a multi-material virtual prototyping system
TRANSCRIPT
A multi-material virtual prototyping system
S.H. Choi*, H.H. Cheung
Department of Industrial and Manufacturing Systems Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China
Received 9 January 2004; received in revised form 30 May 2004; accepted 1 June 2004
Abstract
This paper proposes a multi-material virtual prototyping system for digital fabrication of heterogeneous prototypes. It consists mainly of a
topological hierarchy-sorting algorithm for processing slice contours, and a virtual simulation system for visualisation and optimisation of
multi-material layered manufacturing (MMLM) processes. The topological hierarchy-sorting algorithm processes the hierarchy relationship
of complex slice contours. It builds a parent-and-child list that defines the containment relationship of the slice contours, and subsequently
arranges the contours in an appropriate sequence, which facilitates toolpath planning for MMLM by avoiding redundant tool movements.
The virtual simulation system simulates MMLM processes and provides stereoscopic visualisation of the resulting multi-material prototypes
for quality analysis and optimisation of the processes.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: Multi-material; Layered manufacturing; Virtual prototyping; Visualisation; Toolpath planning
1. Introduction
1.1. Layered manufacturing
Layered manufacturing (LM), also called rapid
prototyping (RP), is an additive manufacturing process
that produces a physical prototype from a 3D CAD model
layer-by-layer. The CAD model can be generated from many
sources, including CAD designs and conversion data from a
3D scanner. LM systems are widely adopted in
manufacturing and medical applications to save costs and
time.
Manufacturers use LM technology to produce prototypes
of products for design evaluation, and as master patterns for
production tools. Lopez and Wright [1] studied the impact
of LM on product design and development processes.
The study shows that LM may help drastically to shorten
product development cycles and enable designers to easily
incorporate mechanical and ergonomic features into new
products.
In medical field, surgeons use LM to plan and explain
complex surgical operations, especially craniofacial and
maxillofacial surgeries. D’Urso et al. [2] described the use
of anatomical biomodels fabricated using LM for education
of patients, diagnosis, surgical planning, and navigation.
Brown et al. [3] used life-sized wax stereolithographic
prototypes of a patient’s pelvis and an interpositioning
template for accurate positioning of the fixation plate.
Harrysson et al. [4] used stereolithography apparatus to
build physical biomodels from computed tomography (CT)
data for planning and rehearsal of complex osteotomies of
severely deformed bones.
Some researchers have explored LM for direct
manufacture of biologically active implants. Das et al. [5]
used micro-computed tomography (micro-CT) technique
and selective laser sintering to fabricate Nylon-6 tissue
engineering scaffolds for human proximal femur trabecular
bones. They attempted to develop a multi-material
deposition system to fabricate tissue engineering scaffolds
with functionally graded material composition.
Currently, commercial LM machines can only produce
single-material prototypes [6,7]. However, there is an
increasing need for multi-material prototypes [8]. This is
because high value-added products often involve advanced
and complex designs, while medical operations are
becoming more complicated and delicate. A typical
example is tissue engineering scaffolds that have to be
tailor-made to fit the individual requirements of each patient
[9]. Indeed, a multi-material prototype that can clearly
differentiate one part from another of a product, or tissues
0010-4485//$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cad.2004.06.002
Computer-Aided Design 37 (2005) 123–136
www.elsevier.com/locate/cad
* Corresponding author. Tel.: þ852-2859-7054; fax: þ852-2858-6535.
E-mail address: [email protected] (S.H. Choi).
from blood vessels or bone structure of a human organ,
will be particularly useful for designers or surgeons,
respectively. Therefore, it would be very desirable and
useful to develop multi-material layered manufacturing
(MMLM) technology.
In general, a heterogeneous object may be composed
of either several materials with varying composition or a
collection of discrete materials [10,11]. Effective
representation of multi-material objects is a vital step
in the development of MMLM technology. Pratt et al.
[12] noted that there is an increasing need to transfer
additional types of information for development of LM
technology, especially material information. However,
a critical problem is that the current CAD models do not
contain any material information. Consequently, most
LM systems treat a model as homogeneous during
down-line process planning [13,14].
Although MMLM technology has attracted much
research interest, practical and viable MMLM machines
have yet to be developed. A MMLM machine may
include mainly two sub-systems, namely a hardware
material-depositing mechanism and a computer control
software system that can process multi-material slice
contours to control the material-depositing mechanism,
as shown in Fig. 1.
The development of MMLM technology is largely a
software issue. The material-depositing mechanism may
consist of an array of nozzles or tools, each of which
deposits a type of material on specific areas of a layer.
The tools may be bundled together or controlled to move
independently. They may be adapted from those used in the
existing single-material LM machines. However, a viable
computer software system for MMLM processes has yet to
be developed. The system should be able to generate
multi-material slice contours, which are subsequently sorted
to relate specific contours to a particular tool, such that the
area is appropriately deposited with the desired material.
Hence, processing of complex multi-material slice contours
and planning of multi-toolpaths are particularly important
issues. They facilitate control of tools to deposit selected
materials at the appropriate contours, and at the same time
to avoid redundant movements and potential collisions of
the tools which move otherwise independently to increase
efficiency.
This is, however, a daunting task because slice contours
are generally random in nature with no explicit topological
hierarchy relationship. As a result, it is very difficult to
identify one contour from another within a slice as required
for multi-toolpath planning. Therefore, it can be seen that
the difficulty of processing slice contours and the lack of
suitable control software largely hinder the development of
MMLM technology.
1.2. Virtual prototyping
Virtual prototyping (VP) is a process of using a digital
prototype, in lieu of a physical one, for testing and
evaluation of specific characteristics of a product or a
manufacturing process. It is often carried out in a virtual
reality (VR) environment, which provides 3D stereoscopic
simulation with innovative input and output. The purpose of
VP is to reduce the iterations of design-prototype-test
cycles. Integration of VP and LM allows a designer to
analyse and visualise the influence of process parameters on
the part quality. Once a VP process is finished, the resulting
digital prototype can be sent via the Internet to customers to
solicit comments, or the process parameters can be used for
optimal fabrication of the physical prototype. This reduces
the number of physical iterations and thereby the
associated manufacturing overheads, leading to faster and
cost-effective product development. Since digital prototypes
are mostly used in VP, there is no need to worry about the
costs and the quality of physical prototypes. However,
few researchers have exploited the strengths of VR to
facilitate LM. Gibson et al. [15] investigated the
contributions of VP and LM towards product development.
They suggested the use of VR as a complementary tool to
LM. Morvan and Fadel [16] linked VR with LM to visualize
the support structures of a part and to help a designer
identify improper support structures. Jee and Sachs [17]
Fig. 1. A layer of a multi-material part being fabricated.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136124
developed a visual simulation system for 3D printing.
However, their system was aimed at providing a visual tool
to examine surface texture only.
The authors have recently developed a VP system for
digital fabrication and visualisation of product prototypes in
a virtual environment [18]. The system allows design
validation and parameter optimisation of a LM process prior
to physical fabrication. By providing 3D stereoscopic
visualisation of the LM process and the resulting digital
prototype, a designer can analyse its quality more efficiently
and take prompt action to improve both the LM process and
the product design. Currently, this system can only simulate
single-material LM processes. As there is now an increasing
need for multi-material prototypes, it is natural and timely to
enhance this system for MMLM processes. The paper,
therefore, presents a multi-material virtual prototyping
(MMVP) system for digital fabrication of heterogeneous
prototypes. The MMVP system may be subsequently
adapted and integrated with a material deposition
mechanism to form a practical MMLM machine.
2. Review of the related works
The benefits of MMLM technology have been widely
recognised. Researchers have recently started working on
MMLM and they have made important contributions.
Their work has been categorically focused on a few areas,
particularly CAD representation of heterogeneous objects
for MMLM, development of experimental systems for
simple objects, and virtual simulation of MMLM processes.
2.1. CAD representation of heterogeneous objects
for MMLM
To fabricate multi-material prototypes with MMLM
technology, both geometric and material information must
be made available. Although STL is now a de facto
industrial standard file format for LM, it contains only
geometric information. Researchers have recently proposed
CAD formats for heterogeneous objects [10,19]. Chiu and
Tan [20] proposed a modified STL format in which a tree
structure was used to represent materials. Pratt [21]
conducted a survey on several methods, including voxel-
based modelling, mesh-based decomposition, and set-based
heterogeneous solid modelling approach, for representation
of material macrostructure. Patil et al. [14] reported the
possible use of R-function for representation of material
composition, and they proposed the rm-object model for
representation of heterogeneous solids. Biswas et al. [22]
proposed a representation scheme for fields defined over
geometric domains. Such proposed formats, when fully
developed and widely adopted, will be useful for MMLM.
However, there are still some major problems to solve.
Indeed, a significant proportion of complex objects,
particularly human organs and bone structures, etc. are not
designed using CAD systems. Instead, they are captured by
laser digitisers, or CT and magnetic resonance imaging
scanners. The digitised images are normally processed to
form a model in STL format or to extract slice contours,
which are random in nature without any explicit topological
hierarchy relationship. Hence, processing of such slice
contours for multi-toolpath planning remains a challenging
obstacle that has yet to be surmounted for the development
of MMLM.
2.2. Development of experimental MMLM machines
for simple objects
A few researchers have recently attempted to develop
experimental MMLM machines for simple objects. Weiss
et al. [23] described building multi-material structures with
pre-fabricated parts, such as advanced tooling and
embedded electronic devices. Qiu and Langrana [11]
reported the development of a MMLM machine for
fabrication of multi-phase electromechanical components
for naval and other military applications. Simple hollowed
circular objects were fabricated. Bondi et al. [24] developed
a laser chemical vapor deposition system, which was used
primarily to deposit carbon to form small 3D and laminated
structures. It could be extended to fabricate multi-material
objects if advanced control software for the system was
available. These systems were among the pioneering
work, although they could only handle relatively simple or
small parts.
2.3. Virtual simulation of MMLM
Cheng and Langrana [13] and Qiu et al. [6] described
a virtual simulator used to simulate fabrication of
multi-material prototypes. Using computer graphics
technology, the simulator could detect and remove errors
of MMLM processes quickly. Zajtchuk and Satava [30]
used VR technology for medical applications. At the US
Naval Research Laboratory (NRL), VR visualisation was
used in a wide range of projects [25]. Although the
applications are focused on relatively simple objects, they
had highlighted the usefulness of virtual simulation of
MMLM.
3. The proposed multi-material virtual prototyping
system
The above pioneering work has made significant
contributions to the development of MMLM technology.
However, there is a lack of integrated effort to tackle the
core software problems of slice contour processing and
multi-toolpath planning for fabrication of complex objects.
Therefore, a multi-material virtual prototyping (MMVP)
system is proposed to fabricate digital multi-material
prototypes. The system consists mainly of a topological
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 125
hierarchy-sorting algorithm for processing slice contours,
and a virtual simulation system for stereoscopic
visualisation and optimisation of MMLM processes.
The topological hierarchy-sorting algorithm constructs the
hierarchy relationship of complex slice contours for
subsequent fabrication of multi-material prototypes [26].
In particular, the explicit hierarchy relationship
facilitates optimisation of toolpaths by avoiding redundant
tool movements. The virtual simulation system uses a
dexel-based approach for digital fabrication of prototypes
[18,27]. A dexel is a hatch vector representing the path that a
tool has to follow within a contour to build a portion of a
layer. By building a volume of a specific height and
width around a dexel, a strip of material may be represented.
Hence, during a digital fabrication process, rectangular
solid strips are laid to form a layer, which is subsequently
stacked up to form a prototype.
Fig. 2 shows the flow of the proposed system. It includes
colouring and slicing the STL model of a multi-material
object, processing and sorting of slice contours,
Fig. 2. Flow of the proposed MMVP system.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136126
generation of multi-toolpaths, and digital fabrication of
multi-material prototypes.
3.1. Representation of multi-material objects
by colour STL models
To build a multi-material prototype, both geometric
and material information should be provided. In the
proposed system, colour STL models are used to
represent multi-material objects. A colour STL model
takes advantage of the two pad bytes that are spared in
each facet in the original monochrome STL format to
store colour indices. The two pad bytes are divided into
16 bits. The first bit indicates whether or not the model
is monochrome. If it is, then no colour will be assigned
to individual facets. Otherwise, each facet will carry a
colour represented by the blue, green, and red indices
stored in the remaining 15 bits. The first 5 bits store the
blue index; the next 5 bits store the green index; the last
5 bits store the red index. Since each colour index has 32
grey levels, ranging from 0 to 31, a digital prototype
may have a maximum of 323 ¼ 32,768 colours, each of
which represents a type of material. Thus, a colour STL
model contains both geometric and material information.
A software package for colouring monochrome STL
models to associate material information has been
developed as an integral part of the proposed system.
Commercially available software systems, such as
Magics RP [28] and FlashTL [29] can also be used to
paint colour STL models. When a colour STL model is
completed, it is sliced to generate multi-material slice
contours that are outputted in a modified common
layered interface file format [26]. A colour representing
a particular type of material is assigned to the related
layer contours.
3.2. Multi-toolpath planning using topological hierarchy
relationship
To fabricate a layer of a multi-material prototype, a set of
tools or nozzles are controlled to deposit specific materials
on the appropriate slice contours. For each tool to
effectively and efficiently deposit a specific material on
the related contours, it is necessary to: (i) identify and relate
specific contours of a slice to a particular tool; and
(ii) arrange the toolpaths of each tool in an appropriate
sequence to shorten build time and to avoid possible
collisions. Therefore, multi-toolpath planning is particularly
important for the subsequent control of tools to deposit
selected materials at the appropriate contours, and at the
same time to detect and avoid possible collisions of the tools,
which move otherwise independently to increase efficiency.
However, the geometry of most STL models tend to be very
complex, and the slice contours do not possess any explicit
topological hierarchy relationship. As a result, it is very
difficult to associate specific contours with a particular tool.
Hence, it is very important to establish the hierarchy
relationship of slice contours. Therefore, this paper proposes
a virtual system for MMLM, which uses a topological
hierarchy-sorting algorithm to process complex slice
contours for multi-toolpath planning. The algorithm first
establishes the hierarchy relationship representing the
connectivity of solid contours and the related cavities,
and then arranges the contours in an appropriate sequence
for subsequent planning of multi-toolpaths. The planning
process is based on two major issues.
1. Identification of slice contours for association with the
related tools.
2. Sequential control of tools to avoid redundant
movements and possible collisions.
A layer of a mechanical gearbox is shown in Fig. 3a.
It has nine contours of four kinds of materials. Four nozzles,
N1 to N4, are assigned to deposit blue, red, yellow, and green
materials on the related contours, respectively. Originally,
the contours are random with no explicit topological
hierarchy relationship. Toolpath planning based on random
contours will result in a considerable amount of redundant
tool movements, as shown in Fig. 3b. In order to optimise
the toolpaths, the contours of the same material should be
appropriately arranged such that the toolpaths will be the
shortest possible. This requires establishment of topological
hierarchy relationship of the slice contours. To do so,
the proposed algorithm first builds a parent-and-child list
that defines the containment relationship of the slice
contours, and subsequently arranges the contours in an
appropriate sequence.
Referring to Fig. 3c, the contours are sorted into five
hierarchy levels. The level number of a contour indicates its
number of parents. For instance, C1, C2, C3, C4, and C6
are Level 0 contours; they have no parent and are therefore
not contained in any other contours. Similarly, a Level 1
contour has one and only one parent, although a parent
can have multiple children. Hence, according to the
parent-and-child list, there are five contour entities, namely
C1, C2, C3, C4, and C6 ! C7 ! C8 ! C5 ! C9. Based on
the relationship of external (solid) and internal (void)
contours, the contour entities can be further grouped into
seven contour families, namely C1 (Family 1), C2 (Family
2), C3 (Family 3), C4 (Family 4), C6 ! C7 (Family 5),
C8 ! C5 (Family 7), and C9 (Family 9). Each contour
family has a specific material property.
Thus, with the hierarchy relationship, it is no longer
necessary to identify and relate contours to a particular
nozzle one-by-one for multi-toolpath planning. Indeed,
only grouping of the outermost contours is required.
For instance, the seven contour families of the gearbox
layer can be treated as seven independent inputs to generate
hatch vectors that define the multi-toolpaths. Hence, seven
toolpaths (PC1, PC2, PC3, PC4, PC6,7, PC8,5, and PC9) can be
generated accordingly. However, these seven toolpaths
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 127
should be further arranged in accordance with the specific
geometric and material information of each contour family,
in order to avoid redundant tool movements. In other words,
each nozzle should deposit the required material on all
related contours before returning to its datum position.
Referring to Fig. 3c and d, the seven toolpaths can be easily
grouped. PC1, PC2, and PC3 are grouped into a toolpath-set
{S1 (PC1 ! PC2 ! PC3)}, which is associated with nozzle
N1 because contour families 1, 2, and 3 are of blue material
property. In the same way, PC6,7 and PC8,5 are grouped into a
toolpath-set {S3 (PC6,7 ! PC8,5)} for association with
nozzle N3. Finally, PC4 and PC9 are grouped into
Fig. 3. (a) A multi-material slice of a gearbox with multiple-inclusion contours. (b) Multi-toolpaths with redundant tool movements based on unsorted slice
contours. (c) Topological hierarchy relationship of the gearbox slice contours in (a). (d) Multi-toolpath planning using topological hierarchy relationship.
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136128
{S2 (PC4)} and {S4 (PC9)}, which are associated with
nozzles N2 and N4, respectively, as shown in Fig. 3d.
This process considerably simplifies the association of
toolpaths with the appropriate nozzles.
However, the toolpaths in each toolpath-set are still
random, because the contours at the same hierarchy level
are random in nature. This may cause redundant tool
movements when the associated nozzle moves from the exit
point of one toolpath to the entry point of another.
Therefore, based on the bounding-box criterion,
the algorithm further sorts the contours at the same
hierarchy level into an appropriate sequence. Contours
with the same level number are arranged according to their
minimum and maximum ðx; yÞ values. The sorted sequence
facilitates optimisation of multi-toolpaths by avoiding
redundant movements. For example, the three toolpaths in
the toolpath-set {S1 (PC1 ! PC2 ! PC3)} are generated for
contours C1, C2, and C3, respectively. They are all at the
same level (Level 0) and the sequence is sorted as
C1 ! C3 ! C2. Accordingly, the toolpaths in S1 are
arranged and linked as {S1 (PC1 ! PC3 ! PC2)}. If the
contours in a toolpath-set are at different hierarchy levels,
the contours at the highest level (with the smallest level
number) are deposited first. In other words, material
deposition starts inwards from the outermost contours.
For example, the two contour families (C6 ! C7) and
(C8 ! C5) in the toolpath-set S3 are at different hierarchy
levels. The contour family (C6 ! C7) is deposited before
(C8 ! C5) because the level of (C6 ! C7) is higher than that
of (C8 ! C5). Hence, the toolpaths in S3 are arranged and
linked as {S3 (PC6,7 ! PC8,5)}. Similarly, the toolpaths in
the remaining toolpath-sets are arranged and linked
as shown in Fig. 3d. According to the hierarchy levels, the
toolpath-sets are subsequently arranged in a sequence of
S1 ! S2 ! S3 ! S4; the movement sequence of the nozzles
is given as N1 ! N2 ! N3 ! N4. Therefore, practical
sequential multi-toolpaths for MMLM can be easily
generated. Fig. 4a shows, in comparison with the
unsorted toolpaths, how these multi-toolpaths can reduce
build time by avoiding redundant tool movements.
Digital fabrication of the slice contours in Fig. 3a is
shown in Fig. 4b. As slice contours become more
complicated, time saving from avoiding redundant tool
movements will become more substantial. Thus, the
importance of topological hierarchy relationship for
multi-toolpath planning is prominent. For an ideal case,
concurrent toolpaths should be generated, which will be
studied in the future.
Fig. 3 (continued )
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 129
3.3. Digital fabrication of multi-material prototypes
After processing slice contours and multi-toolpath
planning, the proposed system simulates a MMLM process
to fabricate a digital multi-material prototype.
Subsequently, it provides vivid stereoscopic visualisation
of the resulting prototype for quality analysis and
optimisation of the MMLM process. The designer can
Fig. 4. (a) Sequential multi-toolpaths without redundant tool movements. (b) Digital fabrication of gearbox layer using sequential multi-toolpaths.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136130
manipulate the prototype using the utilities provided to
visualise the quality of the product prototype that a MMLM
machine will subsequently deliver. The designer can also
navigate around the internal and opaque structures of the
prototype to investigate the product design. Furthermore,
the digital prototype may be superimposed on its STL model
to highlight dimensional deviations. The system also
calculates the maximum and the average cusp heights that
indicate the overall accuracy of the prototype. To study the
dimensional errors, a tolerance limit may be set and any
locations with deviations beyond the limit will be clearly
highlighted. The designer may thus identify and focus on the
parts that need modifications. To improve the accuracy and
the surface quality of specific features of the prototype,
the process parameters, such as orientation of the model,
Fig. 6. A layer of complex slice contours with topological hierarchy relationship.
Fig. 5. Monochrome and colour STL models of the ring. (For interpretation
of the references to colour in this figure legend, the reader is referred to the
web version of this article.)
Fig. 7. Nozzles move sequentially to deposit materials on related contours.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 131
the layer thickness, or the hatch space, may be changed
accordingly.
4. Case studies
4.1. A jewellery product
Hong Kong is a leading exporter of jewellery products.
As these products get more fashion-oriented, innovative
designs become more important. Some jewellery
manufacturers have realised that they should not only rely
on individual skills; they have recently adopted LM
technology to develop high value-added jewellery items.
Hence, the proposed MMVP system will be particularly
useful for building digital multi-material prototypes to help
improve designs and shorten development cycles of
jewellery products at competitive costs. Designers can
view and evaluate their designs with digital prototypes,
instead of physical ones, at minimal costs and time possible.
Fig. 8. Digital fabrication process of a multi-material prototype of the ring.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136132
Furthermore, digital prototypes may be conveniently
transmitted over the Internet to facilitate global
manufacturing.
A ring in Fig. 5 was therefore chosen as an example to
illustrate how the proposed system fabricates digital
multi-material prototypes of jewellery products. The ring
is made of 13 kinds of materials, and its shape is complex.
To produce such multi-material prototypes, an array of
nozzles may be used to extrude materials on the relevant
slice contours in a layer. The topological hierarchy-sorting
algorithm constructs the hierarchy relationship of complex
slice contours for optimisation of multi-toolpaths by
avoiding redundant nozzle movements. The case study
illustrates how the system generates efficient sequential
toolpaths of each nozzle, and subsequently facilitates
quality analysis and optimisation of MMLM processes.
Fig. 9. Ring prototype superimposed on STL model to highlight dimensional deviations. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
Fig. 10. Colour STL model of a pelvis. (For interpretation of the references
to colour in this figure legend, the reader is referred to the web version of
this article.)
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 133
Based on a monochrome STL model of the ring in Fig. 5a,
a colour STL model in Fig. 5b is created by assigning
colours to represent the appropriate materials of the ring.
The colour STL model is sliced to generate slice contours
that contain both geometric and material information.
The layers are relatively rough for clarification purpose.
Fig. 6 shows a layer of the ring. There are totally 52 contours
that are sorted into four hierarchy levels and are grouped
into 34 contour families. Each family has its own material
property. The layer consists of 11 discrete materials.
The contour families with the same material property are
grouped into a set. The 34 contour families are grouped into
11 sets because the layer is made of 11 kinds of materials.
For each set of contour families, toolpaths are generated and
then arranged in an appropriate sequence. Each toolpath-set
is used to control a specific nozzle that extrudes a material
on the related set of contour families. Fig. 7 shows a layer of
the digital fabrication process of the ring prototype.
Each nozzle moves sequentially and deposits a material
on the related set of contour families in the layer by avoiding
redundant movements. More details of the digital
fabrication process of the ring are shown in Fig. 8.
The system also provides stereoscopic visualisation of the
resulting prototype for quality analysis and optimisation of
the MMLM process. Fig. 9 shows the ring prototype being
superimposed on its STL model. The surface texture and the
dimensional deviations are clearly illustrated. In addition, the
system calculates the cusp heights to evaluate the overall
dimensional deviations. In this case, the average and
the maximum cusp heights are 0.097 and 0.176 mm,
Fig. 11. (a) Superimposition of pelvis prototype on its STL model. (b) Pelvis prototype highlighting dimensional deviations beyond accuracy limit.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136134
respectively. Suppose that any deviations more than
0.175 mm are considered unacceptable, the designer may
choose to highlight the areas which are out of the design limit
for subsequent investigation of these important features. The
excessive deviations are highlighted with red or green pins.
The red pins point to the maximum deviations whereas the
green ones point to unacceptable deviations. If unsatisfactory
deviations are located at important parts of the model, the
designer may choose either to change the model orientation
to shift the deviations or to reduce the layer thickness and the
hatch space to improve the accuracy.
4.2. A pelvis
Multi-material prototypes are particularly useful for
visualisation and identification of complex bone structures
to facilitate study and planning of surgical operations.
Therefore, a pelvis model, as shown in Fig. 10, was chosen
to demonstrate possible applications of the MMVP system in
the medical field. For a complex surgery, a patient’s pelvis
was scanned to produce a prototype so that surgeons can
study the bone fractures or injury more clearly. A prothesis or
an inter-positioning template may have to be put into the
patient’s pelvis. A multi-material digital prototype of the
pelvis will be helpful for surgeons to study the bone
structures more clearly. Also, visualisation of the pelvis
prototype and its superimposition on the STL model, as
shown in Fig. 11a, will assist surgeons to select a prothesis
that will fit best to minimise mismatch. It can be seen that the
excessive material is evenly distributed, and the average and
maximum cusp heights are 1.181 and 2.689 mm,
respectively.
Fig. 11b shows the pelvis prototype highlighting devi-
ations over a chosen accuracy limit. Based on the deviation
analysis, surgeons can adjust the process parameters, such as
orientation and layer thickness, to enhance the accuracy of
the target regions. If the prototype is deemed satisfactory,
physical fabrication can be carried out using the process
parameters accordingly. In general, it may not be necessary
to aim at producing a perfect prototype. Instead, an optimal
prototype with good accuracy in target regions will be more
practical and economical.
5. Conclusion
This paper proposes a MMVP system that processes
complex slice contours for planning and validation of
toolpaths for subsequent control of nozzles or tools to
fabricate heterogeneous prototypes. Indeed, multi-toolpath
planning plays a significant role in the development of
practical MMLM machines. The proposed system uses a
topological hierarchy-sorting algorithm to process the
hierarchy relationship of complex slice contours. It builds
a parent-and-child list, and subsequently arranges the
contours in an appropriate sequence. With the topological
hierarchy relationship, sequential multi-toolpaths can be
planned to control nozzles or tools to deposit selected
materials on the appropriate contours by avoiding redundant
movements. Besides, the proposed system simulates a
MMLM process to fabricate digital multi-material proto-
types. Subsequently, it provides vivid stereoscopic
visualisation of the resulting prototypes for quality analysis
and optimisation of the MMLM process. The system can be
adapted and integrated with a material deposition mechan-
ism to form a practical MMLM machine.
Acknowledgements
The authors would like to acknowledge the Research
Grant Council of the Hong Kong SAR Government and the
CRCG of the University of Hong Kong for their financial
support for this project. Thanks are also due to Dr Panos
Diamantopoulos of the University of Sussex for permission
to use the monochrome pelvis model.
References
[1] Lopez SM, Wright PK. The role of rapid prototyping in the
product development process: a case study on the ergonomic
factors of handheld video games. Rapid Prototyping J 2002;8(2):
116–25.
[2] D’Urso PS, Thompson RG, Atkinson RL, Weidmann MJ, Redmod
MJ, Hall BI, Jeavons JS, Benson MD, Earwaker WJS. Cerebrovas-
cular biomodelling: a technical note. Surg Neurol 1999;52(5):
490–500.
[3] Brown AG, Milner B, Firoozbakhsh K. Application of computer-
generated stereolithography and interpositioning template in acet-
abular fractures: a report of eight cases. J Orthop Trauma 2002;16(5):
347–52.
[4] Harrysson OLA, Cormier DR, Marcellin-Little DJ, Jajal K. Rapid
prototyping for treatment of canine limb deformities. Rapid
Prototyping J 2003;9(1):37–42.
[5] Das S, Hollister SJ, Flanagan C, Adewunmi A, Bark K, Chen C,
Ramaswamy K, Rose D, Widjaja E. Freeform fabrication of
Nylon-6 tissue engineering scaffolds. Rapid Prototyping J 2003;
9(1):43–9.
[6] Qiu D, Langrana NA, Danforth SC, Safari A, Fafari M. Intelligent
toolpath for extrusion-based LM process. Rapid Prototyping J 2001;
7(1):18–24.
[7] Zhu WM, Yu KM. Tool path generation of multi-material assembly
for rapid manufacture. Rapid Prototyping J 2002;8(5):277–83.
[8] Shin KH, Natu H, Dutta D, Mazumder J. A method for the design
and fabrication of heterogeneous objects. Mater Des 2003;24(5):
339–53.
[9] Landers R, Pfister A, Hubner U, John H, Schmelzeisen R, Mulhaupt R.
Fabrication of soft tissue engineering scaffolds by means of rapid
prototyping techniques. J Mater Sci 2002;37(15):3107–16.
[10] Kumar V, Rajagopalan S, Cutkosky M, Dutta D. Representation and
processing heterogeneous objects for solid freeform fabrication. Sixth
IFIP WG 5.2 International Workshop on Geometric Modelling:
Fundamentals and Applications, Tokyo, Japan, The University of
Tokyo; 1998.
[11] Qiu D, Langrana NA. Void eliminating toolpath for extrusion-based
multi-material layered manufacturing. Rapid Prototyping J 2002;8(1):
38–45.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136 135
[12] Pratt MJ, Bhatt AD, Dutta D, Lyons KW, Patil L, Sriram RD. Progress
towards an international standard for data transfer in rapid prototyping
and layered manufacturing. Comput-Aided Des 2002;34(14):
1111–21.
[13] Cheng TH, Langrana NA. A system approach in extrusion-based
multi-material CAD. In: Bourell DL, et al., editors. Solid Freeform
Fabrication Symposium, Austin, Texas. The University of Texas;
2001. p. 313–21.
[14] Patil L, Dutta D, Bhatt AD, Jurrens K, Lyons K, Pratt MJ, Sriram RD.
A proposed standard-based approach for representing heterogeneous
objects for layered manufacturing. Rapid Prototyping J 2002;8(3):
134–46.
[15] Gibson I, Brown D, Cobb S, Eastgate R. Virtual reality and rapid
prototyping. In: Warwick K, et al., editors. IEE Workshop on Virtual
Reality in Engineering, London. 1993. p. 51–63.
[16] Morvan SM, Fadel GM. IVCES, interactively correction.STL files in a
virtual environment. In: Bourell DL, et al., editors. Solid Freeform
Fabrication Symposium, Austin, Texas. The University of Texas;
1996. p. 491–8.
[17] Jee HJ, Sachs E. Visually simulated surface texture models for 3D
printing. In: Campbell RI, editor. Seventh European Conference on
Rapid Prototyping and Manufacturing, Aachen, Germany.
The University of Nottingham; 1998. p. 49–62.
[18] Choi SH, Chan AMM. A virtual prototyping system for rapid product
development. Comput-Aided Des 2004;36(5):401–12.
[19] Morvan SM, Fadel GM. Heterogeneous solids: possible
representation schemes. In: Bourell DL, editor. Solid Freeform
Fabrication Symposium, Austin, Texas. The University of Texas;
1999. p. 187–98.
[20] Chiu WK, Tan ST. Multiple material objects: from CAD represen-
tation to data format for rapid prototyping. Comput-Aided Des 2000;
32(12):707–17.
[21] Pratt MJ. Modelling of material property variation for layered
manufacturing. In: Cipolla R, Martin R, editors. The mathematics of
surfaces IX. London: Springer; 2000. p. 486–500.
[22] Biswas A, Shapiro V, Tsukanov I. Heterogeneous material
modeling with distance fields. Comput Aided Geometr Des 2004;
21(3):215–42.
[23] Weiss L, Merz R, Prinz FB, Neplotnik G, Padmanabhan P, Schultz L,
Ramaswami K. Shape deposition manufacturing of heterogeneous
structures. J Manufact Syst 1997;16(4):239–48.
[24] Bondi SN, Johnson RW, Elkhatib T, Gillespie J, Mi J, Lackey WJ.
Multi-material and advanced geometry deposition via laser chemical
vapor deposition. Rapid Prototyping J 2003;9(1):14–18.
[25] Lanzagorta M, Rosenberg R, Rosenblum LJ, Kuo EY. Rapid prototyping
of virtual environments. Comput Sci Eng 2000;2(3):68–73.
[26] Choi SH, Kwok KT. Hierarchical slice contours for layered-
manufacturing. Comput Ind 2002;48(3):219–39.
[27] Choi SH, Samavedam S. Visualisation of rapid prototyping. Rapid
Prototyping J 2001;7(2):99–114.
[28] Materialise. http://www.materialise.com/magics-rp/main_ENG.html;
2003.
[29] TNO. http://www.ind.tno.nl; 2003.
[30] Zajtchuk R, Satava R. Medical applications of virtual reality.
Commun ACM 1997;40(9):63–4.
S.H. Choi is an associate professor in the
IMSE Department at the University of Hong
Kong. He obtained both his BSc and PhD
degrees at the University of Birmingham. He
worked in computer industry as CADCAM
consultant before joining the University of
Hong Kong. His current research
interests include CADCAM, advanced manu-
facturing systems and virtual prototyping
technology.
H.H. Cheung gained his B.Eng. degree from
the IMSE Department at the University of
Hong Kong. He continued his postgraduate
research study in the Department, and
his research interest is in virtual prototyping
technology.
S.H. Choi, H.H. Cheung / Computer-Aided Design 37 (2005) 123–136136