Professional Documents
Culture Documents
TECHNICAL NOTE
2 OBJECTIVES .............................................................................................................. 9
3 DEFINITIONS........................................................................................................... 10
4 APPLICABILITY ...................................................................................................... 13
4.1 General 13
4.2 Probabilistic Criteria 13
4.3 Single Value Criteria 14
4.3.1 Individual Risks 15
4.3.2 Group Risks 15
7 EXPLOSION OVERPRESSURE.............................................................................. 39
7.1 Introduction 39
7.2 Background 39
8 SMOKE...................................................................................................................... 59
8.1 Introduction 59
8.2 Composition of Smoke 59
8.3 Effects of Carbon Monoxide 60
8.3.1 Effects of Carbon Dioxide 62
8.3.2 Occupational exposure limits for CO2 65
8.4 Effects of Oxygen Starvation 66
8.5 Effects of High Temperature 66
8.6 Criteria 67
9 ASPHYXIANTS ........................................................................................................ 68
9.1 Literature Review 68
11 REFERENCES........................................................................................................... 74
The following table summarizes the main recommendations in this Technical Note, but the
reader is advised to confirm that any value in the table below is appropriate for the work being
carried out by reading the background for the selection of these values before using it in client
work.
The main updates in revision 3 of this technical note include the following.
o Updates related to effects of Carbone Monoxide (section 8.3)
o New section related to effects of Carbone Dioxide (section 8.4)
o Calculation of fatalities for offshore facilities (section10)
o Updated suggested approach to impairment and incapacitation due to smoke (app.F)
1
For evaluation of fatalities on Offshore structures, the criterion for onshore “ Onsite” applies.
2
. These fatality rates applies for onshore sites and includes the effect of the explosion pressure only, In a VCE, almost
everyone in the flammable portion of the cloud will be expected to die, due to flame contact, regardless of the
overpressure generated No fatalities related to the VCE cloud it self (heat) or effect of potentially secondary fires due to
rupture of equipment or piping containing flammable materials. These effects must be included additionally.
3
The inside figures are based on the fatality probability of a person in a typical brick dwelling (/100/). Special explosion
rated buildings (e.g. control rooms, special process buildings etc) has to be evaluated separately with respect to indoor
fatality rates.
4
The effects related to neighbouring areas are very dependant on the physical layout of the installation and should be
subjected to a separate evaluation for each installation In many cases an area classed as neighbouring might not be
separated from the initial area by a barrier which reduces the overpressure to a point where it is negligible. Separate CFD
analysis could be used for this purpose. In the absence of CFD use can be made of TNO curves or simpler rule sets..
5
These are figure used in a number of DNV studies and should be used as indicative. There exists no specific reference to
this type of fatality fraction but the fraction should be subjected to a platform individual evaluation.
follow and participate in the discussions which might develop thinking on human impact from
toxic, flammable and explosive materials.
The probit relationship used in any particular analysis should be discussed and agreed with a
client before carrying out the analysis. If a client requests the use of impact criteria which are
different from those recommended in this technical note, then they should be used (although it
may be appropriate to state in the report that the relationship used is not that recommended for
use by DNV internal guidance. It may also be that offices carrying out work for clients where
the ambient temperature is closer to 25C than 10C would wish to use the probits with a
reference temperature of 25C; this is perfectly acceptable.
2 OBJECTIVES
The objective of this Technical Note is to provide a consistent set of human impact criteria for
use in Quantified Risk Analysis (QRA) both onshore and offshore in DNV. Some
information on structural impact is also included primarily because it may result in a
secondary hazard to humans in the vicinity (eg. from falling debris, etc). For offshore studies,
TN 8 should be consulted for further information regarding structural impact and impairment
of safety functions.
The impact criteria contained in this Technical Note relate to impact from:
• toxic gases and vapours;
• short and long duration thermal radiation;
• overpressure; and
• smoke.
The background and recommendations for both single value and multiple value criteria are
given for these hazards.
3 DEFINITIONS
The following are working definitions of terms commonly used in human impact studies.
Probit
Y = A + B Ln (Cn.t)
Y is the probit (or probability measure), A, B and n are constants, C is a concentration and t is
time. The probit, Y, has the properties of the Normal distribution, with a Variance of 1 but a
mean of 5. Its probability values (eg. percentage fatality) can be determined by reference to
any Normal distribution table (single-tailed) by adding 5 to the standard deviation value, or by
reference to published tables (/56/). The most commonly used table is given as Table I.1 in
Appendix I. For further details of probit analysis methodologies see Appendix I.
LD50
A dose which will cause the deaths of 50% of an exposed population. The dose is a function
of concentration raised to some power and duration of exposure. It is not, in this context, the
amount taken into the body expressed in terms of mg/kg of body weight.
LDLo
The lowest published dose to cause a fatality.
LCLo
The lowest published concentration which, for a given duration of exposure, caused a fatality.
Toxic dose (TD)
TDLo
The lowest published toxic dose.
Other Concentrations
TLV(/57/)
Threshold limit values published by the American Conference of Governmental Industrial
Hygienists in 1990. The values given are normally 8 hour time weighted averages (TWA) for
a working lifetime. Shorter periods are also specified though less commonly used: Short
Term Exposure Limit (TLV-STEL) and Ceiling (TLV-C).
It is normally assumed that the SLOT/Dangerous Dose is 1% mortality. Thus the SLOT-value
is the toxic load, Cnt, at which 1% mortality is observed.
4 APPLICABILITY
4.1 General
The impact criteria in this TN are applicable to all QRAs in which the effect on people from
an accidental release of hydrocarbon or other hazardous material (excluding nuclear radiation)
is to be evaluated.
Pr = A + B ln C nt
Where
A,B and n are constants defined for each material (unit specific)
Probits for thermal exposure and harm from explosion overpressure follow the same general
form.
Since the probit values express the relations ship between the dose exposure and expected
human response in terms of a normal distribution, N(5,1), the translation from probit to
percentage of the population affected is not obvious. The following calculation guides may
be helpful:
1/n
2.67 - A
C=
e B
t
2.67 - A
e B
t= /3/
C
1/n
5B- A
C=
e
t
5- A
eB
t= /5/
C
Single value criteria described here may have been derived from a probit relationship
according to the method described in Appendix G. If a more detailed approach to fatality
impacts is chosen (e.g. multiple exposure levels or a more detailed approach to societal risk),
the relevant probit should be consulted.
The use of single-value criteria is coarse and the values recommended here are by no means
universally accepted. The aim is to define these criteria in a systematic way so that QRAs are
performed in a uniform manner by DNV.
It may be argued that personnel on offshore platforms are relatively young and fit, have better
health and health care etc. compared to an average group of people in society (and therefore
are less susceptible to harm that the average person), and old people are more susceptible.
The derivation of a single value criterion generally ignores this distinction.
Y = A + B ln(Cnt)
Where Y is the probability measure, A, B and n are constants, C is the concentration either in
mg/m2 or in ppm, t is the time in minutes or seconds. DNV generally uses the units of ppm
and minutes.
There are three main publishers of toxicity data and probits referenced in this TN, the Purple
Book (PSG 3 /97/), and the Handleiding Risicoberekeningen /96/(referred to as the ‘New
Purple Book’), the HSE (/91/ dated 2008) and the CCPS (/62/, dated 1980). In Table 5-1, an
overview is given of probits from these three sources, PhastRisk 6.53.1 and the Northern
Belgian required probits (/92/). For each substance the probit recommended for use in DNV is
highlighted (in bold).
Handleiding Risicoberekeningen (/96/() and the CCPS (/62/) give probits for some substances
directly. The HSE publishes toxicity data in the form of SLOT and SLOD values (/91/),
which have been used to derive probit constants.
Different sources use different units for the concentration (mg/m3 or ppm). The concentration
in ppm is dependent on the temperature. The method to convert between units is given in
Appendix D of this TN. It should be noted that in PhastRisk 6.53.1, the probits are given for
concentrations in ppm at 9.85°C.
In the previous revision of this TN, the recommended probit was selected based on different
probits found in the literature and used in previous DNV reports. In this edition, the
recommended probit is chosen based on the following:
If the substance is not included in the Handleiding Risicoberekeningen (/96/(), but HSE has
SLOT and SLOD values for a substance (/91/). The recommended probit is calculated based
on the HSE data.
If a material is in neither the Handleiding Risicoberekeningen (/97/) nor the HSE (/91) then
the CCPS probit is used (/62/).
It should be noted that some countries have regulations as to which probits should be used. In
Netherlands, the probits in the Handleiding Risicoberekeningen (/97/) are mandatory. In
Northern Belgium the probits listed below published by the “Departement Leefmiljeu, Natuur
en Energie” (/92/) are mandatory. For a complete overview of the Northern Belgium probits
see /92/.
Issued for RiskNet HUMAN IMPACT CRITERIA Ref: T15 Revision No. 03
Table 5-1 Overview of probits from different sources with DNV recommendation in boldface
Substance New Purple Book (/96/) HSE (/91/) CCPS (/62/) PhastRisk 6.53.1 Northern Belgium (/92/)
A B n ppmn*min C in ppm C in ppm @ 25°C C in ppm @ 9.85°C C in mg/m3
ppm
MW Recom- ppm @
Name g/mol mendation mg/m3 @ 25°C 9.85°C SLOT SLOD A B n A B n A B n A B n
Acetonitrile 41.7 HSE 8.10E+04 1.60E+05 -36.02 3.42 1
Acrolein 56.1 PB -4.1 -3.27 -3.22 1 1 4.20E+02 1.68E+03 -7.48 1.68 1 -9.931 2.049 1 -3.21 1 1 -11.7 2 1
Acrylamide 71.08 HSE 1.30E+05 5.20E+05 -17.12 1.68 1 -10.26 1.678 1
Acrylonitrile 53.1 PB -8.6 -7.59 -7.52 1 1.3 9.60E+03 2.52E+04 -19.47 2.41 1 -29.42 3.008 1.43 -7.52 1 1.3 -7.27 0.86 1.3
Allyl Chloride 76.5 HSE 7.80E+04 3.12E+05 -16.26 1.68 1 -20.20 1.82 1.1
Allylalcohol 58 PB -11.7 -9.97 -9.86 1 2 6.30E+03 2.52E+04 -12.03 1.68 1 -15.1 2 1
Ammonia 17 PB -15.6 -16.33 -16.21 1 2 3.78E+08 1.03E+09 -43.24 2.32 2 -35.9 1.85 2 -16.21 1 2 -34.70 1.85 2
Aniline 93.1 HSE 1.16E+04 4.66E+04 -13.01 1.68 1 -16.9 2 1
Arsine 78 PB -11.2 -8.88 -8.78 1.61 1.24 3.71E+03 5.90E+04 -4.25 0.84 2 -11.20 1.61 1.24
Azinphos-methyl 317.3 PB -4.8 0.33 0.43 1 2
Benzen 78.1 CCPS low acute toxicity not relevant to assessment work -109.78 5.3 2 -109.8 5.3 2 -116.00 5.3 2
Benzyl Chloride 126.6 HSE 2.25E+03 9.00E+03 -10.30 1.68 1 -10.28 1.678 1
Bromine 159.8 PB -12.4 -8.65 -8.54 1 2 1.60E+01 2.60E+01 -10.64 4.80 1 -9.04 0.92 2 -12.36 1 2
1,2-dichloroethane 99 HSE 9.00E+04 3.60E+05 -16.50 1.68 1
Carbon Disulfide 76.14 HSE 9.60E+04 3.84E+05 -16.61 1.68 1 -16.52 1.674 1 -23.44 2.33 0.86
Carbon Monoxide 28 PB -7.4 -7.26 -7.21 1 1 4.01E+04 5.70E+04 -67.68 6.64 1 -37.98 3.7 1 -7.21 1 1 -17.9 1.43 1.4
Carbon Tetrachloride 153.8 HSE 7.20E+05 2.60E+06 -21.80 1.81 1 -6.29 0.408 2.5 -17.9 0.71 2.84
Chlorine 70.9 PB -6.35 -4.89 -4.81 0.5 2.75 1.08E+05 4.84E+05 -15.33 1.55 2 -8.29 0.92 2 -4.81 0.5 2.75 -10.31 0.92 2
Crotonaldehyde 70.1 HSE 3.19E+03 1.05E+04 -13.13 1.96 1 -12.83 1.72 1.16
Epichlorohydrin 92.5 HSE 1.04E+05 1.30E+05 -117.95 10.44 1 -20.28 2.38 1
Ethylene dibromide 187.9 HSE 1.60E+04 4.44E+04 -19.43 2.28 1 -16.30 1.73 1.16
Ethyleneimine/Aziridine 43.1 PB -11.6 -10.47 -10.36 1.77 1.13 1.16E+03 8.25E+03 -5.69 1.19 1 -11.60 1.77 1.13
Ethylene Oxide 44.1 PB -6.8 -6.21 -6.15 1 1 4.68E+04 1.87E+05 -15.40 1.68 1 -6.15 1 1 -18.6 2 1
Fluorine 38 HSE 3.80E+05 1.50E+06 -19.13 1.70 2 -9.15 1 2
Formaldehyde 30 HSE 5.70E+03 8.10E+03 -54.67 6.63 1 -12.24 1.3 2 -12.53 1.3 2
Hydrazine 32.05 HSE 1.51E+04 6.05E+04 -13.48 1.68 1 -13.45 1.676 1 -19.02 2.48 0.81
Hydrogen Chloride 36.5 PB -37.3 -35.82 -35.62 3.69 1 2.37E+04 7.65E+04 -17.36 1.99 1 -16.85 2 1 -15.69 1.69 1.18 -18.20 2 1
Hydrogen Cyanide 27 PB -9.8 -9.56 -9.43 1 2.4 1.92E+05 4.32E+05 -32.28 2.87 2 -29.42 3.008 1.43 -5.8 1.08 1.85 -81.5 6.7 2
Hydrogen Fluoride 20 PB -8.4 -8.7 -8.62 1 1.5 1.20E+04 2.10E+04 -36.44 4.16 1 -25.87 3.345 1 -8.62 1 1.5 -8.40 1 1.5
Hydrogen Peroxide 34.01 HSE 8.60E+04 3.44E+05 -16.43 1.68 1 -16.39 1.678 1
Hydrogen Sulphide 34.1 PB -11.5 -10.87 -10.76 1 1.9 2.00E+12 1.50E+13 -30.08 1.16 4 -31.42 3.008 1.43 -8.53 0.44 4.55 -32.92 3.01 1.43
Methanol 32.04 HSE 8.02E+05 2.67E+06 -23.67 1.94 1 -6.347 0.664 1
Methyl Bromide 95 PB -7.3 -5.81 5.75 1 1.1 3.00E+04 4.80E+04 -48.44 4.96 1 -56.81 5.27 1 -5.74 1 1.1 -64.42 5.27 1
Methyl Isocyanate 57.5 PB -1.2 -0.6 -0.57 1 0.7 7.50E+02 1.68E+03 -16.46 2.89 1 -5.642 1.637 0.653 -0.57 1 0.7
Methyl Mercaptan 48.1 PB -17.8 -16.44 -16.33 2.05 0.98
Nitric Oxide 30.01 HSE 2.09E+04 2.43E+04 -151.10 15.46 1 -150.8 15.43 1
Nitrobenzene 123.1 HSE 8.54E+04 3.41E+05 -16.44 1.68 1 -19.13 2 1
Nitrogen dioxide 46 PB -18.6 -16.26 -16.06 1 3.7 9.60E+04 6.24E+05 -11.61 1.24 2 -13.79 1.4 2 -16.05 1 3.7 -15.65 1.4 2
Issued for RiskNet HUMAN IMPACT CRITERIA Ref: T15 Revision No. 03
Substance New Purple Book (/96/) HSE (/91/) CCPS (/62/) PhastRisk 6.53.1 Northern Belgium (/92/)
n
A B n ppm *min C in ppm C in ppm @ 25°C C in ppm @ 9.85°C C in mg/m3
ppm
MW Recom- ppm @
Name g/mol mendation mg/m3 @ 25°C 9.85°C SLOT SLOD A B n A B n A B n A B n
O-cresol 108.1 HSE 1.20E+03 4.80E+03 -9.25 1.68 1 -15.50 1 2
Parathion 291.3 PB -6.6 -1.64 -1.54 1 2
Perfluoroisobutene 200 HSE 5.40E+01 1.24E+02 -8.51 2.80 1
Phenol 94.1 HSE 1.50E+04 6.00E+04 -13.49 1.68 1 -13.46 1.678 1 -15.2 1 2
Phosgene 98.9 PB -10.6 -7.81 -7.69 2 1 1.08E+02 3.48E+02 -6.65 1.99 1 -19.27 3.686 1 -7.69 2 1 -29.60 4.55 1
Phosphamidon 299.7 PB -2.8 -1.05 -1.01 1 0.7
Phosphine 34 PB -6.8 -6.14 -6.03 1 2 6.90E+03 1.08E+04 -43.30 5.20 1 -9.06 1.63 1.23
1,2-propylene oxide 58.1 HSE 5.16E+04 2.04E+05 -15.72 1.70 1 -7.415 0.509 2 -15.69 1.692 1
Sulphur dioxide 64 PB -19.2 -16.89 -16.76 1 2.4 4.66E+06 7.45E+07 -10.23 0.84 2 -15.67 2.1 1 -16.75 1 2.4 -27.90 1.14 3.7
sulphuric acid 98.1 HSE 1.30E+04 2.08E+05 -5.29 0.84 2
Tetraethyl lead 323.4 PB -9.8 -4.64 -4.53 1 2 1.52E+03 3.79E+03 -16.01 2.55 1 -9.81 1 2
Toluene 92.14 CCPS -6.79 0.41 2.5 -6.794 0.408 2.5
2,4-Toluene diisocyanate 174 HSE 1.76E+02 4.80E+02 -9.34 2.32 1 -27.14 2.43 2
Vinyl Chloride 62.5 HSE 3.39E+06 1.36E+07 -22.55 1.68 1 -22.5 1.674 1
The minor differences between cols 11,12 and cols 17,18 are due to rounding. Most of the values in /96/ are also in /97/ and /16/
5.1.2 Single value criterion for hydrogen sulphide for use offshore
Appendix G gives a methodology for defining single value criteria based on a probit. Using
the following assumptions possible criteria are presented in Table 5-2:
• Time to escape is set (personnel have to secure their work or may not be able to
escape immediately) as either 2, 5 or 30 minutes.
• Incapacitation dose is set to 75% of lethal dose
For offshore studies using a single-value criterion for H2S, 1550 ppm is recommended based
on the use of the probit recommended for use (the Purple Book) with a an exposure duration
of 5 minutes. Note however that this value may well be somewhat higher than values derived
by other methods, and higher than a value derived from the effects given in Table 5-6.
Background
5.2.1 Ammonia
The following table illustrates the effects likely to be experienced by humans exposed to
various concentrations of ammonia (/52/). It is recognised that the concentrations for
fatalities may not agree with those predicted using the probits given above. They are useful,
however, as a general guide.
Table 5-3 Human Impact from Ammonia Exposure
Ammonia Effects Exposure Duration
Concentration (ppm)
50-53 Odour detectable by most Prolonged repeated exposure produces
persons. no injury.
100 No adverse effects for average Maximum allowable concentration for 8
workers hour working day.
300 Maximum tolerated without 1 hour.
serious disturbance.
400-700 Nose and throat irritation. Eye Infrequent, short (1 hr) exposures
irritation with tearing. ordinarily produces no serious affects.
2000-3000 Convulsive coughing, severe No permissible exposure.
eye irritation.
5000-7000 Respiratory spasm, rapid No permissible exposure. Rapidly fatal.
asphyxia.
5.2.2 Chlorine
The following table illustrates the effects likely to be experienced by humans exposed to
various concentrations of chlorine (/52/). It is recognised that the concentrations for fatalities
may not agree with those predicted using the probits given above. They are useful, however,
as a general guide.
Table 5-4 Human response to Chlorine exposure
Chlorine Effect
Concentration
(ppm)
1 Minimum concentration causing slight symptoms after several hours.
3.5 Minimum concentration detectable by odour.
4 Minimum concentration that can be breathed for 1 hour without damage.
15 Minimum concentration causing throat irritation.
30 Minimum concentration causing coughing.
40-60 Concentration dangerous in 30 minutes.
1000 Concentration likely to be fatal after a few deep breaths.
6 THERMAL RADIATION
6.1 Introduction
When flammable materials are burned a number of different forms of combustion can take
place. These, for QRAs, are commonly categorised as flash fires, pool fires, jet fires and
fireballs. This Technical Note concentrates on the effects of thermal radiation from these fire
types on humans but gives some information on the effects on structures and/or materials.
Probits are then reviewed and single impact criteria are given.
6.2 Background
6.2.1 The Effects of Thermal Radiation
The degree of harm to people exposed to thermal radiation is a function of both the intensity
of thermal radiation and the time for which the individual is exposed. As the radiation
intensity is increased, the time to feel pain reduces (see Table 6-1). Data suggests that the
degree of harm is best correlated by a thermal dose of the form:
Where
Injuries from burns result from prolonged exposure to radiation, and are usually characterised
as:
First degree burns - superficial burn giving red dry (painful) skin
Second degree burns - blister formation with the epidermis being affected
Third degree burns - full thickness burn when the dermis and nerve
extremities are affected resulting in a dry skin which
has no feeling.
Second and third degree burns can lead to disability and there is a possibility of mortality.
As the dose is increased above that required to give third degree burns, ignition of clothing
could be anticipated. Certain effects of different doses of thermal radiation are given in
Table 6-2 and effects of different heat fluxes and exposure times are shown in Table 6-1.
Based on the data given in Table 6-1, API produced recommendations for flare radiation
(excluding solar radiation) as given in Table 6-3.
Table 6-1 Exposure Time Necessary to Reach the Pain Threshold (API, /63/)
Radiation Intensity Time to Pain Threshold
kW/m2 BTU/hr/ft2 (s)
1.74 550 60
2.33 740 40
2.90 920 30
4.73 1500 16
6.94 2200 9
9.46 3000 6
11.67 3700 4
19.87 6300 2
Table 6-3 Recommended design thermal radiation personnel (API 521, /63/)
Btu/h/ft2 kW/m2 Conditions
3000 9.5 Maximum radiant heat intensity at any location where urgent emergency action
by personnel is required. When personnel enter or work in an area with the
potential for radiant heat intensity greater than 6,31 kW/m2 (2 000 Btu/h·ft2),
then radiation shielding and/or special protective apparel (e.g. a fire approach
suit) should be considered.
SAFETY PRECAUTION — It is important to recognize that personnel with
appropriate clothing a cannot tolerate thermal radiation at 6,31 kW/m2 (2 000
Btu/h·ft2) for more than a few seconds.
2000 6.3 Maximum radiant heat intensity in areas where emergency actions lasting up to
30 s can be required by personnel without shielding but with appropriate clothinga
1500 4.7 Maximum radiant heat intensity in areas where emergency actions lasting 2 min
to 3 min can be required by personnel without shielding but with appropriate
clothinga
500 1.6 Maximum radiant heat intensity at any location where personnel with appropriate
clothinga can be continuously exposed
a
Appropriate clothing consists of hard hat, long-sleeved shirts with cuffs buttoned, work gloves, long-legged pants and work shoes.
Appropriate clothing minimizes direct skin exposure to thermal radiation.
In addition to the incident thermal flux and the duration of exposure, a number of other factors
also affect the degree of harm. These include:
• Source of thermal radiation e.g. hydrocarbon fire or thermonuclear device
• Area of skin exposed
• Age of person
• Type of clothing
• Speed and type of medical attention after exposure
Further, a significant factor in the consideration of the impact of thermal radiation when
carrying out a QRA is whether a person is considered to be outdoors or indoors, and further if
a person originally outdoors can find shelter and if person originally indoors tries to escape
from the building.
It should be noted that there is an ongoing discussion relating to human tolerance to fire
radiant heat exposure. In recent studies (/93/ and /94/), Raj questions the US regulatory
criterion of 5kW/m2 for 30 seconds to define the safety distance for LNG plants. The author
himself was exposed to heat levels from 3.5kW/m2 to 5kW/m2 for several tens of seconds
without any injury or pain. He also shows that using dummies to measure the effect of
radiation on skin temperature is not plausible due to the protective systems of the human
body. The experiments also suggest that normal clothing can reduce the heat flux to the skin
by a factor of 2-3. This is valid for the general public; most installations will have protective
clothing for personnel, which one must suppose will reduce this even further. Raj also
questions the use of exposure times as high as 30 seconds stating that it should be possible for
a grown person to move up to 40-80 meters in this time span, taking into account reaction
time of untrained personnel.
Lees (/71/) reports work by Lawson and Simms (/72/) who give relationships between
incident radiation and ignition of wood. The relationship for spontaneous ignition is:
(I - Is) t4/5 = k1
where I is the incident radiation (kW/m2), Is is the critical intensity for spontaneous ignition
(equal to 25.4 kW/m2), t is time in seconds and k1 is a constant (equal to 6.73 kJ/m2 s1/5).
Other sources (/73/) suggest that the values for k1 and k2 are in error and should have values,
based on American Whitewood, of 167.6 kJ/m2s1/5 and 118.4 kJ/ m2s1/3 respectively. The
source also gives values of Is and Ip of 25.6 kW/m2 and 14.7 kW/m2 respectively.
Table 6-4 and Table 6-5 give critical radiation intensities for ignition and are taken from the
TNO report (/59/) and the PhastRisk User Manual (/74/).
The probit is based on thermal radiation from nuclear explosions. TNO (/16/) present the
probit as
From a later version of the vulnerability model (/66/), a thermal radiation probit for
hydrocarbon fires is derived and presented in the TNO report as:-
Or
It will be noted that the thermal dose required for a given lethality level is lower for
hydrocarbon fires than for nuclear explosions. Both these probits are for naked human flesh;
a probit for lethality of clothed personnel exposed to hydrocarbon pool fires has been derived
as (/67/):
or
Probits have also been developed for impact other than lethality. The Eisenberg Vulnerability
Model (/65/) presents the limit value for first degree burns as:
For hydrocarbon fires, probits for first and second degree burns are presented as16:
As an illustration the LD50 has been calculated using the Eisenberg probits for naked and
clothed person in Table 6-6.
Table 6-6 LD50 using the Eisenberg probits
Exposure time 30s 20s 10s
Naked person 14 19 33
Clothed person 27 36 61
In some countries, the use of probabilistic criteria using a probit is mandatory, see Section
6.3.1. Where probits are not mandatory DNV may use either a probit or single value criteria
(see section 6.3.2).
In PhastRisk 6.53, human impact criteria for thermal radiation from pool and jet fires can be
defined in two ways:
By use of one or several single-value criteria. For example, 100% mortality inside the
25kW/m2 radiation limit, 50% mortality between the 12.5kW/m2 and 25kW/m2, no fatalities
outside the 12.5kW/m2 radius (these numbers are examples only).
In the Netherlands (/16/) and Northern Belgium (/92/), it is required to use the criteria in
Table 6-7 for pool and jet fires, where Plethal is found using the probit below and an exposure
time of 20s.
The different fire types have historically been associated with different forms of criteria:
• for vapour cloud fires a specific concentration of vapour in air (usually LFL, but
sometimes 0.5 LFL) is used;
• for stable fires of `long' duration (i.e. more than about one minute) a criterion thermal
radiation flux (kW/m2) is used;
• for short duration fires (principally fireballs) a combination of thermal radiation flux
and time is used - this would be either in the form of kJ/m2 or (kW/m2)4/3s (generally
termed a thermal dose).
Different criteria have been used for various studies for various reasons. From a review of
these criteria it is recommended that certain fixed criteria are used for the various fire types
together with a value for fatality for people within the criterion level. A short comment gives
the background to the selection of the value.
6.3.2.1 Flash Fire
When a flammable cloud is ignited a flame travels through the cloud and combustion occurs
in such a way that the hot gases produced rise; there is little expansion horizontally.
Consequently people within the cloud are subject to flame engulfment for a short period of
time; people outside the cloud are not engulfed in flame and are only subject to thermal
radiation for a short time. For the cloud to burn as a whole (sustained combustion) and burn
back to the source of release, the concentration needs to be above 0.9 LFL (/68/). At lower
concentrations (down to 0.6 LFL) combustion gives only pockets of flames. The LFL
envelope is recommended as a criterion. For people located outdoors, the mortality within this
limit is expected to be high, however buildings give good protection against flash fire. Thus a
mortality probability of 1 for people outside and 0.1 for people inside is often applied when
using this criterion. The use of fire retardant overalls, training and ease of movement in a
horizontal direction enable lower fatality probabilities to be used for workers who are
outdoors. For buildings such as workshops where there are large open doors, it may be more
appropriate to treat them as outdoors rather than indoors. Some buildings (eg with positive
pressure ventilation and automatic shutdown on flammable gas detection) will give virtually
complete protection against flash fire (fatality probability of zero).
Table 6-8 Human impact from flash fire
Fire Type Type of Criterion Recommended Fatality Comments
QRA Probability
(Range of Values)
Outdoors Indoors
Flash Fire Onshore and LFL 1.0 0.1 A high probability of fatality
Offshore (0.5-1.0) (0.01-0.3) for people outdoors but within
the burning cloud. Buildings
provide good protection from
flash fire.
The effects of pool and jet fires vary extremely widely depending on the circumstances. The
first consideration is whether the person of interest is trained and clothed for work on the
installation (onsite) or a member of the public (offsite). One can assume a higher vulnerability
for the latter both because of protective equipment and better reaction level due to training
and awareness.
In PhastRisk, it is possible to model fatalities from thermal radiation using a probit (the TNO
probit is the default) or by using one or more criterion level combined with a fatality
probability for each.
For QRAs applying a single criterion level the thermal radiation levels and fatality
probabilities for people onsite and offsite used historically are summarised in Table 6-9.
To define these criteria the method in appendix G for defining single-value criteria based on a
probit was used together with the following:
• Use of the Eisenberg probit for clothed personnel (onsite) and naked skin (offsite)
• Time taken to escape is greater than 30 seconds. Onsite personnel may have to secure
their work prior to exit from the module.
• Incapacitation occurs at 75% of LD50
• North Sea standard clothing for onsite personnel
It should be noted that for installations in warmer climates and with different safety
regulations, the type of clothing could be very different. In these cases a lower criteria for
radiation intensity should be considered especially for people onshore.
Table 6-9 Human impact from heat loads from pool and jet fires
Criterion Level Fatality probability Likely Effects at Used for people
Criterion Level
Outdoors Indoors
20 kW/m2 1 1 Incapacitation, leading to Onsite
fatality unless rescue is
effected quickly
12.5kW/m2 0.7 0.1 Extreme pain within 20 Offsite
seconds and movement to
shelter is instinctive.
Limiting flux for
secondary fires.
Historically several other single-value criteria have been used in DNV, Table 6-10 gives an
overview of these criterion levels. In defining the single value criterion it is important to
remember that people subject to thermal radiation levels just below the criterion level will be
assumed to be unaffected. This may be of significance for large populations just outside the
criterion envelope.
Table 6-10 Human impact from heat loads from pool and jet fires
Criterion Level Comment
40kW/m2 Immediate fatality
37.5 kW/m2 Used in the United States
25 kW/m2 Used for personnel onsite on onshore installations.
This is the intensity which is often used to indicate that
normal buildings can be ignited spontaneously
6.3.2.3 Fireballs
DNV has for many years used a dose value in kJ/m2 rather than (kW/m2)4/3s. Although data
support the use of the latter form, use of kJ/m2 instead may not be significant in the context of
the overall modelling. Two criterion values have historically been used (Table 6-11)..
Table 6-11 Human impact from short duration radiation doses (fire balls)
Criterion Level Effect at Criterion Fatality probability Recommended for
(kJ/m2) Level (Range of values)
Outdoors Indoors
375 3rd degree burns (0.7-1.0) (0.1-0.3)
250 2nd degree burns 0.7 0.2 Onsite and offsite
(0.7-1.0) (0.1-0.3) personnel
Note there are many of combinations of flux intensity and impact level which are basically
equivalent. Choosing a low flux / low impact combination is good where societal risk is a
major factor and populations are farther away. Where individual risk to site workers is
important, a high flux / high impact combination may be more appropriate.
Note also that the impact is based on the average intensity over the whole impact area, and
users must not confuse impact levels corresponding to those at the outer envelope boundary,
with those for an average impact over the entire envelope. As thermal intensity falls off at
faster than the square of distance, there might actually be less difference in average impact
from 250 to 375 kJ/m2 than might initially be envisaged.
100
90
80
70
Percentage Fatality
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100 110 120
2
Thermal Radiation (kW/m )
100
90
80
70
Percentage Fatality
30
20
10
0
0 10 20 30 40 50 60 70 80
Thermal Radiation (kW/m2)
100
90
80
70
Percentage Fatality
30
20
10
0
100 1000 10000
Thermal Dose (kW/m2)(4/3)s
7 EXPLOSION OVERPRESSURE
7.1 Introduction
Overpressures which can be hazardous to people may be caused by a number of mechanisms.
These include condensed phase explosions, vapour cloud explosions (VCE), thermal
explosions etc. The characteristics of the overpressures generated by these means are slightly
different. The hazardous effects described in this section are biased towards VCE effects,
although the principles are appropriate for all types of explosion effects and most data have
been derived from condensed phase or nuclear explosions.
This section gives an indication as to the possible impact of different levels of overpressure on
both people and structures, gives probit equations which have been derived for different levels
of impact and single value impact criteria.
7.2 Background
7.2.1 The Effects of Overpressure
The effects of overpressure on humans are normally categorised as follows:
Tertiary injury as a result of the body being thrown bodily by the explosion wind
and impacting on stationary objects or structures
If the explosion is accompanied by combustion of a vapour cloud (as in a VCE), people in the
vicinity of the explosion may also be affected by this combustion directly or by fires which
occur as a result of loss of containment from equipment containing flammable material and
subsequent ignition.
When examining the effects of overpressure it is common for the effect to be related to peak
side on overpressure (pressure at a surface) as this is relatively easy to both calculate and
measure and is the most important single parameter to determine the effects on structures.
This simplification is therefore reasonable for many effects, but in some circumstances e.g.
whole body translation, control room design, it may be necessary for other wave
characteristics to be determined. It should be emphasised that due to reflection a structure or a
person will be exposed to a higher explosion load (factor 1.5- 3) than what you ideally could
measure if you looked at the pressure wave passing.
Primary Effects
The human body is capable of adapting to pressure changes. However, organs can be
damaged if the change is sudden. The lung is generally regarded as the most susceptible
organ which is affected by overpressure and damage to it can lead to death. The ear is more
sensitive but damage to it does not lead to fatality. Peak overpressures which cause damage
to these organs are given in Table 7-1. The values given for VCEs are derived assuming a
similar impact for a peak overpressure which is three times that of the condensed phase
explosion (/81/). It should be emphasised that in an explosion scenario, people in the vicinity
of the accident may also be affected by the burning VCE cloud as covered in section 6.3.
Table 7-1 Human Injury Levels Resulting from Blast Waves, /81/
Injury Type Injury Level Peak Overpressure (bar)
Condensed Phase Vapour Cloud
Explosion
Eardrum Threshold 0.35 1
Failure 50% 1 – 1.4 3 – 4.2
Lung Damage Threshold 0.7 – 0.8 2.1 – 2.4
Lethality Threshold 2.1 - 2.9 6.3 - 8.7
50% 2.9 - 3.9 8.7 - 11.8
Virtually 3.9 - 5.5 11.8 - 16.6
100%
Secondary Effects
Secondary effects cause an impact on people either because of structural e.g. building
collapse, because of missiles e.g. glass fragments or secondary fires from ruptures of
equipment containing flammable material. The former effects are generally of most
significance for people offsite and for people located inside buildings on a process plant,
while the latter one is more relevant for people in the vicinity of an accident and for people on
offshore installations.
Buildings or process equipment can collapse or be damaged when subject to blast loadings far
lower than those required to produce primary effects on humans.
Empirical data relating the degree of damage of a building and equipment subject to an
overpressure wave are generally given in terms of peak overpressure and are usually from
observations following condensed phase explosions. Typical examples of these data are given
in Table 7-2 and Table 7-3. These lists are oversimplifications as different explosive masses
give different overpressure values for similar effects and wave reflection which can cause
significant damage may not be taken into account.
Table 7-2 Damage Produced by Blast (After Clancey, /75/)
Bar (pressure) Psig (pressure) Damage
0.0014 (0.02) Annoying noise (137 dB), if of low frequency (10 - 15 cps).
0.0021 (0.03) Occasional breaking of large glass windows already under strain.
0.0028 (0.04) Loud noise (143 dB). Sonic boom glass failure.
0.0069 (0.1) Breakage of windows, small, under strain.
0.01 (0.15) Typical pressure for glass failure.
0.02 (0.3) "Safe Distance" (probability 0.05 no serious damage beyond this value).
Missile imit. Some damage to house ceilings; 10% window glass broken.
0.028 (0.4) Limited minor structural damage.
0.034-0.069 (0.5-1) Large and small windows usually shattered occasional damage to
window frames.
0.048 (0.7) Minor damage to house structures.
0.069-0.138 (1-2) Corrugated asbestos shattered. Corrugated steel or aluminium panels,
fastenings fail, followed by buckling. Wood panel (standard housing)
fastenings fail, panels blown in.
0.090 (1.3) Steel frame of clad building slightly distorted.
0.138 (2) Partial collapse of walls and roofs of houses.
0.138-0.207 (2-3) Concrete or cinder block walls, not reinforced, shattered.
0.159 (2.3) Lower limit of serious structural damage.
0.207 (3) Heavy machines (wt 3000lbs) in industrial building suffered little damage.
Steel frame building distorted and pulled away from foundations.
0.207-0.276 (3-4) Frameless, self-framing steel panel building demolished.
Rupture of oil storage tanks.
0.276 (4) Cladding of light industrial buildings ruptured.
0.345 (5) Wooden utility poles (telegraph etc) snapped. Tall hydraulic press
(40000lbs wt) in building slightly damaged.
0.345-0.483 (5-7) Nearly complete destruction of houses.
0.483 (7) Loaded train wagons overturned.
0.483-0.552 (7-8) Brick panels, 8 - 12" thick, not reinforced, fail by shearing and flexure.
0.621 (9) Loaded train box-cars completely demolished.
0.689 (10) Probable total destruction buildings. Heavy (7000lb) machine tools
moved and badly damaged.
Very heavy (12000lb) machine tools survived.
20.68 (300) Limit of crater lip.
Doors and
window frames 0.09 (1.3) 0.059 (.86) 0.053 (77)
may be blown in
Category Ca
damage 0.124 (1.8) 0.079 (1.15) 0.076 (1.10)
Category Cb
damage 0.276 (4.0) 0.165 (2.4) 0.154 (2.3)
Primary missiles Limit of travel 0.014 (.20) 0.01 (.14) 0.008 (.12)
When buildings collapse people inside can be expected to suffer injuries ranging from minor
to fatal. Experience with earthquakes (/16/) has indicated that between 20 and 50% of people
would be expected to be killed. In VCE incidents higher probabilities of fatalities have been
recorded e.g. Flixborough where everyone in the control room at the time of the incident was
killed, although not all fatalities were necessarily due to blast. Factors which are likely to
affect the probability of fatality/survival of a person within a collapsed building are building
size (larger buildings give higher fatalities), age of the person (young and old are more
susceptible), the age of the building (old housing causing higher fatalities than new houses)
and the basic construction of the building.
API RP 752
100
Serious injury/fatality %
B1, B2, B4
10 B3
B5
1
0 200 400 600 800 1000
Overpressure mbar
API RP 752 (/98/) is currently under revision, so the above may be changed in the
new version. Also note that API RP 753 (/99/) has superseded API RP 752 with
regard to locating portable buildings (building type B1). /99/) does not give any
overpressure-lethality relationship for such buildings, but the effects of overpressure
are given:
major
damage and
may fail.
Wall and the
roof
components
not facing
the blast
sustain up to
major
damage.
Window
breakage
and falling
overhead
items are to
be expected.
Data from Pressure Levels for Siting Wood Trailers Using the API RP 752
Addendum Simplified Approach, BakerRisk Paper No 760-110-06 Sept 8 2006.
Note that the above effects are for VCEs with long durations (200 ms) and the weakest type of
wooden trailer construction likely to be encountered in processing facilities in the USA.
Metal trailers, wood trailers with wider studs or other wall construction may be stronger.
Typical office block: four storey, concrete frame and roof, brick block
2 wall panels.
1
Fatality Probability
CIA1
CIA2
0.1
CIA3
CIA4
0.01
0.01 0.1 1
Overpressure bar
These curves have been taken from WS Atkins report (/101/) and are for screening purposes
only. It is recommended that /101/ is consulted for detailed analyses or for analyses of
building types not included in the graph.
Wilfred Baker Engineering, Inc. (BakerRisk) carried out work for the Industry Technology
Cooperative. Relationships have been produced for the following building types in terms of a
relationship between explosion parameters and damage severity levels.
Steel framed structure with metal panels for roof and wall cladding 1
Steel framed structure with metal wall panels and a reinforced concrete roof. 2
Steel framed structure with unreinforced masonry (CMU or brick) infill walls (non-load 3
bearing) and a reinforced concrete or metal roof.
Steel framed structure with transite siding walls and a transite panel or metal deck roof. 4
Steel framed building with reinforced concrete walls panels and a reinforced concrete roof 5
deck.
Steel framed building with reinforced masonry infill walls and a reinforced concrete roof. 8
Reinforced masonry building with load bearing walls and a reinforced concrete roof. 9
Unreinforced masonry building with load bearing walls and a reinforced concrete roof. 10
Reinforced concrete frame structure with unreinforced masonry infill walls and a reinforced 11
concrete roof.
Reinforced concrete frame structure with reinforced masonry infill walls and a reinforced 12
concrete roof
Trailer or modular building (light construction consisting of wood framing with steel or 13
aluminium cladding)
Unreinforced masonry building with load bearing walls and a wood roof. 15
A multi-story steel frame building constructed from wide flange steel beams that support heavy 16
equipment loads. Typically with walls constructed from light metal siding and girts and heavy
steel diagonal bracing members. Floors may be constructed from concrete or steel grating.
Light Moderate 2A
Heavy Moderate 2B
Major 3
Collapse 4
The latest information was published in 2006. This represents the latest thinking on building
vulnerabilities and supersedes previous information from Baker Risk. The 2006 information
is confidential to the members but it is likely that member companies will be using this
information in the analyses they carry out. These vulnerabilities are likely to be more
accurate that the vulnerabilities given previously in this TN.
Tertiary Effects
Injury results from a person being displaced by the wind associated with the blast and a part
of the body impacting on a fixed/solid structure. The extent/seriousness of injury depends on
the velocity of impact, the hardness or shape of the structure and the part of the body involved
in the impact. The most critical body part is the base of the skull for which the threshold
value of velocity for fatality is 4m/s (velocities below 3m/s regarded as being safe).
For a VCE which has a positive phase duration of 100ms the threshold overpressure value is
approximately 0.7 bar. It is noteworthy that experience from VCE incidents (/78/) indicates
that prior to 1983 there had been no recorded fatalities for people outside the flammable cloud
as a result of primary/tertiary overpressure effects.
and
Y = -12.6 + 1.524 ln Ps
Probit functions for the probability of fatality after whole body impact are given as (/80/):
Y = 5.0 - 2.44 ln S
7,38 * 10 3 1,3 * 10 9
where S = +
Ρs Ρs i s
Y = 5.0 - 8.49 ln S
This probit is shown, assuming positive phase durations of 100 and 200m sec, in Figure 7-3.
Note this assumes whole body and head translation and impact onto a hard surface.
100
90
80
70
Percentage Fatality
60
40
30
20
10
0
0,5 0,6 0,7 0,8 0,9 1 1,1
Overpressure (bar)
Figure 7-3 Percentage Fatality for Impact of the Head from Overpressure
Probits for fatality because of secondary effects are dependent on building damage.
Eisenberg (/65/) presented probits for structural damage and glass breakage as
follows:
and
The TNO report (/16/) presents probits for damage to houses and apartment buildings. For
houses and apartment buildings up to four storeys high, the probits are:
Y = 5 - 0.26 ln V
where
3,9 5
4600 110
V =
+
Ρs is
Y = 5 - 0.26 ln V
where
8, 4 9, 3
17500 290
V =
+
Ρs is
Y = 5 - 0.22 ln V
where
7, 4 11,3
40000 460
V = +
Ρs is
This probit is shown in Figure 7-4 assuming (i) 20% fatality and (ii) 50% fatality in collapsed
two-storey buildings. Collapsed high rise buildings (eg. site offices as at Flixborough would
give worse results). It is of interest that the second term in the equation for V i.e (460/is)11.3 is
negligible compared with the first term for positive phase durations of 100 and 200ms i.e. the
term may be neglected.
100
90
80
70
Percentage Fatality
60
30
20
10
0
0,1 1 10
Overpressure (bar)
For apartment buildings over four storeys, the minor and major damage probits are the same
as those for houses given above. For collapse, the probit for a pressure wave is:
Y = 5 - 2.14 ln V
where
1,9 2, 5
1,25 3
V = +
Ρ i
and
Ρ
Ρ= Pst = static absorbable load (Pa)
Ρst
isω
i= ω = angular frequency (s-1)
Ρst
and
This probit is based on fatalities in VI explosions in World War II, corrected to remove the
effects of bomb shelters, and hence takes account of the collapse of average brick-built
houses. Another probit for fatality used in the same study was:
with Ps in psig.
100
90
80
70
Percentage Fatality
60
50
40
30 Pr = 2,47 + 1,43lgPs
Pr = 1,47 + 1,37lnPs
20
10
0
0,1 1 10 100 1000
Overpressure (bar)
100
90
80
70
Percentage Fatality
60
50
40
30
20 Upper limit
Mean
10 Mean
Lower limit
0
0,1 Overpressure
1 (bar) 10
Figure 7-6 Band of Fatality for People Inside Traditional Dwelling Houses
Explosion consequence models for injuries and fatalities of people within buildings have been
reviewed (/82/). These models are based on analyses of World War II damage, subsequent
trials and recent exchange of information between NATO countries. The band within which
these predictions fall is given in Figure 7-6, and these data are applicable to people inside
housing built to traditional UK standards.
In Northern Belgium (/92/), the following probit is mandatory for calculations of fatalities due
to secondary effects of overpressure for people inside:
For tertiary effects (people outdoors) the probit suggested by HSE for use is:
Pr = 5.0 - 8.49 ln S
2,43 * 10 3 4 * 108
where S = +
Ρs Ρs is
is = impulse = ½ Ps tD
For secondary effects (people indoors) the three probits identified and the range of predictions
given in Figure 7-6 also give a wide variation, (see Table 7.4) particularly at high
overpressures. At overpressures usually associated with offsite housing there is slightly less
variation between lowest and highest probits. None of these probits is specifically
recommended for use as the basis for them has not been reviewed by DNV.
Table 7-5 Percentage Fatality at Various Overpressures (From Probits)
Probit Overpressure (bar)
0.05 0.1 0.2 0.35 0.5 1 2 5
Y = 1.47 + 1.37 ln Ps <1 <1 2 10 21 55 86 100
Y = 2.47 + 1.43 log10 Ps <1 1 3 6 10 19 33 55
Y = 5 – 0.22 ln [40000/Ps]7.4 <1 1 7 23 32 47 50 50
Y = 5 – 0.22 ln [40000/Ps]7.4 <1 1 3 9 15 19 20 20
Figure 7.4 Lower Limit <1 <1 <1 2 3 8 14 (23)
Figure 7.4 Upper Limit <1 <1 9 36 55 100 100 100
Figure 7.4 Mean <1 <1 5 19 29 (54) (57) (62)
These fatality rates includes the effect of the explosion pressure only, No fatalities related to the VCE cloud it self (heat) or
effect of potentially secondary fires are included.
For other explosion models it is suggested that single value criteria are derived as follows:
• For people outdoors use the probit given in section 7.3;
• For people indoors use the mean line on Figure 7-6 up to a value of 0.7 bar (the use of
the mean above this value is questionable).
+ These figures are based on the fatality probability of a person in a typical brick dwelling (/100/). Special explosion rated
buildings ( e.g. control rooms, special process buildings etc) has to be evaluated separately with respect to indoor
fatality rates
Analysts are reminded that explosion models are an idealized representation of the real
situation. Impact criteria must be used to some extent to put back the reality into predictions.
Thus in a VCE, almost everyone in the flammable portion of the cloud will be expected to die,
due to flame contact, regardless of the overpressure generated. This effect is not included in
the table above. Neither is the effect of fatalities from secondary fires due to rupture of
equipment or piping containing flammable materials. These effects must be modelled in
accordance with the criteria given in section 6.3.
In escalation scenarios when an explosion escalates to other equipment within the same area,
or to other areas, the fatality rates may increase dramatically to an increased flame volume.
The capacity of deck and firewall structures is usually given in the design basis and escalation
may be evaluated against these design accidental loads (DALs). In addition to the expanded
flame size, high explosion overpressures may make additionally fatalities for outside the
critical radiation level generated by the fire. The explosion capacity for process equipment is
however often not given. A common criterion of 0.3 barg6 is therefore used, meaning that
offshore process equipment designed in accordance with recognized codes will withstand a
pressure of 0.3 barg before escalation take place. This is an average value, associated with
large uncertainties and there will be variations from plant to plant depending on the specific
design
6
This figure is estimated by DNV explosion and structural experts,
The main parts of the body directly susceptible to the damaging effects of overpressure are
normally the eardrums and lungs. Lung damage can be fatal and an example of the While
personnel offshore are typically out of doors, or inside the TR it would be expected that a
lower fatality expectation exists than for a domestic situation. However, there are significant
items of plant and equipment that would be available to provide missiles in the event of an
explosion and in a situation where an explosion has escalated to other equipment inside the
same area such items is expected to be fatal.
For offshore installations this means that the fatality probability due to explosions may be
split up into the following.
Table 7-7 Recommended criteria for overpressure, offshore structures
Hazard Fatality fraction
Explosion Pressure
overpressure Criteria Scenario Thermal effect effects
VCE with subsequent fire, no escalation,
< 0.1 bar low explosion load ref. sec. 6.3 N.A
VCE with subsequent fire, no escalation,
0.1 – 0.3 bar medium explosion load ref. sec. 6.3 0.01
Initial Area VCE escalated to other equipment,
Secondary large fire within initial area. Effect from an
> 0.3 bar High explosion load escalated fire 1
No escalated Fire. Flames from the initial
< wall area may impact neighbouring areas Potential effects
capacity giving potential fatalities
Neighbouring Escalated VCE , flames from the initial
areas7 > wall area may impact neighbouring areas
capacity giving potential fatalities 0.5-18
7
The effects related to neighbouring areas are very dependant on the physical layout of the installation and should be
subjected to a separate evaluation for each installation In many cases an area classed as neighbouring might not be
separated from the initial area by a barrier which reduces the overpressure to a point where it is negligible. Separate CFD
analysis could be used for this purpose. In the absence of CFD use can be made of TNO curves or simpler rule sets.
8
These are figure used in a number of DNV studies and should be used as indicative. There exists no specific reference to
this type of fatality fraction and the fraction should be subjected to a platform individual evaluation.
8 SMOKE
8.1 Introduction
This section considers the effects of smoke on people by examining the effects, separately, of
the main constituents of smoke which can be harmful, i.e. carbon monoxide (CO), carbon
dioxide (CO2), oxygen starvation and high temperatures of inhaled gases.
For process accidents in most QRAs we perform, the fire is assumed to occur outside in well-
ventilated areas and smoke is a negligible contributor to fatalities. It is therefore common
practice not to calculate impact from smoke. For fires indoors, smoke is known to be a main
contributor to fatalities. For example, half of casualties in dwelling fires in the UK are in the
category “overcome by smoke and toxic gases” (/95/). Thus when modelling fires indoors,
one has to take smoke into account. In addition, smoke is a factor when calculating
impairment of safety functions offshore, see TN 8.
Below background information concerning smoke is given. A method for the evaluation of
lethal concentrations is given in Appendix F, and single value criteria for incapacitation
derived by using this method are presented below. It is recommended that this method is used
when examining the impact of smoke etc but the single value criteria given in Table 8-7 are
only for guidance. The method in Appendix F is the method presented in the SFPE Handbook
of Fire Protection Engineering (/95/) and in ISO standard ISO/TS 13571:2002.
The proportion of toxic gases in the smoke depends primarily on the chemical structure of the
burning materials and the combustion conditions. For most hydrocarbons the chemical
structure is of far less importance than combustion conditions and in practical situations
modelled in QRAs ventilation has the main effect on combustion products. Fires in which the
ventilation is restricted or enclosed e.g. in modules or compartments, are called under-
ventilated or ventilation-controlled fires, because the availability of ventilation determines the
rate of combustion. Conversely, fires in well ventilated areas are fuel controlled, i.e. the rate
of combustion is determined by the availability of fuel. In general, reduced ventilation greatly
increases the percentage of CO, while the O2 and CO2 remains more or less unaffected.
Typical gas concentrations close to the fire are given in Table 8-1 (collated by Bonn
(1993)(/83/), based on Johnson (1992)(/84/), Tewarson (1990)(/85/) and Wighus
(1991)(/86/)).
Table 8-1 Initial Gas Concentrations in Smoke (Bonn, 1993)
GAS CONCENTRATION IN SMOKE (%)
WELL VENTILATED FIRE UNDER VENTILATED FIRE
(fuel controlled) (ventilation controlled)
gas fire liquid fire gas fire liquid fire
CO 0.04 0.08 3 3.1
CO2 10.9 11.8 8.2 9.2
O2 0 0 0 0
The effects of smoke can be toxicity (CO2, CO), asphyxiation, high temperature and reduced
visibility. The combination of effects is complex and here assumed to be cumulative. Each
separate effect is discussed below.
CO uptake and intoxication is extremely insidious, producing minimal effects until a critical
dose has been accumulated, which rapidly leads to collapse. Therefore, victims may not
realise the danger until it is too late to escape. The rate of uptake is greatly increased at
higher activity levels.
The effects of exposures to different levels of CO for different lengths of time are given in
Table 8. (Stensaas, 1991 /88/). The CO concentration causing unconsciousness or death is
inversely proportional to the exposure time. The critical toxic effect is therefore best
measures as a dose in the form of:
D = C *t
where
Table 8-3 effect of CO increase in blood (B. Karlsson and J. Quintiere, Enclosure Fire
Dynamics)
% of COHb (carboxy- Volume concentration effect
haemoglobin) in blood [%] of CO present in
the air
0-10 <0.008 Indifferent phase:
None
10-20 0.008-0.015 Compensated phase:
Increased ventilation and heart rate, loss of
efficiency (fatigue, memory, judgement)
20-40 0.015-0.04 Dizziness, nausea, possible paralysis
Excess CO2 causes "hypercapnia", characterised at first by increased respiration. Since CO2
itself is not very toxic, the most important effect of CO2 in smoke is to increase the uptake of
other toxic gases, although CO2 may also have a beneficial effect by increasing the oxygen
uptake.
CO2 concentrations of 5% triple the rate of breathing (TIL, 1991) (/87/). At this
concentration, breathing becomes difficult for some individuals, although it can in general be
sustained for up to an hour without serious after-effects (Stensaas, 1991) (/88/). 5% is also
the IDLH concentration for CO2 (NIOSH, 1990), from which escape is considered possible in
30 minutes without any escape-impairing or irreversible effects.
CO2 concentrations above 10% cause loss of consciousness in times ranging from 30 minutes
at 10% to 1 minute at 15% (Stensaas, 1991) (/88/).
The acute health effects that are seen following inhalation of high concentrations of CO2 are
presented in Table 8-4 and should be used as reference values for safety risk assessments [1],
[2]
.
Table 8-4 Acute health effects of high concentrations of inhaled CO2 /106/
CO2 Concentration
Exposure Effects on Humans
in Air (% v/v)
Guidance Note [1]: The reference source on which the above table was based /106// does not
provide specific details of the age, sex, fitness or general well-being of the people exposed.
But a conservative and reasonable assumption would be that people with underlying health
concerns would not have been selected as a volunteer in tests. Therefore the exposure limits
may need to be adjusted for use in risk assessment should the exposed population being
considered include vulnerable people (e.g. those with a respiratory illness).
The harm level expressed by a given substance in the air is influenced by two factors, the
concentration in the air (c) and the duration of exposure (t). The following expressions have been
defined by the HSE for CO2, ref. Figure 8-1[3]:
25%
SLOD DTL
SLOT DTL
CO2 in Air at 1 Atmosphere (%)
20%
15%
10%
5%
0%
0 5 10 15 20 25 30 35 40 45 50 55 60
Guidance Note [2]: CO2 is not a chemical asphyxiant like hydrosulphide or carbon monoxide in that it does not
prevent the efficient transfer of oxygen via the blood.
Guidance Note [3]: With reference to Figure 8-1 the number of people injured (serious and minor) by the
release may be approximated by the number people estimated to be between the SLOD and SLOT DTL contours
(i.e. the SLOT DTL contour is taken as a pragmatic limit for human harm).
Guidance Note [4]: With reference to Figure 8-1 the number of fatalities caused by a CO2 release may be
approximated by the number people estimated to within the SLOD contour (i.e. the SLOD DTL contour is taken
as a pragmatic limit for fatalities).
The OEL given in Table 8-5 should be applied as reference for evaluating occupational
hazards.
Oxygen concentrations below 15% by volume produce oxygen starvation effects such as
increased breathing, faulty judgement and rapid onset of fatigue.
Oxygen concentrations below 10% cause rapid loss of judgement and comprehension,
followed by loss of consciousness, leading to death within minutes (Stensaas, 1991)( /88/).
Concentrations down to 5% lead to death within few minutes (Sax, 1984) (/89/).
Table 8-6 Effects on Humans from Elevated Temperatures (Bryan, 1986) (/87/)
Temperature Physiological Response
(Celsius)
127 Difficult breathing
140 5-min tolerance limit
149 Mouth breathing difficult, temperature limit for
escape
160 Rapid unbearable pain for dry skin
182 Irreversible injury in 30 seconds
203 Respiratory systems tolerance time for less than
four minutes with wet skin
8.7 Criteria
All of the above effects are important and should be taken into account when considering
smoke effects. Appendix F gives an approach for determining smoke incapacitation, but there
is no recommended single value criterion for smoke. Using the method in Appendix F and
the smoke compositions given in Table 8-1 produces the following values for smoke
concentrations for impairment within one minute (Table 8-7):
Table 8-7 Incapacitation Doses from Smoke
GAS WELL VENTILATED FIRE UNDER VENTILATED FIRE
(fuel controlled) (ventilation controlled)
gas fire liquid fire gas fire liquid fire
Lethal
concentrations 71% 70% 48% 45%
for 1 minute
exposure
The relatively high production of CO for ventilation controlled fires means that the effects of
smoke exposure to these fires are dominated by CO. For fuel controlled fires the production
of CO is very low, and the smoke works almost entirely as a simple asphyxiant (i.e. the lack
of oxygen being the dominant factor). The concentration of smoke causing fatalities in fuel
controlled fires is consequently quite high.
Temperature effects are not included in the above numbers and should also be evaluated along
the same methodology as the effects of the gases, see Appendix F.
9 ASPHYXIANTS
A gas or vapour which is hazardous to health as a result of it displacing the oxygen in the air
is termed an asphyxiant. Asphyxiants are non-toxic and are typically inert gases, such as
nitrogen, although they could also be flammable gases, such as methane, propane etc. The
normal oxygen content in air is approximately 21%. Nitrogen is the other primary component
of air at 78%, with other trace components totalling approximately 1%. The presence of any
additional gas in the air, with the exception of oxygen, dilutes the oxygen concentration and
can create an oxygen-deficient atmosphere. As the oxygen concentration in the air diminishes,
the human body becomes oxygen deprived and asphyxiation occurs. Effects can range from
giddiness, mental confusion, loss of judgment, loss of coordination, weakness, nausea,
fainting to death.
100% fatality rate is assumed for personnel inside the 20 kW/m radiation contours. The
2
fatality rate for a given hydrocarbon event source area is assumed to be equal to the ratio of
the area of the 20 kW/m2 radiation contour to the area of the module. Personnel outside of the
contour are assumed to survive.
A
I initial = min rad ,1 × N i (1)
Ainitiation
Where:
Fatalities due to the explosion overpressure may be as a result of ingestion of hot combustion
products, the effects of overpressure on the body, or impacts with solid objects following full
body translation.
Delayed ignition events, considered as flash fires also including explosions not escalating to
equipment, (Pexp < 0.3 bar), 100% fatalities within the LFL gas cloud.
Where Freduction is a reduction factor to take account of directional and shielding effects. A
value has to be determined for “adjacent” modules and other “modules”.
Flame Wholly Within Module Area Shielded by Obstruction Flame Directed Outward From Module
Source
Source Source
This process will result in the estimated fatalities within the following categories
Persons surviving the initial fire or explosion may be killed when escaping to a temporary
safe area. Proper escape from the process areas onboard a facility is provided by the
dedicated escape ways. The fatality fraction estimated for escape takes into account personnel
not able to find their way around the vessel due to smoke, heat and physical obstructions
within the escape routes.
After the estimation of immediate fatalities, calculation of the fatalities during escape shall be
determined based on the following logic:
o Personnel killed by the immediate effects of the event itself are immediate fatalities.
o Those not killed by the immediate event will attempt to muster (escape) to the TR.
o Fatalities during escape to TR is calculated as follows:
o Heat radiation above 12.5 kW/m2.
To accommodate the fact that personnel may be shielded from the fires by large structures and
equipment the calculated theoretical fatality fractions is reduced to 50%. The criterion shall be
evaluated together with the potential for incapacitation due to smoke exposure.
Incapacitation due to smoke
Incapacitation due to temperature has to be evaluated in addition to the above given criteria
A smoke temperature of 230 oC will with 1 minute exposure cause incapacitation (see
appendix F)
The evacuation fatalities are related to scenarios where the evacuation is impaired and
personnel cannot use the evacuation means. Scenarios where the evacuation means fail on
demand and rescue and recovery of personnel becomes relevant are also included here.
Fatalities during evacuation may be due to:
o Collapse of the vessel hull leading to the vessel capsize or sinking prior to the
completion of evacuation
o Unavailable evacuation means
o Failures/ accidents during launching of the evacuation bridge, lifeboats or tertiary
evacuation means.
o Evacuation fatalities is evaluated within the 500 zone, and does not include transfer
from evacuation means to rescue vessel or other safe location.
Fatalities during evacuation are estimated based on the evacuation fatality rate for the
different evacuation scenarios and probability for unsuccessful evacuation, which is taken
from the EER study. For personnel in the TR, there will be no fatalities if the TR remains
unimpaired.
o All fires and explosions escalation through a fire separation or explosion barrier
o Loss of TR integrity
o Hull failure
A separate escape, evacuation and rescue (EER) study is normally performed, and will
provide input to the assessment of evacuation fatalities. The EER study will consider integrity
of escape ways, evacuation and rescue systems to meet requirements from operations in arctic
conditions, taking into consideration sea ice, icing, low temperatures, long period of darkness
and remoteness.
11 REFERENCES
Withers, R.M.J. The Lethal Toxicity of Ammonia - A Report to the MHAP, IChem E
/1/ N.W. Branch Papers, 1986 N 1, Manchester, January 1986.
ten Berge W.F., Zwart A. and Appelman L.M, Concentration - Time Mortality
/2/ Response Relationship of Irritant and Systemically Acting Vapours and Gases, J. Haz.
Materials, 13 (1986) 301-309.
DCMR Steering Committee, Study into the Risks from the Transportation of Liquid
/3/ Chlorine and Ammonia in the Rijnmond Area. Selection of Probit Equations for Acute
Toxic Gas Exposure, F291 Memo, Technica Ltd, July 1984.
HSE, Canvey - an Investigation of Potential Hazards from operations in the Canvey
/4/ island/Thurrock Area, HMSO, London, 1978.
HSC Advisory Committee on Dangerous Substances, Major Hazard Aspects of the
/5/ Transport of Dangerous Substances, HMSO, London, 1992.
Withers, R.M.J. and Lees F.P., The Assessment of Major Hazards. The Lethal Toxicity
/6/ of Chlorine, Part 2. Model of Toxicity to Man, J. Haz. Materials, 12 (1985), 283 - 302.
DNV Technica Report C2710, September 1991.
/7/
de Weger D., Pietersen C.M. and Reuzel P.G.J., Consequences of Exposure to Toxic
/8/ Gases Following Industrial Disasters, J. Loss. Prev. Process Ind., 1991, Vol 4, July.
ten Berge W.F. and Vis van Heemst M., Validity and Accuracy of a Commonly Used
/9/ Toxicity Assessment Model in Risk Analysis, 4th Int. Symp. Loss Prevention,
Harrogate, 1983.
Zwart, A. and Woutersen R.A., Acute Inhalation Toxicity of Chlorine in Rats and Mice
/10/ : Time - Concentration Mortality Relationships and Effects on Respiration, J. Haz.
Mats., 19, 1988, 195 - 208, Elsevier Science Publishers B.V., Amsterdam.
DNV Technica Report C1672.
/11/
Mudan, K., Use of Toxicity Data in Quantitative Risk Assessment of HF Alkylation
/12/ Units, AIChE Summer National Meetings, August 1989.
DNV Technica Report C1519, January 1989.
/13/
Perry, W.W. and Articola, W.J., Study to Modify the Vulnerability Model of the Risk
/14/ Management System, CG-D-22-80, US Coast Guard, Washington DC, 1980.
VROM, Private Communication from Ministerie van Volkhuisvesting Ruimtelijke
/15/ Ordening en Milieubeheer, the Netherlands, 1981.
VROM, Methoden voor het bepalen van mogelijke schade (Green Book), PGS1,
/16/ Voorburg, December 2003.
(http://www.vrom.nl/pagina.html?id=20725, 03 October 2008).
623 - 631.
Barcroft J., The Toxicity of Atmospheres Containing Hydrocyanic Acid, J. Hyg., 31
/36/ (1931) 1 - 34.
NIOSH, Registry of Toxic Effects of Chemical Substances, 1983
/37/
Lehmann K.B., Experimentele Studien über den Einfluss Technisch und hygeinisch
/38/ Wichtiger Gase und Dämpfe auf den Organ ismus, Arch. Hyg. 14 (1892) 135 - 189.
/39/
Snamprogretti, Research Reports on MTBE : Toxicological Data, 1980.
/40/
Torkelson T.R., Oyen F. and Rowe V.K., The Toxicity of Bromochloromethane as
/41/ Determined on Laboratory Animals, Amer. Ind. Hyg. Assoc. J., 21 (1960) 275 - 286.
ten Berge W.F., The Toxicity of Methylisocyanate on Rats, J. Haz. Materials, 12 (1985)
/42/ 309 - 311.
Hine C.H. Meyers F.H. and Wright R.W., Pulmonary Changes in Animals Exposed to
/43/ Nitrogen Dioxide, Effects of Acute Exposures, Toxicol. Appl. Pharmacol., 16 (1970)
201 - 213.
Smith L.W., Gardner R.J. and Kennedy G.L., Short Term Inhalation Toxicity of
/44/ Perfluoroisobutylene, Drug Chem. Toxicol. 5 (3) (1982) 295 - 303.
BCGA (1984), A Method for Estimating the Offshore Risk from Bulk Storage of
/45/ Liquefied Oxygen, A Report by British Compressed Gas Association /HSE/SRD
Working Group, 1984.
Adams E.M., Spencer H.C., Rowe V.K., McCollister D.D. and Irish DD, Vapor
/46/ Toxicity of Trichloro ethylene Determined by Experiments on Laboratory Animals,
AMA Arch. Ind. Hyg. Occup. Med., 4 (1951) 469-481.
/51/ Treon J.F., Dutra F.R., Cappel J., Sigmon H. and Younker W., Toxicity of Sulphuric
Acid Mist, Industrial Hygiene and Occupational Medicine, Vol. 2, pp716-735, 1950.
/52/ Matheson, Matheson Gas Data Book, Sixth Edition, December 1980.
/53/ ICI, Codes of Practice for Chemicals with Major Hazards: Hydrogen Fluoride
(Anhydrous), January 1974.
Kimmerle G., Aspects and Methodology for the Evaluation of Toxicological
/54/ Parameters During Fire Exposure, Combustion Toxicology, Vol. 1, February 1974.
van Heemst V., Estimating Chlorine Toxicity Under Emergency Conditions, Chlorine
/55/ Safety Seminar, Brussels, November 14-15, 1990.
Finney DJ (1971), Probit Analysis, 3rd Edition, Cambridge University Press
/56/
ACGIH, 1990 - 1991 Threshold Limit Values for Chemical Substances and Physical
/57/ Agents and Biological Exposure Indices, ACGIH, Cincinnati, 1990.
AIHA, Emergency Response Planning Guidelines, AIHA, Akron, 1989.
/58/
Kinsman P., Major Hazard Assessment : A Survey of Current Methodology and
/59/ Information Sources, HSE 1991, HMSO.
HSE, EH40/92 Occupational Exposure Limits 1992, HMSO, 1992.
/60/
HSC, The Control of Substances Hazardous to Health (COSHH) Regulations, HMSO
/61/ 1988.
C.C.P.S. and AICHE, Guidelines for Chemical Process Quantitative Risk Analysis, 2nd
/62/ edition, Wiley-Interscience, 2000.
American Petroleum Institute, Pressure-reliving and Depressuring Systems, API STD
/63/ 521, 5th Ed., Washington DC, 2008.
Hymes I., Boydell W. and Prescott B.L., The Physiological and Pathological Effects of
/64/ Thermal Radiation, SRD Report R275, December 1992.
Eisenberg N.A., Lynch C.J. and Breeding R.J., Vulnerability Model : A Simulation
/65/ System for Assessing Damage Resulting from Marine Spills, Enviro Control Inc., US
Coast Guard Report CG-D-B5-75, June 1975.
Tsao C.K. and Perry W.W., Modifications to the Vulnerability Model, US Coast Guard,
/66/ USCG-D-38-79, March 1979.
Pieterson C.M., Consequences of Accidental Releases of Hazardous Material,
/67/ Proceedings of the Loss of Containment Conference, September 1989, London.
Evans A. and Puttock J.S., Experiments in the Ignition of Dense Flammable Gas
/68/ Clouds, Loss Prevention Symposium, Cannes, 1986.
DNV TN: Offshore Criteria.
/69/
DNV TN: Offshore QRA.
/70/
Mannan, Lees’ Loss Prevention in the Process Industries, Elseviere Butterworths, 2005.
/71/
Lawson D.I. and Simms D.L., The Ignition of Wood by Radiation, Br. J. Appl. Phys.,
/72/ 13, 288, September 1952.
Private Communication.
/73/
DNV Technica, PHAST User Manual, Appendix VI - Modelling Guidance, October
/74/ 1992.
Clancey V.J., Diagnostic Features of Explosion Damage, 6th Int. Meeting of Forensic
/75/ Sciences, Edinburgh, 1972.
Scilly N. and High W.G., The Blast Effect of Explosions, Int. Symp. on Loss
/76/ Prevention and Safety Promotion in Process Industries, Cannes, 1987.
Jarrett D.E., Derivation of the British Explosives Safety Distances, Annals of the New
/77/ York Academy of Sciences.
Wiekema B.J, Vapour Cloud Explosions - An Analysis Based on Accidents, Jal Haz
/78/ Matls 8 (1964). 295-311.
Hirsch F.G., Effects of Overpressure on the Ear - A Review, Annals of the New York
/79/ Academy of Sciences, 1968.
Baker W.E., Cox P.A. and Westine P.S. et al, Explosion Hazards and Evaluation,
/80/ Elsevier Scientific Publishing Company, 1983.
Hazards of Chemical Rockets and Propellants Handbook, Volume 1, General Safety
/81/ Engineering Design Criteria, AD 889763, CPIA/194, May 1972.
Merrifield, R, Simplified Calculations of Blast Induced Injuries and Damage.
/82/ Specialist Inspector Report no. 37, HSE, 1993.
Bonn (1993): Smoke & Gas Detectors for Temporary Refuges, R. Bonn, BP
/83/ Exploration, 4th International Conference on Offshore Loss Prevention - A Systematic
Approach, Aberdeen, Scotland, 3-5 March 1993, published by the BHR Group.
Johnson (1992): Airborne Observations of the Physical and Chemical Characteristics of
/84/ the Kuwait Oil Smoke Plume, D. W. Johnson et al, Nature vol 353, pp. 617-621.
Tewarson (1990): Generation of Heat and Chemical Compounds in Fire, A. Tewarson,
/85/ The SFPE Handbook on Fire and Protection Engineering, NFPA, 1990.
Wighus (1991): Smoke Hazard in Offshore Platform Fires, R. Wighus, Ø. Meland and
/86/ B. Vembe, SINTEF report STF25-A91007, 1991.
TIL (1991): Conditions Within the Muster Station on the Alba Development,
/87/ Confidential Report to Chevron, DNV Technica Indecone Project I671, October 1991.
Stensaas (1991): Toxicity, Visibility and Heat Stresses of Fire Effluents - Human
/88/ Tenability Limits, SINTEF Report STF25-AP1022, May 1991.
Sax (1984): Dangerous Properties of Industrial Materials, 6th edition, N. Irving Sax,
/89/ Van Nostrand Reinhold Company Inc, 1984.
Spieth (1982): A Combined Hazard Index Fire Test Methodology for Aircraft Cabine
/90/ Materials, Spieth, Gaume, Luoto and Klinch, vols I and II, DOT/FAA/CT-82/36-1 and
DOT/FAA/CT-82/36-11, Department of Transportation, Atlantic City, April 1982.
HSE, Assessment of the Dangerous Toxic Load (DTL) for Specified Level of Toxicity
/91/ (SLOT) and Significant Likelihood of Death (SLOD),
http://www.hse.gov.uk/hid/haztox.htm, 24. September 2008.
Departement Leefmilieu, Natuur en Energie Afdeling Milieu-, Natuur- en
/92/ Energiebeleid, VR-richtlijnenboek -Omgevingsveiligheidsrapport - OVR, Revision 3,
January 2006.
Raj (2007): A review of the criteria for people exposure to radiant heat flux from fires,
/93/ Journal of Hazardeous Materials 159 (2008) 61-71, September 2007
Raj (2008): Field tests on human tolerance to (LNG) fire radiant heat exposure, and
/94/ attenuation effects of clothing and other objects, Journal of Hazardous Materials 157
(2008) 246-259, December 2007.
Purser (2002): Toxicity Assessment of Combustion Products, D. A. Purser, The SFPE
/95/ Handbook of Fire Protection Engineering 3rd edition, Chapter 2-6, NFPA, 2002.
Handleiding Risicoberekeningen Bevi Module B Algemeen Versie 3. 1 Jan 2008
/96/ (English version Reference Manual Bevi Risk Assessments version 3.0 Jan 2009).
Guidelines for Quantitative Risk Assessment. “Purple Book”, CPR 18E. VROM.
/97/ Publication Series on Dangerous Substances (PSG 3). Dec 2005.
American Petroleum Institute Recommended Practice 752. Management of Hazards
/98/ Associated with Location of Process Plant Buildings. Second Edition, Nov 2003.
American Petroleum Institute Recommended Practice 753. Management of Hazards
/99/ Associated with Location of Process Plant Portable Buildings. First Edition, June
2007.
Chemical Industries Association. Guidance for the Location and Design of Occupied
/100/ Buildings on Chemical Manufacturing Sites. (Nov 2003).
W S Atkins Science and Technology. Derivation of Fatality Probability Functions for
/101/ Occupants of Buildings subject to Blast Loads. Phase 4. Contract Research report
151/1997.
“Indicative human vulnerability to the hazardous agents present offshore for application
/102/ in risk assessment of major accidents” HSE hazardous Installation directorate doc.
SPC/Tech/OSD/30. January 2008
Dangers of Oxygen-Deficient Atmospheres, Safetygram-17, Air Products
/103/
Hazards of Nitrogen Asphyxiation, Safety Bulletin No. 2003-10-B, June 2003, U.S.
/104/ Chemical Safety and Hazard Investigation Board.
Book: Confined Space Entry and Emergency Response, By D. alan Veasey, Wiley
/105/ InterScience. Published by Wiley-Interscience, 2006. ISBN 0471787655,
9780471787655, 486 pages
U.S. Environmental Protection Agency, Carbon Dioxide as a Fire Suppressant: Examining the Risks,
/106/ February 2000 (http://www.epa.gov/Ozone/snap/fire/co2/co2report.pdf)
Pipeline Safety Design, DNV, Report, 09.12.2008
/107/
APPENDIX A:
PROBIT ANALYSIS
A.0 PROBITS
The range of susceptibility in a population to a harmful consequence can be expressed
mathematically using a criterion in the form of an equation which expresses the percentage of
a defined population which will suffer a defined level of harm (normally death) when it is
exposed to a specified dangerous load. This is a `Probit' equation which has the form:
Y = A + B Ln (Cn.T)
Where Y is the probit (or probability measure), A, B and n are constants, C is a concentration
and t is time. The probit, Y, can be related to percent fatalities using published tables. The
most commonly used table is given as Table A.1. The probit is, in statistical terms, a single
tailed, normal distribution with the values increased by a value of 5 to maintain positive
values of the probit function over the sensible fatality range.
The relationship between a probit and the normal distribution is shown in Figure A.1. Here, a
number of probits have a common intersection at the LC50 (this is shown by following the
dotted line down to the lower, fatality curve and reading across to the 50% fatality value) but
would predict very different toxicological behaviour at other toxicity values. All the lines are
based on the equation given above but with different values for A and B. As can be seen, for
a given LC50, the effect of B is only to change the LC90/LC10 ratio. Similarly, if the
concentration-time units of the dose (CnT) are altered, the probit relationship is affected only
in the A parameter.
The sensitivity of a risk analysis to the choice of probit function can be illustrated by
reference to toxicity probits for chlorine. Table A.2 illustrates the concentrations predicted by
three probits for increasing durations of exposure:
Table A- 1 Chlorine Probit Comparison
TIME LC50 (ppm)
(min)
IChemE, Withers and Zwart and ten Berge et al
Lees Woutersen (Dutch)
10 433 1,931 775
20 306 1,172 602
30 250 690 520
It can be seen that the choice between the IChemE and Zwart and Woutersen probits could
significantly change the predicted risk levels and is likely to produce marked differences in
the number of fatalities calculated as a result of any release.
For toxic effects a logarithmic relationship generally gives rise to a normal distribution.
Consequently the cumulative response curve is generally a normal sigmoid shape (see Figure
A.3). It is the transformation of this sigmoid curve to a straight line relationship that is at the
centre of the probit methodology.
In 1933, Gaddum proposed to measure the probability (P) of response on a transformed scale,
the normal equivalent deviate (NED), defines as:
z
1
P=
2π ∫ exp(-0. 5
-∞
u2
) du
Thus the NED (z) of any value of P between 0 and 1 is defined as the abscissa corresponding
to a probability P in a normal distribution with a mean of 0 and a variance of 1. The NED
however, is a negative if P is less than 0.5 (i.e. 50%). To avoid these negative values Bliss2
proposed a slightly different formulation. He proposed the probit of P as y where:
Y-5
1
P=
2π ∫ exp(-0. 5
-∞
u2
) du
The probit is thus simply the NED increased by 5 with the effect that the probit remains
positive unless P is exceedingly small. The relationship between the probit and percentage (or
probability) of response1 is illustrated in Figure A.4 and Table A.1 and the effect of the probit
transformation on the cumulative response curve is illustrated in Figure A.5.
y = A + B lnx
Where the toxic load function comprises more than one variable the equation becomes:
y = A + B1 ln x1 + B2 ln x2 etc.
where A, B1, B2, etc are regression coefficients. For most toxic substances the variables under
examination are the concentration (c) and the time (t), so the equation becomes:
y = A + B1 ln C + B2 ln t
y = A + B2 ln Cnt
where
n = B1
B2
To obtain the parameters from experimental data a graphical method may be employed but
more rigorous determinations usually involve such as the maximum likelihood estimation
(MLE) method or the method of minimising x2 (x2 is the weighted sum of squares of the
difference between the experimental and expected probit values).
The MLE method1 involves an iterative process which is usually carried out by means of
successive weighted linear regressions. Such regressions generally rapidly converge with the
limit being the maximum likelihood estimate of the probit values. The weighting of the
regression involves an expression of the form nw where n is the number of animals in a
sample and w is a weighting coefficient, values of which are symmetrical about y = 5 and
decrease towards y = 0 and y = 10 (see Table A.3).
A measure of the limits within which the true parameter lies is generally of value and the
method of fiducial limits (usually the 95% fiducial limit) is the one usually adopted. In this
method there is said to be a fiducial probability (F) that the true value of a parameter lies
between upper and lower limits which would not be contradicted by a significance test at the
1-F probability level. To a reasonable approximation the fiducial limits can be calculated as:
+_t V
where t is the normal deviate for the probability to be used (eg 1.96 for 95% probability, see
Table A.4) and V is the variance. For example:
2
11 (m- X )
V(m) = 2 [ + ]
b ∑ nw ∑ nw(x- x )
2
2
g t 1- g (m- x )
m+ (m- x ) + _ [ + ]
1- g b(1 - g) ∑ nw ∑ nw(x- x )2
where
t
g=
b ∑ nw(x- x )
2 2
If g < 0.1 the approximation based on the variance formula does not give rise to significant
error.
If the fiducial limits are plotted as bands along a probit regression line they will form two
hyperbolic curves, convex to the line and approaching most closely at 50% response (eg. see
Figure A.6).
A.2 REFERENCES
/108/ . Finney DJ (1971), Probit Analysis, 3rd Edition, Cambridge University Press
/109/ Bliss CI, Estimating the Dosage-Mortality Curve, J Econ Ent 28 (1934) 646-7
Table A- 2 Transformation of percentages to probits
% 0 1 2 3 4 5 6 7 8 9
0 - 2.67 2.95 3.12 3.25 3.36 3.45 3.52 3.59 3.66
10 3.72 3.77 3.82 3.87 3.92 3.96 4.01 4.05 4.08 4.12
20 4.16 4.19 4.23 4.26 4.29 4.33 4.36 4.39 4.42 4.45
30 4.48 4.50 4.53 4.56 4.59 4.61 4.64 4.67 4.69 4.72
40 4.75 4.77 4.80 4.82 4.85 4.87 4.90 4.92 4.95 4.97
50 5.00 5.03 5.05 5.08 5.10 5.13 5.15 5.18 5.20 5.23
60 5.25 5.28 5.31 5.33 5.36 5.39 5.41 5.44 5.47 5.50
70 5.52 5.55 5.58 5.61 5.64 5.67 5.71 5.74 5.77 5.81
80 5.84 5.88 5.92 5.95 5.99 6.04 6.08 6.13 6.18 6.23
90 6.28 6.34 6.41 6.48 6.55 6.64 6.75 6.88 7.05 7.33
- 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
99 7.33 7.37 7.41 7.46 7.51 7.58 7.65 7.75 7.88 8.09
Table A- 4 Distrubution of t
Degrees Probability
Of .90 .70 .50 .30 .10 .05 .02 .01 .001
Freedom
1 .16 .51 1.00 1.96 6.31 12.7 31.8 63.7 637.
2 .14 .44 .82 1.39 2.92 4.30 6.96 9.92 31.6
3 .14 .42 .76 1.25 2.35 3.18 4.54 5.84 12.9
4 .13 .41 .74 1.19 2.13 2.78 3.75 4.60 8.61
5 .13 .41 .73 1.16 2.02 2.57 3.36 4.03 6.86
6 .13 .40 .72 1.13 1.94 2.45 3.14 3.71 5.96
7 .13 .40 .71 1.12 1.90 2.36 3.00 3.50 5.40
8 .13 .40 .71 1.11 1.86 2.31 2.90 3.36 5.04
9 .13 .40 .70 1.10 1.83 2.26 2.82 3.25 4.78
10 .13 .40 .70 1.09 1.81 2.23 2.76 3.17 4.59
12 .13 .40 .70 1.08 1.78 2.18 2.68 3.06 4.32
14 .13 .39 .69 1.08 1.76 2.14 2.62 2.98 4.14
16 .13 .39 .69 1.07 1.75 2.12 2.58 2.92 4.02
18 .13 .39 .69 1.07 1.73 2.10 2.55 2.88 3.92
20 .13 .39 .69 1.06 1.72 2.09 2.53 2.84 3.85
22 .13 .39 .69 1.06 1.72 2.07 2.51 2.82 3.79
24 .13 .39 .68 1.06 1.71 2.06 2.49 2.80 3.74
26 .13 .39 .68 1.06 1.71 2.06 2.48 2.78 3.71
28 .13 .39 .68 1.06 1.70 2.05 2.47 2.76 3.67
30 .13 .39 .68 1.06 1.70 2.04 2.46 2.75 3.65
40 .13 .39 .68 1.05 1.68 2.02 2.42 2.70 3.55
60 .13 .39 .68 1.05 1.67 2.00 2.39 2.66 3.46
120 .13 .39 .68 1.04 1.66 1.98 2.36 2.62 3.37
In this example several probits (with A values of -40, -30, -25, -20, and -10) all give
the same LC50 value as the B coefficients have compensated.
Figure A-1 Example of the Relationship Between a Probit Function and the Percentage
Fatality Curve
In this example several probits (with A values of -40, -30, -25, -20, and -10) all give the same
LC50 value as the B coefficients have compensated.
The normal sigmoid curve in Figure A.2 is transformed to a straight line when the ordinates
are measured on a scale linear in probits instead of in percentages
APPENDIX B:
ERPG VALUES
B1 ERPG VALUES
Toxicity data of some form exist for a wide range of substances and it would be, clearly,
impractical to present data for each of them here. Consequently data are presented on
substances which are more commonly studied by DNV. ERPG values published by the AIHA
are provided in Table B-1
Table B-1 Extract from Current AIHA ERPG values (2008)
Chemical MW
Name CAS Number g/mol ERPG-1 ERPG-2 ERPG-3
1,2-dichloroethane 107-06-2 99 50 ppm 200 ppm 300 ppm
1,2-propylene oxide 75-56-9 58.1 50 ppm 250 ppm 750 ppm
2,4-Toluene diisocyanate 584-84-9 174 0.01 ppm 0.15 ppm 0.6 ppm
Acrolein 107-02-8 56.1 0.05 ppm 0.15 ppm 1.5 ppm
Acrylonitrile 107-13-1 53.1 10 ppm 35 ppm 75 ppm
Allyl Chloride 107-05-1 76.5 3 ppm 40 ppm 300 ppm
Ammonia 7664-41-7 17 25 ppm 150 ppm 750 ppm
Arsine* 7784-42-1 78 NA 0.5 ppm 1.5 ppm
Benzen* 71-43-2 78.1 50 ppm 150 ppm 1000 ppm
Benzyl Chloride 100-44-7 126.6 1 ppm 10 ppm 50 ppm
Bromine 7726-95-6 159.8 0.1 ppm 0.5 ppm 5 ppm
Carbon Disulfide 75-15-0 76.14 1 ppm 50 ppm 500 ppm
Carbon Monoxide* 630-08-0 28 200 ppm 350 ppm 500 ppm
Carbon Tetrachloride 56-23-5 153.8 20 ppm 100 ppm 750 ppm
Chlorine 7782-50-5 70.9 1 ppm 3 ppm 20 ppm
Crotonaldehyde 4170-30-3 70.1 0.2 ppm 5 ppm 15 ppm
epichlorohydrin 106-89-8 92.5 5 ppm 20 ppm 100 ppm
Ethylene Oxide 75-21-8 44.1 NA 50 ppm 500 ppm
Fluorine* 7782-41-4 38 0.5 ppm 5 ppm 20 ppm
Formaldehyde* 50-00-0 30 1 ppm 10 ppm 25 ppm
Hydrazine 302-01-2 32.05 0.5 ppm 5 ppm 30 ppm
Hydrogen Chloride* 7647-01-0 36.5 3 ppm 20 ppm 150 ppm
Hydrogen Cyanide 74-90-8 27 NA 10 ppm 25 ppm
Hydrogen Fluoride** 7664-39-3 20 2 ppm 20 ppm 50 ppm
Hydrogen Peroxide* 7722-84-1 34.01 10 ppm 50 ppm 100 ppm
Hydrogen Sulphide 7783-06-4 34.1 0.1 ppm 30 ppm 100 pmm
Methanol 67-56-1 32.04 200 ppm 1000 ppm 5000 ppm
Methyl Bromide 74-83-9 95 NA 50 ppm 200 ppm
Methyl Isocyanate 624-83-9 57.5 0.025 ppm 0.25 ppm 1.5 ppm
Methyl Mercaptan 74-93-1 48.1 0.005 ppm 25 ppm 100 ppm
Nitrogen dioxide 10102-44-0 46 1 ppm 15 ppm 30 ppm
perfluoroisobutene 382-21-8 200 NA 0.1 ppm 0.3 ppm
Phenol 108-95-2 94.1 10 ppm 50 ppm 200 ppm
Chemical MW
Name CAS Number g/mol ERPG-1 ERPG-2 ERPG-3
Phosgene 75-44-5 98.9 NA 0.2 ppm 1 ppm
Phosphine 7803-51-2 34 NA 0.5 ppm 5 ppm
Sulphur dioxide 7446-09-5 64 0.3 ppm 3 ppm 15 ppm
3 3 3
sulphuric acid 7664-93-9 98.1 2 mg/m 10 mg/m 30 mg/m
Toluene 108-88-3 92.14 50 ppm 300 ppm 1000 ppm
Vinyl Chloride 75-01-4 62.5 500 ppm 5000 ppm 20,000 ppm
Note: this information is correct as of January 1 2008. It is possible that an ERPG under
review by the Committee will be balloted and approved in 2008, making it eligible for
inclusion in the 2009 ERPF Document Set.
To convert ERPG values from ppm to mg/m3 use the following equation:
ERPG(mg/m3) = ERPG(ppm) MW / 24.45
B2 REFERENCES
/110/ American Industrial Hygiene Association web page,
http://www.aiha.org/1documents/Committees/ERP-erpglevels.pdf, 29 September 2008
APPENDIX C:
GENERATION OF TOXICITY
PROBITS
The `HSE' approach has the advantage of using data points at several mortality levels to
produce `real' probit constants whilst the second, TNO, approach may be useful if only a
single LC50 datum exists for a particular substance.
Other effects, including mutagenicity, carcinogenicity and teratogenicity, may also arise as a
result of a single exposure. However, for most Major Hazards substances, single exposure
dose-effect data are not available for effects such as these.
Experience has shown that commonly used secondary sources of information may be
unreliable, in that the toxicological values given may be inaccurate representations of the
original results, or that the primary sources of such values are either difficult to verify or of
doubtful quality. Therefore, all the data used in the assessment of individual substances
should be obtained from the original reports.
However, for most substances, existing reliable data on acute effects arising from a single
exposure in humans are sparse.
For a few substances some information is available from their use in warfare (eg, chlorine,
phosgene). Nevertheless, for most substances the data are limited to a few reports of
accidental exposures, often involving only a few people and rarely containing accurate
measurements or even estimates of exposure concentrations and times.
In general, extrapolation from laboratory animals to humans with any degree of accuracy and
reliability is fraught with difficulties, principally because of the inadequate information /2/.
Even so, for most substances it is be necessary to make the assumption that results from
animal experiments will be representative of effects on the human population, in terms of both
the nature of the effects produced and the dose-effect relationships observed.
In some cases the paucity of data on certain substances will make any analysis extremely
tenuous, and in these situations further experimental work by manufacturers or their trade
associations would be advisable if important decisions depend on the results.
From the available data, the most sensitive animal species and strain would normally be used
to represent the prediction of human responsiveness, ie the lowest LC50 value and associated
exposure time will be taken as the corresponding values for the human population, unless
there is information indicating that other animal results will serve to model human
responsiveness more reliably.
analysis, then species variation will also exert an unknown degree of influence on the derived
relationship between C and t.
Acute inhalation toxicity experiments in laboratory animals, performed in the early years of
this century on a limited number of gases, obtained mortality results suggesting the following
relationship:-
More recently, a literature survey of more than 30 substances, for which LC50 values had been
determined in the same species for at least three different exposure periods, suggested two
groups of substances, showing empirical relationships of:-
Overall, experimental observations suggest that in many cases the following general
relationship may hold for acute lethality:-
The requirement at this stage is, therefore, knowledge of the value of n which will relate
variations in C and t to a constant, experimentally observed level of mortality within an
individual animal species. The most suitable reference point is usually 50% mortality, since
exposure conditions relating to this level of mortality are most readily available. A plot of 1n
c against 1n t, each pair of c and t values relating to the production of 50% mortality, will
permit derivation of n from the slope of the resulting line (-1/n) since
Cnt = constant, k
In some cases the data available on a substance may be insufficient to permit the derivation of
n. In such situations risk analyses using the two values of n commonly observed, 1 and 2,
should be compared. Using the value of n and concentration/time combinations for two
different mortality levels, the values for `A' and `B' can be determined.
C t (Va )
D"=
Ar
where
C = concentration (mg/m3)
Lung area and breathing rate can be derived from body weight7
Va = u W0.7
Ar = v W0.92
To account for variations in absorption efficiencies of the nasal passages of test animals and
the mouth of humans (humans breathe more through the mouth) an arbitrary safety factor of 5
is introduced. This is then multiplied by 2 to account for the fact that the test animals would
be at rest whereas humans, exposed to an accidental release of gas, would not.
For systemically acting substances the breathed in dose is expressed in terms of body weight
D1 = u W-0.3
Using average body weights for several different test animals species and humans, and the
factors determined above, extrapolation factors are produced for both locally acting and
systemically acting substances. The conclusion is drawn by TNO /112/ that there is no
significant difference between the two.
It is our opinion, however that this is a tenuous conclusion to draw when the data are based on
arbitrarily chosen safety factors which have no proven basis.
Pr = A + B ln (Cnt)
The interspecies differences are assumed to affect only the `A' term and not the `B' or `n'
terms. The `n' term is either known or assumed to be 2. The `B' term is known to lie between
1.1 and 6.1 although values outside this range are given in tables later in the TNO report. A
value of B = 1 is chosen for all probits for all substances. This is described as conservative
below LC50 concentrations since it takes into account the greater spread in sensitivity in a
human population as compared to laboratory animals.
Using the above assumptions, the determination of a probit is reduced to the determination of
the `A' term. This is achieved using the figure reproduced here as Figure C.1. A table of
probit functions derived for a selection of materials is given in Table C.1. Please note that the
units of concentration employed are mg/m3 not ppm.
It should be noted that this mechanistic treatment, although attractive to engineering studies,
is not at all supported by toxicologists. In fact it should only be adopted in extremes where
there is very little data available. The great uncertainty under these circumstances must be
noted in the study report.
C4 REFERENCES
/111/ Haber F., Funf Vortrage aus den Jaren 1920 - 1923, Springer - Verlag, Berlin 1924.
/112/ TNO, Methods for the Determination of Possible Damage to People and Objects
Resulting from Releases of Hazardous Materials, CPR 16E, Voorburg, December 1989.
/113/ ten Berge W.F. and Guldemond C.P., Required Knowledge for the Evaluation of Health
Hazards from Acute Inhalatory Exposure of Humans, International Process Safety
Group, Cannes, September 1986.
- o0o -
APPENDIX D:
CONVERSION OF PROBITS
FOR DIFFERENT UNITS
(MG/M3 AND PPM)
298
volume = 22.4 x = 24.45 m3
273
X kg 24.45 X m3 24.45 X
Concentration = 6 3
= 6 3
= ppm
10 m MW 10 m MW
or
MW mg
1 ppm =
24.45 m3
Therefore;
Pr = B ln (C n t) + A C as mg/ m3
MW n
Pr = B ln (Cn t) + B ln +A C as ppm
24.45
MW n
Pr = B ln (C n t) + B ln
+ A C as ppm
24.45
Consequently, only the effective value of A changes since B, n and MW are constants. In
PhastRisk and Phast, a reference temperature of 9.85°C is used at which temperature a kg mol
occupies 23.21m3. - o0o -
APPENDIX E:
PROBITS OF MIXTURES
If a mixture at concentration Cmix in air contains a mole fraction, X, of toxic gas (C1) and the
standard probit for the toxic gas is:
Pr = A1 + B1 ln(C1n t)
Pr = A1 + B1 ln ([X.Cmix]n t)
One approach is to add the concentrations of the toxic elements of the mixture together and
treat this new concentration as the concentration of the most toxic component. This technique
is thought to be conservative and works tolerably well when the most toxic element of the
mixture is also the most abundant. However, if the more toxic element makes up only a small
proportion of the mixture, then the results can be overly conservative.
An alternative approach is to assume that the toxicities of the individual elements of the
mixture are not additive and to treat them as separate sources of toxicity. This however may
understate the extent of hazard of a toxic mixture.
If single point toxicity criteria are being used (eg TLV, ERPG) then the toxic criteria for the
mixture has been modelled as:
1 X1 X2 Xn
= + + ........... +
Toxicity mix Toxicity1 Toxicity 2 Toxicity n
For conservatism, the method of adding concentrations is recommended until a more rigorous
method is published.
- o0o -
APPENDIX F:
SUGGESTED APPROACH TO
IMPAIRMENT AND
INCAPACITATION DUE TO SMOKE
F1 INTRODUCTION
This appendix contains a suggested approach for calculation of cumulative effect from smoke,
both with respect to impairment and incapacitation. The approach attempts to take account of
the synergetic effect between CO and CO2 as well as additive effects from toxic gases,
oxygen depletion and high temperature.
Equations combining fractions of dose for each smoke components and temperature to a
cumulative dose have been developed in section 8 above. These are in the following used to
generate suggested criteria for typical smoke compositions.
A similar approach is suggested for incapacitation (and subsequent death), i.e. the fraction of
the incapacitation doses are added up to form a total dose. The basic complication here is that
the available equations for dose fractions expresses relationships between concentrations and
impairment, but not for incapacitation. This is suggested allowed for by introducing
reduction factors for the individual dose fractions. The reduction factors are fixed so that
lethal doses give dose fractions equal to 1.0 for each individual contributor.
Refer to Section 8 for descriptions of smoke effects and limits for exposure causing death.
F2 APPROACH
The effects of toxic gases depend on the dose of product effectively received . The method
described in this Appendix is documented in the SFPE Handbook for Fire Protection
Engineering (/94) and in the ISO/TS 13571:2002 technical specification. It is used in the
DNV software SOQRATES (an excel based tool for offshore QRA) for impairment of TR
calculations.
It is based on the use of the fractional effective dose (FED) method which assumes that the
different components of fire work in an additive way and also that the effect is additive over
time. For this method, the received doses for small periods of time during a fire are expressed
as a fraction of the dose causing the toxic effect, and these FEDs are added together to give an
estimation of the combined effect. When the dose causing the toxic effect becomes greater
than 1 impairment is assumed to occur
The following equation gives the FED for CO gas taken into account the increased breating
rate due to increased CO2 levels. This approximation is only valid for exposure up to one
hour.
ISO/TS 13571:2002 gives the following relation for the fraction of the incapacitating dose:
FEDnarcotics = ∑ +
(
C CO in * 10 4 exp C HCN in * 10 4 / 43 ) * ∆t * V
60 CO 2 F1
∆t 35,000 220
Neglected
Where
∆t = time increment [s]
CCO in = average concentration of CO [%] over the time increment ∆t
( C HCN in = average concentration of HCN [%] over the time increment ∆t )
VCO 2 = frequency factor due to hyperventilation,
C
VCO 2 = 1 if C CO 2 in < 2%, VCO 2 = exp CO 2 in if C CO 2 in > 2%
5
CCO * 104 1
FCO = * VCO 2 * * t F2
35000 60
Where
t Is time in seconds
CCO 2
= exp( ) if CCO2 >2%
5
For short exposures at high concentrations when the blood concentration is well below saturation
level, the Stewart-Peterson equation has been derived from experimental human exposures (Stewart et
Al, 1973):
∆t
( ) (
%COHb = ∑ 3.317 * 10 −5 * C CO in * 10 4 )
1.036
( )
* Bactual * 60 * 10 3 *
60
F3
∆t
Where:
%COHb = approximate concentration in blood of carboxyhaemoglobin [%]
C CO in = average concentration of CO [%] over the time increment ∆t
Bactual = Bnominal * VCO 2 , actual breathing rate [m3/s]
Where:
Bnominal = nominal breathing rate supplied [m3/s]
C
VCO 2 = 1 if C CO 2 in < 2% , VCO 2 = exp CO 2 in if C CO 2 in > 2%
5
[L/min] [m³/s]
Rest 8.5 0.000142
Light effort 25 0.000417
Heavy effort 50 0.000833
Table F.1 typical breathing rates (B. Karlsson and J. Quintiere 1999)
Hence,
∆t
( )(
)
%COHb = 3.317 *10 −5 * 60 *103 * Bnominal * CCO in *10 4( )1+ 0.036
*
*VCO 2 F4
60
(
%COHb = (3.317 * 10 −5 )* (60 * 10 3 )* Bnominal * (35,000 * FEDnarcotics ) * CCO in *10 4 )
0.036
F5
Therefore, the fraction of the incapacitating dose at time t can be related to the
concentration of carboxyhaemoglobin present in blood at time t :
%COHb(t )
FEDnarcotics (t ) ≈ F7
95,000 * Bnominal
Where:
%COHb = approximate concentration in blood of carboxyhaemoglobin [%]
Bnominal = nominal breathing rate supplied [m3/s]
Equation F7 is used as basis for the smoke integrity model in SOQRATES. Equation
F2 may be used as they give approximately the same answer see section F3
EXAMPLE Calculation
The impairment dose for carbon dioxide is expressed as a fraction of the smoke impairment
dose:
exp(0.5189 * C CO2 ) 1
F CO 2 = * *t F8
475 60
Where
t is time in seconds
The impairment dose for oxygen starvation is expressed as a fraction of the smoke
impairment dose:
Equation F3
exp(0.54 * (20.9 − C O 2 )) 1
F O2 = * *t F9
3395 60
Where
O2 is concentration of oxygen in
smoke/air mixture in %
t is time in seconds
t
F Temp = (5.1849 -0.0273xC)
F10
60 x e
Where
t is time in seconds
Relevant equations for impairment and incapacitation are used to calculate the respective dose
fractions.
F3 EXAMPLE CALCULATION
The approach to calculate impairment or incapacitation from smoke is to:
• Determine smoke composition near fire
• Calculate details of dispersion of smoke (concentrations at points of interest)
• Calculate dose fraction for each component
• Add dose fractions and compare to criteria of 1.0
In the following, example compositions for smoke are presented, and calculations of
impairment and incapacitation doses are shown as examples
Case
Table F- 1Error! AutoText entry not defined. Initial Gas Concentrations in Smoke (Bonn,
1993)
GAS WELL VENTILATED/ UNDER VENTILATED/
FUEL CONTROLLED VENTILATION CONTOLLED
Gas fire Liquid fire Gas fire Liquid fire
%CO 0,04 0,08 3 3,1
%CO2 10,9 11,8 8,2 9,2
%O2 0 0 0 0
Particulates 15 47 29 70
(dB/m)
VCO 2 = 1 if C CO 2 in < 2% ,
C
exp CO 2 in if C CO 2 in > 2%
5
8.38
This gives VCO 2 = exp = 5.34
5
∆t
( )( )
%COHb = 3.317 *10 −5 * 60 *103 * Bnominal * CCO in *10 4 ( )
1+ 0.036
*
*VCO 2
60
( ) ( )
(
%COHb = 3.317 * 10 −5 * 60 * 10 3 * 2.94 * 10 − 4 * 0.0568 * 10 4 )
1+ 0.036
*
60
* 5.34
60
%COHb = 2.39
%COHb(t )
FED narcotics (t ) ≈
2.39
= = 0.08
95,000 * Bnominal 95,000 * 2.95 E − 4
C CO * 10 4 1 0.0568 * 10 4 1
FCO = * VCO 2 * * t = * 5.34 * * 60 = 0.087
35000 60 35000 60
Using the above equation for different exposure times we get the graph as presented in Figure
F-1. Implementation of the equations for CO CO2 and O2 for a 37oC smoke the total FED for
different smoke concentrations are given in Figure F-2 and F-3
Please note that the temperature comonent is not included in figure F-2 and F-3 and the effect
of high smoke temperature must be evaluated separatey. The incapasitation dose as a function
of smoke temperatre is given in Figure F-4
0.9
0.8
60sec. FC gas
300sec. FC gas
0.7
900sec. FC gas
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Smoke concentration ( fraction)
1.2
60sec. FC gas
60sec. FC oil
0.8 300sec. FC gas
300sec. FC oil
900sec. FC gas
F total
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Smoke concentration ( fraction)
Figure F- 2 Fractional effective dose for exposure of 37oC smoke from fuel controlled
fires (Fco +Fco2 + Fo2) ( effect of high smoke temperature is not included
1.2
60sec. VC gas
60sec. VC oil
0.8 180sec. VC gas
180sec. VC oil
300sec. VC gas
F total
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Smoke concentration ( fraction)
Figure F- 3 Fractional effective dose for exposure of 37oC smoke from ventilation
controlled fires (Fco +Fco2 + Fo2) ( effect of high smoke temperature is not included
1
0.9
60 sec exposure
0.8
30 sec exposure
120 sec exposure
0.7
Fractiopn of incapasitation
0.6
0.5
0.4
0.3
0.2
0.1
0
0 50 100 150 200 250 300
Temperature (C)
APPENDIX G
DERIVING SINGLE VALUE CRITERIA FROM PROBITS
G1 INTRODUCTION
This Appendix describes the approach used to derive single value criteria in this TN. The
approach is also recommended for application to deviations from the single value criteria in
the main report.
A single value criterion refers to the level of exposure (concentration, heat load etc.) which
cannot be sustained for a defined period of time, without causing death, (or alternatively
another defined level of harm).
G2 METHODOLOGY
The time taken to escape should be the average time taken for a person to escape from the
areas studied. Large differences in escape time may warrant several criteria to be developed.
The single value criterion developed in this TN relates to the risk to a person with a 50-
percentile vulnerability, i.e. a person who would be killed by the LD50 dose.
It may be argued that personnel on offshore platforms are relatively young and fit, have better
health and health care etc. compared to an average group of people in society. If this
argument is applied, they should be more resistant to the exposure of hazards and a higher
lethal dose may consequently be chosen, e.g. LD75 or LD90.
Using this level of exposure as the fatality criterion may however be optimistic, because the
methodology takes little account of the synergy between time to escape and effects of
exposure. Incapacitation (inability to escape from exposed area) will occur for lower levels of
exposure and/or shorter exposure times than the lethal dose would imply. Thus in reality,
although survival would be the result if escape had been completed, incapacitation could
occur, thereby causing extended exposure and a dose in excess of lethal dose.
The fatal level of exposure should therefore reflect the dose which causes incapacitation (and
thereby death) within the time to escape. It may be assumed that immediate rescue of
incapacitated personnel within the affected zone is not practicable. In reality, this may
depend on the severity of the accident and the availability of rescue equipment.
- o0o