Heat transfer data were obtained using the transient method and correcting for radial conduction along the ceiling and losses through the insulating material. The heat transfer coefficient of the ceiling was based on the adiabatic ceiling beam temperature (recovery temperature) reached after a long time. A simple parameter for estimating the ceiling heat transfer coefficient was confirmed by the experimental results.
An understanding of this phenomenon will be of great help in evaluating rna.quantities of heat transfer in the development stages of fires in buildings. All heat transfer measurements were transient in nature and attention was focused on the first 7. However, this experimental research also set various limitations, the elimination of which will be of great help in gaining a clearer understanding of the heat transfer process in the ceiling.
EXPERIMENTAL AND ANALYTICAL MODEL A. General Description
Experiments were carried out with different types of heat sources, to find the most suitable source that would meet the following requirements: .. i) produce a stable and steady asymmetric buoyant plume; .. ii) have negligible momentum flux compared to the upward flux (to simulate a floor level fire); iii) provide a simple but reasonably accurate method for determining the heat released, i.e. the source strength; . iv) The source strength should be adjustable to produce 1 kW tol.SkW. Although this resulted in a slightly less wandering flame, the plume still appeared to be quite unstable. A further disadvantage of this type of heat source is that the alcohol must be kept at a constant level and the amount of alcohol consumed must be measured in order to calculate the source strength.
This method proved not to be very attractive due to a large amount of radiant heat flow. Stich a burner was built at almost no cost, and it had the added advantage that by just measuring the volumetric flow rate of propane and using the heating value, the source. The dimensions of the burner were determined by the source strength (Q) and the initial momentum flux (M), which had to be kept as low as possible.
PRESSURE I
INSTRUMENT A 'I' ION A. Heat Source Strength
No special instruments were needed to determine the amount of heat released by the burner. The two flow meters were calibrated for the flow of gas and air respectively, and it was necessary to ensure that during calibration the outlet pressure was as similar as possible to that in the actual experiment. Once adjusted to the regulated value, these flow rates remained constant during each experiment and no further adjustments were required during the course of the experiment.
In order to reduce errors in the measured temperatures due to radiative heat transfer and conduction along the thermocouple leads, the wire diameter had to be as small as possible. 005" iron and constantan thermocouple wires coated with a thin layer of Teflon (commonly known as a J-type joint). This provided a good electrically conductive joint with no added heat capacity due to the extra metal used in forming the joint.
In order to record these temperature measurements on a strip chart recorder, the output from the thermocouple had to be amplified. Experimental plume temperature measurements confirmed the expected fluctuating nature of the temperature in the plume, mainly due to the turbulent nature of the floating plume and the entrainment of the cooler ambient air. These temperature fluctuations varied in magnitude, depending on the position in the plume and the height.
A closer examination of these temperature fluctuations revealed the existence of very low frequency fluctuations (about 0.5 cycles/min) of relatively small amplitude, and superimposed on these were higher frequency fluctuations (about 2–4 cycles/min sec) of large amplitudes , ranging from 5 °C at the outer edge of the plume to approx. 30 - 40 °C on the center line of the plume (decreasing with an increase in height above the source). Since this part of the study was mainly about defining the plume, based on average temperatures, and also due to the fact that at these higher frequency fluctuations the thermal capacity of the thermocouple junction affects the measured temperature, a filtering circuit was built. into the amplifier circuit described above. However, it was important not to filter out the lower frequency fluctuations, as it was suspected that these could be due to large movements of the plume as a whole.
It was therefore decided to use a filter circuit that would filter out frequencies exceeding 1 cycle/s.
INPUT
IOkn
OUTPUT
The thermocouple probe is mounted on a traversing mechanism, enabling temperature measurements at any position in the plume. A time average of the recorded temperatures at each position was used as the average plume temperature (a function cf. position). purpose of obtaining the ceiling jet temperature distribution, a rake consisting of 8 thermocouple probes as described above was used (see sketch below).
Large temperature gradients near the ceiling, observed during preliminary measurements, necessitate a closer spacing of thermocouple probes in the upper case. This rake is mounted on a transverse mechanism, and by moving radially inward and outward directly under the ceiling, the ceiling beam.
8 · STAINLESS
RECORDER I
To ensure matching of recorded data, each set of thermocouples is calibrated through its own recording channel. In addition, these thermocouple leads were extended tangentially to the plate for at least a 1" distance from the junction to further reduce the effect of through-wire conduction losses. It is shown in Appendix C that the conduction effect is indeed negligible. in this case.
The gas temperature in the ceiling jet was measured with a thermocouple below each appropriate ceiling temperature measurement position (see Section II.E). These thermocouple contacts were connected to those on the ceiling in sequence so that the ceiling temperature and the corresponding gas temperature were recorded in sequence. In this way, it was possible to record these two temperatures almost simultaneously for each spot on the ceiling, at regular intervals of 44 seconds.
It should be noted that during these measurements the filter circuit of the amplifier could not be used due to the characteristic time constant associated with the circuit.
TYP. THERMOCOUPLE POSITION
SWITCH
CHART RECORDER
RESULTS AND DISCUSSION A. Plume Temperature Distribution
In developing the appropriate sealing parameters for the buoyant plume (see Section II.. C. 1) it was assumed that the plume has a straight edge, which implies a linear one. At 3 em, the maximum amplitude of the temperature fluctuations of the ceiling beam (measured at r = 14 em and y em below the ceiling) was about 15 - 20 percent of the average value, compared to 25 - 30 percent for the plume centerline. temperature fluctuations at the corresponding altitude. This reduction was as expected, due to.. i) the presence of a solid boundary (the ceiling itself); .. ii) a decrease in maximum temperature with an increase in radius; . iii) the layered nature of the ceiling jet, i.e. the hot gas is confined by the ceiling face.
These measurements were also made for two different powers of the heat source (Q = 1.17 kW and 1.53 kW). The thickness of the ceiling current, based on temperature measurements, is characterized by the Gaussian width .t. For further information about the thickness variation of the ceiling jet, it is essential to make at least some velocity measurements to determine the thickness of the velocity boundary layer (see Alpert(!)).
The results are presented in Figure 7. These graphs show the evolution of the thermal boundary layer on the ceiling. In the early stages of the experiment, up to about 176 seconds, the temperature at the wall remained almost constant, with . a very pronounced thermal boundary layer due to heat transfer to the ceiling. Thus, the equality of these two temperature parameters at r = 0 confirms the validity of the assumption in the derivation of the model for the ceiling jet, in which it was assumed that the ceiling jet starts with the maximum cloud temperature measured at the corresponding height.
The values of the air, ceiling and insulation properties used in the calculation of the heat transfer rates are presented in Table I. By considering the slope of the temperature history curve at each location on the ceiling, the increase in the internal energy of the ceiling was obtained according to Eq. Surprisingly, a maximum of only 18 - 20 percent of the heat supplied by the source was transferred to the ceiling at t = 1 min, and this value decreased to 7 - 10 percent after 7 minutes.
Although such a steady state was reached, it was found that at the positions r = 0 and r = 7 em. the ceiling temperature was approximately 5-10°C lower than the ceiling jet temperature, indicating that heat transfer to the ceiling still occurred. The non-dimensionalization of the coefficient of thermal conductivity of the ceiling is discussed in Chapter II. The main contribution of the experimental work reported in this dissertation is to present the values of the ceiling heat transfer coefficient in the axisymmetric case for the region 0 :;;; r/H:;;; 0.
SUPPORT
BURNER
FJG.5
BUOYANT PLUME CENTRELINE TEMPERATURE VARIATION
32 _____..-ADIABATIC CEILING TEMP
POSITION WHERE MAX. TEMP
TEMPERATURE: COMPARISON BETWEEN SMALL & LARGE SCALE EXPERIMENTS
THEOR. CURVES BY ALPERT I)
The strength of the heat source· can be calculated from the measured flow rates of gas and air and from the heating value of propane. Therefore, five moles of oxygen (ie 5 moles of air) are needed for each mole of propane. Assuming now that the gas and air are mixed at the same temperature (ambient temperature), then.
Assuming that the stainless steel grid covers a maximum of 30 percent of the nozzle exit area, then. The specific heat of R 11 can now be calculated by calculating the weight ratio of glass and air, and combining their specific heats. The effect of the conduction along a single thermocouple lead on the local temperature and 8T I at measurement was analyzed by McMahon(l4 .. He considered a thermocouple wire 0 perpendicular to a thin sheet G) (refer to sketch below).
Thus, calculating the second term between the square brackets of (C-6) will yield the error involved in measuring the plate temperature. To get an idea of the magnitude of this error, consider the case where the plate and the wire are made of the same material (in this case, sheet steel).
