You are on page 1of 33

Design Rule Checker for Sheet Metal Components using Medial Axis Transforms and Geometric Reasoning

Raj Radhakrishnan, Araya Amsalu, Mehran Kamran and B.O. Nnaji


Automation and Robotics Laboratory Department of Industrial Engineering and Operations Research University of Massachusetts at Amherst

Abstract
Design Rule Checker is an essential component of design automation and for making design, process planning and manufacturing concurrent. In this paper, a set of algorithms are presented for detecting design violations in sheet metal components. Violations in sheet metal components occur when holes, slots, or other features are too close to one another. These violations, if undetected during design, increase the design life cycle of the product. Thus it is very essential to detect, if not eliminate these violations at the early stages of design. Typically sheet metal components have a number of features serving various purposes which include ventilation and fastening. As the contour of the parts gets complicated and the number of features on the sheet metal part increases it is di cult to check the rules exhaustively as it may lead to combinatorial explosion.

The algorithm presented here, uses Medial axis transformation to decompose a given domain thus preventing a combinatorial explosion, and also provides a means to integrate the rule checker with a design adviser.

1 Introduction
Industry is in continual search for ways of reducing the cost of manufacturing of the products in its realm. For years e ort has been concentrated on implementing the latest production methods and increased use of automation. Many analyses of product costs in the development cycle indicate about 70% of the total product cost is already committed by the time the design is completed though less than 10% is spent. This indicates that bringing manufacturing considerations to bear on the design process is the logical venue to reduce cost and product realization time. As designers have become estranged from manufacturing knowledge due to specialization, incorporating manufacturing considerations into the design has become more and more di cult. With most of today's design conducted on CAD systems, using computerized tools that will provide input on manufacturability and other lifetime issues in the design stage helps overcome this di culty. Manufacturing issues can be brought to the design stage through manufacturability evaluation of the design. In evaluating manufacturability the design is checked against the process capability as applied to the material used. The common practice in industry is to express the critical relationships between design requirements and process capability attributes as design rules. Thus checking the design against these rules,i.e., design rule checking, is a means of bringing manufacturing issues to the 2

design stage. Design rule checking, thus, presupposes a set of design rules and an e cient rule checking algorithm. The design rule checker presented in this paper forms the core of the design advisor for ProMod-S; a product modeling framework intended to provide a concurrent engineering environment for sheet metal application. We will describe the ProMod-S evaluation and advisory scheme in the next section.

2 The ProMod-S Evaluation and Advisory Scheme


The ProMod-S design rule checker builds upon previous work done by researchers at the Automation and Robotics Laboratory of the University of Massachusetts at Amherst Nnaji91], Nnaji92], Yeh92]. Figure 1 shows how the design rule checker ts into the overall scheme of sheet metal design evaluator and advisor. The scheme, some parts of which are still in development, will ideally operate as follows: There are two ways in which to enter the ProMod-S environment. The user may enter directly by using the ProMod-S feature based modeler, or enter using an external modeler. The output from an external modeler must rst be passed through a neutral graphics translator to obtain an Initial Graphics Exchange Speci cation (IGES) le. The IGES representation of the part then undergoes geometric validation inside the ProMod-S system. This validation prepares the design for feature recognition and extraction. During the validation process, edges which have only one end point connected 3

Figure 1: Sheet Metal Design Evaluator and Advisor in Product Modeling System

ProMod-S Environment
Neutral Graphics Translator Geometric Validator Feature Extractor
Features

Feature File

Design Rules Database

Feature Based Modeler

Tools Database

Tool Matching Engine

Cost Estimator

Design Rule Checker

Basic Cost Database External Modeler Results Display Cost,Advice,Violations

Design Advisor

User

Violations

Cost

to the greater part (dangling edges) are removed, and collinear and cocurve edges are detected and replaced by equivalent single edges. The removal of collinear and cocurve edges is necessary to ensure an unambiguous representation of the part. Regardless of the method of entry into the ProMod-S environment (through the built-in feature-based modeler or through an external modeler), the part must be passed to the feature extractor for feature recognition and extraction. We de ne a feature as a set of surfaces together with speci cations of the bounding relationships
between them which imply an engineering function. Nnaji91] Features are context

dependent; it is therefore not assumed that features with which a part has been designed using the feature-based modeler are the same features needed for subsequent processes. Save for deep drawing and coining, sheet metal operations do not change the uniform sheet thickness of the stock. The change in sheet thickness caused by coining is negligible, and deep drawing processes are beyond the scope of this paper and the ProMod-S system. A uniform sheet thickness assumption on the part of the system simpli es many reasoning processes immensely. It allows the system to consider only one side of the sheet metal part in question. This reduces many three dimensional problems to two dimensions. Making use of this assumption, a surface-oriented model is created as the rst step in recognition and extraction. The surface-oriented model is used for all reasoning activities that follow. After recognition and extraction the features are placed in a feature le where they can be accessed by the Design Rule Checker, the Cost Estimator, and the ToolMatching Engine. The Design Rule checker accesses rules from the Design Rules 5

Database to identify violations; intrafeature violations caused by parameters of a single feature, and interfeature violations caused by the relative locations of features and features and contour edges. These violations are then passed to the Design Advisor. The Cost Estimator accesses cost information from the Basic Cost database. This basic cost information is based on the type, size, and location of each feature on the part. It is calculated assuming a standard method for producing the feature. For example, the Cost Estimator assumes that each hole is produced with a single punch operation. It also receives cost updates from the Tool Matching Engine. Cost updates are necessary because the assumption that there exists a standard method for producing a feature may be faulty. This can happen, for instance, if the exact tool needed to punch a hole is not available to the manufacturer, yet the hole may still be punched using a combination of tools. The additional information needed to estimate the cost of a part ( in this example the number of tools required to make the hole, total number of punching operations, the total number of tool changes needed, etc. ) is passed to the Cost Estimator from the Tool-Matching Engine. The Cost Estimator in turn passes all cost information to the design advisor. The design advisor passes cost information, violations, and solution alternatives for correcting the violations to the user through the Results Display.

3 Previous Work
Luby et al. Luby] explored the possibility of automatic design evaluation during the creation process to reduce the design cycle time with their experimental system CASPER. Luby discusses issues related to capturing design violations and even cap6

tures some intrafeature violations (which can be avoided by designing features using variational geometry), but does not address the problem of interfeature violations that may occur in many mechanical components. Cutkosky Cutk90], points out the possibility of interfeature violation but, it is not clear how they detect it automatically or otherwise. However, they are able to capture intrafeature violations using geometric constraints. Malmqvist Malmq] focused on considering manufacturing requirements early in the design process using parametric design approach. In this approach the researcher imposes the minimum dimensions of the feature parameters as constraints during the creation process thus eliminating the possibility of intrafeature violation. DeFazio Defaz] stressed the general need for better quality designs of manufactured products and indicated (as many others have) that 70% or more of the life cycle cost of a product is determined during the early design process. As a step towards the solution he lists design rule checking as one of the important properties of a design for manufacturing and assembly systems. Gadh et al. Gadh91] addressed various issues in knowledge-based manufacturability analysis. In their approach Gadh and his team use Di erential Depth Perception Filter (DDPF) which is analogous to human discernation of object outlines in recognizing features, during which they check for violation in feature parameters. Nielsen et al. Niels91] continued the work done by Luby in automatic design evaluation. This work was one step towards responding to some of the criticism by Shah and Wilson. This was an extensive work done in capturing both intrafeature and interfeature violations. In their method Neilsen and his group model the Geometric 7

intent as a collection of restraints and geometric attributes of the design's form. Their restraint is a collection of scalar restrictions, with each restriction associated with a distinct source and importance level. This method lends itself to geometric modi cations, however it relies on the designer to input values that are associated with interfeature violations. As the number of features increase the rules associated with them could increase exponentially, making it di cult for the designer to remember all the rules. Also, as the interfeature violations are detected using position constraints, in a part with complex contour, this would mean over constraining or would even result in a failure to detect the violation. Due to the nature of the approach, even if a design con ict is detected, the system cannot resolve it and has to rely on the designer. Most of the work in the literature deals with intrafeature violations. Interfeature violations, on the other hand, received little attention. In the works reviewed above, it is not clear how interfeature violations are captured automatically or otherwise, and/or the design rules to capture these violations are input by the user as constraints in other cases. It becomes a tedious task for the user to remember all the rules, as the number of features on the part increases. It is more e cient and tractable that the user should have an environment to check for interfeature violations automatically with ease and e ciency in a consistent and accurate manner, after having created the part in any CAD system. It is also important that the methodology used to detect interfeature 8

violations as mentioned above should also provide some means for automatically xing the violations. This essentially calls for a methodology which can understand the geometry of the contour and its interior during the process of violation detection. Medial Axis Transformation satis es the above criterion and has been a very useful tool in this research.

4 Design Rules and Violations


Design rules express the critical relationship between design requirements and process capability attributes. The process capability data is compiled by manufacturing engineers and organized to constitute the design rules. These rules provide the limiting conditions that cause infeasibility, high cost, low quality and longer cycle time. Each industry maintains process capability data in the form of handbooks, trade magazines etc. It is up to the manufacturing engineer to synthesize the rules from the process capability data and the industrial experience. A stamping design rules database is compiled in this research and is presented in this section. People in the sheet metal stamping industry have compiled design rules from process capability data over the decades. Most of the explicit work in the area is not readily available since the results are considered as industrial secrets. One has to scavenge die design literature to obtain some detailed information. In this research, in addition to rules collected from sheet-metal enclosure industries, the design rules were collected mainly from; 1. SME's Die Design Handbook Die90], 9

2. Tool and Manufacturing Engineers Handbook, Volume 2 TMEH2] 3. Eary and Reeds's Techniques of Press Working Sheet Metal Eary74], 4. Federico Strasser's Functional Design of Metal Stamping Stras71] and Practical Design of Sheet Metal Stampings Stras59], and 5. ASM's Metals Handbook, Volume 4 Metal69]. These design rules are expressed in terms of sheet metal features such as holes, slots, bends, ribs, dimples, lances, louvers etc. The rules are mainly explicit statements of the process constraints, dimension and geometric tolerances that can be obtained. The rules provide the critical dimensions that should be maintained to avoid violations i.e low manufacturability. The rules are generally of two types in their application; the intrafeature rules and interfeature rules.

4.1 Intrafeature Rules


Interafeature rules are those rules that involve a single feature only. These rules give critical dimensions to the parameters of the feature. Two most common intrafeature rules involve holes, slots and bends.

4.1.1 Holes and Slots


Holes and slots are very common features in sheet metal stampings. The minimum diameter of the hole should be; D (min) = "T" or 0.04 in, whichever is greater as shown in Figure 2. 10

Figure 2: Minimum dimension for holes and slots This rule expresses a technological constraint on punching process. As the hole diameter goes below the stock thickness, the unit compressive stress on the punch increases dramatically(about four times the shear stress on the sheet) causing the thin and weak punch to buckle or break. The same rule applies to slots with the minor modi cation of substituting the minor dimension instead of the diameter in the rule.

4.1.2 Bend
Bends are used in sheet metal parts mainly for forming the required shape contour and sometimes for sti ening. The minimum bend radius should be: R(min) = 1.5 T 11

If the bend radius is smaller than that given by this rule, cracks will occur on the external surface of the bend due to the high tensile stress.

4.2 Interfeature Rules


Interfeature rules describe the relationship between features or features and the part contour i.e edges of the blank.The features may be of the same type (like two holes) or di erent (like hole and dimple).Some examples of interfeature rules follow.

4.2.1 Hole to Edge


If a hole is too close to an edge, the edge will distort forming a bulge as shown in g. The minimum distance a hole should have from an edge is a function of the material thickness as shown in Figure 3.

4.2.2 Hole to Bend


If a hole is too close to a bend, the hole will distort during bending. This will occur if the hole is punched before bending. The other alternative, punching after bending will be more expensive due to added cost of xtures. The minimum distance a hole has to maintain from a bend is a function of the material thickness and the bend radius as shown in the Figure 4.

12

Figure 3: Minimum hole to hole/ hole to edge distance

13

Figure 4: Minimum distance between a hole and a bend

14

4.3 Generic Rules


More than 65 design rules(both interfeature and intrafeature) are collected . These rules are by no means complete. The rules, expressed in terms of features, require the feature to be explicitly recognised before they can be applied. This restricts the rule checking to common features. If the user de nes a new feature, manufacturability can not be evaluated unless the rules are input for the feature. A close study of the rules was undertaken to nd higher level relationships that will deal with this problem. The rst step in analyzing the rules is identifying the principles behind the rules. It is found that many rules stem from few basic principles and are duplicated to express the situation case by case for features. It is observed that: edges are basic elements from the manufacturing point of view and there is a one to one correspondence between edge types and manufacturing processes. Thus expressing the rules in terms of edges will provide the basis for higher level relationship and avoid the duplication. most of the rules deal with proximity of features to other features or the external contour of the part. Using the previous observation, the rules were reorganized to deal with proximity of edges. only the edges in the outermost loop of a feature need to be considered in interfeature rules. most of the rules apply to features in the same face. 15

Table 1: Generic Rules EDGE 2 Minimum Distance Form 10T+R1+R2 Bend 4T+R1+R2 Ext. edge 4T+R1 Hole 3T+R1 Bend Bend Ext. edge 2T+R1 Hole 2T+R1 Ext. edge Ext edge Hole T < 0:032in; 0:060in :032 < T < :125in; 2T T > :125in; 2:5T Hole Hole T < 0:032in; 0:060in :032 < T < :125in; 2T T > :125in; 2:5T The analysis of the collected rules in light of the above observations resulted in a compact set of generic rules expressed in terms of edges shown in Table 1. EDGE 1 Form

5 The Rule Checking Algorithm


Typically sheet metal components have a number of features serving various purposes which often include ventilation and fastening. As the contour of the part gets complicated and the number of features on the sheet metal part increases, it is di cult to check the rules exhaustively as it may lead to combinatorial explosion. Thus, an e cient algorithm is needed for rule checking. As observed in section 4, most rules apply to features in the same face of a part. Each face consists of the external contour of the face and the features inside it. The face pro le or the face contour is the outermost contour and thus contains all other 16

contours within. All the other contours internal to the face contour belong to the features in the face. A feature may have more than one pro le/contour, but the pro le of interest here, called feature pro le, is the outermost pro le of the feature that lies on the face. Using these facts with the face oriented model, each face of the model forms a multiply-connected polygon. Since most rules deal with proximity, the rule checking problem can thus be modeled as a proximity problem of a multiplyconnected polygon. Medial Axis Transform (MAT) provides a convenient way of decomposing a polygonal face into regions associated with edges. Any point inside each of these regions is closer to the edge that is associated with the region than any other. The boundaries of these regions carry distance information. Thus MAT can provide distance information, on the geometric interior of the faces, that is essential for the design advisor during violation correction. A rule checking algorithm that utilizes MAT to reduce the search space to computationally viable level is developed in our laboratory. The algorithm proceeds one face at a time, and it could be applied every time a feature is inserted on a face or after completing all work on a face. The rule checking algorithm has three steps : Generation of medial axes. Extraction of loops that enclose features. Examination of the minimum distance information against the rules in the manufacturing knowledge base.

17

Figure 5: Medial axis or skeleton of a rectangle

5.1 Generation of Medial Axes


Before discussing the steps of generating MAT, let us rst brie y discuss the concept of MAT.

5.1.1 Medial Axis Transformation


The Medial Axis Transformation is a technique rst proposed by Blum Blum] as a means to describe a gure. Lee Lee82] formally de ned it as follows: given an object represented, say by a simple polygon G, the medial axis M (G) is the set of points

fqg internal to G such that there are at least two points on the object's boundary that are equidistant from fqg and are closest to fqg. More explicitly, every point p
on the plane containing the shape may be associated with a nearest point such that the Euclidian distance from a point p to the boundary set B is the distance from p to a nearest point P on B , 18

Figure 6: External medial axis of a square

d(p; B ) = minfd(p; P ) : P 2 B g
Such a nearest point exists because the shape is a closed subset of the Euclidian space. For a particular set of points the minimum distance is not achieved uniquely. Such points are equidistant from two or more points on the boundary contour. The set of points together with the limit points of this set constitute the medial axis or skeleton of a given shape as shown in Figure 5. This de nes what we call internal MAT for the purpose of this paper. The external medial axis is based on the same concept as the internal medial axis except that the concept is applied to the region outside the polygon as shown in the Figure 6. MAT is closely related to Voronoi diagrams. The MAT and Voronoi diagram of convex polygons are the same. In the case of non-convex polygons, polygons with reentrant vertices, however, the MAT will be a subset of the Voronoi diagram. The Voronoi diagram will have two more edges perpendicular to the edges that intersect at each re-entrant vertex. As is done commonly, the terms Voronoi diagram and MAT are used interchangeably here.

19

5.1.2 Generation of Medial Axis Transforms


In section 5, it was discussed that each face can be modeled as a multiply-connected polygon. Thus, algorithms for generating MAT of multiply-connected polygons could be used to generate the MAT for each face. Generating Voronoi diagrams for multiply-connected polygons involves three main steps; generating the internal MAT of the external boundary, generating the external MAT of the internal boundaries and blending the external and internal MAT's generated henceforth. The algorithm, developed by Srinivasan and Nackman srini87] is used to compute the Voronoi diagram of a multiply-connected polygonal domain, V OD(
H @ j =0

),

where H is the number of polygons. The algorithm does this by computing the individual Voronoi diagrams V OD(@!j ); j = 0; :::; H , and then merging them. The complete algorithm runs in O(n log n + H ) time, where n is total number of edges in the polygonal domain and H is the number of polygons. Consider the multiply connected polygon in Figure 7. The Voronoi diagram, i.e. internal medial axis of the external boundary, is computed using the algorithm developed by Lee Lee82]. A modi cation of Lee's algorithm was used to compute external medial axis of the internal boundary. After the internal and external MATs are computed separately, they are superimposed and their merge curve is computed. The merge curve is the bisector between the boundaries whose Voronoi diagrams are being merged. For sheet parts, the pseudo code to apply the above steps for any given face is : 20

Figure 7: Multiply-Connected Polygon


Sort the face loop and the feature loops Generate the internal MAT for the face loop For featn=1, n is the number of features on the face i

Begin
Generate the external medial axis of the current feature Find the start point to blend the external medial axis of the current feature with previous medial axes generated and blend it

End The most important property of Voronoi diagrams of multiply-connected polygons that forms the basis of the rule checking algorithm is :
merge curves are simple, closed curves.

21

Figure 8: An external MAT superimposed and ready for blending Thus each feature is enclosed by a loop which is the locus of points equidistant between the feature pro le and other adjacent pro les. The distances associated with the vertices of the loop are calculated in the process of generating the merge curve. These distances provide the proximity information of the feature under consideration during rule checking. In Figure 9 the face has four features and the blended medial axes. For this particular face if one were to search for design violations exhaustively it would mean that all the four features have to be checked against each other (the complexity of which is a function of the total number of edges of the features) and all four features have to be checked against each of the four edges of the face. However, this search can be reduced by examining the loop that surrounds each feature. But the loop should be extracted before examining it. The next section discusses loop extraction. 22

Figure 9: Medial axis of a multiply connected polygon

5.2 Loop Extraction


As it can be seen in Figure 8 showing blending results in a merge curve that consists of two types of Voronoi edges: Voronoi edges between the edges of di erent pro les(face pro le or feature pro le) and Voronoi edges between the edges of the same pro le. The Voronoi edges of interest, those that form the loop we want to extract, are the ones that belong to the edges of two di erent pro les. The multiply connected polygon shown in Figure 9 has many loops formed by Voronoi edges. Many of these loops do not enclose any features and are generated partly during the external medial axis generation and partly during blending. The loops that are of interest to us are generated completely during blending and should enclose only one feature. In many cases there may be loops that enclose more than one feature which should be neglected. 23

To extract the loops of interest the data structure of the Voronoi edges is designed to hold the feature or pro le ID's of the elements that generate them. For example in Figure 9, the Voronoi edges starting from the circles would have the same ID for both the left and right feature ID's, as the elements that generate these Voronoi edges belong to the same feature. The Voronoi edges shown in Figure 10 would have di erent ID's as the elements that generate these belong to di erent features or pro les. The following enumerate the loop extraction algorithm steps.

Let vij be the ith Voronoi edge of feature j For vij (i = 1::n) f If vij ! leftid = vij ! rightid Continue Call Function (Extract Voronoi edge(vij )g repeat the above steps for all the features on a face.

Applying the above algorithm to the medial axis of the multiply connected polygon shown in Figure 9 results in the extraction of the loops that surround single features as shown in Figure 10.

6 Scanning for violation


The last part of the rule checking algorithm compares the distance information stored in the vertices of the merge loop with the rule base to detect violations. As mentioned earlier the use of medial axis reduces the search space for violation detection. This is especially useful as the number of features and edges on a sheet metal component are many in number. 24

Figure 10: Loops of interest in the medial axis of a multiply-connected polygon After the loops are extracted from the other Voronoi edges that surround the features, the loops (see Figure 10) are scanned for violation that may exist between the elements that generate them. Two consecutive vertices on the merge curve are chosen at a time till the entire loop is traversed. If d is the distance between the vertex on the merge curve and either of the parent elements and dmin is the minimum distance required between the two features under consideration, two cases may arise: 1. If 2 d < dmin , which is the worst case scenario, then the distances between the vertices of the elements, dv1 and dv2,are calculated (Figure 11). If either

dv1 < dmin or dv2 < dmin then the features have violated the rule or else the two
features are a safe distance apart. 2. If 2 d dmin , then the two elements that generate the merge curve are safe distance apart. 25

Figure 11: Closeness of non-parallel Voronoi elements

Figure 12: Violations captured using the loops If two elements are found to be closer than allowed by the design rule, the edges are highlighted. In Figure 12 the loop that surround the right side obround hole is scanned and this feature is found to be too close to the left side obround hole. The violation is due to the fact the two highlighted edges are too close. Similarly other loops are scanned and violations are highlighted as shown in Figure 12.

7 Violation correction
An extensive discussion of violation correction is given in Kamr94]. The purpose of this paper is to address the violation detection aspects of the of the design advisory system. One of the options not covered in Kamr94] is that of moving the feature to 26

Figure 13: Zooming on Vc another location. This option is discussed here.

7.1 Moving Features


As discussed in section 5 most of the manufacturability rules involve proximity conditions. Thus, moving, i.e, repositioning the features is one way to correct violations. But there are two concerns; 1. before moving a feature the direction and magnitude of the necessary displacement should be known, and 2. there should be a way to ensure the feature will not create new violations at the new location. 27

As discussed in section 5.1, the merge loop of a feature contains it's proximity information. By comparing this proximity information with the minimum distance provided by the manufacturability rule, the direction and distance of moving can be derived. The steps required to achieve this are described as follows; (Figure 13 is used to illustrate the steps.) 1. As mentioned in section 6, the edges of the feature that violated the rule are highlighted. Among the endpoints of these edges, nd the vertex,Vc, which is closest to the edge with which the violation has occurred. This vertex is the worst case of the violation. It has the maximum shortfall from the minimum distance, dmin , speci ed by the rule. Referring to Figure 13, vc is the point with

r = max(dmin ? d) where r is the shortfall, dmin is the minimum distance the edges should be apart to avoid violation as
speci ed by the rule, and d is the distance between Vc and Ev , the edge with which the violation occurred. Notice d should be less than dmin for a violation to occur. 2. Resolve r into its rectangular components, along the local coordinates X and Y,

rx and ry respectively. Referring to Figure 13, it can been seen that repositioning Vc anywhere along(and beyond) line l1 which is parallel and a distance of dmin
away from Ev will correct the violation. If Vc is to be moved along X only, a distance of X is required to reach l1 i.e correct the violation. This distance can be calculated by

r2 X=r x
28

(1)

Figure 14: Zooming on Vf Similarly for Y,

r2 Y=r y
movement satis es,

(2)

If Vc is to be moved in an arbitrary direction, the violation is corrected if the

My

XM + Y Y x

(3)

where My is the Y component of the movement,and where Mx is the X component of the movement. Thus, the rst concern, direction and magnitude of the move that is necessary to correct the violation is established. The steps that follow address the second concern, and check if moving the feature will not create new violations. 29

3. Traverse the merge loop to nd the vertex, Vf , on the feature, with the minimum remaining distance. The remaining distance is the di erence between the distance and dmin . Thus, referring to Figure 14, Vf is the vertex where

r = d ? dmin is minimum. Resolve r into its components rx and ry . If Vf is


moved along X, the maximum distance it can move before a violation occurs ,

Rx, is given by: r2 Rx = r x


Similarly for Y, Ry is given by: (4)

r2 Ry = r y

(5)

4. If Rx, is greater than X , or if Ry is greater than Y , then the feature can be moved safely to correct the violation. Otherwise, moving is possible i there could be found a movement M , with components Mx and My that satisfy;

My

Rx M + R Ry x y

(6)

8 Conclusion
Using the algorithms presented in this paper, the design rule checking module of Promod-S is able to o er manufacturing advice to a human designer at an early stage of the design process. While the design process is going on, the entire part can be checked for the occurrence of common design violations based on manufacturing constraints. Medial Axis approach reduces the search space considerably while checking 30

for design violations in complex sheet metal parts. It also provides a way to correct certain violations by moving the feature. This is possible because Medial Axis Transform provides plausible moving directions for the features which violate the design rules. The design rule checking module works in conjunction with the other Promod-S modules in order to bridge the gap between design and manufacturing in the sheet metal domain. Other modules include a feature-based design creation module, and a feature extraction/recognition module. Modules currently under development include a cost estimator and tool matching engine.

References
Blum] Blum H., \A Transformation for Extracting New Descriptors of Shape", Models for the Perception of Speech and Visual Form, ed: Weinant WathenDunn MIT Press, 1967, pp. 362-381. Cutk90] Cutkosky,M.R., and Tenenbaum, J.M., \Features in Process Based Design of Features," Mechanism and Machine Theory ,Vol 25-3, pp. 365-81, 1990. DEC90] Sheet Metal Designer's Handbook-Customer's Edition, Digital Equipment Corporation, 1990. Defaz] Defazio, T.,L., etal \ A Prototype of Feature-Based Design for Assembly ",ASME Advances in Design Automation 1990, DE-Vol 23-1, pp. 9-16. Die90] Smith, D., Editor Die Design Handbook , Society of Manufacturing Engineers, 1990. Eary74] Eary,D. F.,Techniques of Pressworking Sheet Metal, Second Edition, Prentice Hall Inc., New Jersey, 1974. Gadh91] Gadh, R., Hall, M.A., Gursoz,E.L., Prinz, F.B., Sudhakar, A.,M., \Feature Abstraction in a Knowledge-Based Critique of Designs" Manufacturing Review, Vol 4, No 2, June 1991. IGES] Initial Graphics Exchange Speci cation (IGES) Version 4.0, U.S. Department of Commerce, National Bureau of Standards, June 1988. 31

Kamr93] Kamran, M., \Feature-Based Design Evaluation and Modeling for sheet metal", Ph.D. Dissertation, University of Massachusetts at Amherst, February 1993. Kamr94] Kamran, M., Yeh, S., Terry,J.M.E, and Nnaji, B.O., \ A Design Advisor for Sheet Metal Fabrication ", IIE Transactions, In Press. Lee82] Lee, D T., \Medial Axis Transformation of a Planar Shape",IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. PAMI-4, NO. 4, pp. 363-369, July 1982. Luby] Luby,S.,c., and Dixon,J.,R., and Simmons,M.,K.,\Creating and using a features database ",Comput. Mech. Eng. Vol 5, No 3,1986. Malmq] Malmqvist, J., \ A Design System For Parametric Design of Complex Products ",ASME Advances in Design Automation 1990, DE-Vol 23-1, pp. 1724. Metal69] Lyman, T., Editor, Metals Handbook, 8th Edition, Vol.4, Forming, American Society for Metals, Metals Park, Ohio, 1969. Niels91] Nielsen, E.H., Dixon, J.R., Zinsmeister, G.E. 1989. \Capturing and Using Designer Intent in a Design-with-Features System,"Proc. Design Theory and Methodology Conf., ASME DE-Vol. 31, pp.95-102 Nnaji91] Nnaji, B.O.,Kang, T.S.,Yeh, S.C., and Chen, J.P., \Feature Reasoning for sheet metal components", International Journal of Production Research, Vol 29, No 9, 1991, pp. 1867-1896. Nnaji92] Nnaji, B.O., \Feature-Based Product Modelling for Sheet Metal components", National Science Foundation Research Proposal, October 1992. Srini87] Srinivasan V., Nackman L R., \Voronoi Diagrams for Multiply Connected Polygonal Domains I: Algorithm",IBM Journal of Research and Development, Vol. 31, No. 3, May 1987, pp. 361-372. 1, pp 563-9. Stras59] Strasser,F., Practical Design of Sheet Metal Stampings, Chilton CompanyBook Division, 1959. Stras71] Strasser,F., Functional Design of Metal Stampings, Society of Manufacturing Engineers, 1971. TMEH2] Wick, C. and etal, Editors, Tool and Manufacturing Engineers Handbook, Vol.2, Forming, Society of Manufacturing Engineers, Dearborn, Michigan, 1984. 32

TMEH6] Bakerjian, R., Editor, Tool and Manufacturing Engineers Handbook, Vol.6, Design for Manufacturability, Society of Manufacturing Engineers, Dearborn, Michigan, 1992. Yeh92] Yeh, Shuicheh, \CAD-Based Automatic Object Recognition and Determination of Symmetric Properties of Objects", Ph.D. Dissertation, University of Massachusetts at Amherst, October 1992.

33

You might also like