materials
Article
The Mechanical Investigation of Filament-Wound CFRP
Structures Subjected to Different Cooling Rates in Terms
of Compressive Loading and Residual Stresses—An
Experimental Approach
Wojciech Błażejewski, Michał Barcikowski
Michał Stosiak and Grzegorz Lesiuk






Citation: Błażejewski, W.;
Barcikowski, M.; Lubecki, M.; Stabla,
P.; Bury, P.; Stosiak, M.; Lesiuk, G. The
Mechanical Investigation of
Filament-Wound CFRP Structures
Subjected to Different Cooling Rates
in Terms of Compressive Loading and
Residual Stresses—An Experimental
Approach. Materials 2021, 14, 1041.
https://doi.org/10.3390/ma14041041
Academic Editor: Xiaozhi Hu
Received: 21 January 2021
Accepted: 18 February 2021
Published: 22 February 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/).
Materials 2021, 14, 1041. https://doi.org/10.3390/ma14041041
, Marek Lubecki * , Paweł Stabla , Paweł Bury ,
Faculty of Mechanical Engineering, Wroclaw University of Science and Technology, PL 50-370 Wroclaw, Poland;
wojciech.blazejewski@pwr.edu.pl (W.B.); michal.barcikowski@pwr.edu.pl (M.B.); pawel.stabla@pwr.edu.pl (P.S.);
pawel.bury@pwr.edu.pl (P.B.); michal.stosiak@pwr.edu.pl (M.S.); grzegorz.lesiuk@pwr.edu.pl (G.L.)
* Correspondence: marek.lubecki@pwr.edu.pl
Abstract: Although cooling at ambient temperature is widely used and is said to be safe and
convenient, faster cooling may have an influence not only on the time of the manufacturing process
but also on the mechanical response, especially the residual stress. The study aimed to investigate the
influence of the cooling rate after curing on the mechanical response of filament-wound thick-walled
carbon fiber reinforced polymer (CFRP) rings. Three cooling rates were taking into consideration:
cooling with the oven, at room temperature, and in the water at 20 ◦ C. The splitting method was
used to examine the residual strains. In the radial compression test, the mechanical response was
investigated between the rings with different cooling regimes. The FEM analysis of the compression
test in elastic range was also performed. Both the splitting method and the radial compression test
showed no significant difference in the mechanical response of the CFRP rings. The presented results
showed that the fast-cooling rate slightly decreases the mechanical performance of the filament-
wound rings.
Keywords: composites; residual stresses; cooling rate; filament winding
1. Introduction
The manufacturing of composite structures is inherently linked to the mechanism
of residual stress (RS) forming. These stresses affect the mechanical properties of the
manufactured composite structures, essentially, as evidenced in works [1–6]. In particular,
this problem is strongly observed in the manufacturing of composite materials and their
cylindrical structures commonly used in the oil and gas industry [7,8], the automotive
sector including pressure vessels [9–11] as well as in the aerospace industry [12,13]. In
recent decades a significant interest in the matter can be observed. In the study [14] it was
found that the generation of residual stresses strongly fluctuates within the cooling rate. A
higher cooling rate increases the initial residual stresses. Therefore, it was recommended
to first reduce the cooling rate during the curing process to avoid possible cracks in the
matrix when the ambient temperature is reached. In another work [15] the residual stresses
were calculated for different cooling conditions at temperatures 0 ◦ C and 25 ◦ C. It has
been shown that the significant differences in temperature between the curing and cooling
conditions resulted in direct thermal shock, and a considerable amount of residual stress
was created. It resulted in the formation of significant amounts of residual stress, which
may accelerate the destruction of the structure. Samples cooled at 25 ◦ C showed less
residual stress than samples cooled at 0 ◦ C.
The obtained results suggest the necessity of further work related to the assessment of
the influence of carbon fiber reinforced polymer (CFRP) cooling conditions of structures
https://www.mdpi.com/journal/materials
Materials 2021, 14, 1041 2 of 14

on their mechanical properties. Based on these considerations, the key role is played by
residual stresses, which can be estimated using experimental-numerical analysis.
In recent years, the subject of the study experienced a regression—especially since
the available measurement and numerical technique [16,17] allows for efficient estima-
tion of stress levels. One of the most widespread experimental methods is the slitting
method [6,18–20] which is used as a primary investigation method in the presented paper.
In the light of the works [14,15], the nature of the process optimization possibilities arises,
not so much from a change in the isothermal conditions of the cooling, but from a process
medium that allows receiving heat at different speeds, thus clearly influencing the overall
cooling conditions. Recently, in the paper [20] a comparative analysis was performed, but
only for flat components. However, the main aim of the study was to investigate the effect
on fracture resistance expressed in the GIC test. A positive impact of slow cooling has been
demonstrated in increasing resistance against fracture.
Although there is a lack of such analyses in the literature for complex elements
which are an elementary case of a cylindrical structure, this article attempts to bridge a
literature gap in the field of the possible impact of cooling rate on cylindrical compressive
strength of CFRP cylindrical structures. For this purpose, three types of cooling systems are
investigated; quickly in the water at 20 ◦ C (fast cooling, FC), in the air at 23 ◦ C (intermediate
cooling, IC), and slow cooling condition with an oven (slow cooling, SC).

2. Materials and Methods

Carbon fiber reinforced plastic composite cylinders were manufactured to investigate
the mechanical response after different cooling regimes. The general information about the
fiber and matrix is presented in Table 1 [21,22].

Table 1. Fiber and matrix characteristics.

Parameter Value
Fiber characteristics
Name Torayca T720SC
Type Carbon fiber
Filament 36k
Tensile strength 5880 MPa
Tensile modulus 265 GPa
Density 1.8 g/cm3
Matrix characteristics
Resin name Araldite LY 1564
Hardener name Aradur 3474
Resin type thermoset
Formulation 100:26
Pot life at 40 ◦ C 60–70 min
Tensile strength 80 MPa
Tensile modulus 2900 MPa
Density 1.1 g/cm3

2.1. Temperature Gradient Evaluation

The first step was to check the temperature gradient across the cylinder wall cross-
section during cooling. For this purpose, three cylinders with an inside diameter of
104 mm, a length of 400 mm, and a wall thickness of 19.2 mm were made, corresponding
to 42 layers. During the winding, 6 thermocouples were placed at different wall depths.
The arrangement of the sensors is shown in Figure 1.
Only the circumferential layers were wound up. The objects were made of Araldite
LY 1564 epoxy resin with Aradur 3474 hardener reinforced with Torayca T720SC carbon
fiber (material properties are presented in Table 2). All cylinders were cured in accordance
with the manufacturer’s instructions for 5 h at a temperature of 80 ◦ C. After this time, one
 42 layers. During the winding, 6 thermocouples were placed at different wall depths. The
arrangement of the sensors is shown in Figure 1.
Materials 2021, 14, 1041 3 of 14

of the cylinders was cooled quickly in water for 1 h at 20 ◦ C (fast cooling, FC), the other
Materials 2021, 14, 1041 was cooled for 5 h in air at 23 ◦ C (intermediate cooling, IC), and the third was cooled 3 of 15
as
the oven cooled down for 12 h (slow cooling, SC). To simulate the conditions of cooling
present in filament wound tanks (cooling from the outside layer in) specimens were cooled
with a polyamide
42 layers. During the mandrel with
winding, a low thermal were
6 thermocouples conductivity.
placed atThe final product
different afterThe
wall depths. the
manufacturing
arrangement process
of the is shown
sensors in Figure
is shown 2. 1.
in Figure

Figure 1. Diagram of the arrangement of temperature sensors (all dimensions in mm).

Only the circumferential layers were wound up. The objects were made of Araldite
LY 1564 epoxy resin with Aradur 3474 hardener reinforced with Torayca T720SC carbon
fiber (material properties are presented in Table 2). All cylinders were cured in accordance
with the manufacturer’s instructions for 5 h at a temperature of 80 °C. After this time, one
of the cylinders was cooled quickly in water for 1 h at 20 °C (fast cooling, FC), the other
was cooled for 5 h in air at 23 °C (intermediate cooling, IC), and the third was cooled as
the oven cooled down for 12 h (slow cooling, SC). To simulate the conditions of cooling
present in filament wound tanks (cooling from the outside layer in) specimens were
Figure 1. Diagram
Diagram of
ofthe
thearrangement
arrangement of temperature sensors (all
(alldimensions in mm).
Figure
cooled1.with a polyamide mandrel of temperature
with sensors
a low thermal dimensions
conductivity. Thein final
mm).product after
the manufacturing process is shown in Figure 2.
OnlyTable
the circumferential layers
2. Material data used forwere
FEM wound
analysis.up. The objects were made of Araldite
LY 1564 epoxy resin with Aradur 3474 hardener reinforced with Torayca T720SC carbon
Table 2. Material data used for FEM analysis.
E1 (MPa) E2 (MPa) E3 (MPa) ν12 properties
fiber (material ν13 are presented
ν23 G (MPa)2). All
in12Table Gcylinders
13 (MPa) were G23cured ρ (g/cm3 )
(MPa)in accordance
E1 (MPa) 4988
114,000 E2 (MPa) 4988 E3 (MPa)
with 0.33 ν12 0.33
the manufacturer’sν13 instructions
ν23 0.4G12 (MPa)
for 5 2152 G13 (MPa)
h at a temperature
2152 of 80G23°C.
(MPa)
After thisρtime,
1076 (g/cm 3)
1.5 one
114,000 4988 of the cylinders
4988 was0.33
0.33 cooled0.4quickly2152
in water for 1 h at 20 °C (fast cooling,
2152 1076 FC), the1.5
other
was cooled for 5 h in air at 23 °C (intermediate cooling, IC), and the third was cooled as
the oven cooled down for 12 h (slow cooling, SC). To simulate the conditions of cooling
present in filament wound tanks (cooling from the outside layer in) specimens were
cooled with a polyamide mandrel with a low thermal conductivity. The final product after
the manufacturing process is shown in Figure 2.

Table 2. Material data used for FEM analysis.

E1 (MPa) E2 (MPa) E3 (MPa) ν12 ν13 ν23 G12 (MPa) G13 (MPa) G23 (MPa) ρ (g/cm3)
114,000 4988 4988 0.33 0.33 0.4 2152 2152 1076 1.5

Figure2.2.Filament
Figure Filamentwound
woundcylinder
cylindersubjected
subjectedtotothe
theexperimental
experimentalcampaign.
campaign.

2.2.
2.2.Slitting
SlittingMethod—Experimental
Method—ExperimentalAnalysis
AnalysisofofResidual
ResidualStrains
Strains
The
Theslitting
slittingmethod
methodwaswasused
usedtotoreveal
revealthe theresidual
residualstrains.
strains.The
Thetechnique
techniqueisiseasy
easytoto
implement
implementand andinvolves
involvescutting
cuttingout
outaaslice
sliceofofthe
thespecimen
specimentotofree
freethe
theforces
forcespresent
presentininthe
the
composite
compositeafter
afterthe
themanufacturing
manufacturingand andcooling
coolingprocess
process[23–25].
[23–25].InInthe
themethod,
method,the
thestrains
strains
on
onthe
theinner
innerandandouter
outersurfaces
surfacesofofthe
thespecimen
specimenare aremeasured.
measured.The Thespecimen
specimenisismounted
mounted
strictly and by use of a machining tool, such as a hacksaw, a piece of the ring is removed. In
this investigation, strain gauges were mounted on the specimen in four locations. Two of
them were on the inner circumference, and another two were on the outside circumference
Figure 2. Filament wound cylinder subjected to the experimental campaign.
as can be seen in Figure 3.
2.2. Slitting Method—Experimental Analysis of Residual Strains
The slitting method was used to reveal the residual strains. The technique is easy to
implement and involves cutting out a slice of the specimen to free the forces present in the
composite after the manufacturing and cooling process [23–25]. In the method, the strains
on the inner and outer surfaces of the specimen are measured. The specimen is mounted
Materials 2021, 14, 1041 4 of 15

strictly and by use of a machining tool, such as a hacksaw, a piece of the ring is removed.
strictly and by use of astrain
In this investigation, machining
gaugestool,
weresuch as a hacksaw,
mounted a piece ofinthe
on the specimen ring
four is removed.
locations. Two
Materials 2021, 14, 1041 In this investigation, strain gauges were mounted on the specimen in four
of them were on the inner circumference, and another two were on the outside locations. 4Two
of 14
of them were on the inner circumference,
circumference as can be seen in Figure 3. and another two were on the outside
circumference as can be seen in Figure 3.

Figure
Figure 3.
3. Diagram
Diagram of
of the
the strain
strain gauges’
gauges’ locations
locations in
in the
the slitting
slitting method.
method.
Figure 3. Diagram of the strain gauges’ locations in the slitting method.
To
To cut the ring a steelsteel hacksaw
hacksaw waswas used.
used. When
When the bending moment moment was large
enoughTo cut
enough the ring
to close a steel
the ring, hacksaw
another
another cut wasto
cut next
next toused.
the When
thefirst
firstone the
wasbending
onewas performed
performedmoment was
totoeasily
easily largea
remove
remove
enough
aslice tothe
sliceofof
theclose the ring,
specimen
specimen another
so so
that cut
thethe
that next
slitting to thewas
section
slitting first
section one
free.
was wasPhotos
performed
Photos
free. of the thetoring
of ring easily
before remove
and
before after
and
aafter
slice
the of the
experiment specimen
are so
shown that
in the slitting
Figure
the experiment are shown in Figure 4. 4. section was free. Photos of the ring before and
after the experiment are shown in Figure 4.

(a) (b)
(a) (b)
Figure
Figure 4.
4. Photos
Photos of
of the
the tested
tested ring:
ring: (a)
(a) before
before cutting
cutting and
and (b)
(b) after
after cutting.
cutting.
Figure 4. Photos of the tested ring: (a) before cutting and (b) after cutting.
2.3.
2.3. Radial
Radial Compression
Compression Test
Test
2.3. Radial Compression Test
Five
Five samples fromfast
samples from and
fast andintermediate cooling
intermediate andand
cooling six from slowslow
six from cooling were were
cooling used
in Five
the samples
radial from fasttests.
compression and intermediate
All samples cooling
were and
new sixhad
and fromnoslow cooling
previous were
load used
history.
used in the radial compression tests. All samples were new and had no previous load
in the radial compression tests. All samples were new and had no previous load history.
history. The tests were carried out with the MTS 810 (load cell 100 kN) testing machine,
at a compression rate of 3 mm/min. Figure 5 shows a diagram (a) of the test, as well as a
view of the specimen mounted on a machine (b). The reaction force was measured.
 Materials 2021, 14, 1041 5 of 15

The tests were carried out with the MTS 810 (load cell 100 kN) testing machin
compression
The tests were carried outrate of 3the
with mm/min. Figure
MTS 810 (load5 cell
shows100a kN)
diagram (a) of
testing the test,at
machine, as awell as
Materials 2021, 14, 1041 5 of 14
of the specimen mounted on a machine (b). The reaction force was measured.
compression rate of 3 mm/min. Figure 5 shows a diagram (a) of the test, as well as a view
of the specimen mounted on a machine (b). The reaction force was measured.

(a) (b)
(a)
Figure 5. Radial compression of composite rings: (a) scheme(b) of the experimental boundary conditions and (b) view of t
sample mounted
Figure 5. Radial compression on the machine.
of composite rings: (a) scheme of the experimental boundary conditions and (b) viewconditions
of the
Figure 5. Radial compression of composite rings: (a) scheme of the experimental boundary
sample mounted on the machine.
and (b) view of the sample mounted on the machine.
2.4. Finite Element Method Analysis
2.4.
2.4.Finite
FiniteElement
ElementMethod
Method Analysis
The numerical
Analysisanalysis was performed in Abaqus/CAE environment using the
The elastic
numerical material
analysis
The numerical analysis was model.
was performed The finite
performed inin element method
Abaqus/CAE
Abaqus/CAE (FEM) using
environment
environment analysis
usingthe theaimed
linearto inve
linear
elastic material whether
model. the
The residual
finite stresses
element may
method influence
(FEM) the material
analysis
elastic material model. The finite element method (FEM) analysis aimed to investigate aimedbehavior
to in the
investigate elastic rang
whether
whether the material
theresidual
residual properties
stresses
stresses may
may presented
influence
influence the in
the Table
material
material2 were
behavior
behaviorobtained
ininthe
the using
elastic
elastic the
range.
range. product
The data
The
material
material [21,22]
properties
properties and
presented the
presented rule
in Table of
in Tablemixture.
2 were They
2 obtained were
were obtained set
using thein a cylindrical
product
using coordinate
datasheets
the product [21,22]
datasheets system to
and the rule
[21,22] andof the
the hoop
mixture.
rule winding
of They
mixture. technology.
wereThey set in Moreover,
a cylindrical
were set the calculated
coordinate
in a cylindrical system to
coordinate residual
reflect
system stresses
theto hoop were app
reflect
the hoop winding technology. Moreover, the calculated residual stresses were applied were
winding the
technology. ring before
Moreover, thethe compression
calculated simulation
residual was
stresses performed.
were applied Theto stresses
the ring to calc
the ring before the compression simulation was performed. The stresses were calculated stres
before the using
compression the linear
simulation dependence
was between
performed. Themeasured
stresses strain
were signals
calculated and
using appeared
the
linear
usingdependence in dependence
the linear the
betweenelastic range,
measured
between namely,
strain Equations
signals
measured (1) and
and appeared
strain signals (2)
stress
and described
level
appeared in the
stressin level
Section 3.4
elastic
range, namely, Equations
implemented. (1) and (2) described in Section 3.4
in the elastic range, namely, Equations (1) and (2) described in Section 3.4 were were implemented.
The assembly in
implemented. Thetheassembly
simulation consisted
in the of theconsisted
simulation ring andof twotheplates:
ring and the two
lower one the low
plates:
fixed and the upper
fixed with
and the
the vertical
upper displacement.
with the vertical The plates
displacement. were
The assembly in the simulation consisted of the ring and two plates: the lower one modeled
The plates as
werea rigid
modeled as
body as can be seen
body in
as Figure
can be 6.
seen in Figure 6.
fixed and the upper with the vertical displacement. The plates were modeled as a rigid
body as can be seen in Figure 6.

(a) (b)
(a)Figureconditions
Figure 6. Boundary 6. Boundary conditions
(a) and (a) and
geometric geometric
model (b)
model
(b) of the ring.(b) of the ring.
Figure 6. Boundary conditions (a) and geometric model (b) of the ring.
The ring was modeled as a 3D object consisting of 19,400 linear hexahedral elements
of type C3D8R. The discrete model can be seen in Figure 7.
Materials 2021, 14, 1041 6 of 15

Materials 2021, 14, 1041 6 of 14

The ring was modeled as a 3D object consisting of 19,400 linear hexahedral elements
of type C3D8R. The discrete model can be seen in Figure 7.

Materials 2021, 14, 1041 6 of 15

The ring was modeled as a 3D object consisting of 19,400 linear hexahedral elements
of type C3D8R. The discrete model can be seen in Figure 7.

Figure 7. A
Figure 7. discrete model
A discrete of the
model analyzed
of the ring.
analyzed ring.

3. Results
3. Results and
and Discussion
Discussion
3.1. Temperature Gradient during Cooling
3.1. Temperature Gradient during Cooling
The diagrams of temperature in the cross-section of the cylinder wall during heat-
The diagrams of temperature in the cross-section of the cylinder wall during heating
ing (Figure 8) and cooling in air (Figure 9) and in water (Figure 10) are presented below.
(Figure 8) and cooling in air (Figure 9) and in water (Figure 10) are presented below.
Thermocouple 3 was damaged during the test and the data were discarded.
Thermocouple 3 was damaged during the test and the data were discarded.
The maximum temperature difference between the outer and inner layers during
heating was 20 ◦ C, during cooling in air it was approximately 11 ◦ C, and during cooling in
water 42 ◦ C.
The set temperature in the oven was obtained after 15 min whereas the temperature
of the composite ring rose significantly slower and obtained the set 80 ◦ C after about 3 h.
Thermocouples 1 and
Figure 7. A discrete model4 had
of thethe lowestring.
analyzed temperature gain. The highest temperature gain
was on the outer circumference of the composite element which indicates thermocouple 6.
3. Results
In the and of air cooling, the highest temperature difference was about 11 ◦ C. The
caseDiscussion
highest temperature
3.1. Temperature loss during
Gradient was obtained
Cooling by thermocouple 6. The temperature in thermo-
couples 2 and 5, which were placed in the middle of the cross-section of the composite
The diagrams of temperature in the cross-section of the cylinder wall during heating
thickness, did not stand out significantly from the temperature on the inner surface of
(Figure 8) and cooling in air (Figure 9) and in water (Figure 10) are presented below.
the cylinder.
Thermocouple 3 was damaged during the test and the data were discarded.

Figure 8. Temperature change in individual layers during cylinder heating.

Figure8.8.Temperature
Figure Temperaturechange
change in
in individual
individual layers
layers during cylinder heating.
heating.
Materials 2021, 14, 1041 7 of 14

For fast cooling in water, the highest temperature difference occurred, at the level of
about 42 ◦ C. A great temperature gradient was obtained. The outer surface of the cylinder
cooled down significantly faster than the middle of the cross-section and the inner surface.
Nevertheless, the difference between the temperature in the middle of the cross-section
and the inner surface was also noticeable.
Materials 2021, 14, 1041 7 ofThe
Unfortunately, the data recorded during the cooling in the furnace was corrupted. 15
Materials 2021, 14, 1041 7 of 15
time needed to reach room temperature, in this case, was about 12 h.

Figure 9.
Figure 9. Temperature
Temperature change
change in
in individual
individual layers
layers during
during air
aircooling.
cooling.
Figure 9. Temperature change in individual layers during air cooling.

Figure 10. Temperature change in individual layers during cooling in water.

Figure 10.
Figure 10. Temperature
Temperature change
change in
in individual
individual layers
layers during
during cooling
coolingininwater.
water.
The maximum temperature difference between the outer and inner layers during
3.2. Radial Compression
The maximum Test
temperature difference between the outer and inner layers during
heating was 20 °C, during cooling in air it was approximately 11 °C, and during cooling
heating was 20
As mention °C, during cooling in air it was approximately 11 °C, and during cooling
in water 42 °C. in the previous chapter, a radial compression test of the composite rings
in water 42
was performed. °C. Figures 11–13
The set temperature in theshow
oven force-displacement
was obtained after 15 diagrams obtained
min whereas during radial
the temperature
The set temperature
compression of composite inrings.
the oven
All was
the obtained
figures showafterpeaks
15 min whereas
and valleys the temperatureof
characteristic
of the composite ring rose significantly slower and obtained the set 80 °C after about 3 h.
of the composite
composite elementsringresulting
rose significantly
from slower and of
the destruction obtained
individualthe set 80 °C
layers ofafter about 3 h.
the material.
Thermocouples 1 and 4 had the lowest temperature gain. The highest temperature gain
Thermocouples
In the 1 and
elastic part 4of
had thethe lowest behavior,
material temperature gain.
there is The highest
great temperature
consistency betweengain
the
was on the outer circumference of the composite element which indicates thermocouple
was on the
samples. outer
The mean circumference
value of the of the composite
maximum force element
was 19.44 which
kN ±indicates
1.08 kN. thermocouple
After the first
6.
6. the composite ring could withstand a significant portion of energy since the curve did
falls, In the case of air cooling, the highest temperature difference was about 11 °C. The
not In dramatically.
fall the case of airGenerally,
cooling, the highest temperature
theredifference was about 11However,
°C. The
highest temperature loss was after the first
obtained bydrop,
thermocouple is no rise in the
6. The force.
temperature in
highest
in the casetemperature
of specimen loss
6, thewas obtained
force rose even by thermocouple
though small 6.
cracks The temperature
occurred. This in
behavior
thermocouples 2 and 5, which were placed in the middle of the cross-section of the
thermocouples 2 and 5, which were placed in the middle of the cross-section of the
composite thickness, did not stand out significantly from the temperature on the inner
composite thickness, did not stand out significantly from the temperature on the inner
surface of the cylinder.
surface of the cylinder.
For fast cooling in water, the highest temperature difference occurred, at the level of
For fast cooling in water, the highest temperature difference occurred, at the level of
about 42 °C. A great temperature gradient was obtained. The outer surface of the cylinder
about 42 °C. A great temperature gradient was obtained. The outer surface of the cylinder
Materials2021,
Materials 2021,14,
14,1041
1041 88of
of15
15
Materials 2021, 14, 1041 8 of 14

Unfortunately, the
Unfortunately, the data
data recorded
recorded during
during the the cooling
cooling in
in the
the furnace
furnace was was corrupted.
corrupted.
The
The
is time needed
time needed
common to reach room
to reachmaterials
in composite temperature,
room temperature, in
where singleincracks this case,
this case, was
maywas about
not about 12
12 h.
affect the h.
global strength of
the component.
3.2. Radial
3.2. Radial
For theCompression
Compression
intermediately Testcooled (IC) rings there is a small, but noticeable difference in the
Test
elasticAspart of the
As mention
mention inexperiment.
in the The mean
the previous
previous value
chapter,
chapter, of the compression
aa radial
radial maximum
compressionforce was
test
test of20.11
of kN ± 0.87rings
the composite
the composite kN.
rings
In the
was performed.
was case of
performed. Figures fast cooled
Figures 11–13 (FC)
11–13 show rings the
show force-displacementmean value
force-displacement diagrams of the maximum
diagrams obtained
obtained duringforce was
during radial
radial
17.84 kN ± 0.75
compression
compression of kN, which rings.
of composite
composite is significantly
rings. All the
All lowershow
the figures
figures than in
show the previous
peaks
peaks and valleys
and twocharacteristic
valleys groups (8% in
characteristic of
of
the case
composite of slow cooling
elements and
resulting11%
fromin the case of
destructionintermediate
of cooling).
individual layers
composite elements resulting from the destruction of individual layers of the material. of the material.

Figure
Figure 11.Load-displacement
Figure11.
11. Load-displacementcurves
Load-displacement curvesfor
curves forradial
for radialcompression
radial compressionof
compression ofslow
of slowcooled
slow cooled(SC)
cooled (SC)rings.
(SC) rings.
rings.

Figure12.
Figure 12. Load-displacement
Load-displacementcurves
curvesfor
forradial
radialcompression
compressionof
compression ofintermediately
of intermediatelycooled
intermediately cooled(IC)
cooled (IC)rings.
(IC) rings.
rings.
Materials 2021, 14, 1041 9 of 14
Materials 2021, 14, 1041 9 of 15

Figure 13. Load-displacement curves for

Figure 13. for fast
fast cooled
cooled (FC)
(FC) rings.
rings.

Table
In the3elastic
summarizes
part of the
themaximum forces obtained
material behavior, there isfor each
great of the samples
consistency as well
between the
as their mean values. Coefficient of variation was calculated to assess
samples. The mean value of the maximum force was 19.44 kN ± 1.08 kN. After the first the dispersion of
results. To compare the behavior of differently cooled rings in the elastic
falls, the composite ring could withstand a significant portion of energy since the curveregion, the radial
stiffness
did not was calculated. This
fall dramatically. procedure
Generally, wasthe
after performed based
first drop, on regression
there is no rise analysis and
in the force.
calculation
However, inofthe thecase
slopeof coefficient
specimen 6,(in the
the range
force ofeven
rose 25–75% of the
though linear
small part).
cracks The average
occurred. This
difference between FC and IC is about 5% and between FC and
behavior is common in composite materials where single cracks may not affect the SC, it is 7%. However,
global
taking
strengthinto
of account the data deviation, there is no statistical difference between the specific
the component.
cooling regimes in terms
For the intermediately of the elastic
cooled (IC)region.
rings there is a small, but noticeable difference in
the elastic part of the experiment. The mean value of the maximum force was 20.11 kN ±
Table 3. Max loads for radial compression of rings.
0.87 kN.
In the case of fast
Max cooled (FC)Mean
rings the mean value of the maximum force was 17.84
Radial Stiffness
kNSpecimen
± 0.75 kN, which is significantly
Compressive lower than in
Compressive the(%)
CoV previous two groups (8% CoV
in the(%)case
(N/mm)
of slow cooling and Load
11% (kN) Loadof(kN)
in the case intermediate cooling).
Table
IC1 3 summarizes18.80 the maximum forces obtained for each of the samples as well as
their mean
IC2 values. 20.93
Coefficient of variation was calculated to assess the dispersion of
IC3
results. To compare 20.88
the behavior20.11 ± 0.87
of differently 11,377.5
4.32 rings in
cooled ± 916.4
the elastic 8.05radial
region, the
IC4 19.87
stiffness was calculated. This procedure was performed based on regression analysis and
IC5 20.08
calculation of the slope coefficient (in the range of 25–75% of the linear part). The average
SC1 between 20.92
difference FC and IC is about 5% and between FC and SC, it is 7%. However,
SC2 19.41
taking into account the data deviation, there is no statistical difference between the
SC3 18.68
specific cooling regimes 19.44 ± 1.07 11,560.5 ± 288.7
18.9 in terms of the elastic region.
5.50 2.50
SC4
SC5 20.52
Table SC6
3. Max loads for18.97
radial compression of rings.

FC1 Max16.55 Mean

FC2 17.89 Radial Stiffness
Specimen Compressive Compressive CoV (%) CoV (%)
FC3 18.49 17.84 ± 0.75 4.20 (N/mm)
10,764.4 ± 1298.7 12.06
FC4 Load 18.11
(kN) Load (kN)
IC1
FC5 18.80
18.18
IC2 20.93
IC3
The damage20.88 20.11
behavior of the ± 0.87 ring may
composite 4.32 be described
11,377.5mainly
± 916.4as delamination
8.05
IC4 19.87
between the filament-wound layer. It can be observed in Figure 14. On the inner circum-
IC5 local buckling
ference, 20.08 areas with fiber breaking also occurred. These processes generally
tookSC1 20.92
place after the maximum force.
SC2 19.41 19.44 ± 1.07 5.50 11,560.5 ± 288.7 2.50
SC3 18.68
 FC4 18.11
FC5 18.18

The damage behavior of the composite ring may be described mainly as delamination
Materials 2021, 14, 1041 between the filament-wound layer. It can be observed in Figure 14. On the10inner of 14
circumference, local buckling areas with fiber breaking also occurred. These processes
generally took place after the maximum force.

Figure14.
Figure 14.Composite
Compositering
ringafter
afterradial
radialcompression
compressiontest.
test.

3.3.
3.3.FEM
FEMAnalysis
Analysis
In
Inthe
theFEM
FEMsimulation,
simulation,the
theforce-displacement
force-displacementcurve
curvewas
wasobtained.
obtained.Since
Sinceno
nodamage
damage
Materials 2021, 14, 1041 model
modelwaswasapplied,
applied,only
onlythe
theelastic
elasticpart
partcan
canbe
bediscussed,
discussed,where
wherethe
theagreement
agreementbetween
11 of 15
between
the experiment and the simulation is satisfactory. The comparison is presented in Figure
the experiment and the simulation is satisfactory. The comparison is presented in Figure 15.
15.

Figure Comparisonbetween
15. Comparison
Figure 15. between experimental
experimental results
results of radial
of radial compression
compression of rings
of rings and analysis
and FEM FEM analysis
for thefor
fastthe fast
cooled
cooled
rings. rings.

The curves
The curves for
forthe
thethree
threecooling
coolingregimes
regimes dodonotnot differ
differ since
since residual
residual stresses
stresses did did
not
not influence the elastic behavior of the composite rings under radial compression.
influence the elastic behavior of the composite rings under radial compression. The stress The
stress fields
fields show show the concentration
the concentration of stress
of stress in the
in the side
side areas
areas ofofthe
thering,
ring,both
both tension
tension and
and
compression (Figure 16). These reflect the real damage mechanism, where in the
compression (Figure 16). These reflect the real damage mechanism, where in the inner
inner
circumference there
circumference there were
were local
local buckling
buckling areas
areas due
due to
to compressive
compressive loads.
loads.
 rings.

The curves for the three cooling regimes do not differ since residual stresses did not
influence the elastic behavior of the composite rings under radial compression. The stress
fields show the concentration of stress in the side areas of the ring, both tension and
Materials 2021, 14, 1041 11 of 14
compression (Figure 16). These reflect the real damage mechanism, where in the inner
circumference there were local buckling areas due to compressive loads.

Figure16.
Figure Stressfields
16.Stress fieldsininhoop
hoopdirection
direction(S11)
(S11)for
forthe
thefast
fastcooled
cooledrings.
rings.

Thedifferent
The different cooling
cooling rates
ratesreflected
reflectedininthethe
residual stresses
residual did did
stresses not influence the stress
not influence the
distribution in the specimens. The only difference was the value of the stress at the set
stress distribution in the specimens. The only difference was the value of the stress at the
displacement of the plates. However, the difference did not exceed 5%.
set displacement of the plates. However, the difference did not exceed 5%.
3.4. Residual Strains Obtained by Slitting Experiment
3.4. Residual Strains Obtained by Slitting Experiment
To reveal the residual stresses after the manufacturing process the slitting method was
used.ToEach
reveal thewas
ring residual stresses
equipped withafter
four the manufacturing
strain gauges whichprocess the slitting
were supposed to method
measure
was used. Each ring was equipped with four strain gauges which were supposed
the strains in the hoop direction. Graphs showing the strains of rings during cutting are to
presented in Figure 17. It is noticeable that strains increased gradually, finally obtain-
ing the maximum values. The final strains of the inner circumference (gauges 1 and 3)
ended up being negative, whereas those on the outer circumference (gauges 2 and 4) had
positive values.
Although the process of cutting was continuous, there are some plateaus areas on the
graphs, as can be seen in Figure 17. It was by the fact that during the process of cutting the
deformation had an abrupt character.
Judging from the residual strains, the residual stresses can be estimated using the
following equations [16]:
σθin = Eθ ∆εin
C (1)
σθout = Eθ ∆εout
C (2)
where ∆εinC and ∆ε C are the hoop strains on the inside and outside surfaces during the
out

slitting method (radial-cut), respectively, and Eθ is the known elastic modulus. The results
of the calculations are presented in Table 4.

Table 4. Estimated residual stresses in fiber direction on the inside and outside surfaces of the rings.

Cooling Regime σθin (MPa) σθout (MPa)

Intermediate cooling −126 76
Slow cooling −150 86
Fast cooling −143 84
 measure the strains in the hoop direction. Graphs showing the strains of rings durin
cutting are presented in Figure 17. It is noticeable that strains increased gradually, finall
obtaining the maximum values. The final strains of the inner circumference (gauges 1 an
Materials 2021, 14, 1041 12 of 14
3) ended up being negative, whereas those on the outer circumference (gauges 2 and 4
had positive values.

(a)

(b)

(c)
Figure 17. The strain
Figureof 17.
theThe
composite
strain ofrings during therings
the composite slitting experiment,
during recorded
the slitting on the recorded
experiment, inside and
onoutside surfaces
the inside
for: (a) slow cooled, (b) intermediately cooled, and (c) fast cooled rings.
and outside surfaces for: (a) slow cooled, (b) intermediately cooled, and (c) fast cooled rings.

4. Conclusions
The current study aimed to investigate the influence of the cooling rate on the me-
chanical response of the filament-wound composite cylinders. The following conclusions
can be drawn from the performed research work and numerical analyses:
The radial compressive strength does not show significant differences between SC
and IC cooling mode. In comparison with the obtained results, only FC exhibited the
smallest value of maximum force recorded in the test for each specimen. The difference in
Materials 2021, 14, 1041 13 of 14

the maximum force during radial compression test is 8% in terms of FC to SC comparison

and 11% in terms of FC to IC comparison. This effect might be caused by microstructural
changes (microcracks) induced during the fast-cooling regime. This hypothesis demands
further investigation.
The lack of statistical difference between the different cooling regimes in terms of
elastic stiffness during radial compression suggests that fast cooled thick-wall composite
rings could perform equally in the elastic and safe range of the loading.
The analysis of the residual strain distribution and the residual stress values obtained
suggest the selection of slow cooling as a method that ensures equally effective radial
compressive strength as IC.
On the contrary, from the practical point of view, the fast-cooling regime does not
decrease the mechanical durability of the composite cylinders in a way that disqualifies the
method from further analyses. In industry, a slight drop of the strength may be acceptable
in light of the possibility of a significant reduction in manufacturing time.
Although static loads seem to indicate quick cooling as the least advantageous treat-
ment after the manufacturing process, it should be pointed out that this mechanism should
be much more strongly emphasized in the studies related to the influence of both time and
stress concentrators (fatigue and fracture). Furthermore, a comparison (despite similar
static results) of IC and SC in fatigue tests seems to be fully reasonable in further work of
the authors’ team.

Author Contributions: Conceptualization, W.B. and M.B.; methodology, W.B. and P.S.; formal
analysis, M.L. and P.S.; investigation, W.B., M.L., P.S., and P.B.; resources, W.B. and M.S.; data curation,
M.B. and P.B.; writing—original draft preparation, M.L. and P.S.; writing—review and editing, G.L.;
visualization, M.L. and P.S.; supervision, M.S. and G.L.; project administration, M.S.; funding
acquisition, M.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Data sharing is not applicable to this article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Favre, J.P. Residual Thermal Stresses in Fibre Reinforced Composite Materials—A Review. J. Mech. Behav. Mater. 1988, 1, 37–53.
2. Adams, M.E.; Campbell, G.A.; Cohen, A. Thermal stress induced damage in a thermoplastic matrix material for advanced
composites. Polym. Eng. Sci. 1991, 31, 1337–1343. [CrossRef]
3. Unger, W.J.; Hansen, J.S. A Method to Predict the Effect of Thermal Residual Stresses on the Free-Edge Delamination Behavior of
Fibre Reinforced Composite Laminates. J. Compos. Mater. 1998, 32, 431–459. [CrossRef]
4. Kim, J.-K.; Mai, Y.-M. Engineered Interfaces in Fiber Reinforced Composites; Elsevier Science: Amsterdam, The Netherlands, 1998.
5. Ghasemi, A.R.; Asghari, B.; Tabatabaeian, A. Determination of the influence of thermo-mechanical factors on the residual stresses
of cylindrical composite tubes: Experimental and computational analyses. Int. J. Press. Vessel. Pip. 2020, 183, 104098. [CrossRef]
6. Amir-Ahmadi, S.; Ghasemi, A.R. Experimental investigation of cooling conditions, MWCNTs and mandrel diameter effects on
the thermal residual stresses of multi-layered filament-wound composite pipes. J. Compos. Mater. 2020. [CrossRef]
7. Sebaey, T.A. Design of Oil and Gas Composite Pipes for Energy Production. Energy Procedia 2019, 162, 146–155. [CrossRef]
8. Prabhakar, M.M.; Rajini, N.; Ayrilmis, N.; Mayandi, K.; Siengchin, S.; Senthilkumar, K.; Karthikeyan, S.; Ismail, S.O. An overview
of burst, buckling, durability and corrosion analysis of lightweight FRP composite pipes and their applicability. Compos. Struct.
2019, 230, 111419. [CrossRef]
9. Roh, H.S.; Hua, T.Q.; Ahluwalia, R. Optimization of carbon fiber usage in Type 4 hydrogen storage tanks for fuel cell automobiles.
Int. J. Hydrogen Energy 2013, 38, 12795–12802. [CrossRef]
10. Kim, B.S.; Kim, B.H.; Kim, J.B.; Joe, C.R. Study on the development of composite CNG pressure vessels. Cryogenics 1998, 38,
131–134. [CrossRef]
11. Blazejewski, W.; Czulak, A.; Gasior, P.; Kaleta, J.; Mech, R. SMART Composite High Pressure Vessels with Integrated Optical
Fiber Sensors. In Sensors and Smart Structures Technologies for Civil, Mechanical, and Aerospace Systems 2010; Tomizuka, M., Ed.;
SPIE: San Diego, CA, USA, 2010; p. 764712.
12. Al-Lami, A.; Hilmer, P.; Sinapius, M. Eco-efficiency assessment of manufacturing carbon fiber reinforced polymers (CFRP) in
aerospace industry. Aerosp. Sci. Technol. 2018, 79, 669–678. [CrossRef]
Materials 2021, 14, 1041 14 of 14

13. Banerjee, S.; Messidoro, P.; Ferrarese, A.; Beard, S.; Banerjee, R. Structural Health Monitoring Technology for Aerospace Composite
Propellant Tank. In Proceedings of the ASME 2011 Conference on Smart Materials, Adaptive Structures and Intelligent Systems,
SMASIS 2011, Scottsdale, AZ, USA, 18–21 September 2011.
14. Zhang, Y.; Xia, Z.; Ellyin, F. Evolution and influence of residual stresses/strains of fiber reinforced laminates. Compos. Sci. Technol.
2004, 64, 1613–1621. [CrossRef]
15. Asghari, B.; Ghasemi, A.R.; Tabatabaeian, A. On the optimal design of manufacturing-induced residual stresses in filament
wound carbon fiber composite cylindrical shells reinforced with carbon nanotubes. Compos. Sci. Technol. 2019, 182, 107743.
[CrossRef]
16. Kang, C.; Shi, Y.; Deng, B.; Yu, T.; Sun, P. Determination of Residual Stress and Design of Process Parameters for Composite
Cylinder in Filament Winding. Adv. Mater. Sci. Eng. 2018, 2018, 1–11. [CrossRef]
17. Liu, C.; Shi, Y. Analytical model for the winding process-induced residual stresses of the multilayered filament wound cylindrical
composite parts. Mater. Res. Express 2019, 6, 105354. [CrossRef]
18. Ghasemi, A.R.; Tabatabaeian, A.; Asghari, B. Application of Slitting Method to Characterize the Effects of Thermal Fatigue, Lay-up
Arrangement and MWCNTs on the Residual Stresses of Laminated Composites. Mech. Mater. 2019, 134, 185–192. [CrossRef]
19. Salehi, S.; Shokrieh, M. Residual stress measurement using the slitting method via a combination of eigenstrain, regularization
and series truncation techniques. Int. J. Mech. Sci. 2019, 152, 558–567. [CrossRef]
20. Umarfarooq, M.A.; Shivakumar Gouda, P.S.; Veeresh kumar, G.B.; Banapurmath, N.R.; Edacherian, A. Impact of Process Induced
Residual Stresses on Interlaminar Fracture Toughness in Carbon Epoxy Composites. Compos. Part A Appl. Sci. Manuf. 2019, 127,
105652. [CrossRef]
21. Toray. Torayca—High-Performance Carbon Fiber. Available online: https://www.torayca.com/download/pdf/torayca.pdf
(accessed on 19 January 2021).
22. Advanced Materials Araldite® LY 1564/Aradur® 3486/Aradur® 3487. Available online: https://www.swiss-composite.ch/pdf/
t-Araldite-LY1564-Aradur3486-3487-e.pdf (accessed on 22 February 2021).
23. Kim, J.W.; Lee, D.G. Measurement of Residual Stresses in Thick Composite Cylinders by the Radial-Cut-Cylinder-Bending
Method. Compos. Struct. 2007, 77, 444–456. [CrossRef]
24. Kim, J.W.; Lee, J.H.; Kim, H.G.; Kim, H.S.; Lee, D.G. Reduction of Residual Stresses in Thick-Walled Composite Cylinders by
Smart Cure Cycle with Cooling and Reheating. Compos. Struct. 2006, 75, 261–266. [CrossRef]
25. Casari, P.; Jacquemin, F.; Davies, P. Characterization of residual stresses in wound composite tubes. Compos. Part A Appl. Sci.
Manuf. 2006, 37, 337–343. [CrossRef]