Docstoc

APPLICATION OF NON-TRADITIONAL OPTIMIZATION FOR QUALITY IMPROVEMENT IN TOOL HOLDERS

Document Sample
APPLICATION OF NON-TRADITIONAL OPTIMIZATION FOR QUALITY IMPROVEMENT IN TOOL HOLDERS Powered By Docstoc
					 International Journal of JOURNAL OF MECHANICAL ENGINEERING
INTERNATIONAL Mechanical Engineering and Technology (IJMET), ISSN 0976 –
 6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME
                           AND TECHNOLOGY (IJMET)

ISSN 0976 – 6340 (Print)
ISSN 0976 – 6359 (Online)
                                                                              IJMET
Volume 3, Issue 3, September - December (2012), pp. 362-377
© IAEME: www.iaeme.com/ijmet.asp                                          ©IAEME
Journal Impact Factor (2012): 3.8071 (Calculated by GISI)
www.jifactor.com



       APPLICATION OF NON -TRADITIONAL OPTIMIZATION FOR
             QUALITY IMPROVEMENT IN TOOL HOLDERS

                                       K. Saravana kumar
        Assistant Professor, Department of Mechanical Engineering, Karpagam University,
                                        Coimbatore, India

                                    Dr.A.K. Shaik Dawood
         Professor, Department of Industrial Engineering, King Khalid University, Abha,
                                          SaudiArabia
                               Email:dawod.shaik77@gmail.com

                                       P.A. Azeem Hafiz
        Assistant Professor, Department of Industrial Engineering, King Khalid University,
                                       Abha, SaudiArabia

                                        R. Karthikeyan
         Assistant Professor, Department of Management Studies, Karpagam University,
                                       Coimbatore, India

 ABSTRACT
          In the present scenario, quality has become an important factor, which determines the de-
 velopment of a company. Initially the companies were going in for 100%inspection of the com-
 ponents for maintaining their quality. Since quality lies in the efficient control of defects, nowa-
 days newer statistical quality control techniques are employed. At present all the companies are
 moving towards six sigma concept. Even then most of the companies are not able to achieve this
 target. This is mainly attributed to the use of machines with poor process capabilities. This
 project aims at improving the process capability of machines by optimizing the control parame-
 ters thereby reducing the number of defects arising.
          This work deals with the problem arising in a special grinding process know as face pro-
 file grinding done in compression rings of a piston. The rejection level for this process was very
 high as the crowning tolerance values were not within the limits. In order to reduce the number
 of defects, initially Taguchi’s Design of Experiments (DOE) is used to find the better set of
 process parameters that minimizes the tolerance values. Then Response Surface Methodology

                                                 362
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

(RSM) is employed to find out the mathematical model, which relates the control parameters
with the performance measure. This model obtained is used as the objective function for per-
forming minimization of absolute value, in Genetic algorithm (GA). On academic interest, Par-
ticle Swarm Optimization (PAO) is also used for minimization and results obtained by GA and
PSO are compared. Finally confirmation experiments are conducted for the results obtained, with
95% Confidence level. Based on the above observations, suggestions have been made on setting
the parameters to improve the quality.

Key Words: Design of Experiments, Response Surface Methodology, Genetic Algorithm.

         I.     INTRODUCTION
         The goal of any industrial experimentation in manufacturing is to devise the ways of mi-
nimizing the deviation of a quality characteristic from its target value. This can be done only by
identifying factors which impact the quality characteristic in question and by changing the ap-
propriate factor levels so that the deviations are minimized and the quality characteristic is on
target. The classical methods for DOE developed by R.A. Fisher, include a full variety of statis-
tical design techniques based on Latin squares. A major problem with Fisher’s approach in man-
ufacturing industry is the time and cost required to learn and use it. Taguchi’s approach utilizes
Robust design and is applied to a range of problems. The Response Surface Methodology (RSM)
is a collection of mathematical and statistical techniques that are useful for modeling and analy-
sis of problems in which a response of interest is influenced by several variables and the objec-
tive is to optimize this response. Here in this problem, the objective is to find the level if cutting
speed, work head speed and the fine feed rate that minimizes the ovality tolerance values. In
most RSM problems, the form of relationship between the response and the independent va-
riables in unknown. Thus, the first step in RSM is to find a suitable approximation for the true
functional relationship between the response and independent variables. It the response is well
modeled by a linear function of the independent variables, then the approximating function is the
first order model. The regression equation takes the form

             y = b0+b1x1+b2x2+……..+bpxp.

Where b0, b1, b2……..bp, called the regression coefficients, are determined from the data.

       II.     LITERATURE REVIEW

      Optimization technique has focused the interest of many researchers during the last 15
years. Following are the overview of the relevant work done earlier related to the problem identi-
fied and the methodology to be adopted to solve the chosen problem for this work. It gives the
description of literature reviewed from the various research papers published in international and
national journals. [1] optimization metal cutting process in manufacturing industries for increas-
ing demand of quality product in the market. In present scenario optimization methods in metal
cutting processes, considered to be a vital tool for continual improvement of output quality in
products and processes include modeling of input–output and in-process parameters relationship

                                                 363
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

and determination of optimal cutting conditions. Authors analyzed several optimization tech-
niques, incorporates the use of one or more of the existing modeling and optimization tech-
niques, making the framework a unified and effective means. [2] A new optimization technique
based on genetic algorithms (GA) for the determination of the cutting parameters in machining
operations. In metal cutting processes, cutting conditions have an influence on reducing the pro-
duction cost and time and deciding the quality of a final product. The authors formed new me-
thodology as the modification of recommended cutting conditions obtained from a machining
data, learning of obtained cutting conditions using neural networks and the substitution of better
cutting conditions for those learned previously by a proposed GA. The authors used several op-
timization technique and they concluded that genetic algorithm-based approach in complex ma-
chining systems and automated process planning system and compared with a number of other
emerging optimization-techniques. [3] a genetic algorithmic approach for optimization of surface
roughness due to use of highly automated machine tools in the industry, manufacturing requires
reliable models and methods for the prediction of output performance of machining processes.
The prediction of optimal machining conditions for good surface finish and dimensional accura-
cy plays a very important role in process planning. In this work deals with the study and devel-
opment of a surface roughness prediction model for machining mild steel, using Response Sur-
face Methodology (RSM) and the experimentation was carried out with TiN-coated tungsten
carbide (CNMG). The authors concluded that genetic algorithm program gives minimum and
maximum values of surface roughness and their respective optimal machining conditions.

[4] a multi-objective genetic algorithm approach for optimization on surface grinding operations
to optimize grinding conditions, viz. wheel speed, workpiece speed, depth of dressing and lead of
dressing, using multi-objective function model with a weighted approach for surface grinding
process. The procedure evaluates the production cost and production rate for the optimum grind-
ing condition, subjected to constraints such as thermal damage, wheel wear parameters, machine
tool stiffness and surface finish. Genetic algorithm optimum results for production cost, surface
finish and material removal rate compared with quadratic programming technique. [5] the ma-
chining process is evaluated in terms of machining rate and surface finish produced. Higher ma-
chining rate and better surface finish are desirable for better performance of any machining
process. Comprehensive qualitative and quantitative analysis of the material removal mechanism
and subsequently the development of analytical model(s) of material removal (MR) are neces-
sary for a better understanding and to achieve the optimum process performance. In use of ad-
vanced machining processes incurs high investment, operating, maintenance, tooling and other
costs. The authors described that in the absence of analytical models, optimum selection of
process parameters requires extensive experimentation, which is time and money consuming. [6]
A new approach for the optimal sub-division of the depth of cut is presented using a genetic al-
gorithm. The total production-cost minimization is achieved by adding the minimum costs of the
individual rough passes and the finish pass. The selection of the depth of cut during optimization
in multi-pass turning is an important activity, along with the selection of the speed and feed. Au-

                                               364
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

thors proposed GAs always yields production-cost values that are less than, or equal to, the val-
ues obtained using other methods. [7] computer vision techniques to inspect surface roughness of
a workpiece under a variation of turning operations. The authors used digital camera for captur-
ing surface image of the workpiece and then the feature of the surface image is extracted and al-
so authors used method called self-organizing adaptive modeling as polynomial network for con-
structing the relationships between the feature of the surface image and the actual surface rough-
ness under a variation of turning operations. As a result, the surface roughness of the turned part
can be predicted with reasonable accuracy if the image of the turned surface and turning condi-
tions. [13] a real coded genetic algorithm optimization of machining parameters in order to ob-
tain better surface quality. Since, surface quality is one of the important indicators of customer
requirement in machining process. There are various methods available for optimization prob-
lems viz calculus based, dynamic programming, artificial neural network, simulated annealing,
etc. the authors concluded from experimental analysis that surface roughness decreases with in-
crease in cutting speed and decrease in feed rate. [14] a multi-objective optimization technique,
based on genetic algorithms. In any optimization procedure identifying the output parameter is of
chief important. Many of authors have determined the optimization in single objective approach-
es only and it has limited value to fix optimal cutting conditions. The objectives are maximiza-
tion of tool life and maximization of production rate using genetic algorithm method. The pro-
posed genetic algorithm was implemented in C++. By using of Pareto frontier graphics, several
different situations may be considered, facilitating the choice of right parameters for any condi-
tion. The proposed micro-GA has obtain several, uniformly distributed points, in order to arrange
the Pareto front, at a reasonably low computational cost. Cost analysis can complement the Pare-
to front information, and it helps the decision-making process.

       III. METHODOLOGY
A. PROCESS FLOW CHART
              The flow chart below shows the series of operations done in the solution metho-
dology




                                     Figure 1 Process Flow Chart

                                                365
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

IV. EXPERIMENTAL DATA
A. MACHINE SPECIFICATION

       Initially the process is studied before experimentation. The specifications of the Puma
2000Y CNC Turn mill center machine are as follows. The photograph of the machine in which
the experiment is conducted is given below in figure 2




                          Figure 2 Puma 2000Y CNC Turn mill center

B. QUALITY CHARACTERISTIC

      The measurements associated with the ring crowning tolerances are detailed below. The
measurements are taken using AE GOETZE face profile (OD) measuring instrument.




                               .

                      Figure 3 Tool Holder Ring 2D dimensional drawing

       The tool holder ring 2D drawing was as shown above. The back of the ring has to be
pushed against the two stops till the end. Once when the end is reached the measurement starts.
The dimension sensor probe travels along the width of the ring. The measurements are taken at
the specified gauging levels (shown in the figure 3) as the probe traverses in the upward direc-
tion. As the probe returns to its original position the readings are listed in a CRT terminal inter-
faced with the instrument. Finish on the periphery should be barrel honed and ground. The
crowning tolerance is 0.002-0.006 mm over a gauge width of 2 mm.


                                                366
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

C. MATERIAL COMPOSITON OF THE RING

       The table 1 shown below is the material Composition of 20MnCr5

                                 Table 1 Material Composition

                                 COMPOSITION FOR 20MnCr5
                               Description       Specification
                                Cast Iron      Major composition.
                                 Carbon            3.7 – 3.8 %
                                 Silicon           2.5 – 2.7 %
                               Manganese:            0.60%
                                 Sulphur         0.05 – 0.08%
                                Phosphors         0.35 - 0.45%
                                Vanadium        Less than 0.1%
                               Chromium               0.2%
                                 Copper               0.1%



D. CONTROL PARAMETERS OF FACTOR DEVELOP
i) CONTR0L PARAMETERS
           In this problem, there are three control parameters (factors). These parameters are se-
lected after a detailed study. The control factors are as follows.
        Work head speed (m/min)
        Depth of cut (mm)
        Fine feed rate (mm/rev)
       The cutting speed is the major factor in a grinding operation. Secondly the work head
speed is one of the major factors which should not be left behind. Among the rough feed rate and
fine feed rate, the fine feed is a major factor affecting the crowning tolerance. Hence these
process parameters are chosen as the control factors.

ii) FACTOR LEVELS
               The next problem is fixing the levels. Currently grinding is done with a cutting
speed of 1500 m/mm, work head speed of 160 rpm and a fine feed rate of 0.7 mm/min. The table
3 shows the various control factors and their levels. The cutting speed is varied between 1400
and 1550 m/min. Since the minimum cutting speed specified in the manual for effective grinding
is 1400 m/min, the cutting speed is decreased only up to 1450 m/min. The depth of cut is varied
between 0.4 to 0.9. The fine feed rate is varied between 0.12 and 0.05 mm/min.




                                               367
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

                               Table 2 Control factors and their levels


                    CONTROL FACTORS            LEVEL 1         LEVEL 2      LEVEL 3

                     Spindle speed (m/min)        1400           1450           1550

                      Depth of cut (mm)           0.4            0.5            0.9

                    Fine feed rate (mm/rev)       0.12           0.08           0.05




D. ORTHOGONAL ARRAY
        In order to conduct an experiment with three control factors and three levels, an L9 (33)
orthogonal array is formed. The array is called orthogonal because the levels of various factors
are balanced and can be separated from the effects of the factors within the experiment. Here, L
represents Latin square, 9-represent number of experiments,
3 represents number of levels and 3 on the superscript represent number of factors.

E. EXPERIMENTAL DESIGN SETUP
        Now the experiment is designed by substituting corresponding values of various factor
levels in the above table. The table 3 gives the experimental design setup for which the experi-
ments are conducted.
                                      Table 3 Experimental design


                                              CONTROL FACTORS
                        Expt
                         No       Cutting Speed    Work head         Fine feed
                                    (m/Min)        speed (rpm)     rate(mm/Min)
                          1            1400              0.4             0.12
                          2            1400              0.5             0.9
                          3            1400              0.9             0.05
                          4            1450              0.4             0.08
                          5            1450              0.9             0.12
                          6            1450              0.5             0.05
                          7            1550              0.5             0.12
                          8            1550              0.4             0.05
                          9            1550              0.9             0.08

E. SELECTION OF RESPONSE
        From the tabulation, it can be understood that at the gauging level (h25), the values ob-
tained are within the tolerance limits, but at the gauging level (h27), some of the values are not
within the tolerance limits. Hence, the crowning tolerance values at gauging level (h27) are taken
as the response.

                                                  368
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

F. NEED FOR DOE
In the CNC Turn Mill center, for each cycle of operation, for rings are fed. Especially, the fol-
lowing problems found while machining the component.
        Runout
        Faceout
        Ovality
At present the component was produced by using the following parameters.
        Depth of cut = 0.05mm
        Feed rate      = 0.12mm/rev
        Cutting speed = 1450 rpm.
As the rejection rate was steady and increasing, an attempt to use Taguchi method to find the
best set of combinations for which the values are within the tolerance limits.
G.NEED FOR PARAMETER DESIGN
        For the problem as stated above, system design cannot be applied. Tolerance design also
becomes costlier. Hence Parameter design is adopted to find a solution for a problem of the
above type. The necessary output required is between 0.005µm at the gauging level (h27), and
some of the values obtained are out of this range. Thus the objective is minimize the mean ob-
tained to a nominal of 0.01
H. EXPERIMENT SETUP USING TAGUCHI DESIGN
i) Rejection chart while using DOE. Table no 5 shown below observation Data chart.
                                Table 5 Observation Data Chart




                                   Figure no 4 Observation Chart

                                               369
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

I. SIGNAL TO NOISE RATIO
        The S/N ratio is an objective performance measure. The S/N ratio is an evaluation of sta-
bility of performance of an output characteristic. The S/N ratio measures a level of performance
and the effect of noise factors on performance.
S/N ratio,                                   ŋ = 10 log10 [µ2 / σ2]
                                Mean response is given by, µ = 1/n *
Sensitivity to noise is given by σ2 = 1/n *

By substituting the values obtained from experimentation in the above formulae, the following
table 6 is arrived

                        Table No 6 Experimental values and parameters




   i)      S/N RATIO MEAN LEVEL

The response table 7 gives the ranking of the importance of the factors on the response variable,
but it does not indicate the relative magnitude of importance. To fine the magnitude of impor-
tance of various factors.

                                 Table 7 S/N Ratio mean level




                                              370
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

   ii)     RESPONSE GRAPHS




                               Figure No 5 S/N Chart for cutting speed




                                Figure No 6 S/N Chart for Fine feed rat

J. INTERPERTATIONS
                From the above graphs, the factor level with maximum S/N ratio is chosen as the
optimum combination for obtaining the required quality characteristic [7]. The optimum set of
control factors found out by employing Taguchi method is listed below.
Work head speed        = 1450 rpm
Depth of cut = 0.4mm
Fine feed rate = 0.12 mm/rev

K. COMMENTS ON RESULTS

        The values of the control factors arrived as a result of Taguchi’s DOE are not satisfacto-
ry, since they did not produce the desired response. The best combination arrived as a result of
Taguchi method is already present in the experimental design setup as experiment number 1. For
the purpose of confirmation, once again an experiment was conducted with the above combina-

                                                 371
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

tion, but he result obtained was the same as that of the results obtained during the initial experi-
mentation. As stated earlier, the desired response is nominal value of 10.5 µm

M. NEED FOR NON-TRADITIONAL OPTMIZATION

        It was decided to apply optimization techniques, in order to find out the possible ways of
minimizing the response by finding better combination, if any. After a detailed study, it was de-
cided that application of non-traditional optimization algorithms was one of the probable ways of
finding a solution for a problem of the above kind. In order to apply optimization algorithms, a
mathematical model of the process is required. Usually Central Composite Design (CCD) of ex-
periments is recommended for obtaining accurate results with RSM. But in this work, CCD is
not employed as it is mostly applied for processes with wide variations in their response [3]. For
a process with minimal variations in its response, the results obtained by modeling with Taguchi
DOE values are quite acceptable. The mathematical model is arrived from the values of DOE
using a technique called Response Surface Methodology (RSM).

V. REGRESSION COEFFICIENT EVALUATION

A. INTERPRETATION OF COEFFICIENTS

        The Estimated regression coefficients in uncoded ubuts guveb avive are tge coefficients
of the various factors in the equation that relates the control factors and the response. The equa-
tion should be interpreted as shown below.
          Ct = 0.1633x1 – 0.978167x2 + 54.0917x3 – 8.4*10-5x1 2 – 13.5625x3 2
        The equation shown above is the mathematical model of the process obtained by using
RSM. The values of R-sq and R-sq (adj) for this model are given below.
        r-sq = 99.8% and R-sq(adj) = 98.1%
where, Ct = Crowning tolerance (µm)
        x1 = Cutting speed(m/min)
        x2 = Depth of cut (mm)
        x3 – Fine feed rate (mm/min)

VI. RESULTS AND DISCUSSION

A.GA result
        The following is the automatically generated GA output file for the above problem ob-
tained by using ‘Export to workspace’ command in GA tool box and by typing ‘gaprob-
lem’,’gaoptions’ and ‘garesults’ in command window of MATLAB 7.0. The GA fitness distribu-
tion plot is given in figure no 7
 Gaproblem =
        Fitnessfcn: @algo
            Nvars:3
          Options : [1x1 strut]
>>gaoptions
Gaoptions =

                                                372
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

           PopulationType: ‘doubleVector’
             PopInitRange: [2x3 doub;e]
           populationSize: 12
              Elitecount: 2
    CrossoverFraction: 0.8000
    MigrationDirection: ‘forward’
    MigrationInterval: 20
    MigrationFraction: 0.2000
          Generations: 300
             TimeLimit: Inf
           FitnessLimit: -Inf
          StallGenLimit: 50
         StallTimeLimit: 20
       InitialPopulation: [ ]
  InitialScoress: [ ]
                PlotInterval: 1
           CreationFcn: @gacreationuniform
        FitnessScaligFcn: @fitscalingrank
           SelectionFcn: @selectionroulette
           CrossoverFcn: @crossoverscattered
           MutatuibFcn: { [1x1 function_handle] [0.5000] }
             HybridFcn: [ ]
              Display: ‘off’
              PlotFcns: { [ 1x1 function_handle] }
            OutputFcns: [ ]
                 Vectorized: ‘off’
>> garesults
garesults =
                 X: [1.4501e +003 169.9772 0.5002]
            Fval: 3.5294
   exitmessage: ‘Optimization terminated: stall generations limit exceeded.’
Output:[1x1struct]




                                               373
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME




                             Figure No 7 Fitness value distribution chart

       Thus the parameters obtained by Taguchi DOE are fine tuned to obtain still more opti-
mized parameters with best results within the range. The results of GA are as follows. The
crowning tolerance value is given by the sum of the minimal absolute value obtained above and
the nominal value
Ovality tolerance, Ct        = 14.0294 µm
Cutting speed, x1            = 1450.1183 mm/min
Depth of cut, x2              = 0.95241mm
Fine feed rate, x3            = 0.5002 mm/rev

B.RESULTS FOR 95 % CONFIDENCE LEVEL
       The MINITAB 15.0 output for the 95 % confidence level is given below.
Predicted Response for New Design Points Using Model for r
                          Table 8 Results for 95% confidence level

                     Point       Fit          SE fit    95%                    CI
                       1      14.0293       0.0219493 13.9595               14.0992
                       2      14.0249       0.0220154 13.9548               14.0950

C.CONFIDENCE INTERVAL
i) FOR GENETIC ALGORITHM
       Confidence interval, Cl = 0.0698
       µconfirmation = 14.0249
Confidence level: 13.9595 ≤ 14.0293 ≤ 14.0992

ii) CONFIRMATION TEST

       With the values obtained by optimization, a confirmation test is performed whenever De-
sign of Experiments (DOE) is carried out, confirmation tests should be performed to check the
correctness and reproducibility of the predicted mean and factor levels in the experimental de-
                                                374
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

sign setup. Since the best combination arrived by Taguchi DOE is already present in the one of
the experiments, the confirmation tests are done only for the values arrived by GA
        While giving input values for the process parameters in the machine, decimal values of
cutting and work head speeds cannot be given as input so they are rounded off to the nearest in-
teger. Similarly, fine feed rate values are accepted only till a single decimal point. Since the val-
ues given by GA and PSO are too close, only one confirmation test is done for the values given
below
Cutting speed = 1450 m/min.
Depth of cut = 0.89 mm.
Fine feed rate = 0.5 mm/rev.

D.VERIFICATION OF 95% CONFIDENCE LEVEL
       The results obtained are compared with the 95% Confidence Interval obtained earlier.
The value obtained is within the 95% confidence level. This proves the correctness of the Expe-
rimental Design, Modeling and Optimization done earlier.
GA: 13.9595 ≤ 14.0 ≤ 14.0992
                               Table No: 9 Final validation Data




                                 Figure 8 Final validations Chart



                                                375
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

E. IMPLEMENTATION
        Based on the above observations, suggestions have been made on setting the parameters
to improve the quality. The confirmed factor levels for the optimum response are then imple-
mented in the company for production. Currently, the machining is done only according to this
set of process parameters. The results obtained are good. The amount of rejections has also re-
duced drastically. The process is further studied for the feasibility of extrapolation of the results
obtained, but the probability of getting a good response prediction is low for the present model.
Hence the machining is carried out with the above set of process parameters itself. The actual
inputs and response are given below.
        Cutting speed = 1450 m/min.
        Depth of cut = 0.89 mm.
        Fine feed rate = 0.5 mm/rev.
        Ovality tolerance = 0.005 µm

VII. CONCLUSION

        The problem related to tolerance arising in a special turning process known as CNC turn-
ing done in compression rings of a tool holder is identified. Initially Taguchi’s Design of Expe-
riments (DOE) has been used to find the better set of process parameters that minimizes the to-
lerance values. In order to find better combination, Response surface Methodology (RSM) is
employed. The mathematical model is obtained by using regression analysis in MINITAB 15.0
which serves as the objective function for optimization with Genetic Algorithm (GA). Finally
confirmation experiments are conducted for the results obtained, with reasonable confidence lev-
el. Based on the above observations, the parameter settings as suggested by promising. This
leads to quality improvement at no additional cost. The main advantage of this solution metho-
dology is that it can be applied to any process in any branch of engineering. Thus it can be as-
sured that by following this method, the terms defects and inspection can be eradicated from the
dictionary of manufacturing.

REFERENCES

   1. Mukherjee.I, and Kumar.P, “A review of optimization techniques in metal cutting
      processes”, Computers & Industrial Engineering, Vol.50, 2006, pp.15–34.
   2. Cus.F, and Balic.J, “Optimization of cutting process by GA approach”, Robotics and
      Computer Integrated Manufacturing, Vol.19, 2003, pp 113–121.
   3. Suresh.P.V.S, Rao.P.V and Deshmukh.S.G, “A genetic algorithmic approach for optimi-
      zation of surface roughness prediction model”, International Journal of Machine Tools &
      Manufacture, Vol.42, 2002, pp. 675–680.
   4. Saravanan.R Asokan.P, and Sachidanandam.M,“A multi-objective genetic algorithm
      (GA) approach for optimization of surface grinding operations”, International Journal of
      Machine Tools & Manufacture, Vol.42,2002,pp.1327–1334.
   5. Jain.N.K and JainV.K, “Modeling of material removal in mechanical type advanced ma-
      chining processes: a state-of-art review”, International Journal of Machine Tools & Man-
      ufacture, Vol .41 ,2001,pp. 1573–1635


                                                376
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976 –
6340(Print), ISSN 0976 – 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) © IAEME

   6. Bhaskara Reddy.S.V , Shunmugam.M.S and Narendran.T.T, “Optimal sub-division of the
       depth of cut to achieve minimum production cost in multi-pass turning using a genetic al-
       gorithm”, Journal of Materials Processing Technology, Vol 79 ,1998 ,pp.101–108
   7. Lee.B.Y and Tarng.Y.S, “Surface roughness inspection by computer vision in turning op-
       erations”, International Journal of Machine Tools & Manufacture, Vol.41, 2001, pp.
       1251–1263.
   8. Dr. S.S.Mahapatra, Amar Patnaik, Prabhina Ku Patnik, July 2006, “Parametric Analysis
       and Optimization of Cutting Parameters for Turing Operations based on Taguchi Me-
       thod”, International Conference on Global Manufacturing and Innovation.
   9. Aman Aggarwal and Hari Singh, December 2005. “Optimization of machining tech-
       niques – A retrospective and literature review”, Sandhana, Volume 30, Part 6.
   10. Jayant A, Kumar V, March 2008, “Prediction of Surface roughness in CNC Turing Op-
       eration using Taguchi Design of Experiments”, The Journal of Institution of Engineers
       (India), Vol88, 19-25.
   11. Gaitonde V N, Karnik S R, March 2007, “Genetic Algorithm based Burr size Minimiza-
       tion in Drilling using Artificial Neural Network Models”, The Journal of Institution of
       Engineers (Incis), Vol 87, 25-30
   12. Dr R C Paul, Dr P Asokan, Dr J Jerals, September 2008, “Multi-Objective Facility
       Layout Problem using Particle Swarm Optimization”, The Journal of Institution of En-
       gineers (India), Vol 89, 30-35.
   13. Srikanth.T and Kamala.V, “A Real coded genetic algorithm for optimization of         cut-
       ting parameters in turning”, International Journal of Computer Science and Network Se-
       curity, Vol.8, 2006, pp.189-193.
   14. Sardinas.R.Q, Santana.M.R and Brindis.E.A, “Genetic algorithm-based multi-Objective
       optimization of cutting parameters in turning processes”, Engineering Applications of Ar-
       tificial Intelligence, Vol.19, 2006, pp.127–133.
   15. Belavendram N, “Quality by Design”, Prentice Hall publications, 42-51, 84-92, 144-
       159, 274-279.
   16. Douglas C Montgomery, Fifth edition, “Design and Analysis of Experiments”, John
       Wiley and Sons Inc., 427-430, 455-466.
   17. Angela Dean and Daniel Voss, 2006, “Design and Analysis of Experiments”, Springer
       Texts in Statistics, 547-549, 569-571.
   18. M.S. Mahajan, 2006, “Statistical Quality control”, Dhanpat Rai & Co. Ltd, 179- 230.




                                             377

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:15
posted:12/6/2012
language:Latin
pages:16