Beruflich Dokumente
Kultur Dokumente
1/CE, 1998
1.
Introduction
Design earthquakes, winds, and floods are based on risk and probability factors. These factors have
been developed and refined over the years from major disasters which have occurred. As hazard =
risk x probability, then values for risk and probability for design earthquake, winds, and floods can be
readily identified in the national design codes.
Not so for fire. Statistics are as yet not gathered in a readily convenient form to enable fire risk
or probability to be quantified. Human risk would include death or injury of building occupants or
firemen. The NZ Fire Service collects data which show that 99% of all fire deaths occur in residential
properties. Although only 1% occur in non-residential buildings, a huge amount of national effort goes
into making non-residential buildings fire safe. Property risk would involve damage to the owners
building, the building contents, and the neighbours building. The NZ insurance industry collects this
information in dollar terms, but does not publish it in a convenient statistical form for fire design
purposes. The probability of each of the property risks would need to be defined for each type of
property, in various firecell sizes, and in terms of, say, 1 in 10, 20 or 50 year risk.
Fire safety design is a new engineering discipline which is growing at a very rapid rate. No doubt
for fire design, appropriate risk and probability factors will slowly develop, as they have done for
earthquakes in New Zealand from those very elementary design approaches used in the 1930s and
1940s. Suitable risk/probability factors for use in New Zealand fire designs are not at present available
to structural engineers, though Narayanan (1) provides a close approach. Reliance has therefore to be
placed on other means which can be used to develop design fires from which the likely behaviour of
structural members in a fire can be assessed.
Design fires can be classified in two forms:
(a)
intensity
(b)
severity
The purpose of this paper is to demonstrate how both forms of design fire can be developed.
Though this paper deals mainly with steel design, the methods illustrated can be used for concrete
design and modified in various ways for timber design.
2. Fire tests
Fire tests are typically carried out in three main forms, as a standard temperature/time tests, actual
burn tests, or calorimeter tests.
Under a standard temperature/time test, a structural member is installed in a 3 m cube furnace and
subjected to a rising temperature/time curve such as that stipulated in BS 476 or ISO 834. The furnace
fuel is generally oil or gas. The members temperature response is recorded up to the point of failure,
after which the test is usually stopped and the fuel turned off. As a general rule, measurements are
taken only during the heat-up phase and not during the cool-down phase.
47
In an actual burn test, the fuel is typically wooden cribs and the test room may vary in size, type of
construction, and degree of openings. As the experiment progresses, a range of measurements are
taken such as time, weight loss, temperature, radiation, gas properties, etc. Both heat-up and cooldown phases can be measured.
In a calorimeter test, one or more full sized products such as a chairs, sofas, or even motor vehicles
are ignited beneath a hood which collects all the fire gases from which the heat release rate with time
is determined. Weight loss, temperature, etc., can also be measured.
FIGURE 1: Mass loss, rate of burning, and temperature results for CIB Test No. 205.
Also of note are the references to the rate of weight loss shown on a typical data form as R80/55, R55/30
and the time to reach 80% of the original crib weight. One of the CIB test conclusions was that
burning averaged over the 80% to 30% crib weight region could be expressed as a rate of pyrolysis
formula:
R80/30 = kp.(AV.hV) = kp.FV
kg/s
(1)
where: R80/30 = rate of pyrolysis between 80% and 30% of the original crib weight in kg/s; kp =
pyrolysis coefficient as per Eq. 3; AV = area of openings in m; hV = height of openings in m; FV =
ventilation factor (= AV.hV) in m5/2.
In its most simple form and in minute units, the pyrolysis coefficient becomes 5.5, that is:
R80/30 = 5.5 (AV.hV ) = 5.5 Fv
kg/min
(2)
This simplistic formula appears time and time again throughout fire engineering literature, but
should be used with some caution, as the pyrolysis coefficient kp does in fact vary with the geometry of
the fire test compartment. It may be as low as 3 for small spaces, to 10 or more for large spaces.
A curve fit from Fig. 5 of Ref. 3 can give the pyrolysis coefficient kp as:
kp =
1
106. FO3 + 5
(3)
where: FO3 = opening factor = FV in m3/2; FV = ventilation factor (= Av.hv1/2) in m5/2; AT3 = total
A T3
internal surface area of the fire compartment (not including floor area AF or the vertical openings AV)
in m, and hv = height of vertical opening in m.
Once R80/30 has been derived from kp and if the nett calorific value of the fuel Hn is known in
MJ/kg, then the total energy E in the firecell and the fire intensity Q80/30 can be derived from:
E = Hn.B.
MJ
(4)
Q80/30 = Hn.R80/30
MW
(5)
48
If the original total mass of fuel B in kg and the total duration of the fire td in seconds are known,
the average rate of pyrolysis R100/0 and average energy release rate Q100/0 can be derived as follows:
R 100 / 0 =
Q 100 / 0 =
B
td
E
kg/s
(6)
MW
(7)
td
The comparisons between R actual, R80/30 and R100/0 for Test No. 205 are illustrated in Fig. 2. A
similar comparison applies to Q actual, Q80/30 and Q100/0 as illustrated in Fig. 7. It will be obvious from
Figs. 2 and 7, one cannot apply the 80/30 rate for the whole duration of the fire. The 80/30 rate applies
only to the burning of half the fuel. Rates of 100/80 and 30/0 apply before and after the time period
that the 80/30 rate covers.
FIGURE 2: R Actual, R 80/30 and R 100/0 curves for CIB test No. 205.
If the rate of pyrolysis and energy release rates can thus be derived from the original fire test data, a
series of graphs can be put together which will give a useful indication of the fires performance. Fig.
3 shows such expanded data for CIB Test 205.
(logt - logt p ) 2
+ T1
f
T2 = T3 . e
(8)
where: T1 = ambient temperature (C); T2 = temperature at any time t (C); T3 = maximum temperature
generated (C); t = time from start of fire (min); tp = time at which T3 is reached (min); and f = growth
factor fg, or decay factor fd.
50
To illustrate the potential of BFD Curves as a design tool, two of the nine temperature time tests
carried out in the Cardington large-scale fire tests are reproduced as Fig 5, along with the BFD Curves
modelling the test results. As can be seen the calculated fit is very close to the test data and certainly
sufficient for most structural engineering design purposes. Examples of BFD Curves used as design
curves are illustrated in Fig. 10 (a), (b) and (c).
Working downwards from the intensity graph, the designer can then produce a severity graph. One
method is to introduce the fire intensity data into the FIRE SIMULATOR computer program in
FPETOOL (15). From this program temperature/time data can be produced. Another is to use the
BFD Curve method (9). From the severity graph, the designer can then derive the temperature
response of the selected structural member and compare it with the arbitrary 550C limit to determine
the time of collapse tc. By changing the structural member size and thickness of fire insulation, the
designer can modify the design to produce the most economic result to meet the F or S ratings.
As a further step, the designer can also determine the member temperature response curve to an
ISO 834 severity fire and hence determine an equivalent time te. The time of collapse tc and the
equivalent time te can then be compared directly.
For a more sophisticated design, the designer can determine the actual critical temperature (if it
differs from 550C) and produce the load capacity versus applied load graph. Methods for doing this
are covered in Refs 16 and 17.
FIGURE 6: Typical fire design approach for a structural steel member (15).
FIGURE 7: Comparison between an actual fire and 80/30 and 100/0 design fires.
FIGURE 8: Various types of design fires. The area under each curve represents the total fuel
energy released in MJ (18).
53
The t2 fires use a growth coefficient (a1) and a decay coefficient (a2). These are chosen by the
designer from experience. The arithmetic for the t2 curves is easy as it is based on the algebra for
parabolas. All the design fires illustrated in Fig. 8 can be readily modelled on a computer spreadsheet
Once the design fires have been determined they can be readily introduced into FPETOOL (15) or
FIRECALC (19) as desired. FPETOOL and FIRECALC are in regular daily use in New Zealand.
6. Conclusions
To structural engineers, the benefit of using design fires based on the BFD Curve and ISO Curve
methods is considerable. With suitable computer spreadsheets able to produce instant graphs, a
designer can not only explore a wide range of structural sizes and insulation thicknesses very quickly,
but can also carry out a sensitivity analysis on the selected design values of the peak temperatures and
the growth and decay coefficients.
54
FIGURE 10(a) Typical design approach to determine critical time tc and equivalent time te for a high
ventilation fire. Note that the peak temperature occurs at the same time as the peak burn rate period.
occurs.
55
FIGURE 10(b): Typical design approach to determine critical time tc and equivalent time te for a
medium ventilation fire. Note that the peak temperature occurs at the end of the steady state burning
period.
56
FIGURE 10(c): Typical design approach to determine critical time tc and equivalent time te for a low
ventilation fire. Note that the peak temperature occurs at the end of the steady state burning period.
57
7. References
1. Narayanan, P (1995) New Zealand Fire Risk Data, BRANZ Study Report No. 64, Building Research
Association of New Zealand, Wellington, New Zealand.
2. Thomas, P H (1974) Fires in Model Rooms - CIB Research Programmes, Building Research Establishment,
Fire Research Station, Borehamwood, CP 32/74.
3. Thomas, P H and Heselden A J M (1972) Fully Developed Fires in Single Compartments CIB Report No. 20),
Building Research Establishment, Fire Research Station, Borehamwood, UK.
4. Harmathy, T Z (1972) A New Look at Compartment Fires Part 1 Fire Technology, 8(3).
5. Buchanan, A H (Editor) (1994) Fire Engineering Design Guide, Centre for Advanced Engineering, University
of Canterbury, New Zealand.
6. Pettersson, O, Magnusson, S E and Thor, J (1976) Fire Engineering Design of Steel Structures, Publication
50, Swedish Institute of Steel Construction, Stockholm 1976.
7. Eurocode (1993) - Actions on Structures Exposed to Fire, Eurocode Actions on Structures, Part 10, Feb 1993
Draft.
8. Kirby, B R, et al (1994) - Natural Fires in Large Scale Compartments- A British Steel Technical, Fire
Research Station Collaborative Report, British Steel Technical, Swinden Laboratories, UK, 1994.
9. Barnett, C R (in press) BFD Curves Theory to be published.
10. Gamble, W L (1989) Predicting Protected Steel Member Fire Endurance Using Spread-sheet Programs,
Technical Note, Fire Technology, 25(3): 256-273).
11. Wade, C A (1991a) Method for Fire Engineering Design of Structural Concrete Beams and Floor Systems,
BRANZ Technical Recommendation No. 8, Building Research Association of New Zealand, Wellington, New
Zealand..
12. Wade, C A (1991b) Fire Engineering Design of Reinforced and Prestressed Concrete Elements, BRANZ
Technical Recommendation No. 8, Building Research Association of New Zealand, Wellington, New Zealand.
13. Fleishmann, C (1988) Analytical Methods for Determining Fire Resistance of Concrete Members. Section 3,
Chapter 7, SFPE Handbook of Fire Protection Engineering, Society of Fire Protection Engineers, Quincy,
MA, USA.
14. Barnett, C R (1992) Macbar Fire Design Code - Code of Practice for Structural Fire Design of Buildings,
Training Manual No. 1, New Zealand Fire Protection Association Auckland, New Zealand.
15. Center for Fire Research (1994) FPETOOL - Fire Engineering Computer Programs, Center for Fire
Research, National Institute of Standards and Technology, Gaithersburg, MD, USA.
16. European Convention for Constructional Steelwork (1985)
Design Method on the European
Recommendations for the Fire Safety of Steel Structures. Report R35, European Convention for
Constructional Steelwork, Brussels, 1985.
17. Milke J (1988) Analytical Methods for Determining Fire Resistance of Steel Members, Section 3, Chapter 6,
SFPE Handbook of Fire Protection Engineering, Society of Fire Protection Engineers, Quincy, MA, USA.
18. Barnett, C R and Simpson M R (1995) Fire Code Review - New Zealands Performance Based Fire Code,
Proceedings, Asiaflam, Hong Kong, Mar 1995, pp 27 - 40.
19. CSIRO Division of Building Construction and Engineering. (1991) FIRECALC - Fire Engineering Computer
Programs, CSIRO Division of Building Construction and Engineering, North Ryde, NSW, Australia.
20. Law, M and OBrien, T (1983) Fire Safe Structural Steel A Design Guide. American Iron and Steel
Institute, Washington DC, USA.
58