Smll-scle oservtions of typicl fire spred cused y the interction of wind, terrin nd fire Json J. Shrples Applied nd Industril Mthemtics Reserch Group, School of Physicl Environmentl nd Mthemticl Sciences, University of New South Wles t the Austrlin Defence Force Acdemy, Cnerr, ACT Austrli. Emil: j.shrples@df.edu.u Domingos X. Viegs, Crlos G. Ross Deprtmento de Engenhri Mechânic Universidde de Coimr Centro de Estudos Sore Incêndios Florestis Associção pr o Desenvolvimento d Aerodinâmic Industril Coimr, Portugl. Emil: xvier.viegs@dem.uc.pt; crlos.ross@dem.uc.pt Richrd H.D. McRe Risk Mngement Section Austrlin Cpitl Territory Emergency Services Agency Cnerr, ACT Austrli Emil: Rick.McRe@ct.gov.u Astrct The interction of wind, terrin nd fire urning in lndscpe cn produce vriety of unusul yet significnt effects on fire propgtion. Moreover, there is strong evidence to suggest tht such effects plyed n importnt role in driving the spred of lrge wildfires such s the Jnury 2003 lpine fires over southestern Austrli, the Ferury 2009 fires in Victori nd the Jnury 2010 fires in the Tinderry Rnges to the southest of the Austrlin Cpitl Territory. In this pper we investigte the interction etween wind, terrin nd fire urning on lee slope through numer of smll-scle experiments conducted within the comustion tunnel t the Lorotório de Estudos Sore Incêndios Florestis in Lousã, Portugl. The experiments involved using n idelised ridge configurtion, which ws ligned perpendiculr to strong wind so s to produce lee (seprtion) eddy. Fire ws then introduced to the lee slope through point ignition nd the fire spred resulting from the interction etween the fire nd the seprted flow ws oserved. The interction etween the fire nd seprted flow ws consistently oserved to produce rpid lterl propgtion of the fire cross the top of the slope. The detils of the rpid lterl spred re discussed long with some prole implictions of this type of typicl fire spred for wildfire risk mngement t the lndscpe scle. Keywords: wind, terrin nd fire interction; typicl wildfire spred; wildfire experiment; comustion tunnel; wildfire risk mngement.
1. Introduction During the lst decde numer of serious wildfires hve urnt in rugged terrin under the influence of strong winds. Exmples include the Jnury 2003 lpine fires over southestern Austrli nd the Ferury 2009 Victorin fires. In prticulr, on 18 Jnury 2003 fires driven y strong winds (verging 30-40 km h -1 ) urnt through rugged terrin to the west of Cnerr nd impcted the city resulting in four ftlities nd the destruction of round 500 houses within few hours (Nirn 2003). Strong winds were lso fctor during the fires on Blck Sturdy (7 Ferury 2009), which cused unprecedented loss of life in rugged terrin in centrl Victori (Tegue et l. 2010). A numer of other exmples from round the world (e.g. the 2009 Jesusit fire in Cliforni) re lso relevnt. Given the complex terrin in which these nd other lrge fires hve urnt, nd the strong winds experienced during their most devstting runs, it is nturl to consider the concept of wind-terrin interction nd to investigte wht significnce interction etween the wind nd terrin, nd dditionl interctions with fire, my hve on fire development. As discussed y McRe (2004), multispectrl line-scn imgery recorded during the Cnerr fires on 18 Jnury 2003 reveled numer of instnces of typicl fire spred tht pper to e driven y interctions etween the wind, the terrin nd fire. Exmples of such instnces cn e seen in Fig. 1. The imges depict the fires t the loctions known s Pig Hill (Fig. 1) nd Broken Crt (Fig. 1). The fire ehviour in these instnces is chrcterised y: 1. Rpid lterl propgtion of the fire, perpendiculr to the wind, cross steep lee slope (including some lterl spotting); 2. Downwind extension of the flming zone for severl kilometres; 3. The upwind edge of the flming zone is constrined y rek in topogrphic slope. N N Events Events 500m 1km Figure 1. Multispectrl line-scn imgery of the Cnerr fires 18 Jnury 2003 showing events t: () Pig Hill, nd () Broken Crt. Source: New South Wles Rurl Fire Service. Following McRe (2004), numer of cses of this type of typicl fire spred (23 in totl) were studied y Shrples et l. (2010). This study considered numer of possile mechnisms tht could produce the typicl spred in light of the ville evidence. By comining the ville evidence with n nlysis of terrin chrcteristics nd the results
of previous study of wind regimes over complex terrin, Shrples et l. (2010) were le to discount numer of mechnisms nd therey deduce the most likely (hypotheticl) mechnism driving the oserved typicl fire spred. It ws concluded tht the typicl fire spred ws most likely driven y the interction of the fire with lee (seprtion) eddy. In this scenrio the fire enters region of seprted flow (the lee eddy) nd is propgted lterlly. Incresed turulence rising from the interction then results in incresed genertion of emers, some of which re incorported into the synoptic winds ove the eddy nd re deposited downwind. Conducting fire experiments involving fire ehviour t the scles represented in the line-scn imgery in Fig. 1, is intrctle nd dngerous. Moreover, the complexities of the physcil environment mke investigtion nd isoltion of the key driving processes through field experimenttion difficult tsk. We therefore sought to further investigte the physicl mechnisms driving the typicl fire spred through numer of smll-scle experimentl fires. The experimentl fires were conducted within the comustion tunnel t the Lortório de Estudos Sore Incêndios Florestis ner Lousã in Portugl. 2. Experimentl Set-up To test the hypotheticl mechnism discussed in Shrples et l. (2010) numer of experiments were performed in the comustion tunnel t the Centro de Estudos Sore Incêndios Florestis lortory. The experiments involved the use of n idelised ridge configurtion comprised of two djoining slopes. The windwrd slope, mde of wood, hd dimensions of 1.65m 2m nd ws inclined t n ngle of β. The lee slope, mde of metl plte covered with mesh, hd dimensions of 1m 2m nd ws inclined t n ngle of α. The inclintion α ws tken sufficiently lrge to ensure seprtion of the irflow t the ridge-line nd the formtion of lee (seprtion) eddy. Strw fuel ws plced on the lee slope with n rel density of pproximtely 0.6 kg m -2. The ridge-line ws ligned perpendiculr to the comustion tunnel wind direction. See Fig. 2 for n illustrtion of the experimentl ridge configurtion. The width of the comustion tunnel (2.8 m) ws greter thn the width of the experimentl rig nd it ws decided to plce the rig up ginst one side of the comustion tunnel, resulting in gp of round 80 cm etween the experimentl ridge nd the other side of the tunnel. While this resulted in some symmetry in the flow over the ridge, its effect on the experimentl results ws miniml. β Figure 2. Schemtic digrm (cross section) of the experimentl ridge configurtion nd the pproximte ignition point. In series of experiments, point ignitions were mde on the right side of the lee slope (15cm in from edge), in the centre of the slope nd on the left side (lso 15cm in from the edge). All ignitions were mde 30cm from the ottom of the slope. The ignitions were mde in the presence of 4 ms -1 wind lowing t right ngle to the ridge. For comprison point ignitions were lso mde on the right side of the slope nd in the centre of the slope in Fuel 4 ms 1 wind Cmer view α
the sence of wind. The defining chrteristics of the fire experiments conducted re listed in Tle 1. The spred of the experimentl fires ws recorded using two video cmers, one in the position indicted in Fig. 2, nd one positioned to cpture the trnsverse (olique) view. Still photogrphy using digitl cmers ws lso used to record the oserved fire spred. Experiment No. α β Speed (ms -1 ) Ignition 1 30 18 4 R 2 35 20 4 R 3 35 20 0 R 4 35 20 4 C 5 35 20 0 C 6 35 20 4 L Tle 1. Experimentl prmeters: R = right side ignition, C = centrl ignition, L = left side ignition. 3. Results The oserved fire spred ws distinctly different in the presence of the 4 ms -1 wind, when compred to the no wind cses. In the presence of wind, fire ws consistently oserved to spred upslope until it reched the region of seprted flow ner the top of the ridge. At this point the fire ccelerted significntly, exhiiting rpid unstedy spred cross the top of the ridge in the form of finger of turulent flme. Figs. 3 nd 3 illustrte the difference in lterl spred chrcteristics for the wind nd no wind cses when the point ignition ws mde on the right hnd side of the lee slope. Figs. 3c nd 3d demonstrte tht similr typicl spred occurred when the ignition (with wind) ws mde on the left side nd centre of the slope. In fct, for the centrl ignition the rpid lterl spred ws oserved to occur in oth directions long the top of the slope. This indictes tht the rpid lterl spred evident in Fig. 3 cnnot simply e due to n symmetry in the flow. c d Figure 3. () Exp. 3: right side ignition, no wind; () Exp. 2: right side ignition with 4 ms -1 wind; (c) Exp. 6: left side ignition with 4 ms -1 wind; () Exp. 4: centrl ignition with 4 ms -1 wind.
Fig. 4 illustrtes the rpid lterl spred cross the slope in time sequence. Fetures of interest include the formtion of spot fires in Figs. 4f nd 4h, nd the distinctive pttern of smoke highlighted in Fig. 4e. The turulent nture of the flmes is evident throughout the sequence. The lterl spred chrcteristics resulting from the right side ignitions (Exps. 1, 2 nd 3) re quntified in Fig. 5. The significnt lterl ccelertion of the wind-driven fires fter reching the top of the lee slope is evident in Fig. 5 from the chnge in grdient. Note tht the verticl lck dshed lines in Figs. 5 nd 5 mrk the time when the wind-driven fires reched the ridge-line t the top of the lee slope (pproximtely 20-22s fter ignition in 9:51:00, Ignition + 12s 9:51:08, Ignition + 20s c d 9:51:10, Ignition + 22s 9:51:14, Ignition + 26s e f Drker Smoke on dvncing flnk Spot Fire 9:51:18, Ignition + 30s 9:51:20, Ignition + 32s g h Spot Fires 9:51:22, Ignition + 34s 9:51:26, Ignition + 38s Figure 4. Time sequence of photogrphs from experiment 1 (right side ignition with 4 ms -1 wind). The ignition ws mde t 09:50:48. The wind is lowing towrds the cmer.
ll experiments). Similrly, Fig. 5 shows tht fter reching the top of the slope the winddriven fires exhiited unstedy lterl rtes of spred of etween 1 m min -1 nd 6 m min -1, which were etween 2 to 12 times fster thn the (ner constnt) lterl spred rte of out 0.5 m min -1 oserved in the no wind cse. Mximum lterl rtes of spred occurred immeditely fter the fires reched the top of the slope. The lterl fire spred in the presence of the 4 ms -1 wind exhiited the sme qulittive ehviour for oth vlues of the inclintion prmeter (α = 30 nd 35 ), though some quntittive differences were pprent. 1.8 1.6 6.0 5.0 No w ind + 35 slope + 30 slope + 35 slope 12 10 Lterl Distnce (m) 1.4 1.2 0.8 0.6 Lterl RoS (m/min) 4.0 3.0 8 6 4 Nondimensionl RoS 0.4 No w ind + 35 slope 2 0.2 + 30 slope + 35 slope 0.2 0.4 0.6 0.8 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1.2 Figure 5. Right side ignitions experiments 1, 2 nd 3: () Mximum lterl distnce trvelled plotted ginst time since ignition, () Dimensionl nd nondimensionl instntneous rte of spred plotted ginst time since ignition. Fig. 6 shows the experimentl spred dt rising from ignition on the left side of the slope. The wind-driven fire gin exhiits mrked increse in lterl spred rte upon reching the top of the slope, with similr qulittive structure to tht seen in the right side ignition cses. The spred rtes for the left side ignition were slightly lower (mximum rte of spred of round 4 m min -1 s opposed to 6 m min -1 in the right side ignition cse) ut were still up to 8 times greter thn the qusi-stedy lterl spred rte for the no wind cse. The mximum lterl spred rte ws gin found immeditely fter the fire reched the top of the slope. The smller lterl spred rte for the left side ignition cse could e due to n symmetry in the flow creted y the comustion tunnel wll on the right side of the experimentl rig. We note lso tht the no wind dt in Fig. 6 is from the right side ignition cse; due to the symmetry of the experimentl configurtion in the sence of wind, it ws deemed cceptile to compre these dt with those from the wind-driven left side ignition cse. The lterl spred chrcteristics for the centrl ignition cses (Exps. 3 nd 4) re quntified in Fig. 7. In these cses spred to oth the right nd left sides of the slope hve een ccounted for. As expected, in the sence of wind the lterl spred in oth directions ws pproximtely constnt with vlue of out 0.5 m min -1. In the presence of wind, the lterl spred rtes were initilly pproximtely the sme s tht found in the no wind cse. However, fter reching point just elow the ridge-line the fire exhiited significnt lterl ccelertion, this time in oth directions simultneously. As in the
experiments lredy discussed, the spred towrds the right side of the slope ws slightly less thn tht towrds the left, proly due to n symmetry in the flow cused y the wll of the comustion tunnel. Mximum lterl rtes of spred were gin mximum (in oth directions) immeditely fter the fire reched the top of the slope. Clcultion of the grdients of the dt curves in Fig. 7 indictes tht the wind-driven fires spred lterlly with rte of spred of out 4.75 m min -1, which is pproximtely 9.5 times fster thn the spred oserved in the no wind cse. Fig. 7 shows tht the lterl spred rte chrcteristics were qulittively similr for ech of the different ignition ptterns. This indictes tht the physicl mechnism driving the rpid lterl spred is independent of the point of ignition. 1.8 1.6 5.0 4.0 No w ind 10 9 8 Lterl Distnce (m) 1.4 1.2 0.8 0.6 0.4 No w ind 0.2 0.2 0.4 0.6 0.8 1.2 1.4 1.6 1.8 Lterl Distnce (m) 0 0.2 0.4 0.6 0.8 1.2 1.4 Figure 6. Left side ignition experiment 6: () Mximum lterl distnce trvelled plotted ginst time since ignition, () Dimensionl nd nondimensionl instntneous rte of spred plotted ginst time since ignition. Note tht the no wind dt shown is tht rising from the right side ignition cse. 3.0 7 6 5 4 3 2 1 Nondimensionl RoS 0.9 1.8 0.8 1.6 Lterl Distnce (m) 0.7 0.6 0.5 0.4 0.3 0.2 Spred to L + no w ind Spred to R + no w ind 0.1 Spred to L + w ind Spred to R + w ind 0.2 0.4 0.6 0.8 1.2 1.4 1.6 1.8 Lterl Distnce (m) 0.2 0.4 0.6 0.8 1.2 1.4 Figure 7. () Mximum lterl distnce trvelled plotted ginst time since ignition for the centrl ignition cses - experiments 4 nd 5, () Comprison of mximum lterl distnce trvelled versus time since ignition for the different ignition loctions. 1.4 1.2 0.8 0.6 0.4 0.2 Right Ignition Left Ignition Centrl Ignition
4. Discussion nd Conclusions A numer of smll-scle experiments involving fires urning on lee slope were conducted in comustion tunnel. The experiments confirmed the existence of complex interction etween the fire, the wind nd the terrin tht results in the rpid lterl propgtion of the fire cross the slope in ccordnce with the hypothesis dvnced in Shrples et l. (2010). The oserved ehviour in the smll-scle experiments ws qulittively similr to some oservtions of the Cnerr ushfires, which urnt in rugged terrin under the influence of strong winds. For exmple, the typicl fire spred evident in Fig. 1 ers remrkle resemlnce to the flme structure seen in the initil stges of the lterl spred development oserved in the lortory. Oservtions resemling the lortory fire ehviour hve lso een mde in connection with other ushfires urning in rugged terrin under extreme fire wether conditions. For exmple, Fig. 8 shows two instnces of typicl usfire spred. Of prticulr note is the distinctive drker smoke on the dvncing flnks of the fires in Fig. 8. A similr pttern of smoke cn e seen in Fig. 4e. Spred direction Spred direction Figure 8. Instnces of typicl wildfire spred: () fire spreding lterlly cross slope during the Tinderry fires southest of Cnerr, Jnury 2010 (Photo y Mr. Steve Fores, ACT Emergency Services Agency); () fire spreding lterlly elow ridge-line during the Cnerr fires, 18 Jnury 2003 (Photo y Mr. Stephen Wilkes, NSW Rurl Fire Service). The wind-driven experimentl fires exhiited lterl rtes of spred tht were 6-10 times fster thn tht oserved in the sence of wind. Mximum lterl rtes of spred were consistently found to occur immeditely fter the fire reched the top of the lee slope. The physcil mechnism driving the typicl lterl spred ws found to e independent of the point of ignition. The lortory fires lso exhiited some spotting ehviour, though this ws limited y the nture of the (strw) fuel. However, it is entirely plusile tht in rel ushfire sitution (involving fuels tht redily produce firernds such s cn e found in euclypt forests), emers incorported into the min flow ove the seprtion eddy would result in multiple spot fires downwind of the region of lterl spred nd the formtion of extensive zones of ctive flme like tht seen in Fig. 1. Such process could e trigger for the formtion of pyro-cumulonimus, which re of mjor concern for ushfire risk mngement. The rpid lterl spred indicted in the lortory fires lso poses significnt thret to fire-fighter sfety nd could seriously compromise fire suppression tctics. In prticulr, the possiility of rpid lterl spred should cst serious douts over flnk ttck
options on lee slopes when winds re strong, nd could cuse ctstrophic decy in the sfety of fire-fighters lredy conducting flnk ttck opertions on steep slopes. Further collortive work will involve comustion tunnel studies of lee slope fires under vrile wind speeds nd directions nd urning on slopes of vrile inclintion. Repeting the experiments with different fuel types (e.g. Pinus pinster) will lso permit etter definition of the lterl spred chrcteristics nd the driving mechnsims. Acknowledgements This work ws supported y n Austrlin Acdemy of Science Scientific Visits to Europe Progrm grnt. The uthors re indeted to Mr. Dvid Dvim, Mr. Ricrdo Oliveir nd Mr. Jorge Rposo for their invlule ssistnce with the comustion tunnel experiments, nd to Mr. Antonio Crdoso for design nd construction of the experimentl rig. Dr. J. Shrples is grteful to Prof. John Dold (University of Mnchester, UK) nd Dr. Mlcolm Gill (ANU, Austrli) for their keen interest nd illuminting discussions on the topic of this pper. The uthors would lso like to cknowledge the New South Wles Rurl Fire Service, which provided the line-scn imgery. References McRe, R. (2004) Breth of the drgon oservtions of the Jnury 2003 ACT Bushfires. In: Proceedings of 2004 Austrlsin Bushfire Reserch Conference, My 2004, Adelide. Nirn G (Chir) (2003) A Ntion Chrred: Inquiry into the Recent Austrlin Bushfires. The Prliment of the Commonwelth of Austrli, Cnerr. Shrples, J., McRe, R., Weer, R., Wilkes, S. (2010) -terrin effects on the propgtion of wildfires in rugged terrin: fire chnnelling. Interntionl Journl of Wildlnd Fire, under review. Tegue, B., McLeod, R., Pscoe, S., (2010) 2009 Victorin Bushfires Royl Commission Finl Report. Prliment of Victori. Aville from: http://www.roylcommission.vic.gov.u/commission-reports/finl-report