A virtual lab for Ni/PU hybrid foams: stochastic micromechanical identification and efficient numerical simulations

该项目的主要目标是建立一个开孔泡沫金属的虚拟实验室，该实验室可以根据几何和力学参数的介观和微观分布来预测宏观力学响应及其变化。一个离散的细观结构模型将充当虚拟实验室。它将考虑支撑连通性的变化、支撑尺寸的变化以及支撑材料参数的变化。该项目的创新之处主要集中在实验技术和随机参数辨识上，该泡沫的模型制作和随机几何和材料参数的实验辨识将针对最近开发的镍/PU杂化泡沫进行，该泡沫的成本大大低于其他开孔杂化金属泡沫。主要创新可以明显地区分为四个子目标。首先，一种实验程序，测量作为局部应变场函数的各个支柱以及几个孔的力。其次，将开发一个三维梁模型来描述金属泡沫的离散细观结构。光束模型将作为虚拟实验室，并将通过采用基于蒙特卡罗的方法来解释变化，在该方法中，随机参数被多次实现。单个支柱将由一系列一维梁来表示。一维梁的形式将是几何非线性的，并能够预测单根压杆的弹塑性响应。此外，完整的三维梁模型还将能够预测单个支柱的失效，并处理支柱与支柱之间的接触(梁与梁之间的接触)。此外，还将使用贝叶斯推理的逆建模过程来识别由一系列梁模拟的单根压杆的材料参数的分布。贝叶斯推理确保了一致的方法，将材料参数视为随机变量，而不是确定性变量。最后，基于细观尺度(几个孔隙)的实验结果，对三维光束模型(即虚拟实验室)进行了验证。为此，必须详细了解样品的支柱连通性和支柱尺寸。因此，这两个因素必须明确确定。一个成功的项目将导致一个完整的3D梁模型，能够解释泡沫结构中的随机效应，以改善通过数值模拟对宏观泡沫行为的预测。

The main goal of the project is a virtual lab for open-cell metal foams that can predict macroscopic mechanical responses and their variations, based on the mesoscopic and microscopic distributions of the geometrical and mechanical parameters. A discrete mesostructural model will act as the virtual lab. It will account for variations of the strut connectivity, variations of the strut dimensions and variations of the strut material parameters. The model developments and the experimental identification of the stochastic geometrical and material parameters will be performed for a recently developed Ni/PU hybrid foams, which are substantially cheaper than alternative open-cell hybrid metal foams.The innovations in the project largely focus on experimental techniques and stochastic parameter identification. The main innovations can clearly be distinguished in four sub-aims. First, an experimental procedure to measure forces as functions of local strain fields for individual struts, as well as for several pores. Second, a 3D beam model to represent the discrete mesostructure of metal foams will be developed. The beam model will serve as the virtual lab and will account for variations by employing a Monte Carlo based approach, in which numerous realisations of the stochastic parameters are made. Single struts will be represented by series of 1D beams. The 1D beam formulation will be geometrically nonlinear and able to predict the elastoplastic response of single struts. The full 3D beam model will furthermore be able to predict failure of individual struts and deal with strut-to-strut contact (beam-to-beam contact). Furthermore, an inverse modelling procedure using Bayesian Inference will be used to identify the distributions of the material parameters of single struts, when they are modelled by a series of beams. Bayesian Inference ensures a consistent approach to treat the material parameters as stochastic, and not deterministic, variables. Finally, the 3D beam model (i.e. the virtual lab) will be validated based on experimental results at the mesoscale (several pores). The strut connectivity and strut dimensions of the samples must be known in detail for this. Therefore, both factors must be expliciteöy determined.A successful project will result in a full 3D beam model able to account for stochastic effects in the foams structure to improve the prediction of the macroscopic foams behaviour by numerical simulations.