Learning constitutive relations from indirect observations using deep neural networks
Not provided.
Computer Science↗
SEARCH · Search NASA
Results for “constitutive relations”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Remarks on turbulent constitutive relations
The paper demonstrates that the concept of turbulent constitutive relations can be used to construct general models for various turbulent correlations. Some of the Generalized Cayley-Hamilton formulas for relating tensor products of higher extension to tensor products of lower extension are introduced. The combination of dimensional analysis and invariant theory can lead to 'turbulent constitutive relations' (or general turbulence models) for, in principle, any turbulent correlations. As examples, the constitutive relations for Reynolds stresses and scalar fluxes are derived. The results are consistent with ones from Renormalization Group (RNG) theory and two-scale Direct-Interaction Approximation (DIA) method, but with a more general form.
Shih, Tsan-Hsing↗
Nonlinear Constitutive Relations for High Temperature Application, 1984
Nonlinear constitutive relations for high temperature applications were discussed. The state of the art in nonlinear constitutive modeling of high temperature materials was reviewed and the need for future research and development efforts in this area was identified. Considerable research efforts are urgently needed in the development of nonlinear constitutive relations for high temperature applications prompted by recent advances in high temperature materials technology and new demands on material and component performance. Topics discussed include: constitutive modeling, numerical methods, material testing, and structural applications.
Source record↗
MULTISTEP AND CONTINUOUS PHYSICS-INFORMED NEURAL NETWORK METHODS FOR LEARNING GOVERNING EQUATIONS AND CONSTITUTIVE RELATIONS
In this work, we investigate the applicability and relative merit of discrete and continuous versions of physics-informed neural network (PINN) methods for learning unknown governing equations or constitutive relations in a nonlinear dynamical system. In the case of unknown dynamics, entire right-hand-side (RHS) equations of the ordinary differential equations are unknown. In the case of unknown constitutive relations, however, the RHS equations are known up to the specification of constitutive relations (that may depend on the state of the system). We use a deep neural network to model unknown governing equations or constitutive relations. The discrete PINN approach combines classical multistep discretization methods for dynamical systems with neural-network-based machine learning methods. On the other hand, the continuous versions utilize deep neural networks to minimize the residual function for the continuous governing equations. We use the case of a fedbatch bioreactor system to study the effectiveness of these approaches and discuss conditions for their applicability. Our results indicate that the accuracy of the trained neural network models is much higher for the cases where we only have to learn a constitutive relation instead of all dynamics. This finding corroborates the well-known fact from scientific computing that building as much structural information as is available into an algorithm can enhance its efficiency and/or accuracy.
97 MATHEMATICS AND COMPUTING↗
Learning Constitutive Relations From Soil Moisture Data via Physically Constrained Neural Networks
Abstract The constitutive relations of the Richardson‐Richards equation encode the macroscopic properties of soil water retention and conductivity. These soil hydraulic functions are commonly represented by models with a handful of parameters. The limited degrees of freedom of such soil hydraulic models constrain our ability to extract soil hydraulic properties from soil moisture data via inverse modeling. We present a new free‐form approach to learning the constitutive relations using physically constrained neural networks. We implemented the inverse modeling framework in a differentiable modeling framework, JAX, to ensure scalability and extensibility. For efficient gradient computations, we implemented implicit differentiation through a nonlinear solver for the Richardson‐Richards equation. We tested the framework against synthetic noisy data and demonstrated its robustness against varying magnitudes of noise and degrees of freedom of the neural networks. We applied the framework to soil moisture data from an upward infiltration experiment and demonstrated that the neural network‐based approach was better fitted to the experimental data than a parametric model and that the framework can learn the constitutive relations.
54 ENVIRONMENTAL SCIENCES↗
A constitutive relation for the viscous flow of an oriented fiber assembly
A constitutive relation for an equivalent, homogeneous fluid is developed for the anisotropic viscous flow of an oriented assembly of discontinuous fibers suspended in a viscous fluid. The anisotropic viscous compliance matrix can be expressed in terms of three constants by assuming the equivalent fluid to be incompressible and the microstructure to have axial symmetry (transversely isotropic). By means of a micromechanics analysis, the three terms of the constitutive relation are expressed in terms of the viscosity of the matrix fluid, the fiber aspect ratio, and the fiber volume fraction. A comparison of the viscosity terms reveals that the elongational viscosity in the fiber direction varies as the square of the fiber aspect ratio and a complex function of the fiber volume fraction. Furthermore, the ratio of the axial elongational viscosity to the transverse elongational viscosity and both axial and transverse shear viscosities was shown to be 10 exp 4 - 10 exp 6 for fiber aspect ratio of 100-1000, except at extreme values of the fiber volume fraction.
Pipes, R. B.↗
Constitutive relations associated with the Mott-Smith distribution function
It is shown that the distribution function assumed by Mott-Smith determines a unique relation between heat flux, stress, and fluid velocity given by q = (3/2) tau u - i.e., it provides a constitutive relation for heat flux, and it also determines a simple expression for the ratio of third-order central moments. These expressions allow the equation of transfer for (C sub x) squared to be cast in a form that yields a nonlinear constitutive relation for stress. The results obtained from the Mott-Smith ansatz are compared with the theory of Baganoff and Nathenson (1970) and results from a numerical solution of the Boltzmann equation for shock-wave structure obtained by Hicks et al. (1972).
Nathenson, M.↗
Toward a Turbulence Constitutive Relation for Rotating Flows
In rapidly rotating turbulent flows the largest scales of the motion are in approximate geostrophic balance. Single-point turbulence closures, in general, cannot attain a geostrophic balance. This article addresses and resolves the possibility of constitutive relation procedures for single-point second order closures for a specific class of rotating or stratified flows. Physical situations in which the geostrophic balance is attained are described. Closely related issues of frame-indifference, horizontal nondivergence, Taylor-Proudman theorem and two-dimensionality are, in the context of both the instantaneous and averaged equations, discussed. It is shown, in the absence of vortex stretching along the axis of rotation, that turbulence is frame-indifferent. A derivation and discussion of a geostrophic constraint which the prognostic equations for second-order statistics must satisfy for turbulence approaching a frame-indifferent limit is given. These flow situations, which include rotating and nonrotating stratified flows, are slowly evolving flows in which the constitutive relation procedures are useful. A nonlinear non-constant coefficient representation for the rapid-pressure strain covariance appearing in the Reynolds stress and heat flux equations consistent with the geostrophic balance is described. The rapid-pressure strain model coefficients are not constants determined by numerical optimization but are functions of the state of the turbulence as parameterized by the Reynolds stresses and the turbulent heat fluxes. The functions are valid for all states of the turbulence attaining their limiting values only when a limit state is achieved. These issues are relevant to strongly vortical flows as well as flows such as the planetary boundary layers, in which there is a transition from a three-dimensional shear driven turbulence to a geostrophic or horizontal turbulence.
Ristorcelli, J. R.↗
On constitutive relations at finite strain - Hypo-elasticity and elasto-plasticity with isotropic or kinematic hardening
Nagtegaal and de Jong (1982) have studied stresses generated by simple finite shear in the case of elastic-plastic and rigid-plastic materials which exhibit anisotropic hardening. They reported that the shear stress is oscillatory in time. It was found that the occurrence of such an 'anomaly' is not restricted to anisotropic plasticity. Similar behavior in finite shear may result even in the case of hypoelasticity and classical isotropic hardening plasticity theory. The present investigation is concerned with the central problem of 'generalizing' with respect to the finite strain case, taking into account the constitutive relations of infinitesimal strain theories of classical plasticity with isotropic or kinematic hardening. The problem of hypoelasticity is also considered. It is shown that current controversies surrounding the choice of stress rate in the finite-strain generalizations of the constitutive relations and the anomalies surrounding kinematic hardening plasticity theory are easily resolvable.
Atluri, S. N.↗
Robust deep learning framework for constitutive relations modeling
Modeling the full-range deformation behaviors of materials under complex loading and materials conditions is a significant challenge for constitutive relations (CRs) modeling. Here, we propose a general encoder-decoder deep learning framework that can model high-dimensional stress-strain data and complex loading histories with robustness and universal capability. The framework employs an encoder to project high-dimensional input information (e.g., loading history, loading conditions, and materials information) to a lower-dimensional hidden space and a decoder to map the hidden representation to the stress of interest. We evaluated various encoder architectures, including gated recurrent unit (GRU), GRU with attention, temporal convolutional network (TCN), and the Transformer encoder, on two complex stress-strain datasets that were designed to include a wide range of complex loading histories and loading conditions. All architectures achieved excellent test results with an root-mean-square error (RMSE) below 1 MPa. Additionally, we analyzed the capability of the different architectures to make predictions on out-of-domain applications, with an uncertainty estimation based on deep ensembles. The proposed approach provides a robust alternative to empirical/semi-empirical models for CRs modeling, offering the potential for more accurate and efficient materials design and optimization.
36 MATERIALS SCIENCE↗
On the development of constitutive relations for metallic powders
This article describes the development of elastic and plastic constitutive relations as functions of relative density for partially consolidated -100 mesh aluminum powder. First, measurements of yield stress as a function of stress state and relative density are described. Measurements of the plastic strain increments associated with yielding in unconstrained compression tests and elastic properties, both as functions of relative density, are also described. The experimental results are combined with the associated flow rule to show that the yield surface is asymmetric with respect to hydrostatic tension and compression. Second, it is shown that the yield stress results can be represented by a two-part (capped Drucker-Prager) yield surface. The consolidation yield surface moves along the hydrostatic stress axis during densification, while the shear yield surface approaches the Mises yield surface. For the Al powder used in the present investigation, superposition of shear stress on a hydrostatic stress state aids the densification process. However, the hydrostatic stress requirement was found to be reduced by only about 20 pct for relative densities below 0.98.
Watson, T. J.↗
Constitutive relations for stress and heat flux in a shock wave
Gas passing through shock wave, investigating constitutive relations for stress and heat flux using Maxwell moment method
Baganoff, D.↗
The concept of constitutive relations and failure criteria and their essential role in design.
Stress states, structure deformation, failure criteria and constitutive relations in structural design noting tensile, strip and lap shear test and continuum model
STRESS CALCULATION↗
Nonlinear Constitutive Relations for High Temperature Applications, 1986
The purpose of the symposium was to review the state-of-the-art in nonlinear constitutive modeling of high temperature materials for aeronautics applications and to identify the need for future research and development efforts in this area. Through this symposium, it was recognized that considerable research efforts are urgently needed in the development of nonlinear constitutive relations for high temperature applications. In the aerospace industry this need is further prompted by recent advances in high temperature materials technology and new demands on material and component performance.
Source record↗
Constitutive relations for stress and heat flux in a shock wave
Mathematical analysis of shock wave structure incorporating Maxwell moment methods and Boltzmann equation for stress and heat flux constitutive relations
Baganoff, D.↗
Learning constitutive relations using symmetric positive definite neural networks
In this work, we present a new neural-network architecture, called the Cholesky-factored symmetric positive definite neural network (SPD-NN), for modeling constitutive relations in computational mechanics. Instead of directly predicting the stress of the material, the SPD-NN trains a neural network to predict the Cholesky factor of the tangent stiffness matrix, based on which the stress is calculated in incremental form. As a result of this special structure, SPD-NN weakly imposes convexity on the strain energy function, satisfies the second order work criterion (Hill's criterion) and time consistency for path-dependent materials, and therefore improves numerical stability, especially when the SPD-NN is used in finite element simulations. Depending on the types of available data, we propose two training methods, namely direct training for strain and stress pairs and indirect training for loads and displacement pairs. We demonstrate the effectiveness of SPD-NN on hyperelastic, elasto-plastic, and multiscale fiber-reinforced plate problems from solid mechanics. The generality and robustness of SPD-NN make it a promising tool for a wide range of constitutive modeling applications.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
On the Application of Deformation Kinetics to Nonlinear Constitutive Relations at Higher Temperatures
A single phenomenological constitutive equation is derived theoretically from first principles and applied to aluminum, tin and lead. The theory is based on deformation kinetics of steady creep in which the fundamental mechanism is atomic transport over potential barriers whose conformation is distorted by the application of a stress field. The form of the functional dependence of barrier distortion and stress over the entire temperature range is found to be a sigmoidal curve which tends to straight lines of a unit slope in the small and high stress regions. With this form of barrier distortion, the constitutive equation prediction the steady creep behavior of aluminum, tin and lead over a wide range of temperature and stress.
Valanis, K. C.↗
Constitutive Relation in Transitional Granular Flows
To study the constitutive behavior of granular materials, the presence of gravity is detrimental. Although empirical relations have been obtained for engineering designs to control granular flows on Earth, it is not known how well these Earthbound relations can be used in another gravity field. Fundamental understanding must be derived to reliably design for granular flows in space exploration. There are two extremes of granular flows of which significant amount of knowledge is available. One deals with a dense and quasi-static situation where the deformation rate nearly vanishes. The other deals with dilute and rapidly fluctuating grain velocities where particle inertia dominates. This project, funded by the NASA Microgravity Fluid Physics Program, aims to study this transitional regime via physical experiments and computer simulations. A conceptual model has been established as described below. There are two natural time scales in a granular flow. One is the travel time between two consecutive collisions and the other is the duration of a collision contact. At a very low shear-rate, the shear-induced particle velocity is low. Hence the travel time between collisions is longer than the contact time between colliding particles. Binary collisions prevail. As the shear-rate increases, the traveling time between collisions reduces and the probability of multiple collisions goes up. These particle groups disperse shortly after and new groups form. When shear-rate is further increased, clusters grow in size due to an increasing chance for free particles to join before groups have the time to disperse. The maximum cluster size may depend on the global concentration and material properties. As the solid concentration approaches zero, the cluster size goes to one particle diameter. The maximum possible cluster size under any condition is the container size, provided that the shear flow is inside a container. The critical shear-rates that dictate the initiation of the multiple contacts, and the size and lifetime of the collision clusters, are functions of the concentration also. A 'regime' theory has been proposed by Babic et al. This theory suggested that both the solid concentration, C, and the non-dimensional shear-rate, B, are important in determining the regimes of the granular constitutive law. Additional information is included in the original extended abstract.
Shen, Hayley H.↗