Replacement of Reformed Gas Waste Heat
Boiler at Terra Courtright Ammonia Plant
The Reformed Gas Waste Heat Boiler (C110) at Terra Courtright facility, a Steinmueller designed
equipment, was in service since 1985. During the 22 years, it had experienced one catastrophic
failure, in 1995 after 9 years of service, which required replacement of the first section of the boiler.
In the early years, several metal dusting related problems on non-pressure components were reported
as well, mostly in the second section of the boiler, which were eventually solved by adopting different
metallurgy. Subsequent to first compartment replacement in 1995, the boiler was in operation for the
next 8 years without any signs of known failure. In 2003 and 2005, during the routine inspection of
the boiler in turnarounds, it was observed that the outlet tube sheet of the first compartment
developed numerous cracks in the ligament sections. This paper describes the type of failure observed
on the tube sheet and the reasoning that went into the decision to replace the equipment in the year
2007. The paper also highlights the activities that were involved in the design, procurement, onsite
assembly & fabrication and installation of the waste heat boiler.
Bhaskar Rani
Terra International (Canada) Inc.
Courtright, Ontario, Canada, N0N 1H0
amine-based chemicals with an oxygen scavenger,
supplied by NALCO Canada Inc.
Description
First Compartment
T
erra International Inc., operates a 1120
MTPD ammonia plant at Courtright, On- Prior to 1995, the first compartment had 559 tubes
tario, Canada. Engineered and built by of 39 mm OD with 1,050 mm long ferrules with a
UHDE in 1985, the plant is based on ICI’s AMV shell OD of 1,994 mm. The first compartment was
design. Since then, the plant has been consistently replaced in 1995 subsequent to a catastrophic fail-
operating above the design rates at 115-120% ca- ure of tube-tube sheet weld on the inlet side. De-
pacity, with > 98% on stream time. tails of this failure were presented at the 1997
AIChE symposium. The reason for the failure was
The original design of C-110 waste heat boiler was attributed to the operating conditions at higher than
a thin flexible tubesheet with two compartments allowable heat flux and the phenomenon of DNB
connected by an intermediate channel. It is basi- (departure from nucleate boiling).
cally a shell and tube type heat exchanger (Figure
1). Hot process gases at 950-955 ºC flow through During the 1995 replacement job the first section
the tube side and are cooled to 690-710 ºC outlet was re-designed by Steinmueller to reflect the
the first compartment and to 525-550 ºC at the out- 115% operating rate of the plant. The revised de-
let of the boiler. The shell side is connected to a sign of the first compartment had two different
steam drum (F102) operating at 125 barg. About sizes of tubes. The small tubes (38 mm OD, 5.6mm
165-175 TPH of steam is generated from the sys- thick), were arranged in approximately two outer
tem. BFW treatment is a zero solids program, using circumference rows around the large tubes (42.4
2008 227 AMMONIA TECHNICAL MANUAL
mm OD, 6.3 mm thick). The number of tubes was Materials of Construction of C110
increased to 613 (445 large tubes and 168 small Materials of C-110 boiler are as follows:
tubes) with a shell OD of 1954 mm and a ferrule • Shell Gas section = SA387 Gr.12 Cl.2 (al-
length of 410 mm. The effective tube length be- loy steel Cr-Mo)
tween the tube sheets was 900 mm. The thickness • Shell Water section = SA533 Gr.B Cl.1
of the tube sheets is 30 mm. (Mn-Mo-Ni alloy steel, quenched and tem-
pered)
All of the shell section, exposed to the gas, was re- • Nozzles Gas section = SA 182 F12 (forged
fractory lined in two layers. The outer layer was a fitting for high temperature service)
castable material and the inner layer was brick lin- • Nozzles Water section = SA508 Gr.B
ed. Only the inlet tube sheet in the first compart- (quenched and tempered vacuum treated
ment was refractory lined to protect it from high carbon alloy steel)
temperature and to hold the ferrules in place. The • Internal Riser Pipe = SA 106 Gr.B.
outlet tube sheet was not refractory lined.
Second compartment First Compartment Outlet Tube
The second compartment had 415 tubes of 38 mm Sheet Failure – 2003 & 2005
OD and 5.6 mm thick with 950 mm long ferrules.
The central gas bypass section (a set of 12 tubes, Routine inspection of the boiler is carried out in all
114.3 mm OD, 13.5 mm thick) was for exit gas biannual turnarounds. In 2003, cracks were discov-
temperature control through the operation of three ered in the ligament area of the outlet tube sheet of
dampers at the outlet end. The effective tube length first compartment. The tube sheet is 30 mm thick
between the tube sheets was 4106 mm. The thick- made of 1 Cr- 1/2 Mo material. About 18 cracks
ness of the tubesheets was 23 mm. The shell sec- were found. They were all ground to the depth (the
tion exposed to process gas was refractory lined depths ranged from 1.5 mm to 6.9 mm) and left
similar to the first compartment. The inlet since the depths did not impinge on the allowable
tubesheet of the second compartment was also re- minimum wall thickness (20.6 mm). Based on cal-
fractory lined, presumably to hold the ferrules in culations it was recommended that, if the crack ex-
place. The second compartment was in service ceeded 9.4 mm in then weld repair were required.
since 1985.
Upon recommendation from UHDE, refractory was
Metal dusting phenomenon was experienced at the installed on the tube sheet to reduce the effect of
damper sections of this chamber and required re- higher than design operating temperature in that
pairs in every turnaround. Inconel 600H material section (design skin surface temperature was 450
was used in this section with slightly better success ºC, operating bulk gas temperature = 690 ºC)
than the originally used material Incalloy 800H.
The ferrules used in C110 are also Inconel 600H. In 2005 more than 150 similar ligament area cracks
were discovered despite the installation of refrac-
Both the compartments are connected to steam tory two years earlier. The refractory was in good
drum by a series of risers and downcomers, which condition. The depths of cracks ranged from 2 mm
facilitate natural circulation. to 8 mm. Again, since none of the cracks exceeded
the critical depth of 9.4 mm, no weld build was re-
quired. All cracks were ground to depth and left as
is. Refractory was reinstalled and equipment re-
turned to service.
AMMONIA TECHNICAL MANUAL 228 2008
Magnetic Particle Testing, utilizing Magnaflux Y-6 The data for 1 Cr-1/2 Mo material indicates that
Yoke instrument was utilized for the inspection. the tensile strength at 700 ºC is 27% of its room
temperature value of 16,000 psi. A combination of
Failure Analysis - 2003 turnaround examina- the deep tube holes, thin ligaments and potentially
tion high operating temperature suggest a complex
A ‘V’ sample collected in 2003 from a ligament at thermal stress pattern that results in stresses that
row 21/tube 10 revealed that the crack is trans- exceed the tensile strength of the material. Maxi-
granular and exhibited a scale on its surface. (Fig- mum design temperature for 1 Cr-1/2 Mo steel is
ure 2). Examination of the ligament sample surface approximately 648 ºC, while recommended operat-
indicated that there was no evidence of carburiza- ing temperature is 480 ºC or below. Thus the liga-
tion on the crack surface itself. ment cracking was a result of thermally induced
stresses.
The section did not contain randomly spaced
graphite nodules. As per the metallurgical engineer, Failure Analysis - 2005 turnaround examination
graphite nodules are a definite indication that mi- Despite installation of refractory on the tube sheet
crostructural graphitization has occurred. Graphite in 2003, a large number of ligamental cracks
nodules form as the pearlitic phase in the micro- (>150) were found in 2005 turnaround inspection.
structure decomposes into ferrite and graphite. This (Figure 5). It is clear that we had a serious problem
decomposition occurs over long times at tempera- that needed further study to assess the integrity and
tures ranging between 425 and 550 ºC. life of the boiler. The next section of this paper de-
Examination of the microstructure revealed that it scribes this analysis in detail.
was ferrite and spheriodal carbide with some in-
tergrnualar cracking along the surface. (Figures 3
and 4). The carbide had coalesced into large globu- Thermal Evaluation of C110 boiler
lar carbide and the grain boundary carbide has UHDE was contacted to perform a full thermal
spheroidized as evidenced by lack of granular evaluation of the existing C110 boiler. UHDE en-
boundaries. Spheroidization in 1 Cr-1/2 Mo occurs gaged the boiler specialist services of Borsig to do
after prolonged service at 565 to 745 ºC. The nor- this evaluation. The basis of the study was actual
mal microstructure for this material would be fer- operating data at 1350 MTPD, and the scope of the
rite, pearlite and finely dispersed carbide particles. study included the following:
• Heat transfer calculations of the existing
Conclusions from the 2003 examination C110 boiler
The surface of the tube sheet had been exposed to
• Natural circulation ratio for the two com-
temperatures in the range of 565 to 745 ºC. The
partments of the boiler
temperature through the cross section appears to
• Temperature distribution in the outlet tube
drop off rapidly as indicated by the condition of
sheet of the first compartment (without in-
microstructure at the tip of the section removed. A
sulation)
high temperature gradient is expected in boiler type
• Recommendation for possible improve-
equipment where the temperature differential
ments and/or a solution to the problem
across the cross section of the tube sheet can be
great (in this case 690-710 ºC on the process side
Heat transfer analysis
to 325-330 ºC on the water side).
The temperature difference between the tubes and
shell is a measure for the differential thermal elon-
The major crack in the ligament was found to be
gation of the two parts. Thus it is one of the main
transgranular, a characteristic of high stress failure.
parameters that influences the load on the flexible
2008 229 AMMONIA TECHNICAL MANUAL
tube sheets. The results of the study indicate that compartment outlet tube sheet. The modeled parts
these temperature differences are rather large for included the tube sheet, the outlet ends of the
both compartments as shown in Table 1.0. For the tubes, the forged ring and parts of the shell and the
first compartment, the larger difference is compen- intermediate channel. The boundary conditions ap-
sated a little by the very short tube length, thus the plied were as follows:
actual differential elongation between the tubes and • gas and water bulk temperatures and their
shell is very small. respective film coefficients
• thermal resistance of the inside insulation,
Table 1.0 - Heat Transfer Analysis • shell side heat transfer coefficient as a
Heat Trans- Uni Fouled Clean
fer ts function of surface temperature (to account
Compart- 1st 2nd 1st 2nd for heat flux).
ment
Heat Duty K 20,138 33,54 21,483 36,648
W 1 The profiles generated by the Borsig FEA study
Heat Ex- M2 71.4 209.3 71.4 209.3
change Sur-
(Figures 6 and 7) indicated that the maximum tem-
face area peratures were approximately 620 ºC for the larger
o
Mean Tem- C 531.4 267.5 524.7 229.7
perature Dif-
tubes and 580 ºC for the smaller tubes. These val-
ference ues are significantly higher than the design tem-
Overall HT W/ 530.8 599.1 573.4 762.1
coefficient m2
perature of 380 ºC listed on the original drawing by
K Steinmueller. The larger tube temperatures were
Heat Flux at W/ 333.1/314.9 275.3 360.6/339.2 346
Tube sheet mK
also close to the maximum allowable temperature
Metal Tem- of 648.8 ºC for 1Cr-1/2Mo material.
peratures
o
Tube inlet, C 428.9/416/5 404.2 437.3/423.4 423.2
inner wall * * Table 2.0 - Natural Circulation Ratio
o
Tube inlet, C 397.2/390.1 381.8 402.7/394.7 394.6
mean wall * * Mass flows Units Fouled Clean
o
Temperature C 56.4/49* 31 60/51.8* 33.6
difference
Compartment 1st 2nd 1st 2nd
* Where two values are listed, the first one refers to lar-
ger tubes and the second one to the small tubes. Steam mass
Kg/h 63,309 105,448 67,540 115,216
flow
Natural Circulation Ratio Total mass
Kg/h 353,264 877,328 354,585 882,555
The circulation ratio is defined as the total circula- flow
tion mass flow divided by the steam mass flow.
Circulation
The calculated circulation ratio for the first com- - 5.58 8.32 5.25 7.66
ratio (average)
partment is 5.58 and for second compartment is
Flow Veloci-
8.32, for fouled conditions. (Table 2.0). Both these ties
values are lower than the desirable value of 10 for
Downcomers M/s 1.84 0.98 1.85 0.99
the operating pressure range of the boiler. The flow
velocities in the down comers and risers are all in Riser
M/s
2.67/1.5
4.52*
2.77/1.5
4.73*
8”/10”/14” 2/2.51 8/2.61
the usual range.
* The second compartment has only one set of risers
Temperature distribution
Due to the complicated three-dimensional structure Conclusions of Thermal Evaluation
and the uneven temperature distribution, Borsig The conclusion of the study was that the allowable
used the FEA (Finite Element Analysis) program strength of the material for the calculated tempera-
ANSYS to perform the modeling and calculate the ture is very low and it is more than doubtful that
heat transfer and metal temperatures of the first
AMMONIA TECHNICAL MANUAL 230 2008
the tube sheet can withstand the load due to pres- ing reviews carried out concluded that this equip-
sure and temperature. Coupled with this, the lower ment will eventually fail. Also, there was a strong
circulation ratio in both compartments does not opinion that the boiler was in the impending failure
foresee a continued operation of the boiler without mode.
encountering a serious failure in the near future. To
understand more on this and the efficacy of repairs Alternatives:
carried out in 2003 and 2005 a FEA stress analysis An alternative considered was to replace the first
was recommended. compartment only, as was done in 1995. While this
option may have addressed the metallurgical con-
cerns of the tubesheet, it would not have resolved
Finite Element Analysis of Tube
Sheet Load Stresses Table 3.0 – Results of Stress Analysis
Case - No Refrac- Case – Refractory
Surface tempera-
Units tory on outlet Tube installed on outlet
Domain Engineering Inc., of Tulsa, OK, were con- tures
sheet Tube sheet
tracted to carry out the thermal and physical stress Inlet Tube sheet – o
C 411-424 411-424
gas side
analysis on the first compartment. They adopted a Inlet Tube sheet – o
C 367 367
FEA approach. The results of this study revealed water side
Outlet Tube sheet
the same conclusion for the tubesheet problem as – gas side
o
C 620 466
the thermal evaluation. The tube sheet is subjected Outlet Tube sheet – o
C 519 422
water side
to above design temperatures and the cracking is Bare Tube – gas o
C 376 375
the result of thermally induced stress. The study side
Bare Tube – water
also revealed two critical points. First the Von side
o
C 423 422
Mises stresses at the tube sheet surface were re- Inlet Tube sheet –
combined loading
duced by 48% with the addition of refractory on the stress (mechanical
Psi 38,022 42,967
outlet tubesheet, but the surface stresses were still and thermal)
TWICE the yield point at the ligament of the outlet Outlet Tube sheet –
combined loading
Psi 108,246 56,572
tube sheet. Second, the tubesheet surface tempera- stress (mechanical
and thermal)
tures were higher than the design temperature. (Ta-
ble 3.0 and Figures 8, 9 and 10) the operational condition of low water circulation
on the tube sheet face. In other words, it could re-
It is important to note that ASME Section VIII Di- sult in a repeat of the thermally induced cracks.
vision 2 code does not allow surface temperatures This option was not recommended since it did not
to exceed 482 ºC for 1 Cr – ½ Mo. resolve the root cause.
The conclusion from this study was replacement of The accepted alternative was to replace the steam
the boiler with a design that addressed the issues drum, risers, downcomers and the boiler as a com-
revealed in the thermal evaluation and the stress plete unit in the 2007 turnaround.
analysis study.
Due to long delivery times involved in the manu-
facture and transport of waste heat boilers, discus-
Replacement of the boiler sions were initiated in early 2006 with UHDE.
Since the original manufacturer of the boiler,
Justification Steinmueller, was out of business UHDE recom-
Following the discovery of cracks on the tubesheet mended other suppliers with which they experience
face, and that the cracks reappear even though they for other ammonia plant projects.
are ground out and refractory added, the engineer-
2008 231 AMMONIA TECHNICAL MANUAL
Design
The redesign covered the replacement of the C110 The central gas bypass section (a single large tube
boiler, the Steam drum F102 and the interconnect- 457 mm OD, 81 mm thick) is used for exit gas
ing piping (down comers and risers). The design temperature control through the operation of three
parameters were explained in detail to the potential dampers at the outlet end in a mixing chamber. The
vendors. That is, the new boiler was to be designed upper bearing, including the stuffing box and the
for the current operating conditions, while taking central bypass flap, are cooled by 40 barg steam,
into consideration the need for desired circulation which exits via a hole in the damper shaft into the
rates (>10) and high heat flux to reflect the actual process gas chamber. The effective tube length be-
operation of the plant at 115% rate. tween the tube sheets is 6900 mm. The shell OD is
1883 mm and 74 mm wall thickness. The shell sec-
Actual plant operating data at maximum produc- tion exposed to process gas is refractory lined. The
tion rate (1350 MTPD) was supplied to UHDE and inlet tube sheet is refractory lined, and the outlet
the potential vendors. Additional constraints were tube sheet is not.
that the new boiler had to fit on the existing foun-
dation, mate to the existing plant piping, and two The new F102 steam drum weighs 58 tons (empty),
large vessels, the secondary reformer upstream of has a volume of 33 m3, and has an OD of 2400
C1110 and the steam super heater downstream of mm and 93 mm wall thickness. The drum has
C110, could not be moved. steam/water separation chamber fitted with set of
cyclone separators and demister elements.
Based on the new design philosophy, the potential
vendors that UHDE recommended proposed a sin- Two end spool pieces with enough room for field
gle compartment design. Following the engineer- adjustments were supplied as a part of the package.
ing, design review, and quotations from various These spool pieces joined the C110 boiler with the
vendors, plus the tight delivery schedule, UHDE upstream secondary reformer and the downstream
selected Borsig GmbH of Germany to supply the steam super heater.
equipment. See Figure 11 for the Borsig proposal.
A Hazard and Operability study was carried out at
New C110 and F102 the site for the proposed equipment. Minor
The new C110 weighs 72 tons (empty), has a sur- changes, mostly to accommodate operational flexi-
face area of 412.8 m2, 520 tubes of 38 mm OD and bility, were suggested in some piping and instru-
6.3 mm thick, and 315 mm long ferrules. The thin mentation. These changes were included in the
and flexible tube sheets have a thickness of 35 mm overall scope of the project.
and are connected to the shell via special shaped
forged rings. FEA was used in the design of these Procurement and Shipping
thin flexible membrane tubesheets considering all The purchase order was placed with UHDE on
thermal and internal pressure loads – primary and June 05 2006, with delivery scheduled for FOB
secondary membrane stress, bending and peak site, Aug 14, 2007. This included 13 months manu-
stress on the membrane area and around the bundle facturing and about 6 weeks for transportation.
tubes.
The equipment came in several shipments. The
The tube to tubesheet connection is a full penetra- boiler and steam drum were shipped separately. All
tion weld eliminating any crevice between the tube the interconnected piping, risers and downcomers
and tubesheet. This weld is exposed to waterside also came in a separate box. The welding consum-
cooling. ables required for critical assembly components
AMMONIA TECHNICAL MANUAL 232 2008
were supplied by BORSIG. Due to weight and During the turnaround the existing boiler and drum
transportation considerations, refractory was not were removed and new equipment installed. It took
installed at the shop. Karrena, the refractory de- 6 days to cut the piping, separate the old C110 and
signer and contractor contracted by Borsig, shipped F102, and lift them out of place. The foundation
the bricks and mortar separately. was cleaned to install the PTFE sliding supports
and the main field inlet and outlet joints for C110
Onsite Assembly & Installation were prepared.
The equipment arrived at Hamilton, ON port on
Aug 4th and was delivered to site on Aug 10, 2007, A 500-ton crane was employed for the main lift of
four days before the scheduled date. the assembled C110-F102 structure. The combined
weight of boiler and steam drum including refrac-
Onsite assembly and installation were not included tory is about 160 tons. The actual lift itself took
in the BORSIG’s scope, although they supplied de- only about 2 hrs (Figure 14). Other than a few mi-
tailed instructions– installation sequence, specifica- nor during the placement of the boiler on the foun-
tions of weld gaps & PWHT details, radiography dation shims the installation went smoothly.
and hydrotest conditions. These activities along
with commissioning were carried out by Terra with Some extra hours of welding time was required
local contractors. After reviewing the site layout, during the welding of the liner in the inlet spool
the local contractor decided to build a temporary piece (secondary reformer to C110 boiler). The
foundation base and frame structure, in close prox- narrow enclosed space made this weld more diffi-
imity to the existing C110 boiler. The foundation cult than expected.
and frame enabled the assembly and connection of
C110 and F102. (Figures 12 and 13). The mechanical installation was completed by Nov
07, 2007, two days beyond the scheduled turn-
The alignment, assembly and welding of risers around period. It took a total of 12 days to install
went smoothly, however two of the eight down the new equipment and connect all the piping.
comers did not align properly. Their alignment was
off by more than 50mm. After consultation with Commissioning and Start Up
metallurgical experts and BORSIG, the two pipe Chemical cleaning and boil out of the system (util-
bends were heated locally (wrapping by heat pads) izing TSP, a dispersant and a surfactant, supplied
to 510 ºC and force applied to bend them to fit. The by NALCO) was included as a part of the pre-start
affected bend was stress relieved along with the up activity, which took 14 hours. Following this
connecting welds, following the recommended boil out the steam drum internals (demisters) were
PWHT. installed, which required vessel entry and some ex-
tra associated time of 10 hrs. Curing of the refrac-
After the assembly was completed, the refractory tory in the inlet and outlet spools was the final ac-
was installed and cured onsite. The majority of the tivity, which took 56 hours.
insulation work was also completed during this
pre-turnaround activity. Ammonia production was resumed on Nov 15,
2007. The original estimated time for the turn-
All the above activities were completed by Oct 15, around was set over a year ago at 21 days for am-
2007, in 65 days. monia plant product stop to product start. In the
end, the sheer size and complexity of the C110 re-
Turnaround Activities (Oct 20-Nov 05, 2007) placement caused the ammonia plant to be down an
extra 6 days.
2008 233 AMMONIA TECHNICAL MANUAL
Conclusions
Figure 3 - Microstructure and crack in the liga-
The plant has returned to full normal operation ment
since installation of C110-F102. The maximum
production achieved is around 1348 MTPD.
At lower plant rates (<1325 MTPD), the bypass
damper is required to be kept fully open to meet
the downstream super heater requirements. While
this low rate operational issue is currently being re-
viewed with UHDE and BORISG, the overall per-
formance of the boiler is satisfactory.
The entire boiler replacement project – review of
failure, design, procurement, installation and all the Figure 4 – Microstructure at Crack Tip
timely assistance provided by UHDE on this job
has been excellent and professional.
Figures
Figure 1 - Old C110 twin compartment boiler and
F102 Steam drum
Figure 2 – As removed Ligament Sample and
crack
AMMONIA TECHNICAL MANUAL 234 2008
Figure 5 – Map of Ligament cracks found 2005 in-
spection Figure 7 - Borsig FEA results Water side and Gas
side Thermal Profile
Figure 6 - Thermal Boundary Conditions of the
Borsig FEA model
2008 235 AMMONIA TECHNICAL MANUAL
Figure 8 - Domain Engineering FEA Model Figure 10 - Domain Engineering FEA Results –
Von Mises Stress Contours on Outlet Tubesheet
Figure 9 - Domain Engineering FEA Results –
Overall Von Mises Stress Contours
Figure 11 - New C110 boiler with F102 drum –
UHDE/Borsig Proposal
AMMONIA TECHNICAL MANUAL 236 2008
Figure 12 – The Assembly
Figure 13 – The misalignment
Figure 14 – The Big Lift
2008 237 AMMONIA TECHNICAL MANUAL
AMMONIA TECHNICAL MANUAL 238 2008
