sugar dust explosion protection techniques and technologysugar dust explosion protection techniques...
TRANSCRIPT
Julian Turner
Sugar Dust Explosion Protection Techniques And
Technology
5th April 2018
British Society of Sugar Technologists
European legislation
Explosion protection is controlled by the ATEX Directive.
Employers must protect their employees from the harmful
effects of an explosion when processing combustible
powders.
Suppliers have to demonstrate to an independent Notified
Body that their explosion suppression systems will deliver the
performance that are claimed. (Pred) This is achieved by
generating calibrated explosions in 10 Bar vessels and
measuring the results.
Explosion isolation must be proven too using real explosions.
Suppliers must demonstrate that their equipment is not an
ignition source in our customers’ processes.
Dust Explosions
• Any combustible material will burn with a speed that
increases with decreasing particle size
• For a dust explosion to propagate we require:
– Oxidising medium; usually oxygen
– Ignition source
– Adequate fuel concentration
ExplosionFast CombustionSlow Combustion
Explosion venting
▪ Standard
▪ Flameless
Explosion Suppression
▪ Unique dynamic detection (MEX)
▪ Range of suppressor sizes and features
Explosion isolation
▪ Active and passive
▪ Optical detection
Products / Systems
Confined Dust Explosions
• Maximum pressure (Pmax) dependent on fuel and initial pressure
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Time / s
Pre
ssure
/ b
ar(
g)
20 m3
6 m3
2 m3
Vessel Volume
STKVdt
dP 3/1
dP
dt
Example explosion values:
Kst values
Coal 75
Flour 100
Starch 150
Sugar 180
Pharmaceutical 240
Washing powder 280
Aluminium flake 555
Kg values
Methane 55
Propane 100
Hydrogen 550
Explosion Protection Options
Vessel Protection
Containment
•10 bar strong vessels
•Very expensive
•Not practical for large vessels
Explosion Protection Options
Vessel Protection
Containment Venting
• Very common
• Least expensive option
• Very common due to low cost and simplicity
• Weak panel on vessel – yields ~0.1bar
• Pressure relief only – does not extinguish flame
• Post-explosion fires likely
• Area of exclusion required - including flameless vents
• Flame ejection from vent – 8 x vessel volume
Explosion Venting
10
60m3 Vented Explosion
Explosion Protection Options
Vessel Protection
Containment
Pmax = Pred > 8bar
Suppression
Relieves pressure
Extinguished flame
Contains explosion
Venting
Relieves pressure
Inexpensive
Does not extinguish flame
Fireball 8x vessel volume
Explosion Suppression in detail
Ignition:
Time = 0 ms
Pressure = 0 bar(g)
Pressure wave ~300m/s
Flame front ~ 10m/s]
Detection:
Time = 70 ms
Pressure = 0.05 bar(g)
Suppressors Actuate:
Time = 80 ms
Pressure = 0.08 bar(g)
Suppression Complete:
Time = 100 ms
Pressure = 0.25 bar(g)
V0=10m3, Kst= 150 bar.m/s
Pressure Time Characteristics
Pa = Activation Pressure of Pressure Sensor
Ps –Plant Strength
Explosion Pressure [bar]
Maximum Explosion Pressure Pmax
Normal Explosion Development
Reduced Explosion Pressure Pred
Deployment of explosion suppressant
Pa
taTime [milliseconds]
1. Suppressant is a fine sodium bicarbonate powder
2. Suppressant has a large surface area per mass delivered.
3. Suppressant achieves extinguishing of the fireball by extracting heat . We aim to reduce the temperature of the fireball below the Auto Ignition Temperature of the product.
4. Suppressant isolates unburnt product from burning product.
How does suppressant work?
Radiation
Two unburnt coal particles
Coal particle
burns at 1,800
deg C
Unburnt particle
will self combust at
500 deg C (coal)Two burning particles – flame
propagation continues
Suppressant
engulfs
burning
particles
Suppressant forms a
barrier between
burning and unburnt
particles.
All remaining
particles
cooled to <500
deg C
The explosion is
extinguished
Validation by Experimental Data
100 150 200 250 300 350 400 450 500
0.5 m³ Vessel
50 mbar
100 mbar
150 mbar
100 150 200 250 300 350 400 450 500
Vessel Volume 0.5m3
100 150 200 250 300 350 400
100 150 200 250 300 350 400 450 500 100 150 200 250 300 350 400 450 500
Vessel Volume 25m3
Vessel Volume 250m3Vessel Volume 10m3
100 150 200 250 300 350 400 450 500
0.5 m³ Vessel
50 mbar
100 mbar
150 mbar
100 150 200 250 300 350 400 450 500
0.5 m³ Vessel
50 mbar
100 mbar
150 mbar
100 150 200 250 300 350 400 450 500
0.5 m³ Vessel
50 mbar
100 mbar
150 mbar
Kmax – bar.m/s
Kmax – bar.m/sKmax – bar.m/s
Kmax – bar.m/s
Tim
e (d
t)Ti
me
(dt)
Tim
e (d
t)Ti
me
(dt)
17
Suppression system design is critical to prevent
failure.
Principle of Explosion Detection
t
p t0
t0
p = pt0
p agw
t 0t
p
exp
losio
n p
ressu
re p
[b
ar]
time t [ms]
a
- t
- t
P static alarm pressure
p dynamic alarm pressureP =p pressure at the time of activation
P pressure at the time before activation
t dynamic alarm timet time of activation
t time before activation
agw
t0 a
t0-dt
0
t0-dt
Explo
sio
n P
ressure
[ba
r]
Time [milliseconds]
dt
dP
Introduction to dynamic explosion pressure
detection
Principles of dynamic explosion detection
• Pressure is monitored 1,000 times per second.
• The dynamic detector is programmed to respond
to a pressure rise of a known kst value in a vessel
of a given volume and aspect ratio (surface area).
• Including aspect ratio accounts for flame stretch
in an elongated vessel. This effects explosion
suppression and active explosion isolation
performance.
• dP is a value set by the designed. dT is calculated
by research based unique algorithm.
• Where process pressures rise and fall the
dynamic detector will assume a new reference
point.
• Typical set point is 50mBar in 80mSec.
Dynamic detection – the window of detection
dt
dP
Alarm
Ignore
Fault
Fault : Alarm : Ignore
Kreal
dt
dP
Alarm
Kmax
Slow explosion
“dPslow” - criteria
Pred Kreal< Pred Kmax
Ignore
Process pressure
fluctuation
Fault
e.g. Electrical noise
View in
presentation
mode
Dynamic detection – the window of detectionE
xp
losio
n P
ressu
re [b
ar]
Time [milliseconds]
dt
dP
dt
dP
Fault : Alarm : Ignore
dt
dP
Dynamic detection – the window of detectionE
xp
losio
n P
ressu
re [b
ar]
Time [milliseconds]
View in
presentation
mode
Dynamic Explosion DetectionP
roce
ss P
ressu
re [m
ba
r] 50mbar Static Pressure Detection Limit
Pneumatically conveyed batch processing
Dynamic pressure detector will not respond to this
overpressure because dt has not been met or exceeded.
Event Captured Pressure Time DataStrong explosion in bucket elevator HRD discharge only
EMC interface due to poor installationExplosion in large spray drier
All vessel protection options require explosion isolation to
minimise the risk of explosion propagation
Flame Transfer between connected plant leads to enhanced
explosion severity – Flame Jet Ignition
Explosion Isolation
Explosion Isolation much more complicated than Suppression!!
Introduction to Explosion Isolation
How to Isolate? Passive or active?
• Passive Isolation
– Explosion pressure actuates mechanical device;
• Active Isolation
– Mechanical – fast acting valves
– Chemical – Suppressant barriers
29
air flow ~16 m/s
DN500 duct - ~30 m long
vent panel
26 m3 vessel9.6 m3 vessel
DN500 air inlet
vent panel
DN500 to cyclone
V1V2
Active Explosion Isolation Tests at FSAFSA: Forschungsgesellschaft für angewandte Systemsicherheit und Arbeitsmedizin
FSA are a notified body for ATEX approvals
air flow ~16 m/s
DN500 duct - ~30 m long
vent panel
26 m3 vessel9.6 m3 vessel
DN500 air inlet
vent panel
DN500 to cyclone
V1V2
Explosion Isolation System
• System Design Premise – isolation device location
• Probability and consequence of flame propagation
dependent on primary vessel explosion protection used
– Vented and contained vessels ~95% chance of
flame propagation into a duct
– Suppressed vessels ~45% chance of flame
propagation into a duct
Based on actual research.
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
Why Isolate?
No Explosion Isolation: Pressure vs. time
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Time / s
Pre
ssure
/ b
ar(
g)
V1
V2
Flame Entry into Duct Flame Exit from Duct
Explosion Isolation: Pressure
vs. time
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Time / s
Pre
ssure
/ b
ar(
g)
V1
V2
Flame Entry into Duct
45
Explosion Isolation: Pressure vs. time
46
During this test programme, IEP Technologies performed 81 full-
scale interconnected vessel explosion tests.
In every case, flame entered the duct and propagated several
metres
77 out of 81 tests showed flame propagation over 21m
Thus for these geometrical configurations, dust concentration,
turbulence levels and ignition locations, flame transfer between
vessels is very likely.
In most cases, severe flame acceleration was observed leading to
enhanced explosion severity in the connected vessel.
Flame Propagation Probability
Schematic of an Explosion Isolation System
Detector, ta
Duct transit, tdDuct entry, te
Controlling Equation
ta + tb < te + td
d
tb
Active Explosion Isolation
0
50
100
150
200
0 50 100 150 200
experimental tff / ms
model t ff
/ m
s
perfect fit
1 m3 data
4.25 m3 data
9.4 m3 data
Maize Starch Explosions
?
Importance of Ignition Location
CFD simulation - 180ms from ignition as a function of ignition location
Importance of Ignition Location
Isolation Calculation Tool - SmartIS
SmartIS™ assumes worse case ignition locationPressure only assumes ignition close to duct mouth – long isolation
distances required
Pressure and Optical detection significantly reduces isolation distance
since ignition close to the duct mouth is detected very early by the optical
device
Isolation Calculation Tool - SmartIS
Passive Explosion Isolation Valves
Typical Passive Isolation valve▪ Uses the pressure of the explosion to
close the flap.
▪ Typical duct diameters of DN 160mm up
to DN 1000mm
▪ Typical values of Pred = 1.0 bar g. Must
therefore be used in combination with
explosion venting or explosion
suppression.
▪ Suitable only for low dust concentrations
▪ Can only be installed in the horizontal.
▪ Must be installed in line with the
manufacturer’s guidelines.
Thank You
Back-up Slides
Isolation
Full-Scale Testing
DN300 – 30m long
Vent Area=0.5m2
Vent Area=0.26m2
Fan
V1=9.6m3
V2=4.4m3
Air velocity ~16m/s
Air in
Air out
DN300 – 30m long
Vent Area=0.5m2
Vent Area=0.26m2
Fan
V1=9.6m3
V2=4.4m3
Air velocity ~16m/s
Air in
Air out
DN300 – 30m long
Vent Area=0.5m2
Vent Area=0.26m2
V1=9.6m3
V2=4.4m3
Air velocity ~16m/s
Control
Panel
DN300 – 30m long
Vent Area=0.5m2
Vent Area=0.26m2
V1=9.6m3
V2=4.4m3
Air velocity ~16m/s
Control
Panel
Explosion Isolation Trials at FSA
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
No Isolation Isolation 1 x HRD
65
No Isolation Isolation 1 x HRD
66
No Isolation Isolation 1 x HRD
67
No Isolation Isolation 1 x HRD
Optical Detectors
Optical detection dependent on observing radiation from combustion
zone – line of sight
Limited emission wavelength signatures – black body
Dust laden environments bring difficulties
Typically not used for detection within process vessels
Are useful for flame detection within connections – duct work.
Explosion isolation systems
1.3 Products / Systems
Understanding Application of Optical Detectors
Optical detection efficacy will be dependent on the radiation obscuration,
detector sensitivity and duct diameter.
Dust concentration: 100g/m3 of Maize gives ~50% obscuration for every ~10cm!
Dust colour: Black dust attenuates radiation ~2 times more than white dust
Dust median particle size: 35mm attenuates radiation ~3 times more than 70mm
Other factors – cleanliness of optics and detector sensitivity
ALL OF THE ABOVE DETERMINE DETECTION EFFICACY
1.3 Products / Systems
70
Advanced Calculation Tools: CFD
We have at our disposal advanced calculation tools and expert
knowledge of flame propagation
This allows us to generate more advanced in-house calculations tools
and more secure design guidance based on this data
71
During this test programme, IEP Technologies performed 81
full-scale interconnected vessel explosion tests.
In every case, flame entered the duct and propagated several
metres
77 out of 81 tests showed flame propagation over 21m
Thus for these geometrical configurations, dust concentration,
turbulence levels and ignition locations, flame transfer
between vessels is very likely.
In most cases, severe flame acceleration was observed leading
to enhanced explosion severity in the connected vessel.
Flame Propagation Probability