Technical note | Refinement of Out of Circularity and Thickness Measurements of a Cylinder for Finite Element Analysis

Abstract

The as-built out-of-circularity (OOC) deviations from a perfectly round shape and plate thickness variations of a cylinder that is subjected to external pressure are known to have a significant effect on the collapse strength and must be accurately represented in finite element analysis to obtain accurate results. Often it is necessary to interpolate measurements from a relatively coarse grid to a refined finite element model and methods that have wide general acceptance are Fourier series and bi-linear interpolation. These methods have been implemented in an Excel spreadsheet to produce node and element data that can be used directly in the Strand7 finite element package, but the methods employed are suitable for other software as well.

Executive Summary

The as-built out-of-circularity (OOC) deviations from a perfectly round shape and plate thickness variations of a cylinder that is subjected to external pressure are known to have a significant effect on the collapse strength. Non-linear finite element analysis (NL FEA) may be used to predict the collapse strength but the true shape and plate thickness of the cylinder must be modelled before it can be expected to obtain accurate results.

Often the OOC and plate thickness are measured only at discrete grid points on the cylinder surface so suitable methods of interpolation between these measured points are needed for model refinement so that modelling accuracy is maintained. Proven methods that have wide general acceptance are Fourier series and bi-linear interpolation and these methods have been implemented via an Excel spreadsheet to produce node and element data that can be used directly in the Strand7 finite element package through the inbuilt ‘Online Editor’ facility.

The methods employed are also suitable for other finite element software and a test case is used to demonstrate that the spreadsheet implementation provides accurate interpolation between measured data points.