International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) Lean T hinking journal homepage: w ww.thinkinglean.com/ijlt Prediction and Control of Cutting Tool Vibration in Cnc Lathe with Anova and Ann S. S. Abuthakeer * P.V. Mohanram G. Mohan Kumar Department of Mechanical Engineering, Department of Mechanical Engineering, Park College of Engineering and PSG College of Technology, Coimbatore, PSG College of Technology, Coimbatore, Technology, Avinashi Road, Kaniyur, 641 004, India 641 004, India Coimbatore 641 659, India E-mail Adres: [email protected] A B S T R A C T K E Y W O R D S Cutting tool vibration, Machining is a complex process in which many variables can Passive damping pad, Data acquisition, deleterious the desired results. Among them, cutting tool ANOVA, ANN. vibration is the most critical phenomenon which influences A R T I C L E I N F O dimensional precision of the components machined, functional Received 20 February 2011 behavior of the machine tools and life of the cutting tool. In a Accepted 23 February 2011 machining operation, the cutting tool vibrations are mainly Available online 24 February 2011 influenced by cutting parameters like cutting speed, depth of cut and tool feed rate. In this work, the cutting tool vibrations are controlled using a damping pad made of Neoprene. Experiments were conducted in a CNC lathe where the tool holder is supported with and without damping pad. The cutting tool vibration signals were collected through a data acquisition system supported by LabVIEW software. To increase the buoyancy and reliability of the experiments, a full factorial experimental design was used. Experimental data collected were tested with analysis of variance (ANOVA) to understand the influences of the cutting parameters. Empirical models have been developed using analysis of variance (ANOVA). Experimental studies and data analysis have been performed to validate the proposed damping system. Multilayer perceptron neural network model has been constructed with feed forward back-propagation algorithm using the acquired data. On the completion of the experimental test ANN is used to validate the results obtained and also to predict the behavior of the system under any cutting condition within the operating range. The onsite tests show that the proposed system reduces the vibration of cutting tool to a greater extend. ________________________________ * Corresponding Author S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) 1. Introduction The modern trend of machine tool development is required to produce precise, accurate and reliable product which are gradually becoming more prominent features. The monitoring of manufacturing processes and equipment conditions are the essential part of a critical strategy that drives manufacturing industries towards being leaner and more competitive (Al-Habaibeh and Gindy, 2000; Frankowiak et al., 2005). In a machining operation, vibration is frequent problem, which affects the machining performance and in particular, the surface finish and tool life. Severe vibration occurs in the machining environment due to a dynamic motion between the cutting tool and the work piece. In all the cutting operations like turning, boring and milling, vibrations are induced due to the deformation of the work piece, machine structure and cutting tool. In a machining operation, forced vibration and self-excited vibration are identified as machining vibrations. Forced vibration is a result of certain periodical forces that exist within the machine, bad gear such as drives, misalignment, and unbalanced machine tool components, etc. Self-excited vibration is caused by the interaction of the chip removal process and the structure of the machine tool, which results in disturbance in the cutting zone. The self-excited vibration affects the production capacity, reliability and machining surface quality (Luke and Joseph, 2001). Researchers have been trying to demonstrate tool condition monitoring approach in an end- milling operation based on the vibration signal collected through a low-cost, microcontroller-based data acquisition system (Julie and Joseph, 2008). Today, the standard procedure adopted to avoid vibration during machining is by careful planning of the cutting parameters and damping of cutting tool. The methods adopted to reduce vibration are based on experience as well as trial and error to obtain suitable cutting parameters for each cutting operation. Many sensors were used for tool condition monitoring system namely, touch sensors, power sensors, vibration sensors, temperature sensors, force sensors, vision sensors, flow sensors, acoustic emission sensors and so on (Jemielniak, 1999; Dimla, 2000; Xiaoli, 2002). Tool wear sensing techniques are broadly classified into two categories: direct and indirect as shown in Table 1. The direct tool wear monitoring methods can be applied when cutting tools are not in contact with the work piece. However, direct methods of measuring tool wear have not been easily adaptable for shop floor application. Indirect tool sensing methods use relationship between cutting conditions and response of machining process which is a measurable quantity through 2 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) sensor signals output (such as force, acoustic emission, vibration, or current) and may be used to predict the condition of the cutting tool (Kurada and Bradley, 1997). Table 1. Tool wear sensing methods Direct methods Indirect methods Electrical resistance Torque and power Optical measurements Temperature Radio-active Vibration & acoustic emission Contact sensing Cutting forces & strain measurements These indirect methods are used extensively by various researchers and detailed analyses have been carried out in the past two decades. Nowadays, availability of computational power and reliability of electronics help in the development of a reliable condition monitoring system by using indirect methods. However, a problem in TCM system is selection of proper sensor and its location. The sensors have to be placed as close as possible to the target location (close to the tool tip) being monitored. It is interesting to note that an indirect TCM system consists of four steps: (i) collection of data in terms of signals from sensors such as cutting force, vibration, temperature, acoustic emission and/or motor current, (ii) extraction of features from the signals, (iii) classification or estimation of tool wear using pattern recognition, fuzzy logic, neural networks, or regression analysis, and (iv) development of an adaptive system to control the machining process based on information from the sensors (Kakade et al., 1995). The researchers determine mean amplitude of vibration using accelerations in both directions along the axes (Kirby and Chen, 2007). A computer program was developed using Visual Basic programming language in order to analyze one and two degree of freedom of machine tool chatter vibrations (Choudhury et al., 1996). On-line vibration control system for turning operation uses a closed-loop feedback circuit which measures the relative vibration between the cutting tool and the work piece (Taskesen, 2005). There have been many investigations on vibration prediction and controlling based on periodic measurements of various machining conditions using accelerometer and active vibration controller. Two generic techniques used for solving these vibration control problems are modifying stiffness or the fundamental natural frequency of the specified components/subsystems, and their damping (Eugene, 2007). 3 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) Damping is the capacity of a mechanical system to reduce the intensity of a vibratory process. The damping capacity can be due to interactions with outside systems or due to internal performance- related interactions. The damping effect for a vibratory process is achieved by transforming (dissipating) mechanical energy of the vibratory motion into other types of energy, most frequently heat, which can be evacuated from the system. Effects of damping on performance of mechanical systems are due to reduction of intensity of undesirable resonances; acceleration decay(settling) of transient vibration excited by abrupt changes in motion parameters of mechanical components; prevention or alleviation of self-excited vibrations; prevention of impacts between vibrating parts when their amplitudes are reduced by damping; potential for reduction of heat generation, and thus for increase in efficiency due to reduced peak vibratory velocities of components having frictional or micro impacting interactions; reduction of noise generation and of harmful vibrations transmitted to human operators and more. Passive damping is now the major means of suppressing unwanted vibrations. The primary effect of increased damping in a structure is a reduction of vibration amplitudes at resonances, with corresponding decreases in stresses, displacements, fatigue and sound radiation. Designed in- passive damping for any structure is usually based on one of four damping mechanisms: viscoelastic materials, viscous fluids, magnetics or passive piezoelectric (Johnson, 1995). Based on the literature survey, approximately 85 percent of the passive damping treatments in actual applications are based on viscoelastic materials, with viscous devices being the second most actively used (the use of viscous devices is greater for isolation and shock). In the present work attempt has been made to predict and suppressing the vibration level of cutting tool in CNC lathe, by using passive damping pad of viscoelastic material of neoprene. The study is extended to analyze the influence of cutting parameters on the tool vibration during machining. The results obtained have shown the effectiveness of the proposed solution that have been analyzed and discussed in detail. 4 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) 2. Experimentation Workpiece C utting tool Tail stock Impact hammer support PXI 1042 Q with monitor Figure 1. Experimental setup The experimental setup is shown in Figure 1. It includes a CNC -Galaxy –MIDAS-0 turning center, a CCGT-09T30FL (Taegu Tec) turning insert, tool holder SCLC L2020 K09 T3(Taegu Tec), a work piece (Al 6063 aluminum, Diameter 38 mm x 70mm length) without any cutting fluid. The tool is instrumented with two accelerometers (Bruel & Kjaer 9.88mV/g- type 4517). The accelerometers signals are taken to NI PXI 1042 – Q Data Acquisition Card system using LabVIEW software. The vibration data was captured by Data Acquisition Card system. This system included hardware selection, circuit design and implementation, hardware interface, circuit troubleshooting, filtering, computer software programming, system integration, and testing in real CNC turning processes. The following three sections describe the development of the hardware system, software system, and integrating and testing of the data acquisition system along with the vibration data analyses. 2.1 Hardware system Vibration signals are important for monitoring tool condition in turning process. Accelerometers were mounted in the cutting tool, one in the tangential direction of the tool holder and the other one was placed in the axial direction of the tool holder for measuring vibration amplitude in terms of accelerations (g-levels). A computer code has been developed in LabVIEW for data acquisition, data storage and display. Fast Fourier Transform (FFT) computation algorithm was included in the computer program to extract the vibration amplitude in the time and frequency domain, which will be explained in software development section.  Accelerometers: Converts the physical acceleration into a voltage signal.  Signal conditioning circuit: Amplifies the voltage signal and improves the resolution. 5 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011)  Personal computer: Runs the program, stores and display at any desired instant of time. 2.2 Software system The software in this system consists of the following components.  An NC program that directs the CNC turning machine to cut the work piece.  Vibration data analysis and Fast Fourier Transform (FFT) analysis. Main objective of the research work is to monitor the vibration level of cutting tool. So it is assumed that the condition of the machine and its components is good in all other aspects such as foundation of the machine, rigidity of the machine components (such as bed, spindle, tail stock etc.) and so on. The simplest vibration analysis is conducted through collecting the “overall” vibration amplitude Root Mean Square (RMS) value and plotting the vibration data in time domain and frequency domain. The “overall” signal represents the total energy content of all vibration sources at all frequencies. 2.3 Integration and testing of the data acquisition system The Integration and testing of the data acquisition system is shown in figure 1. When tested in a machining work piece, the sensor was protected to prevent any interference caused due to machining chips. 2.3.1 – Modal Analysis – With and without damping pad Any physical system can vibrate, the frequencies at which vibration naturally occurs, and the modal shapes which the vibrating system assumes are properties of the system, and can be determined using modal analysis. Modal analysis is frequently utilized to abstract the modal parameters of a system, including natural frequencies, mode shapes and modal damping ratio. Since these parameters depend only on the system itself but dominate the response of the system to excitations, modal analysis is the fundamental response analysis and has therefore gained increasing attentions. The free vibration tests were carried out for the given cutting tool without any damping pad. In the free vibration analysis test, an impact hammer (PCB-086C03) was used to excite the cutting tool. An accelerometer was mounted on the tool holder and interfaced with a data acquisition card and LabVIEW software to record the response of the cutting tool in time and frequency domains. The impact pulse indicating the magnitude of input force was generated by the impact hammer. 6 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) The frequency domain response was obtained by using signal analyzer available in sound and vibration toolkit of Lab VIEW. The response of the tool holder captured in time and frequency domains as shown in Figure 2. o Accelerati n(g) Time (Sec) o Accelerati n(g) Frequency (Hz) Figure. 2 Vibration signal for response of the accelerometer of free vibration test (without damping pad) From the Figure 2, it is evident that, the fundamental natural frequency of the tool is about 3.4 kHz, acceleration of 12.5g and it takes about 0.95 seconds to settle down. The damping ratio is calculated using Bandwidth method and the value is obtained as 0.0149 ( =  -/ 2 ). The 2 1 n free vibration tests were carried out for the given cutting tool using damping pad made of neoprene. The experimental modal analysis was repeated for the damping condition. The response of the cutting tool is shown in Figure 3, from the figure, the fundamental natural frequency of the cutting tool were found to be about 2.150 kHz, and it takes about 0.4 seconds to settle down. The damping ratio was calculated as 0.06976. on(g) Accelerati Time (Sec) n(g) o Accelerati Frequency (Hz) Figure 3. Vibration signals for response of the accelerometer of free vibration test (with damping pad) 7 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) 2.3.2 Dynamic Analysis - without damping pad The vibration analysis was done without any damping pad under actual machining conditions. In this analysis, a set of experiments were conducted with the cutting tool held in the tool holder as shown in Figure 4.a. The two accelerometers mounted in both the tangential and axial directions were used to collect the vibration signals. The LabVIEW acquires the vibration signals and stored the signals continuously frame by frame at every stage of cutting in on-line. The vibration data given in Figure 4 b is obtained while turning with cutting speed of 250m/min, depth of cut of 0.5mm and feed rate of 0.1mm/rev. The dynamic response of accelerometer without any damping pad is given in table 2. Accelerometer in tangential direction Accelerometer in axial direction Turning insert Figure 4a. Cutting tool without damping pad Figure 4b. Cutting tool vibration signals without damping pad 2.3.3 Dynamic Analysis with damping pad In this set of experiments, the cutting tool is clamped with damping pad made of rubber material called neoprene is shown in Figure 5 a. Same set of experiments were repeated as given in previous section and vibration signals were collected with the use of damping pad. The cutting tool vibration signals with damping pad at cutting speed of 250 m/min, depth of cut of 0.5 mm and feed rate of 0.1 mm/rev is shown in Figure 5 b. The dynamic response of accelerometer with damping pads is given in Table 2. 8 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) Accelerometer in tangential direction Accelerometer in axial direction Damping pad (Neoprene) Turning insert Figure 5a. Cutting tool with damping pad Figure 5b. Cutting tool vibration signals with damping pad 3. Experimental Design Experimental design approach is selected for the investigations of varying three controllable parameters at three levels, since 3k factorial design is efficient to study the effects of two or more factors. Without loss of generality three levels of factor are referred as low, intermediate and high and levels are designed by digits 0.1 and 2. Each treatment combination in the 3k design is denoted by k digits where the first digit indicates a level of factorial A (cutting speed), B (depth of cut), indicates the level of factorial second and C (Feed) indicates the level of three. These factors as well as their levels identified are given in Table 3. Table 2. Input parameters and dynamic response of accelerometers with and without damping pad 9 S. S. ABUTHAKEER, P. V. MOHANRAM, G. MOHANKUMAR /International Journal of Lean Thinking Volume 2, Issue 1 (June 2011) Table 3. Identified control factors and their levels Variables or parameter Parameter designation Level 1 Level 2 Level 3 Cutting speed(m/min) A 150 200 250 Depth of cut(mm) B 0.5 0.75 1 Feed(mm/rev) C 0.1 0.2 0.3 10