This website uses non-intrusive cookies to improve your user experience. You can visit our cookie privacy page for more information.

The incidents
Power Distribution Failure - 29th May 2000

Competent Authority investigation

As a result the HSE set up an investigation team to investigate the causes of the incident. BP independently carried out an investigation following their major incident investigation guidelines and cooperated fully with the HSE investigation. BP also contracted the services of an independent consultant in order to assist in their enquiry (Kirsner Consulting Engineering Inc. Atlanta, Georgia, USA). The BP investigation made a number of key proposals for corrective actions as a result of the incident investigation

The HSE investigation focused on the pressure systems aspects of the failure and associated management issues. The incident was not considered to be a defined COMAH major accident. The investigation did however consider the wider (COMAH related) view of steam as an essential utility service to the Complex, and system vulnerabilities and knock-on major accident related consequences, as a safety report issue.  These issues were also addressed by the BP report into the incident.

The HSE and BP investigation teams worked closely together, especially at the early stages, and also with the external consultant appointed by BP to assist in identifying the technical causes. The BP and HSE investigation teams compared their analysis of the causes and findings in order to check alignment and they were generally close.

The investigation immediately following the incident concentrated initially on the failed tee-piece and around the area close to the failure. Sections of damaged pipework were removed on 22nd June and sent to the Health and Safety Laboratories in Sheffield for detailed metallurgical analysis. The tests indicated that the pipework was in satisfactory condition and the cause of failure was extreme internal overpressure significantly in excess of the design pressure of the pipework.

The investigation determined that the pipework installation was around 25 to 30 years old and generally in very good condition.

Initial inspection of the section of damaged pipeline, which showed that the pipe had almost unwrapped, suggested that considerable energy was involved in the failure but the precise cause of failure was not immediately apparent.

Figure 12 - Steam Main Photo

Figure 13 - Rupture Photo

As far as could be determined there were no process or steam generation upsets just prior to the incident and no direct activity (e.g. opening or closing valves near the point of failure) at the time of the incident.

Pipework and Supports

Observation of the pipework revealed some pipe movement and damage mainly to cladding and supports and bulge distortion to the MP steam main in the West Gemec culvert. Valves in the supply system on the South Side were disassembled and found not to be providing full isolation but allowing some leakage.

Steam Traps

The two steam traps in the section of pipework, which ran through the culvert and which discharged directly to atmospheric drain, were tested and found to be functional but that they had been isolated. Only one of these traps was identified as having been recently isolated for the purposes of the culvert inspections following the flooding. Calculations subsequently showed that the capacity of this trap, which was situated at the lowest point of the line and closest to the point of failure, should have been sufficient to cope with condensate being produced in the line, had it not been isolated.

Mechanism of Failure

Due to the evidence of condensate being vented (30th May) and water hammer being identified in pipework close to the rupture point the most likely cause of failure was determined at an early stage to be water hammer. However there was an absence of steam flow in the dead leg which is a critical component of conventional water hammer (i.e. steam velocity was not present).

It is not unknown, but less well appreciated however that water hammer can also be caused by the rapid collapse, or imploding, of a steam bubble and the mechanism and theory of this are referred to as "condensation induced water hammer".

This mechanism occurs when steam comes into contact with relatively cold condensate or when condensate meets a trapped steam bubble. Due to temperature differences heat transfer between the two is rapid, causing a collapse of the steam (significant reduction in volume occurs as steam vapour changes to a liquid condensate). This is followed by an inrush of more condensate at high velocity to fill the void. The nature of this mechanism is such that it inherently involves much larger volumes of water because lines need to be flooded to isolate a steam bubble. Thus when water hammer occurs a large volume, and hence mass of water travelling at high speed is rapidly decelerated. The high energy dissipated as a result of the rapid deceleration results in a pressure wave travelling at high speed back through the water column.

Due to the isolation of the last remaining functional steam trap in the West Gemec culvert for inspection purposes (the other one was already isolated) this effectively removed the ability to release condensate from this section of the steam supply system. Therefore as the steam cooled and condensed the condensate built up whilst more steam was still available from the North Side supply main to replace the steam that had condensed. As a result of the U configuration of the pipework beneath the road when the condensate built up to the point where it flooded the horizontal section the steam present on the South Side was trapped between the condensate in the culvert and the isolation valves on the South Side. This effectively created a trapped steam bubble.

Figure 14 - Culvert Diagram

Figure 15 - Culvert Diagram

Since the North Side was connected to the steam supply MP steam fed into the dead leg to the culvert continued to cool and condense and thus the condensate level in the culvert U bend arrangement continued to rise up the vertical legs whilst still continuing to cool. Due to the fact that the isolation valves on the South Side were leaking, some steam was passed back through the valves and into the leg of the steam main south of the culvert maintaining the pressure and temperature of the steam bubble.

The geometry of the system was such that eventually the condensate level rose to a point where it ran underneath the steam bubble, creating a much larger surface area in contact with the steam. By this time the condensate had cooled significantly leading to rapid heat transfer and collapse of the steam bubble. The collapsing steam bubble created a differential pressure across the accumulated condensate causing it to accelerate into the void. The mass of condensate involved was in the region of 4 tonnes and this was stopped abruptly when it reached the isolation valves. The resulting pressure wave is believed to have caused the failure of the tee-piece.

Figure 16 - Culvert Diagram

The investigation team concluded that the most likely reason for the pipe line failure was internal overpressure caused by "condensation induced water hammer".

2010-12-13