(सतह ख रदर पन नय ण क लए स एनस मश न ग म कट ई क प र म टस क य ग मक ज च) Dheeraj Soni

Transcription

1 Experimental Investigation of Cutting Parameters in CNC Machining For Controlled Surface Roughness (सतह ख रदर पन नय ण क लए स एनस मश न ग म कट ई क प र म टस क य ग मक ज च) Dheeraj Soni Thesis Master of Technology In Mechanical Engineering (CAD/CAM) 2015 Department of Mechanical Engineering College of Technology and Engineering Maharana Pratap University of Agriculture and Technology, Udaipur 1

2 MAHARANA PRATAP UNIVERSITY OF AGRICULTURE & TECHNOLOGY COLLEGE OF TECHNOLOGY AND ENGINEERING, UDAIPUR CERTIFICATE I Dated: 18/04/2015 This is to certify that Mr. Dheeraj Soni had successfully completed the comprehensive examination held on 25/08/2014 as required under the regulations for the Master of Technology in Mechanical Engineering (CAD/CAM). (Dr. S. Jindal) Professor and Head Department of Mechanical Engineering C.T.A.E., Udaipur 2

3 MAHARANA PRATAP UNIVERSITY OF AGRICULTURE & TECHNOLOGY COLLEGE OF TECHNOLOGY AND ENGINEERING, UDAIPUR CERTIFICATE II Dated: 18/04/2015 This is to certify that this thesis entitled Experimental Investigation of Cutting Parameters in CNC Machining For Controlled Surface Roughness submitted for the degree of Master of Technology in the subject of Mechanical Engineering, (CAD/CAM) embodies bonafide research work carried out by Mr. Dheeraj Soni under my guidance and supervision and that no part of this thesis has been submitted for any other degree. The assistance and help received during the course of investigation has been fully acknowledged. The draft of the thesis was also approved by the advisory committee on.. (Dr. S. Jindal) Professor and Head Department of Mechanical Engineering C.T.A.E., Udaipur (Dr. B. P. Nandwana) Major Advisor Department of Mechanical Engineering C.T.A.E., Udaipur (Dr. B.P. Nandwana) Dean C.T.A.E., Udaipur 3

4 COLLEGE OF TECHNOLOGY AND ENGINEERING MAHARANA PRATAP UNIVERSITY OF AGRICULTURE AND TECHNOLOGY UDAIPUR CERTIFICATE III Dated: - 18/04/2015 This is to certify that this thesis entitled Experimental Investigation of Cutting Parameters in CNC Machining For Controlled Surface Roughness submitted by Mr. Dheeraj Soni to the Maharana Pratap University of Agriculture and Technology, Udaipur in the partial fulfillment of the requirement for the award of the degree of Master of Technology in the subject of Mechanical Engineering after recommendation by the external examiner was defended by the candidate before the following members of the examination committee. The performance of the candidate in the oral examination is satisfactory. We, therefore, recommend that the thesis be approved. (Dr. B. P. Nandwana) Major Advisor (Dr. S. Jindal) Advisor (Prof. M.A.Saloda) Advisor (Dr. Mahesh Kothari) DRI Nominee (Prof. S. Jindal) Professor and Head Department of ME C.T.A.E., Udaipur (Dr. B. P. Nandwana) Dean C.T.A.E., Udaipur Approved 4

5 Director, Resident Instruction MPUAT, Udaipur COLLEGE OF TECHNOLOGY AND ENGINEERING MAHARANA PRATAP UNIVERSITY OF AGRICULTURE AND TECHNOLOGY UDAIPUR CERTIFICATE IV Dated:- 18/04/2015 This is to certify that Mr. Dheeraj Soni student of Master of Technology in the subject of Mechanical Engineering, Department of Mechanical Engineering, College of Technology and Engineering has made all corrections/modifications in the thesis entitled Experimental Investigation of Cutting Parameters in CNC Machining For Controlled Surface Roughness which was suggested by the external examiner and advisory committee in the oral examination held on... The final copies of the thesis duly bound and corrected were submitted on 18/04/2015 are enclosed here with for approval. (Dr. S. Jindal) Professor and Head Department of Mechanical Engineering C.T.A.E., Udaipur (Dr. B. P. Nandwana) Major Advisor & Dean Department of Mechanical Engineering C.T.A.E., Udaipur 5

6 ABSTRACT Mass Customization is an attempt to provide unique values to the customers in an efficient manner. In the present work, that unique value chosen is Desired surface quality instead of Best quality which has become a choice of today s customized market. Prediction of surface roughness and dimensional inaccuracies is an essential prerequisite for any unmanned computer numeric controlled (CNC) machinery. Poor control on the desired surface roughness generates non conforming parts and results in increase in cost and loss of productivity due to rework or scrap. Surface roughness value is a result of several process variables among which vibration is of great significant. In this work, experimentation were carried out to investigate the effect of vibrations on surface roughness of machined parts and it is observed that the quality of surface finish can be predicted within a reasonable degree of accuracy by taking the amplitude of vibration of tool holder into account. Surface roughness and tool vibrations are found to be critical factors which influence the quality of the machined parts. The experimentation was performed on CNC for turning of 6061-T6511 Aluminum alloy rod using diamond shaped carbide tool insert. Experiment was designed using fractional and full factorial design of experiment (DOE) and experimental runs were conducted for various combinations of cutting parameters. Analysis of variance (ANOVA), correlation and regression techniques were employed to study the performance characteristics at different conditions according to DOE. In order to analyze the response of the system, experiments were carried out at various levels of spindle speeds, depth of cut and feed rates. The results obtained by this research will be fruitful for various industries and researchers working in this field. Key words: CNC; Turning; DOE; Surface Roughness; ANOVA. 6

7 ABSTRACT (In Hindi) म स क टम इज़ शन एक क शल तर क स हक क लए अ त य म य द न करन क एक य स ह वत म न क म म, स सतह ग णव क बज य व छत सतह ग णव क अ त य म य क तरह च न गय ह ज क आज क अन क लत ब ज र क एक वक प बन गय ह सतह ख रदर पन और आय म अश य क भ व यव ण कस भ म नव र हत क य टर स य मक नय त (स एनस ) मश नर क लए एक आव यक शत ह व छत सतह ख रदर पन पर कमज र नय ण ग र अन प भ ग क उ प न करत ह और प रन वश प क वजह स ल गत म व एव उ प दकत म न स न ह त ह सतह ख रदर पन कई चर क रक क प रण म ह जनम स क प न एक मह वप ण क रक ह इस क म म, मश न कय ह ए भ ग क सतह ख रदर पन पर क प न क भ व क ज च करन क लए य ग कय गय और यह प य गय क ट ल ह डर क क प न क आय म क य न म रखत ह ए सतह क ग णव क सट कत स एक उ चत ड क भ तर भ व यव ण क ज सकत ह सतह ख रदर पन और उपकरण क प न, मश न कय ह ए भ ग क ग णव क भ वत करन व ल मह वप ण क रक प ए गए ह ह र क आक र क क ब इड ट ल उपय ग म ल कर, स एनस पर ६०६१-६५११ ए य म नयम म ध त क छड पर ट न ग क य ग कय गय य ग आ शक एव प ण भ य डज ईन (ड ओइ) क उपय ग करत ह ए डज ईन कय गय थ और य ग मक म पद ड क कट ई क व भ न तर क लए आय जत कय गय य ग मक डज ईन क अन स र, व भ न प र थ तय म दश न वश षत ओ क अ ययन करन क लए एन व, सहस ब ध, और तगमन तकन क क नय जन कय गय ण ल क त य क व षण करन क लए, तकल ग त, कट क गहर ई, एव फ ड क व भ न तर पर य ग कय गए इस श ध स प रण म क इस म क म कर रह व भ न उ ग और श धकत ओ क लए उपय ग ह ज एग क ज श द: स एनस ; ट न ग; ड ओइ; सतह ख रदर पन; एन व 7

8 INTRODUCTION CHAPTER-I This chapter presents an overview of the motivation towards the present work, the problem statement, objectives and scopes of the present work. A brief overview is presented in this chapter about the steps towards the initiative for the present research work. 1.1 MOTIVATION Manufacturing sectors are facing a new challenge of customized demand fulfillment at a fast pace. Virtually all executives today recognize the need to provide outstanding services to their customers. The changing needs of market call for the flexibility in manufacturing where as high quantities asks for mass production. Companies throughout the world have embraced Mass Customization in an attempt to provide unique value to their customers in an efficient manner. Thus, with Mass Customization, the manufacturing enterprises are forced to make use of the modern machineries with programmable operation and high rate of output. The present age computer numeric controlled (CNC) cells provide answer to most of these challenges (still the human involvement, at planning and programming stages slows down the speed if sufficient prediction about processes are not available). The automated factories employing flexible manufacturing system are catering to the needs of many industrial sectors like automobile, cement, steel, heavy engineering, petroleum, textile and many more, which require number of parts with large variety (which is again changing at fast pace). These automated and semi automated factories make use of variety of computer controlled machines like CNC-Turning, Milling, Planning, Drilling, Boring, Coordinate measuring etc. Out of these machines the turning machines are considered to be more versatile as a large number of operations can be performed on these. Almost all the parts with axis-symmetric geometry and many others with various geometries can be worked on turning machines, making them as the most popular and largely used machines. 8

9 The quality parameters of the parts considered critical are to be taken care of while working with these machines. Modern philosophy focus on providing Required quality in place of Best quality as it always increase extra cost and time, which unfortunately are not available in the competition economy. To attain the Required or Desired surface quality, the machining parameter which affect the quality most must be identified and controlled so as to keep the output quality within the acceptable limit. Mathematical technique applied on experimental research data often provides predictive models for pre-estimation of output values of a system with given set of inputs. For a large number of small and medium enterprises (SME s) working as supporting industries or ancillary to large manufacturing companies, the results of the research done by academia or R & D organization has proved to be a ready available solution to many of the problem. Present study also aims at developing such a prediction model for estimation of surface roughness in metal machining with the given set of machining parameters. 1.2 METAL MACHINING Metal machining is one of the major area offering new challenges to the manufacturing organizations. In the last three decades, there have been immense technical revolutions in the manufacturing segment through use of integration of computers in manufacturing industries. The companies are moving towards use of automated unmanned CNC cells for increasing the production rate and decreasing labor cost. In today s global manufacturing environment, all the companies are competing to produce high quality product at lowest cost. In order to achieve this goal, the companies are needed to modernize themselves. Special attention is given to the dimensional accuracy, surface finish, high production rates, less wear on the cutting tools, economy of machining in terms of cost saving and increased performance of the product with reduced environmental impact. Out of these necessities, the ability to control the process for required surface quality of the final product is of paramount importance because a desired quality machined surface significantly improves the various properties like the sealing property, corrosion resistance, fatigue strength and creep life etc. 9

10 From among all machining operations, turning is the most common machining operation which is responsible for final surface attributes of machined parts. In a typical turning operation, certain critical machine parts may require desired surface roughness. The mechanism behind the formation of surface roughness is very dynamic, complicated, and process dependent which need continuous operator monitoring else that may lead to defects. Such machined parts may be bearing surfaces, sealing surfaces, block deck and cylinder head surfaces to seal the head gasket, couplings, aircraft fittings, valves and many more. These products demand several properties like durability, light weight, corrosion resistance etc. along with the desired surface roughness. Thus, Aluminum alloy is proven to be best material which matches the above discussed properties and applications. Aluminum Alloy 6061-T6511 is commonly used wrought alloy in most of the industries dealing in manufacturing of these components. There are a number of factors which affect the surface roughness but out of that only a few are significant factors. Several researchers considered different factors for studying about surface quality. In addition to different machining conditions, the machining parameters like spindle speed, depth of cut, feed rate, workpiece hardness, time of cut, nose radius and different cutting tool material have been found to affect surface quality to great extent. Significant effects of cutting tool vibrations were also identified on the surface roughness. Effect of radial vibrations on surface roughness had been studied by many researchers and it was established that both amplitude and frequency of vibration has strong effect on surface topography. In machining operation, it is crucial task to select optimum values of the cutting parameters for achieving better performance. Usually, these parameters are determined on the basis of experience or by the use of Machinery s Handbook. Some of the machine operator employs trial and error method to set-up turning machine cutting conditions. These methods are not effective and efficient and the achievement of a desirable value is a repetitive and empirical process that can be very time consuming. In place of these, 10

11 predictive models developed with the help of statistical techniques can be used for optimizing the machining conditions, cutting parameters and tool selection for the desired quality. The models developed, further, can be verified with properly designed experimentation via fractional and full factorial design of experiment. Furthermore, analysis of variance (ANOVA), correlation and regression techniques can be applied which are suitable statistical approaches to find out the best combination of independent variables in order to predict output values. 1.3 OBJECTIVES Looking to the need of studies in the field of predictive modeling of surface roughness in the turning operation the following objectives are shortlisted: To study the effect of cutting parameters on surface roughness in conjunction with induced vibration in CNC turning process. To perform the experiment for full factorial design at different cutting parameters and to collect the respective response data for each run. To establish significant degree of correlation between the parameters and response. To analyze statistically the significant effect between the response and the cutting parameters using the Analysis of variance (ANOVA) to obtain best set of cutting parameters. 1.4 THESIS OVERVIEW The present work aims to achieve predicted surface roughness in turning operation at CNC cell using statistical techniques. The present thesis work is organized in the following manner as discussed below. This chapter presents an overview of motivation towards present work, problem statement and objectives along with the scopes of the work. In Chapter 2, a review of literature related to the present work is presented and a brief overview is presented about the research gap that justifies the objectives of the present work. 11

12 The material and methods including the type of material, experimental setup along with their technical specification, data collection systems, methodology used for completion of this task and overview of fractional and full factorial design of experiment approaches are discussed in Chapter 3. Chapter 4 describes about the experimental results for both fractional and full factorial design of experiment approach. These results include data collection, their analysis and developing the manual as well as software based graphs relevant to collected data. Furthermore, the level of significance is checked using variance analysis and surface roughness values are predicted using regression analysis. Finally the accuracy of the regression technique is found by obtaining the rate of error between actual and predicted surface roughness values. Finally, the thesis concludes with the conclusions of this study which is given in Chapter 5. Furthermore, at last references related to literature of the thesis along with a brief about recommendations for future work, summary and abstract of the thesis has been presented. 12

13 REVIEW OF LITERATURE CHAPTER-II The literature survey is usually carried out to assess the past studies done on the subject to steer the research work towards its objectives. It basically covers previous studies related to research interest which involves the objectives of the studies, experimental details, outcomes and conclusions along with the recommendations for further work. Thus with the help of these literatures, a Research gap can be identified among previous studies and the present work which is going to be accomplished. 2.1 CURRENT AND PREVIOUS STUDIES Current and previous studies are basically reviewed to analyze the gap between conventional technologies and current technologies. By in-depth analysis of these studies it can be easily judged that how the technology is moving ahead and still what is the research gap that can be a step ahead towards future extension of that particular work. Thus, that previous work and literatures provide a path and guidance for future work so that objectives can be planned Manufacturing and Machining Technology The Manufacturing Segment plays a crucial role in global economic development throughout the world. It is only possible by driving innovation and delivering customer s solutions which are keys to sustainability of economic growth. The economic prosperity of a nation is directly linked to the manufacturing capabilities of the nation. The prosperity of the nation and the quality of life of the people depend on the manufacturing capabilities. Manufacturing indeed has positive correlation with GDP growth of the country. In India also, since independence, the Indian Manufacturing Segment has gone through various phases of development (Confederation of Indian Industries (CII), Manufacturing, 2014). It has grown at robust rate over the past 10 years and has been a best performing manufacturing economy at third position throughout the world economy. According to a report of Union Minister of Commerce and Industries, Manufacturing 13

14 contributes 15% of India s GDP which is undoubtedly a good figure (CII National Quality Summit, 2014). But, this share of manufacturing in the Indian GDP is low compared to other developed countries. There are many reasons for this situation. Indian engineering goods are generally inferior in design and quality and costly compared to global standards and are not according to customized demand. All these deficiencies are due to the fact that the Indian manufacturing technologies and machine tools has not kept pace with the developments in other countries. Globalization and liberalization of Indian economy has thrown great challenges to Indian manufacturing industries. Till a decade ago, the industrialists had a captive market and could sell what they could produce at the prices they quote. But, today the situation has changed a lot. The present scenario is of Customer Market rather than Captive Market. The technological improvement is needed to keep pace with customized demand. Only the Flexible automation via Computer Numerical Control Machines (CNC) is answer to these changing requirements. In manufacturing segment, Mechanical engineering industries specially Machining Industries is a very wide and diverse sector. Since 18 th century, the machining technology has progressed in different aspects including improvements in machine tool structures, dynamics, power and stiffness etc. (Childs et. al., 2000). It is estimated that 15% of the value of all mechanical components manufactured worldwide is derived from machining operations (Mukherjee, 2006). In machining, numerous operations are involved such as turning, facing, drilling, boring, shaping and many more but out of that Turning plays a crucial role in affecting final finishing touch to the machined components and it is also a cost deciding factor of that component. In customized market, the major emphasis is always given to the end result of the product that are in machining case may be surface finish, precision, quality of machined parts etc. Hence, in this competition market, all the companies are trying to fulfill customer s needs by innovation and improvements in their machine tools and machining technologies or by applying certain level of automation to the machineries. 14

15 2.1.2 Surface Metrology As discussed in previous sections, in customized market, the customers demand for Required surface quality rather than Best surface quality. Because certain time the best surface quality i.e. highly smooth surface is of no use for particular component. Thus, the issue of Smoothness is one that is worrisome to manufacturers today. For the nearly 35 years period since 1970 to 2005, over 300 research papers were published in accredited journals, and more than 100 patents were issued on topics related to surface roughness (Scientific Bulletin, 2006). These data of survey shows that the interest of the researchers seems to be growing exponentially in the field of surface metrology. Several researchers are exploring the best way to obtain desired surface roughness by various techniques and certain researchers working in this field are trying for achieving smooth surface finish. Tzeng et. al., (2009) in their study for optimization of turning operations also commented upon the utility of required surface attributes. They stated that, required surface attributes of the machined parts is very critical for their proper functioning and long life and also it is a measure of the technological quality of a product that greatly influences manufacturing cost. Same statement regarding the improvement of the component s life due to required surface roughness was given by Fuh et. al., (1995); while working for analysis of the surface quality. Thus it can be concluded that much research for optimization of surface quality according to customer demand is underway throughout the world. According to the most of sealing product manufacturers and suppliers, the surface finish of Cylinder head and Engine block to seal the Head gasket has a decisive effect on the performance potential of cylinder head gaskets (Victor, 2010). Principally, the better a surface is, the better will be the sealing effect. However, if a component surface that is too smooth can results in leakage and offers no adherence points for the elastomeric coating, so that adequate micro sealing cannot be ensured. For this reason, the surface roughness of cylinder head and engine block is kept in desired range. This example shows the significance of Required surface roughness instead of Smooth surface finish. 15

16 Mostly gasket manufacturers recommend that if the surface roughness is kept in desired range will provide good cold sealing and long term durability. For many years, most of gasket manufacturers have recommended a surface finish of 55 to 110 microinches of average surface roughness, when engines were made up of cast iron; those were having conventional soft-faced head gaskets on the cylinder head and deck of block (Cylinder Head Gaskets, 2005). But, recently some gasket manufacturers have changed their recommendations. Engines and castings have become lighter and less rigid and made up of aluminum rather than cast iron. There have also been changes in head gasket materials and designs. So the recommendations for some engines now require a much smoother mirror like surface finish of only 30 to 60 micro-inches of average surface roughness. Latest Ford Engines such as the 4.6L-V8V made up of multilayer steel (MLS) requires very smooth surface finishes in range of the 20 to 30 micron-inches average surface roughness (or even less). These examples show that how critically the required surface quality is needed in today s market to fulfill the requirements which may vary in the range from highly rough surfaces to mirror like smooth surfaces (Cylinder Head Surfacing, R & L Engines, n.d.). As already seen that, it is not possible to achieve desired level of surface roughness automatically by judgment of the machine tool or by trial and error method. It will need an in-depth analysis of the response values with respect to cutting parameters which affect the surface roughness. To accomplish this task, numerous technological advancements have been progressed by adding various levels of automation in the machine tools. But unfortunately, even after broad modernization, the today s unmanned CNC machine is not intelligent enough to produce a part with required surface roughness. To achieve desired level of surface roughness, previously numerous prediction systems have been designed and developed using variety of sensors. Those studies were accomplished by considering different types of sensors including dynamometers for force and torque (Azouzi et. al., 1997), accelerometer based vibration monitoring sensors (Abouelatta et. al., 2001), acoustic emission sensors (Sundaram et. al., 2007), current 16

17 probes for measuring power of the spindle (Sundaram et. al., 2008) etc. to obtain the required surface roughness. A more detailed inclusive discussion was presented by Benardos and Vosniakos (2003) who reviewed numerous surface roughness prediction systems in their study. According to a few researchers working in this field, the required surface quality can be achieved by development of proper tool condition monitoring system which can control desired level of accuracies (Endres, 1995); (Merchant, 1998); (Mukherjee, 2006). But, now with technological advancement, researchers and engineers have developed highly reliable cutting tool condition monitoring system using various techniques like online inspection techniques, experimental approach, modeling techniques, statistical process control & other mathematical techniques (Ulsoy, 1993); (Dimla, 2002). A study for surface roughness prediction suggests that, although tool wear is a major factor which affects the surface roughness but the operator can hear the chattering sound of the worn out tool and see the tool condition or can use tool condition monitoring system to measure the tool wear thus not considered as a key factor while studying about surface roughness (Kirby et al., 2004). Surface roughness is not an independent parameter which can be controlled directly. It is an uncontrolled factor which is completely based on the values of cutting parameters and some other factors. Hence, in most of the studies, major emphasis had been on the cutting parameters which affect surface quality rather than surface roughness. There are a number of factors which affect the surface roughness but out of that only a few are the significant factors. Numerous researchers considered different numbers of factors which may affect surface roughness. According to Lascoe and Nelson (1973), the most readily controlled factors in a turning operation are feed rate, cutting speed, and depth of cut; each of which have significant effect on surface finish. Choudhry et. al., (1997) and Wang et. al., (2002) in their work, considered four most readily available controlled parameters in turning operations i.e. cutting speed, depth of cut, feed rate, and insert nose radius and shown to have positive effect of machining process. 17

18 Various researchers suggested that cutting speed, feed rate, and insert nose radius have shown a significant effect on surface roughness, whereas depth of cut has shown either little or no influence on surface roughness (Yang et. al., 1998); (Lin et. al., 2000); (Davim, 2001); (Fong et. al., 2006). Numerous reviews have been accomplished related to theoretical and experimental studies on surface roughness of machined products which show that cutting conditions such as cutting speed, feed rate, depth of cut, tool geometry, and the material properties of both the tool and work piece and many more significantly influence surface quality of the machined parts (Thomas et. al., 2000); (Abouelatta et. al., 2001); (Serope et. al., 2002); (Risbood et. al., 2003); (Tim, 2003). Tamizharasan et. al., (2005) in their research work for analysis of surface finish and tool wear in turning operation, considered 18 different machining conditions including tool material, different tool grades, workpiece hardness and various cutting parameters including speed, feed, depth of cut etc. Yadav et. al., (2012) also investigated the effect of cutting parameters i.e. feed rate, depth of cut and spindle speed on surface roughness experimentally in CNC turning operation. Thus, it can be identified from these literatures that there are numerous machining parameters which affect the machining and furthermore surface quality. Out of that only a few factors have shown their positive or significant effect on surface attributes Tool Vibrations Numerous cutting parameters have been taken into consideration in previous studies which affect surface attributes. Those factors are controlled or independent factors which can be controlled easily according to the requirements. But, there may be certain factors which cannot be controlled directly but may affect machining conditions as well as surface attributes. Those factors are known to be uncontrolled noise factor. A number of studies have shown the effect of such uncontrolled factors on surface quality. 18

19 Thomas et. al., (1995) and Safeen, (2007) in their work for investigating effect of tool vibrations on surface roughness in turning process concluded that cutting tool vibrations play a vital role in affecting surface roughness as an uncontrolled factor. According to Dimla, (2002), the features of vibration like its frequency & time are highly correlated with various machining processes including turning. Still it is not clearly identified that how cutting tool vibrations affect surface roughness. Certain researchers stated vibration in single axis or double axes affect the roughness while a few stated that it has tri- axial effect. Lin and Chang, (1998) studied the effect of radial vibration on surface finish and concluded that both amplitude and frequency of vibration has strong effect on surface topography. In a study for surface roughness prediction with induced vibrations accomplished by Huang et. al. (2001) and Risbood and Dixit (2003) has found that the vibration in single axis i.e. in the direction of feed affect the surface roughness positively. Kassab and Khashmaw, (2007) correlated well their work with cutting tool vibration for prediction of surface roughness along with consideration of effect of tool length for inspecting the changes in tool vibrations in radial as well as feed axes. Amerego and Brown, (1969) in their book titled Machining of Metals argued that cutting tool vibrations occur in all the three axes that affect the cutting conditions and surface quality during turning process. Thus, in their book, they suggested to consider the effect of tri-axial tool vibration while studying about surface roughness In a study, tool vibration was treated as a controlled factor which affects directly the surface quality (Cirstoiu, 2005). In controversy of that, Sunderam and Lambert, (1981) and various other researchers (Lascoe et. al. 1973); (Fang and Wang, 2003); (Vernon and Ozel, 2003) (Ozel et. al., 2005); stated that, there may be several controlled factors which may directly affect the surface roughness but tool vibrations can never be treated as a controlled factor. Hence, numerous studies had proven cutting tool s vibration as a key parameter affecting cutting conditions and surface roughness as an uncontrolled factor. 19

20 2.1.4 Statistical Tools For achieving the desired goal or objective while studying in particular field of interest, it will need some kind of appropriate methodology or suitable technique for accomplishment of that task. Hence, according to the literature reviewed it is investigated that, for many years statistical or mathematical techniques had proven to be a best choice of the researchers for optimization of their results or for predicting the behavior of different parameters with the response value. Such techniques may involve design of experiment (DOE), analysis of variance (ANOVA), artificial neural network (ANN) technique, correlation technique, Taguchi method, regression technique and many more. These techniques are extremely needed for optimization or for obtaining best set of results because it is difficult to predict the effect of different parameters on the response values in an accurate manner without such techniques (Tobias, 1965). Several existing studies had explored the effect of various cutting parameters on response values which are based on the use of statistical and mathematical techniques (Fang and Wang, 2003); (Vernon and Ozel, 2003); (Cirstoiu, 2005); (Ozel, 2005). Numerous studies show that, Robust Design for the engineering is a better methodology for obtaining best set of results which are minimally sensitive to the numerous causes of variation to produce best quality products at least cost. Design of experiment (DOE) by fractional design approach or Taguchi analysis is an important tool for such kind of robust design which offers simple and systematic approach to optimize the design data (Bendell et. al., 1988); (Park et. al., 1996); (Ghani et. al., 2004); (Julie et. al., 2007); (Kopac et. al., 2007); (Kuram et. al., 2010). Berger and Maurer, (2001) in their study explained the importance of design of experiment in management applications and argued that Taguchi design can optimize the results through setting of design parameters as per requirement effectively and this method can be operated consistently and optimally over a variety of conditions. As conclusions drawn from the various studies, the Taguchi approach is easy to adopt and apply for users with limited knowledge of statistics but to determine the best design it requires use of a strategically designed experiment (Ersan et. al., 2006); (Nalbant et. al., 2006). 20

21 Numerous informative studies related to optimization of cutting parameters in turning have been accomplished by using Taguchi parameter design method with different combinations and levels of cutting speed, feed rate, depth of cut, cutting time, workpiece length, cutting tool material, cutting tool geometry, coolant, and other machining parameters (Lin and Chang, 1998); (Davim, 2001); (Davim, 2003); (Vernon and Ozel, 2003); (Manna and Bhattacharya, 2004); (Lin, 2004); (Yih, 2006). Analysis of variance has also been proven a best statistical tool for obtaining significance level and contribution of each variable in affecting response parameter. Thomas et. al. (2003), in their study for identifying effect of model parameters of cutting tool in turning process suggested analysis of variance (ANOVA) technique for obtaining the level of significance among various parameters involved in the study. On the other hand, ANOVA was employed to recognize the most significant variables and their interaction effects along with their percentage contribution in affecting the response value i.e. surface roughness in turning operation (Henderson, 2006); (Nalbant et. al., 2006). In similar manner a few studies were made in which ANOVA was employed to develop empirical models to analyze response of the system (Chelladurai, 2008); (Rawangwong, 2012). Several studies suggest that, improving quality and reducing cost thus optimizing the system, is only possible by choosing optimum cutting parameters or by selecting best set of input parameters. This task cannot be accomplished by Trial and error method; it will need to develop some Prediction model that can help in achieving desired outcomes (Asilturk and Kunkas, 2011). Kirby, (2006) in his study related to parameter design for turning operation using Taguchi method, formulated the prediction equation using regression approach to obtain the desired level of surface roughness values. Gopalsamy et. al., (2009) also used regression approach for process parameter optimization in hard machining of hardened steel and predicted the surface roughness values with better approximations. 21

22 Thus, from the literature reviewed it has clearly identified that, statistical tools has become first choice of the researchers and scholars to optimize their work and these techniques serve all the intended purpose to fulfill their objectives. 2.2 RESEARCH GAP The Research Gap identified from any previous work is a further step towards new and improved work. The improved work gives major emphasis on reducing all that shortcomings which can improve the outcomes of the present work. Hence, in similar manner, for identifying a particular direction for present work objectives, a number of literatures have been reviewed as discussed in above section. Through the previous literature it has been clearly identified that this is the era of Mass Customization and the scenario of market is Customized market instead of Captive market. Thus, to supply the product as per customer demand has been a basic need of the today s industry. As discussed earlier, final surface attributes as surface finish or surface roughness, surface texture etc. of a machined component is an end result which is responsible for its demand in the customized market. Thus in the present work major emphasis will be on the Desired surface roughness rather than smoother surface finish in Turning operation using CNC cell. Achieving desired surface roughness in the present work will require to consider a set of controlled cutting parameters which affect surface attributes. A number of studies have been in this field and various literatures are presented related to this work. The literature suggests that there are a wide number of parameters which affect the surface roughness but out of that only a few have their significant effect on surface roughness. After in-depth analysis about those all factors, for the present study three highly significant controlled factors have been chosen i.e. feed rate, depth of cut and spindle speed with their specified levels according to the range of machine tool being used for the experimentation. 22

23 As numerous aforementioned literatures suggest that cutting tool vibration as an uncontrolled factor affects the surface roughness in either single or double axes. No any study has been accomplished which considers its tri-axial effect. Thus, in the present study its tri-axial effect would be considered to see how it affects surface attributes and in this way rate of error can be reduced if it may be proven to be a significant factor. For accomplishment of the present work, design of experiment (DOE) approach will be employed using both fractional and full factorial design method because no any study has been previously which employed both of these approaches simultaneously. Thus, by using these both approaches in present study, an easy comparison can be done among the end results obtained from both parameter designs. Further, study will be accomplished by applying appropriate statistical tools such as variance analysis (ANOVA), regression analysis etc. for analysis of the experimental data and for formulation of predictive equation. 23

24 MATERIAL AND METHODS CHAPTER-III This chapter presents a meticulous description about selection criterion for the CNC Lathe for experimentation along with its specifications, material for workpiece and other equipments for vibration and surface roughness measurement. A brief methodology for present experimentation work is also presented in this chapter. 3.1 TURNING OPERATION Present study basically involves the use of Turning Operation on CNC Lathe as it is most widely adopted method for machining and also plays a crucial role in affecting surface roughness. Thus, for accomplishment of the present work, the experimental setup is needed to select which involves a set of machineries which are used for completion of the work along with the material used for workpiece. Thus, the requirements of experimental setup are as follows: CNC turning centre Sample work pieces Tri-axial accelerometer In-process vibration measurement setup Post-process Surface roughness tester A brief description of all these components has been presented below CNC Lathe Selection In accordance with objective of the present work, the complete set of experimentation was to perform at CNC turning centre. The machine tool selection is a crucial factor which affects the outcome of experimental work. Thus, it should be selected in such a manner that it incorporates the basic needs of the present study like desired range of spindle speed, depth of cut and feed rate etc. Therefore, DX-200 Series slant bed CNC lathe was proven to be best choice which fulfills all the necessities of the present study which is as shown in Figure

25 This is a bi-directional turner lathe which incorporates a 30 slant bed setup and some other special features that help in better machining and surface finish. The major technical specifications of the CNC turning centre are given in Table 3.1. In order to maintain constant machining and vibration condition, this experiment is performed with dry cutting i.e. without use of coolant. Fig 3.1: Layout of CNC Turning Centre- DX200 25

26 Table 3.1: Technical Specifications of CNC Turning Centre Sr. No. Specifications Range 1 Slant bed angle 30 2 Spindle speed rpm 3 Spindle motor power kw 4 Maximum Turning length 300 mm 5 Maximum turning diameter 265 mm 6 Controller Fanuc- Siemens 7 Number of station of turret head 8 8 Weight 3200 kg Material Selection As discussed in Chapter 1, the Aluminum Alloy 6061-T6511 had proven to be best material which matches the desired material characteristics for present experimental work and also it is the commonly used wrought alloy in most of the industries. It is a material at which easy detection in variation in surface roughness profile can be detected with changes in cutting parameters. Thus proper analysis for the present study can be made. Mechanical properties and chemical composition of T6511 Aluminum Alloy is as shown in Table 3.2 and Table 3.3 respectively. The work pieces cut for the experimentation were of similar dimensions and cut from 1.0 inch (0.254 m) diameter rod. Each work piece was roughly cut prior to the final finish cut in order to maintain dimensional accuracies and proper measurement of vibrations at varying cutting parameters. Characteristics of Aluminum Alloy There are some key characteristics of this material which have proven this material to be very important for industrial use. These properties are its durability, light weight, corrosion resistive, excellent joining characteristics and good acceptance of applied coatings, relatively high strength and good workability. These properties make it useful for various industrial segments like couplings, aircraft, pistons, fittings, and bike frames etc. 26

27 Table 3.2: Mechanical Properties of T6511 Aluminum Alloy Tensile Yield Young s Thermal Hardness Elongation Strength Strength Modulus Conductivity 95 BHN % 310 MPa 276 MPa 69 GPa 167 W/mK (Source: Metals Handbook, 1990) Table 3.3: Chemical Composition of T6511 Aluminum Alloy, % Weight (Source: Metals Handbook, 1990) In-Process Vibration Measurement Setup There are basically three characteristics of the mechanical vibrations by which its intensity can be judged i.e. vibration frequency, vibration amplitude and phase of the vibration. Out of these, each characteristic is employed to study different types of vibrations caused in different machineries. The first major characteristic of vibration is the Frequency, which is just a gauge of the number of total revolutions that take place in a specific period of time. The second characteristic of vibration is the Phase which is simply an assessment of relative time difference between two sine waves. Although, Phase is in fact a time difference, but it is always represented in terms of angle, either in radians or degrees. But, after all, if the machine is running well and smoothly, recording the frequencies or phase difference of vibration is not imperative. The extent of vibration or how smooth or rough the machine vibration is expressed by its vibration amplitude. Vibration amplitude of any machine can be expressed and measured in many ways as in terms of displacement, velocity or acceleration. Al Cr Cu Fe Mg Mn & Ti Zn Max 0.8- Max each 5 For the present study, amplitude in terms of displacement had been taken into consideration which is proven to be an emphasizing analysis data affecting the surface roughness. The vibration displacement is generally calculated in units identified as micrometer or micron (µm). In metric units, the one micrometer or micron is equal to one thousand of a millimeter, i.e., 1 µm = mm. In inches system, this unit is called mils, where one mils equals one thousandth of an inch i.e. 1 mils = inch. In this study, 27

28 inches system has been adopted. The complete measurement of these vibration data was made using an In- Process vibration measurement setup. The setup for in-process vibration measurement involves an accelerometer and a frequency analyzer. Usually the frequency analyzer connects to the machine via an accelerometer to the point where vibration is to be measured. Further, these vibration signals are amplified and stored in PC for further analysis task. A brief description of these components is presented below: a) Vibration Accelerometer It is basically a type of vibration transducer which is when held or attached to a machine, converts its mechanical vibration signals into electrical signals. Further, that signals can be processed by the associated instrument into measurable characteristics of the vibration. Presently, it is available in all types of the vibration meters, analyzers and data collectors. It is a self generating device that produces a voltage output proportional to vibration acceleration. The reasons for their popularity include their compact, rugged and light weight design as well as a broad range of frequency response. The accelerometers are generally connected to the external surface of machine. The installation and application of an accelerometer is made carefully for having reliable and accurate vibration reading. Typically, following three methods for mounting of accelerometer for vibration sensing applications are employed in rotating machines; Stud Mounting Adhesive Mounting Magnetic Mounting Out of three, magnetic mounting type of accelerometer has been used in present work for recording the vibration data in machining operations because it provides more accurate data and its position can easily be changed. In present work, a tri-axial piezoelectric accelerometer was employed. While machining, it measures and amplifies the cutting tool s vibration signals along all the three axes i.e. X (Radial), Y (Tangential) and Z (Feed) axes in which vibration signals are 28

29 represented by Vx, Vy and Vz respectively. For amplification of the signals, tri-axial accelerometer sensor was used that is PCB Piezotronics 356-B-08 sensor as shown in Figure 3.2. The accelerometer was mounted by Magnetic mounting stud at the machine for vibration analysis. All the signal data were recorded and digitized for analysis purpose using frequency analyzer and data were recorded to the PC attached with the setup for further analysis. There are numerous properties of Tri-axial accelerometer which make it useful for measurement of vibration. The key properties are as follows: Eliminates the use of tri-axial mounting blocks and associated errors. Permits measurements on smaller objects with greater spatial accuracy. Operates with low cost, constant-current signal conditioners. Through-hole mounting adaptor and single cable hookup simplify installation. Low profile, lightweight, hermetically-sealed, titanium housing. Fig 3.2: Tri-Axial Accelerometer b) Vibration Frequency Analyzer For machine s vibration measurement, the vibration frequency analyzers are available in a wide range of features and capabilities. All vibration frequency analyzers currently available can be categorized as either; Analog or swept filter frequency analyzers Digital FFT frequency analyzers 29

30 Analog frequency analyzers basically work in the same manner as Radio having a tuner which is tuned to specific frequency. But, now-a-days these are generally considered out of date by today s technological standard, thus not considered for the present work. The present study was made using Digital Frequency Analyzer, which is based on the mathematical calculation made by Fast Fourier Transform (FFT). It is a fully digitized system connected with a PC system which stores the vibration data and provides to the user as per requirement Surface Roughness Tester After completion of machining process, the surface roughness was needed to calculate for each turned workpiece in present work. There are various surface roughness measurement techniques, by which in-process and post-process measurements can be taken. Usually, the techniques of measuring roughness of machined parts can be classified into two categories: a) Non-Contact Surface Roughness Measurements Technique Monitoring surface roughness is often done by manual contact inspection of part surface. However, manual contact inspection is time-consuming. Thus, several noncontact measurement techniques have been developed to reduce measurement time, such as fiber optics, ultrasonic, machine vision, etc. These can be applied in the surface roughness measurement of in-process machining. The above mentioned techniques are basically used to measure the surface roughness of a part being machined. However, they are not good for in-process measurement, since in the non-contact inspection technique, such as in machine vision, the measurement apparatus is needed to be mounted on the machines. The coolant or chips may cover the measured surface of the machined parts during machining, resulting in difficulty for the non-contact technique to measure surface roughness in process. b) Contact Surface Roughness Measurements Technique This is the second kind of technique for measuring the surface roughness in which either the workpiece or the instrument is kept in contact with each other. This is a post process surface roughness measurement technique. For the present study, the task for 30

31 surface roughness measurement was accomplished using this post process technique. It is basically a Stylus Profiler approach in which a Stylus Profilometer is used to measure post process surface roughness by contact inspection technique. This instrument uses a diamond stylus profiler as shown in Figure 3.3 to directly touch the part s surface and measure its surface roughness value by averaging the value in the cutoff. The stylus profiler is one of the most widely used instruments to measure surface roughness in industry and academic laboratories. Fig 3.3: Diamond Tip Stylus Among its many advantage are that it is easy and quick to use, has good repeatability, and is relatively inexpensive. The surface roughness value from a profiler is usually employed as a basic index, which is taken and compared with other surface roughness measurement techniques. Thus, in the present work, after final finish cut, the surface roughness was measured for each work piece at each 90 incremental interval along the circumference as shown in Figure 3.4. This gives a better approximation of the result for surface roughness and after that their mean value is used for final result and analysis purpose. The measurements were taken using Mitutoyo Surface Roughness Tester SJ-210 as shown in Figure 3.5. The measurements were obtained with the help of movement of stylus with diamond tip over the surface along the z- axis. For measurement of the surface roughness along z-axis, the finished part is mounted over a specially designed V-shaped fixture at which the workpiece can be placed easily for movement of diamond tip stylus over the surface of workpiece. 31

32 Fig 3.4: Measurement Section of Surface Roughness Fig 3.5: Mitutoyo Surface Roughness Tester: SJ-210 The instrument measures various forms of surface roughness amplitudes according to the particular industrial use i.e. Arithmetic mean of roughness (Ra), Root Mean Squares (RMS) of roughness (Rq), and Maximum peak to valley values of roughness (Rz or Rmax). But out of those, the average surface roughness (Ra) is most widely adopted because it provides the average value of surface roughness which is an easy interpretation of roughness profiles. While, the other two parameters either provide RMS value or maximum value of surface roughness. Thus, present study was accomplished by measuring average value of surface roughness (Ra) for each turned workpiece sample. The average roughness can be defined as the integral of the absolute value of the roughness profile height over the evaluation length or the area between the roughness profile and its mean line shown in Figure 3.6 (Surface Metrology Guide, 2009). The value of average surface roughness is given by the equation 3.1: 32

33 R a = (3.1) Where: R a = Deviation of the arithmetic average from the centre line, L = Total length of the sampling profile, H= Height of the profile above and below the centre line Fig 3.6: Typical Surface Roughness Profile 3.2 DESIGN OF EXPERIMENT This section presents a detailed description about experimental strategy, design of experiment (DOE) via fractional and full factorial approach, and data collection system for the present work. A brief review about these techniques employed in the study also has been introduced in this section Experimentation Strategy The present work is divided basically into three phases. In the first phase, the design of experiment is carried out according to both fractional and full factorial design approach. In the second phase, the experiments are performed for individual run at different levels and all the data are collected. In the final phase, analysis task is performed over the collected data using various mathematical and statistical tools for optimization of the results. In the present work, the complete above mentioned strategy has been implemented in following two stages: In the first stage, for fractional design of experiment, Taguchi L9 orthogonal array 33

34 with the objective of least experiment (nine runs) was employed. Signal to Noise (S/N) ratio, analysis of variance (ANOVA) and regression analysis are taken up to get the desired optimal levels of the controlled cutting parameters for obtaining predicted surface roughness. In the second stage, experimentation is planned using full factorial design of experiment for the three input parameters and their different levels. DOE method is indeed an effective tool in developing a reliable and accurate model and has a broad range of applications in optimizing problems (Oktem et. al., 2005); (Rao Mohan, 2009); (Yang, 2009). Overall thirty six (2x6x3=36) experiments were carried out in order to let the statistical detection of two parameters interactions of all three independent factors. Analysis of variance (ANOVA) was carried out for identifying the significant factors and their effect on the response variables. Regression technique was used to predict the surface roughness values with optimized cutting parameters Fractional Design Approach: Taguchi Design of Experiment Taguchi method is a statistical method to improve the quality of manufactured goods, and more recently this technique is widely applied to engineering problems. These are the optimization tools by which best set of results can be obtained by conducting minimum number of experiments thus called Fractional design approach. Taguchi method uses a special set of arrays called Orthogonal Arrays. These standard arrays stipulate the way of conducting the minimal number of experiments which could give the full information of all the factors that affect the performance parameter. The bottom of the orthogonal arrays method lies in choosing the level combinations of the input design variables for each experiment. The technique of laying out the conditions of experiments involving multiple factors was first proposed by the Fisher, (1925). The method is popularly known as the factorial design of experiments. A full factorial design will identify all possible combinations for a given set of factors. Since most industrial experiments usually involve a significant number of factors but, a full factorial design results in a large number of experiments. To reduce the number of experiments to a practical level, only a small set from all the possibilities is selected. Thus, partial or fractional design of experiment is a 34

35 method of selecting a limited number of experiments which produces the most information via Taguchi method. This method involves a technique of reducing the variation in a process through robust design of experiments. The overall objective of the method is to produce high quality product at low cost to the manufacturer and was firstly developed by Dr. Genichi Taguchi of Japan who maintained that variation. This method allows for the collection of the necessary data to determine which factors most affect product quality with a minimum amount of experimentation, thus saving time and resources. a. Steps for Taguchi Method The design of an experiment by Taguchi method involves the following steps: 1) Define the process objective 2) Selection of independent variables 3) Selection of number of level settings for each independent variable 4) Selection of orthogonal array 5) Assigning the independent variables to each column 6) Conducting the experiments 7) Analyzing the data 8) Inference Brief descriptions of the major steps involved in the Taguchi Method are as follows: 1) Define the process objective: It involves a strategy to choose a target value for a performance measure of the process. The target of a process may be to maximize or minimize the output. 2) Determine the design parameters affecting the process: Parameters are the variables within the process that affect the performance measure that can be easily controlled. The number of levels that the parameters should be varied at must be specified. Increasing the number of levels of the parameters will increase the number of experiments to be conducted. 35

36 3) Create orthogonal arrays: It involves a step for selecting the parameter design indicating the number of end conditions for each experiment using Orthogonal Array approach. The selection of orthogonal arrays is based on the number of parameters and the levels of variation for each parameter. Before selecting the orthogonal array, the minimum number of experiments to be conducted shall be fixed based on the total number of degrees of freedom. The minimum number of experiments that must be run to study the factors shall be more than the total degrees of freedom available. The number of degrees of freedom associated with each factor under study equals one less than the number of levels available for that factor. 4) Assigning the independent variables to columns: The order in which the independent variables are assigned to the vertical column is very essential. In case of mixed level variables and interaction between variables, the variables are to be assigned at right columns as stipulated by the orthogonal array. Finally, before conducting the experiment, the actual level values of each design variable shall be decided. It shall be noted that the significance and the percentage contribution of the independent variables changes depending on the level values assigned. Hence, proper experimental levels should be chosen to conduct the study. 5) Conduct the experiments: Once the orthogonal array is selected, the experiments are conducted as per the level combinations. The interaction columns and dummy variable columns shall not be considered for conducting the experiment, but are needed while analyzing the data to understand the interaction effect. 6) Complete data analysis: Since each experiment is the combination of different factor levels, it is essential to segregate the individual effect of independent variables. This can be done by summing up the performance parameter values for the corresponding level settings. In other words, by conducting the analysis of variance (ANOVA), one can decide which independent factor dominates over other and the percentage contribution of that particular independent variable. Depending on the complexity of the analysis, a pictorial depiction of the above discussed steps and additional possible steps are presented as shown in Figure

37 b. Experimental Levels and Orthogonal Array Selection The effect of various parameters on the performance characteristic in a condensed set of experiments can be examined by using the orthogonal array experimental design proposed by Taguchi. Once the parameters affecting a process that can be controlled have been determined, the levels at which these parameters should be varied must be determined. Determining the levels of a variable requires an in-depth understanding of the process, including the minimum, maximum, and current value of the parameter. Also, the cost of conducting experiments must be considered when determining the number of levels of a parameter to be included in the experimental design. Typically, the number of levels for all parameters in the experimental design is chosen to be the same to aid in the selection of the proper orthogonal array. Knowing the number of parameters and the number of levels, the proper orthogonal array can be selected. Using the array selector table as shown in Table 3.4 and the name of the appropriate array can be found by looking at the column and row corresponding to the number of parameters and number of levels. Once the name has been determined (the subscript represents the number of experiments that must be completed), the predefined array can be looked up. There is one more approach by which the array can be selected directly by statistical software approach i.e. MiniTab or JMP SAS statistical software. This is also a quite easy approach in which just by entering the number of factors and their corresponding number of levels, the software will itself provide us the suitable type of orthogonal arrays for particular study. These statistical software are 30 Days Trial- Versions available at free of cost on the internet for academicians and scholars. For present study major emphasis has been on using MiniTab-17 statistical software [Minitab-17 Free Trial, 2014]. For the present work, the experimental levels for the controlled factors are as shown in Table 3.5, where all the three controlled factors i.e. spindle speed (SS), depth of cut (DC) and feed rate (FR) has three levels. 37

38 Determine the Factors and their levels Identify test conditions Identify Control and noise factors Design the experiment Define the data analysis procedure Conduct experimentation Analyze data using S/N ratio and ANOVA Fig 3.7: Pictorial Depiction of Steps in Taguchi 38

39 Table 3.4: Orthogonal Array Selector Thus, for the present experimental levels, from the given orthogonal array selector table, L9 orthogonal array was chosen which requires nine experimental runs to be conducted to test all the factors to analyze the results. The combinations of these cutting parameters for experimentation were obtained by Taguchi design of experiment through MiniTab-17 statistical software as shown in Table 3.6. Cutting Parameters Table 3.5: Experimental Levels of Cutting Parameter Units No. of Levels Values For Each Level Level 1 Level 2 Level 3 SS rpm DC inches FR ipr c. Experimental Procedure For completion of the present study, a randomized schedule of runs was created at various combinations according to Taguchi design of experiments as shown in Table 3.6. The workpieces from the bar were turned with specified cutting conditions. The dry turning was performed for each run in order to get accurate vibration signals. During the finish turning, in-process tri-axial vibration data were collected using vibration data collection system as discussed earlier. After completion of all the runs, the surface roughness of all the work pieces was measured using SJ-210 Surface roughness tester. 39

40 The complete line diagram of experimental setup for the present work is as shown in Figure 3.8. Fig 3.8: Line Diagram of Experimental Setup 40

41 d. Analysis Methodology Once the experimental design is determined and the trials have been carried out, the measured performance characteristic from each trial can be used to analyze the relative effect of the different parameters. To determine the effect of each variable on the output, the signal-to-noise (S/N) ratio needs to be calculated for each experiment conducted. In S/N ratio, the signal is representing the desirable value i.e. mean of the output characteristics while the noise represents the undesirable value i.e. squared deviation of output characteristics. Table 3.6: Design of Experiment via Taguchi Method Spindle Speed (rpm) L-1: 2500 L-2: 3000 L-3: 3500 FR (ipr) L-1: L-2: L-3: DC (in.) L-1: 0.01 Run # L-2: 0.02 L-3: 0.03 L-1: Run # L-2: 0.02 Run # Run #3 Run #6 L-3: 0.03 L-1: 0.01 L-2: Run #5 Run # L-3: 0.03 Run # Run # While analyzing through Taguchi method, there are three categories for analysis of S/N Ratio i.e. the smaller is better, the larger is better and the nominal is best. The S/N ratio for each level of process parameter is computed by S/N analysis. Regardless of the category, the larger S/N ratio is always recommended for better performance. Thus, the optimal parameter for any factor is the level having highest S/N ratio. Usually, the objective of the first category is to optimize the system when the response is as small as possible while the second category is used when the response is as large as possible and the objective of the third category is to reduce the variability around the specific target. The following equations are used to calculate S/N ratio [MiniTab-17 Free Trial, 2014]: 41

42 1) The smaller is better : η = -10log 10 ( y ) (3.2) 2) The larger is better: η = -10log 10 (( )/n) (3.3) 3) The nominal is best: η = 10log 10 (3.4) Where: η = Signal to Noise ratio n = No. of repetitions of the experiment y i = Measured value of the quality characteristics s = Variance The S/N ratios are expressed on the decibel scale. For the present experimental analysis, the first category i.e. The smaller is better was chosen to apply while calculating the values of S/N ratio using MiniTab-17. The first category was chosen to obtain the optimization conditions for minimization of surface roughness which is usually a desired condition for turned machined parts. After that the table for Rank value of the controlled factors, its main effect plots and ANOVA table for S/N ratio can be generated for observation of results obtained by these analyses. Thereafter, Regression equation was formulated to predict the desired surface roughness value. The meaning of the word Regression is the act of returning or going back. This term was firstly used by Galton, (1877), while studying the relationship between heights of fathers and sons. This term is basically used after finding out the correlation between the data. The regression equation is usually formulated using the factors affecting significantly the response value. These highly significant factors are analyzed using Pearson s Correlation Technique which shows percent contribution of these factors on the response value (Data Analysis, 2014). 42

43 In the present study, this task of formulating the regression equation was accomplished by Minitab-17 statistical software. As stated above, the equation involves only the significant variables which have positive correlation with the value to be predicted. Thus, before conducting regression formulation, Pearson s correlation method was applied among all the data variables and the response value. Pearson s correlation method basically follows the three steps: 1) Determining whether a relation exists or not. 2) Testing whether it is significant or not. 3) Establishing the cause and effect relation, if any. The Pearson s method (Data Analysis, 2014) also known as the Karl Pearson s Method is most widely employed in the engineering practices. Pearson s coefficient is usually denoted by the symbol r. The value of this coefficient is usually obtained either by mathematical formula or can be found by various mathematical or statistical available software. The value of r always remains between +1 to -1. Between this ranges the following cases may arise: 1) If r = +1: There is perfect positive correlation. 2) If r = 0 : There is no correlation 3) If r = -1: There is perfect negative correlation Thus by using this method, the value of r can be identified and significance of the parameter relation can be shown by the above stated relations. After identification, the factors having positive correlation are only considered for the further study and analysis purpose. Furthermore, the regression technique was applied using MiniTab-17 software by entering the highly correlated values in the equation, and finally the rate of error was measured between the actual response values and the predicted response values. 43

44 The following equation was used to check the accuracy in terms of percentage error or rate of error: Rate of Error (δ) = x 100 % (3.5) Where: R a : Measured surface roughness for specific run Ra p : Predicted surface roughness for specific run, n : Total number of measurements i : Measurement being done for specific run. Since, the experiment involves a smaller number of experimental runs, thus a validation run was also conducted to test the accuracy of the system and to validate the results. As discussed above, the rate of error was also calculated for the validation model using the same Equation Full Factorial Design Approach In statistics, a full factorial experiment is an approach whose design consists of two or more factors, each with discrete possible values or levels, and whose experimental units take on all possible combinations of these levels across all such factors. A full factorial design may also be called a fully crossed design. Such an experiment allows the investigator to study the effect of each factor on the response variable, as well as the effects of interactions between factors on the response variable. As discussed in the Taguchi method, if the number of combinations in a full factorial design is too high to be logistically feasible, a fractional factorial design may be done, in which some of the possible combinations are omitted. But, Fisher, (1925) in his book The Design of Experiments stated that the factorial designs are more efficient than studying certain factors at a time in fractional design approach. 44

45 a. Steps in DOE Although, the full factorial design approach is very easy to perform because all the data combination is involved for the study. But still there are certain steps which should be followed to complete the task in a proper way. These steps are as follows: 1) Identify factors of interest and response variables 2) Determine appropriate levels for each independent variable. 3) Determine a design structure according to full factorial design approach. 4) Conduct the experiment and collect the data. 5) Analyze the data and organize the results in order to draw appropriate conclusions. b. Experimental levels and Run selection It is an attempt to experimentally investigate the effect of various cutting parameters on the surface roughness along with its vibration signals. A randomized schedule of runs will be created for all the possible sets of parameter levels being used in the experiment for a number of samples of work piece having same material. The experiment will involve three independent variables (spindle speed, depth of cut & feed rate) and four dependent variables (R a and V x, V y, V z ). For conducting the experiment each controlled factor or independent variable will have certain fixed levels as shown in Table 3.7 and this will result in total of (3x6x2) 36 runs (i.e. full factorial design) to be conducted to test all the possible combinations of parameter levels. In case of fractional design approach, the results obtained are only relative and do not exactly indicate what parameter has the significant effect on the performance characteristic value because it does not involve all the interaction effects. But, In case of full factorial design approach, although number of runs for the experimentation is increased but the results obtained will be the best one and will show the exact parameter value which will affect the performance significantly. The full factorial design approach will include each and every combination of possible experimental run. Hence, the numbers of experimental runs are highly increased 45

46 compared to fractional design approach. Accordingly, MiniTab-17 software all the possible run for full factorial design is as specified in Table 3.8 which results in total of 36 runs to be conducted. Cutting Parameters Table 3.7: Experimental Levels of Cutting Parameters Units No. of Levels Values For Each Level SS rpm FR inches DC ipr Table 3.8: Design of Experiment for Full Factorial Data Collection Spindle Speed (rpm): L-1: 2500 L-2: 3000 L-3: 3500 FR (ipr) DC (in.) L-1: 0.01 L-2: 0.02 L-1: 0.01 L-2: 0.02 L-1: 0.01 L-2: 0.02 L-1:0.002 Run #1 Run #2 Run #13 Run #14 Run #25 Run #26 L-2:0.003 Run #3 Run #4 Run #15 Run #16 Run #27 Run #28 L-3:0.004 Run #5 Run #6 Run #17 Run #18 Run #29 Run #30 L-4:0.005 Run #7 Run #8 Run #19 Run #20 Run #31 Run #32 L-5:0.006 Run #9 Run #10 Run #21 Run #22 Run #33 Run #34 L-6:0.007 Run #11 Run #12 Run #23 Run #24 Run #35 Run #36 c. Experimental Procedure A randomized schedule of runs will be created at various combinations according to full factorial design approach as shown in Table 3.8. Similarly to the experimentation performed in Taguchi method, the work pieces from the bar were turned with specified cutting conditions. The dry turning was performed for each run in order to get accurate vibration signals. During the finish turning vibration data were collected on each axes using vibration data collection system. 46

47 After completion of all the runs, the surface roughness for each workpiece was measured in the similar manner as discussed in fractional design approach and the measured roughness values were collected in the given data table in appropriate column. d. Analysis Methodology Once the experimental design was determined and the trials were carried out, the measured performance characteristic from each trial can be used to analyze the relative effect of the different parameters. To determine the effect of each variable on the output, firstly analysis of variance (ANOVA) was applied for observing the effect of all the controlled factors over the vibration signals in each axes separately. Analysis of variance frequently referred by the contraction ANOVA, is a collection of statistical models or statistical technique used to investigate and model the relationship between a response variable and one or more independent variables. It is also used to analyze differences between group means and their associated procedures (such as variation among and between groups). It is specially designed to test whether the means of more than two quantitative populations are equal. Each explanatory variable called the factor and consists of two or more categories called the levels. This technique was developed by R.A. Fisher in 1920 s is capable of fruitful application to a diversity of practical problems. Basically, it consists of classifying and cross classifying statistical results and testing whether means of a specified classification differ significantly. In this way it is determined whether the given classification is important in affecting the results. This variance technique is based on the F-test named after R.A. Fisher. Usually, the larger the F- values, there will be a greater effect on the performance due to varying cutting parameters. Fowlekes and Craveling, (2006) suggested a method for looking at the F-ratios calculated in the ANOVA table for each parameter to find out its level of significance. Because the main purpose of this tool is to find the significance of the parameters. Hence, the following criteria have been suggested by the above mentioned researchers: a. If F<1: Control factor effect is insignificant (the experimental error outweighs the controlled factor effect) 47

48 b. If F 2: Control Factor has only a moderate effect compared with experimental error. c. If F>4: Control Factor has a strong (clearly significant) effect. ANOVA consists of simultaneous hypothesis tests to determine if any of the effects are significant or not. Several calculations are usually made for each main factor and interaction term. But, for the present study the complete task for developing the ANOVA tables including the values of sum of squares (SS), degree of freedom (df), mean square (MS), mean square error (MSE) and F-ratio were measured with the help of MiniTab-17 Statistical software. Thereafter, regression technique was applied to formulate the prediction equation as discussed in fractional design approach and in similar manner rate of error was obtained between actual and predicted surface roughness values using Equation 3.5. The next chapter will involve all the work related to data collection using the proposed experimental setup and methodology and further their analysis using the statistical techniques as discussed in this chapter. 48

49 RESULTS AND DISCUSSION CHAPTER-IV After thorough analysis for various cutting parameters and their levels, the experimental run tables have been developed for both fractional and full factorial design of experiment (DOE) as discussed in Chapter 3. There was a major role of MiniTab-17 statistical software in developing these data table and providing appropriate level values for each experimental run. In this chapter, a description is presented about the task of data collection and their analysis according to proposed methodology as discussed in Chapter 3. This complete task has been divided between two stages: 1) Fractional design of experiment 2) Full factorial design of experiment Hence, the experiments were performed according to these set of data levels and further these data were used for analysis and formulation of regression equation to get predicted surface roughness values. 4.1 FRACTIONAL DESIGN OF EXPERIMENT This section basically deals with the collection and analysis of the data using fractional design approach. All the techniques and methodology discussed in the previous chapter is involved in the present section Experimental Run As discussed in the previous chapters, fractional design approach is fully based on the Taguchi DOE. Experimental run table already has been generated using orthogonal array and now after performing the experimentation as proposed in the Table 3.6, the results of individual run including mean vibration signals and surface roughness were collected as shown in Table 4.1. This table also shows the value of S/N ratio obtained by MiniTab-17 software employing The smaller is the better approach. 49

50 Sr. No. Table 4.1: Factors and Response Data for Individual Experimental Run Factors SS DC FR Ra S/N ratio Response Analysis of variance, Rank Value and Main Effect Plots Vibration Amplitudes Vx Vy Vz It can be seen by the collected data in Table 4.1 that, the amplitude of vibrations on all the three axes and the surface roughness is affected by all the controlled factors i.e. spindle speed, feed rate and depth of cut. But it is not clearly justified that which factor has significant effect. Hence, Rank table and analysis of variance (ANOVA) table were developed for S/N ratio to explore these observations using MiniTab-17 software. The rank table basically shows that which factor has the highest impact over the response while the ANOVA shows their significance with the help of F-values as discussed in previous chapter. These data values are as shown in Table 4.2 and Table 4.3 respectively. It can be seen from Table 4.2 and the corresponding rank value for each factor that the feed rate (Rank 1) is the highest influencing factor which has strongest effect on the surface roughness followed by spindle speed (Rank 2) and last by depth of cut (Rank 3). Hence, while designing the system, major emphasis should be on the feed rate. The identical sets of results have been obtained by analysis of variance (ANOVA) as shown in Table 4.3 by which most significant variable can be analyzed. From this table, it can be seen that the feed rate is a highest significant factor (F-value = 5.57), 50

51 while effect of depth of cut and spindle speed has not been found statistically significant (F<1). Thus, Feed rate strongly affect the S/N Ratio while the other two factors (F<1) have insignificant effect. Table 4.2: Rank Table for S/N Ratios with Cutting Parameters Level SS DC FR Delta Rank Table 4.3: Analysis of Variance for S/N Ratio Source DF Seq SS Adj SS Adj MS F P Percentage Contribution SS % DC % FR % Residual error % Total S = R-Sq = 87.2% R-Sq (adj) = 49.0% Therefore, based on S/N ratio and variance analysis tables, the optimal cutting parameters for surface roughness are the Feed rate at level 1, the Spindle speed at level 2 and the Depth of cut at level 3. Further, the same results can be validated by main effect plot for S/N ratio as shown in Figure 4.1. It can be analyzed from the main effect plots of S/N Ratio that in order to obtain optimized value of surface roughness, the feed rate should be set to its lowest value (0.002 ipr) while the spindle speed and depth of cut to their highest values i.e rpm and 0.03 inch respectively because for optimization, largest S/N ratio should be 51

52 employed. Although, Depth of cut has third rank, so it s any level can be chosen because it will not influence the surface roughness. Fig 4.1: S/N Ratios vs. Cutting Parameters (SS, DC, FR) Plots Furthermore, a chart was also developed which shows the percentage contribution of each factor on the S/N ratio as shown in Figure 4.2. This was developed with the help of ANOVA values as shown in Table 4.3. It is clearly seen that, only feed rate has nearly 70% effect on the response and 30% effect is shared by remaining factors including error Regression Analysis Regression analysis is performed to formulate predictive equation for most significant data i.e. data having highest correlation coefficient so that it will create more robust model. For that, Pearson s correlation coefficient is measured for each factor with respect to the surface roughness as shown in Table 4.4. The positive correlation is analyzed as discussed in the previous chapter and a pie chart is also developed which shows the percentage correlation with surface roughness as shown in Figure 4.3. Finally, the regression model is created with the help of highly correlated factors to get prediction equation. As it can be analyzed from Table 4.4 and Figure 4.3 that, the feed rate and mean amplitude of vibrations in y & z axes (i.e. Vy and Vz) have positive correlation with the surface roughness so these most significant variables can be used in regression analysis to 52