical related conclusions The most extensive theoretical studies were directed at determining the breakage mechanism and, subsequently, developing an understanding of the breakage mechanism. The temperatures difference within the fire compartment and the resultant thermal stresses were derived to predict this cracking point difference The tabular results presented in Table 1 shows that there is an inverse correlation between the time to cracking and the vent size. These results basically support the theory that glass breakage is direct related glass temperature or the temperature difference. The larger the door vents, the greater the heat flux blown towards the window, and consequently the rapid increase in glass temperatures leading to lesser breakage time. All resultant patterns of glass breakage show a rather qualitative trend in the data. Although the breakage was experienced in all but Run 3 and 12, there are notable differences between the breakage tests. There are single and multiple bifurcations in the tests conducted. It seems that multiple bifurcations are the most conducive type of window collapse. A similar pattern of glass breaking as a function of the compartment temperature can be noted. Although the time duration to glass cracking and the fire compartment temperature vary depending on the fuel size and the door vent size, the glass temperatures at crack initiation do no depict this same pattern. The temperatures recorded are probably a function of other variables as window size, however there seems direct correlation between time for glass cracking initiation and the corresponding temperatures. The breakage patterns depicted in the window breaking tests carried out in this experiment represent a crucial qualitative trend in the data recorded. Although there was glass breakage in all but one experiments in all but two experiments (tests 3 and 10), there are vital differences between the glass breakage patterns for the tests. Due to the complexities between all the factors considered and the available resources and time, a detailed research analysis is not feasible 5.1. Recommendations for Future Research The most significant challenge encountered when carrying out the tests is the difficulties of irradiative heating which affect negatively the thermocouple reading. This was possible due to the absorption properties of the thermocouple covering. The problem can be sorted out by shrouding the thermocouple with a reflective material such as aluminium foil. Another level of experiment would be to determine if there are cases of uneven heating of the glass window. The theory developed in this experiment is based on the assumption that the window is heated to uniform temperature. To determine heat uniformity, a series of thermocouples would be spread across the glass window to provide local heating temperatures. Read More

Table 1: The relation between the occurrence of glass cracking and compartment gas temperature 4.1 Glass Breaking patterns The figures in table 1 above shows the breakage patterns recorded from the ten window cracking tests where cracking occurred. In all the tests, the glass cracks initiated from the edge of the glass window and propagated rapidly, such that the entire cracking was finished in less than a one second. In all the tests apart from tests1 and 11, the figures show that there were multiple cracking patterns, with cracks spreading throughout the entire sample and some later joining together.

It seems that when the fuel is located at the back of the compartment, cracking pattern tend to follow the shape of flame “V” shape, runs 4, 9, and 10. As this is not seen in any published literature, further study is needed to confirm it. In all scenarios the window glass initially cracks at points of stress concentration, on the edge of glass. This is characteristically at locations where there exist imperfections, such as notches that occur after cutting the glass, or locations where the glass has undergone stresses, for instance on points where glazing points bind the glass in place.

This was observed in Run9 where the glazing sealant (putty) was attached to the glass window for a whole week. For instance when the glass starts to fail, it starts to develop a bifurcating fracture pattern followed by multiple cracks propagating from the edge of the glass. Also, it was noted that back fuel position runs showed different Crack patterns forming ‘V’ shape specifying and following the flames. 4.2 Glass cracking relative to heat produced by fire in different locations In terms of fire location, glass cracking in configuration with fire at the middle of the compartment and with either 150mm or 50mm vent occurred at almost the same time (Runs 1 and 2 see Figure 16 below).

It was noted several times earlier that fire locations appears insignificant to glass cracking events and this is still true because although indirect, the heat and gases produced by fire regardless of location are also responsible for glass cracking. As shown in Figure 14 gas temperatures recorded at times 1,440 (or 24 minutes) and 1,380 (or 23 minutes) in sequence, were from hot gases produced by fire. Similarly, a 150mm2 PMMA Slab fire source can easily raise the temperature of a compartment area that is roughly 0.08645m3. Note that the type of fuel selected for this small-scale experiment as stated in Section 3.5.3 can provide steady heat of combustion and fire growth similar to those of a furnished compartment thus rapid hot gases production can be expected.

Figure 1: Thermocouple arrangement relative glass window and fire location By analysis, the experimental result indicate that glass cracking in relation to heat produced by fire in different locations has no direct effect. For instance, as shown in the thermocouple arrangement below, TC01 to TC04 is at the back of the compartment and very near to the glass window. Logically, if the heat from burning fire located near TC01 (Chan 1) and directly under or attached to the wall TC02 (Chan 2), TC03 (Chan 3), and TC04 (Chan 4) can actually advance the glass cracking and may also lead to complete fall out of glass window which never been recorded with in this experimental study.

This is may be because of flame impinging on the glass surface directly. The gas temperatures recorded by TC01 (or Chan 1) is significantly lower than the rest TC02, 3 and 4 (or Chan 2, 3 and 4) which indicates the lower part of the fire box has much lower temperature compared with the upper part. This complies with the general understanding of the existence of a hot layer and a cold layer in a compartment fire. When calculating the average temperature, only the hot layer measurements were considered.

As TC1 is judged to be in the cold layer, gas temperatures taken by TC 1 were not included in the average temperature calculation in all temperature data collected throughout.

Read More