Partial Differential Equations (PDE) are fundamental to model different phenomena in science and engineering mathematically. Solving them is a crucial step towards a precise knowledge of the behaviour of natural and engineered systems. In general, in order to solve PDEs that represent real systems to an acceptable degree, analytical methods are usually not enough. One has to resort to discretization methods. For engineering problems, probably the best known option is the finite element method (FEM). However, powerful alternatives such as mesh-free methods and Isogeometric Analysis (IGA) are also available. The fundamental idea is to approximate the solution of the PDE by means of functions specifically built to have some desirable properties. In this contribution, we explore Deep Neural Networks (DNNs) as an option for approximation. They have shown impressive results in areas such as visual recognition. DNNs are regarded here as function approximation machines. There is great flexibility to define their structure and important advances in the architecture and the efficiency of the algorithms to implement them make DNNs a very interesting alternative to approximate the solution of a PDE. We concentrate in applications that have an interest for Computational Mechanics. Most contributions that have decided to explore this possibility have adopted a collocation strategy. In this contribution, we concentrate in mechanical problems and analyze the energetic format of the PDE. The energy of a mechanical system seems to be the natural loss function for a machine learning method to approach a mechanical problem. As proofs of concept, we deal with several problems and explore the capabilities of the method for applications in engineering.
Mountain regions worldwide present a pronounced spatiotemporal precipitation variability, which added to scarce monitoring networks limits our understanding of the generation processes involved. To improve our understanding of clouds and precipitation dynamics and cross-scale generation processes in mountain regions, we analyzed spatiotemporal rainfall patterns using satellite cloud products (SCP) in the Paute basin (900–4200 m a.s.l. and 6481 km2) in the Andes of Ecuador. Precipitation models, using SCP and GIS data, reveal the spatial extension of three regimes: a three-modal (TM) regime present across the basin, a bimodal (BM) regime, along sheltered valleys, and a unimodal (UM) regime at windward slopes of the eastern cordillera. Subsequently, the spatiotemporal analysis using synoptic information shows that the dry season of the BM regime during boreal summer is caused by strong subsidence inhibiting convective clouds formation. Meanwhile, in UM regions, low advective shallow cap clouds mainly cause precipitation, influenced by water vapor from the Amazon and enhanced easterlies during boreal summer. TM regions are transition zones from UM to BM and zones on the windward slopes of the western cordillera. These results highlight the suitability of satellite and GIS data-driven statistical models to study spatiotemporal rainfall seasonality and generation processes in complex terrain, as the Andes.
SUMMARYA homogenization method is proposed for sandwich structures consisting of two plates interlaced with beams and shells in a periodic, lattice structure. The proposed method is a quasi-continuum approach where the constitutive response is obtained from the generalized forces of the interlacing elements. Buckling is studied as part of this model. Comparison of the homogenized model with fully discrete models show reasonable to very good agreement.
SUMMARYThe work focuses on the presently existing families of ÿnite elements with embedded discontinuities and explores the possibilities of obtaining symmetric statically consistent ÿnite elements that alleviate the stress-locking problem. For this purpose, mixed (reduced integration) and assumed enhanced strain techniques are applied to the basic symmetric four-noded element. Numerical simulations show the e ectiveness of the proposed measures.
SUMMARYA methodology to model shear bands as strong discontinuities within a continuum mechanics context is presented. The loss of hyperbolicity of the IBVP is used as the criterion for switching from a classical continuum description of the constitutive behaviour to a traction-separation model acting at the discontinuity surface. The extended finite element method (XFEM) is employed for the spatial discretization of the governing equations. This enables the shear bands to be arbitrarily positioned within the mesh. Examples that study the shear band progression within a rate-independent material are presented.
A simple methodology to model shear bands as strong displacement discontinuities in an adaptive meshfree method is presented. The shear band is represented by a displacement jump at discrete particle positions. The displacement jump in normal direction is suppressed with penalty method. Loss of material stability is used as transition criterion from continuum to discontinuum. The method is twoand three-dimensional. Examples of complicated shear banding including transition from brittle-to-ductile failure are studied and compared to experimental data and other examples from the literature.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.