Modelling and predicting the deformed geometry of thick-walled pipes subjected to induction bending

Induction bending offers a rapid, cost-effective method of producing complex convoluted pipework. The resultant bends, however, typically show unwanted geometric deformations which include wall thinning at the extrados, wall thickening at the intrados, awkward transitions on going from tangent to bend, and wrinkling at the intrados surface. All forms of geometric deformation are worse depending on the tightness of the bend and the thickness of the original pipe. Predicting the final geometry of the bend is a non-trivial problem. This article first considers the use of simple analytical models to predict the final deformed geometry of induction bends in thick-walled pipe. It then goes on to compare the predicted geometry obtained from those analytical models with the outputs from empirically derived charts and computational models. The article concludes that empirically derived charts do not offer accurate predictions of wall thinning, are not currently available for intrados wall thickening, and cannot, with confidence, be extrapolated to produce accurate results for very tight bends. Numerical models are developed which predict wall thickening and thinning for any combination of outside diameter/wall thickness/bend radius combination; however, it is concluded that their limited accuracy makes their use questionable. Elastic-plastic FEA is developed which is shown to provide the most consistently accurate means of predicting both wall thickening and wall thinning, and, while not providing a comprehensive prediction of the post-bend geometry did, nevertheless, to go some way towards predicting the overall deformed geometry including transition ramps and intrados surface wrinkling.