Optimize the corrosion behaviour and mechanical properties of AISI 316 stainless steel under heat treatment and previous cold working

: Improving corrosion resistance in alloys made of stainless steel is an important innovation on the petroleum trade. The e ﬀ ect of heat treatments (HT) and cold working on the corrosion behaviour, surface hardness, and micro-structure of 316 stainless steel was investigated experimentally. The corrosion environment is seawater and crude oil. The corrosion rates (CRs) were obtained using the mean loss of weight approach, which was then optimised using the Taguchi method. The specimens used in this study are made of 316 stainless steel rod, which is ﬁ rst annealed to obtain the qualities of the raw material before being put through a tensile test to assess the mechanical characteristics of the metal. After cold working, the hardness test, the corrosion test utilising the lost weight method, and the microstructure test are all carried out. By performing these tests, the metal show excellent mechanical properties such as yield stress, tensile stress, and hardness; in the corrosion test, the raw metal show higher resistance in both seawater and crude oil, while in cold working and HT with cold working, samples show higher corrosion The HT samples had the lowest corrosion resistance as the cold working percentage increased. In this work, the input parameters such as ultimate corrosion media, HT and cold work (CW) are optimised utilising a multiple objective optimisation approach that uses weighted grey relational analysis. Two objectives, that are CR and Hardness (H), are simultaneously optimised. We suggested a quantitative approach to establish the weight factors of various responses for grey relational analysis called weighted grey relational analysis. The optimum input parameters were determined using weighted grey relational analysis, and the outcomes showed that HT is the most relevant parameter. Cold working has been observed in association with stress-related twinning and austenite phase deformation, resulting in fast grain splitting and the production of a microstructure that resembles a ribbon composed of auste-nite and ferrite.


Introduction
Austenitic stainless steels (SS) are commonly used in many instances of industrial and construction industries due to their passivation feature and ability to tolerate environmental deterioration.Their outstanding mechanical performance and corrosion resistance at both high and low temperatures are largely responsible for this wide range of applications.However, intergranular corrosion poses a serious threat to them [1][2][3].The majority of biomedical applications for austenitic stainless steels include implants, fittings, medicines, and surgical instruments.The biocompatibility, low cost, ease of manufacture, exceptionally high mechanical strength, and resistance to corrosion of 316L stainless steel contribute to its widespread application [4][5][6].
Heat treatment (HT) cannot harden 316 stainless steel.Rapid cooling can be used to perform annealing or solution treatment after heating to 1,010-1,120°C [7].
Rather than being heated, such stainless steel can be toughened by cold working.Intergranular corrosion can result from heat treating stainless steel as well as chromium and C depletion at grain boundaries [8].However, by limiting the diffusion channel for carbon together with chromium in the fine granules, stainless steel grain refinement could significantly enhance corrosion resistance.Because of this, there is less chance of chromium depletion at grain boundaries, which would encourage the development of an identical passive oxide coating on the surface and prevent 316L stainless steel from dissolving [8][9][10].
The 316L steel has a very low strength.HTs cannot strengthen these steels since they do not go through any phase transformations.However, they may be made stronger by reducing the grain size to submicrometer levels [11,12].
Plastic deformation has the ability to alter the characteristics of a material by increasing the dislocation density, decreasing the percentage of coincidence site lattice barriers, introducing deformation bands, and even changing the phase [13,14].The corrosion and cracking resistance of austenitic stainless steels in settings would be impacted by these modifications to the material and mechanical characteristics.Austenitic stainless steel achieves a passive state with high stability and strong corrosion resistance when chromium is present in some quantity [15,16].It has been found that increasing the quantity of cold-working produced tensile and yield strength improvements for austenitic stainless steels.This can be done as a percentage, such as 10, 20%, and so on [17][18][19].
316 austenitic stainless steel has been the subject of several studies on corrosion rates (CRs); however, there has been little study on how pre-cold working and annealing temperature affect 316 CRs.These impacts at various levels and in diverse corrosive settings are made clear by the current investigation.In addition, each environment's mechanical properties and microstructure were examined, and the most efficient elements of the process were investigated.

Experimental work 2.1 Raw materials and HT
Three 316 stainless steel rods were selected, each measuring 1.87 m in length and 10 mm in diameter.As indicated in Table 1, a PECTROTEST TXC25 spectrometer was utilised to ascertain the chemical composition for a double check.
According to the ultimate corrosion media (CM), either saltwater or crude oil, the samples were split into two primary groups, each of which was further separated into three parts (without HT, annealing at 1,160°C and quenching with water, and annealing at 1,160°C and quenching with oil).Three parts of each specimen group are divided into three smaller groups, each of which is subjected to 10, 20, and 30% cold working, respectively, according to ASTM E8M [19] utilising a universal tensile machine.Each group from 20 groups contains three specimens according to the cold working and immersion media environments as shown in Table 2.
The hardness test was performed in accordance with ASTM E384 [20].After machining the metal as shown in Figure 1, in the first two cases, the specimens were kept unprocessed simply by being placed in seawater and crude oil, while the other examples were subjected to 10, 20, and 30% cold working, respectively, and annealing.The process of annealing was used to create 316 stainless steel.After being submerged at 1,160°C, Type 316 is quenched in either water or air.

Aggressive environments
The impact of annealing and pre-cold on the corrosion resistance of the 316 stainless steel was examined using two corrosive media.The corrosive environment consists of seawater and crude oil.Table 3 lists the characteristics of seawater, while Table 4 lists the characteristics of the crude oil utilised in the experiment.
The majority of the oil produced in Iraq is exported via the Arabian Gulf; therefore, sea water measurements from the Arabian Gulf must be taken.This is because there are numerous export platforms (including Khor al-Zubayr in the Basra Governorate and the port of Umm Qasr).

Corrosion test
The corrosion of the samples was examined using the lossweight method in conformity with ASTM G31 [21], a standard.The specimens were cleaned correctly, with the use of chemical and mechanical cleaners to eliminate any contaminants and any oxidisation layer (if present).To avoid the inaccurate and deceptive findings of short-period tests, a certain container was used, and the appropriate test period was picked.The tests were set up as cases submerged in settings including seawater and crude oil.
The container was shut tightly and kept for a week.To determine the weight lost, the specimens were taken out and cleaned every week.The CR was calculated using the following equation [22]: where ρ is the metal density in grams per cubic centimetre, A is the specimen's area in square centimetres (cm 2 ), t is the exposure duration in hours, m is weight loss in milligrams, and CR is the CR in millimetres per year (MPY).In Table 5, a portion of the weight loss calculations is shown.

Analysis and design of experimental data
The total number of the input parameters, their involvement in research, as well as their levels, influence the design of experiments (DOE) that will be utilised to perform the experiments.The L20 array is created in the current experiment using the Taguchi approach, and subsequent processing is progressed correspondingly.Furthermore, the strength of the chosen design is guaranteed.A powerful   design is one that noise or other uncontrolled events have the least impact possible on the response variables.The next sections provide specifics on the experiment and analytic method utilised in the current study for multi-response optimisation.

Optimisation parameters
The input parameters used in the current analysis are ultimate CM, HT, and cold work (CW).The parameters, units, and levels as indicated in Table 6.

Response variables
CR and hardness (H) were examined as two response variables.CR minimisation and Hardness (H) maximisation were the goals.Table 7 displays the response variables along with the abbreviations and units.

Experimental data array
With fewer experiments, the experimental data array may be used to examine how a system's or process's input parameters impact the response variables.The Taguchi technique is used to do this.The number of input parameters utilised in the experiment, together with their levels, determine the experimental data array.The L20 data array was used for the current investigation, as indicated in Table 8.
The data analysis may be carried out using a standard statistical programme like Minitab; in the current study, the tool used for this was Minitab18.

Signal-to-noise ratio calculation and analysis
The signal-to-noise (S/N) ratio for each factor level combination is calculated.Since the goal was to reduce the CR, the smaller-is-better criteria was applied, and Eq. ( 2) is utilised to obtain the S/N ratio.Similarly, as the goal was to maximise hardness (H), the larger-is-better criteria was applied, and the S/N ratio is determined using Eq. ( 3).
The following is the S/N ratio for the smaller-the-better characteristic: The S/N ratio for the larger-the-better feature is also written as follows: where y i stands for the observed response values for each run, and n stands for the experimental sets.Table 9 shows the S/N ratio for CR and H depending on Eqs. ( 2) and (3).

Grey-Taguchi optimisation technique
We have used a multi-objective optimisation approach since two response parameters are thought to be optimised concurrently.In the current experimental study, we are using grey and Taguchi to optimise the process parameters, and depending on their grade, Taguchi approach is then used to optimise them.

Grey relational technique
Grey relational methodology is a technique for combining many quality factors into one, allowing for the implementation of Taguchi single objective optimisation techniques and multiobjective quality parameter conversions.To do this, grey relational analysis (GRA) is used to determine the grey relational grade (GRG).Less data and multifactor analysis are its defining qualities, and these two traits might outweigh the drawbacks of statistical regression analysis.This single goal optimisation strategy makes advantage of the performance feature known as GRG.A step-by-step description of the GRA process and its outcome is presented [22,23].

GRA methodology
The following stages are the order in which GRA is conducted.

Normalising the S/N ratios
The study's initial data, which is used to convert the original sequence into an identical sequence, is prepared by normalising the S/N ratio in Taguchi-based GRA.The information in the 0-1 range of values, often known as the "grey relational generation," are transformed by normalising the S/N ratio.In this study, the smaller the better criterion for CR and the larger the better criterion for normalisation of Hardness (H) are employed, respectively, utilising the equations found in Eqs. ( 4) and ( 5) [23].Smaller the better: Larger the better: where max z i (p) is the largest value of z i (p) for the pth respond, min z i (p) is the lowest value of the S/N ratio, u i (p) is the outcome of grey relational generation.The normalised data are presented in Table 10.

Determining the deviation sequence
The difference between the reference sequence y o (p) and the comparability sequence y i (p), following normalisation, is represented by the deviation sequence Δ oi [23].Eq. ( 6) is used to determine it as follows: The deviation sequence is presented in Table 11.

Grey relational coefficient
The grey relational coefficient (GRC) is a measure of the connection between the ideal (optimal) and actual normalised S/N ratio for all sequences.The two sequences have a grey relational coefficient of 1 if they agree at every points [23,24].Eq. ( 7) may be used to represent the grey relational coefficient GRC.
where φ is the distinctive coefficient, which has a value between 0 and 1, but is often considered to be 0.5.The value of Δ oi are Δ min (smallest value) and Δ max (biggest value).Table 12 presents the grey relational coefficient (GRC) (Tables 13 and 14).

Weight factor calculations
For an actual engineering challenge, different approaches have varying degrees of relevance.The GRG varies significantly when different responses have differing weights, indicating that weight considerations are crucial for attaining the best outcomes.In most cases, researchers employ equal weight to calculate the GRG of numerous replies [25,26] or use a weight to subjectively emphasise or de-emphasize the aim.
To provide acceptable values to various responses under optimisation, a sensible criterion for weight factor computation can be utilised [27].To estimate the weight factors, this technique depends on how much the alterations in the parameters have an impact on the responds.Using Eqs. ( 8) and ( 9) correspondingly, the grey relational coefficient ranges and weight factors are calculated.The weighting variables for each response are also shown in the last row of Table 15.
j = 1,2…, p, i = 1,2…, m, and k = 1,2…, l. w, weights, m, responses, p, parameters, l, experimental levels, R, ranges of grey relational coefficients, K, average grey relational coefficients for each parameter at each level of each response, and m, responses.There are specified weighting parameters for the responses, and the calculation for the GRG is provided by where GRC CR stands for the grey relational coefficient of CR and GRC H is that for the hardness.Table 16 lists the values for two responses grey relationship grades.

GRG and rank determination
The GRG provides the foundation for the overall assessment of the many performance aspects.The highest rank is given to the grade with the highest value.The computation of the GRG, which acts as a basis for the overall assessment of the multiple-performance feature, is the next stage of GRA and is performed using Eq. ( 11) [23,28]: where w p is the weighted factor for each grey relational coefficient and n is the total number of response variables.For each of the response variables, the total weighting factors should equal 1.

Microhardness
Results for the microhardness of the 316 austenitic stainless steels received and after HT and cold working are shown in Figure 2 for the groups in Table 2.At room temperature, the microhardness values were determined.As a result, cold work was the sole factor in all of the microhardness variations.The percentage of cold working affects the work-hardening capacity [22,29].
When the hardness values are observed, they increase directly with the value of cold operation and reach their maximum when the cold working rate is 30%, as seen in groups 5 and 8 with hardness value 229 and 231, respectively.316 stainless steel hardness usually reduced relatively considerably when annealing up to 1,120°C, when it dropped significantly due to recrystallisation and the formation of equated grains [30].

Microstructure
The chemical, physical, and mechanical properties of a specimen can be significantly influenced by its microstructure.For metallographic investigation, samples were taken from the experimental AISI 316 stainless steel.Each of the samples were done by grinding with emery paper measuring 250, 600, 1,000, and 1,200, then polishing with 3 µm diamond, and etching for 50 s in a certified and tested hood with 10 mL HNO 3 , 10 mL acetic acid, 15 mL HCl, and 2-5 drops glycerin to define the microstructure according to ASTM E 407 [28].As shown in Figure 3a, the microstructure of the specimen as it was received was mostly made up of equiaxed austenite grains, with just a little amount of annealing twins.Parallel lamellar structures and many dislocations were produced by cold working.The lamellar structure is extended mostly in the cold working direction [31].When demonstrated in the insets in Figure 3b and when cold working increases, the deformation and lamellar structure become increasingly apparent as a difference between Figure 3c and d.
Furthermore, as cross-sectional metallographic structures indicated, distortion morphologies became thinner and more uniform.Cold working causes considerable deformation and the production of a lamella rough grain, as shown in Figure 3d.The density of dislocations increases over time with cold working, which can improve micro-hardness.

Surfaces analysis
Photography of high resolution to the surface morphology of the 316 stainless steel samples was examined using a NanoSEM 8  Haider Mahdi Lieth et al. the differences in the morphology of polished surfaces for steel that has had 10, 20, and 30% cold working in addition to HT. Examining the micro scanner images makes it clear how coldworking in various proportions has affected the internal structure and the HT process.As the metal is cold worked more quickly, it deforms more and changes its mechanical properties, especially its ability to resist, i.e., corrosion.The internal structure of the metal is more significantly affected by HT, and this is seen by the metal's ability to resist corrosion.

CRs
It was previously explained how the sample surfaces experienced a mass difference due to the hostile environment.Time affects how much each specimen's mass changes during a    Corrosion behaviour and mechanical properties of AISI 316 stainless steel  11 corrosion test.Calculations of weight loss offer an accurate assessment of CRs.Using Eq. ( 1), the weights before and after exposure to the corrosive liquid were computed.
The CR for different cold working percentage with and without HT for oil and water as aggressive environments are shown in Figures 5-7.According's to Figure 5, where it is clear that the received sample experienced the greatest weight loss, this graph investigates the impact of cold working alone, without consideration of HT conditions.

GRG analysis using Taguchi
The main effects analysis is utilised to examine the influence and impacts of input parameters on the GRG, as shown in Figure 8.It is clear from Figure 8 that as CM changed from sea water to crude oil, the S/N ratio decreases [33].Also the S/N ratio decreases when the HT is applied.But in the case of cold work, the S/N ratio also decreases and then increases at 30%. Figure 8 shows how to estimate CM at level 1, HT at level 1, and cold work at level 4, which means CM1-HT1-CW4 will concurrently give the optimal output characteristics (CR and surface hardness).

Signal-to-noise ratio response table
Based on the rank value shown in Table 17, the response table shows that the control factors affecting the response variable (GRG) follow the prescribed sequence in descending order: HT > CM > CW.

Analysis of mean
The optimal amount for the process parameters was calculated using the analysis of means (ANOM).The collection of input parameters is presented in Figure 9's ANOM graph for response variable optimisation (GRG).We must go with the higher values of input parameters under the larger, better criteria used for optimisation of GRG since we are improving the process parameters under multi-objective optimisation such that sea water may be used as the corrosive medium, without HT, and with 30% cold work, which is the ideal combination of input parameters.
The response table indicates that the control variables that affect the response variables follow a decreasing order HT > CW > CM based on the rank value presented in Table 18.

Variance analysis for GRG
The P value for the HT parameters is less than 0.05, which is considered significant (lower probabilities give greater proof against the null hypothesis), and its percentage of contribution is also high, being 38.39%.This indicates that HT is the most important variable, followed by CW and CM, which have 16.87 and 8.64% influence, respectively, in the GRG ANOVA table.The F-value, a test statistic used to establish if a term is related to a response, also demonstrates that the HT is the aspect that has the greatest impact on the GRG response (Table 19).

Conclusions
In this work, two techniques were used to strengthen AISI 316 stainless steel.In the first instance, the specimens were immersed in two hostile environments (crude oil and seawater) after having gone through many cold working actions on the tensile machine.In the second instance, the specimens were sliced before being heated.These samples were afterward subjected to the same circumstances as the initial case.The most important conclusions are as follows: 1. Cold action has been accompanied by stress-related twinning and austenite phase deformation, which resulted in fast grain splitting and the creation of an austenite and ferrite ribbon-type microstructure.2. The hardness values increase in direct proportion to the intensity of cold working.Since a decline in hardness values was noticeable as a result of the microstructure changes that took place throughout the annealing process, it is also possible to see how the annealing HT affected hardness values.3. The multi-objective optimisation technique we suggested reveals that HT is the element with the greatest influence, followed by CW and CM. 4. The order of the descending importance of the control variables impacting the response variable (GRG) is as stated: CW > HT > CM. 5.The combined set of the resultant parameters (CR and surface hardness) will be the best possible when CM are at level 1, HT is at level 1, and cold work is at level 4, which provided by CM1-HT1-CW4.

Figure 2 :Figure 3 :Figure 4 :
Figure 2: Hardness of 316 stainless steel of the as-received and after heat treatment, cold working/aggressive environments classified by groups.

Figure 5 :
Figure 5: Corrosion rate for different cold working without heat treatment.

Figure 7 :
Figure 7: Corrosion rate for different cold working quenching in water.

Figure 6 :
Figure 6: Corrosion rate for different cold working quenching in oil.

Figure 8 :
Figure 8: S/N ratio for GRG main influence plot.

Figure 9 :
Figure 9: Mean of means of GRG main effect plot.

Table 1 :
Lists the chemical constitution of 316 stainless steel in percentages

Table 2 :
Specimen classification based on cold working and immersion media

Table 3 :
Properties of sea water

Table 5 :
Weight loss for sea water samples Sample

Table 6 :
Parameters of input and their levels

Table 7 :
Response variables

Table 8 :
Experimental data array

Table 9 :
S/N ratio for CR and H with their calculated values

Table 10 :
S/N ratio normalised value for CR and H

Table 11 :
Deviation sequence for CR and H

Table 12 :
GRC for CR and H

Table 13 :
Mean value for the CR parameter at each level

Table 14 :
Mean value for the H parameter at each level

Table 15 :
Grey relational coefficients weighing factors

Table 16 :
Grey relational grade and rank

Table 17 :
S/N ratio response table (larger is better)

Table 18 :
Response table for means