You are on page 1of 4

Comparison of

TNO Multienergy and


Baker–Strehlow–Tang Models
Ali Sari, Ph.D., P.E.
Atkins Oil and Gas, 920 Memorial City Way, Suite 700, Houston, TX 77024; absasari@gmail.com (for correspondence)

Published online 22 December 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/prs.10424

Vapor cloud explosions (VCEs) cause considerable models [1]. However, there are limitations to the ex-
problems in the chemical and petrochemical indus- perimental data available for making such predictions
tries. They generate damaging levels of overpressure because large-scale explosion experiments are pro-
and the possibility of human injury/death, and build- hibitively expensive to perform [2]. Therefore, a vari-
ing/equipment damage. Predicting the possible conse- ety of prediction methods have been developed to
quences of VCEs is important to ensure the safe estimate the air blast parameters at any given distance
design of existing and new structures. Prediction of from a possible explosion source, to assess explosion
the overpressures resulting from a VCE is typically hazards, and to design safer structures.
done using simplified (empirical) models, phenome- Prediction of the overpressures resulting from a
nological models, and computational fluid dynamics VCE is typically done using one of three categories of
models. The focus of this article is on two of the most models: simplified models, phenomenological mod-
frequently used simplified prediction methods: TNO els, and computational fluid dynamics models. As
multienergy and Baker–Strehlow–Tang models. These Baker–Strehlow–Tang (BST) and the TNO multie-
models are compared in terms of structural response nergy (ME) models are the most frequently used blast
and vulnerability of damage caused by an explosion. overpressure simplified prediction methods, these
Ó 2010 American Institute of Chemical Engineers Pro- methods are compared in this study. TNO ME and
cess Saf Prog 30: 23–26, 2011 BST models are also compared in terms of structural
Keywords: vapor cloud explosions; blast overpres- response and vulnerability of damage caused by an
sure; structural response; structural damage; explo- explosion.
sion hazard
TNO ME MODEL
The multienergy concept [3–6] assumes that only
INTRODUCTION the part of the combustion energy present in the
Any release of a flammable fluid in a petrochemi- flammable cloud which is confined or obstructed
cal facility has the potential to generate a flammable contributes to pressure generation in the explosion.
vapor cloud that, if ignited, could produce a vapor In this way, the ME model takes account of the posi-
cloud explosion (VCE). Past accidents have demon- tive feedback mechanism of a gas explosion. The
strated that vapor cloud explosions are the most amount of energy released during a VCE is limited ei-
severe threat to refining and petrochemical industries. ther by the volume of the partially confined portion
Predicting the possible consequences of gas explo- of the flammable vapor cloud (if the flammable vapor
sions occurring in these industries is important to cloud is larger than the partially confined region) or
ensure the safe design of existing and new structures, by the volume of the vapor cloud (if the vapor cloud
and the risk assessment must be considered in the is smaller than the volume of the portion of the par-
development of their design. The predictions through tially confined space). In either case, the volume of
the assessment of such explosions are improved by the cloud within the partially confined space can be
carrying out experiments and by using theoretical converted into a hemisphere of equal volume. The
model treats the hemispherical cloud as a homogene-
Ó 2010 American Institute of Chemical Engineers ous, stoichiometric mixture of flammable gas and air,

Process Safety Progress (Vol.30, No.1) March 2011 23


with combustion energy of 3.1 3 106 J/m3 (the aver- vapor clouds [10]. With this method, the strength of
age heat of combustion of a stoichiometric mixture of the blast wave is proportional to the maximum flame
hydrogen and air) [7]. speed achieved within the cloud. Thus, each curve is
TNO has used a flux-corrected transport code to marked with a flame velocity, which is presented in
numerically simulate the explosion of a hemispheri- the form of a Mach number, Mf. The appropriate
cal, homogeneous, stoichiometric cloud, with con- flame speed (Mach number) for the specific situation
stant flame speed. The blast charts that determine being modeled is selected [10,11].
peak overpressure and the duration of the positive The fuel reactivity and the congestion or obstacle
phase form a family of curves relating the dimension- density are used in this calculation. The congestion
less overpressure to the combustion energy-scaled density is low if the area blockage ratio is below
distance [8]. The source strength index that is an 10%, medium if it is in the range of 10–40%, and high
overpressure level expected to occur in the if the blockage ratio is higher than 40%. Finally, the
obstructed region, varying from one for weak explo- confinement or flame expansion must be taken into
sion in an unobstructed and unconfined region to 10 account. A single plane is 2D. A pipe, culvert, or area
for detonation, is assigned a value to determine the with a roof and two vertical walls would be 1D. The
curves to use. The source strength depends on num- 2.5D confinement is defined as an area with limited
ber, type, and orientation of the obstacles present in 2D confinement, such as a roof and/or partial walls
obstructed region as well as the fuel reactivity. that might be expected to fail quickly and provide
The TNO ME model curves consist of positive venting, or an area without total 2D confinement
pressure and time duration as functions of distance. such as an elevated fin fan cooler [7,9].
The parameters were also nondimensionalized using
Sach’s scaling law with pressure, distance, and the
COMPARISON OF TNO ME AND BST METHODS
positive phase duration. These curves relate dimen-
In both the BST and the ME models, the source
sionless overpressure (overpressure divided by
energy is defined by a stoichiometric flammable
atmospheric pressure) to combustion energy scaled
cloud that is in a congested or partially confined
distance, which is calculated as shown in Eq. 1.
region [12]. Both models are based on dimensionless
overpressure and positive impulse as a function of
R
R ¼ ð1Þ energy-scaled distance. Although Eq. 1 is used to cal-
ðE=P0 Þ1=3 culate the scaled distance for both methods, the defi-
nition of R differs. R is defined as the distance from
where R is the dimensionless combustion energy- the center of the hemispherical explosion in the TNO
scaled distance, R is the distance from the edge of ME model. R is the distance from the actual conges-
the hemisphere in m, P0 is the ambient atmospheric tion center for the BST model. The flame speed for
pressure in J/m3, and E is the total available energy the BS model or initial explosion strength for ME
in J. The peak side-on overpressure DPS (N/m2) can model is determined by empirical approaches based
be calculated using the following relationship: on the degree of confinement and obstruction with
the source region as well as the distance available for
flame acceleration [9,12,13]. Both methods require the
DPS ¼ DPS =P0 ð2Þ
user to estimate the strength of the explosion as a
function of the reactivity of the flammable material
where DPS is the dimensionless peak overpressure. and the degree of confinement or congestion present
The tt time duration of the positive phase can be cal- in the cloud. This information is then used to deter-
culated using the following equation: mine which strength curve (in the case of the TNO
model) or flame speed (in the case of the BST
tt C0 model) is used to calculate the overpressure of the
tt ¼ ð3Þ
ðE=P0 Þ1=3 explosion as a function of distance from the center of
the explosion.
where C0 is atmospheric sound speed in m/s. The major difference between the ME and the BS
models is the method used to construct the graphical
relationship between dimensionless overpressure and
BST METHODOLOGY
combustion energy-scaled distance. The curves used
This model has some similarities with the TNO ME
in the BS model are based on numerical modeling of
model. Similar to TNO ME model, this model is based
constant flame velocities and accelerating flames
on dimensionless overpressure and positive impulse
spreading through spherical vapor clouds. With this
as a function of (a) energy-scaled distance from the
method, the strength of blast wave is proportional to
gas blast center, (b) the maximum flame speed using
the maximum flame speed, which is presented in the
the method based on the fuel reactivity, (c) degree of
form of a Mach number, achieved with the cloud [14].
confinement, and (d) obstacle density [9]. This
method uses the graphical relationship between
dimensionless and combustion energy-scaled dis- VULNERABILITY STUDY
tance, R. The curves used in the BST model are based This section includes the estimation of the vulner-
on numerical modeling of constant velocity flames ability of persons and structures to the physical
and accelerating flames spreading through spherical effects of a determined magnitude, which can be cal-

24 March 2011 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.30, No.1)
Table 1. Variation of the overpressure, impulse, vulnerability of a person, and the structural damage with the
distance—TNO MEM.

Ear Drum Death from Lung Structural


R0 (m) R0 (ft) P (psi) I (psi ms) Rupture (%) Hemorrhage (%) Damage (%)
10 33 58.8 940.5 100.0 100.0 100.0
60 197 5.9 125.4 45.1 0.0 98.5
150 492 2.2 82.3 2.2 0.0 24.7

Table 2. Variation of the overpressure, impulse, vulnerability of a person, and the structural damage with the
distance—BST model.

Ear Drum Death from Lung Structural


R0 (m) R0 (ft) P (psi) I (psi ms) Rupture (%) Hemorrhage (%) Damage (%)
10 33 21.3 318.2 99.1 54.4 100.0
60 197 6.6 135.8 54.1 0.0 99.4
150 492 2.6 69.8 4.8 0.0 44.0

Figure 1. Variation of the overpressure with the dis-


tance. [Color figure can be viewed in the online issue, Figure 3. Variation of vulnerability of persons with
which is available at wileyonlinelibrary.com.] distance—eardrum rupture.

Figure 2. Variation of the impulse with the distance.


[Color figure can be viewed in the online issue, Figure 4. Variation of vulnerability of persons with
which is available at wileyonlinelibrary.com.] distance—deaths from lung hemorrhage.

Process Safety Progress (Vol.30, No.1) Published on behalf of the AIChE DOI 10.1002/prs March 2011 25
structural damages—especially in short distan-
ces—with TNO ME are usually higher than with
BST.

LITERATURE CITED
1. J.S. Puttock, M.R. Yardley, and T.M. Cresswell,
J Loss Prev Process Ind 13 (2000), 419–431.
2. R.S. Cant, W.N. Dawes, and A.M. Savill, Annu Rev
Fluid Mech 36 (2004), 97–119.
3. W.P.M. Mercx and A.C. van den Berg, The explo-
sion blast prediction model in the revised CPR
14E (Yellow Book), Process Safety Prog 16 (1997),
152–159.
4. A.C. Van den Berg, J Hazard Mater 12 ( 1985),
Figure 5. Variation of structural damage (%) with dis- 1–10.
tance. 5. A.C. Van den Berg and A. Lannoy, J Hazard
Mater, 34 (1993), 151–171.
6. A.C. Van den Berg and N.H.A. Versloot, J Loss
culated with TNO ME and BST methods. In this arti- Prev Process Ind 16 (2003), 111–120.
cle, the Probit (probability unit) method is used. 7. J. Lobato, P. Cañizares, M.A. Rodrigo, C. Sáez,
and J.J. Linares, A comparison of hydrogen cloud
Scenario explosion models and the study of the vulner-
In this study, we assume a volume of 11,000 m3 ability of the damage caused by an explosion of
of ethylene, and a high congestion and 3D flame H2, Int J Hydrogen Energy 31 (2006), 1780–1790.
expansion are assumed for the overpressure and 8. AIChE, Guidelines for Evaluating the Characteris-
impulse predictions. A fuel burning velocity of 80 cm/ tics of Vapor Cloud Explosions, Flash Fires, and
s is used for ethylene (high reactivity). The explosion BLEVEs, American Institute of Chemical Engineers,
energy is estimated to be 3.85E 1 10 and 7.14E110 J New York, 1994.
for TNO MEM and BST methods, respectively. The se- 9. Q.A. Baker, M.J. Tang, E.A. Scheier, and G.J. Silva,
verity level for TNO MEM is determined as 10 and the Vapor cloud explosion analysis, Process Safety
estimated flame speed is Mach 5.2 for BST method. Prog, 15 (1996), 106–109.
The explosion type is a detonation for this scenario.
10. A.J. Pierorazio, J.K. Thomas, Q.A. Baker, and D.E.
Impulse and overpressure values for different distan-
Ketchum, An update to the Baker–Strehlow–Tang
ces are calculated by using TNO ME and BST models.
vapor cloud explosion prediction methodology
The percentages for eardrum rupture, death from lung
flame speed table. Process Safety Prog, 24 (2005),
hemorrhage, and structural damage are determined and
59–65.
compared using the probit equations.
11. Q.A. Baker, M.J. Tang, E.A. Scheier, and G.J. Silva,
Results and Discussion Vapor cloud explosion analysis, Proceedings of
Tables 1 and 2 summarize the variation of the over- the 28th AIChE Annual Loss Prevention Sympo-
pressure, impulse, vulnerability of persons, and the sium on Risk Analysis and Process Safety Manage-
structural damage with the distance R0 . It should be ment, Atlanta, GA, 1994.
noted that R0 is the distance from the congested area 12. M.J. Tang and Q.A. Baker, A new set of blast curves
edge. Figures 1 and 2 illustrate the variation of the over- from vapor cloud explosions, Proceedings of 33rd
pressure and impulse with distances, respectively. The AIChE Loss Prevention Symposium, Paper 29e,
variation of vulnerability of persons (ear drum rupture American Institute of Chemical Engineers, Hous-
and death from lung hemorrhage) and structural dam- ton, TX, March 14–18, 1999.
age with distance is provided in Figures 3–5. 13. Q.A. Baker and W.E. Baker, Pros and cons of TNT
equivalence for industrial explosion accidents, Pro-
CONCLUSIONS ceedings of the International Conference and
It is observed that the TNO ME method predicts Workshop on Modeling and Mitigating the Conse-
overpressure values significantly higher than BST quences of Accidental Releases of Hazardous Mate-
model at short distances. Nevertheless, a relatively rials, New Orleans, Louisiana, May, 1991.
good agreement is achieved between the TNO ME 14. D.J. Park and Y. S. Lee, A comparison on predic-
and BST models at longer distances. It is also tive models of gas explosions, Kor J Chem Eng,
found out that the calculated vulnerabilities and 26 (2009), 313–323.

26 March 2011 Published on behalf of the AIChE DOI 10.1002/prs Process Safety Progress (Vol.30, No.1)

You might also like