Internal pressure testing of open-ended filament-wound cylinders using radial expansion of elastomeric inserts

Abstract

A novel approach for generating internal pressure in filament-wound cylinders using the compressed elastomer method is investigated. Pressure is exerted on the inner surface of the cylinder by radially expanding elastomeric discs inside the composite cylindrical shell. This is accomplished by applying a uniaxial compressive load to the discs with the aid of rigs placed between compression platens in an ordinary universal testing machine. The digital image correlation (DIC) technique is employed to assist the tests in mapping both strain and displacement fields during the entire testing process. Cylinders with six different winding angles (±35°, ±45°, ±55°, ±65°, ±75°, and ±90°) are prepared and analysed. Their load-carrying capacity increases up to an angle of ±75°, accompanied by systematic changes in deformation and failure mechanisms. The suggested testing setup offers benefits in evaluating the impact of processing parameters on the internal pressure strength of filament-wound cylinders. It serves as a secure and cost-efficient laboratory-scale alternative to conventional hydrostatic pressure testing.

Keywords

Internal pressure test digital image correlation open-ended cylinders uniaxial compression filament winding

Introduction

Fibre-reinforced cylindrical shells have proven to be effective as load-bearing elements in pipes^1–4 and pressure vessels^5–8 for transmitting and storing pressurised fluids. They are especially advantageous for operations in harsh environmental conditions^9,10 and in applications where weight is critical,^11,12 as in hydrogen storage.^13–15 The safe operation of these structures under internal pressure is closely related to their radial deformation, which is controlled by hoop tensile properties, including tensile stiffness, ultimate strain, and strength.¹⁶

Filament winding (FW) is among the most important composite manufacturing techniques,¹⁷ being particularly well-suited for producing hollow solids of revolution, where reinforcing fibres are positioned in an oriented pattern for enhanced structural efficiency.¹⁸ The process is not only cost-effective, with minimal material waste,¹⁹ but also meets the high fibre volume requirements of high-performance structures.²⁰ Nevertheless, the filament-wound layer is inherently intricate, exhibiting a distinctive regular mosaic pattern, consisting of triangular-shaped two-ply units with alternating $\pm ϕ$ and $\mp ϕ$ reinforcement arranged around the circumference and along the axis of rotation.²¹

As usual for laminates with continuous fibres, the mechanical behaviour of filament-wound parts is strongly dependent on the fibre direction.²² Moreover, FW manufacturing parameters, such as fibre tension,²³ and possible deviations from design specifications (e.g., layer thickness and winding angles)²⁴ exert significant influence on the mechanical response of these structures.²⁵ These uncertainties and variability may affect the range of operating pressure, possibly compromising the strength of the composite.²³ Therefore, pressure vessels for critical applications demand at least one valid burst test for each batch of cylinders, and the difference between burst pressure and maximum service pressure determines the true margin of safety.²⁶

There are some experimental methods available for determining burst pressure and hoop tensile properties of composite vessels, including hydrostatic pressure and ring specimen tests. In the former, the structure is filled in with a liquid, usually water or oil, and the pressure increases up to failure. While this method yields reliable data, it carries notable risks when a violent burst occurs. In addition, it is costly due to the need for specialised equipment, including a pump, thick flanges with O-rings for sealing, and axial beams to prevent the axial displacement of the cylinder and the flanges from being pushed off the sample.^27–29 In ring tests, a tensile load is applied on a ring specimen extracted from cylinders,^30–33 being much simpler and less expensive than hydrostatic pressure tests. However, they present an important limitation related to the difficulty in assessing the behaviour of helical layers of filament-wound cylindrical shells.³⁴

An alternative approach was proposed by Mosley,³⁵ a burst pressure test in which the cylinder is partially pressurised by an elastomeric insert inside it. The insert is axially loaded by two pistons, causing its radial expansion and contact with the cylinder under testing. This was the basis for establishing the standard test method ASTM C1819.³⁶ The accuracy of this method was investigated by Carter³⁷ using a steel control sample, and the internal pressure was calculated from linear elastic and hyper-elastic equations. The values were checked with a pressure value found via a Lamé cylinder analysis of the experimental hoop strain. The values from linear elastic, hyper-elastic, and Lamé solutions showed sharp agreement, and notably high pressures, exceeding 340 MPa.

Some works in the literature have exploited this experimental approach. Vandeput³⁸ used a polyurethane (PU) insert to assess the hoop tensile strength distribution of cemented tungsten carbide tubes. The results were correlated with the transverse strength distribution from three-point bending tests using statistical fracture theories. Hong et al.³⁹ reported a miniaturised version of the method for tubular ceramic specimens of commercial alumina and silicon carbide. A PU cylindrical rod was utilised, and statistically consistent strength data compared to diametrical loading was obtained. Jacobsen et al.⁴⁰ measured the hoop strength of tubular silicon carbide fibre-reinforced composite specimens using C-ring and compressed elastomer tests, and the results showed agreement within a 6% margin. Shapovalov et al.⁴¹ assessed different test methods for open and closed ends tubes of reinforced silicon carbide matrix composite. They measured pressure at the proportional limit and at ultimate failure and analysed the strain fields using DIC. While the closed-end hydraulic burst test reached the proportional limit at higher pressure, the compressed elastomer method yielded higher elastic modulus, hoop strain, and ultimate failure pressure. Nevertheless, the results were considered consistent given the typical scatter in properties for ceramic composites.

Optical methods using DIC are an excellent tool to study the behaviour of composites under mechanical loading. The DIC technique is most used to measure deformation and displacement fields of parts by tracking the grey value pattern in small neighbourhoods, referred to as subsets, during straining. Castro et al.⁴² introduced a methodology for measuring geometric imperfections of 3D filament-wound cylinders with DIC and only one pair of cameras. The developed procedure is depicted in detail allowing straightforward implementation to other structures and includes proper lighting, focus adjustment, calibration, environment, stitching process, and surface reconstruction. Later, the authors exported the geometric imperfections to finite element (FE) software to generate FE meshes using experimentally measured imperfections.⁴³

Given the safety concerns associated with hydrostatic pressure testing and the inherent restrictions of ring specimen tests, it is worth exploring alternative testing methods. The compressed elastomer technique shows desirable features, although it has not yet been properly employed for filament-wound cylinders. Therefore, this work proposes and evaluates a laboratory-scale testing setup for pressurising such cylindrical shells by the radial expansion of elastomeric discs that undergo compression loading through rigs in an ordinary universal testing machine. The deformation response is monitored by DIC, and several winding angles are examined and discussed.

Experimental details

The design of composite cylindrical shells is developed in CADWIND^® software,⁴³ with two antisymmetric angle-ply $(+ ϕ / - ϕ)$ layers for six different angles: ±35°, ±45°, ±55°, ±65°, ±75°, and ±90°. The helical windings (±35° to ±75°) are carried out with a winding pattern of two. The ±90° winding is accomplished by the hoop method, using a very small helical angle, with a nominal deviation of 0.4° from pure hoop winding.⁴⁴

The cylinders are manufactured by winding carbon fibre tows pre-impregnated with epoxy resin (SIGRAFIL^® C T24-5.0/270-E100, from SGL Carbon) around a cylindrical stainless-steel mandrel with a 50 mm diameter, using the desktop filament winder machine X-Winder, model 4X-23 (Figure 1). Before each winding, a release agent (S-31, from Jost Chemicals) is applied to the surface of the mandrel. Shrink tape is wrapped after the end of the winding to aid consolidation of the wound layers. The system is cured in a horizontal oven with air circulation at a heating rate of 3°C/min up to 140°C and held at that temperature for 45 min. After being cured, the composite is cooled down to room temperature and extracted from the mandrel. Cylindrical specimens are cut using a diamond saw with water jet cooling to prevent microcracks and higher temperatures. The samples are then painted to produce a speckled pattern to allow surface deformation tracking and correlation matching to make them well-suited for DICtracking analysis.

Figure 1.

Filament winding manufacturing process.

The proposed pressure test procedure is an adaptation of the ASTM C1819 standard. The test consists of applying a uniaxial compressive load only on elastomeric discs concentrically inserted inside the cylinder, without any direct compressive force on the composite specimen. Radial expansion of the elastomeric inserts is achieved through the compressibility effect, transmitting the applied load to the cylinder wall. In the analysis of the partially pressurised cylinder, the unpressurised ends are assumed infinitely long. When the cylinder is pressurised in the central part of its length, the hoop stress varies along the length and bending stresses occur in the axial direction. Moreover, shear stresses act at right angles to the cylinder axis and compressive stresses due to the pressure may be present solely in the pressurised length. Despite this complex stress state, it is noteworthy that the hoop stress remains the largest among the three stress systems (see Figure 5 in Reference 35).

The elastomeric inserts must have their quasi-incompressibility checked to reach Poisson’s ratio of 0.5 or as close to it as possible. The discs are produced by casting a two-component ADDV-42 silicone rubber (40 Shore A hardness) into 3D-printed open-top moulds, being 10 mm in height and 49.8 mm in diameter. To obtain discs with a smooth and uniform surface finish, a glass plate is carefully positioned on the top of the moulds during the solidification process. Before being inserted into the cylinder, the discs are covered with talcum powder to minimise friction effects. Two pushrods are then fitted to the ends of the cylinder, with a clearance of 0.05 mm. The pushrods are machined from SAE 1040 steel, with a black oxidation surface treatment.

The test rig assembly is schematically illustrated in Figure 2, in which the pressure testing rig is positioned between two compression platens on a Zwick Roel universal testing machine equipped with a 100 kN load cell. Tests are conducted at a displacement rate of 10 mm/min, selected to minimise the potential effects of slow crack growth.

Figure 2.

Proposed internal pressure test rig.

During the tests, a high-resolution ARAMIS 3D camera system, consisting of two 12-megapixel cameras and a blue light projector, is utilised to acquire images on about 60° of the cylinder’s surface. The sampling frequency is set to 5 Hz. These images are then post-processed using GOM Correlate Pro software (ZEISS Quality Suite) to measure global displacements and strain fields through the DIC method. Figure 3 shows the full experimental setup used.

Figure 3.

(a) Test rig coupled to the testing machine with the DIC system. (b) Zoom-in highlighting the speckle pattern, and (c)discretisation of the speckled image onto a surface mesh.

Dimensional requirements of test

The effective length of the pressurised region, located halfway in the cylinder and filled with elastomeric inserts, as well as the unpressurised ends, where the pushrods are introduced, must be determined for the success of this test. When appropriately dimensioned, the unpressurised ends allow neglecting the stresses induced by edge effects, and the stresses over the pressurised region remain reasonably constant. Furthermore, the dimensional changes at the ends of the cylinder do not significantly affect the pressurised region, as these are limited to lengths very close to the pressure surface of the pushrods, decaying steeply when moving away from them.⁴⁵ The lengths of the pressurised and unpressurised regions are determined based on the parameter $β$ , proposed by Mosley³⁵:

β = \sqrt[4]{\frac{3 (1 - ν^{2})}{r_{i}^{2} t^{2}}}

(1)

where

ν

is the Poisson’s ratio of the cylinder material,

r_{i}

is the internal radius of the cylinder, and t is the wall thickness of the composite specimen.

Previous studies (^35,37,46) have shown that there is a stabilisation of both hoop and axial stresses when the length of the pressurised portion is greater than 9/ $β$ . To achieve stress conditions that cannot be distinguished in practice from those produced by an infinitely long cylinder, the length of the overhung parts (unpressurised ends) should not be less than 3.5/ $β$ . Mosley³⁵ has conducted a comprehensive investigation and experimentally confirmed (using mild steel cylinders with three different $r_{i} / t$ ratios) that the proposed minimum lengths (9/ $β$ and 3.5/ $β$ ) satisfy the requirements of the test method. Since equation (1) is parameterised by the test material property ( $ν$ ) and the geometric features of the cylinder ( $r_{i}$ and $t$ ), it can be applied to cylinders in general. It should be emphasised that the general deformation kinematics of the proposed test is similar to the hydrostatic one (when the cylinder is loaded only in the hoop direction), except that the internal pressure is nonlinearly related to the axial force due to the hyperelastic response of the elastomer, present even in low-strain regimes.

Given the in-plane Poisson’s ratio of the laminates as 0.3, and the inner radius and average thickness of the manufactured cylinders of 25 mm and 1 mm, respectively, the resulting parameter $β$ is 0.26 (the mechanical properties of laminates produced using the same towpreg of this work are documented in Reference 47). Therefore, the pressurised region should be longer than 35 mm, while the unpressurised end on each side must be at least 14 mm in length. To meet these dimensional criteria, samples are 110 mm long and six silicone discs are inserted into the cylinder, forming a 60 mm long pressurised region. The remaining length of the cylinder, 25 mm on each side, corresponds to the unpressurised ends. The sample length is also chosen to accommodate a complete diamond pattern in the filament winding part (see pattern explanations in References^48,49). The spacer (Figure 2) allows for the adjustment of the extent to which the cylinder penetrates the bottom pushrod and is removed before testing.

Calculation of testing parameters

The following parameters are calculated in accordance with the ASTM C1819 standard guidelines:

(i) Internal pressure exerted on the cylinder by an expansion of the elastomeric inserts, assuming incompressibility of the elastomeric discs:

p = \frac{F}{π r_{i}^{2}}

(2)

where

p

is the internal pressure and

F

is the uniaxial compressive force.

(ii) Hoop strength, that is, the maximum hoop stress that the material can sustain:

S_{h u} = η_{m} p_{\max} \frac{2 r_{i}^{2}}{r_{o}^{2} - r_{i}^{2}}

(3)

with

S_{h u}

being the hoop strength,

η_{m}

the stress factor,

p_{\max}

the maximum internal pressure, and

r_{o}

the outer radius of the cylinder. The stress factor

η_{m}

is a correction to account for differences between analytical, numerical, and experimental results for hoop tensile stresses in cylinders that are pressurised using compressed elastomer inserts. The value of

η_{m}

in any cylinder is a function of

r_{i} / t

and

ν

. A wide range of materials, with Poisson’s ratios ranging from 0.15 to 0.45, and inner radius to wall thickness ratios ranging from 1 to 50, were covered in the work by Mosley (for detailed information, see Appendix 1 in Reference 35). Here, the adopted value is

η_{m} = 1.09

The hoop stiffness of the cylindrical shells can be determined in the linear-elastic range of the load × displacement curves using linear regression. To assess the statistical significance of the experimental data, normality and homogeneity of variances are first examined through the Shapiro–Wilk and Levene tests, respectively. In cases where these criteria are not met, outliers are identified and removed using a quantile-based approach. One-way analyses of variance (ANOVA) are then conducted, and if the null hypothesis is rejected, Tukey’s honestly significant difference (HSD) test is employed for mean comparisons. The statistical analyses are conducted with a significance level of 5%.

Results and discussion

The results of the internal pressure test, comprising five specimens for each winding angle, are summarised in Table 1. Some variability is observed among the values obtained for cylinders wound at the same angle. Despite that, when comparing the average results across different angles, the hoop stiffness of the cylinders increases with the angle, with the highest relative scatter found for the ±35°. Regarding failure pressure and hoop tensile strength, both exhibit an increase up to ±75° and subsequently decrease for the hoop-oriented cylindrical shell. Variability also tends to increase with the winding angle.

Table 1.

Summary of the internal pressure test results.

Winding angle	Failure pressure (MPa)	Axial strain @ failure (%)	Hoop strain @ failure (%)	Hoop stiffness (kN/mm)	Hoop tensile strength (MPa)
±35°	1.83 ± 0.07	−0.64 ± 0.13	1.26 ± 0.18	1.85 ± 0.30	48.83 ± 1.82
±45°	2.20 ± 0.13	−0.89 ± 0.12	1.10 ± 0.10	2.66 ± 0.18	58.67 ± 3.47
±55°	5.19 ± 0.19	−1.87 ± 0.25	1.14 ± 0.06	5.73 ± 0.46	138.64 ± 5.10
±65°	18.93 ± 0.90	−5.33 ± 0.49	2.33 ± 0.22	10.70 ± 0.96	505.77 ± 23.95
±75°	22.19 ± 2.03	−2.00 ± 0.52	1.21 ± 0.10	20.70 ± 1.01	592.69 ± 54.16
±90°	18.23 ± 4.22	−0.86 ± 0.04	0.76 ± 0.14	23.59 ± 1.65	486.91 ± 112.79

The deformation and failure processes vary according to the angle of the fibres. For cylinders wound at ±35°, the magnitude of hoop strain is higher at the failure pressure level than axial strain. The opposite is seen for cylinders with winding angles from ±55° to ±75°, where axial strain becomes predominant. Meanwhile, cylinders with fibres oriented at ±45° and ±90° display similar strains in both directions. The variation in strains is more significant in the axial direction, except for the hoop winding.

Figure 4 presents average load × displacement curves, describing simultaneously the monotonic compressive force (left axis) applied to the elastomeric discs and the corresponding internal pressure (right axis, calculated by equation (2)) resulting from their radial expansion. The sequence of events during the loading history shows an initial accommodation phase followed by a linear load increase. Within the linear regime, the stiffness of the cylinders is determined through linear regression in the 25%−50% range of ultimate load (dashed lines). The regression lines nicely fit the experimental data demonstrating a good correlation, with a coefficient of determination (R²) exceeding 0.99 for all curves. The increase in stiffness for higher winding angles is also seen in Figure 4.

Figure 4.

Applied axial compressive force and corresponding internal pressure as a function of the axial displacement for different winding angles.

As the internal pressure increases, cylinders wound at angles of ±35°, ±45°, and ±55° exhibit a clear transition to nonlinear behaviour. Cylinders wound at ±65° show a less pronounced deviation and a lower degree of nonlinearity. In contrast, cylinders wound at ±75° and ±90° demonstrate almost linear-elastic behaviour up to failure. This behaviour can be justified by the orientation of carbon fibres closer to the hoop direction of the cylinders. The fibres play a crucial role in withstanding the generated stresses during pressurisation. Being aligned in the hoop direction, where the highest stresses occur, the fibres effectively reinforce the structure, leading to an increase in stiffness and a more linear-elastic response. The observed pattern is in agreement with the work of Rosenow²⁷ for the hoop pressure loading of glass fibre-reinforced polyester filament-wound pipes, who reported an increase in the load-bearing capacity of cylinders up to ±75°, declining for ±85°.

The statistical significance of the experimental data is analysed in relation to hoop tensile strength, which is directly proportional to the failure pressure (see equation (3)). Figure 5 illustrates the comparison among different winding angles, with the letters indicating the angles that share an average hoop tensile strength that is not statistically different from each other.

Figure 5.

Hoop tensile strength of the studied cylinders.

The radial displacement ( $U_{r}$ ), hoop ( $ε_{θ}$ ), and axial ( $ε_{z}$ ) strains are measured through DIC. Figure 6 shows the maximum strains captured at the mid-length of the pressurised region for each cylinder. In all cases, the axial strain is negative and the hoop strain positive. Typical deformation fields for each winding angle are presented in Figures 7–12 at three pressure levels: pre-failure (ca. 75% of the failure pressure), failure, and post-failure. Analysis of these deformation fields reveals that the highest strains occur in the fibre tows. This indicates that the matrix is no longer significantly contributing to load-carrying of the cylindrical shells and most of the load is being taken up by the fibres. Furthermore, the unpressurised ends are performing as intended, that is, they do not influence the strain distribution in the pressurised region.

Figure 6.

Pressure × strain for cylinders wound at (a) ±35°, (b) ±45°, (c) ±55°, (d) ±65°, (e) ±75°, and (f) ±90°.

Figure 7.

The DIC results for the ±35° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Figure 8.

The DIC results for the ±45° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Figure 9.

The DIC results for the ±55° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Figure 10.

The DIC results for the ±65° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Figure 11.

The DIC results for the ±75° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Figure 12.

The DIC results for the ±90° cylinders: radial displacement $U_{r}$ , hoop strain $ε_{θ}$ , and axial strain $ε_{z}$ .

Cylinders wound at ±35°, ±45°, and ±55° do not undergo burst failure. Instead, large nonlinear deformations are observed in both hoop and axial directions. This behaviour is explained by Evans and Gibson.⁵⁰ When the initial fibre angle deviates from the stable one, the application of strain induces fibre rotation, resulting in nonlinear stress–strain relationships. As the matrix fails, the effective stiffness of the composite reduces, allowing the fibres to rotate with small increments of pressure and increasing strains until they reach a stable angle at which they can withstand the load without strain-induced rotation. The proportional limit of these cylindrical shells is determined by the intersection point between the pressure × hoop strain curve and a line parallel to the linear response, which is horizontally offset by 0.25% in hoop strain. This intersection of each curve identifies the end of the linear behaviour.

For the ±35° cylinder, this point is around 1.8 MPa of pressure, with 1.2% and −0.6% of the hoop and axial strains, respectively (a hoop-to-axial strain ratio of 2:1). In this case, the end of the linear regime also means the structural failure of the cylinder, as there is practically no increase in load for higher strains. Despite the pronounced bulging experienced by the cylinder, as seen in Figure 7, no visible damage is observed in the fibre tows.

The end of the linear regime for cylinders wound at ±45° occurs at a slightly higher pressure (2.2 MPa) than those at ±35°. In contrast to the previous case, residual strength is still observed in the nonlinear regime. However, strains propagate rapidly with small pressure increments, defining a potentially hazardous operating range. Therefore, structural failure is also linked to a significant reduction in stiffness after the end of the linear-elastic regime. Hoop and axial deformations occur in a more balanced manner, with magnitudes around 1% in each direction, resulting in a more uniform expansion along the pressurised region, as depicted in Figure 8. Once again, despite the significant rotation of the fibre tows, no associated macroscopic major damage is identified.

The ±55° winding angle is the recommended angle to withstand internal pressurisation in closed-end vessels, where the walls are subjected to a 2:1 hoop-to-axial stress ratio. However, for the proposed test setup, consisting of cylinders with open and unconstrained ends, the stress state is different, with hoop stress being the highest and the others negligible in comparison. It is interesting to note that the strains resulting from pressurisation follow an approximate 1:2 ratio in the hoop-to-axial direction. The proportional limit is reached at pressures around 5.2 MPa. Nonetheless, there is a significant remaining strength where the effective stiffness decreases with the continuous increase in pressure up to ca. 16 MPa. In this process, progressive failure due to matrix shear dominates,⁵¹ resulting in a small lateral displacement of the specimen during the test (Figure 9). Analyses of the measured strain fields for these three angles align well with the findings presented by Spencer and Hull.⁵²

Filament-wound shells with angles of ±65° (Figure 10) and ±75° (Figure 11) experience burst failure, unlike the previous ones. In both cases, small strains are recorded in the hoop direction before the abrupt collapse of the cylinder walls, highlighting the better performance of the reinforcing fibres oriented at these angles. Larger strains are observed in the axial direction, implying shrinking of the cylinders. Deviations from linearity are also more pronounced in the axial direction. The cylinders wound at ±75° achieve the highest capacity to withstand internal pressure, followed by the ±65° sample.

Cylinders wound at ±90° do not form a mosaic pattern. The failure pressure for this winding angle falls between those observed for the ±55° and ±65° angles. However, the failure mode is entirely different from the other tested cylinders. It occurs along a circumferential line parallel to the fibre tow at the pressurised end of the cylinder. In this region, bending stresses in the axial direction are non-negligible, as indicated by the axial strain gradient in Figure 12. The difference in terms of fibre architecture, lacking interweaving regions, is detrimental to the hoop pressure ability of the cylinders.⁵³ Similar results regarding pressure capacity variation with the winding angle were reported by Martins et al.²⁹

Conclusions

This paper explores an alternative laboratory-scale test method to conventional hydrostatic pressure tests and tests based on ring specimens for filament-wound cylindrical shells. The featured test proved effective for several winding angles and, compared to hydrostatic tests, show advantages like good loading control, easy assembly, reduced time and costs, improved safety, and the possibility to use DIC for monitoring strain and displacement fields of the specimens. Also, considering that it allows the evaluation of cylinders with helical windings, the proposed test does not have that limitation faced by ring-based tests.

The hoop stiffness of the cylinders increased with the winding angle up to ±90° (hoop winding). The load-carrying ability increased with the winding angle up to ±75°. Specimens wound at ±35°, ±45°, and ±55° angles exhibited matrix-dominant failure, while those wound at ±65°, ±75°, and ±90° showed abrupt brittle failure. The cylinder with hoop-oriented winding, nevertheless, failed along a circumferential line parallel to the fibre tow at the end of the pressurised region, where bending stresses in the axial direction are non-negligible.

Footnotes

Declaration of conflicting interests

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

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: CAPES/DAAD (PROBRAL project Nos. 88881.198774/2018-01 and 57447163), FAPERGS (Inova Clusters Tecnológicos project No. 22/2551-0000839-9), and CNPq/EQUINOR (project No. 440115/2019-1). HA is supported by the Royal Academy of Engineering, UK, under the Research Fellowship scheme (Grant No. RF/201920/19/150).

ORCID iDs

José Humberto S Almeida

Sandro C Amico

Data Availability Statement

Data available on request from the authors.*

References

Azevedo

Almeida Jr

JHS

Lisboa

Scherer

Spickenheuer

Amico

. Combining filament winding with tailored fiber placement in composite cylinders locally reinforced with fibrous patches. J Reinf Plast Compos 2024; 0: 0–0. DOI: 10.1177/07316844231168542, In press.

Gemi

Şahin

ÖS

Gemi

. Experimental investigation of axial compression behavior after low velocity impact of glass fiber reinforced filament wound pipes with different diameter. Compos Struct 2022; 280: 114929. DOI: 10.1016/j.compstruct.2021.114929.

Almeida Jr

JHS

Lisboa

Spickenheuer

St-Pierre

. A sequential finite element model updating routine to identify creep parameters for filament wound composite cylinders in aggressive environments. Comput Struct 2023; 276: 106939. DOI: 10.1016/j.compstruc.2022.106939.

Colombo

Vergani

. Optimization of filament winding parameters for the design of a composite pipe. Compos B Eng 2018; 148: 207–216. DOI: 10.1016/j.compositesb.2018.04.056.

Rafiee

Torabi

. Stochastic prediction of burst pressure in composite pressure vessels. Compos Struct 2018; 185: 573–583. DOI: 10.1016/j.compstruct.2017.11.068.

Rafiee

Rashedi

Rezaee

. Theoretical study of failure in composite pressure vessels subjected to low-velocity impact and internal pressure. Front Struct Civ Eng 2020; 14(6): 1349–1358. DOI: 10.1007/s11709-020-0650-3.

Lisboa

Almeida Jr

JHS

Spickenheuer

Stommel

Amico

Marczak

. FEM updating for damage modeling of composite cylinders under radial compression considering the winding pattern. Thin-Wall Struct. 2022; 173: 108954. DOI: 10.1016/j.tws.2022.108954, In this issue.

Alam

Yandek

Lee

, et al. Design and development of a filament wound composite overwrapped pressure vessel. Composites Part C: Open Access 2020; 2: 100045. DOI: 10.1016/j.jcomc.2020.100045.

Azevedo

Almeida Jr

JHS

Flores

, et al. Influence of mosaic pattern on hygrothermally-aged filament wound composite cylinders under axial compression. J Compos Mater 2020; 54(19): 2651–2659. DOI: 10.1177/0021998319899144.

10.

Eggers

Almeida

JHS

Lisbôa

, et al. Creep and residual properties of filament-wound composite rings under radial compression in harsh environments. Polymers 2020; 13(1): 18–33. DOI: 10.3390/polym13010033.

11.

Hilburger

Nemeth

Starnes

. Shell buckling design criteria based on manufacturing imperfection signatures. AIAA J 2006; 44(3): 654–663. DOI: 10.2514/1.5429.

12.

Wang

Almeida

JHS

Ashok

, et al. Lightweight design of variable-angle filament-wound cylinders combining Kriging-based metamodels with particle swarm optimization. Struct Multidiscip Optim 2022; 65(5): 140. DOI: 10.1007/s00158-022-03227-8.

13.

Kim

Park

. The effect of composite damage on fatigue life of the high pressure vessel for natural gas vehicles. Compos Struct 2011; 93(11): 2963–2968. DOI: 10.1016/j.compstruct.2011.05.007.

14.

Zhang

Kang

, et al. A literature review of failure prediction and analysis methods for composite high-pressure hydrogen storage tanks. Int J Hydrogen Energy 2019; 44(47): 25777–25799. DOI: 10.1016/j.ijhydene.2019.08.001.

15.

Barthelemy

Weber

Barbier

. Hydrogen storage: recent improvements and industrial perspectives. Int J Hydrogen Energy 2017; 42(11): 7254–7262. DOI: 10.1016/j.ijhydene.2016.03.178.

16.

Zhao

Druzhinin

Ivens

, et al. Split-disk test with 3D Digital Image Correlation strain measurement for filament wound composites. Compos Struct 2021; 263: 113686. DOI: 10.1016/j.compstruct.2021.113686.

17.

Meyer

Pignagnoli

Potts

, et al. Lightweighting matters in energy storage. Reinforc Plast 2014; 58: 20–23. DOI: 10.1016/S0034-3617(14)70176-6.

18.

Antunes

Almeida Jr

JHS

Amico

, et al. Curing and seawater aging effects on mechanical and physical properties of glass/epoxy filament wound cylinders. Compos Commum 2020; 22: 100517. DOI: 10.1016/j.coco.2020.100517, In this issue.

19.

Almeida Jr

JHS

Tonatto

MLP

Ribeiro

Tita

Amico

, et al. Buckling and post-buckling of filament wound composite tubes under axial compression: Linear, nonlinear, damage and experimental analyses. Compos B Eng 2018; 149: 227–239. DOI: 10.1016/j.compositesb.2018.05.004.

20.

Wang

, et al. Prediction of void growth and fiber volume fraction based on filament winding process mechanics. Compos Struct 2020; 246: 112432. DOI: 10.1016/j.compstruct.2020.112432.

21.

Morozov

. The effect of filament-winding mosaic patterns on the strength of thin-walled composite shells. Compos Struct 2006; 76(1–2): 123–129. DOI: 10.1016/j.compstruct.2006.06.018.

22.

de Menezes

EAW

Lisbôa

Almeida

JHS

, et al. On the winding pattern influence for filament wound cylinders under axial compression, torsion, and internal pressure loads. Thin-Walled Struct 2023; 191: 111041. DOI: 10.1016/j.tws.2023.111041.

23.

Błachut

Wollmann

Panek

, et al. Influence of fiber tension during filament winding on the mechanical properties of composite pressure vessels. Compos Struct 2023; 304: 116337. DOI: 10.1016/j.compstruct.2022.116337.

24.

Azizian

Almeida

JHS

Azizian

. Efficient strategies for reliability analysis and uncertainty quantification for filament-wound cylinders under internal pressure. J Compos Mater 2023; 57(11): 1863–1874. DOI: 10.1177/00219983231164221.

25.

Zhang

Zhao

Wang

. Burst pressure prediction and structure reliability analysis of composite overwrapped cylinder. Appl Compos Mater 2018; 25(6): 1269–1285. DOI: 10.1007/s10443-017-9665-x.

26.

Ellul

Camilleri

Grech

, et al. Filament wound composite pressure vessels and pipes subject to an internal pressure: an experimental and material characterization study. J Pressure Vessel Technol 2016; 138(6). DOI: 10.1115/1.4032506.

27.

Rosenow

MWK

. Wind angle effects in glass fibre-reinforced polyester filament wound pipes. Composites 1984; 15: 144–152. DOI: 10.1016/0010-4361(84)90727-4.

28.

Onder

Sayman

Dogan

, et al. Burst failure load of composite pressure vessels. Compos Struct 2009; 89(1): 159–166. DOI: 10.1016/j.compstruct.2008.06.021.

29.

Martins

LAL

Bastian

Netto

. Reviewing some design issues for filament wound composite tubes. Mater Des 2014; 55: 242–249. DOI: 10.1016/j.matdes.2013.09.059.

30.

Hwang

Park

Kim

. Evaluation of fiber material properties in filament-wound composite pressure vessels. Compos Part A Appl Sci Manuf 2012; 43(9): 1467–1475. DOI: 10.1016/j.compositesa.2012.04.005.

31.

Kim

Choi

Kim

, et al. Ring burst test of filament wound composites for environmental resistance. J Compos Mater 2016; 50(18): 2507–2521. DOI: 10.1177/0021998315607611.

32.

ASTM D2290-19 . Standard test method for apparent hoop tensile strength of plastic or reinforced plastic pipe. West Conshohocken, PA: ASTM International, 2019. DOI: 10.1520/D2290-19A.

33.

Kim

. Design of a segment-type ring burst test device to evaluate the pressure resistance performance of composite pressure vessels. Compos Struct 2020; 242: 112199. DOI: 10.1016/j.compstruct.2020.112199.

34.

Rafiee

. Apparent hoop tensile strength prediction of glass fiber-reinforced polyester pipes. J Compos Mater 2013; 47(11): 1377–1386. DOI: 10.1177/0021998312447209.

35.

Mosley

. The stressing for test purposes of Materials in tubular form using Elastomeric inserts – experimental and theoretical development. Proceedings of the Institution of Mechanical Engineers 1982; 196(1): 123–139. DOI: 10.1243/PIME_PROC_1982_196_014_02.

36.

ASTM C1819-21 . Standard test method for hoop tensile strength of continuous fiber-reinforced advanced ceramic composite tubular test specimens at ambient temperature using elastomeric inserts. West Conshohocken, PA: ASTM International, 2021. DOI: 10.1520/C1819-21.

37.

Carter

. Compressed elastomer method for internal pressure testing,” ARL-TR-3921. Washington, DC: Aberdeen Proving Grounds, 2006.

38.

Petersen

Link

Vandeput

. Strength testing of brittle materials in tubular form using elastomeric inserts. J Test Eval 1995; 23(4): 261–266. DOI: 10.1520/JTE10423J.

39.

Hong

Byun

Snead

, et al. Evaluation methods for the hoop strength of small-sized tubular ceramic components. J Mater Res 2009; 24(4): 1422–1434. DOI: 10.1557/jmr.2009.0158.

40.

Jacobsen

Stone

Khalifa

, et al. Investigation of the C-ring test for measuring hoop tensile strength of nuclear grade ceramic composites. J Nucl Mater 2014; 452(1–3): 125–132. DOI: 10.1016/j.jnucmat.2014.05.002.

41.

Shapovalov

Jacobsen

Alva

, et al. Strength of SiCf-SiCm composite tube under uniaxial and multiaxial loading. J Nucl Mater 2018; 500: 280–294. DOI: 10.1016/j.jnucmat.2018.01.001.

42.

Castro

SGP

Almeida Jr

JHS

St-Pierre

, et al. Measuring geometric imperfections of variable–angle filament–wound cylinders with a simple digital image correlation setup. Compos Struct 2021; 276: 114497. DOI: 10.1016/j.compstruct.2021.114497.

43.

Almeida

JHS

St-Pierre

Wang

, et al. Design, modeling, optimization, manufacturing and testing of variable-angle filament-wound cylinders. Compos B Eng 2021; 225: 109224. DOI: 10.1016/j.compositesb.2021.109224.

44.

Henriquez

Mertiny

. Filament winding applications. In: Comprehensive composite materials II. Beaumont

PWR

Zweben

(eds). Amsterdam, Netherlands: Elsevier, 2018, pp. 556–577. DOI: 10.1016/B978-0-12-803581-8.10313-3.

45.

Kraus

. Thin elastic shells: an introduction to the theoretical foundations and the analysis of their static and dynamic behavior. Hoboken, NJ: John Wiley & Sons, 1967.

46.

Salem

, et al. Burst testing of triaxial braided composite tubes, 2014. NASA/TM-2013-216615.

47.

Scherer

Almeida

Azevedo

, et al. Effect of geometry on physical and material properties of filament wound laminates. In: Proceedings of the 5th Brazilian conference on composite materials. Tita

Tarpani

Ribeiro

(eds). São Carlos: USP, 2021.

48.

Dalibor

Lisbôa

Marczak

, et al. A geometric approach for filament winding pattern generation and study of the influence of the slippage coefficient. J Braz Soc Mech Sci Eng 2019; 41(12): 576. DOI: 10.1007/s40430-019-2083-2.

49.

Lisbôa

Almeida

JHS

Dalibor

, et al. The role of winding pattern on filament wound composite cylinders under radial compression. Polym Compos 2020; 41(6): 2446–2454. DOI: 10.1002/pc.25548.

50.

Evans

Gibson

. Composite angle ply laminates and netting analysis. Proc R Soc Lond A 2002; 458(2028): 3079–3088. DOI: 10.1098/rspa.2002.1066.

51.

Hull

Legg

Spencer

. Failure of glass/polyester filament wound pipe. Composites 1978; 9(1): 17–24. DOI: 10.1016/0010-4361(78)90513-X.

52.

Spencer

Hull

. Effect of winding angle on the failure of filament wound pipe. Composites 1978; 9(4): 263–271.

53.

Cox

Flanagan

. Handbook of analytical methods for textile composites. Washington, DC: NASA, 1997.