Abstract
Fires in underground mines may cause dangerous phenomena to mining personnel. One of these phenomena is the throttle effect, reducing the mass flow. This paper investigates the nature of the throttle effect. Data was provided from fire experiments in a model-scale mine drift. With an increasing heat release rate the reduction in the mass flow will increase. An increasing ventilation velocity may initially cause a reduction of the mass flow. With a further increase of the flow velocity the mitigating effect of the forced flow will increase. A dimensional analysis resulted in an equation where the mass flow reduction could be reasonably well described. It was found that the flow velocity had a weak effect on the mass flow reduction. Nevertheless, the flow velocity influence the initiation of the reduced mass flow. A full-scale flow velocity of 3.5 m/s was found to prevent the throttle effect for typical fires underground.
Keywords
Introduction
A fire in an underground mine poses several risks to the personnel underground, where being trapped by smoke due to unforeseen smoke spread caused by the fire is one of the risks. As a fire develops underground, it will interact with the ventilation air flow and cause disturbance and changes in the flow pattern. The interaction and disturbance may lead to phenomena such as the throttling of the air flow. The throttle effect may hamper the smoke extraction and in turn the evacuation and fire and rescue operation. The throttle effect could also lead to undesired and unforeseen ventilation flow directions which could endanger the personnel underground. The throttle effect would lead to a reduction in the mass flow rate in the affected mine drift compared with prior to the fire. If a certain mass flow rate is required to achieve a desired smoke extraction or flow pattern, the power of any fan upstream of the fire will have to be increased to mitigate the effects of the throttle effect.
This paper focuses on the disturbance of the ventilation air flow during a fire in an underground mine drift, causing a reduction in the mass flow rate from a blower fan positioned upstream of the fire. When can a reduction in the flow rate be expected and how severe will it be? Being able to detect and quantify the effect of the fire on the mine ventilation system is of utmost importance with respect to the safety of the personnel underground. The experimental data were obtained from earlier performed fire experiments in a model-scale mine drift presented by Hansen and Ingason (2012) as the data from these experiments were found to fit very well the scope of the work presented here. The fuel load in these experiments was positioned in the mine drift in a similar way as for example mining vehicles.
The purpose of this paper is to investigate the nature and the extent of the flow rate reduction for fires with transient heat release rates, varying flame spread rates and flame lengths, at varying longitudinal ventilation velocities and where the ventilation flow is provided by a blower fan upstream of the fires. The paper also investigates the impact on the smoke extraction operation and potential thresholds and benchmarks to be used during pre-planning as well as during an ongoing fire intervention.
Better knowledge of the impact on the ventilation air flow would increase the knowledge on the fire behaviour and its effects in underground mines, which is desirable due to generally few studies in the field (Hansen 2015).
The relatively few earlier works on the disturbance of the flow rate have mostly been aimed at cases with steady state heat release rates, fire sources consisting of for example a gas burner or where the walls of the drift were lined with wooden slabs and where the longitudinal ventilation flow was provided by an exhaust fan. Hwang and Chaiken (1978) performed an analysis on the coupling between a fire and the ventilation airflow in a duct where the longitudinal ventilation flow was provided by an exhaust fan, relating the air velocity prior to the fire to the air velocity during the fire. The volumetric flow rate was assumed to be constant at the exhaust fan. The ratio of the intake air velocity during the fire to the intake air velocity prior to the fire would be a function of a mass injection parameter, the ratio of the gas temperature downstream of the fire to the gas temperature upstream of the fire, the duct geometry, the heat transfer coefficient and the fan-operating conditions. If the convective heat losses from the hot gas to the surrounding walls downstream of the fire were accounted for, the ratio of the intake air velocity prior the fire to the intake air velocity during the fire would decrease for a given mass injection parameter and temperature ratio of the gas downstream to the gas upstream. The given explanation of the decrease was that as the mass flow rate at the fan inlet will be higher for a larger convective heat loss, this will decrease the gas temperature and increase the gas density. As the air density upstream of the fire was assumed to be constant, a higher mass flow will imply a decrease in the air velocity. If the convective heat losses are neglected, the ratio of the intake air velocity during the fire to the intake air velocity prior to the fire was said to increase due to increasing temperatures and larger mass injection.
Lee et al. (1979) studied the interaction between duct fires and ventilation flow in terms of fire throttling effects and backlayering. Mass flow rates, velocities, pressures, and temperatures in a model tunnel network were obtained before and during fires at various ventilation air velocities. An exhaust fan was used during the experiments to provide a longitudinal ventilation flow. It was found that the fires increased the flow resistances of the ducts through fuel mass injection and high temperatures. The ventilation velocity was found to be throttled to less than half of its initial value prior to the fire. When the fan was maintained at a constant speed, the mass flow rate was found to decrease by 50% due to the throttling effect of the fire. When increasing the ventilation velocity, it was found that the throttling effect increased as the increase in ventilation velocity caused an increase in the fire growth and heat release rate.
Litton et al. (1987) performed fire experiments in an intermediate-scale fire tunnel with an exhaust fan and a large-scale gallery to determine the effects on the ventilation flow. The experimental fire in the fire tunnel was provided by a natural gas burner mounted into the floor and providing a heat release rate in the range 8.5–120 kW. The overall effects were found to be substantial, with a 10–11% maximum reduction in the total airflow and 29% reduction in the intake airflow. An estimate on the induced flow resistance was provided for larger fires, based on the measured CO2 concentrations from the fire.
During full-scale fire experiments in a tunnel Ingason and Lönnermark (2005) observed reductions in the centreline longitudinal ventilation velocity upstream of the fire compared with the corresponding ventilation velocities prior to the fire. The reduction in the ventilation velocities was attributed to the flow resistance of the fire and the thermal stack effects as the experiments were performed in an inclined tunnel.
Vaitkevicius et al. (2016) performed a CFD modelling study to demonstrate the impact of the throttle effect on tunnel fires and presenting the need for further research into the throttle effect.
In the following, the fire behaviour and mass flows during a fire in a mine drift are briefly described. Earlier performed model-scale experiments are described and the resulting data related to the reduction in flow rates are presented. Furthermore, the flow rate results are analysed and discussed with respect to the underlying mechanisms and the extent of the reduced flow.
Fire behaviour and mass flows in a mine drift
A fire in an underground mine will undergo several different phases from ignition to – finally – extinction. The appearance of the heat release rate curve of a fire will vary depending on several factors such as the type of fuel/s involved, the possible ignition of adjacent fuel items, ventilation conditions etc. Figure 1 displays an example of a heat release rate curve of a fire, with the various phases as described by ISO/TC 92 (ISO 16733 2015).
The heat release rate of a fire, with the included phases.
Ignition is followed by the incipient phase which is distinguished by a fire positioned at the site of origin, gradually getting less dependent of the ignition source but still sensitive to random fluctuations in the immediate environment. The fire during the incipient phase may either be a flaming fire or a smouldering fire. At a certain point in time, the fire may reach a stage where it becomes stable and independent of the random and natural fluctuations of the near environment. The fire at this time enters the growth phase where the heat release rate starts to distinctly increase and display less marked fluctuations. During the growth phase the fire spread relies more on the radiative heat transfer mechanism through flame radiation and less on the conductive heat transfer. The fire growth rate is highly dependent upon factors such as the fuel continuity, fuel configuration and the presence of a longitudinal ventilation. At the end of the growth phase the fire attains heat release rates at the same level or close to the maximum values. During the fully developed phase, the fire displays the highest heat release rates. The peak heat release rate may be dictated by the access of oxygen, if the fire is limited by the availability of oxygen. In mine drifts with longitudinal ventilation, the fires will generally have ample access to oxygen and the heat release rate will generally be unlimited with respect to oxygen availability. At the later stage of the fully developed phase, the access to fuel will decrease markedly as the combustion progresses. The fire at this stage enters the decay phase, which in the end will result in extinction. The decay phase is distinguished by a declining heat release rate where the decreasing access to fuel dictates the fire development.
Connected with the heat release rate curve and the fire development is the mass flow in the fire area, where the smoke spread, and smoke behaviour are visible aspects of the mass flows. A fire in a mine drift will generate high temperatures which will cause buoyancy forces, with smoke rising towards the roof. The differences in density will determine the buoyancy force. When the hot smoke encounters the roof, the smoke will be deflected, spreading along the underside of the roof. With increasing heat release rate, the buoyancy force will increase as well. The highest buoyancy effects will thus occur during the final parts of the growth phase and during the fully developed phase. With increasing mine drift height, the temperature of the smoke will decrease and the buoyancy force as well.
The longitudinal ventilation in a mine drift will also have a profound impact on the smoke behaviour; together with the heat release rate of the fire and the ensuing heat losses largely determining the occurring smoke stratification along flow direction in the mine drift. The stratification will be highest in the near vicinity of the fire, due to the distinct temperature gradient caused by the fire. With increasing distance from the fire – resulting in an increasing heat loss from the smoke – and with increasing longitudinal ventilation flow, the stratification will decrease. Eventually, a uniform temperature will be found in the mine drift with no stratification and smoke covering the mine drift from floor to roof.
The rising, hot smoke encountering the roof will be pushed along the mine drift in the direction of the ventilation flow. But some of the smoke will flow in the opposite direction with respect to the ventilation flow, resulting in backlayering. The hot smoke will eventually cool off, descend towards the floor, mix with in the intake air and reverse direction, flowing in the same direction as the ventilation flow.
In the case of a large fire in a mine drift with an inclination, the resulting heat from the fire will cause a temperature increase, resulting in a decrease in the smoke density downstream of the fire. The decrease in density will enhance the ventilation in rising drifts and cause disturbances and even reversal of the ventilation flow. The phenomenon is known as the buoyancy effect.
A fire with a considerable heat release rate causes an increase of the air masses as they pass the fire. The increase in volume, in turn, causes an additional pressure loss known as the throttle effect. The throttle effect will be noticed by a blockage in the ventilation flow at the fire site.
Backlayering, buoyancy effect and throttle effect require a certain heat release rate and will not exist or be noticeable during the incipient phase. The largest effects of these three phenomena will occur during the final part of the growth phase or during the fully developed phase.
Fire experiments in a model-scale mine drift
Fire experiments were conducted in a model-scale mine drift with longitudinal ventilation (Hansen and Ingason 2010). A total of 12 experiments was carried out using a single or multiple pile of wooden pallets as fire load. The model-scale mine drift was in scale 1:15, with a length of 10 m, a width of 0.6 m and a height of 0.4 m. The model-scale mine drift had no inclination and no buoyancy effect was therefore anticipated.
The varying parameters during the experiments were the number of wooden pallet piles (which would therefore result in varying heat release rates) and the longitudinal ventilation velocity. The following longitudinal ventilation velocities were applied during the experiments: 0.3, 0.6 and 0.9 m/s. See Figure 2 for the layout of the mine drift – long-section – and the position of thermocouples, probes and instruments. At the end of the model-scale mine drift, an exhaust duct could be found. The exhaust duct was equipped with a thermocouple and a bi-directional probe. Pile A and B were thermocouple piles with thermocouples at different heights to give a picture of the vertical temperature distribution. The first pile of pallets was always positioned at the same location, i.e. on a scale, which can be seen above the letter ‘W’ in the layout. The remaining piles of pallets were positioned at various distances from the first pile. The Schmidt-Boelter gauge is a sensor, which measures the heat flux from the fire.
Layout of the model-scale mine drift (long-section) and the position of thermocouples, probes and instruments (Ingason 2005).
The longitudinal ventilation was established using an electrical axial fan attached to the entrance of the model-scale mine drift. Thus, the conducted experiments presented a case with a fan blower provided upstream of the fire. The longitudinal velocities were obtained by adjusting a frequency regulator and a constant volumetric flow rate was provided at the fan outlet.
The wooden pallets used in the experiments were scaled down pine pallets. Each pile consisted of five individual wooden pallets. Four experiments were conducted with a single pile of wooden pallets, where experiment #2 actually consisted of four piles of wooden pallets but where only the first pile took part in the fire. In the remaining eight experiments, the fuel load consisted of four piles of wooden pallets and where all four piles took part in the fire. In the experiments with multiple piles of pallets, the piles were positioned at different distances to obtain varying ignition times of the piles and heat release rate curves with varying appearances.
The following parameters – with respect to the mass flow or fire behaviour – were either measured or calculated during the experiments:
Fire gas temperatures, using either single thermocouples predominantly along the ceiling of the mine drift or thermocouple piles (see Figure 2). Being aware that Figure 2 does not display the thermocouple in the exhaust duct. Concentrations of O2, CO and CO2, measured just downstream of thermocouple pile B at the end of the mine drift. The centreline pressure difference, which was measured with bi-directional probes. Figure 2 displays the bi-directional probe installed in the exhaust duct. The centreline flow velocity, calculated using Equation (1). The heat release rate, calculated based upon a method by Newman (1984).
The centreline flow velocity was determined using the measured pressure difference – 
[Pa] – for each bi-directional probe and the measured gas temperature. The centreline flow velocity – 
[m/s] – was calculated using the following equation (Hansen and Ingason 2010):
is a calibration coefficient (set equal to 1.08), 
is the gas temperature [K], 
is the density of the ambient air [kg/m3] and 
is the ambient air temperature [K].
The mass flow rate was calculated applying the following expression (Hansen and Ingason 2010):
is a mass flow correction factor (set equal to 0.817) and 
is the cross-sectional area of the mine drift or duct [m2].
Results from the model-scale fire experiments.
– which relates to the full scale – will be 15 in this case and the index 
– which relates to the model scale – will be 1 in this case.
List of scaling models (Ingason 2005).
Where 
is the heat release rate [kW], 
is the length [m], 
is the time [s], 
is the energy [kJ], 
is the heat of combustion [kJ/kg] and 
is the mass [kg].
Results and discussion of mass flows in fire experiments
In the calculations, the mass flow rate in the exhaust duct was calculated to obtain the mass flow rate downstream of the fire. As the blower fan was equipped with a frequency regulator, providing a constant volumetric flow rate and the air temperature at the fan was equal to the ambient air temperature throughout the experiments, the mass flow rate upstream of the fire was close to constant throughout each experiment (some minor deviations in the volumetric flow rate could be expected). The mass flow rate upstream of the fire was set equal to the average mass flow rate at the bi-directional probe upstream of the fire prior to the ignition of the fire.
The heat release rates of the fire experiments displayed a rapid fire development, with a considerable fire growth rate and with a short incipient phase. Figure 3 displays the resulting heat release rate of experiment #5.
The heat release rate of fire experiment #5.
For experiments #1 and #2, the heat release rate was not calculated using the method by Newman (1984) as backlayering occurred at the bi-directional probe upstream of the fire which resulted in erroneous results. As the pile of pallets in these two experiments was positioned on a scale, the heat release rate of these two experiments was instead calculated based on mass loss data. In these cases, the time difference between the scale results and the bi-directional probe/thermocouple results in the duct was accounted for.
The decreasing mass flow rate during a fire
Figures 4–6 display the resulting mass flow rates of experiment #1, #4 and #12. The experiments consisted of only one pile of pallets and the longitudinal ventilation velocity was varied: 0.3 m/s (experiment #1), 0.6 m/s (experiment #4) and 0.9 m/s (experiment #12). The ignition of the fire occurred after 2 min.
The mass flow rate of fire experiment #1. The mass flow rate of fire experiment #4. The mass flow rate of fire experiment #12.
As can be seen from the resulting curves, the mass flow rate starts to decrease a certain time period after ignition. In experiment #12 the mass flow rate can be seen to increase at a later stage, getting close to the mass flow rates measured prior to the fire.
If taking the average mass flow rate during the initial two minutes prior to ignition and comparing with the average mass flow rate value following upon ignition, the decrease in percentage is 2.1% in experiment #1, 3.8% in experiment #4 and 4.5% in experiment #12. But to draw the conclusion that the throttle effect will increase solely due to an increasing ventilation velocity is incorrect as the maximum heat release rate did not stay constant but varied in the three experiments. As can be seen in Table 1, the maximum heat release rate increased with increasing ventilation velocity due to more rapid flame spread along the fuel surfaces and a higher fire growth rate. Instead – if keeping everything else constant – an increasing ventilation velocity should decrease the throttle effect as it is directed in a perpendicular direction compared with the buoyancy flow. But with an increasing heat release rate, the buoyancy forces will increase, and the buoyancy forces will possibly have a larger impact on the resulting throttle effect compared with the amplitude of the longitudinal ventilation velocity.
What will happen if the ventilation velocity is kept constant but the heat release rate of the fire is varied? Figure 7 displays the mass flow rate of experiment #5 with a longitudinal ventilation velocity of 0.6 m/s where the fire load consisted of four piles of wooden pallets resulting in a maximum heat release rate of 467 kW. When comparing experiment #5 with experiment #4 (Figure 5) with a maximum heat release rate of 154 kW, it is clear the throttle effect increases with increasing heat release rate. The decrease in the average mass flow rate of experiment #5 – comparing with the average mass flow rate before ignition – was 12% and therefore significantly higher than experiment #4. The increase in heat release rate will increase the buoyancy force and the throttle effect which is seen in the increase in mass flow rate reduction between experiment #4 and #5.
The mass flow rate of fire experiment #5.
Lee et al. (1979) claim that the throttle effect will be higher with higher ventilation velocities as the ventilation flow intensifies the fire. This is partially true as the heat release rate increases with increasing ventilation flow and result in an increased throttle effect, but only up to a certain level. Eventually, the heat release rate may start to level off and even decrease due to the blow-off effect, despite an increasing ventilation velocity. The blow-off effect causes a slower flame spread along the fuel surfaces. Therefore, at higher ventilation velocities the throttle effect may very well start decreasing with increasing air flow and declining heat release rate.
What is the explanation of the reduced mass flow rate in the experiments? Assuming a compressible flow along the mine drift, the following expression for the mass flow rate upstream and downstream will be valid:
is the density of the gas fluid [kg/m3].
Calculating the change in the velocity and density respectively, the results of experiment #1 during the time period when the reduced mass flow rate was distinct can be seen in Figure 8. The decrease in the gas density is considerably larger than the increase in the flow velocity, therefore resulting in a reduced mass flow rate as can be seen from Equation (9). Furthermore, as the mass flow decreases the volume flow exiting the mine drift increases, which is in line with the findings of McPherson (1993).
The change in velocity and density of fire experiment #1.
The reduction in the downstream mass flow rate could be compared with the compressible, isentropic and subsonic mass flow through a convergent-divergent nozzle in aerodynamics (Anderson 2001). As the exit pressure of the mass flow through the nozzle is decreased, the flow velocity in the throat increases and the gas density decreases. But the downstream mass flow rate increases as the increase of the flow velocity in percentage is larger than the decrease in gas density. Thus, leading to a direct opposite result compared with the throttled mass flow in the mine drift.
Influence of mass injection
Hwang and Chaiken (1978) list a mass injection parameter as one of the influencing parameters on the ratio between the air velocity during the fire to the air velocity prior to the fire. The mass injection parameter is defined as the ratio between the mass flow from the fire to the mass flow from the intake. During the combustion the fire contributes to the total mass flow through the mass loss of the fire, where the emitted mass flows along the mine drift and is part of the mass flow downstream of the fire. The question is how much will the mass loss of the fire contribute to the total mass flow along the mine drift? Assuming an ideal gas law applies and using the measured values from the gas analysis downstream of the fire, the added mass flow of carbon in the CO and CO2 molecules into the air flow was calculated. It was found that an average of approximately 2% of the total mass flow downstream originated from the mass loss from the fire. Thus, the mass flow injected from the fire will only be a small fraction.
Calculating the mass injection from the fire during full-scale fire experiments in a mine drift resulted in similar mass fractions. Full-scale fire experiments on a drilling rig and a loader were conducted in a mine drift in an underground mine (Hansen and Ingason 2013). As opposed to the model-scale experiments where the fuel load consisted only of wooden pallets, the fuel load during the full-scale experiments consisted of several types of fuel such as tyres, hoses, cables, diesel and hydraulic oil. The average mass flow fraction from the fire was less than 1% in the case of the drilling rig experiment and approximately 3% in the case of the loader. A large part of the mass loss during the loader experiment occurred during the later parts of the smouldering decay phase.
Even though the mass injection from the fire will not be a dominating parameter, the mass loss of the fire will still be accounted for through the heat release rate of the fire in this paper. The heat release rate as a function of the mass loss rate using the following equation:
is the mass loss rate of the fire [kg/s] and 
is the effective heat of combustion [kJ/kg].
The fire load in the model-scale experiments was solely wooden pallets throughout all experiments. Therefore, the effective heat of combustion stayed constant throughout the experiments, as the effective heat of combustion varies depending on the type of fuel (Tewarson 2008). Any future experiments should be conducted where a composition of various types of fuel loads are used, further analysing the influence on the mass injection (and thus also the mass flow rate reduction). The effective heat of combustion will also depend on the combustion efficiency of the fuel load. Any future experiments should also investigate the possible impact of a varying combustion efficiency on the mass flow rate reduction.
Dimensional analysis
Performing a dimensional analysis on the reduction of the mass flow rate along the model-scale mine drift, the following governing parameters of the fire as well as the surrounding conditions were included based upon earlier studies where the pressure drop across the fire area, the change in centreline flow velocity or the fire behaviour in a mine drift was analysed (Hwang and Chaiken 1978; Lee et al. 1979; Litton et al. 1987), (Hansen 2019): 
, 
, 
, 
, 
, 
, 
, 
is the mass flow downstream of the fire [kg/s], 
is the centreline flow velocity upstream of the fire [m/s], 
is the hydraulic diameter of the mine drift [m], 
is the gas density upstream of the fire [kg/m3], 
is the mass flow rate upstream of the fire [kg/s], 
is the specific heat [kJ/kg·K] and 
is the gravitational constant, which was set to 9.81 m/s2.
Hwang and Chaiken (1978) lists the downstream gas temperature as a governing parameter but given that the downstream gas temperature will be a function of the heat release rate and the upstream centreline flow velocity, the gas temperature was omitted in the ensuing analysis. Hwang and Chaiken (1978) lists a mass injection parameter as a governing parameter but as mentioned in Section ‘Influence of mass injection’, the mass loss rate of the fire correlates with the heat release rate of the fire. The hydraulic diameter of the mine drift was selected as the length parameter when considering the influence of mine drift geometry. The convective heat transfer coefficient is also listed by Hwang and Chaiken (1978) as a governing parameter. Given that the heat transfer coefficient is a function of the flow velocity, the hydraulic diameter, the gas density and the dynamic viscosity of the gas, the parameter was omitted as the dynamic viscosity in turn is a function of the gas temperature. The heat transfer coefficient is also a function of the thermal conductivity of the solid surface, but as the type of solid surface remained the same through the experiments it was also omitted in the analysis. Nevertheless, future experiments where the surface material is varied would be of interest.
The normalized mass flow rate:
Continuing with the following normalized mass flow rate:
Setting a dimensionless heat release rate of Equation (13) equal to:
Figure 9 displays the mass flow rate ratio of experiment #1 as a function of the dimensionless heat release rate. Initially the mass flow rate ratio starts to decrease and eventually levelling off with increasing dimensionless heat release rate. At the transition from the growth phase to the fully developed phase, the mass flow rate ratio displays a loop motion as can be seen for the higher dimensionless heat release rate values in Figure 9. During the decay phase, the mass flow rate ratio starts to increase (with decreasing dimensionless heat release rate) and work itself up towards the values at the time of ignition. Due to the meandering type of curve, it was decided to focus on the incipient phase and growth phase of the different fire experiments. The incipient phase and the growth phase are generally the phases of most interest as they will encompass the evacuation, possibly the initial fire suppression and rescue operations and the initiation of the smoke removal operation. The incipient phase and the growth phase will also cover the transition from the maximum values of the mass flow rate ratio to the minimum values. Figure 10 displays the mass flow rate ratio of experiment #1 for the incipient phase and the growth phase.
The mass flow rate ratio of experiment #1 as a function of the dimensionless heat release rate. The mass flow rate ratio of experiment #1 as a function of the dimensionless heat release rate, encompassing the incipient phase and growth phase.
When proceeding with the work on a correlation, Equation (12) was used instead of Equation (13) to investigate the influence of the centreline flow velocity separately from the heat release rate. A correlation between the ratio of the two mass flow rates and the dimensionless group containing the heat release rate of Equation (12) was studied. Using an exponential function to fit the calculated values:
and 
are coefficients. Resulting in the following expression:
Continuing the analysis with the second dimensionless group found in Equation (12) containing the centreline flow velocity. During the analysis it was found that the factor:
The following correlation finally resulted from the regression analysis:
The measured mass flow rate ratio for the 12 fire experiments was plotted in Figure 11 as a function of the corresponding mass flow rate ratios calculated by Equation (17). An The scatter plot of the experimental mass flow rate ratio versus the calculated.
value of 0.81 resulted from the regression analysis. The plot in Figure 11 shows that the mass flow rate ratio can be reasonably well described by Equation (17). A total of 1582 data points is included in Figure 11. Values of experiment #12 with lower heat release rates were not included in the scatter plot as the mass flow rate ratio did not start to drop off until a certain threshold was exceeded. This is further analysed and discussed in Section ‘Initiation of the throttle effect’.
Initiation of the throttle effect
Especially experiment #12 – with a centreline flow velocity of 0.9 m/s – displayed a delayed initiation of the decline in the mass flow rate ratio as can be seen in Figure 12. Figure 12 displays the mass flow rate ratio as a function of the dimensionless group containing the heat release rate of Equation (12). After an initial oscillation, the mass flow rate ratio stays constant at a unity value despite an increasing heat release rate. This appearance of the curve distinctly differs from the results of the other experiments where the mass flow rate ratio started its decline more or less at the ignition of the fire. The high centreline flow velocity of experiment #12 obviously counteracts the buoyancy force of the fire and stalls the throttle effect. The question here is whether a threshold criterion can be established, pinpointing the initiation of the throttle effect at higher centreline flow velocities? Lee et al. (1979) presented the following Froude number:
The mass flow rate ratio of experiment #12 as a function of the dimensionless heat release rate of Equation (12).
is the height of the mine drift [m] and 
is the density of the hot fire gases in the fire zone [kg/m3].
The density of the hot fire gases can be calculated using the following expressions where the convective part of the heat release rate was assumed at 2/3 (based on the findings from earlier full-scale fire experiments in a mine (Ingason et al. 1994) where wood cribs were one of the involved fuel types):
is the bulk average temperature increase across the fire zone [K].
The Froude number contains the ratio of the buoyancy head of the fire to the kinetic head of the longitudinal ventilation flow and thus is highly applicable to a throttle threshold. By applying the values of the different experiments where the mass flow rate ratio initiated its decline, an average value of the Froude number of 0.8 was calculated. The throttle effect would occur for Froude numbers exceeding this threshold. This average Froude number could serve as a possible threshold criterion for the initiation of the throttle effect. Still, difficulties were encountered when trying to determine the point when the decline started as there were generally some oscillations at the early stages of the experiments. Given the uncertainties and the limited number of experiments with higher flow velocities, additional experiments and more research into this field is needed.
The plateau type of curve found in Figure 12 also indicates the measure to counter the throttle effect, i.e. by increasing the intake velocity of the ventilation flow. By increasing the flow velocity, the throttle effect will initiate at higher heat release rates.
The dimensional analysis indicated that the centreline flow velocity had a weak effect on the mass flow rate ratio. But for cases below the threshold limit the centreline flow velocity will have a decisive effect on the mass flow rate ratio. Above the threshold, the centreline flow velocity will be overtaken by the heat release rate. Be aware that the plateau values of experiment #12 were not included in the dimensional analysis due to the delayed decline of the mass flow rate ratio.
Litton et al. (1987) proposed the following criterion based on a modified Froude number as a threshold for a throttling effect:
is the initial average flow velocity upstream of the fire.
The throttle threshold of Equation (21) indicates when the reduction in ventilation velocity was approximately constant and independent of fire size. One must bear in mind that the experiments conducted by Litton et al. (1987) were based on an exhaust fan where the ventilation flow velocity at the exhaust was pre-set and the changes in the intake velocity were measured during the experiment. The experiments by Litton et al. (1987) resulted in reduced ventilation velocities upstream of the fire, while the model-scale experiments of this paper resulted in reduced mass flow rate downstream of the fire. Thus, the conditions and flow results differed between the two sets of experiments.
When assuming that the initial average flow velocity could be set equal to the initial centreline flow velocity, the calculated threshold of the different experiments was not found to fit the threshold of Equation (21). When applying the measured average fire gas temperatures at thermocouple pile A for the experiments involving a single pile of pallets and thermocouple pile B involving multiple piles (see Figure 2), the resulting values still did not fit the threshold value of Equation (21). Besides the different conditions and results of the two sets of experiments, there may also be some uncertainties using Equation (19) when calculating the bulk average temperature increase as opposed to the measured values of Litton et al. (1987). In addition, the thermocouples were positioned differently, and the fire gas temperatures were measured at different distances from the fire zone in the two sets of experiments. Thus, more research is required with respect to throttle thresholds.
Equivalent full-scale values
The longitudinal ventilation velocities of the model-scale experiments corresponded to the following full-scale velocities: 1.16 m/s (0.3 m/s), 2.32 m/s (0.6 m/s) and 3.49 m/s (0.9 m/s). The model-scale heat release rate of experiment #12 at the time when the throttle effect initiated corresponded to a full-scale heat release rate of approximately 56 MW. A 56 MW fire represents quite a severe fire and most likely involving more than one larger mining vehicle. The corresponding full-scale heat release rate threshold of experiments #1 and #2 would be approximately 9 MW and for the cases with a longitudinal flow velocity of 0.6 m/s (2.32 m/s full-scale) it would be approximately 16 MW. The 16 MW fire would be close to the peak heat release rate of a loader fire (Hansen and Ingason 2013). As noted, during the jump from 2.32 to 3.49 m/s (full-scale) the throttle threshold is increased more than three times. A longitudinal flow velocity of 3.5 m/s (the 3.49 m/s value rounded to 3.5 m/s) would prevent the throttle effect from occurring for fires with typical maximum heat release rates involving a single large mining vehicle (Hansen and Ingason 2013). The longitudinal ventilation velocity of 3.5 m/s can also be compared with the typical upper limit for backlayering of 3 m/s below which no backlayering occurs (Ingason et al. 2016). The upper limit for backlayering thus coincides with the threshold of the throttle effect for fires typically found underground.
Application during a fire
During a fire in a mine drift, the throttle effect should be mitigated or minimized by either an already existing longitudinal ventilation flow velocity or a flow velocity which is ramped up to successfully counter the increasing buoyancy force of the fire. The heat release rate of the fire and thus also the buoyancy force may at least initially increase due to the increasing flow velocity. Eventually the peak heat release rate will be attained and with an increasing flow velocity the mitigating effect on the mass flow rate ratio will increase.
The throttle effect may also affect the mitigation of backlayering during a fire. As the throttle effect will result in a reduced mass flow rate, the ventilation flow velocity may have to be ramped up accordingly to prevent the backlayering. The interaction between the throttle effect and the backlayering effect will have to be investigated further.
The model-scale mine drift in the experiments had no inclination and the so-called buoyancy effect would therefore not occur. The findings were therefore applicable to a horizontal mine drift and any interaction between the throttle effect and a buoyancy effect will have to be looked into further.
The upstream flow velocity during the conducted fire experiments was constant throughout each experiment. During a fire in an underground mine drift the ventilation flow velocity may have to be ramped up to counter the increasing heat release rate of the fire. The question is how this transient ventilation flow and a transient heat release rate will affect the resulting mass flow rate ratio? Further experiments are needed with respect to this issue.
Conclusions
The mass flow during a fire in a mine drift with longitudinal ventilation flow was analysed, applying data from fire experiments in a model-scale mine drift. The focus of the study was foremost to investigate the nature and the extent of the flow rate reduction for fires with transient heat release rates and at varying longitudinal ventilation velocities and where the ventilation flow is provided by a blower fan upstream of the fires. It was found that:
With an increasing heat release rate the buoyancy force of the fire and the reduction in the mass flow rate increases as well. An increasing longitudinal ventilation velocity may initially cause an increasing heat release rate and reduction of the mass flow rate. Eventually the peak heat release rate will be attained and with a further increase of the flow velocity the mitigating effect of the forced flow on the mass flow rate ratio increases. The reduced mass flow rate downstream in the fire experiments can be explained by a larger decrease in the gas density compared with the increase in the flow velocity, resulting in a reduced mass flow rate. Automatically following upon this, the volume flow exiting the mine drift increases. A dimensional analysis resulted in an equation where the mass flow rate reduction during the incipient phase and the growth phase could be reasonably well described. It was found that the centreline flow velocity had a weak effect on the mass flow rate reduction compared to the heat release rate. At higher centreline flow velocities, the initiation of the reduced mass flow rate was clearly delayed. A potential threshold for the initiation of the reduced mass flow rate was investigated, using a Froude number. A full-scale longitudinal flow velocity of 3.5 m/s was found to prevent the reduction in mass flow rate from occurring for fires involving a single large mining vehicle.
The findings of the analysis increase the knowledge of the throttle effect's impact on the ventilation air flow. This knowledge would aid during the pre-planning of the ventilation system or during an actual fire underground, mitigating or minimizing the throttle effect.
Footnotes
Acknowledgment
The author would like to thank and acknowledge the support from the Sustainable Minerals Institute, The University of Queensland.
Disclosure statement
No potential conflict of interest was reported by the author(s).