Surface Roughness Modeling in Turning: Amit Samdariya

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 45

Project Code: P04

SURFACE ROUGHNESS MODELING IN TURNING

Amit Samdariya
(2007ME10471)

Dhiraj Kumar Singh


(2007ME10486)

Dr. Sudarsan Ghosh Dr. Jayanta Kumar Dutt


(Supervisor) (Supervisor)

Dr. Sunil Pandey


(Examiner)

Department of Mechanical Engineering


INDIAN INSTITUTE OF TECHNOLOGY
Delhi
April, 2011
Certificate

This is to certify that this project entitled “SURFACE ROUGHNESS MODELING IN


TURNING” submitted by Amit Samdariya (2007ME10471) and Dhiraj Kumar Singh
(2007ME10486) in requirement for MED411 course is a bona fide work carried out by them.
They have worked under our supervision and guidance and have fulfilled the requirements for
submission of this project report.

The work contained in this report, to the best of our knowledge and belief, has not been
submitted in part or full to any other university or institution for award of any degree/diploma.

Dr. Sudarsan Ghosh Dr. Jayanta Kumar Dutt


(Supervisor) (Supervisor)

i
Acknowledgement

We would like to express our gratitude to Dr. Sudarshan Ghosh and Dr. Jayanta Kumar Dutt for
their supervision, advice, and guidance from the very early stage of this project as well as giving
us extraordinary experiences throughout. Above all and the most needed, they provided us
unflinching encouragement and support in various ways. Their experience in the field of
production technology and mechanical vibrations inspired us & enriched our growth as students.

We are especially grateful to Mr. Andriya Narasimhulu for guiding us in different matters
regarding the topic. He helped us a lot in gathering different information, collecting data and
guiding us from time to time in making this project. In spite of his busy schedule, he gave us
different ideas in making this project unique. The experimentation on the whole would not have
been possible without his guidance and immense patience.

We would like to acknowledge the help of the lab assistants Mr. Subhash Chand, in-charge of
Metrology Lab and Mr. Dhulichand Meena, in-charge of Machine Tools Lab for helping us out
in various jobs.

Finally we would like to thank IIT Delhi and especially the Mechanical Engineering Department,
in particular, for providing us all the facilities needed for our project.

Amit Samdariya
(2007ME10471)

Dhiraj Kumar Singh


(2007ME10486)

ii
Abstract

Surface Roughness is one of the most crucial requirements in machining operations as it affects
the interaction of a mechanical component to its environment. An optimal selection of cutting
conditions is the key to achieve the desired surface finish. Unfortunately, roughness is often
determined by a complex interplay of cutting parameters, material properties as well as the
resultant vibrations. No wonder, even today, it remains difficult and expensive to predict and
control surface quality and is a matter of concern to the scientific community. This project
focuses on utilizing finite element and mathematical modeling techniques for predicting surface
profile of a turned workpiece. A finite element model of the workpiece has been developed in
ANSYS which has been subjected to experimentally recorded cutting forces. The transient
response of the tool has been used to generate the surface profile of the workpiece using
mathematical modeling in MATLAB. A statistical reconstruction of the random forces would
make this model self-sufficient in roughness prediction, without requiring to conduct the turning
experiments and using sophisticated instruments.

Keywords: surface roughness; tool vibration; roughness modeling; turning; finite element model

iii
Contents

Certificate ........................................................................................................................................ ii

Acknowledgement .......................................................................................................................... ii

Abstract .......................................................................................................................................... iii

Contents ......................................................................................................................................... iv

List of Figures ................................................................................................................................ vi

List of Tables ................................................................................................................................ vii

Nomenclature & Abbreviations ................................................................................................... viii

Chapter 1. Introduction ............................................................................................................... 1

Chapter 2. Literature Review & Problem Statement .................................................................. 3

Chapter 3. Experimental Setup and Design ................................................................................ 5

3.1 Experimental Setup .......................................................................................................... 5

3.2 Experimental Design ........................................................................................................ 7

3.3 Experimental Observations .............................................................................................. 8

3.3.1 Observations regarding Cutting Forces .................................................................... 8

3.3.2 Observations regarding Surface Roughness ........................................................... 11

Chapter 4. CAD Modeling of Tool ........................................................................................... 12

Chapter 5. Finite Element Modeling of Tool ........................................................................... 14

Chapter 6. Surface Profile Generation ...................................................................................... 17

6.1 Generation of tool tip‟s trajectory .................................................................................. 17

6.2 Generation of Surface Profile ......................................................................................... 19

6.3 Simulation Results.......................................................................................................... 22

Chapter 7. Conclusions and Discussion ................................................................................... 25

iv
7.1 Conclusions .................................................................................................................... 25

7.2 Limitations of model ...................................................................................................... 26

Chapter 8. Scope for future work ............................................................................................. 27

REFERENCES ............................................................................................................................. 32

APPENDICES .............................................................................................................................. 33

Appendix-I: MATLAB program surface profile ...................................................................... 34

Appendix II: User Manuals ....................................................................................................... 36

1. System Requirements for Matlab R2008b & ANSYS 12 .......................................... 36

2. User manual for ANSYS ............................................................................................ 36

v
List of Figures
Figure 3.1: Schematic of the experimental setup prepared at Machine Tools Lab, IIT Delhi ........ 5
Figure 3.2: Cutting tool insert geometry (www.kennametal.com) ................................................. 6
Figure 3.3: CNC lathe, Talysurf surface profiler and Dynamometer ............................................. 6
Figure 3.4: Variation of cutting forces with feed rate at different cutting speeds .......................... 9
Figure 3.5: Variation of cutting forces with speed at different feed rates .................................... 10
Figure 3.6: Variation of surface roughness with (A) feed rate (B) cutting speed ......................... 11
Figure 4.1: Solid model of the tool holder and insert assembly (trimetric view) ......................... 13
Figure 4.2: Engineering drawings of the tool assembly (all dimensions in mm) ......................... 13
Figure 5.1: SOLID285 element geometry (source: ANSYS) ....................................................... 15
Figure 5.2: Meshing of the tool model with refinement towards tip ............................................ 15
Figure 5.3: Displacement response of tool tip in 3 directions (X‟, Y‟, Z‟) .................................. 16
Figure 6.1: Shape of tool tip (A) pointed, used for generating surface profile and (B) circular,
actually used in experiments ......................................................................................................... 17
Figure 6.2: (A) Rotating workpiece, (B) Motion of tool relative to workpiece............................ 18
Figure 6.3: Helical path of tool tip relative to workpiece ............................................................. 19
Figure 6.4: Generation of surface profile; black circles are points from helical trajectory of tool
tip, circles are intersection points between cutting edges at two subsequent tool locations ......... 20
Figure 6.5: Deteremination of average roughness (Ra) by calculation of enclosed areas ............ 21
Figure 6.6: Comparison of experimental and simulation roughness with varying feed rate ........ 23
Figure 8.1: Relative frequency histograms of forces in 3 directions Fx, Fy and Fz ..................... 27
Figure 8.2: Relative frequency histogram for four sections of force data to check stationarity ... 28
Figure 8.3: Normal probability plot experimental cutting force Fy (dashed line is for normal
distribution while other is for recorded forces)............................................................................. 29
Figure 8.4: Regenerated force vs time plot (120m/min speed and 0.12mm/rev feed) .................. 30
Figure 8.5: Amplitude spectra for (A) experimental force (B) regenerated force ........................ 31

vi
List of Tables
Table 2.1: Gantt Chart..................................................................................................................... 4
Table 3.1: Experimental design ...................................................................................................... 7
Table 3.2: Mean and standard deviation (SD) values of experimental cutting forces .................... 8
Table 3.3: Experimentally measured surface roughness............................................................... 11
Table 5.1: Mechanical properties of tool holder and insert (www.memsnet.org) ........................ 14
Table 6.1: Surface roughness results predicted from the simulation ............................................ 22
Table 6.2: comparison of experimental and simulated surface roughness values ........................ 22
Table 8.1: comparison of statistical properties to check stationarity ............................................ 29
Table 8.2: Comparison of skewness and kurtosis between normal distribution and measured
force .............................................................................................................................................. 30

vii
Nomenclature & Abbreviations

d Depth of cut (mm)

f Feed rate (mm/rev)

Fx Force measured in the X direction by dynamometer (N)

Fy Force measured in the Y direction by dynamometer (N)

Fz Force measured in the Z direction by dynamometer (N)

R Radius of workpiece (mm)

Ra Mean roughness (μm)

v Cutting velocity (m/min)

X’ Nodal displacement along X‟ axis

Y’ Nodal displacement along Y‟ axis

Z’ Nodal displacement along Z‟ axis

ω Angular speed (radians/sec)

γ Rake angle (degrees)

viii
Chapter 1. Introduction

Surface roughness is used as the critical quality indicator for the machined surfaces and has
influence on several properties such as wear resistance, fatigue strength, coefficient of friction,
lubrication and corrosion resistance of the machined parts. It is an important parameter to
evaluate the productivity of machine tools as well as machined components. Achieving the
desired surface quality is of great importance for the functional behavior of the mechanical parts.

In today‟s manufacturing industry, special attention is given to dimensional accuracy and surface
finish. Thus, measuring and characterizing the surface finish can be considered as the predictor
of the machining performance. Turning is the primary operation in most of the production
processes in the industry. The turning operation produces the components, which have critical
features that require specific surface finish. Due to inadequate knowledge of the complexity and
factors affecting the surface finish in turning operation, an improper decision may cause high
production costs and low machining quality. A proper selection of cutting tools and process
parameters for achieving high cutting performance in a turning operation is a critical task. Hence
a proper estimation of surface roughness has been the focus of study for several years.

The surface finish can be characterized by a number of parameters, average roughness (Ra),
maximum peak to valley height (Rt), root mean square roughness (Rq), etc. Among these Ra is
the used most commonly which is generally defined on the basis of the ISO 4287 norm as the
arithmetical mean of deviations of surface profile from the centre-line along the measurement.

This is expressed mathematically in Eq. 1.1

(1.1)

where L is the sampling length, z is coordinate of the profile curve.

The surface roughness is produced by a complex interplay of various cutting parameters. In


general, the relationship between the surface roughness and independent machining variables can
be defined as in equation 1.2

1
(1.2)

where; Ra is the surface roughness in μm; V, f, d and r are the cutting speed (m/min), feed rate
(mm/rev.), depth of cut (mm), and tool nose radius (mm), respectively. C, n, m, p, l are constants
and ε is random error.

A number of models have been proposed to predict the surface roughness under varying
conditions. Theoretical and experimental approaches are two master pillars of this research. With
the rapid development of the computer technologies, computer modeling and simulation
approach too has emerged as a third master pillar.

Computer simulation techniques, such as finite element method (FEM), have been widely used
in scientific study; in some circumstances, they can be reasonable substitutes to the physical
experiments. Hence, they are often referred to „computer simulation experiments‟.

FEM is a numerical technique for finding approximate solutions of partial differential equations
as well as of integral equations. In this method, a distributed physical system to be analyzed is
divided into a number (often large) of discrete elements. The complete system may be complex
and irregularly shaped, but the individual elements are easy to analyze. The elements may be one
dimensional, two dimensional (triangular or quadrilateral), or three dimensional (tetrahedral,
hexahedral, etc.); and linear or higher-order.

Different models are available to model mechanics, acoustics, thermal fields, electromagnetic
fields, etc., or coupled problems. For the current problem structural elements have been used.
More detaied discussion will be taken up on the chapter dealing with it.

2
Chapter 2. Literature Review & Problem Statement

In the field of manufacturing, especially in engineering, the exact degree of roughness can be of
considerable importance, affecting the functioning of a component as well as its cost. The
performance as well as the longevity of the component is determined by the surface finish. This
makes roughness one of the decisive parameters while choosing machining process and the
process parameters. In addition, intense international and domestic competition has now focused
the attention of manufacturers on to control/improve the surface quality of machined products.
The study of surface roughness and vibration and the interactions between the two is now fast
becoming an integral part of mechanical engineering around the world.

Therefore, it is important to determine or predict, in numerical terms, how rough a surface will
be. Unfortunately, surface roughness is also difficult and expensive to control/improve. Selection
of appropriate process parameters is the most critical step involved, and is still being explored.
Because the degree of required surface roughness partially forms the basis for selection of
cutting process, extensive attention has been given to modeling & prediction of roughness.

Davim et al. (2008) have developed surface roughness prediction models using artificial neural
network (ANN), which emulate the biological connections between neurons, to investigate the
effects of cutting conditions during turning of free machining steel. The ANN model of surface
roughness parameters (Ra and Rt) was developed with the cutting conditions such as feed rate,
cutting speed and depth of cut as the affecting process parameters.

Lin and Chang (1998) developed a surface topography simulation model to simulate the surface
finish profile generated after a turning operation. This simulation model incorporates the effects
of the relative motion between the cutting tool and the workpiece with the effects of tool
geometry to simulate the resultant surface geometry. It is found that the vibration frequency ratio
is a more important vibration parameter than the vibration frequency on the characterization of
the surface finish profile. The vibration frequency ratio is the ratio between the vibration
frequency and the spindle rotational speed.

3
Bouacha et al (2010) have analyzed and modeled the relationship between cutting parameters
(cutting speed, feed rate and depth of cut) and machining output variables (surface roughness,
cutting forces) through the response surface methodology. The model was used to find optimum
values of machining.

Abouelatte and Madl (2001) have worked to find a correlation between surface roughness and
cutting vibrations in turning and to derive mathematical models for the predicted roughness
parameters based both on cutting parameters and machine tool vibrations.

The aim of this project is to develop a finite element model of the cutting tool and couple it with
a mathematical model for tool-workpiece interaction to predict the resulting surface
roughness/profile. Hence, this project involve three significant aspects namely, laboratory
experiments to record cutting forces, ANSYS finite element modeling to obtain tool response
and MATLAB mathematical modeling to generate surface profile.

Since surface roughness is a complex interplay of various cutting conditions, tool and material
parameters, it is possible to make the model more accurate by further incorporating these effects.
The model could also be made self-sufficient by evolving statistical models to reconstruct the
random force data thus removing the need to conduct experiments every time.

Table 2.1: Gantt Chart

4
Chapter 3. Experimental Setup and Design

The current work requires certain data (cutting forces) to be fed into the simulation to create a
model and generate surface roughness profile, while another set of data (experimental roughness)
is required to establish the validity of the models. Hence, the work began with a series of
experiments at the Machine Tools Laboratory, IIT Delhi where a mild steel workpiece was
turned on a CNC lathe machine. The cutting forces operating during this process were recorded
with a dynamometer and a Talysurf surface profiler was used to measure the surface roughness
of the turned workpiece. Sections 3.1 and 3.2 describe in detail the experimental setup used and
the set of operating conditions chosen for this purpose.

3.1 Experimental Setup


As mentioned an experimental setup was prepared at the Machine Tools Laboratory, IIT Delhi
which consisted of a CNC lathe machine for turning, cylindrical mild steel workpiece,
dynamometer for recording the cutting forces and a talysurf surface profiler (separately) for
roughness measurement. Figure 3.1 gives a schematic of the aforementioned setup.

Figure 3.1: Schematic of the experimental setup prepared at Machine Tools Lab, IIT Delhi

5
As depicted in the figure 3.1, for a variety of cutting parameters like cutting speeds and feed rate,
a mild steel workpiece was turned on Leadwell T-6 CNC lathe machine. Two workpieces made
of the workpiece were turned, each being 160mm long and having 50mm initial diameter.

The cutting tool was a Widia Kennametal CNMG120404 carbide insert (shown in figure 3.2)
mounted on a tool holder made of high carbon steel. The rhombic insert had 800 nose angle with
0.8 mm nose radius. It had 00 relief angle while the rake angle was - 60.

Figure 3.2: Cutting tool insert geometry (www.kennametal.com)

Kistler Multicomponent Dynamometer 9129AA was used for recording the three components of
the resultant cutting force acting on the tool. The dynamometer was mounted on the tool holding
fixture of the turret using a machine adapter type 9129AD. Kistler multichannel charge amplifier
type 5070A was used to convert measurement signals into an electrical voltage. Kistler
DynoWare type 2825A software was used for data acquisition.

After turning was over, the workpiece was taken to the Metrology Laboratory, IIT Delhi for
measuring the surface roughness. For this purpose, Taylor Hobson Form Talysurf profiler was
used. This contact type instrument has a diamond stylus with 2μm radius conisphere. Figure 3.3
presents pictures of the three instrumemnts used in these experiments- Leadwell T-6 CNC lathe
machine, Taylor Hobson Form Talysurf and Kistler Multicomponent dynamometer.

Figure 3.3: CNC lathe, Talysurf surface profiler and Dynamometer

6
3.2 Experimental Design
The experiments involved turning of mild steel workpiece on CNC lathe machine in Machine
Tool Laboratory, IIT Delhi. As shown by Davim et al (2008), depth of cut does not have
significant influence on surface roughness, only cutting speed and feed rate were chosen to be
the cutting parameters. The depth of cut was maintained constant throughout the experiment at
1mm.

For the experiments, 4 levels were selected for feed rate as well as cutting speed as shown in the
Table 3.1 below:-

Table 3.1: Experimental design

Cutting Speed Feed Rate


(m/min) (mm/rev)
Level 1 60 0.08
Level 2 90 0.12
Level 3 120 0.16
Level 4 150 0.20

7
3.3 Experimental Observations

3.3.1 Observations regarding Cutting Forces


All the three components of the cutting force were recorded using the dynamometer with
sampling frequency 1000Hz. As expected the forces showed fluctuations around a mean value
which can be attributed to the vibrations and other irregularities arising during the process. Table
3.2 below presents the mean and deviations of the cutting forces for every cutting condition.

Table 3.2: Mean and standard deviation (SD) values of experimental cutting forces

Speed Feed Fx (N) Fy (N) Fz (N)


(m/min) (mm/rev) Mean SD Mean SD Mean SD
60 0.08 28.4 7.5 116.1 34.2 93.9 19.2
0.12 30.4 8.4 137.6 28.5 110.3 17.1
0.16 38.9 11.2 159.6 39.8 123.6 21.8
0.20 46.1 13.5 172.3 52.0 129.1 29.2
90 0.08 21.8 6.2 78.8 19.1 57.9 12.0
0.12 29.1 7.3 101.1 22.5 74.5 11.9
0.16 34.7 8.3 116.2 22.8 84.2 16.3
0.20 33.0 4.2 124.9 9.7 86.6 6.4
120 0.08 21.3 4.5 71.7 12.2 50.2 8.0
0.12 27.0 7.3 89.2 11.2 62.9 8.9
0.16 30.5 2.4 103.6 7.2 71.5 5.5
0.20 33.7 1.9 118.3 7.4 81.6 6.4
150 0.08 22.2 5.7 66.8 8.1 48.5 5.8
0.12 28.8 2.3 85.0 5.9 60.1 5.2
0.16 30.7 2.5 96.8 6.9 68.0 7.0
0.20 38.2 4.1 115.6 9.3 82.5 10.5

From the table 3.2 it can be inferred that as the feed rate is increased, higher forces are required
for cutting. This trend is also illustrated in figure 3.4 on the next page.

8
Vc = 60m/min Vc = 90m/min
200 200

150 150
Force (N)

Force (N)
100 Fx 100 Fx
Fy Fy
50 50
Fz Fz
0 0
0.08 0.12 0.16 0.2 0.08 0.12 0.16 0.2
Feed rate (mm/rev) Feed rate (mm/rev)

Vc = 120m/min Vc = 150m/min
200 200

150 150
Force (N)

Force (N)

100 Fx 100 Fx
Fy Fy
50 50
Fz Fz
0 0
0.08 0.12 0.16 0.2 0.08 0.12 0.16 0.2
Feed rate (mm/rev) feed rate (mm/rev)

Figure 3.4: Variation of cutting forces with feed rate at different cutting speeds

It can also be observed that at higher cutting speeds smaller forces are required. This can be seen
in figure 3.5

9
f = 0.08mm/rev f = 0.12mm/rev
200 200

150 150
Force (N)

Force (N)
100 Fx 100 Fx
Fy Fy
50 50
Fz Fz
0 0
60 90 120 150 60 90 120 150
Cutting speed (m/min) Cutting speed (m/min)

f = 0.16mm/rev f = 0.20mm/rev
200 200

150 150
Force (N)

Force (N)

100 Fx 100 Fx
Fy Fy
50 50
Fz Fz
0 0
60 90 120 150 60 90 120 150
Cutting speed (m/min) Cutting speed (m/min)

Figure 3.5: Variation of cutting forces with speed at different feed rates

10
3.3.2 Observations regarding Surface Roughness
Surface roughness of the turned workpiece was measured using a Taylor Hobson Talysurf. Table
3.3 presents the roughness values for all the operating conditions.

Table 3.3: Experimentally measured surface roughness

V= 60m/min V= 90m/min V= 120m/min V= 150m/min


Feed Ra (µm) Feed Ra (µm) Feed Ra (µm) Feed Ra (µm)
(mm/rev) (mm/rev) (mm/rev) (mm/rev)
0.08 1..21 0.08 1.18 0.08 1.14 0.08 1.12
0.12 1.78 0.12 1.67 0.12 1.61 0.12 1.58
0.16 2.31 0.16 2.27 0.16 2.21 0.16 2.11
0.20 2.74 0.20 2.68 0.20 2.67 0.20 2.56

From the roughness data presented in table 3.3 it can be observed that for a particular cutting
speed, roughness increases with feed rate while cutting speed does not affect the roughness value
significantly. These have been presented in figure 3.6 below.

3 60 3

2.5 2.5 0.08


90
2 2 0.12
Ra (μm)

Ra (μm)

1.5 120 1.5 0.16


1 1
150m/ .2mm/
0.5 min 0.5 rev

0 0
0.08 0.12 0.16 0.2 60 90 120 150
Feed rate (mm/rev) Cutting speed (m/min)
A) B)

Figure 3.6: Variation of surface roughness with (A) feed rate (B) cutting speed

11
Chapter 4. CAD Modeling of Tool

The cutting forces and corresponding surface roughness values for every combination of process
parameters has been recorded through a series of experiments. The next step was to create a
simulation comprising of solid as well as mathematical models which would utilize the recorded
forces to predict surface roughness and experimental roughness values for validation.

This started with a solid model of the cutting tool which was created using commercial software
SolidWorks Premium 2010. In this case, the tool was a combination of a carbide insert with a
high carbon steel holder. The insert was Widia Kennametal CNMG120404 with 800 nose angle
and 0.8mm nose radius. It had 00 relief angle and - 60 rake angle. Other dimensions have been
shown in figure 3.2. Figure 4.2 below too shows important dimesions of the insert as well as the
holder.

Solid models for the insert and holder were created separately in SolidWorks. This is because the
two have different material properties which greatly affect the tool‟s response to cutting forces.
The models were saved in .IGES (initial graphics exchange specification) file format. These
models were then imported into ANSYS and subsequent analysis done as detailed in chapter 5.

Figure 4.1 shows the assembly view of the tool in which the dark portion corresponds to the tool
holder while the grey portion is for the insert. Figure 4.2 presents the engineering drawings.

12
Figure 4.1: Solid model of the tool holder and insert assembly (trimetric view)

Figure 4.2: Engineering drawings of the tool assembly (all dimensions in mm)

13
Chapter 5. Finite Element Modeling of Tool

The solid models created in SolidWorks were imported into ANSYS, specialized software for
finite element modeling. As the insert and the tool holder have different material properties,
relevant properties were assigned for each separately. For the tool holder made of cemented
carbide Young‟s Modulus of 210 GPa, Poisson‟s Ratio of 0.29 and density of 7900 kg/m3 were
taken. Similarly, for the insert which was made of cemented carbide, 660 GPa Young‟s Modulus,
0.24 Poisson‟s Ratio and 15800 kg/m3 density were taken. These properties have been
summarized in table 5.1 below.

Table 5.1: Mechanical properties of tool holder and insert (www.memsnet.org)

Property unit Tool holder Insert


Material - High carbon Cemented
steel carbide
Young‟s Modulus GPa 210 660
Poisson‟s Ratio - 0.29 0.24
Density kg/m3 7900 15800

After defining material characteristics of the two tool components, the next step was selection of
appropriate solid element for meshing. There various criteria for selection of relevant solid
element including analysis attributes (mechanical, thermal, magnetic, etc.), dimensions (1D, 2D,
3D), degrees of freedom as well as geometric complexities involved.

This particular work required analysis of a three dimensional mechanical system where isotropy
was assumed throughout the bulk of both insert and the holder. Considering the geometric
complexity especially at the insert tip, a four node tetrahedral element (SOLID285) was selected.

SOLID285 is a 3 dimensional structural solid element. It has four nodes (I, J, K and L as shown
in figure 5.1) giving it a tetrahedral shape. It exhibits four degrees of freedom including three
translational degrees of freedom and hydrostatic pressure. Figure 5.1 shows a schematic diagram
of this element (SOLID285)

14
Figure 5.1: SOLID285 element geometry (source: ANSYS)

The solid model was then meshed with these elements. Considering higher geometric complexity
around the tool insert as well as high forces acting during the course of cutting and higher
expected deformations, the model was meshed relatively fine towards the insert while a coarser
meshing was done for the tool holder. Such biased meshing has been widely known to be
efficient in cases where some parts are bulky while others intricately shaped. It reduces the
number of elements which in turn has direct effect on reducing the processing time, while at the
same time not affecting the accuracy significantly. Figure 5.2 shows a trimetric view of the tool
assembly which has been meshed with SOLID285 elements.

Figure 5.2: Meshing of the tool model with refinement towards tip

15
The meshed model was then subjected to the experimentally recorded forces. As described in
earlier chapters, a dynamometer was used to record the three components of the cutting forces
while the mild steel workpiece was being turned on a CNC lathe. These forces were then applied
on a node at the tip of the tool assuming the forces concentrated here though in reality they are
spread over a finite contact area. But the area contact has not been considered for simplicity.

After meshing the model, experimentally recorded forces were applied on it. For this purpose,
the time-domain forces were first imported into ANSYS and applied on a node near the tip of the
insert. Figure 6.1 below shows displacements of the node on tool tip as predicted by ANSYS.

Figure 5.3: Displacement response of tool tip in 3 directions (X’, Y’, Z’)

16
Chapter 6. Surface Profile Generation

6.1 Generation of tool tip’s trajectory


The displacement response of the tool tip obtained from the finite element analysis as a response
to cutting forces was used to generate the surface profile which would be inscribed on the
workpiece. It has been assumed that the movement of the tool relative to the workpiece alone
dictates the resulting surface profile. Also, though in practice the tool had 0.8mm nose radius
(figure 6.1(B)), it has been considered having pointed tip (figure 6.1(A)) with the nose angle
same as in real tool, i.e. 80 degrees.

Figure 6.1: Shape of tool tip (A) pointed, used for generating surface profile and (B) circular, actually used in experiments

Ideally, in absence of any vibrations, the tip moves along a helical trajectory with respect to the
workpiece. But, due to vibrations, it keeps on fluctuating around its mean position thus giving
rise to a rough profile on the workpiece.

Though the tool remains fixed while the workpiece rotates, it is convenient to visualize the
opposite, i.e. keeping the workpiece fixed and the tool moving relative to it. Hence, in reference
frame of workpiece, tool rotates around it as well as moving along the feed directions. Thus, the
tool traverses a helical trajectory relative to the workpiece. The vibration in the process give rise
to fluctuations in this helix making it non-smooth.

Figure 6.2(A) shows the condition where tool is held fixed and workpiece rotates, this is what
happens in reality. On the other hand, figure 6.1(B) shows the motion of tool relative to the
workpiece. (X‟, Y‟) is the coordinate representing the displacement of the tool tip as obtained
from ANSYS. (X, Y) is the global coordinate system.

17
Figure 6.2: (A) Rotating workpiece, (B) Motion of tool relative to workpiece

Now, from finite element analysis already done, (X‟, Y‟, Z‟) coordinates are known. Coordinates
of the tool tip relative to the (X, Y, Z) coordinates at any time t can be represented as:

(6.1)

(6.2)

(6.3)

where

R = radius of workpiece (mm), d = depth of cut (mm), ω = angular speed of workpiece


(radian/sec), f = feed rate (mm/rev)

The equations 6.1, 6.2 & 6.3 were used to generate the trajectory of the tool tip using a program
written in MATLAB. The program for this operation is provided in Appendix-III. As expected,
the tool tip traversed a helical path relative to the workpiece with fluctuations because of the
vibrations. Figure 6.3 presents a portion of the trajectory of the tip as predicted from the analysis
done so far.

18
Figure 6.3: Helical path of tool tip relative to workpiece

6.2 Generation of Surface Profile


Now it is assumed that the surface profile to be generated on the workpiece would be solely
dictated by the motion of the tool tip and its shape. To generate the surface profile, all the points
lying on the helical trajectory for any particular angular position have been considered. Such
points are illustrated by the black dots in figure 6.3 above and figure 6.4 below. These points
represent the location of the tool tip while to get the whole surface profile, intersection points
between major and minor cutting edge angles of the tool for two subsequent positions of the tool
are required. These intersection points are illustrated as small circles in figure 6.4 below.

A mathematical expression for the determination of intersection points has been derived which is
presented in equation 6.4 and 6.5

(6.4)

(6.5)

19
Figure 6.4: Generation of surface profile; black circles are points from helical trajectory of tool tip, circles are
intersection points between cutting edges at two subsequent tool locations

Here,

α = 900 – working minor cutting edge angle = 860


β = 900 – working major cutting edge angle = -60

This method has been used to locate all valley (dark dots) and peak points (small circles) at a
particular angular position, then join them with straight lines and thus create a zigzag profile to
obtain the finished surface of the workpiece.

With the surface profile known, another MATLAB program has been used to determine the
corresponding roughness values. First of all, the meanline ( was determined using formula
6.6 and 6.7 shown below. Figure 6.5 shows surface profile and the way it has been used to
calculate roughness.

(6.6)
where is the area below the curve defined by the generated surface profile and n is the number
of points.

(6.7)

20
Figure 6.5: Deteremination of average roughness (Ra) by calculation of enclosed areas

Now, with known, Ra was determined by calculating the total area enclosed between the
generated profile and the meanline. To calculate this area, the areas enclosed between line
joining two subsequent peaks and valleys and the meanline were calculated separately and added
together to et the total area as in formula 6.8

(6.8)
If the two points lie on opposite sides of the meanline, areas of the two triangles is calculated
using formula 6.9

(6.9)

It the two points lie on the same side, area of trapezium is calculated using formula 6.10

(6.10)

After calculating the total area, Ra is determined using equation 6.11

(6.11)

21
6.3 Simulation Results
These formulae were used to create a program in MATLAB and calculate R a values for all
combinations of process parameters. Table 6.1 shows a summary of this.
Table 6.1: Surface roughness results predicted from the simulation

V= 60m/min V= 90m/min V= 120m/min V= 150m/min


Feed Ra (µm) Feed Ra (µm) Feed Ra (µm) Feed Ra (µm)
(mm/rev) (mm/rev) (mm/rev) (mm/rev)
0.08 1.42 0.08 1.41 0.08 1.41 0.08 1.41
0.12 2.12 0.12 2.12 0.12 2.11 0.12 2.11
0.16 2.83 0.16 2.82 0.16 2.82 0.16 2.82
0.20 3.53 0.20 3.52 0.20 3.52 0.20 3.53

Table 6.2: comparison of experimental and simulated surface roughness values

Cutting Speed Feed Rate Ra (µm) from Ra (µm) from RMS


(m/min) (mm/rev) Experiments Simulation Error (%)
60 0.08 1.21 1.42 17.1
0.12 1.78 2.12 19.0
0.16 2.31 2.83 22.2
0.20 2..74 3.53 28.5
90 0.08 1.18 1.41 19.4
0.12 1.67 2.12 26.2
0.16 2.27 2.82 24.0
0.20 2.68 3.52 31.1
120 0.08 1.14 1.41 23.3
0.12 1.61 2.11 30.7
0.16 2.21 2.82 27.2
0.20 2.67 3.52 31.3
150 0.08 1.12 1.41 25.5
0.12 1.58 2.11 32.6
0.16 2.11 2.82 33.2
0.20 2.56 3.53 37.2

22
Vc = 60m/min
4
3

Ra (μm)
2
1 Experiment
0 Simulation
0.08 0.12 0.16 0.2
Feed rate (mm/rev)

Vc = 90m/min
4
3
Ra (μm)

2
1 Experiment
0 Simulation
0.08 0.12 0.16 0.2
Feed rate (mm/rev)

Vc = 120m/min
4
3
Ra (μm)

2
1 Experiment
0 Simulation
0.08 0.12 0.16 0.2
Feed rate (mm/rev)

Vc = 150m/min
4
3
Ra (μm)

2
1 Experiment
0 Simulation
0.08 0.12 0.16 0.2
feed rate (mm/rev)

Figure 6.6: Comparison of experimental and simulation roughness with varying feed rate

23
It can be observed from table 6.1, 6.2 and figure 6.5 that the simulation predicts a surface
roughness (Ra) which is higher than the experimentally determined roughness with root mean
squared error ranging from 17% upto 37%. It is known that roughness is a weak decreasing
function of cutting speed. This has been observed experimentally and the simulation also predicts
exhibits a similar relationship between the two. Another factor considered in this work was feed
rate which has a relatively strong effect on roughness which increases with feed. This
experimental observation is also reflected by the simulation predictions as shown in figure 6.5
above. It can be concluded that the proposed model has been able to predict roughness values
with small errors and also shows trends of roughness with process parameters similar to
experimentally established pattern.

24
Chapter 7. Conclusions and Discussion

7.1 Conclusions
The present work has focused on prediction of surface roughness by reconstructing the finished
surface profile. This has been done using finite element and mathematical modeling of the tool
which was subjected to the cutting forces recorded during laboratory experiments. The surface
roughness of the finished workpiece, also measured in laboratory, was used to validate the results
and simulation.

Certain assumptions have been taken for this work:-

1. Machine vibrations have been neglected while analyzing the finite element model and
only the vibrations generated as response of cutting tool to dynamic cutting forces have
been considered.
2. Imperfections and anisotropy in the workpiece and tool have been neglected.
3. Surface roughness has been considered to be determined solely by the tool geometry and
its vibrations while other factors like built-up edge, tearing of work material, chip flow,
inaccuracies in machine tool movement, etc. have been neglected.
4. While generating the surface profile on the workpiece, nose radius of the tool insert has
been neglected and it has been taken as sharp pointed as shown in figure 6.1
5. It has been assumed that the geometry determined by working cutting-edge angles and
tool vibrations as shown in figure 6.4 generates identical surface profile on workpiece.

Certain observations can be made regarding the model:-

1. It has been able to predict roughness values with certain errors and also shows trends of
roughness with process parameters similar to experimentally established pattern.
2. Another significant feature is that the model takes into account the fluctuations in cutting
forces which result into tool vibrations and increased surface roughness.
3. Since cutting forces recorded experimentally have been used to predict roughness, the
effect of other factors like coolant, tool wear, tool-workpiece material properties, etc. are
inherent to the force pattern. Hence, when the forces are used in the model, these factors
are automatically taken care of. To remove the dependence on experiments to record
forces, statistical data reconstruction techniques can be adopted as described in chapter 8.
4. The effects of tool geometry and material properties can be easily studied by making
small alterations at the modeling stage.

25
7.2 Limitations of model
The present work involves a number of experimental work and finite element and mathematical
modeling, each having its own set of imitations. These can be summarized as:-

1. The tool has been modeled as a sharp pointed one while neglecting the nose radius. This
has the effect of providing inflated roughness prediction. To get a better prediction close
to reality, the tool needs to be modeled as shown in figure 6.1(B).
2. The geometry of the tool gets altered in machining due to wear. But in the present model,
the tool has been modeled such that its geometry undergoes no change.
3. The present model is dependent upon experimentally recorded forces to predict the
surface roughness. This has limiting effect on its practicable utility in the present form.
But, if this can be coupled with statistical data reconstruction techniques (illustrated in
chapter 8), the need to record forces experimentally can be eliminated.
4. In the present wok, effects of only cutting speed and feed rate have been analyzed. The
modeling needs to be further extended to include other factors like coolant, tool wear, etc.

26
Chapter 8. Scope for future work

Looking at the whole work it appears that the single most important factor inhibiting a
practicable use of the proposed model is its dependence on experimentally recorded cutting
forces. But clearly, this is only the first step towards making the whole process self-reliant.

A statistical analysis of the recorded forces reveals so much about their nature which could be
gainfully utilized in future. Force data corresponding to various cutting conditions were analyzed
and numerous useful observations were recorded. As an illustration, a simple analysis for the
operating condition (120m/min speed and 0.12mm/rev feed) is being presented here to serve as
an easy reference to future researchers. To start with, a relative frequency histogram shows that
the distribution for all the three components of the force resembles normal distribution.

Figure 8.1: Relative frequency histograms of forces in 3 directions Fx, Fy and Fz

27
Another test for stationarity of the forces can be applied. A stationary random process is one
whose statistical properties do not vary with time. This check is important to ensure that the
statistical properties of the recorded force data for any given cutting condition do not vary
whenever experiments are conducted. To check it, data samples for different time intervals at the
same speed and feed were taken and their statistical properties compared. Figure 8.2 shows
relative frequency histograms of for four different sections of the cutting force (Fy).

Figure 8.2: Relative frequency histogram for four sections of force data to check stationarity

Table 8.1 presents a summary of other important statistical properties for the four sections. It has
been observed that the statistical properties for the different sections of force data do not vary
much for any given operating conditions. This indicates that the forces are stationary to a good
extent. This property can be utilized to regenerate the force statistically.

28
Table 8.1: comparison of statistical properties to check stationarity

Mean (N) Standard


Deviation
Section 1 88.66 11.56
Section 2 89.15 11.93
Section 3 88.23 12.01
Section 4 86.43 11.88

Till now, it has been observed that the forces resemble a normal distribution and are stationary
also. To visualize the extent to which the forces exhibit normal distribution, normal probability
plot was constructed for Fy which has been shown in figure 8.3

Figure 8.3: Normal probability plot experimental cutting force Fy (dashed line is for normal distribution while other is
for recorded forces)

To quantify the closeness of force data to normal distribution, skewness and kurtosis were
compared. Skewness is the degree of asymmetry, or departure from symmetry, of a distribution
which is mathematically expressed as in equation 8.1

(8.1)

Kurtosis is the degree of peakedness of a distribution which is mathematically expressed in


equation 8.2

29
(8.2)

Where is the moment about the mean as shown in equation 8.3

(8.3)

To check if a normal distribution can be used to regenerate cutting forces, their extent of
closeness was analysed. Skewness and Kurtosis of the measured force data were determined and
compared to those of an ideal normal distribution. The results of this comparison are shown in
table 8.2 below.

Table 8.2: Comparison of skewness and kurtosis between normal distribution and measured force

Skewness Kurtosis
Normal Distribution 0 3
Measured Force 0.15 2.95

The tests mentioned here indicate that the force data have striking closeness to normal
distribution and that while statistically regenerating the cutting forces their distribution can be
taken as normal distribution. Using this property, cutting forces were regenerated for cutting one
condition 120m/min speed and 0.12mm/rev feed. Mean and deviation values of 88.12N and
11.89N respectively were used for this. Figure 8.4 shows the regenerated force (Fy) with time.

Figure 8.4: Regenerated force vs time plot (120m/min speed and 0.12mm/rev feed)

30
A pictorial comparison of the fourier transforms of the experimental and regenerated cutting
forces has been presented in figure 8.4 below. The fourier transform shows the amplitudes of
frequency components present in the signal which also show similarity.

Figure 8.5: Amplitude spectra for (A) experimental force (B) regenerated force

Hence, it has been shown that the cutting forces can be easily regenerated by using a normal
distribution. A wide range of experiments needs to be conducted to determine the mean and
standard deviation values for varying process parameters. These results would then aid force
regeneration. Such regeneration would eliminate the need of bulky experiments while enabling
off-field roughness prediction and selection of appropriate conditions for machining.

31
REFERENCES

1) Abouelatta, O., & Madl, J. (2001). Surface Roughness Prediction Based on Cutting
Parameters and Tool Vibrations in Turning Operations. Journal of Materials Processing
Technology , v 118, pp 269-277.

2) Bouacha, K., & Yallese, M. A. (2010). Statistical analysis of surface roughness and
cutting forces using response surface methodology in hard turning of AISI 52100 bearing
steel with CBN tool. International Journal of Refractory Metals & Hard Materials, v 28,
pp 349–361.

3) Carroll, J. T., & Strenkowski, J. S. (1988). Finite Element models of orthogonal cutting
with application to single point diamond cutting. International Journal of Mechancal
Science , v 30, N.12, pp 899-920.

4) Davim, J. P., Gaitonde, V., & Karnik, S. (2008). Investigations into the effect of cutting
conditions on surface roughness in turning of free machining steel by ANN models.
journal of materials processing technology , v 205, pp 16–23.

5) Kadirgama, K., Noor, M., Rahman, M., Rejab, M., Haron, C., & Abou-El-Hossein, K.
(2009). Surface Roughness Prediction Model of 6061-T6 Aluminium Alloy Machining
Using Statistical Method. European Journal of Scientific Research,v 25, N.2,pp 250-256.

6) Ling, S., & Chang, M. (1998). A study on the effects of vibrations on the surface finish
using a surface topography simulation model for turning. International Journal of
Machine Tools & Manufacture , v 38, pp 763–782.

7) Ozel, T., & karpat, Y. (2005). Predictive modeling of surface roughness and tool wear in
hard turning using regression and neural networks. International Journal of Machine
Tools & Manufacture , v 45, pp 467–479.

8) Taskesen, A., & Ercan, Y. (2005). Computer Aided Non-linear analysis of Machine Tool
Vibrations and a developed Computer Software. Journal of Mathematical and
Computational Applications , v 10, N. 3, pp 377-385.

32
APPENDICES

33
Appendix-I: MATLAB program surface profile

A mathematical model was created which used tool‟s displacement response to fluctuating cutting forces
as determined by another finite element model. This information was used to generate the surface profile
on finished workpiece and determine roughness values. The MATLAB code for surface profile generation
is presented in this Appendix.

r_wp=input('input the value of radius of workpiece (mm) ');


w=input('input the value of angular speed (rpm) ');
f=input('input the value of feed (mm/rev) ');
d=input('input the value of depth of cut (mm) ');
T=60/w; %time period(s)of rotation of workpiece
xo=xlsread('UX',-1); %import x component(m) of tool vibration
yo=xlsread('UY',-1); %import y component(m) of tool vibration
zo=xlsread('UZ',-1); %import z component(m) of tool vibration

t_incr=.05; %time increment(s)used in ANSYS analysis


q=2; %no of roughness samples on the w/p
t=[0:t_incr:((length(xo)-1)*t_incr)]';
wt=w*pi*t/30;

%locus of tool tip w.r.t. workpiece frame


x=-(r_wp-d)*cos(wt)- 1000*(xo.*cos(wt)- yo.*sin(wt));
y=(r_wp-d)*sin(wt)- 1000*(yo.*cos(wt)+ xo.*sin(wt));
z=-f*wt/(2*pi)+ 1000*zo;

%transforming (x,y,z) to (r,theta,z)


r=sqrt(x.^2 +y.^2);
theta=atan(y./x);

Ra=[];
n=round(T/t_incr); %number of data points in one cycle
p=round(T/(t_incr*q));
for l=1:p:((q-1)*p+1)
R1=[]; Z1=[];
for i=l:n:length(t)
R1=[R1;r(i)]; %r coordinate of data points at particular theta
coordinate
Z1=[Z1;z(i)]; %z coordinate of data points at particular theta
coordinate
end
a=86; %angle alpha
b=-6; %angle beta

%Creating peaks in surface profile from tool geometry


R2=[];
Z2=[];
g=length(Z1);
for i=1:(g-1)
Z2=[Z2 ((R1(i)-
R1(i+1))*tand(a)*tand(b)+Z1(i)*tand(a)+Z1(i+1)*tand(b))/(tand(a)+tand(b))];
R2=[R2 (Z1(i)-
Z1(i+1)+R1(i)*tand(b)+R1(i+1)*tand(a))/(tand(a)+tand(b))];
end

34
Z=[];R=[];
for i=1:(2*g-1)
if mod(i,2)==1
Z=[Z Z1((i+1)/2)];
R=[R R1((i+1)/2)];
else
Z=[Z Z2(i/2)];
R=[R R2(i/2)];
end
end
m=length(R);
Area=0; A=0;
for j=1:(m-1);
Area=Area+.5*(R(j)+R(j+1))*(Z(j+1)-Z(j));
end
ravg=Area/(Z(m)-Z(1)); %mean line for defining Ra
Rm = R - ravg;
for k=1:(m-1)
if (R(k)-ravg)*(R(k+1)-ravg)<0
A = A + .5*(Rm(k)^2 +Rm(k+1)^2)*(Z(k+1)-Z(k))/(abs(Rm(k)-
Rm(k+1)));
else
A = A + .5*abs(Rm(k)+Rm(k+1))*(Z(k+1)-Z(k));
end
end
Ra=[Ra;A/(Z(m)-Z(1))]; %Ra values at different theta coordinates
end
Ra_micron=Ra*1000 %Ra values in micron
Ra_avg_micron=sum(Ra)*1000/q %average of Ra values

subplot(2,1,1),plot3(x,y,z)
subplot(2,1,2),plot(Z,R)

save('Roughness.mat','Ra_micron','Ra_avg_micron')

35
Appendix II: User Manuals

1. System Requirements for Matlab R2008b & ANSYS 12


Operating System Processors Disk Space RAM
®
Intel Pentium
Windows® XP
(Pentium 4 and above)
(Service Pack 1, 2 or 3)
Intel Celeron 510 MB* 512 MB
Windows Server 2003
Intel Xeon (MATLAB® (1024 MB
(Service Pack 1 or 2, R2)
Intel Core only) recommended)
Windows Vista™ (Service
AMD Athlon 64**
Pack 1)
AMD Opteron

2. User manual for ANSYS


User is instructed to follow the following steps to create and analyse tool model:

• Make solid model of tool in SolidWorks


•Save the model as *.IGS file and import to ANSYS
1

•ANSYS Main Menu>Preprocessor>Element Type - Select element type


•ANSYS Main Menu>Preprocessor>Material Props -Define material properties
•ANSYS Main Menu>Preprocessor>Meshing - Perform meshing
2
•ANSYS Main Menu>Solution>Analysis Type>New Analysis>Transient
•ANSYS Main Menu>Solution>Define Loads>Structural>Displacement - define
boundary condition
•Using command prompt define a variable as table and import force data
3 available in *.txt format
•Use *do loop to apply force on aparticular node and specify time interval and
time increment.

•Export the displacement response of a node near the tip in *.xls file format
•Import the data in Matlab while running the m-file for roughness value
4 prediction

36

You might also like