Different types of pyrolysis coil failure

Rob Gommans M.Sc.

Gommans Metallurgical Services
Stevensweert, NL

ABSTRACT

The most occur r ing f ailur e mechanisms ar e explained f or t ubes, bends

and out let par t s in pyr olysis coils. The f ir st dominant f ailur e mechanism

f or r adiant t ubes is t he combined act ion of car bur isat ion and cr eep

duct ilit y exhaust ion. This r esult s in bulging, bending and ovalisat ion of t he

t ubes. The second dominant f ailur e mechanism is br it t le f r act ur e dur ing

f ur nace t r ips, which can r esult in lar ge, longit udinal cr acks on many t ubes.

The pr ocess t echnological causes will be explained f or t hese dominant

f ailur e mechanisms. The main causes, over heat ing and mis-oper at ion, can

be pr event ed.

The met allur gical backgr ound of t he combined act ion of car bur isat ion

and cr eep duct ilit y exhaust ion will be explained. Car bur isat ion is

det er mined by t he pr esence of pr ot ect ive oxide scales, and t he nickel

cont ent of t he mat r ix. Cr eep duct ilit y exhaust ion is det er mined by t he

number of cycles (st ar t / st op- and decoke cycles) and t he nat ur e (or

sever it y) of t hese cycles.

The lif e assessment t echniques and inspect ion st r at egies should t ake

account of all r elevant f ailur e mechanisms.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

 Content

1. Macroscopic and microscopic description of failure mechanisms

1.1 Radiant tubes
1.2 Bends and outlet parts
2. Metallurgical background of the two main failure mechanisms
2.1 Carburisation and ductility exhaustion
2.2 Brittle fracture caused by furnace trips
3. Process technological background
4. Relation to inspection and life assessment
5. Conclusions

References

1. Macro- & microscopic description of failure mechanisms

1.1. Radiant tubes

The two main failure mechanism for radiant tube failure are :
- ductile failure caused by carburisation and ductility exhaustion; and
- brittle failure caused by thermal shock.

The ductile failure can be reconised by a bulge on the tube and a (short)
longitudinal crack on top of the bulge. The crack is not longer than the bulge.
In the micro-structure creep voids can be observed between matrix and
carbides. Further explanation about carburisation and ductility exhaustion is
given in par.2.1.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

The brittle fracture can be recognised by a long longitudinal crack which
“ends” in a fork-like apearance. Sometimes, the cracks result in circum-
ferential rupture or “windows” that fall out of the tube. The cracks can be many
meters long, and many times, a thick coke layer is present inside the tube.
In the micro-structure can be observed that the carbides have split. This is a
marked difference to the ductile fracture and can be easily recognised.

Creep elongation (also called stretching) occurs because of creep by the self
weight of the tube and the coke layer present in the tube and is influenced by
temperature, the load carrying cross section of the tube, and the material
used. It is generally known that HP-micro types creeps less than 45/35 types.
There is also a difference of the heigth of the firebox. In general, modern
higher furnaces suffer more from stretching than older smaller furnaces.

Another failure mechanism is overheating, which results in local melting or

overall melting of the tubes. Such an overheating can happen because of lack
of flow, coke blockage or burner problems (flame impingement). Lack of flow
can occur when inlet valves fail or in case of compressor problems.

Above 1100°C nitriding occurs at the OD of the radiant tube (flue gas side),
because nitriding kinetics are faster than oxidation kinetics. The rough as-cast
surface disappears and the surface becomes a smooth and glazed
appearance. Also, the lack of oxygen in the flue gas plays a rôle. Under
reducing conditions (many times caused by flame impingement by badly
adjusted burners) this can result in severe loss of wall thickness by alternating
oxidation and nitriding. The nitrides then causes spallation the oxides. As a
result a thick layer of oxides (up to 10-20 cm thick !) can be found on the
furnace floor. Sometimes, this is called oxide shedding.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

1.2. Bends and outlet parts

Erosion can be observed in 90° or 180° bends or in Y-pieces. This is caused

by hard coke particles during the decoking procedure (most accepted theory).
Some investigators believe that this erosion is caused by coke particles,
which are present during normal operation. The remedy is to modify the
decoking pocedure, so that the coke is gasified instead of being spalled.
Second remedy is to lower the gas velocity during decoke below 200 m/s.
Third remedy is to apply “internally stepped fittings”, which have been applied
sucessfully on many occasions.

Thermowells also suffer from erosion by coke particles. Succesfull solutions

include :
- thermowells fully made of Stellite
- rotate the thermowell at regular intervals
- apply “diamond shape” thermowells

As in the radiant tubes, carburisation also occurs in fittings (bends, Y-pieces,

tetra fittings etc.). In general, such carburisation is not life limiting for the coils,
because the radiant tubes will fail before the fittings do. This is because of the
high wall-thickness of the fittings.

Thick-walled bends and outlet fittings (Y-piece, tetra-fittings, flanges) may

suffer from carburisation and thermal fatigue. The carburised zone is brittle
and already cracks at low strains. The thermal stresses during start/stop-
operation and decoke cycles cause thermal fatigue at the inside of such thick-
walled components. A distinct feature is that the cracks are oriented in many
directions and are opened widely. Some people call this “crazy cracking” or
“mud cracking” because of this marked appearence. This type of cracking is
not considered to be life limiting of the component.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

2. Metallurgical background of the main failure mechanisms

2.1. Carburisation and ductility exhaustion

Carburisation is the carbon enrichment and carbide formation in the tube

material under influence of the presence of carbonaceous gases and high
temperatures. The carburisation rate is related to the carbon activity of the
gas and progresses exponentionally in relation to temperature. The material
and it’s ability to form and maintain an oxide layer is essential to prevent
carburisation.

There are several mechanisms known how an oxide layer can fail :
- diffusion distance thru the de-chromised zone is too large ;
- failure of the Cr-oxide layer by formation of Cr-carbide above 1050°C ;
- structural defects in the oxide (cracks, formation of spinels etc.).

According to the author’s opinion the mechanism described by

Ramanarayanan [ref.1] describes the degradtion mechanism best, see fig. 1.
It describes the process of carburisation after failure of the oxide scale.

The resistance of materials against carburisation is given by the Ni-content

and the presence of Si [ref.2], which forms a silica sub-scale, see figure 2.
Therefore, modern materials have a high Ni- and Cr-content and contain
1½ - 2 %Si. The most modern Cr-oxide forming material in this respect is
45Ni/35Cr-material (such as ET45 Micro). A step forward in oxidation and
carburisation resistance would be an Al-oxide forming material.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

 Figure 1 . Oxide degradation and carburisation mechanism
for pyrolyis s tubes according to Ramarayanan [ref.1]

Figure 2 . Influence of Ni- and Si-content on carburisation resistance [ref.2]

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

Creep ductility exhaustion is a complex mechanism in which the creep
ductility is exhausted by cyclic operation (such as : decoke stops, plant shut-
downs and trips). During normal operation a coke layer is deposited at the ID
of the tube. At end-of-run (EOR) such coke layers can be up to 20 mm thick.
During the decoking the coke is gasified, with the purpose that the tube is
clean after the decoke.
However, during the change between normal operation and decoking a
temperature drop occurs. Because the thermal expansion coefficient of the
metal is much higher than the coefficient of the coke, the metal-tube shrinks
on the “coke-tube”. Because of the high compressive strength of coke, the
metal does noet crush the coke. This causes high tensile stresses in the tube
metal, which relax during the (on-line) decoking procedure. The strain range
that occurs is proportional to the difference in themal expansion coefficient
and the temperature range according to the equation given in figure 3.

α - metal ≈ 19 µm/m/K

α - coke ≈ 4 µm/m/K

∆ε = ∆α · ∆T

Figure 3 . Tube metal and coke layer during a temperature drop

During a normal decoke such a temperature drop can be 100-200°C, which

causes a strain range of 0.15-0.30 % which correspond to high stress levels.
During the subsequent de-coking procedure these high stresses relax
because of creep. The damage mechanism is thus cyclic creep relaxation.
During each cycle the tube creeps a small amount; and at end-of-life the
material reaches it’s creep ductility. That is were the name “creep ductility
exhaustion” comes from. The mechanism described is similar to the failure
mechanism for reformer tubes [ref.3].

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

Tube life is thus dependent on :
- the number of de-coke cycles (n) ;
- the severity of the cycle (start/stop, on-line or off-line decoke, ∆T) ; and
- creep parameters (temperature, material, creep rate, creep ductility).

The combination of carburisation and creep ductility exhaustion is the

“normal” failure mechanism for pyrolysis tubes in ethylene plants. It results
in bulges, ovalisation and sometimes, tube bending as has been described in
par.1.1. Each single occurrence is relatively simple to explain, but the
complete process is complex and not yet fully understood.

2.2. Brittle fracture caused by furnace trips

As described in the previous paragraph the metal tube shrinks on the “coke
tube” during a temperature drop. During a normal decoke the temperature
drop is limited to about 100-200°C. During a furnace trip such a temperature
drop can be 500-1000°C. The strain during such a furnace trip is then
0.75-1.5%. This equals the rupture ductility of aged and carburised material
between RT and ~600°C. Since materials tend to crack when their rupture
ductility is reached … furnace tubes also crack. Because aged and carburised
furnace tube materials are brittle at these temperatures, they crack brittle by
splitting the carbides and subsequently the tubes. These brittle cracks can
extend for many meters.

The risk for brittle cracks is dependent upon :

- the severity of the trip ; meaning the temperature drop ;
- the thickness of the coke layer inside the tube ; thin coke layers at
SOR will be crushed, but thick coke layers at EOR will not ;
- the degree of brittleness of the material, which is determined by the
amounts of ageing and carburisation.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

As a guideline brittle fracture occurs most frequently when the tubes are
already a few years old and when the furnace is in the second half of it’s
operation run. This “dangerous area” is given in the diagram below.

coke thickness
EOR
DANGEROUS
AREA

SOR
0 % carb 100 % carb
coil life

Figure 4 . “Dangerous area” for brittle cracks

3. Process technological background

The underlying problem for both failure mechanisms of radiant tubes is

deposition of coke at the ID of the tube. The coke deposition causes
an increase of the thermal gradient over the tube wall, which results in
higher tube wall temperatures. For a tube life of ~6 years in high severity
cracking EOR-temperatures can be as high as 1100°C for HP-materials
(such as G4852 and derivatives) and 1125°C for 45/35 materials (such
as ET45 micro). At these high temperatures the materials carburise
and creep harder.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht

Therefore, many problems are related to furnace operation. Overheating,
flame impingement, nitriding, oxide shedding, reducing flue gases, erosion,
and specially brittle fracture ….. they can all be prevented by proper furnace
operation.
Creep elongation, carburisation, and creep ductility exhaustion ….. they can
be kept within limits by proper furnace operation, proper materials choice and
good design.

4. Relation to inspection and life assessment

In the ideal case the inspection and life assessment tools take account of the
failure mechanisms involved. However, there are no such tools available yet.

It can be helpful to measure carburisation. Several companies offer traditional

carburisation meters based on magnetic principles (permeability, eddy current
etc.). Recently, Shell Global presented a new pulsed Eddy Current technique,
which showed promissing results. However, a fully carburised tube can have
a remnant life of 1-2 years (if no furnace trips occur).

Dimension measurements can be helpful as well. If creep elongation

(stretching) is life limiting for the furnace, it can be monitored by the position
of guide tubes. In most cases creep in circumferential direction (tube swelling,
bulging and ovalisation) is life limiting. Diameter measurements by simple
strapping can be helpfull. However, there will be tubes that fail at low
ductilities and tubes that will fail at higher ductilities (up to 15% has been
observed ! ). This is dependant on the type of material (HK40, HP-Nb or
45/35), but also significant scatter between the individual tubes exists.

S+C 3rd European Ethylene Symposium 21-22 May 2001 @ Nümbrecht