You are on page 1of 4


Why do waste heat

boilers fail?
D. Martens and M. Porter of Porter McGuffie, Inc. and L. Stern of Stern Treating & Sulphur
Recovery Consulting, Inc. examine the main causes of waste heat boiler failures in Claus
sulphur recovery units and discuss what lessons can be learned from past WHB failures.

hy do waste heat boilers in sul- bottlenecked for needed increased sulphur The ability to accurately simulate and
phur plants fail? We can look to capacity, it was determined that increas- monitor the furnace core gas temperatures
the auto racing industry to find the ing the oxygen content to the burner would is often a problem that can result in exces-
answer. Engine size in Formula 1 race cars allow increased capacity with the same sive operating temperatures. It is common to
has continually reduced over the years as furnace and boiler. Improved burner tech- use both pyrometers and thermocouples to
engineers have found ways to make the cars nology allowed even greater increases in provide the best possible temperature meas-
go faster and faster with a given engine size. oxygen enrichment and, consequently, more urements. However, during shutdown inspec-
On its super speedways, NASCAR places sulphur was produced by the same plant. tions the refractory and ferrule materials
restrictor plates below the carburettor to limit This added oxygen technology has become will often indicate that the operational tem-
engine output. Similar efforts take place in a standard offering for new SRU units. peratures were actually above 1,650C. At
almost all types of racing. Why? Because However, the higher temperatures encoun- the same time, the process control system
the limiting factor is the ability of the driver tered with oxygen enriched operations has temperature measurement historical data
to react quickly enough to safely control the resulted in cases where we are bumping up often does not indicate temperatures above
car. Waste heat boilers fail because we try to against the sulphur plants speed limit. 1,540C and sometimes not above 1,425C.
operate them at levels beyond which we can The continuous operating temperature One strategy for high temperature shut-
adequately design and safely control them to limit for modern, well-designed and installed down protection is to use thermocouples
provide reliable operation. thermal protection systems (both refrac- in the furnace with the shutdown set at
Typically waste heat boilers (WHBs) fail tory and ferrule systems) is approximately 1,540C for five minutes. The thermocou-
due to three factors: excessive tempera- 1,540C. A well designed system (please ples, which measure the hot face of the
ture, excessive mass flux rate and exces- note the emphasis on the word system) refractory in the furnace, indicate a lower
sive water-side fouling1. Similar to the racing can operate successfully for brief periods temperature than the core gas tempera-
industry, the sulphur industry has increased of time at temperatures above 1,540C, ture entering the WHB. This differential
temperatures and mass flux (process flow) but only by sacrificing reliability. If a 3-4 year temperature is typically 110C or more.
rates to obtain greater unit capacity. How- life is desired for the thermal protection It should be noted that pyrometer tem-
ever, this push has exceeded reasonable systems, the design operating temperature perature measurements may be highly
bounds, to the extent that reliability of a should be somewhat less than 1,540C. A influenced by the process gas analysis
unit can be and has been compromised. To 100C buffer between the normal and maxi- change (such as occurs with oxygen enrich-
maintain acceptable discharge environmen- mum operating temperature is about the ment). Pyrometers set for air only will
tal criteria, often the sulphur recovery units minimum that can be used to protect the normally read low by as much as several
(SRUs) must operate with significant vari- thermal protection system from negative hundred degrees. As a caution, we would
ance in acid gas flow rates variances that (i.e., >1,540C) conditions. With this small suggest that there are inherent inaccu-
are not controllable by the SRU operators. margin for error, it takes a well-designed racies of all temperature measurement
Water-side fouling is potentially affected by and calibrated temperature measurement devices due to installation, location, cali-
these same parameters and becomes a sig- system, a process control system, and bration, interference, maintenance, etc.,
nificant factor for reduced reliability. vigilant operators to control the temperature issues. Therefore, any specific plant read-
and avoid reducing the reliability of these ing can be off by as much several hundred
Excessive temperature thermal protection systems. Increased tem- degrees C. This discrepancy is normally
peratures affect the reliability of the WHB lower than the actual temperature.
Reaction furnaces (e.g., in refinery service) by increasing the temperature of the metal Hot standby operations have the poten-
were originally operated with a combination parts, which can increase corrosion. It also tial to produce excessive temperatures;
of amine acid gas and sour water acid gas increases the heat flux through the tubes, therefore, tempering of the sub-stoichio-
burning in a sub-stoichiometric combus- which can result in a Leidenfrost steam blan- metric hydrocarbon or hydrogen combus-
tion environment using atmospheric air keting condition that usually occurs at the tion is necessary1. Short term hot standby
as the oxygen source. As plants were de- end of the ferrules2,3. operation of much less than an hour with-

Sulphur 354 | September - October 2014 1


out adequate tempering can significantly potential for the tube sheet and tube welds. Mass flux rates
impact the reliability of the refractory and Significant changes of the WHB steam pres-
ferrule systems. Temperatures in excess sure can potentially over-stress the tubes in Mass flux refers to the mass flow rate
of 1,650C can occur without adequate compression or tension depending on the through the tube set. It is the total mass
tempering. A stoichiometric high tempera- developed temperature differential between flow through the unit divided by the
ture can result in significant ferrule system the relatively thin tube wall versus the much open area of all the tubes. As with tem-
failure and/or reduced reliability. thicker shell. For example, the lowering of peratures, common practice has been to
Sudden changes in process temperature the steam pressure during a unit shutdown increase the mass flux rate as a means
can be detrimental to the WHB tube sheet has been identified as overstressing and of achieving increased capacity without
protection system, resulting in a loss of cracking the tube-to-tube sheet welds that increasing the tube area. It is common
system reliability and increased corrosion were already thinned due to corrosion. to have design and operating mass flux
rates today that are twice the design and
Fig 1: Severely corroded and non-corroded tube sheets operating mass fluxes from two to three
decades ago. Excessive mass flux will
result in a significantly increased pres-
sure drop at the entrance to the ferrules.
This pressure drop can be expected to
increase the gas bypass of an individual
ferrule, leading to significant Increases
in the metal temperature and resulting
It is important to note here that the ther-
mal protection of the tubes and tube sheet
is almost entirely governed by the paper/
board between the ferrule and the metal.
Fig 2: Proposed Claus SRU service sulphidation corrosion curve for carbon steel The typical ceramic ferrules are not good
thermal insulators. Their primary function
is to protect the paper/board from the gas
10 1 mil/year
flow. The temperature drop is almost all
2 mils/year across the paper.
3 mils/year This higher temperature results in
5 mils/year accelerated sulphidation corrosion of the
tube sheet and tube welds. Thus, the
10 mils/year
increased mass flux rate adversely affects
15 mils/year the WHB tube sheet protection systems
20 mils/year reliability. This is true for both removable
25 mils/year ferrule and non-removable ferrule systems.
Figure 1 illustrates a tube sheet on the
30 mils/year
left where severe corrosion has occurred
40 mils/year
after only two years of service. The welds
50 mils/year between the tubes and tube sheet have
mol-% H2S

0.1 almost been corroded away and some are

leaking. After five years in the same ser-
vice, the replacement WHB tube sheet on
the right looks almost pristine5,6. The differ-
ence is that the tube sheet on the left has
seen much higher temperatures in opera-
tion than the one on the right although the
0.01 normal process operating temperature was
essentially the same. However, the mass
flux was reduced significantly.
The rate of corrosion in a waste heat
no corrosion boiler is a function of the metal tempera-
ture. ASM7 has published a series of curves
by Couper-Gorman that relate the rate of
0.001 corrosion in carbon and stainless steels for
500 600 700 800 900 1,000 1,100 several refinery environments. It has been
the authors experience that these Couper-
temperature, F Gorman curves are somewhat conservative
for the SRU environment. For a number of

2 Sulphur 354 | September - October 2014


Fig 3: Tube partial collapse at end Process-side tube fouling unit has been operated at high tempera-
of ferrule due to steam tures. Corrosion characteristics observed
blanketing conditions Fouling on the process side often occurs during inspection may indicate a local prob-
first in the hot pass tubes which reduces lem or a general problem that may have
the heat transfer. This duty is passed on differing root causes. For example, inspec-
to the next pass or exchanger often with tion should be conducted after a thorough
increased potential of sulphidation of removal of all scale including the I.D. of
unprotected channels and tube sheets at the tubes within the ferrule length to assist
the cold end of the first pass. The fouling in determining the apparent corrosion rate
of the hot pass tubes has been observed and support the remaining life and root
years, therefore, a modified Couper-Gor- to cause failure of the downstream pass cause analysis.
man curve has been used that correlates or exchanger. Review of historical operating data is
better to the actual experience in SRUs. one of the most effective tools for input
Figure 2 illustrates this modified Couper- Learning from WHB failures into root cause analysis of WHB failures.
Groman curve for SRU service. Such reviews often confirm operating con-
The increased mass flux rate also When a WHB does not provide reasonable ditions that are distinctly different from
raises the heat flux through the tubes, reliability, it is imperative to understand those the plant personnel understood was
which can result in a Leidenfrost steam the root cause(s) of the failure. The princi- occurring. For example:
blanketing or tube dry-out condition that ple inputs for evaluation are the inspection Was the unit started up and shut down
usually occurs at the end of the ferrules. observations, the actual operating condi- appropriately and per standard operat-
When this happens, the tube metal tem- tions and procedures, installed materials ing procedures?
perature can go up by 300C or more in a and installation procedures. What were the principle operating
matter of minutes. This can result in short Inspection activities are critical to infor- parameters of temperature and mass
term creep, as illustrated in Fig. 3. This mation gathering for reliability considera- flux during normal and abnormal operat-
condition occurs more frequently in kettle tions and failure analysis. All too often, the ing conditions?
type boilers but can occur in those with a inspection activities do not gather critical Comparison of WHB steam production
separate steam drum. information necessary for further review to the mass flux and temperature to
There is no universally agreed upon and root cause analysis During an un- confirm data is reasonable.
limit for the mass flux rate to achieve relia- scheduled or scheduled unit shutdown, the Reported process temperatures as
bility in a waste heat boiler. However, expe- unit inspection is often conducted quickly compared to inspection observations.
rience has shown that boilers operating at to determine the scope of possible nec-
a mass flux of less than 12.2 kg/m2-sec essary repairs before returning to service. The review should include the total time of
have runs exceeding eight years without This cursory examination does not provide operation since the last thorough inspec-
failure. Units operating at 25 kg/m2-sec or sufficient input for a root cause analysis. tion, or a minimum of two years of opera-
above have often exhibited failures within As already described, if inspection indi- tion. This type of review is necessary to
two years or less. Based on information cates ferrule and refractory glazing, the capture the abnormal operating conditions
collected, the closer the unit can be oper-
ated at a mass flux of 12.2 kg/m2-sec, Fig 4: Corrosion at end of ferrule due to water side fouling
the greater the reliability of the tube sheet
protection system to maintain acceptable
metal temperatures and less potential for
Leidenfrost conditions.

Water-side tube fouling

The increase in both operating tempera-
ture and mass flux increase the tube heat
flux, which can increase the potential for
water-side fouling with the same quality
boiler feed water. The boiler water quality
can also become compromised by inad-
equate blow down and chemical treatment
additions for increased steaming rate with
the same tube surface area6. The tube OD
fouling reduces the heat flux on the out-
side of the tube by insulating it from the
water, resulting in higher tube tempera-
tures. The higher temperatures, in turn,
promote sulphidation in the tube as illus-
trated in Fig. 4.

Sulphur 354 | September - October 2014 3


that an SRU unit has been subjected to, able metal temperatures can be studied References
such as a short duration high mass flux with these engineering tools. The effect 1. Brimstone 2011 Top five fundamental integrity
required due to facility operational condi- of the burner flame pattern and resulting issues for SRU wasteheat exchanger (boiler)
tions, or hydrocarbon carry-over in the acid refractory and ferrule temperatures can panel discussion and technical papers:
gas feed stream. The ability of the opera- also be studied with these engineering Mosher A.: Burner flame temperature dur-
tors and the process control systems to tools and help verify the true temperatures ing warm up and hot standby; Misale D.:
Ferrule design and installation for SRU
maintain the operational reliability of the reached. The CFD analysis2,3 can indicate
tubesheets; Stern L.: Boiler water level
SRU is often impacted by variations in the what physical and process parameters safety considerations & tube collapse; Mar-
feed stream for rates and multiple unit changes could be possible to improve the tens D. Tube and tube weld corrosion and
load sharing requirements. In addition to reliability of the WHB and can include both tube collapse; Lamar J.: SRU overpressure
the measurement accuracy of the instru- the process and water services of a WHB. in a waste heat boiler failure.
mentation, the response time of the pro- 2. Porter M.: A means of avoiding sulphur
recovery furnace fired tube boiler failures,
cess control systems and analyser feed New and replacement WHB Brimstone 2009.
back to the control system for these feed
variations is critical to achieving a reliable
specifications 3. Porter M.: A means of avoiding sulfur recov-
ery reaction furnace fired tube boiler failures,
WHB service result. There is no industry consensus document ASME PVP paper number 78073 (2009).
A thorough engineering analysis of that an owner and operator can refer to 4. Porter M.: Computational fluid dynamics
a WHB using state-of-the-art tools can when specifying the details necessary for investigation of a high temperature waste
heat exchanger tube sheet assembly, ASME
provide the insight as to root cause of a a reliable and robust WHB. Guidance can
paper number PVP 2005-71143 (2005).
failure mechanism. These analysis tools, be found in published technical papers 5. Martens D.: Robust SRU waste heat boiler
such as computational fluid dynamics and obtained from various subject matter design, Brimstone 2012
(CFD)8, can also assist in determining the experts5. When combined with your exist- 6. Combining CFD derived information and
actual metal temperatures and mass flux ing units reliability and operating param- thermodyamic analysis to investigate waste
operating conditions for a specific WHB to eter experience, this guidance information heat boiler characteristics, ASME paper
number PVP 2011-57625 (2011).
provide reliable service. The limitations of can provide the foundation for establishing
7. Metals Handbook Ninth Edition Volume 13
heat flux in the turbulent area at the end the physical and operational criteria and Corrosion Copyright 1987 ASM International
of the ferrule and the ability of the tube specifications for the detailed design of a 8. McGuffie S.: ASME PVP Conference Tutorial
sheet protection system to maintain suit- reliable and robust waste heat boiler. Use of CFD in design (2012).

Porter McGuffie, Inc.

Engineering Measurement & Analysis Services
544 Columbia Drive, Suite 19 Lawrence, Kansas 66049-2348 USA
phone:01.785.856.7575 fax:01.785.856.7577 web:

4 Sulphur 354 | September - October 2014