Fire Safety Journal, 10 (1986) 67 - 68 67

Letter to the Editor

Comments on "Fire Sizes and Sprinkler Effectiveness in Offices - - Implications for Smoke Control Design"

M. LAW

Ove Arup Partnership, 13 Fitzroy Street, London WIP 6BQ (U.K.)

(Received August 5, 1985)

The paper by Hansell and Morgan on fire sizes in offices [ 1] gives data which are of interest in relation to fire grading and the assessment of the benefits of sprinklers. For large fires, that is, ones of area 10 m 2 or more, the data fit simple exponential relationships, for the propor t ion P (%) exceeding a given area A (m2):

P = 60A -°'63 for sprinklered offices

P = 180A -°'Ts for non-sprinklered offices

However when there are no sprinklers, the data for dayt ime fires are significantly differ- ent f rom those for night-time fires giving approximately:

P = 130A -°'~s day

P = 340A -°'Ts night

Making the assumption that most people are in the building during the day, it can thus be said that sprinklers offer less benefit , in terms of fire size, when people are about the building than when they are not. Whether sprinklers are of benefit to life safety is diffi- cult to ascertain f rom these data. There is no correlation between numbers of deaths or casualties and final fire area.

The data relate to mainly non-atrium build- ings. The authors suggest tha t fire spread would be similar in buildings with atriums, but pre- sumably this would be affected by the nature of the enclosure overlooking the atrium. It would be interesting if the authors were able to extend their calculations of fire behaviour to deal with this point. However, if escape takes place during the early stages of a fire, then the final fire size at tained is less relevant,

SO far as the design of smoke control for pro- tect ion of escape routes is concerned.

The authors discuss the concept of a design fire size, but it is not obvious that there should be a single value. If life safety is the object, it could be related to the number of people at risk, giving a higher standard of safety for a high-rise building, for example. Neither is it obvious that the relative f requency adopted for shopping centres is directly relevant to offices: the type of people, the fire load and the escape provisions are all different, as the authors admit. Nor is the shopping centre value based on an assessment of acceptable risk to life. It should also be noted that the 0 - 4% relative frequency quoted for shopping centres is from a paper by Morgan and Chan- dler in which these authors explain that it is based on a small sample and an " informed guess".

Turning to the discussion of fuel bed/venti- lation controlled fires: eqn. (1) gives the effec- tive durat ion of the fire assuming a total burn out of the fuel, while the t quoted from the paper by Thomas and Theobald is the actual durat ion of a fire a t tended by the fire brigade, no t necessarily a total burn out. Indeed this point is ment ioned later in Section 6.2.2. Thus one t cannot necessarily be substi tuted for the other. In addition, the second t is based on industrial fires, not office fires.

However, accepting the suggested values of 30 min for durat ion and 57 kg/m 2 for fire load per unit f loor area, and assuming a calo- rific value of 13 MJ/kg, as used by Thomas and Theobald, the heat ou tpu t at the demar- cation between ventilation control and fuel bed control is given by:

57A~ Q ~ - x 1 3 x 1 0 3 =411Af (kW)

3O X 60

This value of Q~ conflicts with Section 6.2.2, in which it is suggested that for fuel- bed control , at the same fire load per unit f loor area, Qf = 260Af (kW). Can the authors explain the reason for the difference ? It also appears that they have misinterpreted the graph by Theobald, which they reproduce as

0379-7112]86]$3.50 © Elsevier Sequoia/Printed in The Netherlands

68

Fig. 3. This graph is fo r fire load per uni t area covered by fuel, which in the crib exper iments was 8 .65 m 2. The to ta l f loor area, and the fire area, were 28.5 m2; the values of fire load gave 7.5, 15, 20 and 60 kg/m 2 f loor area, bu t 25, 5 0 , 1 0 0 and 200 kg/m 2 fuel area. It is the la t ter values which are shown in Fig. 3. There- fore a value o f 57 kg/m 2 f loor area canno t be en te red di rec t ly in Fig. 3. I t should be en te red near the top end o f the curve. Qf then be- comes d i f fe ren t f r om the value quo t ed by Law, which was ci ted by the au thors as corro- bora t ion . (Law's value relates to a fire load per uni t f loor area o f a b o u t 25 kg/m 2 in terms o f w o o d cribs.)

The au thors suggest t ha t one th i rd of the hea t genera ted would be carried by the gases leaving the c o m p a r t m e n t . Heselden [2] , anal- ysing the same crib exper iments showed tha t the p r o p o r t i o n o f hea t in the e f f luen t gases

varied with A u N / H and that , in these experi- ments at least, the p r o p o r t i o n es t ima ted varied f rom 47% to 67%. Would the au thors explain w h y these values were n o t t aken in to acco u n t ?

To sum up, it seems tha t more considera- t ion is needed be fo re these da ta cou ld be appl ied to the design of smoke con t ro l systems in of f ice buildings, with or w i t h o u t atr iums.

REFERENCES

1 H.P. Morgan and G. O. Hansell, Fire Safety J., 8 (3) ¢1984/85) 187.

2 A. J. M. Heselden, Parameters determining the severity of fire, Paper No. 2 of Symposium No. 2, Behaviour of Structural Steel in Fire; Ministry of Technology and Fire Offices' Committee Joint Fire Research Organization, London, HMSO, 1968.