This action might not be possible to undo. Are you sure you want to continue?

M. S. Patil1

Department of Mechanical Engineering, Gogte Institute of Technology, Belgaum 590008, Karnataka, India e-mail: mspatil_git@rediffmail.com

1

Introduction

Jose Mathew

Professor e-mail: josmat@nitc.ac.in

P. K. Rajendrakumar

Professor e-mail: pkrkumar@nitc.ac.in Department of Mechanical Engineering, National Institute of Technology, Calicut 673601, Kerala, India

Sumit Karade

Department of Mechanical Engineering, Gogte Institute of Technology, Belgaum 590008, Karnataka, India e-mail: sumitkarade@gmail.com

The presence of defect in the bearing (outer race, inner race, or ball) results in increased vibrations. Time domain indices such as rms, crest factor, and kurtosis are some of the important parameters used to monitor the condition of the bearing. Radial load and operating speed also have an important role in bearing vibrations. The interaction between the defect size, load, and speed helps to study their effect on vibrations more effectively. Response surface methodology (RSM) is a combination of statistical and mathematical techniques to represent the relationship between the inputs and the outputs of a physical system. But so far, the literature related to its application in bearing damage identiﬁcation is scarce. The proposed study uses RSM to study the inﬂuence of defect size, load, and speed on the bearing vibrations. Kurtosis is used as response factor. Experiments are planned using Box Behnken design procedure. Experiments are performed using 6305 ball bearings and the results have been presented. MINITAB statistical software is used for analysis. It is seen from the analysis of the experimental results that the defect size, interaction effect of defect size and load, and interaction effect of defect size and speed are signiﬁcant. Response surface method using Box Behnken design and analysis of variance has proved to be a successful technique to assess the signiﬁcant factors related to bearing vibrations. ͓DOI: 10.1115/1.4002520͔ Keywords: condition monitoring, kurtosis, response surface methodology

1 Corresponding author. Contributed by the Tribology Division of ASME for publication in the JOURNAL OF TRIBOLOGY. Manuscript received September 17, 2009; ﬁnal manuscript received July 19, 2010; published online October 7, 2010. Assoc. Editor: Shuangbiao ͑Jordan͒ Liu.

Rolling element bearings are one of the most essential parts of rotating machinery. A machine could be seriously jeopardized if defects occur in the bearings during service. Early detection of the defects, therefore, is crucial for the prevention of damage and a total failure of the associated large system. Different methods are used for detection and diagnosis of bearing defects; they may be broadly classiﬁed as vibration and acoustic analysis, temperature measurements, and wear debris analysis. Among these, vibration analysis is the most widely used technique. Vibration signature based diagnostics are mainly concerned with the extraction of those features from a diagnostic signal, which can be related to a good or a defective state of the component. Vibration produced due to defects is sensed using the velocity transducer or accelerometers. The signals collected from the rolling contact element are analyzed. Various signal processing techniques involving time, frequency, and statistical methods have been used to detect and check the progress of the incipient fault. The extraction of meaningful information from this data is challenging and, therefore, it calls for different approaches to analyze the data. Time wave form analysis includes the visual inspection of the time history of the vibration signals, time wave form indices, probability density function, and probability density moments. A time wave form index is a single number calculated based on the raw vibration signal and used for trending and comparisons. The indices include peak value, mean value, rms ͑root mean square͒ value, and peak-to-peak amplitude. The probability of ﬁnding the instantaneous amplitude value from a vibration signal within a certain amplitude range can be represented by probability density function. The shape of probability density function will be similar to a Gaussian or normal probability distribution for a bearing in good condition. Alfredson and Mathew ͓1͔ reported obtaining a new Gaussian distribution for some damaged bearing. Dyer and Stewart ͓2͔ discovered that the probability density of acceleration of a bearing in the perfect condition has a Gaussian distribution where as a damaged bearing results in non-Gaussian distribution with dominant tails. Even moments ͑second and fourth, standard deviation, and Kurtosis͒ are proportional to the spread of the distribution. The most useful is the kurtosis, which is sensitive to the impulsiveness in the vibration signal and, therefore, sensitive to the type of the vibration signal generated in the early stage of a rolling element bearing fault. Dyer and Stewart proposed the use of kurtosis for bearing fault detection. Martin and Honarvar ͓3͔ used kurtosis and skew is used to compare the signal obtained from the damaged and undamaged bearing. Tandon and Nakra ͓4͔ extracted other features such as rms level and crest factor for diagnosing the damaged bearing. Many researchers ͓5–11͔ have shown the effectiveness of kurtosis in defect detection. Heng and Nor ͓12͔ discussed the use statistical parameters derived from the Beta distribution function. The results are compared using rms, crest factor, and kurtosis. A paper by Kar and Mohanty ͓13͔ discusses the possibility of application of the Kolmogorov–Smirnov ͑KS͒ test in diagnosing rolling element bearing defect identiﬁcation. Theoretical predictions based on experimental observations mark the essence of useful research. Proper use of statistical methods greatly improves the efﬁciency of the experiments and helps to draw meaningful conclusions from the experimental data. There are two basic aspects of concern in scientiﬁc experimentation: the design of the experiment and the statistical analysis of the data. Successful experimentation requires knowledge of the important factors that inﬂuence the output. Design of experiments ͑DOE͒ ͓14͔ helps to determine the factors, which are important for explaining a process variation. Interactions are the driving forces in many processes and proper understanding of the process may be difﬁcult or impossible if important interactions remain undetected. DOE also helps to understand how the inﬂuence facOCTOBER 2010, Vol. 132 / 044505-1

Journal of Tribology

Copyright © 2010 by ASME

Downloaded From: http://tribology.asmedigitalcollection.asme.org/ on 04/28/2013 Terms of Use: http://asme.org/terms

then the responses can be represented as a function of levels and variables.e. The cutoff value often used is 0. It involves performing an experiment. and ﬁnally. The symbols b0. when the other ones are kept constant. Factors are the variables of interest that inﬂuence the response ͑output͒ of a system. Kurtosis is considered to be a good parameter to measure defectiveness in the bearings but irregularity in variation of kurtosis makes it difﬁcult for judging.1 Basic Deﬁnitions. Y = f ͑X1u. There is an interaction ͑dependence͒ between the variables when the effect of one factor depends on the behavior of another. It is common to declare a result signiﬁcant if the p-value is less than 0. ANOVA ͓14͔ gives a summary of the main effects and interactions. The present work uses the Box Behnken design. The RSM is widely used as an optimization. . N represents N observations in the factorial experiment and Xiu represents the level of ith factor in uth observation.. The p-value represents the probability of making a type-I error or rejecting the null hypothesis when it is true. bi. 1 Box–Behnken design 2 Introduction to Response Surface Method Eu measures the experimental error of the uth observation. The designs of the RSM are those in which problems are modeled and analyzed. Effect of a factor is deﬁned as change in the response produced by a change in the level of the factor. and also allows for the study of interactions between the factors.05.3 Response Surface Methodology. and ball. the regression coefﬁcients. experiments were planned and analyzed using DOE approach to study the inﬂuence of the operating conditions on kurtosis.tors interact with the system. Factorial designs allow for the simultaneous study of the effects that several factors may have on a response. and Taguchi techniques can be used for planning the experiments. Time domain indices such as rms. and kurtosis are generally used for statistical analysis. reject the null hypothesis when the p-value is less than 0. . Figure 1 shows the Box–Behnken design for three factors in coded form.org/terms . Here. optimal values of the dependent variable. This is consists of correlating the k variables put into action through a second-degree polynomial expression of the following form: −1 k Y = b0 + ͚ik=1bixi + ͚ik=1biixi2 + ͚ik=1 ͚ j=i+1bijxix j 2. The designs were developed by the combination of two level factorial designs with incomplete block designs. and speed on the bearing performance. The design is obtained by the combination of 22 design with a balanced incomplete block design having three treatments and three blocks. OCTOBER 2010 ͑2͒ where Y is the dependent variable and xi is the factor or variable with which we wish to correlate it. in these problems. The advantages of these designs include the fact that they are all spherical designs and require facTransactions of the ASME Downloaded From: http://tribology. The experiments are performed on bearings having defect on outer race. The RSM designs are classiﬁed into central composite design and Box Behnken design. 2 . varying the levels of the factors simultaneously rather than one at a time. The p-value ͓14͔ in the ANOVA analysis helps to determine which effects ͑factors and interactions͒ are statistically signiﬁcant. . the inputs or the variables that inﬂuence the behavior of the system are called factors or variables and the outputs represent the response that generates the system under the causal action of the factors. 2. p-values are often used in hypothesis tests where you either accept or reject a null hypothesis. The results explained are in context with the experiments performed for this study. 2. load. Effect. Response surface methodology ͓14–16͔ is a collection of mathematical and statistical techniques that are useful for modeling and analysis of problems in which a response of interest is inﬂuenced by several variables.2 Analysis of Variance (ANOVA). those in which the response variable is measured for all the possible combinations of the levels chosen of the factors.e.asmedigitalcollection. . it is considered necessary to explore the relationship between the factor and dependent variable within the experimental region and not only at the borders.. crest factor. Factorial design. X2u. . The smaller the p-value. Levels are called as “low” and “high” and deﬁned on a coded scale from “−1” and “+1” ͑the low and high levels of the factors͒. If all the input parameters represent quantitative variables. i. Speciﬁc values for the factors at which the experiment is performed. For the present study kurtosis of the signal has been used as the response parameter.1 Description of the Method of the Response Surface. Response surfaces are recommended for these types of factorial designs for their effectiveness and quick execution. and improvement technique for processes based on the use of factorial designs—that is. The response surface methodology emerged in the 1950s ͓17͔ within the context of chemical engineering in an attempt to construct empirical models that are able to ﬁnd useful statistical relationships between all the variables making up an industrial system. response surface method ͑RSM͒. The experimental values are adjusted to the above equation by a polynomial regression and the usual statistics can be used to determine the goodness of the ﬁt. The function f is called the response function. inner race. after the signiﬁcant factors affecting the response have been identiﬁed. and bij are constants. The application of the RSM becomes indispensable when. . Methods such as factorial design. The expression contains a ﬁrst-degree term that represents a linear relationship considered as the principal.05. Box and Behnken ͓18͔ introduced designs for three level factors that are widely used in response surface methods to ﬁt second-order models to the response.asme.3. . development. the response of interest is inﬂuenced by different variables. the smaller is the probability that you would be making a mistake by rejecting the null hypothesis. The residual 044505-2 / Vol. another term in which the variables cross each other to represent the inﬂuence of some over others. . 2. Xku͒ + Eu ͑1͒ where u = 1 . Fig. These designs are referred to as Box–Behnken designs. Hence. and the p-value. a second-degree term that reﬁnes the previous one and gives maximums and minimums—i. 132. Level.org/ on 04/28/2013 Terms of Use: http://asme. SKF 6305 ball bearing having artiﬁcially induced defect of different sizes have been used for the experiments. Factors. The present work explores the application of RSM method using the time domain values of the bearing vibration signal for analyzing the effect of defect size. The main effect of a factor is deﬁned as the variation in response caused by a change in the level of the factor considered.05.

9232 2.008a 0.523 Coefﬁcient 5.631 2.5 mm artiﬁcially induced on outer race.9998 3.5 1.6011 3. and display the time domain signal are installed in the computer used for this work.5 1 1 1 1 1 1 1 Shaft speed ͑S͒ ͑rpm͒ 800 800 1200 1200 1000 1000 1000 1000 800 1200 800 1200 1000 1000 1000 Load ͑L͒ ͑kN͒ 1 1 1 1 0 0 2 2 0 0 2 2 1 1 1 Kurtosis ͑K͒ Outer 3.9495 2.6957 2.638 Fig.85042 0.72991 0.56958 0.000a 0.17927 0.004a 0. a single-row deep groove ball bearing ͑SKF 6305͒. 132 / 044505-3 Downloaded From: http://tribology. Precaution was taken during ﬁtting and removal of the bearing. and the ball.02500 Signiﬁcant.859 0.786 2. Electric discharge machine ͑EDM͒ was used to create the defect.43999 −0. 3 Experimental Details 3.41775 −0.013a 0.017 0.660 2.5119 4.1 Experimental Setup. 2 „a… Experimental setup and „b… hydraulic loading arrangement tors to be run at only three levels. separately.7804 3.37490 8.7116 3.300 Coefﬁcient 3. A piezo-electric accelerometer with a sensitivity of 50 mV/g is used to measure the vibrations.020a 0.5 mm. It is calculated using the formula ͓6. Table 1 Factors and levels Levels/coded Factors ͑unit͒ Defect size ͑mm͒ Shaft speed ͑rpm͒ Load ͑kN͒ Notation D S L −1 0.07025 −0.335 0. xi is the amplitude of vibration.009a 0.4978 2.436 0. It is consists of a shaft supported on two bearings and driven by a variable speed motor.10195 0.11430 0.646 0.2 Factors and Their Level. Signals were ﬁltered using Butterworth ﬁlter.27825 p-value 0.39661 0. inner race. The signiﬁcant terms in the model were found by analysis of variance at 5% level of signiﬁcance ͑95 % conﬁdence level͒.145 0.001a 0.4171 2.3304 2.asme.9395 2.5 0.9481 3.53945 4.org/terms . and is the standard deviation. and load.7͔ Kurtosis͑K͒ = 1 N xi − ͚ N i=1 ͩ ͪ 4 ͑3͒ where N is the number of data points. Table 1 shows the factors and their levels in coded and actual values. Kurtosis is considered as the response parameter. store it. it is important that the correct mounting method be used. Journal of Tribology OCTOBER 2010.0116 3.000a 0. and 1.008a 0.4116 3. the factors that are of interest are defect size.5 1200 2 4 Results The MINITAB statistical package was used to analyze the experimental data and response parameters.07885 0.5 1.104 3.000a 0. Yet another advantage of these designs is that there are no runs where all factors are at either the +1 ͑highest value͒ or −1 ͑lowest value͒ levels. The vibration signals are sampled at 9 kHz with a sampling size of 4096 ͑212͒ samples.7055 2.22015 0.0349 3.33282 −0.35920 0.3054 4.38838 0.9443 3.695 2.709 0.034 3.2741 3.0656 Ball 2.5472 3.012 0. The loading arrangement is placed between these two bearings.003a Ball defect Coefﬁcient 7.11079 0. before analysis.5472 3.5 0. and ball.414 3. The gap between the bearing housing cover and the body was maintained the same for all the experiments.5 1.354 3. available in MATLAB.org/ on 04/28/2013 Terms of Use: http://asme. 2͑b͒. 3.5 0.90775 0.02835 0.488 0. Vol. 1 mm.312 Inner 2.246 0.3304 2.904 3. shaft speed.00347 1.asmedigitalcollection. The experimental setup used for this study is shown in Fig.5259 3.000a 0. the inner race. Table 2 depicts the design matrix and the kurtosis ͑average of replicates͒ for defect on the outer race. Table 3 shows the regression coefﬁcients and p-value.03882 0.041a 0. The accelerometer is connected to the charge ampliﬁer.94200 p-value 0.3227 3.9125 3.358 0.09852 −0.0823 3.5 1.37475 0. is placed on the nondrive end of the shaft and a double-row selfaligning ball bearing is placed on the drive end side. It is assumed that all the bearings have the same surface properties.0823 2.5 800 0 0 1 1000 1 +1 1.000 0. is the mean deviation. The experiments have been performed on separate test bearings having defects of sizes 0.0546 3.08300 −0. A hydraulic loading arrangement is shown in Fig. The regression coefﬁcients are obtained using the coded units.185 0.7175 3.9572 3. The relevant hardware and the software required to acquire the data. The bearing was ﬁtted using hot oil bath technique and was removed using the bearing puller.Table 2 Design matrix „uncoded factors… Defect size ͑D͒ ͑mm͒ 0. the output of which is connected to a computer. Feeler gauge was used to measure the gap.6428 2. For this study. The test bearing. It is mounted on the housing of the test bearing.01225 3.043a 0.4062 3. 2͑a͒.123 0. Table 3 Regression coefﬁcients and p-value Outer race Term Constant D S L D2 S2 L2 DϫS DϫL SϫL a Inner race p-value 0. In order for a bearing to function properly.49543 0.

load.13709 Residual error 5 0.73 208. It is also observed from the table that the shaft speed. It is observed in Figs.11% 4.002 0.00250 F-value 81.00530 0. Table 3.020 0. Figures 5͑a͒–5͑c͒ show the variation in the kurtosis due to change in the defect size.32017 Square 3 0. 4͑b͒͒. It is clear from Table 3 that the defect size and the interaction between the defect size and the speed is signiﬁcant ͑p-value Ͻ0. 4 „a… Contour plot and „b… surface plot for outer race defect 044505-4 / Vol. It is seen in Fig.43345 1. OCTOBER 2010 Transactions of the ASME Downloaded From: http://tribology. respectively.05͒.Table 4 ANOVA table for outer race defect Source DF SS MS 0.00500 Total 14 3.00717 0.87 p-value 0. 3 Interaction plots „data means… for kurtosis for outer race defect 4. the increase in the defect size results in an increase in the kurtosis and hence the bearing vibrations.05.asmedigitalcollection.32% R2-͑adj͒ = 98.44383 Interaction 3 0.14794 0. This indicates that the variables ͑predictors͒ excellently explain the amount of variation in the observed value of the kurtosis.02652 Lack-of-ﬁt 3 0. 3͑a͒ and 3͑b͒ that kurtosis at higher defect size and a higher load is high. load.000 0. It is observed that with the increase in the defect size.269 Regression 9 3. Fig. which shows the estimated regres- Fig. In Fig. 5 Interaction plots „data means… for kurtosis for inner race defect all values of the loads.05 for ͑D ϫ S͒ ͑Table 3͒ indicates that the interaction effect between defect size and speed is not signiﬁcant. which inﬂuences the response.04570 0.10672 0. it is clear that the kurtosis increases with an increase in the defect size for Fig. The p-value greater than 0.92761 R2 = 99.000 0. The p-value closer to zero for the interaction effect in ANOVA ͑Table 5͒ indicates that interaction effect is most signiﬁcant.9932 and 0. 4. and the speed. Figures 6͑a͒ and 6͑b͒ show the contour plot and the surface plot for the interaction between the defect size and the shaft speed for inner race defect. The curved contour lines and the twist in the response surface indicate that there exists a signiﬁcant interaction between the defect size and the load.62 2. 132. The interaction effect between the defect size and load is also shown using the contour plot ͑Fig. Table 4 shows the result of ANOVA of the kurtosis for outer race defect.68 27.1 Outer Race Defect.org/terms .2 Inner Race Defect.org/ on 04/28/2013 Terms of Use: http://asme. which indicate that the inﬂuence of linear and square effect related to the defect size are more signiﬁcant. The interaction plots are shown below. Regression analysis for the case of outer race defect ͑Table 3͒ indicates that the individual effect of defect size is statistically signiﬁcant. 3͑a͒. The p-values corresponding to linear and square term are closer to zero in the ANOVA analysis ͑Table 4͒. the kurtosis increases. For the bearing with inner race defect.9811. and interaction effect between the defect size and load also have signiﬁcant contributions since the corresponding p-value is Ͻ0. 3͑c͒ that there is a slight variation in the kurtosis for all defect sizes at lower speeds but the inﬂuence of speed on kurtosis is less at higher speeds. The analysis shows that for a bearing having outer race defect.3 Ball Defect. 4͑a͒͒ and surface plot ͑Fig.90 8.90109 Linear 3 3. The R2 and R2-͑adj͒ for the outer race are 0.02152 Pure error 2 0.asme. the interaction between the defect size and the speed also plays a signiﬁcant role on bearing vibrations. respectively.

000 0.org/ on 04/28/2013 Terms of Use: http://asme. 8͑a͒͒ and surface plot ͑Fig.71% R -͑adj͒ = 99.112 Regression 9 2. 7͑a͒–7͑c͒ shows that the effect of the defect size on the kurtosis is more as compared with the effect of the load and the speed.00048 Total 14 2. Comparison of Figs.040 0. It is seen in Fig.86728 Linear 3 3.013 0.001267 0. The value of the kurtosis with the increase in the load remains almost constant.87887 2 2 R = 99.39 2.14 6.value 193. The chances of contact may be more if the defect size is more.244777 0.003298 0.14 23. It can be observed that the kurtosis does not change much with the increase in the speed when the defect size is large. Fig. Vol.Fig.024962 0. Therefore.00634 Lack-of-ﬁt 3 0. it is clear that ball defect kurtosis mainly depends on the defect size. 8͑b͒͒.211 Fig.89 p-value 0.org/terms .176 0.429698 0. 7 defect Interaction plots „data means… for kurtosis for ball Regression 9 3. 132 / 044505-5 Downloaded From: http://tribology. the kurtosis increases. The ANOVA of kurtosis for the ball defect shown in Table 6 also predicts that the square term is the most signiﬁcant. 6 „a… Contour plot and „b… surface plot for inner race defect Table 5 ANOVA table for inner race defect Source DF SS MS 0.008658 0. it is clear that with the increase in the defect size and load.20% Table 6 ANOVA table for ball defect Source DF SS MS 0. It may be noted that the rate of increase of the kurtosis is more for larger defect sizes.005753 0.217344 0.07489 Residual error 5 0. 8 Contour plot and surface plot for ball defect size and load Journal of Tribology OCTOBER 2010.000 0. the kurtosis increases. 9͑a͒ and 9͑b͒.65203 Interaction 3 0.20299 Linear 3 2.13 p-value 0.14036 Square 3 0.09171 Square 3 0.20933 2 2 R = 99. It is also observed that the interaction of defect size with the load and speed is also signiﬁcant.asmedigitalcollection. 5 Conclusions The following conclusions can be drawn from this study.002318 0. It can also be seen from Fig.16% sion coefﬁcients and the p-value for the ball defect also. which means that with the increase in the defect size.029317 0. 7͑a͒ that with the increase in the defect size the kurtosis increases.00170 Total 14 3.5 mm.13 8.77 3. The p-value for these terms is less than the signiﬁcance level ͑0.48 93. From the contour plot ͑Fig.77 10.032 0. kurtosis becomes less sensitive to the load. indicates that the square effect of defect size is the most signiﬁcant factor.01159 Lack-of-ﬁt 3 0.asme.000 0.05͒.000240 F.001952 0.08795 Residual error 5 0. Similarly observations can be made from Figs.83 6.02334 Interaction 3 0.00586 Pure error 2 0.002 0. 8͑a͒ that as the defect size approaches 1.00989 Pure error 2 0. It is unlikely that the defect on the ball makes contact with the raceway during every revolution of the ball.70% R -͑adj͒ = 99.007780 0.000848 F-value 185.

53͑13͒. A. DETC2009-87333. Int. “Application of Statistical Moments and Spectral Analysis in Condition Monitoring of Rolling Element Bearings. M.. Li. P. ͓6͔ Kiral. 31–36. K.. 533–541. Montgomery.” Mech. For the outer race defect. and Kurtaran. pp. the factors inﬂuencing the kurtosis are the defect size and the speed. 1999... pp. Austral. T. Nurul Amin.. 15͑5͒. Williams... C. and Dalpiaz. Billington. 102–107.. 271–282.” Technometrics.asme.. 1–45. 25͑204͒.. T. H. Erzurumlu. R... Kurtosis increases with the increase in the defect size. “Time Domain Methods for Monitoring the Condition of Rolling Element Bearings. the increase in defect size and load results in an increase in kurtosis. For the inner race defect. ͓5͔ Igarashi. R. and Honarvar. “Detection of Rolling Element Bearing Damage by Statistical Vibration Analysis. pp. 1982. 979–993..” European Journal of Scientiﬁc Research. Karacay. and Stewart. 2007. G. 21͑3͒. 1951..org/terms .asmedigitalcollection.” Journal of Institution of Engineers ͑India͒—Mechanical Engineering Division.. pp. Soc. 2001.” Journal of the Royal Statistical Society. 994–1001. M. Malagò. 170. S. J. pp.. D. H. T. Wiley. Eng. pp.. pp.. 439–454. and Nakra. 13. “Some New Three Level Designs for the Study of Quantitative Variables. M.” Tribol. Signal Process.. Process. Liang. 2009. J..” J. 044505-6 / Vol... “Application of KS Test in Ball Bearing Fault Diagnosis.. B. T. “Rolling Element Bearing Diagnostics in Run-to-Failure Lifetime Testing.. Mech. 455–475. M. and Mohanty.. N. Mathew. and Wilson. ͓7͔ ͓8͔ ͓9͔ ͓10͔ ͓11͔ ͓12͔ ͓13͔ ͓14͔ ͓15͔ ͓16͔ References ͓1͔ Alfredson. R. pp. 484–492. pp...” ASME J. 28͑4͒. A. H.. and Karagulle. and Lajis. Kar. 211–226.. K. “Studies on the Vibration and Sound of Defective Rolling Bearings ͑First Report: Vibration of Ball Bearings With One Defect͒. ͓2͔ Dyer. JSME. S..” Bull. W. 2005... 1993. Heng.. pp. Small variation in the kurtosis value is observed at lower speed but at higher speeds. T. C. Box.” J. D. A.. pp. 1995. Inst. pp. H. Inﬂuence of defect size is more signiﬁcant.. G. Sound Vib.” Appl. F.” J. 132. Des. pp. Int. J. Acoust. M. Sci. 42͑6͒. Design and Analysis of Experiments. OCTOBER 2010 Transactions of the ASME Downloaded From: http://tribology. H. 73. R. ͓4͔ Tandon.. For the ball defect. T. Ribadeneisa. pp. and Mathew.. A. “Tool Life Prediction by Response Surface Methodology in End Milling Titanium Alloy Ti-6Al-4V Using Uncoated WC-Co Inserts.org/ on 04/28/2013 Terms of Use: http://asme. K. “Simulation and Analysis of Vibration ͓17͔ ͓18͔ Signals Generated by Rolling Element Bearing With Defects. 36. D. 12͑1͒. and Kurfess. 2004. C. E.. S.. “Experimental Diagnostics of Ball Bearings Using Statistical and Spectral Methods. Y. 667–678.. and Nor. pp. Singapore. Shiroishi. there is not much variation in the kurtosis.. G. B.” Appl.. P. W.Fig. ͓3͔ Martin. 9 Contour plot and surface plot for ball defect size and shaft speed ͑a͒ ͑b͒ ͑c͒ ͑d͒ Response surface method using Box–Behnken design and ANOVA has proved to be a successful technique to assess the signiﬁcant factors related to bearing vibrations.. Technol. “Detection of Defects in Rolling Element Bearings by Vibration Monitoring. 1978. E. 2009. Mucchi. 100͑2͒. 44.. D. Z. pp. and Nizami. 5th ed. “Application of Statistical Moments to Bearing Failure Detection. 67–77. L. Mater. and Rajendrakumar. defect size is more dominant.” International Journal of COMADEM. S. 2. Oktem.. J. “On the Experimental Attainment of Optimum Conditions. Patil. P. H. Mohd Radzi. “Vibration Analysis for Bearing Outer Race Condition Diagnostics. and Kurfess. 229–235.. Syst. J.” J. and Hamada. 11–16. Kurtosis becomes less sensitive to load and speed for larger defect sizes. “Condition Monitoring and Diagnosis in Heavy-Duty Wheels: A First Experimental Approach. 2009. 2009. Box. “Application of Response Surface Methodology in the Optimization of Cutting Conditions for Surface Roughness. pp. R. M... “Statistical Analysis of Sound and Vibration Signals for Monitoring Rolling Element Bearing Condition. C. M. 269.” Tribol... 21–46. E. 1985.. Braz. 1998. 1960. 2003. X.. Acoust. Ginta.. Danyluk. Mech. T. and Behnken.. 10͑2͒.... 836–843.” ASME Paper No. B.

- Verbal-and-Non-Verbal-Reasoning-by-r-s-Agarwal.pdf
- The Constitutuion of India
- Optimal Operation of Multi-reservoir System by Multi-elite Guide Particle
- Extracting Elitepairwiseconstraintsforclustering
- view_ga
- Fitness Function Calculation
- Finite element analysis instructions manual
- fgdddddddddddddddddddddddddddddddddddddddddddddd gggggggggggggggggggggggggggggggggggggg
- concept of base Ayres Differential Equations
- gggggggggggggggggggggggggggggggggggggggggggggggg
- hjkgbhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
- basic concept of Machine Design 01
- genetic algorithm for volume optimization
- Origin Universe
- Genetic algorithm.ppt
- NTN_TR71_en_P008
- fundamentals of Cad cam cim
- Theory of machine_Geneva Mechanism
- Geneva Mechanism
- Ref 12_A Review of Optimization Techniques in Metal Cutting Processes
- artificial bee colony algorithm
- Chapter 8 - Fault Identification and Monitoring
- Strength of Materials
- Knowledge Based System for Design

Sign up to vote on this title

UsefulNot usefulsfdhgsdhdfsdfhdshdfhsdhdfhshsd gsddddddddddddddddddddddddddddddddddddds

sfdhgsdhdfsdfhdshdfhsdhdfhshsd gsddddddddddddddddddddddddddddddddddddds

- R-Practicals Part I
- Ind. Project in R Analysis
- documents.mx_ps-work-booksolution.doc
- Abid ResearchMethods Vadodra
- BUS 308 Week 3 ANOVA and Paired T-test
- Determining the Factor Settings for Optimal Download Speed at Home
- Ch 10 (Formula Sheet )
- Hypothesis Test Format
- Price
- Price
- Cross Tabs
- Spss ancur
- Hypothesis Testing
- 241086544 Industrial Training Presentation Nbc
- Performing t Tests Using STATA
- Backtesting - Campbell Harving Nov 22 2013
- statistik
- Harvey and Liu
- H834P019
- description
- description
- Spss.anova
- The Ball Displacement in Angular Contact Ball Bearings for Static Load
- Eckel_Grossman_TeamIdentity
- Lab Exercises Answer
- Tutorial Solutions Week 7-2012_2013
- Chi Square Tests
- description
- jeas_0414_1059.pdf
- Bearing en a017-020
- sgdfsssgdhgsdgfdgs

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd