Finlay, J P (2004) Numerical methods for the stress analysis of pipe-work junctions. Doctoral thesis, Liverpool John Moores University.

Preview

Text 403282.pdf - Published Version Download (41MB) | Preview

Abstract

Pipe junctions arc a regular feature of piping and pressure vessel systems and are often the subject of multiple loads. acting simultaneously and at irregular intervals. Due to the nature and complexity of the loading. the subject has received a significant amount of study from designers and stress analysts to resolve some of the difficulties in stressing pressure structures. An extensive finite element (FE) analysis was carried out on 92 reinforced buttwelded pipe junctions manufactured by the collaborating company. Spromak Ltd. After comparing the resulting effective stress factor (ESF) data with ESFs for un-reinforced fahricated tee (UFT) it was concluded that, for the majority of loads, reinforced branch outlets appear better able to contain stresses than their un-reinforced counterparts. The linear FE study was followed by the inelastic analysis of three reinforced branch junctions. The purpose of the research was to investigate the potential use of such analysis as a tool for estimating the bursting pressure of pipe junctions and satisfying customer requirement for proof of a products performance under internal pressure. Results obtained showed that small displacement analysis is unsuitable for estimating the bursting pressure of a pipe junction, whilst the large displacement results were similar to those obtained using a hand-calculation. Ultimately, the study concluded that inelastic analysis was too expensive, offering little by way of insight into the problem than could be found by using classical stress analysis techniques. Following on from the study of reinforced branch outlets, this thesis described work undertaken with British Energy Ltd. to extend their current capability of stress prediction in UFT junctions using a FE based neural network approach. Upon completion of training new neural networks, the PIPET program was tested against new, previously unseen, FE data generated for this study with good results. The program was further evaluated by comparing the output from PIPET with FE data obtained from reviewed literature. For the pressure load case, a significant proportion of the data obtained from said literature was within the PIPET predicted stress ranges. with the new version of PIPET tending to calculate slightly lower stresses than the original program. However, whilst the pressure load case comparisons proved useful, the branch bending cases showed less concordance with PIPET's predicted stress ranges.

Item Type:	Thesis (Doctoral)
Subjects:	T Technology > TA Engineering (General). Civil engineering (General)
Divisions:	Engineering
Date Deposited:	02 Mar 2017 12:30
Last Modified:	03 Sep 2021 23:29
DOI or ID number:	10.24377/LJMU.t.00005632
URI:	https://researchonline.ljmu.ac.uk/id/eprint/5632

View Item