International Journal of Production Research
International Journal of Production Research
International Journal of Production Research
International Journal of
Production Research
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/tprs20
An intelligent feature-based
process planning system for
prismatic parts
Lalit Patil & S. S. Pande
To cite this article: Lalit Patil & S. S. Pande (2002): An intelligent feature-based
process planning system for prismatic parts, International Journal of Production
Research, 40:17, 4431-4447
This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make
any representation that the contents will be complete or accurate or
up to date. The accuracy of any instructions, formulae, and drug doses
should be independently verified with primary sources. The publisher
shall not be liable for any loss, actions, claims, proceedings, demand, or
costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
int. j. prod. res., 2002, vol. 40, no. 17, 4431±4447
Today, feature-based design has become a core technology for solid product
modelling primarily because it can capture the designer’s intent and provide the
capability to satisfy the informational needs of the down-line application tasks for
CAD/CAM integration. This paper reports the development of an intelligent
environment, IFPP (Intelligent Feature-based Process Planning), for the
feature-based design synthesis and process planning of prismatic parts to be
produced on CNC machining centres. IFPP consists of two functional modules,
namely Feature Based Modeller (FBM) and Automatic process Planner
Downloaded by [University of Guelph] at 07:32 26 July 2012
1. Introduction
Modern manufacturin g faces several challenges, such as sti global competition,
low volume, large variety production, the requirement of high productivity and
product quality, and shorter lead times from design to manufacturing . During the
last two decades, CAD/CAM technology has been extensively developed to auto-
mate and integrate various stages in the design and manufacturing cycle (Lee 1999).
Despite these e orts, di culties exits in the integration of CAD and CAM software
domains mainly due to their di erent informational needs. CAD focuses on part
speci®c geometry and topology while CAM needs process-speci®c features and their
accuracies. Integration e orts thus attempt to augment or translate information
between the two domains.
In this background, today, feature-based (supported) design has become a core
technology for solid product modelling primarily because it o ers the following
important characteristics.
. Capability to capture the designer’s intent in terms of product shape, toler-
ances, surface ®nish and manufacturing processes, right at the design stage.
. Capability to satisfy the informational needs of the down-line application
tasks.
International Journal of Production Research ISSN 0020±7543 print/ISSN 1366±588X online # 2002 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/00207540210155855
4432 L. Patil and S. S. Pande
2. Literature review
The literature documents that, in the last two decades, several research e orts
have been focused towards feature-supporte d part modelling, automated process
planning and CAD/CAPP/CAM integration.
For feature-supporte d part modelling, two directions were pursued, namely auto-
matic extraction of features from solid models, and feature-based part modelling.
Several di erent approaches, such as rule based (Vandenbrande and Requicha 1993),
syntactic pattern based (Kulkarni and Pande 1995, Prabhu and Pande 1999) and
graph based (Joshi and Chang 1988, Gavankar and Henderson 1990) recognition,
have been reported for the automatic feature extraction from B-Rep/CSG/wireframe
CADD models. Various domain-speci®c feature modelling systems, such as PRI-
CAPP (Pande and Walvekar 1990), QTC (Chang 1990), BLOCK CAD/CAM
(Hoshi and Hanada 1992), OMEGA (Sabomin and Villeneuve 1996), PRISPLAN
(Karadkar and Pande 1996), CFACA (Liu 2000) and VITool (Maropoulos et al.
2000), have been reported with a focus on prismatic machined parts. These feature-
modelling systems attempted to capture part geometry and process data at the
modelling stage and were thus found to be more suitable for integration with auto-
mated process planning systems.
These systems vary in their capabilities in terms or strategies for part geometry
input and process planning stages, such as operation extraction and sequencing,
machine/tool/cutting parameter selection and NC code generation. The various
strategies for input of part geometry information have been a ected by the growth
in the use of feature technology as mentioned in the previous paragraph. Further,
various steps in the process planning domain, such as operation extraction and
sequencing, machine selection, tool selection, cutting parameter selection and NC
code generation have been dealt with using di erent strategies in di erent systems. In
the domain of process planning, most of the research e ort has been devoted to the
automatic extraction and sequencing of operations. PRICAPP (Pande and Walvekar
1990) relies on a heuristics (industry) based policy for the sequencing. BLOCK
CAD/CAM (Hoshi and Hanada 1992) optimizes the machining operations and
their sequence by using certain heuristics based on best machining practices.
OMEGA (Sabomin and Villeneuve 1996) uses production rules to de®ne operation
sequences that are then subsequently grouped into set-ups based on the nature of the
tolerances. Process selection in QTC (Chang 1990) is done by utilizing a knowledge
Intelligent feature-base d process planning system 4433
base that comprises results from research and inputs from industry. The operations
are sequenced in order to optimize the number of set-ups and tool changes. Systems
such as QTC (Chang 1990), OMEGA (Sabomin and Villeneuve 1996), PRISPLAN
(Karadkar and Pande 1996) and CFACA (Liu 2000) focus on CAPP for CNC
machining centres. These systems claim to generate CNC code to fabricate the
object, but do not describe the approach and the quality (optimality) of the code.
These systems have thus achieved a restricted integration between part synthesis and
CAPP for speci®c part geometry arid machine tool con®gurations.
Much commercial CNC programming software is semi-automati c in operation,
requiring several inputs from process planners for identi®cation of features to be
machined, tool path approach/retract geometry, boundaries of features, etc. Such
software often ignores advanced programming utilities provided by the controllers,
such as Parametric Programming, special cycles (macros), feature pattern symmetry
aids (rotation, mirroring), etc. As a result, the CNC codes produced by many CAM
Downloaded by [University of Guelph] at 07:32 26 July 2012
The part model can be conceived as having a gross shape (a 2.5D sweep primi-
tive) on which a variety of local features such as holes, pockets, steps, slots and their
individual shape derivatives can be modelled as interacting local part features. The
object Oriented Programming Strategy (OOPS) is used to represent the geometric
and process related information of the features on the part solid model.
Various design and implementation issues of the feature-based modeller are
brie¯y discussed here.
. Gross features. These resemble a raw stock for CNC machining from which
various feature shapes are carved out. In the present work, 2.5D linear sweep
primitives have been implemented. The exact shape of the gross feature
depends on the two-dimensional (2D) cross-sectional contour designed by
the user.
. Local features. These represent families of features having varying geometries
but the same topological characteristics (connectivity). All local features are
considered to be of depression type as the primary manufacturing operation is
CNC machining.
Based on the manner in which features appear on the faces, the local features are
further classi®ed as follows:
. Faced-based features. Holes, pockets, feature patternsÐArrays of holes.
. Edge-based features. Slots.
. Corner-based features. Steps.
attributes of the feature. It provides polymorphic user interfaces to input data for
these attributes.
The Local object deals with the dimensional attributes of the feature (e.g. length.
width, and the height for the rectangular slot). It also records the depth status (blind/
through) of the feature. Information, such as tolerances and surface ®nish, is stored
as process data by this object. Polymorphic methods are used to validate the feature
based on design and manufacturabilit y issues.
The object oriented design strategy provides the capability to organize and repre-
sent the feature information for easy message processing. Further, it o ers the ¯ex-
ibility to modify the de®nition of an object, its structure, message and linkup without
a ecting the rest of the system con®guration. This is a major advantage of the object
oriented design paradigm.
The data structure of FBM makes it easier to hook up applications onto the
model, thus enhancing the portability and extensibility of the system.
designer models the part in terms of the manufacturabl e feature primitives outlined
in section 4.1. Feature taxonomy has been worked out by considering manufactur-
able feature shapes and not the design features. FBM has been provided with a set of
rules to validate intelligently the part features right at the stage of creation. The
feature validation checks include the input of vital dimensions for feature construc-
tion, tolerance speci®cations, location and orientation characteristics (datums) with
proper values; interference checks between the, current feature being modelled and
the previous features/raw stock; process constraints on size, tolerance, accuracy. etc.
These rules have been collected from the literature (Bralla 1986) and from the study
of actual component drawings.
Typical principles for feature validation are as follows.
(1) Topological checks. These basically include the check’s on the number of
constituent faces, e.g. a blind slot has four faces; a corner step has three
faces; a through hole has one face etc.
(2) Geometry checks. These include checks to maintain the topological and
feature-type constraints, e.g. position and width of slot to ensure that
some material surrounds it.
(3) Process checks. These include limits on hole spacings, tolerance/®nish, sizes
of features and transitional features such as arcs/®llets etc.
The CAD model is created by the user in an incremental fashion. Therefore, the
system keeps checking for validity on creation of each new feature and displays
critical areas/features by highlighting them. Many times, the validity checks detect
and display possible feature interactions (e.g. a hole in a slot). The user is given the
option to override the result and proceed further with the modelling.
Figure 4(a) shows an example component that can be modelled using the FBM.
The features to be machined on this component are step, rectangular pocket, slot and
an array of holes inside the pocket.
Figure 4(b) shows a feature graph for the example component considered in
®gure 4(a). The arc joining G and L1 represents the information about the location
(x and y coordinates) of the local feature L1 on the gross feature G. The existence of
the array of holes (L6 ) on the face of the pocket (L5 ) is represented by an arc from L5
to L6 . Similarly, other relationships are maintained and represented in the FBM. The
geometry of every feature is further stored as a B-Rep structure by the solid mod-
eller. It is linked to the Feature object in the FBM data structure. As shown in
section 4.2 and in ®gure 3, all the information related to a feature is stored in a
data structure compatible with object oriented design.
Intelligent feature-base d process planning system 4439
Thus, the FBM stores feature information along with the B-Rep structure of the
solid model. Process information, such as tolerances, is also associated with the
geometric information. This repository of the feature information is then utilized
for the down-line application of process planning and CNC code generation.
Feature 7! Process
Hole 7 ! Centre drilling, drilling, reaming, boring, tapping, milling
Slot 7 ! Slot milling (pro®ling)
Step 7 ! Step milling (pro®ling)
Pocket 7 ! Centre drilling and pocket milling (pro®ling)
Cross 7 ! Face/pro®le milling
This mapping is governed by several factors, such as the type of feature, its size
and dimension, tolerance and surface ®nish requirements, and accepted manufactur-
Downloaded by [University of Guelph] at 07:32 26 July 2012
ing practices (Halevi and Weil 1995). Typical operation selection strategies for the
machining of holes are included here as an example.
Rules have been written to match the feature characteristics with hole machining
strategies as above. They are further broken down as follows:
Table 1 represents the decision logic applied in selection of the processes and
sequencing to machine a hole. The process capabilities of the various hole-making
processes are stored in the database, which can be edited by the user to account for
shop ¯oor practices. In AutoPlan, rules have been written to encode process knowl-
edge derived from actual shop studies and from the literature. This rule base can be
edited by the user to tune the mapping strategy.
After identifying candidate operations, the operations are sequenced. If any
con¯icts arise during the planning, they are taken care of based on a broad strategy
as follows.
Intelligent feature-base d process planning system 4441
L=D < 4
and D < 50
and 11 < tol < 13 Twist drill
and 5 < Ra < 80
(1) All operations generally follow the sequence constraint shown in ®gure 6.
(2) Reaming is preferred to boring as it provides better surface ®nish and size
tolerances as compared to boring. This also reduces the number of tool
changes, thereby the non-productive time and thus also the machining cost.
(3) In the case of counterbored holes, if a counterboring tool is hot available,
then the feature is considered as two holes on the same axis.
(4) Cycle time is reduced by minimizing the number of operations and tool
changes. The number of set-ups per machine is maximized.
(5) If more than one process appear to qualify as the ®nal ®nishing process, then
these processes are given the same priority and the user has the option to
choose, or some other criteria (cost, availability of tool, availability of
machine, etc) is used to resolve the con¯ict.
Similar strategies are developed and implemented for milling operation for gen-
erating contoured features such as slots, steps, pockets, etc. Parameters that govern
the operation selection of these features include the dimensions of the feature, sur-
face ®nish requirements, corner radius, cutter overlap, etc. Issues such as cutter
selection, single/two-stage process operations, cut size determination, etc, are
planned, based on the recommendations from shop ¯oors and the literature (Patil
1998).
5.3.1. Approach
In standard CNC programming systems, the tool path planning strategy is gov-
erned by the set of three surfaces, termed as drive, part and check surfaces, which are
derived from the part geometry. The cutter location (CL) path data are obtained by
o setting the surface to be machined, as per the type of tool and its contact condi-
tions, with the part. The tool path planning activity in such software primarily
focuses upon the geometry of surfaces, tool type and contact conditions. The geo-
metry and topology of the part feature being machined is not considered in a
bundled manner and the CNC tool path apparently bears no direct relationship
with the feature in a gross sense.
In contrast, the approach followed in AutoPlan uniquely maintains the relation-
ship between the feature and its associated manufacturing strategy (tool path) in a
generic fashion. The object oriented framework in FBM permits encapsulation of
this part±process information. This entails that the hierarchical organization
of features in the part model translates directly into the hierarchical organization
of feature speci®c CNC codes, arranged as per the sequence of manufacturing of
features in each set-up.
This feature-CNC code integration will provide an environment (Variant Design
or Variant Process Planning) by which changes in feature parameters (size, location,
orientation, etc) on the part model will be automatically mapped to changes in CNC
codes without the requirement of generation of CNC code de novo. Thus, a standard
master CNC code could be created in a generic fashion that will cater to a group of
parts belonging to a part family. This strategy will be very useful in the rapid
creation of error-free, standard CNC code. Since the facilities of parametric pro-
gramming and special cycles provided by the controller are incorporated into the
feature manufacturing strategies, the code is deemed to be compact and e cient.
Intelligent feature-base d process planning system 4443
In AutoPlan, this concept mentioned above has been implemented. A sample tool
motion planning strategy for the machining of slots is discussed.
AutoPlan does not produce a `hard coded’ CNC program but generates a parametric
CNC code to enable a link between feature parameters, CL data and the G/M codes.
Parametric programming is widely used by expert process planners during manual
part programming to create generic CNC codes. The parametric programs o er
several bene®ts, such as ¯exibility, standardization of shop-speci®c practices and
improved productivity (Lynch 1995).
No standard CAM software, to our knowledge, is known to provide the facility
of generating feature-speci®c CNC code from a feature-based CAD model. Several
bene®ts will accrue if this methodology is implemented. Important among them are
the following.
. The feature-code binding will permit quick creation of error-free CNC code in
a feature-based ¯exible CAD environment.
. Standardizatio n of code that incorporates shop-speci®c (better) manufacturing
practices and expertise.
. Use of advanced facilities/cycles provided on the controller leads to better
utilization of the machine capabilities. Most CAM software creates code
using only the basic tool motion commands such as G01, G02, etc.
. Creation of compact and e cient code.
. Less expertise is needed on the part of the CNC programmer in using para-
metric programming.
With these objectives, AutoPlan was developed to demonstrate the strategy for
automatic generation of feature-speci®c parametric CNC code. This, we believe, is
the primary contribution of the present research work.
get re¯ected in the CNC code. It thus incorporates the ¯exible feature-spe-
ci®c parametric CNC programming strategy.
(2) Subroutines that build in the modularity in the CNC code are automatically
generated. They can be effectively used for repetitive tasks such as machin-
ing the external gross, or the internal face of a pocket in multiple passes by
increasing the depth of cut in every pass. Multiple calls to the same sub-
routine can be used, if necessary. Facilities such as coordinate axes rotation,
mirroring, ®xture offsets etc are also included and are useful if they are
provided directly on the controller.
(3) The modal nature in the operation of preparatory words (G01,
G02; . . . ; G81), standard ISO canned cycles and cutter radius compensation
(G40, G41, G42) is accounted for to avoid redundancy and verbosity in the
generated code.
Downloaded by [University of Guelph] at 07:32 26 July 2012
6. Conclusions
The IFPP (Intelligent Feature-based Process Planning) system provides an inte-
grated environment for the feature-based design synthesis and manufacture of pris-
matic parts. The FBM (Feature Based Modeller) provides a designer friendly
environment for the synthesis and validation of a feature-based solid part model.
The intelligent process planner (AutoPlan) provides an integration of the feature to
process knowledge in order to demonstrate a feature-speci®c parametric CNC pro-
gramming strategy.
IFPP was extensively tested for several prismatic parts from industries by mod-
elling, planning and actual machining on the CNC machining centre with a FANUC
controller. The integrated environment in IFPP was found to provide error-free,
compact and e cient CNC code to enable a quick turnaround from design to
manufacture in a rapid product development environment.
References
Bralla, J. G. (ed.), 1986, Handbook of Product Design for Manufacturing (New York:
McGraw-Hill).
Chang, T. C. , 1990, Expert Process Planning for Manufacturing (Reading, MA: Addison-
Wesley).
Gvankar, P. and Henderson, M. R. , 1990, Graph-based extraction of protrusions and
depressions from boundary representations. Computer Aided Design, 22(7), 442±450.
Halevi, G. and Weil, R. D., 1995, Principle of Process Planning: a Logical Approach
(London: Chapman & Hall).
Hoshi, T. and Hanada, T. , 1992, Block like CAD/CAM system for fully automated CAM
processing. Annals of the CIRP, 41(1), 551±556.
Joshi, S. and Chang, T. C. , 1988, Graph-based heuristics for recognition of machined fea-
tures from 3D solid model. Computer Aided Design, 20(2), 58±66.
Kale, M. S. and Pande, S. S. , 1998, Intelligent setup generation for automated ®xture plan-
ning. Proceedings of the Eighth International Flexible Automation and Intelligent
Manufacturing (FAIM-98) Conference, Portland, OR, pp. 545±557.
Karadkar, R. B. and Pande, S. S. , 1996, Feature based automatic CNC code generation for
prismatic parts. Computers in Industry, 28(2), 137±150.
Kulkarni, V. S. and Pande, S. S. , 1995, A system for automatic extraction of 3D part
features using syntactic pattern recognition techniques, International Journal of
Production Research, 33(6), 1569±1580.
Lee, K., 1999, Principles of CAD/CAM/CAE Systems (Reading, MA: Addison-Wesley).
Liu, X. , 2000, CFACA: component framework for feature-based design and process planning.
Computer Aided Design, 32(7), 397±408.
Intelligent feature-base d process planning system 4447
Vandenbrande, J. H. and Requicha, A. A. G., 1993, Spatial reasoning for the automatic
recognition of machinable features in solid models. IEEE Pattern Analysis and Machine
Intelligence, l5(12), 1269±1285.