Test-based equivalent-material method for collapse qualification of helically wound and layered cylindrical structures Authors: Yuteng Zhang, Mohsen Saneian, Yong Bai, Pan Fang, Peng Cheng, Liang Zhao Published: 2026 ABSTRACT Reliable qualification of collapse resistance is critical across pressurised cylindrical shells ranging from offshore risers and subsea cables to aerospace tanks and even biological ducts. Existing approaches either rely on costly hyperbaric chambers or oversimplified geometric models that neglect the forming-induced residual stresses and imperfections. Here, we propose a back-inferred equivalent-material method (EMM), which transforms a simple benchtop flat-plate compression test result into a nonlinear constitutive law that implicitly embeds cold work, residual stress, friction, and small-scale compliance. Embedding this equivalent law into a homogeneous finite-element model reproduces hydrostatic collapse without the need for full-scale testing. We demonstrate the method on flexible-pipe carcass layers and steel-strip reinforced thermoplastic pipes, showing close agreement with explicit-geometry simulations and hyperbaric measurements. More broadly, the geometry-agnostic formulation enables rapid, low-cost, and physically faithful qualification across diverse layered shells, offering a general route to replace hyperbaric testing in industries spanning energy, aerospace, and biomedical engineering. Introduction Cylindrical shells subjected to external pressure are prevalent in engineering and natural systems, sharing a common vulnerability whereby collapse is triggered by hydrostatic loading or environmental pressure1. Even subtle imperfections or forming-induced residual stresses can drastically reduce their loading capacity2, making collapse resistance one of the most uncertain and safety-critical parameters throughout their service life. Helically wound and layered cylindrical structures are found in biological ducts, rocket tanks, submersibles, subsea cables, and umbilicals (Fig. 1a). In offshore engineering, unbonded flexible pipes represent a typical example of these multilayer systems, linking floating production systems to seabed infrastructures3,4. In the layered architecture, the metallic carcass prevents collapse, armour layers carry radial/axial loads, and polymer sheaths provide sealing and wear protection5 (Fig. 1b). Across these diverse applications, hydrostatic buckling remains highly sensitive to manufacturing-induced defects6, generating large uncertainty in collapse resistance and further driving designs toward costly conservatism or resulting in unexpected failure7 (Fig. 1c).Fig. 1Collapse of pressurised cylindrical structures.a Representative layered pressurised cylinder in engineering and biology. b Typical architecture of an unbonded flexible pipe. c Collapse uncertainty in multilayer shells caused by constitutive complexity and geometric imperfections. I.C. internal carcass, I.S. internal sheath, P.A. pressure armour, A.T. anti-friction tape, T.A. tensile armour, O.S. outer sheath. Collapse of pressurised cylindrical structures. a Representative layered pressurised cylinder in engineering and biology. b Typical architecture of an unbonded flexible pipe. c Collapse uncertainty in multilayer shells caused by constitutive complexity and geometric imperfections. I.C. internal carcass, I.S. internal sheath, P.A. pressure armour, A.T. anti-friction tape, T.A. tensile armour, O.S. outer sheath. Among the various failure mechanisms of flexible risers, the collapse of the carcass layer under external pressure is of primary concern8. This issue is especially severe in the touchdown zone, where curvature exacerbates structural vulnerability9. The critical collapse pressure not only governs the selection of the structural wall thickness but also strongly influences manufacturing cost and installation feasibility10. Carcass collapse is typically categorised as dry (outer sheath intact; carcass and pressure armour share resistance) or wet (outer sheath breached; carcass resists nearly alone), with the latter being far more severe11. Outer-sheath damage is frequently observed in service, which increases the risk of wet collapse12. Typically, during the installation stage, the riser is effectively unpressurised (i.e., no internal pressure)13, representing the most dangerous service condition14. Yet, existing design standards provide no explicit methodology for evaluating carcass collapse pressure, resulting in conservative wall thicknesses and correspondingly higher costs15. These limitations emphasise the importance of establishing high-fidelity and robust methodologies to assess the collapse resistance of the carcass structure reliably. Since the 1970s, critical collapse pressure has been predicted chiefly by many scientists16–18. Full-scale experimental testing is the most direct method to determine the critical collapse pressure; however, such hydrostatic tests are often cost-prohibitive19. Similarly, full-dimensional 3D finite element (FE) simulations that explicitly capture the interlocked profile and internal contact mechanics are computationally intensive, limiting their routine use in early-stage design or parametric studies20. To overcome these challenges, equivalent modelling approaches have been widely adopted, in which the complex carcass geometry is idealised as a homogeneous cylindrical shell governed by ring-buckling theory21. Several equivalent-layer methods (ETMs) have been proposed, differing mainly in how an effective thickness is defined. Area-equivalent approaches conserve cross-sectional area, assuming that redistributing the same material into a regular shell yields comparable collapse behaviour22. This assumption neglects the cubic dependence of bending stiffness on thickness (I ∝ t3), thereby limiting its ability to represent the flexural contribution to the collapse response. To reflect the bending-dominated character of collapse, subsequent work matched bending stiffness-either per unit area23 or per unit length24-with refinements that introduce correction factors for interlock discontinuities and sectional non-uniformity25. Beyond these, energy-based or hybrid analytical techniques infer thickness from cross-section approaches26, and classical shell stability theory provides the analytical foundation27, sometimes coupled with finite-element calibration under realistic constraints28. Predictions are boundary-condition sensitive and can overestimate bending stiffness, leading to non-conservative collapse pressures. More recently, Tang et al.29 proposed a strain energy equivalence in which the effective thickness is chosen such that the equivalent shell stores the same strain energy as the actual carcass under hoop loading. Building on this idea, Li et al.30 combined strain-energy and membrane-stiffness equivalences to represent bending and in-plane response when direct evaluation for a profiled contacting section is infeasible31. While valuable, existing strategies are restricted by geometric idealisation, the neglect of interlock contact and friction, and the exclusion of forming-induced cold work and residual stresses. Consequently, their predictions of collapse behaviour may be biased or inaccurate under realistic manufacturing conditions. Data-driven surrogates also complement classical ETMs. de Sousa et al.32 used symbolic regression (Eureqa33) on 3D FE ring datasets to derive orthotropic equivalents whose shell models matched collapse tests, and later proposed semi-empirical formulas for dry/wet collapse based on FE datasets in which ovality, inter-layer gaps, cold work, and residual stress were systematically varied, providing dimensionless sensitivity insights34. Yan et al.35 developed a Kriging-modified buckling model that improves the predictive accuracy of classical shell formulas and analysed sampling layout/size effects. Li et al.36 used symbolic regression trained on FE-generated ring-compression datasets to obtain closed-form equivalent-thickness relations for carcass layers. While fast once trained, these surrogates rely on FE-generated data; their fidelity therefore remains bounded by the modelling assumptions (imperfections, contact/friction, residual stress) used to create the training sets. Yet, the real carcass does not arise as an idealised structure but as the outcome of a complex manufacturing sequence. As illustrated in Fig. 2a, the manufacturing of the carcass layer involves a sequence of unavoidable forming operations that progressively shape flat steel strips into interlocked profiles. As the strip passes through roll forming, edge folding, and interlock closure, it is bent repeatedly over small radii, inevitably leaving plastic strain gradients, work hardening, and residual stresses37. The numerical reproduction of this process (Fig. 2b) confirms that plasticity and residual stress are not incidental by-products but inherent consequences of manufacturing38. These imprints, together with geometric scatter from roll forming and closure, persist in the carcass and ultimately govern its radial stiffness and collapse resistance39.Fig. 2Carcass manufacturing and forming sequence.a Multi-station manufacturing process, including roll forming, locking/sizing, and helical winding. Repeated bending and interlock closure introduce plastic strain, work hardening, and residual stress. b Finite-element reproduction of the sequence: (01-08) roll forming of the S-profile; (09) wrapping on a cylindrical mandrel; (10) interlocking transition (open-S → half-locked-S → locked-S); and (11) final interlocked carcass. Carcass manufacturing and forming sequence. a Multi-station manufacturing process, including roll forming, locking/sizing, and helical winding. Repeated bending and interlock closure introduce plastic strain, work hardening, and residual stress. b Finite-element reproduction of the sequence: (01-08) roll forming of the S-profile; (09) wrapping on a cylindrical mandrel; (10) interlocking transition (open-S → half-locked-S → locked-S); and (11) final interlocked carcass. These effects, however, are systematically neglected in existing equivalent-thickness methods (ETMs), which treat the carcass as a smooth shell. To bridge this gap, we here propose a back-inferred equivalent-material method (EMM) that implicitly embeds manufacturing-induced strengthening into the material description, thereby providing a physics-based route to predict collapse capacity without recourse to full-scale hyperbaric testing. Specifically in the current research, flat-plate compression of a short carcass segment provides a load-displacement curve that, through the proposed EMM, is translated into a nonlinear constitutive law embedding the manufacturing imprints (i.e., cold-work plasticity, residual stress, and geometric compliance). The back-inferred stress-strain relation is then assigned to a homogeneous ring model, from which the load-displacement response and collapse capacity are recovered. Validation against flat-plate calibration on carcass specimens and steel strip-reinforced flexible pipes (hereinafter referred to as SSRFPs40), together with an extra full-pipe hyperbaric test on SSRFPs, shows close agreement, confirming the fidelity of the proposed EMM. Unless otherwise stated, the finite-element simulations in this study were performed in Abaqus/Standard 202341. More broadly, EMM generalises to other helically wound reinforcements and layered cylindrical shells (e.g., Fig. 1) as the calibration is independent of detailed lay-up or profile. By embedding manufacturing effects into a simple equivalent law, EMM provides a practical route for efficient qualification, design optimisation, and imperfection sensitivity assessment without recourse to hyperbaric testing facilities. Methods The proposed framework is built on the principle that the complex response of pressurised structures under external loading can be reformulated in an incremental equivalent-material form. Rather than relying on costly hyperbaric tests or detailed geometric simulations, the essential constitutive behaviour is inferred directly from simple compression tests on short specimens. Figure 3a outlines a conventional workflow for a helically wound layer. Specimens are prepared (Fig. 3a1), constitutive properties are derived from material tests (Fig. 3a2), together with a detailed geometric model constructed (Fig. 3a3). Classical equivalent thickness methods (ETMs) are hence introduced for comparison. Yet these ETMs remain fundamentally limited by the absence of the manufacturing process and the true constitutive response. Moreover, their accuracy is highly case-dependent and remains difficult to evaluate (Fig. 3a4).Fig. 3ETM practices and the proposed EMM workflow.a Conventional ETM-based route: a1 constituent material testing; a2 conversion to constitutive properties; a3 detailed FE modelling; a4 equivalent-thickness baseline construction. b Proposed EMM-based route: b1 flat-plate compression of a short ring; b2 calibration of an equivalent nonlinear material law from the load--displacement response; b3 numerical analysis emplying the calibrated EMM material; b4 hyperbaric validation where available. ETM equivalent thickness method, EMM equivalent material method, FE finite element. ETM practices and the proposed EMM workflow. a Conventional ETM-based route: a1 constituent material testing; a2 conversion to constitutive properties; a3 detailed FE modelling; a4 equivalent-thickness baseline construction. b Proposed EMM-based route: b1 flat-plate compression of a short ring; b2 calibration of an equivalent nonlinear material law from the load--displacement response; b3 numerical analysis emplying the calibrated EMM material; b4 hyperbaric validation where available. ETM equivalent thickness method, EMM equivalent material method, FE finite element. In contrast, the proposed EMM is anchored in experiment rather than geometric idealisation, as illustrated in Fig. 3b. A short ring or repeat unit is compressed between flat plates to generate a representative load-displacement curve (Fig. 3b1). Through the EMM framework, this curve is mapped to a nonlinear equivalent material that naturally captures cold-work plasticity, residual stress, contact/friction, and small imperfections (Fig. 3b2). Assigning this law to a homogeneous shell allows direct simulation of external-pressure response and collapse (Fig. 3b3), with hydrostatic tests providing validation where available (Fig. 3b4). The EMM proceeds incrementally: at each loading step, the measured load–displacement increment is mapped to the two state variables that define the evolving constitutive response—(i) the tangent modulus, Et, representing the instantaneous stiffness of the equivalent material, and (ii) the associated strain increment, εinc—as follows: 1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{t}=f({F}_{{{{\rm{inc}}}}},{X}_{{{{\rm{inc}}}}})$$\end{document}Et=f(Finc,Xinc)2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varepsilon }_{{{{\rm{inc}}}}}=f({F}_{{{{\rm{inc}}}}},{X}_{{{{\rm{inc}}}}})$$\end{document}εinc=f(Finc,Xinc)where Finc is the increment of the measured reaction force from the testing machine, and Xinc is the increment of the applied longitudinal displacement. The coupling between these quantities is governed by equilibrium and compatibility of the underlying curved-beam model, which together lead to a differential equation: 3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f({E}_{t},{\varepsilon }_{{{{\rm{inc}}}}},{F}_{{{{\rm{inc}}}}},{X}_{{{{\rm{inc}}}}})=0$$\end{document}f(Et,εinc,Finc,Xinc)=0 By iteratively solving this equation, a complete equivalent stress-strain curve is reconstructed, embedding not only the elastic-plastic response but also manufacturing imprints (i.e., residual stresses, cold work, interfacial friction, and geometric imperfections). This back-inferred law therefore reflects the true structural behaviour far more faithfully than idealised elastic-plastic models. Once obtained, the equivalent material curve can be embedded into classical stability criteria - such as ring buckling or shell collapse formulations - by substituting the evolving tangent modulus Et in place of the constant elastic modulus. In doing so, the framework connects local material plasticity with global instability, providing a direct route to estimate plastic collapse pressure (for the full derivation and detailed equations, see Supplementary Information Section 1). Collectively, the test-anchored incremental constitutive description provides a transferable route to qualify external-pressure capacity. The framework links benchtop measurements to full-scale collapse resistance, reduces reliance on hyperbaric testing, and enables rapid assessments that remain consistent with the governing material and stability mechanisms. Results Carcass: experiment-based EMM calibration – input data preparation To characterise the constituents, dumbbell-shaped specimens of high-density polyethylene (HDPE) and steel strip were prepared from the raw material (Fig. 4a1) and tested in uniaxial tension (Fig. 4a2) following refs. 42,43. Nominal and true stress-strain curves are shown in Fig. 4b1 and Fig. 4b2 for PE and raw carcass steel, respectively.Fig. 4EMM calibration benchmark and input data preparation.a1, a2 Uniaxial tensile tests of HDPE and steel strip (raw material). b1, b2 Nominal and true stress–strain responses of HDPE and steel strip. c1–c4 Specimen preparation: formed carcass wrapped with HDPE, sectioned into Groups A and B, and compressed between flat plates. d Cross-sectional geometry and key dimensions of the formed carcass. e Post-test deformation shapes of Sp-A1, Sp-A2, Sp-B1 and Sp-B2. f Load–displacement responses (on a consistent load-per-length basis, n = 4 independent tests). HDPE high-density polyethylene, Sp specimen, s.d. standard deviation. EMM calibration benchmark and input data preparation. a1, a2 Uniaxial tensile tests of HDPE and steel strip (raw material). b1, b2 Nominal and true stress–strain responses of HDPE and steel strip. c1–c4 Specimen preparation: formed carcass wrapped with HDPE, sectioned into Groups A and B, and compressed between flat plates. d Cross-sectional geometry and key dimensions of the formed carcass. e Post-test deformation shapes of Sp-A1, Sp-A2, Sp-B1 and Sp-B2. f Load–displacement responses (on a consistent load-per-length basis, n = 4 independent tests). HDPE high-density polyethylene, Sp specimen, s.d. standard deviation. A 6-inch carcass segment (Fig. 4c1) was overwrapped with HDPE (Fig. 4c2) to suppress slip and ensure uniform load transfer during compression44. To examine the potential length effects relative to the helical pitch, the pipe was divided into four rings of two different lengths (Fig. 4c3). Following the protocol in ref. 45, each ring was then compressed radially between stainless-steel plates on a universal testing machine (Fig. 4c4). The cross-sectional profile and key dimensions of the carcass sample are summarised in Fig. 4d. After compression, all four specimens evolved from near-circular to ovalised sections (Fig. 4e), with hinge-like flattening appearing at the crown and the invert regions, typically leading to an eight-shape. This ovalisation pattern (crown/invert flattening with sidewall bulging) depicts the characteristic deformation mode of rings under diametral compression. The corresponding load–displacement curves are demonstrated in Fig. 4f. For comparison between the two ring lengths, the force is reported on a load-per-unit-length basis (F/L), enabling a consistent ring-level comparison across groups. The curves collapse into a tight band with nearly identical initial stiffness and smooth nonlinear softening, indicating good repeatability and that the response is insensitive to specimen length over the investigated range. Carcass: FE model-based EMM validation – collapse pressure prediction To translate the measured flat-plate response into an equivalent constitutive description, the experimental load-displacement curves (Fig. 4f) were mapped into the equivalent-material method (EMM). At each displacement, the local slope defined the incremental stiffness (Sinc), from which the tangent modulus (Et) and associated strain increment (εinc) of the EMM material can be obtained. Solving the resulting differential relation yields an equivalent stress-strain law that implicitly embeds residual stresses, cold-work plasticity, and small-scale geometric compliance inherited from the forming process. The inferred constitutive response shows an initial elastic region, followed by strain hardening (Fig. 5a), consistent with the behaviour of cold-worked steel46.Fig. 5EMM material validation and hydrostatic collapse modelling.a Back-inferred EMM material. b Boundary conditions for the flat-plate FEM. c Contact definitions, model surfaces and mesh details. d Comparison between experimental and EMM-model responses. e1 Quarter-ring FEM of the ideal carcass encased in PE. e2 Equivalent-layer model replacing the carcass with a homogeneous layer. f Contact representation. g1, g2 Load and boundary condition for both models. h Comparison between ideal and EMM-based carcass models. EMM equivalent material method, FEM finite element model, PE polyethylene, RP reference point, Eq. equivalent. EMM material validation and hydrostatic collapse modelling. a Back-inferred EMM material. b Boundary conditions for the flat-plate FEM. c Contact definitions, model surfaces and mesh details. d Comparison between experimental and EMM-model responses. e1 Quarter-ring FEM of the ideal carcass encased in PE. e2 Equivalent-layer model replacing the carcass with a homogeneous layer. f Contact representation. g1, g2 Load and boundary condition for both models. h Comparison between ideal and EMM-based carcass models. EMM equivalent material method, FEM finite element model, PE polyethylene, RP reference point, Eq. equivalent. The equivalent law was then implemented in a flat-plate FE benchmark by replacing the detailed carcass profile with a homogeneous ring of the same thickness. The plates were modelled as analytical rigid bodies; a prescribed displacement was applied to the top plate, while the bottom plate was fixed in all degrees of freedom, as shown in Fig. 5b. All interfaces were modelled with surface-to-surface contact (Fig. 5c): hard normal contact with separation allowed and penalty friction in the tangential direction (μ = 0.1)44. A static general algorithm with an elastoplastic constitutive framework and C3D8R elements was adopted here. The simulated load-displacement curves align closely with the experimental results across the full range, with only minor mid-range deviations (Fig. 5d). Specifically, the agreement is further quantified using the mean absolute error (MAE = 2.49N/mm) and the root mean square error (RMSE = 2.85N/mm), corresponding to a normalised RMSE (NRMSE) of 2.41%. These values indicate that the overall discrepancy remains within approximately 3% of the experimental mean response. In other words, the EMM ring successfully reproduces both the initial stiffness and the subsequent nonlinear softening observed in tests, demonstrating that the manufacturing imprints embedded in the equivalent law successfully govern the carcass mechanics at the ring scale (more details of radial compression FE model are provided in Supplementary Information Section 2). Hydrostatic collapse was assessed using two ring-scale FE models discretised with C3D8R elements: an explicit “ideal carcass” that preserved the interlocked profile and a homogeneous “equivalent-layer” ring assigned with the EMM-calibrated material. As ring models are known to reproduce full-pipe collapse behaviour47, the analysis was performed on a quarter-ring with symmetry to reduce computational cost (Fig. 5e1, e2). In the detailed model, both PE-carcass interaction and carcass self-contact were fully represented (Fig. 5f), while in the equivalent model, the PE sheath acted directly on the homogeneous layer. Boundary conditions were identical for both models (Fig. 5g1, g2): symmetry constraints prevented rigid-body movement, and a uniform external pressure was applied to the outer PE surface. Collapse was traced using the Riks method, which follows the unstable equilibrium path and ensures consistent comparison between detailed and equivalent representations (for more details of hydrostatic FE models, see Supplementary Information Section 3). Figure 5h summarises the hydrostatic collapse predictions of the FE models and highlights the role of manufacturing imprints. The ideal carcass model (PE + practical carcass profile) neglects forming history and therefore results in a relatively higher collapse pressure (Pcr = 5.59 MPa). In contrast, the EMM-based model incorporates forming-induced compliance (cold work, residual stress, and contact effects) into the equivalent constitutive law, yielding to a more conservative collapse pressure (Pcr = 4.12 MPa). Overall, considering manufacturing imprints through EMM reduces the predicted collapse capacity by ~ 26% relative to the idealised geometry, highlighting the importance of manufacturing effects. SSRFP: reinforcement calibration/validation and full-scale hydrostatic test To obtain a direct experimental validation of the workflow, the methodology was extended to a steel strip-reinforced flexible pipe (SSRFP) comprising an inner PE liner, four helically wound steel strip layers, and an outer PE sheath (architecture in ref. 48). A 4-inch SSRFP was sectioned into four short rings (Fig. 6a, b) and tested under flat-plate compression using the same displacement-control protocol (Fig. 6c) as for the aforementioned carcass samples. The resulting load-displacement curves (Fig. 6e) form a uniform family with consistent elastic slopes and smooth nonlinearity, providing a robust basis for calibration.Fig. 6Calibration and validation workflow for SSRFP.a Specimens cut from the reel. b Flat-plate setup with layer architecture. c Specimen deformation. d FE mesh for SSRFP. e Load--displacement responses. f Back-inferred EMM material. g Comparison of model responses. h Imperfection sensitivity of the EMM pipe. i Hyperbaric test setup: 1, pressurisation system; 2, hydrostatic chamber; 3, pipe specimen; 4, water tank; 5, electronic gauge; 6, recording system; 7, pumping tube; 8, drain tube; 9, data cable. j SSRFP specimens after hydrostatic testing. SSRFP steel strip reinforced flexible pipe, FE finite element, PE polyethylene. Calibration and validation workflow for SSRFP. a Specimens cut from the reel. b Flat-plate setup with layer architecture. c Specimen deformation. d FE mesh for SSRFP. e Load--displacement responses. f Back-inferred EMM material. g Comparison of model responses. h Imperfection sensitivity of the EMM pipe. i Hyperbaric test setup: 1, pressurisation system; 2, hydrostatic chamber; 3, pipe specimen; 4, water tank; 5, electronic gauge; 6, recording system; 7, pumping tube; 8, drain tube; 9, data cable. j SSRFP specimens after hydrostatic testing. SSRFP steel strip reinforced flexible pipe, FE finite element, PE polyethylene. The flat-plate response was reproduced in FE using the tested SSRFP geometry and materials (Fig. 6d), showing close agreement with the experiments across the full range (SSRFP modelling details follow ref. 49). Unlike the carcass-only condition, however, the inner and outer PE layers contributed substantially to radial stiffness (around 40–60%50). Consequently, treating the whole pipe as a single homogeneous layer would overestimate its capacity. To address this, calibration was decoupled from the assembly process51: The steel contribution was first isolated by removing the PE layers in the FE setup and extracting a steel-only load-displacement curve (more FE details are available in Supplementary Information Section 4). This curve was then supplied to the equivalent-material method (EMM) to back-infer a nonlinear constitutive law for the steel-strip stack (Fig. 6f). Finally, a three-layer model-inner PE/equivalent steel/outer PE-was re-assembled and shown to reproduce both the measured and simulated flat-plate responses of the complete SSRFP (Fig. 6g). It is noteworthy that a small discrepancy remains between the specimen-average curve (Sp. Ave) and the EMM-calibrated response (Eq. Material) at intermediate displacements. This difference stems from the fact that the EMM is calibrated against the FE model rather than the experimental data in order to maintain consistent boundary conditions. As the FE model does not incorporate initial imperfections (e.g., ovalisation), it inevitably yields a slightly higher response than the specimen-average curve. Nevertheless, the calibrated model captures both the initial stiffness and the overall nonlinear response with good fidelity. Quantitative comparison between the FE model and the EMM-calibrated response shows close agreement, reporting a mean absolute error (MAE) of 0.63 N/mm, a root mean square error (RMSE) of 0.81 N/mm, corresponding to a normalised RMSE (NRMSE) of 1.77 %. The small deviations further confirm the accuracy and reliability of the EMM-calibrated material representation. External-pressure resistance was then assessed at the pipe scale. The three-layered EMM model was subjected to uniform hydrostatic loading, with imperfection sensitivity summarised in Fig. 6h, yielding a critical pressure of Pcr = 2.38 MPa. In parallel, two full-scale hyperbaric tests on SSRFP specimens were conducted (Fig. 6i) following the external-pressure protocol of refs. 52,53. Collapse occurred at 2.13 and 2.20 MPa after stable ramping, accompanied by buckling (Fig. 6j), and corresponding to an average Pcr of 2.17 ± 0.05 MPa (mean ± s.d., n = 2). The simulated collapse pressure remains in close agreement with the experimental measurements54, confirming that the EMM framework provides an accurate prediction of SSRFP hydrostatic capacity (for more information, see Supplementary Information Section 5). This validation shows that EMM extends beyond carcass collapse, offering a generalised and cost-effective alternative to hyperbaric testing for pressurised structures. By linking benchtop calibration with structural-scale collapse prediction, the framework provides a transferable route for rapid assessment of external-pressure capacity across a broad class of multilayer pressurised systems. Discussion For comparison, several classical thickness-based approaches (ETMs) were also evaluated, with their governing expressions. While convenient, these baselines neglect interlock mechanics and forming-induced compliance as mentioned above. The post-processed responses of the ideal carcass model and the equivalent model (using the EMM representation here as an example) are shown in Fig. 7a, b, respectively. Mirror visualisation is adopted in panels a1 and b1 to improve clarity of the ring configuration. Panels a2-a4, b2-b4 show the von Mises stress fields at 25%, 50%, and 100% of the corresponding critical pressure Pcr, with all contours plotted using a deformation scale factor of 20.Fig. 7Comparison between EMM and ETMs.a1 Ideal carcass FEM. a2–a4 von Mises stress fields of the ideal carcass at 25%, 50%, and 100% of Pcr. b1 Equivalent FEM based on EMM. b2–b4 von Mises stress fields of the EMM model at 25%, 50% and 100% of Pcr. c Responses predicted by various models. d EMM and ETM formulas with their resulting values. e Capability map of equivalence strategies. EMM equivalent material method, ETM equivalent thickness method, FEM finite element model, Eq. equivalent, Teq equivalent thickness, Pcr critical pressure, UR radial displacement. Comparison between EMM and ETMs. a1 Ideal carcass FEM. a2–a4 von Mises stress fields of the ideal carcass at 25%, 50%, and 100% of Pcr. b1 Equivalent FEM based on EMM. b2–b4 von Mises stress fields of the EMM model at 25%, 50% and 100% of Pcr. c Responses predicted by various models. d EMM and ETM formulas with their resulting values. e Capability map of equivalence strategies. EMM equivalent material method, ETM equivalent thickness method, FEM finite element model, Eq. equivalent, Teq equivalent thickness, Pcr critical pressure, UR radial displacement. The full pressure-displacement responses are shown in Fig. 7c, in which classical thickness-based methods exhibit a noticeable dispersion in their predicted responses relative to the proposed EMM-based model. The area-equivalent model (Model A)22 follows the overall trend but underpredicts the initial stiffness and fails to reproduce the nonlinear regime. The inertia-based equivalents (Models B and C)23,25 predict relatively higher collapse pressures, mainly due to their larger effective thickness and the fact that profile-level compliance associated with interlock contact and relative slip is not directly embedded in these equivalence formulations. Further note that Model B’ uses only the equivalent-thickness expression from ref. 23, while omitting the associated equivalent modulus assignments. It is hence included as a thickness-only reference case for comparison with other equivalent-thickness methods. By contrast, the strain-energy match (Model D)29 yields a lower effective stiffness and thus a more conservative collapse prediction, consistent with the findings reported by Li et al.30,31. The ideal carcass model also predicts a stiffer response than EMM because it assumes a perfect geometry and does not include forming history (cold work/residual-stress-induced compliance), both of which reduce collapse resistance in practice. As outlined above, the proposed EMM establishes a direct link between benchtop mechanical characterisation and the collapse behaviour of the structural ring. Once the equivalent material law is obtained through EMM calibration, the original profiled carcass structure can be represented as a homogeneous ring endowed with the EMM-derived constitutive response. In the present study, the critical pressure is primarily evaluated using nonlinear FE analysis with the Riks method. Moreover, the equivalent ring formulation permits alternative solutions; in particular, the critical pressure can be evaluated analytically following the theoretical approaches reported in refs. 52,55. The corresponding theoretical prediction is included in Fig. 7c and is obtained from the intersection of the two red curves. To characterise deformation, radial displacements were averaged along a reference line on the inner bore, providing a representative measure for collapse. For the present geometry, the resulting equivalent thicknesses Teq, collapse pressures Pcr, and displacements UR for each model are summarised in Fig. 7d. In specific, the critical point was identified at the point where the applied pressure starts ceasing to increase with radial displacement. The ability to capture key sources of nonlinearity between the proposed EMM and classical ETMs is contrasted in Fig. 7e. Note that a stand-alone equivalent-thickness expression does not intrinsically embed geometric imperfections or material nonlinearity. Such effects can be further incorporated when coupled with an appropriate finite-element framework; accordingly, they are indicated in Fig. 7e as conditionally accounted for rather than directly embedded in the equivalent-thickness formulation itself. EMM, on the other hand, incorporates both geometric and material imprints through test-based inference, providing a more direct and physically grounded representation of the sample-dependent and process-dependent structural response. Overall, the EMM is not limited to carcasses or SSRFP alone. Its geometry-agnostic formulation makes it applicable across a wide range of pressurised systems, from flexible pipes and umbilicals to thin-walled shells and even biological ducts. Across these systems, variability in collapse resistance often stems from the combined influence of nonlinear material behaviour and small imperfections, whose overall effects are implicitly incorporated into the equivalent material law through the calibration. By embedding these effects in a transferable constitutive representation, the framework provides a unified approach to collapse assessment across different structural scales and geometries. More broadly, it suggests a pathway for bridging benchtop characterisation and system-level stability prediction in pressurised structures. Supplementary information Article File (PDF) Supplementary Information Description of Additional Supplementary Files Supplementary Data Article File (PDF) Supplementary Information Description of Additional Supplementary Files Supplementary Data