Laser peening (LP) is a surface enhancement technique that has been applied to improve fatigue and corrosion properties of metals. The ability to use a high energy laser pulse to generate shock waves, inducing a compressive residual stress field in metallic materials, has applications in multiple fields such as turbomachinery, airframe structures, and medical appliances. In the past, researchers have investigated the effects of LP parameters experimentally and performed a limited number of simulations on simple geometries. However, monitoring the dynamic, intricate relationships of peened materials experimentally is time consuming, expensive, and challenging.
With increasing applications of LP on complex geometries, these limited experimental and simulation capabilities are not sufficient for an effective LP process design. Due to high speed, dynamic process parameters, it is difficult to achieve a consistent residual stress field in each treatment and constrain detrimental effects. With increased computer speed as well as increased sophistication in non-linear finite element analysis software, it is now possible to develop simulations that can consider several LP parameters.
In this research, a finite element simulation capability of the LP process is developed. These simulations are validated with the available experimental results. Based on the validated model, simplifications to complex models are developed. These models include quarter symmetric 3D model, a cylindrical coupon, a parametric plate, and a bending coupon model. The developed models can perform simulations incorporating the LP process parameters, such as pressure pulse properties, spot properties, number of shots, locations, sequences, overlapping configurations, and complex geometries. These models are employed in parametric investigations and residual stress profile optimization at single and multiple locations.
In parametric investigations, quarter symmetric 3D model is used to investigate temporal variations of pressure pulse, pressure magnitude, and shot shape and size. The LP optimization problem is divided into two parts: single and multiple locations peening optimization. The single-location peening optimization problems have mixed design variables and multiple optimal solutions. In the optimization literature, many researchers have solved problems involving mixed variables or multiple optima, but it is difficult to find multiple solutions for mixed-variable problems. A mixed-variable Niche Particle Swarm Optimization (MNPSO) is proposed that incorporates a mixed-variable handling technique and a niching technique to solve the problem.
Designing an optimal residual stress profile for multiple-location peening is a challenging task due to the computational cost and the nonlinear behavior of LP. A Progressive Multifidelity Optimization Strategy (PMOS) is proposed to solve the problem. The three-stage PMOS, combines low- and high- fidelity simulations and respective surrogate models and a mixed-variable handling strategy. This strategy employs comparatively low computational-intensity models in the first two stages to locate the design space that may contain the optimal solution. The third stage employs high fidelity simulation and surrogate models to determine the optimal solution. The overall objective of this research is to employ finite element simulations and effective optimization techniques to achieve optimal residual stress fields.