Collapse prediction of pipe subjected to combined loads

Abstract

Prediction of tubular performance has over the last decades improved with models that are more accurate. There is a trend in the oil and gas industry where traditional uniaxial modelling is given less importance and the triaxial consideration is gaining ground. The American Petroleum Institute (API) added a formula to the standard to consider the effect of internal pressure on the collapse strength in the early 1980s. In 2015, API issued an addendum to the API Technical Report 5C3 (TR 5C3) where the triaxially based collapse strength method was incorporated. This was a more accurate method of incorporating the effect of internal pressure and axial stress on collapse strength. The validity of the formula was demonstrated by collapse strength tests with simultaneous internal pressure by an API work group – API WG 2370 (Greenip, 2016).

In 2007, API/ISO presented an ultimate strength (ULS) method for predicting collapse. The new calculation, referred to as the Klever & Tamano model, was developed by API/ISO Work Group 2b (WG2b) under the Steering Committee 5 (SC5) for tubular goods. Following 2986 collapse tests of quenched and tempered tubular specimens; the Klever & Tamano (K&T) model has since been presented as the most accurate ULS model for collapse prediction.

This paper compares the collapse prediction performed by the Klever & Tamano model with the 2015 API model using the triaxial collapse tests performed by Greenip for API. Comparison of the K&T (ULS) model and the traditional API (minimum performance) model requires some considerations to establish common ground before the results can be compared. The K&T model builds on a probabilistic estimation of the pipe properties while key components of the API prediction is empirical. The resulting collapse prediction for the entire batch is 3.11% lower than actual for K&T and 20.9% for API. Using two standard deviations, the collapse prediction of K&T is 14.7% lower than actual. Increasing to three standard deviations, the K&T model coincides with the API triaxial model from 2015 for the investigated pipe. No figures reported include any design factors. These results support that slimmer tubular designs can be made, exercising detailed control of safety margins to collapse. A generic example shows a reduction of $47,000USD per well for a typical 13 000 ft long well in an 8.6 lb/gal (1.03 sg) pressure gradient.