Study and quantitative characterisation of 3D printed costal cartilages for highly realistic clinical simulators

Medicine is evolving at an accelerating pace, continuously enhancing the quality of treatments and the performance that clinicians can offer their patients. This progress is increasingly supported by biomedical engineers, whose role is to optimise medical practice through the design and improvement of engineered procedures and tools. In this context, the development of highly realistic simulators plays a crucial role, allowing surgeons to practise interventions in a safe and controlled environment, refine their techniques and determine the best approach. For this purpose, it is essential that the training medium accurately reproduces the human body both in terms of physical behaviour and visual realism. The realistic modelling and prototyping of thoracic cage simulators exemplify this need for fidelity in surgical training.

The thoracic cage constitutes a shield for the heart and lungs, and, in addition, it guarantees, thanks to a certain degree of mobility, the respiratory function. From a skeletal point of view, the structure comprises the thoracic section of the vertebral column, the sternum and 12 pairs of ribs, of which the first seven are called “true” ribs, while the remaining ones are “false” ribs. This distinction is due to the different way in which the ribs connect to the rest of the skeleton: both types are articulated posteriorly to the corresponding vertebrae, but only the true ribs are connected anteriorly to the sternum via costal cartilage (CC) (Standring, 2016). The first false ribs (eighth, ninth and 10th pairs) are connected directly to the upper ribs through CC, while the last two pairs are called “floating” ribs, as they do not articulate anteriorly with other bones.

The length of CC segments increases from the first to the seventh pair of ribs and decreases at the lower costal levels (Standring, 2016). CC is hyaline cartilage with a biphasic structure: the internal elastic matrix (or mid-substance) is rich in proteoglycans and is surrounded externally by the perichondrium, a more rigid and fibrous sheath that confers robustness to the composite.

CC should be considered the main element responsible for the mobility of the rib cage, vital for respiration and for energy absorption during impacts. Studies such as that of Schaffer (Schaffer et al., 2025) confirm that it is the elasticity of CC that allows the thorax to deform, significantly reducing the probability of rib fracture.

Characterising the mechanical properties of this elastic portion that connects the bony elements of the thoracic cage, therefore, makes it possible to predict how it deforms due to respiratory movements and/or under the action of an external force, as in the case of medical intervention. It is therefore of critical importance to quantitatively characterise the behaviour of CC so as to be able to replicate it in clinical simulators and offer those who practise on them support as close to reality as possible.

This study aims to enhance the mechanical performance and aesthetic realism of clinical simulators by focusing on CCs. Based on a comprehensive literature review and analysis of existing models, a parametric three-dimensional (3D)-printable model was developed and validated through computational and experimental testing.

Among the surgical procedures that most benefit from high-fidelity thoracic simulators are the correction of pectus excavatum via the Nuss and Ravitch procedures, in which surgeons must manipulate and sense the resistance of CCs to achieve the desired thoracic reshape, as well as thoracotomy and video-assisted thoracoscopic surgery (VATS) approaches, which require trainees to develop tactile familiarity with rib and cartilage compliance (Servi et al., 2025). Paediatric thoracic procedures present an additional challenge, given the higher elasticity and smaller dimensions of cartilage in younger patients, highlighting the need for patient-specific and age-adapted simulation tools. The model proposed in this work directly targets these training scenarios.

The literature contains several studies focused on CC: the earliest concerned animal cartilage [mostly of porcine origin, as in Roy et al (2004)]; the most recent ones instead characterise mechanical properties of human CC such as the tensile elastic modulus, the flexural elastic modulus, the shear modulus and/or the failure loads for these same tests.

A first group of studies is distinguished by the choice to remove the perichondrium from the specimens before testing. Their results are collected in Table 1. The reason for separating these results from those of subsequent studies lies in the work of Forman and Kent (2011), which analyses the contribution of the perichondrium to the mechanical properties of the structure. Comparing specimens stripped of the fibrous outer layer with those in which it was left intact, he reveals its significant influence on overall elasticity: in particular, maintaining the perichondrium corresponds to an almost doubled stiffness.

Table 2 reports data from studies in which the perichondrium was preserved. A comparison between Tables 1 and 2 reveals substantial discrepancies in the reported values of the investigated properties. These differences can be attributed to several influencing factors, including the dimensions of the CC (length of the compressed segment, cross-sectional shape and width) and the resulting structural anisotropy, as well as subject-related variables such as age, degree of calcification and the presence of thoracic pathologies, including pectus excavatum (Feng et al. 2001).

In most of the aforementioned studies, ribs and CCs are modelled as tubular structures with elliptical cross-sections. This shape, which is not symmetric with respect to the centre, and a possibly nonuniform perichondrium thickness, can be identified as the cause of the difference between flexural resistance in the dorso-ventral and cranio-caudal directions (Gradischar et al., 2022; Lebschy et al., 2024).

Moreover, as shown by Sandoz et al. (2013), the ratio between the length of the cartilaginous portion and the total length of the costal arc (called “costal index” in the study) is not constant: it varies depending on the costal level considered and decreases progressively with age. This last aspect is particularly significant in cases where the thorax of a paediatric subject must be modelled.

Beyond geometric considerations, Guo et al., (2007) analyse the variation of the mechanical properties of CC in relation to sex and the progression of years. In particular, differences were found between infantile CC to which the maximum values correspond and those of the other groups. Quite the reverse, no statistically relevant differences are observed between adolescents and adults, as well as between subgroups of the same age range.

A significant age-related factor is the degree of calcification of CC. Indeed, like other articular cartilages, it is subject to a progressive ossification process with the passage of time, even before the age of 40 (Hawellek et al., 2024). This entails a nonnegligible variation in the stiffness of the structure: in particular, the study of Forman and Kent (2014) attests to an increase ranging from 2.3 to 3.8 times when the relative volumes of calcified CC go from 0% to 24%.

What has been highlighted so far may be considered as a first characterisation of the mechanical properties of CC and of their influencing factors; however, the premise of many of the studies considered was that of operating and performing tests on healthy tissues, which greatly reduces the field of applicability of the results. Recalling, in fact, that the design of the CC model that is the subject of this work is aimed at application in a surgical simulator, excluding the possibility that the modelled anatomy may be affected by malformations is a limiting factor. On the contrary, it could be particularly interesting to observe and quantify the alterations in the mechanical properties of CC for pathological subjects, because they are specifically the target of a medical intervention.

Several studies whose objective was analogous to the one proposed in the present work have been carried out. Most of them aim to model CC starting from actual data and focusing on mechanical characterisation.

Lebschy (2021), for example, proposes a finite element model (FEM) of the entire thorax (see Figure 1), in which CC is represented by a composite: a more rigid outer layer, symbolically representing the perichondrium, and a more elastic filling, the equivalent of the mid-substance. The values attributed to the respective elastic moduli are those that, in the finite element analyses (FEA), most faithfully reproduced the desired values, namely, those obtained from tests performed on samples of human CC and on the whole thorax.

Similar is the approach of Zhang et al. (2022) (see Figures 2 and 3), who design a corrugated structure intended as a CC substitute in patients requiring a prosthesis. The component is manufactured in polyetheretherketone (PEEK) by means of fused deposition modelling (FDM) and subjected to mechanical verification tests. The particularly interesting aspect of the study, in relation to the aims of this work, lies in the parametric approach adopted: once the geometry of the structure is defined, the authors identify three influencing variables (thickness [d], amplitude [h] and wavelength [λ] of the corrugated trajectory) and study the trend of stresses and of the equivalent elastic modulus in relation to these geometric variables via FEA and repeated dynamic tests on printed specimens. Zhang thus arrives at the conclusion that, through his model, by playing with the parameters, it is possible to cover the entire range of elastic variability of real CCs.

These studies, however, present certain limitations in relation to the aims of this work. Lebschy (2021) proposes an FEM, useful for computer simulations, but not for immediate practical application. Nevertheless, the study provides consistent geometric and mechanical data, along with an interesting structural proposal involving a composite component. The prosthetic model of Zhang and Kang (Kang et al., 2022; Zhang et al., 2020), on the other hand, is partially inaccurate, as the studies refer to elastic modulus and resistance values from Forman and Guo (Forman and Kent, 2014), which do not consider the nonnegligible contribution of the perichondrium. Furthermore, the high cost of PEEK, which is more than justified in in vivo applications, is not so for the purpose of training mannequins. The flat, corrugated structure proposed by Zhang and Kang, while efficient in replicating CC mechanically, is aesthetically not realistic, and would not be suitable for the application in surgical simulators that motivates this work; it nonetheless constitutes an excellent starting point in terms of a parametric approach to the problem.

Considering what emerged from the research, the design proposed in this work is aimed at producing a model that embodied precise physical and mechanical characteristics: the stiffness of the specimen had to be easily adaptable, to replicate that of the real structure; a differentiated flexural response in the dorso-ventral and cranio-caudal directions was necessary (it was imperative that deformation occur principally and more easily on the transverse plane, to best simulate the movement of the thoracic cage during respiration or under external stress); and the external profile of the specimen had to be modifiable, to faithfully reproduce the appearance of the cartilage.

A model of the CC is therefore conceived and realised. It consists of two elements:

1. An elliptical helical spring, with major and minor axes A and B of the trajectory, pitch P and wire diameter D as variable parameters.

2. An internal core, also elliptical, with axes a and b, integral with only one of the bases and free on the other side.

The material chosen for the 3D printing of the load-bearing structure is acrylonitrile butadiene styrene (ABS). The complete model is encased in a silicone matrix to achieve anatomical appearance.

The helical geometry was selected over alternative structural forms based on three concurrent design requirements that emerged from the clinical application context. Firstly, the need for anisotropic flexural compliance, approximately double in the dorso-ventral compared to the cranio-caudal direction, is naturally achieved by an elliptical helix trajectory, because the ratio of the second moments of area in the two bending planes can be tuned through the ellipse axes. A flat corrugated structure, such as that of Zhang et al. (2020), which constitutes the primary literature reference for this work, achieves anisotropy only in a planar sense and cannot be directly embedded in a 3D encasement volume without geometric discontinuities. Secondly, the cylindrical-envelope geometry of the helix provides a smooth internal void that accommodates silicone casting without undercuts, which is essential for the moulding process described in Section 3.6. Thirdly, and critically for the simulator application, the helical spring enclosed in a silicone matrix closely reproduces the external morphology of actual CC, a requirement explicitly identified as absent from the Zhang and Kang models (Section 1.3) (Kang et al., 2022; Zhang et al., 2020). It is acknowledged that a systematic cross-geometry exploration (e.g. comparing helical, lattice and corrugated structures within the same FEA framework) was not performed, and this constitutes a limitation of the present work. The parametric study reported in Sections 2.2 and 3 addresses design-space exploration within the chosen structural concept.

Given the large number of data available for characterisation and their variability, both in qualitative and quantitative terms, the work of Lebschy and Gradischar (Lebschy, 2021; Gradischar et al., 2022; Lebschy et al., 2024) is used as a reference, as it provides the most complete and coherent set. These studies, in fact, provide both the measurements of mechanical properties and of geometric properties for each analysed sample.

The initial proportions of the model were therefore defined starting from the mean values of the geometric characteristics, so that the load-bearing structure (i.e. the assembly of spring and core) as a whole would fit within the elliptical cylinder volume corresponding to those of the actual cartilage. It remained necessary to verify that this sizing was also appropriate from a mechanical standpoint; thus, in keeping with the approach of Zhang (Feng et al., 2001; Zhang et al., 2020), the trend of the mechanical properties of the structure in relation to the parameters was studied. Multiple alternative configurations of the model were created, each characterised by one of the possible combinations of the values assigned to the considered variables (see Figure 4).

The possible combinations are identified by the alphanumeric code R_xD_yP_z (see Table 3), where the indices x, y and z identify the value of the parameter to which they refer and where R is the ratio between the major and minor axes of the elliptical cross-section (1.25, 1.50 or 1.75), D is the wire diameter (4.0 or 4.5 mm) and P is the pitch (6, 7 or 8 mm).

The dimensions of the internal core, a and b, were obtained by offset from the axes of the helix trajectory, maximising the filling of the internal space while leaving a small margin to avoid interference between the two structures. The tested configurations are listed in Table 4.

Once defined, the different configurations were subjected to a series of FEA, performed through the simulation module of SolidWorks, with the aim of finding the ideal combination of parameters that would produce a component as close as possible to the real element. The objective was to characterise the mechanical behaviour of the model subjected to: axial tension; flexion in the cranio-caudal direction; and flexion in the dorso-ventral direction, in relation to the values of the geometric parameters of the structure.

Given the particular geometry of the model, which would make it complex to define elastic moduli, rather than operating in terms of stress and strain (σ − ε), it was chosen to study the relationship between applied force and resulting displacement (F − δ). Lebschy (2021), in fact, also provides the force–displacement graphs for each test performed.

Replicating the modus operandi of Lebschy and Gradischar (Lebschy, 2021; Lebschy et al., 2024), the tensile tests were set up on SolidWorks Simulation as dynamic tests, with force linearly increasing from 0 to 40 N (see Figure 5). This value represents, in fact, the upper limit of the linear elastic region for the cartilage, identified by the study as the interval of forces between 20 and 40 N.

One of the two bases was therefore immobilised with a fixed constraint, while the lateral faces of the other were constrained with sliders so that the only displacement resulting from the application of the loading was in the axial direction. In this way, the constraint conditions of a specimen during a mechanised tensile test were reproduced, in which displacements are confined to the sole loading direction.

Given the nature of the spring, characterised by a linear behaviour of the type F = −kδ, where k is the equivalent spring constant of the elliptical spring, to best approximate the desired F − δ trend, a stiffness was needed such that a displacement on the order of a few millimetres would be observed at the maximum applied force. The force–displacement behaviour of the reference human CC (see Figure 6) shows a linear region between 20 and 40 N that serves as the calibration target.

Starting from the results of the tensile simulations (see Table 5), only the configurations that showed maximum axial displacements (S_x) between 3.5 and 5.5 mm were considered satisfactory. In particular, configuration R1D2P3 was identified as optimal for tension, to which the smallest S_x corresponds.

As for the flexural tests, the simulations were set up by assimilating the specimen to a cantilever beam (see Figure 7). The base on which the loading was applied was constrained by sliders, disposed on the lateral faces of the base in the case of cranio-caudal flexion, while on the upper and lower faces for dorso-ventral flexion. In this way, displacements in directions other than the one being tested were prevented. A dynamic test was then set up, imposing a force linearly increasing from 0 to 4 N (Lebschy, 2021; Lebschy et al., 2024), directed along the two axes of interest in the respective tests.

It was precisely from the first preliminary set of simulations in which only the helical structure was tested that the necessity arose to add an internal core to the spring. On its own, the spring was too flexible, producing maximum displacements an order of magnitude greater than those expected, which are in the range of 4–8 mm for the dorso-ventral direction and approximately 2–4 mm for the cranio-caudal direction. Furthermore, despite the elliptical trajectory for the windings, implemented in the model precisely to obtain a differentiated response to flexion in the two directions, the discrepancy was not satisfactory: from preliminary tests, a difference of only about 10% was obtained between displacements of the same configuration in the cranio-caudal and dorso-ventral directions, whereas the flexural stiffness of CC in the frontal plane is about double that in the transverse plane. The reference force–displacement curves from the human CC specimens are shown in Figure 8.

It was therefore decided to add an internal core, integral with only one of the two bases and, on the other side, inserted in a slot cut into the base itself. In this way, its presence would not alter the tensile behaviour, as it would be free to slide within the slot, but would provide the necessary support to the structure during flexion, and its elliptical shape would differentiate the static moment in the two directions of interest.

As for the results of the final flexural tests (see Table 6), configurations with a cranio-caudal displacement (S_y) between 3 and 5 mm, and a dorso-ventral displacement (S_z) between 6 and 8 mm, were considered satisfactory. The configuration optimal for flexion was thus identified as R2D1P3, to which the smallest S_y and S_z among those considered acceptable correspond.

The FEA, beyond helping in identifying the most promising configurations, provided an overview of the relative trend of loadings and displacements with respect to the identified variables. In particular, from the tensile tests it emerges that, all other parameters being equal, S_x decreases if: the ratio R is reduced; the diameter D and/or the pitch P are increased. These conclusions are consistent with the mechanical properties of a helical spring with circular cross-section, whose spring constant increases with larger wire diameter and fewer active coils.

Analogously, in the flexural tests, the results suggest that displacements reduce, that is, stiffness is greater, if the ratio R and/or the pitch P increase; and the diameter D decreases. This behaviour can be traced back to the sizing system adopted for the core, which, as already noted, is the main element responsible for the flexural behaviour of the structure. Increasing R and/or decreasing D results in greater internal clearance and thus more space for the core, whose axes a and b increase, increasing its static moment and therefore its flexural stiffness.

Considering the reference values and the general trends of S_x, S_y and S_z in relation to the geometric parameters, the configurations R1D2P3 and R2D1P3 were identified as tensile and flexural optima, respectively; however, these had unsatisfactory results in the opposite test. The idea was therefore to “hybridise” the two structures, producing a third one, called R1*D2P3, composed of the spring body of R1D2P3 and of a core as close as possible, geometry permitting, to that of R2D1P3. In this way, the tensile behaviour of the optimum would have been maintained, as it is not influenced by the core, while optimising the flexural response at the same time. The parameters of the hybrid configuration are summarised in Table 7.

FEA were then carried out on the hybrid configuration as well, which returned excellent results: the displacements in all three directions (S_x = 3.90 mm, S_y = 3.80 mm and S_z = 7.70 mm) fall within the previously defined acceptance margins.

The load-bearing structure (spring with core) of the hybrid configuration R1*D2P3 was printed by means of an Arburg freeformer 750-3x, it being the most promising among the structures tested in the FEA.

The choice of this industrial fused filament fabrication (FFF) platform and of ABS as a build material was driven by specific technical constraints of the design. The structure involves internal clearances between helical windings and an elliptical core of 0.2–0.6 mm: the Arburg freeformer operates with a 0.2 mm nozzle diameter, enabling reliable reproduction of these fine features (ARBURG GmbH + Co KG, 2023). Standard desktop FFF printers, which typically operate at 0.4 mm nozzle diameter, would not provide sufficient resolution to guarantee the independence of the two structural elements, a prerequisite for the intended mechanical behaviour. Furthermore, the complex internal geometry of the spring-core assembly requires a support material that can be removed without mechanical stress that could damage the fine structure. The freeformer uses Armat 11, a water-soluble support, enabling complete removal by soaking in heated water rather than by mechanical action. ABS was selected for its well-characterised elastic behaviour in the small-displacement regime, its compatibility with the freeformer process, and its substantially lower cost compared to PEEK. Photopolymer resins (stereolithography/digital light processing) were considered but set aside due to their tendency to exhibit viscoelastic behaviour and brittleness at small cross-sections, both of which would compromise the spring-like response and long-term mechanical repeatability of the structure.

The accessory components of the specimen were optimised to facilitate the execution of tests: the size of the lateral bases was increased, their cross-section changed from elliptical to rectangular and two holes were added to each, to allow the anchoring of the structure and the application of forces more easily. The nozzle diameter, and therefore the thickness of each material layer, was 0.2 mm.

The preliminary print, however, revealed a limitation in the initial model: the clearances between core and windings, as well as the slot between core and free base, proved too thin, producing specimens in which the ABS of theoretically separate surfaces was in fact bonded, making it impossible to stretch the spring. The clearances were therefore increased: the dimensions a and b of the core were reduced, and the slot width (L) was increased from 0.2 to 0.6 mm. The modified configuration R1*D2P3 was then reprinted without adhesion issues (see Figure 9).

FEA were re-run on this modified configuration: the axial displacement remained unchanged (as expected, because the core does not contribute to the tensile response), while the flexural displacements increased slightly (S_y = 4.67 mm and S_z = 9.44 mm), with the cranio-caudal value still within the acceptance range and the dorso-ventral value slightly exceeding it.

A support structure in PLA was designed in SolidWorks and printed with a Prusa MK3S+, with a geometry such that it could be used for all three tests (tension and flexion in both directions), simply changing the resting side (see Figure 10). Two specimens of load-bearing structure were tested under tension (ST1 and ST2) and two under each flexural direction (SFCC1, SFCC2 for cranio-caudal; SFDV1, SFDV2 for dorso-ventral).

Tensile tests imposed axial forces increasing from 10 to 40 N with a step of 5 N; flexural tests imposed forces increasing from 1 to 4 N with a step of 0.5 N. Displacements were measured by tracking a reference point on the loaded base through high-resolution video recording, using Tracker software to reconstruct the point’s trajectory over time and calculate the resulting displacement by subtracting the initial position (see Figure 11).

Interpolating the displacement values for each applied force, force–displacement curves were traced for each specimen (see Figure 12). The obtained trends are approximately linear for all three loadings. In the flexural tests, maximum displacements close to those predicted by the FEA were observed, while in the tensile tests, the structures showed a stiffness lower than expected.

A plausible explanation for the increased displacements lies in the manufacturing method itself: 3D printing, in fact, produces parts whose integrity and structure at the microscopic level and therefore all the mechanical properties deriving from them are not the same as those of a solid component, which is however the reference adopted during simulations. The predictable degradation of resistance and stiffness properties with respect to FEA is precisely the reason why those configurations producing the smallest possible displacements were selected as optimal: in this way, the possibility that, once actually printed, the structures would be less rigid than expected was accounted for.

Given that both computational and experimental results had generally produced displacements slightly larger than desired, for the soft matrix a silicone of intermediate hardness was chosen, which could contribute, at least in part, to the stiffness of the structure. A mould in PLA was designed and printed, consisting of a central chamber in the shape of an elliptical cylinder into which the silicone would be poured and two lateral recesses, in which to position the bases of the specimens. Two silicones were tested: Ecoflex 00-50 (Smooth-On, Inc, 2023), slightly softer, which produced specimen PE; and Dragon Skin 10, which produced specimen PD. The complete specimens are shown in Figure 13.

The mechanical tests on the complete specimens followed the same modalities as those on the bare structures. The resulting force–displacement curves (see Figure 14) once again show an approximately linear trend. The displacements of the complete specimens are reduced compared to those of the corresponding bare structures and are closer to the reference values relative to the deformability of CC itself, and can therefore be considered a good approximation of them.

No significant differences are observed between the two types of silicone; however, given that the desired displacements are generally smaller than those obtained, it would seem more reasonable to opt for the silicone of greater hardness, that is, Dragon Skin 10. A summary of all displacement results is provided in Table 8.

This work aimed to study and quantitatively characterise the mechanical behaviour of CCs to develop a standardised 3D-printed model suitable for implementation within highly realistic surgical simulators. From the literature review, multiple studies on CC provided an overview of the geometric and mechanical properties of interest, as well as influencing factors, and two studies were selected as references for their completeness and coherence. Based on these references and a review of pre-existing models, a composite model was conceived, featuring a spring with an elliptical trajectory and an integral elliptical core attached to one base, encased in a silicone matrix to enhance geometric and aesthetic realism. The mechanical response of the load-bearing structure relative to geometric parameters was characterised through repeated FEA, allowing identification of the configuration that most accurately reproduced the expected behaviour. This optimal design was 3D-printed, encased in silicone and experimentally validated, showing displacements close to the reference values from human CC.

The proposed model is particularly well-suited to support surgical training in procedures that require direct manipulation of CCs. Primary target contexts include:

• Corrective surgery for pectus excavatum via the Nuss and Ravitch procedures, in which the trainee must develop an accurate sense of the resistance and compliance of the costal arches during thoracic reshaping.

• Thoracotomy and VATS, where familiarity with the anisotropic flexural behaviour of cartilage is essential for safe rib retraction.

• Paediatric thoracic procedures, for which the higher elasticity and smaller cartilage dimensions require dedicated training tools.

The parametric nature of the model directly supports these applications: by adjusting the geometric parameters R, D and P, the mechanical response can be tuned to match different costal levels, patient age groups (including infants, for whom CC stiffness differs markedly from adults) and pathological conditions such as calcified or malformed cartilages that are the specific target of corrective surgery.

The parametric nature of the model allows adaptation to specific design requirements, enabling a patient-specific approach that considers influencing factors such as age and pathology. The dual-component design, with a load-bearing structure and soft silicone matrix, not only mimics the biphasic nature of actual CC but also compensates for compliance introduced by the FDM printing process, ensuring mechanical fidelity while maintaining aesthetic realism. Compared to existing thermoplastic polyurethane (TPU)-based training mannequins and PEEK prosthetic models, the proposed approach offers systematic tuneability of mechanical properties, lower cost and enhanced visual realism.

Two limitations of the present work should be noted. Firstly, the design-space exploration was conducted within a single structural concept (the elliptical helix); a systematic comparison with alternative geometries such as lattice infill or corrugated flat structures was not performed, and this represents an avenue for future investigation. Secondly, the use of an industrial FFF platform (Arburg freeformer 750-3X) required for the fine clearances of the design limits immediate accessibility. Future work should verify whether desktop FFF printers with 0.2–0.25 mm nozzles or high-resolution resin printers can achieve equivalent results with comparable mechanical reliability, which would substantially improve the generalisability and adoption of the approach.

This paper describes the design, computational characterisation, fabrication and experimental validation of a parametric 3D-printed model of human CC designed for use in high-fidelity surgical simulators. The proposed model, consisting of an ABS-printed elliptical helical spring with an internal elliptical core enclosed in a silicone matrix, successfully replicates the anisotropic mechanical behaviour of actual CC: the optimised configuration (R1*D2P3, Dragon Skin 10 encasement) achieved axial displacements of 4.8 mm, cranio-caudal flexural displacements of 3.8 mm and dorso-ventral flexural displacements of 7.2 mm under the reference loading conditions. All dorso-ventral compliance was roughly double that in the cranio-caudal direction, which is consistent with the anisotropic behaviour of actual CC described in the literature.

The systematic parametric examination of 18 geometric configurations using FEA revealed obvious design criteria that relate the spring aspect ratio, wire diameter and pitch to the tensile and flexural mechanical responses. These design rules allow for direct customisation of the model’s stiffness to match specific patient profiles, costal levels or pathological conditions, distinguishing this approach from existing uniform-infill TPU mannequin components and PEEK prosthetic models that focus on in vivo rather than training applications.

Based on these findings, the model is recommended for use in surgical simulation platforms aimed at thoracic procedures, specifically corrective surgery for pectus excavatum (Nuss and Ravitch procedures), thoracotomy and VATS, which are the primary contexts in which realistic haptic feedback from CC is clinically relevant. Future validation studies should include the component’s integration into a full thoracic cage simulator with rib and sternum surrogates, as well as face-validity and haptic-fidelity assessments with surgical trainees and experienced thoracic surgeons, to ensure that the mechanical and aesthetic realism achieved in this work translates to effective and meaningful training.