Predicting weld hardness

A comparative study of the methods of predicting the heat-affected-zone (HAZ) hardness is presented in this thesis. There are some five methods suggested by researchers for this purpose --- the Jominy-end-quench method proposed by Doan and Stout [1], a weld-test method suggested by these same authors [2], a mathematical model proposed by Bastien et al [3,4,5], and two similar mathematical models contributed by Arata et al [6].A "Welding Hardness Diagram " is proposed for the Jominy and weld-test methods to simplify the task of predicting hardnesses. Computer programs have been generated for the mathematical models to eliminate the necessity of time consuming calculator computations.Hardnesses predicted by all five methods are compared to measured values from the author's experiment, experimental values from the work of Graville [7] and experimental values from the work of Doan and Stout [1]. It is found that Vickers hardnesses measured with a 500 g load are on average 45 points (VPN) higher than those measured with a 10 kg load. The weld-itest method and the Arata α model provide the most accurate predictions where the load is 10 kg, while the Arata β and Bastien methods show less accuracy in predicting 10kg load hardness results. While the Jominy - end-quench method results (Vickers — 10kg) are seemingly acceptable, basic experimental difficulties and uncertainties suggest that this technique has limited applicability.A discussion of the relative merits of the prediction methods is included. Because the predicted hardnesses based on the Bastien method differed considerably from the experimental results (Vickers --- 10 kg), a sensitivity  analysis was carried out to help rationalize the differences.