Page 32 - Failure Analysis Case Studies II

Failure Analysis Case Studies II
D.R.H. Jones (Editor)
0 2001 Elsevier Science Ltd. All rights reserved 19

BREAKUP OF THE FIREWALL BETWEEN THE B AND C
MODULES OF THE PIPER ALPHA PLATFORM-I.
ANALYSIS BY HAND CALCULATION

A. C. PALMER
Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 IPZ, U.K.
(Receiued 2 July 1997)
Abstract-In the Piper Alpha disaster, the initial explosion in module C destroyed the firewall between modules
C and 9, and fragments caused secondary damage which lead to a fire. This paper describes the dynamic
analysis of the explosion response of the firewall which was not intended to resist blast and was relatively
lightly constructed. The analysis is based on hand calculation of idealised elastic and elastic-plastic models.
(r7 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Disasters, mechanical connections, offshore failures, overload, structural failures.

1. INTRODUCTION
An explosion occurred on the Piper Alpha production platform in the North Sea on 6 July 1988.
In the subsequent fire 167 lives were lost, and the platform was destroyed. The incident was the
subject of a major public inquiry and extensive investigations and litigation. The inquiry concluded
that the initial explosion occurred in module C, and resulted from a flammable cloud created from
a hydrocarbon release, at a blind flange fitted where a pressure safety valve had been removed for
maintenance.
Observers saw a fireball emanate from module B at the west face of the platform some 10 s after
the initial explosion. The appearance of the fireball was consistent with a condensate release. The
inquiry concluded that the B/C firewall, between the C module and the B module, had been at least
partially destroyed by the explosion, and that fragments from the firewall had been projected into
module B and had ruptured a 4 in. pipeline carrying condensate.
The accident had unusual features. No wreckage could be examined, because most of the platform
collapsed into the sea during the fire, and only the two accommodation modules were recovered. It
was decided that any attempt to recover additional wreckage would not have a forensic value that
would justify the risks and costs that would be incurred, since it would no longer be practicable to
distinguish damage that had occurred during the original explosion from damage during the fire,
collapse and subsequent recovery. On the other hand, some revealing photographs were taken less
than half a minute after the explosion.
This paper describes additional research into possible explosion damage to the firewall, carried
out as part of subsequent legal proceedings. The analysis is complicated by the fact that the firewall
was not intended to resist pressure, and that it contains elements of widely different strengths, with
a limited capability to redistribute loads.
Complex structures of this kind are nowadays almost invariably analysed by finite-element
methods. However, for lcgal reasons that regrettably cannot be examined here, there were two
independent analyses, one by hand calculations alone and the other by the finite-element method.
The structure is too complicated for analysis by hand calculation to give a complete picture, but a
hand calculation does give useful results which will answer the key questions that are important to
an understanding of what happened. In particular, it can tell us:
1. whether or not the wall would break up;
2. when it would break up, in relation to the timing of the explosion;
3. if it broke up, how fast the fragments would be moving.

Reprinted from Engineering Failure Analysis 5 (l), 57-67 (1998)