Case Study on the Failure of a Centrifugally Cast Reformer Outlet Header in

a Hydrogen Generation Unit

Dasari Sukumar Reddy P Nagendra
YASREF CPCL (Indian Oil company)
Yanbu Chennai
King Abdul Aziz Road, Yanbu Tamil Nadu
Saudi Arabia India

Aditya Divvela Samarendra Das

Indian Oil Corporation CPCL (Indian Oil company)
Chennai Chennai
Tamil Nadu Tamil Nadu
India India

ABSTRACT
Hydrogen Generation Unit (HGU) serves as a critical process plant in refineries, playing a pivotal role in
meeting stringent environmental regulations governing the sulfur content in gasoline and diesel. At the
heart of this unit is the Hydrogen reformer, employing Steam Methane Reforming for hydrogen
production. The reliability of the reformer is critical for ensuring the safety and profitability of the refinery.

This paper presents a detailed case study on the failure of a Centrifugally Cast Hot Collector Outlet
Header constructed from Centralloy 4859-Cast -20% Cr-32% Nickel alloy, after nearly 20 years of
operational service. A comprehensive investigation methodology encompassing visual inspection,
stereoscopic imaging, microscopic examination, electron probe microanalysis, and analysis of process
parameters was meticulously conducted. The investigation conclusively attributed the failure of the outlet
header to High-Temperature Creep Damage. Salvaging of other portions of the Hot collector header was
undertaken, followed by a Remaining Life Analysis (RLA) which included High Temperature Tensile
Testing, Accelerated Creep Rupture testing.

RLA results indicated an estimated remaining life of 8 years, highlighting the localized nature of Creep
damage in Hot Collector header. Furthermore, this paper offers valuable insights into inspection
strategies and reformer spare part strategies crucial for enhancing reliability.

Keywords: Hydrogen Generation Unit, Steam Methane Reforming, Centrifugally Cast Hot Collector
Outlet Header, Failure Analysis, High-Temperature Creep Damage, Remaining Life Assessment,
Accelerated Creep Rupture testing, Reliability, Safety, Refinery Operations, Inspection Strategies, Spare
Part Strategies.
INTRODUCTION

A top-fired reformer, operational since 2003, consists of 264 reformer tubes arranged in six rows with 44
tubes in each row. After nearly 20 years in service, a failure occurred in one of the reformer’s outlet hot
collector headers. A detailed investigation was conducted on the failed section though it had served 20
years ;nearly twice the design creep life of 100,000 hours (11.7 years).

PROCESS DESCRIPTION
The desulfurized feed gas (Naphtha/RLNG) at approximately 360°C is combined with superheated
process steam at 286°C. The mixed feed is then heated in the convection coil of the Feed/Steam
Superheater to around 540°C.
Subsequently, the superheated feed gas is distributed over the catalyst tubes in the radiant section of
the steam reformer, where hydrocarbons are converted into hydrogen, carbon monoxide, and carbon
dioxide in the presence of steam over a nickel-based catalyst.
The overall heat effect of the reforming reactions is strongly endothermic, meaning heat must be supplied
externally to achieve the required conversion. This heat is provided by the combustion of PSA purge gas
and fuel naphtha from the battery limit.
The reformed gas exits the reformer tubes at approximately 856°C and 25.5 kg/cm²g.
The reformer is a top-fired, forced draft furnace consisting of two sections: the radiant section, where the
hydrocarbon/steam reforming reaction takes place, and the convection section, where the mixed feed is
preheated, steam is generated, and combustion air is preheated by the flue gases. Purge gas from PSA-
I (205-X1), supplemented by fuel naphtha, is burnt to provide the necessary heat for the reforming
reaction.
The burners are in the arch of the reformer, arranged in rows along the sides and between the catalyst-
filled tubes. The overall reaction in the reformer results from the following three equations: (followed by
the equations you mentioned).
Reforming: CnHm + nH2O -----> nCO + (2n + m)/2 H2
CH4 + H2O -----> CO + 3H2
Shift: CO+ H2O -----> CO2 +H2
 Figure 1: Process flow diagram (left). Schematic diagram of the reformer section (right).

The reformer effluent, or syngas, is an equilibrium mixture predominantly consisting of hydrogen, carbon
dioxide, carbon monoxide, steam, and some residual methane. At approximately 856°C and 25 ksc
pressure, the syngas exits the reformer tubes and reaches the hot collector header (outlet manifold) via
the outlet pigtails. The syngas from the hot collector is then cooled in the process gas boiler and sent to
the high-temperature shift reaction, followed by process gas heat recovery and pressure swing adsorption
(PSA) sections.
Table 1: Design details of Outlet Hot Collector header(Outlet Manifold)

Parameter Details
Design Code API 530
Service Process Gas
Operating Pressure 24 Ksc
Operating Temperature 840 Deg C
Design Pressure 28.5 Ksc
Design Temperature 880 Deg C
MOC Centralloy 4859 (20% Cr 33 Ni Nb)
Dimensions 10 Inch Dia
Operating Life 20 years

VISUAL OBSERVATIONS
During the normal operation of the reformer, a leak was observed from the 10-inch hot collector header
(outlet manifold). After removing the insulation, the following observations were made:
 The hot collector header (Centralloy 4859-Cast - 20% Cr, 32% Ni) had been in service since
commissioning in 2003.
 Axial cracks were observed on the hot collector header between the D1 and D2 pigtail weldolets.
 A 2% creep (OD growth) was observed near the leaky location.
 Crack
location

Fig 2: General Arrangement Dwg Depicting Failure Location (Left) & Close-up of Failure (Right)

C row

D row

D2 D1

Cracks were observed on Main header (before Axial Cracks were observed on Main header
buffing) between D1 & D2 weldolet (After buffing & DPT)

Figure 3: Close-up of Failure Location (Left) and DP Test Revealing Axial Cracks on Header (Right)
Comprehensive Inspection of Reformer Outlet Header System

Considering the failure location of the hot collector, a comprehensive inspection of the entire reformer
outlet header was carried out. The following key observations were made:
 OD Growth: A 2.8% increase in outer diameter was detected in the AB hot collector header at the
terminal end (north side) of the hot collector. To further assess this deformation, radiographic
testing (RT) was conducted at the bulged location, which revealed the presence of cracks in the
radiographic film.
 RT inspection: RT inspection also identified cracks in the EF hot collector, located on the same
side (north end) of the header.
 South End Inspection: No abnormalities, either visual or radiographic, were detected on the south
end of the hot collector header, indicating the issue was confined to a specific area.

Inference: These findings suggest that all observed damage was concentrated on the north side of the
hot collector header, while the south end remained unaffected

Summary of observations are furnished in the figure below

Fig 3A : Comprehensive Inspection and Key Observations of the Reformer Outlet Header
MACROSCOPIC OBSERVATIONS
The failed hot collector was cut longitudinally into two sections: a failed sample and an un-failed sample.
The failed sample was subjected to failure analysis, while the un-failed portion was subjected to remaining
life analysis.

 Failed Sample exhibited multiple cracks aligned in the axial direction on both the inner and outer
surfaces, as shown in the figures.
 A cross-sectional thickness (CT) examination of the failed sample revealed numerous cracks
throughout the section, as illustrated in the figure 6.
 Macro etching (using boiling 50% HCl and 50% water) was performed to reveal the crack
morphology. The macro-etched sample clearly shows a 100% dendritic structure with severe
cracking on the outer surface, inner surface, and CT section, as depicted in the figure 6.

Fig 4 :-Multiple axial directional cracks on OD surface(Left) & ID surface ( Right)

Fig 5 : Macro etched photographs of the sample on OD surface(Left) & ID surface ( Right)
 Figure 6: Visual Inspection CT section (Left) and Macro-Etched CT Section (Right)

CHEMICAL ANALYSIS
The chemical analysis of Hot Collector sample was carried out using Optical emission spectroscopy at
core region & results are tabulated below.

Table 2: Results of chemical analysis obtained through optical emission spectroscopy

Composition, measured Required as per

Elements value (weight per cent) Centralloy G4859

Carbon 0.085 0.06 -0.12

Sulfur 0.0073 0.030 max
Phosphorous 0.021 0.030 max
Manganese 0.87 1.50 max
Silicon 0.52 1.00 max
Chromium 21.02 19.00 – 22.00
Nickel 32.81 31.00 - 35.00
Molybdenum 0.0096 0.50 max
Copper 0.02 0.15
Niobium 1.2 0.50 – 1.35

Stereoscopic Analysis
Stereoscopic imaging was carried out on the inside suface of the tube sample to see the cracking pattern.
This analysis clearly indicated significant cracking on the tube inner surface as well as OD surface as
shown in Fig 7.
 Fig 7: Stereoscopic imaging of the sample (OD & ID surface)

Microscopic Examination
As the component has gone through long service at very high temperature, service induced
microstructural degradation / changes are likely to set in. To comprehend the same, CT (Cross section
thickness) sections of hot collector header samples were prepared for microscopic examination. These
samples were polished up to mirror finish and then etched with glycerigia solution.

Microstructure at all locations (inner, mid and outer sections) show severe creep related micro voids
adjacent to the primary carbides along with cracking. Numerous parallel cracks were observed at mid
section of the hot collector. Based on the microstructural evidences gathered, it is inferred that the sample
has gone severe creep damage (tertiary stage) as shown in Figs. 8 (A, B, C, D).

Fig 8A Fig 8B
 Fig 8C Fig 8D

Fig 8-A,B,C&D Showing optical microscopic images of the sample at Mid, ID & OD surface

Electron Probe Micro Analyser (EPMA)

Electron Probe Micro Analyzer (EMPA) was used to characterize the carbides morphology, composition,
creep voids and cracking pattern. The analysis emphasized voids initiated just adjacent to the primary
carbides, which is generally seen in creep damage mechanism as shown in Fig. 9(a-b). The different
types of carbides (chromium carbides, Niobium carbides) were identified in Backscattered electron
images with the help of WDX spectroscopy as shown in Fig. 10(a-b). Grey colour carbides indicates
chromium carbides, Niobium carbides are whitish in nature.

Fig 9A
 Fig 9-B

Fig 9A-B : EPMA images shows cracking pattern just adjacent to primary carbides

Fig 10A
 Fig 10B

Fig 10A-B: EPMA Analysis shows carbides composition at position 1, 2, & 3.

SCANNING ELECTRON MICROSCOPY

SEM was conducted at after metallography surface adjacent to cracked location & reveals presence of
creep voids/cavities adjacent to Metallic carbides.

Fig 11 SEM images revealed Creep Voids adjacent to Grain boundaries

HARDNESS ANALYSIS
Hardness examination of hot collector sample was carried out using micro & macro hardness tester.
Hardness reading was taken using Brinell Hardness Number (BHN). Hardness of the sample found to be
in the range of 142 – 159 BHN as shown in the Fig 11. Hardness of the sample found to be within the
acceptable range.
 Fig 12 Hardness pattern of Hot collector sample

FAILURE ANALYSIS SUMMARY

 Failed sample shows multiples axial directional cracks on inner, outer and cross section surface
of the sample.
 Macro etched sample also clearly shows 100% dendritic structure, with severe cracking in outer
surface, inner surface and CT section of the sample.
 Stereoscopic imaging also revealed multiples cracks at inside surface as well as out side
surface of the sample.
 Microstructures at inner, mid and outer sections shows severe creep micro voids adjacent to the
primary carbides along with cracking. Multiple parallel cracking were observed at mid section of
the hot collector.
 EPMA analysis emphasized voids initiated just adjacent to the primary carbides, which is
generally seen in creep damage mechanism.

REMAINING LIFE ASSESSMENT

The other half portion of the failed HOT collector header was salvaged and subjected to a remaining life
assessment (RLA) through high-temperature tensile testing and accelerated creep rupture testing.

Fig 13 Other Half portion of Hot Collector salvaged for RLA

High Temperature Tensile Test

As a part of RLA assessment, Tensile test at design temperature (880 °C) was carried out on a test
specimen drawn from the sample. The results are shown in Table 3.

Table 3: Tensile test at elevated temperature result:

Required as per
Parameter Measured Values Centralloy
G4859
Test Temperature (°C) 880 -
Diameter of specimen (mm) 12.51 -
Cross sectional area (mm2) 122.92 -
Original gauge length 50.00 -
Final gauge length (mm) 58.85 -
Ultimate Load (N) 17100 -
Yield strength (N/mm2) 97.62 80 Min
U.T.S. (N/mm2) 139.13 135 Min
% Elongation 17.70 -

Fracture Within Gauge Length

Inference: The results meet with the requirement as per specified in datasheet of Centralloy G4859

ACCELERATED CREEP RUPTURE TEST

As a part of RLA, Accelerated creep rupture tests (ACRTs) were carried out on the specimens drawn
from the sample. The tests were scheduled for exposure period of 100 hours and after 1 hour soaking.
The test parameters and the results of both the tests are shown in Table 4.
Table 4: Results of Accelerated creep rupture test.

Parameter Description Parameter Value

Specimen Specimen 1
Test Temperature 880°C
Applied stress uniaxial 55 MPa
Constant as per material G4859 18.3
Calculated LMP 23.40
Anticipated rupture 100
Actual test duration 48 Hours 15 minutes
Total strain 15.30%
Result Specimen ruptured after completion of 48 hours
and 15 minutes.
Constant ‘C’ determined based on 18.61
ACRT
Inference: Results indicates a lower than estimated rupture time indicating reduction in creep strength
most likely due to metallurgical ageing.

The test parameters for ACRT were selected from datasheet of Centralloy G4859. The specimens
ruptured after 48 h and 15 minutes in front of expected rupture time 100h. With this consideration, the
rupture life is calculated as under:

Table 5: Results of Remaining Creep Rupture life

MOC Centralloy G4859

Nominal/original thickness 24.00 mm
Minimum measured thickness 25.35 mm
No. of years in service 20 years
Evaluation Thickness 24.00 mm
Nominal outer diameter 254 mm
Design internal pressure 28.5 Kg/cm2g (2.793 MPa)
Induced hoop stress 13.38MPa
LMP at stress G4859 28.5

LMP constant (based on ACRT) 18.61

28500= (880+273) x (18.61+Log t)
Remaining Life => t= 106.10 = 1,282,704 hours

Inference: The results reveal specimen ruptured earlier against the anticipated rupture duration of 100
hr which indicates reduction in creep strength of material. Based on the test results, the remaining creep
life is calculated/judged to be 1,282,704 hrs, which on practical aspect, considered to be 8 years or more.

DISCUSSION AND ANALYSIS

As received photographs of the sample shows multiple cracks aligned to axial directional on inner / outer
surface as well as on cross section surface of the sample. Macro etched sample indicated 100% dendritic
structure, with severe cracking in outer, inner and CT section of the sample. Stereoscopic imaging also
reconfirmed multiples cracks at inside surface as well as out side surface of the sample.

Microstructures at inner, mid and outer sections shows severe creep induced micro voids adjacent to the
primary carbides along with cracking. Multiple parallel cracking were observed at mid section of the hot
collector. EPMA analysis emphasized voids initiated just adjacent to the primary carbides, which is
generally seen in creep damage mechanism.

Therefore, on the basis of the detailed investigation carried out on the failed sample of the hot collector
header of HGU unit, it is inferred that the component has gone through severe (tertiary stage) “Creep
induced degradation”. It has caused significant microstructural changes, formation of voids and
dendritic cracks resulting into failure of the component.

As part of the Remaining Life Assessment (RLA) of the other cut portion of the hot collector, located on
the opposite side of the failed section, the results of the accelerated creep rupture test indicate a reduction
in creep strength. However, the available thickness does not show any wall loss, and the estimated creep
life at design parameters is calculated to be more than 8 years, despite the effects of metallurgical aging.
This indicates the localized nature of creep, with more pronounced creep on one side and less on the
other side of the same hot collector header.

The subject reformer is a top-fired furnace with refractory flue gas tunnels on the radiant floor to facilitate
the flow of flue gas from the radiant section to the convection section.
The reformer has experienced multiple refractory flue gas tunnel failures in the radiant section between
2013 and 2021 & continued operation even with flue gas tunnel failures. These failures have significantly
impacted the distribution of flue gases within the radiant section.

Fig 14 Flue gas Tunnel failure inside the radiant section between 2013 to 2021

As identified in the comprehensive inspection of the reformer outlet header, all the damage to the header
was concentrated at one end of the hot collector. There are no thermowells or skin thermocouples
available to monitor temperature trends in individual reformer outlet hot collector systems. However,
based on the observations, it can be inferred that the hot ends on one side of the reformer are exposed
to higher temperatures than the other.
This temperature imbalance may be attributed to the previous damage to the refractory tunnels, which
likely caused maldistribution of flue gases within the radiant section. This maldistribution could have
resulted in uneven distribution of flue gases inside the reformer, thereby affecting the temperature
distribution within the hot collector header & resulting in failure of hot Collector header due to creep.

CONCLUSIONS

Creep-Induced Degradation: The failure analysis of the hot collector header reveals that the component
underwent severe (tertiary stage) creep-induced degradation. This degradation caused significant
microstructural changes, including the formation of voids and dendritic cracks, ultimately leading to the
failure of the component.
Localized Creep Behaviour: The Remaining Life Assessment (RLA) of the section opposite the failed
portion shows reduced creep strength despite no wall loss being observed. The estimated creep life at
design parameters exceeds 8 years, indicating that creep damage is highly localized. Creep degradation
was more severe on one side of the hot collector header, suggesting an uneven thermal and mechanical
load distribution within the system.
Impact of Refractory Tunnel Failures: Historical failures of refractory flue gas tunnels in the radiant
section between 2013 and 2021, followed by continued operation under compromised conditions,
significantly impacted flue gas distribution. This maldistribution likely led to uneven heat flux and
temperature gradients in the radiant section, contributing to the localized damage in the hot collector
header.
Temperature Imbalance Across the Hot Collector Header:- The absence of thermowells and skin
thermocouples on individual hot collector systems limited the ability to monitor and detect abnormal
temperature trends. However, based on the inspection findings, it is inferred that the hot ends on one
side of the reformer experienced elevated temperatures compared to the other. This temperature
imbalance likely exacerbated creep damage, particularly on the north end of the hot collector header.

Failure Mechanism: The combination of high-temperature exposure, creep-induced microstructural

degradation, and uneven flue gas distribution resulted in the failure of the hot collector header. The
interplay between refractory tunnel failures and temperature gradients further accelerated The
degradation process, making creep a dominant failure mechanism in the affected region.

Recommendations for Future Operation: To prevent recurrence, it is recommended to monitor

temperature trends more effectively by installing thermowells on individual hot collector headers and
providing skin thermocouples on pigtails. Additionally, timely repairs of refractory tunnels and regular
inspections of the hot collector system are essential to ensure even flue gas distribution and mitigate
localized thermal stresses that can lead to creep damage.

Furthermore, priority should be given to the inspection of hot collectors, including OD inspections and
random radiography, to assess the condition of the reformer hot collector system.

ACKNOWLEDGEMENTS

The authors thank the management and technical teams of Chennai Petroleum Corporation Limited &
Indian Oil Corporation R&D for their support, as well as the operations, maintenance, and laboratory staff
for their assistance with data, inspections, and testing. Special thanks to colleagues and reviewers for
their valuable feedback.
REFERENCES

1. Saad Al-Dhafiri, Jadee Ben Eid, Tareq Al-Foudari, Thammer Al-Ansari and Laxma Reddy “Creep
Rupture of a Hydrogen Production Reformer Outlet Header of Nickel Base Alloy and Preventive
Measures to Reduce Downtime,” CORROSION/2013, paper no. 2369

2. API 942 A (2014) :- Materials, Fabrication, and Repair Considerations for Hydrogen Reformer Furnace
Outlet Pigtails and Manifolds

3. API 571 (2020):- Damage Mechanisms Affecting Fixed Equipment in the Refining Industry

4. ASM Metals handbook, Volume 11, “Failure Analysis and Prevention”.

5. Mechanical Integrity of Reformer Outlet systems ,Compressed gas Association.