Chemical Engineering June 2010 - 41

IncIdent no. 3: BP RefIneRy,
texas cIty, tex., July 8, 2005
I
n the design layout of a duplex heatexchanger
arrangement (Figure 7) in the
resid-hydrotreater unit of the BP Refinery in
Texas City, Tex., the designer duplicated the
fabrication dimensions of the 90-deg fabricated
elbow-spool assemblies shown in Figure
7 as Elbows 1, 2, and 3. While the pipe
sizes and equipment nozzle sizes were the
same, permitting an interchangeability of the
fabricated elbow spool assemblies, the service
conditions prohibited such an interchange.
The shell side conditions on the upstream
side (at Elbow 1) were 3,000 psig at 400ºF.
The shell side conditions on the downstream
side (at Elbow 3) were 3,000 psig at 600ºF.
The intermediate temperature at Elbow 2
was not documented. In the initial design,
the material for Elbow 1 was specified as
carbon steel, Elbow 3 was specified as a
1 - 1/4 chrome/moly alloy. The reason for
the difference in material of construction
(MOC) is that carbon steel is susceptible to
high temperature hydrogen attack (HTHA)
above ~450°F at 3,000 psig, therefore the
chrome/moly alloy was selected for the
higher temperature Elbow 3.
High-temperature
hydrogen to furnance
Preheat gas
Heat
exchanger A
Elbow 3
(failure location)
Elbow 2
Bolted flange
(typical)
11/4 chrome
alloy piping
Elbow 1
carbon steel
Heat
exchanger B
Preheat gas
to separator
Low-temperature
3,000 psig
hydrogen feed
At 3,000 psig and temperatures above
450°F hydrogen permeates the carbon steel
and reacts with dissolved carbon to form
methane gas. The degradation of the steel's
tensile strength and ductility due to decarburization, coupled
with the formation of methane gas creating localized stresses,
weakens the steel until it ultimately fatigues and ruptures.
In January 2005, scheduled maintenance was performed on the
heat exchanger assembly. The piping connected to the heat exchangers
was dismantled and stored for the next 39 days. After
maintenance was completed, the piping was retrieved from storage
and reinstalled.
The elbows of different material were not marked as such and
the maintenance contractor was not warned of the different
MOC for the elbows. Elbows 1 and 3 were unknowingly installed
in the wrong locations. On July 8, 2005, approximately
five months after re-installing the piping around the heat exchangers,
the elbow in the #3 position catastrophically failed as
shown in Figure 8.
11/4 Chrome alloy pipe
Carbon steel pipe
Figure 7. Heat exchanger flow diagram
The one thing you can take
away from this incident is: Do
not dimensionally replicate
piping spools or assemblies of
different materials. The other
underlying, but significant
component you can also take
away is this: In the initial design
of a plant facility the engineer
of record will routinely
hold formal design reviews
Figure 8. Severed 8-in.
NPS hydrogen piping
Figure 9. Fragments of
the failed 8-in. NPS carbonsteel
spool
As you can see in Figure 9 the carbon steel, after becoming
progressively weakened by HTHA, fractured on the inside of
the pipe and catastrophically failed. The incident injured one
person in operations responding to the emergency and cost the
company $30MM.
Point 4. During a fire, another source
for valve leakage is by way of stem
packing and body seal, as mentioned
earlier. Leakage, at these seal points,
can be prevented with valves that are
not necessarily fire-rated, but contain
stem packing and body seal gasket
material specified as an acceptable
form of graphite (flexible graphite,
graphoil and so on). This is a fire-safe
material which is readily available in
non-fire-rated valves.
Point 5. The valve on the bottom of
the vessel should be fire-rated with a
fusible link or a fail closed position.
that will include all key personnel with vested interest in the project.
In doing so, include, among the attendees, key operations
and management plant personnel from one of the owner's operating
facilities, if available. These individuals typically bring a
lot of insight and knowledge to a review. Whereas the designers
may not have the wherewithal to think along the lines of issues
that might pertain to a facility turnaround, the plant personnel
will. These are issues that they normally think long and hard
about. Make use of this resource.
❏
Relying on an air or electric operated
valve actuator may not be practical. A
fusible link is most certainly needed
on a manually operated valve. The
contents of a vessel containing a hazardous
liquid needs to get pumped to
a safe location during a fire until such
time as the fusible link is activated,
closing the tank bottom valve, or the
pump fails. All valved gage and instrument
connections (SG-1) mounted on
a vessel should have a graphite-type
stem packing and body-seal-gasket
material at a minimum. Flange gaskets
at these gage and instrument connections
should be of a spiral-wound
fire-safe gasket type similar to those
mentioned earlier.
Specialty tankbottom
valves (XV-2) should be given
special consideration in their design
by considering a metal-to-metal seat,
or a piston valve design along with
fire-rated seal material.
Point 6. As mentioned in Point 2, the
residual fluid in Line A, after flow has
been stopped, should be drained to
the vessel. To help the liquid drain,
the pipeline should be sloped toward
the vessel. The intent, as mentioned
above, is to prevent sections of any
ChemiCal engineering www.Che.Com June 2010 41
http://www.Che.Com
Chemical Engineering June 2010

Table of Contents for the Digital Edition of Chemical Engineering June 2010
