Evaluation of Fatigue Performance of Press Hardening Steel Joints Welded by GMAW-CSC and PAW Processes

Abstract

The application of press hardening steels is spreading from automobiles, where crashworthiness demands are critical, to other segments, such as the agricultural and road transport industries. However, the operational conditions to which such equipment is exposed requires the application of heavier sheet gages and adapted joining processes. In this context, fatigue is recognized as the critical failure mode. The present article describes the procedures and results of fatigue testing performed on GMAW-CSC and PAW butt-welded specimens of 1500 MPa press hardening steel. Both methods are suitable alternatives to laser welding when joining relatively heavy-gaged components. The obtained fatigue results are also related to heat-affected zone characteristics and weld bead surfaces. Additionally, some ground-flush GMAW-CSC specimens were tested. The test results indicate that both welding procedures provide suitable fatigue performance. As-welded GMAW-CSC joints on this ultra-high strength steel accomplished a fatigue performance similar to welds of conventional steel. However, a remarkable performance increase was observed after flush grinding the weld beads. The plasma welding process presented less good results due to the more extensive manufacturing and geometric variability. The results indicate that adopting a suitable arc welding process, in association with post-weld geometry improvement, provides a fatigue performance that is competitive with laser-welded press hardening steels.

1. Introduction

As a result of extensive research and development efforts, current automobile structures provide a high safety level regarding crashes [1]. Such a performance is related to several design aspects but mainly originates from the current availability of high-performance structural materials [2,3,4,5]. Consequently, the right material and gage can be assigned to each specific position or function in the car body’s structure. However, joining a set of considerably dissimilar steels, some of them presenting critical weldability, may be a significant challenge. Projection and spot-welding processes, which are consolidated in the automotive industry, could be effective for joining somewhat dissimilar steels. However, these processes are typically applied in an overlap joint configuration, and, additionally, the connection between parts is attained by individual weld nuggets. Both features contribute to stress concentration and eventually force designers to specify larger plate gages, which opposes current targets. Therefore, it is important to employ joining processes which are able to promote smooth load paths in the structure when using very high strength materials.

One of the techniques capable of accomplishing these demands is laser welding [6]. The resulting joints present a narrow heat-affected zone and a low level of internal and surface defects. However, as in conventional welding processes, the geometry aspects, as well as the presence of crack-like defects at the weld toe, cause the fatigue performance of the joint to be a concern [7,8,9,10].

Despite all of the difficulties, the results obtained by employing advanced structural materials and joining processes in the automotive industry are evident [11,12]. For instance, due to the exceptional capacity of absorbing crash energy yet preventing significant intrusion, press hardening steels (PHS) are currently being used in critical parts of the body structure, forming a survival cell. The very high ultimate strength (in the order of 1500 MPa or greater) results from the martensitic microstructure, which is obtained during the hot forming and quenching process.

The same requirements that have been compelling the automotive industry to adopt press hardened steels, i.e., the necessity of lighter components and structures, are now being adopted by other industrial sectors. Using press-hardened parts in road transport and agricultural equipment, manufacturers could significantly reduce weight by using lighter-gaged longitudinal beams and fewer cross members [13]. Newly developed PHS alloys guaranteeing a fine-grained heat-affected zone microstructure after laser welding were shown to allow the manufacturing of relatively heavy-gaged parts [14,15]. Therefore, despite the apparent cost and technical challenges, road transport and agricultural equipment manufacturers became interested in employing such a material.

Nevertheless, there are important differences between passenger vehicles compared to agricultural or road transport applications. Passenger cars are primarily developed to attain high velocities on relatively smooth pavements. Thus, crashworthiness is one of the most important structural requirements, which, in turn, supports the adoption of PHS components. In contrast, heavy trucks and tractors are developed to withstand extreme operational conditions and loadings during their life cycle. Accordingly, heavier wall gages need to be employed, and fatigue is the prevalent failure mode.

Heavy wall gages cause some drawbacks in applying laser welding. High-power lasers, which are expensive, will be demanded, and, in many instances, a double joint configuration will be necessary to attain a suitable joint quality [15,16,17,18,19]. In contrast, conventional arc welding processes produce a higher heat input and, consequently, generate a comparably large heat-affected zone comprising the deteriorated material properties. Additionally, the use of established filler wire materials leads to the undermatching of the weld metal versus the exceptional mechanical strength of the base metal.

The present work investigates conventional arc welding processes parameterized to provide an as-low-as-possible heat input as a potential alternative to laser welding. The respective processes are plasma arc welding (PAW) and controlled short circuit gas metal arc welding (GMAW-CSC). The first process was autogenous, whereas in the second one, an AWS ER120-equivalent filler metal was employed. A single joint configuration was targeted for both processes. However, in the case of the PAW, a suitable joint quality could not be attained, and a double joint configuration had to be implemented instead. Fatigue tests conducted on these arc-welded samples were compared to the performance of laser-welded samples published previously by Almeida et al. [16]. Additionally, the current results were also benchmarked against the corresponding IIW reference curves, representing the fatigue performance of welded joints in conventional structural steels. Considerations related to the extent of the generated heat-affected zones will be made, and the applicability of such arc welding processes for joining PHS parts will be discussed.

2. Materials and Methods

2.1. Base Metal Properties

A modified version of the 22MnB5 steel consisting of a chemical composition including niobium and molybdenum (

Table 1. Chemical composition of the hot-rolled press hardening steel (weight percent).

C	Si	Mn	P	S	Cr	Ni	Mo	Nb	Ti	B
0.22	0.22	1.22	0.006	0.001	0.21	0.03	0.15	0.048	0.04	0.001

), for reasons that were detailed in a previous publication [15,16], was used as base metal. Accordingly, microalloying of niobium generates a population of nano-sized particles in the microstructure, restricting grain coarsening during the forming process and welding. Molybdenum, in particular, is able to tolerate moderate cooling rates in the press hardening process, which is an important criterion in the case of heavier plate gages. When compared to conventional 22MnB5 steel, formability and mechanical properties are somewhat superior, making the modified alloy especially suitable for application in road transport and agricultural components.

The material was supplied in uncoated hot-rolled condition with 3 mm sheet gage. The applied hot stamping procedure was detailed in a previous work [15], aiming to reproduce similar microstructure and mechanical properties as those obtained under industrial conditions. The quenched plates were subsequently processed to samples with lateral dimension of 75 × 150 mm². Figure 1 shows a representative martensitic microstructure of the quenched steel after hot stamping.

For characterizing the mechanical properties of the base metal, five tensile specimens were tested in the hot-rolled condition and five in the quenched state according to ASTM E8/E8M-15a [20].

Table 2. Base metal mechanical properties (YS_BM: yield strength, TS_BM: tensile strength, El_BM: elongation).

	YSBM [MPa]	TSBM [MPa]	ElBM [%]
Hot-rolled	612	739	18
Quenched	1124	1536	7

compares the obtained results. The specimens presented a somewhat higher mechanical strength than those reported by Almeida et al. [15], which is likely related to the comparably thinner sheet gage of the current samples, resulting in higher cooling rates during quenching after hot forming.

2.2. Preparation of the Welded Specimens

For exploiting the potential of ultra-high strength steels in service, the butt-weld configuration is the most recommended choice as it provides the smoothest load transfer across welded joint. The butt-welded plates had dimensions of approximately 150 × 150 mm².

2.3. Description of the GMAW-CSC Procedure

The chosen filler metal, classified as ER120S-G in accordance with the AWS A5.28 standard, came as a solid wire with diameter d = 1 mm. The shielding gas (C25) consisted of 75% argon + 25% CO₂. The mechanical properties of the filler metal in the as-welded condition, according to the manufacturer, are presented in

Table 3. Mechanical properties of the filler metal in the as-welded condition (YS_WM: yield strength, TS_WM: tensile strength, El_WM: elongation).

	YSWM [MPa]	TSWM [MPa]	ElWM [%]
ER120S-G	920	940	18

, whereas the chemical composition is shown in

Table 4. Mass percent composition of the weld wire.

C	Mn	Mo	Si	Ni	Cr
0.081	1.75	0.533	0.80	2.22	0.41

.

2.4. Parameterization of the GMAW Process

Press hardened steels are not represented in the material groups covered by AWS D1.1 [21] and ISO 15608 [22] codes, nor by the IIW recommendations [7]. Consequently, a prequalified procedure was not available, and an experimental approach had to be pursued for determining the welding parameters.

Despite the uncommon characteristics of the base steel, the AWS D1.1 [21] code served as a reference to define the joint preparation parameters. Thus, the following specifications were adopted:

(i)

One welding pass suffices for obtaining a complete penetration joint;

(ii)

Bevel angles are not necessary;

(iii)

Root opening: R_o ≅ 1.4 mm.

The torch displacement for welding was actuated by a robot. A power source with arc control strategy was used with base metal and shielding gas as input data. Welding optimization consisted of a step-by-step adjustment of wire feed speed and arc length (trim parameter). At this stage, visual inspection was performed as qualifying parameter. The contact tip to the work distance was preset to CTWD = 12 mm in the robot software. Visual inspection was also used to establish the welding speed. During the welding process, after a short period of arc stabilization, the voltage and current values were recorded using a data acquisition system with specific sensors. The following parameters provided the best results: wire feed speed v_f = 3 m/min, arc length (trim) a_l = 1.5, and welding speed v_w = 230 mm/min. The correspondingly recorded mean voltage and current values were V_m = 17.5 V and I_m = 81 A. Lastly, the adopted shield gas flow was Q = 16 L/min.

During the parameter adjustment process, it was observed that for avoiding undercut in the toe region, a relatively high reinforcement would be necessary. Consequently, fine-tuning of the welding parameters aimed to promote a balance between undercut and reinforcement height.

After an appropriate bead aspect was achieved, joints were cross-sectioned at randomly chosen positions for preparing macro images of the obtained sections. In both steps, the complete penetration and absence of important defects were the main evaluation criteria. Adopting BS 7910 code [23] and IIW recommendations [8] as references, the following maximum dimensions were established for internal and surface defects:

(i)

Void diameter: dv_max = 0.3 mm;

(ii)

Lack of fusion (crack-like defect in the fusion line): hc_max = 0.15 mm;

(iii)

Undercut: u_max = 0.3 mm (u/t ≤ 10%);

(iv)

Misalignment between plates (or eccentricity): ml_max = 0.15 mm (ml/t ≤ 5%).

These limits, which correspond to an intermediate joint quality level, were employed to ensure that the fatigue process would neither be governed nor mainly be affected by the defects. Figure 2 shows the cross section of a typical obtained GMAW-CSC joint.

2.5. Description of the Plasma Arc Welding (PAW) Procedure

The autogenous PAW procedure was performed with a tungsten electrode using a tip angle of 30°. The electrode tip was positioned approximately 1 mm backwards from the nozzle edge. Pure argon was used both as shielding and plasma gas. The gage t = 3 mm corresponds to an intermediate class between thin and thick sheets. Both methods had to be tested prior as the melt-in operational mode is recommended in the thin sheet case, and the keyhole technique is suitable in the other case. The first attempt was performed with a single welding pass using the keyhole mode. Nevertheless, arc instability occurred and, consequently, the melt-in mode had to be chosen. As complete penetration was not reached, a double joint configuration was finally adopted. The same procedure for joint quality evaluation, as in the GMAW-CSC case, was applied.

2.6. Parameterization of the PAW Process

In the adjustment procedure for obtaining a sound weld joint, secondary parameters were fixed:

(i)

CTWD = 6 mm;

(ii)

Shield gas flow: Q = 12 L/min;

(iii)

Plasma gas flow: Q_plasma = 3 L/min.

No bevel angle was applied similar to GMAW-CSC welding. The faces to be joined were put into contact, and extension plates were added to the joint to provide enough space for arc opening, stabilization, and closing. Afterwards, the terminal parts of the joint were cut off, leaving a proper welded plate from which the test specimens were extracted. Figure 3 shows a typical cross section of the obtained PAW joints.

2.7. Characterization of the Welded Joints

The following test procedures were used for joint evaluation:

(i)

Microstructural analysis of the heat-affected region;

(ii)

Tensile tests;

(iii)

Vickers (HV0.5) hardness profiling;

(iv)

Surface laser scanning;

(v)

Fatigue tests for obtaining the respective S-N diagrams.

3. Results and Test-Related Aspects

3.1. Microstructural Analysis

Before presenting the different microstructures observed inside the heat-affected zone, it is worth mentioning that the press hardening process induced a decarburization layer at the surface of the base metal. Medium- and thick-gaged press-hardening steels are usually provided without surface coating so that both oxidation and decarburization can occur. Optical measurements of the cross-section samples showed that the decarburization layer thickness remained in the range of 50 to 90 μm. Although such a soft layer would significantly impact the mechanical properties in the case of thin sheets, no relevant difference was noticed in the thicker gages investigated in the present study.

As observed in Figure 2 and Figure 3, both arc welding processes generated similar heat-affected zones, however, with different dimensional extension. The weld metal in the fusion zone of the GMAW-CSC weld (Figure 2) presents the typical columnar grain morphology of a solidification structure. Dilution of the filler metal occurs resulting in a higher carbon content in the weld metal than in the original wire. The coarse-grained HAZ has an extension of approximately 2 mm. The fine-grained HAZ ranges from ~3.5 to ~5 mm. Implemented thermal measurements (explained in Section 3.4) allowed us to estimate the size of the intercritical HAZ.

With the exception of the autogenous fusion zone of the PAW process, i.e., without dilution, the configuration of the HAZ was similar to that observed in the GMAW-CSC joints. However, a larger heat-affected zone was obtained due to the higher heat input, and the first welding pass was tempered by the heat of the second pass.

The significant microstructural heterogeneity seen in Figure 2 and Figure 3 can be relevant to fatigue performance. Thus, it was crucial to measure the strength variation inside the heat-affected zone using tensile and hardness tests.

3.2. Tensile Test Results

Tensile tests were carried out aiming to evaluate the influence of the joints on the strength under static stress. The specimens were based on the ISO 4136 (2013) standard. The reinforcements were removed from the GMAW-CSC welding specimens before testing. In the case of the autogenous plasma-welded samples, no finishing procedure was applied. Sets of five specimens were tested for each case. The obtained mean values are listed in

Table 5. Mean tensile test results of the welded specimens.

	YS [MPa]	TS [MPa]	El (%)
GMAW-CSC	945	1028	16
PAW	1161	1235	11.8

.

In the GMAW-CSC process, three specimens experienced final fracture in the light-contrasted region corresponding to the intercritical HAZ. The remaining specimens fractured in the fusion zone, achieving an ultimate strength close to the value of the undiluted welding wire material listed in Table 3 (TS_WM = 940 MPa). Hence, the characteristic of these joints was undermatching relative to the original base metal (TS_BM = 1536 MPa). It is worth mentioning that the fusion zone and the outer part of the HAZ presented a comparably low strength level. The PAW welded joints were approximately 200 MPa stronger than the GMAW-CSC welds. This is due to the autogenous nature of the fusion zone. However, the tempering effects by the second weld pass prevent the fusion zone from achieving the original strength of the base metal. Two of the five specimens fractured in the outer part of the HAZ and three in the fusion region.

3.3. Hardness Testing across the Joints

Hardness profiles were taken across the weld area using the HV0.5 scale. Figure 4 shows the obtained profiles for the GMAW-CSC process. As expected for such a martensitic steel, the original hardness of around 500 HV is significantly reduced by tempering or recrystallization effects. The lowest hardness is observed in the intercritical reheated zone where new ferrite is formed during the weld cycle, and existing martensite is heavily tempered. The fusion zone still achieves a relatively high hardness due to the alloy content of the wire promoting full martensite transformation. However, the absolute hardness is lower than in the base steel due to the lower carbon content in the fusion zone.

Figure 5 shows the hardness profile obtained for the PAW technique. Compared to the GMAW-CSC process, the heat-affected zone is much wider, and the fusion zone is softer. Even at the start and end positions, i.e., at +/− 15 mm from the weld center, tempering of the original martensite is observed. However, the hardness variation across the weld seam shows a smoother behavior. By the interaction of the two welding passes, the fusion zone also does not develop a fully martensitic microstructure and is, thus, much softer than that of the GMAW-CSC process. As the PAW process was autogenous, the fusion zone did not contain the substantial amount of hardenability-improving alloying elements as in the GMAW-CSC welds.

3.4. Temperature Measurement during the Welding Process

A measuring procedure was implemented to estimate the temperature range attained during the welding process in the light-contrasted columns shown in Figure 2 and Figure 3. Four type-K thermocouples were fixed in the opposed face to the melting pool. They were transversally disposed in relation to the joint central plane. A data acquisition system converted the signals, which were recorded using a computer. The measurements were made on eight welded plates for each welding technique. Some results had to be discarded because of unexpected incidents such as the dropping of the thermocouple terminal. Figure 6a,b shows the obtained results. The fitting curves were superposed to the acquired points, providing an approximate mean temperature-to-distance relation. It is worth mentioning that the adopted measuring procedure was not able to generate extremely accurate values but provided comparable data sets instead.

The zone with bright contrast in Figure 6 represents a mixed microstructure of heavily tempered martensite and newly formed ferrite as heating occurred in the temperature range between the A_c1 and A_c3 temperatures. It is particularly this zone that showed the lowest hardness values in Figure 4 and Figure 5. Moving away from the A_c1 limit, the original martensite was tempered to a decreasing degree, i.e., the hardness increased. In the area where the peak metal temperature was above A_c3, a normalized (fine-grained) microstructure was produced followed by a coarse-grained zone closer to the fusion line. These zones were narrower for the GMAW-CSC process compared to the double-pass PAW process. It should be emphasized that the metallurgical effects, occurring in the heat-affected zone of the laser-welded PHS and described by the authors in more detail in [15], apply principally in the same way to the current arc welding processes. Only the dimensions of the various subzones in the HAZ are larger due to the higher heat input by the arc welding processes.

3.5. Geometry of the Bead Surfaces

Joint geometry is one of the most important aspects regarding fatigue performance [7,8,9,10,24]. The bead surface profiles of the five specimens of each welding process were laser-scanned before fatigue testing. Only the GMAW-CSC process generated significant reinforcements, for which the mean values were:

• At the weld face: length l = 5.03 mm and elevation h = 1.72 mm, corresponding to the ratios h/l ≅ 0.34 and h/t ≅ 0.57;
• At the weld root: length l = 2.22 mm and elevation h = 1.48 mm, resulting in the following ratios h/l ≅ 0.67 and h/t ≅ 0.49.

The PAW joints did not comprise the reinforcements. Thus, a related stress concentration factor at the weld toe was not expected. Nevertheless, microcracks were eventually present in this specific region, regardless of the employed welding process. The sum of effects was unfavorable for the GMAW-CSC process. Consequently, a poor fatigue performance would be expected without the removal of the reinforcements. The fatigue damage process is multifactorial, and some additional effects may become active [7,24,25,26]. However, the GMAW-CSC specimens exhibited lower scatter in the measured geometry parameters, indicating that the process is more stable and provides higher manufacturing repeatability. In contrast, the topography of the PAW joints was more irregular, exhibiting sparse and subtle undercuts.

3.6. Fatigue Testing of GMAW-CSC Samples

A preliminary set comprising 21 flat specimens, with the same geometry as used in the tensile tests (dimensions of the neck cross section: thickness t = ~3 mm, width w = 13 mm, and transition radius between the neck and the grips r = 25 mm) was assigned for the fatigue evaluation. The set of specimens was divided into two subgroups:

(i)

Thirteen specimens with original as-welded joint geometry;

(ii)

Eight specimens with improved geometry by grinding the reinforcements’ flush.

The reinforcement removal consisted of a manual grinding process, applied in such a way that the generated cutting grooves remained nearly aligned with the load direction of the fatigue tests. Three ground-flush specimens experienced fracture inside the grip area due to a fretting process. Two specimens fractured in a section far from the heat-affected zone, i.e., in the linear-to-circular transition zone.

To mitigate these issues, an improved sample geometry was established for the remaining specimens. The modification consisted of a lower admissible neck width in accordance with the ISO 14345 standard [27] (that is: w ≅ 3 t) and a greater transition radius in order to lower the stress concentration factor. Additionally, the grip region was enlarged as far as possible. This optimized sample geometry is shown in Figure 7.

A batch of 18 specimens was manufactured with the optimized geometry, and all of the specimens were assigned for the evaluation of samples after flush grinding. It is worth mentioning that in both the original and final samples’ geometries, the stress concentration factor at the end of the neck is unitary according to Wilson and White [28]. An additional investigation using a microscope and roughness tester was applied at the transition points to identify potential machining defects that could promote the onset of the fatigue damage process. However, the surface was similar to that observed inside the neck region. An extremely careful sanding process at the transition region was then employed to all the specimens.

To avoid fretting damage, two aluminum plates were fixed to each gripping area, protecting them from indentation during the clamping force application. Despite all of the adopted precautions, fretting and fracture at the transition region occasionally occurred in some tests. In the first case, the specimen data were discarded because fretting strongly promotes premature failure. However, it was further observed that the transition region competed with the heat-affected zone as the failure initiation site, specifically for the ground-flush GMAW-CSC samples. Following a conservative approach, both populations were aggregated to compose the fatigue diagram.

All S-N diagrams shown in the following sections were elaborated in accordance with the ASTM E739-10 standard [29]. A load ratio of R = F_min/F_max = 0.1 was applied during the tests. Figure 8 shows the obtained S-N diagram (specimens without surface improvement). The obtained line inclination is m ≅ 3.76, and the stress range corresponding to 2(10⁶) cycles over the 95% survival line is Δσ_2E6 ≅ 101.8 MPa. No runout was observed for these samples, and the damage process always occurred in the weld toe region.

For transverse butt-welded joints made in a flat position in a workshop environment and with a weld reinforcement height lower than 10% of the thickness value, the inclination is m = 3 according to the IIW recommendations [8]. The stress range corresponding to 2(10⁶) cycles, over the 95% survival line, is Δσ_2E6 = 90 MPa. The code assigns such a curve to all structural steels with yield stress up to 960 MPa. Consequently, the results obtained in the GMAW-CSC case are considered adequate for general purpose structures. However, such results corroborate the IIW recommendations, according to which the stress concentration at the weld toe generates and governs the failure process. In such a case, material aspects are considerably less important than joint geometry effects.

The uniformity in the fracture process of the as-welded specimens was not observed in the ground-flush samples, where the final failure occurred in two different regions:

(i)

Inside the welded joint (eight specimens);

(ii)

Far from the heat-affected zone (seven specimens).

In this last case, the prevalent failure sites were the neck extremities at the beginning of the transition radius. Two specimens fractured in a region relatively distant from the neck region yet not inside the gripping zone. Figure 9 shows the S-N diagram for this case. The slope of the S-N curves is m ≅ 5.9, and the stress range associated with 2(10⁶) cycles in the 95% survival line is Δσ_2E6 ≅ 288 MPa. A group of three specimens experienced failure in the gripping region because of fretting damage. These results were discarded.

As observed, four run outs were obtained. The condition to interrupt the test and consider a run-out case was not sharply determined before the test. An estimate of life, associated with the 5% survival line, was previously computed taking into consideration the precedent results. So, the test was interrupted when the number of cycles far exceeded the previously estimated value.

Additionally, it was observed that two specimens achieved the run-out condition in the stress range Δσ ≅ 456 MPa. However, one specimen tested in a lower stress range, i.e., approximately 440 MPa and attained only ~600.000 cycles. As no particular difference was observed in the fracture surfaces in this last case, the two run-out specimens were tested again, using a higher stress range value (Δσ ≅ 540 MPa). The main idea behind this was to qualitatively evaluate whether these specimens had been produced with a higher integrity level inside the natural scatter of manufacturing. However, the newly obtained results (indicated by arrows in the diagram) were near to the expected value. Nevertheless, it is reasonable to assume that the initial tests with the lower stress range, in which the run outs occurred, might have already generated some fatigue damage. Accordingly, the retesting results were not considered for computing the S-N curves of Figure 9.

3.7. Fatigue Testing of Plasma Arc-Welded Samples

A group of 16 specimens with the same geometry (Figure 7) was employed in this case. Although a relatively irregular surface was produced in the fusion zone, no severe protrusions or undercuts were observed. As a result, no surface improvement process was applied to the specimens.

Figure 10 shows the obtained S-N diagram after plasma arc welding. The obtained line inclination is m ≅ 3.49, and the stress range corresponding to 2(10⁶) cycles over the 95% survival line is Δσ_2E6 ≅ 111.3 MPa. All specimens experienced the damage process in the heat-affected zone. Two run outs occurred, and, similar to the previous case, the specimens were retested at higher stress ranges. Both samples achieved lifecycles near to the expected value. However, the results were not used for computing the S-N lines.

4. Discussion

4.1. Comparison of Fatigue Results for Different Welding Processes

Figure 11 summarizes the 95% survival curves corresponding to the cases described in the previous sections and superposes the data of the laser-welded joints of the same steel reported by Almeida et al. [16]. Later, the obtained S-N curves will be compared with the IIW reference lines correlated to the butt-welded joints. Following the IIW procedures, the S-N lines of Figure 11 were extended to N = 10⁷ cycles, where an inflexion point separates the high from the very high cycle life regions. According to Hobbacher [8], the recommended line inclination beyond the inflexion (knee) point has an m-value of 22.

A comparison of fatigue test results in welded specimens should take the specific characteristics of each studied case into consideration.

Table 6. Main characteristics of the joints with reference data.

Joint	Weld Form	Pass No.	Geometry	Δσ2E6 [MPa]	m
GMAW-CSC	filler metal	1	high reinforcement	~102	3.76
GMACSC/GrFl	filler metal	1	ground-flush	~288	5.9
PAW	autogenous	2 opposed	subtle irregularities	~111	3.49
Laser [11]	autogenous	2 opposed	small reinforcement	~135	3.86
Reference curves
IIW80	transverse butt weld not satisfying the IIW90 conditions			80	3
IIW90	transverse butt weld made in shop in flat position. Reinforcement < (0.1) t			90
IIW112	transverse loaded butt weld ground flush to plate			112

summarizes some relevant parameters and aspects of the joints analyzed here and where a larger m-value indicates a superior fatigue performance. The corresponding IIW reference curve for each case is also mentioned. It is worth observing that such reference curves are not supposed to be applied in the case of very high strength steels which are not covered by the IIW recommendations or structural codes such as Eurocode BS EN 1993-1-9:2005 and AWS D1.1 (2020) [21]. However, such reference curves were inserted in Table 6 to give support to the subsequent discussion of the results. Figure 12, Figure 13 and Figure 14 separately display the obtained S-N lines along with the respective IIW reference curves.

4.2. Interpretation of the Results

The results obtained on the arc-welded samples in comparison to the fatigue performance of the laser-welded samples of the same steel investigated earlier [16] indicate that:

(i)

In GMAW-CSC welded samples with original joint geometry, the fatigue damage process is mainly controlled by the weld reinforcement geometry. All specimens experienced the onset of the fatigue damage process at the weld toe. Figure 12 indicates that the common IIW80 curve could be satisfactorily applied to this case, although non-conservative values are observed in the region near to the low cycle fatigue life, specifically at Δσ ≥ 264 MPa (N ≤ 55,000 cycles).

(ii)

The set of ground-flush GMAW-CSC specimens exhibits scattered results. Some specimens fractured in the fusion zone, whereas others failed in the part of the HAZ comprising tempered martensitic microstructures, and the remaining samples fractured away from the heat-affected region. These three regions apparently presented a similar severity level for fatigue failure. Such a case provided better fatigue performance with the obtained S-N line, presenting a notably smaller slope value and a higher position in relation to the IIW112 reference curve (Figure 13). In fact, the IIW160 curve, presenting a slope of m = 5 and assigned to non-welded rolled or extruded products, would still be applicable to this case. This result obviously indicates that by avoiding issues related to joint geometry, the remarkable mechanical strength of the PHS base metal can be exploited in terms of providing a superior fatigue performance.

(iii)

The autogenous PAW joints exhibited an unsatisfactory fatigue performance, similar to that observed for the original GMAW-CSC specimens (Figure 13). This was due to a severe geometry-related stress concentration factor. Despite small irregularities that could be observed at the faces of the heat-affected region, all PAW specimens experienced fatigue failure in the fusion zone, indicating that the joints presented an inadequate quality level. As mentioned before, the original idea was to apply a single-sided welding pass. However, the necessity of an additional welding pass on the opposite side for attaining a full penetration joint eventually caused a lack of quality inside the fusion zone. To some extent, this observation agrees with the statement made in the precedent item: when the stress concentration generated by the weld geometry is irrelevant, the material region with the lower strength (the fusion zone in the present case) represents the weakest link for fatigue performance. In addition, the extension of the HAZ of the PAW joint is notably larger. This fact may cause difficulties when designing a complex-shaped structure. Consequently, and considering the results obtained so far, PAW is not recommended for joining PHS parts. However, it is recognized that the analyzed sample base is rather limited. The optimization of the plasma welding process probably could promote better weld quality and, consequently, an improved fatigue performance.

(iv)

The laser-welded specimens tested previously by Almeida et al. [16] presented an acceptable fatigue performance (Figure 14). With conservative contemplation, the obtained results can be considered to be matching the corresponding IIW90 reference line. However, the same study indicated that superior fatigue performance can be achieved when the reinforcements are removed. In this line of thought, it is interesting to observe that even in the case of very high-quality joints such as those obtained by laser welding, the fatigue results are inferior to the GMAW-CSC ground-flush specimens (Figure 11) despite the more substantial microstructural changes in the HAZ in terms of severity and geometrical extension. This observation emphasizes the relevance of joint geometry aspects and indicates that even small and smooth reinforcements are detrimental to fatigue performance.

(v)

Codes and recommendations related to general welded structures address only ordinary and high-strength structural steels. If, for these steel classes, an ordinary quality level is maintained in the manufacturing process, the geometry-related aspects are prevalent for the fatigue damage process. All results reported in the present study indicate that such an approach could be extended in the case of PHS parts. However, the results also indicate that if an effective geometry improvement procedure is applied to the joints, the fatigue performance of the welded PHS components could be notably superior to the IIW160 curve, which is actually assigned to the non-welded parts of ordinary and high-strength structural steels.

5. Conclusions

The current study analyzed the fatigue performance of arc-welded weld joints in heavier gage press hardening steel (PHS) in comparison to laser-welded joints that have been previously investigated by the authors. As welded joints unavoidably contain crack-like defects resulting from the manufacturing process, the propagation of these cracks is mainly influenced by metallurgical and geometric features. From a metallurgical point of view, some microstructures formed in the heat-affected zone have a brittle character, which can enhance crack propagation. On the other hand, the joint geometry causes stress concentration also promoting crack propagation. The results obtained from this study support the following conclusions:

• Press hardening steels can be reliably welded using electric arc processes. This is a very important result for industrial sectors that may apply these steels with relatively heavy gages and where the application of laser welding is not feasible.
• The changes in metallurgical and mechanical properties caused by electric arc welding processes are qualitatively similar to those by laser welding. The main difference relates to the dimensional extension of the heat-affected zone.
• The fatigue performance of laser-welded joints is not immune to geometric issues. Even very small reinforcements cause a decrease in fatigue life.
• An arc-welded joint subjected to a geometry improvement procedure has a better fatigue performance than laser-welded joints without improvement.
• The advantages of applying laser welding instead of conventional arc welding are apparently reduced to:

(i)

Production aspects;

(ii)

Low thermal distortion in the case of thin-gaged parts;

(iii)

The generation of a very narrow heat-affected zone.

This last aspect is important from a design point of view. Due to the decay of properties in the heat-affected zone, as well as the related stress concentration, welded joints represent the weakest link in the structure. Therefore, joints should not be positioned in highly stressed areas. Hence, a welded joint with a narrow heat-affected zone causes fewer constraints regarding the design of complex structures.