AWS A9.5 PDF
This guide introduces computational weld mechanics methodology through an overview of the current technology. It presents current practices for heat transfer, microstructure, residual stress, and distortion calculations. In addition, a framework for developing verification and validation (V&V) procedures for these models is presented through an example related to the prediction of thermo-mechanical conditions. This document establishes the foundation for future V&V operations to allow for other emerging computational weld mechanics tools.
Although this guide is not written with mandatory requirements, mandatory language, such as the use of “shall”, will be found in those portions of the document where failure to follow the instructions or procedures could produce inferior, misleading, or unsafe results.
Preface.
Computational models have been used routinely to great advantage for more than three decades. This technology has been used in many industries to analyze and assist in the design of many items. From architecture to telecommunications, computational analyses (structural to thermal to fluid) have been used to develop objects from the most complex to commonplace everyday items. Numerical analysis has given engineers the capability to make products better, safer, and more functional with less development costs. The growth in the use of computational models shows that commercial industries have confidence in the accuracy of the codes to reduce costs and delivery times while improving quality. In manufacturing, computational solid mechanics (CSM) and computational fluid mechanics (CFM) have been fully adopted; yet, the use of computational weld mechanics (CWM) has not. It has been suggested that the same level of confidence in CWM analyses does not exist due to relative newness of the tools and the lack of experience in their use. In comparison, CWM is quite complex involving a coupled phenomena of thermal and nonlinear, transient structural analyses. Information regarding material responses due to thermal inputs, microstructure evolution, and to stresses and strains are needed to perform this type of analysis. It is for these reasons that CWM has emerged about two decades later than CSM.
The process to develop confidence in computational modeling can be expedited by a process called verification and validation (V&V). Verification testing ensures that a computational code solves the mathematical state equations that describe the phenomenon with sufficient accuracy, robustness, and reliability. Validation tests that a particular computational model predicts a particular event with accuracy and reliability. Such V&V has been developed for computational solid mechanics [1]1 with Figure 1 illustrating a typical methodology used to develop V&V documents for creating models throughout the design process. This approach can be applied to CWM as well. In welding, the relevant phenomenon might be distortion, residual stress, microstructure, or risk of in-service failure. Once through this process, a computational model can be used repeatedly without physical experimentations with confidence that the output will be accurate and reliable.
Computational models fitted to experimental data before being used are called calibrated models. These types of models cannot be used to predict a particular outcome unless the input values are the same range as the original ‘calibrated’ model. Any changes in the model require the repetition of the calibration process resulting in multiple computational models and experimental tests. As one can easily see, computational approaches that utilize validated codes and verified approaches are much more expansive and usable in multiple cases than their calibrated computational model counterparts.
1 See Annex A for in-text citations.
