Dn/1

Assessment of Machining Models : Progress Report

R.W. Ivester (2), M. Kennedy, M. Davies (1)

1
National Institute of Standards and Technology
Gaithersburg, MD 20899, USA
Tel: 301/975-8324, Fax: 301/975-8058, e-mail: [email protected]

R. Stevenson, General Motors, Warren, MI, USA

J. Thiele, Caterpillar Incorporated, Peoria, IL, USA
R. Furness, S. Athavale, Ford Motor Company, Dearborn, MI, USA

Abstract
Machining involves extremely localized and nonlinear physical phenomena that occur over a wide range of
temperatures, pressures, and strains. The complexity of the system has hindered progress in predictive
modeling of machining processes. Many different types of models ranging from theoretical to empirical have
been developed, but the wide variety of the models makes performance assessment difficult. The difficulty
in assessing the performance of machining models has been cited by industry as the major factor that limits
the use of modern machining models in industry. Thus, the current practice in industry is either to use
conservative machining settings and tool-change policies, or to conduct costly empirical studies for a limited
selection of tools and coolants. Either practice may lead to sub-optimal process performance. The goal of
the Assessment of Machining Models project is to assess the ability of modern machining models to predict
the outputs of machining processes based upon data typically available on the shop floor. In order to
achieve this goal, the participating laboratories plan to develop and provide a clear, consistent, well-
measured and relevant data set, and use that data set to benchmark the predictive capability of machining
models in blind tests. This paper presents the project motivation, goals, and some representative results.

Keywords: machining, tool wear, temperature, force, modeling, validation

1 MOTIVATION representative of machining production in the United

In 1998, Merchant estimated that 15 % of the value of all States the financial impact of the use of sub-optimal
mechanical components manufactured worldwide is tooling selection, process parameters, and tool-change
derived from machining operations [1]. Other studies policies is staggering.
have found that total U.S. expenditures on machining are An alternative approach to empirical testing and
between 3 % and 10 % of the annual U.S. gross experience is the development of predictive models that
domestic product (GDP): between $240 to $850 billion are based upon the fundamental physics of the
dollars for 1998 [2]. However, despite its obvious machining process. The advantage of this approach is
economic and technical importance, machining remains that predictions are made from the basic physical
poorly understood. Parameters are chosen through properties of the tool and workpiece materials together
empirical testing and the experience of machine with the kinematics and dynamics of the process. Thus,
operators and programmers. This process is expensive after the appropriate physical data is determined, the
and time-consuming. Furthermore, while large empirical effect of changes in cutting conditions (e.g., tool
databases have been compiled [3][4][5] to aid in geometry, cutting parameters, etc.) on industrially
process design, these databases lose relevance as new relevant decision criteria (e.g., wear rate, geometric
tool materials, machines, and workpiece materials are conformance, surface quality, etc.) can be predicted
developed. For example, in the development of high- without the need for new experiments. If robust
speed machining centers over the past ten years, predictive models can be developed, this approach
speeds and feed rates have increased by an order of would substantially reduce the cost of gathering
magnitude, rendering databases and handbook tables empirical data and would provide a platform for a priori
essentially useless. Setup/Changeovers
Lubrication 9%
A number of recent industrial studies have illustrated the 27%
shortcomings in the current empirical methods of Coolant Other 1% Start Up
designing machining processes. For example, in an Downtime
internal study, Kennametal gathered data over a one-
year period (1992-1993) on the global use of cutting 3% Speed10%
tools. This study found that in the United States: (1) the 2%
incorrect cutting tool is specified more than 50 % of the 2%
time; (2) tools are not used at the rated cutting speed in 8% Blocked/Starved
42 % of applications; and (3) tools are not used to the Fixtures 38%
end of life in 62 % of applications. An internal study at a
major automotive manufacturer tracked the downtime on Tooling
111 different machine tools. The results of this study, related
detailed in Figure 1, indicate that 35 % of this downtime
can be attributed to deficiencies in the performance of
Figure 1: Downtime on 111 machine tools at a
the machining process. If these studies are truly
major U.S. automotive company

1
Official contribution of the National Institute of Standards and Technology; not subject to copyright in the United States.
optimization of machining process parameters based the earliest scientific report on the formation of a chip
upon the physics of the system. was presented in 1881 by A. Mallock [9] in the
The difficulties in realizing true predictive models for Proceedings of the Royal Society of London. Nearly
machining arise from the extreme physical phenomena coincident with the publication of Mallock’s article, F. W.
inherent in the system. Machining generates a highly Taylor was appointed foreman of the machine shop at
inhomogeneous plastic flow where local stresses the Midvale Steel Company, and over the next 25 years
6 -1
generate high rates of plastic deformation (up to 10 s ) produced his now famous study of machining [3]. Thus,
th
that give rise to inhomogeneous thermal fields, high before the beginning of the 20 century, notable
temperatures (1200 ºC in machining steel), and high empirical and theoretical studies of machining were
pressures (10 MPa). This type of complex plastic flow is already underway.
difficult to predict even with sophisticated numerical The next period of development occurred in the 1930’s
software, and the basic data on material behavior under and 1940’s. The study of machining mechanics was for
such conditions is nonexistent for most materials of the first time placed on a solid physical and
practical interest [6][7]. These difficulties have forced mathematical foundation by the work of Piispanen [10],
model development to rely on various levels of empirical Ernst [11], and Merchant [12]-[15]. Since then, four
input data taken from machining tests (Figure 2) in order formal categories of cutting models have emerged: (1)
to model process variables of industrial interest. The analytic models; (2) slip-line models; (3) mechanistic
limitations imposed on the applicability of machining models; and (4) finite element models. Each approach
models by their reliance on empirical input data has has certain advantages and shortcomings. The choice
limited their industrial use, particularly in smaller of a particular cutting model depends on the information
operations which are unable or unwilling to perform desired, the required accuracy of this information, and
extensive validation testing. the available resources. Availability of laboratory
The goal of the Assessment of Machining Models effort equipment enabling accurate process measurements is
is to provide an unbiased and anonymous assessment of paramount importance to the development of accurate
of the ability of current machining models to predict the models. The literature on experimental measurements
practical behavior of machining processes. We make no in machining is extremely vast and therefore even a
attempt to restrict the definition of a model, other than to description of the most important results is beyond the
state that the models should be clearly defined in terms scope of this paper. However, comprehensive review
of the input data needed to make a prediction. For the articles and textbooks can be consulted for further
purposes of the effort, we define a correct prediction as information and references [6][7][8].
one that agrees with an experimental result to within a Analytic models [16-22] establish relations between
well-quantified experimental uncertainty. Thus to define force components (e.g., between cutting and thrust
an accurate prediction, the uncertainty inherent in forces and normal and tangential forces) based on the
machining systems must first be assessed by conducting cutting geometry. These models are easy to use, but
experiments at multiple labs on different machines. The require prior knowledge of the shear angle, mean-friction
intent of the effort is to generate an experimental data angle, and chip-flow angle. These quantities must be
set encompassing the inherent uncertainties associated determined experimentally, which limits the applicability
with multiple laboratories and machining centers, provide and the accuracy of these models.
an unbiased report of current capabilities for predicting Slip-line models [23-33] depend solely on material
the practical behavior of machining operations, and properties, rather than on experimental data. These
develop a roadmap for future directions in machining models predict mechanical response and temperature
modeling research. distributions and are compatible with strain, strain-rate,
and temperature-dependent models. However, the
2 BACKGROUND geometry of the slip-line field in the shear zone must be
assumed.
Machining research is driven by both ardent scientific
curiosity and tremendous practical and financial utility. Mechanistic modeling is a semi-empirical method
This dichotomy of motivations pervades the published capable of accurate prediction of cutting forces in a wide
literature. However, even though these two motivations range of complex machining operations [34-47]. This
are often at odds, neither can truly advance without the approach is based on the assumption that cutting forces
other. Without the practical utility, scientific studies are are proportional to the uncut chip area. The constant of
simply curiosities, and yet without a systematic scientific proportionality, called the specific cutting energy,
approach, the repeatability and overall utility of empirical depends on the workpiece material, the cutting
machining studies is severely limited. conditions, and the cutting geometry. The form of the
function relating the specific cutting energy to the cutting
geometry and conditions is assumed. The actual
3D, Empirical, Mechanistic
function is then determined by fitting experimental data
in a process called calibration. Calibration can be based
3D, Analytic on simple orthogonal or oblique machining set-ups;
geometric transformations can then be applied to predict
2D, FEM cutting forces for a complex, three-dimensional
machining process [43]. This simplifies the calibration
2D, Thick Zone, Slip-line set-up, but the need for testing is not eliminated and
other important quantities such as tool chip interface
2D, Thin Zone, Analytic temperatures are not predicted.
Finite element models for machining processes were
introduced in the early seventies [48,49]. Stevenson et
1937 1947 1957 1967 1977 1987 1997 al. conducted pioneering work in thermal finite element
analysis of machining [50,51]. Lajczok [52] proposed a
Figure 2: Timeline showing the history of the cutting model based on plane strain assumptions.
development of models. Natarajan and Jeelani used a viscoplastic model to
The research literature on machining problems is vast predict chip geometry [53]. Usui et. al. employed an
(see for example Komanduri [8], Shaw [6]). Probably incremental, elastic-plastic finite element model, starting
from a steady-state solution [54,55]. Strenkowski and complex shape). Unfortunately, these complexities and
Carroll [56,57,58] also did pioneering work in finite uncertainties often represent an unacceptable risk for
element simulation of machining processes. More the potential end-users of such models, who currently
recently Marusich et al. [59] and Ceretti et al. [60,61,62] resort to heavily empirical (and costly) but reliable
have developed more sophisticated schemes with models such as Taylor tool-life curves.
adaptive remeshing that are capable of predicting non-
steady chip formation. Today, the use of numerical
cutting models is no longer limited to research; industry 3 OBJECTIVES
is starting to adapt these methods to practical design In an attempt to focus and better define the current state
applications. For example, Wayne et. al. make use of a of machining research the following question was raised
st
commercial finite element code to predict chip flow [63]. at the 1 International CIRP Workshop on Modeling of
Currently a major limit to the use of finite element Machining Operations:
software is the lack of fundamental materials data “After 100 years of research in machining, why
relevant to the conditions that occur in machining [64]. does industry still rely on 100 year old models to
There are two main approaches to formulating finite make predictions of what will be seen on the
element models for machining, Lagrangian models and shop floor?”
Eulerian models. Lagrangian models track a fixed body - Rich Furness, Ford Motor Company.
of material as it approaches and passes the tool, while
Eulerian models trace the flow of material through a fixed The Assessment of Machining Models (AMM) effort
spatial domain in the vicinity of the tool. arose from discussions motivated by this question. The
goal of the project is to assess the ability of state-of-the-
Lagrangian methods are well suited to simulating the art machining models to make accurate predictions of
entry and exit phases of chip formation as well as the behavior of practical machining operations based
intermittent and discontinuous machining processes. upon the knowledge of machining parameters typically
They are compatible with elastoplastic material models available on a modern industrial shop floor. It is
[65,66]. On the other hand, Lagrangian models suffer a important to note that from the industrial point-of-view,
number of disadvantages. The large plastic approximate predictions that can be made from
deformations that occur in machining can generate machining parameters typically available on the shop
unacceptable mesh distortion. It is difficult to construct a floor would be more useful than more precise predictions
Lagrangian model for continuous tool advance through based on less readily available model parameters.
the workpiece; node-splitting methods [56-58,63,66] or
continual remeshing [65, 59,60,62] are required to Critical assessment of model performance and
address this problem. Artificial parting criteria are robustness is necessary before such models can see
required in node-splitting methods to determine when to widespread industrial use. In addition to assessing
advance the tool through the material [56-58,63,66]. It is current prediction capabilities, the Assessment of
also difficult to maintain grid alignment across slip Machining Models effort will assist in evaluating how
interfaces, a problem that can lead to errors in close the modeling community is to the goal of practical
predictions of thermal response. Lagrangian models are predictive modeling of machining and where future
incompatible with direct steady-state solutions and research efforts in this area should be focused. In
relatively small time steps are required in transient designing the effort it was asserted that any meaningful
solutions to capture the rapid changes that occur in the evaluation of model performance must have two
vicinity of the cutting tool. These features make characteristics. First, it must be based upon an accurate
Lagrangian finite element models computationally assessment of the uncertainty inherent in machining
expensive. operations. Second, predictions must be conducted
blindly (i.e., the predictor must have no a priori
Eulerian models avoid problems of mesh distortion, knowledge of the test results).
since the mesh is constructed on a fixed domain.
Relative material motion across a contact interface can To achieve the goal of the project, the following plan was
be accommodated without sacrificing mesh compatibility developed by the project participants:
[56-58,67-75]. Eulerian models capture the continuous 1. Provide a clear, consistent, well-measured, and
flow of material around the tool, enabling a physically relevant data set consisting of two components:
realistic model of machining of ductile materials, without a calibration data set that is fully disclosed; and
remeshing or node-splitting. Direct steady-state a validation data set that is not disclosed until
solutions can be computed with Eulerian models, so they predictions have been submitted. Assess the
are computationally far less expensive than Lagrangian uncertainties in the measurements.
models for this class of problems. In transient problems, 2. Release the calibration data set and the
the solution at a fixed point in an Eulerian frame evolves parameters for the validation data set.
more gradually than the solution at a fixed particle in a
Lagrangian model. Therefore, larger time steps can be 3. Solicit and collect voluntary blind predictions of
used without loss of accuracy in transient Eulerian the validation data set.
models. On the other hand, Eulerian models require 4. Provide an unbiased and anonymous reporting
custom software development and are less well suited to of the results. Organize a workshop to present
modeling intermittent machining, entry and exit phases, and discuss the results of the project, and
and discontinuous chip formation (combined Eulerian- develop a roadmap for future research in
Lagrangian (ALE) approaches [76] might address these predictive modeling of machining.
problems). These steps are discussed in more detail below.
The range and complexity of machining models across Data Generation: All data will be generated concurrently
these four formal categories of analytic, slip-line, at the four different labs (NIST, Ford Motor Company,
mechanistic and numerical models makes it extremely General Motors, and Caterpillar) and carefully
difficult to compare their utility and robustness. The compared prior to the solicitation of predictions. Lab-
quality of the output of models is uncertain, and there to-lab consistency provides a filter for identifying and
are no universal definitions of model inputs and outputs correcting inconsistent or spurious experimental data
(e.g., some models may predict specific cutting energy and thereby ensuring that the data is accurate. In
from basic materials properties while other models use addition, the mean and standard deviation of the data
specific cutting energy to predict the forces on a tool with
 will be calculated and used to define an accurate and (3) the workpiece and tool materials are easily
model, i.e., a model is considered accurate if it can obtainable in the configurations necessary for the tests.
consistently predict the standard deviations and the
means of the validation data set. 4.2 Test Configuration and Conditions
Data Release: Once consistent calibration and validation Three types of experiments are being performed: (1)
data sets are obtained, the calibration data set and orthogonal machining using coated and uncoated inserts
the parameters for the validation data set will be to generate a “calibration” data set; (2) orthogonal
publicly released via web site presentation, Internet machining using machining parameters that fall both
downloads, and data storage media (Zip Disks,
2 between and outside the parameters chosen for the
CD’s etc.). calibration data set as part one of a validation data set;
and (3) turning tests with uncoated inserts and with
Solicitation and Collection of Blind Predictions: coated grooved (chip-breaker) inserts as part two of a
Solicitations for predictions will be sent to all parties validation data set. These experiments were chosen first
who have expressed interest, and a general to provide data to calibrate any models as necessary
solicitation will be posted on the web. Predictions will and second to provide a validation data set that tests
be submitted on a voluntary basis. The due date for model capabilities against problems of increasing
submission of predictions will be six months following difficulty.
the release of the data.
Given the current understanding of the physics of
Reporting of Results: An unbiased report of the results machining, it is likely that prediction of changes from
will be developed based on comparison of the orthogonal to non-orthogonal or non-orthogonal with a
predictions and the experimental data. This report grooved insert is possible. However, because of
will provide the focus for a workshop aimed at changes in the tool-chip interface conditions, even if the
developing a roadmap for future work in the modeling properties of the coating are available, extrapolation from
of machining operations. The full calibration and uncoated to coated conditions is much more difficult.
validation data sets will be made available over the The coated chip breaker cutting tests are most difficult of
Internet for use in future modeling efforts. all to predict, yet they reflect current industrial practice.
Currently, we are nearing the end of the data generation
phase of the work. Approximately 30 international 4.2.1 Workpiece Material and Geometry
research groups have already expressed interest in The workpiece material was AISI 1045 steel bar
participating in this study based upon responses to obtained from a single batch/heat. The material was
presentations of the planned activities at the machined to produce bars and tubes at NIST (National
International Mechanical Engineering Congress and Institute of Standards and Technology) and then
Exhibition (IMECE) 1998 in Anaheim, the second CIRP distributed to the industrial laboratories. The dimensions
modeling meeting in Nantes, France, January 1999, the of test workpieces were:
North American Manufacturing Research Conference Tubes - 152.4 mm overhang, 101.6 mm
(NAMRC) 1999 in Berkeley, California, and IMECE 1999 diameter, 1.6 mm wall thickness
in Nashville, Tennessee. Release of the data is
Bars - 101.6 mm overhang, 101.6 mm
scheduled for the end of the summer, 2000. During and
diameter.
after each of these presentations, the plan was
strengthened in response to the comments and The material was subjected to chemical, mechanical,
questions that arose. The remainder of the paper and metallurgical analysis as detailed in the next section.
describes the design of the experiments for the effort 4.2.2 Tools and Toolholders
and presents some representative samples of the
experimental results. Kennametal CTAPR 123 B toolholders were chosen for
the experiments. These toolholders have three major
advantages: (1) they are commercially available and so
4 DESIGN OF THE EXPERIMENTS allow others to easily repeat the experiments in the
As stated above, the first step of the plan is to develop future; (2) they have a simple geometric configuration
an accurate data set relevant to a common machining with a 5º rake angle and a 0º included angle; and (3)
process used in industry. The choice of the process was they could be used for each of the cutting configurations.
primarily based upon input from the three industrial The inserts used were Kennametal grade K68. These
participants in the effort. All experiments will be done inserts have the advantages: (1) they are commercially
without coolant and will be repeated twice at each lab in available; (2) they consist of a very simple WC/Co
order to obtain a statistical sampling of behavior. material; and (3) they come in configurations compatible
with the toolholder and suitable for the desired cutting
4.1 Choice of Process conditions. Custom ground angle plates were
The process chosen was turning of American Iron and manufactured to mate with the toolholder and distributed
Steel Institute (AISI) 1045 steel using a general purpose by Kennametal to establish the desired rake angles of
tungsten carbide / cobalt (WC/Co) unalloyed carbide +5º, 0º and -7º without changing the toolholder.
grade insert. The simplest grade of carbide was chosen Observations of the initial tool edge radii for each of the
to simplify tool-material modeling. Both uncoated and types of inserts provide insight into the effect of tool
titanium nitride (TiN) coated inserts are used. The merits coating on edge radius. Tool replications produced
of this process are: (1) the machining of AISI 1045 steel using a dental replicant (Dentsply Hydrosil) were sliced
has significant relevance in the automotive and heavy and examined under an optical microscope. Pictures of
equipment industries; (2) the material properties of AISI the sliced tool replications will be made available as part
1045 steel and general grade carbide are well known; of the data release.
4.2.3 Machine type and characterization
2
Identification of specific commercial products is The experiments were carried out simultaneously in the
included in order to provide a complete description of the four test laboratories using CNC turret lathes as detailed
experimental work and does not imply endorsement by in Table 1. While the results of several internal round
the National Institute of Standards and Technology, nor robin tests done at one of the participating laboratories
does it imply that the commercial products identified are indicate that consistent machining behavior can be
necessarily the best for the purpose.
 Lab Machine Force Tool Wear conditions for test 8 in Table 2 are likely to lead to rapid
Industrial OKUMA Kistler Wild Heerbrug crater growth.
Lab 1 Cadet- MPS 11 4.2.5 Validation Data Set Parameters
L1420 Optical
The validation data set consist of two major portions,
X650 CNC Microscope
orthogonal cutting and turning tests. The orthogonal
(3500 rpm,
tests were conducted with the parameter values given in
22 kW)
Table 3. These parameters will test both the interpolation
Industrial Lelond Kistler Optical and extrapolation capability of the models. The turning
Lab 2 Makino 9257B Microscope tests are also chosen to require interpolation and
Baron 60 and extrapolation of the orthogonal cutting data. The cutting
Lathe (379403 conditions for these tests are summarized in Table 4.
kW) mount Each experiment will be conducted using uncoated flat-
Industrial Okuma Kistler Wyko white- faced inserts and TiN coated grooved inserts. As
Lab 3 LC-20 9257 B light mentioned above, this battery of experiments was
Lathe and Interferometer, chosen to provide a progressively increasing level of
(3000 rpm,
9403 SEM modeling difficulty.
15 kW) mount
4.3 Measured Quantities
NIST Hardinge Kistler Wyko white- Forces, temperatures and wear were measured using
Superslant
9257 B light the methods detailed below for the conditions detailed in
(4000 rpm,
and Interferometer, Tables 2-4. For all of the calibration tests in Table 2
7.5 kW) 9403 SEM, Air except test 5 and all of the validation orthogonal tests
mount Bearing LVDT (Table 3) no wear results were measured. Chip
Table 1: Experimental equipment at each of the four morphology measurements were made for the
labs. orthogonal cutting tests in the calibration data set.
These include measurements of the chip thickness and
width after machining. Because measurements of
Test Cutting Speed Feed Rake contact length can be ambiguous, contact length was not
No. (m/min) (µm/rev) Angle measured; however, it could be inferred for some of the
1 200 150 -7 tests from the wear patterns that occur (see section 5.5).
2 200 150 +5 4.3.1 Forces
3 200 300 -7 Forces were measured using Kistler 3-axis piezoelectric
dynamometers as detailed in Table 1. The tools were
4 200 300 +5 mounted to the dynamometer using Kistler toolholder
5** 300 150 -7 mounts as detailed in Table 1. The rake angles were
300 150 +5 established using custom ground angle brackets from
6
Kennametal. An example of the dynamometer
7 300 300 -7 toolholder set-up is shown in Figure 3.
8 300 300 +5 4.3.2 Temperatures
Table 2: Parameters for orthogonal cutting tests for Average temperatures were measured using two
model calibration. methods. All tests were measured using an intrinsic
thermocouple at industrial lab 3 and NIST, and selected
obtained on different equipment, it has been reported
measurements are being made at NIST using IR
elsewhere that variations in dynamic behavior of
microscopy.
machine tools can affect results. Particularly, rates of
tool wear have been reported to be affected by the
dynamics of the structural loop even under stable (non- Test Cut Feed Rake
chattering) cutting conditions. Thus, the dynamics of No. Speed (µm/rev) Angle
each machine tool will be characterized by NIST (with (m/min)
the dynamometer in the structural loop) using tool-tip
frequency response functions measured with a 1 225 200 0
piezoelectric hammer, signal analyzer, and 2 275 250 -7
accelerometer. 3 400 400 +5
4.2.4 Calibration Data Set Parameters 4 400 100 -7
A total of 16 orthogonal cutting tests were done to Table 3: Parameters for the orthogonal cutting portion of
generate the calibration data set. The parameter values the validation tests.
for both the coated and uncoated tools are shown in
Table 2. Test 5 is an accelerated wear test in which tool
wear was measured as described below. The wear test Test Cutting Feed Rake Depth
is intended to provide data that can be used by modelers No. Speed (µm/rev) Angle of Cut
to “calibrate” the parameters in diffusion wear models. (m/min) (mm)
The parameter values were chosen to span a range of 1 180 125 0 1
behaviors. According to data in Trent [7], all of these
2 250 225 -7 2
conditions will be free of a built-up edge, which would
lead to unpredictable behavior and be impractical. 3 350 375 +5 1
According to Trent [7], machining of a low carbon steel 4 350 125 -7 3
with uncoated carbide under most of these conditions
will lead to slow steady crater growth. Only the Table 4: Parameters for turning portion of the validation
tests.
 some limited thermal maps of machining AISI 1045 steel
for the AMM project.

4.3.3 Wear
Wear measurements were conducted according to ISO
3685. The following methods of measuring wear
patterns were investigated and compared: (1) air bearing
linearly variable differential transducer (LVDT) (crater
depth); (2) white-light interferometer (500 micrometer
vertical travel with 10 nm resolution and 5 mm by 5 mm
field of view using a 2.5x objective); (3) replication
techniques; (4) optical microscope; and (5) scanning
electron microscope. The first three methods were used
to quantitatively measure crater geometry. The optical
microscope was used to measure crater depth and flank
Figure 3 : Photograph of the tool, dynamometer, and the wear shape. The SEM was used primarily for qualitative
intrinsic thermocouple. imaging of worn tools.
The intrinsic thermocouple uses the bi-conducting tool- 4.3.4 Chip Morphology
chip interface as a thermocouple junction. First the For each of the tests the machined chips were collected
toolholder was electrically isolated from the remainder of and classified according to type as specified in ISO 3685
the machine using a Bakelite spacer between the Table G1. For the orthogonal cutting tests, the thickness
dynamometer and the Kistler toolholder mount. Bakelite and width of the chips were measured after machining.
washers were also placed between the screw heads and By combining these measurements and the forces a
the toolholder. One wire was then mounted to a slip ring table (see below) of relevant physical quantities (e.g.,
on the back of the spindle and the other to the tool shear angle, shear and friction forces, etc.) was
holder. The calibration procedure for the intrinsic populated using the classical Merchant model.
thermocouples involves placing the Kennametal
toolholder in contact with a workpiece and observing the
output voltage while heating the toolholder and
workpiece at the contact point and measuring the
temperature with a conventional thermocouple. The
intrinsic thermocouple systems have not yet been
calibrated. As such, only comparisons of the voltages
measured by NIST and industrial lab 3 can be made until 1
the systems have been calibrated. While this method
has been used for some time in experimental research 4
[77,78,79,80] it does have the following documented 5
2
shortcomings: (1) it reports some weighted integration of
the temperatures across the tool-chip interface surface;
(2) it is affected by other bi-conductor interfaces in the
measurement loop; and (3) it is affected by (unknown)
fluctuations in tool-chip contact area. However, this 3
method remains one of the most robust and reliable
methods for assessing changes in mean tool-chip
interface temperatures. Therefore, it was chosen as the
main method of temperature measurement for this effort.
(a)
As part of a separate project at NIST a thermal imaging
micro-pyrometer (MIPY) was constructed using a Chip Tool
commercial 128 by 128 indium antimony (InSb) focal V
plane array (FPA) with an all-reflective 0.5 numerical
aperture (NA) and 15x microscope objective that directly
focused the image of the object on the FPA. The InSb Ω
detector was used with a broad-band spectral filter,
which transmits from 3 P WR  P LQ ZDYHOHQJWK DQG
the spatial resolution of the system was estimated to be
approximately equal to the wavelength of the detected
light from using the Rayleigh criterion. The individual
pixels in the FPA were deposited 50 P DSDUW UHVXOWLQJ Shear
in a spatial resolution of at best < 5 PSHUSL[HODQGWKH Zone
spatial resolution was further verified using a chrome-on- Feed
(f)
glass USAF 1951 resolution target. The system was
calibrated using a NIST miniature blackbody with a 2 mm
aperture. The emissivity of the steel was measured as a (b)
function of temperature using a reflectance technique.
Due to the nonlinear calibration curve of the detector and Figure 4: (a) Photograph of the experim ental system
the nonlinear temperature variation of the emmissivity, showing: (1) air bearing spindle; (2) AISI 1045 steel
an iterative procedure was required to obtain material tube; (3) m onolithic tool post; (4) zero rake angle
temperature from the measured signals. To reduce error tungsten carbide insert; (5) m icro-pyrom etry system .
motions and maintain focus the thermal microscopy (b) Schem atic diagram of orthogonal cutting, a cutting
system was mounted on a high-load capacity air-bearing configuration that generates a nearly two-dim ensional
spindle configured to conduct orthogonal cutting as plastic flow of m aterial.
shown in Figure 4a. This system will be used to provide
5 GENERATION OF THE CALIBRATION DATA SET
This section details the generation of the calibration data
set and provides some sample results that demonstrate
the approach and the level of lab-to-lab consistency
obtained.
5.1 Material Characterization
The material was ordered as a single batch of AISI 1045
steel, 30.5 m (100 ft) of 101.6 mm (4 in) diameter rod in
5 rods, each 6.1 m (20 ft) long. The 5 rods were cut into
a combination of 152.4 mm (6 in) and 203.2 mm (8 in)
long bars, with a 50.8 mm (2 in) thick circular sample
removed from each end. The ten samples were
subjected to chemical and metallurgical analysis using
American Society for Testing and Materials (ASTM) test
procedures E3-95, E407-93, E112-95, E1019-94, and
E415-95a. Grain sizes were measured quantitatively
using an optical microscope. Additionally, Brinell Figure 5: Micrographs of polished and etched
hardness measurements and a series of ultrasonic longitudinal and transverse sections of bar caps.
nondestructive evaluations were performed on a Element Mean Std Mil- ISO
sampling of the bars. spec
Err C45 E4
The chemical analysis results are shown in Table 5. The
Carbon 0.445 0.005 0.450 0.42-0.50
chemical content did not vary significantly among the ten
samples as indicated by the average and standard Manganese 0.677 0.009 0.700 0.50-0.80
deviation of the measurements given in Table 5. The Phosphorus 0.010 0.001 0.009 0.035
material conforms to the military specification (milspec)
for AISI 1045, and satisfies the more stringent ISO C45 Sulfur 0.011 0.001 0.008 0.035
standard except for copper content, which is slightly Silicon 0.247 0.005 0.260 0.10-0.40
high. The grain size ranges were measured for each Nickel 0.080 0.001 n.a. 0.200
sample. The lower and upper bounds of these ranges
averaged 3.4 and 7.55, with standard deviations and Chromium 0.144 0.005 n.a. 0.150
repeatability uncertainties of 0.7 and 0.1, respectively, Molybdenum 0.010 0.001 n.a. 0.050
for both the upper and the lower bounds. The resulting Copper 0.234 0.005 n.a. 0.200
±2s expanded uncertainty [81] on the upper and lower
bounds is ±0.71. Four representative pictures of Aluminum 0.020 0.001 n.a.
longitudinal and transverse cross-sections from two of Nitrogen n.a. n.a. n.a. 0.003-
the samples are shown in Figure 5. 0.008
Brinell Hardness measurements were performed on two Table 5: Military specification (milspec), mean and
of the 150-mm long workpiece bars with a 29.4 kN load standard error of chemical concentrations and the
and 10-mm ball. For each of the bars, six specification given in ISO 3685 for workpiece material
measurements were made on one face approximately AISI 1045.
1 cm away from the center of the face in a circular The results of the chemical analysis, hardness tests, and
pattern around the center. The diameters of the ultrasonic examinations establish that the original
resulting impressions left on the bars were measured at material was mechanically and chemically suitable for
4.3 mm each with a measurement precision of ±0.025 use in the machining tests.
mm. This equates to a Brinell Hardness Number (BHN)
with a ±2s expanded uncertainty of 196 ±5 [81]. 5.2 Preparation of workpieces
Ultrasonic examinations of the workpiece material were The material arrived at NIST in 5 bars each
also performed on four of the 150-mm long bars before approximately 6 m in length. The bars were rough-
they were machined into tubes. Two different types of sawed into shorter bars 150 mm and 200 mm long for
pulse-echo ultrasonic contact tests were performed with use in the orthogonal and turning tests, respectively.
an orientation parallel to the axis of the cylinder. One The short bars were then faced and numbered in the
using longitudinally oriented waves, and the other using order they were cut from the long bars. Parts were
transverse waves. The results of the two pulse-echo stamped with a 3 digit number, with the first digit for the
ultrasonic tests enable the speed of sound through the bar number out of the original 5 and the second and third
material to be estimated, and thus the consistency of the digits for the part number within the bar (01-36). In order
bulk elastic modulus and density can be characterized. to address the possibility of variation among the tubes by
On each of four bars, measurements of length were location within the original bar, the bars were distributed
taken with a caliper, and elapsed time between to the four labs in a staggered order. The outer diameter
ultrasonic pulses and echoes were recorded on an of the 200 mm bars was finish turned to remove scale.
oscilloscope. The measured values of the speed of The 150 mm bars were machined into tubes for
longitudinal and transverse ultrasonic waves in the bars orthogonal cutting through a series of operations
were found to be in reasonable agreement with the including drilling, rough and finish boring of the internal
handbook values [82] for steel, 5,900 m/s and 3,230 m/s diameter, and finish turning of the outer diameter. The
respectively. cores of the 150 mm bars were drilled out to a depth of
Additionally, the transducer was moved continuously in 100 mm on a manual lathe using a 31.75 mm (1.25 in)
circular rings around the center of the bars’ circular diameter 118° included angle tapered shank drill bit.
ends. No distinguishing features were observed among The drilled hole was then rough bored at 245 rpm
the four bars judging from the echo amplitude pattern. spindle speed, 22.4 mm/min feed rate, and 2.54 mm
Had there been any large differences in the density or depth of cut. Finish boring was performed at 165 rpm
microstructure of the material a visible difference would spindle speed, 38.1 mm/min feed rate and 0.56 mm
have been observed in the echo pattern. depth of cut. All boring was performed with a Carboloy
S16-SCLCL-3 boring bar and a SECO CCMT 32.52-F2 Figure 9 shows the averages and ±2s expanded
TP30 insert. Finish turning of the outer diameter was uncertainties of the intrinsic work-tool thermocouple
performed with a Kennametal CN42 80° Diamond K68 voltage outputs for each of the calibration data set tests
grade coated insert at 300 rpm spindle speed, 38.1 with uncoated inserts. The clearest trend in the changes
mm/min feed rate, and 0.89 mm depth of cut. The length of thermocouple voltage with test number are between
of cut for all operations was 101.6 mm. Wall thickness test 1-2 and 3-4, where the rake angle changes from –7°
measurements were performed to ensure consistency. to +5°, and between tests 1-4 and tests 5-8, where the
The ±2s expanded uncertainty [81] for the 1.6 mm wall changes between the ordered pairs from these sets (1
thickness was ±0.05 mm. and 5, 2 and 6, etc.) are an increase in cutting speed
from 200 m/min to 300 m/min.
5.3 Characterization of Experimental Apparatus
Figure 10 shows a representative thermal profile of
The tool holder and mount and dynamometer were machining with the cutting conditions indicated in the
assembled as shown in Figure 3. The dynamic stiffness caption. While these conditions are not as aggressive
of both of the lathes to be used for the experiments at as those used in the calibration data set, they do serve
NIST was measured. The frequency response function to demonstrate the resolution and capability of the
(FRF) for the Hardinge Superslant is given in Figure 6. system. The white lines on the figure represent the tool
The ±2s expanded uncertainties for the peak/valley and workpiece surfaces in the plane. The peak
heights and frequencies are ±10 % and ±5 %, temperature is approximately 650 °C and occurs at a
respectively. SRLQW DERXW  P IURP WKH FXWWLQJ HGJH RI WKH WRRO
5.4 Forces This is consistent with other findings as well as tool wear
patterns and is a result of the highly localized plastic flow
While the raw data for the calibration cutting experiments that occurs at the boundary of the tool and chip. The
will be released in its entirety, full presentation of the raw ±2s expanded uncertainties in these measurements from
data would be impractical in this paper. Instead, all sources are less than ±30 °C.
representative samples of the wear progressions are
given, and averages of the forces and intrinsic
thermocouple voltages are given for the full data set. The 6 FUTURE WORK AND TIMETABLE
average forces for the orthogonal cutting tests with Currently, the calibration data set is nearly complete.
coated and uncoated flat inserts for experiments The future steps in the effort are described below.
conducted at all four laboratories are shown in Figure 7.
In this graph, the cutting forces are negative and the 6.1 Release of Calibration Data
thrust forces are positive, as produced by the Kistler The calibration data will be released during the summer
Dynamometer in the cutting setup used. The lines of 2000. The primary method of release will be through
represent ±2sI expanded uncertainty for the set of all downloads from the project’s web site, located at
force averages from the four labs for test condition i. The http://www.nist.gov/amm/. The web site is used for
consistency between the four laboratories is sufficient to reporting progress, disseminating the model reference
allow differences in forces due to parameter variations to data and input parameters as described above, and
be discerned. collecting prediction reports. The calibration data set will
5.5 Accelerated Wear Test be provided in a spreadsheet format that will include a
template for submitting results. At the time of data
Figure 8 shows two wear progressions for experiments release, all groups who have expressed an interest in
conducted at NIST. Cutting was stopped every 5 mm of submitting predictions will be contacted via email.
tube length and the insert was examined with an optical
microscope, an LVDT air bearing indicator, and a white- 6.2 Solicitation and Collection of Predictions
light interferometer. Wear measurements were taken All parties who have expressed interest in AMM at
following procedures detailed in ISO 3685. The two various workshops will be invited to provide predictions.
wear progressions shown are for nominally identical After the release of the calibration data set, a six-month
cutting conditions and demonstrate a high degree of deadline for submission of predictions will be
repeatability in the wear progression. The squares and established. By that deadline all prediction reports will
diamonds represent LVDT measurements of peak crater be electronically submitted to NIST via the project web
depth. The triangles represent measurements of peak site.
crater depth from the interferometer. The five
interferograms used are shown inset on the figure 6.3 Evaluation and Reporting of Results
adjacent to the corresponding data points. A committee of representatives from the four industrial
5.6 Chip Morphology participants will meet to discuss the results. At this
meeting the results will be discussed and a final report
For each of the calibration tests in Table 2, the final chip will be generated. The report will objectively and
thickness and width was measured, and the shear angle impartially document the results and make suggestions
was calculated based upon the assumption that the for future work in the area of modeling of machining.
0HUFKDQW PRGHO ZDV YDOLG  7KH VKHDU DQJOH  ZDV Also at this time, a workshop at NIST will be held to
FDOFXODWHG DV DWDQ>U FRV U VLQ @ ZKHUH  LV discuss future work. All of the modeling participants will
the rake angle, and r is the chip thickness ratio. It is be invited to attend the workshop. Some of the modeling
important to note that the chip width varies considerably, participants may be invited to make presentations at this
which violates the assumptions for the Merchant model. workshop.
The ±2s expanded uncertainty for NIST measurements
is less than ±100 µm. The averages of chip
measurements are given in Table 6. 7 SUMMARY
5.7 Representative Temperature Data The project plan and experimental procedure were
designed to result in acceptable control of experimental
The primary temperature measurement method was the consistency. Based on the results presented in this
intrinsic thermocouple. The signals measured through paper, the lab-to-lab consistency is acceptably good.
the coated inserts were not found to be sufficiently This effort will result in an unbiased assessment of the
consistent to be a meaningful representation of the current capabilities of machining models.
average temperature at the tool-chip interface surface.
ACKNOWLEDGEMENTS
We would like to take this opportunity to thank the many
people who have contributed to the development and
implementation of this effort, including Ranajit Ghosh,
Ming Leu, Vivek Chandrasekharan, Bob Polvani, Tony
Schmitz, Dieter Ebner, Wendy Shefelbine, and Gerry
Blessing. This work was supported in part by National
Science Foundation Grant Number 9909130.
Test Uncut Cut Shear Width Lab
No. Chip Chip Angle (µm)
Thick- Thick- (deg)
ness ness
(µm) (µm)
1 150 523 15.38 2050 Lab 2
513 15.65 2060 Lab 2
2 150 501 17.03 1956 NIST
465 18.30 1949 NIST
462 18.41 1839 Lab 2
3 300 889 17.83 2637 Lab 2
894 17.74 2611 Lab 2
4 300 671 24.86 1891 NIST
682 24.50 1990 NIST
721 23.27 2108 Lab 2
5 150 904 9.17 2626 Lab 2
437 18.11 1976 Lab 2
6 150 392 21.52 1778 NIST
340 24.56 1742 Lab 2
503 19.96 1930 Lab 2
7 300 n.a. n.a. n.a. n.a.
8 300 787 21.44 2154 Lab 2
757 22.24 1986 Lab 2
9 150 460 17.29 2108 NIST
454 17.50 2061 NIST
472 16.89 2113 Lab 2
475 16.79 2098 Lab 2
10 150 441 19.25 1870 NIST
416 20.35 1826 NIST
417 20.30 1875 Lab 2
478 17.82 1890 Lab 2
11 300 803 19.53 2696 NIST
819 19.19 2743 NIST
772 20.22 2490 Lab 2
757 20.57 2576 Lab 2
12 300 683 24.46 1953 NIST
620 26.71 2012 Lab 2
676 24.70 2037 Lab 2
13 150 435 18.18 2032 NIST
425 18.56 2003 NIST
478 16.70 2047 Lab 2
452 17.57 2063 Lab 2
14 150 375 22.43 1781 NIST
371 22.66 1809 Lab 2
366 22.95 1758 Lab 2
15 300 660 23.15 2464 NIST
743 20.90 2499 NIST
16 300 600 27.51 1852 NIST
640 25.96 1870 NIST
666 25.04 1895 Lab 2
671 24.86 1961 Lab 2
Table 6: Measurements of chip thickness and width.
Shear angles are calculated from a Merchant model
assumption.
 Frequency Response Function for Lathe at NIST

0.0000003

0.0000002

0.0000001
Real (m/N)

-0.0000001

-0.0000002

-0.0000003
1.00E-06 100 200 300 400 500
Frequency (Hz)

Frequency Response Function for Lathe at NIST

-0.0000001

-0.0000002
Imaginary (m/N)

-0.0000003

-0.0000004

-0.0000005

-0.0000006

-0.0000007
1.00E-06 100 200 300 400 500
Frequency (Hz)

Figure 6 : Real and imaginary components of the frequency response measured at the tool tip on the NIST Hardinge lathe.
The x- y- and z- components are shown in solid, dashed, and dash-dotted lines, respectively. The ±2s expanded
uncertainties on the peak/valley heights and frequencies are ±10 % and ±5 %, respectively.
 AMM Cutting Data NIST, Cat, GM
Tangential (Cutting) Forces, Uncoated Inserts

1100

1000

900

800

700

600

500

400

300

200

100

0
Cutting Force (N)

-100

-200

-300

-400

-500

-600

-700

-800

-900

-1000

-1100

-1200

-1300

-1400
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Test Number

Figure 7 : Force data for the calibration tests from all four labs. NIST data is shown as diamonds, industrial labs 1, 2, and
3 are shown as squares, circles, and crosses, respectively. The lines bounding the data points represent ±2s expanded
uncertainty on the data points for all four labs. The upper trace is thrust force, the middle trace is the out of plane force,
and the lower (negative) force is the cutting force.
 140

120

100
Crater Depth (um)

80

60

40

20

0
0 20 40 60 80 100 120 140 160
Length of Cut (m)

Figure 8: Measured tool wear progression in two orthogonal wear tests showing white light interferometer measurements
(triangles and overlays) and LVDT measurements (squares and diamonds). Overlays are three-dimensional crater depth
plots, with the cutting edge at the top. The depth scale for these overlays is shown on the right. Expanded uncertainties
(±2s) on LVDT measurements and interferometer measurements are ±5µm.

Thermocouple Voltage vs. Test Number

0.016

0.014
Thermocouple Voltage (V)

0.012

0.01

0.008

0.006

0.004

0.002

0
1 2 3 4 5 6 7 8
Test Number

Figure 9: Average thermocouple voltages for uncoated calibration tests. Diamonds and crosses represent measurements
performed at NIST and industrial lab 3, respectively. Dashes represent ±2s expanded uncertainties on data points from
both labs.
 Tool

T (°C)
Y (µm)

Workpiece

X (µm)

Figure 10: Temperature map of orthogonal cutting with a zero-rake cutting tool using a cutting
speed of 3.7 m/s, and an uncut chip thickness of 50 micrometers.

REFERENCES [16] Ernst, H., Merchant, M.E., 1941, Chip formation,

[1] Merchant, E., 1998, Proceedings of the CIRP friction and high quality machined surfaces, Surface
International Workshop on Modeling of Machining Treatment of Metals, Vol. 29, p. 299.
Operations, Atlanta, GA, USA. [17] Merchant, M.E., 1945, Mechanics of the cutting
[2] Soons, H.A., Yaniv, S.L., 1995, Precision in process, Journal of Applied Physics, Vol. 16, p. 267 -
machining : research challenges, Gaithersburg, MD : 318.
National Institute of Standards and Technology, [18] Lee, E.H., Schaffer, B.W., 1951, The theory of
NISTIR 5628. plasticity applied to a problem of machining, Journal
[3] Taylor, F. W., 1907, Trans. ASME 28:31. of Applied Mechanics, Vol. 18(4), p. 405.
[4] Koenigsburger, F., 1964, Design Principals of Metal [19] Rubenstein, C. ,1983, The mechanics of continuous
Cutting Machine Tools, Peragamon, Oxford. chip formation in oblique cutting in the absence of
chip distortion, part 1 – theory, International Journal
[5] Machining Data Handbook, 3rd Edition, 1980, Vol. 1 of Machine Tool Design and Research, Vol. 23, No.1,
& 2, Metcut Reseach Associates Inc., Cinncinnatti, pp. 11-20.
Ohio, USA.
[20] Lau, W. S., Rubenstein, C., 1983, The mechanics of
[6] Shaw, M. C., 1984, Metal Cutting Principals, Oxford continuous chip formation in oblique cutting in the
Press, Oxford, UK. absence of chip distortion, part 2 - comparison of
[7] Trent, 1991, Metal Cutting, Butterworth-Heinemann, experimental data with the deductions from theory,
Oxford. International Journal of Machine Tool Design and
[8] Komanduri, R., 1993, Appl Mech Rev, 46(3), 80-129. Research, Vol. 23, No.1, pp. 21 -37.
[9] Mallock, A., 1881, Proc. Royal Soc (Lon), 33:127- [21] Grzesik, W., 1990, The mechanics of continuous
139. chip formation in oblique cutting with single-edged
tool - part I. theory, International Journal of Machine
[10] Piispanen, V., 1937, Teknillan Aiakakauslenti 27:315. Tool Manufacturing, Vol. 30, No. 3, pp. 359-371.
[11] Ernst, H., 1938, Physics of Metal Cutting, Machining [22] Grzesik, W., 1990, The mechanics of continuous
of Metals, ASM, 1-34. chip formation in obliquecutting with single-edged
[12] Ernst, H., Merchant, M. E., 1941, Trans. ASME tool - part II. experimental verification of the theory,
29:299. International Journal of Machine Tool Manufacturing,
[13] Merchant, M. E., 1945, Trans. ASME, 66:A65-A71. Vol. 30, No. 3, pp. 373-388.
[23] Palmer, W.B., Oxley, P.L.B., 1959, Mechanics of
[14] Merchant, M. E., 1945, J. App. Phys.,16:267.
metal cutting, Proceedings of Institute of Mechanical
[15] Merchant, M. E., 1945, J. App. Phys., 16:318. Engineers, Vol. 173, p. 623.
[24] Oxley, P.L.B., 1961, A strain hardening solution for
the shear angle in orthogonal metal cutting,
 International Journal of Mechanical Sciences, Vol. 3, [43] Endres, W.J., DeVor, R.E., Kapoor, S.G., 1993, A
p. 68. dual-mechanism approach to the prediction of
[25] Johnson, W., 1962, Some slip-line fields for swaging machining forces, Parts 1 and 2, Proc., ASME Sym.
or expanding, indenting, extruding and machining on Modeling, Monitoring and Control Issues in
with curved dies, International Journal of Mechanical Manufac., PED-Vol. 64, pp. 563-593.
Sciences, Vol. 4, p. 323. [44] Jawahir, I. S., 1986, Ph.D. Thesis, University of New
[26] Oxley, P.L.B., Hatton, A.P., 1963, Shear angle South Whales.
solution based on experimental shear zone and tool- [45] Amarego, E.J.A., Brown, R.H., 1969, The Machining
chip interface stress distribution,International Journal of Metals, Prentice Hall, Engelwood Cliffs New
of Mechanical Sciences, Vol. 5, p. 41. Jersey.
[27] Oxley, P.L.B., Welsh, M.J.M., 1963, Calculating the [46] Budak, E., Altintas, Y., Amarego, E.J.A., 1996, J.
shear angle in orthogonal metal cutting from Man. Sci. Eng., 118: 216-224.
fundamental stress, strain, strain-rate properties of [47] Amarego, E.J.A., Uthaichaya, M., Journal of
the work material, Proceedings 4th International Engineering Production 1(1):1-18.
Machine Tool Design and Research Conference.
Pergamon, Oxford, p. 73. [48] Okushima, K., Kakino, Y., 1971, The residual stress
produced by metal cutting, Annals of CIRP, Vol.
[28] Usui, E., Hoshi, K., 1963, Slip-line fields in metal 20(1), p. 13.
cutting involving centered fan fields, International
Research in Production Engineering ASME, p. 61. [49] Klamecki, B.E., 1973, Incipient chip formation in
metal cutting - a three dimension finite element
[29] Kudo, H., 1965, Some new slip-line solutions for two- analysis, Ph.D. Dissertation, University of Illinois at
dimensional steady-state machining, International Urbana-Champaign.
Journal of Mechanical Sciences, Vol. 7, p. 43.
[50] Tay, A.O., Stevenson, M.G., Davis, G.D.V., 1974,
[30] Fenton, R.G., Oxley, P.L.B., 1968, Mechanics of Using finite element method to determine the
orthogonal machining : allowing for the effects of temperature distribution in orthogonal machining,
strain-rate and temperature on tool-chip friction, Proceedings of Institute of Mechanical Engineers,
Proceedings of Institute of Mechanical Engineering. Vol. 188, p. 627.
[31] Dewhurst, P., Collins, I.F., 1973, A matrix technique [51] Stevenson, M.G., Wright, P.K., Chow, J.G., 1983,
for constructing slip-line field solutions to a class of Further developments in applying finite element
plane-strain plasticity problems, International Journal technique to the calculation of temperature
of Numerical Methods in Engineering, Vol. 7, p. 357. distributions in machining and comparisons with
[32] Dewhurst, P., 1978, On the non-uniqueness of the experiment," Journal of Industry, Vol. 105, p. 149.
machining process, Proceedings of Royal Society, [52] Lajczok, M.R., 1980, A study of some aspects of
London, Vol. A360, p. 587. metal machining using finite element method, PhD
[33] Dewhurst, P., 1979, The effect of chip breaker Dissertation, North Carolina State University.
constraints on mechanics of machining, Annals of [53] Natarajan, R., Jeelani, S., 1983, Residual stresses in
CIRP, Vol. 28, p. 1-5. machining using finite element method, Computers in
[34] Martelloti, M.E., 1941, An analysis of the milling Engineering, Computer Software and Applications,
process, Trans. ASME, Vol. 63, p. 667. Vol. 3, p. 79.
[35] Martelloti, M.E., 1945, An analysis of the milling [54] Shirakashi, T., Usui, E., 1974, Simulation of
process II : down milling, Trans. ASME, Vol. 67, p. orthogonal cutting mechanism, Proceedings of
233. International Conference of Production Engineering,
[36] Koenigsberger, F., Sabberwal, A.J.P.,1960, Chip Tokyo, p. 535.
section and cutting force during the milling operation, [55] Usui, E., Shirakashi, T., 1982, Mechanics of
Anals of CIRP. machining from descriptive to predictive theory, On
[37] Kline, W.A., DeVor, R.E., 1983, The effect of runout the Art of Cutting Metals - 75 Years Later, ASME
on cutting geometry and forces in end milling, Publication PED, Vol. 7, p. 13.
International Journal of Machine Tool Design and [56] Strenkowski, J.S., Carroll III, J.T., 1986, An
Research, Vol 23, No. 2/3, p. 123-140. orthogonal metal cutting model based on an Eulerian
[38] Nakayama, K., Arai, M., Takei, K., 1983, Semi- finite element method, Manufacturing Processes,
empirical equations for three components of resultant Machines and Systems, Proceedings of the 13th
cutting force, Anals of CIRP, Vol. 30, No. 1, p. 33. NSF Conference on Production Research and
Technology.
[39] Fu, H.J., DeVor, R.E., Kapoor, S.G., 1984, A
mechanistic model for the prediction of the force [57] Carroll III, J. T., 1986, A numerical and experimental
system in face milling operations, ASME Journal of study of single point diamond machining, PhD
Engineering for Industry, Vol. 106, p. 81-88. Thesis, North Carolina State University.
[40] Subramani, G.R., Suvada, R., Kapoor, S.G., DeVor, [58] Carroll III, J.T., Strenkowski, J.S., 1988, Finite
R.E., Meingast, W., 1987, A model for the prediction Element Models of Orthogonal Cutting with
of force system for cylinder boring process, Proc. Application to Single Point Diamond Turning,
NAMRC 15, pp. 439-446. International Journal of Mechanical Sciences, Vol.
30, p. 899-920.
[41] Smith, S. and Tlusty, J., 1988, Modeling and the
simulation of the milling process, ASME, PED-Vol. [59] Marusich, T. D., Ortiz, M., 1995, Int J. Numer.
33, Proc. Winter Annual Meeting, pp. 17-26. Methods Eng. 38:3675.
[42] Chandrasekharan, V., Kapoor, S.G., DeVor, R.E., [60] Ceretti, E., Fallbohmer, P., Wu, W.T., Altan, T., Int. J.
1993, A mechanistic approach to predicting the Mat. Proc. Tech., 59:169.
cutting forces in drilling : with application to fiber- [61] Maekawa, K., 1999, Application of 3D Machining
reinforced composite materials, ASME, MD-Vol. Modeling to Cutting Tool Design, Proceedings of the
45/PED-Vol. 66, Proc. Machining of Advanced nd
II CIRP Int’l Workshop on Modeling of Machining
Composites, pp. 33 - 51. Operations, Nantes, France.
[62] Ceretti, E., 1999, Numerical Study of Segmented [77] Herbert, E.G., 1926, The Measurement of Cutting
Chop Formation in Orthogonal Cutting, Proceedings Temperatures, Proceedings of the Institute of
nd
of the II CIRP Int’l Workshop on Modeling of Mechanical Engineers, pp. 289-329.
Machining Operations, Nantes, France. [78] Boston, O.W., Gilbert, W.W., 1935, Cutting
[63] Wayne, S.F., Zimmerman, C., O’Niel, D.A., 1993, Temperatures Developed by Single-Point Turning
Aspects of advanced cutting tool and chip-flow Tools, Trans. of the ASM, vol. 23, pp. 703-726.
modeling, Proceedings of the 13th International [79] Trigger, K.J., 1948, Progress Report No. 1 on Tool-
Plansee Seminar, Eds. H. Bildstein and R. Eck, Chip Interface Temperatures, Transactions of the
Metallwerk Plansee, Reutte Vol. 2, p.64. ASME, Vol. 70, pp. 91-98.
[64] Childs, T., 1997, Material Property requirements for [80] Stephenson, D.A., 1993, Tool-Work Thermocouple
modeling metal machining, J. Phys. IV France 7 Temperature Measurements – Theory and
Colloque C3, suppl. Au J. Phys. III. Implementation Issues, ASME Journal of Engineering
[65] Camacho, G., Marusich, T., Ortiz, M., 1993, for Industry, Vol. 115, pp. 432-437.
Adaptive meshing methods for the analysis of [81] Taylor, B.N., and Kuyatt, C.E., 1994, Guidelines for
unconstrained plastic flow, Advanced Computational Evaluating and Expressing the Uncertainty of NIST
Methods for Material Modeling, Eds. Benson, D. J. Measurement Results, National Institute of Standards
and Asaro, R. J., ASME AMD-Vol. 180/PVP-Vol.268, and Technology Technical Note (TN) 1297.
pp. 71 - 83.
[82] Nondestructive Testing Handbook, 1991, 2nd Edition,
[66] Lin, Z.C., Chu, K.T., Pan, W.C., 1994, A study of the Volume 7, Ultrasonic Testing, American Society for
stress, strain and temperature distributions of a Nondestructive Testing.
machined workpiece using a thermo-elastic-plastic
coupled model, Journal of Materials Processing
Technology, Vol. 41, pp. 291-310.
[67] Strenkowski, J.S., Moon, K.J., 1990, Finite Element
Prediction of Chip Geometry and Tool/Workpiece
Temperature Distributions in Orthogonal Cutting,
ASME Journal of Engineering for Industry, Vol. 112,
p. 313-318.
[68] Moon, K.J., 1988, A thermo-viscoplastic model of
orthogonal cutting including chip geometry and
temperature prediction, Ph.D. Thesis, North Carolina
State University.
[69] Strenkowski, J.S., Moon, K.J., 1989, A thermo-
viscoplastic model of the cutting process including
chip geometry and temperature prediction, Advances
in Manufacturing Systems Integration and Processes,
15th Conference on Production Research and
Technology.
[70] Hsu, J.M., 1990, Three-dimensional finite element
modeling of metal cutting, Ph.D. Thesis, North
Carolina State University.
[71] Chern, S.Y., 1991, An improved finite element model
of orthogonal cutting with special emphasis on a
separate shear zone model, Ph.D. Thesis, North
Carolina State University.
[72] Hsu, H.C., 1992, An elasto-viscoplastic finite element
model of orthogonal metal cutting for residual stress
prediction, Ph.D. Thesis, North Carolina State
University.
[73] Chien, W.T., 1992, A numerical and experimental
investigation of the sticking and sliding contact
regions at the chip-tool interface in machining
Aluminum 6061-T6, Ph.D. Thesis, North Carolina
State University.
[74] Yang, J.A., 1992, A predictive model of chip breaking
for groove-type tools in orthogonal machining of AISI
1020 steel, Ph.D. Thesis, North Carolina State
University.
[75] Athavale, S.M., 1994, A damage-based model for
predicting chip breakability for obstruction and
groove tools, Ph.D. Thesis, North Carolina State
University.
[76] Liu, W.K., Chang, H., Chen, J.S., Belytschko, 1988,
"Arbitrary Lagrangian-Eulerian Petrov-Galerkin Finite
Elements for Nonlinear Continua," Computer
Methods in Applied Mechanics and Engineering, Vol.
68, pp. 259-310.