Case Study of The Use of API 581 On HK and HP Material Furnace Tubes
Case Study of The Use of API 581 On HK and HP Material Furnace Tubes
Case Study of The Use of API 581 On HK and HP Material Furnace Tubes
1 Introduction Reformer tubes are generally made from cast creep resistant aus-
tenitic steel HK grade 共25 Cr, 20 Ni, 0.4 C兲 or HP grade 共26 Cr, 35
Inspection to detect potential failures has been an important
Ni, 0.4 C兲. Although the furnace tubes are usually designed for a
activity in industry for more than 100 years. The traditional view
normal life of 100,000 h 共11.4 years兲, their actual service life,
of inspection is that it should reduce the probability of failure of however, varies from 30,000 to 180,000 h, which is 5–20 years
the components being inspected. Design codes and standards in depending on the service conditions and the quality of materials
this area are written to deal with structural integrity but they were 关1兴.
not as helpful for managing inspections.
Risk in the formal definition is the product of consequence and
frequency. The failure of pressure equipment can have several
undesirable consequences such as fires, toxic releases, major en- 2 Semi-Quantitative RBI Analysis Furnace Tubes
vironmental damage, and business interruption losses. Given this, Semi-quantitative RBI analysis here is done using the organized
it should be the intention that not only the frequency of failure but methodology developed by API as outlined in the API Base Re-
also the consequence of failures should be reduced by inspection/ source Document API 581. API RBI Appendix J considers the
maintenance activities. furnace tube failures because of creep or overheating. Materials
In the 1990s it became apparent that the lack of uncomplicated data provided in API RBI Appendix J are ferritic steel 共carbon
and suitable quantitative guidance for determining risk was a ma- steel and low alloy steels up to 12 Cr兲 and austenitic stainless steel
jor problem for any pressure system user trying to determine the 共types 304, 316, 321, and 347兲.
priorities of an inspection program. Guidance on inspection pro- In this case study reducing gas furnaces tubes have designed
gramming had to be practical and had to take account of damage based on API 530 procedures but the tubes are made of materials
mechanism, inspection techniques, inspection effectiveness, rate which are not considered in API 581, namely HK40 and HP. Pro-
of damage progression, and other technical factors. In addition an cess and design data are shown on Table 1.
approach to considering the consequences of a failure had to be
developed. 2.1 Likelihood Analysis. Likelihood analysis is the result
The American Petroleum Institute 共API兲 initiated the Risk of a number of factors but in semi-quantitative RBI only the tech-
Based Inspection 共RBI兲 Project in May 1993. The intention of API nical module subfactor 共TMSF兲 is used to determine the probabil-
RBI is to be a useable and understandable tool at a plant inspec- ity of failure. The TMSF identifies damage mechanisms and
tion level by simplification of complex models using a practical equipment age in current service to estimate damage rate. The
procedure standardized for petrochemical plants. The RBI analy- technical module subfactors can range as high as 1000 or more,
sis is an attempt to analyze risk and inspection effectiveness quali- while the other subfactors are relatively small 共⬍10兲 关2兴.
tatively and quantitatively and looks at the factors required to The key to understanding TMSF is found in paragraph 9.4.3 of
API 581. Here it becomes clear that if TMSF is less than 10 then
develop an inspection plan including both the likelihood and con-
there is very little damage occurs. Over 10 and the damage rate
sequence of failure. The technical modules in API 581, the re-
becomes more significant and there are increasing demands on
source document for RBI, include thinning, stress corrosion crack-
inspections and other preventative measures 共see Tables 9-15 to
ing, high temperature hydrogen attack, mechanical fatigue 共piping
9-17, of API 581兲. With TMSF above 500 there is a concern that
only兲, long-term creep, short-term over heating, brittle fracture,
lifetime may have been exhausted.
equipment linings, and external damage.
Likelihood analysis starts with calculating tube current diam-
This paper is an exploration of the difficulties with applying
eter and determining tube elastic temperature limit and elastic
API RBI to furnace tubes of gas generation plant. Note that be-
stress limit. If actual tube metal temperature 共TMT兲 is less than
cause API uses British units, these units are used in some of our
elastic temperature limit or the actual tube stress 共which is calcu-
tables. Because of the complexity of some of the equations in-
lated by considering operating pressure and tube diameter兲 is less
volved, conversion of all the units to Systeme International is not than the elastic stress limit, then long-term creep is not a concern
necessarily appropriate.
In this study we are examining some typical reformer tubes. TMT⬍T elas and S⬍S elas
The actual tube metal temperature, which was determined by
Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the PVP
direct measurement, is TMT⫽905°C. API 581 defines elastic
Division August 30, 2004; revision received September 7, 2004. Review conducted metal temperature limit or critical tube metal temperature as ‘‘a
by: S. Zamrik. temperature at which a tube would have 100,000 h design life
Journal of Pressure Vessel Technology Copyright © 2005 by ASME FEBRUARY 2005, Vol. 127 Õ 49
Larson–Miller parameter HK HP
lm 44.16 57
LMdelta 0.8 0.3
LMavg 46.02481–1.88469 ln S 65.05956 –2.17933 ln S
Long-term creep analysis HK HP
FFLT 共failure factor兲 1.00E⫺300 0
TMSFLT 共technical module subfactor兲 0.55 0.55
W g 共 sonic兲 ⫽C d A P 冋冉 冊 冉 冊
KM g c 2
RT 144 K⫹1 册
k⫹1/k⫺1 1/2
(10)
冉 冊 k/k⫺1
consequence.
k⫹1 Considering toxic consequences in this case, it was necessary to
P trans⫽ P a (9)
2 consider that hydrogen sulphide is very toxic by inhalation and
Table 4 Reducing gas composition and properties for radiant and convection section
may cause damaging effects to the nervous system. Reducing gas than just a comparative importance between different parts of the
includes 25 ppmv H2 S. The exposure limit value for H2 S is iden- same plant. Clearly equipment operating in the H or high risk area
tified from the material safety data sheet to be 10 ppm. From this could potentially be dangerous. If we look at the major hazard
it is found that hydrogen sulphide may cause toxic effects after regulations for Victoria Guidance Note 19 关8兴 also gives a similar
release. Release rate and duration was used to determine conse- graph in Appendix 1. Here, however, the upper reaches of the
quence area for this toxic gas. The results of consequence analysis graph are identified as not being acceptable ‘‘societal risks’’ in the
are shown in Table 7. state of Victoria. The results of the work based on all clauses of
API 581 are also shown on Table 8. The simplified method of
3 Risk Calculation Appendix A only is also shown.
API 581 provides a semi-quantitative method of assessing the
risk for each component. This analysis uses a matrix with fre-
quency on the vertical axis 共listed in categories 1–5 with 5 the
highest frequency event兲 and consequences on the horizontal axis 4 Inspection Intervals
共listed as A – E with E having the highest consequences. Parts of To plan inspection intervals API 581 recommends inspections
the matrix is indicated to have a high, medium high, medium, and intervals for furnace tubes based on TMSF less or greater than 10.
low risk, which is shown by the letters H, M H, M, L on risk If the current TMSF is less than 10, increment t i 共operating hours兲
matrix. This matrix is a reinterpretation of the Farmer curve first by 10,000 共⬃1 year兲 until a TMSF of 10 is reached. The number
published as a risk criterion in 1967 关7兴. of increments is the time to the next inspection, T insp 关2兴. In this
The risk matrix for this case for four different types of conse- case TMSF is less than ten by increasing operating hours the
quences is shown in Fig. 2. The result of this analysis is not really TMSF was not changed. Considering API RBI Table 9-16 the
discussed in enough detail in API 581 though the result has more inspection plan is shown in Table 9.
5 Discussion • using risk matrix which considers both consequence and like-
lihood to assign risk
5.1 How Simplified is the Method? API 581 is a simplified
method for furnace RBI analysis. The document is very useful but Some gaps are that materials such as HP and HK40 which are
has some gaps. used to manufacture furnace tubes are not modeled in API 581. A
Some good features are: number of activities such as searching the literature, using manu-
facturer’s data were used in this paper to generate the necessary
• a predefined set of hole sizes is used for the likelihood data. It is noticeable that RBI analysis in this way is time-
analysis consuming method and the data may not be accurate.
• a consequence model of flammable, toxic, environmental, API 581 states that general thinning mechanism H2 S/H2 for
and business interruption events is included. This is very high temperature should be considered to find out the effect on
helpful for such modeling is not easily available in open lit- wall thickness and accordingly on actual stress. This calculation
erature. The consequence model includes:
actually turns out to be quite complicated because it has to be
–a simple method for modeling the release
repeated several times during life. A computer program is avail-
–an approach for modeling releases of mixtures
able but we have not included these calculations in the current
–a simple procedure to determine the mass of fluid that is
released in the event paper.
–a method for determining the area that is effected by each 5.2 Inspections. Risk in API 581 is used to recommend in-
release type spection frequency based on both the consequences and likelihood
–standardized event trees are used of failure. API RBI introduces inspection intervals based on
–developing a simplified methodology for assessing the TMSF or likelihood analysis. The main factor in finding TMSF is
effectiveness of various types of mitigation systems mean Larson–Miller parameter, which cannot be accurately deter-
–providing simple guideline to calculate the inventory mined by using manufacturer’s data.
In the formulas for calculating TMSF for long term creep the
only factor which changes during time is ‘‘operating hours’’ which
Table 8 Inspection methods and intervals for furnace tubes is used for finding time to next inspection. In the case studied in
based on TMSF this paper the operating hours up to 100 years was calculated but
Inspection Intervals the TMSF was approximately constant at less than 1. As a result
for furnace tubes with medium risk highly effective inspection is
Inspection Nature of inspection recommended only at 20 years. This is a long period of time and
20 years Highly effective would be attractive to operators. However, the question remains
10 years Usually effective what exactly is being proposed.
5 years Fairly effective The highly effective inspection proposed in API 581 is defined
as UT thickness measurements of all tubes, and strapping at UT
Table 9 Summary of RBI results measurement locations and FMR at various locations. Questions
concerning these inspection regimes immediately present
Semi-quantitative results 共Appendix B兲 HK HP themselves.
FFLT 1.00E⫺300 0
TMSFLT 0.55 0.55 • FMR is not defined in API 581. FMR is used in a paper by
TMSFST 0.02 0.02 Valbuena et al. 关9兴 to mean field metallographic replication. This
Flammable consequence category 共equipment兲 C C is a method of detecting creep damage but in one of the authors
Flammable consequence category 共fatalities兲 D D 共J.W.H.P.兲 experience it is also highly unreliable unless used fre-
Toxic consequence category E E
Likelihood category 1 1 quently, say at one year interval and also only if a sample of
Simple qualitative results 共Appendix A兲 original, noncreep affected material is available.
Overall consequence category D D • UT thickness measurements are not normally a measure of
Likelihood category 2 2 creep damage, so the question of what is being tested for is not
clear.