Preliminary analysis of the effects of thermal cycling on the mechanical properties of Kevlar®/Epoxy tubular braided composites for use in satellite components

Abstract

Satellites in low Earth orbit (LEO) are exposed to several hazardous conditions including large temperature fluctuations. Materials with non-zero coefficients of thermal expansion (CTE) can develop thermal stresses that lead to failure. Tubular braided composites (TBCs) are a customizable type of fiber reinforced polymer composite (FRPC) and are manufactured with high-strength fibers and polymer resins to create a cylindrical structure optimal for load bearing applications with a near-zero CTE. Despite this, TBCs have not been used in satellites, nor have they had the effects of thermal cycling on their mechanical behaviour evaluated. This preliminary work studied the effect of relatively low amounts of thermal cycling on the mechanical performance of Kevlar®/Epoxy TBCs to pre-emptively assess their viability in satellite structural applications. Samples made with one of three braid angles (35°,45°, and 55°) were exposed to either zero, five, or 10 cycles between −50 $℃$ and 120 $℃$ , before being tensile tested to find their elastic modulus and ultimate tensile strengths (UTS). Thermal cycling decreased the average elastic modulus and UTS of the 35° TBCs by 16.6% and 13.8% after five cycles, and 17.0% and 33.5% after 10 cycles respectively. The 45-degree TBC’s average elastic modulus and UTS increased by 8.9% and 7.3% after five cycles, then decreased by 25.8% and 22.2% after 10 cycles. The 55° TBC’s average elastic modulus and UTS decreased by 9.3% and 25.1% after five cycles, and 30.7% and 36.2% after 10 cycles. Normality and equal variance of the data were verified with Kolmogorov-Smirnov and Bartlett’s tests respectively. Single factor ANOVA tests were also used to verify the significance of change between the mechanical property averages and were succeeded by Tukey’s test to determine which specific pairs of averages were significantly different. Before TBCs can be used in satellites, further evaluation must be performed. This may include variation in materials, greater cycle counts, alternate cycle durations, or different temperature change rates.

Keywords

composites space satellites temperature thermomechanical testing braided composites mechanics of materials thermal cycling coefficient of thermal expansion

Introduction & literature review

Satellites serve a significant role in transmitting vital information across Earth. They allow for global positioning systems, long range media broadcasting, weather forecasting, and imaging of both Earth and deep space.^1,2 Due to their importance, satellites must be designed to best resist the harsh environmental conditions of Low Earth Orbit (LEO), where many of them are located. Hazards present in LEO include space debris impacts, vacuum, ionizing particles, cosmic radiation, and large temperature gradients. Ambient temperatures in LEO range from −65 $℃$ to 125 $℃$ , with satellites experiencing anywhere from 780 to 6000 cycles between these temperatures annually based on their altitude.³

Rapid cycling between these temperatures can develop thermal strains within a material. The magnitude of the thermal strains is determined by a material’s coefficient of thermal expansion (CTE). At critical locations, such as joints, mismatched CTEs between components comprised of two materials can lead to thermal stress development at connection points, causing premature failure.⁴

An important consideration of satellite design is material selection. Aluminum alloys were used to create the earliest satellites due to their ease of manufacture and desirable properties. Aluminum alloys had high specific strength which improved fuel efficiency for deployment and were also corrosion resistant providing protection from atmospheric and chemical wear.^5–7 However, they were not suitable for applications requiring tight thermal tolerancing due to their large positive CTEs of 23.2 $μ$ mm/mm $℃$ .⁸

Fiber reinforced polymer composites (FRPCs) have high strength and comparably lower weight than aluminum alloys, but with the added benefit of having low CTEs. This is due to the thermomechanical behavior of the commonly used fibers, such as carbon, glass, and aramid fiber.^4,9–11 Aramid and carbon fibers exhibit small but negative CTEs while coupled with polymer matrices that have positive CTEs, allowing the overall composite structure to achieve a near-zero strain with temperature changes. FRPCs are typically used in tape and sheet woven forms for aerospace applications. Some applications implement woven FRPCs but in general the use of textile composites has been limited.⁴

Textile composites provide the same high strength-to-weight ratio of typical FRPCs, but with the ability to be widely customized and designed to different shapes. A unique configuration of textile composites are tubular braided composites (TBCs). These are highly customizable cylindrical composites that are applicable to structural satellite components such as load bearing shafts. TBCs are manufactured in two stages. The first involves the formation of a braided structure composed of pure fiber called a preform, which is made using either a maypole braider, a radial braider, or even a 3-D braider for 3-D TBCs. After a preform is fabricated, a polymer matrix is applied to the structure and cured to form the full composite.¹² A TBC can be customized through material selection, fiber volume fraction, and braid angle. Braid angle measures the angle between the principal axis of the composite and the direction of the braid strands and heavily affects the mechanical performance of the composite.^13,14 Braided composites have been used in sectors such as aerospace, automotives, medicine, and sports equipment.¹⁵ Although studies have thoroughly investigated the mechanical properties of these materials, some gaps remain in fully understanding their thermal behaviour.

Similar to other textile composites, TBCs have seen limited application to satellites, resulting in no evaluation of their performance in simulated LEO conditions. Before they can be used in satellite structures in any form, TBCs need to have their behaviour in LEO conditions evaluated and understood. This study aims to quantify the previously unexplored effects of thermal cycling typical in LEO on the mechanical properties of Kevlar®/Epoxy TBCs. This work experimentally investigates how a limited number of thermal cycles affect the elastic modulus and ultimate tensile strength of TBCs manufactured at three different braid angles. The results from this preliminary study aim to verify the potential viability of this material and configuration for TBCs intended for use in satellite support structures. Additionally, it would provide grounds for an expansion of the tests to larger thermal cycle counts. However, if the work determines significant negative changes at low cycles, the selected configurations are not acceptable for satellite use. In that event, modifications to either material selection or geometric configuration may need to be made before TBCs can be suitable for this application.

The environmental demands of Low Earth Orbit (LEO) have increased the desirability of lightweight, strong, and thermally stable materials.^4,16 To meet these requirements, studies have investigated composite materials as viable alternatives to typical metallic alloys and ceramics. Advanced composite materials have already been used in spacecraft structures. NASA has used composites to design liquid hydrogen tanks to reduce craft weight by up to 30%.¹⁷ Deployable mechanisms such as antennas, solar panels, observation instruments, parabolic reflectors, and booms have also been fabricated using composite materials.^18–21 Additionally, Lightweight ionizing shielding has been made using composites as well.²²

Temperature cycling causes repetitive expansion and contraction of the individual material constituents in a composite structure at dissimilar rates. This results in the development of stresses within the composite structure. To better understand the influence of temperature cycling on the structural damage of composite materials, there have been numerous investigations into FRPCs and their behavior in environments with temperature fluctuations. The investigations evaluated any potential thermally induced stress development within composites and characterized the types of damage observed.

Several studies have discussed the effects of thermal cycling on the mechanical performance of fiber reinforced polymer composites (FRPCs). Park et al. experimentally investigated the effects of a simulated LEO environment on three different types of carbon fiber prepreg tapes. The tapes were thermally cycled between −175 $℃$ and 120 $℃$ at a vacuum pressure of 1.3 × 10⁻³ Pa for either 0, 500, 1000, 1500, or 2000 cycles. Following their exposure, the samples were evaluated for their interlaminar shear, flexural, and tensile strengths. At 2000 cycles, shear, flexural, and tensile strength decreased by 14.7%, 13.3%, and 11.1% respectively across the three sample types.²³ Another thermal cycling study completed by Boccaccini et al. investigated the effects of thermal cycling on Silicon-Carbide fiber reinforced glass matrix composites. Composites were cycled between 25 $℃$ and 700 $℃$ for 1000 cycles while the change in the composite’s elastic modulus was monitored using a non-destructive forced vibration technique. Their results displayed that the elastic modulus decreased by nearly 4% as samples approached 1000 cycles, and this reduction was attributed to oxidation of the fibers, softening of the glass matrix, and microstructural degradation caused by thermal aging.²⁴

In contrast to these findings, Qu et al. investigated the impact of cold temperature thermal cycling on carbon fiber laminates intended for use in cryogenic launch vehicle tanks. Experimental laminae samples were composed of graphene oxide nanosheet modified epoxy and carbon fabric. The samples were cycled between room temperature and −196 $℃$ for either 50 or 100 cycles and were compared to a non-cycled control group. The main mechanical property of interest was the intralaminar shear strength (ILSS) due to its importance in cryogenic applications. After 50 cycles, the ILSS was found to have initially increased from the control group but then decreased below the control group after 100 cycles. The observed increase in ILSS after 50 cycles was attributed to the release of internal thermal stresses by light debonding.²⁵ Another cryogenic fuel application study was completed by Tarpani et al. Within their work, they evaluated the performance of a laminate consisting of a mix of woven carbon fiber and woven glass fiber layers with a polyphenylene sulfide matrix. Samples were exposed to thermal cycling between 100 $℃$ and −196 $℃$ , using boiling water and liquid nitrogen, which replicated the conditions of the fuel containers. Nine samples were exposed to 150, 300, or 500 temperature cycles, and were compared to a control group. Additionally, microscopy was used to qualitatively describe transverse section damage. The study found that temperature cycling induced intralaminar cracking, and the frequency of cracks increased with cycle count. This damage was attributed to the mismatch of CTEs between the fibers and matrix.²⁶

An additional study investigating composite application to cryogenic fuel tanks was completed by Cao et al. In this work, the effects of thermal cycling were studied on carbon-epoxy prepreg laminates by conducting transverse tensile tests to evaluate changes in mechanical behaviour. Samples were divided into three groups that each underwent a different case of thermal cycling. Two of the three groups were subjected to 20-min cycles between either 25

℃

and −196

℃

(case 1) or 80

℃

and −195

℃

(case 2). Samples in these groups would be exposed to either 5, 25, or 50 cycles. In the third group, samples underwent 12-h cycles between 25

℃

and −196

℃

, and were cycled either 2, 4, or 10 times. All the cases were compared to an untreated control group. After thermal cycling, tensile tests were performed on the samples to obtain the transverse tensile strength (TTS) and transverse modulus. The results of the study are summarized in Tables 1–3. Across all three cases, exposure to the greatest number of cycles led to the largest reduction in transverse tensile strength and modulus. TTS decreased by 23.37% to 47.59% across three cases, while transverse modulus decreased by 17.90% to 27.10%. Furthermore, changing the temperature gradient and length of thermal cycle exposure caused differing degrees of property degradation.²⁷ The investigation found that both the range and duration of thermal cycling affected the level of degradation in mechanical properties experienced by the laminate samples. The reduction in TTS was most pronounced by the wide temperature range seen in case 2, while the transverse modulus was impacted by the extended duration of the thermal cycle in case 3. Cycling in case 1 caused the smallest reduction in both TTS and transverse modulus in the samples.²⁷

Table 1.

Reductions in Transverse Tensile Strength (TTS) after exposure to thermal cycling in cases 1 and 2.²⁷.

Cycle count	Case 1 TTS% loss	Case 2 TTS% loss
5	1.65	1.72
25	16.27	42.28
50	23.37	47.59

Table 2.

Reductions in transverse modulus after exposure to thermal cycling in cases 1 and 2.²⁷.

Cycle count	Case 1 transverse modulus% loss	Case 2 transverse modulus% loss
5	7.20	10.50
25	11.98	20.10
50	17.90	22.90

Table 3.

Reductions in TTS and transverse modulus after exposure to thermal cycling in case 3.²⁷.

Cycle count	Case 3 TTS% loss	Case 3 modulus% loss
2	28.77	6.90
4	29.46	11.98
10	39.86	27.10

Another investigation into the effects of varying thermal cycle ranges on various types of carbon/epoxy laminates was completed by Qiu et al. with intended application of these laminates to antenna reflectors. In this study, samples were exposed to three different cycle conditions. Condition 1 exposed samples to temperature fluctuations between −60

° C

and 60

° C

, condition 2’s samples were fluctuated between −60

° C

and 120

° C

, and condition 3’s samples between 0

° C

and 120

° C

. After exposure to thermal cycling, the samples were tested in tension and in-plane shear to obtain the residual values of the tensile and shear moduli. Each temperature range group had varying cycle counts of either 100, 200, 500, or 1000, and were all compared to an unexposed control group. The study found that residual tensile and in-plane shear moduli of the specimens decreased as the number of thermal cycles increased before plateauing around 500 cycles. Condition 1 resulted in the greatest reduction in residual tensile modulus after 1000 cycles, while condition 2 caused the largest reduction in residual shear modulus.²⁸ Table 4 compares the residual tensile and in-plane shear moduli for one of the laminate types after 1000 cycles of each condition.

Table 4.

Residual tensile and in-plane Shear Moduli and the standard deviation of T300/MT0982 uniform laminate samples after 1000 cycles of exposure.²⁸.

Cycle condition	Residual tensile modulus (%)	Residual shear modulus (%)
1	95.51 ± 0.50	95.19 ± 0.47
2	96.05 ± 0.57	84.43 ± 0.82
3	98.19 ± 0.45	90.47 ± 0.78

While there have been several investigations into the degradation of composite materials caused by thermal cycling, the bulk of this work has been completed on laminates and prepreg sheets. Investigation into the behaviour of textile composites in simulated LEO conditions are very limited, including their behaviour under thermal cycling.

Zhang et al. investigated the effects of thermal cycling on the impact behaviour of carbon/epoxy triaxial braided composite panelling. Samples were cycled between −55 $° C$ and 120 $° C$ for up to 160 cycles, and X-ray computed tomography (CT) and acoustic emission were used to observe microcrack formation and distribution. From the characterization methods, microcracks were first observed after 50 cycles were applied and were primarily aligned in the fiber orientation direction.²⁹ An additional study by Gigliotti et al. focused on a non-crimp 3-D orthogonal woven carbon/epoxy composite. Within their work, specimens were subjected to thermal cycling between 50 $° C$ and −50 $° C$ using constant heating and cooling rates of 4 $° C$ /min for either 200, 600, or 1400 cycles. X-ray microcomputed tomography was then used to study and reconstruct 3-D models of the composite samples and search for defects. From their results, the presence and propagation of cracks increased with the number of cycles.³⁰

Microcomputed tomography (μCT) was also used in a study by Cao et al., who investigated thermally cycled 3-D carbon-epoxy interlocked woven fabric composites. Samples in this study were cycled between 200 $° C$ and −20 $° C$ in 40-min durations for either 150, 300, and 450 cycles before being scanned with μCT. From the scans, they found that not only did a greater number of thermal cycles increase the presence of damage in the composite cross-sections, but it also changed the type of damage experienced. At lower thermal cycles, damage in the composites was observed to be due to matrix cracking. At higher thermal cycles, intra-yarn cracks and delamination also became more common.³¹

Using mechanical testing to evaluate the degree of damage, Azimpour-Shishevan et al. completed two investigations. The first studied carbon/epoxy woven textiles cycled between 120°C and −40°C in 20-min cycles, with cycle counts of 0, 20, 40, 60, and 80 cycles. After exposure, the samples underwent tensile and bend testing to evaluate their respective moduli and ultimate strengths in each loading scenario. The mechanical tests revealed that the elastic modulus first saw a slight decrease from zero to 20 cycles before increasing with each subsequent cycle count. This resulted in the elastic modulus at 40, 60, and 80 cycles being higher than the control group. With ultimate tensile strength, 20 cycles caused a large decrease in the property, with each subsequent set of cycles gradually increasing it until it was nearly as high as the control group at 60 and 80 cycles. The behaviour in both cases was attributed to a post curing effect that occurred after the first 20 cycles. From the bend testing, the flexural modulus fluctuated, starting with a slight decrease at 20 cycles, an increase at 40 cycles and 60 cycles, then a larger decrease at 80 cycles. A similar pattern to tensile strength was observed with the flexural strength but with a much smaller decrease at 20 cycles. In both flexural properties, the values of the cycle groups did not deviate far from the control group. Overall, thermal cycling was believed to have caused both degradation and further crosslinking to take place in the composite, leading to the fluctuations in properties observed during testing.³² A repeat of the above experiment was completed on Basalt fiber textile with identical cycle conditions, but they found that thermal cycling seemed to gradually increase the elastic and flexural moduli, as well as the tensile and flexural strengths.³³

Outside of the hazards posed purely by thermal cycling, the exposure of braided composites to elevated temperatures for extended periods of time can have additional adverse effects through thermo-oxidative aging. Long et al. studied carbon/epoxy 3D braided composites that were held at 180 $℃$ for durations of 4, 8, 16, or 32 days using both $μ$ CT to study microcrack development.³⁴ Additionally, a molecular simulation was used to model the behaviour of oxygen molecules as they entered fiber/matrix interfaces. The simulations contained oxygen molecules placed onto a simulated graphene/epoxy interface at either 25 $℃$ or 180 $℃$ . From the $μ$ CT detection results, interface cracks in the braided composites increased in frequency and depth. The development of cracks created channels for oxygen molecules to penetrate deeper into the interior of the material and further degrade resin. The results from the molecular simulations supported this, finding that interface cracking was greater at 180 $℃$ as oxygen molecules moved with greater intensity, diffusing into the simulated structure.³⁴

From the literature, investigations into thermal cycling and its effects on textile composites have been limited, with no specific studies investigating the behaviour of 2D and 3D TBCs or their degradation under thermal cycling. Studies have been especially limited with regards to direct mechanical testing after thermal cycle exposure with textile composite. Although TBCs share the same component materials as other FRPCs used in space applications, their use has been limited as their behaviour in LEO conditions are poorly understood. Accordingly, the objective of this work is to experimentally investigate the effects of thermal cycling in LEO on the tensile strength and the elastic modulus of TBCs intended for space applications. As this is a preliminary investigation introducing TBCs to this method of testing, the selected cycle counts are relatively small at just five and ten total thermal cycles.

Experimental methods

Materials

TBCs were created using 1420 denier Kevlar® 49 Aramid fiber (DuPont, Wilmington, Delaware, USA) and EPON 826 (Hexion Inc., Ohio, USA) mixed with Epikure W (Miller-Stephenson Chemical Co., Danbury, USA). Aramid fiber was selected for this work over carbon fiber due to its generally lower density, reducing the weight of a satellite during and improving a launch vehicle’s fuel efficiency during its deployment.³⁵ Glass fiber was not considered as well due to its higher average density than aramid fiber, as well as its larger positive CTE that makes it less desirable for applications requiring high thermal stability.^9,36 Relevant mechanical and thermal properties of these materials are shown in Table 5.

Table 5.

Thermal and mechanical properties of Kevlar® 49 aramid fiber and Epoxy resin.

Material property	Kevlar® 49⁴¹	Epoxy resin^38,42,43
Longitudinal elastic modulus, $E_{1}$ (GPa)	112.4	0.99–3.17
Ultimate tensile strength (MPa)	3600	23.72–54.09
Longitudinal CTE, $α_{1}$ ( $μ$ mm/mm $℃$ )	−4.86	60

TBC sample manufacturing

Preforms were manufactured using a Maypole braider (K80/HS80 Series 1; Steeger GmbH and Co., Wuppertal, Germany) that formed braids over a 7/16″ diameter steel mandrel. The Maypole braider was loaded with 36 carriers that braided in a diamond braid configuration (one over, one under) onto the mandrel, as the mandrel was drawn with a built-in take up mechanism. Samples were manufactured at three different braid angles (35°, 45° and 55°) as these angles have been thoroughly investigated mechanically and demonstrate a wide range of properties of TBCs.¹³ Lower braid angles (<45°) place fibers in a direction closer to parallel with the axial direction of the TBC, which improves composite’s tensile strength and stiffness. Higher braid angles (>45°) place the fibers further from parallel, resulting in poorer tensile but improved torsional strength. Figure 1 shows a schematic of the setup used to manufacture the braided preforms.

Figure 1.

Annotated schematic of Maypole Braider forming a braided preform over mandrel.

Once manufactured, samples were transferred to a PTFE rod for curing. EPON 826 was mixed with EPIKURE W with a mass ratio of 100:27.2 and then applied to the PTFE mounted preforms via hand lay-up. Upon application of the resin mixture, the TBCs were cured in a programmable oven (Thelco 31,480, GCA/Precision Scientific, Chicago, IL) with a schedule of 1 h at 80

℃

, 1 h at 121

℃

, and 2 h at 177

℃

, following the manufacturer data sheet.³⁷ After curing, TBC samples were cut to a length of 7 inches in a custom mitre cutting box. Prior to testing, each manufactured braid had its inner and outer diameter (

d_{i}

and

d_{o}

, respectively) measured to calculate cross-sectional area using equation (1). Samples were randomly selected from each braid angle for each thermal cycling group. The braid angle of the samples was determined after manufacture by taking images of the braids at a set perpendicular distance using a stereo camera. The angle was then measured digitally on captured images using an angle measurement tool in ImageJ (ImageJ, National Institutes of Health, Bethesda, MD). Table 6 contains the measured braid angles and diameters for each sample, and Figure 2 contains a box plot visualizing the spread of braid measured braid angles across the three groups.

A = \frac{π}{4} (d_{o}^{2} - d_{i}^{2})

(1)

Table 6.

Measured inner diameters, outer diameters, surface areas, and true braid angles for all samples.

Targeted braid angle	Cycle count	Inner diameter (mm)	Outer diameter (mm)	Measured braid angle (°)
35 Degrees	Control (zero cycle)	11.68	12.37	35.95
		11.60	12.39	35.62
		11.62	12.26	34.13
		11.37	12.39	35.85
		11.69	12.42	35.02
	Five cycle	11.58	12.23	36.30
		11.62	12.38	35.72
		11.61	12.35	35.79
		11.51	12.37	35.84
		11.55	12.47	35.32
	Ten cycle	11.33	12.29	34.85
		11.49	12.21	33.96
		11.33	12.14	34.51
		11.62	12.27	36.67
		11.44	12.37	34.06
45 Degrees	Control (zero cycle)	11.57	12.46	43.24
		11.62	12.38	46.79
		11.53	12.56	47.01
		11.71	12.43	45.80
		11.63	12.36	45.01
	Five cycle	11.50	12.33	45.69
		11.54	12.15	44.69
		11.54	12.17	45.98
		11.76	12.26	43.87
		11.66	12.42	45.75
	Ten cycle	11.05	12.33	45.65
		11.56	12.34	46.25
		11.56	12.36	46.17
		11.53	12.24	46.13
		11.24	12.33	43.67
55 Degrees	Control (zero cycle)	11.53	12.42	55.76
		11.50	12.37	57.00
		11.50	12.47	55.63
		11.57	12.48	55.45
		11.58	12.21	55.03
	Five cycle	11.47	12.49	54.64
		11.59	12.41	55.75
		11.53	12.54	54.52
		11.48	12.39	55.16
		11.66	12.58	56.22
	Ten cycle	11.11	12.51	55.18
		11.18	12.22	55.50
		11.22	12.45	55.25
		11.06	12.66	54.62
		11.17	12.55	55.30

Figure 2.

Box plot of the true measured braid angles for each specific braid angle group.

Previous investigations involving Kevlar®/epoxy TBCs completed by Lepp and Carey, and Melenka et al. found average fiber and void volume fractions for TBCs through the use of resin burn-off tests and micro-CT respectively. Samples in this experiment were manufactured in an identical process to the aforementioned studies. Average fiber volume fractions for each braid angle are shown in Table 7, and the average enclosed void volume fraction for TBCs measured from Melenka et al.’s micro-CT investigation was found to be 1.04%.^38,39

Table 7.

Average fiber volume fraction of each braid angle, found using a resin burn off tests.³⁸.

Braid angle	Fiber volume fraction (%)
35 Degrees	25.3 ± 4.3
45 Degrees	33.7 ± 5.1
55 Degrees	46.4 ± 9.9

Thermal cycling

TBCs underwent thermal cycling within a Thermotron™ environmental chamber (SE-1000-6-6 Environmental Chamber, Thermotron, Holland, Michigan, USA). A single thermal cycle in this work was defined as:

• Cooling from ambient temperature to −50 $℃$ at a rate of −1.5 $℃$ /min

• A hold at −50 $℃$ for 1 min

• Heating to a temperature of 120 $℃$ at a rate of 5 $℃$ /min

• A hold at 120 $℃$ for 1 min

• Cooling to room temperature at −1.5 $℃$ /min

A graphical representation of the thermal cycling used in this work is shown in Figure 3. Within the experiment, samples were manufactured at one of three braid angles. For each braid angle, they were then divided into three groups (n = 5) that underwent zero, five, and ten thermal cycles. The temperature rates of change were selected based on the operational capacity of the apparatus.

Figure 3.

Graph of chamber temperature over time of a single thermal cycle.

Tensile testing

To assess the influence of thermal cycling on the mechanical properties, after samples were cycled, tensile testing was completed in accordance with a modified version of ASTM D3039.⁴⁰ This standard was slightly modified to satisfy the unique geometry of TBCs, with the process having been completed and fully outlined in previous work by Melenka and Carey, as well as Lepp and Carey.^13,38 Due to the unique geometry of the samples used within this experiment, TBCs were affixed to the two-part epoxy coated conical end tabs instead of standard test grips. These tabs were designed to allow for a smooth transition of load from the end tabs themselves to the center of each sample’s gauge length, and to allow for continuous strain from the end tabs to the gauge length with fewer discontinuities. As a result, edge effects were minimized during the tensile tests.^13,38 As a result of the design of the end tabs and the smooth distribution of load across the TBCs during testing, the calculation of stress only required the applied load and cross-sectional surface area of each sample.

Sample preparation

Samples were prepared for tensile testing using the following steps:

• Bonding of TBC samples to end tabs with two-part epoxy (Loctite E-20 HP; Hysol, Henkel, Rocky Hill, Co.)

• Spray painting samples with matte black paint (Painter’s Touch Flat Black, Rust-Oleum Corp, Concord, On)

• Speckling with white paint (5212 Opaque White, Createx Airbrush Colors, Createx Colors, East Granby CT) using an airbrush (Custom Micron B, Iwata Medea Inc., Portland, OR)

End tab application allowed for proper sample mounting in the universal testing machine (Instron 1000, Instruments and Systems for Advanced Materials Testing, Canton, Massachusetts, USA). Painting and speckling were completed to allow for deformation tracking using digital image correlation (DIC) during tensile testing. Figure 4 visually outlines the tensile testing sample preparation process, while Figure 5 shows an image of a speckled TBC during a tensile test.

Figure 4.

Process of painting and speckling of TBC samples for DIC.

Figure 5.

Image of speckled TBC obtained during tensile testing.

Tensile testing

Quasi-static tensile tests were performed in a universal testing machine (Instron 1000, Instruments and Systems for Advanced Materials Testing, Canton, Massachusetts, USA). Load data was collected using a data acquisition system (NI-USB 6211 DAQ, National Instruments, Austin, Texas). Tensile tests were conducted at a displacement rate of 2 mm/min until ductile failure was visually observed. This displacement rate was selected to ensure that samples failed within 1–10 min as specified by the ASTM D3039 standard that was adapted in this work.⁴⁰

To measure sample strain, 3D stereo DIC was used. Two stereo cameras (AVT GT3400, Allied Vision Technologies, Stadtroda, Germany) were mounted symmetrically at an angle of separation of 22° and focused between the two crossheads where test samples would be mounted. Additionally, the experimental area was illuminated to provide better image quality and brighter contrast on the sample speckle pattern. Prior to testing, the cameras were calibrated to translate pixel coordinates to spatial coordinates. An in-house MATLAB® code was then used to begin the test, collecting load data and capturing the images from the cameras. Images were collected at a rate of 0.5 Hz, and load data was collected at 100 Hz. Figure 6 features an annotated image of a mounted sample during testing.

Figure 6.

Image of a sample mounted in the universal testing machine with dual stereo cameras.

Digital image correlation

Digital image correlation was completed using DaVis™ (DaVis version 8.2.0 StrainMaster 3D, LaVision GmbH, Gottingen, Germany).

• Image sets of the sample were merged from both cameras, filtered with a gaussian sliding average non-linear filter, then processed with 3-D DIC.

• The processed images were used to make a strain map which visually displayed the magnitude and location of strain on the sample.

• A built-in plotting tool was used to convert data from the strain map into a plot of average strain in each subsequent image.

Data analysis

To create stress-strain curves, load data from MATLAB™ and strain data from DaVis were combined. The collected load data was filtered in MATLAB script using the “filtfilt” function. Load for each sample was converted into stress by using the cross-sectional surface area of each sample, which was calculated using the inner and outer diameters of each sample. Individual stress values were calculated using equation (2).

σ = \frac{P}{A}

(2)

This stress data was then plotted against the output strain values from DaVis to produce stress strain curves for each sample.

Ultimate tensile stress (UTS) and elastic modulus were obtained from these plots. UTS was determined by taking the peak value of stress from each curve, while the elastic modulus was calculated from the slope of the linear-elastic region within the stress-strain curves.

Results & discussion

Elastic modulus and ultimate tensile strength (UTS) results

Average stress-strain curves for each braid angle and cycle group are displayed in Figures 7–9 to provide a visualization of the effects of thermal cycling on the TBCs.

Figure 7.

Average stress-strain curves for the 35-degree TBCs.

Figure 8.

Average stress-strain curves for the 45-degree TBCs.

Figure 9.

Average stress-strain curves for the 55-degree TBCs.

Bar graphs displaying the results and the variability between the average elastic moduli and UTS are displayed in Figures 10 and 11.

Figure 10.

Bar graphs of the average elastic modulus for each braid angle and cycle group.

Figure 11.

Bar graphs of the average ultimate tensile strength for each braid angle and cycle group.

The elastic moduli (E) for each braid angle and cycle count were calculated and averaged. Tables 8–10 contain the average elastic moduli of the samples and their standard deviations, grouped by braid angle and cycle exposure.

Table 8.

Average elastic modulus and change from control samples for 35-degree TBCs.

35 Degrees
Cycle exposure	Average E (GPa)	Change from control group (%)
Control (zero cycles)	5.67 ± 1.09	—
(n = 5)	5.67 ± 1.09	—
Five cycles	4.73 ± 0.24	−16.6
(n = 5)	4.73 ± 0.24	−16.6
Ten cycles	4.70 ± 1.02	−17.0
(n = 5)	4.70 ± 1.02	−17.0

Table 9.

Average elastic modulus and change from control samples for 45-degree TBCs.

45 Degrees
Cycle exposure	Average E (GPa)	Change from control group (%)
Control (zero cycles)	3.11 ± 0.61	—
(n = 5)	3.11 ± 0.61	—
Five	3.39 ± 0.66	+8.9
(n = 5)	3.39 ± 0.66	+8.9
Ten	2.22 ± 0.48	−25.8
(n = 5)	2.22 ± 0.48	−25.8

Table 10.

Average elastic modulus and change from control samples for 55-degree TBCs.

55 Degrees
Cycle exposure	Average E (GPa)	Change from control group (%)
Control (zero cycles)	2.46 ± 0.67	—
(n = 5)	2.46 ± 0.67	—
Five	2.23 ± 0.19	−9.3
(n = 5)	2.23 ± 0.19	−9.3
Ten	1.70 ± 0.27	−30.7
(n = 5)	1.70 ± 0.27	−30.7

Similarly to the elastic modulus, UTS values were obtained from each trial, averaged, and compared to the control groups with percentage change. Tables 11–13 contain the average UTS of the samples and their standard deviations grouped by braid angle and cycle exposure.

Table 11.

Average UTS and change from control samples for 35-degree TBCs.

35 Degrees
Cycle exposure	Average UTS (MPa)	Change from control group (%)
Control (zero cycle)	109.33 ± 16.03	—
(n = 5)	109.33 ± 16.03	—
Five	94.22 ± 15.02	−13.8
(n = 5)	94.22 ± 15.02	−13.8
Ten	72.70 ± 7.15	−33.5
(n = 5)	72.70 ± 7.15	−33.5

Table 12.

Average UTS and change from control samples for 45-degree TBCs.

45 Degrees
Cycle exposure	Average UTS (MPa)	Change from control group (%)
Control (zero cycle)	62.06 ± 18.66	—
(n = 5)	62.06 ± 18.66	—
Five	66.57 ± 8.96	+7.3
(n = 5)	66.57 ± 8.96	+7.3
Ten	48.27 ± 9.83	−22.2
(n = 5)	48.27 ± 9.83	−22.2

Table 13.

Average UTS and change from control samples for 55-degree TBCs.

55 Degrees
Cycle exposure	Average UTS (MPa)	Change from control group (%)
Control (zero cycle)	31.12 ± 6.44	—
(n = 5)	31.12 ± 6.44	—
Five	23.31 ± 1.77	−25.1
(n = 5)	23.31 ± 1.77	−25.1
Ten	19.86 ± 4.42	−36.2
(n = 5)	19.86 ± 4.42	−36.2

Similarly to the average elastic moduli, the 35- and 55-degree sample groups had a decrease in the average UTS with increasing thermal cycles while the 45-degree TBCs had the average UTS increase at five trials then decrease at 10 trials.

Statistical verification

To confirm the influence of thermal cycling on mechanical properties, single factor ANOVA tests were conducted on the 35-, 45-, and 55-degree trial groups, with the factor of interest being thermal cycle count. Prior to the completion of ANOVA, the experimental data needed to be verified for normality and equal variance.

Normality was verified using the Kolmogorov-Smirnov test with a 0.95 confidence level (

α

= 0.05). This test was completed on the elastic modulus and UTS results for each braid angle and cycle count group. With the selected confidence level and the sample count of five across all test groups, the Kolmogorov-Smirnov scores for each group needed to be less than 0.5633 in order fail to reject the null hypothesis and have normal distributions. Tables 14 and 15 contain the Kolmogorov-Smirnov critical values for elastic modulus and UTS respectively and found that all the sample groups were normally distributed.

Table 14.

Kolmogorov-Smirnov scores for elastic modulus results.

Elastic modulus
Braid angle	Cycle count	Kolmogorov-Smirnov output
35 $°$	Zero	0.1665
	Five	0.2482
	10	0.1536
45 $°$	Zero	0.1739
	Five	0.2109
	10	0.2612
55 $°$	Zero	0.2110
	Five	0.1636
	10	0.1601

Table 15.

Kolmogorov-Smirnov scores for UTS results.

UTS
Braid angle	Cycle count	Kolmogorov-smirnov output
35 $°$	Zero	0.2791
	Five	0.1605
	10	0.1998
45 $°$	Zero	0.1644
	Five	0.2306
	10	0.2236
55 $°$	Zero	0.2079
	Five	0.2311
	10	0.1666

Equality of variance was determined using Barlett’s test, which was completed for the elastic modulus and UTS of the samples grouped by braid angle. With a 0.95 confidence level and two degrees of freedom, the chi-squared test statistic for this test needed to be less than

χ_{2, 0.05}^{2}

= 5.991 to fail to reject the null hypothesis that the variances are equal. Results of the test are shown in Table 16, and equal variances were found for each braid angle.

Table 16.

Bartlett’s test critical values for mechanical test results grouped by braid angle.

Braid angle	Elastic modulus $χ^{2}$	UTS $χ^{2}$
35°	5.600	1.924
45°	0.3459	2.059
55°	5.136	4.051

With the baseline assumptions confirmed, single factor ANOVA tests were completed for the 35-, 45-, and 55-degree TBCs, with the thermal cycle count being the factor of interest. All the ANOVA tests were completed with a confidence level of 0.95 (

α

= 0.05), and the results of the ANOVA tests are shown in Table 17, where significant p-values are bolded.

Table 17.

Results of the ANOVA tests for each braid angle group with significant p-values bolded.

Braid angle	Average elastic modulus p-value	Average UTS p-value
35°	0.245	0.00731
45°	0.0577	0.168
55°	0.0736	0.0140

With the 35- and 55-degree groups displaying significant differences between mean UTS values according to the ANOVA test, Tukey’s test was completed on the data to determine the specific group means that contained significant differences. Results for the 35- and 55-degree UTS groups can be found in Tables 18 and 19, and mean differences that exceeded the respective Tukey critical value are bolded.

Table 18.

Results for Tukey’s test on the 35-degree UTS means.

35 $°$ UTS (Tukey critical value = 25.14)
Compared group means	Group difference
Five - zero	15.11
Ten - zero	36.63
Ten - five	21.52

Table 19.

Results for Tukey’s test on the 55-degree UTS means.

55 $°$ UTS (Tukey critical value = 8.719)
Compared group means	Group difference
Five - zero	7.811
Ten - zero	11.26
Ten - five	3.447

Discussion

The aim of this preliminary investigation was to evaluate the effects of thermal cycling on the mechanical properties of Kevlar®/Epoxy TBCs, with comparisons of these effects across different cycle counts and braid angles. The 35-degree braids exhibited the highest average elastic moduli and UTS across all cycle counts, followed by the 45-degree then the 55-degree braids. This behaviour is corroborated by previous TBC experimentation investigating the effects of braid angle on the tensile properties of TBCs.¹³ Smaller braid angles place the stronger fibers more parallel to the axial direction of the TBC, which allows for greater resistance to tensile deformation. These properties decrease with increasing braid angle. Figure 12 displays how the braid angle is measured on a TBC.

Figure 12.

Visual representation of braid angle on a cured TBC.

The results of this study show that thermal cycling must be accounted for to implement TBCs in structural applications in LEO. From the data, thermal cycling significantly influenced the UTS of the 35- and 55-degree TBCs. In the 35- and 55-degree TBCs, both the average elastic modulus and average UTS decreased with an increasing thermal cycle count. This was expected due to the mismatch of CTE’s between the fibers and epoxy. As the temperature changed, the two material components expanded and contracted at different rates. This difference in expansion and contraction between materials is further emphasized with the use of Kevlar® fibers which feature a negative CTE.⁴¹ The mismatch in properties likely caused the development of stress between material constituents, and the fluctuation in temperature created cyclical loading within the TBCs. As a result, higher cycle counts caused a greater level of degradation to occur within the material, resulting in poorer mechanical performance. To verify this structural degradation, visual characterization of sample surfaces using SEM is needed such as in Qiu et al.’s study of thermally cycled carbon/epoxy laminates.²⁸ This overall decrease in mechanical performance for the 35- and 55-degree sample groups agrees with the findings from Cao and Qiu et al., who also compared mechanical property degradation across varying thermal cycle count and found the same relationship.^27,28

DIC strain maps revealed a concentration of higher strain occurring in resin rich regions of the TBC samples, with this strain occurring earlier in the photosets of samples that were weakened by higher exposure to thermal cycling. This pattern of strain concentration has been observed previously in work completed by Lepp et al. in which they studied the localized degradation of TBCs under quasi-static tensile testing.³⁸ The earlier occurrence of this strain at higher cycle counts may have been the result of weakening at the fiber-matrix interfaces within the samples. An example of this is shown through the comparison of strain maps of a 45-degree control sample and a 45-degree 10 cycle sample shown in Figure 13, where a decrease in the average elastic modulus and UTS was also observed after 10 cycles.

Figure 13.

Comparisons of concurrent strain maps of two 45-degree TBCs during tensile testing at (a) 20 s, (b) 40 s, (c) 60 s, (d) 80 s, and (e) 100 s. Images on the left are the strain maps of a 45-degree TBC sample that has undergone no thermal cycling, while images on the right side correspond to a 45-degree TBC that has experienced 10 thermal cycles. The development of strain occurs sooner in the sample that has experienced 10 cycles, which corresponds to the lower average elastic modulus and ultimate tensile strength.

To further verify the weakening of the TBC structure due to mismatched CTE, images of the points of failure from the tensile tests were analyzed as well. Figure 14 displays images of TBCs that failed in tension where the point of failure was captured by the stereo cameras.

Figure 14.

Failure locations of two 55 $°$ TBCs after (a) zero and (b) 10 thermal cycles. Both samples exhibit a contraction of the TBC diameter resembling ductile necking, but in reality it is the result of brittle failure of the matrix phase.

The observed failure mode within the TBC samples in this work resembles a “pseudo-necking” pattern. Within this pattern, the brittle failure of the polymers within the TBC caused the fibers of the TBC to fail to maintain their defined structure, resulting in a contraction of the TBC’s diameter as the freed fibers gradually moved closer to each other. The weakening of the interactions between the fiber and matrix of the TBCs caused this effect to occur sooner in the tensile tests within the 35- and 55-degree TBCs that had undergone a higher degree of thermal cycling.

The 45-degree samples appeared to partially contradict expected behavior, as both the average elastic modulus and UTS saw an increase from the control to the five cycle groups before decreasing after 10 cycles. This conflicts with the expected decrease of the properties seen in the 35- and 55-degree groups as well as experimental findings from Cao and Qiu.^27,28 Additionally, the increase in mechanical performance from zero to five cycles would potentially suggest that the cyclical loading degradation did not occur with this group.

This increase is hypothesized to be the effect of post-curing within the TBCs caused by thermal cycling. This effect was observed within the work of Azimpour-Shishevan et al., in which the elastic modulus seemed to initially decrease after 20 cycles, but then increased with each successive group of 20 cycles until the eighty-cycle sample group featured an average elastic modulus exceeding that of the control group. When evaluating the ultimate tensile strength, they found an initial steep degradation after 20 cycles, but another gradual increase with each additional set of 20 cycles.³² Investigations were completed with Carbon/Epoxy and Basalt/Epoxy, and this behavior was found in both experiments.^32,33 In their work, thermal cycling was believed to have caused some post-curing effect that perhaps outperformed the potential thermal cycling induced damage. While it could explain the pattern observed in the 45-degree group, this effect was not observed in the 35- and 55-degree groups tested in this work.

A speculative reason for the occurrence of the post-cure strengthening occurring exclusively in the 45-degree samples may be due to the uniqueness of their geometry when compared to the 35- and 55-degree sample groups. 35-degree TBCs are typically manufactured with a more open mesh overall than 45- and 55-degree variants, which results in fewer regions for the epoxy matrix to bind fibers together. This could have caused a post-curing effect to fail to overcome the gradual weakening of the TBC structure during thermal cycling exposure. While 55-degree samples also have a very tight mesh that increases the area for matrix and fiber bonding, they place the reinforcement fibers within the TBC at a disadvantageous angle for tensile loading, possibly resulting in these samples still seeing a reduction in mechanical performance with increasing thermal cycle count. As the 45-degree TBCs balance a closed mesh for adequate matrix coverage and void reduction with a braid angle that can provide greater strength in tensile loading, this may be a hypothetical cause for the increase in average elastic modulus and UTS after the initial five thermal cycles. However, this cannot be confirmed without verification of a change in the matrix’s glass transition temperature and the collection of curing degree data.

From the ANOVA tests, it was found that there were insignificant differences between the average elastic moduli of all braid angle trials, and the UTS of the 45-degree trials. However, the ANOVA tests did reveal a significant difference in the average UTS for the 35- and 55-degree sample groups across cycle count. The completion of Tukey’s tests for the 35- and 55-degree UTS datasets revealed that the source of significant difference in the means was the result of the difference between the control and 10 cycle groups within both braid angles. Variability among the results appeared to be largest among the 35-degree samples and decreased as the braid angle increased. This was due to the difficulty in maintaining the structure of the TBCs with lower braid angles during the manufacturing process. 35-degree braids will have a more open mesh than 45- and 55-degree braids. This results in a larger likelihood of braid strands being shifted slightly out of place during the transfer of preforms from the mandrel to curing rods, as well as during the hand lay up of epoxy into the braids.

Conclusion

This preliminary study investigated the effects of thermal cycling on Kevlar®/Epoxy tubular braided composites (TBCs) and introduced this class of composite materials to thermal cycling test methods. This was done to evaluate the viability of TBCs in space applications where they have seen limited use using a relatively small number of thermal cycles. Particularly, TBCs designed for this experiment were intended for use in structural reinforcement applications. Samples were manufactured at braid angles of either 35°, 45°, or 55°. Of each braid angle, samples were further categorized into a control group or were exposed to either five or 10 cycles of thermal cycling between −50 $℃$ and 120 $℃$ . After thermal cycling, they underwent tensile testing with digital image correlation to evaluate the change in elastic modulus and ultimate tensile strength (UTS) after treatment.

The 35-degree average elastic modulus decreased by 16.6% and 17.0% after five and 10 cycles respectively. The 45-degree average elastic modulus increased by 8.9% after five cycles, then decreased by 25.8% after 10 cycles. The 55-degree average elastic modulus decreased by 9.3% after five cycles then decreased by 30.7% after 10 cycles.

With the average UTS, the 35-degree samples saw a decrease of 13.8% after five cycles then a 33.5% decrease after 10 cycles. The 45-degree samples saw a 7.3% increase in average UTS after five cycles and a decrease of 22.2% after 10 cycles. The 55-degree sample group’s average UTS decreased 25.1% after five cycles followed by a 36.2% decrease after 10 cycles.

After using Kolmogorov-Smirnov tests to verify normality of the results and Bartlett’s test to confirm equal variances, single factor ANOVA tests were conducted on the elastic modulus and UTS data with a 0.95 confidence level. The tests revealed that significant differences were found in the average UTS values of the 35- and 55-degree sample groups. Tukey’s test was then used to identify which pairs of cycle groups exhibited a significant difference within the braid angle groups. This post-hoc test found that in the both the 35- and 55-degree sample groups, the means were significantly different between the control and 10 cycle average UTS values.

The degradation in mechanical properties was expected in the 35- and 55-degree sample groups and agree with much of the established literature. However, the 45-degree group saw an unexpected increase after five cycles due to a hypothesized post curing interaction. As the 45-degree samples did not exhibit a statistically significant difference in the average mechanical property values, this configuration of TBC will require additional evaluation with higher cycle counts to fully verify their suitability in satellite structural applications. With an overall reduction in the stiffness and strength of the 35- and 55-degree TBCs after relatively small durations of cycle exposure, these samples will need to have their designs further developed to minimize the effects of thermal cycling, and these configurations are unsuitable for use in satellite structures. While their thermal expansion as structures may be small, testing different material combinations and braid angles may be necessary to design an TBC that may be best suited for use in satellites.

Footnotes

ORCID iDs

Daniel Gye

Ahmed S. Ead

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

References

Canadian Space Agency . Monitoring and protecting our ecosystems from space. Canadian Space Agency, 2022. https://www.asc-csa.gc.ca/eng/satellites/everyday-lives/monitoring-and-protecting-our-ecosystems-from-space.asp (Accessed 26 March 2025).

Canadian Space Agency . Space serving the arctic and the great Canadian north. Canadian Space Agency, 2019. https://www.asc-csa.gc.ca/eng/satellites/everyday-lives/space-serving-the-arctic-and-the-great-canadian-north.asp (Accessed 26 March 2025).

Plante

Lee

. Environmental conditions for space flight hardware – a survey. Nasa Electronic Parts and Packaging Program, 2004.

Finckenor

. Chapter 6: materials for spacecraft. Aerospace Materials and Applications.

Bell

Williams

. Sputnik 1. NASA space science data coordinated archive, 2022. https://www.nasa.gov/image-article/sputnik-1/ (Accessed 26 February 2025).

Williams

Starke

. Progress in structural materials for aerospace systems. Acta Mater 2003; 51: 5775–5799. https://doi.org/10.1016/j.actamat.2003.08.023

Xia

. Overview of aerospace materials and their applications. Theor Nat Sci; 13: 20240808. https://doi.org/10.54254/2753-8818/13/20240808. Epub ahead of print 2023.

Materials

AZO

. Aluminium/aluminum 6061 alloy (UNS A96061). AZO Materials 2012. https://www.azom.com/article.aspx?ArticleID=6636 (Accessed 23 November 2025).

Chu

Huang

, et al. Mechanical and thermal expansion properties of glass fibers reinforced PEEK composites at cryogenic temperatures. Cryogenics 2010; 50: 84–88. https://doi.org/10.1016/j.cryogenics.2009.12.003

10.

Altin

Gökkaya

. A review on machinability of carbon fiber reinforced polymer (CFRP) and glass fiber reinforced polymer (GFRP) composite materials. Def Technol 2018; 14: 318–326. https://doi.org/10.1016/j.dt.2018.02.001

11.

Pradere

Sauder

. Transverse and longitudinal coefficient of thermal expansion of carbon fibers at high temperatures (300–2500 K). Carbon N Y 2008; 46: 1874–1884. https://doi.org/10.1016/j.carbon.2008.07.035

12.

Melenka

Hunt

van Ravenhorst

, et al. Manufacturing processes for braided composite materials. Handbook of advances in braided composite materials: theory, production, testing and applications. Woodhead Publishing, 2017, pp. 47–153.

13.

Melenka

Carey

. Experimental analysis of diamond and regular tubular braided composites using three-dimensional digital image correlation. J Compos Mater; 51: 0021998317695418. https://doi.org/10.1177/0021998317695418. Epub ahead of print 2017.

14.

Carey

Melenka

Hunt

, et al. Introduction to braided composite material behavior. Handbook of advances in braided composite materials: theory, production, testing and applications. Elsevier Inc., 2017, pp. 207–237.

15.

Carey

. 1 - introduction to braided composites. In: Carey

(ed). Handbook of advances in braided composite materials. Woodhead Publishing, pp. 1–21.

16.

Fouchal

Ramirez

Beloreshka

, et al. Comparative evaluation of spacecraft materials properties under simulated and true space environments. J Astronaut Sci 2024; 71: 53. https://doi.org/10.1007/s40295-024-00476-1

17.

Johnson

Sleight

Martin

. Structures and design phase I summary for the NASA composite cryotank technology demonstration project. In: Collection of technical papers - AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics and materials conference, Boston, Massachusetts, 8–11April 2013.

18.

Yost

. 6.0 structures, materials, and Mechanisms– state of the art of small spacecraft technology. National Aeronautics and Space Administration.

19.

Fernandez

. Advanced deployable shell-based composite booms for small satellite structural applications including solar sails. Inter Sympos Solar Sail; 1.

20.

Ming

Liu

, et al. Space deployable parabolic reflector based on shape memory polymer composites. Compos Struct 2023; 304: 116327. https://doi.org/10.1016/j.compstruct.2022.116327

21.

Keller

Laket

Francist

, et al. Development of a deployable boom for microsatellites using elastic memory composite material. In: Collection of technical papers - AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics and materials conference, Palm Springs, California, 19–22 April 2004. https://doi.org/10.2514/6.2004-1603. Epub ahead of print 2004.

22.

Zeng

Kang

Duan

, et al. Development of polymer composites in radiation shielding applications: a review. J Inorg Organomet Polym Mater; 33(8): 1–49. https://doi.org/10.1007/s10904-023-02725-6. Epub ahead of print 2023.

23.

Park

Choi

, et al. An experimental study into aging unidirectional carbon fiber epoxy composite under thermal cycling and moisture absorption. Compos Struct 2019; 207: 81–92. https://doi.org/10.1016/j.compstruct.2018.08.069

24.

Boccaccini

Strutt

Vecchio

, et al. Behavior of Nicalon^TM-fiber-reinforced glass-matrix composites under thermal cycling conditions. Compos Part A Appl Sci Manuf 1998; 29: 1343–1352. https://doi.org/10.1016/s1359-835x(98)00063-3

25.

Xiao

Huang

, et al. Effects of cryo-thermal cycling on interlaminar shear strength and thermal expansion coefficient of carbon fiber/graphene oxide-modified epoxy composites. Compos Commun 2022; 32: 101180. https://doi.org/10.1016/j.coco.2022.101180

26.

Tarpani

ACSP

Goyo Brito

Tarpani

. Effects of simultaneous hygrothermal aging and thermal shock cycling on carbon fiber-reinforced polyphenylene sulfide composite performance. Polym Compos 2025; 46: 11922–11940. https://doi.org/10.1002/pc.29719. Epub ahead of print 10 September 2025.

27.

Cao

Zhang

, et al. Experimental investigation of thermal cycling effects on the transverse tensile properties of T800 carbon fiber/epoxy composites. J Compos Mater 2025; 0: 00219983251379934. https://doi.org/10.1177/00219983251379934

28.

Qiu

Zhang

, et al. On the mechanical behavior of carbon fiber/epoxy laminates exposed in thermal cycling environments. Thin-Walled Struct 2024; 196: 111481. https://doi.org/10.1016/j.tws.2023.111481

29.

Zhang

Binienda

Morscher

, et al. Experimental and FEM study of thermal cycling induced microcracking in carbon/epoxy triaxial braided composites. Compos Part A Appl Sci Manuf; 46: 006. https://doi.org/10.1016/j.compositesa.2012.10.006. Epub ahead of print 2013.

30.

Gigliotti

Pannier

Sinchuk

, et al. X-ray μcT based assessment of thermal cycling induced cracks in non-crimp 3D orthogonal woven composite materials with porosity. In: IOP conference series: materials science and engineering, Melbourne, Australia, 15–18 September 2018. https://doi.org/10.1088/1757-899X/406/1/012008. Epub ahead of print 2018.

31.

Miao

Zhiwei

Qiaoyi

, et al. Anisotropic thermal expansion behaviors and damage evolution mechanisms of 3D angle-interlocked woven composites subjected to thermal cycling. Polym Compos 2025; 42(2): 1661–1675. https://doi.org/10.1002/pc.70243. Epub ahead of print.

32.

Azimpour-Shishevan

Akbulut

Mohtadi-Bonab

. Thermal shock behavior of twill woven carbon fiber reinforced polymer composites. J Compos Sci 2021; 5: 33. https://doi.org/10.3390/jcs5010033. Epub ahead of print 2021.

33.

Azimpour-Shishevan

Akbulut

Mohtadi-Bonab

. Effect of thermal cycling on mechanical and thermal properties of basalt fibre-reinforced epoxy composites. Bull Mater Sci 2020; 43: 88. https://doi.org/10.1007/s12034-020-2059-y. Epub ahead of print 2020.

34.

Long

Sun

, et al. Multi-scale dynamic behaviors of oxygen entering carbon fiber/epoxy resin interfaces under thermal environment. Compos Sci Technol 2023; 242: 110165. https://doi.org/10.1016/j.compscitech.2023.110165

35.

Ștefania

Cerbu

Hadăr

. Characteristics of carbon and kevlar fibres, their composites and structural applications in civil engineering—A review. Polymers; 16: 16010127. https://doi.org/10.3390/polym16010127. Epub ahead of print 2024.

36.

Loos

Abetz

Schulte

. Composites. Carbon Nanotube Rein Compos 2015; 212: 37–72.

37.

Miller-Stephenson. EPIKURE W. Miller-Stephenson. https://miller-stephenson.com/chemicals/epoxy-resins/epikure-curing-agents/epikure-w-composites/ (2025, Accessed 26 November 2025).

38.

Lepp

Carey

. An examination of initial structural degradation in tubular braided composites through region-by-region strain analysis. J Eng Fibers Fabrics; 15: 1558925020978325. https://doi.org/10.1177/1558925020978325. Epub ahead of print 2020.

39.

Melenka

Lepp

Cheung

BKO

, et al. Micro-computed tomography analysis of tubular braided composites. Compos Struct 2015; 131: 057. https://doi.org/10.1016/j.compstruct.2015.05.057. Epub ahead of print 2015.

40.

International

. D3039/D3039M-08: standard test method for tensile properties of polymer matrix composite materials. Adv Stand Trans Market 2014; 4; 1–13.

41.

DuPont . Kevlar aramid fiber technical guide. DuPont.

42.

West System . Compare epoxy physical properties. West System, 2025. https://eu.westsystem.com/compare-epoxy-physical-properties/ (Accessed 25 November 2025).

43.

Wang

Moghaddam

Moreno

, et al. Coefficient of thermal expansion of Nd-Fe-B magnetic particle polymer composites – experiments and stochastic finite element modeling. Results Phys 2025; 74: 108310. https://doi.org/10.1016/j.rinp.2025.108310