virtual and physical prototyping
TRANSCRIPT
-
8/13/2019 Virtual and Physical Prototyping
1/16
This article was downloaded by: [North Dakota State University]On: 17 August 2013, At: 22:59Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House37-41 Mortimer Street, London W1T 3JH, UK
Virtual and Physical PrototypingPublication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/nvpp20
Fabrication of lightweight structural panels through
ultrasonic consolidationJ. George
a& B. Stucker
a
aDepartment of Mechanical Engineering, Utah State University, Logan, Utah, 84322-4130,
USA
Published online: 16 Feb 2007.
To cite this article:J. George & B. Stucker (2006) Fabrication of lightweight structural panels through ultrasonic
consolidation, Virtual and Physical Prototyping, 1:4, 227-241, DOI: 10.1080/17452750601106799
To link to this article: http://dx.doi.org/10.1080/17452750601106799
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of thContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon ashould be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveor howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
http://www.tandfonline.com/action/showCitFormats?doi=10.1080/17452750601106799http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditionshttp://dx.doi.org/10.1080/17452750601106799http://www.tandfonline.com/action/showCitFormats?doi=10.1080/17452750601106799http://www.tandfonline.com/loi/nvpp20 -
8/13/2019 Virtual and Physical Prototyping
2/16
Fabrication of lightweight structural panels through ultrasonicconsolidation
J. GEORGE and B. STUCKER*
Department of Mechanical Engineering, Utah State University, Logan, Utah, 84322-4130, USA
A fundamental investigation of the feasibility of producing lightweight structural panels
using ultrasonic consolidation (UC) was undertaken. As a novel solid freeform
fabrication technology, UC utilizes both additive ultrasonic joining and subtractive
CNC milling to enable the creation of complex aluminum structures with internal
geometry at or near room temperature. A series of experiments were performed to
understand the issues associated with sandwich structure fabrication using UC, including
peel test experiments which evaluated the bond strength for various geometric
configurations. The honeycomb lattice was found to offer the best core configuration
due to its ability to resist vibration from the sonotrode and provide adequate support for
pressure induced by the sonotrode. UC was found to be capable of producing lightweight
and stiff structures, including honeycomb and other sandwich panels, without the use of
adhesives. An effective manufacturing process plan for fabricating structural panels was
developed. A case study was performed on a deck built for the TOROID small satellite
spacecraft. The fabricated deck was tested for mechanical integrity. Finally, the cost and
benefits of utilizing UC for lightweight structural panels versus traditional fabrication
methods are discussed.
Keywords: Ultrasonic consolidation; Honeycomb; Sandwich panel; Lightweight panels;
Composite panel
1. Introduction
In mechanical design, a common design goal is to achieve
maximum load bearing capability or stiffness from a given
structural panel with the least amount of weight. This is
true throughout applications in the transportation and
aerospace industries. One application where this is particu-
larly true is in the area of small satellite design. Due to the
exorbitant costs associated with launching satellites, de-
signers seek greater functionality from their satellites at
reduced overall mass, volume and cost, and with less time
needed for design, production and testing (Kingston 2005,
Rodgers 2005). Over the past few years, small satellites have
emerged as a potentially disruptive technology for many
space missions (Lewin 2004). The potential disruptiveness
of small satellites, however, has been significantly hindered
by the cost and time involved when using traditional
manufacturing techniques for satellite fabrication, which
still predominate in industry (Panetta 1998). Traditional
methods of machining and assembly make every satellite
produced one-of-a-kind. Advanced additive manufacturing
techniques provide a potentially important shift in satellite
design and manufacturing, if an automated and repeatable
process can be developed for manufacturing satellite
systems (Mosher 2004).
One additive manufacturing technique that has tremen-
dous potential for fabricating the lightweight structural
panels needed for small satellites is ultrasonic consolidation
(UC). This technology, developed by Solidica Inc, USA,
uses a sonotrode to join layers of metal foils using
ultrasonic vibration (White 2002). The UC machine is an
*Corresponding author. Email: [email protected]
Virtual and Physical Prototyping, Vol. 1, No. 4, December 2006, 227 241
Virtual and Physical PrototypingISSN 1745-2759 print/ISSN 1745-2767 online # 2006 Taylor & Francis
http://www.tandf.co.uk/journalsDOI: 10.1080/17452750601106799
-
8/13/2019 Virtual and Physical Prototyping
3/16
integrated machine tool that incorporates an additive
ultrasonic joining head with a computer numerically
controlled milling machine (CNC). Because features may
be machined into the deposited layers and subsequently
covered with more layers, it is possible to create parts with
internal features. This is desirable, as sensors, electronics,
thermal regulators, and simple voids can be integrated to
create a multifunctional lightweight panel. In theory, it ispossible, using UC, to create a printed satellite which
offers reproducibility and functionality never before seen in
the satellite industry. This vision for printing satellite
panels, as illustrated in figure 1 where a functional satellite
panel is fabricated in a series of steps, was the impetus for
the current investigation. As the first step toward this
vision, the capabilities of UC with respect to fabrication of
lightweight structural panels were investigated.
2. Background
2.1 Lightweight structural panel design and fabrication
Over the past 40 years, many designs for lightweight
structural panels have been developed. Vinson (1999) has
applied a methodology to create minimum-weight panels.
He presented the idea that a panel contains many failure
modes, any of which could cause failure of the entire panel.
Different features of the panel have an associated weight
which varies directly with its load carrying capabilities.
When a failure occurs at any one location, any portions
which have not failed are essentially dead weight. Thus it
is apparent that a minimum-weight panel is one in which all
of the failure modes occur simultaneously.Another important aspect of structural efficiency and
weight was presented by Osgood (1966). He noted that an
optimum structure would weigh nothing and posses infinite
strength. Because neither of these is attainable, it is
necessary to define the method which will make optimiza-
tion possible. In many cases, the loading condition can be
well defined, thus imposing a constant strength require-
ment. Since the strength requirement is defined, the weight
must be the variable parameter which will enable optimiza-
tion.
There are two common solutions to the general design
problem of creating structural panels with high buckling
strength relative to their weight (Larson 2003). The first
solution is the use of a milled pattern, such as an isogrid or
orthogrid pattern, in aluminum or titanium plate metal (a
so-called rib-on-plate solution). The second solution is asandwich panel. Composites are an appealing solution for
sandwich panels due to their light weight and stiffness. They
do, however, present many difficulties due to the level of
required expertise, specialized equipment and expensive
molds needed for manufacturing.
The most common composite solution for lightweight
applications is a honeycomb sandwich panel. This type of
panel provides a large surface area and has a high ratio of
stiffness to weight (Osgood 1966). A simple form of the
sandwich construction consists of two thin, stiff, strong
sheets of dense material (facings) separated by a less stiff
and strong central core (Allen 1969). Generally, the core is
much thicker than the facings to prevent shear deformation
in the panel. The facings of a sandwich panel act similarly
to the flanges in an I-beam: they take the bending load with
one facing in compression and the other in tension. In a
typical I-beam, the flanges cannot be extremely thin
because of buckling at the flange tips. With sandwich
panels, however, the numerous webs which compose the
core support the flange tips, and thin facings will work,
even to their full material yield stress, without buckling
(Bitzer 1997).
The structural efficiency of a honeycomb sandwich panel
as compared to a solid metal sheet has been illustrated by
Hexcel (1999). This manufacturer has shown that asandwich construction twice the thickness of a solid metal
sheet can increase the stiffness by 7 times and the strength
by 3.5 times, while only increasing the weight by 3 percent.
A sandwich thickness of four times that of a solid metal
sheet increases the stiffness by 37 times and the strength by
9.25 times, with only a 6 percent increase in mass.
Honeycomb is widely used in the aerospace industry as a
core material for sandwich panels. Satellites requiring large
Figure 1. a) Metal layers are deposited by a UC machine. b) Milling operation removes material, leaving thin ribs. c) Sensors,
electronics, thermal regulators, and fiber reinforcement are embedded and covered with a facing. d) Final procedure mounts
solar cells and external mechanisms.
228 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
4/16
surface areas for solar cells almost always use some form of
honeycomb sandwich construction. It is typically produced
using one of two methods. The most common method is by
expansion (Hexcel 1999). As shown in figure 2, the
expansion process connects sheets of material with adhesive
lines. The resulting block is then cured and sliced to the
proper thickness for a core. A final procedure expands the
sliced block into a lattice of connecting cells that are very
thin. Later, thin sheets of another material can be glued to
the core as facings to form a sandwich panel.Though honeycomb core can be produced in very high
volumes, there are also many drawbacks to this type of
sandwich construction. Traditional methods require preci-
sion in assembly, as the process is sensitive to variation. In
addition, any bolted or riveted joints cause stress concen-
trations, and special potted inserts are required to prevent
local failures of bolts (Shirgur 2000), adding to the mass
and manufacturing time of the panel. In addition, this
design discourages modularity and increases both time and
cost when any modification to the panel is required. For
this reason, an automated process such as ultrasonic
consolidation is viewed as a valuable solution for creating
sandwich panels.
2.2 Ultrasonic consolidation
Though ultrasonic welding has been performed since the
late 1950s (Daniels 1965, Weare 1959), it has only recently
become a useful additive manufacturing technique. UC
emerged as a direct metal manufacturing technique and
rapid prototyping technology in the late 1990s. Research by
Johnson (1998) at Tufts University found that ultrasonic
welding could be used to make prototypes similar to other
rapid prototyping machines with the added benefits of low
energy consumption, modest space, and no emission offumes. In addition, he found that ultrasonic metal welding
had many advantages over other rapid prototyping methods
due to the fact that bonds can be formed between dissimilar
metals. He also noted that since there is no melting,
dimensional accuracy could be very high. Finally, he noted
that off the shelf materials could be used, which lowers cost.
Johnsons work involved the integration of a simple
ultrasonic metal welder and a high-speed cutter to make
very simple three dimensional dog bones for testing. His
work was followed by Gao (1999) who analyzed the
mechanics of ultrasonic metal welding during rapid proto-
typing. He used analytical modeling, finite element analysis,
and experimental data acquisition to look at static and
dynamic effects in the elastic and plastic flow regions during
welding.
White (2002) was the first to develop a commercial
machine tool to fabricate 3-dimensional structures using
ultrasonic welding. This ultrasonic consolidation (UC)machine uses a custom computer software tool to slice
the three-dimensional CAD model of the component to
be built into a number of horizontal layers, whose
thickness is equal to the metal foils used. These layers
are systematically created and stacked from bottom to
top, producing a three-dimensional object. Figure 3a
shows the UC machine installed at Utah State University
and figure 3b illustrates the basic UC additive manufac-
turing process. In the process a rotating ultrasonic
sonotrode travels along the length of a metal foil placed
over the substrate. The foil is held in contact with the
substrate by applying a normal force via the rotatingsonotrode. The sonotrode oscillates perpendicular to the
direction of welding at a frequency of 20 kHz and at a
user-set oscillation amplitude. The combination of normal
and oscillating shear forces results in the generation of
dynamic interfacial stresses between the two mating
surfaces (White 2003, Daniels 1965, OBrien 1991). The
stresses produce deformation of surface asperities, break-
ing up the oxide film, producing relatively clean metal
surfaces in intimate contact, establishing a metallurgical
bond. After depositing a strip of foil, another foil is
deposited adjacent to it. This process repeats until a
complete layer is placed. After placing a layer, a CNC
mill shapes the layer to its slice contour. This milling can
occur after each layer or, for certain geometries, after
several layers have been deposited. Once the layer is
shaped, chips are automatically blown away using com-
pressed air and foil deposition starts for the next layer. A
number of studies have been conducted to determine the
optimum process parameters for welding Al 3003 using
UC (Kong et al. 2004, Janaki Ram et al. 2006).
Figure 2. Conventional honeycomb production by expansion.
229Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
5/16
2.3 UC design issues
By utilizing UC, a number of potential benefits of tradi-tional honeycomb panels can be adopted without some of
the inherent manufacturing drawbacks of more traditional
processes. A UC panel can conceivably utilize a sandwich
honeycomb configuration of a thick core composed of thin
webs along with rigid facings. Because an integrated
machine tool is used to fabricate the panel directly from
CAD, the process eliminates some of the expertise and
precision required for traditional honeycomb manufactur-
ing. In addition, a UC-built honeycomb panel can adopt
modular bolt patterns, often used in orthogrid configura-
tions, but without the expensive, time-consuming process of
manually potting inserts for each bolt location. Finally, aUC built panel can integrate multifunctional capabilities by
embedding components, including electronics, wiring har-
nesses and thermal management devices, during the build
(Clements 2006). Figure 4 shows potential design ap-
proaches for lightweight structural panels. UC can be
considered a useful fabrication technique for structural
panels if the stiffness to mass ratio can exceed that of an
isogrid panel. It is not necessary for a UC panel to exceed
the stiffness to mass ratio of honeycomb panels for
usefulness, as the UC built panel will also have multi-
functional capabilities which are not possible when using
traditional honeycomb panels.
One of the critical design aspects of a honeycomb-type
sandwich panel is the achievable height to width ratio of the
webs within the core. Due to the physics of the UC process,
a newly deposited layer will not adhere to a previously
deposited layer if the stiffness of the previously deposited
layer is not sufficient to resist the vibrational forces of the
sonotrode (thus producing the differential motion necessary
between the newly deposited layer and the previously
deposited layer to induce plastic deformation at the inter-
face). A tall, thin rib will not resist sonotrode vibration as
well as a short, thick rib, for instance. As the rib height
increases, a cantilever effect allows the part to vibrate freely
(figure 5). Thus there exists a maximum height-to-width
ratio for a freestanding rib above which UC bonding will
not occur (Robinson et al. 2006). The maximum height-to-
width ratio of 1:1 found by Robinson for Al 3003 alloys is
considered inadequate for thin webs in a structural
sandwich panel core, and thus one of the major objectivesof this current work was to ascertain whether patterns of
ribs, as opposed to single, freestanding ribs, could achieve
height-to-width ratios significantly above 1:1.
The height and width of webs in the core are directly
related to the overall mass and stiffness of a sandwich
panel. The thinner the web, the lower the panels mass. The
taller the web, the higher the panels stiffness. This is due to
the effect of core thickness, c, on the bending stiffness, D, of
a sandwich panel (ASTM C 393-00). If the modulus
of elasticity, E, the width of the panel, b, and the sand-
wich thickness, T, are all held constant, the stiffness
increases rapidly with increasing core thickness, as shown
in equation (1).
DE (T3 c3) b
12(1)
Additionally, the core shear stress, t, and the facing bending
stress, s, can be defined (ASTM C 393-00) by:
Increasing Stiffness
Solid
Panel
Ortho
grid
Isogri
d
Honeycom
b
Absolut
eMini
mum-
Weigh
tStru
ctures
Figure 4. Scale to quantify usefulness of new fabrication
technique (assumes identical weight).
Figure 3. a) Solidica Form-ation ultrasonic consolidation
machine. b) Schematic of the ultrasonic consolidation
process.
230 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
6/16
tN
(T c) b(2)
s
N L
2 tf(T c) b(3)
where N is the load placed upon the midpoint of the
sandwich panel, andtfis the facing thickness. If c is allowed
to increase independent of the other variables, the stresses
experienced in the panel decrease. It is therefore evident
that a thicker core or web structure is stiffer and capable of
withstanding more stress.
3. Experimental approach
In order to determine the feasibility of creating lightweight
structural panels using UC, a series of experiments were
designed. These experiments were designed to determine the
types of core geometries which were most readily fabricated
using UC as well as the effects of build parameters on the
bond strength between the facing and the core, and
involved: (1) development of a repeatable method for
evaluating the bond strength between face sheets and the
core, (2) determination of a benchmark peel strength based
on established UC optimum parameters, (3) investigation of
the importance of heating the baseplate on bond strength,
(4) an evaluation of the effects of core rib direction on bond
strength between face sheets and the core, (5) and anevaluation of the effects of core lattice shape on bond
strength.
Following these experiments, a case study was performed
where the developed optimum core geometries and process
parameters were used to design and fabricate a functional
structural panel for the TOROID spacecraft. This panel
was modelled and tested to determine its mechanical
properties and its suitability for the launch environment.
3.1 Peel test apparatus development
Typically, honeycomb specimens undergo compressive test-
ing and plate shear testing (Bitzer 1997). These tests help in
measuring the compression modulus as well as the honey-comb shear strengths and moduli. Their applicability to UC
built specimens, however, may be minimal since the core is
not produced by gluing thin pieces of aluminum together.
Another series of tests, such as the flatwise tension test
and climbing drum peel test, are performed on assembled
sandwich panels to test the effectiveness of the bond
between the honeycomb core and the thin facings. The
flatwise edge test pulls the facings in tension to separate
them from the core. The climbing drum peel test peels off a
facing by rolling it around a drum. The failure modes in
both tests are revealed as core tearing, cohesive failure of
the adhesive, or failure of the adhesion to the honeycomb
or facing. Both of these tests are excellent ways ofevaluating the integrity of a honeycomb sandwich panel
(Bitzer 1997).
Kong (2005) found that a test method used for measuring
the resistance of adhesives to peeling was an effective
method for determining weld quality for specimens built
with UC. From his research, he found that as the number
and size of contact points within the welded interface
increased, so did the average resistance to peeling. Though
the peel test results were not as consistent as those for
adhesives, they revealed general trends for weld effective-
ness. Kongs peel test is very similar to the climbing drum
peel test, and thus a modified version of the climbing drumpeel test was adopted for our studies, resulting in an
apparatus much like Kongs (2003).
The Standard Test Method for Floating Roller Peel
Resistance of Adhesives (ASTM D3167-03a) was used to
create a fixture for specimens created on the UC machine.
Some deviations in the dimensions of the specified test
fixture to accommodate plates used in the Solidica machine
were required. Also, the speed was changed from 152 mm/
Figure 5. Effects of different height to width ratios.
231Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
7/16
-
8/13/2019 Virtual and Physical Prototyping
8/16
lightweight properties. Other designs include triangles, sine
waves, and squares. Based on results from the rib direction
studies, discussed below, it was determined that the core
lattice should avoid patterns where the ribs are oriented
parallel to the roller traverse direction, and instead max-
imize ribs perpendicular to the roller traverse direction. Of
the standard core configurations, hexagons and triangles
best meet these criteria. By argument it was difficult to
determine if the hexagon was a better core lattice than the
triangle so an experiment was designed to compare the two.
The dimensions of the patterns were chosen such that both
the triangle and the hexagon enclosed an area of 0.635 cm.
This area was chosen, as it was near the maximum area of a
triangle where the sonotrode would always be over at least 2
ribs at any point during its travel (so as to give it adequate
support). Ribs 1.016 mm thick were used for both pattern
geometries. The specimens were created by milling the
patterns into a 1.27 cm thick aluminum plate to a depth of
2.8 mm. A skin consisting of one tape was applied to each
specimen as shown in figure 8. The tapes were then removed
in a peel test.
4. Experimental results and discussion
4.1 General peel test apparatus results
A number of experiments were performed using the test
apparatus shown in figure 6 to determine whether the setup
gave repeatable and consistent results. A typical baseplate
for the UC machine is 35-by-35 cm, and allows all edges of
the plate to be firmly bolted to the heated platen using
8 fasteners. When using a smaller 10-by-35 cm baseplate
that is capable of fitting in the test apparatus, only the ends
of the baseplate (and not its sides) are bolted to the heated
platen. This reduces the rigidity of the smaller baseplate
during deposition. As three tapes can be consolidated side-by-side on the smaller baseplate, a study was performed to
see if location on the baseplate affected bond strength. This
study determined that the bond strength of tapes consoli-
dated on the outer edges of the small baseplate were always
inferior to the center tape bond strength. Thus, any valid
comparison of bond strength must be done for specimens
fabricated at the same location on the baseplate. This result
is not expected to be true of a firmly bolted, typically-sized
baseplate.
Another specimen preparation aspect that was found to
be of great importance was the direction of the flat-pass
milling operation used to prepare the surface of thebaseplate prior to consolidation. A flat pass milling
operation in the roller traverse direction was found to give
smooth peel data, whereas milling operations perpendicular
to the traverse direction caused spikes in the data.
For bond strength peel tests, there were three failure
mechanisms, based on the strength of the bond. The
weakest bonds would allow a smooth peel where the
resistance to peeling could be observed over the length of
Figure 7. Test specimens for determining the effect of rib
direction on peel strength. Figure 8. Test specimens for comparison of bonding for
hexagons and triangles.
233Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
9/16
the experiment without any tearing. In specimens with
extremely good bonding, the aluminium tape would tear
before significant peeling occured. This can be seen in figure
9. On the load plots for such peel tests, there was typically a
sharp incline, a short peak, and a rapid fall to zero. The
third reaction to the peel test was where a small portion of
the tape would tear due to a variation in weld quality across
the tape. In the load versus extension plots for this type of
reaction, the load would increase to a peak and then slowly
slope down to zero. This is because the tear decreases the
effective cross section being tested and all data after the
peak was therefore not comparable with other results.
4.2 Benchmark test
Figure 10 shows the results from peeling two different tapes
consolidated at the optimum parameters to a solid base-
plate. The maximum bond strengths measured were 193 N
and 195 N. The following experiment on the heat effect also
provided a peak value of 195 N. Using the three values, the
average maximum obtainable bond strength for theseparameters was approximately 195 N, which was taken as
a benchmark value for the discussions which follow. It
should be noted that the initial extension length prior to a
load increase was due to the amount of slack present in the
fixture. In order to compare results, the maximum load
reached and the shape of the load-versus-extension curve
were the primary comparison mechanisms, whereas the
absolute extension value was meaningless.
4.3 Effects of baseplate heating on bond strength
A comparison of results for a heated platen versus a room
temperature platen are shown in figure 11. The average
peak value for specimens processed at 150 8C was 195 N.
The room temperature specimen had a peak value of 71 N,
which was the same as the peak of 71 N reported by Kong
(2005) during his room temperature peel tests. From these
results, it was clear that heating the baseplate to 150 8Ccreated a bond nearly three times as strong as a room
temperature bond. The peak value was the same as the
other benchmark results since the test setup and ultrasonic
consolidation parameters were similar.
4.4 Rib direction
An overlay of three results from the rib direction experi-
ments are found in figure 12. The ribs parallel to the
traversing direction of the sonotrode (08) did not bond well,
as the load necessary to peel the tape was close to zero. The
458 ribs provided a relatively weak bond, but the tapes did
stick to the rib pattern. As had been expected, the ribs
perpendicular to the traversing direction provided a sub-
stantially better bond with a peak load of approximately
182 N before failure. The data shows a repeating pattern of
peaks, which indicate each time the tape pulls away from a
rib and the peel test advances to the next rib. The tape
eventually failed in a manner similar to that shown in figure
Figure 9. Failed specimen after being consolidated to a baseplate and peeled in a peel test.
0
50
100
150
200
250
0 2 4 6 8 10 12 14 16
Extension (mm)
Load(N)
Trial 1
Trial 2
Figure 10. Peel test data for maximum bond strength (150
8C, 16 mm, 28 mm/s, 1750 N).
0
50
100
150
200
250
0 2 4 6 8 10 12 14 16
Extension (mm)
Load(N)
150 Degrees C
27 Degrees C
Figure 11. Peel test data for heat effect (16 mm, 28 mm/s,
1750 N).
234 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
10/16
9, illustrating that a facing could be bonded to a rib just as
well as to a solid baseplate under certain conditions.
As illustrated by Robinson et al. (2006), producing tall,
thin free-standing ribs presents a challenge for the UC
process. What this present study illustrates is that an
effective way to minimize web deflection during vibrationis to orient ribs such that the mechanical oscillation of the
sonotrode is applied along the stiffest direction of a rib. As
the sonotrode oscillates perpendicular to the direction of
travel, ribs which lie perpendicular to the traversing
direction are most resistant to vibration, however, rib
patterns in this direction present problems as well, as the
sonotrode will dip into the channels between the ribs if
there are no other structural support members present,
creating a wavy surface. A 458 rib angle relative to the
traversing direction creates a smoother surface and provides
some resistance to vibration, but at a lower overall bond
strength than ribs at 908.
The results of this experiment, as alluded to earlier,
narrowed the options for practical core configurations.
When considering 08ribs, squares, hexagons, and triangles,
a series of perpendicular lines provide rigidity in only one
direction within a sandwich panel, and also produce a wavy
surface. A lattice of squares would provide rigidity in both
directions but only members oriented at 908 would bond
well to the facing. This left hexagons and triangles as the
likely best potential core configurations.
4.5 Core lattice geometry results
The results when comparing hexagonal and triangularpatterns (figure 13) show that the effective bond strength
was similar for both the hexagonal and triangular lattices,
and reached a maximum value of approximately 100N. In
the case of the triangle, the peak load occurred at the
location of maximum bond width (i.e. the width of a tape,
which was 2.4 cm (see figure 8)).
The hexagon, however, had a maximum bond width of
only 1.04 cm. Because this area was only 44% of the area of
a full tape, the equivalent normalized bond strength of the
hexagon was 231 N. This far exceeded the value obtained in
the benchmark test and showed that very good bonding can
occur between segments of honeycomb and a facing.
Because hexagonal structures are more efficient they
use the least amount of material to create a lattice of cellswithin a given volume for any common cell pattern
hexagons represent an excellent shape for a core within a
UC-fabricated sandwich panel. Based on the results from
the rib direction study, a good rule of thumb when using
UC to fabricate honeycomb sandwich panels is to ensure
that the hexagons are oriented such that no cell walls are
parallel to the traversing direction of the sonotrode.
5. Integrating experimental results into design
Once the honeycomb core was confirmed as the best coregeometry for UC-fabricated sandwich panels, solid me-
chanics theory was used to optimize the core geometry to
ascertain the best compromise between what is ideal and
what is realistic for fabrication in the UC machine. Though
the height to width ratio of a rib plays an important role
when fabricating a free standing rib, it has little impact on a
lattice of connecting cells, as the overall stiffness of the
lattice itself is the key to resisting vibration when adding
facings. From a practical standpoint, the thickness of the
panel, therefore, is limited to the maximum allowable
thickness for the application or the maximum depth the
Solidica machine can mill.Other geometric factors that must be determined are the
size of the honeycomb cells and the thickness of the cell
walls. Though the ideal core would posses very thin walls,
the wall thickness in a UC-built panel must be thick enough
to resist buckling under the applied normal force of the
sonotrode. The critical buckling load of the lattice is
determined by the second moment of inertia of the walls
of the cells, which is defined by Gibson (1988) as:
0
20
40
60
80
100
120
0 5 10 15 20 25 30 35 40 45
Extension (mm)
Load(N)
Hexagonal Pattern
Triangular Pattern
Figure 13. Peel test results for hexagonal vs. triangular
pattern (150 8C, 18 mm, 13 mm/s, 1750 N).
0
20
40
60
80
100
120
140
160
180
200
0 10 20 30 40 50 60
Extension (mm)
Load(N)
0 Degrees
45 Degrees
90 Degrees
Figure 12. Peel test results for variation in rib direction
(150 8C, 18 mm, 13 mm/s, 1750 N).
235Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
11/16
PK E
(1 n2)
t3
l(4)
where K is a constraint factor, E is the modulus of elasticity,
v is Poissons ratio, t is the wall thickness and l is the length
of a single cell wall. Because the UC machine operates at a
specified load during consolidation, and since K, E, and v
are constants which depend on the geometry chosen, t can
be solved as a function of l. The length of a single cell wall
ultimately determines the size of the cells, which is limited
by the need to have the sonotrode always straddling at least
two cell walls in order to provide the sonotrode a flat
surface to which a tape can be consolidated (to help avoid
waviness in the facings). A cell wall size of 0.787 cm was
chosen as an optimal value for maximizing the area of a
honeycomb under these constraints.
The parameter K can be approximated as 4 (Gibson
1988) based on the fact that the honeycomb cell is neither
completely free nor rigidly clamped. Equation (4) can then
be rewritten to represent the elastic collapse stress s3. For
regular hexagons and v/0.3, the formula becomes:
s3
E5:2
t
l
3(5)
The elastic collapse stress is equal to the force applied by
the sonotrode divided by the area of the ribs underneath the
sonotrode. The sonotrode has a contact area of approxi-
mately 5/24 mm. This area, however, acts only on the
honeycomb rib line enclosed by the area. For a cell wall
length of 0.787 cm the minimum required area can be
calculated as:
Alt2 t (0:197 t)
sin(60)
(6)
Plugging in 68.95 GPa for E, 1750 N for the force, and
0.787 cm for the cell wall length gives a minimum cell wall
thickness of 0.61 mm. To allow for a factor of safety, a cell
wall thickness of 1.0 mm was chosen as a conservative wall
size for a honeycomb lattice.
Finally, the dimensions of the facings were determined.
The standard thickness used in regular honeycomb of 0.635
mm corresponded well with the thickness of four consoli-
dated layers and was therefore used. During builds where
facings were consolidated to honeycomb cores, it was noted
that the first couple of layers contained minor defects due
to the sharp interface between the facing and the core whenapplied at high amplitudes. The third and fourth layers,
however, contained negligible defects and therefore repre-
sents the minimum facing thickness for a well built
sandwich panel.
A defined geometry now enables one to use an effective
or equivalent properties method to predict the properties
of the sandwich panel. This method uses the geometry of
the facings and core lattice to create an equivalent solid core
and skin which approximates the properties of the real
sandwich panel. The equivalent single skin plate method
outlined by Paik (1999) considers the rigidity of panels, with
equal facing skin thickness, separately for in-plane tension,
bending, and shear. This method was used to analyze the
deflections in the panel due to single point loading and the
results are compared with experimental data as part of
the case study below.
6. Case study: TOROID
6.1 Panel design
As a demonstration of the capabilities of UC for fabrication
of a lightweight panel, a deck panel was fabricated for the
TOROID spacecraft. Utah State Universitys entry into the
4th University Nanosatellite Competition is the Tomo-
graphic Remote Observer of Ionospheric Disturbances
(TOROID). TOROID will demonstrate both scientific
and technological capabilities as the satellite is fabricated,
tested, and eventually put into orbit around the earth. Thescientific mission of TOROID is to observe scintillations in
the low latitude ionosphere with increased fidelity. This
data will provide the scientific and military communities
with a greater understanding of the morphology and
equatorial phenomena which currently impede accurate
space-based geolocation.
There were several reasons a deck was needed in the
structural design of the satellite. First, the existing Utah
State University Satellite (USUSat) design emphasized the
importance of modularity by using panels. Components for
the various subsystems were attached to the panels which
were, in turn, assembled into a boxlike structure. This
allowed each panel to be tested individually for vibration
and thermal effects. Due to this design, there was very little
space for mounting a new payload, such as the TOROID
science instrument, to an external panel. This was a
significant problem in that the science instrument required
a large area and cantilever support (figure 14). While the
inside of the panels were covered with components and
electrical harnessing, the majority of the interior volume of
the satellite was empty. This empty space, however, left
room to install a horizontal deck panel, upon which the
science instrument could be mounted. It was proposed to
use UC to fabricate this deck as an integral part of the
TOROID design. As the deck employed new fabricationtechniques and the possibility for future multifunctional
capability (i.e. embedded wiring harnesses, thermocouples
or other devices) its development comprised one of the
technological objectives of the TOROID mission.
One major design feature of the USUSat Bus is a
standard bolt pattern. This pattern is repeated on the side
panels, which are orthogrid panels with reinforced, tapped
holes every 3.24 cm. For the UC panel, reinforced cylinders
236 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
12/16
were input into the CAD model to allow holes to be
machined and tapped. The pattern was aligned such that
the fasteners used to secure the deck to the satellite bus were
attached to the rim of the deck for added support. These
fastening points were placed on an edge perpendicular to
the direction of the tapes. This was done so that when the
deck panel is loaded, the tapes will not be stressed at theirabutting interfaces.
The rim around the perimeter of the deck panel acts as a
stiffener in the satellite. This helps maintain rigidity in the
in-plane axis. The overall dimensions of the final deck panel
were governed by the footprint of the TOROID spacecraft.
The maximum thickness of the panel was limited to
1.27 cm. Small holes were used to perforate the honeycomb
sections, as completely enclosed cavities have a tendency to
rupture in space due to low pressure in the space environ-
ment. This could have also been easily accomplished by
milling a tiny channel through the centroid of the honey-
comb cell walls during fabrication. The final CAD model ofthe TOROID deck panel without its top facing is shown in
figure 15.
6.2 Manufacturing process plan
Solidicas proprietary software, RPCAM, was used to
generate the G-code for the toolpaths and tape lays for
the solid model. A configuration file in the software enabled
the user to modify the weld speed, amplitude of oscillation,
and force for each model. To prepare the machine, an
ultrasonic couplant was applied to one face of the
aluminum baseplate. This couplant enhanced thermal
conduction between the heated platen and the aluminum
baseplate while mitigating differential motion between thesurfaces. The plate was bolted to the heated platen and a
flatpass milling operation was used to clean the surface of
the plate and to zero the plate with respect to the machine,
as shown in figure 16a.
The files for the first model were uploaded into the
controller for the Solidica machine and the process was
initiated. Thirteen foils were welded side-by-side to form a
layer (figure 16b). For each new layer deposited, the tapes
were offset to avoid the creation of a seam. Every four
layers the machine was programmed to use the milling head
to trim excess tape. After building up sufficient thickness
for the panel, the honeycomb pattern and bolt pattern were
milled into the UC-deposited structure. After a final
flatpass (figure 17a), the facing was consolidated to the
honeycomb core (figure 17b) and trimmed.
After adding the facing, the UC machine proceeded to
drill the holes for the bolt pattern and mounting brackets.
The perimeter of the deck was purposefully built larger than
necessary, in order to enable a final trim toolpath to remove
any poorly bonded edges, as our practical experience and
initial modelling efforts indicate that the edges of structures
are prone to non-optimal bonding (Li 2006). The trimmed
deck is shown in figure 18a. It should be noted that the
pattern visible in this figure is due to the fact that in
locations where there is no material below the facing, thedeposited tape remains shiny, whereas in the regions directly
above the honeycomb and perimeter pattern, the resistance
to vibration causes the sonotrode to roughen the surface of
the deposited layer, thus turning the surface white instead
of shiny. In addition, by running your finger over the
surface, you can feel the underlying honeycomb structure
through the facing, as the web locations are higher than the
open regions.Figure 15. CAD of final deck panel.
Figure 14. TOROID spacecraft with simplified science instrument.
237Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
13/16
Finally, excess baseplate material was removed by turning
the plate over and milling the back down until the desired
facing thickness remained on the deck panel, as shown in
figure 18b. This operation also resulted in a flat surface
suitable for mounting payloads. The deck was then removed
from the Solidica machine and cleaned up using a band saw
and manual mill. A secondary operation involved threading
the mounting points and installing helicoils for addedstrength within the bolt pattern. The final deck panel for
the TOROID spacecraft is shown in figure 19.
6.3 Mechanical Testing
The three beam-flexure test is often used to evaluate overall
sandwich panel performance (Hexcel 1999). This test, often
called the 3-point bend test, is particularly important since
it verifies how the core and facings work together to achieve
the overall mechanical performance of the panel. The test
can be performed with a single or double point load. The
stiffness of the panel can be calculated using the imposed
force and deflection at the mid span of the panel (Bitzer1997). The ASTM Standard Test Method for Flexural
Properties of Sandwich Constructions can be used to
determine the properties of flat sandwich constructions
subjected to flatwise flexure. Such an experiment can be
carried out in a quasi-static manner with a very low loading
speed (Paik 1999).
With the deck panel fabricated, it was possible to test the
panel for comparison with the finite element results. The
load cell used was a Tinius Olsen load cell with an 50 kN
capacity. It had the capability of measuring a force to within
9/0.5% of the indicated load. The extension was accurate to
within 9/0.01 mm. A 3-point bend test was performed by
placing the panel on two supported cylinders which created
a simply supported line in the same location where it was
applied in the finite element model. A third supported
cylinder was attached to the load cell and brought down tothe middle of the panel. The force and extension were
referenced at zero and the machine was programmed to
lower at a rate of 0.254 mm/min. The resulting data is
shown in figure 20.
The plot shows a nonlinear stiffness for the first 0.75 mm
and then a linear trend for the remainder of the test. The
nonlinear portion was due to the fact that the apparatus
was not touching the deck when the experiment
was initiated in addition to some minor settling in the
deck and fixture. The linear region showed a stiffness of
1511 N/mm.
For comparison with the finite element model, the
deflection at 1334 N was noted. The correct deflectionwas obtained by noting when the force measurements began
in the recorded data and using that as the reference for zero
deflection. The experimental data showed a deflection of
1.285 mm. The finite element analysis predicted 1.458 mm.
This is a difference of 12%. The deflection at 890 N for the
3-point bend test was 0.99 mm, compared to a predicted
value of 0.97 mm, a 2% difference. Differences were
expected due to assumptions made in the equivalent skin
Figure 16. a) Clean baseplate, b) First layer of consolidated aluminum tapes.
Figure 17. a) Milled honeycomb core. b) First layer of top facing consolidated on honeycomb core.
238 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
14/16
method for the finite element model as well as discrepancies
between the setup of the model and the setup of the 3 point
bend test and due to settling in the fixture. For the purposes
of showing a general trend between experimental and
numerical results, however, the results give sufficient
correlation to enable some level of usefulness for predicting
UC sandwich panel performance using an equivalent skin
method.
6.4 Economic Considerations
The total time for the deck panel build was approximately
56 hours. As much of the build was performed unattended,
the deck was completed within a 40 hour work week. The
majority of the time was in machining the honeycomb grid
and bolt pattern. These tasks consumed a significant
amount of time due to the inefficiency of the milling
apparatus on the UC machine at Utah State University.
This machine is not optimized for removing large amounts
of material, as it possesses a small spindle. The bolt pattern
also took considerable time due to excessive lengths fortoolpaths. The time required for both of these operations
could be greatly reduced by performing them in a CNC
with a more powerful spindle.
The total cost of the build was $1597 USD. This cost was
comparable to the cost incurred when a professional
machine shop machined one of the panels of the USUSat
bus. An informal estimate was done by a commercial small
satellite producer as a comparison. They estimated that a
panel similar to the deck plate but fabricated out of
honeycomb composite would cost $2200 to $3200 USD.
Material costs would have been similar but labor costs
would have greatly surpassed those for fabricating the
TOROID deck. This illustrates one of the main advantages
of the UC-built panel over a traditionally fabricated panel,namely the lack of touch-labor required.
The mass was estimated for an equivalent composite
panel with potted inserts as 0.22 kg compared to 0.626 kg
for the UC panel. This illustrates one of the disadvantages
of UC-built aluminium honeycomb panels. In small satel-
lites, such as TOSOID, however, this mass difference has
very little impact on its handling and launch costs, as the
satellite is inherently small and light. The UC built deck has
the added benefit that it can very easily be made into a
multifunctional structure.
7. Conclusion
Ultrasonic consolidation has been demonstrated as a useful
technique for fabrication of lightweight structural panels.
The peel test has been found to be an effective method for
evaluating bonding between an aluminum tape and a solid
or lattice substrate. A benchmark load of 195 N has been
measured as a benchmark bond strength when using
Figure 18. a) Deck panel with bolt pattern milled out. b) CNC mill removing excess material from baseplate.
Figure 19. Finished deck panel.
0
500
1000
1500
2000
2500
0 0.5 1 1.5 2 2.5
Displacement (mm)
Load
(N)
Figure 20. Stiffness of prototype deck panel from 3-point
bend test.
239Fabrication of lightweight structural panels through UC
-
8/13/2019 Virtual and Physical Prototyping
15/16
optimum parameters (Jankai Ram 2006). Heating the
baseplate to 150 8C during consolidation has been found
to enhance bond strength by 3X over room temperature
bonding. Core rib direction has a significant impact on the
bond strength and ribs perpendicular to the traversing
direction of the sonotrode have been found to give the best
bond. This has led to the implementation of an optimized
honeycomb lattice for the core of sandwich panels. Thisconfiguration has been found to allow benchmark quality
bonding between the facing and core while maximizing the
volume within the cells.
The design of the TOROID UC sandwich panel has
integrated the experimental results and solid mechanics
theory. Though this design does create a panel with greater
mass than a similar panel made with composite honey-
comb, its method of fabrication is possible without the use
of significant touch-labor, epoxy or inserts. UC also has the
added benefit of enabling the embedding of useful compo-
nents. The TOROID deck panel was found to interface well
with other structural elements in the satellite and has
provided significantly more mountable area for scienceinstruments. Its testing has shown a stiffness of 1,511 N/mm
and its suitability for the launch environment.
8. Future work
There is a significant amount of future work possible to
improve the quality of structural panels produced using
UC. The most significant may be the implementation of a
support material apparatus in the UC machine. Support
material would allow each layer of the facing to bond fully
to the layer below it. Currently bonding between facing
layers is only achieved directly over the honeycomb cell
walls. In addition, support material would enable the use of
thinner web structures (as they would support against
buckling) and remove some of the honeycomb size restric-
tions.
The facings could also benefit from fiber reinforcement.
Utilization of a fiber-reinforced tape as the facing would
make the sandwich panel much stiffer. Also, the overall
mass of the panel could be reduced by using a stronger
material, such as a 6061 aluminum alloy. In addition, 6061
is more accepted in the aerospace community due to its
extensive use in successful missions.
The time it takes to produce a deck panel could be cutfrom one week to approximately one day with the use of a
more powerful mill, more machinable alloys, and with
optimization of the toolpaths for machining as well as when
laying tapes.
One additional area for future work is the embedment of
subsystems into a structural deck panel. Items such as heat
pipes, antennas, wiring, thermocouples, low profile heaters,
embedded computers, connectors, and printable batteries
are being tested for integration into functional deck panels.
Their integration may one day be automated using direct
write technologies and pick-and-place heads, greatly in-
creasing the automation possible when fabricating future
multi-functional panels.
ReferencesAllen, H.G., Analysis and design of structural sandwich panels, 1969
(London: Pergamon Press).
ASTM, Standard test method for floating roller peel resistance of adhesives,
ASTM D 3167-03a (2004), 2004 (ASTM International, West Consho-
hocken, PA).
Bitzer, T.,Honeycomb technology: materials, design, manufacturing, applica-
tions and testing, 1997 (London: Chapman & Hall).
Clements, J. and George, J., Rapid manufacturing of satellite structures and
heat pipes using ultrasonic consolidation,Proceedings of the 20th AIAA/
USU Conference on Small Satellites, Logan, UT, 2006.
Daniels, H.P.C., Ultrasonic welding, Ultrasonics, 1965, October
December,
190196.
Gao, Y., Mechanical analysis of ultrasonic metal welding for rapid
prototyping, Masters diss., Tufts University, 1999.
Gibson, M.F. and Ashby., M.F., Cellular solids: Structure and properties,1988 (Elmsford, New York: Pergamon Press, Inc.).
Hexcel,Hexweb honeycomb attributes and properties, 1999 (Pleasanton, CA:
Hexcel).
Janaki Ram, G.D., Yang, Y., George, J. and Robinson, C. eds., Improving
linear weld density in ultrasonically consolidated parts, 17th Solid
Freeform Fabrication Symposium, 2006.
Johnson, N., Rapid prototyping using ultrasonic metal welding, Masters
thesis, Tufts University, 1998.
Kingston, J., Modular architecture and product platform concepts applied
to multipurpose small spacecraft, 19th AIAA/USU Conference on Small
Satellites, Logan, UT, 2005.
Kong, C., Investigation of ultrasonic consolidation for embedding active/
passive fibres in aluminium matrices. PhD. Thesis, Loughborough
University, 2005.
Kong, C., Soar, R.C. and Dickens, P.M., Optimum process parameters forultrasonic consolidation of 3003 aluminium, Journal of Materials
Processing Technology, 2004, 146, 181187.
Kong, C., Soar, R.C. and Dickens, P.M., Characterization of aluminium
alloy 6061 for the ultrasonic consolidation process,Materials Science and
Engineering: Part A, 2003, 363, 99106.
Larson, W. J., Spacecraft structures and mechanisms: From concept to
launch, 2003 (El Segundo, CA: Microcosm, Inc.).
Lewin, A.W., Evaluating the present and potential future impact of small
satellites,18th Annual AIAA/USU Conference on Small Satellites, Logan,
UT, 2004.
Li, L. and Chunbo, Z. A study of static and synamic mechanical behavior
of the substrate in ultrasonic consolidation (review), 17th Solid Freeform
Fabrication Symposium, 2006.
Mosher, T. and Stucker, B., Responsive space requires responsive manu-
facturing, 2nd Responsive Space Conference, Los Angeles CA, April 19
22, 2004.
OBrien, R.L., Welding Handbook, 8th ed., vol. 2, 1991 (American Welding
Society: Miami).
Osgood, C., Spacecraft structures, 1966 (N.J.: Prentice Hall, Inc.).
Paik, J.K., Thayamballi, A.K. and Kim, G.S., The strength characteristics
of aluminium honeycomb sandwich panels,Thin-Walled Structures, 1999,
35, 205231.
Panetta, P.V., Culver, H., Gagosian, J. and Johnson, M. eds., NASA-GSFC
nanosatellite technology development, Proceedings of the 12th AIAA/
USU Conference on Small Satellites, Logan, UT, 1998.
240 J. George and B. Stucker
-
8/13/2019 Virtual and Physical Prototyping
16/16