7/28/2019 a seminar topic http://slidepdf.com/reader/full/a-seminar-topic 1/18 Heterogeneous representation and fabrication using additive manufacturing
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 1/18
Heterogeneous representation and
fabrication using additive
manufacturing
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 2/18
1. Introduction
A heterogeneous object is referred to a solid component consisting of two or more
material primitives distributed continuously or discontinuously within an object. Modeling and
manufacturing of heterogeneous object (HO) have been paid much attention recently as the
advent of rapid prototyping manufacturing technology, which makes it possible to fabricate the
heterogeneous object. As the continuously variation of material composition produces gradient in
material properties, they are often known as functionally gradient materials (FGM), shown in
Fig. 1(a). For example, a component contains two compositions, metal and heat resistance
material (such as ceramic); the material distribution is illustrated in (b). From the figure we cansee that metal increases its fraction gradually from one side to another (the red line), while the
heat resistance material linearly reduces its fraction (the green line), which can avoid the stress
concentration because of the thermal stress relaxation in transition of two materials, shown in (c).
A discontinuous change in material composition generates distinct regions of material in the
solid, which is usually called multi-material object (MMO) such as composite materials, as
demonstrated in Fig. 2 (ZCorp (2005)). MMO has been extensively used in industry for a long
time, while FGM has shown tremendous potential in many fields, such as aeronautics and
astronautics, biomedical engineering, and nano-technology, etc.[4]
Fig. 1. Model of functionally gradient material.[4]
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 3/18
Solid modeling of objects forms an important task in design and manufacturing. Recent
developments in the field of layered manufacturing have shown potential for the physical
realization of heterogeneous (multi-material) objects. Thus, there is a need to represent material
information as an integral part of the CAD model data. Information models for the representation
of product data are being developed as an international standard informally called STEP (ISO
10303). However, the current application protocols focus on the representation of homogeneous
objects only.
Fig. 2. A multimaterials blade.[4]
In the past few decades, considerable attention has been devoted to FGM representations
(Computer-Aided Design), design validation (Computer-Aided Engineering), fabrication
(Computer-Aided Manufacturing) and material heterogeneity optimization. Recently, several
FGM modeling and representation methods have been reported. These methods have allowed
designers to design not only the geometry, but also the material composition of an object.
However, all of these methods have mainly focused on simple-shaped FGM objects with simple
gradient schemes. It is very difficult (or impossible in some cases) to use these methods to
process arbitrary-shaped FGM objects with authentic 3D gradients.
2. Heterogeneous solid modeling
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 4/18
Heterogeneous solid modeling aims to incorporate material distribution information
along with geometry into the CAD model. This section summarizes results reported in the
literature. The following are several possible representations of the macrostructure of the
material for modeling of heterogeneous objects. [3]
• r m object model,
• Tetrahedral decompositions,
• voxel-based representation,
• R-function method.
2.1 rm object model:
This model is based on the exact representation of geometry and material distribution
function. It forms the basis of the proposed representation to develop a standard for heterogeneous modeling.
The product space T = E3× _ R n forms the mathematical space to model heterogeneous
objects. Material points are restricted to lie in the material space V ⊂ R n. Each point p in the
object S is a combination of n primary materials and is specified by the volume fractions of these
primary materials. The material composition of any point p is represented as a material point v in
R 3, with each dimension representing exactly one primary material. As these volume fractions
must sum to unity, the space of material points (material space) is defined as:
V= {v ∈ R n | || v || 1 ≡ ∑ vi = 1 and vi ≥ 0} [3]
where vi represents the volume fraction of the ith primary material. Thus, each point p ∈ S can be
modeled as a point ( x ∈ E3 ; v ∈ V) in T, where x and v represent the geometric and material
points respectively.
Material r-set (rm set) An r m set is defined as a subset D ≡ ( P , B) of T where P ⊂ E3 is an r-set
and B ⊆ V assigns material to the r-set P .
1. The set B is specified by a material function F . Thus, an r m set can also be defined as the pair
( P , F ) where the subset B is defined implicitly through its material function as F ( P ).
2. An r m set is undefined for all the points lying in the exterior of P .
3. If F =1 & n = 1, then it is a single material r m set.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 5/18
4. To avoid the need for modeling discontinuities, it is assumed that the material function F is C ∞
continuous in P .
Material object (rm-object) an r m-object is then defined as a finite collection of r m-sets ( Pj, Bj)
such that the following conditions hold true:
1. The r m -sets are geometrically interior-disjoint.
2. The r m -sets are minimal.
2.1.1 Processing of Heterogeneous Solid Modeling for SFF:
There are three tasks that typically need to be performed before a design can be process-planned
and fabricated. These are:
i. Selection of a growth direction (or part orientation during fabrication).
ii. Construction of a support structure that enables overhang features in the design.
iii. Decomposition of the supported model into simple features suitable for automatic
tool-path (or deposition-path) generation.[3]
2.1.2 Extension of Modeling Entities:
In this section, we present the entities that are used to describe a decomposed design. These
entities (called SFF-Compacts and SFF-Objects) can be thought of as extensions or “derived
classes” of the heterogeneous modeling entities described earlier.
SFF-Compacts: SFF-Compacts are extended r m-sets that implement the above concept. In
addition to geometry and material information which is inherited from the r m-sets, compacts alsohave a unique growth direction, a build order field (which is used to assemble SFF-Compacts
into SFF-Objects), and a material precedence field (which is used to implement merging and
splitting algorithms that operate on SFF-Objects).[3]
SFF-Objects: SFF-Compacts do not have much significance in and of themselves, and are
partial object prescriptions. In order to represent a real object, they need to be assembled into a
collection called an SFF-Object. An SFF-Object can also be thought of as an extended r m-object,
where each member is an SFF-Compact, and whose minimality condition (C2) has been
dropped. Also, some constraints and dependencies are enforced between the geometry, material,
growth axis, build-order and precedence fields of all SFF-Compacts within a single SFF-Object.
Some of the important constraints are:[3]
i. The SFF-Compacts that constitute an SFF-Object partition the SFF-Object into a finite number
of pair-wise disjoint regions.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 6/18
ii. The closure of the interior of all the SFF-Compacts in an SFF-Objects is exactly equivalent to
the union of all the compacts (i.e., there are no extraneous “voids” within an object - internal
cavities are explicitly modeled with material type set to “air”, for example).
iii.All SFF-Compacts within an SFF-Object are similarly oriented (i.e. share a common growth
axis direction).
iv. An SFF-Object is fully supported (or encapsulated in support structure) and no two SFF-
Compacts with consecutive positions in the build order have the same material type (i.e. the
compacts are minimal as long as geometric and material restrictions on a solid are not violated).
v. The build order of SFF-Compacts in an SFF-Object is sequential, and decompositions that
enforce cyclic ordering are invalid (i.e. they need to be further decomposed until cyclicity is
destroyed). See Fig. 3.
Fig. 3. A decomposition that violates the cyclicity constraint[3]
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 7/18
2.1.3 Example
RM-OBJECT
RM-SET 1 RM-SET 2
{Wheel geometry, Epoxy} {Shaft geometry, Polyurethane}
Fig. 4. Heterogeneous Solid Model of a Multi-Material Part [3]
The object is modeled as an r m-object (HSM) with two r m-sets. The wheel is modeled as an r m-set,
the material being epoxy. The second r m-set models the shaft made of polyurethane. It is
processed further for fabrication. In Fig. 5, the support structure is generated, and the model is
decomposed into SFF-Compacts that form an SFF-Object. The compacts are ordered according
to their build order. This SFF-Object can be further processed by the deposition and tool path
planner for fabrication by a hybrid process like SDM.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 8/18
HETEROGENEOUS SOLID MODEL
SUPPORTEDMODEL
DECOMPOSED MODEL
Modeled as r m-objectWith three r m-sets
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 9/18
SFF-Object with 7 SFF-Compacts (in buildorder)
Fig 5. Process planning for the wheel/shaft example using Design Decomposition[3]
In Fig. 6, the decomposed model (SFF-Object) is shown sliced to generate planar layers
(traditional slicing) using an adaptive slicing algorithm . The SFF-Object has been decomposed
into layers which can be manufactured by any purely additive process (e.g. DMD, SLS).
Toolpaths for a sample slice is also shown.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 10/18
Fig. 6. Sliced model and a sample heterogeneous slice for the wheel shaft example[3]
2.2 Decomposition methods:
2.2.1 The tetrahedral model
In this representation a solid model created on a state-of-the-art CAD system is meshed
into finite elements (tetrahedra). The topology is maintained using the cell-tuple structure as a
graph of cells. Every cell is then associated with information about the composition and the
geometry. Material space is defined as M , spanning the dm materials available to the LM
machine. The material composition of the model is represented as a vector valued function m( x)
defined over the interior of the model. The designer specifies the overall variation in terms of
distance from a particular feature. This expression is used to obtain the volume fractions at the
vertices of each tetrahedron. The composition in the interior of the cell is then obtained in terms
of a set of control points and control compositions blended with barycentric Bernstein
polynomials. One of the major advantages of decomposing the model into tetrahedra is that the
model can directly be used for finite element analysis.
However, the following are the major drawbacks associated with this representation as
compared to the representation in Section 2.1:
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 11/18
1. Any modification in the material distribution function, m( x) will necessitate regeneration of
the mesh because the mesh generation depends on m( x).
2. The exact material distribution function is used to obtain the compositions at the vertices of
very tetrahedron. However, the composition at other points in the tetrahedron is calculated by
interpolation. The advantage of increase in the computational speed may be offset by the
approximation in representation.
3. The approximation in the shape due to meshing may lead to inaccurate dimensions and errors
in the required surface finish. Note that this approach approximates the object geometry and
material distribution function as well.
2.2.2 Voxel-based model
This is a special case of cell decomposition. The cell is cubical in shape and is located in
a fixed grid ( A voxel (x,y,z) in a 3D discrete space is defined by a unit cube centered at (x,y,z)).
Voxelization is the process of converting a geometrically represented 3D object into a voxel
model defined by a set of voxels. The voxelization is such that the voxel size is uniform and
every voxel is small enough to be considered as a homogeneous lump.
It is observed that the voxelization is independent of the material distribution function.
This representation provides to the designer a unique ability to selectively assign materials to
individual voxels. This is also better suited for fabrication using layered manufacturing as each
individual slice can be represented as a collection of voxels.The following constitute some of the limitations associated with this representation as
compared to the representation in Section 2.1:
1. This model also faces limitations like the one in Section 2.2.1 because of the approach of
decomposition.
2. Geometry-dependent function distributions are not easily applicable because of approximation
of the geometry in the voxel object.
3. This method of representation is not compatible for any kind of data transfer amongst CAD
systems. Research is being carried out to construct a solid model from voxel models
4. This technique does not seem to be suited for finite element analysis where tetrahedral
structures are preferred to cubical ones.
2.3 The R-Function approach:
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 12/18
An R-function (which should not be confused with an r-set) is a real-valued function
whose sign is completely determined by the signs of its arguments. Such functions provide
analogies to the Boolean logical functions. Simple examples are provided by min( x1, x2) and
max( x1, x2), which are analogous to Boolean ‘and’ and ‘or’, respectively, if we take + and -
values of the arguments to correspond to the logical values of TRUE and FALSE.
The analogy with Boolean functions allows any closed shape model in 2D or 3D,
expressed in terms of Boolean combinations of half-spaces, to be defined in terms of a single
implicit function, by composition of appropriate R-functions. It can be arranged for this function
to be positive inside the shape, and its value will be zero on the boundary. More generally, an
extension of the R-function approach allows the generation of functions on such a domain that
interpolate continuous distributions of function values or derivatives on its boundary. The
method may therefore be used to model distributions, e.g., of material properties, in the interior
of the shape.
Till now we have seen some of the representation schemes for heterogeneous objects
(functionally graded materials) .In the next session we are going to deal with their fabrication
methods. The RP technologies which are mainly used for making metal– metal or metal–ceramic
FGMare Selective Laser Melting, laser cladding-based techniques, Laminated Object
Manufacturing and Ultrasonic Consolidation (UC) while for fabricating polymer–polymer or
polymer–ceramic, ceramic–ceramic FGM, Selective Laser Sintering, Three Dimensional Printing
and Inkjet Printing have been mainly used. In this paper we are going to focus on two fabrication
techniques for functionally graded (heterogeneous) materials those are Freeze-form extrusion
process.
3. Freeze-form extrusion:
Freeze-form Extrusion Fabrication (FEF) is a novel, environmentally friendly, additive
manufacturing process that builds a 3D part layer-by-layer by computer controlled extrusion and
deposition of aqueous based colloidal pastes. Unlike most other extrusion freeform fabrication
processes, which use organic binders to bond the ceramic powders, the organic binder content is
only 2–4 vol% in this process, while the paste solids loading is 45–50 vol% or higher. Also,
unlike Robocasting which fabricates parts at room temperatures, FEF builds a ‘green’ (before
postprocessing) part in an environment below the freezing point of water to solidify the paste
after the deposition of each layer during the fabrication process. This enables relatively large
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 13/18
parts to be built compared with Robocasting. Cones and other monolithic ceramic components,
including alumina (Al2O3) and zirconium diboride (ZrB2) parts, have been fabricated using the
FEF process; for example, see Fig. 7.[1]
Fig. 7. Sintered ceramic cones fabricated using the FEF process for (a) Al2O3 and (b)
ZrB2.[1]
The part fabrication process involves computer control of flows of multiple aqueous pastes (each
controlled separately), the mixing of these pastes, and the extrusion of the mixed paste to
fabricate a 3D part layer-by-layer according to a CAD model with pre-specified material
compositions.
3.1. Process and system concept:
An FEF machine equipped with three servo controlled extruders and an inline static
mixing unit to merge the pastes through a single orifice has been designed and developed. This
mixing technique results in a natural transition between composition changes; however, it
introduces a transport delay. This delay must be repeatable and accurately predicted in order for
the path planning algorithm to deposit material in the desired location properly. The system
transport delay t is modeled using linear relationships between the paste volumetric flow rate Q
and the combined internal volume of each segment of the static mixer V:
t =V/ (A1v1+ A2v2 + A3v3) [1]
Ai is the cross sectional area of the i th cylinder, and vi is the velocity of the i th plunger. The
combined flow rate from all three extruders, Q, is equal to the sum of the individual flow rates,
Q1, Q2 and Q3. The ratio of Q1:Q2:Q3 represents the ratio of the three pastes in the material
composition.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 14/18
The three pastes are extruded simultaneously by a triple extruder mechanism, as
illustrated in Fig. 8. Continuous control over the material compositions and their gradients during
the part building process can be achieved by planning (with time delay taken into consideration)
and controlling the relative flow rates of the different pastes. As an example, assuming that the
three cylinders containing the three different pastes have the same cross-sectional area, a desired
paste mixture consisting of 20% paste A, 30% paste B, and 50% paste C can be achieved by
controlling the three plunger velocities with the ratios of v1:v2:v3 = 2:3:5, where v1, v2, and v3 are
the plunger velocities for pastes A, B, and C, respectively.[1]
Fig. 8. Triple-extruder mechanism design.[1]
3.2 System design and development:
The triple-extruder mechanism is designed using three stainless steel cylinders, each
containing a paste driven by an individual plunger whose movement is controlled by a DC servo
motor; see fig 9, The paste flow rate in each cylinder is controlled by the plunger’s velocity, and
the force exerted on the plunger is measured by a load cell. The FEF system uses a static mixer
to blend the three different pastes and mixes them into a homogeneous stream as they pass a
series of mixing blades positioned at alternating angles.[1]
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 15/18
Fig. 9. The triple-extruder FEF system in a temperature-controlled enclosure: three
Servo motors control linear cylinders for paste extrusion and a three-axis gantry
system controls nozzle movement.[1]
The triple-extruder mechanism is mounted on a gantry system, which consists of three
orthogonal linear drives, each with a 508 mm travel range. The X-axis consists of two parallel
slides and is used to support the Y-axis. The Z-axis is mounted on the Y-axis, and the extrusion
mechanism is mounted on the Z-axis. Four DC servo motors, each with a resolver for position
feedback at a resolution of 1000 counts per revolution, drive the various axes.
The part fabrication process is conducted in a freezing environment, which could becontrolled to as low as -20 oC using a liquid nitrogen injection system. This enables the aqueous
paste to solidify at temperatures below the freezing point of water after it is extruded to solidify
the paste, thus avoiding part deformation during the fabrication process and enabling fabrication
of larger parts. A heating jacket is used to keep the paste temperature above the freezing point of
water until it is deposited.
3.3 Test Results:
Control of grading between two pastes:
A cylindrical part with a 50 mm diameter was fabricated using two extruders filled with
limestone (CaCO3) pastes, one with a green color and the other with a pink color. Fig.10
demonstrates that a part can be built with desired material gradients by varying plunger velocities
for extrusion of different pastes. The color of the fabricated part starts at pink (A) and shifts to
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 16/18
brown (B), then green (C), then brown (D), then pink (E), and finally green (F). The color
distribution of the part is consistent with the velocity profiles of the two plungers.[1]
Fig. 10. Extrusion of pink and green colored CaCO3 pastes for tests with and without
mixing. [1]
4.0 Ultrasonic Consolidation
It is another important technique to fabricate heterogeneous objects. Out of all RP
techniques, UC gives an advantage (such as solid state bonding, low temperature processing and
surface texture retaining) to fabricate precisely a graded layer metal–metal FGM which has
potential to furnish a metallic product with pre-defined properties gradient. Foils are joined by
ultrasonic welding using a UC machine named Formation, manufactured by Solidica Inc., USA.
Formation is a hybrid additive layer manufacturing machine which consists of following main
parts: (1) a computer program to generate toolpaths for welding and cutting as per an STL file of
a 3D CAD, (2) a base plate and attached heater, (3) an automatic foil-feeding system to lay foils,
(4) an ultrasonic welding system to join foils, (5) a 3-axis CNC milling machine to shape and
trim joined foils.[2]
Process description:
The machine works by joining two foils (layers) using mechanical vibrations of a
welding head (sonotrode) at 20 kHz and removing the extra part of the foil/build by a milling
machine. For joining foils, a foil is laid on a base plate which could be maintained at a higher
temperature, if needed. Another foil is manually fixed on top of it as automatic feeding is used
only for aluminium foils. A rolling welding head, forced on the foils travels along the length of
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 17/18
the foils. During travel, the head vibrates ultrasonically with small amplitude normal to the
direction of the travel. This vibration plus normal force helps foils bond metallurgically. Fig.11
shows the principle of the bonding of foils in a UC machine. The machine is equipped with a
sonotrode of size 5.8’’ of titanium alloy; Foils are taken from the following materials: stainless
steel (SS) 316L, annealed, thickness 0.00400; Cu 110 annealed, 0.00500; Al 3003 H19, 0.00600
and Al 1100 annealed, 0.00200. Al 1100 foil is used only to facilitate joining of two SS foils by
placing it between the two foils. [2]
Fig. 11. Principle of the bonding of foils in a UC machine[2]
Summary:
In this paper we studied new approach of various representation schemes of heterogeneous
objects with an example of wheel and shaft. A Freeze-form Extrusion Fabrication (FEF) process
and ultrasonic consolidation process aimed at fabricating 3D parts with functionally graded
materials are presented in this paper.
References
1) Ming C. Leu , Bradley K. Deuser , Lie Tang , Robert G. Landers , Gregory E. Hilmas ,
Jeremy L. Watts, “Freeze-form extrusion fabrication of functionally graded materials”,
CIRP Annals - Manufacturing Technology,61, pp. 223–226, 2012.
7/28/2019 a seminar topic
http://slidepdf.com/reader/full/a-seminar-topic 18/18
2) S. Kumar, “Development of Functionally Graded Materials by Ultrasonic
Consolidation”, CIRP Journal of Manufacturing Science and Technology, 3, pp. 85–87,
2010.
3) Vinod Kumar, Sanjay Rajagopalan, Mark Cutkosky and Debasish Dutta, “Representation
and Processing of Heterogeneous Objects for Solid Freeform Fabrication”, IFIP WG5.2
Geometric Modelling Workshop, Tokyo, 1998.
4) Xiao J. Wu, Wei J. Liu and Michael Y. Wang, “Modeling Heterogeneous Objects in
CAD”, Computer-Aided Design & Applications, Vol. 4, No. 6, pp. 731-740, 2007.