MPWT19-15243

Investigation of Platformer Reactor Nozzle Cracking

Pranav Agrawal
The Bahrain Petroleum Company (Bapco)
P.O. Box 25555, Awali
Kingdom of Bahrain

Abdulla Ali Juma

The Bahrain Petroleum Company (Bapco)
P.O. Box 25555, Awali
Kingdom of Bahrain

ABSTRACT

This paper details with the unexpected cracking encountered in the outlet nozzles of all three reactors for
Platformer unit, during a scheduled shutdown. The unit was commissioned in 1957 and the reactors
metallurgy is as per withdrawn ASTM standard, A301 Gr. B (1Cr-1/2Mo). The isolated cracks were
located at the upstream edge of the inset nozzle and running vertically down. The cracking in the high-
pressure fixed bed reactor outlet nozzles was successfully repaired in-situ. This paper reviews the
circumstances which led to these failures and highlights the lessons learned from each failure.

Key words: Cracking, platformer, thermal fatigue, stress concentration, low alloy steel, thermal cycles.

INTRODUCTION

The “Platforming” process was so named to reflect the words platinum and reforming, and the image of
elevated octane. The first Platforming unit in the world was commissioned in October 1948(1).

The purpose of the Platformer is to increase the Research Octane Number (RON) of the gasoline feed
stock and increases its value for motor gasoline blending. When feeding Medium Straight Run (MSR)
Naphtha stock, the Platformer can feed 16,000 BPD to give a 98 RON platformate. However, the
maximum feed rate depends on the octane required and the condition of the catalyst.

The Platforming unit was commissioned in 1957 at Bahrain Refinery.

Cracking was observed in all three reactors of the Unit, during the planned shutdown. The reactors’
material of construction is A301 Grade-B. It is 1Cr-½Mo (equivalent to P12) low alloy steel and this
standard is withdrawn from ASTM. The inspection findings, investigation and learnings from this failure
are reported.

BACKGROUND

There are three reactors (38V1, 38V2 and 38V3) that operate in series. The typical maximum inlet
temperatures are 990 °F for 38V1, 970°F for 38V2 & 38V3. The Platforming reactions are endothermic
in nature and the typical maximum endotherm across 38V1, 38V2, and 38V3, is 80°F, 25°F and 20°F,
respectively. There is an intermediate heater between the first and second reactors, and another between
the second and third reactors to maintain the process temperature. The typical operating pressure at inlet
of 38V1 is 400 psig and outlet of 38V3 is 330 psig. It is summarized in the table below.

Table 1
Typical operational parameters of reactors

38V1 38V2 38V3

Maximum Inlet Temperature (°F) 990 970 970
Temperature Drop, ΔT (°F) 80 25 25
Operating Pressure (psig) 400 345 330

The shutdown frequency of the unit is six years. However, its operations also depend on the market
demand for platformate. Hence, in cases of lower demand, the unit is run either at lower throughput or is
shut down.

Brief Description of Platformer Unit Operation

The Unifiner, which is a naphtha hydro-treater, purifies the Platformer section feedstock. This is done
because of the sensitivity of the Platforming catalyst. It is susceptible to deactivation by sulfur, nitrogen
and halides, and poisoning by organometallic compounds in the naphtha feed.

The Platformer is a catalytic reformer, employing a special catalyst, which upgrades the feed in the
presence of hydrogen and in the process produces fresh hydrogen.

The Platformer (hydrotreated naphtha) feed is injected with a few parts per million of water and chloride.
The feed is then mixed with hydrogen-rich recycled gas and passed over the catalyst at the correct
temperature and pressure. The catalyst is held in three reactors and the feed temperature drops when
passing through a reactor. There is an intermediate heater between the first and second reactors and
between the second and third reactors to maintain the process temperature. The hydrogen and vaporized
feed enter a reactor at the top, then pass down the inside wall of the reactor where the mixture is
dispersed through scallop screens to pass through the catalyst bed, crossing from the outside through to
a central outlet pipe.

The reaction mix (mixture of hydrogen and reformed gasoline) leaves from the bottom of the third reactor,
is cooled, then sent into the product separator drum; here hydrogen rich gas is taken off. The condensed
liquid product (gasoline), called platformate, is fed from the separator drum into a stabilizer column
where it is stabilized to the required Reid vapor pressure (RVP), and sent to storage.
 Figure 1: Platformer process flow sketch.

Typical inspection carried on Reactors (38V1, 38V2, and 38V3):

• 100% internal Visual Inspection (VI).

• Ultrasonic Testing (UT) gauging at all Thickness Monitoring Locations (TML).
• External grit blasting on the shell weld seam joints and vessel skirt-to-shell weld, and MPT.
• Reactor external attachments grit blasting and MPT.
• Phased Array Ultrasonic Testing (PAUT) for accessible welds of shell seams and shell-to-head
seam joints, at random locations.
• 100% Magnetic Particle Testing (MPT) on shell, dish heads and nozzles internal welding.
• 100% Penetrant Testing (PT) of gasket face for top & bottom Nozzle ORJ [Octagonal or Oval
Ring Joint, also called Ring Type Joint (RTJ)].
• Positive Material Identification (PMI) of reactor parts.
• Removal of refractory bricks from bottom head for VI, PT & UT.

When these reactors were opened, and catalyst removed, internal inspection revealed cracking of the
outlet nozzles of all three reactors. The following additional inspection activities were performed:
• 100% bottom nozzles internal and external PT.
• 100% bottom nozzles internally and externally UT scanned with 0° and 45° angle probes,
wherever accessible.
 Figure 2: Sketch of reactor 38V1.

Damage History

The cracks were observed earlier in Jan’74 for the outlet nozzle of 38V3. The visual examination of the
outlet nozzle showed two cracks 1/2" deep and running 4" straight down the bore of the nozzle. Several
fine shallow cracks were also found on the end of the nozzle. Micro samples were cut from both types of
cracks. All cracks were grounded out and checked clear using dye penetrant and ultrasonic crack
detection techniques. Both straight cracks down the nozzle bore found to propagate through the full
length of the nozzle. Nozzles were weld repaired with preheat of 300°F and PWHT for 4 hours at 1350°F.
All reactor nozzles were sandblasted, ultrasonically crack detected, and visually examined. No further
cracks or other significant defects found.

After such failure in 1974, the cracks were found only during the March 2018 T&I (Turnaround &
Inspection) on the nozzles of 38V1, 38V2 and 38V3. The details are as following:

38V1 and 38V2

The cracking of the Reactor outlet nozzles of both reactors was in the parent material and unassociated
with the weld. The isolated crack was located at the upstream edge of the inset nozzle. It was
approximately ¾” in length on the top face (i.e. in the thickness direction) and approximately 1.25” long
crack running vertically down (i.e. along the bore of the nozzle).
 Figure 3: Crack on 38V1 (Left) and on 38V2 (Right).

38V3

The crack on 38V3 was close to or at the heat affected zone (HAZ) of the previous weld repair carried
out in Jan 1974. The isolated crack was located at the upstream edge of the inset nozzle. It was
approximately ¾”-1” in length on the top face (i.e. in the thickness direction) and approximately 0.5”- 1.25”
long crack running vertically down.

Figure 4: Crack on 38V3.

Repair Procedure

The following repair procedure during March 2018 is outlined here. Both ends of the internal crack in the
bottom nozzle were “VEE” drilled to avoid propagation during grinding. Then, the cracks were
successfully removed by grinding and dye penetrant tested to ensure the full removal of the cracks. The
weld repaired as per below procedure:

• Preheat to 300°F.
• Weld using ENiCrFe-2 (SMAW process).
• Maintain the inter-pass to 300°F.
 • Finally cover with Kaowool (insulation) for slow cooling.
• Full PWHT to 1350°F for 4 hours at controlled ROH (Rate of heating) & ROC (Rate of cooling).
This is in-line with API1 RP 934-E, which suggests that the use of PWHT temperatures higher
than the Code minimum requirements helps to prevent reheat and LCD cracking.
• Visual, DPT, UT and hardness testing (215 BHN maximum allowed). This re-inspection of welds
post-PWHT is also due to possibility of stress relief cracking.
• Smooth flush grinded to original profile followed by final DPT, UT and hardness tested. This is to
minimize the stress concentration in the HAZ.

The similar welding procedure was used in Jan 1974 for the same nozzle of 38V3. This welding repair,
which was carried out under technical service provider’s expert supervision, lasted for about 44 years
with no issue. Our technical service provider also provided a response which detailed their proposed
approach to weld repair, and this was in line with the company’s repair procedure.

It was decided that the corner would be radiused to reduce the stress concentration at the shoulder of
the nozzle. Hence, the 90° shoulder of the upstream end of the outlet nozzle of the three reactor nozzles
were ground to a radiused configuration during the 2018 T&I repairs.

INVESTIGATION FINDINGS

Operational Findings

The design pressure of all three reactors was re-rated in 1996 from 645psig to a lower pressure of
450psig. In the period from 2000 to 2005, the Platformer was frequently shutdown based on the
economics. The trends below show unit shutdown/start-up history.

Figure 5: PI data depicting frequent shutdown/ startup of platformer.

The Platformer operating temperature and pressure was checked between the last two consecutive
shutdowns and found to be consistent over this period, and within maximum design limits. There was no
‘long-term’ increases of temperature or pressure over the last two shutdowns of the Platformer Unit.

1 American Petroleum Institute (API), 200 Massachusetts Avenue NW Suite 1100, Washington, DC 20001-5571
 Figure 6: Example of relatively steady temperature and pressure of reactors.

Nozzle Design Shortcoming

One of the common factors of this failure for all three reactors was the location, i.e. at the outlet nozzle,
and at the shoulder on the upstream end of the nozzle. It is believed that the sharp 90-degree corner
configuration contributes to the onset of thermal fatigue cracking. It was therefore decided during the
2018 T&I that the corner would be radiused to reduce the stress concentration at the shoulder of the
nozzle. This is believed to be a minor contributor to the incident. This was done by grinding to a radiused
configuration during the 2018 T&I.

16” Sch 120

900# RTJ WN Flange

Figure 7: Sketch for outlet nozzle (Bottom of reactor).

Possible Mechanisms for Cracking Failure

The various possible causes were evaluated. By process of elimination and available facts, the
assessment was carried out.

Temper Embrittlement

The reduction in toughness due to a metallurgical change (primarily impurity segregation at grain
boundaries) in some low-alloy steels, as a result of long-term exposure in the temperature range of about
650 °F to 1070 °F, is called Temper Embrittlement. The loss of toughness is not evident at operating
temperature. However, embrittled equipment may be susceptible to brittle fracture during start-up,
shutdown, or hydrotesting after weld repair. It is highlighted in published literature that the 1¼Cr-½Mo
and 1Cr-½Mo steels are less susceptible. It is worth mentioning that 2¼Cr – 1Mo is the most susceptible.
The startup and shutdown operating procedures for the 38V1, 38V2 and 38V3 have the MPT integrated
into them. It clearly states that “the pressure of reactor should not exceed 150 psig when the shell and
nozzle skin temperatures are below 350°F. This is the Minimum Pressurization Temperature requirement
(MPT).”

Several events were identified where the MPT criteria were not met during the last two operating cycles.
However the variations in the operating temperature were in the range of 5-10°F from the MPT limit,
which was considered negligible.

For startup and shutdowns, there is a further restriction on rate of temperature increase/decrease at
100°F/hr. During a shutdown of the unit the heater fires are pulled out when the reactors reach a
temperature of 750°F. They are then allowed to cool under recycle gas circulation. Random examples of
startup and shutdown of all three reactors were checked, and the rate of reactor temperature change had
not exceeded the limits.

High Temperature Hydrogen cracking

HTHA is a time-temperature-pressure dependent phenomenon. This means the longer that a piece of
equipment is exposed to temperatures and hydrogen partial pressures exceeding its resistance limit, the
more damage will accumulate to the steel. Inspection for this is relatively specialized and includes
Advanced Ultrasonic Backscatter Testing (AUBT) or Ultrasonic attenuation testing (UAT).

The Platformer reactor 38V1, which has the highest temperature, operates with hydrogen partial pressure
in the range of 240 – 290 psia and with an operating temperature in range of 960 – 995°F. It was plotted
on the Nelson curve(5).

OPERATING ENVELOPE

Figure 8: Operating envelope (RED box) of reactors plotted on Nelson curve

There is safety margin of about 25°F. However, the reactors operate very close to the limit for the material
of construction, beyond which HTHA is likely to occur, and therefore the HTHA was considered as a
potential damage mechanism.

Hydrogen damage (blisters) was found on 38V2 in Nov’1974. Ultrasonic attenuation Testing (UAT) had
been carried out by external specialist in 1986. The replicas for microstructure analysis were taken in
1990, 1993, and 1996. The HTHA was not reported to be evident at that time. The boat sample also was
taken from all three reactors for metallurgical analysis in 1996. This was made possible by rerating of the
design pressure. Since then, there has been no evidence of hydrogen damage (hydrogen blistering) of
the reactors.
The HTHA was ruled as a likely mechanism for current failure, also because of the following reasons:
• HTHA is generally a damage mechanism that affects the bulk of the material, internally (though
this would lead eventually to surface cracking).
• Cracking is observed in the outlet nozzles, which has lower temperature than the inlet nozzles
(as the reactions are endothermic), and HTHA is more pronounced at highly stressed components
and at higher temperature.
• In addition to above, 38V2 and 38V3 operate at much lower temperatures than 38V1 and hence
has a higher safety margin than 38V1 against HTHA. However, cracking was observed in all three
reactor bottom nozzles.

Creep

Creep is the time-dependent deformation of a material under load. It is a generally slow developing form
of degradation occurring in metals under stress and higher pressures. Whereas, Low Creep Ductility
(LCD) cracking, or Creep Embrittlement, is a form of reheat cracking occurring due to long-term exposure
to temperatures >825 °F for 1¼Cr-½Mo and 1Cr-½Mo steels(4). It involves creep cracking of welds or
weld heat–affected zones (HAZ) due to carbide precipitation and diffusion of certain elements (As, Cu,
P, B, Sb, Sn, Nb, V, Ti, etc.) to grain boundaries.

The material of construction for all three reactors is ASTM A301 Gr. B (1Cr-1/2Mo), and all three reactors
operate in the creep range for this material of construction.

As per published literatures(3) and discussion with a process licensor, industry experience indicates that
cracking has occurred on Reformer Reactors even after 6 to 10 years of service. The cause of the
cracking has typically been creep embrittlement. The damage has occurred at stress concentrations,
such as the heat affected zone of weld areas. The cracking phenomenon may also be experienced due
to less than optimal post-weld heat treatment (PWHT) of these vessels during original fabrication. Low
PWHT temperatures and short PWHT time have been linked with the origin of the crack formation.

Cracking was found internally on Reactors 38V1 and 38V2 on the support ring-to-shell weld in Jan’1974,
and externally at the outlet nozzle of 38V3 in Nov’1974. The cracking on the outside surfaces of the
reactors at support welds and/or reinforcing pad welds had been observed many times since 1990
(reactors service life of 33 years then). The outside consultant was hired on these occasions for
metallurgical examinations and life assessment of equipment. The investigation work carried out during
the 1996 T&I on 38V1, 38V2 and 38V3, showed that there was no evidence of creep damage at that
time.

Although creep damage cannot be eliminated completely for this failure, it is highly unlikely. Most often
creep damage occurs on the outer surfaces of the reactors and is generally associated with inlet nozzle
(highest temperature) weld areas. The failure due to LCD (or creep embrittlement) or due to sub-optimal
PWHT is also ruled out, as such cracking manifests in welds or its HAZ. In this incident the cracking
occurred on the “inner” corner surface of the outlet nozzle. Furthermore, cracking was in the parent
material, unassociated with welds, for two out of the three reactors. Therefore, creep or creep
embrittlement were deemed unlikely to be the cause of the cracking of the reactors.

Thermal Fatigue

Thermal fatigue is a type of fatigue failure that is induced by cyclic stresses from repetitive fluctuations in
the temperature of equipment. The extent of damage will depend on the frequency and magnitude of
these temperature swings. Damage from thermal fatigue manifests itself in the form of cracking of the
equipment material. Cracks start on the surface and can propagate through the material. Unless cracks
are repaired, they may propagate through the material.

The prevention of thermal fatigue includes minimizing thermal stresses and cycling in the operation of
the equipment, reducing stress raisers, controlling temperature fluctuations (especially during shutdown
and start-up), and reducing thermal gradients. Unfortunately, thermal fatigue cannot always be
prevented.

The heating-up and cooling-down of such a thick-walled, high-pressure reactor system induces stresses
on the material which can be significant. The damage is accumulative and can lead to thermal fatigue
cracking.

There have been 160 startup and shutdown events of the Platformer unit in the period from Feb’1983
until April’2017. There is no available record of events which may have happened from 1957 until 1983
(26 years), therefore by extrapolation it was estimated that 122 such events have taken during this period.
Therefore, the total cyclic start up and shutdown events would likely have been in excess of 300 events
since commissioning in 1957.

A detailed review of an ESD trip that caused an emergency trip of the Recycle Gas Compressor (RGC)
was carried out. When the gas compressor trips, the temperature of the outlet nozzle of the reactors
initially drops, as there is no circulation. However, upon re-start of the compressor, due to the residual
heat from the catalyst and feed interheater, there is a rapid increase in the outlet nozzle temperature
(refer to PI data in Figure below). The temperature change for 38V3 (worst case) was of almost 600°F in
2 minutes after the compressor was restarted. This provides some insight into why it has been the outlet
nozzles that have experienced thermal fatigue damage.

Figure 9: Example of temperature gradient after recycle gas compressor trip.

The review indicated that the trip of the RGC, or a trip with the same consequences, has occurred 17
times since 2006. The thermal stress caused by the compressor trip (short-term anomalies) suggests
that a degree of accumulated thermal damage could have occurred during the RGC startup after a trip,
or some other similar circumstance. It is considered, however, that the high temperature gradient would
(for practical purposes) be unavoidable, without the lengthy cool down process of the bulk of the reactor
catalyst.

The location of the cracking was at the internal surface and unassociated with a weld. The geometry of
the upstream shoulder of the outlet nozzle was not ideal for resistance to thermal fatigue (square versus
radiused or chamfered). Hence, the thermal fatigue was considered to be the most likely
mechanism to have caused the cracking. The process of elimination of the other potential causal
factors and consultation with the process licensor and our technical service provider confirm that the
most likely mechanism of the cracking to be thermal fatigue.

Metallurgical Analysis

Since the cracks were found during T&I, and the reactors overhaul was in the critical path, there was a
time constraint in arranging specialists for In-situ Metallography (Replica test) before repairs. Since
nozzles are an integral part of reactors, no sample was available for further analysis. The repairs were
successfully completed, and the catalyst loaded inside the reactors. This failure and subsequent
investigation highlighted the need for carrying out the In-situ Replica and making this provision in T&I
work scope.

CONCLUSIONS

The cracking of the outlet nozzle of all three Platformer Reactors is attributed to thermal fatigue. The
reactors have experienced long term exposure (~61 years) to cycles of temperature and pressure from
planned/un-planned startup and shutdown activities, which has resulted in Thermal Fatigue of these
nozzles.

The geometry of the upstream shoulder of the outlet nozzle is not ideal for resistance to thermal fatigue
(square versus radiused or chamfered). However, this is believed to be a minor contributor to the incident.

Due to the lack of practical measures for predicting when thermal fatigue might manifest itself as cracking
(apart from minimizing such start-up/shut-down cycles and following standard operating practices), the
viable action is to monitor and repair during each T&I, if the cracking recurs. Hence, the inspection scope
of work for T&I, was amended to include –
a) Advanced Ultrasonic Backscatter Technique (AUBT) for surface & subsurface HTHA fissures.
b) In-situ metallography of the internal surfaces of the outlet and inlet nozzles for signs of long-term
damage, and
c) Shear Wave Ultrasonic Testing (SWUT) or Phased Array Ultrasonic Testing (PAUT) on outlet and
inlet nozzle-to-shell welds.

The above includes the recommendation from the process licensor for determination of cracking in the
reactor vessel shell walls and nozzles.

ACKNOWLEDGMENTS

The authors wish to thank the Management of the Bahrain Petroleum Company (Bapco) for their
permission to present this paper. The authors are grateful to Mahmood Bucheeri (Manager, Reliability
Engineering Department) for his support. The authors would like to thank the Technical Service
Department and Operational Excellence Department for their expert assistance.

REFERENCES
1. S. Metro, “IDEAS FOR RENT: The first fixed bed reforming unit history”, https://www.uop.com/ideas-
rent-fixed-bed-reforming-unit-history/ (July 10, 2019).
2. API RP 571 (2011), “Damage Mechanisms Affecting Fixed Equipment in Refining Industry”, (NW,
Washington DC: API Publishing Services).
3. API TR 934-D (2010), “Technical Report on the Materials and Fabrication Issues of 1¼Cr-½Mo and
1Cr ½Mo Steel Pressure Vessels”, (NW, Washington DC: API Publishing Services).
4. API RP 934-E (2018), “Recommended Practice for Materials and Fabrication of 1¼Cr-½Mo Steel
Pressure Vessels for Service above 825 °F (440 °C)”, (NW, Washington DC: API Publishing Services).
5. API RP 941 (2016), “Steels for Hydrogen Service at Elevated Temperatures and Pressures in
Petroleum Refineries and Petrochemical Plants”, (NW, Washington DC: API Publishing Services).
6. W. T. Becker, R. J. Shipley, Failure Analysis and Prevention - Volume 11, (Materials Park, OH: ASM
International: 2002).