Papers by Keyword: Parameter Identification

Abstract: Machine learning (ML) algorithms have been studied in literature as an inverse method to predict material constitutive parameters. However, these approaches are often dependent on the mesh discretisation settings applied during numerical simulations, and then difficulty model adaptation to experimental digital image correlation (DIC) subsets. Although a recent study explores the use of an interpolation-based approach to achieve experimental adaptation from numerically-based trained ML models, the proposed methodology lacks evaluation using experimental data. As a follow-up, this study proposes a new evaluation approach. Numerical data is DIC-levelled via MatchID software and then submitted to interpolation. An XGBoost algorithm is then trained on interpolated DIC data and evaluated for parameter prediction, comparing the obtained results with those obtained from the model trained on interpolated numerical data. Overall, the proposed DIC-levelling and interpolation pipeline yields an excellent predictive performance, with results comparable to those obtained when training on interpolated numerical data. The largest deviations are observed for the hardening exponent, while the remaining parameters are predicted with consistently high accuracy. These findings validate the practical applicability of the interpolation-based strategy to reduce the subset scheme dependency of ML models trained on real experimental data.

Abstract: In this work, a new methodology for the identification of the CPB06 yield criterion parameters is presented. This methodology is based on the application of Machine Learning models and the Levenberg-Marquardt optimization algorithm. The proposed methodology relies on data obtained from a biaxial tensile test with a cruciform specimen and aims to overcome some of the challenges usually faced during material characterization with the CPB06 yield criterion. The predictive performances achieved were positive overall, when comparing the yield surfaces obtained for testing cases, highlighting the potential of the proposed methodology.

Abstract: This study proposes a Bayesian data assimilation approach to estimate material model parameters based on deformation fields measured via digital image correlation in a biaxial tensile test using a cruciform specimen. The anisotropy parameters and exponent of the Yld2000-2d yield function for a A5052P-H32 aluminum alloy are identified. The results indicate that the proposed method can estimate parameters with high accuracy—comparable to those identified via conventional multiaxial testing methods—while requiring only a single biaxial test. The proposed method offers an efficient framework for material modeling by minimizing a cost function via Bayesian optimization, enabling parameter identification from a single biaxial tensile test for sheet metal forming applications.

Abstract: Forming processes significantly influence the product properties of a formed workpiece. Next to the effects of work hardening and residual stresses, the influence of ductile damage determines the final performance of a formed component. Thus, precise damage models are crucial for designing new forming process sequences. In general, this is achieved by modelling the evolution of damage as a function of hydrostatic and deviatoric stress, characterized by the stress triaxiality and the Lode-parameter. However, calibrating damage models to the effects of triaxiality and the Lode-parameter is not trivial, since experiments usually represent a combination of both influences. A recent experimental approach by the authors offers the possibility to vary the Lode-parameter in extrusion experiments while keeping the triaxiality constant. This paper aims to use this data of the isolated deviatoric effect on damage to calibrate a damage evolution equation. The model is calibrated to void area fraction measurements obtained by scanning electron microscopy of extruded case-hardening steel 16MnCrS5. For validation, the model predictions for non-constant Lode-parameter histories are compared to corresponding experiments. The model and experiments are in good agreement.

Abstract: This study contributes to the development of more resilient and responsive control systems for industrial robotics. Industrial robot arms are subject to various vibrational forces during various operations, which can limit their accuracy and response time. This paper studies the vibration characteristics of a robotic arm through real world measurement and Finite Element Analysis (FEA). The robot arm is the MELFA RV-2SDB15. In this paper, the authors determine the dynamic parameters of the examined manipulator. Experimental measurement is carried out with a modal approach. Optimization techniques are employed to develop an accurate CAD model of the robotic arm.

Abstract: Today, the vast majority of design tasks are based on simulation tools. However, the success of the simulation depends on the accurate identification of the constitutive parameters of materials, i.e., its calibration. The classical parameter identification strategy, which relies on homogeneous tests, does not provide accurate and robust results required by the automotive and aerospace industry. Recently, numerical inverse methods, such as the Finite Element Model Updating (FEMU) and the Virtual Fields Method (VFM), have been developed for identifying constitutive parameters based on heterogeneous tests. Although these methods have proven effective for linear and non-linear models, the parameter identification process is complex, making it computationally expensive. In this work, a machine learning algorithm (XGBoost) is used to pursue the goal of parameter identification of non-linear models using heterogeneous tests. A statistical analysis is conducted to identify the correlation between the training dataset size, mechanical tests results and the material parameters. The goal is to understand the importance of the different inputs and to reduce the computational time.

Abstract: Advanced hardening models accurately describe the transient plastic behavior (reyielding, stagnation, resumption...) after various strain-path changes (reverse, orthogonal...). However, a common drawback of these models is that they usually predict monotonic loading with lower accuracy than the regular isotropic hardening models. Consequently, the finite element predictions using these models may sometimes lose in accuracy, in spite of their tremendous theoretical superiority. This drawback has been eliminated in the literature for the Chaboche isotropic-kinematic hardening model. In this work, a generic approach is proposed for advanced hardening models. Arbitrary models could be successfully compensated to preserve rigorously identical predictions under monotonic loading. A physically-based model involving a 4^th order tensor and two 2^nd order tensors was used for the demonstration. The parameter identification procedure was greatly simplified by rigorously decoupling the identification of isotropic hardening parameters from the other parameters.

Abstract: The reliability and predictive accuracy of forming simulation depend on both the material constitutive model and its inherent parameters. As opposed to conventional sheet metal material testing, heterogeneous mechanical tests provide more complex strain and stress states. Heterogeneous mechanical tests can be used to efficiently predict the material behavior in forming processes due to an improvement in the time required and accuracy in the identification of the parameters. The present work aims at identifying the Swift hardening law parameters of a dual-phase steel by means of an optimum-designed interior notched specimen that presents several strain and stress states simultaneously. The finite element model updating (FEMU) technique was used for the identification of parameters, by comparing a DIC-measured virtual material with numerical results iteratively DIC-filtered.

Abstract: This work aims to evaluate the predictive performance of various Machine Learning algorithms when applied to the prediction of material constitutive parameters, particularly the parameters of the Swift hardening law. For this, datasets were generated from the results of the numerical simulations of uniaxial tensile tests. The Machine Learning algorithms considered for this study are: Gaussian Process, Multi-layer Perceptron, Support Vector Regression, Decision Tree and Random Forest. These algorithms were used to train metamodels based on training sets considering different numbers of materials and input parameters, which were then used to predict the hardening law parameters. The Gaussian Process algorithm achieved the overall best predictive performances. The results obtained show the potential of Machine Learning algorithms for application on the identification of material constitutive parameters.

Abstract: Timber drying consists of reducing the moisture content up to a level required by the intended application of the wood product. A proper drying operation is essential to reduce time and energy, as well as to prevent defects. Numerical simulation of this class of problems constitutes an important tool available to the process engineer to define the best drying schedule. However, a successful prediction requires knowledge of the wood properties and additional process parameters. This work is inserted within this framework and aims at discussing strategies do determine material and process parameters using inverse problem techniques. The timber drying process accounts for the fully coupled solution of the heat and mass (moisture) transfer problem, whereas the inverse problem is solved within the time domain based on population-based optimization techniques.