Efficient materially nonlinear [Formula: see text]FE solver for simulations of trabecular bone failure

A cluster-based incremental potential approach for reduced order homogenization of bones

We develop a cluster-based model order reduction (called C-pRBMOR) approach for efficient homogenization of bones, compatible with a large variety of generalized standard material (GSM) models. To this end, the pRBMOR approach based on a mixed incremental potential formulation is extended to a clustered version for a significantly improved computational efficiency. The microscopic modeling of bones falls into a mixed incremental class of the GSM framework, originating from two potentials. An offline phase of the C-pRBMOR approach includes both a clustering analysis spatially decomposing the micro-domain within an RVE and a space-time decomposition of the microscopic plastic strain fields. A comparative study on two different clustering approaches and two algorithms for mode identification is additionally conducted. For an online analysis, a cluster-enhanced version of evolution equations for the reduced variables is derived from an effective incremental variational formulation, rendering a very small set of nonlinear equations to be numerically solved. Several numerical examples show the effectiveness of the C-pRBMOR approach. A striking acceleration rate beyond 104 against conventional FE computations and that beyond 103 against the original pRBMOR approach are observed.

Comparison of simplified bone-screw interface models in materially nonlinear μFE simulations

Micro finite-element (μFE) simulations serve as a crucial research tool to assist laboratory experiments in the biomechanical assessment of screw anchorage in bone. However, accurately modelling the interface between bone and screw threads at the microscale poses a significant challenge. Currently, the gold-standard approach involves employing computationally intensive physical contact models to simulate this interface. This study compared nonlinear μFE predictions of deformations, whole-construct stiffness, maximum force and damage patterns of three different computationally efficient simplified interface approaches to the general contact interface in Abaqus Explicit, which was defined as gold-standard and reference model. The μCT images (resolution: 32.8 μm) of two human radii with varying bone volume fractions were utilized and a screw was virtually inserted up to 50% and 100% of the volar-dorsal cortex distance. Materially nonlinear μFE models were generated and loaded in tension, compression and shear. In a first step, the common simplification of using a fully-bonded interface was compared to the general contact interface, revealing overestimations of whole-construct stiffness (19% on average) and maximum force (26% on average), along with inaccurate damage pattern replications. To enhance predictions, two additional simplified interface models were compared: tensionally strained element deletion (TED) and a novel modification of TED (TED-M). TED deletes interface elements strained in tension based on a linear-elastic simulation before the actual simulation. TED-M extends the remaining contact interface of TED by incorporating neighboring elements to the contact area. Both TED and TED-M reduced the errors in whole-construct stiffness and maximum force and improved the replication of the damage distributions in comparison to the fully-bonded approach. TED was better in predicting whole-construct stiffness (average error of 1%), while TED-M showed lowest errors in maximum force (1% on average). In conclusion, both TED and TED-M offer computationally efficient alternatives to physical contact modelling, although the fully-bonded interface may deliver sufficiently accurate predictions for many applications.

Development of a density-based topology optimization of homogenized lattice structures for individualized hip endoprostheses and validation using micro-FE

Prosthetic implants, particularly hip endoprostheses, often lead to stress shielding because of a mismatch in compliance between the bone and the implant material, adversely affecting the implant's longevity and effectiveness. Therefore, this work aimed to demonstrate a computationally efficient method for density-based topology optimization of homogenized lattice structures in a patient-specific hip endoprosthesis. Thus, the root mean square error (RMSE) of the stress deviations between the physiological femur model and the optimized total hip arthroplasty (THA) model compared to an unoptimized-THA model could be reduced by 81 % and 66 % in Gruen zone (GZ) 6 and 7. However, the method relies on homogenized finite element (FE) models that only use a simplified representation of the microstructural geometry of the bone and implant. The topology-optimized hip endoprosthesis with graded lattice structures was synthesized using algorithmic design and analyzed in a virtual implanted state using micro-finite element (micro-FE) analysis to validate the optimization method. Homogenized FE and micro-FE models were compared based on averaged von Mises stresses in multiple regions of interest. A strong correlation (CCC > 0.97) was observed, indicating that optimizing homogenized lattice structures yields reliable outcomes. The graded implant was additively manufactured to ensure the topology-optimized result's feasibility.

Efficient nonlinear homogenization of bones using a cluster-based model order reduction technique

We present a reduced order model for efficient nonlinear homogenization of bones, accounting for strength difference effects and containing some well-known plasticity models (like von Mises or Drucker-Prager) as special cases. The reduced order homogenization is done by using a cluster-based model order reduction technique, called cluster-based nonuniform transformation field analysis. For an offline phase, a space-time decomposition is performed on the mesoscopic plastic strain fields, while a clustering analysis is employed for a spatial decomposition of the mesoscale RVE model. A volumetric-deviatoric split is additionally introduced to capture the enriched characteristics of the mesoscopic plastic strain fields. For an online analysis, the reduced order model is formulated in a unified minimization problem, which is compatible with a large variety of material models. Both cortical and trabecular bones are considered for numerical experiments. Compared to conventional FE-based RVE computations, the developed reduced order model renders a considerable acceleration rate beyond10 3 , while maintaining a sufficient accuracy level.

Investigation on primary stability of dental implants through considering peri-implant bone damage, caused by small and large deformations: A validated non-linear micro finite element study

Primary stability of a dental implant is defined as its ability to resist the applied load without showing excessive damage in peri-implant bone, which is a prerequisite for secondary stability, and consequently for implantation success. The main goal of this study was to develop a validated micro-finite element (μFE) approach to assess the primary stability of dental implants in terms of stiffness, stiffness reduction, and irreversible displacement of the bone-implant system, subjected to an increasing step-wise quasi-static compressive loading-unloading test. The μFE models were generated based on the μCT images of bone, taken from extracted bovine tibia trabecular bone samples after drilling and implantation. A tissue constitutive model was considered for trabecular bone by describing elasto-plasticity with a modified von Mises yield criterion and element deletion technique to account for trabecular bone damage behavior. Then, the obtained force-displacement curves from the simulation were compared with the in-vitro mechanical test curves to evaluate the validity of the model. The results showed that the proposed μFE model could be properly predict the bone-implant system mechanical response in terms of irreversible displacement (R2 = 0.99), stiffness (R2 = 0.77), and stiffness reduction (R2 = 0.72) of the bone-implant construct for all the applied displacements without a significant difference from the unit slope and zero intercept of the QQ-plot (p-value<0.05). Moreover, a qualitative agreement was seen between the peri-implant bone damage predicted by the μFE model and the observed from μCT images. The adopted methodology used in this study can predict the mechanical failure response of the bone-implant system, which can be employed as a representative tool to study the effects of various dental implant design parameters on the primary stability with the ultimate goal of optimizing dental implants design.

Comparison of linear and nonlinear stepwise μFE displacement predictions to digital volume correlation measurements of trabecular bone biopsies

Digital volume correlation (DVC) enables to evaluate the ability of μFE models in predicting experimental results on the mesoscale. In this study predicted displacement fields of three different linear and materially nonlinear μFE simulation methods were compared to DVC measured displacement fields at specific load steps in the elastic regime (Step_El) and after yield (Step_Ult). Five human trabecular bone biopsies from a previous study were compressed in several displacement steps until failure. At every compression step, μCT images (resolution: 36 μm) were recorded. A global DVC algorithm was applied to compute the displacement fields at all loading steps. The unloaded 3D images were then used to generate homogeneous, isotropic, linear and materially nonlinear μFE models. Three different μFE simulation methods were used: linear (L), nonlinear (NL), and nonlinear stepwise (NLS). Regarding L and NL, the boundary conditions were derived from the interpolated displacement fields at Step_El and Step_Ult, while for the NLS method nonlinear changes of the boundary conditions of the experiments were captured using the DVC displacement field of every available load step until Step_El and Step_Ult. The predicted displacement fields of all μFE simulation methods were in good agreement with the DVC measured displacement fields (individual specimens: R²>0.83 at Step_El and R²>0.59 at Step_Ult; pooled data: R²>0.97 at Step_El and R²>0.92 at Step_Ult). At Step_El, all three simulation methods showed similar intercepts, slopes, and coefficients of determination while the nonlinear μFE models improved the prediction of the displacement fields slightly in all Cartesian directions at Step_Ult (individual specimens: L: R²>0.59 and NL, NLS: R²>0.68; pooled data: L: R²>0.92 and NL, NLS: R²>0.94). Damaged/overstrained elements in L, NL, and NLS occurred at similar locations but the number of overstrained elements was overestimated when using the L simulation method. Considering the increased solving time of the nonlinear μFE models as well as the acceptable performance in displacement prediction of the linear μFE models, one can conclude that for similar use cases linear μFE models represent the best compromise between computational effort and accuracy of the displacement field predictions.

High performance multi-platform computing for large-scale image-based finite element modeling of bone

BACKGROUND

Image-based finite element (FE) modeling of bone is a non-invasive method to estimate bone stiffness and strength. High-resolution imaging data as input allows for inclusion of bone microarchitecture but results in large amounts of data unsuitable for traditional FE solvers. Bone-specific mesh-free solvers have been developed over the past 20 years to improve on memory efficiency in simulated bone loading applications. The objective of this study was to provide linear performance benchmarking for a bone-specific, mesh-free solver (FAIM) using µCT and HR-pQCT image data on Mac, Linux, and Windows operating systems using both single- and multi-thread CPU and GPU processing.

METHODS

The focus is on the linear gradient-descent solver using standardized uniaxial loading of bone models from µCT, and first- and second-generation HR-pQCT scans of the radius and tibia. Convergence, speedup, memory, and batch performance tests were completed using CPUs and GPUs on three laboratory-based systems with Windows, Linux, and Mac operating systems.

RESULTS

Although varying by system and model size, time-per-iteration was as low as 0.03 s when an HR-pQCT-based radius model (6.45 million DOF) was solved with 3 GPUs. Strong scaling was achieved with GPU and CPU parallel processing, with strong parallel efficiencies when models were solved using 3 GPUs or ≤ 10 CPU threads. Errors in force, strain energy density, and Von Mises stress were as low as 0.1% when a convergence tolerance of 10^-3 or smaller was used.

CONCLUSION

The results of this study indicate that to maximize computational efficiency and minimize model solution times using FAIM software under the standardized tested conditions using µCT, XCT1 and XCT2 HR-pQCT image data, convergence tolerance set to 10^-4, and 10 threads or 2 GPUs are sufficient for efficient solution times. Less strict convergence tolerances will improve solution times but will introduce more error in the outcome measures.

Mechanical Properties of Robocast Glass Scaffolds Assessed through Micro-CT-Based Finite Element Models

In this study, the mechanical properties of two classes of robocast glass scaffolds are obtained through Computed micro-Tomography (micro-CT) based Finite Element Modeling (FEM) with the specific purpose to explicitly account for the geometrical defects introduced during manufacturing. Both classes demonstrate a fiber distribution along two perpendicular directions on parallel layers with a 90∘ tilting between two adjacent layers. The crack pattern identified upon compression loading is consistent with that found in experimental studies available in literature. The finite element models have demonstrated that the effect of imperfections on elastic and strength properties may be substantial, depending on the specific type of defect identified in the scaffolds. In particular, micro-porosity, fiber length interruption and fiber detaching were found as key factors. The micro-pores act as stress concentrators promoting fracture initiation and propagation, while fiber detachment reduces the scaffold properties substantially along the direction perpendicular to the fiber plane.

Nonlinear micro finite element models based on digital volume correlation measurements predict early microdamage in newly formed bone

Bone regeneration in critical-sized defects is a clinical challenge, with biomaterials under constant development aiming at enhancing the natural bone healing process. The delivery of bone morphogenetic proteins (BMPs) in appropriate carriers represents a promising strategy for bone defect treatment but optimisation of the spatial-temporal release is still needed for the regeneration of bone with biological, structural, and mechanical properties comparable to the native tissue. Nonlinear micro finite element (μFE) models can address some of these challenges by providing a tool able to predict the biomechanical strength and microdamage onset in newly formed bone when subjected to physiological or supraphysiological loads. Yet, these models need to be validated against experimental data. In this study, experimental local displacements in newly formed bone induced by osteoinductive biomaterials subjected to in situ X-ray computed tomography compression in the apparent elastic regime and measured using digital volume correlation (DVC) were used to validate μFE models. Displacement predictions from homogeneous linear μFE models were highly correlated to DVC-measured local displacements, while tissue heterogeneity capturing mineralisation differences showed negligible effects. Nonlinear μFE models improved the correlation and showed that tissue microdamage occurs at low apparent strains. Microdamage seemed to occur next to large cavities or in biomaterial-induced thin trabeculae, independent of the mineralisation. While localisation of plastic strain accumulation was similar, the amount of damage accumulated in these locations was slightly higher when including material heterogeneity. These results demonstrate the ability of the nonlinear μFE model to capture local microdamage in newly formed bone tissue and can be exploited to improve the current understanding of healing bone and mechanical competence. This will ultimately aid the development of BMPs delivery systems for bone defect treatment able to regenerate bone with optimal biological, mechanical, and structural properties.

Loss of longitudinal superiority marks the microarchitecture deterioration of osteoporotic cancellous bones

Osteoporosis (OP), a skeletal disease making bone mechanically deteriorate and easily fracture, is a global public health issue due to its high prevalence. It has been well recognized that besides bone loss, microarchitecture degradation plays a crucial role in the mechanical deterioration of OP bones, but the specific role of microarchitecture in OP has not been well clarified and quantified from mechanics perspective. Here, we successfully decoupled and identified the specific roles of microarchitecture, bone mass and tissue property in the failure properties of cancellous bones, through μCT-based digital modeling and finite element method simulations on bone samples from healthy and ovariectomy-induced osteoporotic mice. The results show that the microarchitecture of healthy bones exhibits longitudinal superiority in mechanical properties such as the effective stiffness, strength and toughness, which fits them well to bearing loads along their longitudinal direction. OP does not only reduce bone mass but also impair the microarchitecture topology. The former is mainly responsible for the mechanical degradation of bones in magnitude, wherever the latter accounts for the breakdown of their function-favorable anisotropy, the longitudinal superiority. Hence, we identified the microarchitecture-deterioration-induced directional mismatch between material and loading as a hazardous feature of OP and defined a longitudinal superiority index as measurement of the health status of bone microarchitecture. These findings provide useful insights and guidelines for OP diagnosis and treat assessment.

A Damage Model to Trabecular Bone and Similar Materials: Residual Resource, Effective Elasticity Modulus, and Effective Stress under Uniaxial Compression

Experimental research of bone strength remains costly and limited for ethical and technical reasons. Therefore, to predict the mechanical state of bone tissue, as well as similar materials, it is desirable to use computer technology and mathematical modeling. Yet, bone tissue as a bio-mechanical object with a hierarchical structure is difficult to analyze for strength and rigidity; therefore, empirical models are often used, the disadvantage of which is their limited application scope. The use of new analytical solutions overcomes the limitations of empirical models and significantly improves the way engineering problems are solved. Aim of the paper: the development of analytical solutions for computer models of the mechanical state of bone and similar materials. Object of research: a model of trabecular bone tissue as a quasi-brittle material under uniaxial compression (or tension). The new ideas of the fracture mechanics, as well as the methods of mathematical modeling and the biomechanics of bone tissues were used in the work. Compression and tension are considered as asymmetric mechanical states of the material. Results: a new nonlinear function that simulates both tension and compression is justified, analytical solutions for determining the effective and apparent elastic modulus are developed, the residual resource function and the damage function are justified, and the dependences of the initial and effective stresses on strain are obtained. Using the energy criterion, it is proven that the effective stress continuously increases both before and after the extremum point on the load-displacement plot. It is noted that the destruction of bone material is more likely at the inflection point of the load-displacement curve. The model adequacy is explained by the use of the energy criterion of material degradation. The results are consistent with the experimental data available in the literature.

Non-invasive prediction of the mouse tibia mechanical properties from microCT images: comparison between different finite element models

New treatments for bone diseases require testing in animal models before clinical translation, and the mouse tibia is among the most common models. In vivo micro-Computed Tomography (microCT)-based micro-Finite Element (microFE) models can be used for predicting the bone strength non-invasively, after proper validation against experimental data. Different modelling techniques can be used to estimate the bone properties, and the accuracy associated with each is unclear. The aim of this study was to evaluate the ability of different microCT-based microFE models to predict the mechanical properties of the mouse tibia under compressive load. Twenty tibiae were microCT scanned at 10.4 µm voxel size and subsequently compressed at 0.03 mm/s until failure. Stiffness and failure load were measured from the load-displacement curves. Different microFE models were generated from each microCT image, with hexahedral or tetrahedral mesh, and homogeneous or heterogeneous material properties. Prediction accuracy was comparable among models. The best correlations between experimental and predicted mechanical properties, as well as lower errors, were obtained for hexahedral models with homogeneous material properties. Experimental stiffness and predicted stiffness were reasonably well correlated (R2 = 0.53-0.65, average error of 13-17%). A lower correlation was found for failure load (R2 = 0.21-0.48, average error of 9-15%). Experimental and predicted mechanical properties normalized by the total bone mass were strongly correlated (R2 = 0.75-0.80 for stiffness, R2 = 0.55-0.81 for failure load). In conclusion, hexahedral models with homogeneous material properties based on in vivo microCT images were shown to best predict the mechanical properties of the mouse tibia.

A Review on Multiscale Bone Damage: From the Clinical to the Research Perspective

The investigation of bone damage processes is a crucial point to understand the mechanisms of age-related bone fractures. In order to reduce their impact, early diagnosis is key. The intricate architecture of bone and the complexity of multiscale damage processes make fracture prediction an ambitious goal. This review, supported by a detailed analysis of bone damage physical principles, aims at presenting a critical overview of how multiscale imaging techniques could be used to implement reliable and validated numerical tools for the study and prediction of bone fractures. While macro- and meso-scale imaging find applications in clinical practice, micro- and nano-scale imaging are commonly used only for research purposes, with the objective to extract fragility indexes. Those images are used as a source for multiscale computational damage models. As an example, micro-computed tomography (micro-CT) images in combination with micro-finite element models could shed some light on the comprehension of the interaction between micro-cracks and micro-scale bone features. As future insights, the actual state of technology suggests that these models could be a potential substitute for invasive clinical practice for the prediction of age-related bone fractures. However, the translation to clinical practice requires experimental validation, which is still in progress.

High-Performance Computing Comparison of Implicit and Explicit Nonlinear Finite Element Simulations of Trabecular Bone

BACKGROUND AND OBJECTIVE

Finite element models built from micro-computed tomography scans have become a powerful tool to investigate the mechanical properties of trabecular bone. There are two types of solving algorithms in the finite element method: implicit and explicit. Both of these methods have been utilized to study the trabecular bone. However, an investigation comparing the results obtained using the implicit and explicit solvers is lacking. Thus, in this paper, we contrast implicit and explicit procedures by analyzing trabecular bone samples as a case study.

METHODS

Micro-computed tomography-based finite element analysis of trabecular bone under a direct quasi-static compression was done using implicit and explicit methods. The differences in the predictions of mechanical properties and computational time of the two methods were studied using high-performance computing.

RESULTS

Our findings indicate that the results using implicit and explicit solvers are well comparable, given that similar problem set up is carefully utilized. Also, the parallel scalability of the two methods was similar, while the explicit solver performed about five times faster than the implicit method. Along with faster performance, the explicit method utilized significantly less memory for the analysis, which shows another benefit of using an explicit solver for this case study.

CONCLUSIONS

The comparison of the implicit and explicit methods for the simulation of trabecular bone samples should be highly valuable to the bone modeling community and researchers studying complex cellular and architectured materials.

A Review on Recent Advances in the Constitutive Modeling of Bone Tissue

PURPOSE OF REVIEW

Image-based finite element analysis (FEA) to predict and understand the biomechanical response has become an essential methodology in musculoskeletal research. An important part of such simulation models is the constitutive material model of which recent advances are summarized in this review.

RECENT FINDINGS

The review shows that existing models from other fields were introduced, such as cohesion zone (cortical bone) or phase-field models (trabecular bone). Some progress has been made in describing cortical bone involving physical mechanisms such as microcracks. Problems with validations at different length scales remain a problem. The improvement of recent constitutive models is partially obscured by uncertainties that affect overall predictions, such as image quality and calibration or boundary conditions. Nevertheless, in vivo CT-based FEA simulations based on a sophisticated constitutive behavior are a very valuable tool for clinical-related osteoporosis research.