Sei sulla pagina 1di 8

Journal of Materials Processing Technology 107 (2000) 127134

Integration of tool selection with design


Part 1. Feature creation and selection of
operations and tools
P.G. Maropoulos*, R.P. Baker
School of Engineering, Design and Manufacturing Research Group, University of Durham,
South Road, Durham DH1 3LE, UK

Abstract
Tool selection is one of the most signicant process considerations when deciding how a machined product will be manufactured. The
tool selection activity is rooted in feature geometry but it also has a direct impact on machinability and machine tool performance.
Consequently, the integration of tool selection with design is a key step towards the goal of a seamless integration between computer-aided
design and computer-aided process planning. This integration requires that design be carried out using features. A feature based design
environment was created with associated validation methods for updating and managing feature dimensions, type descriptions and dealing
with feature interactions. The next stage of automated tool selection involves converting the information associated with each feature into
machining operation requirements, which in turn is used to query a database to extract all tools that may be used to complete a given
operation. In this paper, which is the rst of two parts, the feature creation and validation methods are presented together with the
methodologies for operation and tool selection. In Part 2, a method for the estimation of machining times is presented together with results
from testing the methods using real components. # 2000 Elsevier Science B.V. All rights reserved.
Keywords: Computer aided design; Computer aided process planning; Tool selection

1. Introduction
Is now widely accepted that an important objective of
modern manufacturing planning is to carry out process
planning as early on in the design process as possible [1].
In particular, this should be done in a time-phased manner,
allowing individual methods to be initiated as soon as the
data they require becomes available. The aggregate, management and detailed (AMD) process planning architecture
is a framework that facilitates the early, time-phased application of process planning functionality [2]. According to
the AMD classication, ``aggregate'' methods are used to
identify possible processing options during conceptual and
embodiment design. ``Management'' methods are carried
out during detailed design and are used to rationalise the
options initially identied during the aggregate stage. Once
the design has been completed, ``detailed'' process planning
is used to prepare approved data for machine control and
scheduling purposes.
*

Corresponding author. Tel.: 44-191-374-2574;


fax: 44-191-374-2550.
E-mail address: p.g.maropoulos@durham.ac.uk (P.G. Maropoulos).

Among the highly varied and complex tasks of process


planning [3], the tool selection functionality can be considered particularly important [4], since the inuence of
tooling related issues affects most aspects of process planning. Indeed, it is possible to provide a formal categorisation
of tooling issues and their effects on various process planning activities within a framework called the ``ve levels of
tool selection'' [5]. Fig. 1 shows the mapping of the AMD
architecture and the ve levels of tool selection to the main
stages of design and the effect on processing options as time
progresses.
Using features that act as containers for manufacturing
and other technological information [6,7], is a prerequisite
for the vertical integration of tooling methods with design
and process planning as shown in Fig. 1. This paper will
show that features dened during the early stages of design
(that is, features dened with minimal product information)
can contain adequate manufacturing data to enable comprehensive tool selection to take place during the aggregate
stage of process planning. During these early stages of
design, the key tasks are feature creation, management of
feature types and dimensions (otherwise known as feature
validation), identication of operation requirements, and the

0924-0136/00/$ see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 4 - 0 1 3 6 ( 0 0 ) 0 0 6 8 6 - 5

128

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

Fig. 2. Possible tool approach directions for a drilled through-hole.

Fig. 1. Vertical integration of tooling methods within design and process


planning.

selection of tools to carry out the operations. The aim of this


paper is to present a framework for early tool selection using
minimal data, providing the background necessary to carry
out machining time estimation which is discussed in Part 2.
The methods described in both parts of this paper have been
implemented within a CAD software module called VITool,
which is currently being developed at Durham University.
2. A feature based design environment
This section describes how the seven feature types presently implemented within VITool are dened and validated.
Firstly, the common properties required by all features are
discussed followed by the introduction of each of the feature
types together with the parameters required for their creation. The objectives and methods of the validation of each of
these features will be presented and nally, rules for extracting alternative versions of a feature (the same feature
machined from a different approach direction) will be
discussed.
2.1. Common properties of form features
The common properties of form features provide the data
necessary to dene the operations required to manufacture
them. These are feature type, tool approach direction, position, dimensions, surface nish and tooling options. The
feature type property provides an indication as to what kind
of operations and tools will be required to manufacture the
feature. The seven types implemented within VITool are
drilled hole, closed pocket, slot, shoulder, full slot, full
shoulder and face.
The next property common to all features is the tool
approach direction which serves two purposes: rstly to help
describe the orientation of the feature and secondly to
describe the direction of depth cutting. This can be one
of six values (Z, Z, X, X, Y and Y) indicating the

main face of a block starting-feature into which the tool will


be cutting. As can be seen from Fig. 2, the example through
hole can be considered as two features: one with an approach
direction from Z and an alternative with an approach
direction from Z. The position and dimensions of a given
feature initially are nominal values specied by the user.
However, these are subject to change as a result of automatic
validation methods. Two dimensions of particular interest
are cutting depth and tool depth. Cutting depth is simply the
depth of material removed, whereas tool depth represents the
required engagement of the tool needed to avoid a collision
of the milling machine with the work piece. The surface
nish property consists of two Ra values: one for the surface
nish of the base of the feature (Rab) and one for the surface
nish of the feature walls (Raw). Features that extend beyond
the component, such as through holes, do not have Rab
values, and facing features do not have Raw values. The
tooling options attached to each feature are created when the
user requests tool selection to take place. These contain a
sequence of all the operations required to make the feature
and a list of all possible tools that might be available on the
shop oor to carry out each operation.
2.2. Form features of VITool
Within VITool there are seven different feature types,
which can be grouped into two distinct feature classes. The
``drilled hole'' feature belongs to a class of its own, whereas
the other six feature types belong to what shall be referred to
as the ``pocket'' class, since these are simply special cases of
a closed pocket. Although a hole can also be considered as a
pocket variant, it has been decided to categorise drilled holes
within a separate class, because of their strict validation
requirements (discussed in the next sub-section). Form
features are created from a list of parameters. For drilled
holes these include approach direction, centre point, depth
and diameter. The nominal position of the centre point is
simply an indication of where the central axis of the hole
should pass, since the actual start point of the hole is
determined during the validation process.
The pocket class features are instantiated with an
approach direction, a corner point, a length direction, a
width direction, a length, a width, a depth and a corner

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

129

Fig. 3. Constructor overloading of pocket class features.

radius. The corner point, length, width and depth are all
subject to change as a result of automatic validation. During
validation, the feature type (closed pocket, slot, shoulder,
full slot, full shoulder of face) will also be determined,
although as shown in Fig. 3, limiting the parameters supplied
forces VITool to assess rstly the type of pocket feature
required and secondly to determine the values of the missing
parameters. In object oriented terms, this is referred to as
``constructor overloading''. Fig. 3a shows that the only
parameters required to create a facing feature are the tool
approach direction and the depth of cut. In Fig. 3b, a slot is
created from a tool approach direction, a length direction, a

length, a width, a depth, a corner radius and a centreline


point, which is any point through which the central lengthaxis of the slot passes. A full slot requires all the slot
parameters, except the length and the corner radius, as
shown in Fig. 3c. Shoulder features do not require a position
to be specied since this can be inferred from the length,
width and tool approach directions. As can be seen in Fig. 3d,
the only other parameters that need to be specied are the
length, the width, the depth and the corner radius. Finally, if
the shoulder is to extend right across the work piece as in
Fig. 3e, then the length direction, the length and the corner
radius parameters can be omitted.

130

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

2.3. Validation of features


Feature validation is the process of managing the feature
class and dimensions and detecting changes in these when a
feature is added, edited or interacts with other features [7,8].
A methodology has been developed for carrying out this
process, which is implemented in a Brep geometry modeller
requiring a great deal of information to be known about each
face of the work piece [7,8]. However, if some of the initial
design tasks take place in a non-feature-based environment,
in which information about each face is not stored, then
these methods will become impractical to implement. To
address this issue, the feature validation requirements of
VITool are twofold. Firstly, a new data structure is required
to identify the differences between original features and the
resulting component [9]. Secondly, methods are needed to
manage and correct features added by VITool for a component not built completely from VITool features. The rst
requirement is satised through the implementation of the
feature properties described in the previous two sub-sections. The second requirement is satised using the validation methods described in the next two sub-sections. These
methods work with abstract shapes which are created
and modied using propriety CAD object libraries [10]
that deal with generic ``shape'' properties. Hence, if a design
is read into VITool from an external source and converted
into a ``shape'' (which can be done using the above-mentioned object libraries), then the following methods will
work.
2.3.1. Validation of drilled hole features
Three stages are required for the validation of drilled hole
features: (i) existence check, (ii) XY validation and (iii)
position and depth validation. The rst stage requires a
Boolean test to be carried out to determine if the feature
alters the component. This is particularly important, as
adding one feature may well erase another; this erased
feature needs to be excluded from the analysis. The second
stage of validation is carried out in the local XY plane which
is the plane perpendicular to the approach direction. The
checks carried out during this stage test whether or not the
feature overlaps any of the outer edges of the initial component and also checks whether or not there is a partial
overlap with any other drilled hole features. These checks
are carried out, because of the risk that a rotating drill bit that
is not fully engaged with the work piece will break. If the
feature fails either of these tests, then it is disallowed.
However, a drilled hole is allowed to overlap the edges of
a pocket class feature, provided the edge is not an initial
component edge. This is because, these features could be
sequenced in such a way that the pocket would be machined
rst, thus preventing the increased risk of drill bit breakage.
The nal validation process for drilled holes corrects both
the nominal position and the nominal depth value of the
feature to obtain precise tool engagement and cutting depth
values for tool selection. This is done using Boolean ``feel-

ing'' processes with a thin disc having a diameter equal to


that of the hole. The disc is fed along the feature until it
intersects with the component. If feeling for the top of the
hole, the disc is incremented down from just outside the
initial stock material. When the component is found, the
start position of the hole is adjusted to reect the start point
of the drilling process. When feeling for the bottom of the
feature, the disc is incremented up the hole from the position
of the end point originally dened. When the component is
found, the hole dimensions are adjusted to give the correct
cutting depth and tool depth. Cutting depth is the depth of
material removed plus the depth of the drill nose and tool
depth is the cutting depth plus the safe height.
2.3.2. Validation of pocket class features
The validation of pocket class features consists of the
same three steps as that for drilled hole features. The rst
check is to test whether or not the feature actually exists. The
second validation process is XY validation, which is considerably different to the XY validation of drilled holes. This
second stage of validation is purely corrective so that if the
pocket has passed the rst check (that is, if it exists), then it
cannot fail XY validation. Rather, the dimensions of the
pocket feature are adjusted to the actual dimensions required
for tool selection.
An additional task of pocket XY validation is to identify
which of the six pocket types the feature is. An example of
position, length and width adjustment of a shoulder is shown
in Fig. 4. As can be seen, thin rectangular probes having a
depth equal to the nominal depth of the pocket are used to
carry out incremental Boolean tests to nd the component in
much the same way that depth validation takes place for
drilled hole features. First of all, feeler A with a width equal
to the nominal width of the feature (the width initially
specied by the designer) is incremented along the length
direction until the component is touched. This gives the ``felt
width'', which is the width of feeler B used in the width
direction to determine the ``felt length''. Felt width and felt

Fig. 4. Feeler movement for XY validation of a shoulder.

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

131

2.4. Extraction of alternative features

Fig. 5. Dimensions of validated shoulder.

length are then adjusted to give the true length and width
of the validated pocket class feature as shown in Fig. 6.
Note that in the remainder of this paper, unless otherwise
specied, length and width refer to ``felt length'' and ``felt
width''. Further probing identies the feature type, but in
this case, the probe shape is ``pin'' having a radius equal to
the corner radius of the pocket class feature and a depth
equal to the nominal, pre-validated depth of the feature.
These pins are inserted into each corner of the feature
and tested for an intersection with the component. The
conguration of the number of lengthways and widthways
moves using feelers such as those in Fig. 5, along with
the number of pins intersecting the component is enough
to identify the pocket class feature type. Fig. 6 presents
various congurations for pocket type recognition. Once the
feature has been identied, depth validation takes place.
This is done in precisely the same way as depth validation
for drilled holes, with the only difference being that the
probe shape, instead of being a disc, resembles the pocket
shape.

An approach direction initially specied by the user may


not be the only one, which can be used by a tool to create the
feature. An example is a through hole, which has two
approach directions that are opposite to each other. As well
as providing a greater number of set-up options, the identication of alternative features also has tool selection implications since the dimensions of the tool required may be
different depending on which approach direction is used.
This allows a greater number of tools to be selected, thus
increasing the number of processing options available.
Hence, each time VITool detects that a feature extends right
through the component, an alternative version of that feature
is created from the opposite approach direction and validated to provide parameters for tool selection. Similarly, an
alternative version of a full shoulder has an approach
direction perpendicular to the original feature, whereas a
face has four alternative features: each one is a full shoulder
with an approach direction perpendicular to that of the
original face.
3. Operation and tool selection
After the creation and validation of features and the
extraction of alternative features, it becomes possible to
identify the operations necessary to create the component
and use these to select cutting tools. In VITool, a tool list is a
group of tools that can complete a specic operation. The
information about a particular operation and the operation's
tool list are together known as an option list. A feature's

Fig. 6. Identication of pocket types during XY validation.

132

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

tooling options are the option lists required to manufacture


it. This section describes the rules used to create option lists
and explains how tool lists are populated.
3.1. Operation selection
The rst task is to dene generic tool types and the
properties of the operations they will be required to complete. The individual tool types presently considered within
VITool are u-drills, twist drills, drilling end mills, square
shoulder tools, end mills, solid end mills and facing tools.
The three types of milling operation that may be used to
create the VITool features are drilling, roughing and proling. Drilling refers to a cutting process along the axis of the
approach direction. Tools that may be assigned to a drilling
operation include u-drills, twist drills and drilling end mills.
The only features requiring drilling operations are drilled
holes and closed pockets that need a hole to enable roughing
tool access. A roughing operation is a cutting process that
removes the bulk of material contained within the feature.
The tools that may be used to carry out roughing operations
are facing tools, square shoulder tools, end mills, solid end
mills and drilling end mills. The nal operation type is
proling which is used to complete the feature. The tools
that may be used for proling are the same as those used for
roughing, with the exception of facing tools, which are
assumed to have non-perpendicular inserts and are hence
unable to provide a suitable surface nish on the wall of the
feature. Pocket class features are the only ones requiring
roughing and proling operations.
Having established the tool types that may be used to
carry out each of the three operation types, it is now possible
to assign operations to features. The dimensions of the tools
that are used to carry out the operation are extracted from the
feature properties. For all operations, there are only three
tool dimensions that are of interest at this aggregate stage of
tool selection: cutting depth, working engagement and tool
diameter. The cutting depth and working engagement of
tools for any operation are the cutting depth and the tool
depth of the feature being machined. The tool diameter, on
the other hand, varies depending on the operation taking
place, but in the case of drilled holes, this is simply the
diameter of the feature. Facing features are completed using
a single roughing operation. Although as already discussed,
a roughing operation refers to a mass material removal
process, if a ne surface nish is required on the base of
the feature, then the additional ``skimming'' depth of cut can
also be considered as part of the roughing operation, because
the same tools may be used. The only difference is that to
achieve a ne surface nish, different cutting conditions are
required. The diameter of a tool used to carry out a facing
operation is 1.35 times the width of the feature [11] (it is
assumed that cutting takes place along the length of the
feature and that the width is smaller than or equal to the
length).

Cutting a full shoulder is very similar to creating a face,


except that an additional proling operation is required. The
diameter of the tools for both the roughing and proling
operations is, like facing operations, 1.35 times the width of
the feature. The main difference between full shoulder
roughing and proling, as described earlier, is that roughing
operations may use facing tools. The operations required to
manufacture a full slot are precisely the same as those
required to create a full shoulder, the only difference being
that the diameter of the tool is that of the slot minus a margin
of 1 mm to leave material for the proling operation. Unlike
full slots and full shoulders, slots and shoulders which are
closed at one end need the diameter of their proling tools to
match the corner radius of the feature. However, to avoid a
situation where the tool is ``sucked'' into the component
through inadvertent climb-milling, the radius of the tool
should be about a millimetre less than that of the corner. The
roughing operations for slots use tools with a diameter no
greater than the width of the feature, whereas for shoulders,
the tool diameter can be no greater than 1.35 times the width
of the feature.
Closed pockets are created in a similar manner to slots and
shoulders, with the additional requirement for a drilling
operation to allow the roughing tool access into the component. This, however, imposes a constraint in that the diameter of the roughing tool is dependent on the diameter of
the drilling tool used. Since at this stage, the nal choice of
machines and tools is not yet determined (the objective is
merely to identify as many options as possible), it becomes
necessary to make assumptions about what ought to be
available on the shop oor. As a result of visits to the
collaborating companies on the VITool project, and from
the range of drills supplied by the tooling manufacturers
[1113], it can be assumed that drills up to 50 mm are
readily available. Thus, if a third of the width of the closed
pocket is 50 mm or less, then an access hole with a diameter
of a third of the pocket width should be drilled (minimum
diameter is set to 20 mm). A third of the width is used as the
benchmark value to ensure that enough material remains for
the subsequent roughing operation. If the pocket width is
larger than 150 mm, then an access hole of 50 mm should be
drilled. The roughing and proling operations that follow are
precisely the same as for slots and shoulders, except that the
diameter of tools for the roughing operation are constrained
by the diameter of the drilled access hole. In the above rules,
a number of common parameters were used to describe
operations. These were: (i) operation description (drilling,
roughing or proling), (ii) a list of possible tool types, (iii)
tool diameter (feature and operation dependent), (iv) working engagement (feature tool depth) and (v) cutting depth
(feature cutting depth). These parameters along with surface
nish and a tool list make up an option list for a particular
operation. The surface nish of an option list may be
different to that of the feature being machined. For example,
drilling operations always have a default surface nish.
Roughing operations require only a base surface nish to

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134

be specied, whereas proling operations require both the


base and wall surface nishes. Sometimes, a proling
operation may be used between the roughing operation
and the nishing proling operation, particularly if the
corner radius of the feature is very small and the initial
roughing operation uses a tool with a comparatively large
diameter [8]. For these special proling operations, the wall
surface nish will be set to a default value, thus postponing
the skimming cut until the nal proling operation. The
other important option list parameter is the tool list, which is
obtained by carrying out tool selection.
3.2. Tool selection
The aggregate selection of tools is carried out according to
the type and geometry requirements of the operations
selected to create a particular feature. The objective of tool
selection is to populate the tool list properties of each option
list. A database storing tools available on the shop oor need
at this stage only have the following elds: (i) the machine
the tool is allocated to, (ii) generic tool type, (iii) diameter
(or diameter range in the case of u-drills which have variable
diameters), (iv) maximum working engagement depth, (v)
maximum cutting depth, and (vi) number of teeth. A query
can now be put together from the information in an option
list to extract records from the tools database based on the
following criteria:
1. The tool type of the selected tool should match one of
the option list's possible tool types.
2. For a drilling operation, the tool diameter should exactly
match that of the option list, otherwise any tool with a
smaller diameter can be selected. (In VITool, after tool
selection, the resulting tool list is rationalised so that
tools with a diameter less than 80% of the largest
diameter selected are deleted.)
3. The maximum working engagement and cutting depth
of the selected tool should be greater than or equal to the
corresponding values required by the option list.
Tool selection for a whole component requires that operation requirements rst be determined for all features and all
alternative features too. These requirements can then be used
to extract tools from a database. Hence tools can be selected
to populate the tool list attached to each option list thus
forming the tooling options for a particular feature. The tools
within each tool list can now be assessed and ranked using
multi-criteria assessment, such as the capability analysis
methods of Baker and Maropoulos [14]. One of the most
important criteria is machining times. Part 2 of this paper
presents methods for the aggregate estimation of these
within the context of the work presented here.
4. Conclusions
Early design and planning integration is crucial for lowering product development time and cost. The early, time-

133

phased integration of process planning with design is facilitated by the specication of the AMD architecture. Since
tooling issues are an important thread linking design to
manufacturing, a key task of early manufacturing decision
making within AMD is the selection of cutting tools. This is
achieved by carrying out design by using a library of
predened features. Managing feature dimensions and type
denitions as the design progresses is called feature validation. Feature validation can be accomplished in a versatile
manner by carrying out topological Boolean feeling operations. The main advantage of feature based design is that
each feature can store information about the operations
required to manufacture it. This information is stored within
the feature's tooling options consisting of option lists that
describe the operations needed to complete it. An option list
consists of a number of properties that dene the tooling
requirements for the operation it describes and a tool list that
match those requirements. After tooling options are created,
the analysis proceeds to the management stage. The objective of this stage is to use tool resource structures [5] to
rationalise tool lists and provide rened tooling options
containing few enough tools for practical optimisation during the detailed stage. One of the main criteria of the
``capability analysis'' carried out during the management
stage is the time taken by each tool to complete the operation
allocated to it. The research described herein has been
implemented within a CAD module called VITool. Within
the context of the research described in this paper, scope for
further work exists in increasing the number of feature types
currently implemented, extending the creation and validation methods for non-discrete directions (presently VITool is
restricted to Z, Z, X, X, Y, Y directions) and
adding rules and tool types for creating option lists to
include more operations. The aggregate estimation of
machining times is presented in Part 2 of this paper.
Acknowledgements
The authors wish to acknowledge the support of the
Engineering and Physical Sciences Research Council
(EPSRC), UK, Grant GR/L07017. Special thanks go to
Matra Datavision, UK, and our industrial collaborators
for their support to this project.

References
[1] B. Goldense, D. Vermette, Driving forces for product development
speed, Rapid News 5 (6) (1997) 1217.
[2] P.G. Maropoulos, A novel process planning architecture for productbased manufacture, Proc. Inst. Mech. Engrs., Part B 209 (1995)
267276.
[3] L. Alting, H.C. Zhang, Computer-aided process planning the
state-of-the-art survey, Int. J. Prod. Res. 31 (9) (1989) 21392160.
[4] C.L. Chen, S.R. LeClair, Integration of design and manufacturing:
solving set-up generation and feature sequencing using an unsuper-

134

[5]
[6]
[7]
[8]

P.G. Maropoulos, R.P. Baker / Journal of Materials Processing Technology 107 (2000) 127134
vised-learning approach, Computer-Aided Design 26 (2) (1994) 59
75.
P.G. Maropoulos, Cutting tool selection: an intelligent methodology
and its interfaces with technical and planning functions, Proc. Inst.
Mech. Engrs., Part B 206 (1) (1992) 4960.
J.X. Gao, X.X. Huang, Product and manufacturing capability
modelling in an integrated CAD/process planning environment, Int.
J. Adv. Manuf. Technol. 11 (1996) 4351.
N.N. Gindy, Y. Yue, C.F. Zhu, Automated feature validation for
creating/editing feature-based component data models, Int. J. Prod.
Res. 36 (9) (1998) 24792495.
C.F. Zhu, CAD/CAPP/CAM integration using feature-based
component data model, Thesis, University of Nottingham, UK,
1996.

[9] G.C. Vosniakos, Feature based product engineering: a critique, Int. J.


Adv. Manuf. Technol. 14 (7) (1998) 474480.
[10] Cascade CAE Development Environment, Matra Datavision, University of Warwick Science Park, Sir William Lyons Road, Coventry
CV4 7EZ, UK, 1998.
[11] Sandvik, Rotating Tools Catalogue, Sandvik Cormant UK, Manor
West, Halesowen, West Midlands B62 8QZ, 1998.
[12] Seco, Milling Catalogue and Technical Guide, Seco Tools UK,
Alcester, Warwickshire B49 6EL, 1998.
[13] Stellram, Stellram Milling Program, Stellram, Hercules Way,
Bowerhill Industrial Estate, Wiltshire SN12 6TS, 1998.
[14] R.P. Baker, P.G. Maropoulos, Manufacturing capability assessment
for cellular manufacturing systems, Int. J. Prod. Res. 36 (9) (1998)
25112527.

Potrebbero piacerti anche