See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281035733

Develop an Excel-Based Modeling Tool to Predict Weld and HAZ Cooling Rate
and Hardness for Pipeline Welding

Conference Paper · July 2015

DOI: 10.1115/PVP2015-45010

CITATION READS

1 1,045

5 authors, including:

Y.P. Yang Tom McGaughy

Edison Welding Institute (EWI) EWI
106 PUBLICATIONS   373 CITATIONS    19 PUBLICATIONS   40 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Y.P. Yang on 17 August 2015.

The user has requested enhancement of the downloaded file.

 Proceedings of 2015 ASME Pressure Vessels & Piping Conference
PVP2015
July 19-24, 2015, Boston, Massachusetts, USA

PVP2015-45010

DEVELOP AN EXCEL-BASED MODELING TOOL TO PREDICT WELD AND HAZ

COOLING RATE AND HARDNESS FOR PIPELINE WELDING

Yu-Ping Yang Zhenning Cao

EWI Advantech GS Enterprises, Inc.
Columbus, OH 43221, USA San Diego, CA, 92121, USA

Jerry Gould Tom McGaughy Jonathan Jennings

EWI EWI EWI
Columbus, OH 43221, USA Columbus, OH 43221, USA Columbus, OH 43221, USA

ABSTRACT years, new formulae [7, 8] have been developed for individual
A Microsoft Excel-based screening tool was developed to steel. Yan et al [7] developed and validated two formulae to
allow an engineer with weld process knowledge to predict predict hardness in the heat affected zone of 9%Ni steel. Oddly
cooling rate and hardness during welding procedure et al [8] have developed a method for determining the hardness
qualifications to screen a combination of materials and welding of a multipass weld of 2.25Cr-1Mo steels. But, this method
process parameters quickly to meet requirements of fabrication requires a TTT diagram, phase diagram, absolute martensite and
and design codes. The material properties for commonly used bainite hardness levels, which are not readily available.
pipeline steels have been built into a database coupled with the Therefore, Nicholas and Abson [9] developed a semi-empirical
screening tool. The Excel-based tool includes a physics-based method of determining the maximum hardness in the heat
laser and arc welding solution which was developed based on affected zone of multipass welds. Recently, new structural steels
Rosenthal’s mathematical equations for a point heat source to with improved properties have been introduced, especially for
predict thermal cycles by inputting welding parameters. A pipelines and offshore structures used in oil and gas production,
reflecting heat source scheme was adapted to model the processing and transportation. To improve the accuracy of
boundary conditions and plate thickness effect on cooling rate. hardness predictions, Nolan et al [10] used an artificial neural
The Excel-based tool also includes a microstructure model network to develop new hardness prediction models that are more
which was developed based on the Ashby model. The accurate than Yurioka’s model for typical in-service welding
microstructure model can be used to predict the distributions of conditions.
individual phases such as ferrite, bainite, and martensite along Significant progress has been made in numerical simulation
with a hardness map across the weld and heat-affected-zone of weld microstructure in the past twenty years. It is now
(HAZ) regions by integrating with the thermal model. possible to predict the microstructure of welds and the resulting
hardness based on a rule of mixture from their chemical
compositions and welding parameters [11-13]. Ion et al [12]
INTRODUCTION developed a microstructure model (Ashby model) and a
Hardness is an important property for determining material hardness prediction method based on the fundamental phase
weldability and service performance, particularly in corrosive transformation thermodynamics and kinetics theories. Yang et al
environments. Increasingly, fabrication codes set up a limit on [13] implemented Ion’s method in a finite element code in an
the maximum hardness allowed in the weld region. A number of online weld modeling software for predictions of temperature,
empirical hardness estimation formulae [1-6] have been microstructure, hardness, residual stress, and distortion.
developed since the 1970s to predict as-welded heat affected To enable a welding engineer to use these microstructure
zone (HAZ) hardness for C-Mn and low alloy steels. In recent and hardness prediction tools, a Microsoft Excel software

1 Copyright © 2015 by ASME

application was developed to predict cooling rate, THERMAL MODEL
microstructure, and hardness for pipeline steels. The cooling For years pipelines have been welded manually using arc
rate was predicted using Rosenthal’s mathematical equation [14] welding processes such as gas metal arc welding (GMAW).
and considering the plate thickness effect with a reflecting heat However, for larger diameter pipeline projects, automated arc
source scheme developed by Cao et al [15]. The microstructure welding systems have been increasingly used. HLAW, a
model based on Ion’s method [12] has been integrated with the combination of laser beam welding and arc welding, was
thermal solution and embedded into an Excel file with VBA developed and demonstrated for pipe welding in a joint
computer programming. It should be pointed out that the industrial project carried out by EWI. This advanced welding
developed Excel modeling tool is less accurate than a finite process combines the advantage of laser welding such as fast
element based tool because of many simplifications in the operation speed with deep penetration and the advantage of arc
thermal model such as a point heat source, constant material welding such as slow cooling rate and the addition of filler
properties, and ignoring the cooling from welding fixtures. metal to tailor weld properties. It avoids the fast-cooling
Finite element based thermal solution will be developed to problem of laser welding and shallow penetration of arc
replace the analytical thermal model in the future version of this welding. A half-inch thick pipe butt joint has been successfully
tool to improve the accuracy of this modeling tool. welded with a full-penetration single-pass HLAW weld using 10
kW of laser power [16].
NOMENCLATURE
 thermal diffusivit y (m 2 /s)
 thermal conductivity (W/m K)
v travel speed (m/s)
CE carbon equivalent
CR cooling rate (C/s)
GMAW gas metal arc welding Figure 1. Coordinate System of Welding Heat Source
H hardness (HV)
Hm hardness (HV) of martensite The physics-based solution for HLAW was developed based
Hi hardness (HV) of microstructure phase i on Rosenthal’s mathematical equation for a point heat source to
predict thermal cycles by inputting welding parameters.
HAZ heat affected zone Rosenthal moving point heat source model for semi-infinite
HLAW hybrid laser and arc welding body (quasi-steady state solution) has been widely used for
LBW laser beam welding thermal modeling of arc welding processes. For the coordinate
m number of point heat sources in laser weld ing system shown in Fig. 1, the arc-welding induced temperature
can be calculated with the following equation:
n number of point heat sources in GMA welding
Qin heat input (W) T Qin , x, y, z, t   T0  T ps Qin , x, y, z, t 
T ps Qin , x, y, z, t   H max Ds
th
Qlaser_i heat input (W) of the i heat source in laser weld ing
QGMA_i heat input (W) of the i th heat source in GMA welding H max 
Qin
(1)
Vm volume fraction of martensite 4 x  y 2  ( z  vt ) 2
2

 v  2 
 x  y 2  ( z  vt ) 2  z  vt 
Vi volume fraction of microstructure phase i
Ds  exp 
t heating time (s)  2   
t transforma tion time (s)
HLAW could be simulated by the superposition of
t1/2
m
critical time (s) that gives a 50% martensite temperature induced by an arc welding heat source and laser
t1/2
b
critical time (s) that gives a 50% bainite welding heat source, as illustrated in Fig. 2. The mathematical
T0 initial temperature (C) implementation for the superposition is expressed in Equation
(2). Equations (2) can be used for modeling GMAW by
x, y, z coordinate (m) inputting zero laser power and laser beam welding (LBW) by
xi , yi coordnates (m) of the i th point heat source inputting zero arc power. For modeling HLAW, the power for
both GMAW and LBW are needed to input to the thermal

2 Copyright © 2015 by ASME

model. Therefore, three welding processes, HLAW, GMAW, and of continuous heating or continuous cooling on the kinetics of
LBW, can be simulated using the developed thermal model. transformation was modeled. The model also calculated the
dissolution and growth rates of carbides during heating and
m
 
T x, r , t   T0   T ps Qlaser_i, x, y  yi , z, t 
considered the effect of micro-alloying elements on the growth
of austenite grains in the HAZ. Transformations during cooling
i 1
(2) were modeled using a representation of experimental continuous
n

 T ps QGMA_i , x  xi , y  yi , z, t  cooling transformation (CCT) diagrams. With the calculated
i 1 microstructure phases: ferrite, bainite, and martensite, the HAZ
hardness was estimated based on the rule of mixtures.
Stand-alone microstructure calculation software for
predicting microstructure and hardness for welding simulation
was developed based on the microstructure model developed by
Ion, Easterling, and Ashby [12]. It has been implemented into
EWI Weld Predictor (an online welding simulation tool) to
predict the distribution of each individual phase such as ferrite,
Figure 2. Illustration of multiple point heat source
bainite, and martensite and the hardness map around the weld
superposition method
area. The software can be coupled with the analytical thermal
solution developed in this paper and the thermal analysis
Equation (1) was obtained by assuming an infinite-size
conducted by commercial finite element software such as
workpiece. In order to consider boundary effects at the top and
ABAQUS.
bottom surfaces for a finite thickness plate, it was assumed that
Implementation of the microstructure model was described
the heat loss from the surfaces was negligible in comparison
in Ref. 13. For completeness, the main equations are briefly
with that due to conduction in an infinite body, i.e.,
discussed here. The calculation of microstructure starts by
 T z  0 . As shown in Fig. 3, to prevent heat loss from the
calculating the carbon equivalent (CE) [17] using Equation (3).
top surface, an auxiliary heat source (referred as 2) with the
same heat input and travel speed as the original heat source was Mn Cr  Mo  V Ni  Cu
placed at the reflected mirror position. Similarly, the auxiliary CE  C    (3)
6 5 15
heat source 3 was placed for the bottom surface. Once the
auxiliary source 3 is imposed, another auxiliary source 4 needs where C is carbon, Mn is manganese, Cr is chromium, Mo is
to be placed above the top surface so that the top surface is molybdenum, V is vanadium, Ni is nickel, and Cu is copper. All
insulated. As such, a number of the heat sources are required to composition elements are defined in weight percent.
obtain a reasonable temperature solution to satisfy the top and The CE can be used to relate to the following critical
bottom boundary conditions. The detailed mathematic cooling rates [12]:
implementation of reflecting heat source scheme can be found in
Ref. 15.
log t1m/ 2  8.79CE  1.52
(4)
log t1b/ 2  8.84CE  0.74

where t1 / 2 is the critical time that gives a 50% martensite and
m

50% bainite structure, and t1b/ 2 is the time that gives a 50%
mixture of bainite and 50% ferrite. From the critical cooling
rate, the final volume fraction of martensite (Vm), bainite (Vb),
and perlite plus ferrite (Vfp) can be calculated from the following
equations [12]:

Figure 3. Heat source refection scheme

MICROSTRUCTURE MODEL
Ion, Easterling, and Ashby [12] developed a microstructure
model to predict the HAZ hardness. In the model, thermal
analysis was conducted using Rosenthal’s solution. The effect

3 Copyright © 2015 by ASME

   t  
2 Based on the steel chemistry input by the user and the
  
Vm  Vmax exp  0.69 cooling rate calculated from the thermal analysis, Equations (3)
 t m  
  1 / 2   through (9) predict the final volume fractions and hardness.

  t  
2 EXCEL INTERFACE
  
Vb  Vmax exp  0.69 V
 t b   m
(5) The closed form solution for HLAW is not as easily used to
  1/ 2  
  predict temperature, microstructure, and hardness because of
V fp  1  (Vm  Vb ) complex calculations. Therefore, a user interface based on
Microsoft Excel was developed for facilitating the applications.
The thermal solution and microstructure model were embedded
where t is the transformation time and Vmax is the maximum into an Excel file with VBA computer programming. The Excel
volume fraction of austenite which can form during a thermal file includes 11 pages: start page, geometry and material page,
cycle with a peak T and be calculated using the following welding process page, calculation page, six report pages, and a
equation [12]: database page.
The start page, as shown in Fig. 4, includes three functions:
Vmax  0 if T p  A1 unit selection, analysis type (temperature only or temperature
T p  A1 C A3  T p and microstructure) selection, and an option for clearing the
Vmax   if A1  T p  A3 spreadsheet. User guidelines and Excel tabs (pages) are also
A3  A1 0.83 A3  A1
descripted in the start page.
Vmax  1 if T p  A3
(6)

Where A1 (K) and A3 (K) can be estimated by the following

equations:

A1  1187.8  9780.3C if C  0.019

A1  999.78 if C  0.019
(7)
A3  937.1  250.91 exp( 1.6636C ) if C  0.76
A3  999.78 if C  0.76

The hardness for alloy steels can be calculated using the

Figure 4. Start page
following empirical equation [12].
The geometry and material page, as shown in Fig. 5, allows
H m  127  949C  27 Si  11Mn  8 Ni  a user to input pipe dimensions, pipe material properties, and
16Cr  21 log CR material chemical composition. By selecting a material name
H b  323  185C  330 Si  153Mn  65 Ni  from a drop-down list, the material thermal-physical properties
144Cr  191Mo  (89  53C  55Si  22 Mn  will be fetched from the user pre-built database page. In the
(8) current software version, a user can only select Ashby model. It
10 Ni  20Cr  33Mo ) log CR is anticipated the Suzuki model [4] will be added in a future
H fp  42  223C  53Si  30 Mn  12.6 Ni  version.
7Cr  19 Mo  (10  19 Si  4 Ni  8Cr  The welding process page allows a user to select either
HLAW or projection welding, as shown in Fig. 6. The
130V ) log CR
implementation of a projection weld model will be introduced
in future publications. If HLAW is selected, welding parameter
where Hm, Hb, and Hfp are the hardness of martensite, bainite, tables for HLAW will be shown to allow a user to input welding
and perlite plus ferrite, respectively. CR is the cooling rate at parameters. After entering the welding parameters, the user can
700 ºC (ºC/hr). The final hardness (H) at each point of the weld move to the calculation page. Note that a user can input zero
can be estimated using the rule of mixtures. laser power to model GMAW or zero GMAW power to model
LBW. The following three types of calculation have been built
H  H mVm  H bVb  H fpV fp (9) in this page for predicting temperature, cooling rate,
microstructure, and hardness:

4 Copyright © 2015 by ASME

 • at points user conducts calculations at both points and on a weld cross
• on a weld cross section section, all six report pages: report at points, report maximum
• on a surface through the pipe thickness. temperature, report temperature map at a time, report cooling
rate between 800C to 500C, report microstructure phases, and
report hardness, will show up.
In addition, a database page was designed to allow a user to
build his/her database for calculation using their material
properties.

ANALYSIS EXAMPLE
An analysis was conducted using the developed Excel
modeling tool to predict temperature, cooling rate from 800C
to 500C, microstructure, and hardness for a bead-on-plate
welding on an X70 plate with filler wire ER70S-6. The
chemical compositions of X70 and ER70S-6 are shown in Table
1. A dilution 0.36 was used to consider the mixing in the weld
between the base material and the filler wire. The modeling
prediction was compared with the experimental results to
validate the Excel modeling tool.

Table 1. Chemical compositions of base material and

filler metal
Material C Mn Si S P Cr Ni Cu V Mo
X70 0.07 1.66 0.30 0.01 0.01 0.01 0.11 0.01 0.00 0.22
ER70S-6 0.09 1.10 0.50 0.01 0.01 0.00 0.00 0.20 0.01 0.00

Figure 5. Geometry and material page

Figure 7. Welding experiment

Figure 6. Welding process page

To carry out a calculation, a user needs to input the initial

temperature or pre-heating temperature. After calculation, report
pages will appear corresponding to the type of calculation: at
points or on a weld cross section or on a surface through
thickness or a combination of points and a weld cross section or
points and a surface. For example, if a user conducts a
calculation at points, the Report-Point page will show up. If a Figure 8. Weld cross-section micrograph

5 Copyright © 2015 by ASME

 Figure 7 shows the as-welded bead-on-plate weld and After 250C, the model predicted slower cooling than measured.
thermocouples (TC) layout. The welding process was GMAW This is because the heat convection coefficient is a function of
with the following welding parameters: current 228A, voltage temperature which cannot be considered in the thermal model.
20.8v, and travel speed 17.5IPM (7.41mm/s). The material The thermal sink from the welding fixture cannot be included in
thickness is 12.7mm. Figure 8 shows the weld cross-section the thermal model as well. Since the cooling rate at lower
macrograph for the weld joint. temperature has limited influence in predicting the hardness of
Figure 9 shows the detailed locations for eight the weld and HAZ, the Excel modeling tool is not overly
thermocouples. The thermocouples were placed at both sides of compromised by this limitation.
the weld and in the holes drilled on the plate. The depth for the Maximum temperatures at all points in the weld cross
A group of thermocouples is 1.59mm and for the B group is section were recorded during the analyses. The maximum
3.18mm. The distance between TC1 and TC5, TC3 and TC7, temperatures can be contour plotted to show the fusion zone
and TC4 and TC8 is 16mm and the distance between TC2 and and heat affected zone, as shown in Fig. 11. It should be pointed
TC6 is 19mm. TC1 and TC3 and TC5 and TC7 should see the out that the entire weld bead is not included in the model. The
same temperature in theory since they have the same distance to part of weld bead above the plate is not included in the model
the weld center. During welding experiment, the bead-on-plate which is a limitation of the Excel modeling tool. This will be
weld was deposited in the middle of both sets of thermocouples. improved in future versions of the modeling tool.
Very similar temperatures were measured between TC1 and The cooling rate between 800C to 500°C has a significant
TC5, TC2 and TC6, TC3 and TC7, and TC4 and TC8, which effect on the material properties of steel welds. The Excel
verifies the thermocouple measurements were accurate. modeling tool predicts the cooling rate from 800C to 500C at
the weld cross section which is contour plotted, as shown in
Fig. 12. The figure provides an overall picture of cooling rate in
a weld region.

Figure 9. Thermocouple locations

The Excel modeling tool predicted temperatures were

compared with the measurement at TC1, TC2, and TC4
locations as shown in Fig. 10. The model can predict the similar
maximum temperature as measured. Prediction and measurement
had good agreement during the heating and cooling to 250C.

6 Copyright © 2015 by ASME

 (a) Martensite distributions

Figure 10. Comparison of thermal cycles at

thermocouple locations T1, T2, and T4 (lines are
measurements and symbols are predictions)
(b) Bainite distributions

(c) Ferrite distributions

Figure 13. Predicted microstructure distributions
Figure 11. Predicted Maximum Temperature
Figure 14 shows the predicted Vickers hardness
distributions. The HAZ hardness is between 300 to 400HV and
the fusion zone hardness is between 200 to 300HV. The
hardness along a line A-A was measured using a 0.5kg load as
shown in Fig. 15 and compared with the model prediction. The
weld cross-section macrograph (Fig. 8) shows a HAZ under the
fusion zone which is similar to the model predicted higher
hardness zone shown with red color in Fig. 14. Analysis results
Figure 12. Predicted cooling rate between 800 to 500C show that the model can predict the hardness in the HAZ and
on a weld cross section the fusion zone with reasonable accuracy. The model under
predicted hardness in the heat un-affected zone since 100%
Figure 13 shows the predicted volume fraction of the ferrite was assumed in that zone. To obtain an accurate
microstructure phases: martensite, bainite, and ferrite in the prediction of the unaltered base metal, a user has to input the
weld region. The HAZ consists of about 70% martensite, as microstructure phases for the base material which may not
shown in Fig. 13a, and about 25% bainite as shown in Fig. 13b, always be known. However, since the initial microstructure
while the fusion zone consists of about 50% bainite, as shown phases do not affect the prediction in the weld and HAZ, the
in Fig. 13b, and about 50% ferrite, as shown in Fig. 13c. In the modeling tool does not require the user to input the initial
calculation, the base material unaffected by the weld thermal microstructure phases. A note has been added in the reporting
cycle was assumed to be 100% ferrite, as shown in Fig. 13c. pages to remind users of this assumption and limitation.
This assumption does not influence the modeling prediction in
the weld and HAZ.

7 Copyright © 2015 by ASME

 To improve the accuracy of this excel modeling tool, a
standalone finite element based thermal-analysis solver will be
implemented in future version to replace the analytical thermal
model.

REFERENCES

[1] Bechert, M. and Holz, R., 1973, “Use of Carbon Equivalent

Figure 14. Predicted hardness distributions and Prediction of the Weld Heat Affected Zone Hardness,”
Schweiss Technik, Vol. 23(8), pp. 344-346.
[2] Lorenz, K. and Duren, C., 1981, “Evaluation of Large
Diameter Pipe Steel Weldability By Means of the Carbon
Equivalent”, Proceedings of Conference on Steels for Pipe and
Pipeline Fittings, TMS, pp. 322-332.
[3] Yurioka, N., Okumura, M., Kasuya, T., and Cotton, H. J. U.,
1987, “Prediction of HAZ hardness of transformable steels,”
Metal Construction, Vol. 19(4), pp. 217R-223R.
[4] Suzuki, H., 1985, “A new formula for estimating HAZ
hardness in welded steels,” IIW DOC IX-1351-85.
[5] Cotrrell, C. L. M., 1984, “Hardness equivalent may lead to a
more critical measure of weldability,” Metal Construction, Vol.
16, pp. 740-744.
[6] Mayoumi, M. R., 1991, “Towards a prediction of the
hardness of the heat-affected zone of steel weldments,” Journal
Figure 15. Comparison of hardness between
measurement and prediction of Materials Science, 26, pp. 2716-2724.
[7] Yan, C., Yang, S., Zhao, J., and Li, W., 2012, “Prediction of
SUMMARY Hardness in Heat Affected Zone of 9%Ni Steel”, Advanced
A closed-form thermal model for welding processes, Materials Research, Bol. 455-456, pp. 406-412.
GMAW, LBW, and HLAW, was developed based on Rosenthal [8] Oddy, A. S., Chandel R. S., 1990, “Report MIC-91-01013
analytical solutions. The thermal model was integrated with a (MTL 90-42(TR)),” Canada Centre for Mineral and Energy
microstructure model which was developed based on the Ashby Technology.
model to predict microstructure and hardness distributions in [9] Nicholas, J. N., and Abson, D. J., 2008, "The prediction of
the weld and HAZ. The thermal model and microstructure model maximum HAZ hardness in various regions of multiple pass
were embedded in an Excel interface to facilitate welding welds", Presented at the 17th International Conference of
engineers in calculating the cooling rate between 800C to Computer Technology in Welding and Engineering, Cranfield,
500C along with weld and HAZ hardness. UK.
A welding experiment was conducted to validate the Excel [10] Nolan, D. Sterjovski, Z., and Dunne, D., 2005, "Hardness
modeling tool to predict thermal cycles and hardness. A Prediction Models Based on HAZ Simulation for In-Service
reasonable agreement was achieved between model prediction Welded Pipeline Steels," Science and Technology of Welding
and measurement. More experimental studies are needed to and Joining, Vol. 10 (6), pp. 681-694.
validate the modeling tool for laser welding and HLAW and for [11] Gould, J., Khurana, S., and Li, T., 2006, “Predictions of
other pipeline steels. Microstructures When Welding Automotive Advanced High-
It is anticipated that the Excel modeling tool is less Strength Steels”, Welding Journal, Vol. 85(5), pp. 111s-116s.
accurate than a finite element modeling tool due to the [12] Ion, J. C., Easterling, K. E., and Ashby, M. F., 1984, “A
simplification in the thermal model. However, the Excel Second Report on Diagrams of Microstructure and Hardness for
modeling tool can be run on any computer installed with Heat-Affected Zones in Welds,” Acta Metallurgica, Vol. 32, pp.
Microsoft office without relying any other modeling and 1949-1955.
simulation software. Welding engineers can use this excel [13] Yang, Y. P., Zhang, W., Gan, W., Khurana, S., Xu, J., and
modeling tool as a screening tool to predict cooling rate and Babu, S., “Online Software Tool For Predicting Weld Residual
hardness during the selection of materials, welding process, and Stress and Distortion”, Proceedings of 2008 ASME Pressure
welding parameters to tailor weld properties to meet fabrication Vessels and Piping Division Conference, PVP2008-61123,
and design code requirements. Vol. 6, pp. 279-288.

8 Copyright © 2015 by ASME

[14] Rosenthal, D. and Cambridge, M., 1946, “The Theory of [16] Boring, M., Thompson, B. and Nagy, B., 2011, "Advanced
Moving Source of Heat and its Application to Metal Pipeline Welding Technologies Boost Productivity," Pipeline
Treatments”, Trans. ASME, Vol.68 (11), pp. 849-866. and Gas Journal, Vol. 238(2), http://www.pipelineandgasjournal.com
[15] Cao, Z., Brust, F., Nanjundan A., Dong, Y., Julta, T., 2000, [17] Ginzburg, V. B.; Ballas, R., 2000, “ Flat Rolling
“A New Comprehensive Thermal Solution Procedure for Fundamentals,” CRC Press, pp. 141–142, ISBN 978-0-8247-
Multiple Pass and Curved Welds”, Proceeding of 2000 ASME 8894-0.
Pressure Vessel and Piping Conference, Seattle, Washington,
USA.

9 Copyright © 2015 by ASME

View publication stats