Reduced-Order Modelling of Residual Stresses and Distortions in Welded and Additively Manufactured Components

Abstract

Welding, particularly in its modern forms such as fusion-based additive manufacturing (AM), has been viewed as a key enabler for advancing Construction 4.0 to support environmental sustainability and national security. A major challenge in the design and manufacture of welded structures is how to effectively estimate residual stresses and distortions that have been shown to impact not only manufacturability but also the structure’s fitness-for-service (FFS). Excessive distortions can lead to time-consuming and costly rework in downstream modular assembly and introduce secondary bending stress that can have detrimental effects on fatigue performance. These issues become even more pronounced as lightweight designs are increasingly adopted. Due to the complex multi-physics and multi-scale nature of residual stress generation in welding and AM processes, particularly in structural contexts, direct numerical modeling for engineering applications still remains impractical even with today’s computational power, in addition to the fact that some aspects of high-temperature material behaviors under rapid melting, heating, and cooling conditions are still not fully understood. Available experimental residual stress measurement methods are limited in applicability, particularly for actual components, and are subject to interpretation, resulting in significant variability as documented in the literature. A critical assessment of the state-of-the-art research on residual stress generation mechanisms suggests that a reduced-order modeling procedure should be considered. As such, a physics-based analytical model can be introduced for capturing the lower-order parameters governing residual stress development, while the finite element method can be used to extend the analytical model to complex structural configurations, e.g., additively manufactured components. Along this line, this study demonstrates that a characteristic residual stress distribution for FFS purposes can be captured using an analytical scheme based on two key parameters: (1) the plastic zone (its size and shape), and (2) the limit elastic strain state within the plastic zone. To further improve its applicability to lightweight thin section structures, the analytical method has been further extended by introducing a novel heat source modeling method. The closed-form plastic zone estimation method coupled with a novel analytical thermomechanical model has been shown to provide an effective estimation of full field residual stress distributions for various engineering components, e.g., pressure vessels and piping components. Additionally, a physically consistent two-dimensional finite element modeling procedure is developed to estimate residual stress under three-dimensional welding conditions, which incorporates an analytical heat flow solution and a sub-domain-based model. The framework's effectiveness is validated by time-dependent and sequentially coupled thermomechanical simulations and well-documented experimental data. The presented reduced-order modeling framework provides a physically sound and computationally efficient solution for consistently estimating residual stresses and distortions in welded and AM components. A key contribution of this work is that the resulting residual stress solutions have been selected for adoption by the international FFS code, i.e., API 579 RP-1/ASME FFS-1 Annex 9D for prescribing residual stress distributions for performing fracture mechanics-based FFS assessment. The framework also includes a robust distortion estimation procedure that enables efficient modular design and assembly of advanced lightweight structures. In addition, the reduced-order modeling procedure is shown to be highly effective for characterizing residual stress and distortion development in AM components, which is an important element in achieving rapid qualification and certification for deploying metal AM components in safety-critical applications.