aerospace
Article
Thermoplastic Mandrel for Manufacturing Composite
Components with Complex Structure
Xishuang Jing 1,2 , Siyu Chen 1,2 , Jiuzhi An 1 , Chengyang Zhang 1,3, * and Fubao Xie 2






Citation: Jing, X.; Chen, S.; An, J.;
Zhang, C.; Xie, F. Thermoplastic
Mandrel for Manufacturing
Composite Components with
Complex Structure. Aerospace 2021, 8,
399. https://doi.org/10.3390/
aerospace8120399
Academic Editor: Kyungil Kong
Received: 5 November 2021
Accepted: 13 December 2021
Published: 16 December 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Aerospace 2021, 8, 399. https://doi.org/10.3390/aerospace8120399
1 School of Mechanical Engineering and Automation, Beihang University, Beijing 100191, China;
tom@buaa.edu.cn (X.J.); Xavier_csy@buaa.edu.cn (S.C.); SY2007312@buaa.edu.cn (J.A.)
2 Beihang Hangzhou Innovation Institute Yuhang, Hangzhou 311121, China; xiefb@buaa.edu.cn
3 Key Laboratory of Intelligent Manufacturing Technology for Aeronautics Advanced Equipment,
Ministry of Industry and Information Technology, Beihang University, Beijing 100191, China
* Correspondence: zhangchengyang@buaa.edu.cn
Abstract: This study was to solve the mandrel demolding problem after curing the composite compo-
nent with complex structure. In this paper, a reusable thermoplastic mandrel with heating softening
characteristics was developed by resin transfer molding (RTM). The glass transition temperature (Tg),
surface roughness, and reusability of the mandrel, as well as the shape, surface roughness, thickness
uniformity, and internal quality of the formed structure, were tested. The result showed that the
Tg of the mandrel was between 80 and 90 ◦ C and the surface roughness was less than Ra 0.5 µm.
Additionally, the mandrel can be recycled and can still maintain a good shape after 20 times of
deformation. By using this method, the demolding process can be realized by heating and softening
the mandrel. The profile error of the formed structure was within 0.5 mm, the surface roughness was
less than Ra 0.5 µm, the thickness error was within 0.2 mm, and the average porosity of the upper
and lower halves of composite parts was 0.72% and 0.61%. All those data showed that the formed
part was in good shape and of good quality. The thermoplastic mandrel can solve the demolding
problem of composite materials with complex shapes.
Keywords: composites; composites’ manufacturing; reusable thermoplastic mandrel; mandrel demolding
1. Introduction
Carbon fiber composite was developed rapidly in recent decades. Owing to its light
weight, high specific strength, and specific modulus, it can reduce the weight of an aircraft.
Many composite components are now made of carbon fiber instead of aluminum alloy [1–3].
With the rapid expansion of its application, carbon fiber composite was gradually used
to manufacture components with complex shapes, such as wing spar beams with double
torsion, S-shaped variable cross-section inlet, and other parts [4,5]. However, since the
forming and curing processes of the aforementioned components are completed at the
same time, when the components have cavities, we need to use the mandrel for composite
winding [6–8] or prepreg paving to form such a complex structure [9–12]. When we use the
traditional hard-mandrel to make a complex structure, after curing the prepreg, the state of
the mandrel changes from a viscoelastic state to a solid state, which makes it difficult to
demold and even impossible to make a complex structure or composite component with
inner cavities.
In order to solve the problem of difficult demolding, experts and scholars at home and
abroad carried out relevant research. Forcellese et al. studied the 3D printing technology of
composite materials, formed the composite grid structure with complex shape, and tested
its strength [13]. Hao et al. established a fused deposition modeling (FDM) 3D printing
platform [14]. On this platform, they used the rotating cylindrical mandrel to make the
thermosetting composite a cylindrical grid structure; thus, the use of the mandrel inside
the grid structure was saved. These 3D printing technologies can produce more complex
https://www.mdpi.com/journal/aerospace
 the thermosetting composite a cylindrical grid structure; thus, the use of the mandrel in-
side the grid structure was saved. These 3D printing technologies can produce more com-
plex shapes, but it is difficult to produce structures with curved radian shapes and a cy-
lindrical mandrel is still needed as support when making a cylindrical grid structure.
Aerospace 2021, 8, 399 Researchers have also made attempts at innovating the materials of mandrel. 2 of 12 Lom-
bardi et al. and Vaidyanathan et al. studied the water-soluble mandrel [15,16]. The man-
drel can be dissolved by water, which solves the problem of demolding. However, the
water-soluble
shapes, but it is mandrel
difficult tocannot
producebe reused. with
structures Mohammadi
curved radian M et al. studied
shapes the soft rubber
and a cylindrical
mandrel is
mandrel still Through
[17]. needed asthe support
grooved whenrubber
makingmandrel,
a cylindricalthegrid structure.
prepreg is paved on the groove
Researchers
for curing. Then have also made
the volume of attempts
the rubber at innovating
mandrel isthe materials
reduced by of mandrel. Lom-
compressed air, so that
bardi et al. and Vaidyanathan et al. studied the water-soluble
the rubber core is detached from the cavity to make the cylindrical pyramidal lattice struc- mandrel [15,16]. The
mandrel can be dissolved by water, which solves the problem of demolding. However,
ture. Xie et al. made a rubber mandrel with different thicknesses to make a cap girder and
the water-soluble mandrel cannot be reused. Mohammadi M et al. studied the soft rubber
studied
mandrel the [17].influence
Through the of grooved
the different
rubberrubber
mandrel, mandrels
the prepreg onisthe thickness
paved error for
on the groove of the cap
girder
curing.[18].
ThenLithe et volume
al. studied
of thethe influence
rubber of aisrubber
mandrel reducedmandrel
by compressedwith a air,
middle opening
so that the and
its air pressure on the quality of cap girder through simulation
rubber core is detached from the cavity to make the cylindrical pyramidal lattice structure. [19]. Although this rubber
mandrel
Xie et al. can made beademolded by itswith
rubber mandrel softness, there
different might betouncontrollable
thicknesses make a cap girder deformation
and of
studied
the the influence
external prepreg of the different
during curing,rubber
finallymandrels
affectingon thethe thickness
accuracy oferror of the cap
the formed structure.
girder [18]. Li
Shape et al. studied
memory the influence
polymers with lowerof a rubber
glass mandrel
transition with a middle opening
temperature (Tg) and
provide us
with a new way of demolding. That is, heating can soften the mandrel so rubber
its air pressure on the quality of cap girder through simulation [19]. Although this that it can be
mandrel can be demolded by its softness, there might be uncontrollable deformation of the
extracted from the part. Du et al. studied a shape memory polymer mandrel that can be
external prepreg during curing, finally affecting the accuracy of the formed structure.
used Shape
to form composite parts with complex structures [20,21]. The matrix material of the
memory polymers with lower glass transition temperature (Tg) provide us with
mandrel
a new way is of
styrene,
demolding. and That
the mandrel
is, heatingwillcangradually soften when
soften the mandrel so thatheated.
it can beItextracted
provides a new
idea
from the part. Du et al. studied a shape memory polymer mandrel that can be the
for the manufacture of composites, but its strength cannot support usedcuring
to and
forming of external
form composite parts prepreg,
with complexfinally affecting
structures the shape
[20,21]. accuracy.
The matrix material of the mandrel
is styrene,
To utilizeand the themandrel
specificwill gradually
polymers soften
with when and
heating heated. It provides
softening a new idea for
characteristics and com-
the manufacture of composites, but its strength cannot support
bined materials with sufficient strength that can support modeling, a thermoplastic man- the curing and forming of
external
drel prepreg,of
composed finally
carbonaffecting the shape accuracy.
fiber-reinforced plastic polymers was made by RTM. The ther-
To utilize the specific polymers with heating and softening characteristics and com-
moplastic mandrel can support the paving of prepreg at room temperature, complete the
bined materials with sufficient strength that can support modeling, a thermoplastic mandrel
curing and molding under high-pressure during high temperatures, and realize heat-sof-
composed of carbon fiber-reinforced plastic polymers was made by RTM. The thermoplas-
tening demolding
tic mandrel can support by itsthe
low Tg. The
paving Demolding
of prepreg at room principle of thecomplete
temperature, thermoplastic mandrel is
the curing
shown in Figure 1. This paper mainly studies a thermoplastic
and molding under high-pressure during high temperatures, and realize heat-softening de- mandrel for manufacturing
composite
molding bycomponents
its low Tg. The with complex
Demolding structures.
principle of theThrough
thermoplasticthe manufacturing
mandrel is shown ofincomplex
composite parts including the fabrication of a reusable mandrel and the investigation of
Figure 1. This paper mainly studies a thermoplastic mandrel for manufacturing composite
components
the mandrelwith complexprocess,
demolding structures. theThrough the manufacturing
applicability of complex
of a thermoplastic composite
mandrel for manu-
parts including the fabrication of a reusable mandrel
facturing excellent composite components with complex structure was verified. and the investigation of the man-
drel demolding process, the applicability of a thermoplastic mandrel for manufacturing
excellent composite components with complex structure was verified.

1. Demolding
Figure 1.
Figure Demoldingprinciple
principleof of
thethe
thermoplastic mandrel.
thermoplastic mandrel.
2. Material and Sample Manufacturing
2.2.1.
Material
Materialsand Sample Manufacturing
2.1. Materials
The mandrel was composed of reinforcing fiber and matrix material. The carbon fiber
plainThe
fabric (H3K-CP200,
mandrel RELIEYSEX,
was composed ofShanghai, China)
reinforcing was
fiber adopted
and matrixtomaterial.
ensure theThe
strength
carbon fiber
of the mandrel at room temperature. The matrix material mainly adopted thermoplastic
plain fabric (H3K-CP200, RELIEYSEX, Shanghai, China) was adopted to ensure the
material with low Tg; thus, the mandrel can remain softened without air leakage at high
temperature [22–24]. The matrix material consisted of polymethylmethacrylate (PMMA,
Elium 188, ARKEMA Co., LTD., Shanghai, China), epoxy resin (NPEF-170, Nan Ya Plastics
Corp, Taiwan, China), and liquid nitrile butadiene rubber (LNBR, Lanzhou Petrochemical
 extract the mandrel, the RTM mold was generally designed in the form of multi-part com
bination.
Before the RTM process, the matrix material was prepared, referring to the matr
Aerospace 2021, 8, 399 material mentioned in Section 2.1. The materials in the formula were dried3 of in12a vacuu
oven for 8 h before use. Finally, the resin mixture was centrifuged and deaerated by ce
trifuge for 10 min, and then it was used to prepare the mandrel by the RTM process. T
Mandrel preparation
Company flow Shanghai,
of Petro china, chart is shown
China). in
TheFigure
matrix 2a.
material was stirred and mixed in
Because the mandrel often has complex cavities, itcorresponding
the resin reaction kettle at the rate of 60 r/min. Then, the is necessarycomposite parts
to use a water-solub
were formed by injecting the matrix material into the mold with the paved carbon fiber
core to form the cavity. First of all, we laid twill carbon fiber cloth (TORAY INDUSTRIE
plain fabric.
INC., Tokyo, Japan)parts
Composite on the
were core and then
planned put itby
to be made into the RTM
prepreg pavingmold. Next, the
and curing. As formatrix m
terial,the
which was prepared
composite material, the incarbon
the previous step
fiber/epoxy (cured
resin at 60
prepreg, °C for 4 h), (VTC401-
VTC401-C650T was injected in
the RTMC650T, Resinas
mold. Castro,
After thePontevedra,
resin wasSpain)cured,was selected,
the mandrel which is commonly
was obtainedused in the
by dissolving t
manufacture of composite parts.
core. The RTM process for manufacturing the mandrel, whose thickness was about 3 m
is shown in Figure 2b.
2.2. Thermoplastic TheManufacturing
Mandrel gray part on the mandrel is the attached Teflon film. Aft
separatingThe the resin was
mandrel matrix of manufactured
mainly the mandrelbyand RTM,theasprepreg by Figure
shown in the the film, the prepreg
2. Firstly, the w
directly laid on the mandrel.
metal RTM mold needed to be designed according to the shape of the mandrel. To extract
the mandrel, the RTM mold was generally designed in the form of multi-part combination.

(a)

(b)
Figure 2. RTM process for manufacturing mandrel: (a) The flow chart of mandrel preparation; (b) Mandrel prepared by
Figure 2. RTM process for manufacturing mandrel: (a) The flow chart of mandrel preparation; (b) Mandrel prepared by
RTM process.
RTM process.
Before the RTM process, the matrix material was prepared, referring to the matrix
material mentioned in Section 2.1. The materials in the formula were dried in a vacuum
oven for 8 h before use. Finally, the resin mixture was centrifuged and deaerated by
centrifuge for 10 min, and then it was used to prepare the mandrel by the RTM process.
The Mandrel preparation flow chart is shown in Figure 2a.
Because the mandrel often has complex cavities, it is necessary to use a water-soluble
core to form the cavity. First of all, we laid twill carbon fiber cloth (TORAY INDUSTRIES,
INC., Tokyo, Japan) on the core and then put it into the RTM mold. Next, the matrix
material, which was prepared in the previous step (cured at 60 ◦ C for 4 h), was injected
into the RTM mold. After the resin was cured, the mandrel was obtained by dissolving
the core. The RTM process for manufacturing the mandrel, whose thickness was about
3 mm, is shown in Figure 2b. The gray part on the mandrel is the attached Teflon film.
After separating the resin matrix of the mandrel and the prepreg by the film, the prepreg
was directly laid on the mandrel.

2.3. Manufacturing of Composite Parts

According to the properties of the mandrel, we could use it in the curing process of
vacuum bag prepreg to form complex composite components. The end connectors on both
 assembled on both sides of the mandrel.
Because the mandrel remained solid at room temperature, after pasting the Teflon
layer on the outer surface of the mandrel, the prepreg was directly laid on it and the caul
plate (CP) was also laid above the prepreg. A layer of Teflon film was also laid on the
inner side of the CP to avoid the bonding between the mandrel and the prepreg. The sche-
Aerospace 2021, 8, 399 4 of 12
matic diagram of the vacuum bag curing process is shown in Figure 3a,b.
The most important step is to inject air at normal pressure through the ends on both
sides of the mandrel, as shown in Figure 3c. If the air pressure inside and outside the
vacuum
sides of bag is identical
the mandrel to each
were other,tothe
designed mandrel
ensure can
the air maintain
tightness ofits
theshape during thebeing
end connectors cur-
ing of the prepreg.
assembled on bothAfter
sidescuring the prepreg, the mandrel was heated to soften, and then
of the mandrel.
it wasBecause
extractedthefrom the cavity
mandrel part. solid at room temperature, after pasting the Teflon
remained
layerAsonshown in Figure
the outer surface3d,
of we
the used the mandrel
mandrel, for was
the prepreg prepreg paving
directly laidand
on itvacuum
and the bag
caul
curing to prepare
plate (CP) was alsocomposite
laid abovecomponents
the prepreg.with complex
A layer shapes.
of Teflon film The
was mandrel
also laid was sealed
on the inner
by connectors
side of the CP on both sides,
to avoid and the
the bonding inflation
between thewas maintained
mandrel and theinprepreg.
the mandrel’s shape
The schematic
when
diagramthe prepreg was cured.
of the vacuum bag curing process is shown in Figure 3a,b.

(a) (b)

(c) (d)
Figure
Figure3.3.The
Themanufacturing
manufacturingprocess
processby
byvacuum
vacuumbag bagtechnique:
technique:(a)
(a)Schematic
Schematicdiagram
diagramofofvacuum
vacuumbag
bagcuring
curingprocess;
process;(b)
(b)
The manufacturing process of composite parts; (c) the illustration of air pressure balance during part forming; (d) Formed
The manufacturing process of composite parts; (c) the illustration of air pressure balance during part forming; (d) Formed
composite parts.
composite parts.
The most and
3. Experiments important
Tests step is to inject air at normal pressure through the ends on both
sides of the mandrel, as shown in Figure 3c. If the air pressure inside and outside the
Experiments and tests were mainly carried out on the thermoplastic mandrel and the
vacuum bag is identical to each other, the mandrel can maintain its shape during the curing
basic properties of composites were prepared with it.
of the prepreg. After curing the prepreg, the mandrel was heated to soften, and then it was
extracted from the cavity part.
3.1. Mandrel Performance Test
As shown in Figure 3d, we used the mandrel for prepreg paving and vacuum bag
Thetoglass
curing transition
prepare temperature
composite components(Tg), surface
with roughness,
complex shapes.and
Thereusability
mandrel wasof sealed
thermo-by
plastic
connectors on both sides, and the inflation was maintained in the mandrel’s shapeparts.
mandrel directly determine whether it is suitable for forming composite when
Therefore, in was
the prepreg this section,
cured. several experiments were designed to test these three properties.

3.1.1. Glass Transition

3. Experiments Temperature
and Tests
The Tg of the and
Experiments mandrel was tested
tests were mainlyby dynamic
carried thermomechanical
out on the thermoplasticanalysis
mandrel(DMA).
and the
basic
The propertiesused
equipment of composites were prepared
in the experiment with it.
was DMA Q800 (DMA Q800, TA Instruments,

3.1. Mandrel Performance Test

The glass transition temperature (Tg), surface roughness, and reusability of thermo-
plastic mandrel directly determine whether it is suitable for forming composite parts.
Therefore, in this section, several experiments were designed to test these three properties.

3.1.1. Glass Transition Temperature

The Tg of the mandrel was tested by dynamic thermomechanical analysis (DMA). The
equipment used in the experiment was DMA Q800 (DMA Q800, TA Instruments, New
Castle, DE, USA) in tensile mode. The detailed parameters are shown in Table 1 below.
Through experiments, we obtained the change curves of the storage modulus and loss
modulus of the thermoplastic mandrel with temperature at different heating rates. We then
 New Castle, DE, USA) in tensile mode. The detailed parameters are shown in Table 1 be
Aerospace 2021, 8, 399 low. Through experiments, we obtained the change curves of the storage modulus 5 of 12 and
loss modulus of the thermoplastic mandrel with temperature at different heating rates
We then calculated the peak value of loss modulus and the peak point of derivation of th
storage modulus
calculated the peak and
valuefinally
of lossobtained the Tg
modulus and the of thermoplastic
peak mandrel
point of derivation [25,26].
of the storage
modulus and finally obtained the Tg of thermoplastic mandrel [25,26].
Table 1. Dynamic thermodynamics’ experimental equipment and conditions.
Table 1. Dynamic thermodynamics’ experimental equipment and conditions.
Experimental Equip- Heating Rate Temperature
Size (mm) Mode Control Model Frequency (Hz)
Experimental
ment Frequency Heating(°C/min)
Rate Temperature
Range (°C)
Size (mm) Mode Control Model
Equipment (Hz) (◦ C/min) 1 Range (◦ C)
DMA Q800(TA In- Displacement 1
DMA Q800 10 × 3.5 × 2.5 Tensile Displacement 1 2.5 45–140
struments, USA) 10 × 3.5 × 2.5 Tensile Control 1 2.5 45–140
(TA Instruments) Control 5
5

The results of the DMA test of the thermoplastic mandrel are shown in Figure 4. Th
The results of the DMA test of the thermoplastic mandrel are shown in Figure 4.
results showed that the storage modulus decreased more than a thousand times when th
The results showed that the storage modulus decreased more than a thousand times
mandrel material was heated from 45 to 140 °C. When the material was in low tempera
when the mandrel material was heated from 45 to 140 ◦ C. When the material was in low
ture, the resin
temperature, thewas
resinatwas
theatglass state.state.
the glass AfterAfter
the temperature
the temperature rose, thethe
rose, material
materialbecame
highly elastic
became rubber.
a highly elastic In addition,
rubber. with thewith
In addition, increase of the heating
the increase rate, the
of the heating molecular
rate, the mo
tion needed
molecular to overcome
motion needed tolarge internal
overcome largefriction
internaland thermal
friction motion motion
and thermal energy.energy.
Therefore, th
Therefore, the storage
storage modulus modulus distribution
distribution curve moved curvetomoved to thetemperature
the higher higher temperature
side. side.

Figure4.4.DMA
Figure DMAtest results
test of mandrel
results material.
of mandrel material.

The Tg of the thermoplastic mandrel were determined by the peak value of loss modu-
The Tg of the thermoplastic mandrel were determined by the peak value of loss mod
lus and the peak point of derivation of the storage modulus in the DMA test results [25,26].
ulus and the peak point of derivation of the storage modulus in the DMA test result
At different heating rates, the Tg corresponding to the thermoplastic mandrel is shown in
[25,26].
Table At test
2. The different heatingthat
results showed rates,
the the
Tg ofTg
thecorresponding to the80
mandrel was between thermoplastic mandrel i
and 90 ◦ C, which
was 50 to 60 C lower than that of the thermoset composite. The Tg of the thermoplastic 80 and
shown in Table
◦ 2. The test results showed that the Tg of the mandrel was between
90 °C, which
mandrel was 50
could avoid thetoinfluence
60 °C lower than that
of demolding on of
thethe thermoset
formed composite.
composite The Tg of th
components.
thermoplastic mandrel could avoid the influence of demolding on the formed composit
Table 2. Glass transition temperature (Tg) of mandrel material.
components.
The Peak Value of the Loss The Peak Value of Derivative of
Heating
Table Rate
2. Glass transition temperature (Tg) of mandrel material.Storage Modulus
Modulus
1 ◦ C/min The Peak81
Value of the Loss 81.5 of Derivative of Storage
The Peak Value
Heating
◦
Rate
2.5 C/min Modulus
86 86.5 Modulus
51◦ °C/min
C/min 91 81 90 81.5
2.5 °C/min 86 86.5
5 °C/min 91 90
Aerospace 2021, 8, x FOR PEER REVIEW 6 of 1
Aerospace 2021, 8, 399 6 of 12

3.1.2.Surface
3.1.2. Surface Roughness
Roughness
Thesurface
The surface roughness
roughness of theofthermoplastic
the thermoplastic
mandrelmandrel
was mainly was mainly
tested by a tested by a rough
roughness
ness instrument
instrument (EDKORS, (EDKORS, TR110,
TR110, CHN). CHN). The
The roughness roughness
tester tester can
can only measure only measure th
the roughness
roughness
within a smallwithin
lengtha (the
small length (the
sampling sampling
length length of the
of the instrument instrument
is 0.8 is 0.8 mm).
mm). Therefore, we There
tested 20 points on the surface to evaluate its roughness. The points were evenly
fore, we tested 20 points on the surface to evaluate its roughness. The points were evenl distributed
on the surfaceon
distributed of the
the thermoplastic
surface of themandrel, as illustrated
thermoplastic in Figure
mandrel, 5.
as illustrated in Figure 5.

Figure5.5.Roughness
Figure Roughness measurement
measurement process
process of area
of each eachofarea of mandrel
mandrel surface. surface.

The
Thesurface
surfaceroughness
roughnessof the
of thermoplastic mandrel
the thermoplastic was tested,
mandrel and the and
was tested, results
theofresults o
each
eacharea
areaare shown
are shownin Table 3. It 3.
in Table can It be
canseen from the
be seen fromresults that, after
the results theafter
that, mandrel was
the mandrel wa
attached with Teflon film, the roughness of the forming surface was less than Ra 0.5 µm.
attached with Teflon film, the roughness of the forming surface was less than Ra 0.5 μm
According to the definition and classification of roughness in the ISO 1302, the surface
According to the definition and classification of roughness in the ISO 1302, the surfac
roughness of the thermoplastic mandrel was N6 and was close to N5. It means that the
roughness
mandrel had of the degree
a high thermoplastic
of surfacemandrel was N6 and was close to N5. It means that th
smoothness.
mandrel had a high degree of surface smoothness.
Table 3. Measurement results of mandrel surface roughness.
Table 3. Measurement results of mandrel surface roughness.
Point 1 2 3 4 5 6 7 8 9 10
Point
Surface Roughness Ra (µm) 0.35 0.2 0.27 1 0.26 2 0.37 3 4
0.17 5
0.25 60.29 7 0.28 8 0.27 9 10
Surface Roughness Ra
Point 11
(μm) 12 13
0.35 14
0.2 150.27 0.26
16
0.37
17
0.1718
0.25 19
0.29 200.28 0.27
Point 11 12 13 14 15 16 17 18 19 20
Surface Roughness Ra (µm) 0.33 0.19 0.23 0.21 0.31 0.19 0.22 0.17 0.24 0.33
Surface Roughness Ra (μm) 0.33 0.19 0.23 0.21 0.31 0.19 0.22 0.17 0.24 0.33

3.1.3. Reusability
3.1.3. Reusability
In this experiment, we used the reused mandrel to manufacture composite parts and
In this experiment, we used the reused mandrel to manufacture composite parts an
compare the parts’ quality to evaluate the reusability of the mandrel. In this study, the
compare
mandrel wasthe parts’
tested forquality
20 timesto byevaluate the reusability
the processes, of the mandrel.
including softening, deforming,Incooling
this study, th
down, hardening, and heating back. As shown in Figure 6 below, the shape after the first, coolin
mandrel was tested for 20 times by the processes, including softening, deforming,
down,
10th, andhardening, and heating
20th deformations back. As shown in Figure 6 below, the shape after the firs
is shown.
10th,Asand 20thindeformations
shown the Figure 7, we isused
shown.
the return mold to return the thermoplastic mandrel
to the As
original
shown in the Figure 7, we used the the
design shape and manufactured composite
return mold toparts again.
return theBy means of
thermoplastic man
digital
drel tomeasurement,
the original we measured
design shapethe
andprofile error of the composite
manufactured parts parts
the composite manufactured
again. By mean
by the mandrel for the first time and composite parts manufactured after 20 times, to
of digital measurement, we measured the profile error of the composite parts manufac
determine the reusability of the mandrel.
tured by the mandrel for the first time and composite parts manufactured after 20 time
The equipment used in digital measurement experiments was a Blue Light 3D Scanner
to determine
(COMET theSteinbichler
L3D-5M, reusabilityOptotechnik
of the mandrel.
GmbH, Neubeuern, Germany). After spraying
The equipment
the developer, used
the complex in digital
shape measurement
composite componentsexperiments
were fixed onwas a Blue
the black Light 3D Scan
platform,
nerthe
and (COMET L3D-5M,
3D scanner Steinbichler
was used to scan theOptotechnik GmbH, Germany).
composite components at a certainAfter spraying the de
distance.
veloper, the complex shape composite components were fixed on the black platform, an
the 3D scanner was used to scan the composite components at a certain distance.
 Aerospace
Aerospace 2021, 8,FOR
8, x8,
2021, 399 PEER REVIEW 7 of 12 7 of7 12
Aerospace 2021, x FOR PEER REVIEW of 1

(a)(a) (b)(b)

(c)(c)
Figure 6. 6.
Figure
Figure Shape ofof
Shape
6. Shape mandrel
of mandrel
mandrel after
after deformations:
deformations:
after (a)(a)
deformations: Mandrel
Mandrel
(a) after
after
Mandrel the thethe
first
after first deformation;
deformation;
first (b)(b)
Mandrel
(b) Mandrel
deformation; after the after
Mandrel 10th thethe
10th
deforma-
after defor-
10th defor-
mation;
tion; (c)
(c) Mandrel
Mandrel after
after the
the 20th
20th deformation.
deformation.
mation; (c) Mandrel after the 20th deformation.

Figure
Figure7.7.Parts
Figure formed
7.Parts
Parts formedfor
formed the
for
for first
the
the time
first
firsttimebyby
time the
by mandrel
the
the and
mandrel
mandrel andparts
andpartsprepared
partspreparedafter
preparedafter202020
after times
timesofofof
times mandrel
mandr
deformation.
mandrel deformation.
deformation.

Through measurement,
Through measurement, we wewegot the
gotgot point
thethe
pointcloud
cloud ofofthe physical
thethe
physical surface
surface of of
thethe
part,
part, and
Through measurement, point cloud of physical surface of the part, an
and
then then generated
generated the the surface.
surface. Then,Then,
we we compared
compared the the physical
physical digital
digital model
model ofof the
the compo-
then generated
component thedesign
with its surface.
modelThen, we compared
in Geomagic controlthesoftware,
physicalwith
digitalthe model
operation of the
of compo
nent with itsits
design model in in
Geomagic control software, with thetheoperation of of
optimal
optimal alignment. After the two digital models were aligned, we ran the 3D comparison optim
nent with design model Geomagic control software, with operation
alignment.
module to After
alignment. After
check the
the two
the two
errordigital
the models
ofdigital models
surface were
were
along aligned,
aligned,
the normalweweran thethe
ran
direction. 3D 3Dcomparisondigitalmodule
comparison
Through modu
tomeasurement
check the error
to check the and of the
errorGeomagicsurface
of the surface along
Controlalongthe normal
the normal
comparison, direction.
direction.
we verified Through
theThrough digital measurement
digitalofmeasuremen
shape accuracy the
andandGeomagic
Geomagic Control
Control comparison,
comparison, wewe verified
verifiedthethe
shape
shapeaccuracy
accuracy of of
thethe
twotwoparts, and
parts, an
thethe
results are shown in Figure 8. The Figure 8a shows the comparison
results are shown in Figure 8. The Figure 8a shows the comparison results of the fir results of the first
composite
composite parts manufactured
parts manufactured bybythethe
mandrel,
mandrel, and
andthethe
Figure
Figure8b8bshows
shows thethe
comparison
compariso
results of the composite parts prepared after 20 times of mandrel
results of the composite parts prepared after 20 times of mandrel deformation. deformation.
Aerospace 2021, 8, 399 8 of 12

two parts, and the results are shown in Figure 8.

Aerospace 2021, 8, x FOR PEER REVIEW The Figure 8a shows the comparison results
8 of 12
of the first composite parts manufactured by the mandrel, and the Figure 8b shows the
comparison results of the composite parts prepared after 20 times of mandrel deformation.

(a) (b)
Figure
Figure 8.
8. Comparison
Comparison results
results between
between composite
composite parts
parts and
and design
design digital
digital simulation:
simulation: (a)
(a) Composite
Composite part
part first
first formed
formed by
by
mandrel;
mandrel; (b)
(b) Composite
Composite part
part prepared after 20
prepared after 20 times
times of
of mandrel
mandrel deformation.
deformation.

ItIt can
can be
be seen
seen that
that after
after the
the mandrel
mandrel waswas used
used for
for 20
20 times,
times, thethe internal
internal error
error of
of the
the
design
design digital
digital mold
mold (yellow
(yellow part)
part) appeared
appeared in in the
the middle
middle of of the
the outer
outer surface
surface and the
curved
curved parts
parts on
on both
both sides.
sides. ItItshowed
showedthat,that,after
afterthe
the core
core mold
mold waswas reused
reused over
over 2020 times,
times,
the
the mandrel may may not
notbebeable
abletotoreturn
returntotothethe design
design profile.
profile. TheThe
reasonreason is that
is that afterafter the
the core
core mold was reused many times, since the core mold material
mold was reused many times, since the core mold material underwent many transitions underwent many transi-
tions
betweenbetween the glass
the glass statethe
state and and the rubber
rubber state, state, a certain
a certain residual
residual stressstress may accumulate
may accumulate in the
in the middle
middle of thesurface
of the outer outer surface
and theand the curved
curved parts onparts
bothon both
sides, sides,prevents
which which prevents
the mandrelthe
mandrel from returning
from returning to the original
to the original designwhen
design profile profile
thewhen the mandrel
mandrel is heated.is heated.
However, How-the
ever,
overallthemolding
overall molding
error canerror can be controlled
be controlled within 0.3 within 0.3 mm
mm after after 20indicating
20 times, times, indicating
that the
mandrel
that has good
the mandrel reusability
has and canand
good reusability guarantee the forming
can guarantee accuracy
the forming of partsofwithin
accuracy parts
20 times.
within 20 times.

3.2. Performance
3.2. Performance Test
Test of
of Composite
Composite Parts
Parts
In this
In this section,
section, we
we mainly
mainly tested
tested the
the surface
surface accuracy
accuracy and
and internal
internal porosity
porosity of
of the
the
formed composite parts and evaluated the applicability of the thermoplastic
formed composite parts and evaluated the applicability of the thermoplastic mandrel by mandrel by
testing the quality of the parts.
testing the quality of the parts.
3.2.1. Profile Accuracy
3.2.1. Profile Accuracy
The profile accuracy of composite parts was mainly determined by digital measure-
The profile accuracy of composite parts was mainly determined by digital measure-
ment experiments, and the specific process of the experiment and the equipment used were
ment experiments, and the specific process of the experiment and the equipment used
introduced in Section 3.1.3. It was compared experimentally with the profile error between
were introduced in Section 3.1.3. It was compared experimentally with the profile error
the composite part profile formed by the mandrel and the composite part design profile.
between the composite part profile formed by the mandrel and the composite part design
The visual error distribution diagram and error statistics curve of each point on the part
profile. The visual error distribution diagram and error statistics curve of each point on
are shown in Figure 9.
the part arebe
It can shown
seen fromin Figure
Figure9. 9a that most of the errors of the formed parts were within
It can
0.3 mm, be seen
which from Figure
is marked 9a that
in green, andmost of the
part of the error
errorswas
of the formed
0.3–0.5 mm,parts
which were within
is marked
0.3 mm, which is marked in green, and part of the error was 0.3–0.5 mm, which
with yellow, concentrated on both sides of the “C”-shaped feature of the part. The specific is marked
with
erroryellow, concentrated
distribution can be seen on from
both Figure
sides of9b.
theThe
“C”-shaped feature
average error wasof themm,
0.04 part. The
the specific
maximum
error
error was 0.59 mm, and the points with an error within 0.3 mm accounted for moremaxi-
distribution can be seen from Figure 9b. The average error was 0.04 mm, the than
mum
90%. error
As thewaserror
0.59 value
mm, and the points
increased, thewith an error of
probability within 0.3 mm accounted
the remaining for more
errors gradually
than 90%. As
decreased andthetheyerror value
were increased,
nearly the probability
symmetrically of the
distributed remaining
on both errors
sides of gradually
the error mean.
decreased and they were nearly symmetrically distributed on both sides of
The results showed that the shape accuracy of composite parts formed by this thermoplastic the error mean.
The results showed that the shape accuracy of composite parts formed by this thermo-
plastic mandrel was high. The mandrel can effectively solve the demolding problem of
composite parts and form composite parts with a complex structure.
Aerospace 2021, 8, 399 9 of 12

Aerospace 2021, 8, x FOR PEER REVIEW 9 of 12

mandrel was high. The mandrel can effectively solve the demolding problem of composite
parts and form composite parts with a complex structure.

(a) (b)
Figure 9. Manufacturing
Figure 9. Manufacturing error
error of
of parts:
parts: (a)
(a) Error
Error distribution
distribution diagram
diagram of
of digital analog comparison
digital analog comparison between
between parts
parts and
and
design;
design; (b)
(b) error statistics diagram
error statistics diagram of
of digital
digital analog
analog comparison
comparison between
between parts
parts and
and design.
design.

3.2.2.
3.2.2. Internal
Internal Porosity
Porosity
The
The internal porosity
internal porosity ofof composite
composite parts
parts was
was tested
tested by
by industrial
industrial CT
CT (RX
(RX solutions,
solutions,
EasyTom150Micro,
EasyTom150Micro, FR). The ray beam passed through the parts, and different results
FR). The ray beam passed through the parts, and different results were
were
obtained
obtained according
according to
to the
the material
material distribution
distribution inside
inside the
the parts.
parts. According
According to to this
this result,
result,
the
the CT
CT scanning
scanning model
model of
of the
the part
part was
was reconstructed,
reconstructed, and
and the
the defect
defect information
information such
such as
as
porosity was obtained. The porosity calculation formula of composite
porosity was obtained. The porosity calculation formula of composite parts is: parts is:
Porosity = Defect Volume/Material Volume (1)
Porosity = Defect Volume/Material Volume (1)
Due to the large volume of the composite parts, they were evenly divided into upper
Due toparts
and lower the large volume ofCT
for industrial the measurement.
composite parts, they
The were evenly
specific divided
detection into upper
parameters are
and lower
shown parts4.for
in Table industrial
Then, CT measurement.
the scanning The specific
results were processed detection
by the parameters
instrument are
supporting
shown in Table 4. Then, the scanning results were processed by
software, and the porosity information of the parts was obtained. the instrument supporting
software, and the porosity information of the parts was obtained.
Table 4. Industrial CT Detection Parameters.
Table 4. Industrial CT Detection Parameters.
Test Parameters Voltage Electric Current Detector Pixel Projection Number Voxel Scan Duration
Test Figure
Parameters Voltage
100 kV Electric
500 μACurrent Detector
127 μmPixel Projection Number
1440 Voxel
95 μm Scan
56 Duration
min
Figure 100 kV 500 µA 127 µm 1440 95 µm 56 min
Figure 10 shows the distribution of internal pores in the upper and lower parts of the
composite part measured by industrial CT. It can be seen that the internal pores of the
Figure 10 shows the distribution of internal pores in the upper and lower parts of
parts were mainly distributed in the middle of the parts. The maximum pore volume in
the composite part measured3by industrial CT. It can be seen that the internal pores of the
the upper part was 22.46 mm and that in the lower part was 4.55 mm3.
parts were mainly distributed in the middle of the parts. The maximum pore volume in
the upper part was 22.46 mm3 and that in the lower part was 4.55 mm3 .
The internal porosity of the upper half and the lower half of the composite part is
shown in Table 5. The volume of the upper half was 94,798.1093 mm3 and the pore volume
was 683.0574 mm3 . The volume of the lower half was 82,698.5078 mm3 and the pore
volume was 511.0985 mm3 . It was calculated that the porosity of the upper half and the
lower half of the composite parts was 0.71% and 0.61%.

(a) (b)
Figure 10. The distribution of internal pores in the composite part measured by industrial CT: (a) The upper parts; (b) The
lower parts.
 Test Parameters Voltage Electric Current Detector Pixel Projection Number Voxel Scan Duration
Figure 100 kV 500 μA 127 μm 1440 95 μm 56 min

Aerospace 2021, 8, 399
Figure 10 shows the distribution of internal pores in the upper and lower parts of the
composite part measured by industrial CT. It can be seen that the internal pores 10 of ofthe
12
parts were mainly distributed in the middle of the parts. The maximum pore volume in
the upper part was 22.46 mm3 and that in the lower part was 4.55 mm3.

(a) (b)
Figure
Figure 10.
10. The
The distribution
distribution of
of internal
internal pores
pores in
in the
the composite
composite part
part measured
measured by
by industrial
industrial CT:
CT: (a)
(a) The
The upper
upper parts;
parts; (b)
(b) The
The
lower parts.
lower parts.
Table 5. The internal porosity of the upper half and the lower half of the composite part.

Test Area Pore Volume (mm3 ) Material Volume (mm3 ) Internal Porosity
Upper half of part 683.0574 94,798.1093 0.72%
Lower half of part 511.0985 82,698.5078 0.61%

When we used the mandrel to form parts, the internal and external air pressure of the
mandrel was balanced, so the shape of the parts was not extruded. However, through the
analysis of the distribution of porosity, it was found that when the part had a free-form
surface with large area and small curvature, the compaction effect of thermoplastic mandrel
(compared with metal mold) on the prepreg was limited, so there were more fine pores in
the middle of the part. Through analyzing the maximum pore volume and porosity, it was
seen that the molding internal quality of the lower half was higher than the upper half, but
the overall internal porosity of the parts was still far lower than the requirement of 1.5%
porosity in the general composite parts’ manufacturing standard.

4. Conclusions
This paper made a thermoplastic mandrel with heat-softening ability. The mandrel
used carbon fiber as the reinforcement material, PMMA and epoxy resin liquid nitrile as
the matrix material, and LNBR was added to make the thermoplastic mandrel heat and
soften at a certain temperature. Through measurement and experiment on the properties
of the thermoplastic mandrel and composite parts manufactured by the mandrel, it was
concluded as follows:
• The thermoplastic mandrel made in this paper had suitable glass transition tem-
perature, good surface roughness, and the ability to be reused. According to the
experiments of the thermoplastic mandrel, the glass transition temperature of the
mandrel was between 80 ◦ C and 90 ◦ C, the surfaces roughness of the mandrel was
below Ra 0.5 µm, and the mandrel can be reused more than 20 times, because of the
parts manufactured by the mandrel still maintaining high accuracy after the mandrel
was reused for 20 times.
• The thermoplastic mandrel can effectively solve the demolding problem and the
composite parts manufactured by the mandrel have low profile error and internal
porosity. According to the experiments of composite parts, composite parts with
complex structure could be manufactured by the mandrel successfully. The average
error of the molded part is about 0.04 mm and the average porosity of the upper and
lower halves is 0.72% and 0.61%.
Aerospace 2021, 8, 399 11 of 12

The thermoplastic mandrel studied in this paper has suitable Tg, smooth surface
roughness, excellent recycling characteristics, and good part-forming performance. There-
fore, it can be used for the forming of composite components with complex structure, and
makes the application range of carbon fiber composites more extensive.

Author Contributions: Methodology, S.C. and X.J.; Field Investigation, S.C., X.J. and F.X.; Data
curation, S.C., X.J. and J.A.; Writing—Original Draft Preparation, S.C., C.Z. and J.A.; Writing—
Review and Editing, X.J. and C.Z. All authors have read and agreed to the published version of
the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are contained within the article.
Conflicts of Interest: The authors have no conflict of interest.

References
1. Ahmed, S.; Thostenson, E.T.; Schumacher, T.; Doshi, S.M.; McConnell, J.R. Integration of carbon nanotube sensing skins and
carbon fiber composites for monitoring and structural repair of fatigue cracked metal structures. Compos. Struct. 2018, 203,
182–192. [CrossRef]
2. Reis, V.; Opelt, C.; Cândido, G.; Rezende, M.; Donadon, M. Effect of fiber orientation on the compressive response of plain weave
carbon fiber/epoxy composites submitted to high strain rates. Compos. Struct. 2018, 203, 952–959. [CrossRef]
3. Radue, M.; Odegard, G. Multiscale modeling of carbon fiber/carbon nanotube/epoxy hybrid composites: Comparison of epoxy
matrices. Compos. Sci. Technol. 2018, 166, 20–26. [CrossRef]
4. Chung, K.-C.; Shu, M.-H.; Wang, Y.-C.; Huang, J.-C.; Lau, E.M. 3D printing technologies applied to the manufacturing of aircraft
components. Mod. Phys. Lett. B 2020, 34, 2040018. [CrossRef]
5. Abedi, M.M.; Nedoushan, R.J.; Sheikhzadeh, M.; Yu, W.-R. The crashworthiness performance of thin-walled ultralight braided
lattice composite columns: Experimental and finite element study. Compos. Part B Eng. 2020, 202, 108413. [CrossRef]
6. Guo, K.; Wen, L.; Xiao, J.; Lei, M.; Wang, S.; Zhang, C.; Hou, X. Design of winding pattern of filament-wound composite pressure
vessel with unequal openings based on non-geodesics. J. Eng. Fibers Fabr. 2020, 15, 15. [CrossRef]
7. Zu, L.; Xu, H.; Jia, X.; Zhang, Q.; Wang, H.; Zhang, B. Winding path design based on mandrel profile updates of composite
pressure vessels. Compos. Struct. 2019, 235, 111766. [CrossRef]
8. Wang, R.; Jiao, W.; Liu, W.; Yang, F. A new method for predicting dome thickness of composite pressure vessels. J. Reinf. Plast.
Compos. 2010, 29, 3345–3352. [CrossRef]
9. Brodsjo, A.; Dekker, J. An Automated and Digital Approach to Manufacture Complex, One-off Composite Structures. Sampe J.
2020, 56, 34–42.
10. Fal, M.; Abutunis, A.; Chandrashekhara, K.; Dhaliwal, G.S. Investigation of laminate debonding in horizontal axis water turbine
composite blades. J. Renew. Sustain. Energy 2020, 12, 044501. [CrossRef]
11. Ni, X.; Liao, C.; Li, Y.; Zhang, Z.; Sun, M.; Chai, H.; Wu, H.; Jiang, S. Experimental study of multi-stable morphing structures
actuated by pneumatic actuation. Int. J. Adv. Manuf. Technol. 2020, 108, 1203–1216. [CrossRef]
12. Engelhardt, R.; Irmanputra, R.; Brath, K.; Aufenanger, N.; Drechsler, K. Thermoset Prepreg Compaction during Automated Fiber
Placement and Vacuum Debulking. Procedia CIRP 2019, 85, 153–158. [CrossRef]
13. Forcellese, A.; Simoncini, M.; Vita, A.; Di Pompeo, V. 3D printing and testing of composite isogrid structures. Int. J. Adv. Manuf.
Technol. 2020, 109, 1881–1893. [CrossRef]
14. Hao, W.; Liu, Y.; Zhou, H.; Chen, H.; Fang, D. Preparation and characterization of 3D printed continuous carbon fiber reinforced
thermosetting composites. Polym. Test. 2018, 65, 29–34. [CrossRef]
15. Lombardi, J.L.; Vaidyanathan, K.; Artz, G.; Gillespie, J.; Yarlagadda, S. A water soluble mandrel material for fabricating complex
polymer composite components. In Proceedings of the SAMPE 2001, Long Beach, CA, USA, 6–10 May 2001; Volume 46,
pp. 1316–1319.
16. Vaidyanathan, R.K.; Campbell, J.; Artz, G.; Yarlagadda, S.; Gillespie, J.W., Jr.; Dunaj, D.; Guest, B.; Nesmith, K.L. Water soluble
tooling materials for complex polymer composite components and honeycombs. SAMPE J. 2003, 39, 22–33.
17. Mohammadi, M.; Sadeghi, A. The compressive behavior of a composite cylindrical pyramidal lattice structure manufactured
with a new production methodology. Proc. Inst. Mech. Eng. Part E J. Process. Mech. Eng. 2020, 234, 222–231. [CrossRef]
18. Xie, P.; Zhu, S.; Shao, Y.; Peng, W.; Zhan, L.; Li, S. Simulation and experimental analysis of autoclave co-curing CFRP hat-stiffened
panels with silicone airbag mandrels. Iran. Polym. J. 2019, 28, 505–514. [CrossRef]
19. Li, S.; Zhan, L.; Chang, T. Numerical simulation and experimental studies of mandrel effect on flow-compaction behavior of
CFRP hat-shaped structure during curing process. Arch. Civ. Mech. Eng. 2018, 18, 1386–1400. [CrossRef]
Aerospace 2021, 8, 399 12 of 12

20. Du, H.; Liu, L.; Leng, J.; Peng, H.-X.; Scarpa, F.; Liu, Y. Shape memory polymer S-shaped mandrel for composite air duct
manufacturing. Compos. Struct. 2015, 133, 930–938. [CrossRef]
21. Du, H.; Liu, L.; Zhang, F.; Leng, J.; Liu, Y. Shape retainability and reusability investigation of bottle-shaped SMP mandrel. Polym.
Test. 2018, 69, 325–331. [CrossRef]
22. Jang, J.H.; Bin Hong, S.; Kim, J.-G.; Goo, N.S.; Lee, H.; Yu, W.-R. Long-term properties of carbon fiber-reinforced shape memory
epoxy/polymer composites exposed to vacuum and ultraviolet radiation. Smart Mater. Struct. 2019, 28, 115013. [CrossRef]
23. Yan, Y.; Zhou, P.; Wang, H.; Mao, Y. Thermal Effect on Poly(methyl methacrylate) (PMMA) Material Removal in the Micromilling
Process. Polymers 2020, 12, 2122. [CrossRef] [PubMed]
24. Liu, H.; Li, J.; Gao, X.; Deng, B.; Huang, G. Double network epoxies with simultaneous high mechanical property and shape
memory performance. J. Polym. Res. 2018, 25, 1–12. [CrossRef]
25. Zhou, B.; Liu, Y.-J.; Lan, X.; Leng, J.-S.; Yoon, S.-H. A glass transition model for shape memory polymer and its composite. Int. J.
Mod. Phys. B 2009, 23, 1248–1253. [CrossRef]
26. Rasa, K.M.; Aycan, O.O.; Holger, S. Characterization of shape memory polymer estane by means of dynamic mechanical thermal
analysis technique. Smart Mater. Res. 2014, 2014, 250258.