Professional Documents
Culture Documents
ASMEThermal Shock Design Guidleines 2007
ASMEThermal Shock Design Guidleines 2007
net/publication/236170876
CITATIONS READS
9 6,901
1 author:
John W H Price
Monash University (Australia)
111 PUBLICATIONS 1,160 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by John W H Price on 30 May 2014.
Keywords: thermal shock, cracking, initiation, growth, arrest, design, fitness for purpose
Journal of Pressure Vessel Technology Copyright © 2007 by ASME FEBRUARY 2007, Vol. 129 / 125
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
mined using the techniques of ASME Section XI and are shown in
Fig. 4. Note that the combined stress intensities are modified by
plastic zone correction factors.
The key thing to note about Fig. 4 is that while the stress
intensity factor 共“SIF”兲 due to mechanical loads increases with
distance from quenched face, d, the SIF due to the thermal shock
cracking increases then decreases as d increases. The total SIF
thus peaks in the example given at about 2.5 mm and then falls.
The total SIF falls below 50 MPa· m1/2 at about 7 mm. This turns
out to be about the level that indicates a dramatic loss of crack
growth driver.
Results
Many cracks have now been grown on both the horizontal and
vertical rigs and have covered a wide range of conditions. Some
of the factors that have been varied include:
Fig. 1 Distribution of thermal shock and mechanical stress • Peak temperature 共T ° C兲. Temperatures from 240° C to
across a section.
400° C have been used.
• Constant stress applied 共P MPa兲. This is generally either 0
or 90 MPa stress.
The stress intensity factor produced by thermal shock stresses is • Quench time 共Q s兲. This is the time for which cold water is
dependent not only on the stresses indicated in Fig. 1, but also on sprayed on the specimen. From theoretical work, peak stress
the depth of the crack 共including any notch兲. This has been deter- at 3.5 mm was achieved after 7 s.
Fig. 2 „a… Cracking from external corners such as at penetrations. „b… Cracking at internal cor-
ners. „c… Examples of geometrical details affecting thermal shock cracking from power stations.
Left at the intersection of two drain lines. Right in an economizer inlet header.
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Fig. 3 Experiment design. Left, vertical furnace; right horizontal furnace. Full details of the experiment design are given in
Ref. †6‡
Discussion
General Observations. The main observations from this work
are as follows:
• All of the cracks that have been grown have arrested.
Fig. 4 Maximum stress intensity factor profiles during 7 s • Cracks on specimens with primary loadings have at first
shock from 370° C, with and without 90 MPa primary load
Fig. 5 Crack growth rate versus crack length „includes notch depth… for a number of cracks in the vertical
furnace experiments. T = maximum cycle temperature „°C…, P = primary mechanical load „MPa…, Q = quenching
time „time of water application… in seconds. Starting notch depth, ao, is 3.5 mm for all cases. The meaning of
“Best fit” and “Conservative” lines are described in relation to Eq. „3….
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Fig. 6 Data generated during horizontal rig experiments. Crack growth rate ver-
sus crack length for various notch depths, ao.
accelerated in growth rate, reached a plateau of growth, rate Growth Stages of the Cracks. To simplify the picture of what
and then arrested. is being observed, consider Fig. 8. This figure shows two sets of
• Cracks growing without primary loading have tended to data presented for two specimens from the vertical furnace which
grow at slower rates and arrest earlier. differ only in that one has a mechanical stress of P = 90 MPa
• Most cracks have arrested in less than 3 mm growth. A few applied.
cracks growing from the theoretically most severe notch Environmental effects can be seen by the knee in the fatigue
tested 共ao = 3.5 mm兲 have grown further to a maximum of growth curve first presented by Austen and Walker 关9兴 and also
6.2 mm. found in BS 7910 关10兴. Figure 9 presents some of our data on this
curve. Data with higher primary stresses apparently have an in-
Both the physical appearance of the cracks and the growth rate creased environmental effect.
data suggest there are some simple major features of the crack
growth. Initiation of Thermal Shock Cracking
1. Growth in the early stages up to about 1 mm exhibits the A Design Guideline. The circumstances that can lead to ther-
features of plastic tearing—raised and less corroded surfaces mal shock cracking should clearly be avoided during design. In
2. Growth with a large amount of corrosion. addition, existing plants should be assessed for the possible pres-
3. Growth region where the corrosion is limited and there is ence of thermal shock cracking. It is thus important to have a
sometimes branching of the crack 共Paris and arresting兲. guideline for determining the conditions necessary for the preven-
tion of thermal shock cracking. There is currently no clear guide-
line in the standards examined by the author.
For the initiation of thermal shock cracking, the following key
factors are required:
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
thermal shock, is Poisson’s ratio, and k f is the stress concentra-
tion factor due to geometric discontinuities. This has values in the
range 1–5.
With E, ␣, and being material constants, the only remaining
unknowns are ⌬Tm and k f . Methods for determining these param-
eters are given below. Note that in this work, the assumption of
temperature independence of E and ␣ is acceptable as long as the
product of the two does not vary by more than 10% over the
expected temperature range. This is the case for carbon steel be-
tween 100° C and 370° C.
Magnitude of Thermal Shock 共⌬Tm兲. The maximum magnitude
of the thermal shock 共⌬Tm兲 can be determined from design data
based on the worst possible thermal shock condition 共Tmax
− Tmin兲, where Tmax and Tmin correspond to the maximum and
minimum possible temperatures of the process fluid passing
Fig. 9 Smoothed experimental crack growth data plotted through the component during operation.
against a Gabetta et al. †11‡ model prediction, allowing for the Level 1 assumes instantaneous change of temperature at the
effects of environment and primary load. Experimental data for surface of the metal, that is, infinite heat transfer coefficient. This
dissolved oxygen „“DO”… =8 ppm plotted.
is normally approximately correct for water on clean steel where
heat transfer coefficients of around 10 kW/ m2 K are seen. It can
be shown that at this level there is little change in thermal shock
a. a geometrical feature 共stress concentration兲 and stresses as the heat transfer coefficient increases up to infinite. In
b. a number of thermal shock cycles of sufficient level 3 this conservatism could be removed.
magnitude.
Stress Concentration Effects 共k f 兲. To include the effects of ge-
Crack initiation can be complex, so the proposed guidelines ometry and other stress concentrators such as machining marks,
involve a three-level approach as is seen in other codes such as BS suitable values for k f need to be selected. In accordance with the
7910. Each of the levels from 1 to 3 will be intended to have a fatigue design rules of the ASME Boiler and Pressure Vessel
decreasing amount of conservatism. The choice of level to be used Code, Sec. VIII, Division 2 the maximum value for k f that is
in an analysis depends upon the input data available and the con- generated by a severe notch 共or corner兲 is 5.0. Similarly, for a flat
servatism required. The three levels are summarized below. surface free of machining marks, k f can be taken as 1.0. It is
Level 1 is a simplified go/no-go type analysis requiring the recommended that in a level 1 analysis a value of 5.0 be selected
absolute minimum of input data. This approach is the most con- for all discontinuities unless clear evidence of the lack of sharp
servative in its results as all stress concentrators are assumed to machining marks on the component can be made. Values of k f less
behave as worst-case notches. than 5.0 will need to be justified using methods similar to that
Level 2 is a slightly enhanced procedure allowing for the ef- used in a level 2 analysis.
fects of some geometries and the number of thermal shock cycles
expected in a component lifetime. Conservatism at this level is Crack Initiation. The generation of thermal shock cracks is in-
moderate. dicated if the half maximum theoretical stress amplitude 共0.5
Level 3 is an advanced analysis procedure that requires more Smax兲 exceeds the allowable design stress amplitude Sa. The al-
detailed input data. Conservatism is at its lowest level in this lowable design stress amplitude may be taken from the S-N de-
analysis. Improved analysis includes sign curves contained in Sec. VIII, Division 2 of the ASME Boiler
and Pressure Vessel Code 共Fig. 5-110.1 for the case of carbon and
• A detailed temperature profile across a specimen during a low alloy steels兲. This curve, as is well known, is generated from
shock to establish surface stress ranges and entirely different data using mechanical tests. For the purposes of
• Cyclic stress strain relationships at geometrical discontinui- this work the curve is adopted as a useful and convenient input,
ties to accurately allow for the effects of stress concentra- which we have found successful under the conditions described.
tors. Generation of a new curve specifically for thermal shock would be
• Improved fatigue initiation curves 共S-N curves兲 for the par- a large additional task that might be contemplated at level 3.
ticular material and stress state. Example. Using this method the permitted ⌬Tm can be calcu-
lated from transposed equation 共1兲
A procedure for each level is suggested in the following sec-
tions.
Level 1 Analysis. At this level, the maximum theoretical ther- 2共1 − 兲 Sa
⌬Tm = 共2兲
mal stress amplitude developed during a thermal shock 共Sm兲 is E␣ k f
compared to the allowable stress amplitude based on the design With Sa from ASME VIII Div. 2, Fig 5.110.1, typical carbon
considerations of the ASME Boiler and Pressure Vessel Code. The steel values, and a k f of 5, it is found that the permitted ⌬Tm for
effects of geometric discontinuities 共stress concentrators兲 are al- 1000 cycles is 123° C and for 100,000 cycles is 30° C. It is to be
lowed for using a stress concentration factor k f . noted that there are no effects from steady load or environmental
Maximum Theoretical Thermal Stress 共Smax兲. The maximum conditions, and this is to be expected for fatigue initiation. These
thermal stress amplitude generated during a thermal shock tran- results can be compared to the EPRI guidelines for economizer
sient is calculated by Eq. 共1兲: headers 关12兴, which limit ⌬Tm to 21° C. The guidelines presented
here will normally permit much higher levels of ⌬Tm. However,
E␣共⌬Tm兲 these results can still be restrictive since thermal shocks exceeding
Smax = kf 共1兲
2共1 − 兲 100° C or 200° C are often seen in power plants and this forces us
to consider all the conservatisms in the analysis by introducing a
Smax is the maximum thermal stress amplitude generated during a
level 2 and 3 analysis.
thermal shock. E is the elastic modulus, ␣ is the coefficient of
thermal expansion, ⌬Tm is the maximum possible magnitude of Level 2 Analysis. Possible improvements include
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
• Magnitude of thermal shock 共⌬Tm兲 determined from mea-
sured or calculated data at the component surface during a
thermal transient.
• Stress concentration effects 共k f 兲 determined theoretically or
experimentally.
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
1 1 − R da Conclusions
⌬K2c = yE 共5兲
␣ASCR 1 + R dt This work reported here has demonstrated the factors that cause
and drive thermal shock cracks.
␣ASCR is a constant proposed by Gabetta et al. that relates
an “active surface creation rate” 共ACSR兲 to the rate of
change of crack mouth opening distance during a load
cycle, R is Kmin / Kmax, is the cycle time, 7 s in the Design and Operation: Asme Boiler and Pressure Vessel
current case, and da / dt is a crack growth rate with time. Code and EPRI Guidelines. A key interest area is design guide-
Currently the Monash experiments cannot distinguish lines for vessels subject to thermal shock. The guidelines we are
all these factors, but there is enough data to provide the investigating will provide the basis for design and operation of
following interpretation for the conditions. pressure vessels with sharp geometrical features, which may be
damaged by thermal shock. There is little information about this
1−R case currently in the ASME Boiler and Pressure Vessel Code.
⌬K2c = 1.45 ⫻ 103 共6兲
1+R EPRI guidelines 关12兴 for assessing the initiation of thermal
For the current case the criterion for corrosion domi- shock cracking are highly conservative. Using the simplest level
nated growth could be further simplified to the form: of our guidelines 共level 1兲, typical carbon steels with a sharp notch
are permitted; ⌬Tm for 1000 cycles is 123° C and for
725 100,000 cycles is 30° C. If we use level 3 analysis and remove
⌬K ⬎ 共7兲 notches from the component being considered, then the ⌬Tm per-
Kmean
mitted would rise probably by a factor of at least 3, that is, to
In this form it is readily seen that if Kmean is low or 90° C for 100,000 cycles. These results can be compared to the
zero, then no corrosion dominated growth will occur. Us- EPRI guidelines for economizer headers, which limit ⌬Tm to
ing Fig. 4 it can be seen that the condition will only be 21° C, which is a very low level, virtually unachievable in most
satisfied for a range of crack growth in the region of a power stations.
= 5 – 10 mm in these experiments. A key result of this work is that the features such as the sharp
b. Crack growth rates in corrosion-dominated region. Once corners illustrated in Fig. 2 should be avoided in design. Such
corrosion dominated growth occurs, the growth rate is corners are not stress concentrators in the way normally consid-
determined by time of opening and the primary stress. ered in design since they may not be highly stressed in normal
The growth rate does not appear to be related to ⌬K 关15兴. operation. However, in thermal shock they start to behave like
The growth rate in the corrosion dominated region for the sharp notches. Such sharp corners frequently are used simply be-
current experiments, which have a cycle time of about cause that is the cheapest construction method. However, it would
15 min is given by be appropriate for the ASME Boiler and Pressure Vessel Code to
dissuade the use of the features where thermal shocks could occur
da and corners 共such as shown in Figs. 2共a兲 and 2共b兲兲 are found on
= 7 ⫻ 10−7 共8兲
dN many figures in the code.
Growth Guidelines for Thermal Shock Cracking. The dis-
3. Paris law type growth region. This can also be termed the
covery of thermal shock cracking often by opening up a neglected
LEFM growth region. As the crack grows the stresses fall and
part of a plant and observing it visually can be very disconcerting
corrosion 共if it occurs兲 is no longer the dominant growth mecha-
for the operators. Often the immediate response is to carry out
nism since the diffusion lengths are too long. In this region the
growth rate is found to fit the “Paris law” equation 关11兴. expensive repairs or replacements. These could, in fact, be
pointless.
It has long been known that many thermal shock driven cracks
da
= C共⌬K兲m 共9兲 arrest at fairly shallow depths and only a few eventually leak. The
dN experimental work presented here starts to indicate the growth
where C and m are constants that are dependent on the environ- mechanisms involved in thermal shock crack growth and presents
ment, the mean stress, and the stress range. the possibility of fitness for purpose assessments of thermal shock
This equation has been examined and found to fit the data pro- cracks when they are found in-service.
viding some care is taken in defining the stresses involved in ⌬K. This work has identified that there are four mechanisms to be
The values of C and m can be found from testing. considered in the growth of thermal shock cracking.
For the data collected in this study, the “freely corroding ma- 1. Growth due to stress exceeding the tensile strength of the
rine” crack growth equations in BS 7910 关10兴 tend to be ad- material—a high strain fatigue 共“HSF”兲 region.
equately conservative. Alternatively, a curve fitting exercise to the 2. Growth due to corrosion dominated growth mechanisms
data produced the equation. where the growth rate is constant and independent of ⌬K.
Corrosion occurs while the crack is open and not very deep.
da The proposal is that this occurs when the applied range of
= 5.28 ⫻ 10−16共⌬K兲5.89 共10兲
dN stress intensity factor, ⌬K exceeds some critical value ⌬Kc.
4. Final failure. Because this study is dealing with carbon steels The value of ⌬Kc depends on a number of factors not fully
in fairly warm conditions below the creep level, final failure if a investigated in this study such as time of crack opening and
crack keeps growing will be by normal plastic collapse normally oxygen content of water.
detected by leaking. This can happen when mechanical stresses 3. Once the stresses start falling and corrosion rate falls, Paris
are high and so must be considered in an assessment guideline. growth occurs 共affected by environment兲. Thermal shock
The limits that would seem to be appropriate are covered by the cracking is likely to arrest in this region if there are no high
failure assessment diagram 共FAD兲 of BS 7910. There has been no primary applied stresses.
through thickness failure case in the Monash tests. However, it is 4. Final failure occurs only if primary stresses are high enough.
known from industrial case studies that leaks are observed in This occurs by plastic collapse of the remaining ligament in
plants when high stresses are present 共see Ref. 关1兴兲. carbon steels.
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Acknowledgment 关7兴 Vitale, E., and Beghini, M., 1991, “Thermal Shock Fracture Experiments on
Large Size Plates of A533-B Steel,” Int. J. Pressure Vessels Piping, 46, pp.
This work has been conducted with the assistance of an Aus- 289–338.
tralian Research Council grant with contributions from HRL Tech- 关8兴 Marsh, D. J., 1981, “A Thermal Shock Fatigue Study of Type 304 and 316
nology Ltd., Optima Energy, Western Power, Hazelwood Power, Stainless Steels,” Fatigue Fract. Eng. Mater. Struct., 4, pp. 179–195.
关9兴 Austen, I. M., and Walker, F. F., 1977, “Quantitative Understanding of the
Loy Yang Power, Edison Mission, Pacific Power, and EPRI. Effects of Mechanical and Environmental Behaviour on Corrosion Fatigue
Crack Growth Behaviour,” The Influence of Environment on Fatigue, I Mech
References Conf, London, pp. 1–10.
关10兴 British Standards, 2000, BS 7910:1999—Guide on Methods for Assessing the
关1兴 Alexander, C., Frey, J., and Shin, S., 2004, “Evaluation of the failure in the Acceptability of Flaws in Fusion Welded Structures, BSI, London.
Texas Genco W. A. Parish Unit #8 Cold Reheat Line,” 4th Int Conf., Advances 关11兴 Gabetta, G., Rinaldi, C., and Pozzi, D., 1990, “A Model for Environmentally
in Material Technology for Fossil Power Plants, Hilton Head, SC, Oct. 26–28,
Assisted Crack Growth Rate,” Environmentally Assisted Cracking: Science
EPRI, Palo Alto.
and Engineering, ASTM STP 1049, American Society for Testing and Mate-
关2兴 American Society of Mechanical Engineers, 2004, “Boiler and Pressure Vessel
rials, Philadelphia, p. 266.
Code,” ASME, New York. 2004 edition is used unless other wise noted.
关3兴 Dooley, R. B., and McNaughton, W. P., 1996, Boiler Tube Failures: Theory 关12兴 Stevenson, G. G., 1989, Guidelines for the Prevention of Economiser Inlet
and Practice, EPRI, Palo Alto, CA. Header Cracking in Fossil Boilers, GS-5949, EPRI, Palo Alto, CA.
关4兴 Ng, H. W., and Lee, C. K., 1997, “Remaining Life of a Vessel Containing an 关13兴 Price, J. W. H., and Kerezsi, B. B., 2004, “Potential Guidelines for Design and
Internal Corner Crack Under Repeated Thermal Shock,” Proc. Inst. Mech. Fitness for Purpose for Carbon Steel Components Subject to Repeated Ther-
Eng., Part E, 211, pp. 215–219. mal Shock,” Int. J. Pressure Vessels Piping, 81共2兲, pp. 173–180.
关5兴 Yagawa, G., and Ishihara, K., 1989, “Cleavage and Ductile Thermal Shock 关14兴 Skelton, R. P., 1982, “Growth of Short Cracks During High Strain Fatigue and
Fractures of Corner Cracked Nozzles,” ASME J. Pressure Vessel Technol., Thermal Cycling,” Low Cycle Fatigue and Life Prediction, ASTM STP 770,
111, pp. 241–247. American Society for Testing and Materials, Philadelphia, pp. 337–381.
关6兴 Kerezsi, B. B., Kotousov, A., and Price, J. W. H., 2000, “Experimental Appa- 关15兴 Kerezsi, B. B., Price, J. W. H., and Ibrahim, R., 2003, “A Two-Stage Model for
ratus for Thermal Shock Fatigue Investigations,” Int. J. Pressure Vessels Pip- Predicting Crack Growth, Due to Repeated Thermal Shock,” Eng. Fract.
ing, 77共7兲, pp. 425–434. Mech., 70共6兲, pp. 721–730.
Downloaded 01 Feb 2009 to 130.194.10.86. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
View publication stats