Prediction of Residual Stress Random Fields in Selective Laser Melted Aluminum A357 Components Subjected to Laser Shock Peening
This work aims to develop a procedure to simulate laser shock peening treatments more efficiently, and to characterize the major differences in laser peening effects for cast and additively manufactured (selective-laser-melted) metallic specimens fabricated from A357 aluminum alloy. In addition, residual stresses (RS) are to be predicted probabilistically as a random field, allowing rigorous determination of RS values for a desired reliability. Laser shock peening (LSP) is a surface treatment technique that induces compressive RS near the surface of target metal components to improve fatigue life. Developing an LSP process using physical experiments is very expensive and time-consuming. To address this issue, finite element methods (FEM) have been widely used to simulate the LSP process and predict RS. Conventionally, almost all material constitutive models used in LSP prediction of RS involve deterministic parameters. Therefore, the predicted RS profiles do not reflect real-world variations in the material or uncertainties in the LSP process. Moreover, prediction of RS as a random field has not been done. While the effect of LSP on cast alloys has been studied extensively, few researchers have investigated the effects of LSP on metallic specimens produced by additive manufacturing processes such as selective laser melting (SLM). Therefore, the objectives of this research are: (1) Develop a procedure to simulate the LSP process with reduced computational time; (2) Conduct experimental and numerical studies to understand the effects of LSP on SLM A357 aluminum alloy; (3) Create a probabilistic approach to quantify the material constitutive model parameters as a joint probability distribution of correlated random variables; and (4) Demonstrate a technique to efficiently generate stochastic maps of the resulting RS random fields, enabling improved reliability analysis for desired RS values. To increase LSP simulation speed, a new systematic procedure is developed using modal analysis and generalized variable damping profiles with the “single explicit analysis using timedependent damping” (SEATD) FEM approach. To begin understanding the effects of LSP on A357 aluminum alloy specimens produced by SLM, true-stress-strain curves of both as-built (AB) and laser shock peened SLM samples are obtained through transverse tensile tests. An initial hypothesis on the effects of LSP during tension testing is formulated and subsequently tested using SEATD approach. To quantify the plasticity-Johnson-Cook (J-C) material model parameters as a joint probability distribution of correlated random variables for heat-treated (HT) and as-built (AB) SLM A357, the Bayesian inference (BI) probabilistic approach is utilized. Also proposed in this work are two BI-quantified-techniques called, respectively, the Multidimensional-BI method and the Spatial-Posterior-Prior-Probability-Mass-Function (SPP-PMF) method. Both can be used to efficiently predict RS as a random field, thus providing far greater insight into the practical ability to attain desired RS. For identical LSP treatments, it is determined that the material models are significantly different for the SLM and the conventional cast A357 aluminum alloys, resulting in much lower overall magnitude of compressive RS in the SLM-alloy. In addition, stochastic maps of the resulting random stress fields for LSP treatments on specific SLM A357 components are generated using the approach described herein.