~~.

t.\1~'"
FURNACEINCID NT DURINGDECOKING
O~~,,~
~~..
BaJaS. D vakottai,CentralProcessEngineering ~ fr'
RobettM. ritchett,Channelview
PinakinB. istry, Specialty
PlantMaintenance
Engineering
CJarencA. Brewton,Channelview
-Metallurgy
Operations
~ t~
V 1'"
~ ~ .
~
LyondellChemicalCompany
Channelview, Texas
V ~...
~--.;;:--
~/
"'~, \' b
This paper originallypresentedat AIChE2006 Spring NationalMeeting "~~ I ../
Orlando,Florida
April23-27, 2006 ~ ~ ~~..-J;"
~

AIChEshall not be responsi e for statementsor opinionscontainedin papersor printe .~

in its publications. @>~ ...

ABSTRACT
In ethylene furnaces, cok forms in radiant tubes and decoking operations must be
performed regularly. During d coking, the upper convection section is at higher
temperatures than during nor al operation. The upper hydrocarbon preheat tubes are
protected by dilution steam d ring decoking. In one of Lyondell Chemical Company's
ethylene furnaces a small am unt of water was added to the dilution steam to prevent the
hydrocarbon preheat tube ou ets from exceeding the temperature limit of the piping.
During a regular decoking op ration, a convection shock bank tube ruptured, resulting in
positive pressure in the firebo .The excursion damaged the arch area refractory, a few
adjacent convection tubes, a d the convection header box.

This paper will review theldetails of this incident and the changes implemented to
prevent occurrence of similar 'ncidents.

INTRODUCTION
Ethylene furnaces are ty ically designed to handle a wide range of operating
conditions, including the abili to process feeds ranging from light gas feed, such as
ethane, to heavy liquid feeds. Regardless of the furnace type, two important furnace
operations are common to all furnaces: decoking and hot-steam standby. All furnaces
must be decoked at regular i tervals-due to coke in radiant tubes, quench exchangers,
or a combination of both. The frequency of decoking depends on the furnace design,
feedstock and furnace operat ng conditions. While normal furnace operation gets a lot of
attention, often, decoking op rations do not get the attention they deserve.

BACKGROUND
In order to protect the hy rocarbon (HC) preheat tubes of convection section during
the decoke process; it is com on industry practice to inject dilution steam. As shown in a
Lyondell furnace, which has f, ur feed control valves, 200 psig steam is injected into the
HC preheat tubes during dec ke (see Figure 1). To maintain the outlet temperature
below the 800F limit, a small amount of surface condensate was injected into each of
the four HC passes, along wi h 200 psig steam, using four manual valves.

3102/0506

~.
~
 Figu,~ 1. Schematico/ConvectionSection

OPERATINGISSUES
As the feed is purged fro the furnace with dilution steam, surface condensate is
injected into the feed coils to c ntrol the temperatures in the feed preheat section. This
task is a manual operation usi g a single one-inch block valve for each pass, that
operators adjust in very small ncrements as the board operator monitors the preheat
section temperatures on each f the four feed preheat coils.

There is no measurement l or this condensate flow; it is adjusted according to the

preheat temperature changes uring the purging process. Once the preheat
temperatures have stabilized, he furnace continues to purge until all hydrocarbons are
cleared from the furnace pipin and associated equipment before turning the furnace to
the atmosphere for decoking.

As shown in Figure 2, dur f.g every decoke operators tried to control the HC preheat

outlet temperature below 800 by manual injection of surface condensate (flow not

measured) along with 200 psi steam at the HC inlet. This process swung the HC outlet
temperature from 300F to 90 of, which, in turn, caused convection temperature swings
downstream.

It was difficult to control thf HC outlet temperature below 800F with surface

condensate injection. A small hange in the manual valve for surface condensate

resulted in excessive temperat re swings. In addition, a wide variation in the HC outlet

temperature occurred betwee the four passes of convection tubes, resulting in wide
swings of crossover temperatures, which is detrimental to effective decoking operations.

3102/0506
 SEQ~ENCE LEADING TO THE EVENT
On September 1, 2004, '

~valv~On

raaia~nt tubes. l~ ~- ~ -~ ~

~
IMPACT OF INCIDENT
There was no damage to tructural members or support systems, and no injuries
were associated with the incid nt. The incident resulted in unplanned ~!er outage.
during repairs.
As shown in Figure 4, the I C" pass shock ~ank!ube ruptured-, resulting in positive
pressure in the firebox" Ad"ace t s ck bank tub s suffered minor failure resulting in
~cracks in the~besDue to the force of ru ture, part of the header box was ushed off to
one side, away from its bolts.

Cradle
support

Ii.

Sp'it tube on West End of Furnace

at ~radle suppor1
Figure 4. Failed! TubeLooking Upfrom Radiant Arch

3102/0506
 As shown in Figure 5,

refractory aamage to peep do9r~ and other areas in the radiant section.

~ ~

Figure 5. {IJsulationD~mageAround Arch Area

Figure 6 shows the shock ~ ankFAILURE

tube ANALYSISthat
sample had ruptured. Figure 7 shows the

crazed pattern of cracking on 10 of the tube, and Figure 8 shows the corresponding

microstructure

Figure 6. failed Tubefrom Row # 1

3102/0506

~
 Figure 7. Crazfd Pattern of Thermal Fatigue Cracking on ID

Figure 8. Microstructur, and Crack Originationfrom ID Magnified 100X

3102/0506
 ~
Visual examination and icrostructure study were carried out. The inside surface of
this 347H SS exhibited exten ive crazing by thermal fatigue. The microstructure did not
reveal extensive high-temper ture damage.

The shock tube failed by ultiple modes of thermal fatigue and stress rupture.
Thermal fatigue frequently re ults from intermittent wetting of a hot surface by liquid that
has a much lower temperatur .The wetted surface contracts rapidly, whereas the metal
below the surface does not; this produces large biaxial tensile stresses in the wetted
surface. As the water absorbs heat and evaporates, the temperature of the surface
returns to its previous value, a d surface stresses are relaxed. After a sufficient number
of thermal cycles, a crazed pa ern of cracks appears on the surface. Apparently, cyclic
thermal stress initiated cracki g at the ID tube surface. Pressure stresses assisted in
defining the preferential directi n of fatigue-crack propagation (parallel to the tube axis
and transverse through the tu e wall), and stress rupture cracking became the ultimate
fracture process in the weake ed tube. The large displacement, evidenced by the
ruptured tube, indicates subst ntial energy during the rupture event. The excessive,
uncontrolled introduction of su ace condensate before the rupture is thought to have
rapidly increased the tube pre sure before the rupture. The external firebox damage
supports this conclusion of su stantial energy. In this case, th~ raDid introduction of ~
into the hot tube resulted in un ontrolled vaporization and a ra id increase in pressure.

DISCUSSION
The root cause of the tub failure was_e~cessive su!:fa~e cQ!Jd~!Jsate lnje~d_al_o~
with 200 psig steam. Ther was I inadeq~~te ~2n!rol for i[!j~c!!onof!~~~~~
condensate. The surface con ensate is a~ualinie~!on Do~that i contrQII~d ~:{.
~ia~-~p~r~tors at the re u st of the board 0 erator. The board operator controls the
ee pre ea emperatures b monitoring those temperatures in the control room and
then having the outside operat r make flow adjustments with a guard valve.

The in.ection flow was n etered. The flow was managed by E~u_nting the turns of
~ valve handle. The manual alves ~:~~_g~_a_ngedprio~ to decoi5.e:Which had an~-
impact on the opening and clo ing o~ the valyes.~-

Calculations ,to understand! the reason for extreme temperature swings revealed that
, chan,qed th.e ~mount,
a

---figure ---
9.-changing surface co~densate from gpm per pass cha-nged the HC
outlet temperature from 1,1300~ to 315F.

HC preheat temp..F

,/, liquid at HC Inlet

--HC preheat outle

temp calc !
--% liquid

0 112 3 4 5 6 7 8
gpm s~rface condensate per pass

Figur~ 9. Impact ofSurface WaterInjection

3102/0506
 M DIFICATIONAND RESULTS
Rather than modify the c ntrol s stem for surface condensate injection, the following
modifications were made to t e decoking procedure.
Using. " .steam in the HC bank, one of the main

changes d in lowering the lower steam temperature at

the HC ba would directionally ~elp lower the outlet-
~ature ~

In addition, the dilution s usted between the diluti ank and

HC bank to maintain sufficie and stay within metallur

Figure 10 shows the temp rature swing during a decoke with the modified decoking
procedure.

~iI
~~
"'

Figure 10. TemperatuteProfile with Modified DecokingProcedure

This modification eliminated the surface condensate injection, avoiding potential

shock bank failure caused by liq id water carry-over. This modification also resulted in
easier temperature control, with minimum variation between passes and minimized
swings in temperatures.
-~

3102/0506