Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Contents lists available at ScienceDirect
Journal of Loss Prevention in the Process Industries
journal homepage: www.elsevier.com/locate/jlp
Does your facility have a dust problem: Methods for evaluating dust
explosion hazards
Scott G. Davis*, Peter C. Hinze, Olav R. Hansen, Kees van Wingerden
GexCon US, 7735 Old Georgetown Rd, Suite 1010, Bethesda, MD 20814, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 11 January 2011
Received in revised form
3 June 2011
Accepted 7 June 2011
The hazards of dust explosions prevailing in plants are dependent on a large variety of factors that
include process parameters, such as pressure, temperature and ow characteristics, as well as equipment
properties, such as geometry layout, the presence of moving elements, dust explosion characteristics and
mitigating measures. A good dust explosion risk assessment is a thorough method involving the identication of all hazards, their probability of occurrence and the severity of potential consequences. The
consequences of dust explosions are described as consequences for personnel and equipment, taking into
account consequences of both primary and secondary events.
While certain standards cover all the basic elements of explosion prevention and protection,
systematic risk assessments and area classications are obligatory in Europe, as required by EU ATEX and
Seveso II directives. In the United States, NFPA 654 requires that the design of the re and explosion
safety provisions shall be based on a process hazard analysis of the facility, process, and the associated
re or explosion hazards. In this paper, we will demonstrate how applying such techniques as SCRAM
(short-cut risk analysis method) can help identify potentially hazardous conditions and provide valuable
assistance in reducing high-risk areas. The likelihood of a dust explosion is based on the ignition
probability and the probability of ammable dust clouds arising. While all possible ignition sources are
reviewed, the most important ones include open ames, mechanical sparks, hot surfaces, electric
equipment, smoldering combustion (self-ignition) and electrostatic sparks and discharges. The probability of dust clouds arising is closely related to both process and dust dispersion properties.
Factors determining the consequences of dust explosions include how frequently personnel are
present, the equipment strength, implemented consequence-reducing measures and housekeeping, as
risk assessment techniques demonstrate the importance of good housekeeping especially due to the
enormous consequences of secondary dust explosions (despite their relatively low probability). The
ignitibility and explosibility of the potential dust clouds also play a crucial role in determining the overall
risk.
Classes describe both the likelihood of dust explosions and their consequences, ranging from low
probabilities and limited local damage, to high probability of occurrence and catastrophic damage.
Acceptance criteria are determined based on the likelihood and consequence of the events. The risk
assessment techniques also allow for choosing adequate risk reducing measures: both preventive and
protective. Techniques for mitigating identied explosions risks include the following: bursting disks and
quenching tubes, explosion suppression systems, explosion isolating systems, inerting techniques and
temperature control. Advanced CFD tools (DESC) can be used to not only assess dust explosion hazards,
but also provide valuable insight into protective measures, including suppression and venting.
2011 Elsevier Ltd. All rights reserved.
Keywords:
Dust explosion
Risk assessment
Process safety
Risk reducing measures
Dust Explosion Simulation Code (DESC)
Explosion modeling
1. Introduction
Dust explosions are a continuous threat in facilities producing
combustible powders and dust as nal and intermediate products.
* Corresponding author. Tel.: 1 617 407 3300.
E-mail address: [email protected] (S.G. Davis).
0950-4230/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jlp.2011.06.010
Regrettable recent examples include the 2003 explosion at West
Pharmaceutical Services in Kinston, North Carolina (killing 6), the
2008 explosion at Imperial Sugar Plant in Savannah, Georgia
(killing 14), and one year later the explosion in a coal silo at WE
Industries power plant injuring 7 in Oak Creek, Wisconsin (2009).
Along with these serious accidents are many smaller industrial dust
explosion accidents causing limited damage and only minor or no
838
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Fig. 1. Dust layer accumulation under a silo (left) and due to excessive material leaking out of process equipment (right).
injuries. Some of these smaller incidents, however, could have led
to more serious consequences.
Dust explosion risks prevailing in industrial facilities are dependent on a large variety of factors that include process parameters,
such as pressure and temperature, as well as equipment properties,
such as the presence of moving elements, the mechanical strength of
such dust handling equipment, dust explosion characteristics, and
mitigating measures taken, including housekeeping and protective
measures such as explosion venting or suppression.
In this document a semi-quantitative short-cut risk analysis
method (SCRAM) is presented, allowing for the assessment of dust
explosion risks and choosing adequate preventive and protective
measures. The performance of an analysis as described here would
make industry aware of the most hazardous areas in their facilities
and associated consequences in case of an explosion.
The method is described and an application example presented.
The example demonstrates the strength of the method and the
support it offers to industry for choosing appropriate risk mitigating measures. Lastly, examples showing how advanced CFD
codes like DESC can be used to assess explosion hazards and
explosion venting are presented.
Fig. 2. Dust accumulation on elevated, horizontal surfaces such as ducts, beams, cable
trays, etc.
2. Description of the short-cut risk analysis method
This chapter describes the methodology used to determine the
risk for dust explosions in industrial facilities. The risk for a dust
explosion is the product of the probability of a dust explosion
occurring and the consequences of the dust explosion. The consequences can be divided into primary consequences such as failure
Fig. 3. Outdoor lter installation recycling process air back indoors.
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Fig. 4. Bonded and grounded piece of process equipment.
of the piece of equipment in which the dust explosion occurs and
secondary consequences such as an ensuing re and secondary
explosions in connected equipment or in the working area due to
whirling up and subsequent ignition of dust layers there. Fig. 1
shows examples of dust layers that have accumulated outside of
equipment and elevated horizontal surfaces (shown in Fig. 2),
which would be considered when evaluating secondary explosion
risks.
Another process for which secondary explosions should be
considered is when the air of material separators is recycled back
into the facility, with a potential of escalating an explosion in the
equipment back into the facility if explosible atmospheres are
present or isolation is not provided. An example of an outdoor lter,
which recycles the process air stream back into the facility, is
shown in Fig. 3.
2.1. Estimating the probability of an explosion occurring
For a dust explosion to occur, a ammable atmosphere must be
present simultaneously with a competent ignition source. The dust
839
concentration in this atmosphere must exceed a certain threshold
limit, typically 30 g/m3, with a particle size distribution sufciently
small. Dust with particle size distributions from 10 to 40 microns
and concentration ranges from 250 to 1500 g/m3 have been shown
to have the highest propensity to ignite and to produce the most
severe explosions. Finer dust particles might produce explosions
that are more severe if the dispersion process has enough force to
break up the agglomerates and produce a dust cloud consisting of
primary particles.
To be able to quantify the probability for the occurrence of an
explosive atmosphere, properties of the combustible material
should be considered, together with how likely it is that the
combustible material will be mixed with air. The probability of
a specic ignition source being able to ignite the explosive atmosphere is considered based on different criteria, such as the energy
released by the ignition source, the period in which this energy is
supplied, the surface temperature of the ignition source and its size.
For mechanically generated sparks, collision speed, friction, contact
time and physical properties of the colliding materials are included.
For electrical discharges, the facilities practices for grounding and
bonding need to be considered; an example of a bonded piece of
equipment is shown in Fig. 4.
Whether an ignition source is capable of igniting an explosive
atmosphere depends on several properties of the atmosphere, for
instance the fuel concentration and the turbulence level, and the
ignition properties of the explosive atmosphere (normally described
by the minimum ignition energy and minimum ignition temperature).
The factors mentioned above are considered individually and
form the basis for estimating how often an explosion can occur. It is
not possible to give the exact frequencies for an explosion. In a risk
analysis the probability for an explosive atmosphere and the
probability for an ignition source are ranged from I to V, where
I has the lowest probability and V has the highest probability.
Each range (I, II, III, IV and V) describes a range in probability or
frequency.
The probability of an explosion occurring depends on the
probability of the presence of an effective ignition source and the
probability of having an explosive atmosphere. The probability of
an explosion will be the product of these two probabilities (as long
as the two are generated independent from each other). Denitions
and explanations of the values used are described below.
Fig. 5. Examples of equipment improperly vented indoors.
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Consequence
840
IV
III
II
II
III
IV
Table 1
Denition of the probability and consequence for explosions under normal operation.
Probability of the formation of an explosive
atmosphere
Description
Very unlikely
Unlikely
Somewhat likely
Likely
Very likely
Range Da
I
II
III
IV
V
Probability of the formation of an effective ignition
source
Probability
Description
Very unlikely
Unlikely
Somewhat likely
Likely
Very likely
Range Di
I
II
III
IV
V
Fig. 6. Risk matrix.
The probability for a secondary event depends on the probability for the primary event and is normally lower than that of the
primary event.
Probability for an explosion to occur
Range Dx
I
II
2.2. Estimating the consequences of an explosion
The consequence for personnel (Dp) and equipment (De) is
estimated based on the expected effect of the explosion. This is
estimated based on expected damage caused by the heat, pressure
or loose items after the denitions given below. The consequence
for personnel and equipment from an explosion depends on the
explosion pressure and the heat intensity from the explosion.
Pressure build-up in enclosed units might cause the units to
rupture, resulting in heat radiation from ames, dispersion of
pressure waves and ying objects.
The strength of an explosion depends on several factors, such as
the initial conditions of the dust cloud, including the fuel concentration, initial turbulence and the position of the ignition source.
The properties of the combustible material are also important,
including chemical composition. The properties of the explosive
atmosphere will change over time; hence, the time of the explosion
is important for the explosion propagation.
Flames propagating out from a ruptured vessel release heat that
might injure personnel or cause damage to equipment. The
convective heat transfer during an explosion causes the most
severe burns. Burns/damage might be the result if personnel or
equipment are in direct contact with the explosion ame. This can
be especially true for explosions that are vented indoors. While the
explosion vent may adequately protect a given piece of equipment,
it may exacerbate injuries to personnel if vented in occupied areas.
Fig. 5 shows examples where a bucket elevator was improperly
vented indoors in the area of personnel (see ladder) as well as
another piece of equipment vented indoors.
Another concern is that of secondary explosions, which under
certain circumstances can be more severe than the primary
explosion. Secondary explosions are caused when the primary
containment system (duct, vessel, dust collector, etc.) ruptures due
to the primary explosion, and the resulting venting and blast wave
entrains and disperses neighboring dust layers into the air, which
subsequently ignite. Generally, the severity of secondary dust
explosions is related to the thickness and the area occupied by the
dust layers; hence housekeeping plays a key role in reducing the
hazards of secondary explosions.
III
IV
V
Denition
<1/10000 per year
>1/10000 per year
< 1/100 year
Somewhat likely >1/100 < 1/10 per year
Likely
>1/10 year < 1 per year
Very likely
>1 per year
Consequence for personnel and equipment
Range Dp De Description
I
Personnel
Equipment
II
Personnel
Equipment
III
Personnel
Equipment
IV
Personnel
Equipment
Personnel
Equipment
Denition
No injury.
Marginal damage to
process units. Process
shut down.
Limited injury.
Damage to process unit
(<$20,000).
Personnel injury.
Process unit collapse,
possible damage to
other units (>$20,000;
<$200,000).
Serious personnel injury,
possible loss of life.
Signicant damage to
several process units
(>$200,000;
<$2,000,000).
Loss of one or several
lives.
Plant fully damaged
(>$2,000,000).
2.4. Estimating the explosion risk
The explosion risk is the product of the probability of an
explosion occurring and its consequences. The risk level for
explosions can be estimated from the matrix given in Fig. 6 below,
based on the probability and consequence described in the above
section and the denitions provided in Table 1. The risk level
increases from E to A.
Table 2
Risk level e denitions and recommended acceptance criteria.
2.3. Denitions
The probability or the frequency of an explosion occurring and
the potential consequences are estimated from I to V, as described
previously. Denitions and descriptions of the different values are
given below.
Description
Very unlikely
Unlikely
A
B
C
D
E
Risk level
Acceptance criteria
Need for risk reducing measures (RRM)
Very high
High
Medium
Low
Very low
Unacceptable
Unacceptable
Medium
Acceptable
Acceptable
RRM
RRM
RRM
RRM
RRM
must be implemented
must be implemented
should be implemented
can be implemented
are not required
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
841
Table 3
Example of table summarizing the assessment of probability and consequences of a dust explosion in a process unit.
Process
unit
Example
Probability
of
flammable
atmosphere
IV
Probability of ignition
Equipment
(electric and
mechanical)
II
Hot
surfaces
Electric and
electrostatic
sparks and
discharges
Mechanical
sparks
Flames
and
smoldering
combustion
Probability
of
explosion
II
EXPOSURE TO EXPLOSION
PRIMARY EXPLOSION
Probability (injury/damage)
Consequence
Risk
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
II
III
III
SECONDARY INCIDENTS (inclusive explosions)
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
Comments:
EXAMPLE
2.5. Acceptance criteria
The risk level and the recommended acceptance criteria are
selected and based on the probability for human and economical
loss according to Table 1. The selected criteria are given in Table 2. It
should be emphasized that these acceptance criteria are proposed
levels only, and alternate criteria may also be chosen.
Fig. 6), reducing the consequences (going down in the risk matrix)
or a combination of probability and consequence-reducing
measures (most effective method).
In the application example given in this document, the
estimated probabilities and consequences are summarized in
tables(see Table 3 for an example). These tables also include estimates of ignition source probability and of the risk of secondary
incidents/events.
2.6. Risk evaluation and risk reducing measures
Comparing the determined risk, which is based on the acceptance criteria, determined from the probability and consequence of
an explosion as described above, one then decides whether riskreducing measures are necessary. Risk-reducing measures imply
reducing the probability of an event (going left in the risk matrix of
Process unit: The process unit the analysis applies to.
Probability: The estimated explosion probability is the product
of the probability for an explosive atmosphere and competent ignition source (see Table 3).
Consequence: The consequences for an event, considering both
personal injuries and damage to equipment. Both primary and
Fig. 7. Analyzed milk powder spray dryer installation.
842
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Table 4
Explosion properties of milk powder (Beck, Glinke, & Mohlman, 1997).
Explosion property
Value
Maximum explosion pressure Pmax (bar)
Dust explosion constant KSt (bar m/s)
Minimum ignition energy (MIE) (mJ)
Minimum ignition temperature (MIT) ( C)
Lower explosion limit (LEL) (g/m3)
6e7
80e130
>50
450e600
60e150
lter. The dust collected in the cyclones is returned to the uidized
bed by pneumatic transport.
The described spray dryer installation has not been provided/
equipped with any special preventive or protective measures. The
installation is located inside a building. Personnel are around the
installation only occasionally for inspection reasons.
3.1. Analysis
secondary consequences are given. Denitions for explosion
related probability (and consequences) are given in the above
section (see Table 3).
Risk: This is the product of probability and consequence. Both
the risks of primary and secondary events are estimated. See
Table 2 for acceptance criteria.
Ignition source: Probabilities of occurrence of the ve most
common ignition sources are given (see Table 3).
3. Application example: a spray dryer installation for milk
powder
To demonstrate the method an analysis is performed for a spray
dryer installation used for drying milk powder (see Fig. 7). The total
height of the spray dryer is 15 m, and the height of the cylindrical
part is 6.3 m supported by a conical part (angle 60 ). To move dried
powder out of the conical part a pneumatic hammer has been
provided. The temperature of the hot air to dry the milk slurry is
200 C. The temperature of the air leaving the dryer is 90 C. Based
on air and product throughput, the average dust concentration in
the dryer would be 30 g/m3. The dried powder collected in the cone
of the spray dryer is transferred into a uidized bed for further
drying or cooling. The powder taken along with the air ow out of
the dryer is removed from the air by cyclones and a bag lter. Also
the air from the uidized bed is cleaned in cyclones and the bag
The analysis has been performed for the dryer only. To perform
the risk analysis the explosion properties of milk powder need to be
known. Although it is strongly preferred to have these properties
determined for the milk powder in question, the present study was
performed using literature data. This may lead to overly conservative preventive and protective measures resulting from the
analysis, since one would normally base the analysis on the most
conservative published values. On the other hand, an underestimate of the hazards may also be possible, especially for dusts where
only a limited set of explosion properties is available. For milk
powder, the use of literature data is acceptable since there is
a rather large number of well-described data available, with little
variation. The data found for milk powder are presented in Table 4
(Beck, Glinke, & Mohlman, 1997).
In addition to the properties presented in Table 4, it is known
that milk powder stored in bulk might self-ignite when exposed to
a higher temperature over a longer period. Tests have shown that
storage at 80e90 C for a period exceeding 20 h results in selfignition (Le Maillard reaction).
3.1.1. Hazards identication
Under normal operating conditions the average dust concentration in the dryer is below the lower explosion concentration. Locally
in the cone, however, one can expect that ammable concentrations
can be reached intermittently. It has been demonstrated that an initial
local explosion can whirl up dust present on the cone walls causing
Table 5
Summarizing the probabilities and consequences of primary and secondary events in the spray dryer and the associated risks for personnel and equipment.
Process
unit
Spray
dryer
Probability
of
flammable
atmosphere
V
Probability of ignition
Equipment
(electric and
mechanical)
Hot
surfaces
Electric and
electrostatic
sparks and
discharges
Mechanical
sparks
Flame and
smoldering
combustion
III
Probability
of
explosion
III
EXPOSURE TO EXPLOSION
PRIMARY EXPLOSION
Probability (injury/damage)
Consequence
Risk
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
II
III
IV
III
SECONDARY INCIDENTS (inclusive explosions)
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
II
II
IV
Comments:
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
843
Table 6
Summarizing the probabilities and consequences of primary and secondary events in the spray dryer and the associated risks for personnel and equipment after implementation of a CO-detection system.
Process
unit
Spray
dryer
Probability
of
flammable
atmosphere
V
Probability of ignition
Equipment
(electric and
mechanical)
Hot
surfaces
Electric and
electrostatic
sparks and
discharges
Mechanical
sparks
Flame and
smoldering
combustion
II
Probability
of
explosion
II
EXPOSURE TO EXPLOSION
PRIMARY EXPLOSION
Probability (injury/damage)
Consequence
Risk
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
II
II
IV
III
SECONDARY INCIDENTS (inclusive explosions)
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
IV
Comments: A CO-detection system has been included.
a stronger secondary explosion (Siwek et al., 2004). While all
potential ignition sources were evaluated, mechanical sparks were
not considered possible due to the ignition properties of the milk
powder. Self-heating of the layers of milk powder was determined to
be a possible ignition source. Layer accumulation is more frequent
during anomalies associated with the rotating spraying wheel,
distributing the milk slurry against the walls of the cylindrical part of
the dryer. The hot drying air could cause the resulting milk powder
cake to self-ignite. The smoldering material could then come loose
and fall into the cone of the dryer, causing either ignition of a ammable dust cloud, or cause the ammable dust cloud by whirling up
dust that subsequently ignites.
The probability of the latter is relatively high and, based on
historical evidence, an explosion should be expected with a frequency
between 101 and 102 per year (probability class III). Here it is
assumed that the ignition source also causes the dust cloud (a
smoldering cake of milk powder falling into the cone of the dryer).
Another ignition source could be an explosion originating in
other parts of the drying installation and running back into the
dryer. This ignition source, although very realistic, is not considered
here because events originating in other pieces of equipment
would only be included in a full risk analysis of the spray dryer
installation. For the present example, it is assumed that sufcient
preventive and protective measures are taken to prevent this from
happening, i.e., the likelihood of this ignition source occurring is
assumed to be sufciently low.
The consequence of the explosion is most likely the failure of the
dryer, potentially injuring personnel or even causing fatalities if in the
vicinity of the dryer at that very moment (consequence classes III and
IV respectively). Explosion tests reported by Siwek et al. (2004) show
that pressures up to 1 bar are possible, under somewhat conservative
conditions. Moreover there is a possibility that the explosion propagates into the uid bed or the cyclones, and into the bag lter
(secondary incident). This probability is however lower than the
probability of an explosion (probability class II). The consequences are
however more severe: loss of the plant (consequence class IV) and
most likely loss of one or more lives (consequence class V).
The analysis is summarized in Table 5. The table also determines the
risk based on the various probabilities and associated consequences.
3.1.2. Risk evaluation
The results of the analysis of the spray dryer are summarized in
Table 5. The table shows that the risks are either medium (implying
that risk reducing measures should be implemented) or high
Fig. 8. Pressure in dryer with pressure relief panel (black) and without pressure relief
panel (red). Note relief panel opens at approximately 0.25 s. (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of this
article.)
844
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Fig. 9. Flame extent (left) and pressure [in bar] & velocity vectors (right) just before panel opens.
(implying risk reducing measures must be implemented). Hence two
alternatives are investigated: one where a single preventive measure
is introduced, reducing the probability of explosions, and a second
where this preventive measure is combined with protective measures.
3.2. New analysis investigating the introduction of preventive
measures
To reduce the probability of explosions from occurring, it is
proposed to introduce a carbon monoxide-detection system.
Smoldering results in the generation of carbon monoxide (CO) due
to incomplete combustion. A CO-detection system could warn the
operator of ongoing smoldering before a hazardous situation arises
(Steenbergen, Van Houwelingen, & Straatsma, 2007). A new analysis has been performed of the explosion risks of the spray dryer
including this preventive measure.
3.2.1. Hazard identication
The introduction of a CO-detection system will reduce the
probability of an explosion. An early detection of smoldering
combustion is assumed to reduce the probability of explosions by at
least a factor of 10 implying a class II probability of explosion. The
probability of equipment being damaged and personnel being
affected will be reduced accordingly for both primary and
secondary incidents. The consequences are however still similar.
This results in risks as summarized in Table 6.
3.2.2. Risk evaluation
Table 6 shows that risks have been reduced by introducing a COdetection system as compared to Table 5, presenting the original
risks without any preventive or protective measures.
The remaining risks for personnel, which are described as
medium according to the acceptance criteria proposed in Table 2,
should be addressed by introducing further risk reducing measures.
As described in Section 3.1, an additional analysis is presented
where the preventive measure of CO-detection is combined with
protective measures. A combination of explosion venting and
explosion isolation by extinguishing barriers between the dryer
and uidized bed, and the dryer and the cyclones is investigated.
3.3. New analysis investigating the introduction of preventive
measures in combination with protective measures
Reducing the probability of an explosion by introducing COdetection alone still leaves personnel exposed to a medium risk.
Hence additional protective measures are proposed. The effects of
introducing a combination of explosion venting and explosion
isolation (extinguishing barriers) have been investigated.
The probability of explosions, assuming an early detection of
smoldering combustion is implemented as described in Section 3.2,
is considered to be probability class II. The consequences of possible
explosions are, however, reduced considerably. Assuming use of
appropriate venting devices, sufcient venting surface, taking into
account the effect of vent ducts (which are necessary since the
spray dryer is installed inside a building) and adequate installation
distances for the extinguishing barriers (containing sufcient
extinguishing powder to extinguish ames), the risk of explosion in
the spray dryer can be reduced considerably. For example,
advanced CFD explosion models (to be discussed in detail in Section
4) can provide assistance in quantifying the reduction in explosion
consequences for complex dust handling equipment.
Using the Dust Explosion Simulation Code (DESC) CFD software,
an analysis was conducted to optimize the size and possible
conguration of the vent on the spray dryer. The present example
shows the spray dryer both with and without a pressure relief panel
located on top of the dryer with an opening pressure of 50 mbar. The
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
845
Fig. 10. Flame extent (left) and pressure [in bar] & velocity vectors (right) after panel opens.
dust cloud was assumed to only ll the lower section of the cone and
the explosion properties used in the simulation were based on the
literature values of a milk powder with a Kst value of 100 bar-m/s. In
this example, one simulation was conducted without the pressure
relief panel and demonstrated the explosion pressure within the
dryer would be slightly in excess of 1 bar, while the second simulation showed a signicant reduction in overpressure with the top
panel opening at 50 mbar (see Fig. 8). This is due to the venting of
the gases within the dryer as the dust cloud continues to burn (see
Fig. 9 and Fig. 10), thus relieving the overpressure within the dryer.
Table 7
Summarizing the probabilities and consequences of primary and secondary events in the spray dryer and the associated risks for personnel and equipment after implementation of a CO-detection system in combination with explosion venting and explosion isolation toward uidized bed and cyclones.
Process
unit
Spray
dryer
Probability
of
flammable
atmosphere
V
Probability of ignition
Equipment
(electric and
mechanical)
Hot
surfaces
Electric and
electrostatic
sparks and
discharges
Mechanical
sparks
Flame and
smoldering
combustion
II
Probability
of
explosion
II
EXPOSURE TO EXPLOSION
PRIMARY EXPLOSION
Probability (injury/damage)
Consequence
Risk
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
II
II
SECONDARY INCIDENTS (inclusive explosions)
Personnel
Equipment
Personnel
Equipment
Personnel
Equipment
Comments: A CO-detection system has been included combined with explosion venting
and isolation.
846
S.G. Davis et al. / Journal of Loss Prevention in the Process Industries 24 (2011) 837e846
Fig. 11. Example of simulations performed with DESC investigating the effect of the shape of a vent duct on explosion loads generated in a vented dryer due to dust explosions
originating inside.
The consequences of an explosion are now reduced to limited or
no damage both for the primary and secondary events (consequence class I).
3.3.1. Risk evaluation
Introducing explosion protective measures as described reduces
the risks both for the equipment and personnel to acceptable levels.
The reduction of consequences to consequence class I (replacement
of vent panels and relling of extinguishing barriers [neglecting the
costs of loss of some produced milk powder]) results in risk levels E
implying that no further measures would be necessary. Results of
the analysis have been presented in Table 7.
4. Dust Explosion Simulation Code (DESC)
DESC is a CFD-tool that can simulate the course of industrial
dust explosions in complex geometries. DESC was developed by
GexCon as part of a project supported by the European Commission. Apart from GexCon, there were 10 main participants in the
project with additional contributions from three other participants.
DESC can be a valuable tool for engineers designing powderhandling plants, especially when optimizing mitigation measures
such as venting devices, suppression systems or explosion barriers.
For the example provided in Section 3.3, various explosion vent
congurations can be evaluated in a manner that takes into account
equipment details such as duct lengths, as well as a given dust
reactivity. Fig. 11 shows an example of how explosion vent designs,
including duct conguration, were evaluated for an indoor piece of
equipment.
In the approach adopted, dust explosion properties from standardized tests are used as input to the combustion model. Papers
have been published on the DESC project (Skjold, Arntzen, Hansen,
Storvik, & Eckhoff, 2006; Skjold et al., 2005; Skjold & Hansen, 2005)
and a review of the DESC project has been published in Journal of
Loss Prevention in the Process Industries (Skjold, 2007). The DESC
simulations include:
Realistic representation of industrial process plants
Most kinds of explosive dusts used in industry
Input from standardized tests in a 20-L sphere
Flame propagation and rate of pressure rise
Effect of varying the ignition position
Prediction of external blast waves
Dust lifting by ow or shock waves
Identifying worst-case explosion scenarios
Pressure-piling in interconnected vessel systems
Extensive options for output of results, including 2D and 3Dplots
Fast acting valves triggered by sensor points
Vent panels triggered by internal pressure
5. Conclusions
A semi-quantitative short-cut risk analysis method (SCRAM) has
been presented, allowing for the assessment of dust explosion risks
and choosing adequate preventive and protective measures. The
performance of such an analysis makes industry aware of the most
hazardous areas in their facilities and associated consequences in
case of an explosion.
The application example demonstrates the strength of the
method and the support it offers to industry for choosing appropriate
risk mitigating measures. In addition, an advanced tool (DESC) is
discussed, which can greatly help engineers designing powderhandling plants, especially when optimizing mitigation measures
such as venting devices, suppression systems or explosion barriers.
References
Beck, H., Glinke, N., & Mohlman, C. (1997). BIA-report: combustion and explosion
characteristics of dust, HVBG. Berufsgenossenschaftliches Institut fr Arbeitssicherheit. BIA 13/97.
Siwek, R., van Wingerden, K., Hansen, O. R., Sutter, G., Schwartzbach, Chr., Ginger, G.,
et al. (May 31eJune 3, 2004). Dust explosion venting and suppression of
conventional spray driers. In Eleventh international symposium on loss prevention, Prague.
Skjold, T. (2007). Review of the DESC project. Journal of Loss Prevention in the Process
Industries, 20, 291e302.
Skjold, T., Arntzen, B. J., Hansen, O. J., Storvik, I. E., & Eckhoff, R. K. (2006). Simulation
of dust explosions in complex geometries with experimental input from standardized tests. Journal of Loss Prevention in the Process Industries, 19, 210e217.
Skjold, T., Arntzen, B. J., Hansen, O. R., Taraldset, O. J., Storvik, I. E., & Eckhoff, R. K.
(2005). Simulating dust explosions with the rst version of DESC. Process Safety
and Environmental Protection, 83(B2), 151e160.
Skjold, T., & Hansen, O. R. (2005). The development of DESC e a dust explosion
simulation code. In International ESMG symposium, 11e13 October 2005 (pp. 24),
Nuremberg, Germany.
Steenbergen, A. E., Van Houwelingen, G., & Straatsma, J. (2007). System for early
detection of re in a spray drier. International Journal of Dairy Technology, 44(no.
3), 76e79.
