Welding problems encountered in the nuclear industry have been mainly addressed by weldability tests analysis, development of new techniques or improvements through lessons learnt. For the last decade, AREVA has developed a complementary approach based on numerical simulation. Residual stresses inherent in the manufacturing of reactor components do not constitute a major problem at the design stage; even though they may have a strong impact on some types of damage. Numerical welding simulation in the nuclear industry has mainly focused on residual stress prediction. Distortion may also be a source of concern in component design: some structures are slender in spite of their thickness. In the frame of GEN IV project like sodium fast reactor or ITER, the distortion problem has gained in importance. For industrial project considering large structure, DCNS has developed electron beam welding process and built several mock-ups to validate the manufacturing steps. AREVA NP carried out welding simulations on one of these mock-ups (1m × 1m) made of austenitic steel plates and representative of nuclear components. The aim of the numerical simulations was to check the quality of the distortion prediction. Electron beam (EB) welding offers significant advantages for structures with many long welds and high requirements on the final shape of the welded structure. Since distortion values are very low (between 0.1 mm and 1.6 mm), prediction of distortions induced by electron beam welding is difficult. Compared to the size of the mock-up (1 m), distortions represent only 0.1%. Distortions induced by each weld have been computed using a simplified approach (local/global method [1]), which aims at modelling long and numerous welding operations with an acceptable calculation time. After each welding sequence, DCNS measured distortions at some representative points of the mock-up. This paper comments on the difficulties of the validation of the numerical predictions by comparison with measurements: • Measurements are made after each welding step, requiring removing the parts from the chamber. • Distortion values are very low compared to the size of the mock-up that may increase the risk of measurement error. In spite of these problems, the main trends of the experimental deformed shape are well represented by the simulation.

This content is only available via PDF.
You do not currently have access to this content.