The document discusses a fire that occurred at a Port Arthur Light Olefins Unit (LOU) in 2006. The fire was caused by a sudden rupture of a corroded gas warm-up line due to severe corrosion under insulation from intermittent dripping of moisture. This released large amounts of flammable gas and fueled the fire. The investigation found 18 additional piping ruptures from extreme heat. Key learnings included improving inspections for corrosion under insulation, especially on top of pipes, and quickly fixing moisture leaks to prevent accelerated corrosion.
Seismic Hazard Assessment Software in Python by Prof. Dr. Costas Sachpazis
108b Presentation - Killing Rattlesnakes Before They Bite You.ppt
1. Killing Rattlesnakes Before They
Bite you
Review of the Port Arthur Light Olefins Unit Fire
29 April, 2006
Presented by J. Prows at AIChE Spring Conference
29 April, 2009
2. 2
Differentiated Chemicals
Differentiated Chemicals
Who is Huntsman?
Mainly differentiated businesses
Materials and Effects
Advanced
Materials
Design &
Composites
Engineering
Power &
Electronics
Coatings,
Construction &
Adhesives
Polyurethanes
Adhesives
Coatings
& Elastomers
Appliances
Automotive
Construction
Composite
Wood Products
Footwear
Furniture
TPU
Pigments
Titanium Dioxide
Performance
Products
Performance
Specialties
Performance
Intermediates
Maleic Anhydride
& Licensing
Textile
Effects
Apparel
Home textiles
Technical textiles
Base Chemicals and
Polymers
Ethylene
Propylene
PO/MTBE
Cyclohexane
Paraxylene
Polyethylene
Polypropylene
APAO
EPS
Commodity Chemicals
2005 Revenue: $6,164m 2005 Revenue: $8,638m
SOLD
3. 3
You Can’t Fix
What You Don’t Know About
Do you know what lies
underneath your insulation?
Is the pipe, exchanger or
vessel under your insulation in
good shape or is it is corroded
and/or stress cracked?
Do your piping systems
and equipment have
defects (“rattlesnakes”)
that you can’t see that
are ready to bite you at
any time?
11. 11
Port Arthur Light Olefins Unit
History
LOU produces 1.4 billion pounds of
ethylene per year
1978 - Originally designed by Stone and
Webster to produce 1.0 billion
pounds of ethylene per year
1979 - Started production as part of the
Texaco Chemical Company.
1987 - Lummus modernization
completely reconfigured the LOU
cold train
1994- Huntsman purchased the LOU
2006 - LOU Fire in cold train
2007 - LOU sold to FHR
12. 12
Port Arthur Light Olefins Unit Fire
April 29, 2006
7:27 am - Four Huntsman
technicians were outside
troubleshooting a Port Arthur
LOU process upset
They heard a loud popping
noise and saw a large fire ball
rapidly coming toward them.
They were able to outrun
the fire and find shelter!
No fire related injuries/deaths
One post event recordable
injury (emergency responder
suffered heat exhaustion)
13. 13
• Neighborhood in
vicinity of the plant
did voluntary
shelter in place
• No offsite injuries
or fatalities
• $300+ million in
plant damage
• Plant down 14
months for
demolition and
repair
17. 17
The Apollo “Why Drill”
Apollo “Why Drill” Reality
Charting tool used to categorize
and drill into potential causes
Interviewed involved individuals
Access to scene limited at first
Digital photography used
extensively (850 pictures) to
remotely assess scene & to
provide facts for Why Drill
process
17 vessel samples taken
Pipe and bolt fragments
collected
Documented 197 Why Drill
cause/effect blocks
18. 18
• The Incident Investigation Team received immediate and
repeat external visits from the:
• FBI
• EPA
• TCEQ
• OSHA
• Chemical Safety Board
• Insurance Companies
• The CSB came and stayed the first week and ran a
parallel investigation process.
• The CSB complimented the investigation team for the
thoroughness, openness and speed of the investigation.
• A 158 page investigation report was submitted to the
CSB and the above agencies in September 2006.
We Got Lots of Visits and Help
20. 20
Line Fracture / Rupture Why Drill
Line Fracture/Rupture Why Drill (simplified)
Showing Corrosion, Overpressure & Cyclical Pressure Legs
21. 21
• Warm Up Line Guillotine rupture
• Hot Tensile Ruptures of 18 Process Lines
System Pounds of Flammable Gas
Deethanizer 200,000
Ethylene Tower 700,000
Propylene Refrigeration 550,000
Ethylene Refrigeration 80,000
• Compromised Flare System
The flare line sagged and multiple high temperature
ruptures developed in the line feeding the flare
system allowing approximately 980,000 pounds of
flammable gas from inside and outside the damaged
area to fuel the fire.
Plenty of Fuel Released to Really
Warm Things Up!
26. 26
Warm Up Line “Guillotine” Failure
All other Piping Failures Heat Related
A sudden “guillotine” rupture of
the cracked gas warm-up line
ripped a 15 inch long section of
piping from the middle of a 200
foot long piping run.
The 27-year old 8 inch diameter
carbon steel warm up line
ruptured because the top of the
pipe was severely corroded in a
2 ft long area.
It no longer had sufficient wall
thickness to contain even
normal operating pressure.
The pipe was under 674 psi
operating pressure at the time it
failed.
27. 27
Warm Up Line Guillotine Rupture
Both Ends of Lines shown
The
Blow
Torch
29. 29
The Blow Torch
Cracked gas
jetting from a
wide open 8 inch
diameter line at
1.8 million
pounds per hour
initial rate
10F250
18 heat related
piping ruptures (fish
mouth ruptures)
30. 30
10F-250 Propylene
second stage flash
drum - contained
20,000+ gallons of
liquid propylene
Heat input from “Blow
Torch” boiled the
liquid in this vessel
and back fed the fire
through ruptured lines
Pressure in lines and
vessels increased
from massive heat flux
Piping and vesseel
metal weakened and
failed due to extreme
heat
10F250
31. 31
Warm Up Line Failure Analysis
From Dr. D. Scott Harding (Exponent – Failure Analysis Associates):
0.322” nominal wall thickness when new (8” schedule 40 pipe –
API 5L-Grade B reference)
Original 8” pipe (60 ksi tensile, and 35 ksi yield)
Burst pressure 4840 psi
Yield pressure 2824 psi
Back calculating thickness to burst this pipe grade at 674 psi
(relief valve allowed overpressure)
Burst thickness 0.045”
Yield thickness 0.077”
Actual wall thickness in heavy corrosion area on 15” long fragment
from failure area was 0.018 min. to 0.313 max.
Failure mode – Tensile overload in the hoop stress direction.
Overload resulted from a combination of external wall thinning (from
corrosion) and cyclical internal pressure.
32. 32
Failed 8” Warm Up Line
Actual Thickness Measurements
0.045 in. 0.044 in.
0.019 in.
0.025 in.
0.034 in.
0.064 in.
33. 33
Warm Up Line Corrosion – Why?
Corrosion was on the TOP of the warm up line. Why?
Strange place for it to occur.
There was a mechanical integrity program in place for this
line.
1987 Installation data 0.332 wall thickness
1999 First thickness inspection 0.344 wall at Elbow
2004 visual inspection (insulation damage noted)
2009 next scheduled inspection (class 2 piping
system, 10 year thickness, 5 year visual required)
Above the failed 8 inch diameter warm up line were two
non-insulated stainless steel lines that operate from 55 to
80 degrees F. These “dripper” lines provided a means to
routinely and intermittently wet the failed warm up line
section with moisture condensed from the air.
34. 34
Warm Up Line Top of Pipe Corrosion
Caused by “The Drippers”
35. 35
Warm-up Line Guillotine Failure Cause
(CUI Caused by Long Term Overhead Water Drip)
Corrosion on the warm-up line was in a localized two-foot
long section of a two-hundred foot long horizontal run
The rest of the horizontal run was in good condition.
Heavy corrosion in this two-foot area was caused by
condensed water dripping intermittently from a pair of
cold service non-insulated stainless steel lines crossing at
a 90° angle just a few feet above this section of the warm-
up line. These “drippers” were added in 1987 expansion.
The most severe corrosion was on top of the ruptured
warm-up line and is most likely to have been caused by
corrosion under insulation.
Insulation was completely removed from the warm-up line
and most other lines in the fire zone by the fire and high
pressure fire water.
36. 36
Two Team Behaviors
Reduced Serious Injury Potential
LOU Operations Team strictly followed procedures for
denying work permits and access during unit upsets.
Huntsman Emergency Response and Sabine Area
Mutual Aid teams quickly responded to the fire. They:
Effectively contained the large fire to the initial area of
impact and limited softening and collapse of major
structures and vessels.
Allowed non-isolatable fuel sources in the system to
burn and effectively prevented conditions for a vapor
cloud explosion and/or a Boiling Liquid Expanding
Vapor Explosion (“BLEVE”).
37. 37
LOU Fire Why Drill
Cause/Effect Summary
LOU in Upset Condition
Higher than normal temperatures at exit of crack gas
compressor due to hurricane impact on heat
exchanger performance that cools the crack gas
Higher temperature & moisture levels into crack gas
driers. Driers overwhelmed. Moisture breakthrough.
Plugging of cold train with ice/hydrate
High crack gas system pressure
Cyclical pressure (compressor surging)
Mechanical Integrity Failures
Crack gas warm up line suddenly failed (jet fire
impinged dense piping area) due to severe CUI
250 psi steam piping high temperature rupture
Hydrocarbon line high temperature failures (18 lines)
38. 38
The Timing of a Spark
BP Texas City had a large
flammable gas release that
resulted in an explosion,
15 fatalities and many
injuries.
The Port Arthur LOU had a
large flammable gas
release the resulted in a
large fire and NO injuries.
WHY???
The LOU event found an
early spark or heat source,
so a large fire resulted, not
an explosion.
39. 39
What Do We Control?
We DO NOT control the timing of a spark
We DO control:
Process Upsets and their duration
Keeping the chemicals, flammables and water in the
pipes
Choosing not to live with defects
ZERO tolerance for living with “rattlesnakes” (At Risk
Behaviors and Defects)
Learning from unwanted events and sharing the
learning with others
40. 40
No Corrosion on the Bottom & Sides of the Pipe
(Normally Where We Inspect the Most)
The wet calcium
silicate insulation
on a propylene
refrigerant line
sagged away from
the bottom of the
pipe from the
additional weight.
The direct contact
between the top of
the pipe and the
wet insulation
resulted in
CORROSION ON
TOP OF THE PIPE
41. 41
Heavy Corrosion on the TOP
of the Pipe (CUI)
Propylene
refrigerant
line from 10E-
1401 was
found to have
corrosion
under
insulation
(CUI) along
the TOP of
the pipe.
42. 42
MORE CUI on the TOP of the Pipe
(Not Where We Typically Test for Thickness)
43. 43
Insulation Jacketing must be
Inspected for Defects
CUI experienced
on the TOP of the
propylene
refrigerant line
from 10E-1401
likely resulted
from perforations
in the aluminum
insulation jacket
that allowed
water to saturate
the underlying
calcium silicate
insulation.
The photographs
display different
areas on the
same line.
44. 44
Note: The LOU Team were
inspecting 19,000 piping &
21,000 equipment points. Yet
they did not see the snake!
45. 45
CUI Learning from LOU Fire Applied to
PNPP A3 Olefins Unit
Found
corrosion under
insulation (CUI)
on A3 Olefins
Unit propylene
piping.
Hole in pipe
plugged with
ice!
UNIT MUST BE
DOWN AND
WARM TO SEE.
46. 46
A3 Olefins Unit CUI
CUI on A3
Olefins
Unit.
Deep pits
ON TOP of
propylene
line.
48. 48
Key Learning – Mechanical Integrity
The API 570 MI inspection procedure covers pipe that corrodes at a
uniform rate. It covers CUI as well as dead end lines
API 570 is less prescriptive regarding other damage mechanism like
pitting and other accelerated corrosion mechanisms
The warm up line that failed was not:
Painted or coated
Initially insulated in 1979 but insulation was added later.
Recognized as a dead end line because the P&ID showed a
control valve on the demethanizer end of the line. The LOU MI
team was unaware the line is used once in 10 years.
Current Huntsman piping specs require pipe to be cleaned and
painted prior to installing insulation.
Lloyds Register Capstone’s RBMI process and database have been
selected as a Best Practice Science for A3 Olefins and the PNPP
site.
49. 49
Key Learning - Look for Drippers!
Dripping Water is a Rattlesnake
The condensed moisture dripped down onto the warm-up
line insulation cladding or piping below. Constant
exposure to moisture can accelerate the corrosion of both
the pipe where the condensation occurs and the pipe
onto which the water drips.
Past practice was to repair steam and condensate leaks
as maintenance “fill-in jobs.” This can lead to delays in
getting the water leaks fixed when they are small. Water
leaks are snakes.
Take a more aggressive approach to repairing steam and
condensate (and other leaks) to reduce potential damage
to adjacent piping.
50. 50
Recommendations
Water on your piping and equipment is not your friend.
Near constant dripping from non-insulated cold pipe
Water being held against pipe by insulation
Calcium silicate Insulation absorbs water and holds it
against the pipe.
Check straight sections of pipe that operate in the 25° F
to 300° F temperature range, particularly if the pipe is
insulated with a material that will absorb water (i.e.,
calcium silicate).
Look for water leaks, condensate leaks, cladding
damage, insulation damage, bulging and other evidence
of integrity failure should be an ongoing job of all
employees
51. 51
Recommendations – CUI & MI
Consider improvements to your Mechanical Integrity
piping inspection program to better identify
Corrosion Under Insulation, especially for equipment
and piping that operates in the temperature range of
25° F to 300° F.
Inspect the TOP OF YOUR PIPES for thickness and
corrosion, and
Consider using higher quality insulation that does not
wick water and hold it against the pipe surface.
Look for water leaks, condensate leaks, cladding
damage, insulation damage, bulging and other evidence
of integrity failure
When in doubt – STRIP THE INSULATION AND LOOK!
52. 52
Cold Train Plugging Prevention
Dryer Moisture Breakthrough
Low levels of moisture breakthrough from the dryers can
form ice deep in the Cold Train even at a 10 ppm water
level.
Dew point is a better measure of moisture breakthrough
from the dryers vs. ppm moisture level because it will let
the operations team know where moisture is likely to
liquefy and turn to ice/hydrate in the Cold Train system.
53. 53
PPM levels of Water to the Cold Train
will go Deeper Than You Think
Temp. Dew Point Moisture
Temperature (°F) Dew Point Moisture (ppm)
-120 0.8
-100 1.6
-80 2.9
-60 5.6
-50 7.7
-40 10.8
-30 15.0
-20 21.0
-10 29.5
0 41.6
54. 54
Recommendations – Hydrate / Ice Breaking
Methanol Injection (and other anti-freeze solvents) to
Break Hydrate/Ice
As the Cold Train plugging-related events unfolded,
there were attempts to remove the plugging through
methanol injection.
Ice/hydrate formed over a longer period of time will be
frozen more solid and harder to break with methanol.
Methanol injection may not remove hard ice since the
vapor stream carrying the methanol will take the path
of least resistance through the exchanger.
If methanol does not work - shut it down
and warm it up!
56. 56
Project:
What’s the highest grade you can get in EHS performance, Rattlesnake safety?
ZERO
Zero Process Safety Incidents
Zero Injuries
Zero Loss of Containment
Zero Defects
TM
57. 57
PNPP Project ZERO
Wireless WiMAX Subscriber, IONizer and Handhelds
• SOC Database Engine
• Rounds & Checklists
• Defect capture via RFID
• Smart Procedures
59. Killing Rattlesnakes Before They
Bite you
Review of Port Arthur Light Olefins Unit Fire
29 April, 2006
Presented by J. B. Prows at AIChE Spring Conference
29 April, 2009