Journal of Constructional Steel Research 168 (2020) 105921

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Testing of oxyacetylene weld strength

S.A. Civjan a,⁎, T. Guihan b, K. Peterman a
a
University of Massachusetts Amherst, 232 Marston Hall 130 Natural Resources Road, Amherst, MA 01003, USA
b
Tighe and Bond, 53 Southampton Road, Westfield, MA 01085, USA

a r t i c l e i n f o a b s t r a c t

Article history: A testing program is reported to evaluate the capacity of oxyacetylene fillet welds in a fatigue prone detail. His-
Received 28 May 2019 torical context is provided to evaluate the oxyacetylene weld performance and past structural applications. This
Received in revised form 10 December 2019 test program showed that fillet welds exceeded the static design capacities based on current weld design prac-
Accepted 23 December 2019
tice. In addition, fatigue testing of these welds in a Category E detail demonstrated very consistent fatigue capac-
Available online xxxx
ity and significantly exceeded the predicted life for a fatigue prone detail (Category E) for cycles exceeding
Keywords:
200,000. Based on these results, the authors propose that oxyacetylene welds could be considered as a viable al-
Welding ternative to arc weld processes when appropriate, subject to qualification testing. This may be relevant in the case
Oxyacetylene welding of emergency repairs, welding in remote areas or where electrical sources are not present.
Fusion welding © 2020 Published by Elsevier Ltd.
Gas welding
Fatigue
Weld history

1. Introduction (including certified welding rods from various manufacturers) with no

external power requirements. The process is extremely versatile with
Oxyacetylene fusion welding played a significant role in the devel- the ability to utilize the same general setup for cutting, bending, brazing,
opment of welding methods and design, though it is rarely discussed soldering, and welding, requiring only replacing torch tips and adjusting
or utilized as an option for modern structural welding. This is justified gas pressures to move from one process to another. Disadvantages are
on large projects due to the lower weld efficiency as compared to arc- that welding is generally slower than more commonly used methods.
welding methods but has led to the perception that oxyacetylene Applications where this method would be appropriate would in-
welding is an outdated or inferior process. However, there are specific clude individual welds and field repairs of lighter members and connec-
conditions where the method could prove very useful. This paper will tions, retrofit work on cast‑iron members and work in remote areas
present these uses in the context of historical background and provide where electrical sources are not available or reliable. An example of
data from new tests that were conducted to evaluate the static and fa- the latter could include work by NGO organizations such as Bridges to
tigue capacity of fillet welds made by the oxyacetylene weld process. Prosperity or Engineers Without Borders where local metalworking
The purpose is to show that oxyacetylene welds are reliable, of compa- tools are available but power sources may be unreliable, or in post-di-
rable quality to other weld processes, and may be preferable in specific saster recovery efforts.
applications.
Oxyacetylene welding is a method of fusion welding where the base 2. Background
and filler metals are heated by a welding torch until they melt into a
pool of molten metal and fuse when cooled. This is a form of gas Oxyacetylene welding was introduced at the turn of the 20th cen-
welding, with acetylene and pure oxygen used due to the high heat pro- tury and made significant technological advancements in each decade
vided by these gases. Oxyacetylene equipment is common on construc- into the 1930's. Gas welding and electric arc welding were both being
tion sites, typically used for cutting of metals and preheating of steel. advertised and discussed as equivalent options in the 1920's (in con-
Equipment used for oxyacetylene welding is identical to oxyacetylene temporary journals such as the Welding Journal and the Journal of Acet-
cutting commonly performed in fabrication shops and construction ylene Welding and Mackenzie and Card [43]) with authors noting the
sites, with the substitution of a welding torch tip. Advantages are that efficiency of maintaining both oxyacetylene and arc welding equipment
the weld process relies on very basic and readily available equipment in plants to utilize each at what they are best suited for (Anon [8],
Whittemore [57]), though opinions differed on applications where
⁎ Corresponding author.
each was preferable ([8]). Issues of material compatibility, metallurgy
E-mail addresses: kdpeterman@umass.edu, civjan@umass.edu (S.A. Civjan), and fatigue were often mentioned in earlier papers and explicitly re-
TGuihan@TigheBond.com (T. Guihan), kdpeterman@umass.edu (K. Peterman). ported in later research such as Schuster [50] and Driscoll [28], with

https://doi.org/10.1016/j.jcsr.2019.105921
0143-974X/© 2020 Published by Elsevier Ltd.
2 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921

comparable results and concerns reported between both welding pro- (ratio of tension specimen ultimate stresses in welded versus base
cesses. In World War I shortages in gas supplies made for a push toward metal specimens) lower than 100% for a variety of techniques of pre
electric arc welding in England [40,54]. A report by the Bureau of Stan- and post treatments. The study did not limit specimen warping in the
dards [34] likely undermined the oxyacetylene weld market in the weld procedures, leading to large load eccentricity in welded tension
United States due to results that noted the difficulty of maintaining a specimens, and photos indicate poor quality welds by visual inspection,
neutral flame during the weld process due to flashback and wide varia- so the lower reported weld efficiencies are not surprising. Moore [39]
tion in control of gas to oxygen ratios depending on welding torch reported 268 mild steel specimens fabricated at a specialized oxyacety-
(blowpipe) manufacturer, which can both be avoided with modern lene weld shop subjected to static, fatigue or impact load tension testing
welding equipment. After an in-depth evaluation and discussion of var- for base metals up to 25 mm (1 in) in thickness. Static and fatigue tests
ious manufacturer's weld equipment the report noted that sound welds indicated 100% weld efficiency for plates 13 mm (½ in) thick or less and
were not possible since “No oxyacetylene blowpipe on the market to- lower for thicker plates, while impact capacities were reduced. Both of
day fulfills these fundamental conditions” of constantly avoiding flash- these studies are comparable with results from similar testing of the
back (pg. 103) and concluded that “a blowpipe that can not maintain arc welding processes at the time [43,61,63]. A series of explosive fail-
under all operating conditions a neutral flame can not logically be ex- ures of welded pressure vessels led to a large testing program by the
pected to produce sound welds” (pg. 107), but also noted that should British Engine Boiler and Electrical Insurance Co. LTD reported by
a proper blowpipe be manufactured “there is no reason to believe, Schuster [50] and Anon [9] which compared oxy-acetylene and electric
that a weld of clean sound metal can not be made with assurance during arc welds. This research tested “several hundreds” of tension specimens
any welding operation and that such welds will or can be made to pos- of 6 to 25 mm (¼ to 1 in) thickness of various welded splice configura-
sess the proper physical properties” (pg. 108). Further marketing of the tions, weld processes, electrode types and weld operators with the in-
newer arc welding processes in the 1920's and 1930's began to domi- tention of developing British standards for the manufacturing of
nate discussions, predominantly on the basis of welding speed and au- welded pressure vessels that addressed weld strengths and best prac-
tomation (though automation of oxyacetylene weld process was also tice. The provisional rules were specified to apply to either “metallic
advancing at this time). Weld failures encountered in World War II, typ- arc process with a flux-coated electrode or by the oxy-acetylene pro-
ically related to fatigue and welding of thick plate and materials, led to cess” and conclude with a factor of safety of ½ and regulations for
many research projects related to welding [59,60,62,64,65,66,67] but weld quality control [9,50]. Anon [9] points out that similar research
these studied welds made by the electric arc welding processes exclu- for guidelines were being carried out by ASME and the Swiss Steam
sively, perhaps due to the previous Johnston [34] report. The fact that Users' Association, but that those organizations were looking at a
the latter research projects were compiling new data and using im- much narrower scope. The testing was fairly exhaustive, including met-
proved materials and methods suggested, perhaps erroneously, that allurgical investigations, Izod and Brinell hardness tests, tension tests,
the arc process provided better quality welds. Testing of oxyfuel pro- repeated impact tests and bend tests. Bend tests “show how efficient
cesses using similarly improved materials and guidelines does not ap- is a well-made acetylene weld for standing up to such treatment,
pear in the literature. The relegation of oxyfuel welding processes to which accords with what is already recognized” and that when
primarily non-structural applications in modern times contributes fur- quenching was included “the good bending qualities of acetylene
ther to the notion that it is not as reliable or effective a process. There welds is more noticeable still” ([9] p 51). The conclusions stated that
is an unfortunate lack of data on oxyfuel weld quality in structural appli- “a good acetylene weld will show excellent ductility and toughness,
cations using modern materials and methods. when subjected to bend, tensile, Izod, and repeated-impact tests, the re-
In the early decades of the 20th century oxyacetylene welding was sults surpassing those given by any electric weld, except those made
used in a wide array of large scale projects. A significant structural appli- with the very best class of metallic-arc electrode” (pg. 65) but also
cation was a demonstration multistory research laboratory building pointing out that acetylene welds have lower strengths and difficulty
with 20 ton crane load where all typical riveted connections were welding thicker base plate metals. A further series of long term tests
redesigned using oxyacetylene welds which was documented in detail were reported in [12], with the majority of these studies focused on im-
[46,47,48]. Structural applications also included constructing a building proving the quality of electric-arc procedures with some interspersed
from salvaged steel using all oxyacetylene welding for connections [24]. comparisons to oxyacetylene welds, addressing important issues such
An early textbook on welded building construction [32] stated that as heat-sink effects, importance of root inspections and vertical and
buildings had been fabricated using both arc and oxy-acetylene overhead welding. Meslier [36] reports a cost comparison of electric
welding, though electric arc was more common. Other applications in- arc and oxyacetylene welded butt welds of “thin” materials (4 to
cluded major pipeline work of 15 to 150 miles prior to 1922 30 mm (0.16 to 1.18 in)) providing justification for the efficiency of
[11,30,41,53,54,55], fusing of railroad tracks [27] and repair of salvaged oxy-acetylene welding at that time, and alludes to existing data of com-
ships during World War I [25]. The process was also used extensively in parable test results, though the paper discussion notes some bias in the
field and shop repair of heavy equipment, including gears, railroad and comparisons and the difficulty in oxyacetylene welding of thicker sec-
mining parts, indicating an inherent fatigue performance that was relied tions. Driscoll [28] reports comparison testing of welds on cast iron con-
on for these repairs [10]. Wagner [53] specifically noted the oxyacety- ducted by the British Cast Iron Association. Two bending tests, one
lene weld resistance to vibration of pipe manifolds which “had very tension test, weld metallurgy and visual examination of cut welds
good success with welded manifolds, having found no defects in any were completed on samples cut from 11 arc welded and 25 oxyacety-
we have built to date” ([53] pg. 27). In another paper oxyacetylene lene welded plates using a variety of welding rods, with a summary
welding was specifically recommended “for reclamation work where by Whyte [56] stating that “on the whole welding by the oxy-acetylene
welded joints are subjected to unusual strain and vibration” ([8] pg. flame gave rather better and more uniform results than welding by the
32). Papers typically note the efficiency of oxyacetylene welding in thin- electric arc” (pg. 176). Specific recommendations for oxyacetylene
ner plate applications with arc welding preferable in heavier members welding of cast iron are provided by Pierce [45] and Brownlie [23],
[36,49,51], though the definition of “thinner plate” is vague, with with the former noting that arc welding “is not used so widely as the
many oxyacetylene weld applications noting metals up to and exceed- oxy-acetylene process” (pg. 157). While historic cast iron structures
ing 25 mm (1 in) thickness [36,39,46,47]. will be riveted, this indicates that repair of cracks or modifications in
The historical research published on oxyacetylene weld capacities these structures may benefit from the oxyacetylene weld process.
are described here. Two early test programs occurred at the University Though further testing of oxyacetylene welds in structural applications
of Illinois (Whittemore [57] and Moore [39]). Whittemore [57] per- does not appear to have taken place in research laboratories, improve-
formed the welding of specimens himself and reports “weld efficiency” ments in filler materials and methods continued to occur [25,26], with
 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921 3

the process being the subject of entire chapters in many textbooks on Therefore the use of oxy-acetylene welding is allowed, but in the ab-
welding through the years up until present times [7,31,34,38,42,51]. sence of prequalified details is difficult to justify on a project. Currently,
Early structural engineering texts mention both arc and gas weld the AWS C4 Committee has a draft version of AWS C4.7/ C4.7M “Recom-
processes equally, with a shift to exclusive mention of arc welding pro- mended Practices for Safe Oxyacetylene Welding of Steel” [21] which
cesses in structural engineering textbooks after World War II, a parallel focuses on the process (rather than design issues) of oxyacetylene
development to the research programs reported during that time pe- welding.
riod. Hool and Kinne [33] note that “The A.I.S.C. has the following to Based on the literature, it is indicated that there were comparable
say concerning welding: ‘The use of the fusion welding process (some- quality and concerns between oxyacetylene and arc welding processes
times electric arc, sometimes oxy-acetylene) plays an increasing part in in the early years of welding. The shift to arc welding was motivated
the fabrication of structural steel’” (pg. 355) and defers to the American by weld efficiency, especially when welding thicker materials and
Welding Society documents as the definitive reference for design and high number of welds in larger projects and war efforts. After World
fabrication issues. Williams and Harris [58] note that “Specifications for War I, published test data for structural applications focused on the per-
Welded Highway and Railway Bridges and Code for Arc and Gas Welding formance of arc welding. This led to further use and pre-qualification of
in Building Construction, prepared by the American Welding Society, arc-weld details and by the 1960's structural engineering textbooks ex-
are applicable to the arc and gas fusion welding processes. A majority clusively discussed arc welding processes. The lack of information on
of structural welding is done by the electric-arc fusion method.” (pg. the quality and capacity of modern oxyacetylene welds is a barrier to
44) Later texts, such as McGuire [35] and Tall [52] only mention arc consideration of them as a viable weld process for structural applica-
welding processes, while Bresler and Lin [22] state “In structural-steel tions, even for individual welds or repairs.
work, metal arc welding is used almost exclusively… gas welding and
brazing may be used for special parts and fittings requiring small 3. Test program
welds” (pg. 159). AISC Design Guide 21 Welded Connections – A Primer
for Engineers [37] states “years ago, gas welding was used to construct a A series of testing was undertaken at the University of Massachu-
few steel buildings, but the process was found to be too slow and too ex- setts at Amherst to address the lack of data on the performance of oxy-
pensive to be commercially viable” (pg. 6). Throughout this transition it acetylene welds. The purpose was to evaluate the strength, consistency
is important to note that the quality of welds is not mentioned as differ- and fatigue resistance of fillet welds made by this process. In addition,
ing among processes in the literature, but merely efficiency and the re- the learning curve for obtaining quality oxyacetylene welds was
lated development of arc welding as the industry standard. observed.
This change in weld method preference was mirrored in language of
standards, specifications and codes. The 1928 standard AWS “Code for 3.1. Materials and specimens
Fusion Welding and Gas Cutting in Building Construction” [15] specified
“fusion welding is restricted to the arc and gas welding processes” and All welding and testing on this project was performed by the second
provided allowable strengths of welds under static and repeated load author, who previously had some limited experience with other weld
independent of weld process. This standard was followed for the con- processes. The student self-learned the welding process over a two
struction of at least one large structure, a 300 Ton mill building week period in January 2018, using Althouse et al. [7] as a primary
[46,47]. Per [32], “New York City, Chicago, and approximately 15O addi- source along with guidance from experienced welders who had used
tional cities in the United States; London, England, and many other for- the process in the past for non-structural applications. Preliminary
eign cities have adopted welding in their respective building codes.” welding included slight variations in technique until consistent visual
(pg. 5). The First Edition of AISC Steel Construction Manual [2] did not quality of welds were obtained (Fig. 1) and these were then tested
mention welding, while the Second Edition [3] addressed welding of through a modified fillet weld break test. Modification to the standard-
building structures, included excerpts from AWS [15] where “some- ized break test included decisions to test the specific welds that would
times electric arc, sometimes oxyacetylene” welds are specified as ac- be used in testing, specifically welding thinner plate, testing welds in
ceptable. In 1942, the American Institute of Steel Construction the final orientation representative of welds that would be tested in
provided guide specifications for the war effort [6] in which welding the research project where the fillet was along the edge of the thin
criteria for buildings are defined for both arc and gas methods, with al- plate materials (rather than a T section) and testing weld lengths of
lowable weld stresses and detailing specified independent of method only 32 mm (1–1/4 in) length (longer welded plates were saw cut
and referencing the 1934 AWS “Code for Arc and Gas Welding in Build- into specimens of this length). A typical specimen is shown in Fig. 2a.
ing Construction” [16]. Reference to AWS standards continued until the A total of 4 bend tests were completed on two separate welds. All of
Fifth Edition [4], at which point specifications are referenced to “riveted, this preliminary work was performed utilizing scrap materials of mild
bolted and arc-welded construction” with extensively expanded AISC carbon steel and weld materials similar to the test program. All speci-
weld criteria and the introduction of standard welded connections, mens were bent 90 degrees over the weld root (Fig. 2b–d) with no rup-
though still referencing the AWS “Code for Arc and Gas Welding in tures. Further bending to approximately 120 degrees resulted in partial
Building Construction” [17]. The 1947 Welding Encyclopedia [44] weld ruptures with progression of the crack as the bending continued
under “Building Construction” that states “arc and gas welding may be
employed, either alone or in combination with riveting, bolting or
other connection means permitted” for weld types and gives weld al-
lowable stresses that are not dependent on weld method. The title of
the 1946 to 1963 editions of the welding standard published by the
American Welding Society [18,19] changed from “Standard Code for
Arc and Gas Welding in Building Construction” to “Code for Welding
in Building Construction” and the content shows a shift to arc welding
processes as indicated by the prequalified details only encompassing
arc methods in the latter standard. The current AWS D1.1 Structural
Welding Code [20] is similar, specifically stating many accepted weld
processes in sections 3.2.1, 3.2.2 and 4.14.1 though oxyfuel processes
are not included. Sections 3.2.3 and 4.14.2 allow for “other welding pro-
cesses” subject to qualification testing approved by the engineer. Fig. 1. Practice welds submitted for review.
4 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921

Fig. 2. Typical modified bend test results A) Initial specimen B) 90 Degree bend side view C) 90 Degree bend back view D) 90 Degree bend front view.

(Fig. 3). All ruptures were through the center of the weld with no visual with nominal thickness of 2.28 mm (0.0897 in). As delivered specimens
signs of inclusions in the weld fracture plane. Full fusion was observed had an average measured thickness within 1% of nominal (2.24 to
in weld areas of the samples, with some bleed through of metal be- 2.28 mm (0.0882 to 0.0897 in)). Steel strength was determined by fol-
tween the two plates resulting in expected locations of incomplete fu- lowing procedures of ASTM E8 - Standard Test Methods for Tension
sion indicated by the shiny weld material near the intersection of the Testing of Metallic Materials [14]. Five test coupons were cut using a
two plates in Fig. 2C. Note that a typical T section bend test would Flow Mach 2B waterjet machine. Tension tests were carried out on an
never exceed 90 degrees of rotation, but these were forced to rupture Instron 3369 test frame with wedge grips (Fig. 4A). Four out of the
so that the fracture plane could be visually inspected. Full qualification five test specimens were tested to ultimate strength and one specimen
of acetylene welding would require a much fuller test regimen decided was not tested to retain an original sample (Fig. 4B). The base material
upon by the engineer in accordance with AWS D1.1 Section 4 [20], but had an average yield stress (Fy) of 245 MPa (35.5 ksi), ultimate stress
these preliminary results indicated welds of good quality after a rela- of (Fu) 316 MPa (45.9 ksi) and percent elongation of 44.9%. The yield
tively short training period. stress was slightly higher than the supplier's general data but within
Welding supplies for this project were a Victor compatible torch set the ASTM A 1008-18 typical range of 140 MPa to 275 MPa (20 ksi to
with oxygen cylinder Type M and acetylene cylinder Type B, each with 40 ksi).
capacity 1.13 m3 (40 ft3). Welding rod was 1.6 mm (1/16 in) Weldcote
R45 electrodes which meet AWS A5.2-R45 criteria. These rods are duc-
tile and reported by the manufacturer to be “excellent for steel sheets,
plates, pipes, castings and structural shapes.” The smaller tank sizes
used in this project limits the gas flow rates and therefore limited the
size of welds and thickness of base material that could be welded effec-
tively for the specimens. The preliminary welding confirmed published
recommendations [7] of a number 64 (orifice size) welding tip and
maximum 2.5 mm (3/32 in) base material thickness being optimal for
this welding equipment. Thicker base materials resulted in too great
of a heat sink for the flow rate of these specific tanks, resulting in diffi-
culty maintaining a consistent weld pool and fusion. With larger capac-
ity tanks welding of thicker materials would be possible, as was noted in
the literature. Resulting fillet welds on this thin material appear to have
vertical:horizontal leg lengths of approximately 3:5 for completed
welds rather than 1:1 as is often specified (Fig. 2A), partly due to fusion
into the left plate. Nominal design capacities were used assuming a 1:1
weld.
For all further welding and testing in this project, base materials
consisted of 13 gage ASTM A 1008 [13] CS Type B - 15 Prime Cold Rolled
Sheet (general properties reported by the supplier as nominal ultimate
stress Fu = 303 MPa to 358 MPa (43.9 ksi to 51.9 ksi), nominal yield
stress as Fy = 180 MPa to 240 MPa (26.1 ksi to 34.8 ksi) and percent Fig. 4. Base metal tension testing A) Base metal tension coupon installed in Instron 3369
elongation from 42% to 48% though no heat specific data was provided) test machine B) Coupons after testing.

Fig. 3. Further modified bend test results A) 120 Degree bend side view B) Initial cracking front view C) Final failure side view D) Final failure back view E) Final failure repositioned at initial
position.
 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921 5

Fig. 5. Test specimen dimensions.

Test specimens were designed to use a single specimen weld type (2.00 in by 2.50 in) plate on each end connected by two 25.4 mm by
and detailing to evaluate the static strength of fillet welds loaded in 101.6 mm (1.00 in by 4.0 in) plates as shown in Fig. 5. Components
shear and then evaluate the fatigue performance of these specimens are joined by a full depth (2.28 mm (0.0897 in)) fillet welds 25.4 mm
and compare to reported fatigue details. Specifically, a fatigue specimen (1.0 in) long. All specimen material was first rough cut with a band
corresponding to detail type 4.1b “Longitudinal Fillet Welded End Con- saw and then milled to final dimensions using a Bridgeport J-Head ver-
nection” from Table A3.1 in the AISC-360-16 [5] was used. For a base tical milling machine. A welding jig was fabricated to align components
metal thickness of less than 13 mm (½ in) this results in a stress cate- during welding to ensure repeatable accuracy in all test specimens.
gory E while for thicker base metal the category would be E'. This detail Weld capacity was calculated following AISC-360-16 [5] Section J2.4.
was selected to provide a fatigue susceptible detail in order to evaluate It was determined that the capacity of the four, 25 mm (1 in) long fillet
the performance of oxyacetylene produced welds subjected to fatigue welds would have a nominal capacity of 30.5 kN (6.85 kips) neglecting
load, which had very little information in the literature. In an actual con- resistance (ϕ) factors (0.6xFEExx4 × 2.28 × 0.707 = 30.5 kN) (6.85 kips)
nection the welds would have a capacity exceeding the base metal, but and would fail prior to rupture of the base materials (Fux2.28 × 2 = 36.6
capacities were chosen to determine static weld capacities and subject kN) (8.23 kips), block shear of the larger plate (0.6xFyx50.8 × 2.28
welds to higher fatigue stress ranges than would be typical in a design + Fux25.4 × 2.28 = 35.3 kN) (7.94kips) or shear yielding of the base
for this detail. The closest detail to this in the AASHTO [1] specification metal at the weld interface (0.6xFyx50.8 × 2.28 × 2 = 34.0kN) (7.64
is detail 7.2 in Table 6.6.1.2.3-1, also a category E detail with similar de- kips). However, tension yield of the base materials (Fyx2.28 × 50.8 =
sign requirements, though this is specifically for an angle or T section 28.3 kN) (6.37 kips) is expected prior to weld rupture. Therefore, it is
welded along three faces. Specimens consist of a 50.8 mm by 63.5 mm expected that the specimen base metals will see some significant yield-
ing prior to weld failures.

3.2. Test results

Testing of five specimens in direct tension was completed using an

Instron 8501 test machine (Fig. 6) with hydraulic grips, with data col-
lected for three specimens (A, D, F) and results shown in Fig. 7. The
other three specimens (B, C, E) displayed similar results during testing
with respect to duration of the test and observed maximum force
which was the only data collected for these tests. Test B was terminated

Fig. 6. Test specimens in Instron 8501 test machine. Fig. 7. Specimen static tension test results.
6 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921

when partial rupture of a fillet weld was first observed while all others failure. During testing of other specimens there was significant plastic
progressed until no further load was resisted. All specimens failed at deformation of base materials prior to partial weld rupture as shown
two of the four connections by shearing of one or two fillet welds at in Fig. 8D and E, which was predicted with weld failures expected at a
the connection, with an average capacity of 31.9 kN (7.18 kips) (speci- capacity between yield and rupture capacities of the base metal. Fig.
mens A, D and F) defined as the capacity at 2.5 mm (0.1 in) of cross 8F shows a typical non-ruptured weld location with some slight residual
head displacement (representing the shear yield plateau in the test re- base metal deformations. Weld ruptures did not progress through the
sults and limit of usable deformation in a connection). This load repre- root of the welds, but were typically along the interface of the
sents a stress of 279 MPa (40.5 ksi) on the base material (using 25.4 mm (1 in) plate base metal, with significant plastic weld deforma-
measured plate thickness of 2.25 mm (0.0886 in), a value between Fy tions apparent (Fig. 8). Still, AISC provisions predicted weld capacities
and Fu. quite well as 30.5 kN (6.85 kips) nominal design capacity (neglecting
A typical specimen (A) after testing is shown in Fig. 8A. In this spec- ϕ factors) versus experimental results of 31.9 kN (7.18 kips).
imen the front top weld ruptured with minimal base metal distortion A total of fourteen fatigue specimens were loaded cyclically in ten-
(Fig. 8B) while the bottom welds ruptured on the back plate with distor- sion at a maximum load range of 29 to 89% of the base metal yield ca-
tions shown in Fig. 8C. It is likely that the bottom location ruptured sub- pacity (equivalent to 26 to 78% of the 31.9 kN (7.18 kip) static load
sequent to the top rupture and was therefore subject to eccentric load at capacity as defined from the static test results of Fig. 7) in the same

Fig. 8. Typical specimen static tension test photos A) Specimen A overview B) Specimen A top front weld rupture with minimal base metal distortion C) Specimen A bottom rear weld
rupture showing significant base metal distortion D) Specimen E weld rupture with significant base metal distortion E) Specimen F weld rupture with significant base metal distortion
F) Specimen C typical intact welds.
 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921 7

Table 1
Fatigue test results.

Specimen Percent Yield of Base Metal Percent of Static Capacity Load Amplitude Stress Range Test Cycles Cycle Rate Run Time

% % kN (kip) MPa (ksi) # Hz Hours

F89A-1 89 78 12.46 (2.80) 218 (31.6) 21,554 10 0.30

F89B-6 89 78 12.46 (2.80) 218 (31.6) 22,590 varieda 0.29a
F84A-2 84 74 11.79 (2.65) 206 (29.9) 28,002 10 0.39
F84B-7 84 74 11.79 (2.65) 206 (29.9) 52,577 20 0.73
F73A-3 73 64 10.23 (2.30) 179 (26.0) 68,500 10 0.95
F73B-8 73 64 10.23 (2.30) 179 (26.0) 65,659 20 0.91
F73C-14 73 64 10.23 (2.30) 179 (26.0) 128,641 20 1.79
F62A-4 62 54 8.67 (1.95) 152 (22.0) 259,709 10 3.61
F62B-9 62 54 8.67 (1.95) 152 (22.0) 175,835 20 2.44
F62C-13 62 54 8.67 (1.95) 152 (22.0) 227,096 20 3.15
F51A-5 51 45 7.16 (1.61) 125 (18.2) 504,407 20 7.01
F51B-10 51 45 7.16 (1.61) 125 (18.2) 598,287 20 8.31
F40A-12 40 35 5.64 (1.268) 99 (14.3) 1,591,872 20 22.11
F29A-11 29 26 4.09 (0.92) 72 (10.4) 11,321,328 20 157.24

Notea: Specimen 6 investigated the use of different cycle rates, using 50 Hz for the first 659 cycles, 40 Hz for the next 848 cycles and then 25 Hz for the remaining 21,083 cycles. All sub-
sequent tests (7 to 14) used 20 Hz. All prior tests (1 to 5) used 10 Hz. Run time for specimen 6 includes evaluating cycle rates before re-starting the test.

and c is the order of specimen testing and marking on the specimen in

Fig. 10.
Fig. 9 shows very consistent fatigue results of these oxyacetylene fil-
let welds, with an R2 value of 0.9644. The trend line is shallower than
the standardized AISC and AASHTO categories, resulting in significantly
better performance than predicted by design equations at very high cy-
cles, and E category or better beyond approximately 50,000 cycles. Re-
sults are between E and E' for fewer cycles. While US practice is to
assume that failure is controlled by cracking in the base metal, Eurocode
EN 1993-1-9 [29] includes separate checks for shear in the weld. Base
metal criteria in all codes for this detail use a power of 0.333 while the
Eurocode weld check uses a power of 0.200 for weld shear. The data
in Fig. 9 (power of 0.178) is closer to the weld shear requirement,
though as noted below all failures initiated in the base metal. This
may indicate positive effects due to a larger heat affected zone and
resulting residual stress distribution in the oxyacetylene process and is
Fig. 9. Specimen fatigue tension test results. supported by historical applications in conditions with stress concentra-
tions and repair of fatigue loaded components.
Table 2 specifies the observed material failures at each of the four
test frame as the static tension specimens. A preload tension of 0.9 kN connections in a specimen examples shown in Fig. 11. All failures of
(0.200 kips) was applied to each specimen and tensile cyclic loads test specimens exhibited typical failure modes of crack initiation at the
(stress ranges) were then applied, resulting in load ratios R (minimum end weld terminations and extending into the base metal (Fig. 12), de-
stress/maximum stress) of 0.02 or less. Thirteen specimens were tested noted as 1 or 2 in Table 2. Eleven of the samples ruptured through the
until failure and one (tested at 0.29Fy) was discontinued after running 25.4 mm (1 in) plates while two (F51B-10 and F62C-13) ruptured
11.3E6 cycles. Specifics of each specimen and results are presented in through the 50.8 mm (2 in) wide end plates. Even in specimens
Table 1. Results are plotted in Fig. 9 along with category C, D, E and E' de- where weld ruptures occurred, either full or partial, the initiation of fail-
sign lives as designated by both AASHTO [1] and AISC [5]. Specimen no- ure was a fatigue crack propagating through base metal materials. With
menclature is Fab-c where F stands for fatigue test, a is the stress range welds sized to initiate static failures (static test results) the lack of initial
as a percent of base metal Fy (determined by the base material tension rupture in weld materials indicates the acceptable fatigue resistance of
tests), b distinguishes the specimen (A, B, etc. at this stress range) oxyacetylene fillet welds. A typical design would have higher strength

Fig. 10. Fatigue specimens after testing.

8 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921

Table 2
Fatigue test failure observations.

Failure Observeda

Location

Sample Number Top Front Top Back Bottom Front Bottom Rear Notes

F89A-1 1* W 1 1* W* One weld failure through throat

F89B-6 1 1 W* 1* N Some non-fusion in weld. Weld failure at location of final plate fracture
F84A-2 1 W* W 1* 1*
F84B-7 1 2* 2* N 1
F73A-3 N N 1 W One weld failure through throat, some non-fusion at location
F73B-8 1 1 N N
F73C-14 1 1 1* 1*
F62A-4 1 1* N 1 W* Weld failure at location of final plate fracture
F62B-9 1 1* 1* 1
F62C-13 N N 2 W* 2 1* Weld failure at location of final plate fracture
F51A-5 N N 1 1
F51B-10 2 2 1* N
F40A-12 1 W* 1 N N
F29A-11 N N N N
a
1=failure in 25.4 mm (1 in) plate; 2=failure in 50.8 mm (2 in) plate; W=weld failure; N=no sign of rupture; *=initiation of failure but not complete

welds making them even less susceptible to fatigue fractures, though ruptures show significant deformation prior to rupture. Examples of
other connections could put higher relative stresses on the welds. these ruptures are shown in Fig. 13. Note that the apparent lack of fusion
Full weld ruptures were only observed in specimens F89A-1, F84A-2 shown in Fig. 13D is actually showing the irregular edge of fusion under
and F73A-3, though a fatigue crack propagation in base metal was pres- the 25.4 (1 in) plate as the plate pries away from the 50.8 mm (2 in)
ent in all of these specimens and initiated failure. All of these weld plate during the final weld rupture. Specimens F89A-1, F89B-6,

Fig. 11. Fatigue specimens rupture modes A) 25.4 mm (1 in) plate failure B) 25.4 mm (1 in) plate partial failure C) 50.8 mm (2 in) plate failure D) 50.8 mm (2 in) plate partial failure E)
Weld failure F) 25.4 mm (1 in) plate failure and weld partial failure.
 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921 9

F84A-2, F62A-4, F62C-13 and F40A-12 exhibited partial weld ruptures,

with examples shown in Fig. 14. These similarly occurred after signifi-
cant deformations and are considered secondary failures.

4. Conclusion

Oxyacetylene welding has a long history of use in industrial applica-

tions, though the structural applications were not pursued beyond the
1930's due to the greater efficiency of mass production using arc
welding processes. However, this process may be of practical use in lim-
ited applications. This test program showed that 2.5 mm (3/32 in) fillet
welds on mild carbon steel base material exceeded the static design ca-
pacities based on current weld design practice, with very consistent per-
formance. In addition, fatigue testing of these welds in a Category E
detail demonstrated very consistent fatigue capacity and significantly
exceeded the predicted life for a fatigue prone detail (Category E) for cy-
cles exceeding 200,000. Based on these results, the authors propose that
oxyacetylene welds could be considered as a viable alternative to arc
weld processes when appropriate, subject to qualification testing. This
may be relevant in the case of emergency repairs, welding in remote
Fig. 12. Fatigue specimens base metal failures A) Rupture surfaces through 25.4 mm (1 in) areas or where electrical sources are not present. Further testing of
plates B) Rupture surface through 50.8 mm (2 in) plate.
other weld details and thicker materials is recommended to expand

Fig. 13. Fatigue specimen weld ruptures A) Rupture through throat at 25.4 mm (1 in) plate B) Matching rupture through throat at 50.8 mm (2 in) plate C) Rupture at edge of 25.4 mm (1 in)
plate D) Matching rupture at edge of 25.4 mm (1 in) plate with significant weld deformation.

Fig. 14. Fatigue specimen partial weld rupture A) At edge of 50.8 mm (2 in) plate B) At edge of 25.4 mm (1 in) plate C) Partial rupture with block rupture.
10 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921

the potential applications, but based on the literature review of early [8] Anon, Oxy-acetylene welding in electric railway shops, J. A.W.S. Am. Weld. Soc. 1
(1922) 32–35.
20th century welding it is expected that weld quality would be compa- [9] Anon, Fusion-welded Pressure Vessels, British Engine Boiler and Electrical Insurance
rable to arc processes for base metal thicknesses up to 25.4 mm (1 in), Co, 1928 6–121 LTD Technical Report for 1928.
only lacking in efficiency. The process may be very well suited for [10] Anon, Iron mining, Acetylene J. Int. Acetylene Assoc. 30 (8) (1929) 331–332.
[11] Anon, Pipe lines, Acetylene Journal. International Acetylene Association 30 (8)
repairing cast iron structures. (1929) 332.
The following specific conclusions may be drawn: [12] Anon, Welding – Experimental, British Engine Boiler and Electrical Insurance Co, 1931
50–147 LTD Technical Report for 1930-1931.
• The development of oxyacetylene welding was similar to arc welding [13] ASTM Standard A1008/A1008M – 18, Standard Specifications for Steel, Sheet, Cold-
until the 1930's. Larger testing programs at this time focused on arc Rolled, Carbon, Structural, High Strength Low-Alloy, High Strength Low-Allow with
Improved Formability, Solution Hardened, and Bake Hardenable, ASTM Interna-
welding exclusively, therefore only anecdotal evidence of oxyacety-
tional, West Conshohocken, PA, 2018.
lene performance under fatigue and welding of thicker base metal ex- [14] ASTM Standard E8, Standard Test Methods for Tension Testing of Metallic Materials,
ists. ASTM International, West Conshohocken, PA, 2016.
[15] AWS Code for Fusion Welding and Gas Cutting in Building Construction, American
• The preference for arc welding over oxyacetylene welding over time
Welding Society. New York, NY, 1928.
was due to efficiency of welding, though quality of welds from these [16] AWS Standard Code for Arc and Gas Welding in Building Construction, American
weld processes are not differentiated in the literature. Welding Society. New York, NY. 1934.
• Static testing of short fillet welds in shear were completed and indi- [17] AWS Standard Code for Arc and Gas Welding in Building Construction, American
Welding Society. New York, NY, 1941.
cate very consistent static strengths which exceed design code predic- [18] AWS D1.0 Standard Code for Arc and Gas Welding in Building Construction, Amer-
tions. ican Welding Society. New York, NY, 1946.
• Fatigue testing of short fillet welds in shear were completed and indi- [19] AWS D1.0 Code for Welding in Building Construction, American Welding Society.
New York, NY, 1963.
cate design life that significantly exceeds those predicted by design [20] AWS D1.1 Structural Welding Code - Steel, American Welding Society. Miami, FL,
codes for cycles exceeding 200,000. The fatigue testing exhibited 2017.
very consistent performance. [21] AWS C4.7/ C4.7M Recommended Practices for Oxyfuel Gas Welding of Steel –
DRAFT Document, AWS C4 Committee on Oxyfuel Gas Welding and Cutting, Amer-
• Oxyacetylene welding requires very minimal tools and equipment be- ican Welding Society, Miami, FL, 2019 (in press).
yond what would already be on most structural job sites. The training [22] B. Bresler, T.Y. Lin, Design of Steel Structures, John Wiley and Sons, Inc, New York,
required for completing welds of sufficient quality was low, though NY, 1960.
[23] T. Brownlie, Gas welding of cast iron, N. Z. Eng. 5 (11) (1950) 1027–1028Institution
more training would be required as the weld size increases.
of Professional Engineers New Zealand.
• Oxyacetylene welding would be advantageous and should be consid- [24] G.O. Carter, Oxy-acetylene welding and cutting in the structural field, Weld. J. 3 (11)
ered for emergency repairs, welding in remote areas or where electri- (1924) 13–17American Welding Society.
[25] G.O. Carter, The oxy-acetylene industry in 1942, Proceedings of the annual meeting
cal sources are not present.
of the Compressed Gas Manufacturers’ Association 1942, pp. 53–66.
• Oxyacetylene welding is not pre-qualified by AWS, so would require [26] G.O. Carter, Historical development of oxy-acetylene welding and cutting, Sheet
qualification testing for each application. Metal Worker 35 (2) (1944) 26–28.
Oxyacetylene welding is not proposed as a replacement for arc weld [27] R. Cummins, Uses oxy-acetylene process to butt-weld rails in open track, Railw. Eng.
Maint. 33 (12) (1937) 894–904.
processes. It is less efficient, does not have pre-qualified details, and [28] W.J. Driscoll, Report on a Preliminary Investigation of the Welding of Cast Iron,
does not have a structural welding certification program. However, for Transactions of the Institute of Welding. London 4 (1) (1941) 164–173.
very specific applications it may be a very useful tool that should be [29] EN 1993-1-9, Eurocode 3: Design of steel structures - Part 1-9: Fatigue, CEN Euro-
pean Committee for Standardization, Brussels, Belgium, 2005.
considered. [30] I. Gaston, Another oxy-acetylene welded river crossing, Weld. J. 9 (4) (1930) 52–60.
[31] R. Granjon, P.A. Rosemberg, Practical Manual of Autegenous Welding (Oxy-Acety-
lene), Charles Griffinand Co. Limited, London, ENG, 1915 234 Pages.
[32] R.S Hale, Welded Steel Construction, Pitman Publishing Corporation, New York, NY,
Declaration of Competing Interest 1938.
[33] G.A. Hool, W.S. Kinne, Structural Members and Connections, 2nd edition McGraw-
The authors declare that they have no known competing financial Hill Book Company, Inc, New York, NY, 1943.
[34] R.S. Johnston, An Investigation of Oxyacetylene Welding and Cutting Blow-Pipes,
interests or personal relationships that could have appeared to influ-
with Especial Reference to their Design, Safety, and Economy in Operation, Techno-
ence the work reported in this paper. logic Papers of the Bureau of Standards. Report #200, Government Printing Office,
Washington D. C, 1921 , 108 pgs.
Acknowledgements [35] W. McGuire, Steel Structures, Prentice-Hall, Inc, Englewood Cliffs, N.J, 1968.
[36] R. Meslier, Comparison between results obtained with modern methods of oxy-
acetylene welding and electric arc welding, The Twelfth International Congress of
The assistance of Mark Gauthier in the laboratory was invaluable, as Acetylene, Oxy-Acetylene Welding and Allied Industries, London, Caxton Hall,
well as Michael Voight for his donation of time, advice and shop space Westminster 1936, pp. 808–812 , 8-13 June.
[37] D.K. Miller, Welded Connections – A Primer for Engineers. AISC Steel Design Guide
for preparation of specimens. Jim Papritan, emeritus faculty at Ohio No. 21, American Institute of Steel Construction, Chicago, IL, 2006 151 Pages. Re-
State University provided insight into historical context. vised 2015.
This research did not receive any specific grant from funding agen- [38] S.W. Miller, Oxy-Acetylene Welding, The Industrial Press, New York, N. Y, 1916 287
Pages.
cies in the public, commercial, or not-for-profit sectors. [39] H.F. Moore, Tests of Oxyacetylene Welded Joints in Steel Plates. University of Illinois
Bulletin No. 98, University of Illinois, Urbana IL, 1917 26 pgs.
References [40] A.A. Noyes, The nitrogen problem in relation to the war, J. Wash. Acad. Sci. 8 (12)
(1918) 381–394Washington Academy of Sciences.
[1] AASHTO LRFD Bridge Design Specifications, American Association of State Highway [41] Linde Air Products Company, Useful Information and Tables Covering Oxy-Acetylene Pipe
and Transportation Officials, 8th ed, 2017 Washington D. C. Welding and Pipe Data, Linde Air Products Company, New York, NY, 1923 51 pages.
[2] AISC Steel Construction Manual, American Institute of Steel Construction, 1st ed., [42] Linde Air Products Company, The Oxy-Acetylene Handbook: A Manual on Oxy-Acet-
1927 New York, NY. ylene Welding and Cutting Procedures, Ninth edition Union Carbide and Carbon
[3] AISC Steel Construction Manual, American Institute of Steel Construction, 2nd ed., Corporation (Linde Air Products Company), New York, NY, 1951 587 pages.
1934 New York, NY. [43] L.B. Mackenzie, H.S. Card (Eds.), The Welding Encyclopedia, 2nd editionWelding En-
[4] AISC Steel Construction Manual, 5th ed. American Institute of Steel Construction, gineer Pub. Co, Chicago, IL, 1922 , 388 Pages.
1946 New York, NY. [44] L.B. Mackenzie, T.B. Jefferson, The Welding Encyclopedia, 12th edition Welding En-
[5] AISC Steel Construction Manual, 15th ed. American Institute of Steel Construction, gineer Pub. Co, Chicago, IL, 1947 Pages.
2017 Chicago, Il. [45] J.G. Pierce, Welding of cast iron: a review, Transactions of the Institute of Welding.
[6] AISC National Emergency Specifications for the Design, Fabrication and Erection of London 5 (4) (1942) 156–163.
Structural Steel For Buildings, Specifications for the War Production Board, Ameri- [46] H.M. Priest, Design and fabrication of oxy-acetylene welded buildings, Weld. J. 9 (2)
can Institute of Steel Construction Inc, New York, NY, 1942. (1930) 33–43.
[7] A.D. Althouse, C.H. Turnquist, W.A. Bowditch, K.E. Bowdich, M.A. Bowditch, Modern [47] H.M. Priest, Moss, Gas-welding Design and Erection of a 300-Ton Mill Building, En-
Welding, 11th ed. The Goodheart-Willcox Company, Tinley Park, IL, 2013. gineering News Record. 103 (24) (1929) 928–931.
 S.A. Civjan et al. / Journal of Constructional Steel Research 168 (2020) 105921 11

[48] H.M. Priest, Moss, Gas-welding Shop Fabrication of a 300 Ton Mill Building, Engi- [60] W.M. Wilson, W.H. Munse, W.H. Bruckner, Fatigue strength of fillet-weld, plug-
neering News Record. 103 (25) (1929) 964–967. weld, and slot-weld joints connecting steel structural members, University of Illi-
[49] P.L. Roberts, Welding – the modern method of production, Proceedings of The Insti- nois Bulletin No. 380 46 (68) (1949)University of Illinois. Urbana IL. 104 pgs.
tution of Production Engineers. November 1923, pp. 517–521. [61] W.M. Wilson, W.H. Bruckner, T.H. McCrackin Jr., Test of riveted and welded joints in
[50] L.W. Schuster, The Strength and Design of Fusion Welds for Unfired Pressure Ves- low-alloy structural steels, University of Illinois Bulletin No. 337 40 (5) (1942)Uni-
sels, Proceedings – Institution of Mechanical Engineers 118 (1) (1930) 319–417. versity of Illinois. Urbana IL. 76 pgs.
[51] H.S. Smith, A.F. Brennan, Instruction Book on Oxy-Acetylene Welding and Cutting, [62] W.M. Wilson, W.H. Bruckner, J.V. Coombe, R.A. Wilde, Fatigue tests of welded joints
4th Ed The Presto-Lite Co. Inc, Indianaplis IN, 1916 45 Pages. in structural steel plates, University of Illinois Bulletin No. 327 38 (27) (1941)Uni-
[52] L. Tall, Structural Steel Design, 1st edition The Ronald Press Company, New York, N. versity of Illinois. Urbana IL. 86 pgs.
Y., 1964 [63] W.M. Wilson, Tests of welds a report of an investigation, University of Illinois Bulle-
[53] N.E. Wagner, Oxy-Acetylene Welding of Oil Pipe Lines, J. Am. Weld. Soc. (1922) tin No. 21 28 (11) (1930)University of Illinois. Urbana IL. 37 pgs.
20–29. [64] W.M. Wilson, W.H. Munse, E.S. Snyder, Fatigue strength of various types of butt
[54] WeldGuru. Anon, Metalworking and Welding Timeline, 2018 WeldGuru. https:// welds connection steel plates, University of Illinois Bulletin No 384 47 (50)
weldguru.com/welding-history/ Accessed 7/2/2018. (1950)University of Illinois. Urbana IL. 60 pgs.
[55] WeldingHistory.org. Anon, Welding Timeline, 2019 http://www.weldinghistory. [65] W.M. Wilson, W.H. Bruckner, T.H. McCrackin Jr., H.C. Beeded, Fatigue tests of com-
org/whfolder/folder/wh1900.html Accessed 7/2/2018. mercial butt welds in structural steel plates, University of Illinois Bulletin No. 344
[56] A.G. Whyte, The welding of iron, N. Z. Eng. 3 (2) (1948) 176Institution of Profes- 41 (8) (1943)University of Illinois. Urbana IL. 140 pgs.
sional Engineers New Zealand. [66] W.M. Wilson, A.B. Wilder, Fatigue tests of butt welds in structural steel plates, Uni-
[57] H.L. Whittemore, The Strength of Oxyacetylene Welds in Steel, University of Illinois versity of Illinois Bulletin. No, 310 36 (42) (1939)University of Illinois. Urbana IL. 58
Bulletin No. 45 8 (4) (1910)University of Illinois. Urbana IL. 65 pgs. pgs.
[58] C.D. Williams, E.C. Harris, Structural Design in Metals, The Ronald Press Company, [67] W.M. Wilson, W.H. Bruckner, J.E. Duberg, H.C. Beeded, Fatigue strength of filler-weld
New York, NY, 1949. and plug-weld connections in steel structural members, University of Illinois Bulle-
[59] W.M. Wilson, W.H. Munse, Fatigue strength of various details used for the repair of tin No. 350 41 (30) (1944)University of Illinois. Urbana IL. 94 pgs.
bridge members, University of Illinois Bulletin No. 382 47 (30) (1949)University of
Illinois. Urbana IL. 60 pgs.