As part of a substantial research programme related to nuclear plant component justification, Frazer-Nash was asked to simulate a typical multi-pass repair weld made to a full scale welded test specimen.  We demonstrated that the predicted stresses agreed well with the measured data, and were able to identify the key parameters that affected the residual stress distribution and magnitude.  These results, coupled with further validation, now underpin the use of the methodology in support of nuclear plant safety cases.

 
Repair Weld Cracking

A survey of weld repair technologies currently used by EPRI member utilities has found that 40% of all repairs to steam chests, piping and headers resulted in subsequent cracking. It further reports that over 70% of the repairs were performed without implementing post weld heat treatment.  It is reasonable to infer that high residual stresses associated with the repair process probably played an important role in many of these subsequent failures.

Frazer-Nash has developed a world leading capability in the accurate modelling of welding processes and prediction of residual stresses and component distortions.

Repair Weld Modelling

Repair welds are usually introduced into structures either to remedy initial fabrication defects found in castings or welds by routine inspection, or to rectify in-service degradation of components and thereby extend the life and economic operation of ageing engineering plant.

In this project, the principal objective was to validate the novel 3D finite element methods developed by Frazer-Nash to predict residual stresses in welded components.  The project simulated a multi-pass short repair weld in a stainless steel boiler component.

Methodology

Welding is a 3-dimensional time-dependent process.  The resulting distribution of residual stress depends on various factors; long range structural restraint; medium range temperature decay gradients and short range welding start and stop effects. 

A 3-dimensional moving arc weld simulation using a fine mesh is required to capture all three of the above effects. However, we have developed an alternative 3-dimensional procedure where the entire length of the weld bead is deposited at the same time. This approach has been demonstrated to accurately predict the global distribution of stresses for a finite length 'repair weld' and offers substantial savings in solution time for simulating complex multi-pass repairs.

Validation

The validity of the approach is supported by comparison of residual stress predictions with neutron diffraction and deep hole residual stress measurements.