You are on page 1of 6

Three-Dimensional Fracture

Analysis of Circumferentially
Cracked Boiler Tubes

E. Citirik
Advanced Engineer
Cameron International, Inc.,
Houston, TX 77041
e-mail: erc5@lehigh.edu

A comprehensive methodology is developed to understand and characterize the fracture


behavior of circumferentially cracked boiler tubes in this study. Weld overlay is applied on
the coal-fired boiler tubes in order to prevent the degradation of corrosive and erosive
environment that the boiler tubes are exposed to in the power plants. Finite element modeling and analysis are employed for all of the computations including steady-state and transient stress intensity factor (SIF) calculations in this study. Circumferential cracking has
been one of the failure modes in waterwall boiler tubes, which results in high maintenance
and replacement costs. Thermomechanical stresses and corrosive environment are basically the two remarkable contributors that bring about this failure mode. The former one is
investigated and quantified in this study in order to explain the fracture behavior of weld
overlay coatings during the power plant operation. Periodic soot blowing operations cause
cyclic transient thermomechanical stresses on the weld overlay coating that results in crack
propagation and fatigue failure. Three-dimensional fracture analysis of circumferentially
cracked boiler tubes is examined using enriched finite element method in this study.
Transient temperatures and thermomechanical stresses are computed using ANSYS for
five different periodic crack spacing values (h), which are 2, 4, 6, 10, and 20 mm in
the axial direction. 3D fracture analysis was performed, and stress intensity factors were
computed using FRAC3D, which is Finite Element Analysis (FEA) software developed at
Lehigh University. The maximum stress intensity factor is obtained at the deepest penetration of the crack in the model which has the largest periodic axial crack spacing, h 20
mm. The stress intensity factors due to welding residuals decrease as the axial crack spacing, h, decreases. The FEA methodology developed in this research would provide the engineers with the ability to understand the fracture problem and predict component life and
improve the reliability of the weld overlay coated boiler tubes utilized in the power plants.
[DOI: 10.1115/1.4005857]
Keywords: 3D fracture analysis, weld overlay, boiler tube, circumferential crack

Introduction

Boiler tubes are commonly used in chemical plants, refineries,


and power generation plants. These tubes often operate under
extreme temperatures and pressures that can cause them to fail.
Boiler tubes can become weakened by fireside erosion and corrosion. Therefore, weld overlay can be applied to solve this problem. The boiler tube materials are generally SA213/T22, T11, or
T12 type steel. The overlay coating material is usually a high temperature alloy such as S309 steel, Inconel625, or Inconel622.
Weld overlay coatings provide improved resistance to erosion and
corrosion, without adversely affecting the boilers heat transfer
characteristics. Overlays can significantly increase the boiler tube
life. Weld overlay is used to protect the fireside surface, while
also adding additional material to this same surface. Weld overlay
coatings are susceptible to circumferential cracking in low NOx
combustion environments in power plants [1]. A picture of a
cracked boiler tube panel section is shown in Fig. 1.
The soot blowing operations performed to improve the efficacy
of the coal-fired boiler tubes result in thermal shock behavior that
produces significant transient thermal stresses on the boiler tubes.
The transient thermal stress problem in a circumferentially cracked
hollow cylinder was studied using analytical techniques in Ref. [2].
First the transient temperature and stress distributions in an
uncracked cylinder are calculated. The transient thermal stresses
Contributed by the Pressure Vessel and Piping Division of ASME for publication
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received March 1, 2010;
final manuscript received November 28, 2011; published online May 17, 2012.
Assoc. Editor: F. W. Brust.

Journal of Pressure Vessel Technology

are applied on the fracture surface, as the crack surface tractions


and transient SIFs are calculated [2].
The fracture behavior in a hollow cylinder with cladding was
studied in Ref. [3]. It was shown that the SIF calculated for the
crack tip in the elastic base material was reduced with the yielding
of the cladding material [3]. The fracture problem in the coal-fired
boiler tubes is a periodic array of cracks, i.e., it is not a single
crack problem. An analytical approach was employed to characterize the periodic array of cracks subjected to surface tractions in
Ref. [4].
Transverse cracking in the vicinity of long longitudinal fillet
welds has been a serious problem in engineering structures. A
methodology to investigate the fracture behavior in Y-joints that
are often defined in offshore structures is described in Ref. [5].
The SIFs were obtained for a surface semi-elliptical crack along
the crack tip using quarter-point elements using specialized finite
element meshes [5]. Different crack modeling techniques in FE
analysis were explained in Ref. [6]. The fracture behavior of
cracked tubular joints was characterized using more detailed modeling of welds and cracks. Both through-thickness cracks and surface cracks could be characterized by employing the different
type of crack tip models given in Refs. [6] and [7].
The weld overlay coating is applied on the boiler tubes so that it
becomes essential to understand and investigate the welding residual stresses in order to accurately characterize the fracture behavior
of the circumferentially cracked weld overlay coatings. Transverse
cracking in welded structures is strongly influenced by the residual
stresses that arise during welding process. Welding is a reliable and
efficient metal joining process widely used in industry. Research

C 2012 by ASME
Copyright V

JUNE 2012, Vol. 134 / 031202-1

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 1 Picture of the boiler tube panel section, h 5 periodic


crack spacing

has been conducted to determine the temperature distribution and


residual stresses due to welding experimentally in Refs. [810] and
computationally in Ref. [11].
The mode I stress intensity factor, KI, is one of the parameters
that is calculated in fracture mechanics to accurately predict the
stress state near the tip of a crack caused by an applied loading.
For some common geometries and specimens, the analytical
expressions to determine SIFs are available in many handbooks
and other references. However, there is no analytical solution of
the complex geometries examined in this study, circumferentially
cracked weld overlay coatings. Therefore, Finite Element Modeling (FEM) was used to determine SIFs computationally. In this
study, SIFs are determined using enriched 3D crack tip elements.
Commercial software, ANSYS, is used as a preprocessor to generate
Finite Element Analysis (FEA) models for the computation of SIFs
obtained in this study. FRAC3D, to compute stress intensity factors
due to thermal stresses, was employed in fracture mechanics calculations. FRAC3D was previously developed by Ayhan and Nied at
Lehigh University [12]. This FEA code written in FORTRAN provides
users with accurate computation of SIFs in 3D fracture problems.
The enriched finite element approach, which is very effective for
obtaining stress intensity factors for general three-dimensional crack
problems, is implemented in the source code of FRAC3D. Transition
elements are located between the regular elements and the enriched
crack tip elements in order to satisfy the displacement compatibility.
The transitional elements have an additional modifying factor,
which zeros the nonpolynomial asymptotic behavior [12].
Fracture analysis of the boiler tubes considered in this study is
performed in three steps using FRAC3D. The first step is determination of SIFs due to residual stresses caused by welding. The second step is determination of SIFs due to steady-state thermal
stresses. The final one is determination of SIFs due to transient
thermal stresses. Since linear elastic fracture mechanics approach
is employed in these computations, the SIFs obtained from the
three steps could be superposed to find the maximum SIF that the
boiler tube experiences.

Thermal Analysis

The thermal loading of DT 50  C is applied in this study on the


circumferentially cracked boiler tube, i.e., the maximum temperature of 400  C is reduced to 350  C on the fireside of the boiler
tube. All calculations in this study are performed considering
five different periodic axial crack spacing values: h 20 mm,
h 10 mm, h 6 mm, h 4 mm, and h 2 mm. The measured
dimensions of the 2D boiler tube cross section are shown in Fig. 2.
Half of the boiler tube is modeled considering symmetric boundary
031202-2 / Vol. 134, JUNE 2012

Fig. 2 2D cross section of the weld overlay coated boiler tube


geometry utilized in the FEA computations (dimensions are in
millimeter)

conditions. Even though the waterside area in this figure is circular,


the outside of the boiler is noncircular.
The steady-state and the transient thermal analyses on the circumferentially cracked weld overlay coatings are performed using
ANSYS. Heat balance equation obtained from the principle of conservation of energy is employed for thermal analysis in ANSYS.
Temperature is the only degree of freedom in the finite element
solutions performed using ANSYS. The nodal temperatures are utilized to calculate other thermal quantities [13].
Half of the boiler geometry in the lateral direction was modeled
in 3D considering the symmetric boundary conditions for the
steady-state and the transient thermal and stress analyses. The 3D
element used for both steady-state and transient thermal analyses
is a 20-node thermal solid element (SOLID90), which has one
degree of freedom, temperature at each node. Temperature values
computed at each node are applied as thermal loading in the fracture analysis to determine the stress intensity factors for both
steady-state and the transient thermal cases.
The thermal and mechanical material properties used in all thermal and stress calculations are given in Tables 1 and 2 as a function of temperature. The steady-state temperature distributions for
five different axial crack spacing values are computed and all of
them have the same temperature distribution. Definitely, the temperature distribution is independent from the periodic crack spacing. Temperature distribution of h 20 mm model at steady
state is shown in Fig. 3.
There are five different FEA models developed to simulate different axial crack spacing values ranging from 2 mm to 20 mm,
which is determined from the analysis of the field data provided
by Pennsylvania Power Lines (PPL). The largest threedimensional finite element model, which has 14,457 nodes and
Table 1 Temperature dependent thermal properties of the substrate (A213/A) and overlay coating material (309S) (T: temperature, k: thermal conductivity, q: density, c: specific heat)
A213/A
T ( C)
21.1
500
1200

k(w/m )

309S
2

k/qc(m /s)

k(w/m )

42.9
12.44  106
34.1
6.88  106
22.3
5.80  106
q 7900 kg/m3

k/qc(m2/s)

9
3.25  106
13.8
4.2  106
19.3
5.4  106
q 7950 kg/m3

Transactions of the ASME

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Table 2 Temperature dependent mechanical properties of


the substrate (A213/A) and overlay coating material (309S) (T:
temperature, E: Youngs modulus, a: coefficient of thermal
expansion, Sy: yield strength, v: Poissons ratio, Suts: ultimate
tensile strength)
A213/A
T ( C)
21.1
500
1200

E(GPa)

a(1/ C)

205
1.10  105
170
1.50  105
110
1.87  105
Sy 200 Mpa, v 0.30
Suts 400 Mpa

309S
E(GPa)

a(1/ C)

210
1.55  105
120
1.78  105
103
2.00  105
Sy 160 Mpa, v 0.27
Suts 455 Mpa

3000 elements used in the computations, is shown in Fig. 4, which


has h 20 mm axial crack spacing.
The constant temperature is applied on both fireside and waterside of the boiler tube, and convective boundary conditions were
defined on the ambient side, which is free convection, to simulate
the steady-state thermal condition. The waterside temperature is
kept constant at 370  C, and the fireside temperature is reduced
from 400  C to 350  C in 100 s to simulate the soot blowing operation and compute the maximum transient thermal stresses in the
structure. The corresponding transient temperature distributions
obtained from the transient thermal analysis in the 3D boiler tube
model are shown in sequence of pictures in Fig. 5.

Structural Analysis

Three different stresses are considered to obtain the final stress


state that is anticipated to result in surface cracking and fatigue
crack growth. These are residual stresses due to welding, steadystate thermal stresses due to temperature difference between the
waterside and the fireside, and the transient thermal stresses
caused by soot blowing operations.
Thermal stress is introduced within a body that has a gradient
in temperature. The region of material which has a higher temperature generally produces a larger thermal expansion than regions
with lower temperatures. Hence, for the boiler tube, the warmer
side of the tube is in compression and the cooler boundary is in
tension in steady-state thermal conditions. In a soot blowing operation, the temperature of the fireside decreases to a lower value.
The larger the temperature drop or temperature gradient, the more
severe the thermal stresses are.

Fig. 4 3D finite element model of a circumferentially cracked


boiler tube (h 5 10 mm)

In order to compute the stresses in the tube after the thermal analysis, the thermal element should be replaced by an equivalent structural element in ANSYS. Thermal stress solutions are thus obtained
from two consecutive ANSYS analyses; the first analysis solves for
the temperature distribution within the model from the given thermal boundary conditions. Then, the nodal temperatures from the
thermal analysis are directly input to the structural analysis.
The structural boundary conditions for the 3D model are
applied as given in Fig. 6, where u, v, and w are the x, y, and z
components of the displacement, h is the surface crack spacing in
the axial direction of the boiler tube, a is the width of the model
in x direction, b is the thickness of the model in y direction. The
main goal is to calculate the stresses around the crack region in
the axial direction, because it is assumed to be a mode-I fracture
problem, i.e., KI > 0 and KII KIII 0. The axial length of the
boiler tubes in a power plant furnace is much larger than the diameter of the boiler tube, i.e., axial length is much larger than ten
times of the diameter of the boiler tube, and the internal steam
pressure is acting perpendicular to the axis of the boiler tube geometry; therefore, its impact on the mode-I fracture behavior was
assumed negligible in the present work.
Figure 7 depicts the uncracked steady-state axial stresses in the
boiler tube. Since the fireside is warmer than the waterside at steady
state, there is compression at the fireside and tension at the waterside in the axial direction. This compression will result in negative
stress intensity factors for cracks on the outer surface, i.e., cracks
within 0.8 mm from the outer surface are in the compressive stress
region, at steady-state thermal conditions. Crack depth is 0.64 mm
in all of the fracture computations performed in this study that
results in zero stress intensity factors at steady-state thermal conditions. Sequence of images depicting axial stress distribution during
soot blowing operation is shown in Fig. 8 in order to demonstrate
the stress distribution in the uncracked boiler tube geometry. The
temperature of the fireside starts to decrease due to a simulated soot
blowing operation. The compression in the fireside surface becomes
tension as the temperature drops from 400  C to 350  C.

Fig. 3 Temperature distribution (in  C) of h 5 20 mm model at


steady-state

Journal of Pressure Vessel Technology

Fracture Analysis

Three-dimensional fracture analysis for a semicircular surface


crack was performed employing enriched crack tip elements using
FRAC3D, 20 node quadratic hexahedron elements are used in
FRAC3D calculations. In the absence of a crack, FRAC3D is capable
of evaluating displacements and three-dimensional stress states at
the nodal points. When a crack or multiple cracks are present in
the model, additionally, fracture parameters such as stress intensity factors and strain energy release rates are computed at the
nodes along the crack front.
JUNE 2012, Vol. 134 / 031202-3

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 5 Sequence of images depicting temperature distribution during soot blowing operations
in 100 s (The fireside: t 5 0 s is at 350  C and t 5 100 s is at 400  C)

Fig. 6 Mechanical boundary conditions in the x-y plane, front view of the model and in the z-y
plane, right view of the model. u 5 0 and w 5 0 represent the symmetric boundary conditions

Fig. 7

Steady-state axial thermal stresses, rzz, in the boiler tube (units are in pascal)

The first step in the fracture analysis is to calculate stress intensity factors due to residual stresses caused by the welding. Residual stresses in weld overlay operations have been shown to be
equal to the yield strength of the material by simulating the welding process using finite element analysis [14]. The welding residual stress was applied uniformly to the model (with crack) to
calculate the stress intensity factors. The uniform stress of 160
MPa is applied only on the circumferential crack surface of the
031202-4 / Vol. 134, JUNE 2012

weld overlay coated boiler tube geometry in order to calculate the


KI values due to welding residual stresses. It is tensile stress
because the weld overlay coating is assumed to be stress free
when weld material is flowing over the boiler tube during fabrication but once the weld cools down, the weld overlay goes into tension. Therefore, any crack in the weld overlay coating is subject
to tensile stresses due to welding residual stresses that generates
positive mode-I stress intensity factors in the axially spaced
Transactions of the ASME

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 8 Sequence of images depicting stress distribution during soot blowing operation (units
are in pascal)

circumferential cracks. The superposition approach employed in


this study is shown in Fig. 9. The stress intensity factors for five
different axial crack spacing values are computed under applied
welding stresses as the first step in the overall methodology developed in this study. Even though the applied stress and the crack
size are the same for all of the five models, the decreasing axial
crack spacing results in reduction in the magnitudes of the KI values as shown in Fig. 10. The largest axial crack spacing, which is
20 mm, results in the largest KI value from the fracture analyses.
However, the 20 mm and 10 mm axial crack spacing KI results are
identical, which points out that the KI values are not sensitive to
the axial crack spacing above a certain axial crack spacing. The

Fig. 9 Superposition approach for computing KI values due to


residual stresses from welding

Journal of Pressure Vessel Technology

maximum stress intensity factor at p/2, which is the deepest point


of penetration, is shown in Fig. 11 for different axial crack spacing values.
The simulated crack depth is 0.64 mm with the crack width of
7.2 mm in the circumferential direction of the boiler tube in this
study, these specific values were accepted as the baseline crack
size to characterize the fracture behavior, i.e., the sensitivity of
the fracture behavior to varying axial crack spacing values.
The stress intensity factors due to steady-state thermal stresses
are computed next. Since the fireside is in compression at steadystate thermal conditions, the crack surfaces are closed. The values
computed for steady-state condition are very small negative values around zero. Therefore, they can be accepted as zero for
steady-state thermal conditions.
Finally, the stress intensity factors due to transient thermal
stresses were computed. Transient thermal analysis was performed
to obtain time dependent temperatures at the nodes from ANSYS.
These time dependent temperatures at the nodes are input to FRAC3D
in order to determine the stress intensity factors for the transient conditions. The computations were performed for five different h (axial
crack spacing) values, which are 2 mm, 4 mm, 6 mm, 10 mm, and
20 mm. The KI values are varying for different crack spacing

Fig. 10 KI values along the circumferential crack front due to residual stresses caused by welding residual stresses. (h/2 5 5 mm
and h/2 5 10 mm results are identical)

JUNE 2012, Vol. 134 / 031202-5

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 11 KI values due to residual stresses caused by welding at


the deepest point of crack penetration (a; crack depth 5 0.64 mm,
h; crack spacingvarying from 2 mm to 20 mm)

conditions. Transient thermal analyses were performed using five


different FEA models with varying axial length to obtain the temperature and stress distributions due to soot blowing operations. All
computations were performed using reference DT, which was
50  C, in transient calculations. Coupled structural analyses were
performed to simulate the thermal stresses. High tensile thermal
stress values, those were close to two thirds of the yield strength,
were obtained for 50  C temperature difference due to cooling temperature boundary conditions on the fireside of the boiler tube.
Both temperature and stress results were shown as contour plots.
Stress intensity factors were generated for five different crack
spacing for both welding residual stresses and transient thermal
stresses. The maximum values were obtained at the deepest point
of penetration for cracks with the greatest periodic axial spacing.
As the axial crack spacing distance increases, the values of computed stress intensity factors also increase.
In conclusion, DK value is obtained in order to assess the
remaining life of circumferentially cracked weld overlay coatings,
i.e., crack growth. Tensile transient thermal stresses cause positive
DK, which is the key parameter in fatigue crack growth calculations. A systematic computational methodology was developed in
order to investigate the circumferential cracks in the weld overlay
coated boiler tubes, i.e., first obtain the SIFs due to welding residual stress, and then obtain the SIFs due to transient thermomechanical stresses. The parameters computed for the cracks will be
utilized as input for fatigue life calculations, e.g., DK could be
employed in the Paris law to determine the remaining life for a
given crack configuration, i.e., for a given crack size and axial distance between the circumferential cracks.

Acknowledgment
The work was supported by Pennsylvania Power and Light
(PPL), First Energy Corporation and Virginia Power, while the
author was at Lehigh University. The author is also greatly
indebted to Professor H.F. Nied for his suggestions and technical
discussions to complete this study.
Fig. 12 Transient stress intensity factors for axial crack spacing,
h 5 20 mm

models because the KI value becomes smaller as the total number of


cracks increase in the axial direction, i.e., the axial crack spacing
distance becomes smaller, that results in redistribution of the stresses
around the crack tip in the weld overlay coated boiler tube geometries. The transient KI values are computed for all of the models considered in this study. The computational results for only h 20 mm
axial crack spacing distance are shown in Fig. 12 because the other
models show the same behavior qualitatively, i.e., t 0 s is the
steady-state thermal condition that has negative KI values, as the
temperature on the fireside decreases from 400  C to 350  C in
100 s, the largest KI values are computed at the coldest fireside temperature at t 100 s. Therefore, the maximum tensile stress values
were obtained at t 100 s. The
fracture toughness value, KIC, of the
p
overlay coating is 120 MPa m. The maximum KI that will be
observed in the boiler tube fracture problem examined in this study
could be calculated
p as KI KDT Kres, which results in approximately 12 MPa m (from Figs. 10 and 12). The calculated maximum SIF is much smaller than the KIC of the weld overlay coating
material that points out that the failure is not in tearing mode, i.e., it
is a fatigue crack growth problem. Therefore, it can be concluded
that the crack was stable, because the KI values calculated for welding residual and transient thermal stresses are not sufficient enough
to tear the structure, and the failure mode is fatigue crack growth
due to transient thermal loading during soot blowing operations.

Conclusions

Full 3D models were used to examine the mechanism of cracking


in weld overlay boiler tubes by employing symmetric boundary
031202-6 / Vol. 134, JUNE 2012

References
[1] Luer, K., DuPont, J. N., Marder, A. R., and Skelonis, C., 2001, Corrosion Fatigue of Alloy 625 Weld Claddings in Combustion Environments, Mater. High
Temp., 18, pp. 1119.
[2] Nied, H. F., and Erdogan, F., 1983, Transient Thermal Stress Problem for a
Circumferentially Cracked Hollow Cylinder, J. Therm. Stresses, 6, pp. 114.
[3] Nied, H. F., 1984, Thermal Shock in a Circumferentially Cracked Hollow Cylinder With Cladding, Eng. Fract. Mech., 20(1), pp. 113137.
[4] Nied, H. F., 1987, Periodic Array of Cracks in a Half-Plane Subjected to Arbitrary Loading, ASME J. Appl. Mech., 54, pp. 642648.
[5] Chiew, S. P., Lie, S. T., Lee, C. K., and Huang, Z. W., 2001, Stress Intensity
Factors for a Surface Crack in a Tubular T-Joint, Int. J. Pressure Vessels
Piping, 78, pp. 677685.
[6] Cao, J. J., Yang, G. J., Packer, J. A., and Burdekin, F. M., 1998, Crack Modeling in FE Analysis of Circular Tubular Joints, Eng. Fract. Mech., 61, pp.
537553.
[7] Taylor, D., 1996, Crack Modeling: A Technique for the Fatigue Design of
Components, Eng. Failure Anal., 3(2), pp. 129136.
[8] Murugan, S., Kumar, P. V., Raj, B., and Bose, M. S. C., 1998, Temperature
Distribution During Multipass Welding of Plates, Int. J. Pressure Vessels
Piping, 75, pp. 891905.
[9] Murugan, S., Rai, S. K., Kumar, P. V., Jayakumar, T., Raj, B., and Bose, M. S.
C., 2001, Temperature Distribution and Residual Stresses Due to Multipass
Welding in Type 304 Stainless Steel and Low Carbon Steel Weld Pads, Int. J.
Pressure Vessels Piping, 78, pp. 307317.
[10] Sarkani, S., Tritchkov, V., and Michaelov, G., 2000, An Efficient Approach
for Computing Residual Stresses in Welded Joints, Finite Elem. Anal. Design,
35, pp. 247268.
[11] Fricke, S., Keim, E., and Schmidt, J., 2001, Numerical Weld ModelingA
Method for Calculating Weld-Induced Residual Stresses, Nucl. Eng. Des.,
206, pp. 139150.
[12] Ayhan, A. O., and Nied, H. F., 2002, Stress Intensity Factors for ThreeDimensional Surface Cracks Using Enriched Finite Elements, Int. J. Numer.
Methods Eng., 54, pp. 899921.
[13] ANSYS, Inc., 1999, Thermal Analysis Guide, ANSYS Release 5.6.
[14] Yildirim, B., 2000, Nonlinear Thermal Stress/Fracture Analysis of Multilayer
Structures Using Enriched Finite Elements, Ph.D. thesis, Lehigh University,
Bethlehem.

Transactions of the ASME

Downloaded From: http://asmedigitalcollection.asme.org/ on 01/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like