A P M APPROACH FOR REFINER PLATE DESIGN AND PRODUCTION
Transcript of A P M APPROACH FOR REFINER PLATE DESIGN AND PRODUCTION
A PARAMETRIC MODELING APPROACH FOR REFINER PLATE DESIGN AND PRODUCTION
By:
Boonserm KulvatunyouGraduate Research Assistant
Department of Industrial & Manufacturing EngineeringThe Pennsylvania State University
University Park, PA 16802
Timothy W. SimpsonAssistant Professor
Departments of Mechanical & Nuclear Engineering andIndustrial & Manufacturing Engineering
The Pennsylvania State UniversityUniversity Park, PA 16802
Corresponding AuthorPhone/fax: (814) 863-7136/4745
Email: [email protected]
Erik HalbergGraduate Research Assistant
Departments of Mechanical & Nuclear EngineeringThe Pennsylvania State University
University Park, PA 16802
Barry HodgeManager Engineering
Durametal CorporationMuncy, PA 17756
To appear:
104th Casting Congress and AFS TransactionsApril 8-11, 2000Pittsburgh, PA
Submission Deadline:
November 1, 1999
A PARAMETRIC MODELING APPROACH FOR REFINER PLATE DESIGN AND PRODUCTION
Boonserm Kulvatunyou, Timothy W. Simpson, Erik Halberg, and Barry Hodge
ABSTRACT
Parametric and feature-based design is a modern approach for product modeling to save time and
money during the product development and manufacturing process, especially in the preliminary
stages of design. Parametric modeling can be used to improve customer satisfaction by
facilitating rapid product customization to achieve specific customer requirements. In this paper,
we discuss a parametric and feature-based modeling approach that we are developing to facilitate
design verification, reduce lead-time, and improve product customization when designing and
producing a variety of refiner plate patterns for sand casting. Our primary focus in this paper is
on formalizing our parametric modeling approach and comparing its implementation in two
CAD modeling systems. A comparison of the cost and product lead-time using the proposed
solid modeling approach and the traditional 2-D drawing approach to generate a CNC machined
pattern for sand castings is also presented.
Keywords: Parametric modeling, feature based design, CAD, CAM, refiner plate
Word Count: 5,686
1
1 FRAME OF REFERENCE: REFINER PLATE DESIGN
Product customization is a value-added activity that can significantly increase sales by
increasing customer satisfaction (Anderson and Pine, 1997; Pine, 1993). However, the time and
costs of individual product customization limit all but the most agile manufacturers from
providing this service to their customers. There are several reasons associated with this
hindrance including the lack of supporting technology, design standards, product modularity and
trained support personnel. The investment to eliminate this insufficiency is an additional burden
that often prevents companies from providing customized goods and services.
In the refiner plate design and casting industry, the traditional process of customizing a
refiner plate pattern for sand casting is shown in Figure 1a. First, salespeople meet with
customers to identify their particular pulp requirements, and then designers develop a new 2-D
CAD model. Finally, craftsmen or production engineers develop prototypes for customer
verification. If the product does not satisfy the customer, it is then redesigned and the process is
repeated, causing significant delays in product lead-time. When the customer is finally satisfied
with the design, pilot production is used to verify the product performance. If this verification
does not meet expectations, the entire process is repeated until performance verification is
achieved and full-scale production starts. In this traditional product development and
customization process, a large amount of time is spent translating customer requirements into
CAD drawings and prototype development for customer design verification. To further
complicate matters, many companies still use two-dimensional CAD software which can cause
delays when the product is complex and subject to miscommunication between the design
engineer and the prototype engineer due to ambiguous design specifications. Furthermore,
several prototypes are usually required before all of the customer requirements are met and
2
customers are completely satisfied. The lack of associativity between the two-dimensional
model and the manufacturing process can also cause further delays. Every time there is a design
change, the numerical control (NC) code for prototyping must be manually developed or edited.
Hand-crafted prototype orManually generated NC code
DesignVerification
Pilot Production
PerformanceVerification
Production
2-D CAD Drawing
CustomerRequirements
Rapid prototype orNC machine from CAM
Computer-basedVerification
Pilot Production
PerformanceVerification
Production
Generate solidparametric model
CustomerRequirements
Cur
rent
Foc
us
(a) Traditional (b) ProposedApproach Approach
Figure 1 Product Design Customization and Production Process
We propose a new development and customization process to reduce design lead-time
and the number of physical prototypes developed prior to full-scale production. As shown in
Figure 1b, the customer requirements are turned directly into a solid three-dimensional CAD
model by changing design parameters in a parametric, feature-based design. By doing so, the
amount of time required to develop the product model is significantly reduced. Design
verification and computer analyses (e.g., mechanical and thermal analysis) can also be performed
readily once the solid model has been created. By employing advanced solid modeling software,
a significant portion of customer requirement verification can be done electronically on the
3
computer before any real prototypes are produced, resulting in the reduction in the number of
prototypes. After the customer is satisfied with the three-dimensional solid model on the
computer, a solid prototype can be developed to verify requirements that cannot be easily
analyzed on the computer. In this new development and customization process, parametric CAD
and CAM expedites model development, prototype development, and design verification by
associating them with only one model: the product model. Moreover, in the parametric
associative software tools, a change in the product model will automatically propagate changes
in other engineering models such as the casting die and patterns and NC programs for making the
tooling and machining the castings. Additional advantages of parametric modeling and feature-
based design for casting are discussed in (Lichtenberg, 1998) while Shah and Mantyla (1995)
discuss the advantages of parametric modeling and feature-based design in general.
In this paper, we focus on a novel approach to exploit feature-based and parametric CAD
modeling to facilitate design verification, reduce lead-time, and improve product customization
when designing and producing a variety of refiner plate patterns for sand casting. The details of
the proposed approach are presented in the next section. The approach is then implemented in
Section 3 to compare and contrast the capability to two commercially available CAD parametric
modeling systems: Pro/Engineer and AutoCAD Mechanical Desktop. Finally, Section 4 contains
a brief summary of the work along with some closing remarks.
2 PARAMETRIC MODELING AND FEATURE-BASED DESIGN APPROACH
Parametric and feature-based modeling was introduced in the mid- to late-80s as a means
to incorporate design and manufacturing intent into a product model. In parametric and feature-
based modeling, every part consists of design features and manufacturing features. Design
features include protrusions, cuts, blends, freeform surfaces, etc. while manufacturing features
4
include surface drafts, surface offsets, (thru) holes, etc. Features are characterized by parameters
and geometrical and topological relations among them. These parameters include geometric
parameters (e.g., length, width, height, and location), manufacturing parameters (e.g., surface
finish, machine stock allowance, and draft angle), and analytical parameters (e.g., temperature,
pressure, force, and flow). Changing a parameter value in the parametric model will
automatically update portions of or the entire part geometry based on the relations related to that
parameter. Thus, the relation is a mean to incorporate design intent into the product model.
Furthermore, just as design and manufacturing features can be combined to create a parametric
model, new complex features can be constructed from the primitive design and manufacturing
features to create reusable features for product modeling.
There are two steps involved in developing a parametric, feature-based CAD model.
Step 1 – Identifying features in a design and developing new types of reusable features for the
feature library.
Step 2 – Assembling the features into a parametric model.
The implementation of these steps is formalized and described in the following sections. To
facilitate the discussion of these two steps, examples for refiner plate design are used
intermittently to help explain concepts and terminology.
2.1 Identifying Features and Creating Reusable Features
There are many issues involved in creating reusable features; however, all of them are
directed towards saving time and money. This is especially true when the reusable feature issues
are important in the early design stage, i.e., when the function(s) of the product has not yet been
finalized. In our work, we start considering reusable features when a physical product or a
version of the product model already exists. The process of creating reusable features parallels
the process of feature-based design which begins by identifying features on the part. Instead of
5
looking for primitive features (such as protrusions, holes, and sweeps), we are looking for a set
of function-oriented features which can be used to describe the design. In other words, we are
not trying to define the part using primitive features but function-oriented features that have been
formalized into complex features which can be reused. After function-oriented features are
derived, they are converted into design features, which are grouped into modules that serve
specific design functions. Selected modules are then formalized and archived as reusable
features with respect to time, cost, and recurrence justifications.
Feature identification and modular design techniques have been adapted for the process
of creating these reusable features. Modular design techniques are of interest here because the
results of the modular design are self-contained functional components, which are corresponding
to the function-oriented features. Ulrich (1995), and Ulrich and Eppinger (1995) investigate
product architecture and modularity and its impact on product change, variety, component
standardization, performance, and development management. Pahl and Beitz (1996) decompose
the overall function of a product into sub-functions (less complex functions) and search for
solution principles for the individual sub-functions based on concept variants and the cost of
individual modules versus overall function. The solution principles indicate whether the sub-
functions will be standard, reconfigurable, or design specific. Feng, et al. (1996) suggest that
studying the relationship and dependency between functions will indicate how sub-functions can
be combined into one functional module.
Shah and Mantyla (1995) state that feature identification should first be considered from
two points of view: (1) the product (or finished part drawing) and (2) the design process. In the
product point of view, geometric entities are identified and grouped into features. In the process
point of view, design intent is inferred from the process modeling framework and transformed
6
into design constraints and parameter relations. The information from both points of view is then
consolidated and reconciled to form a parametric representation of the product. Their
methodology is well-defined; however, gathering a designer’s intent during the design process is
error prone due to misinterpretation if this portion of the process has not been formalized.
Furthermore, the parametric model is subject to the limitations of the CAD modeling system’s
capability to store design states (i.e., how the design evolves). In other work, Rosen (1998)
describes on-going research at capturing and representing a designer’s intent for the purposes of
rapid tooling and prototyping.
To overcome some of these limitations, we have developed a four step approach that
integrates feature identification and modular design techniques to create reusable features based
on an existing design (i.e., functions are fixed), see Figure 2. These four steps are (1) function
decomposition, (2) feature identification, (3) formalization, and (4) archiving; the steps and their
associated sub-steps are described in detail in the following sections.
Existing Design
Step 1 Identify Features and Create Reusable Features
Step 2 Assemble Features into Parametric Model
Rapid DesignCustomization
1.1 Decompose Overall Function into Sub-Functions
1.2 Identify FeaturesA. Assign function-oriented features to sub-functionsB. Group function-oriented features into design modulesC. Identify reusable modules (reusable features)
1.3 Formalize Reusable FeaturesA. Identify parameters that define each reusable moduleB. Classify the parametersC.Develop relations and constraints
1.4 Archive Reusable Features
Figure 2 Parametric Modeling Approach using Reusable Features
7
2.1.1 Step 1.1 – Decompose Overall Function into Sub-Functions
Function decomposition is the process of sub-dividing or breaking down the overall
function into sub-functions and forming a function hierarchy. This process is continued until the
sub-functions are self-contained and cannot be further broken down. Note that by sub-function,
we mean a more specific function. The function hierarchy is represented by a hierarchical graph
in which each node is a function. The graph expands outward from a single node representing
the overall function to individual, specific sub-functions. The top three levels in Figure 3
illustrate one function hierarchy for a refiner plate design whereby the overall function is to
transform wood chips into fiber during pulp processing.
Function-orientedfeatures
(Step 1.2.A)
Designmodules
(Step 1.2.B)
Functiondecomposition
(Step 1.1)
Confine & HoldWood Chips
Groove
Receive Energy From Rotor
Apply Energy toMake Fiber
Feed Wood ChipThrough Plate
Receive WoodChip From Input
Bar Groove HoleC’ Bore
Bar Feedin Bar
C’ Bore
Plate C’ Bored Hole
CutWood Chips
TransportWood Chips
FeedinGroove Plate
Convert Wood Chips to Pulp
Hold Plate tothe Rotor
Figure 3 Function Hierarchy, Feature Assignment, and Design Modules
As illustrated in Figure 3, the overall function is decomposed into three specific sub-
functions: (1) cut wood chips, (2) transport chips, and (3) receive energy from the rotor. It is
worth noting that unlike the methodology in (Shah and Mantyla, 1995) which infers design intent
from design process, the function decomposition is not subjected to misinterpretation because it
is done regardless of how the part is actually modeled in the CAD system.
8
2.1.2 Step 1.2 – Identify Features
After the function hierarchy is developed, the next step is to identify features. This step
consists of three sub-steps to facilitate the identification of function-oriented features.
Step 1.2.A – Assign function-oriented features to sub-functions
In this step, function-oriented features are extracted from the existing part or design and
are assigned to the lowest level sub-functions obtained in Step 1.1. If there are sub-functions that
do not correspond to any function-oriented features, either the function hierarchy or the function-
oriented features must be reconsidered, i.e., the function hierarchy may not be properly
decomposed or the function-oriented features may not be listed in their entirety. A function-
oriented feature is a group of geometric entities that performs operating functions when the part
is being used. A function-oriented feature may consist of a group of primitive features or may be
a primitive feature itself, depending on what primitive feature are available in the CAD modeling
system. Returning to our refiner plate design example, the bar on a refiner plate is one of the
function-oriented features illustrated in the fourth level of Figure 3. When two bars are rotating
in the opposite direction, they peel and cut the wood chips into fibers. The bar is a function-
oriented feature that consists of three types of primitive features: (1) protrusions, (2) drafts, and
(3) rounds. The counter-bored hole is another example of a function-oriented feature. Its
function is to hold the plate to the rotor. If the CAD modeling system has a counter-bored hole
as a primitive feature, then the counter-bored hole is a function-oriented feature consisting of
single primitive feature. Otherwise, the counter-bored hole is a function-oriented feature
consisting of a thru-hole and a blind-hole.
9
Step 1.2.B – Group function-oriented features into design modules
A function-oriented feature may or may not be a design feature that is used to model the
part. Some function-oriented features may be features that are derived from other features. An
example is a groove feature on a refiner plate which is formed from two adjacent bars.
Therefore, the function-oriented features are grouped to form design modules that serve different
design functions. We refer to these design features as modules because they serve specific
design functions that are assembled into a final design. The association of function-oriented
features into design modules has some degree of dependency on the CAD system being used
because one should keep in mind that the module should be easy to implement and use in the
CAD modeling system. For instance, the bar and the groove can be associated into either a bar
or a groove design module as shown in the lowest level of Figure 3. The bar module is a positive
design module while the groove module is a negative design module. The bar module may be
selected because a positive module is typically easier to define in a CAD modeling system, and
the bar module is more function-oriented in the design.
Step 1.2.C – Identify reusable modules (reusable features)
Once the function-oriented features are grouped into design modules as shown in the
lowest level in Figure 3, decisions must be made as to which modules are worth being
formalizing into reusable modules. This decision is made based on the economic cost of
formalizing the design module and its utility (i.e., recurrence). These two criteria are highly
coupled and therefore difficult to consider separately. A sophisticated design module, which has
a large number of parameters, can be configured into several slightly different features in a wide
range of design. The recurrence of this sophisticated module may be high, but if it requires a
longer time to formalize, then a simpler module defined by fewer parameters might be more
10
useful. In addition, different function decompositions and function-oriented feature associations
may result in higher module utilization. Other than the time and cost justification, the
sophisticated design module is required when the archive becomes too large.
The utilization of reusable modules also relies on its recognizability by the designer.
Consider the bar feature that is selected to be a reusable module because there are many bars on
each refiner plate. The bar also exists in other plate sizes and other application specific plates.
However, a bar is a complex feature and has many variants on a single plate and even more
variants across different plate designs. For simplicity of discussion, we will consider only four
variations, the ends of the bar, the width, the height, and the length. It is relatively easy to define
a bar that always has both ends rounded (Figure 4a) which can be varied in its width, height, and
position. However, a bar may also have one end rounded and the other end chamfered (Figure
4b). Formalizing both types of bars into a single reusable module such that its ends can be
rounded or flattened requires complicated geometric relations and is a sophisticated task.
Although the utilization of the bar module will increase, it is easiest to resolve this problem by
having several types of reusable bar features form which to chose provided the resulting library
of options is a reasonable number.
(a) Bars with both end rounded (b) Bars with chamfered and rounded ends
Figure 4 Example of Reusable Bar Features
11
2.1.3 Step 1.3 – Formalize Reusable Features
In this step, reusable design modules are formalized within the CAD modeling system.
Formalization involves three steps which include: (1) identifying the parameters that define the
module, (2) classifying the parameters, and (3) developing appropriate relations and constraints.
These steps are described in detail as follows.
Step 1.3.A – Identify parameters that define the module
The first step is to identify the parameters that define the geometry and topology of each
design module. There are usually many ways to define a module; however, it is best to select
parameters that directly map to the functionality of the design module. In this manner, the
design module can be solely reconfigured with these parameters for different functional
requirements. For example, the bar module is similar to a cube so either a combination of width,
length, and height or a combination of width, length, and diagonal height can define it.
Furthermore, either an (X, Y) pair or a polar coordinate can position the module. However,
designers can increase the bar width when they want to increase the strength, change the bar
height and vary its angular orientation to change the wood chip flow rate. Thus, the parameters
of the bar should be formalized to the width, height, length and polar coordinate position.
Step 1.3.B – Classify the parameters
Three distinctions are used when classifying the parameters used to define each module:
(1) standard, (2) independent, and (3) dependent parameters. Standard parameters maintain a
constant value throughout the module utilization and parametric modeling process. Meanwhile,
a designer assigns different values to the independent parameters in different designs to
customize the parametric model to satisfy different customer requirements. Finally, the
12
dependent parameters change with respect to the independent parameters and dimensional
constraints (alignments). In general, when a parameter does not correspond to any module or
function, it should be classified as either a standard or dependent parameter. A complete list of
parameters for the bar (see Figure 5) include both geometric and topological parameters:
• Geometric parameters: the bottom width, top width, height, draft angle, length,
chamfer, and radii of the rounds
• Topologic parameters: the distance from a reference point, angularity placement from
a reference plane, and datum references.
Figure 5 Parameterization of a Reusable Bar Feature
Using the representation shown in Figure 5, the draft angle is the only standard parameter
in this bar feature. The independent parameters are the bottom width, height, chamfer, distance
13
from reference point, angularity placement, and datum references. The dependent parameters are
the top width and the radii of the rounds.
Step 1.3.C – Develop relations and constraints
The final step required to formalize the reusable feature is to develop relations among
independent, dependent, and standard parameters for the reusable modules. Constraints must
also be developed in order to prevent modules from becoming geometrically and functionally
invalid or infeasible. In addition, auxiliary constraints (e.g., for manufacturability or castability)
should be included as needed to reduce the amount of redesign required before production.
Relations and constraints are usually represented in algebraic form (some CAD modeling
systems might support predicate logic as described in (Anderl and Mendgen, 1995)). Relations
are listed sequentially so that all of the dependent parameter values can be determined. The
following list shows examples of relations and constraints that represent the reusable bar feature,
using the independent parameters identified in the previous section. Note that the bottom width
and height are designer’s input.
Relations:top_width = bottom_width – [2 * height * tan (draft angle)]
Constraints:Functionality constraints:
height/bottom_width ≤ strength_limitangularity_placement ≤ discharge_limit
Geometric constraints and Manufacturability constraints:bottom_width ≥ milling_process_limitheight ≤ tool_length_limit
Figure 6 Relations and Constraints for a Reusable Bar Feature
Note that the sophisticated relations and constraints such as conditional relations may be
included for a versatile reusable module depending on the CAD modeling system’s capability
and justification.
14
2.1.4 Archive Features
In this last step, the reusable modules are implemented in the CAD modeling system and
stored in a feature library. The modules should be created corresponding to the parameter
classifications identified in Step 1.3.A. In particular, the standard parameters should be default
while the independent parameters must be input by the designer. Then the relations are codified
to specify the dependent parameters. Finally, the constraints are formalized to validate the whole
module. The module should be named by designer vocabulary so that the designer can easily
recognize its utility. When the number of modules in the library grows large, some outdated and
nearly duplicated modules should be disposed from the archive. However, if the library is large,
a feature taxonomy should be created to increase the accessibility (Shah and Mantyla, 1995).
2.3 Assemble Features into a Parametric Model
After a library of reusable features has been created, all features including all functional
features identified in previous step and manufacturing features (casting feature and machining
feature, e.g., draft and stock material) are assembled to make a parametric model. But before
doing this, the standard parameter, the independent parameters—which will drive design
customization, dependent parameters, and their relations must be planned up front. This will
allow the designer to input the part parametric relations while sketching profiles or adding
features; thus, the solid model can represent design intent. Generally, the techniques used to
identifying parameters, classifying the parameters, and developing parametric relations follow
Step 1.3 (formalization), with an addition of manufacturing features and other inter-feature
parameters (e.g., number of bar instances) consideration. It is worth noting that the independent
parameters should represent part functional requirements, which are affected by customer
15
requests. To demonstrate the implementation of the proposed approach, the design of a low-
consistency refiner plate for pulp processing is described in the next section.
3 EXAMPLE PROBLEM: LOW-CONSISTENCY REFINER PLATE DESIGN
In this example, our goal is to develop a solid model for a low-consistency, twin-flow
refiner plate that is scalable from 20” up to 58” in diameter. This parametric model will affect
casting pattern lead-time, cost of making the pattern, and the shortage of the wood pattern
craftsman. This will improve customer satisfaction through rapid response to changing customer
requirements.
3.1 Step 1.1 – Functional Decomposition
The inputs to the refiner plates are the raw material of pulp and energy generated by
motor, and the output is the fiber product based on the customer’s request. The overall function
of a refiner plate is to transform the raw material to final fiber product by imparting the rotational
energy to shred the pulp fibers. The overall function is sub-divided into five sub-functions as
shown previously in Figure 3 and described as follows.
1. Confine & Hold Wood Chips: Confine the pulp between the plate for a certain amount of
time and hold it against another plate process.
2. Apply Energy to Make Fiber: Transfer the energy coming from motor to peel of the fiber in
the pulp by the media of shear force generated between the two plates and the raw material.
3. Feed Pulp through Plate: The refiner plate also serves as a “pump” that draws in the pulp and
forces it out of the refiner.
4. Receive Wood Chip from Input: Open large area to receive the input from feeding area
5. Hold Plate to the Rotor: Assemble the plate to the whole machine
16
3.2 Step 1.2 – Feature Identification
From the existing 2D drawing, corresponding function-oriented features are assigned to
each sub-function. These function-oriented features are Bar, Groove, Feedin groove, Plate, Hole,
and Counter Bore as shown in Figure 3. These features are grouped into four design modules
(design features) that are Bar, Feedin Bar, Plate, and Counter-bored hole.
The next step is to determine reusable features based on complexity and potential reuse.
Since the bar slightly varies within a plate design and from plate to plate, it is worthwhile to
assign them as a reusable feature to save modeling time. Similarly, the counter-bored hole
module is also selected to be a reusable module. The plate design module is left to be a design
specific module because it is a simple sweep protrusion, and it is used only one time in each
plate design. The feed-in bar module varies largely among plate families; therefore, it is a design
specific module as well.
3.3 Step 1.3 – Formalization
Based on the methodology mentioned above, parameters for each module are identified
and classified in Table 1. These correspond to the design modules listed in Figure 3.
Table1 Parameter List of Design Modules
Parameter Type Design Modules Parameters
Bar Width, Height, Angularity,Linear, bottom fillet Position
Independent Counter-bored Hole Inside Dia., Outside Dia,Angular Position, Inner gripdepth, Linear Position
Standard Bar draft angle, end rounds
Counter-bored Hole draft angle
Dependent Bar Length
Counter-bored Hole Counter bore depth.
17
3.4 Step 2 – Assemble Features to Create a Parametric Model
The refiner plates that we are currently considering are for Twin Flo, low-consistency
refiners. Casting patterns for these plates are primarily sized first by plate configuration and then
by bar configuration. Plate configurations includes: (1) plate size, ID and OD dimensions, (2)
plate angle, the segment angle and field angle, (3) bolt hole sizes and locations, and (4) stock
material thickness. Bar configuration includes: (1) bar angle, (2) bar width, (3) groove width,
and (4) the number of continuous bars. These parameters represent functional requirements that
impact customer satisfaction. The plate configurations serve different sizes of refiner,
throughput, and fiber characteristics. The bar configurations affect the fiber characteristics, plate
life, and productivity. Not all of the parameters can be listed in this paper; thus, only normal
bars and plate associated parameters are listed as an example in Tables 2 and 3.
Table 2 Independent Parameters for Low-Consistency Refiner Plate Parametric Model
Feature Parameter name Definition
Bpad_thickness butt pad thickness(should be zero if the plate has no overhang)
IR Inside radius
OR Outside radius
TR The radius where the bar stop extending (terminated) before the barfeedin plane
Segment_angle plate segment angle
Field_angle angle per bar pattern/field
Plate
Nominal_overall_thickness overall thickness before stock material
Inner_grip bolt hole id depth
Hole_id bolt hole id
Hole_od bolt hole od
Num_of_holes number of holes on the plate (maximum is 3 holes)
Inner_hole_loc radial distance of the middle bolt hole in case of 1 hole or 3 holes
Bolt_Hole
Outer_hole_loc radial distance of the two other holes in the case of 2 or 3 holesSide_stock the side stock material thickness
Stock_draft the draft angle on the stock materialMachine Stock
Vert_stock stock material thickness on the top of the bar and bottom of the plate
Bar_angle the bar pumping angle
Nominal_bar_width bar width before stock material
Nominal_bar_height Bar heightbefore stock material
Nominal_groove_width Groove width before stock material
Bar
Nominal_bar_od_chamfer_depth the depth of the od chamfer
18
Table 3 Dependent Parameters Used to Create a Scalable Parametric Model
Parameter name Parametric relation DefinitionOREB OR - BPAD_THICKNESS Outside radius exclude butt pad
thickness representing real patternboundary
BAR_END_ANGLE BAR_ANGLE+FIELD_ANGLE Summation of bar_angle and field_angle
OVERALL_THICKNESS NOMINAL_OVERALL_THICKNESS + VERT_STOCK*2 Overall thickness after adding stockmaterial
BAR_WIDTH NOMINAL_BAR_WIDTH-2*VERT_STOCK*TAN(BAR_DRAFT) The width of the bar after adding stockmaterial
BAR_HEIGHT NOMINAL_BAR_HEIGHT+VERT_STOCK The height of the bar after adding stockmaterial
GROOVE_WIDTH NOMINAL_GROOVE_WIDTH+2*VERT_STOCK*TAN(BAR_DRAFT)
The width of the groove after addingstock material
PITCH BAR_WIDTH+GROOVE_WIDTH Pattern incremental distance
NUM_OF_BARS_COPIED_LEFT FLOOR((OREB-IR)*SIN(BAR_ANGLE)/PITCH) Max number of bars from the fieldreference line to the left (to ID) beforethe pattern will drop out of the field (fieldreference line is the line parallel to barand passing through the top-left cornerof the field)
CORNER_LEFT_OVER The projected distance from the upperright corner of the field to the last bar ofthe field
Max number of bars the pattern shouldhave from the field reference line to theright (to OD) before the pattern will dropout of the field
NUM_OF_BARS_COPIED_RIGHT
NUM_OF_BARS_COPIED_RIGHT=FLOOR(2*OREB*SIN(0.5*FIELD_ANGLE)*COS(BAR_ANGLE+0.5*FIELD_ANGLE)/PITCH)CORNER_LEFT_OVER=2*OREB*SIN(0.5*FIELD_ANGLE)*COS (BAR_ANGLE + FIELD_ANGLE * 0.5) -(NUM_OF_BARS_COPIED_RIGHT) * PITCHIF CORNER_LEFT_OVER <BAR_WIDTH/SIN(BAR_ANGLE)*SIN(BAR_END_ANGLE)NUM_OF_BARS_COPIED_RIGHT=NUM_OF_BARS_COPIED_RIGHT-1CORNER_LEFT_OVER = 2*OREB*SIN (0.5 *FIELD_ANGLE)*COS (BAR_ANGLE+ FIELD_ANGLE*0.5)-(NUM_OF_BARS_COPIED_RIGHT) * PITCHENDIF
The implementation of the parametric model was done on both Pro/Engineer (ProE) and
AutoCAD Mechanical Desktop (MD) parametric modeling package. We found that the most
significant weakness of the MD, especially in refiner plate design, is its patterning definition
(arraying) which does not allow any dimension of feature change with respect to the pattern
dimensions (driving dimension) and constraints. In other words, the copy and pattern function in
MD only duplicate the geometry. This is illustrated in the sketch of a constrained protrusion
shown in Figure 7.
19
Direction copied
Co-linear edge
Co-linear edge
parent
Figure 7 An Array of Sketched Features Does Not Maintain Co-Linearity in MD
Other weaknesses of MD lie in its limited geometric handling capability, such as the
inclusion of free form curves in the sketch. To achieve a scalable model, in which the extensible
bars with respect to the plate size and the ability to control the number of bar instances on the
plate are required, array features as shown in Figure 8 are used. A bar that is so long enough that
after arraying, it will remain crossing over the entire field. Then intersect and cut to obtain the
required bar geometry. This method causes several downstream problems. They are (1) the
fillet at the bottom of the bar cannot be included in the pattern, (2) the scalability of the model is
limited to a single field instead of the entire plate, and (3) the use of multiple cut and intersect
features results in an error prone model.
Figure 8 An Array of Bar Crossing Over the Entire Field
The resulting refiner plate model from MD is shown in Figure 9. It can be seen that the
fillet features at the bottom of the bar are not successfully incorporated into the model.
20
Although, the features can be automatically included in the tool profile when machining the
casting pattern, the incomplete model will disable or cause inaccuracy in downstream
engineering analysis. Note that if the model is not created with this scalability regard, all
features should be able to incorporate into the model.
45ta3f42ta4f
Figure 9 Two Customized Model in Mechanical Desktop, Left-3 Fields, Right 4 Fields
Another difference between MD and ProE worth discussing is the definition of reusable
features. MD’s reusable feature, namely toolbody, is a model created in part mode. Toolbody is
imported into assembly mode then used to Join/Cut/Intersect with the part creating in the
assembly mode. MD treats toolbody as a geometry that is added to the model; thus, any
extended cut or intersect operation do not affect the current model. This allows reusable features
to be easily created in MD without having to worry about the features used to create it. ProE’s
reusable feature, namely user-defined feature, consists of a group of features. Adding a group of
features to a model means adding that sequence of features to it. Thus, any extended cut or
intersect operation included in the user-defined feature will effect the model. Therefore, reusable
feature is more difficult to create and use in ProE than in MD.
ProE patterning definition allows the size of patterned geometry change with respect to
constraints and dimension incremental values as shown in Figure 10. This capability facilitates
21
much of the refiner plate design task and increases model scalability. The scalability of the
parametric model lies heavily on the well-established dimensioning and constraint schemes as
well as the datum system. These tasks must be planned prior to the creation of any reusable
features.
Bar is aligned tothis datum lineand od curve
Bar is automaticallyextended becausePROE maintain thealignment constraint.
Figure 10 Bar End Alignment to Datum References during Patterning in ProE
Even though ProE appears to be a robust parametric solid modeler, some limitations and
underlining errors still exist. In the refiner plate parametric design, we found that using a
composite datum curve (continuous curve) to end the bar is efficient because it allows the bar to
be pattern over the entire plate. Consequently, the number of bar instances is easy to control.
However, this also results in ambiguity in constraint scheme. As shown in Figure 11, the bar can
intersect three points on the curve in the box at the line position. ProE selects the solution
inconsistently, where in this case we do not always want the bar to cross the curve. However, a
refined dimensioning scheme may be able to prevent such problem.
22
Multiplesolutionsregion
θ
Figure 11 Example of Multiple Constraint Solutions (θ = bar angle) in ProE
These errors and limitations prevent the solid model, especially that of a complex refiner
plate, from being fully parametric (100%). Furthermore, this raises similar issues encountered
when creating reusable features, namely, should we spend a lot of time making a complex, fully
parametric model or take less time to create a simpler parametric model which requires manual
adjustments? Our experience indicates that creating a 90% scalable model requires about 3-4
days while customizing this model requires 0.5-2.5 hours. Creating a 70% scalable parametric
model requires about 1.5-2 days while customizing it takes 3.5-4 hours. Since time and man-
hours in industry are limited, achieving an 80% or 70% scalable model in the refiner plate design
appears most suitable, where the remaining 20% to 30% of design requires manual adjustment
and customization. Figure 12a shows a 70% scalable model where the bars are still crossing the
boundary of the bolt-hole. Figure 12b shows the parametric model after manual customization.
Finally, Figure 13 shows to completed models with the bolt-holes added.
23
(a) A 70% Scalable Model (b) Manually Customize to Clear theBars from the Holes
Figure 12 Customizing a 70% Scalable Parametric Model
45ta3f 42ta4f
Figure 13 Two Customized Models in ProE, Left 3 Fields, Right 4 Fields
3.5 2D Drawing versus Parametric Feature-Based Solid Modeling
In this section we compare the time and cost needed to obtain a casting pattern of a
refiner plate using AutoCAD 2D drawing and a solid parametric model in ProE. A 2D drawing
and IGES surface model from ProE of the same refiner plate were sent to a contractor to get a
quote on time and cost to machine a casting pattern. The time and cost spent to make those 2D
and solid model were estimated with the results summarized in Table 4.
24
Table 4 Cost Comparison: 2D Drawing versus Solid Model
Method Activity Lead time (week) Cost
Making 2D Drawing 1 $140
Pattern Making 3 $1,400
Mounting 1 $400Total 5 $1,940
Making Solid Model and archive a 2D
1 $210
Pattern Making 2 $1,000
Mounting 1 $400Total 4 $1,610
1 $330
Aut
oCA
D2D
D
raw
ing
PR
OE
Sol
id
Mod
eler
Solid Model Saving over 2D
Note: Assumes labor cost is $35 per hour formaking the model in both AutoCAD and ProE.
The results in Table 4 show that using a solid model will save approximately one-week in
lead-time and $330 dollars (17%) over the 2D drawing. This savings is marginal if the cost of
the software is taken into account; however, software purchasing costs can be amortized over
time so that they become negligible. Moreover, a solid parametric model provides downstream
benefits from the software associativity; for instance, open area at each cross sectional radius
point calculation, mass properties (surface area, volume, and weight) calculation, finite element
analysis, etc. In addition, a parametric solid model facilitates design verification in that it allows
designers (and customers) to see exactly what the real product will look like and make changes
prior to prototyping and pilot production. Thus, this will save additional cost and lead-time that
may be incurred from the 2D-model ambiguity.
4 CLOSURE
In the preceding sections we have outlined a parametric modeling approach that we are
developing to facilitate design verification, reduce lead-time, and improve product
customization. We are currently testing our approach for designing low-consistency refiner
plates for pulp processing as described in Section 3, having benchmarked solid model
25
development in two commercially available software packages: Pro/Engineer and AutoCAD
Mechanical Desktop. The advantages and disadvantages of each software package have been
discussed briefly. In addition, a comparison of lead-time and labor costs for utilizing 2-D
drawings and solid modeling are presented. Having formalized a parametric modeling approach
for refiner plate design, our next step is to integrate the aforementioned approach with functional
analyses which relate plate geometry and topology to performance requirements (e.g., average
pulp diameter, flow rate, and processing time) and manufacturing and casting cost estimates to
develop a system-level optimization framework for refiner plate design.
REFERENCES
Anderl, R. and Mendgen, R., 1995, May 17-19, "Parametric Design and Its Impact onSolid Modeling Applications," Proceedings of the 3rd Symposium on Solid Modeling andApplications, Salt Lake City, UT, ACM SIGGRAPH, pp. 1-12.
Anderson, D. M. and Pine, B. J., II, 1997, Agile Product Development for MassCustomization, Irwin, Chicago, IL.
Feng, C., Huang, C., Kusiak, A. and Li, P., 1996, "Representation of Functions andFeatures in Detail Design," Computer-Aided Design, Vol. 28, No. 12, pp. 961-971.
Lichtenberg, J., 1998, "Parametric Software Brings Concurrent Engineering to Casting,"Transactions of the American Foundrymen’s Society, Vol. 106, pp. 323-325.
Pahl, G. and Beitz, W., 1996, Engineering Design: A Systematic Approach, 2Rev.ed,Springer-Verlag, New York.
Pine, B. J., II, 1993, Mass Customization: The New Frontier in Business Competition,Harvard Business School Press, Boston, MA.
Rosen, D. W., 1998, December 1-4, "Progress Towards a Distributed Product RealizationStudio: The Rapid Tooling TestBed," 3rd IFIP WG 5.2, Knowledge Intensive CAD Workshop,Tokyo, Japan, .
Shah, J. J. and Mantyla, M., 1995, Parametric and Feature-Based CAD/CAM, JohnWiley & Sons, New York.
Ulrich, K., 1995, "The Role of Product Architecture in the Manufacturing Firm,"Research Policy, Vol. 24, No. 3, pp. 419-440.
Ulrich, K. T. and Eppinger, S. D., 1995, Product Design and Development, McGraw-Hill, Inc., New York.
CAPTION LIST
Figure 1 Product Design Customization and Production Process
Figure 2 Parametric Modeling Approach using Reusable Features
Figure 3 Function Hierarchy, Feature Assignment, and Design Modules
Figure 4 Example of Reusable Bar Features
Figure 5 Parameterization of a Reusable Bar Feature
Figure 6 Relations and Constraints for a Reusable Bar Feature
Figure 7 An Array of Sketched Features Does Not Maintain Co-Linearity in MD
Figure 10 Bar End Alignment to Datum References during Patterning in ProE
Figure 11 Example of Multiple Constraint Solutions (θ = bar angle) in ProE
Figure 12 Customizing a 70% Scalable Parametric Model
Figure 13 Two Customized Models in ProE, Left 3 Fields, Right 4 Fields
Table 1 Parameter List of Design Modules
Table 2 Independent Parameters for Low-Consistency Refiner Plate Parametric Model
Table 3 Dependent Parameters Used to Create a Scalable Parametric Model
Table 4 Cost Comparison: 2D Drawing versus Solid Model