MANUFACTURING COMPLEXITY EVALUATION FOR ADDITIVE AND

SUBTRACTIVE PROCESSES: APPLICATION TO HYBRID MODULAR TOOLING

O. Kerbrat, P. Mognol, J.-Y. Hascoet
IRCCyN (Institut de Recherche en Communications et Cybernétique de Nantes), MO2P Team
1 rue de la Noë, BP 92101, 44321, Nantes Cedex 03, France

Abstract Reviewed, accepted September 10, 2008

The aim of this work is to determine how to combine a subtractive process (HSM) and an
additive process (SLS) to realize tools (dies or molds). In fact, the design and manufacturing
of tools may be optimized with hybrid and modular points of view. Tools are not seen as
single pieces but as 3-D puzzles with modules; each module is manufactured by the best
process. So a new methodology is proposed: the most complex-to-manufacture areas of a tool
are determined (based on a manufacturability analysis from tool CAD model) and a hybrid
modular tool CAD model with a reduced manufacturing complexity is proposed.

1. Context of the study: hybrid modular tooling

In order to improve competitiveness in modern mass production industry, products have to
be designed and manufactured with the following two goals that are often in opposition:
- Decreasing time and cost;
- Improving quality and flexibility.
These objectives imply two design and manufacturing constraints: a rapid manufacturing
and a high level of reactivity when design evolutions are required. The current field of tooling
(dies and molds) does not break these constraints and one answer to the problem is to design
and manufacture hybrid modular tools, with modular and hybrid points of views.
Modular point of view: Instead of a single-piece tool, it is seen as a 3-D puzzle with
modules realized separately and further assembled. The two advantages are: each module may
be produced simultaneously and few modules may be changed without changing the whole
tool. As it can be seen in the example in Figure 1, the two alternatives of the product may be
advantageously manufactured with the same mold. Only one module of the mold is changed
to provide new part functions.

Fig.1. Two alternatives of a product model (Logicom®).

Hybrid point of view: Each module of the tool is manufactured by the best process, in term
of time, cost and/or quality. Presently, focus is put on comparison between a subtractive
process (HSM: High-Speed Machining) and an additive process (SLS: Selective Laser
Sintering). Another research topic investigates the combination of these two manufacturing
processes. Figure 2 shows an example of a hybrid tool.

Fig.2. Industrial injection hybrid mold [1].

519
 To illustrate the advantages of using hybrid modular tooling, Figure 3 presents an
industrial example, developed at IRCCyN [2]. It is a part from automotive industry,
manufactured by injection molding. In this part, the positions of the circular shapes have
diversifying alternatives, there are a marking that changes with the model and an evolutionary
feature for the seal positioning. The part with its evolutionary areas may be produced with just
one mold creating modules for each changing area of the piece. So modules are designed and
realized with the best process, and changed when the product model is modified.

(a) (b)

(c) (d)
Fig.3. Industrial example of hybrid modular tool [2].
(a) Seal rear door of a vehicle (b) Seal rear door CAD model with evolutionary areas
(c) Seal rear door hybrid modular tool CAD model (d) Seal rear door hybrid modular tool.

2. One important issue in hybrid modular tooling

Hybrid modular tools allow diversifying alternatives of a product and are manufactured at
lower cost in the shortest lead-time. In order to help tooling designers to choose between a
hybrid modular tool and a traditional single-piece tool, the advantages of the hybrid modular
design have to be quantified. In case of flexibility, it is quite evident that creating modules
will help multiple geometric evolutions of the future part to produce.
To evaluate the impact of a hybrid modular design on the time, cost and quality of the tool
is not easy before doing the complete preparation of manufacturing. For example, the
machining time depends on the tool path strategy. And the choice of a strategy is based on the
ability of the manufacturer, with the help of a Computer-Aided Manufacturing software. It
will be more interesting for the designer to be able to choose the best hybrid modular tool
design before complete manufacturing preparation. So criteria have to be elaborated and
utilized at the early stage.
A new methodology is developed to determine which areas of the tool will be the most
difficult-to-manufacture, directly from tool CAD model. The most difficult-to-manufacture
areas are:
- The areas which will impose an increasing manufacturing time;
- The areas which will increase overall tool cost;
- The areas for which it will be difficult to achieve a high quality level.
Examples of complex-to-manufacture areas: the areas with a fine roughness; the back
drafted areas, which obligate particular milling tool orientations (in five axis machining) or
building support (in layered manufacturing).
This paper presents this new methodology, based on the calculation of manufacturing
complexity indexes, in order to determine which areas of the tool are the most difficult-to-
manufacture.

520
3. Manufacturing complexity evaluation
The aim of this section is to determine several parameters which have a great influence on
the time, cost and quality. These parameters provide information on the most difficult-to-
manufacture pieces, or areas of a piece.
Many works on the definition of manufacturability indexes were achieved. In fact, the
determination of a manufacturing process is often based on such indexes [3]. Most typically,
manufacturability indexes include manufacturing cost, product quality and production time.
Some other works use the concept of “effort to produce the final part” as a quantification of
product complexity [4].
Three categories can be distinguished for classifying manufacturability indexes: geometric
parameters, material information and specifications. The following list of parameters is
limited to those which can be determined only with CAD model. So parameters that require a
complete manufacturing preparation analysis (for example: tool path strategy) are not taken
into account to be free from manufacturer skills.

3.1. Geometric parameters

First of all, the geometry and dimensions of the part to realize clearly affect the
manufacturing time, cost and quality. The geometric parameters have not the same influence
in case of a subtractive or an additive process.
If the tool is machined, the geometric parameters that lead a mechanical part difficult-to-
machine are:
- Maximal dimensions: a given machine has its own limitations on each axis;
- Minimal dimensions: if some dimensions are too small, it will be impossible to
machine with traditional milling tools;
- Slenderness: parts with a high slenderness ratio will be more difficult-to-machine than
other ones;
- Geometrical accessibility for the milling tool: machining the bottom of a depth pocket
implies using a long milling tool which can generate a bad quality surface;
- Curvature radius: a convex surface with small curvature radius implies using a milling
tool with a corresponding radius;
- Back drafted areas: surface orientations sometimes obligate particular milling tool
orientations and five-axis machining;
- Free form surface: lots of changes in the surface orientations have a large influence on
the number of feed rate alteration;
- Blank volume: the blank dimensions have an impact on the chip quantity and so a
consequence on the part cost;
- Etc.
In case of layered manufacturing, other geometric parameters are taken into account to
evaluate manufacturing complexity:
- Volume and height: direct influence on manufacturing time;
- Surfaces orientations: the quantity of support has an impact on the material cost,
manufacturing time and surface quality;
- Distance from the centre of the platform: the dimensional error strongly depends on
the distance from the platform centre [5];
- Quantity of skin surface: manufacturing time is higher for skin surfaces than for inner
surfaces;
- Maximal and minimal dimensions, slenderness: same impact than for a machining
process;
- Etc.

521
3.2. Material information
Obviously, the mechanical characteristics of material directly affect manufacturing
process parameters.
As an example, when the material to machine is very hard (50-60 HRC), a special range
of cutting tool materials is required (ceramic metal composites, polycrystalline cubic boron
nitride) with low feed rate [6].
So manufacturability indexes based on material will be defined according to the following
characteristics:
- Hardness;
- Young modulus;
- Ductility;
- Microstructure;
- Thermal conductivity;
- Etc.
And in layered manufacturing, the material choice is limited by the different powders
available in a machine, and melting point temperature is clearly significant.

3.3. Technical specification

The specification of high degree tolerances and surface finish always increase the number
of operations required and more expensive machines.
Of course, the consequence is a rise in the difficulty of manufacturing. Four parameters
are very sensitive with respect to the accuracy and dynamical capability of manufacturing
equipment (in case of machining process or layered manufacturing process) [7]:
- Dimensional tolerance;
- Geometric tolerance;
- Location tolerance;
- Surface finish.

4. Global and local indexes

These three categories of manufacturability indexes are subdivided in two types: global
indexes and local indexes.
Global indexes are defined for the whole tool. As an example, an index may be calculated
from the parameter “Volume”. In fact, volume has a great impact on manufacturing time in an
additive manufacturing process.
Local indexes are defined for each area of the tool. Tool CAD model has to be
decomposed into several elementary elements, and local indexes are calculated for each
element. Several methods exist to decompose a CAD model into basic components. A choice
must be done to determine the most appropriate method in this case.

4.1. Feature decomposition

Many decomposition methods are based on volume segmentation approaches [8],
considering that elementary volumes, named features, form a part. Features usually rely on
one specific field and are used as specific data to automate CAD, CAM, process planning, etc.
As an example, machining features are developed for mechanical product definition for
process planning (ISO 10303-224), but there are still no manufacturing features for additive
process. And for free-form surfaces, usually used in tooling design, machining features do not
bring enough information on the shape. Furthermore, comparing different manufacturing
processes would involve including different manufacturing features which would make the
evaluation of manufacturing complexity difficult. And sometimes there are several ways of
decomposing a part, because some shapes can be interpreted as either two features or one long

522
one [9]. So feature decomposition will not be the best way to decompose a CAD model in
order to obtain a precise view of the tool complexity. It must be obtained with a neutral
decomposition (not rely on one specific manufacturing process), which gives automatically
only one way to decompose the tool CAD model.

4.2. Examples of other CAD model decompositions

Each solid modeling method (CSG, B-rep, decomposition method) has its advantages and
disadvantages relative to the others in term of accuracy, robustness, data structure and
computing time.
Construction Solid Geometry (CSG) method is very popular because this method can
complete Boolean operations of any 3D part model relatively easily and accurately. The
problem in the CSG approach is that it is computationally expensive to represent the parts
with irregular surfaces [10].
A common decomposition method used in layered manufacturing is STL format. A major
problem with STL is on its representation of curved surface, which can only be approximated
by triangular facets [11]. The manufacturing complexity analysis will not be based on this
format, because even if this deviation can be controlled according to user’s requirements on
approximation accuracy, information is lost. In fact, there are often few details of the tool that
can change the manufacturing process choice (a curve radius of a small complex shape, for
example).

4.3. Octree decomposition

An octree is a tree data structure, which represents a three-dimensional object by the
division of space into small cubic boxes, or small parallelepipeds. The size of each box
depends on the local geometric complexity of the object represented [13]. Each box in space
corresponds to a node in the tree and each node is referred to as an octant.
To construct an octree, the object is first enclosed by the smallest box (octant) that can
completely contain the object in any direction. This octant (a cube or a parallelepiped) makes
up the root level of the octree. It is then subdivided into 8 sub-octants (4 sub-octants in case of
a two-dimensional object) which then represent the first level. The octants are classified into
three categories: black (full), white (empty) and grey (partially filled). Black octants are those
that are completely contained in the object of interest, whereas white ones are those that are
completely outside the object. Grey octants are those that are partially inside and outside the
object. The subdivision process is performed on grey octants until a desired resolution is
reached. The specified accuracy is used to determine the final size of the smallest octants [14].
Octree decompositions have been used for several years, first in computer graphics [15].
In mechanical engineering, octree decomposition is used for the verification of numerical
command tool paths [13], for interference detection in five axis machining [14] and in
robotics [16]. In rapid prototyping, octree decompositions of 3D models have been used to
realize approximate prototypes before final machining [17].
The advantages of using an octree decomposition model are:
- It do not rely on one specific manufacturing process;
- Decomposition model can acquire relatively high accuracy;
- The special location of an octant is determined by an index code; with these codes, the
position of each octant could be easily found and the geometric information such as
centre point and edge lengths could thus be calculated.
That is why an octree structure is used in the following methodology for the calculation of
local manufacturability indexes.

523
5. Methodology to evaluate manufacturing complexity

5.1. Interface
A procedure has been developed to evaluate manufacturing complexity from a tool CAD
model. The work has been carried out on a CAD software (SolidWorks 2007) with Visual
Basic language. The interface is presented in Figure 4.

Fig. 4. Interface of the manufacturing complexity evaluation methodology.

5.2. Definition of the manufacturability indexes used in the methodology

Concentration is first put on geometric parameters. Manufacturability indexes have been
developed, according to the previous analysis. Machinability indexes are defined and
presented in Table 1, and layered manufacturability indexes in Table 2. The indexes based on
an analysis of the surface orientations of the tool have not yet been developed. All these
indexes must be calculated according to the help of the tool CAD model, and without a
complete manufacturing preparation.
If the manufacturability index that has been chosen is a global index, the procedure
directly post the index value. The higher the value of the index is, the more difficult-to-
manufacture the tool is.
In case of local indexes, a step of decomposition is done, according to an octree
decomposition algorithm. Then the index value is calculated for each grey or black octant and
a color map of manufacturing complexity for this index is drawn (with automatic or
customized color scale). For each octant, the higher the value of the index is, the more
difficult-to-manufacture the fraction of the tool contained in the octant is. If the accuracy of
the decomposition is not satisfying (the octants are too big compared to the dimensions of the
tool), another level of decomposition is done, only for grey octants. When a sufficient
accuracy is reached, the decomposition is stopped. The accuracy of the octree decomposition
must be carefully determined, because if it is too high, it will dramatically increase computing
time. Nevertheless, it must not be too small with respect to the smallest dimension of the tool.
In this paper, four levels of decomposition have been chosen.

524
 Index Linked to Type
C (d x ) C (d y ) C (d z ) Maximal dimensions Global
C (r ) Tool rigidity Local
C (b) Blank volume Global
C (c) Chip quantity Global
Table 1. Manufacturability indexes for machining process.

Index Linked to Type

C (d x ) C (d y ) C (d z ) Maximal dimensions Global
C (v) Volume Global
C (s) Outside surface Global
C(h) Height Local
C(  ) Distance from the centre of the platform Local
Table 2. Manufacturability indexes for layered manufacturing process.

5.3. Calculation of manufacturability indexes

The manufacturability indexes defined in Tables 1 and 2 are calculated with the help of
the following equations (Equations 1-8):
LX 0
(1) C (d x ) 
LX max
where LX0 is the maximal dimension of the tool in x-direction, and LXmax is the length of
the X-axis of the machine. C (d y ) and C(d z ) are determined with similar equations. The
machines and tool orientation in a machine have to be previously established. For further
examples, Hermle C30U HSM and EOS 250 Xtend machine are used with z-direction as spindle
axis and layer normal orientation.
L
(2) C (r ) 
D
where L is the minimal length of the milling tool that can machine the surface included in
the octant and D is the maximal diameter of the milling tool that can machine the surface
included in the octant.
This index is based on two reports. In most cases, when the milling tool diameter is reduced,
machining time increases. Moreover, when the ratio length/diameter of the milling tool
increases, the quality of the piece realized is reduced. It corresponds to a diminish of the
milling tool stiffness. L is calculated by the difference of height between the top face of the
highest octant and the bottom face of the octant for which C (r ) is being calculated. D takes
into account both curvature radius of convex surface (a small curvature radius limits the di-
ameter of the milling tool that can machine a convex surface) and space between two surfaces
which may limit the milling tool diameter (Figure 5).

Fig. 5. Example of machining a pocket.

525
(3) C (b)  LX 0  LY0  LZ 0
C (b) represents the volume of the blank part. The higher C (b) is, the more expensive the
rough part is and consequently the more difficult-to-machine the tool is considered.
LX 0  LY0  LZ 0
(4) C (c) 
V
where V is the volume of the tool. Chips are considered as material lost, so if C (c) has a
low value, the tool will not require a large quantity of chips and the tool will be considered as
easy-to-machine.
(5) C(v)  V
(6) C ( s)  S ext
where Sext is the area of the whole outside surface of the tool.
(7) C (h)  z  Z 0
where z is the z-coordinate of the centre of gravity of the volume of the tool contained in
the octant and Z0 is the z-coordinate of the bottom face of the tool.
(8) C (  )  ( x  X 0 ) 2  ( y  Y0 ) 2
where x and y are the coordinate in the X-Y plane of the centre of gravity of the volume of
the tool contained in the octant and X0 and Y0 are the coordinate of the centre of the tool,
considering that the tool will be manufactured with its centre exactly at the centre of the
platform.

For each local index, a global one may be calculated (Equation 9) for an easier
comparison between manufacturability indexes.
j (C (ilocal) j  V j )
(9) C (i global) 
j
V j
where Vj is the volume of the fraction of the tool contained in the octant for which
C (ilocal ) j is calculated.

6. Examples of using the manufacturing complexity evaluation methodology

6.1. Comparison of CAD models for one manufacturing process: modular point of
view
This methodology allows comparing different tool CAD models, regarding one
manufacturing process. The most difficult-to-manufacture areas may be improved with a
modular point of view, designing modules in order to decrease the value of manufacturability
indexes in these areas.
As an example, a single-piece test-part CAD model is analyzed in term of local and global
indexes for subtractive process. Figure 6a presents the test-part, which is representative of
dies and molds traditionally made by High-Speed Machining, and Table 3 gives the values of
machining indexes. For C (r ) - tool rigidity index, a map of machining complexity is obtained
(Figure 6b) and most difficult-to-machine areas are thus known.
This map provides an accurate view of the manufacturing complexity of the test-part.
With the example of machining process and C (r ) as manufacturability index, the easiest-to-
manufacture areas are those where there are no limitations for the milling-tool diameter
( C(r )  0 ). The most difficult-to-manufacture areas of the test-part are at the bottom of the
circular boss, with a small curvature radius and the surface between the two high bosses.

526
 0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
(b)
(a) 4.00
4.06
Fig. 6. Example of a test-part.
(a) CAD model.
(b) Map of machining complexity.

Then the aim is to concentrate on these areas to understand why there are complex-to-
manufacture, according to this particular index. The difficulty can be due to a small space
between two bosses that allows only small diameter milling tools. On the other hand, the
difficulty can be due to a high wall that forces milling tools to be long. Alternatively, because
there is a small radius on a convex surface that implies using a milling tool with a small
radius. The further step of the methodology is to take into account modular point of view,
creating modules, manufactured aside and further gathered, to reduce manufacturing
complexity in the previous most difficult-to-machine areas, as it can be seen in Figure 7. In
this example, assembly process is not treated.

0.00
0.50
Module 1 Module 1 1.00
1.50
2.00
2.50
3.00
Module 2 3.50
Module 2
4.00
+ (a) + 4.06 (b)
Fig. 7. Example of a modular test-part.
(a) CAD model.
(b) Map of manufacturing complexity.

For a global comparison of the two CAD models, manufacturability indexes are calculated
for the single-piece test-part and the two modules of the modular one (Table 3). Concerning
the modular test-part, total indexes are calculated with the following equation (Equation 10).
C (imod .1 )  Vmod .1  C (imod .2 )  Vmod .2
(10) C (itotal) 
Vmod .1  Vmod .2
In this methodology, it is still impossible to compare different indexes between
themselves, so a comparison of the evolutions of the values of the machinability indexes is
done between the two tools (Table 3).
So it can be concluded that the modular point of view provides a modular tool with less
manufacturing complexity because the tool rigidity index C (r ) decreases by 40 % whereas
the other indexes evolutions are not significant.

527
 Modular test-part
Single-piece test part Comparison
Module 1 Module 2 Total
C (d x ) 0.185 0.185 0.043 0.178 -4%
C (d y ) 0.133 0.133 0.043 0.129 -3%
C (d z ) 0.100 0.100 0.06 0.098 -2%
C (rglobal ) 1.700 0.998 0 0.998 - 40 %
C (b) 480 000 480 000 20 280 457273 -5%
C (c) 2.35 2.09 2.01 2.09 - 11 %
Table 3. Comparison of C (rglobal ) for the two CAD models.

6.2. Comparison of manufacturing processes: hybrid point of view

Another way of using this methodology is for comparison of two manufacturing processes
(additive and subtractive) for one tool CAD model, in order to determine which parts of the
tool may advantageously be machined or realized by a layered manufacturing process.
This second example is based on the comparison of the same test-part, but with changes in
the pocket dimensions. The test-part presented in Figure 6a has the following pocket
dimensions: 30x50x30 mm, whereas the second test-part shown in Figure 8 has a 20x30x30
mm pocket. Manufacturability indexes are calculated, first for machining process (Table 4).

Fig. 8. Second test-part, with a smaller pocket.

Pocket dimensions: 30x50x30 Pocket dimensions: 20x30x30 Comparison

C (d x ) 0.185 0.185 =
C (d y ) 0.133 0.133 =
C (d z ) 0.100 0.100 =
0.00
1.00
2.00
Machining

3.00
4.00 C(rmax)
C (r ) 5.00 + 100 %
6.00
7.00
8.00 C(rglobal)
C (rglobal )  1.700 8.13 C (rglobal )  2.071 + 22 %
C (b) 480 000 480 000 =
C (c) 2.35 2.09 - 11 %
Table 4. Comparison of manufacturing indexes for machining process.

528
 An analysis of the evolutions in machinability indexes between the two CAD models is
done. In this example, it can be seen that changing the dimensions of the test-part provide new
areas of the second test-part very complex-to-machine, according to the C (r ) index. The
C (rmax ) value is doubled and the C (rglobal ) value increases by 22%, whereas the other values
evolutions are not significant.
Then, manufacturability indexes are calculated for additive process (Table 5).
Pocket dimensions: 30x50x30 Pocket dimensions: 20x30x30 Comparison
C (d x ) 0.48 0.48 =
C (d y ) 0.32 0.32 =
C (d z ) 0.28 0.28 =
C (v) 204 183 229 177 + 12 %
C (s) 36 866 35 557 -4%
3.13
5.00
10.00
Layered manufacturing

15.00
20.00
C(h) 25.00 +7%
30.00
35.00
40.00
45.00
C (hglobal )  14.250 48.44
C (hglobal )  15.259
4.51
10.00
20.00
30.00
C(  ) 40.00 ≈
50.00
60.00

C (  global )  39.018 67.60 C (  global )  39.042

Table 5. Comparison of manufacturability indexes for layered manufacturing.

Changing the dimensions of the test-part provide few evolutions in the values of the
different indexes. So the two test-parts will have the same level of manufacturing complexity
in case of an additive process.
Consequently, for this second test-part, the areas which are the most complex-to-machine
would advantageously be manufactured with an additive process, creating a hybrid part. In
this hybrid part, the areas which are easy-to-machine would be machined and the most
difficult-to-machine would be manufactured by a layered manufacturing process.

7. Conclusion and future work

In this paper, a manufacturing complexity evaluation methodology is exposed.
Manufacturability indexes have been developed, and an interface is created to calculate them
directly from a tool CAD model. This new approach provides an accurate view of which parts
of the tool have to be improved in order to reduce manufacturing difficulties. Then modular
and hybrid points of view allow designing a hybrid modular tool which will be less difficult-
to-machine than the first single-piece tool, and consequently be manufactured at lower cost, in

529
the shortest lead-time and with high degrees of flexibility and quality. Two simple examples
have been treated to illustrate the possibilities of this new methodology.
To have a more detailed view of manufacturing complexity, more accurate
manufacturability indexes may be calculated, with other parameters involved and organized
with fuzzy logic, so further researches will be conducted to develop new manufacturability
indexes (based on material information and specifications). A study has to be done in order to
be able to compare different indexes between themselves.
Assembly constraints generated by a hybrid modular design have also to be taken account,
and the methodology will be applied an industrial die from automotive industry.

References
[1] Mognol P, Jégou L, Rivette M, Furet B. High Speed Milling, Electro Discharge
Machining and Direct Metal Laser Sintering: A method to optimize these processes in hybrid
rapid tooling. Intern. Journal of Advanced Manufacturing Technology 2006; 29(1-2):35-40.
[2] Mognol P, Hascoët J.Y, Rivette M, Kerbrat O. Evaluation of Hybrid Tooling with HSM-
EDM-DMLS: methods for comparison of time and cost using product examples. Euro-uRa-
pid2007, Frankfurt, Germany, December, 3-4 (2007) p. 179-188.
[3] Kuzman K. Nardin B. Determination of manufacturing technologies in mould
manufacturing. Journal of Materials Processing Technology 2004;157-157: 573-577.
[4] ElMaraghy W.H., Urbanic R.J. Modelling of Manufacturing Systems Complexity. CIRP
Annals – Manufacturing Technology 2003;52(1): 363-366.
[5] Pessard E., Mognol P. Hascoet J.Y. Gerometta C. Complex cast parts with rapid tooling:
rapid manufacturing point of view. International Journal of Advanced Manufacturing
Technology. Accepted 9 October 2007.
[6] Coldwell H., Woods R., Paul M., Koshy P., Dewes R. Aspinwall D. Rapid machining of
hardened AISI H13 and D2 moulds, dies and press tools. Journal of Materials Processing
Technology 2003;135: 301-311.
[7] Korosec M. Balic J. Kopac J. Neural network based manufacturability evaluation of free
form machining. International Journal of Machine Tools & Manufacture 2005;45:13-20.
[8] Dong J, Vijayant S. Manufacturing feature determination and extraction – Part I: optimal
volume segmentation. Computer-Aided Design 1997;29:427-440.
[9] Suh S.H, Lee B.E, Chung D.H, Cheon S.U. Architecture and implementation of a shop-
floor programming system for STEP-compliant CNC. Computer-Aided Design
2003;35:1069-1083.
[10] Tarng Y.S, Chang X.S. Dynamic simulation of milling operations. Computer-Aided
Design 1993;25(12):769-775.
[11] Kumar V., Dutta D. An assessment of data formats for layered manufacturing. Advances
in Engineering Software 1997;28:151-164.
[12] Lee S.K, Ko S.L. Development of simulation system for machining process using en-
hanced Z map model. Journal of Materials Processing Technology 2002;130-131:608-617.
[13] Kim J. NC verification using octree. M.Sc in Mechanical Engineering, MIT (1998).
[14] Ding S, Mannan M.A, Poo A.N. Oriented bounding box and octree based global in-
terference detection in 5-axis machining of free-form surfaces. Computer-Aided Design
2004;36:1281-1294.
[15] Szeliski. Real-time octree generation from rotating objects. Technical Report, Cambridge
Research Laboratory (1990).
[16] Wenger P, Chablat D. Definition sets for the Direct Kinematics of Parallel Manipulators.
8th International Conference in Advanced Robotics, Monterey, USA (1997).
[17] Medellin H, Corney J, Davies J.B.C, Lim T, Ritchie J.M. Algorithms for the physical
rendering and assembly of octree models. Computer-Aided Design 2006;38:69-85.

530