NATIONAL RESEARCH COUNCIL CANADA

DIVISION O F BUILDING RESEARCH

SMOKE DETECTION: A STATUS REPORT

by

M, Sultan

O t t a w a

February 1984

TPLELE OF CONTENTS

Page

1. ABSTRACT

2, INTRODUCTION

3. CHMACTERISTICS OF COMBUSTION AEROSOLS

4 . MECWANISM OF SMOKE DETECTOR OPERATION 4.1 Photoe lec tr irType Detectors 4.2 Ionization-Type Detectors 4 . 3 Combined Ionization/Fhotoelectric Detectors

4 .3 .1 The "OR" Detector 4.3.2 Rze "AND" Detector

5. P O S I T I O N I N G OF SMOKE DETECTORS IN RESIDENCES 5.1 Location of Smoke Detector I n s i d e a Room 5.2 Tnstallatfon of Smoke Detectors in Long Halls 5.3 I n s t a l l a t i o n 05 Smoke Detectors Ln C o r r i d o r s 5.4 Detection of Bedroom Fires

6 . WAKLNG EFFECTIVENESS OF SMOKE DETECTORS 6.1 Factors Affecting the Waking Response

7. CAPABILITIES AND LIMITATIONS OF SMOKE DETECTORS

8. CONCLUSION AND RECOMMENDATIONS FOR FUTURE WOKK

9. REFERENCES

20. TABLE

1. ABSTRACT

After a discussLon on the characteristics of combustion aerosols, various aspects of the performance of residential smoke detectors are reviewed w l t h emphasis on operation mechanisms, posltloning, waking effectiveness, and capabilfry and l imitations. Recorn~nendatlons for fu tu re research are made to improve t h e understanding of smoke detect ion practices.

2 . INTRODUCTION

Loss of l l f e by fire is a significant problem th roughout the world. Rright reported that 40 to 50% of t h e p e o p l e k i l l e d in f i r e s each yeas could be saved if ear ly warning f l r e detection devices were i n s t a l l e d 1 . Fire death statistics f o r various countrLes a r e presenter1 i n Flgure I (from Ranks and ~ a r d l n ~ and Banks 3). Despite some improvement In f i re s ta t . l s t ics during t h e l a s t f e w years3, the U n i t e d States and Canada share the hlghest d e a t h rate per millien pe r sons among t h e nat ions covered, almost t w i c e t h a t of any o t h e r country reported.

In their ea r ly stages, most f i r e s are innocuous and can be e a s i l y controlled. The earl ier the detection, the better the chances of escape and the sooner suppression methods can be a p p l i e d .

3. CHARACTERISTICS OF COMBUSTION AEROSOLS

A s reported by Custer and I3 r igh t4 , combusrion releases into t h e atmosphere s o l i d part ic les (aerosols) r a ~ g i n g it? s i z e E rom 5 x 13-+ t o 10 urn. Particles less than 0.3 pm do not scatter I l g h t eEficiectIy, and thus are practically invisible.

In the event of a fire, on ly de tec t lon aL che s t a r t t l f combustlont whether flaming or smouldering, can e n s u r e effective in te r -dent ion , The effectiveness of detectfon depends on the nature of the alarm sEgnal produced by cornbustlon products. Of the avai lab le models of cletectots, two types respond to the presence of particles: o p t i c a l detectors measure smoke concentration either by l i g h t absorption or l i g ? i t scattering, and ionization chamber detectors respond t o the presence of smoke by changed dynamic resistance of an ionization chamber, which resalts in the reduction of the quiescect current produced by a radiolsotoplc source.

In order to enhance t h e performance of a smoke detector, one must know t h e characteristics of combustion aerosols. The a e r o s o l s i z e distribution frorn smouldering and flaming combustion of various materials has been determined by Moliere et al. 5 , ~ i n g - ~ , 13ankston7 and Lee and ~ulholland~. The aerosol distribution curves obtained by Moliere et al, are shown in Flgure 2 for d i f fe ren t combustible materials (Table 1) lnvolved in real f ires . Two distinct sets of

dlstributzons can be observed: one corresponds to a smoulderlng fire, where the aerosol has an average p a r t i c l e diameter of mare than 0.3 p (visible aerosol) and one corresponds to a flaming f l r e wLth subs tan t i a l invisible aerosol (average particle diameter less than 0.2 vm). These d l s t r i b u t l o n s are even more strlking when presented i n the form of a cumulatTve p r e b a b l l l t y sca le f r o m 0 t o 100% f o r each f l re , as shown in Flguse 3.

4 . MECHANISM OF SMOKE DETECTOR OPERATION

There are three t y p e s of smoke detectors: the point- type detector, whfch detects smoke at one pos i r ion , the sampling detector , whLch is a point-type detec tor taking samples from a number of positions, and the b e a r t y p e detector, which covers a length u p to 100 m. Only point-type detectors wLll be considered in th is report.

4.1 Photoelectric-Type Detectors

Photoelectric detection of smoke has been employed for some t i m e , partLcularly where t h e fire Ls expected to generate a s u b s t a n t i a l amount of smoke. T h l s type of detector operates oc the p r i n c i p l e t h a t smoke obscures a light beam or scatters l i g h t i n t o a photocell, as shown tn Figure 4 (from ort the^^). The change In current resalting from partial obscura t lon or l i g h t reflectLon of a photnelectrlc beam by smoke between a recelvlng element and a light source is measured and an alarm Ls tsLpped when t h e obscuration o r s ca t t e r reaches a crltical value. Photoelect rlc detect oss respond qulckly to optically dense smoke, b u t they are far less sensit ive than i o n i z a t i o n chamber detectors to cleao-burning flres.

4.2 Ionization-Type Detectors

An lonization de t ec to r consists of one or two ionization chambers and amplificatton ctrcuits. A i r in t h e lonization chamber is made e lec t r i ca l ly conductive (ionized) by bombardment of the nit rogen and oxygen molecules with alpha particles e m i t t e d by a m i n u t e radioactive source. A voltage appl ied across the ionization chamber causes a very small electric current to flow as the ions travel to the electrode of oppos l t e po lar i ty . When combustion particles enter the chamber they attach themselves to t he i o n s , as schenatPcally shown I n Figure 5 (f rom Northey), and cause a reductLon i n current flow. The reduced current f l a w increases the voltage on the electrodes. When a predetermTned l e v e l 1s reached, an alarm Is activared. One of t h e problems w i t h i on lza t ton de tec tors Is thclr reduced sensltivity to large aerosol particles produced maLnly by smouldering types of f l r e s l O . An o p t i c a l detector cannot prov ide the sensltivity of an lonizat_Lan detector when the particles are small, and conversely an lonlzation detector cannot provlde the sensltlvity of an optical detector if the parttcles are large.

4.3 Combined ~onization/Photoelectric Detectors

Since photoelectric detectors and ionization d e t e c t o r s are sensitive to particles of different size spectrums, combination smoke de t ec to r s (photo-Ion) can be used t o c o v e r the f u l l spectrum of partlclesL1~91 These detectors have both LonizatLon and photoelectric sensors. The circuitry can be arranged so t h a t if ei ther sensor 'seesr smoke an alarm is glven. ' There are two t y p e s of comblned smoke detectors 1 2 .

4 .3 .1 The 'OR' detector. TnLs t y p e of detector , shown in Figure 6 ( f r o m Solomon and I?efssl2), s e t s o f f an alarm when either the ionization or the optical sensor de tec ts smoke; In o t h e r words, t he 'OR' detector is sensitive to both small and la rge particles. The false alarm incidence is approximately the s u m of the false alarms f o r the two individual detectors .

4.3.2 The 'AND' detector. This type of combination de t ec to r , shown in Figure 7 (from Solomon and Reiss), will give an alarm only if both detector sensors detect smoke. The advantage of an 'AND' detector Is Its greater immunity t o f a l s e alarms, I f only one detec tor is activated, the other usually holds off the alarm.

5 . POSITIONING OF SMOKE DETECTORS IN RESIDENCES

There are very little f i e l d data t o back up the current smoke detector lnstallatlon s tandards for resLdences. More r s e a r c h Is required to evaluate the current practice of positioning smoke detectors, as s p e c i f i e d in t h e f i r e p r o t e c t i o n standards,

5.1 Location of Smoke Detector Ins ide a Room

The p o s i t i o n i n g of smoke detectors in residences has been s t u d i e d by W a t e m n et a l . 3 . Roth lonizatlon and photnelectr ic smoke detectors were considered. For ty experiments were performed at t w o different sites. The ambient c o n d i t i o n s included summer conditions a t one s i t e with the building air-conditioned, and wlnter conditions at both sites with the building heated. F i r e was initiated Ln furnishings at varP~us locations throughout the slte. The time between the f i r e in l t ia t ian and the response of smoke detectors at several l o c a t i o n s w a s recorded. The results obtained indicated t h a t wall-mounted detectors responded on an average about three minutes ea r l i e r t h a n t h o s e mounted on the ceiling. These r e su l t s are cansLstent with t h o s e ob ta ined by 13ukowski14 for smoulder ing fires. Bukowski found, however, that in flaming f i r e s , ceiling-mounted detectors respond at approximately t h e same t i m e as wall-moucted detectors. H e a t t r i b u t e d t h L s to t h e fac t that there is a stranger thermal buoyancy i n flaming fires than in smouldering fires. However, as the smoke moves along the ceilir ,g, the particles begin to settle, and therefore wall-mounted detectors are exposed to h i g h e r smoke levels earllcr and respond faster.

5.2 Installation of Smoke Detectors +n Long Halls

The response time f o r smoke detectors may vary considerably depend-ing on the s i z e and configuration of the h a l l . For a long h a l l , the installation of one smoke detector at each e n d would significantly improve the warning time potentla1 in comparLson w i t h on ly one detector

1 6 i n s t a l l e d at one end of the h a l l . C ~ u l t e r ~ ~ and Kennedy et al. among others, sugges tcd tha t L n a long hall a smoke detector should be p l a c e d at least every 9 m.

5.2 Instal lat ion of Smoke Detectors in Corridors

Kennedy et al,16 studied the response of smoke detectors in a cor r idor when the smouldering f i re occurred in an adjacent roam They Eoond that wlth small smouldering flres, t h e rate 05 smoke generatfon w a s n e t high and t he temperature -in the corrldos remained l o w . When the door was open, the room and t h e corridor tended to fill w i t h smoke uniformly so t h a t detection of smoke occurred withln a re lat ive ly narrow band of t i m e at a11 detector locations.

In f l a d n g fire, the time of detec t ion is governed mainly by t h e v e l o c i t y of the smoke f r o n t as it advances down t h e corridor1 6~ 17. Consequently, for f laming f i r e s the positioning of smoke detectors becomes important .

5 .4 Detection of Bedroom Fires

A f l r e that starts in a bedroom will be detected r a p i d l j r by a smoke detector I n an adjacent corridor only lf there Ls an opening through the wall between the room and the corridor. The usual route for smoke through the w a l l Is via an open doorway. plattla found that when the door Is closed, a corridor-meunted smoke d e t e c t o r takes two mlnutes or more t o respond to a flaming f i r e withln the room; a similar detector l n the bedroom slgnals the alarm wlthin 30 seconds. The difference between these two response times is very impor tan t . Based on results of full-size experiments, Kennedy ct a1, l 6 and ~ukowskll 9 l repor ted tha t corridosmounted smoke detec tors respond t o bedroom £Ires behind closed doors only after l e t h a l smoke and gas conditions are reached within the room.

A study has begun at the F f r e Research Section, NRCC, to d e v e l o p methods of providing early warning of a bedroom f lre behind a closed door, using corridorrnounted detectors.

6 , WAKING EFFECTIVENESS OF SMOKE DETECTORS

Although extensive research conducted during the last decade has provided a bet ter understanding of many aspects of smoke detection, the wakLng effectiveness of household smoke detectors has received ILttle a t t e n t i o n . Consumers, architects, builders, f ire f ighters , and profess lonals involved in consumer safety should have available

d e t a i l e d information concesnlng the acoust ic spectral characteristics of alarm signals , energy dfstribution of alarm sfgnals, and their relatLon to hearing sensktivlty against different acoust ic backgrounds.

S l e e p i s a complex mlr iphaae process. The ability of smoke detectors to awaken a s l e e p i n g household has been investigated by Nober ct a1.20 as a functlon of sex differences, hours into s l e e p , and nights of the week. The sound l e v e l s were evaluated from a teti-f oat distance as required by NFPA (85 dBA) , a t pillow site, w i t h bedroom door open (70 dBA) , acd at p i l l o w s i r e , with bedroom door closed (55 dBA) , for seventy college-aged people (18-29 years). TheFr results Indicated t h a t the three detector alarm levels were adequate to awaken young, normal-hearing a d u l t s from sleep at any hour of the nfght. Rackground noise, such as a taped 6 3 dEA noise from an a i r conditioner, slgniffcantly decreased the waking response at t h e 55 and 70 dBA leve ls .

6 .1 Factors Affecting t h e Waking Response

Waking response is dependent on t h e following factors:

I . Time of night

T h e of night has been cited as a varlable a f f e c t i n g response. There is not always agreement among experamenters or- this po in t . Pot example, 21mermaa21 found t ha t respons Lveness of people increased as a functlon of t L m e into n l g h t (hours c r ' sleep). On t h e other hand Wlllfams et and observed a decrease in responsiveness, and Nober et al. 2 4 noted no s l g n i f i c a ~ t effect. Perhaps the time effect was n o t a p p a r e n t in Yober's stud, because I t covered only young, normal-heating a d u l t s ,

2. Age '

Age c l e a r l y affects s l e e p and response d u r i n g sleep, Lukas and ~ o b b s ~ ~ reported that older peop le awaken f a s t e r and have shorter 'grogginess' periods after a rousa l . It would be presumptuous l o p r e d i c t s leep response behaviour of a geriatric population from data collected on college-aged people.

3. Sex

Differences In sex relative t o sleep arousal have also been noted. Lukas and ~ o b b s 2 5 reported t h a t regardless of age, s leep a r o u s a l thresholds were lower in women. In Nober's studies of college-aged s t u d e n t s (18-29 years), 50% were female and 50% m a l e . H e found no slgnlflcant difference in response between men and women.

7. CAPABILITIES AND LIMITATIONS OF SMOKE DETECTORS

As discussed earlier, the sensitivity of fonizatlon-type detectors is hlghes for smoke contaLnlng small partlcles s u c h as are produced by flaming f i r e , whereas the sensitlvity of optical smoke detectors Is h i g h e r for smoke consis tlng of Large part lcles, such as are p r o d ~ c e d by smaaldesing f ire . Kennedy e t al.16, observed that: though ionization detectors dld not operace a t t he very h i g h concentrations of smoke produced by smouldering, they responded rapidly after the fLrst appearance of flames. Apparently the i a w i z a t i o n smoke detector was n o t blocked by the presence of large-particle smoke in flamlng fire.

~ukowski 26 reported tha t residential smoke detectors of e i t h e r t h e ionlzatlon or photoelectrLc type with a s h o r t response time would provide adequate warning f o r most r e s i d e n t i a l fires, if t h e d e t e c t o r s are properly i n s t a l l e d .

In summary, the effectiveness of smoke detection is limited by the following f actors20,27,28:

1. Type of de tec t ion device, due to restricted c a p a b i l i t y to respond to smoke particle size.

2. Alarm level.; f t must be not Less than 55 dBA at t h e sleeperf s p i l l o w 2 0 .

3. Source of power; if b a t t e r y operated, it m s t b e replenished from time t o time.

8. CONCLUSION AND IZECOMMENDATIONS FOR FUTURE WORK

Extensive experlmental research conducted in the last decade h a s provided some i n s l g h t into the fundamentals of de tec t l on . Some areas of the d e t e c t l o n problem, however, need b e t t e r understanding in order to select the rLght detec tor for t h e specif lc applications. The following items are proposed for further s tudy .

1. Since hear ing becomes less sens i t ive a f t e r prolonged exposure to high sound levels, experiments should be performed r o deterdne waking effectiveness of smoke detectors a t d i f f e r e n t noise hackgrounds, such as a i r conditioning and t ra f f i c , d f s c o a n d aircraft nolse, t h e last f o r those who l i v e in the v i c i n i t y of an a i rpor t .

2. The optimum location of a res ident ia l smoke d e t e c t o r should be investigated in order to ensure that t h e warnlng s i g n a l is clearly audible h a l l bedrooms of a dwelling.

3. Research is urgently needed t o determine the b e s t way o f detecting smoke in a corridor when t h e f l r e o r ig ina t e s inside a bedroom with the door closed.

4 . F u r t h e r research is requlred to investigate the contribution 0%

heating and arirconditLonPng system t o the movement of smoke o r i g i n a t i n g from flaming o r srnoulderlng f lres.

REFERENCES

R.G. Bright , "Recent Advances in Residential Smoke Detection", Fire Journal, 6 8 ( 6 ) , 69-77 ( 1974 ) , - J. Ranks and R.L. Rardin, " In t e rna t l sna l Comparison of Fire Loss", Fire Tech. , - 18 (31, 268-279 (1 9 8 2 ) .

J, Banks, "Selected International Comparison of Fire Loss, 1979- 1980", Fire Journa l , - 77 (13 , 36-41, 61-65 (1 983) .

R,L. Custer and R.G. Bright, "Fire Detection: The State-of-the- kt", Natlonal Rureau of S t a ~ d a r d s , Center f o r Fire Research, Washington, D.C., Tech. Note 839 (19741.

J.C. Mollere, M. Pourprlx and G. Roux, "Characteristics of Combustion Aerosols", F i r e International, 6 ( 6 93 , 73-80 (1 9803. -

T.Y. King, "Empirical Relationship Between Optica l Density and Mass Density of Smoke", J o u r n a l of F i r e and Flammability, 6(2), - 222-227 (1975).

C.P. Bankston, "Detailed Measurement of P h y s i c a l Characteristlcs of Smoke Particles Generated by Flaming Material", Journal of Fire and Flammability, - 8 ( 4 ) , 395-411 (1977).

T.G. Lee and G. Hulholland, "PhysFcal Properties of Smokes Perttnent to Smoke Detector Te chnology" , National Rureau of Standards, Center f a r Flre Research, Washington, D . C . , h e . Report 77-1312 (1977).

J.W. Northey, "Smoke Detectors: O p t i c a l or Ionizatian", P i r e Surveyor, - 9 ( 2 ) , 20-23 ( 1 980).

G . Mulholland and T.3. Ohlemiller, "AerosoL Characterlzatlon of a Smouldering Source", Aerosol Science and Technology , 1, 59-71 - ( 1 9 8 2 ) .

J . Kelley, "Optlcal or I o n i s a t i o n Detectors? The Case for a Combination in High L l f e Risks", Fire, *(913), 93 ( 1 9 8 1 ) .

E.E. Solomon and M.H. Reiss, "The Requirement and Design f o r Combined Tsn l za t i on J~ho toe l ec t l r l c Smoke ~etectors", Fire Journal , 73(5), 45-55 ( 1979 ) . - T.E. Waterman, S.W. Harpe and W.J. C h t - L s t i a n , "An Engineering Approach to the PosLtioning of F i r e Detectors I n Residences", Soc le ty of Fire Protectlon Engineers, Technology Report 6, (1977).

R-W, Rukowski, "Investigation of the Ef fec t s of Hearing and A i r Canditiontng on the Performance o f Smoke Detectors in Yobile Homes", N a ~ i o n a l Bureau of Standards, Center for F i r e Research, Washington, D.C., Int. Rep. 1915, (1979).

G.L. Coulter, " F u l l Scale F l r e Tests Lead t o Development of New Standards for Smoke Detector Poslt Lonirig add Smoke Js olatlon", Constructfon Speclfler, - 27(2), 20-27 ( 1 9 7 4 ) .

R.H. Kennedy, K.W.P. Riley and S . P . Rogers, "A Study of t h e Operation and Ef f ectiueness of Fire De tectors Installed .Ln t h e Bedroom and Corr idors of Residential Lnstitutlons", Butlding Research Estlablishaent, UK, Building Research Es t ablishment Current Paper, CP 26 (1978).

A.S. Eastwell, "Calculation of the Effec t iveness of Cor r ido r Smoke Detectors in Detecting Fires In MjolnLng R~oms", Flre Prevention Scfence and TechnoLogy (201, ( 1 9 7 8 ) .

5. P l a t t , "Automatic Flre DetectLon: W i l l Recent Advances Save Lives", Fire, - 74(913), 11-73 (1981).

R.W. Bukowski, " F i e l d Investigation of Residential Smoke Detectors", National Bureau of Standards, Center far Fire Research, Washington, D.C., h r , Rep. 1126, (1976).

E.H. Nober, H, Peirce and A. Well, "WakFng Effectiveness of Household Smoke and Fire Detec tkon Devlccs", F i r e Journal - 7 5 ( 4 ) , 86-91, 130 (1981).

W.G. Zimmerman, "Sleep MentatPon and Auditory Awakening Threshulds'", Psychophysiology , - 6(5), 540-549 (1 970).

H.L. Williams, J.T. Hammock, R.L. Paly, W.C, Dement and A. tub in , "Response to Auditory S t f m l a t t o n , S l e e p Loss and the EEG Stages of Sleep", Electroencephalography and C l l n i c Neurophysiology, 16, 269-279 ( 1964).

H.L. Wflliams, "Effects of Noise on Sleep: A Review", Proceedings of International Congress on NoIse Health Problems (W.D. Wand, e d . ) , Environmental Pro tec t ion Agency, Washington, D.C., No. 550J9-008, 501-511 (1973).

E.W. Nober, H . Fferce, A, Well, C.C. Johnson and C. Cllfton, "Waking Effectiveness of Household Smoke and F i r e Detection DevLces", NatLonal Bureau of Standards, Washington, D.C., GCR-80-284 (1 980).

J.S. Lukas and M.E. Dobbs, "Effects of Gtrcsaft Noise on the S leep of Women", National Aeronautics and Space Adrn-inLstratio~, Contract Report 2041 (1972).

R.W. Bulcowski, " F i e l d Investigation of Residential Smoke Detectors", Fire Journal, - 71 (2), 18 (1 977).

27. C. Pike, "Early Warnlng CapabilLtles and Limitat- ions", Fire Protection, E ( 5 3 0 1 , 19, 22, 24 (1981).

28. R.L. Nailen, "Why Smoke Detectors Don't F u l f i l l Great Expectations For Saving Lives", Fire Engineering, 1 3 3 ( 9 ) , 47 ( 1 980).

TAELE 1 Combustible Materials Tested by Molliere et a l e 5

Fire No.

E t h y l alcohol Wood, s l e w combustion Wood, qulck combus tion Cardboard i n s i d e DEF smoke generator discharger Cardboard l n s l d e box Polyurethane foam PYC cable (by overheating) Neoprene cable (by overheating) Cotton P e t r o l Transformer overload Acetone Damp wood, quick combustion Methyl alcohol Diesel 011 Wood, qulck combustion Rubber cable Damp wood, s low combustion Sodium Magnesf urn Gauchard aerosol generator Geosyl aerosol generator Foam

a 0 0 0 0 a L n -s m cu -

S N O S ~ ~ ~ a o e aoo r r s t c lv la 3 8 1 4

-

r 2,- u -z o w - II: - m 4 y 2 g 024 - d m 0 3 ,,, u 'd a z c

u - ..2 T cc - W 3 - 5 d = L

.u, zz-. u - In- 3 5 z! 5 : ~

L? - L c 2 kZE

L I G H T

. _ _ . _ - - - ._ - _ ' . . , . . -.. .. - :. .-

F I G U R E 4

PHOTOFLEGTRI C - T Y PE D E T E C T O R { F R O h l N O R T H E Y )

F I G U R E 5

I O N I Z A T I O N - T Y P E D E T E C T O R [ F R O h l N O R T W E Y l