Table 1 - uploaded by Yongxiang Hu
Content may be subject to copyright.
Source publication
Accurate numerical modeling of laser shock processing, a typical complex physical process, is very difficult because several input parameters in the model are uncertain in a range. And numerical simulation of this high dynamic process is very computational expensive. The Bayesian Gaussian process method dealing with multivariate output is introduce...
Context in source publication
Citations
... However, FEA has limited applicability due to its large computational load and sensitivity to boundary conditions [14,29]. ...
Shape accuracy is an important quality measure of finished parts built by additive manufacturing (AM) processes. Previous work has established a generic and prescriptive methodology to represent, predict and compensate in-plane (x – y plane) shape deviation of AM built products using a limited number of test cases. However, extension to the out-of-plane (z plane) shape deviation faces a major challenge due to intricate inter-layer interactions and error accumulation. One direct manifestation of such complication is that products of the same shape exhibit different deviation patterns when varying product sizes.
This work devises an economical experimental plan and a data analytical approach to model out-of-plane deviation for improving the understanding of inter-layer interactions using a small set of training shapes. The key strategy is to discover the transition of deviation patterns from a smaller shape with fewer layers to a bigger one with more layers. This transition is established through the effect equivalence principle which enables the model predicting a smaller shape to digitally “reproduce” the bigger shape by identifying the equivalent amount of design adjustment. Besides, a Bayesian approach is established to infer the deviation models capable of predicting deviation of complex shapes along the z direction. Furthermore, prediction and compensation of out-of-plane deviation for 2D freeform shapes are accomplished with experimental validation in stereolithography (SLA) process.
... From the past two decades, Fabbro's model is used for LSP simulations due to accurate prediction of pressure pulse and the same is used in the present study. The Gaussian profile obtained from peak power density calculation is used as the input for the pressure model to draw pressure pulse temporal profile [35]. In the LSP process the material experiences high strain rates of order 10 6 /s, in such conditions material mechanical properties derived at static conditions are no longer valid. ...
Laser Shock Peening (LSP) turned out to be the most efficient surface engineering process for advanced materials to induce beneficial deep compressive residual stress which helps in improving mechanical, fatigue properties and surface damage resistance. But, analyzing the nonuniform distribution of residual stresses in the treated sample with X-ray diffraction (XRD) is much time taking and a costly process. This problem can be resolved with LSP finite element numerical simulation model which is feasible with the realistic experimental process. The FE model allows the user to control the laser parameters in order to achieve the optimal level of all controllable parameters. This study is intended to analyze and optimize the influence of laser processing parameters that assists in inducing the residual compressive stress with minimal surface deformation. A Ti6Al4V material model with Johnson–Cook’s visco-elastic–plastic material behavior law is prepared for LSP simulation. Gaussian pressure profile is utilized for uniform loading of the targeted zone for the proposed model. Taguchi Grey Relational Analysis (TGRA) with L27 orthogonal array is applied to LSP simulation, and the results were analyzed with consideration of multiple response measures. It is noted that surface deformation is increased with the rise in a number of laser shots and pressure pulse duration. Maximum compressive residual stresses are falling for higher levels of laser spot diameter, laser spot overlap and laser power density. The correlation is observed between the FE simulation and the published results. The optimal set of process parameters are obtained for improving the LSP on Ti alloys.
... Simulation based on the first principles has been extensively studied to understand the causal relationship between physical parameters and product quality [9,10,11,12]. However, improving part accuracy based purely on such simulation approaches is far from being effective in practice [13,14]. ...
Geometric accuracy control is crucial to fulfill the promise of additive manufacturing (AM). The control of the out-of-plane deformation has been a challenge task for AM due to its complex underlying physics. We have been establishing a generic and prescriptive methodology to represent, predict and compensate 3D geometric deformation of AM built products based on a limited number of test cases. Built upon our previous study, this work aims at developing a prescriptive approach to better understand the out-of-plane deformation due to complex inter-layer interactions. Experimental investigation is conducted to validate the prescriptive model through a stereolithography process.
... Work by Hu et al. indeed highlights the importance of uncertainty in LP. Their work involved uncertainty using Gaussian processes for parameters of an LP pressure model in an axisymmetric single-spot LP simulation [27]. To study the model form uncertainty, based on measured deviations from the experimental data for LP residual stresses, Park and Grandhi applied the maximum likelihood method [28]. ...
A method is introduced for efficient reliability-based design of laser peening (LP) surface treatment to extend fatigue life of metal components. The method includes nonparametric probability density estimation, surrogate modeling using a new finite element (FE or FEA) approach, and reliability analysis with correlated random variables (RVs). Efficient LP simulation is achieved via a new technique termed single explicit analysis using time-dependent damping (SEATD), which reduces simulation times by a factor of 6. The example study of a three-point bend coupon reveals that fatigue life reliability significantly affects optimal LP design, as 52 laser spots are needed for 99% reliability versus 44 spots for 95%.
... Detailed mathematical models are developed to describe the underlying phenomena in continuous casting: a finite difference method based heat transfer model and a comprehensive model for center line segregation (CLS). To facilitate faster and exploration of wider search space, analogously to the work of Witherell et al. [18] and Hu et al. [19] metamodels have been used as an alternative to the physics-based process models. Data generated using the detailed models are used to create RSMs to predict output parameters as a function of the system variables, further details are presented in Ref. [20]. ...
To compete with other materials and/or contribute toward light-weighting of vehicles, newer grades of steel are continuously invented and experimented upon. Due to the costs and time involved in such developments, manufacture of new grades of steel at an industrial scale is difficult. We propose a method that is useful for steel manufacturers interested in producing a steel product mix with new grades of steels by predicting the required change in the operating set points of each unit operation in the manufacturing chain of products with the new grade of steel. Here, we demonstrate a method to determine the set points of one unit operation, continuous casting which is measured in terms of conflicting objectives including productivity, quality, and production costs. These parameters are sensitive to the operating set points of casting speed, superheat, mold oscillation frequency, and secondary cooling conditions. To ensure targeted performance and address the challenges of uncertainty and conflicting objectives, an integrated computational method based on surrogate models and the compromise decision support problem (cDSP) is presented. The method is used to explore the design space available for casting operations and determine operating set points to meet requirements imposed on the caster from subsequent downstream processes. This method is of value to the steel industry and enables the rapid and cost effective production of a product mix with a new grade of steel.
This study aims to model the effects of multiple laser peening (LP) on the mechanical properties of AA2024-T351 by including the material microstructure and residual stresses using the crystal plasticity finite element method (CPFEM). In this approach, the LP-induced compressive residual stress distribution is modeled through the insertion of the Eigenstrains as a function of depth, which is calibrated by the X-ray measured residual stresses. The simulated enhancement in the tensile properties after LP, caused by the formation of a near-surface work-hardened layer, fits the experimentally obtained tensile curves. The model calculated fatigue indicator parameters (FIPs) under the following cyclic loading application show a decrease in the near-surface driving forces for the crystal slip deformation after the insertion of the Eigenstrains. This leads to a higher high cycle fatigue (HCF) resistance and the possible transformation of sensitive locations for fatigue failure further to the depth after LP. Experimental observations on the enhancement in the HCF life, along with the relocation of fatigue crack nucleation sites further to the depth, reveal the improvement in the HCF properties due to the LP process and validate the numerical approach.
Laser peen forming (LPF) is a flexible forming process that brings many challenges for complex shaping. This study aims to develop an effective optimization method to complete efficient process planning of LPF to generate the desired geometry shape. The eigen-moment is proposed as a new intermediate variable to describe the bending deformation to relate the LPF process parameters and the geometry shape. The governing equation of eigen-moment is derived to characterize the bending deflection with a PDE equation. The numerical computation method is developed by the PDE weak form and finite element method to predict the bending shape. To achieve process planning of complex shaping, the distributed eigen-moment on both plate surfaces is prescribed, and a PDE-constrained optimization is utilized to define the process planning problem of LPF. Then, the eigen-moment field is optimized by adopting the interior-point algorithm to obtain the desired shape. The proposed eigen-moment is verified to be an intrinsic physical quantity to describe the bending behavior of LPF. A complex shape with saddle geometry is used as a typical case to demonstrate the process planning method. The experiments conducted with the planned LPF process parameters are validated to produce a shape consistent with the designed geometry.
Additive manufacturing (AM) is a promising direct manufacturing technology, and the geometric accuracy of AM built products is crucial to fulfill the promise of AM. Prediction and control of three-dimensional (3D) shape deformation, particularly out-of-plane geometric errors of AM built products, have been a challenging task. Although finiteelement modeling has been extensively applied to predict 3D deformation and distortion, improving part accuracy based purely on such simulation still needs significant methodology development. We have been establishing an alternative strategy that can be predictive and transparent to specific AM processes based on a limited number of test cases. Successful results have been accomplished in our previous work to control in-plane (x-y plane) shape deformation through offline compensation. In this study, we aim to establish an offline out-of-plane shape deformation control approach based on limited trials of test shapes. We adopt a novel spatial deformation formulation in which both in-plane and out-of-plane geometric errors are placed under a consistent mathematical framework to enable 3D accuracy control. Under this new formulation of 3D shape deformation, we develop a prediction and offline compensation method to reduce out-of-plane geometric errors. Experimental validation is successfully conducted to validate the developed 3D shape accuracy control approach.
It is known that estimating the wear level at a future time instant and obtaining an updated evaluation of the tool-life density is essential to keeping machined parts at the desired quality level, reducing material waste, increasing machine availability, and guaranteeing the safety requirements. In this regard, the present paper aims at showing that the tool-life model that Braglia and Castellano (Braglia and Castellano, 2014, "Diffusion Theory Applied to Tool-Life Stochastic Modeling Under a Progressive Wear Process," ASME J. Manuf. Sci. Eng., 136(3), p. 031010) developed can be successfully adopted to probabilistically predict the future tool wear and to update the tool-life density. Thanks to the peculiarities of a stochastic diffusion process, the approach presented allows deriving the density of the wear level at a future time instant, considering the information on the present tool wear. This makes it therefore possible updating the tool-life density given the information on the current state. The method proposed is then experimentally validated, where its capability to achieve a better exploitation of the tool useful life is also shown. The approach presented is based on a direct wear measurement. However, final considerations give cues for its application under an indirect wear estimate.
