Comparison between CAD models based on geometric

information
Montasser, Billah Letaief, Mehdi Tlija, Raoudha Gaha, Borhen Louhichi

To cite this version:

Montasser, Billah Letaief, Mehdi Tlija, Raoudha Gaha, Borhen Louhichi. Comparison between CAD
models based on geometric information. CFM 2017 - 23ème Congrès Français de Mécanique, Aug
2017, Lille, France. �hal-03465229�

HAL Id: hal-03465229

https://hal.archives-ouvertes.fr/hal-03465229
Submitted on 3 Dec 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

CAD Model Comparison: Manufacturing cost

estimation Based On Unified Feature Technology

Montasser Billah Letaiefa, Mehdi Tlijab, Raoudha Gahab, Borhen Louhichia

a.LMS, ENISO, University of Sousse, 4023 Sousse, Tunisia
b.LGM, ENIM, University of Monastir, 5019 Monastir, Tunisia
[email protected]

Abstract :
Each product developed by a company enriched its know-how. So this expertise needs to be
highlighted for reuse in modeling of new products. The reuse of acquired knowledge engenders time
and money benefits. Thus, this paper proposes a new approach for CAD model comparison basing on
an unified feature technology, in manufacturing semantics. This feature modeling is used to facilitate
the comparison between a new product and a database of CAD models already mastered. The
comparison model allows the reuse of the CAM data and the cost estimation of the new product. In
this article, the proposed approach is detailed and a case study is presented to highlight the major
contributions.

Keywords: CAD Model; geometric entities, unified feature technology,

comparison; Digital Mock-Up.

1 Introduction
The majority of industrial companies develop their products using different methods. Nerveless, The
employ of a (Digital Muck-Up) DMU, from the geometric modeling phase to the marketing phase,
remains a common used tool. The DMU is based on the geometric definition for multidisciplinary
engineering applications. Over time, the geometric shape evolves according to the customers' needs.
Thus, changes in the product shape into the Computer Aided Design (CAD) and / or subsequent
phases are required. This evolution will also impact the technical data specific to each discipline
(manufacturing, FE calculations, Assembly, etc.). Thus, each modification realized by
multidisciplinary actors enriches the geometric model by information. This redundancy and wealth of
information drives companies to retain and reuse data that describes reliable and proven products, as
well as capitalize on the knowledge gained in the past. The last strategy allows the company to benefit
from expertise acquired during the manufacturing of similar product: reduce the cost and the
development time of new products as well as circumvent the problems already encountered.
Thus, this work presents a new tool, based on an Unified Feature Technology (UFT), allowing the
comparison of geometric models in manufacturing semantics.

1
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

This paper is organized as follows. In section 2, a review of the literature is presented. Then, an
overview of the proposed model is described as well as the assumptions and approaches used. Section
4, a case study is detailed to validate the proposed tool. The conclusions and perspectives for this work
are presented in Sect. 4.
2 State of the art
The need to reduce product development time and cost initiated the establishment of comparison
tool. A decade years ago, numerous researches are interested to the CAD models comparison.
Louhichi et al. developed an algorithm to compare two CAD models (initial Work Package (iWP) and
modified Work Package (mWP))[1]. The algorithm is based mainly on faces and adjacent faces for the
identification of changes between iWP and mWP. This work targets the propagation of changes within
a DMU. Souaissa et al. proposed a new geometrical and topological descriptor to compare and
localize, automatically, changes realized on models [2]. This comparison scenario is used to perform a
partial mesh on modified zones. The complete mesh is avoided [3]. The approach in [2,3] is based on
the comparison of the metric and inertial tensors associated to the Boundary Representation (B-REP).
The comparison approaches are used for different tasks as collaborative engineering [4], finite
element analysis [5],meshing [6] and manufacturing processes [7–12]: Huifen et al. developed a
system, based on FT, to support collaborative engineering, where collaborators can act together from
anywhere. This system is founded on CAD, CAPP and CAM features [4]. These features are used as
information to characterize a product. A feature classes are defined: general, collaboration, design,
manufacturing, shape, process, precision and material. The feature modeling is performed by
instantiation and is applied to all features of the proposed system to form a library. For finite element
analysis, Richard et al. evaluated current CAD systems and provided recommendations for future
CAD systems [5]. There commended architecture of CAD systems respects both manufacturing and
analysis constraints. The new system must be based on the features and supports the multidisciplinary
activities. Each graphical representation must be able to maintain the shape module as well as adjust
and propagate the parameters for the finite element analysis module. W. Duan et al. proposed a solid
modeling tool based on features, because features recognition, after extraction, is a difficult task with
several limitations [6]. This tool comprises two mechanisms: the first allows designers to define a
feature-oriented application; the second provides directly the mesh from the feature definition for
downstream applications.
The manufacturing process is the aim of most researches using CAD models comparison, based on
FT. Zhenbo et al. proposed an approach to generate automatically a 3D assembly dimension chain
based on the feature model [7]. The developed module is integrated into CATIA® system based on the
Feature Attributes Set (FAS) concept.FAS model requires only information about the closed loop
formed by the features, in order to generate automatically the dimension chain. In order to create an
automatic system for holding small parts for the rapid prototyping process, Choi et al. proposed a new
Reference-Free Part Encapsulation(RFPE) technique [8]. The developed program generates a process
plan using the feature-based modeling technology. This software comprises mainly two modules: a
feature-based modeling module and a machining data retrieval module. An additional module
comprising tools database is used for tool selection corresponding to machining feature. So, the tool
database is built firstly. Hoque et al. established a concept to design basing only on manufacturing
features [9]. A library of predefined features is taken into account when modeling a product. Hence,
machining problems will be detected faster and the best manufacturing solutions will be deduced. As a
result, the production cost will decrease. Huang et al. presented an effective approach to retrieving
sub-parts of 3D CAD models in order to be reused in the manufacturing process [10]. This approach,
based on features, evaluates not only the similarity according the geometric level but also the
manufacturing semantics level. For the comparison between features, two filters to search sub-parts.

2
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

The first is a designation code describing the part type, the raw material, etc. The second codes the
manufacturing features coupling, which uses matrix to describe the distribution of the machining
features on a 3D CAD model. The integration of these information on a prototype system, allows the
similarity detection between features. Wong et al. developed a feature-based design system to
integrate CAD and CAPP (Computer-Aided Process Planning) functions to machine prismatic parts
[11]. This approach is based on the conversion of geometric data into input data for the planning
process to avoid the feature recognition task. The method imposes requirements as in the case of part
modeling with a manufacturing semantics, the designer must first create the raw material and then uses
predefined feature to remove material. Holland et al. proposed an approach for feature extraction from
a STEP file, in order to be used in a material deformation process [5]. This approach facilitates
designer’s task, even in the lack of knowledge on metalforming. The method allows estimating the
product manufacturing cost.
Kumar et al. [13,14] presented a quality loss function approach to calculate total cost of product. The
computation model is based on Taguchi’s method. The manufacturing tolerance value are optimized in
order to minimize the product cost. Four criteria are adopted for this optimization: Worst case, RSS,
Spotts and Estimated mean shift. The calculation of the cost is expressed in exponential model. Ghali
et al. [15] proposed an approach allowing tolerance integration into a CAD model, while taking into
account functional and manufacturing requirements in an early DMU phase (DMU). The proposed
approach consists on broadening the tolerances values of difficult machined dimensions while
respecting the functional requirements. Thus, the total cost of assembly decreases. The same
mathematical model cited in [13,14], is used.
According to the works cited above [1–12], the FT is the base of different developed approaches.
However, different definitions of features are assigned and used according the studied discipline of life
product cycle. Thus, a definition of an unified feature is required. In this paper, a CAD model
comparison approach based on unified feature for manufacturing cost estimation is proposed.
3 Proposed model
The proposed approach is based on a new concept, the Unified Feature Technology (UFT), in order to
compare a Reference Model (RM) with a Company Models DataBase (CMDB). The UFT is uniform,
from a structural point of view, and includes multidisciplinary data while adopting the following
assumptions:
CAD data: geometrical and topological structures of CAD model as well as specified
dimensional tolerances. The same platform is used to avoid the loss of information when
passing through standard formats like the STEP format. The geometrical and topological
parameters are standardized. For example, in the case of an open drilling feature (Fig. 1), the
geometrical parameters are the diameter, length as well as the two angles α and β. The
topological parameter is one cylindrical face (Fig. 2).

3
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

Loop 1 Center
Loop 1
α

Length
β
Loop 2 Loop 2 Center
Fig. 1. Geometric parameters of an open drill

Topological Data Open cylinder

Diameter
Length

Loop 1 Centre (X1, Y1, Z1)

CAD Data Geometrical parameters
Loop 2 Centre (X2, Y2, Z2)
α Angle
β Angle

Maximum Tolerance δ Max

Dimensional Tolerances
Minimum Tolerance δ Min

Fig. 2. Standard CAD Data Structure for Drilling Feature

CAM data: All possible manufacturing parameters (cutting parameters, machining strategies,
etc.) are standardized, as quality standard tools and maintenance. For example, standard CAM
data structure for drilling feature is shown in Fig. 3. This standard can be used by all
engineering disciplines. The chronological order of manufacturing operations specified in plan
process is not considered. For the face milling feature, the raw material must be
communicated through an interface. A comparison between the bounding box and the
Reference Model (RM), the feature will be identified with corresponding parameters (the
target face, the extra thickness, etc.). In this work, tree types of CAM features (XF) are fixed:
Face Milling Feature (FMF), Turn Milling Feature (TMF) and Drilling Feature (DF). This
choice is established only to simplify the approach,. Each feature is studied separately, without
taking into account the interaction between features, as realized by Huang et al. [10].

4
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

Machining Plan
Model Material
Machine Definition
Start Point
Tool Path Type (Pointing, Drilling, Bore,…)
Name (Number)
Offset
Diameter
Corner Radius
Useful length
Material
CAM Data Tool Parameters Cutting speed
Feed per Tooth
Tooth Number
Type Tool holder
Spindle Speed
Feed
Type (Drill, Pointer, …)

Clearance
Retraction
Linking Parameters Top of Stock
Depth
Break Trough Amount (on Z)

Fig. 3. Standard CAM Data Structure for Drilling Feature

Two costs are presented for each feature (Fig. 4). The first represents the theoretical average
cost of the feature (Eq. 1) which is computed using the exponential function (Eq. 2). g(δ)
represents the cost of a machining operation with tolerance δ. C0 and C1 are constants obtained
by experimental machining [14]. The second represents the true cost of the feature based on
real results of manufacturing, i.e. After the product manufacturing, the feature cost is deduced
and inserted into the CAM data.
Theoretical average cost based on exponential function
Cost Data
True cost inserted after manufacturing
Fig. 4. Standard CAM Data Structure

g (δ iMin ) + g ( δ iMax )
CiMoy = ; with i is the ith XF. (1)
2

1
g ( δ ) = C0 C1δ
(2)
e

The proposed algorithm comprises four sub-algorithms (Fig. 5): CMDB organization, comparison
based on codes, comparison based on features and manufacturing cost estimation.

5
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

Step 2: Reference
Step 1:
Model

Data
Extraction Organisation de la
BDMC
Features
classification

Model
codification
RM Code Comparison CMK Code
based on codes

CBSR Computation

Selection of FP% of models

with the highest CBSR

Data Recovery of selected

models (CMDBFP% )

Step 3: RM-XFi Comparison based

CMk-XFj C0, C1
on Features

FBSR Computation

Feature/Feature Comparison:
MFBSR Affectation RM-DFi % CMk-DFj
RM-FMFi % CMk-FMFj

GFBSR Computation RM-TMFi % CMk-TMFj

Selection of similar CM to RM

Step 4: Condition 1 XFi and XFj Condition 3

Correspondence?

Condition 2
CAM parameters and CAM parameters and Condition 2
Cost Association Cost Update

Approximate Cost
Determination for all XFi

Cost RM Computation

Fig. 5. Proposed model

CMDB organization: The data of all CAD models, used by the company, are
extracted. According to the features properties, a codification of models is established (Fig. 6).

6
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

Thus, CMDB is defined. The code assigned to each Company CAD Model (CM) becomes the
digital communication support as an identifier of the model. This code expresses the type and
the number of model features. The data aggregation is based on feature classification. The
features are classified into three types, considered in this work (XF: FMF, TMF and DF). Each
feature is defined by CAD, CAM and costing data.
Model 1 Model 2 Model n

Data Data Data

Extraction Extraction Extraction
… … …
Features Features Features
Classification Classification Classification

Model Model Model

Codification Codification Codification

Data Aggregation of
CMDB

Fig. 6. Sub-Algorithm for CMDB organization

Comparison based on codes: CAD data of the RM are extracted. The RM features are
classified. Also, a code will be assigned to the RM that highlights the type and number of
these features. To reduce the runtime of the proposed approach, the code-based comparison is
established. For each CM, the Code-Based Similarity Ratio (CBSR) is computed using Eq. 3;
such as nRM and nCM are the total numbers of RM and CM features respectively. A Filtration
Percentage (FP%) is fixed and chosen according to the CM number of CMDB. Thus, a
recovered data CMDBFP% is obtained by the selection of FP% of the highest CBSR.

nCM − nRM
CBSR = 1 − (3)
nRM
Comparison based on features: This sub algorithm consists on the comparison between
feature CAD properties of RM (RM-XFi) and the CAD data of the same feature type of CM
(CMk-XFj); such as CMk are the kth CM of CMDBFP% and XFi is the ith feature of XF type. A
Feature-Based Similarity Ratio (FBSR) is defined according the feature parameters of RM and
computed for each CM. For each feature type (XF) of the same CM, the maximum FBSR,
denoted by MFBSR, is determined. Indeed, if the RM has k features, then each CM of the
CMDBFP% has k MFBSR. A new Global ratio Feature-Based Similarity Ratio(GFBSR),is
computed for each CM. This ratio represents the average of the MFBSR. Subsequently, the
CM which has the largest GFBSR is the most similar model to the RM.
Manufacturing cost estimation: A correspondence between the Feature of Selected CM
(FSCM) and Feature of RM (FRM) is established according the three following conditions:

7
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

- Condition 1: If the topological and geometrical parameters as well as dimensional tolerance

value are identical, then the CAM parameters and cost of FSCM are directly associated to
FRM.
- Condition 2: If only the topological parameters are identical, then the CAM parameters can
be update according the differences of geometrical parameters and tolerance value between the
FSCM and FRM. The constant C0 and C1 of SCM feature are reused to compute an
approximate manufacturing cost of FRM (Eq. 1).
- Condition 3: If any correspondence between the FRM and FSCM are detected, then a
similarity searching in CMDBFP% based on topology is established. In the case of the
detection of topological similarity, the above condition (condition 2) is satisfied.
4 A case study
In this section, a case study is presented to validate proposed algorithm. Nine CAD models are
chosen to form CMDB (supposed to be used and validated by a company) as shown in Fig. 7. The sub-
algorithm of CMDB organization allows the codification of the 9 models as mentioned in Fig. 6. The
RM features' data are extracted, classified and codified. The RM includes tow drilling, tow face
milling and one turn milling features (2DF-2FMF-1TMF). Thereafter, a comparison based on codes is
established using Eq. 3. At this step, the CBSR will be allocated to each CM of the CMDB. As a
result, a refined database is filtered from CMDB: CMDB33.33%. In this case of study, a FP is chosen
equal to 33.33%, which represent tier of the models, because the number of CM is rather low.
Based on geometric parameters of standard CAD data of each features, the FBSR is computed (Tab.
1). Then, the MFBSR is determined. Each MFBSR represents the maximum value of FBSR for each
FRM (Tab. 2). After that, GFBSR is computed for the three CM (Tab. 3). Subsequently, the CM5 ,
with GFBSR =1.049, is selected as the most similar model to the RM.
A correspondence between the features of CM5 and RM is established according the three
conditions presented previously. This step is essential in order to collect and reuse cost data and
manufacturing parameters. Once the constants C0 and C1 are identified, the machining cost for each
feature is computed (Eq. 1). Finally, the manufacturing cost of RM is deducted by summing the cost
values determined previously.

Tab. 1 . Results of FBSR computation

RM-DF1 RM-DF2 RM-FMF1 RM-FMF2 RM-TMF1

CM4 -DF1 1.000 1.100 CM4 -FMF1 1.026 1.125 CM5 -TMF1 0.991
CM4 -DF2 1.000 1.100 CM4 -FMF2 1.692 1.000 CM6 -TMF1 0.267
CM5 -DF1 1.100 1.225 CM5 -FMF1 0.804 1.125
CM5 -DF2 0.800 0.900 CM5 -FMF2 0.533 0.400
CM5 -DF3 0.800 0.900 CM6 -FMF1 0.693 1.125
CM6 -DF1 1.200 1.350 CM6 -FMF2 1.255 1.150
CM6 -DF2 1.100 1.225

8
 23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

1) 2) 3)

4) 5) 6)

7) 8) 9)
Reference
Model

Data
Extraction

Features
Classification
Organization of
CMDB
Model
Codification
RM Code CMK Code
Comparison based
on codes

CBSR Computation

Selection of 33.33% of
models with the highest
CBSR

Data Recovery of
selected models
(CMDB33.33% )

Comparison based
on Features
C0, C1

FBSR Computation

MFBSR Affectation

GFBSR Computation

Selection of similar CM

Condition 1 XFi and XFj Condition 3

Correspondence?

Condition 2
CAM parameters and CAM parameters and Condition 2
Cost Association Cost Update

Approximate Cost
Determination for all XFi

Cost RM Computation

Fig. 7. Result of the application of proposed algorithm on the case study

9
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

Tab. 2. MFBSR Affectation

RM-DF1 RM-DF2 RM-FMF1 RM-FMF2 RM-TMF1
CM4 1.000 1.100 1.692 1.125 0.000
CM5 1.100 1.225 0.804 1.125 0.991
CM6 1.200 1.350 1.255 1.150 0.267

Tab. 3. GFBSR Computation

GFBSR
CM4 0.983
CM5 1.049
CM6 1.044

5 Conclusion and perspectives

In this paper, a new approach to compare 3D CAD models, in manufacturing semantics, is detailed.
The method is founded on the definition of UFT including multidisciplinary data (CAD, CAM and
Cost data). In order to reuse the company knowledge (CAM and Cost data ), with a reasonable run
time, four sub-algorithms are developed: CMDB organization, comparison based on codes,
comparison based on features and manufacturing cost estimation. As a consequence, time and money
benefits are generated.
The proposed algorithm is flexible to be optimized in future works. The approach might be
generalized to consider other features and environmental constraints. Also, the cost computation
model can be optimized by integrating all standard parameters of each feature.

References
[1] B. Louhichi, L. Rivest, Maintaining consistency between CAD elements in collaborative design
using association management and propagation, Computers in Industry. 65 (2014) 124–135.
[2] K. Souaissa, J.-C. Cuillière, V. François, A. Benamara, H. BelHadjSalah, Automatic modification
retrieval between CAD parts, Mécanique & Industries. 11 (2010) 85–92.
[3] J.C. Cuillière, V. François, K. Souaissa, A. Benamara, H. Belhadjsalah, Automatic comparison and
remeshing applied to CAD model modification, CAD Computer Aided Design. 43 (2011) 1545–1560.
[4] W. Huifen, Z. Youliang, C. Jian, S.F. Lee, W.C. Kwong, Feature-based collaborative design,
Journal of Materials Processing Technology. 139 (2003) 613–618.
[5] R.F. Riesenfeld, R. Haimes, E. Cohen, Initiating a CAD renaissance: Multidisciplinary analysis
driven design. Framework for a new generation of advanced computational design, engineering and
manufacturing environments., Computer Methods in Applied Mechanics and Engineering. 284 (2015)
1054–1072.
[6] W. Duan, J. Zhou, K. Lai, FSMT: a feature solid-modelling tool for feature-based design and
manufacture, Computer-Aided Design. 25 (1993) 29–38.
[7] G. Zhenbo, W. Jing, C. Yanlong, Y. Jiangxin, Automatic Generation of 3D Assembly Dimension
Chain Based on Feature Model, Procedia CIRP. 43 (2016) 70–75.

10
23ème Congrès Français de Mécanique Lille, 28 Août au 1er Septembre 2017

[8] D.S. Choi, S.H. Lee, B.S. Shin, K.H. Whang, K.K. Yoon, S.E. Sarma, A new rapid prototyping
system using universal automated fixturing with feature-based CAD/CAM, Journal of Materials
Processing Technology. 113 (2001) 285–290.
[9] A.S.M. Hoque, T. Szecsi, Designing using manufacturing feature library, Journal of Materials
Processing Technology. 1 (2008) 204–208.
[10] H. Rui, Z. Shusheng, B. Xiaoliang, X. Changhong, H. Bo, Computers in Industry An effective
subpart retrieval approach of 3D CAD models for manufacturing process reuse, Computers in
Industry. 67 (2015) 38–53.
[11] T.N. Wong, K.W. Wong, A feature-based design system for computer-aided process planning,
Journal of Materials Processing Tech. 52 (1995) 122–132.
[12] P. Holland, P.M. Standring, H. Long, D.J. Mynors, Feature extraction from STEP (ISO 10303)
CAD drawing files for metalforming process selection in an integrated design system, Journal of
Materials Processing Technology. 125–126 (2002) 446–455.
[13] R. Sampath, N. Alagumurthi, R. Ramesh, Optimization of design tolerance and asymmetric
quality loss cost using pattern search algorithm, International Journal of Physical Sciences. 4 (2009)
629–637.
[14] S.R. Kumar, N. Alagumurthi, R. Ramesh, Calculation of Total Cost , Tolerance Based on
Taguchi ’ s , Asymmetric Quality Loss Function Approach, American Journal of Engineering and
Applied Sciences. 2 (2009) 628–634.
[15] M. Ghali, M. Tlija, N. Aifaoui, E. Pairel, A CAD method for tolerance allocation considering
manufacturing difficulty based on FMECA tool, The International Journal of Advanced
Manufacturing Technology. (2017).

11