Tank Failures

Download as pdf or txt
Download as pdf or txt
You are on page 1of 65

1

1. Title slide
2
2. Quote from an excellent reference book in chemical process safety.
See reference 1 for full details.
3
3. Tanks come in all colors, shapes and sizes. This first slide is of the
Boston Gas Liquified Natural Gas (LNG) storage tank recently dismantled in Boston
after 20 years of a perfect record. Liquefied natural gas tanks are unique as this
discussion will point out. This tank has been dismantled because Boston Gas is
purchasing most of its natural gas through gas lines. This presentation will focus on
tanks holding more than 50,000 gallons. (Photograph courtesy of Boston Gas,
Boston, MA)
4
4. This is a water storage tank located in Portland Oregon. Several
features common to tanks are shown. Note the manhole for tank entry. Also note
the welded seams. Tanks are constructed from plates of carbon steel
approximately 4 X 8 to 8 X 32 feet welded together. The plates may or may not be
pre-formed to the tank diameter. Finally, note the hold down bolts along the bottom
to keep the tank in place in case of a seismic event.
5
5. This is a fixed roof, cone with column support, tank for fuel oil storage
located in the mid-western part of the U.S.A. The tank is insulated. This tank is 52
feet high and about 240 ft in diameter. It can hold nearly 21 million gallons of
material (500,000 barrels or 80,000 m
3
). Another tank common for the oil industry is
the floating roof tank. The floating roof can be either external or internal
(Photograph courtesy of Chicago Bridge and Iron Company, Oak Brook, Illinois).
6
6. This is a tank farm in which several of the tanks have external floating
roofs. Floating roofs greatly reduce loss by evaporation. For fixed roof tanks,
temperature cycles for a 24 hour period can lead to a large loss of material because
of the "breathing loss" (expansion of saturated air into the surroundings during
daylight and contraction in which fresh unsaturated air is added back into the tank
during evening hours). Another type of a significant material loss for fixed tanks is
"filling loss" in which saturated air is displaced upon filling (a similar everyday
example is the filling of a automobile gas tank). Floating roof tanks significantly
lower breathing and filling losses. If the true vapor pressure of the material to be
stored is between 1.5 to 11.1 psia (10 to 75 kPa) the tank should be equipped with a
floating roof (Ref. 3) (Photograph courtesy of Chicago Bridge and Iron Company,
Oak Brook, Illinois).
7
7. There are five major considerations for tank construction. The majority of
tank failures can be prevented via the use of these major concerns:
1. Adequate base - The foundation must be designed by a qualified engineer and
constructed in accordance with recognized structural practice. A sub-surface investigation
needs to be conducted by a qualified soils engineer. If ground can freeze, consideration
must be given to base depth. Fault lines must be checked. Special conderation has to
given if high winds can exist. Essentially the foundation must provide a stable plane with
very limited settlement after the tank is placed into operation. Adequate drainage around
the foundation is a must.
2. Floor or bottom - must be thick enough to be impermeable to fluid in tank. 1/4" is
standard.
3. Wall or shell - must be thick enough to withstand the hydraulic head. 7/8" thick steel at
the bottom panels is not uncommon, 3/16" at the top. Maximum wall thickness is 1 3/4"
thick and is determined by formulas like those found in API 650.
4. Material - must be suitable for the stored material with minimum corrosion. Carbon steel
(0.2 to 0.25% C in iron or ASTM A36/A36M steel) meets most requirements. Specialized
tanks such as LNG tanks require 9% Ni steel or aluminum as we will learn later. Rubber
lining, stainless steel, aluminum, and clad steel are other materials which can be used for
specialized cases.
5. Roof - Fixed or Floating - must be strong enough to hold 25 pounds per square foot
(typical U.S.A. design standard). In cold climates, the roof must be able to hold several
inches and even several feet of snow. The drainage system must be adequate for floating
roof tanks to handle the severest of rain falls so that the roof doesn't sink into the tank.
Floating roof tanks are generally designed to handle 10" of rainfall in a 24 hour period for a
roof floating above a liquid of sp. gr. of 0.70 (See API 650 Ref. 11).
7
8
8. This view shows a schematic of an external floating roof tank. Note
the articulated roof drain system, the rim seal and bleeder unit. The vacuum
breaker plays an especially important role. Although the roof can withstand some
load, only several inches water vacuum inside can cause a collapse. (Diagram
courtesy of CBI).
9
9. A tank under construction. In this photograph one can see the pre-
formed steel plates being set in place and welded together. A person is standing
near the tank at the bottom to provide a size reference (Photograph courtesy of
CBI).
10
10. Standards are set by a number of associations.
API 650 for welded storage tanks for petroleum storage.
API 620 for low pressure, low temperature, and cryogenic tanks (primarily for design
and construction).
NFPA 59A & 58 for LNG and LPG storage tanks (primarily for location, Q.C., and
testing)
AWWA 100-84 for welded steel tanks for water storage.
UL-142 for steel tanks holding less than 50,000 gallons.
11
11. Large storage tank failures are memorable events. One example is
the 1925 great molasses flood caused by the explosion of a storage tank for
molasses. A wave of molasses twenty feet high knocked down several buildings
and drowned twenty-one people. Clean up took years and some Boston residents
claim that they can still smell molasses on warm days.
In this teaching module we will concentrate on 3 more recent disasters which
demonstrate major storage vessel failures followed by a brief discussion on
common tank and operational failures. Many failures that have and will occur are
the result of poor or inadequate design of the auxiliary systems - relief valves,
venting, purging, foundation heating systems etc. A summary handout (Table 1 in
the notes, Slide 61 in this presenation) describes major tank types and common
failure modes. This presentation concentrates on three major disasters because of
the significance they have made in engineering design and development.
12
12. Definition of BLEVE. BLEVE stands for boiling liquid expanding
vapor explosion. The origin of the word BLEVE came from the Factory Mutual
Insurance Company in the mid 1950's. Originally, the word was used to describe a
pressure vessel explosion which occurred for a vessel holding formalin (37% by
wt% formaldehyde in water) above its normal boiling point. BLEVEs can occur with
any pressure vessel holding a liquid under pressure and at a temperature above its
normal boiling point. It is the rapid phase transition in which a liquid contained
above its atmospheric boiling point is rapidly depressurized, causing a nearly
instantaneous transition from liquid to vapor with a corresponding energy release.
During the 1960's several liquid propane storage vessels failed when the level
inside dropped below an outside flame impingement point. The vessel wall failed. A
large amount of liquid propane then vaporized (entrapping liquid particles as well)
creating a huge vapor cloud. The vapor cloud then exploded. The word BLEVE
was revived to describe these catastrophes. Early fire fighting techniques with
propane storage vessels cost several lives because of lack of knowledge about the
seriousness with fires around liquid propane storage vessels. Through an
educational film produced by NFPA, many fire fighters know that the word BLEVE
means precaution and evacuation first.
13
13. A liquified petroleum gas (LPG) pressure vessel failure. LPG can be
liquified propane, a propane/butane mixture, or liquified butane. In order to achieve
liquid storage at ambient temperature, the material is stored at high pressure.
Pressures can range from 30 psig for liquified butane to 150 psig for liquified
propane. The pressure vessels for LPG storage is often rated for pressures up to
250 psig (LPG can also be stored in low temperature tanks near ambient pressure).
By far BLEVEs are the worst storage vessel failures and the most
spectacular.
14
14. A BLEVE involves several steps.
Simplified steps are:
1. First, for LPG pressure vessels, a flame from a fire must be striking the outside of
the storage vessel. BLEVEs can also occur because of a weakness in a metal wall,
corrosion, or a dent from an impact with another object. For this example we will
concentration on a LPG storage vessel BLEVE.
2. Because energy is being added to the contents, vaporization will occur and the
material will leave the vessel via the relief valve. When the level inside the vessel
drops below the flame impingement point the metal of the vessel wall weakens.
Note that failures can occur even when the vessel pressure is below the design
pressure and relief valve set point.
3. A rupture occurs at this weakened point and a sudden expansion of vaporizing
material occurs. Some liquid is also entrained by the escaping gas in this rapid
decompression process. Exit temperature of the LPG is approximately -40oC, thus
condensing water vapor from the air. A large cloud of material forms around the
vessel.
Note that a BLEVE can occur with water held in a tank at a pressure and
temperature above its normal boiling point. Many serious injuries have resulted
because of BLEVES with water. Of course, when the liquid is a volatile flammable,
the consequences are dramatic.
4. If the vapor is flammable, then the release will ignite. One very large fire ball can
result.
15
15. A diagram slide to go with the description given in 15.
16
16. This slide is a synopsis of one of the worst BLEVE's in terms of lost of
life and property damage on record.
17
17. A 1979 photograph of the tank farm before the accident. Note how
close the storage vessels are. Two more 2,400 m3 spheres were added after this
photograph! The initial accident was a leak from a broken joint or faulty packing in
the pipe line/valve due to poor maintenance, or sabotage???. This release travelled
several hundred meters to an ignition source. Remember gaseous LPG has a
higher density than air thus it tends to move along at ground level. (Slide courtesy
Mr. Olle Johansson, Skandia Group, Stockholm, Sweden)
18
18. An aerial view that shows the point of initial release and ignition
source. The distance involved between release and ignition is 200 m. Ignition
sources may not be immediately apparent, however, they are always free and
available. Most accidents follow a three-step sequence. (Slide courtesy of Skandia
Group)
Initiation: the event that starts the accident (in this case a release from a faulty
valve?)
Propagation: the event or events that maintain or expand the accident (in this case
a release of propane which formed a vapor cloud which expanded over 200 m
before ignition, then ignition occurred, and then...)
Termination: the event or events that stop the accident or diminished it in size (in
this case consumption of combustible material and there was a large quantity!) (See
Ref 15. p 14. for more details about the accident process).
19
19. On the left is an aerial photograph of the PEMEX depot and buildings
in the vicinity before the mishap. (Scale is 1:20,000)
On the right is an aerial photograph of the PEMEX depot and buildings
in the vicinity after the mishap. Note the relocation of many cylindrical vessels.
Also, only two of the 6 original spheres remained. Nearby, all of the residential area
shown in the picture was affected directly by the shock from the explosion, by the
heat from fireballs, or by the rain of droplets of cooled LP-Gas which set people's
clothing and surrounding buildings on fire (portion of text was from the Skandia
Group case study). (Slide courtesy of the Skandia Group) Reference: "Bleve! The
Tragedy of San Juanico," First 1985, Skandia Group International Insurance
Company, Stockholm, Sweden.
.
20
20. Fire fills the sky. The magnitude of a BLEVE is overwhelming. Note
the size in comparison to the electrical transmission tower and the water tower.
(Slide courtesy of the Skandia Group).
21
21. Here we see (lower left horizontal tank) a safety device working
properly. - a relief valve. All that the firemen can do is keep the sphere and tanks
wet and let the contents burn out in a slow controlled manner. (courtesy of the
Skandia Group).
22
22. Two remaining spheres and disaster in the foreground. These
spheres did not explode probably because they were filled to 90% capacity and
consequently contained a large amount of cooled liquid. Also the flame fronts were
not as severe near these spheres. Nevertheless, note that the supports were
weakened. (Slide courtesy of the Skandia Group).
23
23. Believe it or not this cistern (21 m long and 3.5 m in diameter) set a
world's record. It flew across the sky a distance of 1200 m and up an elevation of
50 m destroying a home when it landed. It holds the distance record for 20 ton
bleveing cisterns. Fortunately, the family members of this household were
evacuated one hour earlier. (Slide courtesy of the Skandia Group)
24
24. A photograph which shows an overview of the diaster. 15 LPG
storage vessels bleved. 500 people were killed. This sums up the violent nature of
BLEVES. (Slide courtesy of the Skandia Group).
25
25. How to prevent BLEVES.
Bleveing pressure vessels take time. The minimum known time is approximately 30
minutes for fixed storage vessels and 10 minutes for a tank trucks. The fire chief in
charge must order an evacuation of all non- critical personnel as soon as possible.
26
26. LNG is primarily liquefied methane. LNG storage tanks are near
ambient pressure (1.5 to 2.0 psig). In order to achieve liquid storage, the material is
stored at low temperatures (-260oF).
27
27. The initial interesting fact about LNG tanks are that they are a tank
inside a tank. The outer tank is made out of carbon steel and is under a slight
pressure inside of 1 to 2 psig. Inside is another open roof tank which is constructed
of 9% Ni steel or aluminum. The two tank walls are separated by an annular space
of 3 to 4 feet. This space contains insulation, usually a powdered low density
material called Perlite. Perlite is from a volcanic material which when heated at
very high temperature expands to a highly porous-low density material. It is an
excellent insulating material.
28
28. This is a cutaway of a typical double-walled, flat bottomed tank and its
principal components. The inner tank is suitable for LNG storage at cryogenic
temperatures (-260
o
F or -160
o
C). The outer tank provides a proven means of
protection for the insulation and is vapor tight. Connections to the inner tank must
allow for differential movement because of temperature changes and hydrostatic
loads. (From A.G.A. LNG Information book Figure 3.2 p. 36).
29
29. The Cleveland LNG disaster was quite significant. This was the first
large scale commercial plant in the world. No further commercial LNG tanks were
constructed in the United States after this accident for a period of 20 years. The
inner tank itself was 42 feet high and 70 feet in diameter (1.2 MM gallon or 4,600
m3 capacity). The outer tank was 51 feet high and 76 feet in diameter. Rock wool
was used for insulation in the annular space. Approximately 120 people died and
1500 people were left homeless because of the resultant explosions which occurred
after the tank failed and spilled it contents into the nearby streets and sewers.
30
30. Composite of NY times headlines related to the accident. Note the
headlines: PANIC IS WIDESPREAD! Red Cross is involved. 1,500 homeless. FBI
reported to be studying the possibility of sabotage. Later the investigation ruled that
there was no evidence of sabotage.
31
31. Aerial view of LS & R Plant and No. 2 works, East Ohio Gas Co., and
adjoining property damaged by fire October 20,1944, looking north. Note the upper
right quadrant. Nearly all of the neighboring factory complex was destroyed. In the
lower quadrant one sees that the blocks of 3 to 4 story bricked store fronts and
overhead apartments were completely wiped out. That street is East 61st Street.
Near the middle of this photograph running diagonally towards the bottom right
quadrant is East 62d street. LNG probably flowed into the storm sewer underneath
this street. Eventually the mixture exploded, cracking up the street for 3 blocks from
the original tank whose location is marked on the photograph. (Photograph reprinted
from Bureau of Mines Report of Investigation #3867, Ref 7.)
32
32. Elliot et al. (Ref 7.) in a Bureau of Mines Report carefully developed
the evidence for and against several possible causes. Perhaps the most plausible
reason for the failure is that 3.5 percent nickel steel used for the inner shell may not
have been suitable for the tank. Since the writing of the report, it has been
established in many research laboratories that 3.5 percent nickel steel has low
impact resistance at the service temperature of LNG. (from Ref 6.). Impact
resistance is related to brittle failure something we will discuss in the next case
study in further detail.
33
33. Today LNG tank failures are non-existent. As of 1988, 189 field
erected tanks were in service world wide. The inner wall material is typically 9%
nickel steel (or aluminum) whose strength properties and ductility (impact strength)
actually increases with lower temperature near the cryogenic temperatures of LNG.
Also, some tanks are constructed with aluminum walls. There have been very few
leaks and no ruptures in 28 years since these tanks have been used. The Tokyo
Electric Power Company Inc. has developed an in-tank black and white television
camera with a separate light source for the monitoring of the interior of an LNG tank
while it is in operation. The price is a bit prohibitive presently over $1,000,000.
34
34. Another significant tank failure was because of brittle failure. In this
case, a large inland liquid spill resulted. This spill affected 3 states and 2,000,000
people.
35
35. The day was January 2nd, 1988. A large above ground tank (48 feet
high, 120 feet in diameter, 4.0 MM gallon or 15,000 m3 capacity) was being filled
with diesel fuel. This was the first time that the tank was being filled to near
capacity since it's relocation (disassembled, moved in pieces, and reassembled)
from a site in Cleveland, Ohio to a terminal located in Floreffe, Pennsylvania. The
tank was within 2" of being at its fill capacity when at 5:02 PM a loud noise sounding
like thunder was heard coming from the area near the tank. An operator observed
the roof collapse, and a cloud of mist started to form. A kid playing street hockey
thought that the sound lasted a period of 3 to 4 seconds. Finally, Mrs. Alva
Rogerson who lived across the street heard a rumble for a period of 30 seconds
followed by the observation of a wave of foam splashing over the top of Route 837.
36
36. Unfortunately, the wave of foam observed by Mrs. Rogerson was part
of a 3.9 million gallon diesel fuel spill. 0.75 million gallons of diesel fuel reached the
Monongahela river which flows towards Pittsburgh into the Ohio river. The Ohio
river flows into the Mississippi. 10 states and approximately 2,000 miles of
riverways exist between the spill and the gulf of Mexico. A tremendous national
disaster was in the making and heroic measures by local, county, state, and federal
officials prevented the catastrophe from causing more pervasive consequences (ref.
9 p 1). A round-the-clock clean-up effort was undertaken and 80% of the fuel was
recovered (that still left 511,000 gallons unaccounted for!). The West Allegheny
Municipal Authority was the most seriously affected by the spill. West Allegheny
customers in the Robinson township (10,000 people) went without water service for
a period of 8 days. Other water companies affected included the Midland Borough
Water Authority and the Western Pennsylvania Water Co. Fortunately, no people
died because of this failure.
37
37. The tank was tossed aside as it collapsed. Relocating itself 120 ft from its
original base. The next slide is a picture of the remaining tank. Surprisingly, the
force of the oil spill itself only wiped out one cinder block building. However, several
earthen dikes whose purpose is to contain a spill were washed out.
38
38. A photograph of the collapsed tank. Remember that this tank was 48
feet high. In this picture it is barely 2 times higher than the man walking by.
39
39. The story begins with the need to replace a tank which required
extensive repairs. The terminal decided to replace the tank with a tank located in
Cleveland on a site which was being sold. This tank was 48 feet high and 120 feet
in diameter with a 96,000 barrel capacity. The tank was used for storage of #6 oil
or asphalt. Surprisingly, the date of original construction could not be established.
The name plate on the manhole cover read 1934. But other records indicated 1940.
40
40. This slide shows a sketch of the tank looking northwest showing the
fracture path. The fracture is suspected to have started at the first tier of plates
near a T weld.
41
41. "The collapse could have been and should have been averted. Both the
existence of the flaw and the tendency of the tank steel to react brittlely under normal
regional climatic and service conditions were discoverable through the application of good
engineering, construction, and inspection practices and by compliance with applicable
industry and governmental standards." Ref. 9 p iii.
Unfortunately, the tank was not reconstructed according to current accepted engineering
practice.
1. - Cutting of the old tank over the welds - An engineer specified that the old tank be cut at
the welds. The engineering company who dismantled the tank cut above the old welds. In
fact during the process a cherry picker overturned striking several of the tanks plates.
These were sledge hammered back into round. This, however, did not contribute to the
failure.
2. - Failure to get a fire marshall permit - A fire marshall's permit was never issued because
of lack of communication. The engineer in charge called the fire marshall's office and he
was told over the telephone that the permit was in the mail. No physical permit, however,
was ever issued. Once again, however, this did not have a bearing on the failure.
3. - The foundation for the tank failed compaction tests - As the foundation for the tank was
being built several rain storms occurred which delayed construction. Compaction tests
were done at different levels and several failed. The failed compaction tests were blamed
on weather conditions and the spots were rerolled. However, no retesting of the spots
occurred as the foundation was built up. Also a concrete ringwall foundation was
recommended. Crushed stone was utilized. (It should be noted that crushed stone is
allowed by API 650 and the foundation had no impact on the initial failure).
4. - Radiograph X-ray test did not follow a schematic - Unfortunately, this counted.
Radiographic tests were done on 39 welds. 22 welds were defective. However, closer
inspection showed that these defects were related to the old welds. Nor, was the
radiographic tests charted on a schematic so that one could trace back where the
radiographic results were from on the tank. API 650 requires 160 radiographs. This
emphasizes the need for complete inspections and tests vs. samples.
42
42. The radiographic defects were related to the old welds. This slide lists
the main defects found. Slide 43 is a photograph which demonstrates these
defects.
Inclusion of slag.
Lack of penetration.
Lack of fusion.
or excessive porosity was present.
43
43. This slide goes along with slide 42 to demonstrate different types of
weld defects. All of these defects were present in Tank 1338 and lack of penetration
at one point may have been the cause of the initial flaw where the brittle failure
occurred.
44
44. Full hydrostatic testing was omitted. Recommendations of both API
650 6th and 7th editions require full hydrostatic testing before filling the tank with a
hydrocarbon. Full hydrostatic testing means that the tank is filled with water to it's
rated capacity. The water is held in the tank for a period of time and leaks are
searched for. A purchase order was issued for 3,440,000 gallons of water
($4884.38). Final billing by the water company was for $802.22 worth of water
(500,000 gallons). Only a partial hydrotest was completed in which the water level
in tank was 5 feet and 15/16 inches. Remember that the tank was 48 feet high.
45
45. The cause of tank failure was a brittle failure. Materials have two main
properties; strength and toughness. Strength is the ability of a material to
withstand a force or force per unit area under constant conditions. Toughness is
the ability of a material to absorb energy. A dish is an example of high strength and
low toughness. Under constant force the dish doesn't easily break; however, just
drop it to create a sudden impact and it breaks. Strength failures occur under static
conditions. On the other hand toughness failures are often associated with impacts,
but they can occur under non-impact situations and brittle failure is such an
example. The dropping dish suffers a brittle failure. It is an instantaneous event.
The material cracks instead of deforms.
On this particular day, there was a convergence of three separate but necessary
factors to cause brittle failure. The full tank created high stresses in the first course
tank wall. The cold weather contributed to the increasing probability of brittle failure
because toughness is inversely related to temperature. Finally, latent flaws existed
throughout the old tank at old welds and the tank wall itself. Several tests can be
used to quantify resistance to brittle fracture. The Charpy V-notch test is the most
noted and the first course tank wall material measured 5.9 ft-lbs at 40oF. 13 ft-lbs is
considered a minimum value for tank wall material.
It should be noted that tank standards did not have material toughness
requirements until API 650 7th Edition, 1980. The original tank wall material dated
to the 1930's or 40's.
46
46. These photographs show the suspected defect where brittle failure
occurred. The photographs showed a lack of weld penetration. A gap of nearly a
1/2 wall thickness appears to be the point where the brittle failure started.
47
47. One recommendation can be drawn from this case. Reconstruction of
old tanks must follow current design practices (API 650).
48
48. Fortunately, the large tank failures are few and far between. However,
tank failures of much smaller magnitude and operational errors do occur more often
than admitted. Let us spend some time looking at common failures and how to
avoid these.
49
49. Overfilling: an operational error. This simple slide demonstrates that
storage tanks are not built to hold liquid above the ring wall. In fact, some accidents
have occurred where the tank roof was lifted off of the tank wall because of
overfilling. (Slide courtesy of Roy E. Sanders, PPG Industries)
50
50. Overfilling - A few ways
1. Lack of attention is self explanatory: If gas stations didn't have automatic shut off
nozzles, I wonder how many of us would overfill the gasoline tank in our cars?
2. Wrong setting of valves: Somehow a valve gets left open that should have been closed.
Material flows unnoticed into a tank. Often tanks are at different levels within plants. Any
plant located along a steep river bank may face this situation. One time in a paper mill a
valve was inadvertently left open between a tank storing 5 tons of a clay-water coating at
the coating prep building and a storage tank within the paper mill located in the basement.
Since the coating prep building was located on a side of a hill and the papermill building
along the river bank, the difference between elevation of the two tanks was about 50 feet.
One night material flowed between the two tanks over a period of 18 hours. In the morning
a 5 ton clay-water coating spill was found in the basement of the paper mill.
3. Error in level indicators: Often the experienced operator will know when the indicator is
wrong and issue a work order to correct it. Other times the operator may say, "well the
level indicator has never been correct. I know when the tank is full." Guess what? The
operator doesn't know all of the time.
4. Change in shift duty: A change in shift can mean a missing link in communication. One
crew may be filling a tank and inadvertently forget to tell the crew coming in that Tank 61 is
being filled. The second crew begins a routine shift assuming everything is under control.
Sure enough Tank 61 overfills and warning alarms suddenly change what would have been
a routine shift.
51
51. Some solutions to overfilling problems include: On large tanks usually
two independent methods of level checks are used. Redundant backup is used for
alarms and often the alarms may be of different types. Interlocks can prevent
valves from being left opened after a transfer. In order for the next processing step
to proceed, a valve to or from a storage tank has to be closed. Routine
maintenance is a must; level indicators can fail. Sometimes level indicators are set
to a certain specific gravity. If a lower specific gravity material is added to a tank a
false reading will result. Operators must be forewarned. Use check valves to
prevent back flow. Finally, communications. You will hear a lot about the need for
good communications within the plant environment.
52
52. Over-pressurization
Flat cone fixed roofs above tanks are generally designed to withstand only a few
inches water of pressure. Our lung pressure is about 25 inches water pressure. If
given the time (probably years) and correct piping (check valves etc.), you could
blow the roof off of a 120' diameter tank. The failure usually occurs at the roof/wall
weld. Both liquid and gas can cause overpressurization. For example, many tanks
have N2 atmospheres above a hydrocarbon to prevent explosive mixtures. A
nitrogen regulator can fail and suddenly a rush of high pressure nitrogen surges into
the tank. The relief valve which is poorly maintained may fail. Guess what? The
tank fails because of overpressurization.
53
53. This slide is a diagram to illustrate the point about how little pressure
8" water gauge pressure is.
54
54. The solution to overpressurization is properly designed relief valves,
vents, and vent systems.
55
55. Vacuum is probably the most common way a tank is damaged. The
ways in which it occurs are legion. Great ingenuity is sometimes required. First off,
less negative pressure is required than positive pressure to cause damage. Water
gauge pressure of -2.5 inches, the height of a cup of tea, is just enough vacuum to
suck in many tanks. (Ref. 1 p 73.)
56
56. This photograph, courtesy of Roy E. Sanders, PPG Industries, is a
picture of a tank sucked in. This tank was 8 ft (2.2 m) in diameter and 8 ft (2.2 m)
high and held hydrochloric acid for pH control. During a filling of the tank, it
overflowed and the vent line worked as designed. Then, the alert truck driver
responded and abruptly closed the delivery valve. Next, unexpectedly a partial
vacuum was created by a siphoning action of the overflowing liquid and the tank
was sucked in and totally destroyed (Ref. 17. p 9).
57
57. As mentioned above many methods exist in which a tank can be
sucked in. Here three are mentioned along with solutions.
1. Vents can plug: Birds have a habit of building nests in the most unusual places.
In fact, birds like tanks. Monomers can polymerized in vent lines as well as
polymers building up in non flow relief lines. Routine maintenance and field checks
prevent vents and relief valves from plugging.
2. Vents are covered: Occasionally vents are purposely covered for maintenance.
Alternative venting must be arranged.
3. A sudden change in temperature: A rain storm can do more damage than
imagined, especially if a crew has just finished steam cleaning a tank. The rapid
change in temperature can cause the steam to condense and the tank to suck in if
adequate ventilation (vacuum breaker valves) is not provided. (It has happened see
Ref.1.)
When empty, tanks are large cavernous vessels which require breathing during the
day's normal swing of temperature.
Prevention includes routine visual checks of vents and/or the use of vacuum
breakers. Never cover a vent without proper authorization.
58
58. There are many additional concepts which must be considered when
designing tanks. A few additional concepts are listed in this slide.
59
59. Let's finally conclude
Remember that tanks are fragile. An egg shell can withstand more pressure and
vacuum on a per unit mass basis. If we approach tanks with this concept, fewer
tank failures will occur because we are aware of the care that must be taken with
handling of eggs and tanks. Routine maintenance is a must as well as periodic
inspections. Remember to follow the standards. These standards were written by
practicing engineers who have probably already experienced a catastrophe or two
and know what precautions must be made. Finally, in terms of tank failures I hope
that you will disprove the following famous line:
"History repeats itself"
60
60. This package included contribution from several international experts in
storage tanks. I am indebted to cooperation from insurance representives for use of the
Mexico City disaster (Olle Johansson of Skandia Group) and for my initial education in tank
failures from Bob Nelson (IRI in Hartford Ct). The diesel fuel tank disaster material came
from an excellent report by the Tank Collapse Task Force. This task force was a group of
inter-agency and inter-disciplinary team assembled by Governor Casey of the
Commonwealth of Pennsylvania. The task force was headed by Mr. Keith Welks, Chief
Council for DER with the Commonwealth of Pennsylvania. He served as a reviewer of this
module as well as Mr. Mike Heilman who also served on the task force. I also acknowledge
CBI company for several photographs used in the beginning of this module. The principal
engineer of CBI, Mr. Taavi Kaups, educated me on tank construction and design. In terms
of LNG tanks, I acknowledge the assistance and encouragement of Armand Santacroce of
Distrigas and to Frank Bonomi and Frank Arricale of Boston Gas who allowed me the
opportunity to walk through the Boston Gas tank during dismantling. Roy E. Sanders of
PPG Industries, who is having a book published containing many case histories of tank
failures, is acknowledged for encouragement and direction. Finally Trevor Kletz is
acknowledged. Although I did not meet him, he is one fine engineer. All undergraduate
students should consider reading his books before beginning practice because of their
clarity, brevity, and succinctness. You will become an excellent engineer and manager by
avoiding situations which lead to the cases Dr. Kletz discusses.
The Northeastern Unversity media laboratory (Mr. Terry Beadle) is acknowledged for
assistance in the preparation of many word slides and photographs from references.
Finally, I also acknowledge my loving wife and my two sons who had to suffer lost of
"quality time" while I worked on the project.
Further material:
As I developed this module, I gained more and more information. One discussion was with
Mr. Taavi Kaups of Chicago Bridge and Iron who graciously reviewed this module and
made several helpful comments. One suggestion was to distinguish fail modes for different
types of tanks. The following abbreviated table covers some of the discussion we had.
You may copy this to use as a hand out during the presentation.
61
Summary of Large Storage Vessel Types and Possible Failure Modes

Fixed Floating
Low Cryogenic LPG
Roof Tank Roof Tank
Temperature Storage Tank Pressure Vessel

Applications: #6 Fuel Oil Gasolines Liq.


NH3 LNG LPG
Water Napthas
Butane, LPG O2, N2

Temperature: Ambient Ambient +5


to -50oC -50 to -200oC Ambient

Pressure: ---------atmospheric------
----- ~14 kPag (2 psig) ----- ~1.7 MPa (250 psi)

Standards: AWWA D100-84, H2O API 650


API 620 NFPA 59 A, LNG NFPA 58
API 650, Hydrocarbons
Appendix R API 620 Appendix Q

Possible Failure Modes1:

Foundation: ---settlement after construction---


---heating system fails in foundation system---
structural members shift
floating roof jams bottom buckles

leaks, roof caves in


loss of product
shell and bottom are stressed, leaks

Tank floor: --Corrosion due to H2O or a corrosive component--


Product leakage to the surroundings

Tank shell: ---high winds--- ---high winds--- ---


moist. penetrates ---cryogenic contacts ---flame impact---
shell buckles shell buckles
outer insulation--- outer tank wall--- BLEVE
---vacuum vent clog--- if
floating roof low increased boil off & outer CS wall cracks
---presence of H2S---
shell buckles
possible corrosion insulation packing shifts
metallic stress corrosion

Tank roof: ---hurricane--- ---hurricane---


roof plates lift off floating roof lifted
---vents clog or plug--- ---
roof drains clog---
suck in or over- floating roof sinks
pressurize tank
---overfilling---
roof lifts off wall

Other: ---Lightening---
fire hazard if tank is not properly grounded

1Note: (---Mode---) with possible consequence listed directly below


61
62
References
1. Kletz, T.A. "What Went Wrong? Case Histories of Process Plant Disasters," 1985,
Gulf Publishing Company, Houston, TX.
2. "Fixed Roof Tanks for Liquids," CBI Bulletin No. 3300 1990, Chicago Bridge and Iron
Company, Oak Brook, Illinois.
3. "Evaporation Loss Prevention," CBI Bulletin No. 533 1980, Chicago Bridge and Iron
Company, Oak Brook, Illinois.
4. "Horton Floating Roof Tanks," CBI Bulletin No. 3200 1990, Chicago Bridge and Iron
Company, Oak Brook, Illinois.
5. "Bleve! The Tragedy of San Juanico," First 1985, Skandia Group International
Insurance Company, Stockholm, Sweden.
6. "LNG Information Book 1981," 1981, American Gas Association, Arlington, Virginia
22209.
7. Elliott, M.A., Seibel, C.W., Brown, F.W., Artz, R.T., and Berger, L.B., "Report on the
Investigation of the Fire at the Liquefaction, Storage , and Regasification Plant of the East Ohio Gas
Company, Cleveland, Ohio, October 20, 1944," 1945 U.S. Bureau of Mines, R.I. 3867, Pittsburgh, PA
8. Mounce S. W., "Nine percent Nickel - 28 Years of Reliable Service in Liquefied
Natural Gas Containment," Nickel Development Institute Technical Series No. 10 030, Toronto,
Ontario, Canada M5C 2E3.
9. "Report of the Investigation into the Collapse of Tank 1338" 1988, Tank Collapse
Task Force c/o the Office of Chief Counsel 301 Chestnut St. Suite 400, Harrisburg, PA 17101.
10. "Investigation into the Ashland Oil Storage Tank Collapse on January 2, 1988," 1988
U.S. Department of Commerce National Bureau of Standards, NBSIR 88-3792, Gaithersburg, MD
20899.
11. "API Standard 650, Welded Steel Tanks for Oil Storage, Eighth Edition, November
1988," 1988 American Petroleum Institute 1220 L Street, N.W., Washington, D.C. 20005.
12. "NFPA 58 Production, Storage, and Handling of Liquefied Petroleum Gas (LPG) 1990
Edition," 1990 National Fire Protection Association, 1 Battermarch Park, Quincy, MASS 02269-9101.
62

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy