Available online at www.sciencedirect.

com

Engineering Failure Analysis 15 (2008) 1076–1087

www.elsevier.com/locate/engfailanal

Failure analysis and creep remaining life of hydrogen

reformer outlet pigtail tubes
Chris Maharaj a,*, Clément A.C. Imbert b, John Dear a
a
Department of Mechanical Engineering, Imperial College London, Exhibition Road, South Kensington, London SW7 2AZ, UK
b
Department of Mechanical and Manufacturing Engineering, Faculty of Engineering, The University of The West Indies,
St. Augustine, Trinidad and Tobago

Received 28 August 2007; accepted 24 November 2007

Available online 4 December 2007

Abstract

A failure analysis and creep remaining life assessment on hydrogen reformer outlet pigtail tubes in a methanol plant was
conducted. The tube material was manufactured to ASME SB-407 (UNS N08810) material specifications. The observed
evolution and progression of grain boundary cavity nucleation, coalescence, microcrack formation and eventual cracks
were indicative of long-term creep failure, which occurred before the design life of the tubes. The remaining life evaluation
was performed based on metallurgical evaluation of the outer diameter (OD), middle of wall thickness, and inner diameter
(ID) percentage creep cavities versus tube OD. Very high statistical correlation was achieved by fitting creep data points to
the classic creep strain versus time curve by using third order polynomial equations. The onset of tertiary creep was found
to occur at 9.3% increase in the strain of the tube OD. This work affirms that using percentage area of creep cavities and
the correlation to increase in OD is a robust technique for the reliable and accurate determination of the creep remaining
life of superalloy tubes.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Metallurgical failure analysis; Life assessment

1. Introduction

1.1. Overview

A methanol plant requested a failure analysis and life assessment on its reformer outlet pigtail tubes. The
reforming furnace is heated by burning natural gas as fuel. The 384 catalyst tubes within the furnace contain
the process fluid that transforms methane (CH4) and steam (H2O) to intermediate reactants of hydrogen (H2),
carbon dioxide (CO2), and carbon monoxide (CO). The outlet pigtail tubes transport the process fluid from
the catalyst tubes to eight collection headers. The shape of the pigtail tubes allow for thermal expansion of

*
Corresponding author.
E-mail address: [email protected] (C. Maharaj).

1350-6307/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2007.11.009
 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1077

Nomenclature

A area viewed with 50 magnification

Ac total area of creep cavities viewed within A
ID inner diameter
LOM light optical microscope
OD outer diameter
R2 statistical coefficient of determination
SEM scanning electron microscope
SOP standard operating procedure
x percentage of creep cavities
X average value of x

the collection headers. The tubes also have the added advantage of allowing plant production to continue in
the event of a catalyst tube leak. This is done by clamping shut both the inlet and outlet pigtail tubes. How-
ever, there must be structural integrity of the tubes prior to and after being clamped.

1.2. Plant history

The plant was commissioned in April 1998 and is designed to produce 0.7 billion litres annually. Planned
outages occurred in July 2001 (a turnaround), March 2002 (for separator wash) and April 2003 (for E-305
wash). However in April 2004, after approximately six years of operation, the first unplanned plant outage
occurred with the failure of a reformer outlet pigtail tube. Subsequent pigtail tube failures occurred in October
and December 2004. The reformer temperature and pressure parameters were all controlled to within the stan-
dard operating procedure (SOP) [1] limits during the transient shutdown/startup conditions. Specifically, the
synthesis gas (which leaves the reformer and proceeds onto the waste heat recovery step) did not exceed 880 °C
and 19 bar gauge. Maximum reformer outlet header temperatures did not exceed 900 °C during steady-state
and transient operating conditions as indicated by thermocouples installed on the collection headers. Each
header temperature is a combination of 24 pigtail tube process fluid streams. However, it is possible that indi-
vidual pigtail tube process fluid streams may exceed 900 °C as no thermocouples are installed on the individual
tubes.

1.3. Outlet pigtail tubes

The tubes are designed for a nominal life of 100,000 h (11.4 years) [2] provided that reformed gas pressure
and temperature do not exceed 20 bar gauge and 900 °C, respectively. The nominal initial measurements are
42.16 mm outer diameter (OD) and 29.46 mm inner diameter (ID).
The pigtail tubes are Incoloy 800H material, manufactured to ASME SB-407 (UNS N08810) specifications.
Four different types of tubes (labelled A/C, B, D/G, E/F) are to be found in the existing methanol plant refor-
mer. The types vary slightly in length based on their respective positions in the reformer. The dimensions of
type E/F are shown in Fig. 1.
As found by Dillenback and McCoy [3], tube failures not only result in total plant shutdowns but can also
cause damage to other associated equipment such as catalyst tubes and refractory.
A similarly designed hydrogen reformer experienced outlet pigtail tube failures after approximately seven
years of operation. Microstructural evaluations revealed a grain size number of seven. This relatively small
grain size was attributed to incomplete annealing done during the pigtail manufacturing process as described
by Maharaj [4]. Based on this finding, an entire pigtail tube replacement exercise was executed in March 2004
that also involved an upgrade in material from Incoloy 800H to 800HT.
Met-Star Inc. [5] investigated several pigtail tubes that failed within a few days after furnace startup in a
steam-methane hydrogen reformer furnace. Field metallography and subsequent destructive analysis con-
1078 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087

Fig. 1. Type E/F pigtail tube dimensions (mm).

firmed that the pigtails were improperly heat-treated. Although the heat-treatment used complied with the
minimum grain size requirement for Incoloy 800H, no limit on maximum grain size was set by the purchaser.
This excessively large grain size, while beneficial in terms of creep strength, severely reduced resistance to ther-
mal fatigue.
Mal-distribution was reported [6] in reformers to the extent that reformed gas catalyst tube outlet temper-
ature at one end of the reformer differed by 100 °C from that at the opposite end. This high temperature scat-
ter could only result in a large reduction in the potential life of the reformer and outlet pigtail tubes.
For monitoring and controlling tube temperature, McLaughlin [7] describes a case where exxon mobil
installed thermocouples on every third outlet pigtail tube in their steam reformer heaters. Exxon mobil used
this procedure to startup and balance the furnaces more effectively.
Williamson [8] described a top-fired reformer design in refinery service that had eight rows of 42 tubes and
suffered an outlet pigtail tube creep failure. The tube material was Incoloy 800. The root cause of failure was
not identified in this case.
An Incoloy 800H pigtail tube that had approximately six years of service failed at the bend sidewall loca-
tion [9]. The maximum diametrical growth at this location was 5.1% and the ASTM grain size was found to be
zero. It was found that annealing at 1010 °C after cold bending to avoid grain coarsening would provide per-
formance improvement.

2. Procedure

2.1. Failure analysis

The failure analysis techniques employed were dimensional and microstructural inspection, chemical com-
position analysis, and hardness testing.
The increase in OD of the outlet pigtail tubes was measured by use of a pi-tape. This tape measures the
circumference of the sample tube with an error of ±0.25 mm. The measurement was taken three times and
the average was found. The average circumference was then converted to a diametrical reading. All locations
that were diametrically inspected were described by distance away from a bend.
Epoxy mounted transverse section samples were prepared for microstructural examination and were elec-
trolitically etched with 10% oxalic acid for 20 s. The microstructure was observed using a scanning electron
microscope (SEM).
Vickers hardness measurements from OD to ID were obtained using an automatic hardness indenter
machine with an applied load of 10 kg. Three readings were taken per location and the average found.
 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1079

Inductively coupled plasma OES, and LECO CS444 and TC436AR analysers were used to perform the
chemical composition analysis.

2.2. Remaining life assessment

Electrolitically etched epoxy mounted transverse section samples were prepared for the life assessment
investigation. The microstructure was observed via light optical microscope (LOM).
The following investigative and statistical techniques were applied:

(1) The percentage of cavities, x, in each etched sample image was obtained as shown in Eq. (1):
 
Ac
x¼ 100% ð1Þ
A
where A is the area viewed with a 50 magnification and Ac is the total area of creep cavities observed
within that field of view. Three readings of x were taken at each wall location (ID, middle of wall thick-
ness, OD) and the average value of x (X) determined.
(2) Data points of percentage area of creep cavities X for the ID, middle and OD transverse wall sections of
16 samples versus change in tube outer diametrical creep strain were plotted. Standard deviation bars for
the percentage creep cavities X were also shown. The choice of using transverse section samples and tube
OD measurements was based on results of a previous creep analysis [4] on similar outlet pigtail tubes
that indicated poor correlation with the use of longitudinal sections and tube ID measurements.
(3) A best-fit polynomial equation was applied to each set of data points that modelled the phenomenolog-
ical behaviour of the creep strain versus time curve. Characteristics of this curve include distinguishable
points of inflection in the transition from primary to secondary and secondary to tertiary creep. Due to
the uncertainty of time to failure in the tertiary creep zone, the onset of the tertiary stage was identified
as the end of the remaining useful life of the component.

Vickers hardness tests were also performed at the OD, mid-wall and ID locations. As in the failure analysis
procedure, three readings were taken at each location and the average found. The average hardness was plot-
ted against the pigtail tube outer diametrical strain.

3. Results

3.1. Failure analysis

Samples F39 and E37#5, whose locations are shown in Fig. 2, were used for the failure analysis. Their outer
diametrical growth was 0.4% and 14.8%, respectively. Sample F39 was obtained from the inside section of the
soc-o-let that connects the pigtail to the outlet manifold. While the remainder of samples were exposed to

Fig. 2. Location of samples E37#5 and F39.

1080 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087

time–temperature-stress effects, F39 was only exposed to time–temperature effects. Fig. 3 is a transverse sec-
tion SEM image of sample F39 at the OD location. Fig. 4 is a magnified view of Fig. 3 at region A. It shows
predominant cavity nucleation and coalescence at the grain boundaries. Isolated creep cavity growth, coales-
cence and carbide precipitates are also observed within the grain.
An SEM image is also shown in Fig. 5 of a transverse section of the OD for the E37#5 sample. A tenacious
oxide layer is observed that is approximately 200 lm in thickness in some regions.
Fig. 6 shows a magnified view of crack B observed in Fig. 5. The crack is intergranular in nature and the
surrounding region also reveals cavity nucleation, coalescence and growth at the grain boundaries.
Grain sizes are in conformance with the material specifications (ASTM grain size five or coarser) ranging
from ASTM grain size 3–5. This grain size observation eliminates the possibility of inadequate annealing pro-
cesses being a possible root cause of failure.
The chemical composition analysis is listed in Table 1, tested in compliance with ISO/IEC 17025. Confor-
mance to Incoloy 800H specifications was observed, with the exception of chromium (Cr) being just slightly
below the minimum requirements in sample E37#5.
Vickers hardness (HV) values were obtained at the OD, mid-wall and ID locations. They are given in Table
2. These values are below the minimum hardness value of 142 HV at 25 °C that was obtained on virgin Incoloy
800H material. The lower hardness values obtained may be associated with the combined effects of grain
boundary cavitation and the carbide coarsening/ripening.

Fig. 3. SEM image of F39 transverse section sample at the OD region.

Fig. 4. Magnified SEM view of A shown in Fig. 3.

 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1081

Fig. 5. SEM image of E37#5 transverse section sample at the OD region.

Fig. 6. Magnified SEM view of crack B shown in Fig. 5.

Table 1
Chemical composition analysis of the reformer outlet pigtail tubes
Sample C Si Mn P S Cr Mo Ni Al Cu V Ti N
F39 0.09 0.41 0.74 0.02 0.00 19.2 0.10 30.2 0.46 0.09 0.05 0.50 0.01
E37#5 0.09 0.40 0.66 0.02 0.00 18.5 0.10 30.1 0.44 0.09 0.05 0.49 0.22
Incoloy 0.05- 1.00 0.02 19.0- 30.0- 0.15- 0.75 0.15-
1.50 – – – –
800 H 0.10 Max Max 23.0 35.0 0.60 Max 0.06
Fe-balance.

Table 2
Vickers hardness (HV) values at 25 °C
Sample HVOD HVmid HVID
F39 134 128 126
E37#5 118 117 114
1082 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087

3.2. Remaining life assessment

For the remaining life assessment analysis, 16 pigtail tube samples of varying creep strain were used. It was
observed that the location of the failures was not limited to bends or welds. This suggests that the failures were
largely attributed to temperature and were independent of tube manufacturing operations (e.g. bending and
annealing). Figs. 7 and 8 are representative of the images that were used to calculate the percentage area of
creep cavities (X) for the ID, mid-wall and OD regions of the samples. Figs. 9–11 reveal graphs of the percent-
age area of creep cavities (X) for the ID, middle and OD wall, respectively, versus tube outer diametrical creep
strain. Fig. 12 is a graph of the average of the ID, mid and OD regions percentage area of creep cavities (X)
versus tube outer diametrical creep strain. Curves of the third order polynomial type were used to fit the data
points as they modelled very well the phenomenological effects of the typical strain versus time creep curve. An
associated statistical coefficient of determination (R2) value was calculated for each equation to give an

Fig. 7. LOM images showing F39 sample (a) ID, (b) mid-wall and (c) OD microstructures, respectively. Isolated grain boundary creep
micro-voids observed. Also, some globular precipitation identified within grain interior.
 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1083

Fig. 8. LOM images showing E37#5 sample (a) ID, (b) mid-wall and (c) OD microstructures, respectively. Very severe intergranular creep
cracks and voids observed.

indication of how accurate the equation fits the curve points. An attempt was made to correlate hardness to
creep remaining life but poor statistical correlation (R2 < 0.6) was obtained.

4. Discussion

4.1. Failure analysis

The SEM images suggest that the dominant failure mechanism is creep. At the grain boundaries, creep cav-
ities form, grow and coalesce to create microcracks that can lead to macrocracks and eventual cracking
through the wall thickness. The predominant intergranular nature of the creep cracking indicates that there
were negligible effects of low cycle thermal fatigue that is also validated by the low number of plant outages
documented in the plant history section.
1084 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087

45

40

Creep cavities at ID location (%) 35

30
R2 = 0.934
25

20

15

10

0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Tube outer diametrical creep strain (%)
Fig. 9. Third order polynomial graph of creep cavities at ID location versus tube outer diametrical creep strain.

35
Creep cavities at mid location (%)

30

25
2
R = 0.9506
20

15

10

0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Tube outer diametrical creep strain (%)
Fig. 10. Third order polynomial graph of creep cavities at mid location versus tube outer diametrical creep strain.

Notwithstanding sample F39 did not experience the effects of stress, it was still susceptible to the effects of
creep (cavity nucleation, growth, and coalescence). However the combined effect of stress, temperature and
time on sample E37#5 led to the detrimental effects of through-wall cracking. Chemical composition analysis
confirmed conformance with the material specifications with the exception of Cr being just slightly below the
minimum requirements in sample E37#5. The microstructures in all areas revealed a grain size 3–5, which is
good for optimum creep resistance. This eliminates the possibility of poor factory annealing process.

4.2. Remaining life assessment

A high R2 value of above 0.93 as obtained with the graphs for Figs. 9–12 indicates that the points on the
curve closely follow a typical creep versus time curve, thereby reaffirming the root cause of failure being due to
 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1085

40

Creep cavities at OD location (%)

35

30

25 R2 = 0.9612

20

15

10

0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Tube outer diametrical creep strain (%)
Fig. 11. Third order polynomial graph of creep cavities at OD location versus tube outer diametrical creep strain.

40

35
Average creep cavities (%)

30

25 R2 = 0.9684
20

15

10

0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Tube outer diameterical creep strain (%)
Fig. 12. Third order polynomial graph of average creep cavities versus tube outer diametrical creep strain.

creep. This confirms that tube outer diametrical measurements are very applicable for performing creep life
evaluations. The onset of both secondary and tertiary creep was obtained from Figs. 9–11 by identification
of the points of inflection in the best-fit polynomial curve. The average for the three graphs was found as out-
lined in Table 3.

Table 3
Onset of secondary and tertiary creep for Incoloy 800H based on ID, Mid, and OD wall percentage creep cavities
Location Onset of secondary creep (%) Onset of tertiary creep (%)
ID location (mm) 2.0 9.6
Mid location (mm) 2.0 8.6
OD location (mm) 1.5 9.6
Average creep growth (%) 1.8 ± 0.3 9.3 ± 0.6
1086 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087

5. Conclusions

The evolution and progression of grain boundary cavity nucleation, coalescence, microcrack formation and
eventual cracks were indicative of long-term creep failure. The random nature of the failures (not limited to
bends or welds) and good grain sizes observed suggest that the failures were largely attributed to temperature
and were independent of possible quality inconsistencies in tube manufacturing operations (e.g. welding, bend-
ing and annealing). Chemical composition analysis confirmed conformance with the material specifications
with the exception of Cr being just slightly below the minimum requirements in sample E37#5. Creep life
was most likely compromised by a combination of the adverse effects of extreme localized heating beyond
900 °C.
Knowing that the root cause of failure was solely attributed to long-term creep, a remaining life evaluation
was conducted based on metallurgical evaluation of the OD, mid-wall and ID wall percentage creep cavities
versus tube OD. Very high statistical correlation was achieved by fitting creep data points to the classic creep
strain versus time curve by using third order polynomial equations. Generally, more creep cavities were found
closer to the ID and OD of the tubes as opposed to the mid section. The effects of stress, temperature, and the
external environment would be contributing factors. Specifically, the highest tube stresses and temperatures
occur on the tube ID while the tube OD is exposed to air. A testament to the effects of the external environ-
ment on the tube OD was the observed oxide layer and the higher traces of nitrogen in the E37#5 sample. The
onset of tertiary creep was found to occur at an average of 9.3% increase in the strain of the tube OD. Using
available pigtail tube reading data and the onset of tertiary creep as the rejection criterion, a total tube change-
out was justified based on a desired time between plant turnarounds of seven years (benchmark).
Based on the extreme localized heating conditions and the creep life assessment, it was recommended that
the outlet pigtail tubes be completely changed out and also upgraded from Incoloy 800H to 800HT.
This study affirms that using percentage area of creep cavities and the correlation to increase in OD is a
robust technique for the reliable and accurate determination of the creep remaining life of superalloy tubes.
For a more accurate determination of future remaining life, the following on-line predictive inspection
methods should be explored:

 The installation of thermocouples on individual pigtail tubes located in selected positions in the reformer.
These thermocouple readings would assist in determining the nature of localized heating that takes place.
 To support the thermocouple installation, infrared (IR) thermographs of pigtail tubes should be performed
through the pigtail tube inspection ports. IR thermographs offer a holistic view of the temperature profile of
the tubes.

The incorporation of these on-line inspection measures will assist the operations department in determining
the hot regions in the reformer. The burner firing can be reduced in these high heat regions and this measure
would directly result in prolonged pigtail life.

Acknowledgement

Mr. Maharaj thanks the University of Trinidad and Tobago (UTT) for the sponsorship provided to per-
form part of this work.

References

[1] Saturation and reforming standard operating procedures. Rev. 1. Trinidad: Industrial Plant Services Limited; 1997.
[2] Calculation of heater-tube thickness in petroleum refineries: API recommended practice 530. American Petroleum Institute; 1988.
[3] Dillenback MD, McCoy CS. Major fire in a steam-methane reformer furnace. American institute of chemical engineers. National
meeting 30d, vol. 17. 1985.
[4] Maharaj C. Creep analysis of incoloy 800h reformer outlet pigtail tubes. Mechanical engineering. St. Augustine: The University of the
West Indies; 2001.
[5] Met-star Inc. Case study 5. Pigtail failure evaluation. 2007 [see website at <www.met-star.com/casestudies.html>].
[6] Nuttall K, Reid AR. The case for good distribution in steam reforming furnaces. London, UK: Humphreys and Glasgow Ltd.; 2005.
 C. Maharaj et al. / Engineering Failure Analysis 15 (2008) 1076–1087 1087

[7] McLaughlin J. Corrosion technology week/2004 report technology exchange group 205 refining industry information exchange.
NACE International; 2004.
[8] Williamson J. Risk based life management of catalyst tubes and pigtails. In: Proceedings of fourth annual ammonia and urea
conference. Singapore. 2000.
[9] Hoffman J, Lai G. Metallurgical evaluation of alloy 800HT pigtails. Corrosion/2005. TX, USA: Houston; 2005.