Submerged Arc Welding process: enhancement of production performance based on metallurgical observations

Welding processes are widely used technologies in the industrial context for creating permanent connections between mechanical components. This popularity is due to their versatility, which arises from the numerous available process variants and the multiple advantages they offer compared to other joining techniques. In the manufacturing context, where devices often operate in extreme conditions, the quality of welds becomes a critical factor in ensuring the safety and reliability of the manufactured products. Additionally, the increasingly stringent design speci�cations demanded by customers must be carefully considered. To address these needs and to de�ne the optimal roadmap for the investigated process condition, an experimental investigation was conducted on the Submerged Arc Welding process. The experimental trials involved butt joints of ASTM A516 Gr.70 carbon steel plates with different thicknesses in a �at position, utilizing a U-shaped chamfer and a multi-pass welding technique. For each weldment, the effects of the main process parameters on the qualitative characteristics of the manufactured products was evaluated from a metallurgical perspective. This evaluation included an in-depth metallographic analysis, which measured the dimensions of the heat-affected zones and the amount of ferrite and perlite in the �nal joint microstructure. Furthermore, the joint quality was assessed with regard to mechanical strength through hardness measurements. By analysing the experimental data, the paper provides a valuable contribution for increasing the productivity of the investigated welding process, while simultaneously meeting the speci�ed industrial quality requirements for the products made of medium-thickness carbon steels.


Introduction
Welding processes play a fundamental role in the manufacturing industry, signi cantly contributing to the fabrication of a wide range of products across diverse sectors.These sectors include automotive, aerospace, construction, shipbuilding, consumer goods, industrial machinery, infrastructure, and various applications, spanning from small-scale components to large structures [1,2].
Welding is a primary technique for permanently join materials by applying an appropriate amount of energy.Its primary objective is to create strong and durable connections between metal parts, often achieving a continuity of material at the microstructural level.This capability enables the fabrication of assemblies that require structural integrity.Furthermore, welded structures can be engineered to withstand different environmental conditions, making them suitable for use in a variety of settings.This adaptability is particularly crucial in industries where products are exposed to uctuating temperatures, pressures, and corrosive elements [3].
Ongoing research and innovation in welding techniques contribute to the development of new materials and processes.This innovation propels advancements in manufacturing technologies, resulting in improved product quality and performance.While welding is primarily used for connecting metallic materials, the numerous process variations also allow for the welding of polymeric, ceramic, and composite materials [4][5][6][7].
This study speci cally focuses on a particular arc welding variant, namely the Submerged Arc Welding process (SAW) [8].SAW is commonly employed for joining large-sized components and thick material sections, particularly in industries such as shipbuilding, oil and gas, and structural fabrication.It nds application in various contexts, including the construction of tube and pipeline joints, pressure vessels, nuclear containers, wind tower structures, as well as the fabrication of oil, water, or lique ed natural gas tanks [9][10][11].SAW offers several key advantages, including: i) the high deposition rate and productivity, ii) deep penetration, made possible by the high current intensities that can be utilized, iii) the capability to weld thick and large pieces in a single pass, and iv) the ability to achieve high-quality welds [12,13].
To perform the SAW process, it is essential to establish an arc between a continuously fed electrode (a wire) and the workpiece.The arc's parameters must be set to reach the melting temperature of both the base material and the ller material, allowing the creation of the weld bead.Temperature control plays a fundamental role in the joint formation, as a thermal cycle occurs during the welding process.This cycle induces changes in the metallurgical structure, affecting the material's microstructure, and altering its physical and mechanical properties in the nal joint [9].
The uniqueness of SAW lies in submerging the arc beneath a layer of granular ux, stored in a hopper.This granular ux acts as a protective coating over the molten pool, shielding it from the surrounding atmosphere.Moreover, it aids in reducing splatter and fumes, ultimately improving working conditions compared to other arc welding variants.It is important that the ux has a lower melting temperature and density than the base metal.During the process, a portion of the ux melts and solidi es over the weld bead, while the remaining granular ux is removed through an aspiration system at the end of each pass and then can be reused.The operation of the electric motor is controlled by a unit that ensures a continuous delivery of the wire from the coil through feeding rolls.This controlled wire speed ensures that the electrode wire's feed rate matches its melting rate, resulting in a stable and controlled arc [12,13].The operating principle of the SAW is depicted in Fig. 1 below.
Due to the inherent nature of the welding process, the materials involved are subjected to thermal cycles.A high speci c heat input results in elevated temperatures and reduced cooling rates, which in turn promote the growth of crystalline grains.The cooling rate is also in uenced by the thickness of the plates.In this context, thermal energy is primarily dissipated through conduction across the thickness of the two welded pieces, while the portion dissipated to the external environment through radiation and convection is negligible.The power losses to the external environment are further reduced due to the insulating behaviour of the granular ux which shields the molten pool during the welding process.
Therefore, for the same speci c heat input, the greater the thickness of the joint components, the faster their cooling occurs, leading to a more severe thermal cycle.
In this study, the focus was on a speci c joint con guration and on the analysis of the heat dissipation within the material at the microscopic level was conducted.This analysis involved observing the heat affected zone (HAZ), which represents the portion of the base material adjacent to the melted zone undergoing solid-state transformations.These transformations cause microstructural variations compared to the original base material.The grain morphology and the extent of the HAZ primarily depend on the base material used in the welding process, especially its chemical composition and the initial metallurgical structure.The latter is also in uenced by the manufacturing process adopted for sheet metal production and the parameters that de ne the thermal cycle.These parameters (mainly welding current, arc voltage and travel speed) play a crucial role in the process.Predicting the evolution of the microstructure within the HAZ can be challenging, especially in the case of multi-pass welding, as subsequent passes generate heating effects on the underlying microstructure obtained from previous passes [14].
Saleem and Lakshmanaswamy [15] focused on optimizing input setting parameters to enhance weld bead characteristics.They developed a mathematical model, and their experimental results closely matched the predicted values.Küçüköner et al. [16] used SAW to join two types of steel and assessed the quality and mechanical properties of the welded samples.Microstructure examinations revealed dendritic formation in the welded metal, with larger grains near the melting boundary on the primary material side.
Renwick [17] evaluated ux composition and its impact on the weld bead.Sharma and Chhibber [9] designed various submerged arc welding ux compositions using an extreme vertices design approach and studied their effects on the weld bead and mechanical properties.The ux, in combination with welding parameters, signi cantly in uences arc stability and the nal joint.Datta et al. [18] developed mathematical models to understand the in uence of process parameters on the nal geometry of the weld bead.Prasad and Dwivedi [19] investigated the effect on the microstructure and on the heat affected zone.Rathi [20] highlighted that electrode diameter, trolley speed, and wire feed rate are key factors for controlling weld bead dimensions.Vishwakarma and Dwivedi [21] applied the Taguchi method to optimize ve welding parameters to achieve minimal hardness and maximum tensile strength.
The literature review underscores that the dimensions of the weld bead and all mechanical properties of the welded joint are signi cantly affected by several parameters of the welding process investigated.Understanding how to control the process is crucial for ensuring the mechanical performance of the welding joint and meeting quality standards required by the normative of the industrial sector considered as well as by the customers.
In this study a novel approach was assessed to analyse the interaction of various factors, including the welding parameters, which the heat input mainly depends on, the joint design and the number of passes.
The investigation was aimed at optimizing the process parameters in terms of production performance, while considering microstructure and, consequently, the mechanical properties of the speci c joint con guration under investigation.

Materials
ASTM A516 steel, also known as ASME SA516, was employed for the experimental tests of this work.It is a carbon-manganese steel, representing the standard speci cation for the fabrication of pressure vessels [22].This material is typically chosen for applications with moderate to low operating temperatures due to its excellent notch toughness properties [23].
The ASTM A516 steel plates used in the experiments were obtained through the hot-rolling process.
Speci cally, the dimensions of the plates involved in the experimental tests were 100 mm in width, 400 mm in length and thicknesses of 10 mm and 25 mm.ASTM A516 steel comes in different grades (55, 60, 65, and 70), each offering varying levels of mechanical strength.Grade 70 is the one used to the aim of this work.It has higher yield strength and tensile strength values than other grades, provides higher tensile strength at low temperatures.These characteristics make it suitable for products operating under high-pressure conditions.Key mechanical properties of the material are summarized in Table 1.Elongation at Break in 50 mm (min) 21% Table 2 shows the chemical composition of the weld material.The main alloying elements in ASTM A516 Gr. 70 steel are carbon (C), manganese (Mn) and silicon (Si).The amount of these elements affects the mechanical properties of the material and its microstructure in a different manner.Indeed, C enhances heat treatment hardening and wear resistance in steels, but if it is present in excess it results in the formation of martensite or bainite.These are undesirable structural constituents because they can lead to hydrogen cracking.Mn makes possible to increase the steel's toughness and contrasts the damaging effects of sulfur, but it causes a deterioration of the material's resilience property.In addition, from the microstructural point of view, it was found that the presence of Mn (1.37 wt.%) in ASTM A516 Gr. 70 steel promoted the acicular ferrite formation in the joint molten zone and reduced the grain size of polygonal ferrite in heat-affected regions in multi-pass welds [23].Finally, the presence of Si increases the strength and hardenability of steels and acts as a deoxidizer, however, it negatively affects the toughness and weldability of the material.As for the ller material for the welded joint, a low-carbon steel coded as AWS 5.17 of EH14 grade was used.
This high-Mn, copper-coated solid wire is typically employed in single-pass and multi-pass SAW procedures.
The chemical composition of the electrode used is provided in Table 3 while Table 4 lists its key mechanical properties.The ller wire was selected to achieve a chemical composition and mechanical strength of the joint metal similar to that of the base material.Two different ller wire diameters, speci cally 2.5 mm and 4 mm, were used during welding tests, based on the pass type.
Finally, the covering ux used for the experiments was the ESAB OK 10.61, a highly basic agglomerated ux commonly used for SAW.Its nominal density is 1.1 kg/dm 3 and nominal basicity index is 2.6.

Methods
As mentioned previously, the SAW process is a widely utilized electric arc welding method in heavy industries for joining medium and thick metals.It relies on an electric current passing between the ller wire and the workpiece to generate the thermal energy needed for material melting.Due to its operational characteristics and the equipment involved, the SAW can only be conducted in fully automatic or semi-automatic mode.In this study, the semi-automatic mode was employed.
The tests were conducted using the LINCOLN ELECTRIC's Place Power Wave® AC/DC 1000 SD CE machine.The plates were butt-welded in a at position, and a total of ve passes were performed.Figure 2 illustrates the procedure used for creating the joints.The test execution method involved initially coupling plates of varying thicknesses to create the rst root weld pass (referred to as Pass 1 in Fig. 2) at room temperature.A 2.5-mm-diameter electrode was used for this purpose, providing support for the plates during the subsequent weld bead formation.
A milling operation was then conducted on the opposite side of the previously welded area to create a groove, as indicated in Fig. 2.This groove was then lled by performing four additional passes along the milled channel, using a 4 mm diameter ller wire.It is important to note that no plate preheating or postweld heat treatment was performed.The process parameters set for each pass are the arc voltage (V), the welding current (I) and the travel speed (v).The selected values for these parameters are provided in the next paragraph.These parameters play a crucial role in the welding process and determines the amount of heat input introduced in the workpiece.Higher current values result in increased arc intensity and penetration depth, leading to a larger resistant section of the joint under external loads.The travel speed, on the other hand, is inversely proportional to the thermal input which the material is exposed to.
Both too low and too high travel speeds can result in signi cant defects within the weld bead.
The in uence of current, voltage, and travel speed on the weld bead and on the microstructure has been assessed.To this aim, the transformation phases that occur usually in steels during welding process have been analysed.These phases depend on the speci c heat input, which is a key process variable encompassing all three individual investigated parameters [12], as well as on the cooling rate of the workpiece.The speci c heat input is calculated using the following equation: 1 where η represents the arc e ciency factor, and in case of the SAW it is equal approximately to 0.95 [24], V, I and ws are, as already speci ed, the arc voltage, the welding current and the welding speed, respectively.
The dimensions of the molten and the heat affected zone, as well as of their grain size, depend on the amount of heat (Q) transferred to the material during the welding.Besides, the transient thermal cycle plays a crucial role in controlling the material's cooling rate, which, in turn, is a key factor in determining phase changes and the formation of microstructural constituents.

The experimental plan
For the experimental analysis, speci c working conditions were established.As previously mentioned, a constant number of passes for each sample was considered and it was set at ve owing to the depth of the welding gap.

The experimental procedure
As mentioned, the experimental investigation aimed to assess how process parameters affect joint quality and process e ciency by reducing the time required for welding.To evaluate the process accuracy a metallurgical analysis of the nal joints was conducted, focusing on the thermal effects caused by the process parameters.Speci cally, the HAZ, the microstructure and the mechanical properties of the joints have been examined and discussed below.

Sample preparation
To conduct the required studies, samples have been extracted from the joined plates by cutting them at low temperatures.This was done to avoid the alteration of the microstructure already affected by the welding process.From each weld bead, three samples have been extracted to ensure result repeatability and account for the non-uniformity of the weld bead along the entire joined plates.
Before conducting macro and micro analysis, all specimens underwent a speci c preparation process to eliminate cutting-induced scratches.First, the samples were cold-mounted using acrylic resin to prevent thermal alterations.The drying time for this process was approximately 45 minutes.Once the samples were ready, they were secured in the holder of the Struers Tegramin-25 machine (as shown in Fig. 3a), and their surfaces were prepared for metallographic inspection.The preparation involved the use of abrasive papers with different granularities to remove residual resin and achieve a at surface.Subsequently, speci c cloths were used for mirror polishing (Fig. 3b).The samples were prepared by chemically treating their polished surfaces with a 3% nitric acid-based solution, named Nital, for 25 seconds.This treatment made the macrostructure of the weld beads clearly visible to the naked eye (Fig. 3c), allowing each pass to be distinctly recognized, as well as under the microscope (Fig. 3d).
It is important to note that the immersion time in the solution was determined through experimental observations.Several tests were conducted to establish the right immersion time required to reveal the joint microstructure.Low time (underetching) was insu cient for visualizing the grains, while prolonged immersion (overetching) resulted in a surface burning effect from a microstructure perspective.Metallographic investigations were conducted using the LEICA DM 4000M microscope.The applied method, as outlined in Fig. 4, enabled the analysis of the Heat-Affected Zone (HAZ).For this purpose, at least three pictures were captured close to each welding pass to provide a comprehensive description of the entire joint microstructure.
To determine the amount of microstructural constituents at room temperature, speci cally ferrite and pearlite, micrographs were analysed using a speci c image processing technique that allowed to assumed a pearlitic structure consisting of ferrite plates separated by thin regions of cementite.
Widmanstatten ferrite also nucleated from the grain boundaries of austenite when it no longer transformed into equiaxial ferrite.Being ferrite very low in carbon content, its high level within the investigated region can mainly be explained by two physical phenomena occurring during the steel welding.The rst one is the decarbonisation, i.e. the loss of C from the base material due to its reaction with the oxygen in the air, while the second one is related to the migration of C atoms from a higher to a lower content area.In fact, the material of the HAZ is the ASTM A516 Gr.70 steel and it has a C content of 0.31% by weight, which is higher than that in the melted zone.Indeed, the latter is a mixture of base and ller metal, with a carbon content of about 0.10%, as already speci ed in Table 3. Region 1 (Fig. 6a), on the other hand, was characterised by pro-eutectoid ferrite with a polygonal shape and perlite.
Quantitative analyses were conducted to determine the average extension of the overall HAZ and the size of the regions 1 and 2 for each welding pass.In Fig. 7a, for instance, Pass 2.1 is indicated by a red arrow, denoting the length of the entire HAZ produced at the end of the process, whereas Fig. 7b shows a 5X magni cation micrograph of the HAZ produced by Pass 2.1 for the same specimen.On this latter, the white arrows indicate the extension of Region 1 and Region 2 resulting from the welding.To determine the lengths of both the HAZ and the two regions, multiple equally spaced lengths were measured, considering potential variations along the edge of each welding pass.For each pass the lengths were obtained by means of segments orthogonal to the weld boundary line.Ten length measurements were taken for each welding pass to ensure consistency and repeatability of results.
Additionally, the percentage of ferrite within Region 1 and Region 2 was estimated following the procedure above-mentioned.Data collected from the regions under investigation were then analysed by comparing the different working conditions set in the experimental study (Table 5).First of all, the data related to Pass 2.1 were considered, since, looking for instance at Fig. 7a, Pass 2.2 was covered by Pass 3.1 and Pass 3.2 making an assessment of the corresponding outputs impossible.Just for Test 2, Test 5 and Test 7 it was possible to compare the effect caused by Pass 2.2, as it was not covered by subsequent Pass 3.
Histograms in Fig. 8a display the extension of the HAZ produced by Pass 2.1 for each considered test, while Fig. 8b reports the extension of Region 2. From the histograms observation it is easy to deduce that Region 1 area is wider than that of the Region 2. Figure 8c and Fig. 8d show, respectively, the percentage of ferrite within Region 1 and Region 2.
The data analysis related to Pass 2.1 revealed that both the average size of the whole HAZ (Fig. 8a) and the percentage of ferrite within both regions remained relatively consistent for all the investigated process conditions, and therefore in all the tests.In fact, the corresponding values were uniformly distributed as the process conditions changed.Speci cally, the average length of the HAZ ranged from approximately 1.84 mm in Test 8 to 2.93 in Test 7, with an average value of 2.33 mm.Whereas, the ferrite percentage in Region 1 ranged from 64% in Test 5 to 73% in Test 3 with an average value of 69.22%; while in Region 2 from a minimum of 64% in Test 5 to a maximum of 77% in Tests 0 and 3, with an average value of 73.66%.
Considering the number of tests investigated and the outcomes data distribution under different working conditions, no correlation was identi ed with the process parameters employed for performing the joint.
As previously mentioned, Test 2, Test 5 and Test 7 were compared for evaluating the effect caused by Pass 2.2.The comparison, both from a qualitative and a quantitative point of view, is reported in Fig. 9, where the macrographs and the measurement of the average HAZ extension and the ferrite content just in Region 2, for sake of simplicity, are depicted.The analysis revealed higher values for both the HAZ extension and ferrite percentage related to Pass 2.2 for all the tests, likely due to the different thicknesses of the welded plates.Pass 2.2, in fact, was consistently performed on the 10 mm thick side, resulting in a lower heat dissipation rate.
In addition to that, it is worth pointing out that the difference in the cooling rate of the welded plates, resulting from variations in plate thickness, also had an impact on the microstructure.Speci cally, in Region 2 thicker Widmanstatten ferrite was observed on the thinner plate side (10 mm thickness).For instance, in Test 7, a comparison of micrographs at 50X magni cation revealed the characteristic microstructure of Region 2 produced by Pass 2.2 (Fig. 10a) and Pass 2.1 (Fig. 10b).
The metallographic investigation highlighted that the percentage of ferrite in the two regions remained fairly consistent when modifying the main welding parameters while maintaining a constant speci c heat input.
To further verify the in uence of the process variables on the mechanical properties of the nal joints numerous microhardness measurements were performed across the sample, following the procedure illustrated in paragraph 3.2.To this aim Test 0, which represents the benchmark since satis es the required standard for the speci c application eld, was compared to Test 8, which was performed by setting the highest value of the welding speed, maintaining always constant the speci c heat input.The hardness pro le aimed at showing how the hardness varied across the identi ed zones, which were the welding zone, Region 1, Region 2 and base material, for the two above-mentioned tests.As clearly seen from the histogram displayed below (Fig. 11), in both tests the hardness values are comparable, demonstrating that, the increase in the welding speed, but guaranteeing a speci c heat input, does not affect the mechanical response of the joint.
Following this observation, the e ciency increase, resulting from a higher welding speed, was quanti ed.
More in detail, in Test 0 Passes 2 had a welding speed of 320 mm/min and Passes 3 of 544 mm/min, while in Test 8 Passes 2 had a welding speed of 576 mm/min andPasses 3 had a speed of 768 mm/min.The time necessary to perform Pass 1 was not included in the calculation, as the same parameters were used for all tests, as well as the time between Pass 1 and the next ones, which was also ignored since it was the same in all tests.Therefore, only the active time of the welding process was calculated by considering the four passes and the total length of the plates (400 mm).

Conclusion
This study focused on the Submerged Arc Welding technology employed on carbon steel plates, aiming at demonstrating the process optimization based on substantial experimental evidences.Plates of different thicknesses were joined using multi-pass welding according to an experimental plan.The joints were meticulously analysed from macroscopic and microscopic perspectives to understand the effects of the thermal cycle on the heat affected zone.The key welding process parameters, and the resulting speci c heat input, signi cantly impact on the microstructure of the weld bead rst and, subsequently, on the heat affected areas.The percentage of ferrite and perlite of the welded steel was measured, and the analysis of the results has shown how the process e ciency and, therefore, the productivity can be improved while ensuring the required welding quality.Test 0, in fact, represented the test that satis es the UNI EN ISO 3834 quality standards and, therefore, it was considered as the reference point for both the microstructural analysis and the hardness measurements.All the other tests were carried out gradually increasing the welding speed and the microstructure was analysed comparing the content of ferrite to the one obtained in Test 0. Furthermore, for verifying the in uence of the process parameters on the mechanical properties of the joint, microhardness was measured in different zones.It was demonstrated that, for the same heat input, the welding speed increase does not affect the heat affected zone, as well as the mechanical strength of the joint.Therefore, the productivity of the welding process under investigation may be increased, but the process parameters have to be set maintaining constant the speci c heat input.
These ndings can serve as a guide for future research and practical applications, ultimately leading to enhanced welding practices and the development of more robust and reliable welded joints.

Declarations Figures
Operating

Figure 3a )
Figure 3a) Mounted samples, b) Polished samples, c) Macrograph of the weld bead, d) micrograph of the HAZ

Table 1
Main mechanical properties of ASME A516 Grade 70

Table 2
Chemical composition of the weld material (ASTM A516 Gr.70)

Table 3
Chemical composition of AWS 5.17EH 14 Table 5outlines the designed experimental plan and the values assigned to voltage, current, and welding speed.It is essential to note that Test 0 represents the working condition necessary to achieve a sound joint in accordance with UNI EN ISO 3834 quality standards, therefore it was considered as a benchmark.Consequently, the experimental plan was constructed by maintaining a constant speci c heat input for each pass, based on Test 0. The rst pass was consistent across all tests, as it was performed on the bottom side of the plates to create a foundational layer that supports subsequent welding steps.The second passes (labelled Pass 2.1 and Pass 2.2) had varying parameters in each test, but the nominal speci c heat input for all of them was approximately 2145 J/mm.The third passes (labelled Pass 3.1 and Pass 3.2) were conducted by employing the same process parameters in all tests, aligning with the speci c heat input of Test 0, which was approximately 1790 J/mm.
Table 6 reports the active welding times of the compared tests.Test 0 lasted approximately 4 minutes, while Test 8 approximately 2 minutes and half.The time saving provided by the new process conditions, which guarantee the same mechanical properties of the joint, was around 61%.This represents a signi cant result for such kind of industrial applications where the joints to perform are much larger than the ones herein examined, therefore the competitive industrial advantage provided in these terms becomes signi cant.