Beruflich Dokumente
Kultur Dokumente
Vol. 16/No. 1
1997
Abstract
The range of shapes fxturable with these standard devices is limited. In fact, fixturing considerations place serious constraints on the creativity of the designer.
The determination of a fixturing plan for an
arbitrarily shaped component is a craft that does
not lend itself to efficient automation. Today, the
development of fixture planning systems
remains a major stumbling block in general
automated process planning.
The actual physical construction of the fxture
assembly is a human-intensive, and consequently expensive, process.
In recent years, the problems associated with conventional technology have invited research into
novel, universal workholding techniques. Phase
change based fxturing is a universal workholding
technique that is commonly used for fxturing arbitrarily shaped components. As the name suggests, the
workpiece is immersed in a bath of a molten fixturing alloy. The alloy is then permitted to freeze, thereby trapping the workpiece irrespective of its shape.
Typically, this technique is used in conjunction with
locating dies in a pallet, as shown in Figure 1.
Fixturing has three important functions: immobilization, support, and part location. Unfortunately,
typical phase change based fixturing does not aid in
the location/orientation of the workpiece. Dies are
necessary to determine or ensure the correct orientation of the workpiece within the bath of molten filler.
These dies are component-specific. Different dies
are necessary for different setups, unless the component is symmetric. No robust or economically viable
Introduction
To manufacture a mechanical component on a
milling machine, it is necessary to immobilize, support, and locate it in each setup. This is referred to
as workholding or, alternatively, fixturing. This
paper presents the preliminary concepts of a new
universal workholding technique called Reference
Free Part Encapsulation (RFPE).
The current practice in workholding includes the
use of a number of workholding elements, such as
vises, parallels, clamps, toe-clamps, V-blocks, modular plates, and so on. For each operation, an appropriate workholding configuration must be designed
35
Cutting Tool
Lowmeltipoi
ngnt
~=ece
~omponent
material
Mold
Locating
die
Filler
material
Figure 1
Phase Change Based Fixturing
Basic Concept
Consider the imaginary scenario in which a stock
of metal could be suspended midair using, say, magnetic forces. All faces of the stock would then be
exposed for machining. Any component could be
machined entirely in a single "setup" by an appropriate machine that can access all the sides of the
component. Unfortunately, such a magnetic device
does not exist. As a result, physical forces of contact
must be relied on to immobilize objects.
36
Rller
Stock
Initial block
Machine
features
A'n"
II
Face-mill
Refill with
filler
top
Change
Machine
setup
features
Perfect cube
restored
Restoration
Machining
=: i
Refill with
filler
After ~
melting
Face-mill
top
Ii
Perfectcube
restored
I1~
Figure 2
Steps in R F P E
Practical Issues in
Implementing RFPE
Since the original conception of the underlying
principle of RFPE in 1994, an effort has been made
to develop it to a stage where it can be practiced on
the shop floor. This section presents a preliminary
and qualitative description of some practical issues
in this emerging technology. The development of
quantitative models is in its incipience and is a topic
of ongoing research.
The following sections describe the making of the
initial RFPE blank, the problems related to the material properties of the filler material, and some variant forms of RFPE that mitigate the disadvantages
related to the properties of the filler material.
37
Initial Blank
The initial blank is cast in an open cubical mold.
The mold needs to have high finish and accuracy
and must also facilitate easy retrieval of the blank.
The accuracy of the box directly impacts the accuracy of the machined component in simple RFPE.
"Supported RFPE," described later, makes the accuracy of the RFPE technique independent of the box
tolerances.
Figure 3a shows the creation of the initial blank.
If the machining is to start from an initial rough
stock, where all critical features will be machined
using RFPE, then the exact location of the stock in
the filler cube is not critical. As long as the volume
occupied by the finished component is within the
volume occupied by the rough stock, it does not
matter where the rough stock is exactly. This is
shown in Figure 3b. Once machining is commenced
using RFPE, all following features will be properly
aligned with respect to each other, and because there
are no pre-existing features to reference off, the
component will be accurate.
When the filler block is removed from the mold,
all faces except the top face are nominally accurate
(problems of shrinkage are discussed later). The top
face needs to be machined to make a perfect cube. If
the side faces are not reliably cast, it may be necessary to square the entire filler block before machining is begtm. If the initial workpiece already has
some features on it before RFPE is used, then it is
(a)
Filler
~ll'~
(b)
/Mold
ock
Problems Related to
Material Properties of Filler
The process step that distinguishes RFPE from
classical encapsulation techniques is restoration,
which involves repouring and refreezing filler material. Because the accuracy of the casting is very
important, the behavior of the filler material during
the restoration step is critical to the accuracy of
RFPE. Some potential problems related to the
behavior of the filler material, especially during
restoration, are discussed below.
Eventual
~===~omponent
Rough I ! i l
Ii
Stock ~
II
~,#,l,,,,,,,,,,=,,J I
envelope:
~ S t a n d
Makingthe initial mold.The
top face is machined after
solidification.
Strength~Stiffness
The most obvious problem related to the use of
filler materials such as Cerro alloys and Rigidax
polymers is their lower strength and stiffness when
compared to the tool steel used in conventional fixtures. As shown in Figure 4, the strength of Rigidax
WI Green is more than two orders of magnitude less
38
10
9
8
7
==
5
4
3
2
1
0
0.02
0.04
0.06
Strain
0.08
0.1
0.12
It should be noted that, although to a lesser extent than other polymers, Rigidax~ is in
fact viscoelastic.The data are therefore strain-rate dependent.The representativedata
above were collected at a strain rate of 0.05 ips. Ongoing work is modeling the dynamic
behavior of Rigidax~ with the goal of using it to damp machining vibration.
Figure 4
Stress-Strain Response of RigidaxWI Green
DOC)
ONOC)
Rigk
encap:
Probe
End m ill
Experimental
procedure
Dynamometer
'- -- J
End mill
Probe
.4l
Embedded i Dynamometer i
length . . . . . . . . . . . . . . . . .
Results
Conclusions
For relatively severe cuts (low embedded length, significant cutting forces), Rigidaxe may contribute
upward of 25 microns (0.001") of machining error. However,these errors can be reduced by using
conservative cutting parameters. Metallic fillers are expected to reduce errors further.
Figure 5
Loss of Machining Accuracy with Rigidax WI Green
39
as high as 25 microns in severely undersupported sections. Drift was found to be insignificant with
RigidaxWI Green (less than 10 microns).
The accurate prediction of drift is a difficult problem, and research is currently under way to develop
a reasonable model. In general, however, drift can be
controlled by one or more of the following practices:
Shrinkage
Shrinkage is a primary cause of loss in accuracy
in RFPE. Thermoplastics like Rigidax tend to
shrink more (5%) than fusible alloys like CerroTru
(< 0.05%). Shrinkage, and the resulting warpage of
the initial filler block, may make the reference faces
of the block unreliable.
Thermoplastics and non-eutectic alloys tend to
have higher warpage as a result of the gradual phase
change as the temperature changes. Eutectic alloys
like CerroTru and CerroBend, which have sharp
melting points, are likely to warp less. Furthermore,
because fusible alloys tend to have negligible shrinkage, they are preferable when shrinkage and warpage
are serious problems. In preliminary experiments,
the problems of warpage have been avoided by using
supported RFPE, described in later sections.
Supported RFPE
An important concept in RFPE is the transference
of coordinates between setups through the maintenance of the cubical shape of the filler block. The
accuracy of the dimensions of the cube is critical to
the accuracy of the technique. If the filler material is
not sufficiently hard, it is possible to compromise the
finish of the block during clamping. Furthermore, the
natural problems of shrinkage and warpage during
pouring and freezing tend to make the face finish of
the block unreliable for some materials, like
Rigidax. To combat these problems, a family of
RFPE techniques called supported RFPE have been
[ure
nperature)
Component is
machined in a
given setup
poured to refill
the block
Figure 6
Figure 7
Drift
40
previously
frozen filler
melts at the
interfaces
Component
drifts into the
melt zone
under gravity
Figure 8
Solid-Cage R F P E
Cage RFPE
In cage RFPE, an external cage, or skeleton, is
embedded on the outside of the filler block, as
shown in Figure 8. The mold is used merely to contain the molten filler material and not to provide
dimensional accuracy. As a result, the precision of
the mold does not impact the accuracy of the technique. Furthermore, a certain amount of shrinkage
actually helps because the filler material does not
tend to "bulge" out, thereby not interfering with the
reference surfaces.
An obvious drawback of cage-supported RFPE is
the reduced access to the stock, especially for angled
features, due to the support frame. A solution to this
problem is to use disposable, off-the-shelf stands to
assemble the solid cage. If the strut obstructs access
to a certain feature, it can be machined away. This
introduces the need, during the process planning
stage, to ensure that such operations are carried out
later so that the absence of the struts affects few features. An alternative is to use a partial cage, that is, a
single comer of the original cubical framework. This
reference comer floats in the filler block in a comer
that is not accessed for any feature. This comer can be
used as a reliable reference surface during clamping.
Figure 9
Stock-Enclosed R F P E
2-1/2D RFPE
Stock-Enclosed RFPE
* This idea, along with the gripper design, is due to Mark MacKenzie. 12
41
Vol.16/No.1
1997
~Machined lower
1), features
Retrieve component
~'~ ~
Machinboundary
e
Fixturing Constraints
Figure 10
2-1/2DRFPE
With RFPE, the shape of the component is immaterial. There are no requirements on the order in
which operations must be performed because fixturability can be ensured in any orientation.
Furthermore, the need to locate the component in
every new setup is completely eliminated, thus obviating the need for reference surfaces and datums.
Advantages of RFPE
RFPE was developed to address certain immediate needs in the field of machining. This section
describes the advantages of RFPE. Specifically discussed is how, by expanding the range of parts that
can be fixtured, RFPE dramatically simplifies
design and process planning. Finally, it is shown
how RFPE is amenable to automation.
Process Planning
Process planning may be defined as the determination of a complete set of machining instructions to
manufacture a given component. Lower-level planning tasks include tool selection, path generation,
and cutting parameter selection. Important planning
tasks at the higher level include operation sequencing, fixture design, and setup selection.
The legal sequences in which the machining operations can be performed are constrained by so-called
feature interactions--ordering constraints that
ensure that the requirements of each machining
operation are met. Most interactions can be represented as pairwise ordering constraints and can
therefore be handled by straightforward graph-theoretic means, la These are referred to as local interactions. However, there are two classes of interactions-namely those arising from fixturing concerns
and those arising from component rigidity concerns--that do not lend themselves to such a simple
Rller material
supportsslender
sections
Figure 11
Eliminating Rigidity Concerns Through RFPE
42
Journal ofManufactunngSystems
Vol. 16/No. 1
1997
6ripper
This gripper can be machined in one setup with a two-dimensional version of RFPE
Figure 12
Delicate Gripper with Built-In Flexures
43
Vol. 16/No. 1
1997
Datum 1
Yl
I
Top
Datum2
Datum3
Front
(
Front~
~'~R~ght
Robot elbow
Figure 13
Datumsin MechanicalDesign
a means for automating the CNC machine tool completely to the extent where general three-dimensional parts can be prototyped rapidly "at the touch of a
button?' This section describes a speculative scenario for integrating the operation of RFPE into a
milling system.
The first step in automating RFPE would be the
development of a mechanism that can simultaneously act as a mold and a vise. Such mechanisms are
referred to as mold-vises. A mold-vise concept is
shown in Figure 14. Essentially, a mold-vise makes
it possible to perform the restoration within the vise
holding the blank; by completely closing around the
blank, the mold-vise acts as a mold and prevents
molten filler material from leaking.
Fillerblock
Embedded stock
Bottom
plate
(a)Top view: open
A Mold-Vise Mechanism
44
Machining Tests
Components 1-3 in Figure 16 are robot grippers
that were machined for a separate project at the
/
robot to manipulate
block between setups
(now shown)
Figure 16
45
Part
#
Component
Name
Simple
A1gripper
ABS long
gripper
ABS swivel
gripper
VR headgear
(7 pieces)
Bracket
Escher
part
Trefoil
knot
Computer
casings
(2 pieces)
2
3
4
5
6
7
8
Number
of
Setups
Fixmring
Time
(offline)
Machining
Time
Total
Manufacturing
Time
35a
10
45
50
55
35
15
50
14b
300
200
500
6"
3
150
75
100
75
250
150
150
80
230
100
110
140
46
Acknowledgments
Conclusions
References
47