the relationship between ventilation conditions and ... - CiteSeerX

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

THE RELATIONSHIP BETWEEN VENTILATION CONDITIONS AND CARBON MONOXIDE SOURCE TERM IN FULLY-DEVELOPED COMPARTMENT FIRES Burkhard Forell & Dietmar Hosser Braunschweig University of Technology, Division of Fire Protection, Braunschweig, Germany

ABSTRACT Main species yields of compartment fires can be correlated with the ventilation conditions of the compartment, a well known approach referred to as global equivalence ratio concept (GER-concept). From data of experiments in hoods and in a prototype compartment, Gottuk and Lattimer (SFPE 2002) have presented their idealised correlations of CO yields with the GER. The comparison of re-examined and amended data of ISO 9705 compartments shows reasonable agreement with these correlations; also the described temperature effect can be seen. An important extension of the GER-concept concerns upper layer (wood) pyrolysis as it provides an additional CO source also for well-ventilated conditions. A further extension of the GER-concept is made by the reduction of CO yields by external combustion either by flame extensions or by external burning from under-ventilated conditions. The occurrence of flame extensions can be assessed by CFD fire models; for external burning from under-ventilated conditions Beyler’s ignition index is examined. His simple correlation of a GER to allow for external combustion with the upper layer temperature is extended by the combustion efficiency inside the primary compartment and the overall combustion efficiency including external combustion. The results for the ignition criterion fit much better to given experimental data. The efficiency of external combustion to reduce CO yields depends on the amount of entrained oxygen and thus is a function of the exhaust vent configuration and the gas mixture in front of the vent.

INTRODUCTION By far the most fire deaths in industrialised countries occur from toxic fire effluents. From the large amount of toxic combustion products, carbon monoxide (CO) has shown to account for the largest part of the overall acute smoke toxicity [1]. Depending on the nitrogen content in the fuel, CO is followed by hydrogen cyanide (HCN). From US fire statistics it has been revealed that about 70 % of fire deaths in buildings occurred in fully-developed fires. Regarding the location of the victims, about 80 % of them died in locations adjacent to the room of fire origin. In 14 % of these cases the fire did not spread beyond the room of origin [2]. The composition of fire effluents of fully-developed compartment fires strongly depends on the ventilation conditions, whereas the chemical fuel structure is of minor importance [1]. The state of the art in describing the species yields of fully developed compartment fires is set by the “global equivalence ratio concept” (GER-concept) – a term first used by Pitts 1994 going back to systematic work of Beyler in the middle of the nineteen-eighties. The GER is defined as the actual ratio of fuel mass-flow to air mass-flow divided by the respective ratio under stoichiometric conditions &f m & air m (1) Φ= ⎛m ⎞ &f ⎜ m & air ⎟⎠ stoic ⎝

The ratio

( m& f / m& air )stoic is

a fuel-dependent constant whose reciprocal is commonly termed the

stoichiometric air/fuel mass ratio rair, or which is expressed by the stoichiometric oxygen/fuel mass ratio, rO2 825

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007 ∞ & air ⎞ rair,f = ⎛⎜ m = rO2,f / y O2 ≈ rO2,f / 0.233 & f ⎟⎠ stoic m ⎝

(2)

A GER smaller than 1.0 represents over-ventilated, fuel-lean conditions, a GER of 1.0 corresponds to stoichiometric conditions, and a GER exceeding 1.0 represents under-ventilated, fuel-rich conditions. The term “vitiated” is used independently of the equivalence ratio for atmospheres with depleted oxygen concentrations. In this work, the suitability of the GER-concept to assess CO yields is scrutinised and extensions for the concept are given.

REVIEW OF EXPERIMENTAL WORK ON THE GER-CONCEPT IN SCALED COMPARTMENTS The first widely referenced work to correlate main species production and yields of diffusion flames with the ventilation of an exhaust hood was published by Beyler 1983 [3]. He varied the fuel supply and the distance between the burner and the layer interface to control the entrainment rates and consequently the GER of the hood. By using different modifications of gas burners he found out that the species formation is insensitive of the flame structure but it correlates to the fuel-to-air ratio of the plume that equals the GER for steady state conditions. He also found out that •

for low (Φ < 0.5) and high (Φ > 1.4) GERs the concentrations of CO are quite insensitive to GER variation,

•

the generation of CO under fuel-rich conditions is considerably greater than for fuel-lean conditions, and that

•

under fuel-lean conditions the CO production increases according to aromatics, hydrocarbons (HC), and oxygenated HC - this ranking is reversed under under-ventilated conditions.

Additionally he developed a methodology to assess burning at the interface of the hood layer and the ambient air by introducing an ignition index [4][5]. Zukoski and co-workers also performed hood experiments with different parameter studies. They compared data from test series with different burners that resulted in different interface heights zi, layer temperatures (550 K < Tul < 850 K), and residence times tres, and showed that these parameters are either of minor importance compared to the GER or that the effects on the CO and O2 mole fractions from the coupled variation balance each other out (Fig. 1). In additional experiments Morehart progressively increased the thickness of the hood insulation material to investigate the temperature dependence of the species concentration. He compiled data from experiments of nearly stoichiometric ventilation (Φ ≈ 1.04) and under-ventilated experiments (Φ ≈ 1.45) with the hood temperature varying between 500 K and 870 K. For both ventilation conditions, due to increased temperatures the combustion efficiency increased together with increased fuel (CH4) and oxygen depletion, while the yield of CO only slightly decreased for the nearly stoichiometric case and was indistinct for the under-ventilated case. The clear increase in CO yields with increased temperatures, as found by chemical equilibrium calculations, is not supported by experimental results (Fig. 1). Morehart’s hood also allowed for the injection of additional air into the hood atmosphere, well remote from the plume, to decrease the hood equivalence ratio compared to the plume equivalence ratio. Morehart’s results were reported as mass fractions yi [6] from which species yields were calculated [7]. The normalised yields (Fig. 2) from natural gas correlate well with the hood equivalence ratio and were apparently not influenced by the plume equivalence ratio. For comparison the figure also depicts Toner’s data that was obtained from a smaller hood at higher temperatures. For over-ventilated fires the combustion was consequently more complete than in Morehart’s experiments, while at Φ > 1 the

826



0.14

0.030

0.020

0.12

0.10

CO-Ynorm

Mole fraction CO

800 K

zi ------Tul -----tres -------0.23m 850K 30s 0.10m 700K 60s 0.07m 650K 100s 0.01m 550K 190s Chemical Equilibrium

0.025

0.015

0.010

750 K

650 K 0.5

1.0

1.5

2.0

0.0

3.0

Equivalence ratio

Fig. 1. Mole fractions XCO observed in hood experiments as a function of the equivalence ratio. The solid lines represent the chemical equilibrium concentrations for the temperatures indicated (reproduced from [8]).

10 cm 10 cm 10 cm

0.91 0.50

23 cm 23 cm

Open dots: no air addition

0.00 2.5

1.62 1.09 0.81

2.83 5 cm 1.46 5 cm Toner data

0.02

700 K

0.0

φplume Zinterface 2.17 10 cm

0.06

0.04

0.005

0.000

0.08

0.5

1.0 1.5 2.0 Hood equivalence ratio

2.5

3.0

Fig. 2. Normalised yields of CO observed in Morehart’s hood experiments as a function of the hood equivalence ratio (open dots: without air addition, filled dots: with air addition (Φpl > Φhood). Corresponding results of Toner are added for comparison (calculated from data of [6]).

formation of CO increased due to elevated temperatures. The normalised CO yields from natural gas correlate well with the hood equivalence ratio and are apparently not influenced by the plume equivalence ratio. A large series of room fire experiments was performed at the Virginia Polytechnic Institute and State University with a steel-constructed insulated prototype compartment of 2.3 m3 volume (Fig. 3). The compartment had separated inflow and outflow vents, where the inflow was naturally through a duct with an air distribution plenum situated below the fuel pan. The compartment was partly equipped with an adjacent hallway. In their engineering methodology Gottuk and Lattimer 2002 [9] distinguished between external combustion by “flame extensions” (Fig. 4a) and “external burning from underventilated conditions” (Fig. 4b). “Flame extensions” occur when the fire plume or the resulting ceiling jet is too long to fit in the compartment. This can happen under both well-ventilated and under-

Fig. 3. Schematic of the prototype compartment with Fig. 4a+b. Illustration of external combustion adjacent hallway (adopted from [9]). as (a) flame extensions and as (b) external burning from under-ventilated conditions (adopted from [9]).

827



ventilated conditions. The engineering methodology only accounts for “external burning from underventilated conditions” which they assumed to occur at Φ ≈ 1.6 or which may be assessed by Beyler’s ignition index. In case of external combustion from under-ventilated conditions, the control volume for the GER has to be extended to the external flame tip. For the correlation of CO yields with the GER, they proposed two equations depending on the upper layer temperature

YCO = (0.19 /180) ⋅ arctan(x) + 0.095 ___ for − Tul < 800 ⋅ K

(3)

where x = 10(Φ – 0.8) and arcus tangent in degrees and YCO = (0.22 /180) ⋅ arctan(x) + 0.110 ___ for − Tul > 900 ⋅ K

(4)

where x = 10(Φ – 1.25) and arcus tangent in degrees. Eq. (3) represents the curve fit of Beyler’s hood experiments for hexane. Eq. (4) is an approximate fit to hexane data from Gottuk’s fires [10] in the prototype compartment (Fig. 5). For an upper layer temperature of Tul > 900 K it was shown that the upper layer becomes reactive and CO is oxidised to CO2. Therefore the area of the GER where YCO increases is shifted to higher GERs for Tul > 900 K.

Fig. 5. CO yields according to the recommended Eqs. (3) & (4) depending on the upper layer temperature. For comparison the experimental data of Beyler and Gottuk is given (reproduced from [9]).

EXPERIMENTS IN THE FULL-SCALE ISO 9705 ROOM

Lönnermark, Blomqvist et al. [11]-[13] performed large-scale experiments in the ISO 9705 room within the TOXFIRE-project. The room (l * w * h = 3.6 m * 2.4 m * 2.4 m), made of lightweight concrete, was insulated in its upper half by 50 mm high density mineral wool and was equipped with a constant 0.8 m wide opening vent and constant soffit height of 2.0 mm while the vent height H was altered to 0.89 m, 0.68 m, 0.56 m, and 0.45 m. Different solid and liquid fuels were burned in pans of different size from 0.5 m2 to 1.4 m2. Extensive gas samplings were performed in the opening vent and in the duct of the gas collecting hood (Fig. 6), from which species yields were calculated by the related mass flows. For each test one or more periods of quasi-steady states were defined to correlate the actual GER with species yields. In most cases the GER was determined directly with a phi-meter [14] as the local equivalence ratio from representative sampling in the opening (Fig. 6b). Alternatively, mass flows could be calculated from thermocouple measurements. 828



Duct gas sampling

Gas sampling probe

Thermocouples

wV

Phi-meter probe H Soot sampling probe Opening

Load cell Fig. 6a+b. Schematic of the ISO 9705 compartment with exhaust duct (a) and instrumentation in front of the opening (b) (reproduced from [14]).

The correlations of CO yields with the GERs were presented in [11][12][15] with the GERs determined by the phi-meter. However, the recalculation of the mass flows from phi-meter readings and reported mass loss rates (cf. Eq. (1)) showed that the compartment vent flows would have been considerably higher then commonly assumed by the KAWAGOE equation [16] & air ≈ 0.52 ⋅ w v ⋅ H m

3/ 2

(5)

Hence, the vent mass flows were recalculated based on the reported [11] upper layer temperatures Tul, ambient temperature Ta = 293 K, and orifice coefficient CV = 0.68 by [7] & air = w v ⋅ H m

3/ 2

1 − Ta 2 T ul 2 ⋅ g ⋅ ρa ⋅ ⋅ Cv ⋅ 3 3 1 ⎞ ⎛ 3 T ⎜1 + ul ⎟ ⎜ ⎟ Ta ⎝ ⎠

(

)

(6)

By taking into account the temperature effects the results by Eq. (6) are somewhat more exact than by Eq. (5). With the amended vent flows the GERs were recalculated to establish an amended GER to YCO correlation. Additional experiments in the ISO 9705 room were performed at the iBMB by Hosser et al. [17]. At constant vent height of 2.0 m the width was altered between 0.04 m and 0.80 m. Gas sampling was performed in the exhaust duct. For the tests with three different fuels, polyethylene (PE, 25 kg, 0.5 m2 pool), glycol (22.5 kg, 1.0 m2 pool) and spruce (35 kg, crib), the initial fuel weight and the pool area or crib size respectively were kept the same. Only the opening width wV was altered at each test. For evaluation all measurements were time-averaged that only one data set could be derived from one test. The averaging interval was between 10 % and 80 % mass loss. As the compartment was not insulated, lower temperatures were achieved than in the TOXFIRE project. Calculation of the vent flows from thermocouple tree measurements based on the method by Janssens and Tran [18] gave correlations for the CO yield with the GER [7]. In Fig. 7 the idealised correlations of Gottuk and Lattimer are compared with the ISO 9705 room data from the TOXFIRE (amended) and iBMB project. For the TOXFIRE data the fuel type and number of the experiment, the vent height H and the GER when flames came out of the compartment (FO) are given. In order to avoid influences by external combustion, the TOXFIRE data from opening measurements (cf. Fig. 6) was taken when possible (open dots for Nylon 66 (Ny), chlorobenzene, CB).

829



0.35

PP3 H=0.890m, FO at GER=1.04 PP4 H=0.560m, FO at GER=1.89

0.30


0.25

YCO [g/g]

Ny2 H=0.890m, FO not occured Ny3 H=0.560m, FO at GER=1.98

0.20

Ny4 H=0.680m, FO at GER=0.93 0.15

Ny5 H=0.450m, FO at GER=1.73 CB6 H=0.680m, FO at GER=1.92

0.10

PE, iBMB spruce, iBMB

0.05

glycol, iBMB Temp UL < 800 K

0.00 0.0

0.5

1.0

1.5

2.0

2.5

Temp UL > 900 K

3.0

GER, Φ [ ]

Fig. 7. CO yields correlated with the GER. Comparison of the temperature-dependent idealised Eqs. (3) & (4) [9] with the newly re-examined ISO 9705 room data from the TOXFIRE test series (open dots: opening yields, closed dots: duct yields, data of one test fire is connected) and iBMB test series (different test fires with the same fuel are connected by a dotted line). 1400


1300

PP5 H=0.680m, FO at GER=1.20

1200

Tul [K]

PP6 H=0.450m, FO at GER=1.42 1100

Ny2 H=0.890m, FO not occured

1000

Ny3 H=0.560m, FO at GER=1.98 Ny4 H=0.680m, FO at GER=0.93

900

Ny5 H=0.450m, FO at GER=1.73

800

CB6 H=0.680m, FO at GER=1.92

700

PE, iBMB spruce, iBMB

600

glycol, iBMB

500 0.0

0.5

1.0

1.5

2.0

2.5

3.0

GER, Φ [ ]

Fig. 8. Upper layer temperatures correlated with the GER. ISO 9705 compartment data compiled together from the TOXFIRE test series (open dots: opening yields, closed dots: duct yields, data of one test fire is connected) and iBMB test series (different test fires with the same fuel are connected by a dotted line).

When no reliable opening data was available, the duct data (closed dots for polypropylene (PP)) was taken. External combustion was not reported for the iBMB experiments. Sampling was performed only in the duct. Concerning the TOXFIRE data for PP, the increase from low CO yields at over-ventilated conditions to higher CO yields at under-ventilated conditions is smoother than given by the idealised equations. This seems to be partly an effect of the temperature increase with increased GER (Fig. 8). For nylon the temperatures are always above 900 K and the yields are conservatively described by the high temperature Eq. (4). As the carbon content of nylon is only 63 %, the unnormalised CO yields are always relatively low. For CB high CO yields can be explained by the chemical structure, since chlorine acts as a flame inhibitor in the gaseous phase, where burnout of CO is reduced [19]. Concerning the iBMB data, the early increase of CO yields from PE could again be an effect of the low upper layer temperatures. However, although the upper layer temperatures were below 800 K for spruce and glycol, the increase of the CO yields can be conservatively described by the high temperature Eq. (4). For these fuels, for over-ventilated fires low CO yields are known [20].

830



It must be pointed out that Gottuk and Lattimer’s Eqs. (3) & (4) were established using data from hood experiments and a prototype compartment, while the ISO 9705 compartment was 9 times larger in volume than the prototype compartment and was equipped with window-style (TOXFIRE) or door-style (iBMB) vent. Even so, it can be concluded that the agreement is reasonable. For PP (TOXFIRE) and PE (iBMB) some elevated CO yields are given at Φ ≈ 0.5 for which other effects than the GER were examined. THE EFFECT OF VITIATION AND WOOD PYROLYSIS ON THE CO SOURCE TERM

In contrast to the flows dynamics in hood experiments or in the prototype compartment, the inflow and outflow in real compartments might go through the same opening. In this case shear mixing occurs at the vent and exhaust gases will be fed back into the lower layer. Depending on the vent geometry considerable backflows into the lower layer can be achieved [21]. Additionally net downward wall flows may vitiate the lower layer of compartment fires. The influence of reduced oxygen concentration on the CO yields has been experimentally examined in different apparati with different fuels [22][23]. They consistently show that a significant increase in the CO yields only occurs when the oxygen concentration is reduced to levels close to extinction. Hence, no significant influence on the CO yieldGER-correlation can be expected from effects of vitiation. While vitiation has shown to be of minor importance for CO production, a significant additional CO source term has been observed when combustible lining material was mounted in the upper layer and was pyrolysed by a fire source that was located at the bottom of the compartment. Pitts et al. [24] conducted natural gas fires between 40 and 600 kW ideal heat release rate in a reduced-scale compartment (l * w * h = 1.46 m * 0.98 m * 0.98 m, wv = 0.81 m, H = 0.48 m), where the ceiling and the upper walls were lined with 6.4 mm Douglas fir plywood for the most tests. When the compartment’s surfaces where covered with plywood, a significant increase of the CO levels in the upper layer was shown. The increase correlated with the heat release rates of the gas burner. The pyrolysis and subsequent combustion of the wood linings led to significantly elevated overall heat release rates compared to the gas burner. The heat release rates were also monotonically increasing over & = 1) in time until wood burnout was reached. For the compartment Pitts et al. originally assumed Q(Φ the order of 200 kW [24] and concluded that an extra CO source term would be relevant only for underventilated fires [25]. However, as can be found from Eq. (1) and (5) with ΔHC,O2 = 12.5 MJ/kg for natural gas, the heat release rate to achieve Φ = 1 is above 400 kW for the compartment vent. Based on this value a significant increase in the CO concentrations did occur at well-ventilated fires of Φ ≈ 0.25. CO yields are not reported. Another study on wood pyrolysis in the upper layer has been conducted by Lattimer et al. [26] in their prototype compartment with attached hallway (Fig. 3). 6.4 mm thick sheets of Douglas fir plywood were mounted at the ceiling of the compartment during some tests which were fuelled by hexane fires as primary heat source. Only highly under-ventilated fires were achieved. With the GER made up by the overall mass loss rate, the CO yields (0.22 g/g at Φ = 5.2 and 0.17 g/g at Φ = 5.6) were shown to be in the order of the level predicted by Eqs. (3) & (4). It was concluded that the GER-concept is capable to predict CO, CO2 and O2 yields of compartment fires with oxygenated fuels in the upper layer, like wood, cardboard boxes, and fabrics. For upper layer fuels with no oxygen in their chemical structure, it was supposed that these fuels can not generate additional CO in a hot vitiated upper layer. Instead, upper layer species concentrations might be diluted by added total hydrocarbons (THC). The nominal CO yields would be further decreased, as the additional mass loss rate from upper layer fuel increases the denominator in the yield calculation. From both test series, two conclusions can be drawn with regard to the GER-concept. For low equivalence ratios an additional CO source term from upper layer pyrolysis can occur. Significant wood pyrolysis was reported to start from 280°C [24], a temperature level much too low to expect upper layer oxidation of CO to CO2. In this case upper layer pyrolysis must be considered as an additional CO source. For reactive upper layers (Tul > 900 K), assuming a well stirred upper layer, and with the upper 831



layer fuel mass loss rate taken into account in the GER-calculation, no example is given that the GER CO yield correlations Eqs. (3) & (4) are not valid. ASSESSING EXTERNAL COMBUSTION

In case of external combustion by any of the two phenomena (Fig. 4a+b) the final CO yields are reduced. A simple assumption is that external combustion occurs when the fire goes under-ventilated and excess fuel leaves the compartment. However, this approach on the one hand does not consider flame extensions which allow for an earlier external combustion, and on the other hand it does not account for inerting effects of exhaust gases which prevent external combustion of excess fuel. The occurrence of flame extensions can be found by plume equations for simple geometries. When the flame is bent by the compartment flow, a CFD fire model which predicts the flame shape can be used for a more exact study of the flame location. To assess external burning from under-ventilated fires, Beyler [4][5] has presented a methodology by calculating an ignition criterion he called ignition index, which refers to the hot gases leaving the control volume and mixing with ambient air: Xi ⋅ ΔH C,i

n

I= ∑

i =1 T AFT ,SL,i

∫

T mix

≥ 1 ⎯⎯ → external combustion

(7)

n p ⋅ c p ⋅ dT

The ignition index is the sum of the ratios of the reaction enthalpies of the fuel species i of the exhaust gases (like THC, CO, H2) divided by the heat that can be taken up by the combustion products within the temperature interval of the mixing temperature Tmix of the stoichiometric mixture and the adiabatic flame temperature TAFT,SL,i of the stoichiometric limit mixture of i. The value np accounts for the number of moles of combustion products per mole of reactants. By definition I ≥ 1 allows external combustion of the exhaust gases when mixing with ambient air and an ignition source present. The ignition index was shown to be valid for typical C, H and O containing fuels. As Eq. (7) is difficult to handle, by regarding only one fuel species and assuming complete combustion Beyler [4][5] has derived the relationship between the GER of under-ventilated compartment fires and the temperature of the exhaust gases fulfilling his ignition criterion

⎛ 1 − 1/ Φ I=⎜ ⎝ 1 + 1/ rair

∞ ⎤ ⎞ ⎡ ΔH C,O2 ⋅ yO2 → external combustion ⎥ ≥ 1 ⎯⎯ ⎟⎢ ⎠ ⎢⎣ c p ⋅ (TAFT,SL − Tmix ) ⎥⎦

(8)

Assuming a constant heat capacity, the mixing temperature of the stoichiometric mixture of upper layer gases with ambient air becomes Tmix =

Tul + y f ⋅ rair ⋅ Ta 1 + y f ⋅ rair

(9)

Eq. (8) can be solved for equality conditions. By using semi-universal constants ΔHC,O2 = 13.4 MJ/kg, cp = 1.1 kJ/(kg*K), TAFT,SL = 1700 K, y ∞ O2 = 0.233, Ta = 300 K and a typical rair = 14.3, Beyler has obtained a linear relationship between the GER where external burning from under-ventilated conditions is achieved (ΦEC) and the necessary upper layer temperature Tul (Fig. 9). The illustrated correlation is ΦEC = - 8.135 *10-4 K-1 * Tul + 2.36

(10)

832



ΦEC [ ]

2.4

Beyler's correlation

2.2

Beyler

2.0

Morehart Gottuk

1.8

hoods, gaseous fuels - prototype comp., hexane

Toxfire PP

1.6

Toxfire Ny

ISO 9705 comp., solid fuels

1.4

Toxfire TMTM

1.2

extension 1

χcomp (Tul)

1.0

extension 2

χ = 0.85

300

500

700

900

1100

1300

Upper layer temperature [K]

Fig. 9. Relationship between the equivalence ratio to achieve external burning from under-ventilated conditions and the necessary upper layer temperature. For comparison the results of hood experiments of Beyler (layer burning) [4] and Morehart (ignition with extra air) [6] together with Gottuk’s prototype compartment [10] and the amended TOXFIRE ISO 9705 compartment data (Fig. 7) [7] are illustrated. Two extensions of Beyler’s correlation are also given (extended from [5]).

The comparison with experimental data shows that the criterion is conservative for the hood data, which was achieved at relatively low temperatures and with gaseous fuels. In contrast, for the (prototype) compartment data of Gottuk with hexane and of the TOXFIRE project with solid fuels the correlation is not conservative. Hence, Beyler’s criterion was extended for the combustion efficiency inside the compartment χcomp and the overall combustion efficiency including external combustion χ. In contrast to Beyler’s assumption of ideal combustion, χcomp is smaller than 1 especially for low upper layer temperatures, where combustion does not take place in the upper layer. In this case, the exhaust gases are fuel-richer than assumed by Beyler’s derivation. The overall combustion efficiency is typically 0.80 < χ < 0.95 with gaseous fuels having a higher efficiency than solid fuels [20]. With both parameters, Eq. (8) becomes ∞ ⎤ ⎛ 1 − χ comp / Φ ⎞ ⎡ χ ⋅ ΔH C,O2 ⋅ yO2 → external combustion I=⎜ ⎢ ⎥ ≥ 1 ⎯⎯ ⎟ ⎝ 1 + 1/ rair ⎠ ⎢⎣ cp ⋅ (TAFT,SL − Tmix ) ⎥⎦

(11)

From the hood data [6] the temperature dependent combustion efficiency inside the compartment can be approximated by χ comp = 0.4036 ln(Tul ) - 1.7069 for 500 K < Tul < 850 K (12) and χ comp = 1.4036 ln(Tul ) - 1.7069 for Tul > 850 K

(13)

The solution of Eq. (11) with χcomp according to Eqs. (12) & (13) and χ = 1 is displayed in Fig. 9 (extension 1). The new correlation hits the experimental data from gaseous fuels at low temperatures much better than Beyler’s original correlation. The solution of Eq. (11) with χcomp = 1 and χ = 0.85 is also displayed in Fig. 9 (extension 2). The new correlation is conservative for the data from solid fuels at high temperatures. When external combustion was observed at lower ΦEC than indicated by the extended correlation, this happened by the influence of flame extensions which reduces the assumed χcomp = 1. The efficiency of external combustion to reduce CO yields depends on the amount of oxygen entrained in the flaming region and thus is a function of the exhaust vent configuration. Secondary fire plumes in ambient air have shown to be quite effective to reduce CO yields. Although quantification is not yet possible, it is seen from the TOXFIRE data that the more the extended ignition criterion is fulfilled, the 833



more effective the external combustion becomes. For an attached hallway (Fig. 3) Gottuk and Lattimer [9] proposed to consider external combustion only when the exhaust gases leave the compartment into fresh air below the hallway’s smoke layer. This approach is conservative because it neglects the actual composition of the atmosphere in the attached hallway. Under vitiated hallway conditions, yO2 decreased and Tmix increased (Eq. (11)). As these parameters partly balance each other out in the calculation of the ignition criterion [7], CO reduction is supposed to take place also in vitiated hallways.

CONCLUSIONS AND OPEN QUESTIONS CO yields of fully-developed compartment fires are related to the equivalence ratio to a large extent. The idealised correlations [9] derived by Gottuk and Lattimer from scaled compartment data reasonably fit with ISO 9705 compartment results. However, for Φ ≈ 0.5 and upper layer temperatures below 900 K, elevated CO yields which are not covered by Gottuk and Lattimer’s low temperature correlation were partly observed. On this point additional basic data is needed. An important additional CO source is by (wood) pyrolysis in the upper layer. As this process is mainly temperature dependent, also at wellventilated (here: Φ ≈ 0.25) conditions elevated CO levels occur. It is not known yet, whether CO from wood pyrolysis is reduced in reactive upper layers (Tul > 900 K) under over-ventilated conditions. CO yields might be reduced by flame extensions or by external combustion from under-ventilated conditions. Flame extensions can be assessed by plume equations or by state of the art CFD fire simulation. For external combustion from under-ventilated conditions, the extensions of Beyler’s simplified solution of his ignition index showed to be in reasonable accordance with the given experimental data. A consistent data set over the whole temperature range would allow for a further validation of the ignition criterion. For the efficiency of external combustion also few data is available. However, the use of the extended ignition criterion promises to provide also results on the efficiency of CO reduction.

ACKNOWLEDGEMENTS The authors would like to thank Dr. William Pitts, Dr. Anders Lönnermark and Dr. Per Blomqvist for fruitful discussions about the their experimental data. This work is sponsored by the Deutsche Forschungsgemeinschaft in the framework of the international post-graduate programme “Risk Management on the Built Environment” of the Braunschweig University of Technology and the University of Florence.

REFERENCES [1] [2] [3] [4] [5] [6] [7]

BABRAUSKAS, V.; GANN, R.G.; LEVIN, B.C.; PAABO, M.; HARRIS, R.H.; PEACOCK, R.D.; YUSA, S. 1998: A methodology for obtaining and using toxic potency data for fire hazard analysis. FSJ 31. pp. 345-358. GANN, R.G.; BABRAUSKAS, V.; PEACOCK, R.D. 1994: Fire Conditions for Smoke Toxicity Measurements. FAM, Vol. 18, pp. 193-199. BEYLER, C.L. 1983: Development and burning of a layer of products of incomplete combustion generated by a buoyant diffusion flame. PhD Thesis, Harvard University. BEYLER, C.L. 1984: Ignition and burning of a layer of incomplete combustion products. Combustion Science and Technology (39), pp. 287-303. BEYLER, C.L. 2002: Flammability Limits of Premixed and Diffusion Flames. In: DiNenno, P.J. et al. (Ed.). The SFPE Handbook of Fire Protection Engineering. 3rd Edition, Quincy, Massachusetts, pp. 2-172 – 2-187. ZUKOSKI, E.E.; MOREHART, J.H.; KUBOTA, T.; TONER, S.J. 1991: Species production and heat release rates in two-layered natural gas fires. Combustion and Flame (83), pp. 325-332. FORELL, B. 2007: A Methodology to assess Species Yields of Compartment Fires by means of an extended Global Equivalence Ratio Concept. Dissertation. Braunschweig University of Technology and University of Florence. 2007. Permalink: http://www.digibib.tubs.de/?docid=00020916 834



[8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

LIM, C.S.; ZUKOSKI, E.E.; KUBOTA, T. 1984: Mixing in Doorway Flows and Entrainment in Fires Plumes. National Bureau of Standards, Gaithersburg, MD. GOTTUK, D.T., LATTIMER, B.Y. 2002: Effect of Combustion Conditions on Species Production. In: DiNenno, P.J. et al. (Ed.). The SFPE Handbook of Fire Protection Engineering. 3rd Edition, Quincy, Massachusetts, pp. 2-54 – 2-82. GOTTUK, D.T. 1992: Generation of Carbon Monoxide in Compartment Fires. NIST GCR 92-619. Also PhD Dissertation, Virginia Polytechnic and State University, Blacksburg. LÖNNERMARK, A., BLOMQVIST, P. MANSSON, M, PERSSON, H. 1996: TOXFIRE - Fire characteristics and smoke gas analyses in under-ventilated large scale combustion experiments: Tests in the ISO 9705 room. SP Report 1996:45, ISBN 91-7848-650-5. Boras. BLOMQVIST, P.; LÖNNERMARK, A. 2001: Characterization of the Combustion Products in Largescale Fire Tests: Comparison of Three Experimental Configurations. FAM, 25, pp. 71-81. MANSSON, M.; LÖNNERMARK, A.; BLOMQVIST, P.; PERSSON, H.; BABRAUSKAS, V. 1996: TOXFIRE - Fire characteristics and smoke gas analyses in under-ventilated large-scale combustion experiments. SP Report 1996:44, ISBN 91-7848-649-1. Boras. LÖNNERMARK, A.; BABRAUSKAS, V. 1996: TOXFIRE - Fire characteristics and smoke gas analyses in under-ventilated large-scale combustion experiments: Theoretical background and calculations. SP Report 1996:49, ISBN 91-7848-654-8. Boras. BLOMQVIST, P. 2005: Emissions from Fires - Consequences for Human Safety and the Environment. Dissertation. Lund University, Department of Fire Safety Engineering. DRYSDALE, D. 1998: An Introduction to Fire Dynamics. 2nd Edition, Wiley & sons, Chichester. HOSSER, D.; BLUME, G.; WILL, J. 2000: Brandschutz in Chemikalienlagern -Experimentelle und theoretische Untersuchungen zur Optimierung von Nachweismethoden und Schutzkonzepten. iBMB, TU-Braunschweig, Heft 151, Braunschweig. JANSSENS, M.; TRAN, H.C. 1992: Data Reduction of Room Tests for Zone Model Validation. JFS, Vol. 10, Nov./Dec. pp. 529-555. TROITZSCH, J. 2004: Plastics Flammability Handbook, Principles, Regulations, Testing, and Approval. 3rd Ed. Hanser Publishers, Munich. TEWARSON, A. 2002: Generation of Heat and Chemical Compounds in Fires. In: DiNenno, P.J. et al. (Ed.). The SFPE Handbook of Fire Protection Engineering. 3rd Edition, Quincy, Massachusetts, pp. 3-82 – 3-161. QUINTIERE, J.G. 2002: Compartment Fire Modeling. In: DiNenno, P.J. et al. (Ed.). The SFPE Handbook of Fire Protection Engineering. 3rd Edition, Quincy, Massachusetts, pp. 3-162 – 3-170. MULHOLLAND, G.; JANSSENS, M.; YUSA, S.; TWILLEY, W.; BABRAUSKAS, V. 1991: The Effect of Oxygen Concentrations on CO and Smoke Produced by Flames. Fire Safety Science Proceedings of the Third International Symposium, IAFSS, Edinburgh, pp. 585-594. TSUCHIYA, Y.; MATHIEU, J.F., 1991: Heat, CO and Smoke Releases of Plywood under a Depleted Oxygen Atmosphere: An experimental Study Using an OSU apparatus. Fire Safety Science Proceedings of the Third International Symposium, IAFSS, Edinburgh, pp. 605 - 614. PITTS, W.M.; JOHNSSON, E.L.; BRYNER, N.P. 1994: Carbon monoxide formation in fires by high temperature anaerobic wood pyrolysis. Twenty-Fifth Symposium (International) on Combustion/The Combustion Institute, pp. 1455-1462. PITTS, W.M. 1997: An Algorithm for Estimation Carbon Monoxide Formation in Enclosure Fires. IAFSS, Proceedings of the Fifths International Symposium, pp. 535-546, Melbourne, Australia. LATTIMER, B.Y.; VANDSBURGER, U.; ROBY, R.J. 1998: Carbon Monoxide Levels in Structure Fires: Effects of Wood in the Upper Layer of a Post-Flashover Compartment Fire. Fire Technology, Vol. 34, No. 4, pp. 325-355.

835
