Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.

com, ISSN 1743-3533

Method for establishing machine tool

performance specifications from part tolerance
requirements
R. Callaghan
Independent Quality Labs, Inc., USA

Abstract
Design Engineers are accustomed to using tools to evaluate the effect of
tolerances on their designs. Valve Designers often calculate the tolerance effects
on spring rates and flow paths. Process Engineers, however, have relied on
experience or an "educated guess" and trial and error, to meet the design
tolerance requirements. The capability of an existing or new machine to produce
a given feature to tolerance was essentially unknown. The adoption of two
ASME Standards, B5.54-1992 [ l ] and B5.57-1998 [2], has provided the basis for
new process capability tools. These standards use the "Deterministic Method"
first described by Donaldson [3] and later by Bryan and Loewen [4]. Machine
motions are described by the six degrees of freedom of the linear and rotary axes.
The relationships of the axes are defined by the angles between the linear and
rotary axis average lines. The motions and relationships or parameters are
defmed by errors such as angular (roll, pitch and yaw), spindle (radial, axial and
tilt), straightness, squareness, parallelism and accuracy (linear and angular). This
paper will discuss the method of identifying and describing these errors for a
selection of machine designs. A Machine Error Model will be used to evaluate
the capability of a machine tool to process a sample part. The concept of Part
Tolerance Ratios for establishing the limiting tolerances from a family of parts
will be presented. Machine Tolerance Ratio as a measure of machine capability
will also be discussed. The methods of measuring, recording and controlling
parameters will be addressed.
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

508 Laser Metrology und Muchiize Performance VI

1 Introduction
Machine tool test standards have been evolving for nearly one hundred years. In
1927, Dr. Georg Schlesinger published the first of a series of acceptance test
specifications for machine tools [5]. This book and the subsequent IS0 230
series of acceptance codes for machine tools [6] state that their aim "is to
standardise methods of testing." An accompanying series of machine tool test
conditions, IS0 13041, "establishes the tolerances or maximum acceptable
values for the test results" [7]. These methods and tolerances are utilised by
machine builders and users, to assess the results of their work or condition of
their machines. The tolerances have tightened significantly, to reduce size and
improve efficiency. Their complexity, however, has made it difficult for users to
establish performance specifications based on their own needs.
The B5.54 [l] and B5.57 [2] Standards for CNC machine tools use methods
quite different from Schlesinger. These standards use the "Detenninistic
Method" first described by Donaldson [3] and later by Bryan and Loewen [4].
Machine motions are described by the six degrees of freedom of the linear and
rotary axes. The relationslups of the axes are defmed by the angles between the
linear and rotary axis average lines. The motions and relationslups or parameters
are defmed by errors such as angular (roll, pitch and yaw), spindle (radial, axial
and tilt), straightness, squareness, parallelism and accuracy (linear and angular).
This revised parametric method of describing machines makes it easier for the
Process Engineers to understand the effects of errors.

2 Errors
2.1 Machine errors

2.1.1 Six degrees of freedom - linear motion

Machine elements moving along a linear axis are constrained in space by a
variety of bearing systems. These systems consist of sliding elements and rolling
elements. Sliding systems use anti-friction methods, including solids, fluids and
pressurised fluids. Rolling systems use balls, cylinders and tapered cylinders.
The ways on which these systems ride may be flat, round, or contoured.
Regardless of the design, the systems all have the same motions relative to the
earth. These motions consist of three translation and three rotary motions. These
motions can be described as roll, pitch, yaw, straightness 1, straightness 2, and
linear displacement accuracy. It should be noted that the bearing system
geometry is the primary source of part geometry.
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

Laser Metrology and Machine Performance VI 509

2.1.2 Six degrees of freedom - rotary motion

Rotating machine elements are constrained relative to the earth, in the same
manner as linear systems. The only difference is that the bearing systems are
circular instead of linear. Rotating elements exhibit the same three translation
and rotation motions. Rotating element motions are described as tilt 1 (pitch), tilt
2 (yaw), axial error (straightness l), radial error (straightness 2) and angular
displacement accuracy (roll).

2.1.3 Multi-axis relationships

Machme elements are connected by a variety of frame designs. The relationships
between the bearing systems are defined by the angle between the axis direction
lines of linear axes and the axis average lines of rotary axes [2]. These lines are
through the average centres of the bearing systems.

2.1.4 Dynamic motion

With the advent of CNC (Computer Numerical Control) and DNC (Direct
Numerical Control) the geometry of workpieces can also be affected by the
dynamic capability of the machine control system. Tapered, round and contoured
features can be produced by a number of different methods. These methods
include Linear and Circular Interpolation, managed by the machine control. Tool
paths can also be downloaded directly from a CADJCAM system.

2.2 Thermal and vibration errors

In many material removal processes, temperature can have a significant effect on

the geometry of the finished part. The thermal error can lengthen, shorten or
distort the part. Vibration errors can affect surface finish. These errors may come
from the maclune itself, the environment or the material removal process. Four
tests in the ASME Standards B5.54-1992 and B5.57-1998 provide tools for
measuring and controlling these errors. The TVE (Temperature Variation Error)
Test [l] quantifies the effect of the thermal environment around the machme.
The Relative Vibration Test [l] quantifies the effect of the vibration environment
and the effectiveness of vibration isolation systems. These errors can be reduced
by improving the machine environments.
The Thermal Distortion Caused by Moving Linear Axes [2] and Spindle
Thermal Stability Tests [2] quantify the errors coming from the machine. These
errors can be reduced by process changes.

3 Machine error descriptions

3.1 Machine

The manner in which machine errors combine to affect the part tolerance is
dependent on the machine design. Error Models are custom-designed based on
axis configurations since the offsets used to convert angular errors to
displacements are dependent on the machine construction. Fundamental
differences in machine designs are reflected in the manner in which the axes are
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

5 10 Laser Metrology und Muchine Perfoi.munce VI

arranged or "stacked." There are several schemes in the standards for describing
the stacking of machme axes. Most of them involve the use of numbering
systems, which are difficult to remember. The most convenient method uses a
simple axis, frame (or fixed structure) and spindle nomenclature, first suggested
by Charlton [8].The order in which machine axes are stacked relative to each
other determines the machine description. The part is located frst and a sequence
of axes, frames and spindles is used to describe the machine design. For
example, the 3 Axis Vertical Spindle Machining Centre (Gantry Mill) shown in
Figure 1 is designated by fXYZS because the Part sits on the Frame (0, which is
fixed to the earth, with the X Axis (X) stacked on f, the Y Axis (Y) on X, the Z
Axis (Z) on Y and the Spindle (S) on Z.

3.2 Errors

The direction and sign of machine errors is significant when considering the
effects due to wear or electronic error compensation. None of the performance
standards provides adequate definitions of direction or sign. IS0 841 [g]
standardises the variations in axis designation for motion control. This
standardisation permits the use of the same Part Program on a variety of
machines. The description of machine errors is a similar problem to the axis
control. The resulting solution is to use the same axis and sign conventions for
machine errors as are used for machme motions. This may seem confusing, since
some machine motions are opposite to the error direction. The result however, is
that the errors appear in the part, in the proper direction.

eBy; Y ROLL
eBz; Z PITCH

Figure 1: Vertical spindle gantry mill fXYZS

 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

Laser Metrology and Machine Performance VI 5 11

4 Establishing machine tool performance specifications

4.1 Part feature assessment

Part features and tolerances are well defined by ASME Y14.5M-1994 [10]. The
organisation of tolerance symbols used in this standard is a convenient way to
break down part features. The definitions of size, form, profile, location,
orientation, and run-out are used to relate features with processes.
Part feature tolerance is limited by a machine's capability to produce that
tolerance over a given distance. It is well known that even large machines can
produce better tolerances over shorter distances. T h s capability can be
established by using the Feature Tolerance Ratio (FTR).
The FTR is determined by dividing the feature tolerance bandwidth by the
distance over which it is applied. The FTR can be expressed in inchlinch or
d m m . The FTR can be used to evaluate the limiting tolerances on a single
part or a family of parts. An analysis of FTR is illustrated in Table 1 below. The
part having the smallest FTR for size, form, profile, location, orientation or run-
out is selected for modeling. The Full Volume and Process Models are based on
the processes and the machine axes in motion used to create these features.

Table 1: Feature tolerance ratio

FEATURE TOLERANCE RATIOS
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

5 12 Laser Metrology und Muchiize Performance VI

Table 2: Machine full volume model

FULL VOLUME ERROR MODEL

I Error Descri~tion Error Comments Ang Error Offset Full Error l

Dir (AS) (inlfl) (in) (in)
II eXx lLin Acc 1 X ISystematic Deuation (E) I I 1 0.001600
21 er& [Repeat I x IUnidirectional Repeatability (R+) I 1 0.000200
I ,, I I I I n nnmnn

281 eoAxs ISq SX I X or z loftset Tool Length 1 43 1 0 00025 1 8 1 0 000167

291 eoBys ISq SY I y or z IOffset Tool Length 1 53 1I 0 00031 1 8 1 0 000207
I I I I I I I

4.2 Machine full volume models

Machine errors are modeled here using a Microsoft@ Excel@ spreadsheet. The
nomenclature of the errors uses an e (displacement), er (repeatability), eo
(orthogonality), ei (interpolation), or et (thermal) to describe the type of error. A
capital letter defines the direction of the error, including rotation. A small letter
defines the moving axis or axes. The magnitude of the errors is determined by
the performance measurements of similar sized machmes. An example for the
fXYZS Gantry Mill is shown in Table 2.
The Models described in thls paper assume linearity and use the bandwidth of
bi-directional errors. The use of bi-directional errors considers the effect of
machining in two directions and datum that may be established in directions
opposite to machining. Angular and multi-axis errors are converted to
displacements by using offsets. Errors are weighted for travel and summed to
determine the worst error in the volume.
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

Laser Metrology and Machine Performance VI 5 13

4.3 Process models

The development of a Process Model (Table 3) from the Full Volume Model
involves several steps. The first is to use the FTR to identify the features and
tolerances, wlvch will govern the capability limits of the machine. The second is
to determine which of the machine axes are moved and how far they travel using
the datum reference frames. The errors from the axes not moved in the process
are removed from the Model. T h d , the contributions to part error from the
machine's angular and orthogonal errors are determined using tooling and
machine offsets. The weighted sum of all errors contributing to a feature is
compared to the tolerance bandwidth, resulting in the Part Tolerance Ratio
(PTR). The PTR should be greater than 4 to assure that the machine is
consuming only 25% of the part tolerance. Machine performance specifications
can be adjusted to improve the PTR. If the PTR is between 1 and 4, an in-process
inspection should be made to assure part quality. If the PTR is less than 1, the
machineiprocess should not even be specified for producing the part feature.

Table 3: Process model

FLAT SURFACE MILLED WITH 12 INCH DIAMETER SHELL MILL
PARTAXS MOTONS
X Y

Error Description Error Comments Ana Error OLet Full Error Adj Error Error

Pad Flatness Tolerance =

Part Flatness Est = l
Part Tolerance Ratio =m]
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

5 14 Laser Metrology und Muchiize Performance VI

FIXTURE DATUM

Figure 2: Process datum

4.3.1 Datum reference frame

The use of Coordinate Measuring Machines to inspect parts has improved the
principles of identifying datum features. The effect of datum used in processing
parts is not as well defined and understood. Fixture datums utilise a combination
of planes, pins and balls to fix the part relative to the machine axes. Fixture
datum create part errors, whch are not a function of machine errors. In-process
datum and in-process re-datum (Figure 2) are directly related to the machine's
positioning and measuring capability. In-process datum are simply features that
are created as part of the material removal process. In-process re-datum involve
the machine's ability to measure a feature and re-establish the positioning
coordinate system.

4.3.2 Tooling offset effects

The configuration of tooling has effects beyond the metal removal process. The
length and diameter of tools create offsets on which the machine's angular and
multi-axis errors act. The effect of spindle angular and alignment errors are
shown in Figures 3 and 4. The offsets for these effects may be the distances from
the gage line or centreline to the tool tip, depth of the feature, or the radius of the
tool.

Figure 3: Z axis angular error Figure 4: Spindle alignment error

 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

Laser Metrology and Machine Performance VI 5 15

8 PITCH OFFSE-

SQUARENESS OFFSE

Figure 5: Tooling position Figure 6: Part positioning

4.3.3 Machine offset effects

The angular and multi-axis errors also act on machme offsets to create part
tolerance errors. The offsets are a function of the machme design. Offsets can be
categorised into tool positioning, part positioning and toollpart probing effects.
These effects are illustrated in Figures 5 and 6.

5 Error control
The Process Error Models can be used to determine the parametric errors
limiting machme capability. The Models shown in this paper also determine the
percentage of part tolerance consumed by the machine errors in a given process.
The errors with the highest percentages should be monitored periodically to
maintain control of the process. These errors can be measured and controlled
using the methods described in the ASME standards. Adequate documentation of
the methods should be maintained to assure reproducibility of the parametric
error data.

6 Application experiences
In 1999, a major US manufacturer planned to purchase two large CfWXZS
Vertical Lathes. These machines were multi-axis with interchangeable tools and
heads. The part features included bearing bores and close tolerance alignment
slots. The Performance Specifications were developed in accordance with the
ANSI methods using the Process Models.
The machines were performance tested at the supplier's European facility.
One of the machines was found to be out of specification, requiring design
modifications prior to final acceptance. After completion of the installations in
the US, performance tests were repeated, requiring only minor adjustments to
meet the original specifications. The machines successfully completed the trial
parts withn the first two weeks. Both machines have been operating for over one
year without a single part discrepancy being assigned to the machines.
 Transactions on Engineering Sciences vol 44, © 2003 WIT Press, www.witpress.com, ISSN 1743-3533

5 16 Laser Metrology und Muchine Performance VI

7 Conclusion
The methods of matching part feature tolerances with machine performance
described in t h ~ spaper have been used for over 30 years. Acceptance of this
approach by the users of machine tools has been extraordinarily slow. The
general familiarity with software to process and present data has permitted the
development of simpler modeling tools. These simpler tools have been used to
develop the performance requirements for millions of dollars of machine tool
rebuilds and purchases. The results in all cases were machines that met the
expectations of the users with a minimum of start-up delay.

References
[l] The American Society of Mechanical Engineers, ASME B5.54-1992:
Methods for Performance Evaluation of Computer Numerically Controlled
Machining Centers, ASME: New York, 1993.
[2] The American Society of Mechanical Engineers, ASME B5.57-1998:
Methods for Performance Evaluation of Computer Numerically Controlled
Lathes and Turning Centers ASME: New York, 2000.
[3] Donaldson, R. The Deterministic Approach to Machine Accuracy, Society of
Manufacturing Engineers Symposium, Nov., Golden, 1972.
[4] Bryan, J. & Loewen, E. The Limit of Accuracy of Machine Tools,
International Conference on Production Engineering, Aug. 26-29, Tokyo,
1974. Lawrence Livennore National Laboratory Report # U C U 75645.
[5] Schlesinger, G., Testing Machine Tools, 7" ed. The Machinery Publishing
Co., Ltd: London, 1966.
[6] International Organization for Standardization, Geometric accuracy of
machines operating under no-load or f i s h n g conditions (Part 1).
nternational Standard I S 0 230/1: Acceptance code for machine tools, lSted.
ISO: Geneva, Section l : Scope and field of application, 1986.
[7] International Organization for Standardization, International Standard I S 0
13041: Machine tools - Test conditions for numerically controlled
horizontal turning machines and turning centres with horizontal spindle -
Testing of the accuracy, ISO: Geneva, Introduction, 2000.
[8] Charlton, T. Metrology and Error Budgeting of Precision Machines, Second
International Precision Engineering Conference (Session 2), May 17-20,
Gaithersburg, 1983.
[g] International Organization for Standardization, International Standard I S 0
841: Industrial automation systems and integration - Physical device
control - Coordinate system and motion nomenclature, 2"d ed ISO: Geneva,
1992.
[l01 The American Society of Mechanical Engineers, ASME Y14.5M-1994:
Dimensioning and Tolerancing ASME: New York, 1995.