Scholarly article on topic 'Fuzzy Logic Optimization of Weld Properties for SAW Using Silica Based Agglomerated Flux'

Fuzzy Logic Optimization of Weld Properties for SAW Using Silica Based Agglomerated Flux Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Computer Science
OECD Field of science
Keywords
{Anova / "S/N Ratio" / "SiO2 Based flux" / "Vickers Hardness" / "Impact Strength"}

Abstract of research paper on Materials engineering, author of scientific article — Aditya Kumar, Sachin Maheshwari, Satish Kumar Sharma

Abstract In submerged arc welding flux composition plays a deciding role for good quality weld. In a SiO2 based flux NiO, MnO and MgO are added in three different proportions. By measuring the Vickers hardness and impact strength of the weld, the effect of each flux alloying element was investigated. For multi-objective optimization, a Fuzzy Logic model is proposed and optimal levels of NiO,MnO and MgO were obtained using a single multi-response performance index (MRPI).

Academic research paper on topic "Fuzzy Logic Optimization of Weld Properties for SAW Using Silica Based Agglomerated Flux"

(8)

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Procedia Computer Science 57 (2015) 1140 - 1148

3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015)

Fuzzy Logic Optimization of Weld Properties for SAW Using Silica Based Agglomerated Flux

Aditya Kumara* , Sachin Maheshwarib ,Satish Kumar Sharmac

aAssistant Professor, Deptt of MPA Engg, Netaji Subhas Institute of Technology, Dwarka, New Delhi ,INDIA bProfessor ,Deptt of MPA Engg, Netaji Subhas Institute of Technology, Dwarka, New Delhi INDIA bTRF, Deptt of MPA Engg, Netaji Subhas Institute of Technology, Dwarka, New Delhi INDIA

Abstract

In submerged arc welding flux composition plays a deciding role for good quality weld. In a SiO2 based flux NiO, MnO and MgO are added in three different proportions. By measuring the Vickers hardness and impact strength of the weld, the effect of each flux alloying element was investigated. For multi-objective optimization, a Fuzzy Logic model is proposed and optimal levels of NiO ,MnO and MgO were obtained using a single multi-response performance index (MRPI).

©2015TheAuthors.PublishedbyElsevierB.V.Thisis an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-reviewunderresponsibilityof organizingcommitteeofthe3rd InternationalConferenceonRecentTrendsinComputing 2015 (ICRTC-2015)

Anova, S/N Ratio, SiO2 Based flux, Vickers Hardness, Impact Strength

1. Introduction

The quality of weld is a very important working aspect for the manufacturing and construction industries and heavy duty work. Because of high quality, Submerged Arc Welding (SAW) is one of the reliable metal joining processes for manufacturing industries. The popular areas of application are as ship building, joining of pipe, thick structural work, power system, nuclear and chemical industries, food processing as well as petroleum industries. A strong weld should be resistant to corrosion; hydrogen induced cracking as well as fatigue failure.

* Corresponding author. Tel.: 09968093691; fax: 011-25099022. E-mail address: aditya_rathihere@yahoo. com.

1877-0509 © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of organizing committee of the 3rd International Conference on Recent Trends in Computing 2015 (ICRTC-2015) doi: 10.1016/j.procs.2015.07.403

In SAW, quality of weld depends upon the selection of wire and flux composition along with welding process parameters. It necessitates the need for investigation of mechanical properties of weld for a particular application. Flux is a combination of elements in different composition mixed together to get the better weld properties. These flux dependent mechanical properties of weld mainly are yield and tensile strength, impact strength, elongation and hardness. For a particular application, an optimal combination of these properties is desired. Multiobjective optimization techniques can be very useful for this purpose. Many multi-objective optimization techniques are reported in literature like- GRA, PCA, Fuzzy logic and various intelligent computational techniques. Computational algorithms take large time to simulate whereas GRA need weightage to be defined for each response. In PCA, data loss is a big disadvantage. Fuzzy Logic takes an upper edge over these techniques because of its ability to take care of diversity in data. Fuzzy logic is a natural way of expressing uncertain information that is easy for computation. 2. Literature

As per literature survey, increase in the percentage of SiO2 in the flux increases the arc stability, better slag detachability with good quality weld penetration \ Composition of flux and electrode play an important role in the strength of welded joint 2 Welding of 1.25Cr-0.5Mo steel with SiO2 and TiO2 based flux increases the heat input, lowers the yield strength and decreases the % elongation 3. Manganese fluxes give lower residuals of sulphur where as calcium silicate fluxes were responsible for removal of phosphorus 4. CaO is found to be the most influence factor for controlling acicular ferrite content in the weld metal 5. Because of small grain size and high angle grain boundaries acicular ferrite is desired microstructure in weld for high impact strength and ductility. While using SiO2 flux in TIG welding high arc voltage is required 6. A mathematical programming optimization technique was used to optimize the welding flux composition which is responsible for weld metal properties as a function of welding flux level 7. A non-pre-emptive goal programming model was used for weld metal chemical composition optimization for welding flux ingredients and suggested that this technique will achieve optimum performance for the flux at minimum experimental efforts and cost 8. Fuzzy logic was used to optimize the submerged arc welding process parameters the result of experiment concluded that arc voltage, arc current, welding speed, electrode protrusion and pre heat temperature are optimized with consideration of performance characteristics such as dilution and deposition rate 9 Butt joint made up of AISI 904 L SASS by laser weld are investigated and optimized with the use of fuzzy logic and desirability approach for selected welding parameter 10, 18. Intelligent systems are used in the modern manufacturing units in order to ensure the defect free product and to increase the productivity. A fuzzy logic based system whose elements were determined by training an artificial neural network was employed on the gas metal arc welding to optimise the deposition rate of filler metal 11. Submerged arc welding characteristics such as hardness of metal zone is affected a lot by welding input parameter. This characteristic is required for high productivity and cost effectiveness. Combine effect of TiO2 nano particle along with welding parameter were optimized to develop a fuzzy logic model to predict the hardness of metal zone in submerged arc welding 12, 17. Pressure vessels had to with stand a high pressure across the wall of vessels. Manufacturing of this vessel is done with the help of submerged arc welding. Structural integrity for the performance of weld depends on many factors such as welding parameter. A multi objective optimization of welding parameter is performed on the pressure vessels by the use of fuzzy logic and desirability approach 13, 14, 15. Grey based fuzzy logic method was used to optimize the flux cored arc welding parameter and concluded that this method provides manufacturers an intelligent manufacturing system to facilitate the highest level of automation 16. Tungsten inert gas welding process parameters were optimized to predict the weld pool geometry with the shape of heat affected zone 18. Based on the study of above researcher, fuzzy logic modelling technique gave a conventional approach to model the flux constituent's composition for submerged arc welding. Same methodology is applied in this study to optimize the flux constituents.

3. Design of flux by using Taguchi Method

Taguchi methodology is a simple and systematic approach of experimental design that can reduce number of experiments to an optimal level. In present work taking SiO2 as the base of the flux three alloying elements NiO ,MnO and MgO were varied at three different level. Proper selection of orthogonal array is based upon degree of freedom. Degree of freedom for each factor is one less than level of a factor. So in present design, total degree of freedom for all three factors becomes 6. Since 2 degree of freedom is assigned for each three factor, and 2 degree of freedom is assigned to error term, so total degree of freedom becomes 8. Number of experiments can never be less than the total degree of freedom. Keeping all this in view, L9 orthogonal array was selected which has nine numbers of runs. Nine experiments were run for SiO2 based flux by varying alloying elements in different combination as per the design.

Table 1. Level of factors

Factors Code Level Level Level 3

NiO (Wt in gms) A 60 80 100

MnO (Wt in gms) B 50 70 90

MgO (Wt in gms) C 85 95 105

Table 2. Composition of SiO2 based flux

Constituents Al 2O3 SiO2 CaO CaF2 (Wt %) 18.45 28.0 27.15 7.5

3.1 Preparation of Flux

Based on the above analysis and selection of orthogonal array, 9 fluxes having SiO2 as base were prepared. The Main three constituents that vary in the flux were NiO, MnO and MgO , their level is shown in table 1. Other constituents of flux that were kept constant are shown in table 2. Total 2kg of flux was prepared for each combination by the agglomeration method. All the constituents were dry mixed first in a big container with ten steel balls of 10mm diameter for half an hour. Potassium silicate was heated up to 500C, sprayed into the mixture (1000ml per 500ml of water). Same mixture was then thoroughly mixed again for half an hour. This process was repeated again and again until the binder potassium silicate and other constituents have formed the granules. After that the prepared mixture was dried at room temperature for 48 hrs and then it is baked in muffle furnace at approximately 650°C for 4 hours. After normal cooling, the prepared mixture was passed through the sieve of required size to remove the dust from the flux and kept in air tight containers. The flux was baked again before use to eliminate any possible moisture.

3.2 Experimental Procedure

A constant DC voltage submerged arc welding machine was used for welding the mild steel plate of 300 x 150 x 20 mm using 3.15mm diameter wire of grade EL8 DIN8557: SI by bevelling the plate edges (forming an angle of 450 in between). The machine had the provision for automatic wire feed rate and trolley speed. A backing plate of 10mm with same material grade was attached to the main plates. The welding parameters, as

shown in the table 3, were kept constant for each experiment. Three to four layer weld passes were made for filling the groove in each experiment for 9 fluxes of SiO2 base. After sectioning the plate, three impact test specimen were taken in transverse to weld direction and prepared as per the standard ASTM E 23. Impact strength and Vickers hardness was measured from these prepared specimen and result is shown in table 6.

Table 3 .Welding Condition parameter

Parameter Unit Value

Voltage V 32

Current A 500

Trolley Speed cm/min. 20

Electrode stick out 2.5mm.

Flux SiO2 based

4. Analysis

4.1 Signal to noise ratio

Number of repetition of experiment permits the determination of variance index. This variation of index is called as signal to noise ratio (S/N) . The S/N ratio is used to optimise the signal value. It is basically of three type - lower is better, higher is better and nominal is better. In the present study, Vickers hardness (VHN) and impact strength of each experimental weld were obtained as response. For the weld to be of good quality hardness and impact strength must be on the higher side. A higher value of S/N ratio indicates smaller deviation in response which is desired. The equation used is no. 2 and results are tabulated in table 6.

For smaller is better

— = -10 log10 < N

For higher is better

— = -10 log10

For nominal is the best

— = -i°log10 <

Z (yj - m)

Where Yu Y2 = Result of experimental observation; m = target value; and n = Number of repetitions.

4.2 Fuzzy Logic

A fuzzy logic unit consists of fuzzifier, membership functions, a fuzzy rule base, an inference engine, and a defuzzifier. The fuzzifier makes use of membership functions to fuzzify the signal to- noise ratios that had been generated by the response. Next, the inference engine performs fuzzy reasoning on fuzzy rules to generate a

fuzzy value by this. Finally, the defuzzifier converts the fuzzy value into a multi response performance index (MRPI). The structure of the two-input-one-output fuzzy logic unit is shown in Fig. 1.The input is hardness and impact strength and output is (MRPI). The fuzzy rule base consists of a group of if-then control rules with the two inputs, x1 and x2, and one output y, that is: Rule 1: if x1 is A1 and x2 is B1 then y is C1 else Rule 2: if x1 is A2 and^2 is B2 theny is C2 else

Rule n: if x1 is An and x2 is Bn theny is Cn.

Ai, Bi and Ci are fuzzy subsets defined by the corresponding membership functions.

Impact.trength

Fig 1 . Fuzzy Logic with Mamdani interference for present study

Fig 2 . Membership functions for Hardness, Impact Strength and MPCI.

In the present study, three fuzzy subsets are assigned in the two inputs as well as one output (fig2). Details about the proposed fuzzy model are presented in table 4. Nine fuzzy rules are directly derived based on the fact that the larger the signal-to noise ratio is, the better the performance characteristic shown in table 5.

Table 4 . Details of the proposed fuzzy model

Type of fuzzy inference system (FIS) Mamdani

Inputs/output 2/1

Input membership function types Triangular

Output membership function types Triangular

Number of input membership functions 3/3

Number of output membership function 3

Rules weight 1

Number of fuzzy rules 9

And method Min

Implication method Min

Aggregation method Max

Defuzzification method Centroid

Table 5 . Fuzzy Rule base for MPCI

MPCI Impact Strength

Hardness Small Medium Large

Small Small Medium Medium

Medium Small Medium Large

Large Medium Medium Large

Table 6. Lay out of L9 OA for Vickers hardness and Impact strength with S/N ratio and Fuzzy Multi-response Performance Index.

S.No NiO Levels MnO Levels MgO Levels VHN S/N of VHN IS S/N of IS MRPI

1 1 1 1 161 44.13 52.75 34.44 0.50

2 1 2 2 168 44.51 54 34.64 0.500

3 1 3 3 176 44.91 73.25 37.29 0.852

4 2 1 2 171 44.66 80 38.06 0.868

5 2 2 3 180 45.11 62.5 35.91 0.580

6 2 3 1 180 45.11 51 34.15 0.500

7 3 1 3 174 44.81 65.66 36.34 0.633

8 3 2 1 169 44.56 30.75 29.75 0.141

9 3 3 2 184 45.30 64 36.12 0.6092

5. Results

Proposed fuzzy model which comprises of two inputs and one output fuzzifies the input data using three triangular membership functions for each input and output. Using fuzzy logic rules in mumdani interference, multi-response performance index (MRPI) in numerical value is calculated for each experiment using centroid method of defuzzification as shown in table 6. MRPI of experiment number 4 is highest so can be proposed as experiment with optimal results. For experiment number 4, levels of alloying elements are A2B1C2. Table 7 shows the average MRPI of each flux alloying element for their each level. These average MRPI scores are plotted in graphical form for each level of alloying element (fig 3). This graphical representation also confirms the result of fuzzy model i.e. suggests A2B1C2 as the optimal parameter levels.

Table 7. Average MPCI for each level of NiO, MnO and MgO

Average MRPI

Levels NiO MnO MgO

1 0.617 0.667 0.380

2 0.649 0.407 0.689

3 0.461 0.654 0.688

Le^el of MOJMnO and MgO

Fig 3. MRPI score v/s levels of each flux alloying elements

6. CONCLUSION

From this study following conclusions are drawn-

1. Using Taguchi orthogonal array, 9 SiO2 based agglomerated fluxes are prepared and tested for mechanical properties of weld metal.

2. Multi-objective optimization is done using Fuzzy Logic model for mechanical properties of weld metal.

3. A2B1C2 is suggested as the optimal level of alloying elements in SiO2 based flux by fuzzy model for optimal hardness and impact strength of weld metal.

References

[1] Masao U, Bahaa Z, Wafaa M, Adbe-Monen El, et al. Effect of arc welding flux chemical composition on weldment performance. Trans JWRI. 1995; 24(1):45-53

[2] De Rissone NMR, Faris JP, Souza Bott De I, et al. ANSI/AWS A5.1-91 E6013 Rutile electrodes: The effect of calcite. AWS 2002; 113S-123S

[3] Prasad K and Dwivedi DK. Microstructure and tensile properties of submerged Arc welded 1.25 Cr-0.5Mo steel joint. Materials and manufacturing processes. 2008; 23: 463-468

[4] Mitra U and Eagar TW. Slag metal reaction during submerged arc welding of alloy steels. Metall trans. 1984; 15A: 217-226

[5] Kanjilal P, Majumdar S K and PAL TK. Predication of acicular Ferrite from flux ingredients in submerged arc weld metal of C-Mn steel. ISIJ International. 2005; 45(6) :876-885

[6] Yang C, Lin S, Liu F, Wu L, et al. Research on the mechanism of penetration increase by flux in A-TIG Welding. J.Mater.Sci.techno. 2003;19(1): 225-227

[7] Adeyeye D Ademola and O.A Fes Festus. Weld metal property optimization from flux ingredients through mixture experiments and mathematical programming approach. Material research. 2009; 12(03): pp 339-343

[8] Adeyeye D Ademola and O.A Fes Festus. Optimization of weld-metal chemical composition from welding flux ingredients a non-pre-emptive goal programming approach. Maejo International journal of science and technology. 2010; 4(02): pp 347-359

[9] Tarng YS, Yang WH and Juang SC. The use of fuzzy logic in the taguchi method for the optimisation of submerged arc welding process Int. J Adv Manuf Technology. 2000; 16: pp 688-694

[10] Sathiya P, Abdul jaleel MY, Katherasan and Shanmugarajan B Optimization of laser butt welding parameters based on the orthogonal array with fuzzy logic and desirability approach. Sturt Muktidisc Optim . 2011 ;44 : pp 499-515

[11] Carrino L, Natale U, Nele L, Sabatini ML and Sorrentino A neuro-fuzzy approach for increasing productivity in gas metal arc welding processes. Int. J Adv Manuf Technology. 2007; 32; pp 459-467

[12]Aghakhani M, Ghaderi MR , Jlilian MM and Derakshan A. A Precicting the combined effect of TiO2 nano -particles and welding input parameters on the hardness of melted zone in submerged arc welding by fuzzy logic. Journal of mechanical science and technology. 2013; 27(7): pp 2107-2113

[13] Sateesh M, Edwin Raja Dhas J Multi objective optimization of boiler steel using fuzzy based desirability function. Journal of Engineering Science and Technology Review. 2014; 7(1): PP 29-36

[14] Sateesh M, Edwin Raja Dhas J. Multi objective optimization of flux cored arc welding parameters using fuzzy based desirability function. IJST. Transction of mechanical engineering. 2013; 37: pp 175-187

[15] Edwin Raja Dhas J and Kumanan S ANFIS for prediction of weld bead width in a submerged arc welding process. Journal of Scientific and Industrial Research. 2007;66: pp 335-338

[16] Sateesh M, Edwin Raja Dhas J Multi objective optimization of Flux Cored Arc weld parameters using hybrid grey fuzzy techanique. TJER. 2013; 11(1) : pp 23-33

[17] Aghakhani M, Ghaderi MR , Jlilian MM and Derakshan A. A combined effect of TiO2 nano -particles and welding input parameters on the hardness of melted zone in submerged arc welding by fuzzy logic Int. J Adv Manuf Technology .2014 ; 70 : pp 63-72

[18] Bhujbal VD, Tadamalle AP. Optimization of laser welding process by fuzzy logic technique. International journal of engineering science and innovative technology. 2013; 2(4): pp 221-227

[19] Narang HK, Singh UP, Mahapatra and Jha PK. Prediction of weld pool geometry of TIG arc welding by using fuzzy logic controller. International journal of engineering science and technology. 2011; 3(9): pp 77-85