Effect of Preheating Temperature and Joint Geometry on HAZ Development and Hardness of GMAW-Welded SAE 1045 Steel, and on the Mechanical Performance of Butt Joints

Abstract

Medium-carbon steels are susceptible to heat-affected zone (HAZ) hardening during welding. This study evaluated the effects of preheating temperature and joint geometry on mechanized gas metal arc welded (GMAW) joints in 6.3 mm thick SAE 1045 steel. Butt and fillet joints were welded under conventional single-pass conditions from room temperature to 250 °C, while an additional two-pass procedure was evaluated separately. Metallography and hardness were evaluated for both joint geometries, whereas tensile and Charpy V-notch tests were performed only on butt joints. Significant joint geometry × preheating condition interactions were identified for weld metal (WM) area, HAZ area, WM hardness and HAZ hardness. Average HAZ hardness remained below 300 HV₁₀ for all conventional single-pass conditions. Among the butt joint single-pass conditions, no statistically significant effect of preheating was detected for ultimate tensile strength (UTS) or Charpy absorbed energy, and all tensile specimens fractured in the base metal (BM). The two-pass procedure was treated descriptively because the second deposition simultaneously altered multiple welding variables. Overall, the effects of preheating on WM and HAZ areas and hardness were joint-geometry-dependent under the investigated conditions.

1. Introduction

Medium-carbon steels such as SAE 1045 are extensively employed in the manufacturing of shafts, gears, agricultural machinery components, and structural parts owing to their favourable combination of mechanical strength, wear resistance, machinability, and cost-effectiveness [1]. These characteristics make SAE 1045 steel suitable for applications subjected to moderate-to-high mechanical loads and cyclic service conditions [2]. However, despite their widespread industrial use, medium-carbon steels present lower weldability than conventional low-carbon steels because their higher carbon content and carbon equivalent increase the tendency for HAZ hardening [3]. Although carbon equivalent provides a useful empirical indication of weldability, it does not replace procedure qualification because weldability also depends on factors such as restraint level, heat input, hydrogen content, and cooling conditions.

The thermal cycle imposed during welding strongly influences phase transformations, residual stress development and, consequently, the mechanical performance of welded joints [4,5]. Rapid cooling rates may promote the formation of martensite and other hard transformation products within the HAZ, increasing hardness while reducing ductility and fracture toughness [6,7]. Furthermore, the presence of hard transformation products may increase susceptibility to hydrogen-assisted cracking and compromise the structural reliability of welded components [8,9].

To mitigate these adverse effects, preheating is commonly employed in industrial welding procedures [3,10]. Preheating reduces cooling rates, thereby limiting the formation of hard transformation products within the HAZ. It also facilitates hydrogen diffusion, reducing susceptibility to hydrogen-assisted cold cracking, and decreases thermal gradients, contributing to lower residual stresses [6,7]. The present study focuses on metallurgical observations, hardness, tensile behaviour, and impact toughness associated with the investigated welding conditions. Hydrogen diffusion and hydrogen-assisted cracking were beyond the scope of this investigation and were therefore not directly evaluated.

The effectiveness of preheating, however, depends on the thermal conditions established during welding. Among the variables controlling the welding thermal cycle, plate thickness plays a major role because it governs heat dissipation and structural restraint. Compared with thicker components, 6.3 mm thick plates exhibit different thermal boundary conditions and heat-flow characteristics, which may reduce the influence of conventional preheating on cooling behaviour. Consequently, conclusions obtained under the present conditions should not be directly extrapolated to thicker SAE 1045 components, such as shafts, dies, or other heavily restrained structures, where heat dissipation and restraint conditions differ substantially.

Besides plate thickness, joint geometry also modifies the welding thermal cycle by changing weld volume, heat-flow paths, and thermal accumulation. As a result, different joint configurations may respond differently to the same preheating temperature, affecting HAZ dimensions, microstructural evolution, and mechanical performance [11,12,13,14]. Because preheating increases manufacturing time, energy consumption, and operational costs [15,16], understanding its effects for a given combination of material, plate thickness, and joint geometry remains important from both technical and economic perspectives [17,18]. Nevertheless, industrial preheating recommendations for medium-carbon steels frequently rely on empirical or semi-empirical procedures, and experimental evidence remains dependent on the specific material and welding conditions [1,19].

Although previous studies have investigated the influence of preheating on medium-carbon steels [1,3], these investigations have focused on a single joint geometry, a restricted range of preheating temperatures, or isolated metallurgical and mechanical responses. As a result, they provide limited information regarding how joint geometry modifies the effectiveness of preheating through changes in the thermal boundary conditions established during welding. Furthermore, few studies have evaluated these variables simultaneously using a factorial experimental design capable of identifying interaction effects between joint geometry and preheating condition. Consequently, further investigation is required to determine how joint geometry modifies the response to preheating under controlled welding conditions.

Accordingly, this study evaluates the combined influence of preheating temperature and joint geometry on HAZ development and hardness of mechanized GMAW welds in 6.3 mm thick SAE 1045 steel, supported by qualitative metallographic observations, while the effect of preheating condition on tensile behaviour and impact toughness is evaluated specifically for butt welded joints. Unlike previous studies that evaluated these variables independently, the present work applies a factorial statistical design to quantify not only the individual effects of preheating temperature and joint geometry, but also their interaction, providing a more comprehensive assessment of geometry-dependent responses to preheating in mechanized GMAW of thin SAE 1045 steel.

2. Materials and Methods

2.1. Base and Filler Materials

The BM used in this study was SAE 1045 medium-carbon steel supplied as 6.3 mm thick hot-rolled plates. All specimens were produced from the same batch of material to minimize variations in chemical composition and microstructural condition. The chemical composition of the BM was determined by optical emission spectrometry (OES) and is presented in

Table 1. Chemical composition of the SAE 1045 base metal (wt.%).

C	Mn	Si	P	S	Ni	Cr	Mo	V	Cu
0.45	0.59	0.23	0.01	0.005	0.008	0.01	0.001	—	—

. Prior to welding, the BM exhibited the typical ferrite–pearlite microstructure expected for the as-received condition [3].

The filler metal employed in the welding procedure was an AWS ER80S-Ni1 solid filler wire with a diameter of 1.2 mm. To verify the chemical composition of the deposited weld metal, a multilayer weld pad was produced under controlled welding conditions, and samples were extracted for chemical analysis. The chemical composition of the all-WM deposit is presented in

Table 2. Chemical composition of the all-WM deposit produced with ER80S-Ni1 filler wire (wt.%).

C	Mn	Si	P	S	Ni	Cr	Mo	V	Cu
0.06	1.09	0.52	0.01	0.005	0.98	0.024	0.002	0.002	0.11

.

To provide an empirical assessment of the weldability of the investigated steel, the carbon equivalent (CE) was calculated according to the International Institute of Welding (IIW) formulation (Equation (1)), where C, Mn, Cr, Mo, V, Ni, and Cu represent the respective chemical compositions (wt.%). Based on the chemical composition presented in Table 1, the IIW carbon equivalent (CE_IIW) was calculated as 0.55. This value indicates an increased tendency for HAZ hardening compared with conventional low-carbon structural steels. The calculated carbon equivalent was used only as an empirical indicator of weldability to support the experimental design and was not intended as a substitute for welding procedure qualification.

$${CE}_{IIW}=C+{\displaystyle \frac{Mn}{6}}+{\displaystyle \frac{Cr+Mo+V}{5}}+{\displaystyle \frac{Ni+Cu}{15}},$$

(1)

2.2. Joint Preparation and Welding Procedures

Two joint geometries were investigated in this study: butt joints and fillet joints. The base material consisted of SAE 1045 medium-carbon steel plates with nominal dimensions of 550 mm × 170 mm × 6.30 mm. Prior to welding, the plate surfaces were mechanically cleaned to remove oxides and contaminants. The specimens were clamped to a rigid steel fixture to maintain joint alignment and consistent restraint conditions throughout the experimental programme.

For the butt joints, a single-V groove with a 60° groove angle, 1.0 mm root face, and 0.5 mm root opening was machined and verified before welding to ensure dimensional consistency. The 6.3 mm thick plates were aligned, tack-welded, and welded in the 1G (flat) position. A copper backing bar was positioned beneath the joint to support the molten weld pool and minimize excessive root penetration (Figure 1).

Fillet welds were produced using a T-joint configuration consisting of two 6.3 mm thick SAE 1045 steel plates positioned at 90°. A single-pass fillet weld with a nominal leg size of 8 mm and an overall weld length of approximately 180 mm was deposited without intentional root gap. The welding torch was mounted on a mechanized carriage to produce a rectilinear weld bead at constant travel speed. Welding was carried out in the 2F (horizontal fillet) position using an electrode extension of approximately 18 mm. Samples for metallographic characterization and hardness measurements were extracted from the central portion of the weld to minimize start- and stop-related effects.

Welding was performed using a Lincoln Electric Power Wave S350 constant-voltage GMAW power source coupled with a Power Feed 84 wire feeder (Lincoln Electric, Cleveland, OH, USA). Torch displacement was controlled by a mechanized carriage system (SPS Soldagem, model V1, Florianópolis, SC, Brazil), enabling precise control of travel speed and weld length. Direct-current electrode positive (DCEP) polarity was employed throughout the study.

The filler metal was AWS ER80S-Ni1 solid wire (1.2 mm). Shielding was provided by ISO 14175 M21 gas (82% Ar–18% CO₂) at a flow rate of 16 L/min, and the electrode stick-out was maintained at approximately 18 mm. The nominal welding current, arc voltage, and travel speed were maintained throughout the experimental programme, while the recorded electrical parameters exhibited the normal process variability summarized in

Table 3. Experimental matrix and nominal GMAW parameters for the investigated joint geometries and welding conditions.

Configuration	Welding Condition	Welding Passes
Butt joints	Room	1
	50 °C	1
	100 °C	1
	150 °C	1
	200 °C	1
	250 °C	1
	Two-pass	2
Fillet joints	Room	1
	50 °C	1
	100 °C	1
	150 °C	1
	200 °C	1
	250 °C	1
	Two-pass	2

Note: The overall recorded arc voltage, welding current, welding speed, and heat input across the experimental programme were 28.2 ± 0.3 V, 307.5 ± 10.6 A, 5.4 mm/s, and 1.6 kJ/mm, respectively. These values represent global averages rather than condition-specific averages. Room temperature denotes the condition without external preheating; the measured initial surface temperature was approximately 10 °C. Single-pass conditions correspond to the room temperature and preheating at 50, 100, 150, 200, and 250 °C. Two-pass denotes deposition of a second adjacent bead using the same nominal welding parameters.

. The nominal heat input (HI) was calculated according to Equation (2) using the measured welding current, arc voltage, and travel speed. Because no arc-efficiency factor was applied, the reported values represent nominal heat input and were used exclusively for comparative purposes among the investigated welding conditions. The welding current and arc voltage reported in Table 3 correspond to overall recorded averages obtained across the experimental programme and therefore represent global welding-parameter values rather than condition-specific averages.

$$HI={\displaystyle \frac{U.I}{v}},$$

(2)

In addition to the conventional single-pass welds, a two-pass welding condition was investigated. The second weld bead was deposited immediately adjacent to the first using the same nominal welding parameters. Besides introducing a secondary thermal cycle, the second pass increased cumulative heat input and filler metal volume and modified weld geometry and potentially dilution. Therefore, the two-pass condition was evaluated as a complete welding condition rather than as an isolated reheating treatment. Because interpass temperature and the time interval between passes were not recorded, the contribution of the secondary thermal cycle could not be separated from the effects of increased deposited metal volume, cumulative heat input, weld geometry, and potential changes in dilution. Accordingly, the two-pass condition was treated as an additional welding procedure and was not included as a level of the preheating condition factor in the inferential statistical analyses.

The investigated conditions were selected to provide a systematic range from the room temperature to 250 °C, allowing the influence of increasing initial surface temperature on the investigated welding responses to be evaluated.

Preheating temperature was defined as the main experimental variable. Heating was performed using a multi-flame propane torch until the target temperature was reached. The surface temperature was measured immediately before arc ignition using a calibrated Minipa MT-350 infrared thermometer (Minipa Indústria e Comércio Ltda., São Paulo, Brazil) with a fixed emissivity setting of ε = 0.95 and a resolution of 0.5 °C. The same surface preparation, heating procedure, measurement location, instrument, and emissivity setting were maintained for all specimens to provide a consistent basis for comparison among the investigated conditions. The actual surface emissivity of SAE 1045 steel was not independently determined; therefore, ε = 0.95 was treated as a fixed operational setting adopted for comparative temperature control. The expanded instrument-calibration uncertainty was ±0.2 °C at calibration points close to 50–200 °C and ±0.3 °C at approximately 400 °C (k = 2; approximately 95% confidence). These calibration values describe the instrument response and do not include additional uncertainty associated with the actual surface emissivity. For the room temperature condition, no intentional preheating was applied, and the measured initial plate surface temperature was approximately 10 °C, corresponding to the ambient laboratory temperature during the experimental programme.

Run-on and run-off tabs were attached at both ends of the plates to stabilize the arc and minimize start- and stop-related defects. Welding was performed under controlled environmental conditions. Butt joints were welded in the 1G (flat) position, whereas fillet joints were welded in the 2F (horizontal fillet) position.

Temperature uniformity through the plate thickness and along the welding path was not verified, and thermal cycles and cooling times (Δt_8/5) were not directly measured. Consequently, the reported preheating temperatures correspond to local surface measurements obtained immediately before welding.

Although identical nominal welding parameters were employed for both joint configurations, thermal equivalence between the butt and T-joints was not assumed. Differences in joint geometry, local material volume, three-dimensional heat-flow paths, deposited-metal-to-effective-thickness ratio, and contact with the clamping fixture result in distinct thermal boundary conditions. Consequently, comparisons between joint configurations were based on experimentally measured WM area, HAZ area, and hardness rather than on assumed thermal equivalence.

2.3. Metallographic and Hardness Characterization

After welding, the specimens were sectioned perpendicular to the weld bead using a precision diamond saw cutting machine to expose the transverse cross-sections of the welded joints. Sections were extracted from the central region of each weld to minimize the influence of arc initiation and termination and to provide representative specimens for metallographic evaluation.

The samples were mounted in thermosetting resin and prepared following conventional metallographic procedures. Surface preparation consisted of sequential wet grinding with silicon carbide abrasive papers from #100 to #1200, followed by mechanical polishing with 3 μm diamond paste to obtain a mirror finish. After each grinding and polishing step, the specimens were cleaned with water and ethanol to minimize contamination. The polished cross-sections were etched with 2% Nital to reveal the weld macrostructure and microstructure.

Macrostructural examination was performed using a ZEISS STEMI 2000-C stereomicroscope (Carl Zeiss, Oberkochen, Germany) to evaluate weld bead geometry, penetration profile, HAZ extension, and macroscopic discontinuities. Microstructural observations were conducted using a ZEISS AXIO SCOPE A1 (Carl Zeiss, Oberkochen, Germany) optical metallurgical microscope at different magnifications to identify the characteristic regions and transformation features in the BM, HAZ, and WM.

One representative transverse cross-section from each independently welded specimen was analyzed using z, version 1.4.3.67 (National Institutes of Health, Bethesda, MD, USA) to quantify the WM and HAZ areas. The image scale was calibrated using the dimensional reference included in each macrograph. The WM boundary was defined by the visible fusion line separating the deposited metal from the base material. For HAZ area measurement, the fusion boundary was adopted as the inner limit, whereas the outer boundary was manually traced at the transition in etching contrast between the visibly transformed HAZ and the unaffected BM. All visibly distinguishable HAZ regions contiguous with the fusion boundary were included. The same magnification, image calibration, boundary-selection criterion, and manual tracing procedure were applied to all specimens.

The uncertainty associated specifically with manual boundary segmentation and intra- or inter-operator repeatability was not quantified during the experimental programme. Therefore, the standard deviations reported for WM and HAZ areas represent variability among the three independently welded specimens rather than image-segmentation uncertainty.

Because butt and T-joints differ in cross-sectional geometry and fusion-boundary length, their absolute HAZ areas were not considered geometrically equivalent. Accordingly, the influence of preheating condition was interpreted primarily within each joint geometry, whereas geometry-dependent differences were evaluated through the joint geometry × preheating condition interaction. Normalization by fusion-boundary length could not be performed because fusion-boundary length was not quantified during the original image analysis. Normalization by WM area was not adopted because WM area was itself an experimental response affected by joint geometry and preheating condition, particularly under the two-pass condition.

Vickers hardness measurements were performed on polished cross-sections using a WILSON INSTRUMENT M0 hardness tester (Wilson Instruments, Norwood, MA, USA) with a load of 10 kgf and a dwell time of 15 s. Measurements were conducted in accordance with DIN EN ISO 9015-1 [20] following the indentation scheme shown in Figure 2. Points 1–4 were located in the BM, points 5–8 in the HAZ, and points 9 and 10 in the WM. The same indentation pattern was used for all specimens.

The reported hardness values correspond to the average HV₁₀ values obtained at the predefined locations within each region. This procedure provided a consistent basis for comparison among the investigated welding conditions but was not intended to generate a detailed hardness map or determine the maximum local hardness immediately adjacent to the fusion boundary. Accordingly, the reported HAZ hardness values should be interpreted as regional average responses rather than maximum local hardness values.

2.4. Tensile and Charpy Impact Testing

Mechanical testing was performed exclusively on butt welded joints. Standardized transverse tensile and Charpy V-notch specimens were extracted only from this joint configuration in accordance with ASTM E8/E8M [21] and ASTM E23 [22], respectively. Fillet joints were evaluated only by metallographic characterization and hardness measurements. Consequently, statistical analyses of tensile strength and Charpy impact energy considered only the preheating condition as the independent factor.

Transverse tensile specimens were machined according to ASTM E8/E8M [21], with the weld centerline located at the midpoint of the reduced section (Figure 3). Tensile tests were conducted at room temperature under displacement control using a constant crosshead speed of 0.75 mm/min. Three specimens were tested for each welding condition, and the reported values correspond to the average results obtained. Fracture location was visually confirmed after testing.

In accordance with the scope of the present study, transverse tensile testing was restricted to ultimate tensile strength and fracture location. Yield strength and elongation were not evaluated and are therefore not discussed.

Charpy V-notch impact tests were conducted at room temperature (20 °C), in accordance with ASTM E23 [22]. The notch was positioned at the WM centerline, Figure 4a. For the two-pass condition, specimen extraction and notch positioning followed the same procedure, with the notch referenced to the evaluated WM region of the completed two-pass joint. Figure 4b presents a representative cross-sectional schematic of the two-pass configuration, showing the first and second deposited beads and the approximate Charpy notch location. Because this representation is schematic rather than a metallographic cross section of the specific tested specimen, the indicated notch location should be regarded as approximate. Accordingly, the measured absorbed energy represents the impact response of the completed two-pass welded condition and cannot be assigned to an individual deposited bead or to a specific reheated subregion. For each condition, three specimens were tested, and the absorbed energy was recorded. The reported absorbed energies correspond to the measured subsize specimens. After impact testing, the fractured specimens were macroscopically documented for qualitative comparison among the investigated conditions.

Because the BM thickness was 6.3 mm, ASTM E23 subsize Charpy V-notch specimens (5 mm thickness) were employed. Therefore, the absorbed energies reported in this study correspond to the measured subsize specimens and are presented as raw experimental values. No conversion to equivalent full-size Charpy energies was performed. The specimens were extracted transverse to the welding direction, with the notch centred in the weld metal and perpendicular to the fusion boundary.

2.5. Statistical Analysis

The main factorial experimental design comprised two factors: joint geometry (butt and fillet joints) and preheating condition (room temperature, 50, 100, 150, 200, and 250 °C), considering only the conventional single-pass welding conditions. Two-way analysis of variance (ANOVA) was used to evaluate the main effects of joint geometry and preheating condition and their interaction on WM area, HAZ area, WM hardness, and HAZ hardness. The two-pass condition was not included as a level of the preheating condition factor because the second deposition simultaneously changed the number of passes, deposited filler-metal volume, cumulative heat input, weld geometry, and potentially dilution. The two-pass results were therefore treated separately as those of an additional welding procedure and are reported descriptively.

For the factorial analyses, the experimental unit was an independently welded specimen, with n = 3 independent welding trials for each joint geometry × preheating condition combination. Each welded specimen was sectioned once in its central region, and the reported measurements were obtained from this representative transverse cross-section.

The ANOVA results were reported as degrees of freedom, F values, and p-values. Partial eta squared (${\mathsf{\eta }}_{\mathrm{p}}^2$) was calculated as an effect-size measure. Residual normality and variance homogeneity were assessed using Shapiro–Wilk and Levene tests, respectively. For the hardness responses, for which departures from ANOVA assumptions were identified, heteroscedasticity-consistent HC3 covariance estimates were additionally used as a sensitivity analysis.

When a significant joint geometry × preheating condition interaction was detected (α = 0.05), Tukey’s honestly significant difference (HSD) test was used for post hoc pairwise comparisons among the six conventional single-pass preheating conditions separately within each joint geometry, using the pooled residual error term from the corresponding two-way ANOVA. The post hoc analyses were performed using STATISTICA, version 12.5 (StatSoft, Inc., Tulsa, OK, USA), and Tukey-adjusted p-values were used to determine statistical significance. Results are presented using a compact letter display, in which means sharing at least one letter are not significantly different at α = 0.05.

Ultimate tensile strength and Charpy impact energy, evaluated only for butt welded joints, were analyzed by one-way ANOVA considering preheating condition as the independent factor and including only the six conventional single-pass conditions. Tukey’s HSD pairwise comparisons were additionally performed among these single-pass conditions and are reported using compact-letter groupings for consistency with the other response variables. The two-pass results were excluded from these inferential analyses and are reported descriptively as results of an additional welding procedure.

3. Results

3.1. Effect of Preheating Temperature and Joint Geometry on HAZ Development and Hardness

Figure 5 presents representative optical micrographs of a full-penetration butt joint welded at 50 °C, illustrating the WM, HAZ, BM, and characteristic HAZ subregions distinguishable after conventional Nital etching. The regions identified as coarse-grained (CGHAZ), fine-grained (FGHAZ), and intercritical (ICHAZ) are consistent with conventional metallographic classification of welded steel HAZs reported in previous studies [23,24].

The representative optical micrographs presented in Figure 5 were intended to qualitatively identify the characteristic HAZ subregions observed after conventional Nital etching. Within the resolution of optical microscopy, CGHAZ, FGHAZ, and ICHAZ [12,13] could be distinguished; however, the present observations were not intended to provide detailed phase identification or quantitative microstructural characterization. Consequently, comparisons among the investigated preheating conditions were based primarily on the measured WM area, HAZ area, hardness, and mechanical properties rather than qualitative differences observed in the optical micrographs.

Figure 6 presents the average WM and HAZ areas measured for the butt and fillet joints under the investigated preheating conditions. Because the two joint configurations have different cross-sectional geometries and fusion-boundary lengths, their absolute area values should not be interpreted as geometrically equivalent. Accordingly, the effects of preheating condition were evaluated primarily within each joint geometry, while the factorial analysis was used to determine whether the response to preheating differed between geometries.

Within each geometry, the responses were not uniformly monotonic over the complete preheating range. For the butt joints, HAZ area generally increased with preheating temperature, whereas WM area fluctuated among the single-pass conditions. For the fillet joints, both WM and HAZ areas exhibited condition-dependent variations, with no continuous increase over the investigated temperature range. Therefore, differences among conditions were interpreted using the two-way ANOVA and Tukey HSD results presented in

Table 4. Descriptive statistics, Tukey HSD groups, and two-way ANOVA results for WM area.

Joint Geometry	Welding Condition	Mean ± SD (mm2)	n	Tukey Group
Butt	Room	83.87 ± 1.50	3	ab
Butt	50 °C	71.18 ± 0.80	3	d
Butt	100 °C	82.02 ± 0.90	3	b
Butt	150 °C	84.14 ± 0.29	3	a
Butt	200 °C	84.59 ± 0.29	3	a
Butt	250 °C	78.31 ± 1.50	3	c
Butt	Two-pass	89.14 ± 2.40	3	—
Fillet	Room	70.24 ± 0.35	3	c
Fillet	50 °C	75.63 ± 0.76	3	b
Fillet	100 °C	75.50 ± 0.40	3	b
Fillet	150 °C	75.12 ± 0.55	3	b
Fillet	200 °C	74.88 ± 0.50	3	b
Fillet	250 °C	82.90 ± 0.80	3	a
Fillet	Two-pass	96.63 ± 1.25	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Joint geometry	1	326.60	<0.0001	0.932
Preheating condition	5	61.27	<0.0001	0.927
Joint geometry × preheating condition	5	130.49	<0.0001	0.965
Error	24	—	—	—

Note: Values are reported as mean ± SD (n = 3). The factorial ANOVA and Tukey HSD comparisons include only the conventional single-pass conditions (room temperature and 50–250 °C preheating). The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential statistical analysis. Within each joint geometry, single-pass means sharing at least one Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

and

Table 5. Descriptive statistics, Tukey HSD groups, and two-way ANOVA results for HAZ area.

Joint Geometry	Welding Condition	Mean ± SD (mm2)	n	Tukey Group
Butt	Room	40.39 ± 0.45	3	e
Butt	50 °C	46.72 ± 0.47	3	d
Butt	100 °C	54.76 ± 0.96	3	c
Butt	150 °C	65.81 ± 0.76	3	b
Butt	200 °C	66.65 ± 0.65	3	ab
Butt	250 °C	68.08 ± 1.01	3	a
Butt	Two-pass	73.89 ± 0.42	3	—
Fillet	Room	57.45 ± 1.00	3	e
Fillet	50 °C	106.80 ± 1.21	3	b
Fillet	100 °C	96.86 ± 1.25	3	cd
Fillet	150 °C	96.02 ± 1.00	3	d
Fillet	200 °C	128.34 ± 1.76	3	a
Fillet	250 °C	100.11 ± 1.26	3	c
Fillet	Two-pass	105.13 ± 1.00	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Joint geometry	1	13,583.33	<0.0001	0.998
Preheating condition	5	1406.06	<0.0001	0.997
Joint geometry × preheating condition	5	430.71	<0.0001	0.989
Error	24	—	—	—

Note: Values are reported as mean ± SD (n = 3). The factorial ANOVA and Tukey HSD comparisons include only the conventional single-pass conditions (room temperature and 50–250 °C preheating). The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential statistical analysis. Within each joint geometry, single-pass means sharing at least one Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

. The two-pass condition produced the largest WM area because of the additional deposited filler metal. Because this condition also changed the cumulative heat input, weld geometry, and potentially the dilution conditions, its WM area cannot be directly compared with the single-pass preheating trend as evidence of a secondary thermal cycle effect alone.

Considering only the conventional single-pass conditions, two-way ANOVA identified significant effects of joint geometry, preheating condition, and their interaction on WM area (Table 4). The joint geometry × preheating condition interaction was significant, F(5,24) = 130.49, p < 0.0001, with a large effect size (partial ${\mathsf{\eta }}_{\mathrm{p}}^2$= 0.965), indicating that the response to preheating condition differed between joint geometries. For the butt joints, the corrected Tukey HSD comparisons showed that the room temperature condition did not differ significantly from the 100, 150, and 200 °C conditions, whereas it differed from the 50 °C and 250 °C conditions. Although both the 100 and 150–200 °C conditions were statistically indistinguishable from room temperature, the 100 °C condition differed significantly from the 150 and 200 °C conditions. The 150 and 200 °C conditions exhibited the highest WM areas and belonged to the same Tukey group, whereas the 50 °C condition exhibited the lowest WM area. For the fillet joints, the 50, 100, 150, and 200 °C conditions were statistically indistinguishable, whereas the room temperature and 250 °C conditions formed separate groups. The two-pass procedure was not included in this factorial analysis and is reported separately as an additional welding procedure.

Considering only the conventional single-pass conditions, HAZ area was significantly affected by joint geometry, preheating condition, and their interaction (Table 5). The joint geometry × preheating condition interaction was significant, F(5,24) = 430.71, p < 0.0001, with a large effect size (partial ${\mathsf{\eta }}_{\mathrm{p}}^2$= 0.989). For the butt joints, HAZ area tended to increase with preheating temperature. The 150 and 200 °C conditions were statistically indistinguishable, as were the 200 and 250 °C conditions, whereas the 150 and 250 °C conditions differed significantly. For the fillet joints, the response remained non-monotonic, with the largest HAZ area observed at 200 °C. The two-pass procedure was evaluated separately and was not included in the factorial ANOVA or Tukey comparisons.

Figure 7 presents the average hardness values measured in the BM, HAZ, and WM. BM hardness showed limited numerical variation among the investigated conditions. For the butt joints, average HAZ and WM hardness generally decreased with increasing preheating temperature, although the response was not strictly monotonic at every temperature. In contrast, the fillet joints exhibited condition-dependent variations in HAZ and WM hardness, with no monotonic relationship with preheating temperature. Consequently, the results were interpreted using the statistical comparisons presented in

Table 6. Descriptive statistics, Tukey HSD groups, and two-way ANOVA results for WM hardness.

Joint Geometry	Welding Condition	Mean ± SD (HV10)	n	Tukey Group
Butt	Room	262.15 ± 5.95	3	a
Butt	50 °C	242.65 ± 6.05	3	b
Butt	100 °C	237.30 ± 0.70	3	b
Butt	150 °C	219.75 ± 1.55	3	c
Butt	200 °C	217.60 ± 0.20	3	c
Butt	250 °C	221.50 ± 2.00	3	c
Butt	Two-pass	207.30 ± 1.40	3	—
Fillet	Room	230.60 ± 0.70	3	a
Fillet	50 °C	226.00 ± 1.60	3	ab
Fillet	100 °C	225.35 ± 3.15	3	ab
Fillet	150 °C	216.50 ± 1.70	3	bc
Fillet	200 °C	212.80 ± 2.50	3	c
Fillet	250 °C	219.35 ± 8.65	3	bc
Fillet	Two-pass	201.80 ± 2.90	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Joint geometry	1	84.42	<0.0001	0.779
Preheating condition	5	57.86	<0.0001	0.923
Joint geometry × preheating condition	5	12.87	<0.0001	0.728
Error	24	—	—	—

Note: Values are reported as mean ± SD (n = 3). The factorial ANOVA and Tukey HSD comparisons include only the conventional single-pass conditions (room temperature and 50–250 °C preheating). The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential statistical analysis. Within each joint geometry, single-pass means sharing at least one Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

and

Table 7. Descriptive statistics, Tukey HSD groups, and two-way ANOVA results for HAZ hardness.

Joint Geometry	Welding Condition	Mean ± SD (HV10)	n	Tukey Group
Butt	Room	294.00 ± 1.08	3	a
Butt	50 °C	295.73 ± 1.96	3	a
Butt	100 °C	275.21 ± 1.44	3	b
Butt	150 °C	258.19 ± 0.24	3	c
Butt	200 °C	251.80 ± 1.04	3	d
Butt	250 °C	257.69 ± 0.08	3	c
Butt	Two-pass	254.65 ± 1.50	3	—
Fillet	Room	251.43 ± 16.90	3	a
Fillet	50 °C	250.80 ± 1.38	3	a
Fillet	100 °C	251.83 ± 10.08	3	a
Fillet	150 °C	254.69 ± 1.61	3	a
Fillet	200 °C	237.46 ± 3.66	3	a
Fillet	250 °C	241.11 ± 2.96	3	a
Fillet	Two-pass	239.29 ± 7.91	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Joint geometry	1	150.10	<0.0001	0.862
Preheating condition	5	24.27	<0.0001	0.835
Joint geometry × preheating condition	5	11.53	<0.0001	0.706
Error	24	—	—	—

Note: Values are reported as mean ± SD (n = 3). The factorial ANOVA and Tukey HSD comparisons include only the conventional single-pass conditions (room temperature and 50–250 °C preheating). The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential statistical analysis. Within each joint geometry, single-pass means sharing at least one Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

rather than from the apparent trends in the mean values alone.

For the butt joints (Figure 7a), average HAZ hardness ranged from approximately 294 HV₁₀ at room temperature to approximately 252 HV₁₀ at 200 °C among the conventional single-pass conditions, whereas average WM hardness ranged from approximately 262 HV₁₀ at room temperature to approximately 218 HV₁₀ at 200 °C. The additional two-pass procedure exhibited a lower numerical WM hardness of approximately 207 HV₁₀ and is reported separately. For the fillet joints (Figure 7b), numerical variations were observed in both HAZ and WM hardness; however, the statistical significance of these differences depended on the response considered and was therefore evaluated using the Tukey-adjusted comparisons presented in Table 6 and Table 7.

All average HAZ hardness values measured at the standardized locations remained below 300 HV₁₀ for the investigated single-pass conditions. For contextual comparison, these values were also below the 410 HV₁₀ engineering reference considered in this study [25]. However, this comparison is provided only as an engineering reference and should not be interpreted as demonstrating compliance with a maximum hardness requirement. Because the experimental procedure was not designed to determine the maximum local hardness immediately adjacent to the fusion boundary, no conclusion regarding maximum HAZ hardness or formal hardness qualification can be drawn from these measurements.

Considering only the conventional single-pass conditions, WM hardness was significantly affected by joint geometry, preheating condition, and their interaction (Table 6). The joint geometry × preheating condition interaction was significant, F(5,24) = 12.87, p < 0.0001, with a large effect size (partial ${\mathsf{\eta }}_{\mathrm{p}}^2$= 0.728). Because residual normality was not satisfied, an HC3 robust sensitivity analysis was performed; the interaction remained significant, robust F(5,24) = 10.04, p < 0.0001. For the butt joints, room temperature exhibited the highest average WM hardness, the 50 and 100 °C conditions were statistically indistinguishable, and the 150–250 °C conditions formed a common Tukey group. The two-pass results are reported descriptively and were not included in these inferential comparisons.

Considering only the conventional single-pass conditions, HAZ hardness was significantly affected by joint geometry, preheating condition, and their interaction (Table 7). The joint geometry × preheating condition interaction was significant, F(5,24) = 11.53, p < 0.0001, with a large effect size (partial ${\mathsf{\eta }}_{\mathrm{p}}^2$= 0.706). Because residual normality and variance homogeneity assumptions were not satisfied, an HC3 robust sensitivity analysis was performed. The interaction remained significant, robust F(5,24) = 83.23, p < 0.0001. For the butt joints, Tukey-adjusted comparisons identified significant differences in HAZ hardness among several conventional single-pass preheating conditions. In contrast, none of the pairwise comparisons among the six conventional single-pass conditions was statistically significant for the fillet joints (Tukey-adjusted p > 0.05 for all comparisons). The two-pass procedure was excluded from these inferential comparisons and is reported descriptively.

Residual diagnostics indicated that the ANOVA assumptions were adequately satisfied for WM and HAZ areas considering the conventional single-pass conditions. For WM area, the Shapiro–Wilk and Levene tests yielded W = 0.971, p = 0.441, and F(11,24) = 1.008, p = 0.469, respectively. For HAZ area, the corresponding results were W = 0.969, p = 0.410, and F(11,24) = 0.444, p = 0.919. For WM hardness, variance homogeneity was not rejected, F(11,24) = 1.867, p = 0.097, although residual normality was rejected, W = 0.902, p = 0.0038. For HAZ hardness, both residual normality, W = 0.728, p < 0.0001, and variance homogeneity, F(11,24) = 2.812, p = 0.0165, were rejected. HC3 robust sensitivity analyses were therefore performed for the hardness responses and confirmed the statistical significance of the joint geometry × preheating condition interactions.

3.2. Effect of Preheating Temperature on the Mechanical Performance of Butt Welded Joints

Ultimate tensile strength ranged from 638 to 703 MPa among the conventional single-pass butt welded joints (Figure 8). The two-pass procedure yielded 706.76 ± 20.13 MPa and is reported separately as an additional welding procedure. No statistically significant effect of preheating condition on UTS was detected among the conventional single-pass conditions, F(5,12) = 1.66, p = 0.218, partial ${\mathsf{\eta }}_{\mathrm{p}}^2$= 0.409 (

Table 8. Descriptive statistics, Tukey HSD groups, and one-way ANOVA result for UTS.

Joint Geometry	Welding Condition	Mean ± SD (MPa)	n	Tukey Group
Butt	Room	694.06 ± 12.85	3	a
Butt	50 °C	661.43 ± 44.87	3	a
Butt	100 °C	702.02 ± 11.02	3	a
Butt	150 °C	638.16 ± 65.40	3	a
Butt	200 °C	703.07 ± 24.45	3	a
Butt	250 °C	682.58 ± 1.16	3	a
Butt	Two-pass	706.76 ± 20.13	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Preheating condition	5	1.66	0.218	0.409
Error	12	—	—	—

Note: Values are reported as mean ± SD (n = 3). The one-way ANOVA and Tukey HSD pairwise comparisons include only the six conventional single-pass conditions. The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential analysis. Means sharing the same Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

). Consistent with the ANOVA result, Tukey HSD pairwise comparisons identified no significant differences among the six conventional single-pass conditions, which therefore shared the same statistical group. All tensile specimens (Figure 9), including those produced using the additional two-pass procedure, fractured in the BM.

Figure 9 documents the macroscopic fracture location of the tensile specimens after testing. The images are intended only to document fracture location and the overall macroscopic appearance of the failed specimens and are not presented as fracture-surface characterization. Qualitative observations indicated extensive plastic deformation, consistent with failure occurring in the BM. However, no detailed fractographic characterization was performed.

For the conventional single-pass conditions, Charpy V-notch absorbed energy did not exhibit a systematic relationship with preheating temperature (Figure 8). Because ASTM E23 subsize specimens (5 mm thickness) were employed, the reported absorbed energies correspond to the measured subsize specimens and should not be interpreted as equivalent to full-size Charpy energies. One-way ANOVA restricted to the conventional single-pass conditions showed no statistically significant effect of preheating condition on Charpy absorbed energy, F(5,12) = 2.35, p = 0.105, partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ = 0.494 (

Table 9. Descriptive statistics, Tukey HSD groups, and one-way ANOVA result for Charpy V-notch impact energy.

Joint Geometry	Welding Condition	Mean ± SD (J)	n	Tukey Group
Butt	Room	27.14 ± 4.84	3	a
Butt	50 °C	32.05 ± 6.89	3	a
Butt	100 °C	24.20 ± 1.13	3	a
Butt	150 °C	24.20 ± 0.57	3	a
Butt	200 °C	24.20 ± 3.15	3	a
Butt	250 °C	23.22 ± 1.50	3	a
Butt	Two-pass	58.86 ± 4.28	3	—
Source	DF	F	p	Partial  ${\mathbf{\eta }}_{\mathbf{p}}^2$
Preheating condition	5	2.35	0.105	0.494
Error	12	—	—	—

Note: Values are reported as mean ± SD (n = 3). The one-way ANOVA and Tukey HSD pairwise comparisons include only the six conventional single-pass conditions. The two-pass condition is reported descriptively as an additional welding procedure and was not included in the inferential analysis. Means sharing the same Tukey letter are not significantly different at α = 0.05. Partial ${\mathsf{\eta }}_{\mathrm{p}}^2$ denotes partial eta squared.

). Consistent with the non-significant ANOVA result, Tukey HSD pairwise comparisons identified no significant differences among the six conventional single-pass conditions, which therefore belonged to the same statistical group.

The additional two-pass procedure produced an absorbed energy of 58.86 ± 4.28 J, which was numerically higher than those measured for the conventional single-pass conditions. However, the two-pass procedure was not included in the preheating-condition ANOVA because deposition of the second bead simultaneously changed the number of passes, deposited filler-metal volume, cumulative heat input, weld geometry, and potentially dilution. As illustrated schematically in Figure 4b, the reported notch location refers to the evaluated WM region of the completed two-pass configuration and should be regarded as approximate rather than as documentation of its exact position relative to each individual bead. The higher absorbed energy therefore represents the response of the complete two-pass welding procedure and should not be interpreted as an effect of preheating temperature or of an isolated secondary thermal cycle.

Figure 10 presents qualitative macroscopic observations of the fractured Charpy specimens after impact testing. Because no quantitative fractographic analysis or higher-resolution examination was performed, these images are intended only for qualitative comparison among the investigated conditions and not for definitive identification of fracture mechanisms [5,26].

For the mechanical properties, the inferential analyses were restricted to the conventional single-pass conditions. No statistically significant effect of preheating condition was detected for either UTS (F(5,12) = 1.66, p = 0.218) or Charpy absorbed energy (F(5,12) = 2.35, p = 0.105). Residual normality and variance homogeneity were not rejected for UTS (Shapiro–Wilk W = 0.915, p = 0.103; Levene F(5,12) = 1.955, p = 0.158) or Charpy absorbed energy (Shapiro–Wilk W = 0.943, p = 0.328; Levene F(5,12) = 1.068, p = 0.425). The two-pass procedure was evaluated separately and produced the highest numerical Charpy absorbed energy, but this result represents the combined response of the additional welding procedure and was not included in the preheating-condition ANOVA.

4. Discussion

The present results show that, under the specific welding conditions investigated—including mechanized GMAW, 6.3 mm thick SAE 1045 steel, AWS ER80S-Ni1 filler metal, Ar–18%CO₂ shielding gas, the adopted joint geometries, and the investigated restraint conditions—the measured average HAZ hardness remained below 300 HV₁₀ for all conventional single-pass conditions. For the butt welded joints, no statistically significant effect of preheating condition on UTS or Charpy absorbed energy was detected among the conventional single-pass conditions. These hardness results refer specifically to the regional average hardness values measured at the predefined locations and should not be interpreted as evidence that preheating is unnecessary for controlling maximum local HAZ hardness. It should be noted that the discussion of tensile strength and Charpy impact energy is restricted to butt welded joints, whereas the influence of joint geometry was evaluated through the metallographic and hardness analyses. These conclusions should not be generalized to welding conditions involving different consumables, restraint levels, plate thickness, environmental conditions, or loading regimes.

The representative optical micrographs presented in this study should be interpreted as qualitative metallographic observations intended to identify the characteristic weld regions and HAZ subregions. Because only conventional optical microscopy after Nital etching was employed, no detailed phase identification, quantitative grain-size analysis, or microstructural quantification was performed. Accordingly, the discussion below is based primarily on the measured geometric, hardness, mechanical, and statistical results, whereas the metallographic observations provide only qualitative support.

The WM- and HAZ-area responses cannot be described by a single monotonic trend over the investigated preheating range [27]. In addition, the absolute areas measured in the butt and T-joints should not be interpreted as geometrically equivalent because the two configurations differ in cross-sectional geometry, fusion-boundary length, local material volume, and available heat-flow paths. Therefore, the most meaningful comparisons are those performed among preheating conditions within the same joint geometry.

For the butt joints, HAZ area tended to increase with preheating temperature. The 150 and 200 °C conditions were statistically indistinguishable, as were the 200 and 250 °C conditions, whereas the 150 and 250 °C conditions differed significantly. In contrast, butt joint WM area fluctuated among the single-pass conditions. The fillet joint HAZ and WM areas also exhibited non-monotonic responses. The significant joint geometry × preheating condition interactions demonstrate that the pattern of response to preheating differed between geometries, but they should not be interpreted as establishing geometric equivalence between their absolute area values [7,28].

Normalization by fusion-boundary length could provide a comparison that is less dependent on joint geometry; however, fusion-boundary length was not quantified during the original image analysis. Therefore, the present interpretation is based on within-geometry comparisons and on the geometry × preheating interaction. Differences in heat-flow paths, local material volume, weld configuration, restraint, and contact with the fixture may have contributed to the geometry-dependent responses [18,29]. Because thermal cycles, peak temperatures, and cooling times were not directly measured, the thermal mechanisms discussed herein should be interpreted as plausible explanations consistent with the experimental observations rather than as directly demonstrated thermal histories. Accordingly, the irregular variations are interpreted as experimentally observed interaction effects rather than as evidence of continuous change in cooling rate or thermal accumulation [1,17,29]. In addition, fusion-boundary length, HAZ normalization, and ImageJ segmentation repeatability were not quantified and therefore represent additional limitations when comparing absolute WM and HAZ areas between joint geometries.

The hardness response was dependent on joint geometry and was not uniformly monotonic across the investigated conditions. For the butt joints, average HAZ hardness generally decreased over the investigated preheating range, with significant differences detected among several conventional single-pass preheating conditions. In contrast, although numerical variations were observed in the average HAZ hardness of the fillet joints, none of the Tukey-adjusted pairwise comparisons among the six conventional single-pass preheating conditions was statistically significant. Therefore, these numerical variations should not be interpreted as statistically demonstrated effects of preheating on fillet-joint HAZ hardness. These observations remain qualitatively consistent with the sensitivity of medium-carbon steel transformations to welding thermal conditions [12,17,18]. However, because cooling rates and thermal cycles were not measured, specific transformation mechanisms cannot be directly assigned to each variation in hardness.

The measured average HAZ hardness values were below the 410 HV₁₀ engineering reference considered in this study [25]. However, this comparison is provided solely to contextualize the magnitude of the measured regional average hardness values. The adopted indentation procedure was not designed to determine the maximum local HAZ hardness immediately adjacent to the fusion boundary; therefore, the present results cannot demonstrate compliance with a maximum hardness requirement or establish that preheating is unnecessary for hardness control under a welding procedure qualification framework. The present investigation was not designed to evaluate hydrogen-assisted cold cracking susceptibility, diffusible hydrogen content, residual stresses, different restraint levels, fatigue performance, HAZ fracture toughness, or thermal-cycle parameters such as Δt_8/5. Although the infrared thermometer was calibrated and its measurement uncertainty was documented, the actual emissivity of the steel surface was not independently determined. Consequently, an additional systematic uncertainty associated with the fixed emissivity setting (ε = 0.95) may affect the absolute surface-temperature values. Moreover, the measurements represented local surface temperatures obtained immediately before welding and did not establish temperature uniformity through the plate thickness or along the complete weld path.

The additional two-pass procedure exhibited relatively low numerical average HAZ hardness values and the highest numerical Charpy absorbed energy among the investigated procedures. However, it was intentionally excluded from the preheating condition inferential analyses because deposition of the second bead simultaneously altered the number of passes, cumulative heat input, deposited filler-metal volume, weld geometry, and potentially dilution. Furthermore, neither interpass temperature nor the time interval between passes was recorded. Although the two-pass geometry and approximate Charpy notch location are documented schematically in Figure 4b, the available experimental documentation does not permit the precise notch position relative to each individual deposited bead to be retrospectively established. Consequently, the two-pass results should be interpreted as the response of a distinct welding procedure rather than as an additional level of preheating condition. The higher Charpy absorbed energy cannot be attributed to a particular bead, secondary thermal cycle, localized reheating event, or tempering mechanism [30,31,32].

No statistically significant effect of preheating condition on UTS was detected among the conventional single-pass conditions. However, given the limited number of specimens per condition (n = 3) and the relatively large scatter observed for some groups, this result should not be interpreted as demonstrating equivalence among the investigated conditions. All tensile specimens fractured in the BM, indicating that neither the WM nor the HAZ was the observed fracture location under the investigated transverse static loading conditions [23]. These observations should not be interpreted as establishing that preheating can be omitted from a qualified welding procedure.

No statistically significant effect of preheating condition on Charpy absorbed energy was detected among the conventional single-pass conditions. The additional two-pass procedure exhibited substantially higher numerical absorbed energy; however, this result was treated descriptively and was not included in the preheating condition statistical comparison. Because the second deposition simultaneously changed cumulative heat input, filler-metal volume, weld geometry, and potentially dilution, the observed response cannot be attributed exclusively to the additional thermal cycle [10,33,34]. Since these variables were not independently isolated, and no detailed microstructural or fractographic characterization was performed, the present results do not identify the dominant mechanism responsible for the observed difference in absorbed energy.

Overall, conventional preheating produced geometry-dependent changes in WM and HAZ hardness under the investigated conditions, whereas no statistically significant effect of preheating condition on UTS or Charpy absorbed energy was detected among the conventional single-pass butt welded joints. These responses are consistent with differences in the effective thermal conditions established during welding, although the thermal histories themselves were not directly measured. Given the limited sample size for the mechanical tests, these findings should not be interpreted as demonstrating equivalence among the investigated preheating conditions or as establishing that preheating can be omitted from a qualified welding procedure.

An important contribution of this study is the experimental evidence that the WM- and HAZ-area and hardness responses to preheating depended on joint geometry under the investigated welding conditions, as demonstrated by the significant joint geometry × preheating condition interactions. Future work should investigate thicker sections, higher restraint conditions, consumables with different diffusible hydrogen levels, cold and humid welding environments, residual stress development, fatigue performance, HAZ fracture toughness, repair welding conditions, and direct thermal-cycle measurements (Δt_8/5) to define the limits of applicability of the present conclusions.

5. Conclusions

This study evaluated the combined influence of preheating temperature and joint geometry on HAZ development and hardness of mechanized GMAW welds produced in 6.3 mm thick SAE 1045 steel, supported by qualitative metallographic observations, while tensile behaviour and Charpy absorbed energy were evaluated specifically for butt welded joints. Based on the experimental measurements and statistical analyses, the following conclusions can be drawn:

• (1) Considering the conventional single-pass conditions, preheating condition significantly affected WM and HAZ areas, and the response depended on joint geometry, as demonstrated by significant joint geometry × preheating condition interactions. Because butt and T-joints differ in cross-sectional geometry and fusion-boundary length, direct comparison of their absolute WM and HAZ areas should be made cautiously; the significant interaction indicates primarily that the response to preheating differed between the two joint geometries.
• (2) For the butt joints, average HAZ hardness generally decreased over the investigated preheating range, with statistically significant differences detected among several conventional single-pass conditions. In contrast, no statistically significant differences in HAZ hardness were detected among the six conventional single-pass conditions for the fillet joints. All measured average HAZ hardness values for the conventional single-pass conditions remained below 300 HV₁₀. These values represent regional averages at predefined measurement locations and should not be interpreted as maximum local HAZ hardness values.
• (3) No statistically significant effect of preheating condition on UTS was detected among the conventional single-pass conditions (F(5,12) = 1.66, p = 0.218), and all tensile specimens fractured in the BM. This result should not be interpreted as demonstrating equivalence among the investigated conditions or as establishing that preheating can be omitted from a qualified welding procedure.
• (4) No statistically significant effect of preheating condition on Charpy absorbed energy was detected among the conventional single-pass conditions (F(5,12) = 2.35, p = 0.105). The additional two-pass procedure was evaluated separately and treated descriptively because the second deposition simultaneously altered multiple welding variables; therefore, its Charpy response cannot be attributed specifically to preheating or to an isolated secondary thermal cycle.

Overall, the effects of preheating on WM and HAZ areas and hardness were joint-geometry-dependent under the investigated conditions, emphasizing that the responses should be interpreted in relation to the specific joint configuration and associated thermal boundary conditions.