Multi-response optimisation of wire-arc additive manufacturing process parameters for AISI 4130 steel during remanufacturing process

Wire-arc additive manufacturing (WAAM) has emerged as a critical tool for remanufacturing industrial components. A limited understanding of this technique for quality product manufacturing has hindered its utilisation for industrial applications. This study reports on the optimisation of WAAM process parameters for AISI 4130 steel towards remanufacturing of high-quality products for industrial applications. AISI 4130 steel was selected for this study due to its high strength-to-weight ratio, excellent weldability, and suitability for the WAAM process. Taguchi’s Grey Relational Analysis (GRA) used four factors and three levels in the multiple response optimisation process. The study considered process parameters voltage, current, travel speed and gas flow in the gas metal arc welding (GMAW)-based WAAM technique. Analysis of Variance (ANOVA) results show that voltage, travel speed and gas flow significantly affect material deposition. Voltage had the highest significance (31.61%) compared to other parameters. The optimised process parameters were found to be: voltage –23 V, current –100 A, travel speed −350 mm min−1, and gas flow −10 L min−1. These parameters resulted in tensile residual stresses of 25 ± 74 MPa, microhardness of 171.4 ± 12.2 HV0.3, and a relative density of 98.21%. The microstructural analysis reveals the existence of predominant ferritic and pearlitic colonies. This is due to compounded thermal stresses during the deposition process and alloy composition resulting in tailored microstructure and mechanical properties. The study provides some insights into the WAAM remanufacturing process for producing highly quality industrial components.


Introduction
Additive manufacturing (AM) is an emerging digital technology for fabricating, manufacturing and repairing geometries by layering materials in consecutive layers [1][2][3][4].AM has garnered greater focus than conventional manufacturing procedures due to its various applications in producing high-strength components and its potential to reduce lead time and cost [5].Conventional manufacturing procedures are time-consuming and expensive [6].These demands the experience of the engineers and designers to influence the product quality [7][8][9][10].Time and high production costs of conventional methods have a notable negative impact on the efficiency of repairing broken parts.As a result, industries suffer significant costs due to downtime caused by machine part failure [9].Williams et al [11] compared the cost of wire-arc additively manufactured (WAAMed) components to the traditional machining procedure.According to the study, cost savings for WAAMed parts ranged from 7% to 69%.However, understanding the deposition effects and characteristics is required for this technology to be reliable for component manufacture [12][13][14][15].
WAAM is an additive manufacturing process for constructing large-scale metallic components [16,17].This method employs an arc as a source of heat energy to deposit materials layer by layer, providing a novel approach Component Analysis (PCA).PCA is a reliable statistical method used in multi-purpose optimisation to reduce the size of complicated and interrelated datasets [61].This approach facilitates the discernment of the relative importance of high-quality responses, paving the way for a more advanced strategy in multi-objective optimisation [51].PCA simplifies multi-objective optimisation by lowering the number of dimensions in the data, effectively transforming it into a single-response optimisation problem without losing the critical information encoded in the original dataset [62].

Materials
The elemental composition of the substrate and feedstock wire of AISI 4130 steel, as presented in table 1, forms the basis for this study.
The 0.9 mm diameter crown alloy FH-30 4130 wire was supplied by IOC Welding Supplies, Indianapolis and the 9 mm thick normalised AISI 4130 substrate plate was obtained from ONLINE METALS, United States.
Figure 1 shows the initial microstructure of the substrate steel.The steel exhibited a lamellar columnar structure composed of ferrite (α), pearlite ('P') and carbides ('C').
The mechanical properties of the wire as supplied by the manufacturer are yield strength-896 MPa, tensile strength-1000 MPa, and an elongation-11%.
Table 2 provides a comprehensive overview of the specific properties of AISI 4130 steel adopted in this study.

Experimental procedure 2.2.1. WAAM process
The deposition process was conducted using a custom-made WAAM rig shown in figure 2. The rig consists of an AICO Inverter welder MIG/200 GMAW machine, (AICO, Japan).Filler alloy steel wire, 0.9 mm diameter was deposited on AISI 4130 alloy steel substrate.
The study examined key operating parameters -travel speed, current, arc-voltage, and gas flow during the deposition process of AISI 4130 steel.The operating conditions and levels applied are indicated in table 3.
A Taguchi L 9 Orthogonal Array (OA) [64] was devised with nine trials to carry out experiments.The deposition process adopted constant parameters outlined in table 4. A single track was deposited for each of the nine experiments to ascertain the layer height, width and to visually inspect for any noticeable flaws.The dimensions of the measured bead were used to generate the G-code, which controlled the torch height during the deposition of each layer.A total of ten layers were deposited for a single bead, each starting at the origin and  following a G-code.After the successful deposition of each layer, a natural air-cooling process was allowed for 30 s before commencing the next layer.

Mechanical property evaluation 2.3.1. Residual stress test
In deposited condition, AISI 4130 steel samples were analysed for residual stress at the surface at a 90-degree angle to the deposition path (figure 3) at seven different points using the m-X360 s Residual Stress Analyser (Pulstec Ltd, Tokyo).Electrochemical polishing was systematically executed on the surface of each sample to remove roughly 0.1 mm of surface material for ten successive polishing iterations.Subsequently, residual stress measurements were   acquired after the electrochemical polishing procedure at seven distinct measurement points.The electrochemical polishing was carried out using the Pulstec Electrochemical polisher (Pulstec Ltd, Tokyo) at 1.26 A for 3 min at each iteration according to Pulstec Electrochemical polisher manual [65].The recommended test conditions are presented in table 5.

Microstructure analysis
The microscopic structure of sectioned WAAMed samples were examined.The samples were cut perpendicular to the deposit direction.The sectioned surfaces were then ground and polished for microstructural analysis.Grinding was done using 220, 600 and 1200 grit papers, followed by sequential refining using 3, 1 and 0.25 μm diamond pastes, and then 1 μm colloidal silica suspension according to ASTM standard E3-11 [66].
The polished specimens were treated with a 2%-Nital solution, a combination of 2% Nitric acid and 98% ethanol for 10-15 s.The Olympus BX41M-LED Optical Microscope (Tokyo, Japan) and JOEL JCM-7000 Benchtop Scanning Electron Microscope (SEM) (Tokyo, Japan) were employed to acquire the micrographs.The SEM images were captured utilising an acceleration voltage of 10 kV.In the Taguchi method, a distinctive design of orthogonal arrays to explore the complete process parameter space was used.The L9-orthogonal array was employed for the evaluation of 4 operating parameters at 3 levels.
The determination of the minimum number of experiments to be conducted was calculated using Taguchi's degree of freedom approach illustrated in equation (1) [68].
Where L i is the total of OA applied in the Taguchi method, NV is the number of generations carried in the Taguchi method.Table 6 illustrates the nature and combination of experiments generated in Minitab LLC ® 2023 carried out in the study produced by the orthogonal array.

Signal-to-noise ratio
The signal-to-noise (SN)-ratio is crucial in acquiring the means of the outcome responses.The product condition improvement is to maximise the SN-ratio applicable to the respective product [69].The calculation of the SN-ratio using Taguchi's robust experimental design module is based on the 'Larger is better' type.The 'Larger is better' approach in quality control and product optimisation aims to maximise the SN-ratio, which aids as a measure of quality [68].A high SN-ratio signifies superior product quality without the need for factor adjustments initially.The SN-ratio for assessing the superior quality of responses was developed from the equation (2) [68].
where, the amount of trial test is depicted by n, Y i being the i th numeral of quality characteristics.2.4.3.Grey relational analysis Multiple system performances were analysed using the grey relational method.The various performance parameters were consolidated into a single grey relational grade (GRG) [70].
Step 1: In Taguchi analysis the responses were normalised in the range of 0 to 1 through:

Larger is better
For the GRA, responses were normalised to 'Larger is better' to determine the optimal quality of the material.The GRG was calculated by averaging the grey relational coefficients (GRC) corresponding to each response.[51].The 'Larger is better' value in grey relational was obtained through equation (3) [68].
Where, X ij is the normalised SN-ratio, Y ij is the SN-ratio of analysed by Taguchi, Y ij max and Y ij min are the max and min SN-ratio obtained.

Deviation Sequence
The deviation sequence was obtained using the equation (4): Where ij ∆ is the deviation sequence, defined as the absolute distinction between reference sequence X i and comparison sequence X .j Step 2: In process analysis using the GRG, the obtained values indicate the relational degree between each sequence of operations [70].The GRG of optimal quality is obtained through equation (5).
Where, GC ij is the GRG, j is a distinct coefficient whose value spans from zero to one, Δ min and Δ max are the maximum and minimum absolute differences [43,71].
Step 3: By finding the average grey-relational coefficients (GRC) the GRG is calculated using the equation ( 6): where, GC ij is the GRG, m is the quantity of response variables, G i is the GRG.

Principal component analysis (PCA)
In this paper, PCA was used to estimate the best value of the coefficient, j in the GRC in equation (5), which is in the range 0 to 1 as a weighting factor [71][72][73].
The GRC computation for response variables was utilised to construct a matrix, as per equation (7): Where y n m ( ) is the GRC of all quality responses, m = 1, 2Ks, runs and n = 1, 2Kt, quality outcome.In this study, s = 9 and t = 4.
The correlation coefficient matrix can be produced through the use of equation ( 8): is the sequence covariance of y n m ( ) and y l ; s ( ) n ym ( ) s and l ym ( ) s are standard deviations for the sequence y n m ( ) and y l , s ( ) respectively.s (standard deviation).The eigenvalues and eigenvectors are calculated from R sl array as per equation ( 9) Thereafter, eigenvalues ( t j ) and eigenvectors (V mt ) of square matrix R are employed to ascertain uncorrelated principal components (PC'c) by using equation ( 10) shows the first principal component being Z , s1 the 2nd being Z s2 and continuing.

Results and discussions
The mechanical characteristics of the AISI 4130 substrate plate were measured.The substrate had: microhardness of 170.2 ± 6.9 HV 0.  7 shows the experiment results obtained for each set of runs according to the Taguchi design of tests.The measured responses were: hardness, relative density and residual stress.Notably, the results show that samples sectioned from the deposited region exhibited lower hardness values than that of the substrate.This phenomenon can be linked to the intricate interplay of factors such as the variation of thermal cyclic loading across different layers and the microstructure realignment within the material.Similar results have been documented by Zhou et al [74].According to Hu et al [75], grain size increases with layer height.This was owing to the distinct temperature cycles that the layers underwent.As the successive layers were overlayed, the temperature across the entire workpiece steadily increased.Consequently, the cooling rate reduced, similarly to the hardening stages fraction, causing lesser hardness.
The hardness of the material after deposition exhibited low hardness values near the substrate and gradually increased in the upper deposited layers [38].This trend is primarily due to distinct heat profiles undergone by different regions through the building process.The cooling rate diminishes with a rise in layer height during the multi-laying process [76].The process of cyclic thermal loading may have caused over tempering of the previously deposited layers, which in turn may have reduced the hardness.Low values of relative density were observed for all the samples.The low relative density values may have resulted from varying heat inputs, grain sizes, and entrapment of gases in the deposited region, causing pore formation [75].
Table 8 indicates the SN-ratio of the responses calculated using the equation (2) larger is better option.These SN-ratios provides a basis of the analysis to evaluate each parameter contribution to the responses under investigation.A higher SN-ratio shows a stronger signal relative to noise, which suggests that a specific factor has a more noticeable impact on response variables [77].

Residual stress
Residual stresses refer to interior stresses within a material that persist following the WAAM deposition process.Optimising WAAM parameters to minimise residual stresses is crucial for preventing deformation, warping, and cracking in the printed components, thus improving the overall structural integrity and performance of the WAAMed components.
Figure 4 shows residual stress measurements and results orientation for this study.The results were automatically generated using a residual stress testing machine.The display offers a broad spectrum of valuable insights into the distortion and graphical intensity of stresses.The focus in this section is the Sigma(x) values, which directly represent the residual stresses within the sample.Figure 4 depicts single-point residual stress measurements for experiment samples No. 2 and No. 7 (selected at random), revealing the critical relationship between orientation and the dataset.Positive residual stress measurements denote tensile residual stresses in a material, whereas negative measurements show compressive residual stresses.Attaining a state of zero residual stresses is a desirable objective in manufacturing, as it helps to significantly mitigate the likelihood of deformation and cracking, thereby enhancing the longevity and performance of the component [78].In mining equipment, the ability to withstand compressive stresses is crucial, hence the preference for positive (tensile) residual stresses [79].These positive residual stresses enhance the overall compressive strength of mining drill   bits, particularly against cyclic loading and bending loads, thereby improving their fatigue resistance.However, it is important to note that excessive residual stresses, whether tensile or compressive, can lead to deformation and cracking, thereby reducing the performance and longevity of the drilling bits [80,81].This ensures that they have the mechanical qualities required for reliable performance in compression-dominated and cyclic loading situations common in mining applications.
Figure 5 shows result of the main effect plots for means and SN-ratios for each process parameter investigated on residual stress using the larger is better option.The residual stress increased to a maximum with increasing voltage, and travel speed and decreased to 25 V and 350 mm min −1 respectively.An increase in current and gas flow decreased residual stress from compressive stress towards zero residual stress.The increased current caused an increase in heat input whilst an increase in gas flow caused the weld bead to solidify more quickly, generating higher levels of residual stresses due to the thermal contraction mismatch between the deposited material and the subsequent material layers.The increased heat input might have promoted uniform grain growth, better fusion and bonding lowering tensile residual stresses towards zero [82].
The ANOVA results (table 9) indicate that travel speed had the most substantial contribution (52.37%) on residual stress of deposited layers.The results further show that gas flow, travel speed and current significantly contribute to the residual stress with a p-value less than 0.05 except for voltage having a p-value higher than 0.05.The optimal process parameters towards residual stress were found at 23 V, 100 A, 300 mm min −1 and 20 L min −1 of the experiment No. 4.

Microhardness
Figure 6 shows the AISI 4130 samples deposited at different test combinations based on the L9 OA.Twelve microhardness measurements were done on the deposited layers as shown in figure 7.These hardness results were considered for further analysis in the study.
Figure 8 indicates the SN-ratio analysis of microhardness.The results show that hardness elevated as voltage rise up to 23 V, then decreased with a rise in voltage (25 V).The increase in voltage to 23 V increased heat input leading to improved fusion and bonding between the deposited layers.Research studies have indicated a direct proportionality between voltage and penetration in welding processes [83].This relationship implies that an increase in voltage would result in greater penetration, which in turn, would lead to an increase in the hardness  of the deposited material.As the voltage increased to 25 V there is a decrease in hardness.Cheepu et al [84] discovered a microstructure with a coarser and less uniform arrangement of grains due to excessive heat input, in comparison to the finer grain structure caused by moderate heat input.This observation clearly supports the theory that excessive heat input produces materials with lower hardness when compared to those created under moderate heat input.Excessive high heat input was identified as the main factor leading to a lack of fusion pores, which contributes to the lowering of hardness [24].
The findings suggested that the hardness increased with an increase in welding current and travel speed.The hardness strength lowered with increasing gas flow from 10 l min −1 to 20 l/min.The optimal result of hardness SN-ratio, based on 'larger is better', was found at the combined setting of voltage-23 V. current-140 A, travel speed-350 mm min −1 and gas flow-10 l min −1 .
The ANOVA determines the relative importance of the investigated parameters.Table 10 shows the ANOVA results.The ANOVA indicated that voltage, current, travel speed and gas flow relatively influenced and interacted as assigned in the orthogonal array.Notably, voltage made the most significant contribution, accounting for approximately 59.97% of the observed effects.The statistical analysis reveals that the p-value for both voltage and travel speed is less than 0.05, indicating a statistically significant effect.Furthermore, the F-value is higher than 3.55, which exceeds the nominal F-table value of 3.55 as established by Pratiwi et al [56].This suggests a strong statistical significance for voltage and travel speed, per their findings.On the other hand, the contributions of current (1.12%) and gas flows (7.37%) to the hardness strength of the deposited layers were found to be not statistically significant, as their p-values were greater than 0.05.

Density analysis
The relative density is the density ratio of the deposited layer to the substrate.Optimising WAAM parameters to achieve a high relative density ensures that the deposited material is thoroughly consolidated.This optimisation minimises the presence of voids or porosity, leading to a more uniform and structurally sound product.This is crucial for the remanufacturing process where the material's strength and integrity are paramount.The analysis of the four parameters for density, in terms of plots of main effect for means and SN-ratio for each interaction, is shown in figure 9.The optimal result of relative density SN-ratio, based on the 'larger is better', could be obtained at the overall setting of voltage at 21 V, current at 140 A, travel speed at 300 mm min −1 and gas flow at 10 l min −1 .
Figure 9 indicates that the relative density of the deposited AISI 4130 decreases with increasing voltage and gas flow.Increasing voltage results in higher energy input, leading to increased melting of the feedstock material.When energy input and gas flow are increased during the welding process, it can lead to the formation of gas pockets within the material as it solidifies.These trapped gas pockets can contribute to a decrease in the material's density [24].Relative density reduced to the lowest value with rising current up to 120 A and increases up to 140 A. as the travel speed increases, the relative density of the material also increases, reaching a peak at a speed of 300 mm min −1 .However, when the travel speed further increases to 350 mm min −1 , the relative density starts to decrease.
ANOVA checked the relative importance of the parameters indicated in table 11.The ANOVA shows that all the parameters (voltage, current, travel speed and gas flow) significantly influenced the relative density of the deposited layers tested at 95% confidence level, as indicated by p-values less than 0.05.Voltage had the maximum significant contribution (31.14%), followed by travel speed (30.42%), gas flow (25.97%), and lastly current (9.35%) to the relative density.

Multiple response optimisation
The collective optimisation of WAAM parameters for residual stress, hardness, and relative density ensures that the 3D-printed components meet high performance criteria.The results provide insights into the interactions between different parameters, guiding the remanufacturing process for consistent and reliable results.
GRA and PCA were used to perform multi-objective optimisation, and the findings were validated using confirmatory experiments.This guarantees a solid, data-driven decision-making process for difficult   7. The responses are normalised using equation (3), resulting in a baseline sequence of responses ranging between 0 and 1.After pre-processing the data using equations (3) and (4), the reference and deviation sequence are obtained in table 12.
The eigenvalues, denoted as , j were estimated using PCA in the statistical software Minitab LLC ® 2023 with the aid of the governing equations ( 7)- (10).The outcomes are presented in table 13 together with the scree plot in figure 10.Based on the PCA, voltage was given the highest significance with a weight of 0.786, indicating its substantial influence.Following voltage, relative density was assigned a weight of 0.169, signifying its secondary importance.Residual stress, on the other hand, holds the least weight of 0.045, reflecting its lesser impact in this context.The values of GRC, GRG and the SN-ratios (using the 'larger is better') represented in table 14, were obtained through equations ( 5), ( 6) and (2) respectively.The multiple response optimisation using GRG, the highest GRG was obtained at sample no. 5 characterised by a voltage of 23 V, current of 120 A, travel speed of 350 mm min −1 and 10 L min −1 .Higher GRG signifies a high association with the optimal parameters [69].The study of the WAAM four parameters related to GRG, in terms of the main effect plots for means and SNratios, indicated in figure 11.The finest result of GRG SN-ratio, based on the 'larger is better' option, was found at the combined level of a voltage of 23 V, current of 100 A, travel speed of 350 mm min −1 and gas flow of 10 L min −1 .The GRG increased with voltage up to 23 V and decreased with voltage to 25 V.An increase in both current and gas flow has been observed to cause a decrease in the GRG.Conversely, the GRG tends to increase with a rise in travel speed.
Table 15 shows the summary of means of GRG for individual level and factor computed from the GRG in table 14.The table also depicts the optimal parameter and level for the multiple optimisation process.
ANOVA was conducted on the GRG against the four-process parameter indicated in table 16.Three parameters travel speed, voltage, and gas flow significantly influenced GRG tested at 95% certainty level as indicated by the p-values less than 0.05 and the f-value significantly higher.The current was considered insignificant towards GRG as the p-value is greater than 0.05 and the f-value is too low (0.62).The parameter  with the most influence towards GRG tested by the ANOVA (table 16) was voltage with 31.61% contribution followed by travel speed at 29.28%, gas flow at 13.51% and lastly current with 1.64% which was found to be insignificant.

Probability plots
For the responses indicated in tables 7 and 14, probability graphs were used to assess the spreading of experimental and calculated data [57].The Anderson Darling (ADT) test is a technique employed to find outliers from normality.This tool validates the normality assumption [85].Figure 12 indicated that the test data for all outcomes lie close to the fitted line, except for a single value in the relative density response data.Because the ADT statistical values for hardness, residual stress, and GRG are relatively low, and the p-values of the test are higher than 0.05, this provides proof that the data presented adheres to a normal distribution.A high ADT value for relative density, along with a high p-value, suggests that the distribution of this response is not normal.

Confirmatory experiment
Confirmation experiments on residual stress, hardness and density were conducted using optimal conditions identified by the GRA found at voltage level 2 (V 2 ), current level 1 (C 1 ), travel speed level 3 (T s3 ) and gas flow level 1 (f r1 ).This analysis was done to evaluate and verify any improvement of physical and mechanical characteristics.The predicted GRG value (0.88) in table 17 was computed using the equation (11) [56,57].Where m g refers to the mean GRG, i g refers to the mean GRG at an optimal level of i th factor, and n are the factors that have a significant impact on quality responses.The confirmatory results reveal that the highest values were observed for residual stress and hardness.Conversely, the relative density demonstrated lower values.The GRG improved by 0.3630 (69.99%), which indicates a satisfactory agreement between predicted and experimental values.Based upon the GRA, a significant improvement of residual stress by 239.01%, and hardness by 2.93% was achieved whilst a reduction of relative density by 0.64% was also achieved.The improved GRA results associated with the deposition operating parameters, particularly at larger GRG values, provide strong support for the effectiveness of using the Taguchi technique in conjunction with GRA.This combination is aimed at enhancing the quality of WAAM products.

Microstructure evaluation
Figure 13 shows the microstructure of the sample printed using optimal conditions.The optical images show that the material exhibited mainly ferrite, a ¢ ¢, (light zones) and pearlite, 'P' (dark zones) phases in the deposited regions indicated in figures 13(a), and (b) [86].The ferrite phase found in the deposited regions may have caused the low microhardness values.The ferrite phase typically exhibits lower microhardness due to its relatively soft and ductile nature compared to other phases [86,87].The alpha (α) phase (body-centred cubic (BCC) crystal structure) is recognise to have a negative impact on hardness, while delta (δ) ferrite (face-centred cubic (FCC) crystal structure) has a less pronounced impact [88].Equiaxed grains extending from the interface region into the heat-affected zone (HAZ), are observed in figures 13(c) and (d), highlighting the microstructural changes associated with the WAAM deposition process.This phenomenon is a consequence of thermal gradients and solidification patterns.Figure 13(d) depicts largely the HAZ and showcases the coexistence of three distinct microstructure orientations.The lamellar columnar structure found in the substrate region in figure 13(e), characterised by alternating pearlite and carbides, significantly contributes to the high hardness values observed in this substrate layer.
Figures 13(b), and (c) show pore formation within the deposited WAAM region.The presence of pores offers a significant justification of the low relative density recorded being less than 100% in this specific region (table 7).
Chromium as well as other ferrite-forming elements (W, V, and Nb) influences the mechanical behaviour of steel [89].The presence of chromium in an alloy has a relevant impact on the production of microscopic particles within the material, profoundly changing its behaviour during deformation [90].This impact is caused by chromium's influence on particle balance, composition, and interaction with other elements in the alloy.Consequently, the existence of chromium has a notable influence on the mechanical behaviour and formability of the alloy.AISI 4130 steels' phases can be predicted using Cr-Ni equivalents and the Schaeffler diagram [91].The Cr-Ni equivalents, in wt% are determined using Klueh and Maziasz's equations ( 12) and (13).In steels where Ni is absent, and Cr eq < 10, delta-ferrite does not form.Such steels include only the martensite.For 10 Cr eq 12, the steel comprises both martensite and delta-ferrite; and for Cr eq > 12, only delta-ferrite will form [92]. Using equation ( 12), the calculated Ni eq is 9.55 while equation (13) gives the Creq values as 1.75.According to the Schaeffler diagram depicted in figure 14, when AISI 4130 steel is welded using AISI 4130 filler material, it falls within the martensite phase (M) region.During the WAAM deposition process, the material undergoes varying thermal cycles.This variation inhibits rapid cooling of the material, which in turn, prevents the formation of martensite.The intricate thermal cyclic loading inherent in the WAAM process results in a slow cooling rate for the substrate, interface, and deposited regions.This slow cooling fosters the formation of pearlite and ferrite colonies, a phenomenon that can be elucidated by the Continuous Cooling Transformation (CCT) diagrams [93].Ferrite is a BCC phase that is relatively soft and ductile, whereas pearlite is a lamellar structure made up of alternating layers of ferrite and cementite.The cementite contributes to overall strength and hardness [94].

Ni
The microstructural examination of the deposited AISI 4130 steel reveals a predominant pearlite-ferrite phase, as illustrated in figure 15(a), characterised by its lamellar morphology.This finding aligns with the slow cooling rates of the WAAM deposition process, which foster the formation of pearlite and ferrite colonies.The slow cooling of the deposited regions leads to the transition from austenite to pearlite (comprising ferrite and cementite), resulting in the emergence of ferrite and coarse pearlite phases.The slow cooling of steels from the temperature above and below the material's Ac3 is well reported in literature with the aid of the CCT diagrams [93].In this case, the coarse pearlite-ferrite phases are less dense compared to the interface region in figure 15(b).This less dense characteristic suggests lower microhardness in the deposited region than that of the interface region.The interface region initially experienced faster cooling than the deposited region since the substrate was not preheated prior to deposition process.This faster cooling than the deposited region may have resulted in the formation of finer pearlite-ferrite colonies as compared to the deposited region which comprised of ferrite and coarse pearlite colonies [93].The substrate microstructure of AISI 4130 steel in figure 15(c) shows equiaxed ferrite and lamellar pearlite phases.
Figure 16 shows the hardness of the substrate, interface and deposited regions.The substrate region's hardness of 164.5 ± 11.3 HV 0.3 lowered in comparison to control sample's hardness of 170.2 ± 6.9 HV 0.3 by 3.35%.The substrate region might have undergone an in-process heat treatment, such as tempering.This may have resulted in a slightly lower microhardness compared to the raw substrate plate [95].
As depicted in figure 16, the interface region exhibited the highest hardness of 180.3 ± 12.2 HV 0.3 .This can be attributed to the densely packed ferritic and pearlitic phases as shown in figure 15(b).The hardness of the deposited regions, which was found to be 170.5 ± 6.9 HV 0.3 , is nearly identical to that of the substrate control sample.The hardness values of the deposited region are lower than those of the interface region.This is due to the widely spaced ferritic and pearlitic phases observed in figure 15(a).According to Shim et al [96], the deposited regions tend to have lower hardness values due to the presence of large amounts of ferrite and potentially preserved austenite phases in the microstructure.These phases play a significant role in determining the hardness of the materials.

Conclusion
This study used Taguchi's GRA technique for multiple-response optimisation of WAAM process operating parameters.Four WAAM process operating parameters such as voltage, current, travel speed and gas flow at three levels were considered to investigate the microstructure and mechanical characteristics of WAAMed structures.From this study, the following conclusions were made:   2. The multi-response optimisation results showed that voltage had the highest significance, contributing 31.61% compared to travel speed (29.28%) and gas flow (13.51%).However, the current did not show any significant contribution.The probability plots indicated that all response data followed normal distribution except the relative density, which had an outlier data point suggesting an unusually high presence of pores in the sample no. 8.

3.
Comparison of the mechanical behaviour of the substrate to the optimum confirmatory observations show that residual stress increased from −49 ± 8 MPa to 25 ± 74 MPa, microhardness increased from 170.2 ± 6.2 HV 0.3 to 171.4 ± 12.2 HV 0.3 and density decreased from 7.835 g cm −3 to 7.695 g cm −3 .
4. Microstructural analysis revealed the presence of dense ferrite and pearlite phases in the interface region than all the deposited region, which were associated with higher hardness values, whereas the presence of pores in the deposited regions explains the low relative densities in the deposited regions.
5. The results indicated that hardness of the interface region is higher than the deposited region and substrate region.
The findings of this study provide a rich knowledge base for stakeholders in the industrial remanufacturing processes for high-quality structural and functional components.

Figure 6 .
Figure 6.Microhardness samples for AISI 4130 and the substrate.

Figure 7 .
Figure 7. Summary of hardness test sample and positions of measurement (a) deposited layers, (b) interface layer, (c) substrate layers.

Figure 10 .
Figure 10.Scree plot of hardness, relative density and residual stress.

1 .
The ANOVA reveals that optimum conditions residual stress-25 ± 74 MPa, microhardness-171.4 ± 12.2 HV 0.3 and relative density-98.207%corresponding to voltage of 23 V, current of 100 A, travel speed of 350 mm min −1 and gas flow of 10 l min −1 .

Figure 16 .
Figure 16.Hardness in the WAAM substrate, interface and deposited regions.

Table 1 .
The elemental makeup of AISI 4130 material used.

Table 3 .
The process parameters and the levels considered for the WAAM deposition.

Table 4 .
Summary of constant deposition parameters.

Table 5 .
Residual stress material assessing conditions.

Table 6 .
Design of experiment based on taguchi's L9 OA.

Table 7 .
Design of experiment and responses.Design of experiments: taguchi array L 9 (3 4 )

Table 8 .
SN-ratio for the responses.

Table 9 .
Residual stress analysis of variance for means.

Table 10 .
Hardness analysis of variance for means of AISI 4130.

Table 11 .
Analysis of variance for means of relative density.

Table 12 .
Normalised responses and deviation sequence.

Table 13 .
Eigen and eigenvector analysis of the correlation matrix.

Table 15 .
Response table of GRG.

Table 16 .
Analysis of variance for means of GRG.

Table 17 .
Evaluation summary of confirmatory experiments.