Experimental Study on Flame Merging Phenomenon

No less than 350 housing fires happened each year during 2016-2021 in DKI Jakarta. Fatality is fortunately not a common issue, but it was predicted that 200 billion rupiahs losses annually from the fires. Narrow spaces with a bulk of people housing make the fire spread faster. Fire burns, whichever is the ‘ready-to-burn’ first, so it cannot be expected one house is already burned out first, then the fire will continue to the next house. From this phenomenon, it potentially has more than one hotspot from the fire that can grow bigger with a flame merging situation. For a better understanding of fire merging phenomenon, experimental study was reported in this work. The study was conducted on a laboratory scale experiment with the variation of the separation distance between the fire sources. The separation distance was set to be 0; 5; 10 cm. The fire source was shredded paper which is placed in cylindrical baskets having diameters of 5 cm, 10 cm, and 15 cm. The experimental data were recorded by using K-type thermocouple, visual and thermal cameras, and a scale. The results show that the probability of flame merging increases with the diameter of baskets. The temperature and flame height increase as the flame merging phenomenon occurs. However, flame merging does not have a direct impact on combustion efficiency based on this experiment.


Introduction
Fire events occur almost three times a day based on statistical data by Jakarta Fire Brigade [1].The cause of fires spread from the electrical vault, gas cooking, until the unattended fire.For example, from a bin waste burner or cooking stove represents about 30% of the accident appears in housing fire.More studies about housing fire from Yu Wang et al. explains how informal settlement's fire characteristic is different from the formal settlement.The main difference that has a significant impact on the fire event and frequent occurrence is the collapse of structures in formal settlements [2].Furthermore, the fire load in informal settlements can be more than 1,000 kJ/m2 in one house.A greater fire load represents a higher fire risk [3].This dynamic of housing fire is quite similar to Indonesia, especially Jakarta with packed population density.In addition, there are 64 areas in Jakarta identified as high risk of fire, where other 400 areas are in medium risk [4].Most of them have informal settlement characteristics, such as wood structure and partition, triplex ceilings, and old zinc roofs.The housing in Indonesia has dense population, which makes dense furniture and belonging in one place.A fire will not burn one house completely and then move on to the next house.It will reach whatever is drier or ready to burn first with the temperature and heat flux acquired from the fire source.It can be more dangerous when there are two fires with critical distance merging into one bigger fire.
Flame merging had been studied by D. Kamikawa et al. [5] with propane and wood crib as the main fuel for studying the flame merging phenomenon.They studied different characteristics of propane as a liquid fuel and wood crib as a solid fuel.It is concluded for the propane burner; the higher heat release rate resulted in a higher flame height linearly.The efficiency of combustion was a focused parameter for the wood crib burner, which the combustion efficiency increases with the higher separation distance between wood cribs.Another study was conducted in a scaled tunnel from a heptane pool fire burner with speed variations of wind effect [6].From the study, it was acquired that the fire will decrease at first but rapid increase in temperature and flame coverage after that.It caused merging and worse effects in downwind areas.It is aligned with what Himoto et al. studied by making a simulation from a real urban fire case in Japan, Sakata Fire [7].The flame merging phenomenon had also been studied in the event of ejected fire from the window.The fuel was ignited from an LPG burner, made the room saturated with fire, and then ejected from the window.Correlation between the probability of fire merging, the ratio between two window distances and length of a window had been explained in that paper [8].
However, there has not been any study about flame merging in accordance with housing fire yet.For that purpose, this paper reports a study of the flame merging phenomenon with a mixture of solid and liquid phase material to represent fire in an urban area.The objective of this study is to analyze the flame merging phenomenon and its characteristics.

Methodology
There were 27 experiments conducted.The variation diameter of baskets was 5, 10, and 15 cm.The fuel load consisted of shredded paper and gasoline as reaction accelerator, which was varied.The paper variation in the basket was always intended to be the same density at 0.04 gr/cm 3 .The variation distance between the edge of the basket were measured in 0, 5, and 10 cm.Visual and thermal cameras were used to record the phenomenon of each experiment.Each variation was done three times to validate the data acquired.Temperatures in each basket and in the middle of the basket were recorded with K-Type thermocouple connected to GRAPHTEC DAQ for data logging.The elevation of thermocouples was set in 2,5 cm; 7,5 cm; and 12,5 cm.Zero elevation is measured at the bottom of the basket.All the sensors had been calibrated.The arrangement of the experiment can be seen in Figure 1 and 2. The basket was made of wire-mesh steel materials and partly covered by an aluminum sheet.The side of the basket has a 2 cm height opening area with a total height of the basket of 5 cm.The opening area at the bottom was added for ventilation reasons.The opening area increases the stability of the flame and could enhance the probability of flame merging.Furthermore, the arrangement of paper in the basket also plays a role in the flame merging phenomenon.The arrangement was made to be sloped to the middle area so that the flame merging can happen.This is explained clearer in Figure 3.The variation of the experiment can be seen more clearly in Table 1.There were eight thermocouples placed in the experiments, as explained in the methodology.Thermocouple M1, which was placed at 2,5 cm elevation between the two baskets, was the most representative thermocouple for the flame merging phenomenon.It can be seen in Figure 5 that the separation distance between baskets affects temperature rise.The temperature of the flame merging phenomenon was recorded by M1 thermocouple in the range 600-700 o C. It indicates the further separation distance applied, the lower temperature between the baskets, and the lower probability of flame merging to happen.From Figure 5, it can be seen all these three variations have similar timing in first ignited, peak temperature, and decay moment.First ignited happened in 0.5 -1,5 minutes.Peak temperature happened in 1,5 -3 minutes.The decay moment occurred in 4-6 minutes.Although the temperature trend seems similar in time, the time merged is different between each diameter.The time merged aspect elaborates more in the next part.

Figure 6. Averaged Time Merged
It can be interpreted that the distribution and trend of temperature for all diameter variations are almost the same based on the explanation in section 3.1.It is found that the moment of merging time for each diameter is slightly different.As shown in Figure 6, the smallest diameter needs the longest time to merge compared to the 10 and 15 cm diameter of baskets.Because the amount of fuel will decrease as the fire burns and the mass rate cannot be constantly set, it affects the flame height for comparison between each experiment.This is the basis for measuring the flame height at the same time in seconds after the first ignition happened.It is obvious from this experiment that the size of the basket's diameter does not linearly correlate with the time needed for the flame to merge.The size of the basket's diameter linearly correlates to higher flame height.Another correlation that can be found from the experiment is that the closest separation distance between baskets shows the higher flame height as the flame merging phenomenon happened.An interesting phenomenon in 5 cm separation distance shows lower flame height compared to 10 cm separation distance.It is analyzed that the more significant separation distance resulted in more oxygen entrainment that can boost the flame height accordingly.Another analysis of the lower flame height in a 5 cm separation distance is the energy focused.The energy generated from the flame within a 5 cm distance is used to raise the temperature in the surrounding area rather than to make the flame higher.

Probability of Merging
From 27 experiments conducted, not all the variations could make the flame merge.It was the probability of merging for each diameter and separation distance variable.The calculation is as simple as this notation.Where P(m) is the probability of merging,   is the number of experiments that showed flame merging,   is the number of experiments with each variation that was conducted.

Figure 8. Probability of Merging
From the graphic plot in Figure 8, the larger basket's diameter, the bigger probability of merging happens.Although the time burn was different for each experiment, it shows clearly that the larger fire load linearly increases the probability of merging.The closer separation distance between baskets indeed increases the probability of merging.The best way to explain how big the fire happens quantitatively, exceed from the size of the basket's diameter, is from the heat release rate value [9].Estimated the value of heat release rate based on this equation [10].
Heat release rate in kW to show the fire size quantitatively, ̇ is mass rate burning of the burnt material in g/s, and ∆ℎ is the heat of combustion whose value depends on material type burnt (kJ/g).16,1 kJ/g heat of combustion for paper as cellulosic material and 45,27 kJ/g for gasoline as the catalyst burner [11].The mass rate was estimated from duration length and the material mass burnt in each experiment.Number of averaged heat release rate for each diameter can be found in Table 2.

Combustion Efficiency
Before and after mass of fuel was measured using a scale with 1 gr accuracy to know how much the mass of fuel burnt in each experiment.A study from the experiment shows that the smallest diameter of 5 cm has the most efficient combustion, although the pattern does not clearly show the relation for the other diameters.The separation distance doesn't directly impact combustion efficiency, which relates to the flame merging phenomenon.Contrary to the experiment conducted by W.Weng et al. [12], they showed that bigger separation distance and more fuel load improved the combustion efficiency.In this experiment, the more separation distance between baskets and the bigger basket's diameter, the less combustion efficiency that it gets.

Conclusion
A Series of experiments was conducted in the lab with the variation of basket's diameter (5, 10, and 15 cm) and distance between two baskets (0, 5, 10 cm).It can be concluded in these points.The temperature at 2,5 cm elevation in the middle of two baskets shows the correlation between temperature rise and the flame merging phenomenon.The bigger the separation distance, the lower the peak temperature.
The smallest basket's diameter needs the longest time to merge, but the other diameter does not show a linear correlation with the time merging.Zero separation distance between two baskets shows the highest flame in the experiment, but the lowest flame height can be acquired from 5 cm separation distance rather than 10 cm distance.
The larger diameter relates to the bigger probability of flame merging, whereas the larger diameter also correlates to bigger fuel load and heat release rate estimation.Combustion efficiency doesn't have a direct correlation with the flame merging phenomenon, where the experimental data shows scattered distribution between separation distance and combustion efficiency.

Figure 3 .
Figure 3. (a) 10 cm diameter basket with 2 cm opening area, and (b) sloped paper arrangement in the basket.

Table 1 .
Variable of Experiment

Table 2
Averaged Heat Release Rate