Strain measurements and analyses around the bolt holes of structural steel plate connections using full-field measurements

Toussaint, Évelyne; Durif, Sébastien; Bouchaïr, Abdelhamid; Grédiac, Michel

doi:10.1016/j.engstruct.2016.11.033

Cited by 13 publications

(5 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For this experiment, following parameters used in the experimental work reported in Toussaint et al (2017), we consider a 6-mm-thick plate of S235 grade steel with three holes for the bolts, two at the bottom and one at the top, as shown in Figure 10. A uniaxial tension is applied at the bottom of the plate in the form of an imposed displacement of 1.5 mm.…”

Section: A Plane Stress Problemmentioning

confidence: 99%

Efficient training of physics‐informed neural networks via importance sampling

Nabian

Gladstone

Meidani

2021

Computer aided Civil Eng

113

View full text Add to dashboard Cite

Physics-informed neural networks (PINNs) are a class of deep neural networks that are trained, using automatic differentiation, to compute the response of systems governed by partial differential equations (PDEs). The training of PINNs is simulation free, and does not require any training data set to be obtained from numerical PDE solvers. Instead, it only requires the physical problem description, including the governing laws of physics, domain geometry, initial/boundary conditions, and the material properties. This training usually involves solving a nonconvex optimization problem using variants of the stochastic gradient descent method, with the gradient of the loss function approximated on a batch of collocation points, selected randomly in each iteration according to a uniform distribution. Despite the success of PINNs in accurately solving a wide variety of PDEs, the method still requires improvements in terms of computational efficiency. To this end, in this paper, we study the performance of an importance sampling approach for efficient training of PINNs. Using numerical examples together with theoretical evidences, we show that in each training iteration, sampling the collocation points according to a distribution proportional to the loss function will improve the convergence behavior of the PINNs training. Additionally, we show that providing a piecewise constant approximation to the loss function for faster importance sampling can further improve the training efficiency. This importance sampling approach is straightforward and easy to implement in the existing PINN codes, and also does not introduce any new hyperparameter to calibrate. The numerical examples include elasticity, diffusion, and plane stress problems, through which we numerically verify the accuracy and efficiency of the importance sampling approach compared to the predominant uniform sampling approach. INTRODUCTIONPhysics-informed neural networks (PINNs) leverage recent advances in deep neural networks to calculate

show abstract

Section: A Plane Stress Problemmentioning

confidence: 99%

Efficient training of physics‐informed neural networks via importance sampling

Nabian

Gladstone

Meidani

2021

Computer aided Civil Eng

113

View full text Add to dashboard Cite

show abstract

“…For this experiment, following parameters used in the experimental work reported in [53], we consider a 6 mm thick plate of S235 grade steel with three holes for the bolts, two at the bottom and one at the top, as shown in Figure 10. A uni-axial tension is applied at the bottom of the plate in the form of an imposed displacement of 1.5 mm.…”

Section: A Plane Stress Problemmentioning

confidence: 99%

Efficient training of physics-informed neural networks via importance sampling

Nabian,

Gladstone,

Meidani

2021

Preprint

View full text Add to dashboard Cite

Physics-Informed Neural Networks (PINNs) are a class of deep neural networks that are trained, using automatic differentiation, to compute the response of systems governed by partial differential equations (PDEs). The training of PINNs is simulation-free, and does not require any training dataset to be obtained from numerical PDE solvers. Instead, it only requires the physical problem description, including the governing laws of physics, domain geometry, initial/boundary conditions, and the material properties. This training usually involves solving a non-convex optimization problem using variants of the stochastic gradient descent method, with the gradient of the loss function approximated on a batch of collocation points, selected randomly in each iteration according to a uniform distribution. Despite the success of PINNs in accurately solving a wide variety of PDEs, the method still requires improvements in terms of computational efficiency. To this end, in this paper, we study the performance of an importance sampling approach for efficient training of PINNs.Using numerical examples together with theoretical evidences, we show that in each training iteration, sampling the collocation points according to a distribution proportional to the loss function will improve the convergence behavior of the PINNs training. Additionally, we show that providing a piecewise constant approximation to the loss function for faster importance sampling can further improve the training efficiency. This importance sampling approach is straightforward and easy to implement in the existing PINN codes, and also does not introduce any new hyperparameter to calibrate. The numerical examples include elasticity, diffusion and plane stress problems, through which we numerically verify the accuracy and efficiency of the importance sampling approach compared to the predominant uniform sampling approach.

show abstract

“…On the present work, Digital Image Correlation (DIC) method [6] is used as a noncontact method to detect and quantify displacement and strain field that occurs in the tested specimens. Another noncontact measurement method used in a previous experimental campaign [5], with accurate results, could be used to obtain local deformations of specimen using grid method [7]. However, as the strain levels induced around the slotted hole are high, DIC method was chosen.…”

Section: Digital Image Correlationmentioning

confidence: 99%

01.13: Experimental study of cover‐plate bolted joints with large or slotted holes

et al. 2017

View full text Add to dashboard Cite

Connections in steel structures are of main importance in the design process. Cover-plate connections with circular or slotted holes are of common use in steel structures. The behavior of such connections is characterized by various failure modes. Bearing mode is one of the most complex due to the distribution of stresses and strain around the loaded holes. This paper deals with the characterization of the behavior of cover-plate bolted joints with slotted holes. Numerical simulations using the Cast3M FE package are performed in association with an original experimental method based on fullfield measurements. The strain fields obtained with these methods (grid method and Digital Image Correlation (DIC)) give the possibility to observe the evolution of the local deformations in the loaded zones of the connections. These evolutions are compared to those given by the Finite Element model for each configuration at different loadings stages. Thus, the experimental results are used to validate the numerical model. The comparisons are based on the load-displacement curves, the failure modes and the local deformations in the zones around the bolts holes. Six different geometrical parameters of the holes and their positions are considered in the tests. The validated numerical model is used to evaluate the influences of the dimensions of the slotted hole. The results showed that the geometry has a significant influence on the failure mode of the connections such as net section and bearing. Both numerical and experimental results are compared to those given by the analytical approach of Eurocode 3 to evaluate the accuracy of the analytical prediction. The study shows the influence of various geometrical parameters on the resistance and the failure mode of the connections with slotted holes. The strain fields obtained with the grid method and DIC are observed in the plate around the bolt in the elastic and plastic phases. They give the possibility to observe experimentally the evolution of strains in a wide area of the most stressed zones in order to develop more accurate analytical models based on experimental evolutions of strains.

show abstract

Strain measurements and analyses around the bolt holes of structural steel plate connections using full-field measurements

Cited by 13 publications

References 22 publications

Efficient training of physics‐informed neural networks via importance sampling

Efficient training of physics‐informed neural networks via importance sampling

Efficient training of physics-informed neural networks via importance sampling

01.13: Experimental study of cover‐plate bolted joints with large or slotted holes

Contact Info

Product

Resources

About