Three-dimensional preforming via wet-laid nonwoven technology for ceramic matrix composites

Full text

Journal of the European Ceramic Society xxx (xxxx) xxx
Please cite this article as: Fiona Kessel et al., Journal of the European Ceramic Society, https://doi.org/10.1016/j.jeurceramsoc.2023.04.062
Available online 29 April 2023
0955-2219/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Three-dimensional preforming via wet-laid nonwoven technology for
ceramic matrix composites
Fiona Kessel
a
,
*
, Martin Frieß
a
, Oliver Hohn
b
, Linda Klopsch
a
, Charlotte Z¨
ollner
a
, Cora Dirks
a
,
Matthias Scheiffele
a
, Felix Vogel
a
, Raouf Jemmali
a
a
Institute of Structure and Design, German Aerospace Center e.V., Stuttgart, Germany
b
Institute of Aerodynamics and Flow Technology, German Aerospace Center e.V., Cologne, Germany
ARTICLE INFO
Keywords:
3D Preforming
Ceramic matrix composites (CMC)
Wet-laid nonwoven
Wind tunnel test
Liquid silicon inltration (LSI)
ABSTRACT
In this study, a new 3D preforming method was developed using wet-laid nonwoven technology, for application
in manufacturing ceramic matrix composites (CMC). For this purpose, a process setup was developed and tested
on an example geometry (radome). HTS 45 carbon bers and Nextel610 alumina bers were used for the pre-
forming. The resulting C/C-SiC and OXIPOL materials were mechanically characterized and the microstructure
was investigated. A radome was manufactured from each material and subjected to DLR’s L2K and VMK wind
tunnels. The tests have been successful with the C/C-SiC and OXIPOL radome. Overall the application-oriented
tests show that load-bearing components can be produced with the newly developed preform method and that
they also prove themselves in the application. The knowledge gained, demonstrates the potential of the 3D wet-
laid nonwoven preforming method and represents a new possibility for CMC production with complex shapes.
1. Introduction
Fiber-reinforced composites are a worldwide used group of materials
for applications that demand fracture toughness and high temperature
stability. While it is relatively simple to vary the reinforcing ber types
or switch to different matrix systems it remains difcult to realize
complex preforms which follow the component shape [1–3]. In most
cases the ber reinforcement is aimed to enhance the mechanical sta-
bility and toughness of the component by following its outlines. In
addition, the need for an accurate preform increases dramatically when
using expensive raw material, since waste from the manufacturing needs
to be reduced to a minimum for economic and ecological reasons [4].
For ceramic matrix composites (CMC) those criteria are especially true,
since the material is very costly and used in some of the most challenging
environments. These environments demand superior material properties
(e.g. carbon ber reinforced silicon carbide (C/C-SiC) for rocket nozzles,
silicon carbide ber reinforced silicon carbide (SiC/SiC) turbine shrouds
for jet engines) [5–7]. The group of radomes is one such component
category where CMCs are a favorable material due to its excellent per-
formance. However, accurate preforming is quite difcult to realize for
the complex shape of those structures. For ight stability and reduced air
friction, the tip geometry is best designed pointed, leading ogive-shaped
to the main body. However, this geometry causes high temperatures at
the very tip that range from 900 ◦C to 2000 ◦C depending on the tra-
jectory [6]. The creation of such ogive-shaped preforms suitable for
radome structures is a daunting challenge and the available textile
technologies often come with downsides for the structure.
Frieß et al. demonstrated the creation of a ogive-shaped radar
transparent radomes with different preforming methods and highlighted
some of the difculties in their manufacturing [8]. One radome was
made of several layers of woven fabric, whereby the single layers where
tailored to achieve the desired shape. In consequence the fabrics were
cut to be draped and had multiple cutting edges in every single preform
layer, leading to an interrupted reinforcement. With the woven preform
style manual draping was involved and cutting waste could not be
avoided. The second method was lament winding. While the base
shape of the radome could be produced, the very tip remained difcult.
The laments slipped on the small radius and limited therefore the
shaping possibilities. These difculties were counteracted by placing
two radome-shaped winding body’s opposite each other and connecting
them with a small bar, which allowed a non-slipping winding process.
However, the tip needed to be replaced and was manufactured sepa-
rately by routing it out of a plate of thick CMC material. Later, the two
components were joined to obtain the nal product [8].
* Corresponding author.
E-mail address: ona.kessel@dlr.de (F. Kessel).
Contents lists available at ScienceDirect
Journal of the European Ceramic Society
journal homepage: www.elsevier.com/locate/jeurceramsoc
https://doi.org/10.1016/j.jeurceramsoc.2023.04.062
Received 6 December 2022; Received in revised form 24 April 2023; Accepted 28 April 2023

Journal of the European Ceramic Society xxx (xxxx) xxx
2
Another interesting technology for cone like preforms was described
by Chen at al. which focuses on nonwoven preforms via three-
dimensional needle-punching. They described the newest possibilities
of the needle-punching technology as well as microstructure and me-
chanical properties of C/SiC material [9,10]. Via winding of small
nonwoven tapes on a cone-shaped core a three dimensional preform was
obtained. During the winding process the tapes were needle-punched
and thus the overlapping tapes consolidated. This process evolved in
to robotic assisted needle punching, whereby a single or very few nee-
dles are placed on a robotic arm with ve axes. The arm is then capable
to consolidate a ber eece on a stitching core with the shape of the
desired 3D preform. Still so far, no cone shaped, 3D needle-punched
CMC was manufactured. One possible explanation could be the gener-
ally lower compressibility of needled nonwovens compared to e.g.
woven fabrics caused by bers oriented in the direction of the
compression force (z-direction) [11,12]. The result is a lower ber
volume fraction (FVF) in the nonwoven and a CMC with low fracture
toughness.
Regarding new methods with potential for three-dimensional pre-
forming, a type of nonwovens was investigated in previous work which
offer such possibilities. It is the sub-group of wet-laid nonwoven which
offer a wide range of shaping possibilities due to its unique
manufacturing process and the resulting textile has very few bers in
thickness direction which allows a higher FVF than regular nonwovens.
It was found, that depending on the ber preparation during the fabric
manufacturing, short ber reinforced ceramics as well as monolithic
ceramics are obtainable [13]. The manufacturing of the wet-laid
nonwoven ceramics is summarized in Fig. 1 and shows their two
possible microstructures. During the former studies, an interesting
possibility was found for manufacturing not only at nonwoven fabrics,
but as well to manufacture three dimensional preforms. Originally the
process derived from the paper-manufacturing technology. Here the
process consists of dispersing a cellulose ber mass in water which is
dewatered over a screen resulting in paper or likewise in a wet- laid
nonwoven fabric, (analog to the scheme in Fig. 1). The same process is
applied to receive three dimensional shaped papers or cartonnage. The
commonly best-known example for a product based on this process are
egg trays as well as other tailored product packaging [14]. These carton
packaging have astonishing features like a very versatile shape with
small radii and steep angles between neighboring areas, features that are
desired for textile preforms targeting complex composite components. In
addition, the manufacturing process is very exible, and can be adapted
to different component shapes and can be produced in large quantities in
a short time [15].
Aiming at the three-dimensional preforming of bers, the experience
of the previous work with wet-laid nonwovens was combined with the
technology of the cartonnage shaping to develop such a process. The
technology was developed to target as an example the shape of radome
structures and to evaluate its potential for crucial parts like the tip and
steep angles. A simple laboratory set up was designed to manufacture
the preforms. However, due to the novelty of the 3D-process funda-
mentals regarding control over ber behavior were assessed as well.
This includes the creation of a C/C-SiC at sample and a radome for
characterization, followed by a transition to Nextel 610 alumina bers
(3 M), which are of interest due to its radar transparency. The C/C-SiC
components were manufactured via the liquid silicon inltration (LSI-
process) whereas the Nextel 610 preform was processed via polymer
inltration and pyrolysis (PIP-process) to yield a ceramic material called
OXIPOL. The alumina ber demonstrator was tested in the two wind
tunnels at DLR Cologne (Arc-heated wind tunnel 2 (L2K) and Vertical test
section Cologne (VMK)) to evaluate the material for the radome appli-
cation. The tests showed the impact on the wet-laid nonwoven based
radomes and their load-bearing capacity. For further insights computed
tomography was used to monitor defects and changes in the structures.
2. Material and methods
2.1. Design of the 3D wet-laid nonwoven process
In order to implement the idea of producing wet-laid nonwovens in a
3D process, a laboratory setup was designed with which the textile
preforms could be realized (see Fig. 2a). This consists of a container for a
ber/water suspension, a connected pump circuit and a positive mold
(core) for the 3D textile. The vessel is designed in such a way that
Fig. 1. Upper part: Scheme of the wet-laid nonwoven process and possible nonwoven styles, lower part: Scheme of LSI-process and the microstructure of the resulting
ceramics based on the two wet- laid nonwoven styles [13].
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
3
different cores can be installed and thus the preforms are variable. For
the production of the preforms, the water pumping is activated and the
suspension ows to the core. In the initial setup, the water ow was
designed to form a vortex due to the radial inow, which is intended to
prevent the bers from sedimenting and to continuously ow bers past
the core (see Fig. 2b). The core was initially irregularly covered with
bers, since the suction effect is higher the closer the core openings are
to the suction point. Homogenization of the preform only occurs in the
course of the manufacturing process, when core openings that are
already covered by bers, shift the suction to less covered areas thus
experience a higher suction (see Fig. 3).
The wet-laid nonwoven system is operated with the pump AL-KO Jet
6000/5 premium from the AL-KO Geraete GmbH with a ow rate of
6 m
3
/h. The ber/water suspension is set to 60 liters and a ber con-
centration of 0.17 g/l. To keep this constant, 10 g of bers (carbon -
bers) are added to the process water every two minutes since the bers
are mostly settled on the core after that time (respectively 20 g alumina
bers because of their higher density of the bers). To consolidate the
preforms, carboxymethyl cellulose is added as a binder, which stabilizes
the preform. The ber addition is repeated until no more bers appear to
be deposited on the core. The water recirculation is then stopped and fed
into the recirculation tank. This causes the water container to run dry
and the preform to be exposed. The core with the preform can then be
removed and dried. In this work, two different core structures were used:
a cylindrical one, which yields a textile surface for sheet material, when
cut open and removed from the preform. These sheets have been used
for ceramic characterization. As the second geometry, a radome tip, was
designed to produce the CMC component for wind tunnel testing. This
geometry was used to verify how well complex designs can be imple-
mented using the wet-laid technology. All relevant manufacturing con-
ditions are summarized in Table 1.
2.2. Raw materials
According to the experimental procedure shown in Fig. 4, two ber
materials were investigated in the 3D wet laid nonwoven process:
HTS45 carbon bers (C-ber, with epoxide sizing) from Toho Tenax
Europe GmbH and Nextel 610 alumina bers (Al
2
O
3
-bers, with PVA
sizing) from 3 M Deutschland GmbH. The C-bers were used for tech-
nology testing and for the production of C/C-SiC sheets, analogous to the
two-dimensional wet-laid C/C-SiC from previous work [13]. In addition,
a radome was fabricated from C/C-SiC. Nextel 610 bers were used to
fabricate OXIPOL (oxidic CMC based on polymers) ceramics. The
ceramic ber enables the production of ber-reinforced ceramic mate-
rials that are not electrically conductive and thus radar transparent
which is required for radomes.
As an important step to make the bers processable in the 3D pre-
form process, a sizing has to be applied. Previous work has shown how
important the preservation of the ber bundles is for achieving ber
reinforcement and an acceptable ber volume fraction [13]. Up to now,
attempts have been made to preserve the ber bundle structure exclu-
sively by reducing mechanical action during the wet-laid nonwoven
process. Since the force can only be adjusted or reduced to a limited
extent within the laboratory set-up, the following investigations focused
on ber bundle retention via ber sizing. Polyvinyl alcohol (PVA) was
selected as the ber sizing (PVA, ThermoFisher Scientic). PVA has very
Fig. 2. a) 3D-wet-laid nonwoven device for the manufacturing of three dimensional preforms, b) suspension tank with indication of the direction of the current.
Fig. 3. Formation and ber buildup of the 3D preform in the wet-laid
nonwoven process.
Table 1
Process parameters for the 3D wet-laid nonwoven process.
Parameter Values
pump volume ow 6 m
3
/h
binder concentration 0.03 g/l
ber concentration 0.17 g/l
process time 10–20 min
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
4
good adhesive properties and exhibits solubility in water depending on
the degree of hydrolysis. Thus, it allows the preparation of a simple
water-based coating [16,17]. However, crucial for its suitability is that
thermal aging can increase the degree of crystallization of PVA, thus
worsening its water solubility again [18]. Based on these properties, it
was possible to design a coating concept in which the bers are rst
coated with a self-produced sizing and then it is rendered water insol-
uble by thermal aging. This prevents the bonded ber bundles from
dissolving in the wet-laid process.
According to the experimental proceeding in Fig. 4, the inuence of
the sizing was rst investigated (as received, sized, sized +thermal
treatment). For this purpose, accordingly prepared bers were dispersed
in demineralized water (200 rpm) for 5 min, at a ber/water concen-
tration of 0.167%. Frames of the dissolution process at 10 s (start) and
300 s (end). Allow qualitative comparison of ber bundle dissolution.
This basic study to preserve the ber structure is followed by the
production of the nonwovens for the mechanical characterization,
microstructure analysis and for the wind tunnel tests. This is done
accordingly as described in the section on the design of the 3D wet-laid
nonwoven process. For the production of the CMC plates and radome
components, different matrix systems are used depending on the pro-
cess. For the C/C-SiC samples produced in the LSI process, a carbon
precursor is mandatory as the matrix. A phenolic resin from Hexion
Group is used for this purpose. For the OXIPOL ceramics, a polysiloxane
from Merck KGaA is used. The material is produced via the PIP process
and the component is inltrated several times with the polymer.
2.3. Ceramic manufacturing via LSI-process
The C/C-SiC ceramics are manufactured using the LSI process as
shown in Fig. 1. It includes three process steps: First, the ceramic green
body fabrication. This involves the production of a CFRP body, which is
created by inltrating, or impregnating, a preform in a mold with resin
and then curing it. For all specimens and components, resin transfer
molding (RTM) was used as the molding method. Second, the pyrolysis
of the green body. In this process, the green body is heated to over
900 ◦C in an inert gas atmosphere and the phenolic matrix decomposes
to an amorphous carbon matrix and a crack structure forms in the C/C
body. The third and last step is the siliconization: Here, in a second high-
temperature step under vacuum, silicon granulate is melted at temper-
atures above 1420 ◦C. It then inltrates into the porous C/C body, where
the liquid silicon reacts with the amorphous carbon in the pore in-
terfaces to form silicon carbide [19,20].
2.4. Ceramic manufacturing via PIP-process
For the production of the OXIPOL ceramic, the PIP process is used [6,
8]. For the manufacturing of the ceramic green body a polysiloxane
matrix as precursor is inltrated via RTM. After shaping, an initial py-
rolysis is conducted above 1300 ◦C in inert gas atmosphere. Analogous
to the LSI process, a porous matrix is formed, in composition already
corresponding to the nal SiCO ceramic matrix. However, the porosity is
too high (approx. 30–40%) after only one inltration and pyrolysis
cycle. In order to create a sufcient bond between ber and matrix and
thus improve the structural mechanical properties, the steps of polymer
inltration and subsequent pyrolysis are repeated several times. This
densies the matrix and completes the ceramic manufacturing process
[21]. Four cycles were performed for the fabrication of the OXIPOL plate
and six cycles for the radome.
2.5. Characterization
During material production, the density and porosity of the samples
were determined in all process steps. The Archimedes method according
to DIN EN 993 −1:1995 −04 was used for this purpose. For the pro-
duction of the OXIPOL radomes, the number of cycles required to ach-
ieve a target porosity of less than 10% was determined. In order to
investigate the manufacturing results of the ceramics, the microstruc-
ture of all of them was examined. The Gemeni Ultra Plus scanning
electron microscope (SEM) from Zeiss GmbH was used for this purpose
with an AsB (Angle selective Backscattered electron) detector. Based on
the images, the phase composition was determined using the open
source software ImageJ (Version ImageJ 1.52p, Java 1.8.0_172 (64-bit))
via a gray value distribution.
To test the mechanical strength of the materials, samples (n =5)
were taken from each of the sheet materials for four-point bending tests
according to DIN-EN 658 −3:2002 −11. The samples were cut with a
diamond blade equipped circular saw in the dimensions of
10 mm ×100 mm ×3 mm to have an L/D ratio of 20 for the test setup.
A universal testing machine (Zwick 1494) was used to perform the test
with a controlled cross head speed of 1 mm/min. Samples were tested to
failure and the mean and standard deviation of the test results were
determined. The fracture toughness was investigated using single edge
notched bend testing (SENB) as proposed by Kuntz et al. [22]. Therefore,
a three-point bending test is performed on a pre-notched specimen, on a
Zwick 1494 with a controlled cross head speed of 1 mm/min. The
samples were prepared using abrasive cutting with a saw blade of
300 µm thickness (4.0 mm ×5.0 mm ×22.5 mm, pre-notch 2.0 mm)).
In addition, the interlaminar shear strength (ILSS) was tested according
to DIN EN 658–4: 2003. For this purpose, double-notched specimens
(n =5) were loaded under pressure. For specimen preparation, appro-
priate material was ground at (specimen thickness 3 mm) and then
specimens were machined out using abrasive cutting
(25 mm ×10 mm). The two large specimen sides were notched with a
penetration depth of 1.5 mm (half specimen thickness). The notches are
staggered so that there is a notch distance of 8 mm in the center of the
specimen and the specimen fails interlaminar in this area. The tests were
also performed with a Zwick 1494 and a controlled cross head speed of
1 mm/min.
In order to assess the uniformity of the textile preforms, the radomes
were examined after fabrication using computed tomography. The
measurements were conducted using a high-resolution
μ
CT-System (v|
tome|x L, GE Sensing & Inspection Technologies GmbH, Wunstorf)
consisting of a microfocus X-ray tube with a maximum accelerating
voltage of 240 kV and a 16-bit at panel detector (active area 2048 ×
2048 pixels at 200 µm per pixel). The
μ
CT scans were performed with
the X-ray parameters 200 kV/750
μ
A at an exposure time of 267 ms. A
voxel size of 170 µm could be achieved. The
μ
CT data were visualized
and analyzed with the commercial software package VGStu-dioMax 3.4
(Volume Graphics, Heidelberg). To investigate the inuence of the wind
tunnel test, the radomes were scanned again after the testing.
2.6. Wind tunnel testing
For the qualication of the materials, tests in two different wind
tunnels were conducted with radomes. The Vertical Test Section (VMK)
and the arc-heated wind tunnel L2K of DLR Cologne were used for this
analogous to previous qualication testing of high speed missile ra-
domes [6]. The facilities are described in detail by Triesch and Krohn
[23] (VMK) and Gülhan et al.[24,25]. A combined approach with testing
Fig. 4. Experimental proceeding for the creation of the radome tip for wind tunnel testing.
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
5
in two facilities is necessary to fully validate the radomes, as it is not
possible to directly replicate the ight conditions of the envisaged
missions with ight Mach numbers above Ma =4 and correspondingly
high heat and mechanical loads in only one facility. Though falling short
the real ight conditions, tests in VMK, which can achieve direct ight
conditions up to Mach 3 in an altitude of 4.5 km, give valuable infor-
mation as the radome is subject to both high aerothermal and me-
chanical loads as well as thermal shock. In L2K, on the other hand, ight
relevant heat loads can be imposed on the radome that correspond to
those a radome missile would experience during its planned mission.
Here, however, the density and pressure, and consequently the Reynolds
number, are very low and the aeromechanical loads are accordingly
smaller. The ow conditions of the tests in in L2K and VMK are listed in
Table 2 and Table 3, respectively. In L2K, three different conditions were
tested to simulate cases with a moderate heat ux and longer test
duration, resulting in a high integral heat load (condition FC1), and two
conditions with high and maximum heat uxes but short test durations,
and thus lower integral heat load but higher heat uxes (and potentially
wall temperatures) in the stagnation region of the radome. In addition to
this, the angle of attack was varied (
α
=±10◦in L2K and ±5◦in VMK),
also in combination with different roll angles (0◦and 90◦, to determine
the impact of different ight angles on the thermal behaviour and, more
importantly, an additional bending moment from the aerodynamic
forces in VMK testing. By variation of angle of attack and roll angle, the
direction of the thermal and mechanical loads is changed as well. In both
VMK- and L2K-testing, the temperature distribution on the external
surfaces was measured by infrared thermography and on the internal
wall by thermocouples. The experimental setup is the same as the one
described in [6].
3. Results
3.1. Fiber stabilization
3.1.1. Carbon-ber
The rst tests were aimed at stabilizing the ber bundles. As dis-
cussed previously, the goal is to avoid ber bundle dissolution as much
as possible in order to produce a damage tolerant short ber material. To
investigate the effect of ber sizing on ber dissolution, the bers were
tested ‘as received’, ‘sized (2% PVA)’ and ‘sized (2% PVA) +tempered’
in a laboratory test. The qualitative comparison of the dissolution tests is
shown in Fig. 5a) and shows the dissolution process at the beginning of
the process (10 s) and at the end of the process (after 5 min). Despite
identical ber consistencies, a direct difference in the ber dissolution
can already be observed in the rst 10 s. The newly sized bers tend to
separate directly at the beginning of the exposure process. By the end of
the process, this effect increases signicantly and the bers are present
as fanned-out ber ocks without a bundle structure. In the case of bers
in the as received state, the bundle structure persists longer, and even
towards the end of mixing, isolated intact ber bundles are still present.
Nevertheless, their further disintegration is likely with stronger disper-
sion power, or longer running process times, as was also seen in un-
published data. The smallest change in the bundle structure was
observed in the newly sized bers with subsequent tempering. Here, the
bundle structure remains largely intact, even over the entire dispersion
time. Based on these results, ber preparation in the form of sizing and
subsequent tempering was selected for the production of 3D wet-laid
nonwovens.
3.1.2. Alumina-ber
According to the manufacturer 3 M, the alumina ber is already
sized with PVA. However, as shown in the experiment, the bers
dissolve very quickly into individual laments on contact with water
(see Fig. 5b)). To prevent this, the as received bers were tempered as
well. The results show that, analogous to the carbon bers, the thermal
treatment has a sizing stabilizing effect. Therefore, all alumina bers
were thermal stabilized before the 3D wet-laid nonwoven process.
3.2. 3D wet-laid nonwoven process
The rst preforms produced using the new 3D process were made in
cylindrical geometry. It was found that the ow created a vortex in the
water tank as planned, but resulted in disrupted ber deposition. The
rotation of the ow led to additional forces acting on the bers, which
were to be deposited on the core. As a result, bers were partially torn
from former deposition and the ber bundles were separated too
strongly, a situation that led to the formation of ber agglomerates and
spinnings as described by Hubbe et al. for the regular wet-laid nonwoven
process [26]. To prevent this, pipe segments were inserted into the
bottom plate of the wet-laid nonwoven tank. This allowed the direction
of the water inow to be adjusted. It was found that the formation of a
vortex could be prevented when the water ow was directed directly
towards the core. However, the same unfavorable formation of a vortex
occurred again when the water inow was in any way radially directed.
Therefore, all further experiments were performed with the ow
directed towards the core, which resulted in vortex prevention and thus
homogenized the ber deposition at the same time.
3.3. Manufacturing of the preforms
After a stable process was established, the C-ber preforms were
produced rst. A total of three cylindrical preforms were manufactured
to generate sufcient ber material for the ceramic plate. The cylinders
were cut open in axial direction while still wet and unrolled to obtain a
2D sheet. The masses of the nonwovens produced are summarized in
Table 4. Secondly, the radome preform was produced with the same
machine setup, but with the radome core. Despite the tight radius in the
area of the tip, the preforms were obtained and displayed no defects (see
Fig. 6). Two preforms were needed to achieve a sufcient ber volume
fraction. For this purpose, they were stacked into each other and pro-
cessed as one preform during RTM. For the alumina bers the same
procedure was conducted and again three cylinders for the plate mate-
rial were manufactured. For the radome two preforms were sufcient.
3.4. Ceramic manufacturing
After preform fabrication, these were converted into ber-reinforced
ceramics using LSI and PIP processes. Flat samples were made from the
cylinder preforms and used for microstructure analysis and mechanical
characterization. The radomes were prepared for wind tunnel testing
Table 2
L2K test conditions.
Test condition L2K-1 L2K-2 L2K-3
mass ow [g/s] 110 110 110
hot/cold gas ratio [g/s] 50/60 60/50 65/45
reservoir pressure [hPa] 2000 2360 2570
total enthalpy [MJ/kg] 2,6 3,8 4,8
reservoir temperature [K] 2193 2905 3304
Pitot pressure [hPa] 89 136 597
heat ux [kW/m2] 945 1810 2725
test duration [s] 80 6 6
Table 3
VMK test conditions.
Test condition VMK
Mach number 3
total pressure p
t0
[MPa] 2.1
total temperature Tt0 [K] 700
unit Reynolds number Re
m∞
[10
6
/m] 47.4
test duration [s] 30
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
6
and CT scanned before and after the tests. During the process, the
porosity of the ceramic components was also measured using the
Archimedes method. The results are summarized in Table 5, which also
contains the bulk density in the nal material condition. For the C/C-SiC
samples, data acquisition was carried out to check whether process de-
viations were present. No such phenomena could be detected. For the
OXIPOL samples, there was the additional function of determining, after
how many PIP cycles a target porosity of less or equal 10% could be
achieved. For the plate material, this was already achieved after a total
of four cycles (9.47% open porosity). For the radome geometry, on the
other hand, a total of six cycles were necessary, and yet the target
porosity of less than 10% was not quite reached, but it was determined
that the value was close enough to the set target (10.58% open porosity).
3.5. Characterization
The nished CMC samples were examined under a scanning electron
microscope (SEM) with respect to their microstructure (Fig. 7). The
microstructure shown corresponds to the expectations of wet-laid
nonwoven reinforced C/C-SiC [13]. C/C blocks are formed which are
surrounded by Si-SiC matrix. The block formation does not follow a
regular pattern as it is known from fabric reinforcements, but is instead
irregular. The investigation of the phase composition via the gray scale
distribution showed 58.8% of carbon, 25.3% of SiC and 15.9% of un-
bound Si.
The OXIPOL samples show no gray scale contrast in SEM analysis.
Therefore, no distinction between ber and matrix in quantitative terms
Fig. 5. Testing of ber bundle dissolution for a) C- bers as received, sized and sized +tempered, b) alumina-bers as received and as received +tempered.
Table 4
Preform masses of the manufactured 3D preforms.
Preform 1 Preform 2 Preform 3 ∑Preform
Preform Mass (g) Grammage (g/m
2
) Mass (g) Grammage (g/m
2
) Mass (g) Grammage (g/m
2
) FVF (%)
C-ber / cylinder 49.2 984.0 28.0 560.0 29.2 584.0 40.1
C-ber / radome 53.8 1349.8 40.1 1005.8 44.4
alumina-ber / cylinder 94.0 1880.0 82.7 1654.0 114.8 2296.0 36.5
alumina-ber / radome 113.2 2840.2 92.7 2324.6 43.1
Fig. 6. 3D wet-laid nonwoven preforms manufactured in the laboratory 3D wet-laid nonwoven device.
Table 5
Open porosity of the C/C-SiC and OXIPOL samples during LSI and PIP process and density in the nal material condition.
C/C-SiC Inltration Pyrolysis Siliconization Density
plate 11.49% 24.20% 2.39% - - - - 2,06 g/cm
3
radome 7.05% 27.12% 3.17% - - - - 2,06 g/cm
3
OXIPOL inltration pyrolysis 1 pyrolysis 2 pyrolysis 3 pyrolysis 4 pyrolysis 5 pyrolysis 6
plate 10.11% 36.20% 25.91% 14.13% 9.47% - - 2,06 g/cm
3
radome 12.88% 39.41% 29.14% 23.30% 18.17% 13.62% 10.58% 2,06 g/cm
3
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
7
could be made with this method. The qualitative analysis of the ber
orientation results in an isotropic orientation. The microstructure of the
matrix features dense blocks with partially no contact with the bers due
to shrinkage during pyrolysis. The post-inltration cycles of the PIP
process ll gaps in the matrix and improve the ber matrix bonding. This
layered structure of the matrix is visible in the images. In addition, the
examination of the OXIPOL samples provided interesting insight
regarding the lament cross sections of the Nextel 610 bers. Different
ber cross sections could be detected, although the short bers all came
from the same ordering unit. It is known from studies by Pritzkow et al.
that the ber cross sections of Nextel 610 have a rather round cross
section up to a lament count in roving of 4500, but above 10,000 l-
aments the cross section changes to a kidney shape [27]. Such mixed
cross sections were observed in the short bers used (Fig. 7). Accord-
ingly, the short bers appear to be a mixture of different roving thick-
nesses, which are converted to a uniform cut length. However, it is not
possible to make any statements about the distribution of the roving
parts used, and the effects of the different ber cross sections on the
ceramic have hardly been investigated so far.
After the 4PB tests, the microstructure at the fracture edges was
examined. When analyzing the C/C-SiC samples, it was found that block
pull out is observed wherever bers lie parallel to the load direction
(bers lie parallel to the sample length) (Fig. 7). In areas where the bers
are deected from the load direction, the bers tend to break in a more
brittle manner and apparently contribute less to the fracture toughness.
In addition, the fracture edges show areas where predominantly matrix
is present and thus brittle fracture can be observed. The OXIPOL wet laid
nonwoven exhibits very similar fracture edge characteristics, but with
the difference that increased ber pull out is evident in the place of block
pull out. Analogous to the C/C-SiC specimen, there are also partial areas
with only a few bers in which the matrix is dominant. Here, however,
the matrix breaks more strongly along a large number of pre-cracks,
which can be assigned to the individual matrix layers of the PIP pro-
cess. This results in an increased amount of crack branching and thus
crack energy dispersion even within the matrix-dominated areas.
The analysis of the strength has been carried out by means of 4-point-
bending testing and is supplemented by the evaluation of the fracture
toughness by means of the SENB test and the measurement of the
interlaminar shear strength. The 4PB test shows that the exural
strength of the C/C-SiC material (67.66 ±8.38 MPa on average) is
higher than that of the short ber OXIPOL (41.50 ±5.66 MPa) (Fig. 8a
and b). The same is true for elongation, with C/C-SiC (0.18 ±0.05%)
exceeding OXIPOL (0.10 ±0.03%) almost twofold. Values of the study
are summarized in Table 6. From the SENB test, the stress intensity
factor K
I
/displacement curves were received (Fig. 8c and d). Both ma-
terials show a stepped decrease in strength after reaching a maximum
load, which can be attributed to crack deection or crack stop when
tearing further along the notch. In addition, the K
Ic
value was deter-
mined (Table 6). It allows to compare both materials with each other in
terms of their damage tolerance. For C/C-SiC a K
Ic
of 2.03 MPa*m
1/2
and
for OXIPOL of 3.55 MPa*m
1/2
was achieved, respectively. Accordingly,
the results suggest that the OXIPOL material is more damage tolerant
despite its lower strength. However, this needs to be further investigated
in future research. The results of the shear compression test show that
the C/C-SiC (14.50 MPa) material has a higher interlaminar shear
strength than the OXIPOL (9.83 MPa) variant. This is mainly due to the
SiC matrix reactively bonded with the carbon, which ensures a high
cohesion of the nonwoven layers to each other. However, the in-
homogeneity in the material results in a high standard deviation
(4.19 MPa) and causes the material to lag behind measured values for
fabric-reinforced C/C-SiC (24.8 MPa [28]). Due to the different matrix
properties, the interlaminar shear strength of OXIPOL is low. The matrix
cracks seen in Fig. 7 lead to rapid failure under predominant matrix
loading and the very low percentage of bers in the thickness direction.
Following the material analysis of the at specimens, the radomes
were prepared for the wind tunnel test. To detect irregularities or defects
in the component, the C/C-SiC and the OXIPOL radome were CT scanned
(Fig. 9). Both images show that the contours of the geometry could be
mapped, but that there is signicant inhomogeneity in the wall thick-
ness. This is most extreme in the tip. Here, residues of the matrix are
added to the material inhomogeneities. In the case of the C/C-SiC
radome, areas with delaminations can be seen below the tip, which
are presumably due to the two-layer structure of the preform. While the
OXIPOL radome shows no delamination areas, the irregularity in the tip
Fig. 7. Microstructure of the C/C-SiC and OXIPOL specimen on polished samples and fracture edges after the 4PB testing.
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
8
is extreme. In some areas, the wall thickness of the CMC is less than one
millimeter. Despite these strong variations, it was decided to test both
geometries in the wind tunnel setups to nd out, in case of failure of the
structures, where this is the case and therefore is the most critical.
3.6. Wind tunnel testing
The OXIPOL-radome successfully completed 18 test runs overall in
both wind tunnels. First, in L2K, 12 tests at all three conditions and
angles of attack up to ±10◦were conducted. In consecutive VMK
testing, another 6 tests were done with angles of attack of ±5◦and roll
angles of 0◦and 90◦. Exemplary results from infrared measurements of
the surface temperature are shown in Fig. 10, and photographs of the
radome before and after the tests are shown in Fig. 11.
As the images illustrate, the heating in L2K is much more concen-
trated on the nose region where very high temperatures of more than
1800 ◦C are reached for both low and high heat uxes. In VMK, which
produces an aerodynamically more representative ow, the heating of
the radome is much more extensive to almost the entire surface. The pre-
Fig. 8. Results of mechanical testing a), b) stress/strain curves of tested 4PB specimens, c), d) stress intensity factor K1/displacement curves of SENB samples, e), f)
shear stress/displacement curves of ILSS test.
Table 6
Mechanical properties of C/C-SiC and OXIPOL.
Stress
[MPa]
Strain
[%]
Young’s
modulus [GPa]
K
Ic
[MPa*m
1/2
]
ILSS
[MPa]
C/C-SiC 67.66
±8.38
0.18
±0.05
44.33 ±5.37 2.03 ±0.21 14.50
±4.19
OXIPOL 41.50
±5.66
0.10
±0.03
42.35 ±9.79 3.55 ±0.56 9.83
±2.22
Fig. 9. CT scans of the radomes before and after testing.
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
9
and post-test photographs showed only very little recession of the
radome nose tip. Also the VMK tests with much higher pressure and
correspondingly high shear stress on the surface did not cause obvious
removal of material.
For further post-test analysis, the radome was examined again in the
CT (Fig. 9). Opposed to the appearance in Fig. 11, the images show that
almost complete ablation has taken place in the tip area, especially
where the wall thickness before testing was only one millimeter. At this
point, the tip is almost only closed by matrix deposited on the inside.
Apart from the tip, no other areas with material removal could be
detected. This demonstrates both the strength and durability of the
material as despite the extent of ablation, a critical component failure
was prevented although the radome had been subjected multiple times
to the loads expected in its real application. On the other hand, it also
illustrates that improvements in radome manufacture are imperative,
particularly in this critical area.
The C/C-SiC radome was rst tested in VMK and tests in L2K were
only planned afterwards, simply for timely reasons. However, this
turned out to be very unfortunate, as the radome was destroyed during
the rst wind tunnel test in VMK. After it had successfully passed the
nominal test duration including data acquisition (Fig. 10), the radome
surprisingly detached from the mount during the shutdown of the test
rig. Despite the structure still being subjected to a near Mach 3 ow with
a total pressure of around 17 bar and still being placed completely inside
the Mach rhombus of the ow, the radome moved about 10 mm against
the ow direction before it was pushed back against the model holder
and broke as illustrated in Fig. 12. Supplementary investigations showed
an increase of the internal pressure in the C/C-SiC radome. An analysis
with the aid of CFD computations indicates that it might be possible that
the force induced by this internal pressure exceeds the aerodynamical
force of the oncoming ow. This force, in combination with an imperfect
test setup (glue not properly hardened/dried) could have caused the
failure of the glue connection of the radome to the model holder and
thus the radome to be destroyed. It should be noted that there is no clear
evidence to support this theory but given the circumstances, it seems to
be the most likely explanation.
4. Discussion
Based on the results, the following insights into the potential of the
3d wet-laid nonwoven process were gained. Two elementary re-
quirements were developed for the 3D wet-laid process: First, the sizing
of the ber bundles to keep them stable in the process and, second, the
ow control in the suspension vessel of the wet-laid nonwoven line. The
rst was achieved by stabilizing the bundle structure with a PVA sizing
and its thermal post-crosslinking. For the ow guidance, the original
conguration was modied with a tangentially directed inow into the
vessel. The empirical investigation of different inow angles showed
that the direct inow into the core, i.e. into the center of the container, is
Fig. 10. Exemplary IR-images at end of test time in L2K (left side) and VMK (right).
Fig. 11. OXIPOL radome before rst (left) and after last (right) test in L2K.
Fig. 12. IR-images of destruction of C/C-SiC radome during wind tunnel
shut down.
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
10
best suited for vortex free ow. As a result, ber spinning does not occur
and the ber bundles are deposited on the core as designed. Carbon and
alumina preforms were produced using this process. Fiber deposition
was successful on both, but differences can be seen in the area of the tip
(Fig. 6). The carbon bers were deposited well on the tip (tip radius
3 mm). The alumina bers, on the other hand, do not follow the outline
accurately, since they cannot bend according to the curvature due to
their high stiffness. In addition to the degree of crystallization, the
decisive factor for the high stiffness of the bers is the ber diameter
[29]. This is signicantly higher for alumina bers than for carbon bers
(alumina approx. 14 µm, carbon 6 µm). At such high stiffness, bers can
only be laid down following the contour if the ber length is shorter than
the smallest radius of the geometry. This effect increases signicantly
the more the bundle structure is retained, since this "increased" ber
diameter also increases the stiffness. In order to achieve tight component
curvatures with undissolved ber bundles in the wet-laid nonwoven
process, the bers should be signicantly shorter than the smallest
diameter, to achieve the best preform results.
The shaping of the CMC production then took place via RTM. While
at specimens were produced from the cylinders, the radomes were
manufactured according to the preform geometry. Here, the structure of
the RTM molds in particular had a strong inuence on the quality of the
tip. Fig. 9 shows strong irregularities of the wall thickness in the tip.
These cannot be attributed to the preform. The defects in the tip are due
to the current RTM setup, there preforms are slipped onto the core and
then pushed together into the negative form. However, since a large part
of the radome mold has an almost cylindrical structure, the preforms are
partially trapped between the core and the outer wall when they are
pushed in. As a result, when pushed further, the tip is partially pierced
and the ber material is not transported all the way into the tip. In
particular, since the nonwovens have low strength, they tear quickly and
puncturing becomes more likely. Here, a more suitable setup for mold-
ing needs to be found in the future to accommodate the more fragile
nonwovens compared to, for example, fabric preforms. However, all the
wet-laid nonwoven preforms could successfully be processed via LSI or
PIP process to ber reinforced ceramics, which highlights the possibil-
ities of the new preforming method.
In addition, it was not only possible to demonstrate the
manufacturing possibilities, but also to investigate the materials and
carry out initial application studies. While the results of the micro-
structure investigation show a satisfactory material composition, the
mechanical investigations and the evaluation of the fracture toughness
showed that the existing material still needs to improve signicantly in
terms of strength and fracture toughness. One reason for this is that the
load-bearing properties of a CMC material are dramatically reduced
when bers are not oriented in the load direction [29,30]. Assuming that
within wet-laid nonwovens there is an almost isotropic ber orientation
in the X/Y plane of the nonwoven, only a small proportion of these are
oriented in the load direction. One approach to improve these properties
could be to increase the ber volume content (currently approx. 40%),
since this would also increase the bers oriented in the load direction
across the material. This approach should be further investigated to
increase the mechanical properties of the materials in the future.
The wind tunnel tests have shown that load-bearing components can
also be produced with the current preform and process setup. The
durability and protection of the internal hardware can be made possible
with both material variants. While the OXIPOL radome performed well
and withstood a total of 18 tests, the C/C-SiC radome was destroyed
during the rst VMK run. As described before the radome detached itself
from the sample holder of the wind tunnel during shut down phase of the
rst run due to insufcient hardened glue. Despite the remaining aero-
dynamical force of the oncoming ow it was pushed of the holder and
after the complete detachment accelerated against the holder and
therefore destroyed. The supplementary measurement of the inner
pressure suggests, that the higher pressure inside the radome could have
further supported the detachment. One factor that might have
contributed to the pressure build up could be the low porosity of the C/
C-SiC material (3.17%) as compared to other CMC radomes previously
studied (10–20%). Thus, it has a signicantly less potential for outgas-
sing in the event of possible gas developments or internal pressure build-
up. This is an aspect that has to be investigated more closely and needs to
be carefully considered in future experimental setups or the actual use of
the material for the designated application. In addition, it has to be
considered that when using C/C-SiC, the application of an Environ-
mental Barrier Coating (EBC) would also be benecial to prevent
oxidation effects, which could also lead to failure. Depending on the
application scenario and the resulting maximum ight time, it must be
decided whether it is necessary to provide the radome with this coating.
When using C/C-SiC, however, at least the additional integration of a
radar-transparent window is necessary, in contrast to OXIPOL, which is
radar-transparent as an overall component. However, it is important to
emphasize at this point that it was not the material that led to the failure
of the component, but the connection to the test rig in the form of the
adhesive. Nevertheless, due to the loss of the C/C-SiC-sample during
test, future test campaigns have to shine light on the material’s perfor-
mance regarding, abrasion, oxidation, and material removal, as
happened with the OXIPOL radome.
5. Conclusion
The aim of the work was to nd a suitable preform method with
which complex component shapes can be produced beyond at samples.
Wet-laid nonwoven technology was identied as a potentially inter-
esting technique and a laboratory process was set up with which 3D
preforms can be implemented. The target geometry of this study was
radomes, since these cover problem areas well (small radius tip) and
there was the possibility of subjecting the CMC component to an
application-oriented test.
The developed preforming process opens up new possibilities for
component manufacture in all areas of application in which more
complex geometries have to be realized and no cost-effective preform
manufacture was previously possible. The manufacturing was explored
and suitable process settings were determined. For the low mechanical
strength of the wet-laid nonwoven the RTM green body manufacturing
needs to be adapted, to prevent the tearing of the textile. However,
further studys need to target an increase in strength of the material by
higher the FVF. Specic to the radome the wet-laid nonwoven OXIPOL
material has performed well and is of great interest for this application.
Over all, the 3D wet-laid nonwoven technology was found to be
promising for CMC manufacturing and can make an important contri-
bution to the further commercialization of CMCs, due to its low pro-
duction costs and exibility in component geometry.
CRediT authorship contribution statement
F.K. designed, conducted, and analyzed experiments and wrote the
manuscript. M.F. and O.H. designed and conducted experiments and
edited the manuscript. All remaining authors were involved in data
analysis and edited the manuscript.
Declaration of Competing Interest
The authors declare the following nancial interests/personal re-
lationships which may be considered as potential competing interests:
Patent EP3980388A1 ‘Method for producing a near-net-shape ber
body, ber body, method for producing a ceramic component, and
ceramic component’. Inventors: Fiona Kessel, Linda Klopsch, Martin
Frieß, Charlotte Z¨
ollner. Assignee: German Aerospace Center.
Data Availability
Data will be made available on request.
F. Kessel et al.

Journal of the European Ceramic Society xxx (xxxx) xxx
11
Acknowledgments
This research is funded by the German Ministry of Defence (BMVg)
and part of the DLR project ‘ITEM-FK’ and ‘FK2020+’. The authors
would like to sincerely thank, Daniel Cepli, Stefan Frick, Marco Alex-
ander Smolej, Elena Carcano and Alice Thomas at the DLR-Institute of
Structures and Design for their technical support. Sincere thanks go to
Prof. Dietmar Koch from Augsburg University for his support in data
analysis.
References
[1] A. Mountasir, M. L¨
oser, G. Hoffmann, C. Cherif, K. Großmann, 3D Woven Near-Net-
Shape Preforms for Composite Structures, Adv. Eng. Mater. vol. 18 (2016)
391–396.
[2] C. Weimer, T. Preller, P. Mitschang, K. Drechsler, Approach to net-shape
preforming using textile technologies. Part II: holes, Compos. Part A: Appl. Sci.
Manuf. vol. 31 (2000) 1269–1277.
[3] C. Weimer, T. Preller, P. Mitschang, K. Drechsler, Approach to net-shape
preforming using textile technologies. Part I: edges, Compos. Part A: Appl. Sci.
Manuf. vol. 31 (2000) 1261–1268.
[4] J. Rybicka, A. Tiwari, G. Leeke, Technology readiness level assessment of
composites recycling technologies, J. Clean. Prod. vol. 112 (2016) 1001.
[5] F. Süß, T. Schneider, M. Frieß, R. Jemmali, F. Vogel, L. Klopsch, et al., Combination
of PIP and LSI processes for SiC/SiC ceramic matrix composites, Open Ceram. vol. 5
(2021), 100056.
[6] O. Hohn, B. Esser, J. Klevanski, A. Gülhan, R. Fener, B. Panthen, et al.,
Experimental investigations for the thermal qualication of high speed missile
radomes, 8th Eur. Conf. Aeronaut. Space Sci. (EUCASS) (2019).
[7] F. Breede, S. Hofmann, N. Jain, R. Jemmali, Design, manufacture, and
characterization of a carbon ber-reinforced silicon carbide nozzle extension, Int.
J. Appl. Ceram. Technol. vol. 13 (2016) 3–16.
[8] M. Frieß and S. Denis, "Oxide CMC Components Manufactured via PIP Processing
Based on Polysiloxanes," presented at the ECerS XII, Stockholm, Schweden, 2011.
[9] J. Nie, Y. Xu, L. Zhang, L. Cheng, J. Ma, Microstructure and tensile behavior of
multiply needled C/SiC composite fabricated by chemical vapor inltration,
J. Mater. Process. Technol. vol. 209 (2009) 572–576.
[10] X. Chen, L. Chen, C. Zhang, L. Song, D. Zhang, Three-dimensional needle-punching
for composites – a review, Compos. Part A: Appl. Sci. Manuf. vol. 85 (2016) 12–30.
[11] G. Batch, S. Cumiskey, C. Macosko, Compaction of ber reinforcements, Polym.
Compos. vol. 23 (2002) 307–318.
[12] W. Albrecht, H. Erth, H. Fuchs, Vliesstoffe: Rohstoffe, Herstellung, Anwendung,
Eigenschaften, Prüfung, Wiley, 2012.
[13] F. Kessel, L. Klopsch, V. Jehle, N.-J. Biller, M. Frieß, Y. Shi, et al., Wet-laid
nonwoven based ceramic matrix composites: An innovative and highly adaptable
short ber reinforcement for ceramic hybrid and gradient materials, J. Eur. Ceram.
Soc. vol. 41 (2021) 4048–4057.
[14] R. Wever, D. Twede, The history of molded ber packaging: a 20th century pulp
story, Proc. 23rd IAPRI Symp. . Packag., Windsor, UK September 3–5 (2007).
[15] M. Onilude, T. Omoniyi, B. Akinyemi, K. Adigun, The design and fabrication of a
recycled paper egg tray machine, Int. J. Contemp. Manag. vol. Vol. 3 (2012)
142–151.
[16] F. Khalid, M. Tabish, K. Bora, Novel poly(vinyl alcohol) nanoltration membrane
modied with dopamine coated anatase TiO2 core shell nanoparticles, J. Water
Process Eng. vol. 37 (2020), 101486.
[17] B.H. Musa, N.J. Hameed, Study of the mechanical properties of polyvinyl alcohol/
starch blends, Mater. Today.: Proc. vol. 20 (2020) 439–442.
[18] K.K.H. Wong, M. Zinke-Allmang, W. Wan, Effect of annealing on aqueous stability
and elastic modulus of electrospun poly(vinyl alcohol) bers, J. Mater. Sci. vol. 45
(2010) 2456–2465.
[19] R. Kochend¨
orfer, Liquid Silicon Inltration- A Fast and Low Cost CMC-
Manufacturing Process, 1991.
[20] F.H. Gern, R. Kochend¨
orfer, Liquid silicon inltration: description of inltration
dynamics and silicon carbide formation, Compos. Part A: Appl. Sci. Manuf. vol. 28
(1997) 355–364.
[21] B. Mainzer, M. Frieß, R. Jemmali, D. Koch, Development of polyvinylsilazane-
derived ceramic matrix composites based on Tyranno SA3 bers, J. Ceram. Soc.
Jpn. vol. 124 (2016) 1035–1041.
[22] M. Kuntz, J. Horvath, G. Grathwohl, High temperature fracture toughness of a C/
SiC (CVI) composite as used for screw joints in re-entry vehicles, High. Temp.
Ceram. Matrix Compos. (2001) 469–473.
[23] E.-O.K. Klaus Triesch, The vertical test section (VMK) at DLR in Cologne-Porz vol.
Technical Translation of DFVLR-Mitt-86–22. Germany: European Space Agency,
1986.
[24] A. Gülhan, B. Esser, Arc-heated facilities as a tool to study aerothermodynamic
problems of reentry vehicles, Prog. Astronaut. Aeronaut. vol. 198 (2002) 375–403.
[25] A. Gülhan B. Esser U. Koch K. Hannemann Mars entry simulation in the arc heated
facility L2K L2K 2002.
[26] M. Hubbe, A. Koukoulas, Wet-laid nonwovens manufacture – chemical approaches
using synthetic and cellulosic bers, BioResources vol. 11 (2016) 5500–5552.
[27] W.E.C. Pritzkow, R.S.M. Almeida, L.B. Mateus, K. Tushtev, K. Rezwan, All-oxide
ceramic matrix composites (OCMC) based on low cost 3M Nextel™ 610 fabrics,
J. Eur. Ceram. Soc. vol. 41 (2021) 3177–3187.
[28] Y. Shi, F. Kessel, M. Frieß, N. Jain, K. Tushtev, Characterization and modeling of
tensile properties of continuous ber reinforced C/C-SiC composite at high
temperatures, J. Eur. Ceram. Soc. vol. 41 (2020), 10/02.
[29] Y. Shi, "Characterization and modeling of the mechanical properties of wound
oxide ceramic composites," 2017, 2017.
[30] M. Cordin, T. Bechtold, T. Pham, Effect of bre orientation on the mechanical
properties of polypropylene–lyocell composites, Cellulose vol. 25 (2018)
7197–7210.
F. Kessel et al.