Aws WJ 202009
Aws WJ 202009
SEPTEMBER 2020
WELDING JOURNAL • VOLUME 99 NUMBER 9 • SEPTEMBER 2020
PUBLISHED BY THE AMERICAN WELDING SOCIETY TO ADVANCE THE SCIENCE, TECHNOLOGY, AND APPLICATION OF WELDING
AND ALLIED JOINING AND CUTTING PROCESSES WORLDWIDE, INCLUDING BRAZING, SOLDERING, AND THERMAL SPRAYING
C2 Select Arc.qxp_FP_TEMP 8/5/20 3:02 PM Page C2
Page 1 RPB.qxp_FP_TEMP 8/5/20 3:34 PM Page 1
Page 2 Camfil.qxp_FP_TEMP 8/5/20 2:33 PM Page 2
September 2020.qxp_Layout 1 8/7/20 7:27 PM Page 3
At the American Welding Society (AWS), we are actively following the Novel Coronavirus (COVID-19) outbreak. The welding and
fabrication industries have been determined to be an essential business, especially as it relates to repair and maintenance
of key infrastructure. The content of the September 2020 issue of the Welding Journal is intended to be accurate when pub-
lished, but we recognize that we are in a rapidly changing situation. For AWS’s official statement on COVID-19, as well as the
latest updates and frequently asked questions, please visit aws.org.
DEPARTMENTS
6 Editorial 43 Society News
7 Press Time News 45 Tech Topics
8 News of the Industry 49 Section News
12 Stainless Q&A 51 Guide to AWS Services
14 RWMA Q&A 52 Personnel
20 Product & Print Spotlight 58 Classifieds
39 Certification Schedule 58 Advertiser Index
40 Welding Workbook
EDITORIAL
In honor of its 75th birthday, here’s why the Closing the Curtain on 2020
independent distributor needs GAWDA:
• 24 h a day/7 days a week/365 days a year It is my honor to wish GAWDA a happy
access to consultants 75th anniversary and to thank our members
• Annual convention gathers more than and sister associations for their continued
800 industry leaders collaboration.
• Educational programs through the Asso- The independent distributor needs GAW-
ciation Education Alliance DA, because while we may not be able to fund
• American Welding Society (AWS)/WEM- or source these resources on our own, collec-
CO relationship supports our manufacturers tively we harvest and reap their benefits. Per-
• Access to Compressed Gas Association haps equally as important, GAWDA needs the
(CGA) training modules independent distributor.
• Build consumer safety awareness through With independent distributors, multina-
provided posters and public service announce- tional gas manufacturers, gas and welding
ment videos equipment suppliers, and technology and
• COVID-19 Risk Mitigation Roundtables service providers, GAWDA represents a sym-
• Cross-industry compensation report for biotic industry 75 years strong. WJ
benchmarking
With $25,000 in grant funding, Airport Community A welding student practices shielded metal arc welding in
Schools, Carleton, Mich., intend to expand students’ metal- preparation for the 3G bend test. (Photo courtesy of Airport
working capabilities by adding AWS advanced welder train- Community Schools.)
ing and fabrication classes. Alongside the addition of key
equipment, such as a larger bandsaw, CNC plasma table, and ed that we would like to offer our students the opportunity
sheet metal brake press, students will be able to complete to learn the trade of welding,” said Ryan Irwin, director of
the SENSE Level II Advanced Welder Certificate Program education, Airport Community Schools. “In 2018, our com-
and develop skills required by local industries. munity supported a bond, and we decided to use a portion
“Through multiple community contacts, surveys, and of the bond money to build a 5000-sq-ft facility . . . Now
state data, our district (Airport Community Schools) decid- that we have a great space, we are working within our limit-
STAINLESS Q&A
BY DAMIAN J. KOTECKI
cation of interest.
Q: American Welding Society D1.6/D1.6M:2017, Structural Welding Code — Some restraint is unavoidable in
Stainless Steel, contains two passages that are causing consternation in welded fabrication. Because the weld
our organization. In Section 4, Design of Welded Connections, Clause 4.6 metal shrinks as it cools, and that
ends with the following sentence: “Details shall promote ductile behavior, shrinkage is resisted by the surround-
minimize restraint, avoid undue concentration of welding, and afford ing cold base metal, residual stresses
ample access for depositing the weld metal.” The informative Annex G, up to the yield point can be anticipat-
Nonprequalified Stainless Steels — Guidelines for WPS Qualification and ed in and around all but very small
Use, Clause G2 Nonprequalified Austenitic Stainless Steels (6) concludes welds. These stresses produce distor-
with “Designing the weldment and the welding sequence to minimize re- tion. If distortion is not acceptable,
straint on the solidifying weld metal.” We are concerned with the phrase some applied restraint (such as strong-
“minimize restraint” in both of these sections. If we don’t restrain the backs) can be used to resist distortion,
workpieces before welding, we generally do not end up with the finished but that would appear to be counter to
weldment shape we want. Stainless steel weldments tend to distort the instruction in Clause 4.6. One way
more than carbon steel weldments of the same initial geometry. How are to minimize the use of strongbacks
we to interpret “minimize restraint” and comply with the intent of the and therefore restraint is to design for
Code? welding from two sides, whenever pos-
sible, instead of one side. In addition
A: This cannot be considered as an of- Of these, the most effective method to minimizing restraint while counter-
ficial interpretation of the American for preventing solidification cracks is ing distortion, a double-V or double-U
Welding Society (AWS) D1.6/D1.6M, the use of low heat input to produce joint preparation uses less filler metal
Structural Welding Code — Stainless convex beads and filling craters. Any than a single-V or single-U joint prepa-
Steel. Official interpretations can only time one is welding on fully austenitic ration. Therefore, a two-sided weld
be prepared by full action from the D1 stainless steels (such as 310, 320, or design, instead of a single-sided weld
Committee on Structural Welding. The 330), I would strongly recommend fol- design, minimizes restraint.
following is only my opinion as an ad- lowing those two suggestions. Considering fillet welds, Clause
visor to the D1K Subcommittee on Clause 4.6 is concerned with design 4.3.2.1 of D1.6/D1.6M:2017 states,
Stainless Steel, which is responsible for welding and uses the mandatory “Stress on the effective area of fillet
for D1.6/D1.6M. wording “Details shall . . . minimize re- welds and of welds in skewed joints
The reason for minimizing restraint straint.” There is no objective criterion shall be considered as shear stress, re-
is to minimize the likelihood of crack- that describes how that shall be done. gardless of the direction of applica-
ing. I will consider that later in this In effect, then, the engineer has to tion.” Clause 4.4.2.1 adds, “The effec-
column. again apply his or her engineering tive area shall be the effective weld
First, it is appropriate to decouple judgment, this time as to what “mini- length multiplied by the effective
the two clauses according to their ap- mize restraint” means and how that throat.” Clause 4.4.2.2 asserts, “For fil-
plication. Clause 4.6 is mandatory and can be done in the context of the fabri- let welds, the effective throat shall be
applies to all fabrication under the Code.
Clause G2 is in an informative (non-
mandatory) annex and applies specifi-
cally to fabrication of austenitic stain-
less steels that are expected to solidify
entirely as austenite (no ferrite —
steels such as 310, 320, and 330) and
are therefore somewhat susceptible to
solidification cracking.
Material provided in a nonmanda-
tory annex, such as Annex G, consists
essentially of suggestions. The engi-
neer in charge of fabrication is expect-
ed to apply his or her engineering
judgment as to whether or not a par-
ticular suggestion is appropriate for
the given fabrication. The “minimize
restraint” suggestion is just one of six
in Clause G2. The others suggest using
base metal and filler metal of low im-
purity content, using low heat input to
produce convex beads, maintaining a
low preheat and interpass tempera-
ture, using skip welding to avoid heat Fig. 1 — Fillet weld geometry (Credit: AWS A3.0M/A3.0:2020, Standard Welding Terms
and Definitions, Fig. B25 [A].)
buildup in one area, and filling craters.
the shortest distance from the joint design purposes is T × L. The volume matching filler metals are somewhat
root to the weld face of the diagram- of filler metal deposited is ½(T/0.707)2 less resistant. Least resistant are
matic weld.” × L = 1⁄4 × T2 × L, assuming no rein- martensitic stainless steels and fully
Two small fillet welds, one on each forcement. austenitic stainless steels with match-
side of a lap or T-joint connection, can If this fillet weld is replaced by two ing weld metals. Therefore, the engi-
transmit the same amount of shear fillet welds, one on either side of the neer has more latitude when designing
load as a single fillet weld twice as web, each with an effective throat = a weldment in nominally austenitic or
large on one side of the lap joint or T/2, the effective area of the two fillet duplex stainless steel than in the other
T-joint connection, and the two small welds will be 2 × T/2 × L = T × L, same types of stainless steels. If the weld-
fillet welds will use less filler metal as the single larger fillet weld. But the ment does not experience cracking, I
than the larger single fillet weld. This volume of filler metal deposited to would say the design has sufficiently
is because the shear-load-carrying ca- produce the two fillet welds is minimized restraint. WJ
pacity of a fillet weld is proportional to 2 × ½[(T/2)/0.707]2 × L = 1⁄8 × T2 × L,
the throat dimension, but the volume which is one half as much as required
of the weld metal is proportional to by the single larger fillet weld.
the square of the throat dimension — The resistance to cracking under
Fig. 1. Further, the two small fillet restraint conditions depends upon the
welds counter distortion more than type of base metal and weld metal. DAMIAN J. KOTECKI is president, Damian
the single larger fillet weld, and they Nominally austenitic stainless steel Kotecki Welding Consultants, Inc. He is a
past treasurer of the IIW and a member of
distribute the load more evenly. The base metals and their matching filler the AWS A5D Subcommittee on Stainless
use of two fillet welds, instead of one metals (such as 304L, 308L, 309L Steel Filler Metals, D1K Subcommittee on
larger fillet weld, is a design approach 316L, and 347), which solidify as pri- Stainless Steel Structural Welding, and
that minimizes restraint. mary ferrite, are most resistant to WRC Subcommittee on Welding Stainless
Steels and Nickel-Base Alloys. He is a
As seen in Fig. 1, the effective cracking. They can generally be welded past chair of the A5 Committee on Filler
throat equals 0.707 × (leg length), and under conditions of severe restraint Metals and Allied Materials and served as
the effective area equals the effective without likelihood of cracking. Almost AWS president (2005–2006). Questions
throat times the effective length of the as resistant are duplex ferritic- may be sent to Damian J. Kotecki c/o
fillet weld. Suppose, for example, that austenitic stainless steels and their ap- Welding Journal, 8669 NW 36 St., # 130,
Miami, FL 33166-6672, or via email at
the leg length is T and the effective proximately matching filler metals. [email protected].
length is L, then the effective area for Ferritic stainless steels and their
RWMA Q&A
BY ALLEN M. AGIN
B
A: This is a great question. It also rais-
es additional questions and challenges
to be considered when making consis-
tently good fastener welds on stamp-
ings with an aluminum silicate (AlSi)
coating.
First, we should touch on two dif-
ferent types of steel processing: mill
processing and in-plant processing.
Understanding these are different
processes is the key to some of the
major reasons for inconsistent weld
results.
Mill processing is the processing
done at the steel mill, prior to ship-
• Floor space/footprint chiller requirements mentioned earli- On the other hand, Fig. 6 shows ex-
er, the savings in facility investment amples of the desired HAZ and excel-
• Air filtration would have been huge. lent push-out test results when using
Rework is another major cost to high currents and very short weld
• Electrode life consider. In another case, also due to times. There is no visible HAZ around
upfront cost-saving measures, two au- the projections, or around where they
• Testing tomotive suppliers found themselves pulled.
reworking lots of parts, and paying As you can see, answering your
• Rework more than they had saved on the original question is not simple. It rais-
equipment purchase. They had chosen es many additional concerns that
• Capital investment the MFDC process to weld stampings should be taken into account to create
with AlSi coatings based on equipment consistent and reliable welds.
• Weld quality price. They didn’t recognize a variety In conclusion, one of the most im-
of issues with their in-plant processing portant exercises you should consider
• Welder repurposing (some of which I’ve touched on above), is partnering with a reputable resist-
so production welding yielded incon- ance welding machine builder who has
• Life cycle cost sistent results. Both companies were the expertise, experience, and ability
forced to implement “safety” gas metal to demonstrate a variety of different
• In-plant processing arc welds (GMAW) to their fasteners, processes when welding AlSi-coated
adding more weight to the vehicle and stampings. While this demonstration
Expanding on a few of these fac- much greater production cost due to can take place in the machine builder’s
tors, let’s start with power supply. A longer processing time, as well as addi- welding lab, this demonstration
MFDC system, including the latest tional personnel, welding gas, and should use your production parts and
“fast-rise time” type of transformer, consumables. a range of their production welding
can require up to 1000 ampere (A) of There are at least several other fac- equipment. This equipment should in-
three-phase power and a 2200 A in- tors that I don’t have the space to clude both CD and MFDC welding ma-
verter. CD processes require a small mention here. It is best to discuss all chines with all necessary production
fraction of that power. For example, of these with a resistance welding ma- features. Following this process will
one manufacturer’s CD process only chine builder who has a proven track supply you with the absolute correct
requires 30 A of single-phase 480 volts record on hot-stamped materials. De- answers for your particular applica-
alternating current. Installation costs signing your process for success tion, enabling you to consistently pro-
of the primary power supply, as well as should also include doing lab welding, duce the strongest possible welds. WJ
power cost over the life cycle of the testing, and performing a Design of
capital equipment, are substantial fac- Experiment (DoE) on your stampings Acknowledgments
tors that should be considered. before making a major investment.
Supplying chilled water can also be a I am very grateful for the assistance
significant cost to consider. For exam- I’ve been hearing about the of the following people in answering
ple, MFDC processes require significant heat-affected zone (HAZ). Why your question in more depth: Bob
water cooling for the weld control and does this matter? Kollins, at Technical Sales and Solu-
transformer. A CD welding process that tions, for support with the measure-
does not require water cooling for the The HAZ is a very important, often ments and waveforms shown in Fig. 3;
control or transformer saves at least 8 critical, factor in all resistance welds. and Min Kuo, PhD, at ArcelorMittal
gal per min (30 L per min). In high- The ideal resistance weld utilizes the Global R&D, for the microstructure
duty-cycle CD applications, a small highest possible amount of heat for imaging on our parts shown in Figs. 2
chiller might be recommended to cool the shortest possible amount of time. and 4.
the tooling. There are capital cost sav- Using a process that cannot deliver
ings in buying a much smaller chiller, or weld current in the shortest amount of
not buying a chiller at all, not to men- time can end up heating an area far ALLEN M. AGIN is the Midwest regional
tion the ongoing power savings from outside the weld zone. This may cause sales manager and product manager for
CD welding at Weld Systems Integrators
not running a large chiller. a change in hardness in the base mate- Inc., Warrensville Heights, Ohio; a sustain-
Another often overlooked factor is rial, which can lead to a material fail- ing member of AWS; and an active mem-
floor space. Some capacitive storage ure in the HAZ. Figure 5 shows what ber in RWMA and WEMCO. He has been in-
banks are simply more space efficient. appears to be a good weld based on the volved in the applications, training, and
sales of resistance welding since 1968.
United States-built CD banks are typi- number of weld nuggets, but with se- Agin is also an author and co-author of ar-
cally much smaller than those sourced vere overheating leading to a weak- ticles on the welding of fasteners to hot-
overseas. Recently, a large stamping ened HAZ. In this case, the end user stamped and press-hardened steels with
company purchased CD welding ma- was using a MFDC power supply, and AlSi coatings. He has presented to the
chines based on price alone to keep experiencing much lower push-out American Institute of Steel Engineers, AWS
Sheet Metal Welding Conferences, and
the project cost down. The CD banks values than expected. Simply put, the most recently at the PMA Hot Stamping
were three to five times the size of the metallurgy changed and the part was Experience & Tech Tour. Send your com-
domestic units they didn’t purchase. weakened due to excessive heating. ments and questions to Allen Agin c/o
The overall footprint of each machine Had the HAZ been much smaller, Welding Journal, 8669 NW 36 St., #130,
Miami, FL 33166-6672, or via email at
was almost double what it could have push-out values would have been [email protected].
been. Put together with the smaller much higher.
Fiber Laser Facilitates Welding als. The fiber laser comes with an out-
of Thin, Conductive Metals put power of 4 kW (a 1.5-kW center
beam and a 2.5-kW ring beam that can
The HighLightTM FL4000CSM-ARM be varied and modulated). It is also
fiber laser delivers a dual-beam output equipped with a 15-m length process
to enable the welding of thin, high- fiber, making it easy to integrate in
thermal conductivity metals. The dual any production environment. The
beam, which consists of a center beam fiber laser meets welding require-
surrounded by a ring beam, offers a ments for e-mobility; energy storage;
high brightness (nominal BPP 0.6) to and general electrical interconnects
allow for a smaller spot size and mini- consisting of dissimilar materials, such
mize heating of the material. The small as copper and aluminum, and foil
spot size, coupled with the preheating stacks that require precise control.
effect from the ring beam, improves
Coherent®
weld consistency, which is a challenge
coherent.com
when welding thin, conductive materi- (800) 527-3786
welders more options depending upon to increase at a compound annual materials, production shutdown, cash
weld thickness, the welding wire growth rate of 5%. The 111-page report flow constraints, and import/export),
comes in the following diameters and attributes this growth to the thermal market demand (restrictions/lockdowns,
package sizes: 3⁄32 or 1⁄8 50-lb coil, and 3⁄32 stability provided by laser welding ma- consumer sentiments), and pricing.
or 1⁄8 600-lb Speed-Feed® drum. The chines, which helps in reducing scrap
coil is light enough for one person to metal, as well as their inclusion in var- Market Research Future®
lift and install on a wire feeder reel. ious end-user applications, such as marketresearchfuture.com
The drum offers a larger, bulk package medical, jewelry, automotive, and elec- (628) 258-0071
for high-production environments. tronics. The report also forecasts that
product innovations, advances in man-
The Lincoln Electric Co. ufacturing processes, and technologi- Laser Welding System Offers
lincolnelectric.com cal improvements will continue to a Large Work Area
(888) 935-3877 drive the demand for laser welding
machines. The report concludes with The TruLaser Station 7000 3D laser
an analysis on the impact of the welding system exhibits a beam with a
Consumables Cut Down on COVID-19 pandemic on this industry. laser power of up to 4 kW and a work
Piercing Time for Thick Steel It covers the following topics: econom-
ic, supply chain (price variation of raw — continued on page 57
Thermal Dynamics
thermal-dynamics.com
(866) 279-2628
A
vast part of much marine and
offshore construction, such as
the supporting structures of
platforms and wind towers as well as
pipelines, manifolds, steel waterway
and harbor construction, bulkhead
panels, locks, ship hull repairs, and
more, are to a great extent done under
water. Hence, welding and cutting
activities for maintenance and repair
are bound to take place under water
as well.
Shielded metal arc welding (SMAW)
is a versatile, flexible, and practical
welding process. For this reason, it
is often used for underwater mainte-
nance and repairs. For more than
80 years, this process has been applied
for sealing leaking rivets in riveted
ship hulls. A power source and a cov-
ered electrode as the consumable is al-
most all that is required for this type
of SMAW operation. In this case, the
power source has to be adapted and
Fig. 1 — Welding for qualification test in tank with full diving gear. (Courtesy of the prepared for underwater welding to
Royal Dutch Navy — Diving Group.) meet all safety requirements. The
Since all these applications fall un- Navy. Figure 9 shows the welder/diver
der certain regulations and approvals, with insulating rubber gloves and get-
such as those governed by the Ameri- ting ready to plunge, standing behind
can Bureau of Shipping (ABS), DNV- an underwater workbench with a
GL, or Lloyd’s Register (LR), the weld- clamping device on a platform that can
ing procedure qualifications and the be lowered into the water for practic-
welder performance qualifications ing purposes. These welders have also
have to be carried out in accordance performed underwater cutting with
and under the supervision of these ap- the same electrodes.
proval agencies per their regulations. Using 240 A, a 10-mm plate could
easily be cut into many parts. The cut-
Putting it to the Test ting length is up to 200 mm with one
electrode. The electrode is 3.2 mm in
The Dutch Navy qualified using diameter and 450 mm long. The elec-
UTP Nautica 20 electrodes under LR trode length of 450 mm made it possi-
for welding patches or lap joints onto ble to distinguish them from the re-
naval ship hulls, in case the material pair electrodes used, which were 350
thickness has reduced due to corrosion mm. The welder can easily carry both
or other damage (Ref. 4). To qualify types to be prepared for all applica-
the wet welding procedures and the tions under water. WJ
welders, fillet welds were executed in
plates of 10 mm thickness. This took Acknowledgments
place in an indoor diving tank filled
with 3% saltwater to simulate marine The authors thank Gerrit Weer-
seawater conditions, as shown in stand and his colleagues of the Royal
Fig. 5. Dutch Navy — Diving Group for their
Part of the qualification welds were valuable contributions and practical
fillet welds that were produced in dou- insight on underwater welding as dis-
ble lap joints, prepared on one side cussed in this paper. Thanks also go to
with full fillet welds made in the work- Fred Neessen, an international weld-
shop. The other side was welded in the ing engineering consultant, for the ex-
tank with a 3.2-mm electrode at about pressive graphics (Figs. 2 and 3).
150 A in the downhill position. The References
test weld was a smaller size in order to
test if the specimen would break in the 1. Szelagowski, P. 2015. Unterwa-
weld or in the base material that was teterwasser-Schweißtechnik, Grundlagen
to be tested. - Forschung - Anwendung. DVS Media,
Figure 6 shows the double lap joints Düsseldorf, Germany.
with one side welded in the shop (left) 2. Keats, D. J. 1990. Professional
and the other side welded under water Diver’s Manual on wet-welding. Abing-
(right) showing the specimen broke in ton Publishing, Cambridge, England.
the base material or on the side of the 3. Hilkes, J., and Tuchtfeld, J. 2019.
workshop welds. The result of the Unterwasserschweißen mit umhüllten
multipass fillet weld in downhill posi- Stabelektroden für Schiffbau- und Off-
tion is shown in Fig. 7. The bead ap- shore-Anwendungen, DVS - Große
pearance was very good, especially Schweißtechnische Tagung, 16/17
considering the lack of visibility of the September, DVS Reports, Vol. 355,
welder. The cross section of this weld p. 385-397. DVS Media, Düsseldorf,
in Fig. 8 shows joint penetration, Germany.
which is one of the qualification re- 4. Bos, R. 2001. De kwaliteit van
quirements, proven by a break test een onderwatergelaste verbinding met
that shows the amount of penetration beklede electrode. NIL Lastechniek,
per length of weld. Oktober 2001: 4-8.
Practice Makes Perfect
Open-water training sessions are
organized for the entire underwater
diving and welding team of the Dutch
BY ANDREW PFALLER
Using the right shielding gas and following best practices can
help you save money
U
sing shielding gas flow rates that GTAW process requires an inert at- nozzle, which may restrict the opera-
are too high for gas tungsten arc mosphere to protect the tungsten tor’s ability to complete the weld.
welding (GTAW) applications (where the arc is generated) and, more
can double gas costs, resulting in up- importantly, the molten pool of the Adjusting the Shielding
wards of $3300 in added annual costs metal being welded — Fig. 2. One con- Gas Rate
per welder. Many operations may not sumable used for this in many GTAW
realize there is huge potential for im- applications is a collet body, which To overcome these quality or access
mediate payback and reduced costs by provides shielding gas coverage over issues, operations may take the com-
using a better gas consumable. the part being welded. mon step of increasing the shielding
Improper shielding gas flow can Companies often choose collet bod- gas. However, this only exacerbates
drive up costs through wasted gas and ies because they are a less expensive problems because increasing the flow
added rework. When there is a lack of method of delivering the shielding gas rate increases the instability in the gas
knowledge about shielding gas best compared to other options. However, column. Increasing the flow rate also
practices, operations might also use collet bodies can be the source of prob- causes more shielding gas to be con-
temporary solutions that don’t ad- lems that end up costing time and sumed than before — a point that is
dress the root issue and simply push money — much more than the initial often overlooked.
these hidden costs higher. small amount saved up front. This can drive up costs in the fol-
This article will help you learn the When using a standard collet body lowing three ways:
importance of using the right shield- in GTAW, a considerable amount of • Expenses associated with the
ing gas consumables and how follow- turbulence and atmospheric contami- consumable shielding gas. A 300-ft3
ing best practices can reduce opera- nation are introduced into the stream. tank costs on average $35, or $0.11 per
tional costs. This can cause issues such as porosity, ft3. If an operation uses one tank per
base metal oxidation (also called sug- welder per week, that’s $1750 per year
Standard Collet Bodies for aring on stainless steel), poor welding spent on shielding gas. Some opera-
GTAW performance, and arc flutter — prob- tions studied have used as much as 60
lems that are unacceptable in applica- ft3/h of gas in a high duty cycle appli-
Many operations use GTAW in tions that require high quality. cation, with approximately 50% arc-on
high-profile applications where the The turbulence in the shielding gas time. This would mean shielding gas
greatest weld aesthetics, quality, and flow also limits how far the tungsten costs are about $6600 per welder shift
integrity are required — Fig. 1. The can be extended past the end of the per year. If that company had 25
Shielding Gas:
Lower Your Costs by
Eliminating Inefficiencies
BY DAVID GAILEY
R
emote laser welding was made are joined in the middle. This weld
popular at the turn of the centu- quality problem is often referred to as
ry. The ability to direct a beam false friend — Fig. 1. To overcome this
of light to the workpiece and join ma- problem, the part designer would de-
terials without contact promised to sign the part with more welds than it
dramatically reduce cycle time and actually needed to statistically prevent
lower manufacturing costs. However, part failure if a few false friends were
the technology was not widely adopt- present. Again, this process leads to
ed. Due to the inherently narrow laser excessive part weight, additional cycle
welds, and the large part-to-part toler- time, and increased part cost.
ance of the workpiece, products not What if there were a way to no
only had to be designed specifically for Fig. 1 — False friend — cross section longer worry about part location, and
the remote laser welding process, but shows no weld between the upper position the laser beam exactly on the
fixturing prices increased to better lo- and lower sheets. edge of the part? This would change tra-
cate parts. Ultimately, this increase in ditional lap welds to edge fillet welds.
both part and system cost limited the tion the beam, according to the model. With edge fillet welds, flanges can be re-
financial benefit of decreased cycle However, the real problem is that parts duced, part weight is lowered, part cost
time to only high-volume applications. are rarely placed in the same exact posi- is reduced, and inspectors can see the
In other words, if you were making tion. This inaccuracy of the parts re- top, bottom, and middle of the weld.
hundreds of thousands of the same sulted in the use of lap welds, which What if we no longer needed robust,
exact part per year, the financial justi- placed several millimeters of additional over-designed fixtures to push parts
fication could be made. material to the edge of the part to en- into exact locations and remove small
The challenge of broader industrial sure the weld did not miss the upper gaps* to prevent false friends? Both of
adoption of remote laser welding did sheet. This larger flange subsequently these things are now possible through
not stop there. The bigger problem increases the part weight and cost. utilization of an approach referred to as
was the beam positioning process. The The next problem with the process adaptive remote laser welding.
typical scanner optic is programmed is related to weld quality. With a lap Adaptive remote laser welding tech-
to position the laser beam based on a weld, an inspector can see the weld on nology provides a direct look at the
software coordinate system or comput- the top sheet. In the case of a com- workpiece; determines the location of
er-aided design model. The weld loca- plete joint penetration weld, the in- the edge that needs to be joined, if a
tions are loaded into the computer- spector can also see the weld on the gap is present between the two sheets;
controlled coordinate system, which in bottom of the lower sheet. What they makes on-the-fly adjustments to the
turn adjusts the scanner mirror to posi- cannot be sure of is if the two sheets process parameters; and positions the
*Note: The terms gap and gap bridging are typically referred to as sheet separation. The American Welding Society (AWS)
A3.0M/A3.0:2020, Standard Welding Terms and Definitions, defines sheet separation as “The distance between faying sur-
faces adjacent to the weld once a spot, seam, or projection weld has been produced.”
Laser diode hits the measurement object and is Shape Recognition Cameras
diffusely reflected at an angle such
that it hits a cloud of individual, more Shape recognition technology is a
Line
CMOS sensor
or less bright points of light, onto the comparative process that looks for a
generator camera chip. The individual pixels are shape on the workpiece that matches a
array
filtered and summarized in an applica- predefined shape stored in the soft-
tion-specific integrated circuit chip. ware. The shape is located by the im-
Filters, both software and hardware age processing camera using a technol-
related, will remove the reflections ogy called dark field illumination. This
and other light influences. The data is process uses external lights that flood
then calculated by the seam tracking the part’s surface. When an image is
controller, and the position of the captured, the top surface appears light
Diffuse points of light on the chip are convert- gray while features below the surface,
reflected ed to positional information as it re- (such as holes, part edges, and slots)
laser light lates to the process. A profile of the appear to be darker. This dark image is
joint is assembled and the results are then compared with the stored shape,
linearized and compared with those of and a positive shape recognition is
the customer’s chosen seam profile. determined.
Other supporting parameters are cal- Once the shape is recognized, the
culated with respect to the profile and location of the shape on the image can
are then compared with the defined be determined based on the pixel loca-
values of the parameters set within tion. These parameters are then sent
Fig. 2 — Laser triangulation for optical the motion-control side of the process. to the scanning mirrors, which direct
seam tracking. Offset data is determined and then the laser beam to the correct location
output to the machine control, result- of the actual part being processed —
beam in the exact location required to ing in a modified motion path accord- Fig. 4.
meet weld quality standards. This ing to the offsets found during the Both technologies, laser triangula-
adaptive process is possible by seam tracking process. tion seam tracking and shape recogni-
tracking to position the beam via laser The information output from the tion, are made possible by ultra-fast 1
triangulation sensors or shape recog- sensor package will include informa- GHz on-board communication and
nition cameras. tion such as Y position, Z position, gap, high-powered scanner mirrors capable
mismatch, and angles about the sensor of 1000 Hz. This combination allows
Laser Triangulation datum. Analysis of the data provided process speeds of 1 m/s, which is neces-
Seam Tracking through software algorithms can use sary to make the highly dynamic reori-
the data to not only track the seam but entation movements needed to adapt
With optical seam tracking sensors, also adjust process parameters such as to real-time tracking requirements.
a laser beam is emitted from a diode laser power, spot size, Y-offset, ampli- When the two technologies are
integrated either into the unit or as tudes, and oscillation frequencies of combined into one tool, a new manu-
part of a line generator package — the beam. When combining the control facturing reality becomes possible: the
Fig. 2. This beam is converted by an of all of these process parameters, a replacement of resistance spot weld-
optical line generator into one or mul- true adaptive weld process is created to ing. Resistance spot welding has been
tiple (i.e., three) lines depending on accommodate the variation in the the go-to method for joining stamp-
the make of the sensor. The laser line parts — Fig. 3. ings in many industries, especially the
Fig. 3 — Adaptive remote laser welding with gap* bridging. Fig. 4 — Adaptive shape recognition.
CERTIFICATION SCHEDULE
CERTIFICATION SEMINARS, CODE CLINICS, AND EXAMINATIONS
Note: The 2020 schedules for all certifications are posted online at Location Seminar Dates (cont.)
awo.aws.org/instructor-led-seminars/seminar-exam-schedule/. Sacramento, CA Oct. 4–9
Dallas, TX Oct. 11–16
Denver, CO Oct. 25–30
Certified Welding Inspector (CWI) St. Louis, MO Nov. 1–6
Seminar covers Parts A, B, and C of the CWI exam. Only Part B
New Orleans, LA Nov. 8–13
of the exam is taken following the conclusion of the seminar.
Miami, FL Dec. 6–11
Parts A and C are given at Prometric testing centers.
Phoenix, AZ Dec. 6–11
Location Seminar Dates Part B Certified Welding Educator (CWE)
Exam Date Seminar and exam are given at all sites listed under Certified
Miami, FL Sept. 13–18 Sept. 19 Welding Inspector. Seminar attendees will not attend the Code
Houston, TX Sept. 13–18 Sept. 19 Clinic portion of the seminar (usually the first two days).
Seattle, WA Sept. 13–18 Sept. 19
Benicia, CA Sept. 20–25 Sept. 26 Certified Welding Sales Representative (CWSR)
Louisville, KY Sept. 20–25 Sept. 26 CWSR exams are given at Prometric testing centers. More in-
Minneapolis, MN Sept. 20–25 Sept. 26 formation at aws.org/certification/detail/certified-welding-sales-
New Orleans, LA Sept. 27–Oct. 2 Oct. 3 representative.
Boston, MA Sept. 27–Oct. 2 Oct. 3
Nashville, TN Sept. 27–Oct. 2 Oct. 3
San Antonio, TX Sept. 27–Oct. 2 Oct. 3 Certified Resistance Welding Technician
Miami, FL Oct. 4–9 Oct. 10 (CRWT)
Long Beach, CA Oct. 11–16 Oct. 17 A comprehensive two-day seminar to arm attendees with
Detroit, MI Oct. 11–16 Oct. 17 the knowledge needed to take the exam with confidence.
Tulsa, OK Oct. 11–16 Oct. 17 More information at awo.aws.org/crwt/.
Atlanta, GA Oct. 18–23 Oct. 24
Las Vegas, NV Oct. 18–23 Oct. 24 Location Seminar Dates
Houston, TX Oct. 18–23 Oct. 24 Nashville, TN Oct. 27–28
Kansas City, MO Oct. 25–30 Oct. 31
Pittsburgh, PA Oct. 25–30 Oct. 31
Orlando, FL Nov. 1–6 Nov. 7 Certified Welding Supervisor (CWS)
Indianapolis, IN Nov. 1–6 Nov. 7 CWS exams are given at Prometric testing centers. More infor-
Reno, NV Nov. 1–6 Nov. 7 mation at aws.org/certification/detail/certified-welding-supervisor.
Dallas, TX Nov. 1–6 Nov. 7
Sacramento, CA Nov. 8–13 Nov. 14 Location Seminar Dates
Miami, FL Nov. 8–13 Nov. 14 Atlanta, GA Dec. 6–11
Syracuse, NY Nov. 8–13 Nov. 14
Cleveland, OH Nov. 8–13 Nov. 14
Waco, TX Nov. 15–20 Nov. 21 Certified Radiographic Interpreter (CRI)
Norfolk, VA Nov. 15–20 Nov. 21 The CRI certification can be a stand-alone credential or can
San Diego, CA Dec. 6–11 Dec. 12 exempt you from your next 9-Year Recertification. More in-
Miami, FL Dec. 6–11 Dec. 12 formation at aws.org/certification/detail/certified-radiographic-
St. Louis, MO Dec. 6–11 Dec. 12 interpreter.
Certified Welding Inspector (CWI) Part B Location Seminar Dates Exam Date
Course covers only Part B of the CWI exam. The Part B exam Houston, TX Oct. 5–9 Oct. 10
follows the conclusion of the three-day course.
Certified Robotic Arc Welding (CRAW)
Location Seminar Dates Part B OTC Daihen Inc., Tipp City, OH; (937) 667-0800, ext. 218
Exam Date Lincoln Electric Co., Cleveland, OH; (216) 383-4723
Minneapolis, MN Oct. 21–23 Oct. 24 Wolf Robotics, Fort Collins, CO; (970) 225-7667
Louisville, KY Dec. 9–11 Dec. 12 Milwaukee Area Technical College, Milwaukee, WI;
(414) 456-5454
9-Year Recertification Seminar for CWI/SCWI College of the Canyons, Santa Clarita, CA; (661) 259-7800,
For current CWIs and SCWIs needing to meet education ext. 3062
requirements without taking the exam. Ogden-Weber Applied Technology College, Ogden, UT;
(801) 627-8448
Location Seminar Dates Genesis Systems IPG Photonics Co., Davenport, IA;
Orlando, FL Sept. 13–18 (563) 445-5688
IMPORTANT: This schedule is subject to change without notice. Please verify your event dates with the Certification Dept. to confirm your course status before
making travel plans. Applications are to be received at least six weeks prior to the seminar/exam or exam. Applications received after that time will be assessed a
$395 Fast Track fee. Please verify application deadline dates by visiting our website at aws.org/certification/docs/schedules.html. For information on AWS seminars
and certification programs, or to register online, visit aws.org/certification or call (800/305) 443-9353, ext. 273 for Certification; or ext. 455 for Seminars.
WELDING WORKBOOK
DATASHEET 399
Excerpted from the Welding Handbook, Ninth Edition, Volume 3, Welding Processes, Part 2.
SPPA
ARKING CONNEC
CTIONS –
2020 AW
WS MEMBERSHIPP CHALLENGE
What
h it is:
Be the spark
p that ignites
g s the people you know to
beecome AW
WS members, and get rewards.
Ho
ow it woorks:
Buuild up points throug hout the year for each
AWS
W Member you recr uit: 5 Points per Individual
Meembership and 1 Pointt per Student Membership.
What
W you get:
Grand Prize Winnner: $200 gift card
2nd Plaace: $100 gift card
3rd Plaace: $100 gift card
4th Plaace: $100 gift card
Alll other participants earnning 10 or more points will get
AWS
W branded merchandisse based on points accrued
thrroughout 2020.
The Fine Print: All AWS members in good standinng may participate and are eligiblle to receive rewards based on points accrued
Januarry 1 – December 31 31, 2020
2020. Paarticipant eligibillity is determined at the sole discre
retion of AWS program administrators. AWS staff
members and administrators of commercial / corrporate or educational packaagges that inccllude AWS memberships in the pricing
structure are not eligible to participate. Foor more innformation, visit aws.org/be-the-sppark
8669 NW 36 ST. #130 | Miami, FL 33166-6672
American Welding Society®
MEMBERSHIP
T: 800.443.9353 | F: 305.443.5647 | aws.org Application for Membership
PRIMARY CONTACT INFORMATION (Please Print)
Mr. Ms. Mrs. Dr.
Last Name: First Name: M.I: Birthdate:
Renewing Member
1 Year - Digital Welding Journal ......................................... $88 2 Year - Digital Welding Journal ....................................... $171
1 Year - Print and Digital Welding Journal ........................ $138 2 Year - Print and Digital Welding Journal ........................ $271 $
NEW MEMBER OPTIONAL BOOK SELECTION (Not available to renewals. Choose ONE option ONLY. Domestic .................... $35
Includes shipping & handling.) Visit aws.org/memberships/page/new-member-book-offer to view International ............... $85 $
selections and write your choice here:
TOTAL $
Business (Circle ONE Letter Only)
A Contract Construction F Machinery Except Electric J Transport Equip. — Boats, Ships O Educational Services R Government (Federal, State, Local)
B Chemicals & Allied Products (Incl. Gas Welding) K Transport Equip. — Railroad (Univ., Libraries, Schools) S Other
C Petroleum & Coal Industries G Electrical Equipment, Supplies, L Utilities P Engineering & Architectural
Electrodes M Welding Distributors & Retail Trade Services (Including Assns.)
D Primary Metal Industries H Transport Equip. – Air, Aerospace N Misc. Repair Services Q Misc. Business Services
E Fabricated Metal Products I Transport Equip. — Automotive (Including Welding Shops) (Including Commercial Labs)
Job Classification (Circle ONE Letter Only)
01 President, Owner, Partner, 04 Purchasing 10 Architect Designer 08 Supervisor, Foreman 15 Educator
Officer 05 Engineer — Welding 12 Metallurgist 14 Technician 17 Librarian
02 Manager, Director, 20 Engineer — Design 13 Research & Development 09 Welder, Welding or Cutting 16 Student
Superintendent (Or Assistant) 21 Engineer — Manufacturing 22 Quality Control Operator 18 Customer Service
03 Sales 06 Engineer — Other 07 Inspector, Tester 11 Consultant 19 Other
PAYMENT INFORMATION
Payment can be made (in U.S. dollars) by check or money order (international), payable to the American Welding Society.
Check Money Order
AMEX Diners Club MasterCard Visa Discover Other Application Date:
Name on Card: CC#: / / /
Expiration Date (mm/yy): / CVV: Applicant Signature:
American Welding Society (AWS) case. The reception from users was rials for metal arc welding, in 2014.
A5.36/A5.36M, Specification for Carbon that the specification itself, and the It is recognized that there will be an
and Low-Alloy Steel Flux Cored Electrodes classification designations in particu- impact on the welding community as
for Flux Cored Arc Welding and Metal lar, were too complex. A5.36 provided a result of the decision to withdraw
Cored Electrodes for Gas Metal Arc Weld- a lot of flexibility in how electrodes A5.36, but the clear message was to
ing, has been withdrawn as an AWS were classified — in the as-welded or move back to the previous specifica-
specification. It has also been with- postweld heat-treated condition, with tions as quickly as possible. To aid
drawn as a national specification un- various shielding gases, various users in this transition, three different
der the American National Standards strength levels, and various impact options are listed below. The most ap-
Institute (ANSI). A withdrawn stan- test temperatures. To accommodate propriate one may be selected based
dard is one that has been officially re- that flexibility, the classifications were on the consumable, the application,
moved from publication. It will not be quite complex and did not exactly “roll and the code or contract.
revised and will not be cited in new off the tongue.” The A5.36 designation 1) The most widely used cored wire
AWS standards or publications. This system is shown below. classifications from A5.18 and A5.20
decision, which was not taken lightly have always been included in A5.36.
by the A5 committees or the AWS These E70T- and E71T- classifications
Technical Activities Committee, was still account for the vast majority of
done in response to a clear direction flux-cored and metal-cored electrodes
from the welding community. used today. In these cases, no changes
AWS A5.36 was a classification to existing documentation should be
standard for flux cored arc welding necessary. The classifications will re-
and metal-cored arc welding elec- main the same as in the existing speci-
trodes. The specification was broad in fications, with metal-cored electrodes
scope and covered the following: classified in A5.18 and flux-cored
• All flux-cored electrodes classified electrodes classified in A5.20. These
to AWS A5.20/A5.20M, Specification classifications are listed below. In
for Carbon Steel Electrodes for Flux these cases, the only change that
Cored Arc Welding (E71T-1C, E71T-8, might be necessary is to reference
E71T-1M, etc.) Because of the lack of acceptance by A5.18 or A5.20, rather than A5.36.
• All flux-cored electrodes classified much of the industry, filler metal man-
to AWS A5.29/A5.29M, Specification ufacturers continued to classify elec-
for Low-Alloy Steel Electrodes for Flux trodes to A5.20, A5.29, A5.18, and
Cored Arc Welding (E80T1-Ni1M, A5.28, in addition to A5.36. One prod-
E70T8-Ni2, E91T1-K2C, etc.) uct often had multiple classifications,
• All metal-cored electrodes classi- which complicated the documentation
fied to AWS A5.18/A5.18M, Specifica- on qualification and welding proce-
tion for Carbon Steel Electrodes and Rods dure records and led to confusion
for Gas Shielded Arc Welding (E70C-6M, about which classification to use as an
E70C-3C, etc.) essential variable. Certificates of con-
• All metal-cored electrodes classi- formance could be many pages long,
fied to AWS A5.28/A5.28M, Specifica- and the print on the product labels be-
tion for Low-Alloy Steel Electrodes and came very small.
Rods for Gas Shielded Arc Welding
(E120C-K4, E80C-Ni1, etc.) How to Transition from
2) For electrodes other than these,
A5.36 Back to A5.18, mainly those electrodes used for weld-
Why A5.36 was A5.20, A5.28, or A5.29 ing low-alloy steels, the most efficient
Withdrawn method is to record the A5.18, A5.20,
AWS A5.36 was incorporated into A5.28, or A5.29 designations on
AWS A5.36 was first published in the following documents: AWS D1.1/ records. Qualification and welding
2012 with the thought that welding D1.1M, Structural Welding Code — Steel, procedure records from before 2012
consumable manufacturers and users in 2015; American Society of Mechani- will already carry only these refer-
would gravitate to this specification, cal Engineers, Boiler & Pressure Vessel ences. Records from after 2012 might
and AWS A5.20 and A5.29 could be Code, Section II, Part C, in 2013; and carry both the A5.36 electrode classifi-
withdrawn. However, that was not the CSA W48, Filler metals and allied mate- cation and the classification from
SOCIETY NEWS
A5.18, A5.20, A5.28, or A5.29, since Additional Consumable can be used to specify the details of
these designations were retained on testing, the acceptance criteria, and
most product and certification docu- and Weld Metal Properties whether the testing be done on a spe-
ments by the electrode manufacturers. cific heat or lot of electrode. Terms like
3) For records created since 2012 A5.36 contained provisions to clas- “heat,” “lot,” and “certified material
that carry only an A5.36 classification sify electrodes under a wide variety of test report” are defined so there is no
designation, a bit more detail is neces- welding and testing conditions. Some- confusion regarding the level of test-
sary. Tables B.1, B.2, and B.3 from times information is needed about an ing, the lot classification, or the docu-
A5.36 show equivalents from A5.36 electrode or its weld metal properties, mentation that will be supplied. A5.01
and the corresponding A5.18, A5.20, such as impact testing at low tempera- is a valuable tool for procurement of
A5.28, and A5.29 classifications. If tures or the properties of weld metal any consumables with nonstandard
the A5.36 classification recorded on deposited with a specific shielding gas. testing or reporting requirements and
the qualification or welding procedure In these cases AWS A5.01, Procurement can be used to fill any gaps in informa-
is not listed there, the electrode manu- Guidelines for Consumables — Welding tion that may come about from the
facturer should be consulted to deter- and Allied Processes — Flux and Gas withdrawal of A5.36. WJ
mine the appropriate designation. Shielded Electrical Welding Processes,
TERESA A. MELFI ([email protected]) is a Technical Fellow at The Lincoln Electric Co., Cleveland, Ohio, and chair of the AWS A5
Committee on Filler Metals and Allied Materials.
The Prof. Koichi Masubuchi award, materials joining through research and dation from fellow researchers to Prof.
with a $5000 honorarium, is present- development. Todd Palmer, [email protected]. This
ed to one person, 40 years old or Send a list of your candidate’s expe- award is sponsored annually by the
younger, who has made significant rience, publications, honors, awards, Massachusetts Institute of Technolo-
contributions to the advancement of and at least three letters of recommen- gy, Dept. of Ocean Engineering.
SOCIETY NEWS
TECH TOPICS
Opportunities to Contribute to Committee (C, E, G). Hybrid weld- standards be open to public review for
AWS Technical Committees ing, C7D Subcommittee (G). Welding comment during the approval process.
of machinery and equipment, D14 This column also advises of ANSI ap-
Committee (C, E, G, U). proval of documents.
The following committees welcome
M. Diaz, [email protected], ext. 310. A5.29/A5.29M:20XX, Specification
new members. Some committees are
Resistance welding, C1 Committee for Low-Alloy Steel Electrodes for Flux
recruiting members with specific inter-
(C, E, G, U). Friction welding, C6 Cored Arc Welding. Revised Standard.
ests in regard to the committee’s scope,
Committee (C, E). Automotive weld- $36.00. ANSI public review expired
as marked below: Producers (P), Gener-
ing, D8 Committee (C, E, G, U). Re- 8/31/2020. Contact: R. Gupta, ext. 301,
al Interest (G), Educators (E), Consult-
sistance welding equipment, J1 [email protected].
ants (C), and Users (U). For more infor-
Committee (C, E, G, U). Welding in D17.3/D17.3M:20XX, Specification
mation, contact the staff member list-
the aircraft and aerospace indus- for Friction Stir Welding of Aluminum
ed or visit aws.org/library/doclib/
try, D17 Subcommittee (C, E, G). Alloys for Aerospace Applications. Re-
Technical-Committee-Application.pdf.
S. Hedrick, [email protected], ext. vised Standard. $38.00. ANSI public
S. Borrero, [email protected], ext.
305. Metric practice, A1 Committee review expired 8/24/2020. Contact: M.
334. Definitions and symbols, A2
(C, E). Mechanical testing of welds, Diaz, [email protected], ext. 310.
Committee (E). Titanium and zirco-
B4 Committee (E, G, P). Joining of J1.3/J1.3M:20XX-AMD1, Specifica-
nium filler metals, A5K Subcommit-
plastics and composites, G1 Com- tion for Materials Used in Resistance
tee. Piping and tubing, D10 Com-
mittee (C, E, G). Safety and health Welding Electrodes and Tooling. Addenda.
mittee (C, E, U). Welding practices
committee, SHC Committee (E, G). $27.00. ANSI public review expired
and procedures for austenitic
Welding in sanitary applications, 8/17/2020. Contact: M. Diaz, ext. 310,
steels, D10C Subcommittee. Alu-
D18 Committee. [email protected].
minum piping, D10H Subcommittee.
J. Rosario, [email protected], ext.
Chromium molybdenum steel pip-
308. Procedure and performance
ing, D10I Subcommittee. Welding of
qualification, B2 Committee (E, G). New Standards Approved by
titanium piping, D10K Subcommit-
Thermal spraying, C2 Committee ANSI
tee. Purging and root pass welding,
(C, E, G, U). Oxyfuel gas welding
D10S Subcommittee. Low-carbon
and cutting, C4 Committee (C, E, G). C3.9M/C3.9:2020, Specification for
steel pipe, D10T Subcommittee. Or-
Welding iron castings, D11 (C, E, G, Resistance Brazing. Approval Date:
bital pipe welding, D10U Subcom-
P, U). Railroad welding, D15 (C, E, G, 7/9/2020.
mittee. Duplex pipe welding, D10Y
U). Robotic and automatic welding,
Subcommittee. Joining metals and
D16 Committee (C, E).
alloys, G2 Committee (E, G, U). Reac-
tive alloys, G2D Subcommittee (G). Revised Standard Approved
R. Gupta, [email protected], ext. 301. by ANSI
Filler metals and allied materials, Technical Committee Meetings
A5 Committee (E). Magnesium alloy
All AWS technical committee meet- A5.28/A5.28M:2020, Specification
filler metals, A5L Subcommittee.
ings are open to the public. Contact for Low-Alloy Steel Electrodes and Rods
P. Portela, [email protected], ext.
the staff members listed or call for Gas Shielded Arc Welding. Approval
311. Additive manufacturing, D20
(800/305) 443-9353 for information. Date: 6/30/2020.
Committee (C, E, G). The D1N Sub-
The Fall 2020 D1 Committee and A5.34/A5.34M:2020, Specification
committee on Titanium Structures is
Subcommittees on Structural Welding for Nickel-Alloy Flux Cored and Metal
recruiting all interest groups.
meetings will occur via teleconferences Cored Welding Electrodes. Approval
J. Molin, [email protected], ext. 304.
at the discretion of the D1 Task Groups, Date: 6/30/2020.
Structural welding, D1 Committee
D1 Subcommittees, and D1 Main C3.8M/C3.8:2020, Specification for
(E). Sheet metal welding, D9 Com-
Committee Chairs. Contact: J. Molin, the Ultrasonic Pulse-Echo Examination
mittee (C, G). The D1 Committee is re-
[email protected], ext. 304. of Brazed Joints. Approval Date:
cruiting educators. The D9 Committee
7/9/2020.
is recruiting producers, consultants,
and general interest members.
K. Bulger, [email protected], ext. Standards for Public Review
306. Methods of weld inspection, Amendment Standard
B1 Committee (C, E). Brazing and AWS was approved as an accredited Approved by ANSI
soldering, C3 Committee (C, E, G). standards-preparing organization by
Welding in marine construction, the American National Standards In- D1.4/D1.4M:2020-AMD1, Structur-
D3 Committee (C, E, G, U). High en- stitute (ANSI) in 1979. AWS rules, as al Welding Code — Steel Reinforcing
ergy beam welding and cutting, C7 approved by ANSI, require that all Bars. Approval Date: 7/14/2020.
SOCIETY NEWS
MEMBERSHIP ACTIVITIES
SOCIETY NEWS
Richards Industries Scenic Ridge Co.
4 Fairfield Crescent 2550 Steelton Rd.
Change of Address? Moving?
West Caldwell, NJ 07006 Lancaster, PA 17601
Make sure delivery of your
Welding Journal is not interrupted.
RWP Inspections Ltd. Solid Form Fabrication Inc.
Contact Kim Hugley in the
3995 Hi Mount Dr. P.O. Box 119
Membership Department with your
Victoria, BC 2706 NE Rivergate St.
new address information — (800)
V9C 3X9 Canada McMinnville, OR 97128
443-9353, ext. 204; [email protected].
American Welding Society's District 17 held its annual conference on June 5–6 at the Hampton Inn & Suites in Tyler, Tex. Conference attendees
included District 17 Director J Jones, Bryan Baker, Andy Stormer, Paul Wittenbach, Jim Bridwell, Paul Stanglin, and Donnie Williams. Pictured are
conference attendees and their wives.
(From left) Debra and Paul Stanglin of the AWS North Texas Section
AWS District 17 Director J Jones (right) presented Paul Wittenbach with presented AWS District 17 Director J Jones with a metal-crafted yellow
the District CWI of the Year Award. rose of Texas in appreciation of his service.
SOCIETY NEWS
AWS Member Profile
District 7
District 2 Roger E. Hilty, director
Harland Thompson, director (740) 317-9073
(631) 546-2903 [email protected]
[email protected]
READING District 9
June 20 Michael Skiles, director
Location: Womelsdorf, Pa. (337) 501-0304
Presenter: Lisa Davis, Section scholar- [email protected]
ship chair
Summary: The Reading Section 100 NEW ORLEANS
Millennium Scholarship was presented March 13
to Jack C. Yengo, a student at Pennsyl- Location: Kenner, La.
vania College of Technology pursuing Summary: Section Vice Chair Bruce
a bachelor’s of science degree in weld- READING — Section Scholarship Chair
Hallila arranged for Delgado Commu- Lisa Davis (left) presented Student
ing and fabrication engineering tech- nity College welding students to take a Jack C. Yengo with a $2000 Section
nology. Yengo is a member of both his plant tour of Pellerin Milnor Corp., a Scholarship.
school’s AWS student chapter and the manufacturer of commercial laundry
NEW ORLEANS — Delgado Community College welding students toured Pellerin Milnor Corp.
SECTION NEWS
CENTRAL VALLEY
District 11 District 19 February 26
Phillip Temple, director Shawn McDaniel, director Location: College of the Sequoias,
(734) 546-4298 (509) 793-5182 Tulare, Calif.
[email protected] [email protected] Summary: College of the Sequoias
hosted a Section meeting focused on
District 21
District 15 Sam Lindsey, director
Michael Hanson, director (858) 740-1917
(763) 221-5951 [email protected]
[email protected]
District 22 CENTRAL VALLEY — Section Chair
District 16 Robert Purvis, director Randy Emery (left) presented the
Section Educator Award to Ricardo
Karl Fogleman, director (916) 599-5561
[email protected] Gonzalez.
(402) 677-2490
[email protected]
District 17
J Jones, director
(832) 506-5986
District 18
Thomas Holt, director
(409) 721-5777 COLORADO — Section Chair Jeremy Mowry (far left) and AWS Past President Bob
[email protected] Teuscher (far right) presented awards to Section members (from left) Norma Es-
cobedo, Chad Jackson, and Jessica Lanka.
PERSONNEL
Lockheed Martin Appoints board, which he joined in 2018. Prior president of en-
President and CEO to joining the company, Taiclet was gine services at
with American Tower Corp. since 2001. Pratt & Whitney
James D. Taiclet has been promoted He became CEO in 2003, followed by from 1996 to
to president and CEO of Lockheed chair and president in 2004. He guided 1999. He was also
Martin Corp., Bethesda, Md., a global the company’s transformation from a previously a con-
security and aerospace company. He primarily U.S. business to a global play- sultant at McKin-
succeeds Marillyn A. Hewson, who has er in its industry. Additionally, from sey & Co. Taiclet
served as chair, president, and CEO 1999 to 2001 he was president of Hon- began his career as
since 2014. Taiclet will continue to eywell Aerospace Services, a unit of a U.S. Air Force of-
serve as a member of the corporation’s Honeywell International, and vice ficer and pilot and
served a tour of
J. D. Taiclet
duty in the Gulf
War. He holds a
master’s degree in public affairs from
Princeton University, where he was
awarded a fellowship at the Woodrow
Wilson School, and is a distinguished
graduate of the United States Air Force
Academy with degrees in engineering
and international relations.
2010–2014; vice president, U.S. sales consumer market. His experience also gineering team in the design and de-
operations, 2010; and president and includes corporate management, busi- velopment of a complex IoT-enabled
managing director, Southeast Asia op- ness development, sales, and project product. Prior to Petcube, he held vari-
erations, 2007–2010. management. ous leadership positions within Dolby,
Amazon, and Netflix. He also holds
several patents across a variety of Sili-
Meridian Adhesives Group Industrial Scientific con Valley technology companies.
Adds Global Sales Director Welcomes Vice President
of Engineering
Meridian Adhe- Huntington Ingalls Announces
sives Group, a Saldy Antony Leadership Changes
manufacturer of has joined Indus-
adhesives and trial Scientific,
sealant technolo- Pittsburgh, Pa., a
gies, has hired provider in gas de-
Nicolas Schwarz as tection and con-
global sales direc- nected safety, as
tor of its electron- vice president of
ics division. He engineering. He
joins the company will be responsible
after a long tenure for leading the
N. Schwarz with John P. Kum- company’s engi-
mer (JPK) GmbH, S. Antony neering team
where he held the position of manag- while also devel-
J. Jones G. Fuller
ing director. Prior to his appointment oping and execut-
with JPK, he worked as director of ing the company’s global technology
R&D at Hartmann Group in Heiden- strategy. Antony most recently served Huntington Ingalls Industries,
heim, Germany, where he was tasked as vice president of engineering and Newport News, Va., a military ship-
with developing medical devices, com- product at Petcube, where he led a building company, has made changes
modities, and cosmetics for the global multinational and multifunctional en- to its leadership team at its Newport
News Shipbuilding division. Julia the USS Gerald R. Ford (CVN 78) dur-
Jones was promoted to vice president ing its post-shakedown availability.
of manufacturing and facilities, where
she will be responsible for all manufac-
turing and facilities operations. Jones Obituary
has more than 22 years of shipbuilding
experience, including positions of re- John “Jack” Miaskowski
sponsibility in operations, planning,
manufacturing, nuclear support, and John “Jack” Miaskowski, of J. Miaskowski
corporate strategic planning. She has Strongsville, Ohio, passed away on
served as trades director for surface May 29. He was 82. Miaskowski was in tor at Cryogas International (now Gas-
preparation and treatment, and most advertising/sales for more than 50 world) from 1998 to 2014, where he
recently served as director of opera- years. Highly regarded in the welding worked as an advertising sales manag-
tions integration where she led trans- industry, he was a prominent fixture er. Afterward, he worked as an inde-
formation efforts to improve overall at every welding show and GAWDA pendent sales representative at Label
efficiency and effectiveness at the event. His career began at Penton Pub- Solutions, a provider in the industrial
shipyard. Gary Fuller was promoted to lishing, from 1973 to 1997, where he gas industry, until his passing. Addi-
vice president of fleet support pro- served as the Northeast regional ad- tionally, he was an active community
grams. He brings 38 years of experi- vertising sales manager for the pub- member. He was a member of the
ence in construction and overhaul, as lishing house’s two welding maga- American Welding Society from 1974
well as nuclear and radiological areas. zines, Welding Design & Fabrication and to 1998. He also served as past presi-
Prior to this role, Fuller served as di- The Welding Distribution. He went on dent and honorary District Governor
rector of test engineering in the Nu- to become the publisher of those ti- of Rotary. Miaskowski is survived by
clear Propulsion division. He has also tles. After 24 years, he moved on to his wife of 55 years, Kathleen; daugh-
served as chief test engineer, nuclear open his own advertising/sales compa- ters, Kathy, Lisa, and Maureen; and
superintendent, overhaul control engi- ny named Miaskowski Enterprises, grandchildren, Michael, Patrick, Brian,
neer, trades director, and nuclear con- which he ran for 14 years. At the same Carter, Callie, Max, Joe, Sam, and
struction director, most recently on time, he was an independent contrac- Jake. WJ
America (TMNA), Plano, Tex., for the supply of three jointly Industry Notes
developed SOUTRAC II tailor-welded blank (TWB) laser
welding systems to be installed in its North American manu- • Lamar Institute of Technology, Beaumont, Tex., was
facturing plants. named one of 15 Best Trade Schools in Texas by the World
The TWB laser welding systems are suitable for curved Scholarship Forum. Ranked number eight, the institute
applications. The processing head it contains has been exclu- was chosen, according to the selection criteria, because
sively developed for TMNA. The TWB applications imple- “these programs are focused on benefits offered and career
mented will support Toyota’s New Global Architecture. The outlook. We also considered the quality of education, aver-
fully automated production system will also be capable of in- age earnings of graduates, accreditation, and several other
tegrating automatic guided vehicles that will connect plant relevant factors.” Among its offerings is a welding degree.
logistics systems with warehouse/production equipment.
The new core piece incorporates 15 axes in combination • Keen Compressed Gas Co., Wilmington, Del., recently
with a welding shuttle concept and will be implemented in announced its receipt of ISO 17025:2017 accreditation
the new generation of the SOUTRAC II welding system. from Perry Johnson Laboratory Accreditation Inc. In
2018, the company began construction on a new, automat-
ed $6 million fill plant; completed in the fall of 2019, it’s
ESAB Introduces GCE Specialty Gas and now backed by the ISO accreditation.
Valve Product Lines in North America, • United Performance Metals has installed a new, 8-kW
Opens Texas Configuration Center fiber laser in the company’s Cincinnati, Ohio, headquar-
ters. United now operates four laser cutting machines in its
Ohio laser center: two fiber lasers and two CO2 lasers.
CLASSIFIEDS
email: [email protected]
Phone: (979) 277-8343
Fax: (281) 290-6184
Our products are made in the USA
ADVERTISER INDEX
AWS Certification 5 Gullco International 53
aws.org/cwi_endorsements (800) 443-9353 [email protected] (440) 439-8333
Hypertherm 21
AWS Foundation 52
visit.hypertherm.com/powermaxversatility (800) 737-2978
aws.org/workforcegrant (800) 443-9353
Liburdi 8
AWS Membership 19 liburdi.com (800) 991-2100
aws.org/joinaws (800) 443-9353
Lincoln Electric OBC
Arcos IBC lincolnelectric.com (888) 935-3877
arcos.us (800) 233-8460
Red-D-Arc 10
Camfil 2 reddarc.com (866) 733-3272
campfilapc.com (800) 479-6801
RPB Safety 1
Cor-met Inc. 27 rpbsafety.com (866) 494-4599
cor-met.com (800) 848-2719
RWMA 54
aws.org/rwma (305) 443-935 ext. 220
Diamond Ground 15
diamondground.com (805) 498-3837 Select-Arc IFC
select-arc.com (800) 341-5215
Fischer Engineering 8
fischerengr.com (937) 754-1750 SK Brazing 20
skbrazing.com 82 31-404-8020
Flexovit USA 16
flexovitabrasives.com (800) 689-3539 Triangle Engineering 10
trieng.com (781) 878-1500
Gal Gage Co. 11
galgage.com (269) 465-5750 voestalpine Bohler Welding USA Inc. 9
voestalpine.com/welding (800) 527-0791
Gedik Welding 13
gedikwelding.com (201) 679-5551 Weld Engineering 17
weldengineering.com (508) 842-2224
Gentec/Genstar 14
gentec.com (909) 606-2726 Welder Training & Testing 53
wtti.com (800) 223-9884
WELDING RESEARCH
SUPPLEMENT TO THE WELDING JOURNAL, SEPTEMBER 2020
Sponsored by the American Welding Society
https://doi.org/10.29391/2020.99.022
WELDING RESEARCH
650–950 0
950–1350 1
1350–1650 2
1650–1950 3
1950–2250 4
2250–2500 5
WELDING RESEARCH
A B
A B
Fig. 3 — Weld joint penetration: A — Incomplete joint penetration; B — complete joint penetration.
trained using the stochastic gradient descent optimization ms. The weld pool oscillated and an image was captured,
technique. In our approach, the parameter updating rule one image per pulsing cycle, at 2 ms of the base current to
was mini‐batch stochastic gradient descent (mini‐batch ensure that the arc had become darker. A weld pool sensing
SGD), as shown in Equation 6: system had been designed and established based on the ex-
periment platform proposed in the literature (Ref. 15). As
t+1 = t - t L( t; y(i:i+n), o(i:i+n)) (6) shown in Fig. 1, a 650‐nm‐wavelength 19 × 19 dot-matrix
structured laser pattern was projected onto the weld pool
where t is the learning rate and n is the mini‐batch size. The surface at 30 deg from horizontal. On the other side, the
method used to determine how the parameters are changed, reflected laser light was collected by a screen placed on the
from t to t+1 toward the ones that can minimize Equa- path of the reflected laser pattern. A Point Grey
tion 5, in neural networks is called optimizer. When using GZL‐CL‐22C5M‐C high‐speed camera (camera 1 in Fig. 1)
Equation 6 to change the parameters, the optimizer is to with a 650-nm center wavelength band‐pass optical filter
adaptively adjust the learning rate t based on the computed captured the images from the screen at a speed of 1000
gradient to accelerate the training processes or achieve bet- fps. At each base current period, the high‐speed camera
ter prediction performance.. Widely used optimizers include captured three images of the weld pool surface to use as
momentum (Ref. 45), Adagrad (Ref. 46), Adadelta (Ref. 47), the raw data of the neural networks. In addition, a Point
and Adam (Ref. 48) that have been developed and became Grey FL3‐FW‐03S1C with no optical filter(camera 2 in Fig.
the benchmark works in training neural networks. 1) captured one image of the backside of the weld bead at
the start of the base current period, saved in 8‐bit black-
and-white format. Figure 2A shows a typical reflected laser
II. System Setup and Dataset Building pattern image captured by camera 1 and used as input into
our model. During the welding process, the weld pool sur-
Weld Pool and Weld Joint Penetration Sensing System face was not even, which caused the reflected laser pattern
to be irregular. As discussed, all the information regarding
In our experiments, autogenous, pulsed GTAW spot the backside weld penetration were contained in that im-
welds were conducted on a 0.125-in.-thick 304 stainless age and used as the input data to the convolutional neural
steel. In each pulsing period, the peak current was applied network. Figure 2B is a typical image captured by camera 2
at 60 A for 47 ms and followed by a 20-A base current for 3 and used to identify weld penetration state labels after
WELDING RESEARCH
Data Augmentation
WELDING RESEARCH
Convolutional–1 (5 5 1) 75 = 1950
Pooling–1 (1 1) 75 = 150
Convolutional–2 (75 5 5 1) 50 = 93,800
Pooling–2 (1 1) 50 = 100
Fully connected 2700 500 500 = 1,350,500
Softmax regression 500 6 6 = 3006
Total 1,449,506
WELDING RESEARCH
ing by half if the validation error stopped decreasing for Long term, the goal of this work is to support the imple-
three epochs. Adaptive moment estimation (Adam) (Ref. mentation of this system to control weld joint penetration
48) has become prevalent in recent CNNs. Combining the in an industrial production line setting.
advantages of the RMSProp (Ref. 54) and the AdaGrad
(Ref. 46) methods, this designed adaptive learning rates
for each parameter. The convergence speed was faster References
than the revised mini‐batch gradient descent, but the vali-
dation set performance was slightly worse than GD, so we
did not use the Adam optimizer. 1. Xiao, Y. H., and Ouden, G. D. 1990. A study of GTA weld pool
Batch normalization (BN)(Ref. 55) was used in our exper- oscillation. Welding Journal 69(8): 289-s to 293-s.
iments before the activation function as a kind of datapro- 2. Xiao, Y. H., and Ouden, G. D. 1993. Weld pool oscillation during
cessing, which can speed up the learning process (Refs. 23, GTA welding of mild steel. Welding Journal 72(8): 428‐s to 434-s.
56). By forcing the output of the BN layer to be a designed 3. Chen, W., and Chin, B. 1990. Monitoring joint penetration
distribution (Gaussian distribution in our experiments), the using infrared sensing techniques. Welding Journal 69(4): 181-s to
risk of activation function saturation was reduced. In addi- 185-s.
tion, there was no need to add dropout — one way of con- 4. Nagarajan, S., Banerjee, P., Chen, W., and Chin, B. A. 1992.
trolling overfitting (Refs. 24, 57). Control of the welding process using infrared sensors. IEEE Trans-
actions on Robotics and Automation 8(1): 86–93.
During the training process, the training set was used to
5. Nagarajan, S., Chen, W., and Chin, B. 1989. Infrared sensing
adjust parameters to minimize the loss function as defined for adaptive arc welding. Welding Journal 68(11): 462-s to 466-s.
in Equation 5. Validation error was used to control the 6. Carlson, N., and Johnson, J. 1988. Ultrasonic sensing of weld
learning rate annealing and early stopping, as well as to pool penetration. Welding Journal 67(11): 239-s to 246-s.
compare different algorithms in the design stage. Test data 7. Hardt, D., and Katz, J. 1984. Ultrasonic measurement of weld
were put aside during training and tuning. All the training penetration. Welding Journal 63(9): 273-s to 281-s.
had been run on a personal computer equipped with an 8. Mnich, C., Al‐Bayat, F., Debrunner, C., Steele, J., and Vincent,
EVGA GTX 1080 graphics processing unit (GPU). The com- T. 2004. In situ weld pool measurement using stereovision. Japan‐
plete training process used 171.43 min, with a final test set USA Symposium on Flexible Automation, Denver, Colo.
error of 9.29%. The details of the training process are shown 9. Richardson, R., and Gutow, D. 1983. Coaxial arc weld pool view-
ing for process monitoring and control. Welding Journal 63(3): 43–50.
in Fig. 6.
10. Zhao, C., Richardson, I., Kenjeres, S., Kleijn, C., and Saldi, Z.
2009. A stereo vision method for tracking particle flow on the weld
Testing pool surface. Journal of Applied Physics 105(12): 123104. DOI:
10.1063/1.3143789
The prediction accuracy for a single image was 90.70%. 11. Zhao, D., Yi, J., Chen, S., Wu, L., and Chen, Q. 2003. Extrac-
In our paper, three images were captured for every base tion of three‐dimensional parameters for weld pool surface in
pulsed GTAW with wire filler. Journal of Manufacturing Science and
current period sharing the same label. Therefore, a voting
Engineering 125(3): 493–503. DOI: 10.1115/1.1556400
mechanism was added based on combining results from 12. Agapakis, J. E., and Bolstad, J. O. 1991. Vision sensing and
the three images. This increased prediction accuracy to processing system for monitoring and control of welding and other
97.57%. Further observation of the errors made by the high‐luminosity processes. Optics, Illumination, and Image Sensing
system suggested that errors only occur when the weld for Machine Vision V, pp. 32–38. International Society for Optics
joint penetration state is in the border between two adja- and Photonics: Boston, Mass.
cent labels. 13. Kovacevic, R., and Zhang, Y. 1996. Sensing free surface of
Running the trained CNN on an AMD A6‐3650 2.6 GHz arc weld pool using specular reflection: Principle and analysis. Pro-
takes more than 300 ms to get the predicted label. Running ceedings of the Institution of Mechanical Engineers, Part B: Journal of
on a GPU is much faster, requiring only 1.2 ms with a typi- Engineering Manufacture 210(6): 553–564. DOI: 10.1243/PIME_
PROC_1996_210_154_02
cal GPU (Ref. 58). Currently, system limitations due to the
14. Zhang, Y., Song, H., and Saeed, G. 2006. Observation of a
presence of multiple data collection cards prevents installa- dynamic specular weld pool surface. Measurement Science and Tech-
tion of a GPU. In the future, the welding sensing and con- nology 17(6): L9. DOI: 10.1088/0957-0233/17/6/L02
trol system will be implemented with a GPU as well. 15. Song, H., and Zhang, Y. 2008. Measurement and analysis of
three‐dimensional specular gas tungsten arc weld pool surface.
Welding Journal 87(4): 85-s to 95-s.
IV. Conclusion and Further Research 16. Ma, X., and Zhang, Y. 2011. Gas metal arc weld pool surface
imaging: modeling and processing. Welding Journal 90(5): 85-s to 94-s.
An innovative approach to identify the weld penetration 17. Zhang, W., Wang, X., and Zhang, Y. 2013. Analytical
state using CNN was presented in this paper. An advanced real‐time measurement of a three‐dimensional weld pool surface.
weld pool and weld penetration state sensing system had Measurement Science and Technology 24(11): 115011. DOI: 10.1088/
been designed and implemented to generate training data, 0957-0233/24/11/115011
18. Zhang, W. J., Zhang, X., and Zhang, Y. M. 2015. Robust pat-
and then additional geometric transformations have been
tern recognition for measurement of three dimensional weld pool
used for data augmentation. A six‐layer CNN had been surface in GTAW. Journal of Intelligent Manufacturing 26(4):
implemented using the augmented dataset. The final test 659–676. DOI: 10.1007/s10845-013-0825-z
accuracy rate of the proposed CNN model was 90.70%, with 19. Zhang, W., Liu, Y., Wang, X., and Zhang, Y. 2012. Character-
accuracy improved to 97.57% by using a voting mechanism. ization of three‐dimensional weld pool surface in GTAW. Welding
WELDING RESEARCH
Journal 91(7): 195-s to 203-s. for deep belief networks. Advances in Neural Information Processing
20. Hubel, D. H., and Wiesel, T. N. 1968. Receptive fields and Systems, pp. 1185–1192.
functional architecture of monkey striate cortex. The Journal of 42. Hyvärinen, A., and Köster, U. 2007. Complex cell pooling
Physiology 195(1): 215–243. DOI: 10.1113/jphysiol.1968.sp008455 and the statistics of natural images. Network: Computation in Neural
21. Fukushima, K., and Miyake, S. 1982. Neocognitron: A self‐ Systems 18(2): 81–100.
organizing neural network model for a mechanism of visual pattern 43. Zeiler, M., and Fergus, R. 2013. Stochastic pooling for regu-
recognition. Competition and Cooperation in Neural Nets, pp. 267‐285. larization of deep convolutional neural networks. Proceedings of the
22. LeCun, Y., et al. 1990. Handwritten digit recognition with a International Conference on Learning Representation (ICLR).
back‐propagation network. Advances in Neural Information Process- 44. Bartlett, P. L., and Wegkamp, M. H. 2008. Classification
ing Systems, pp. 396–404. with a reject option using a hinge loss. Journal of Machine Learning
23. LeCun, Y., Bottou, L., Bengio, Y., and Haffner, P. 1998. Gra- Research 9(8): 823–1840.
dient‐based learning applied to document recognition. Proceedings 45. Qian, N., 1999. On the momentum term in gradient descent
of the IEEE 86(11): 2278–2324. DOI: 10.1109/5.726791 learning algorithms. Neural Networks 12(1): 145–151. DOI:
24. Krizhevsky, A., Sutskever, I., and Hinton, G. E. 2012. Ima- 10.1016/S0893-6080(98)00116-6
genet classification with deep convolutional neural networks. Ad- 46. Duchi, J., Hazan, E., and Singer, Y. 2011. Adaptive subgradi-
vances in Neural Information Processing Systems, pp. 1097–1105. ent methods for online learning and stochastic optimization. Jour-
25. Girshick, R., Donahue, J., Darrell, T., and Malik, J. 2014. nal of Machine Learning Research 12(7): 2121–2159.
Rich feature hierarchies for accurate object detection and semantic 47. Zeiler, M. D. 2012. ADADELTA: An adaptive learning rate
segmentation. Proceedings of the IEEE Conference on Computer Vision method. arXiv preprint arXiv:1212.5701.
and Pattern Recognition, pp. 580–587. 48. Kingma, D., and Ba, J. 2014. Adam: A method for stochastic
26. Girshick, R. 2015. Fast r‐cnn. Proceedings of the IEEE Inter- optimization. arXiv preprint arXiv:1412.6980.
national Conference on Computer Vision, pp. 1440–1448. 49. Pinto, N., Cox, D. D., and DiCarlo, J. J. 2008. Why is
27. Ren, S., He, K., Girshick, R., and Sun, J. 2015. Faster R‐CNN: real‐world visual object recognition hard? PLoS Computational Biol-
Towards real‐time object detection with region proposal networks. ogy 4(1): 27. DOI: 10.1371/journal.pcbi.0040027
Advances in Neural Information Processing Systems, pp. 91–99. 50. Simard, P. Y., Steinkraus, D., and Platt, J. C. 2003. Best prac-
28. Zeiler, M. D., and Fergus, R. 2014. Visualizing and under- tices for convolutional neural networks applied to visual document
standing convolutional networks. European Conference. analysis. Proceedings of the Seventh International Conference on Docu-
29. Simonyan, K., and Zisserman, A. 2014. Very deep convolu- ment Analysis and Recognition, p. 958: IEEE Computer Society.
tional networks for large‐scale image recognition. arXiv preprint 51. Glorot, X., Bordes, A., and Bengio, Y. 2011. Deep sparse rec-
arXiv:1409.1556. tifier neural networks. Proceedings of the Fourteenth International
30. Goodfellow, I., et al. 2014. Generative adversarial nets. Ad- Conference on Artificial Intelligence and Statistics, pp. 315–323.
vances in Neural Information Processing Systems, pp. 2672–2680. 52. Glorot, X., and Bengio, Y. 2010. Understanding the difficul-
31. Szegedy, C., et al. 2015. Going deeper with convolutions. ty of training deep feedforward neural networks. Proceedings of the
Proceedings of the IEEE Conference on Computer Vision and Pattern Thirteenth International Conference on Artificial Intelligence and Sta-
Recognition, pp. 1–9. tistics, pp. 249–256.
32. He, K., Zhang, X., Ren, S. and Sun, J. 2016. Deep residual 53. Karpathy, A. 2016. Cs231n: Convolutional neural networks
learning for image recognition. Proceedings of the IEEE Conference for visual recognition. Neural Networks, Vol. 1.
on Computer Vision and Pattern Recognition, pp. 770–778. 54. Tieleman, T., and Hinton, G. 2012. Lecture 6.5‐rmsprop: Di-
33. Abdel‐Hamid, O., Mohamed, A.‐R., Jiang, H., Deng, L., vide the gradient by a running average of its recent magnitude.
Penn, G., and Yu, D. 2014. Convolutional neural networks for COURSERA: Neural Networks for Machine Learning 4(2): 26–31.
speech recognition. IEEE/ACM Transactions on Audio, Speech, and 55. Ioffe, S., and Szegedy, C. 2015. Batch normalization: Accel-
Language Processing 22(10): 1533–1545. DOI: 10.1109/ erating deep network training by reducing internal covariate shift.
TASLP.2014.2339736 International Conference on Machine Learning, pp. 448–456.
34. Sainath, T. N., et al. 2015. Deep convolutional neural net- 56. Wiesler, S., and Ney, H. 2011. A convergence analysis of
works for large‐scale speech tasks. Neural Networks 64: 39–48. log‐linear training. Advances in Neural Information Processing Sys-
DOI: 10.1016/j.neunet.2014.08.005 tems, pp. 657–665.
35. Kim, Y. 2014. Convolutional neural networks for sentence 57. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and
classification. arXiv preprint arXiv:1408.5882. Salakhutdinov, R. 2014. Dropout: A simple way to prevent neural
36. Kalchbrenner, N., Grefenstette, E., and Blunsom, P. 2014. A networks from overfitting. The Journal of Machine Learning Re-
convolutional neural network for modelling sentences. arXiv search 15(1): 1929–1958.
preprint arXiv:1404.2188. 58. Long, J., Shelhamer, E., and Darrell, T. 2015. Fully convolu-
37. Goodfellow, I., Bengio, Y., and Courville, A. 2016. Deep tional networks for semantic segmentation. Proceedings of the IEEE
Learning. MIT Press. Conference on Computer Vision and Pattern Recognition, pp.
38. Gu, J., et al. 2015. Recent advances in convolutional neural 3431–3440.
networks. arXiv preprint arXiv:1512.07108.
39. Maas, A. L., Hannun, A. Y., and Ng, A. Y. 2013. Rectifier non-
linearities improve neural network acoustic models. Proceedings of the CHAO LI, QIYUE WANG, WENHUA JIAO, MICHAEL JOHNSON, and
International Conference on Machine Learning 30(1). YU MING ZHANG ([email protected]) are with the
40. He, K., Zhang, X., Ren, S., and Sun, J. 2015. Delving deep Department of Electrical and Computer Engineering, Univer-
into rectifiers: Surpassing human‐level performance on imagenet sity of Kentucky, Lexington, Ky. LI, WANG, JIAO, and ZHANG
classification. Proceedings of the IEEE International Conference on are also with the Institute for Sustainable Manufacturing,
Computer Vision, pp. 1026–1034. University of Kentucky, Lexington, KY.
41. Boureau, Y.‐L., and Cun, Y. L. 2008. Sparse feature learning
WELDING RESEARCH
https://doi.org/10.29391/2020.99.023
WELDING RESEARCH
and the sound channel can be considered as a cavity covered The purpose of this work is to clarify the effects of the
by the shielding gas atmosphere (Refs. 15–17). According to filler wire intervention on the GTA. For this purpose, a
Lv et al. (Refs. 16, 17), statistical characteristics of the arc novel sensing method of detecting the probe voltages (i.e.,
sound in time domain can be used to distinguish different the voltage signals between the filler wire and tungsten elec-
penetration statuses due to the difference in the arc energy, trode/workpiece) was proposed. Moreover, this work was
and the different penetration statuses will also result in dif- split into two parts for elaboration. This first part aims at
ferent formant distribution of the sound channel because of expounding the principal interaction mechanisms between
the change in the sound channel state (e.g., an excessively the GTA and filler wire, and the second part will discuss the
penetrated weld pool will destroy the sound channel). Zhang behaviors of liquid droplets. In the first part, to simplify the
et al. used the Fisher distance and principal component study, a tungsten probe was used to replace the filler wire
analysis to select the frequency components reflecting the and to interact with the arc through specific experiments.
weld penetration status in pulsed GTAW, then established a This work helped to enhance the understanding of electrical
classification model and successfully distinguished the dif- arc characteristics and laid the foundation for on-line moni-
ferent weld penetration statuses (Ref. 18). toring and control of the GTAW quality when the filler wire
As mentioned above, so much valuable and potential infor- was employed.
mation has been continuously mined, the relevant mecha-
nisms have been deeply explained, and various sensing meth- Sensing Methodology
ods have been proposed to serve on-line monitoring and con-
trol of GTAW quality. However, the reported studies in the aca- In the autogenous GTAW process, electrical arc is estab-
demic field have focused more on autogenous GTAW than lished between the nonconsumable tungsten electrode tip
GTAW with filler wire. Furthermore, researchers have devoted and liquid weld pool surface. Weld pool behaviors will
more attention to studies on weld pool status, often ignoring change arc length and thus change arc voltage (U, i.e., the
the effect of the liquid metal generated from the filler wire (if voltage between the tungsten electrode and the workpiece).
it is employed) on the process. Although the employment of When the filler wire is introduced into the GTA, the feeding
the filler wire will expand the application of GTAW, it will also motion of the filler wire, the resulting metal transfer, and
interfere with the GTA and impact the extraction of character- the pendant liquid droplets that oscillate at the end of the
istic information about the weld pool (e.g., disturbing its regu- filler wire will affect the arc, and the filler wire itself as a
lar oscillation frequency under pulse current), which is not suspended electrical conductor will also become an interfer-
conducive to the welding process stability and monitoring of ence. The resultant interference signals will be coupled into
the weld pool. Yudodibroto et al. studied the effects of filler the arc voltage, thereby obscuring the characteristic signals
wire on weld pool oscillation in GTAW and thought the oscilla- contained in the arc voltage that reflect the weld pool be-
tion frequency could still be measured to reflect the weld pene- haviors. It is difficult to distinguish their respective signal
tration status under the mode of uninterrupted metal transfer components and to clarify the interaction mechanisms be-
(Ref. 19), but the further study has not been reported yet; for tween the filler wire and arc plasma with the arc voltage
instance, the interaction mechanisms between the filler wire alone. Therefore, it may be possible to consider introducing
and GTA are still unclear. additional sensing signals to help explain the relevant con-
WELDING RESEARCH
WELDING RESEARCH
experiments (#1, #2, and #3) have been designed and con- The schematic relative position at different Ds can be seen
ducted. In all the experiments, a tungsten probe with a di- in Fig. 3. During the experiments, arc voltage and probe
ameter of 2.4 mm was immersed into the arc atmosphere in- voltage were collected, and the welding current was meas-
stead of the filler wire, and a relatively small welding current ured to ensure the voltage fluctuations were not caused by
(I) was employed, which can ensure that the probe was not the current fluctuations. Experimental data was filtered by a
burnt as much as possible and the surface of the workpiece simple moving average. Each average was calculated over a
was slightly melted, thus avoiding the metal transfer and dy- sliding window of 151 sample points, which was centered
namic evolution behaviors of the weld pool. This way, the about the element in the current position.
probe can simulate the filler wire to interfere with the arc The specific experimental designs and parameters are re-
plasma, thereby simplifying the objective of this study. Since spectively shown in Fig. 4 and Table 1. Experiment #1 is
the focus of this part of the work is only the inner interac- mainly used to investigate the static effects of a probe on
tion mechanisms between the GTA and filler wire, although the GTA. As demonstrated in Fig. 4A, during Experiment
this simulation is not equivalent to the actual filling process, #1, the welding torch moved horizontally several times to
it can also be considered that the effect of the filler wire as gradually go away from the probe tip until no probe voltage
conductive metal on arc plasma follows similar mechanisms was detected. It moved 1 mm each time (e.g., Da2 Da1 Da1
to that of the probe. Da0 1 mm), and remained stationary for a while (10 s) af-
Figure 2 presents the schematic diagram of the experi- ter each movement. Experiment #1 had been conducted at
mental apparatus. As shown in Fig. 2, a tungsten probe was different Hs, and in this way, the electrical signals can be ob-
suspended parallel to the workpiece, and one end of this tained when the probe is in different spatial positions of the
probe was fixed to the clamping apparatus. This clamping arc column.
apparatus was placed on the welding workbench and isolat- Experiment #2 aimed to study the dynamic effects of an
ed by nonconductive insulating tape. The axis of the welding intermittent/continuous reciprocating probe on the GTA.
torch was perpendicular to the workpiece surface. There are The design of Experiment #2 is presented in Fig. 4B. The
two positional parameters, D and H, in Fig. 2, where D is the welding torch reciprocated equidistantly between position
horizontal distance between the probe tip and the symme- D = 0 mm and position D = Db0 = 0.5 mm. Whenever the
try axis of the electrode, and H is the height difference be- welding torch was horizontally displaced from its original
tween the probe and the workpiece surface. When the hori- position to the new position in the same reciprocating mo-
zontal relative position of the symmetry axis of the elec- tion mode, it stayed at the new position for a period of time
trode and the probe tip is as shown in Fig. 2, D is positive, (T) and then proceeded to the next step. Different Ts
and when the vertical relative position of the workpiece sur- (e.g., Ta > Tb) have been set during Experiment #2. When
face to the probe tip is as shown in Fig. 2, H is positive. T was 0 s, the probe generated a continuous interference,
The general experimental procedure was to keep the and when T was a non-zero value, the probe caused an in-
welding workbench stationary, suspend the probe at a termittent interference.
height of H, and move the welding torch at a welding speed Experiment #3 was about the dynamic effects of a continu-
(v) in the horizontal direction to change D (v is too small to ously moving probe on the GTA at different positions. As
affect the arc shape). When the absolute value of D de- shown in Fig. 4C, the welding torch reciprocated equidistantly
creased, the direction of v was considered positive, indicat- several times between different horizontal positions to con-
ing that the welding torch and the probe were approaching. tinuously generate dynamic interferences. The distance of
WELDING RESEARCH
the probe is directly inserted into the arc column, the conduc-
A
tive cross section in the vicinity of the probe is forcibly re-
duced, which destroys the conductive path with the minimum
energy consumption under the original conditions, thus tend-
ing to change the direction of electron flow. Although the arc
column will resist this change to keep the original conductive
path as much as possible, the conductive path following the
minimal energy principle will still tend to have different de-
grees of deflection (tend to become longer) under new condi-
tions. On the other hand, the probe intervention will alter the
B thermal field of the arc column. Since the probe that is cold rel-
ative to the arc plasma increases the arc heat loss, the arc col-
umn will automatically shrink to reduce the contact area with
the external medium, thereby minimizing the heat loss. It will
also contribute to the reduction in the conductive cross sec-
tion of the arc plasma near the probe, which will increase the
average current density and thus increase the average electric
field strength. In this way, the intervention of the metal probe
causes the arc voltage to rise.
C The relationship between the static probe at different po-
sitions and the stable arc voltage can be obtained by calcu-
lating the mean value of U in the rectangular frames of Fig.
5A, C, and E, and the results are as shown in Fig. 7. With the
increase in D, U declines more and more gently, and the in-
crease in H contributes to the rise of U. This is because the
ionization in the core area of the arc is more sufficient, so
the interference of the metal probe will gradually weaken as
it moves away from the core region of the arc. In addition,
Fig. 4 — Experimental designs (v0 7 mm/s). A — Experiment the cathode is an electron emission source. The closer the
#1; B — Experiment #2; C — Experiment #3. probe is to the cathode, the greater its influence on the GTA
will be. Therefore, the arc voltage will increase as H rises.
each movement was 1 mm, and at intervals, the welding torch It can be further found from Fig. 5 that whenever D is
moved to a farther distance from the probe for a new recipro- changed once, U can quickly level off after a small variation,
cating motion (e.g., Dc2 Dc1 Dc1 Dc0 1 mm). while the changes in U, Upc, and Upa are not completely con-
sistent with one another. During phase A, when U reaches a
new stable level, Upc and Upa are still changing in opposite di-
Results and Discussion rections, and any of them has a greater variation than U.
However, during phase B, the variation in Upc and Upa caused
Static Effects by the change in D becomes modest, and the trend of Upa in
phase B is opposite to that in phase A. Also, the transition
Figure 5 displays the electrical signals measured in Exper- period of Upc and Upa becomes very short or even inconspicu-
iment #1, where Upc and Upa were measured separately. In ous. In the final phase C, although U is almost unaffected by
Fig. 5, D increases stepwise as time goes on. For ease of de- the change in D, the change in D will once again cause Upc
scription, the process was roughly divided into phases A, B, and Upa to change significantly.
and C, which respectively corresponded to the situation As mentioned earlier, the arc voltage signal can reflect the
where the end of the probe was located in the original core overall electrical characteristics of the arc, and the probe volt-
area of the arc (the brightest area), the original noncore area age signal can reflect the electrical characteristics of the local
of the arc, and the area outside the arc (no strict demarca- area of the arc. Such diverse variations in Upc and Upa indicate
tion between the areas). The typical shapes of the core area that the influence of the metal probe on the electrical charac-
of the arc can be seen in Fig. 6. Both the voltage signals and teristics of different areas of the arc at different phases is not
the arc shape were affected by the metal probe. always consistent. During phase A, although the arc above the
When the probe occupies the space area that originally probe maintained its original shape as much as possible, the
belonged to the arc plasma, a plasma sheath will form near lower part of the arc severely deflected. The larger the H is, the
the probe surface and it will become a new partial edge area more severe the deflection will be. This is because the electron
of the arc plasma. Accordingly, the overall arc shape will be flow needs to reform the shortest conductive path between
changed. From phase A to B, U drops stepwise and each drop the anode and cathode due to the physical barrier of the probe.
resulted from an increase in D, while the growth in D did As H rises, the electron flow from the cathode will encounter
not cause such a change in U during phase C. Therefore, the this barrier earlier and thus be deflected more. In addition, the
metal conductor caused an increase in the arc voltage after it decrease in temperature around the probe tends to suppress
had immersed into an area of the GTA. the thermal ionization around the probe. Nevertheless, the
This can be explained from two aspects. On one hand, after neutral particles above the probe can still be directly bombard-
WELDING RESEARCH
A B
C D
E F
Fig. 5 — Electrical signals in Experiment #1. A — Upc and U at H 1 mm; B — Upa and U at H 1 mm; C — Upc and U at H 2 mm; D —
Upa and U at H 2 mm; E — Upc and U at H 3 mm; F — Upa and U at H 3 mm.
ed by electrons from the cathode direction and thus be ion- es, the main conductive path of the arc is gradually restored,
ized. However, in the area below the probe, because of the the arc deflection is reduced, and the gap in ionization be-
physical block by the probe, it will be difficult for the neutral tween the upper and lower parts of the arc is narrowed, so Upa
particles to be ionized by direct collision of electrons above the rises and Upc falls. In phase B, the end of the probe retreated to
probe, and the originally charged ions will even be recombined the original noncore area of the arc, and the main conductive
due to insufficient energy. Therefore, during phase A, the de- path of the original core area was almost restored. The effects
crease in D weakened the ionization of the lower part of the of the probe to the arc were weakened, and the probe no
arc and then reduced the electric field, thus bringing Upa down, longer caused apparent arc deflection. The upper and lower
which corresponded to a weaker heat production capacity, al- parts of the arc presented as a whole, and the trend of Upa and
though the arc length below the probe seemed to be longer. Upc unified with that of U. During phase C, although the inter-
Meanwhile, the heat production capacity of the upper part of ference of the probe was not enough to change the arc voltage,
the arc was enhanced to meet the increased overall heat pro- the edge area of the arc still had a gradient distribution of un-
duction capacity of the arc column, so the ionization degree stable charges, so apparent changes in the probe voltages
and electric field intensity of the upper part of the arc in- could still be detected.
creased, and Upc grew accordingly. In such a way, when the end Even if the physical medium forcibly intervenes in the
of the probe is in the original core area of the arc, as D increas- core area of the arc, the arc can still self regulate in a short
WELDING RESEARCH
Fig. 6 — Typical shapes of the core area of the arc from Fig. 7 — Arc voltage with the static probe immersed in differ-
phase A to C. ent positions of the arc column.
Fig. 8 — Electrical signals in Experiment #2. Fig. 9 — Electrical signals in Experiment #3.
time to present a new state of overall stable heat production along with the arc voltage, was sufficient to reflect the relat-
with the lowest energy consumption. However, in the area ed effects, so the follow up will only focus on probe voltage
near the anode, as the probe interference deepens, the arc Upa to avoid repetition.
stiffness is not as good as before, and it tends to be unsta-
ble. It is conceivable that as the interference increases, the Dynamic Effects
unstable area starting from the anode side will gradually ex-
pand toward the cathode side. Eventually, the arc will extin- In the actual welding process, the effects of the filler wire
guish because it will fail to maintain a stable self-sustained on the arc are more inclined to be dynamic effects. Figure 8
discharge process. shows the electrical signals measured in Experiment #2,
In the actual welding process, the interaction between where the end of the probe occupied the original core area of
the filler wire and arc should conform to the above mecha- the arc. Because the arc plasma possessed heat capacity, it
nisms, which will also cause the similar interference to the can be observed that if T was too short (less than the time
arc (except for metal transfer). To minimize the interference constant), Upa would not reach the stable level, and the fluc-
of the filler wire, it may be recommended to keep the end of tuation of Upa caused by the variation in D would be more
the filler wire melted in phase B. This is because, on the one moderate with a decreasing T.
hand, its interference on the core area of the arc is small; on During the intermittent interference period (e.g., when T
the other hand, if the end of the filler wire is melted in was 4 s), the local maximum of Upa did not appear when U
phase C, although the arc shape is stable, the filler wire may stayed at a lower level (when D was 0.5 mm), but approximate-
be melted intermittently due to the lack of heat. In addition, ly at a point when U was rising (when D was falling from 0.5 to
according to Equation 1, only one of the probe voltages, 0 mm), or probably when U had just risen to the stable level
WELDING RESEARCH
(when D had just completed the drop from 0.5 to 0 mm). Simi- the effects of the filler wire as a metal conductor on GTA
larly, the local minimum of Upa in Fig. 8 generally occurs when based on the proposed arc-sensing method of detecting
U has just completed a drop, rather than when U keeps at a probe voltages. The main conclusions drawn are as follows:
higher level. The above-mentioned changes are mainly due to 1) When the filler wire as a metal conductor is fed into
the variation in D that can cause a sudden change in the con- the arc, the arc voltage (U) will rise due to the increase in the
ductive cross section, which can immediately change the cur- average electric field strength of the arc. The closer the filler
rent density and thus change electric field strength, while the wire is to the core area of the arc, the greater the U will be.
arc itself has thermal inertia. 2) The probe voltages (Upa and Upc) can reflect the electri-
In the core area of the arc, the blocking by the probe tends cal characteristics of different local areas of the GTA. The ef-
to promote the ion recombination below the probe, while the fects of the filler wire as a metal conductor on the different
ion recombination will lag behind the decrease in D due to the areas of the arc at different phases are not always consis-
thermal inertia. Thus, below the probe, the decrease in the tent. When the filler wire is immersed in the original core
electric field strength caused by the ion recombination will lag area of the arc, the arc will be deflected, Upa will drop, and
behind the decrease in D. On the contrary, the decrease in D Upc will grow. When the filler wire is immersed in the origi-
will lead to the sudden decrease in the conductive cross sec- nal noncore area of the arc, the upper and lower parts of the
tion, resulting in an immediate increase in electric field arc will act as a whole, and the trend of Upa and Upc will fol-
strength. Therefore, in the early stage of the decrease in D, Upa low that of U.
will rise because of the immediate increase in electric field 3) When the filler wire as a metal conductor causes dynam-
strength caused by the sudden decrease in the conductive ic reciprocating interference to the arc, the probe voltage does
cross section. If D has enough time to continue to decrease, not strictly change synchronously with the arc voltage, but
the effect of ion recombination below the probe on the electric sometimes reaches peak or trough ahead of the arc voltage due
field strength will gradually emerge, thus reversing the growth to the thermal inertia. Nevertheless, the dynamic effects on
in Upa. Likewise, the gas ionization below the probe will also lag the arc can indeed be reflected in the probe voltage.
behind the increase in D, and Upa will change in the opposite
direction. Therefore, the rate of ion recombination and the Acknowledgments
rate of gas ionization will affect the moments at which the lo-
cal maximum and local minimum of the Upa occur, respectively.
It can be observed that during the intermittent interference,
the initial decline in Upa is steeper than its initial rise, which This study is supported by the National Natural Science
shows that the rate of ion recombination is faster than that of Foundation of China (Grant No.: 51505326), Natural Sci-
gas ionization in the early period after D starts to vary. The ence Foundation of Tianjin (Grant No.: 16JCQNJC04300),
same mechanism is also suitable for the continuous interfer- and the Regional Demonstration Project of Marine Econom-
ence process. Thus, in Fig. 8, Upa usually reaches the peak be- ic Innovation and Development (Grant No.: BHSF2017-10).
fore D falls to 0 mm and falls to the bottom when D has just
risen to 0.5 mm.
Figure 9 shows the electrical signals measured in Experi- References
ment #3 and that the dynamic probe continuously interfered
with different areas of the arc column. It can also be seen that,
in the core area of the arc, Upa can reach the local minimum be- 1. Chen, Z. Y., Chen, J., and Feng, Z. L. 2017. Monitoring weld
fore U drops to the bottom. This is mainly because the change pool surface and penetration using reversed electrode images.
amount of D becomes 1 mm; it can rise to meet the moment Welding Journal 96(10): 367-s to 375-s.
that the decrease in Upa is reversed by the increase in local elec- 2. Fan, C. J., Lv, F. L., and Chen, S. B. 2009. Visual sensing and
penetration control in aluminum alloy pulsed GTA welding. Inter-
tric field strength caused by the gas ionization of the lower national Journal of Advanced Manufacturing Technology 42(1–2):
part of the arc. Therefore, when the continuous interference 126–137. DOI: 10.1007/s00170-008-1587-1
occurs in the original core area of the arc, an impression is pre- 3. Song, H. S., and Zhang, Y. M. 2008. Measurement and analy-
sented that Upa does not strictly change synchronously with U, sis of three-dimensional specular gas tungsten arc weld pool sur-
but often reaches peak or trough ahead of U. However, in the face. Welding Journal 87(4): 85-s to 95-s.
noncore area of the arc, the changes in Upa and U seem to be 4. Zhang, W. J., Zhang, X., and Zhang, Y. M. 2013. Robust pat-
more synchronous, and the dynamic relationship between tern recognition for measurement of three dimensional weld pool
them was more consistent with their static relationship be- surface in GTAW. Journal of Intelligent Manufacturing 26(4):
cause the effects of the probe on the arc were weakened. Addi- 659–676. DOI: 10.1007/s10845-013-0825-z
tionally, the fluctuation of Upa became significant again outside 5. Zhang, W. J., Wang, X. W., and Zhang, Y. M. 2013. Analytical
real-time measurement of a three-dimensional weld pool surface.
the arc, which indicated that there is a large ionization gradi- Measurement Science and Technology 24(11): 115011. DOI: 10.1088/
ent, so that the movement of probe can be detected particular- 0957-0233/24/11/115011
ly clearly. Although the droplet behavior was not covered here, 6. Liu, Y. K., Zhang, Y. M., and Kvidahl, L. 2014. Skilled human
the mechanism of its dynamic interference should also follow welder intelligence modeling and control: Part 1 — Modeling.
the above description. Welding Journal 93(2): 46-s to 52-s.
7. Dong, H., Cong, M., Zhang, Y. M., Liu, Y. K., and Chen, H. P.
Conclusions 2018. Modeling and real-time prediction for complex welding
process based on weld pool. The International Journal of Advanced
Manufacturing Technology 96(5-8): 2495–2508. DOI: 10.1007/
This part of the work mainly studied the mechanism of s00170-018-1685-7
WELDING RESEARCH
8. Cheng, Y. C., Xiao, J., Chen, S. J., and Zhang, Y. M. 2018. In- for pulse GTAW welding dynamic process. Robotic Welding, Intelli-
telligent penetration welding of thin-plate GTAW process based on gence and Automation, RWIA 2014, Advances in Intelligent Systems
arc voltage feedback. Transactions of the China Welding Institution and Computing, Springer International Publishing, Cham, 363:
39(12): 1–4. DOI: 10.12073/j.hjxb.2018390287 249–261.
9. Zhang, S. Q., Hu, S. S., and Wang, Z. J. 2016. Weld penetra- 18. Zhang, Z. F., Wen, G. R., and Chen, S. B. 2018. Audible
tion sensing in pulsed gas tungsten arc welding based on arc volt- sound-based intelligent evaluation for aluminum alloy in robotic
age. Journal of Materials Processing Technology 229: 520–527. DOI: pulsed GTAW: Mechanism, feature selection and defect detection.
10.1016/j.jmatprotec.2015.09.034 IEEE Transactions on Industrial Informatics 14(7): 2973–2983. DOI:
10. Renwick, R. J., Rarson, D. F., and Richardson, R. W. 1983. 10.1109/TII.2017.2775218
Experimental investigation of GTA weld pool oscillations. Welding 19. Yudodibroto, B. Y. B., Hermans, M. J. M., Hirata, Y., and den
Journal 62(2): 29-s to 35-s. Ouden, G. 2004. Influence of filler wire addition on weld pool oscil-
11. Xiao, Y. H., and Ouden, G. D. 1993. Weld pool oscillation lation during gas tungsten arc welding. Science and Technology of
during GTA welding of mild steel. Welding Journal 72(8): 428-s to Welding and Joining 9(2): 163–168. DOI: 10.1179/
434-s. 136217104225012274
12. Shi, Y., Li, C. K., Du, L. M., Gu, Y. F., and Zhu, M. 2016. Fre- 20. Chen, S. J., Zhang, R. Y., Jiang, F., and Dong, S. W. 2018. Ex-
quency characteristics of weld pool oscillation in pulsed gas tung- perimental study on electrical property of arc column in plasma arc
sten arc welding. Journal of Manufacturing Processes 24: 145–151. welding. Journal of Manufacturing Processes 31: 823–832. DOI:
DOI: 10.1016/j.jmapro.2016.08.010 10.1016/j.jmapro.2017.12.028
13. Huang, J. K., Yang, M. H., Chen, J. S., Yang, F. Q., Zhang, Y. 21. Zhang, Y. M., Zhang, S. B., and Liu, Y. C. 2001. A plasma
M., and Fan, D. 2018. The oscillation of stationary weld pool sur- cloud charge sensor for pulse keyhole process control. Measurement
face in the GTA welding. Journal of Materials Processing Technology Science and Technology 12(8): 1365–1370. DOI: 10.1088/0957-
256: 57–68. DOI: 10.1016/j.jmatprotec.2018.01.018 0233/12/8/352
14. Li, C. K., Shi, Y., Gu, Y. F., and Yuan, P. 2018. Monitoring 22. Shan, P., Yi, X. L., Hu, S. S., and Luo, Z. 2003. Detection of
weld pool oscillation using reflected laser pattern in gas tungsten plasma sheath voltage in plasma arc welding. Transactions of the
arc welding. Journal of Materials Processing Technology 255: China Welding Institution 24(2): 19–21.
876–885. DOI: 10.1016/j.jmatprotec.2018.01.037
15. Lv, N., and Chen, S. B. 2011. Investigation on acoustic sig-
nals for on-line monitoring of welding. Robotic Welding, Intelligence
and Automation, Lecture Notes in Electrical Engineering, Springer, SHUANGYANG ZOU, ZHIJIANG WANG ([email protected]), SHENG-
Berlin, Heidelberg, 88: 235–243. SUN HU, GUANCHENG ZHAO, and WANDONG WANG are with the
16. Lv, N., Xu, Y. L., Zhong, J. Y., Chen, H. B., Wang, J. F., and Tianjin Key Laboratory of Advanced Joining Technology, School
Chen, S. B. 2013. Research on detection of welding penetration of Materials Science and Engineering, Tianjin University, Tianjin,
state during robotic GTAW process based on audible arc sound. In- China. YOUQUAN CHEN is with the Department of Mechanical
dustrial Robot 40(5): 474–493. Engineering, College of Humanities & Information, Changchun
17. Lv, N., Xu, Y. L., Fang, G., Zhao, H., and Chen, S. B. 2015. University of Technology, Changchun, China.
Mechanism analysis and feature extraction of arc sound channel
Note that in each issue of the Welding Journal Research Supplement, we are including Direct Object
Identifier (DOI) numbers with each of the papers published in print and online. A direct object identifier is a
unique alphanumeric string assigned by a registration agency (we are using Crossref.org) to identify
content and provide a persistent link to its location on the Internet. Our decision to begin assigning a DOI
for each paper comes directly from a request by the research community.
As part of our obligation to Crossref.org, we are asked to provide DOI numbers, when available, in the
references section of papers. So, if you have submitted a paper to the Welding Journal or are planning on
submitting a paper, we ask that you update your references to include DOI numbers whenever possible.
Thank you.
Annette Alonso
Publisher, Welding Journal
When choosing between collet bodies and gas lenses for GTAW operations, the primary considerations include cost, welding quality, and process efficiency. Collet bodies are less expensive initially but can lead to increased problems like turbulence and contamination, ultimately raising costs through rework and gas consumption. Gas lenses, although costlier upfront, provide improved gas flow stability and coverage, leading to better weld quality and reduced overall costs in the long run .
High shielding gas rates increase costs primarily through the added expense of consumable gas, rework due to instability, and downtime for replacing gas cylinders. These costs can be significant over time, especially in high duty cycle applications. To mitigate these costs, welding operations can adopt proper shielding practices such as using gas lenses, which stabilize the gas flow and reduce consumption and rework needs, resulting in overall cost savings .
Sensing methodologies involving probe voltages can be used to monitor the fill wire's interaction with the arc in GTAW. By detecting voltage signals between the filler wire and the tungsten electrode/workpiece, these methodologies provide insights into the arc's behavior when influenced by the wire. This helps in understanding the electrical characteristics and interactions during welding, ultimately contributing to more precise process control and better quality assurance .
Switching from a collet body to a gas lens consumable improves GTAW outcomes by providing smoother and more stable shielding gas flow, which reduces turbulence and allows the tungsten to extend further. This reduces issues such as porosity and arc flutter, thus lowering the need for rework. Although gas lenses have a higher upfront cost, they deliver significant savings on gas and rework costs due to better welding performance and less downtime .
Filler wires impact the electrical characteristics of the arc by altering the arc voltage; they increase the average electric field strength, especially when positioned close to the core area, and cause arc deflection. This interference affects the sensing signals necessary for monitoring and controlling weld quality. Understanding these effects is crucial for optimizing welding parameters and maintaining weld integrity, as it helps in distinguishing the signal components related to filler wire interaction .
The interaction between filler wire and the arc contributes to arc instability by introducing electrical interference that affects the arc voltage, causing deflection and disturbances. To minimize this effect, the filler wire's end can be kept within certain phases to reduce core arc interference. Maintaining consistent wire feed and managing the melt phase helps stabilize the arc environment, ensuring steady arc behavior and reducing weld defects .
Schlieren photography is a scientific method used to visualize differences in gas flows by capturing the shadows that result from light refraction through gases of varying densities. It was used in studying shielding gas flow in GTAW to observe flow patterns as shielding gas exits the torch. This technique allowed researchers to see how turbulence forms and grows, which helped in understanding the benefits of gas lenses and improving gas flow configurations in welding processes .
Collet bodies in GTAW applications introduce a significant amount of turbulence and atmospheric contamination into the shielding gas stream. This can lead to issues such as porosity, base metal oxidation (sugaring on stainless steel), poor welding performance, and arc flutter, which are unacceptable in high-quality applications. Additionally, this turbulence limits the extension of the tungsten past the nozzle, restricting the operator’s welding ability. Increased shielding gas flow to mitigate these issues only exacerbates the instability, leading to higher costs due to gas consumption, rework, and downtime .
The design of the nozzle plays a crucial role in improving shielding gas flow by promoting a stable gas column. Nozzles with longer lengths help transition plug flow into a developed flow, thereby reducing Kelvin-Helmholtz instabilities. A converging-type nozzle with multiple screens of varying mesh facilitates optimal flow profiles, reduces turbulence, and enhances gas coverage by promoting a more stable shielding gas column around the weld area .