BP – pipe support loads
Alongside project delivery I am currently working as part of the mechanical discipline engineering team. The work I am doing for them is answering Design Technical Queries (DTQ) and Engineering Queries. I started trying to explain both but it became too wordy, the DTQ is explained below:
A DTQ was submitted in relation to the pipe support loads experienced during a blast event along new production flowlines that are due to be installed (some have actually already been installed). The Clair platform has legacy issues relating to blast whereby the original platform design was never designed for blast. As such, all current in-use flowlines and pre-invested (installed but not hooked up) are supported using standard U-bolt pipe supports. The U-bolts are not designed to be able to withstand a blast event since the primary load path for the bolts is vertical. The U-bolt manufacturer only specifies a max vertical load therefore a max lateral load must be assumed.
Wood Group are the contractor designing the new flowlines (to be tied into pre-invested) and are concerned that in the event of a blast a number of the pipe supports will fail due to excessive lateral loads. Their basis for this statement is a rule of thumb that the max lateral load is a nominal 30% of the max vertical load. The manufacturer stated max vertical load is based upon the yield stress of the material. They have asked if they are to replace all the pipe supports since they all fail.
Recognising there are platform wide issues relating to blast, Fraser-Nash Consultants (F-N) were contracted to conduct targeted blast analysis on a complete flowline (one designed IAW original platform design). By modelling the U-bolt failure load using a similar method as Wood Group, F-N came to the same conclusion that the lateral loads are excessive. In order to understand the actual post-yield material characteristics (ie strain hardening etc.), a nonlinear analysis was conducted based upon the U-bolt geometry and material. From this they were able to establish a plastic collapse load. When modelled in this manner all the U-bolts were found to remain within the plastic collapse load. The picture I’ve attached kind of explains this problem through the stress-strain graph.
My response to the DTQ was initially to accept the F-N analysis since the U-bolts do not collapse fully, maintaining some structural integrity. My justification for this was that the design event is blast, a one-off whereby the performance criteria is to maintain primary containment. No-one seems content with this response and I am now stuck trying to provide more justification, any ideas?
Holdfast Training Services Trial Blog 
Could you combine your justification with a cost based analysis of replacing them? Or suggest a replacement as part of current maintenance programme that replaces the existing bolts with essentially a larger cross sectional area bolt (energy absorption is proportionally related to the volume of the bolt section and the tensile strength and ultimate strain of the bolt material) and suggest continued operation of the current bolt under an operational risk assessment? Not sure if that is helpful.
The design solution to replace would be fairly simple. It would require around 500+ supports to be replaced. Unsure as to whether it would require wells to be shut in, I’d probably suggest it does.
The supports are suitable for operational loading. The mechanism in which they ‘fail’ needs to be understood since designing for blast does allow plastic deformation to occur. The analysis suggests the supports are suitable to withstand a blast. The different opinions come from how it has been assessed. Which one is correct?
Hi Chris,
Great topic area! Force Protection Engineering being a particular interest I always feel that the Corps concentrates on structural response at the expense of M&E so it’s fantastic to see some services being designed for blast effects. The challenge you have is that blast pressure has a duration measured in milliseconds and so does not usually deliver a static load i.e. you cannot move from peak pressure straight to force applied and say collapse or not. Strain hardening will not develop in the timeframe in question but neither will the full force by transferred to the bracket. The question becomes one of either quasi static or dynamic response and is therefore better dealt with in terms of energy rather than force. Happily energy is an area that PET (E&M) are usually more comfortable in than the PET(C)! The most important input into blast response is mass of the various components and then capacity of the fixing to absorb that without reaching failure.
I am intrigued as to the nature of the transfer of loads into the pipe system. Are we talking about a blast at one location loading the pipe as it interacts with the blast wave or is this about a detonation in the pipeline itself and shock travelling through the pipework?
Thanks for your comment Rich.
The blast loads are derived from gas cloud accumulation studies based upon a number of worst case events. In this instance, because of location, it would be a gas escape in the wellbay area producing a gas cloud of 9000m3. If this were to ignite the blast wave would exhibit a pressure of 0.27bar E-W and 0.18bar N-S onto the pipework (this is based on the blast wind rather than the shockwave due to the relatively thin CSA of the pipes). Since the exact location of the ignition point is unknown the most common solution is to model the pressure across the whole length of the pipework. In the F-N study the blast load was modelled in a dynamic manner, ie it was ramped up over 30ms and then back down. The support loads were then extracted from this model. In addition to this a submodel of the U-bolts was used to calculated the stiffness and plastic collapse load. The stiffness was used to rerun the analysis with nonlinear springs modelled at the pipe supports. The resultant pipe support loads were then rechecked against the calculated plastic collapse load to assess whether catastrophic failure has occurred.
I am currently looking to assess whether the work done by F-N is applicable to all flowlines and whether or not it satisfies the requirements set out in API 579 ‘Fitness for service assessments’ (I’ve just found out about this document in the last 15 mins!!!).
I hope this answers some of your queries.
Ok thanks Chris.
If you’re using the right words when you say ignite, I think we’re talking deflagration rather than detonation? This would indeed induce a much lower but longer lasting pressure wave and therefore move towards static/quasi-static loading but it is practically impossible to determine the pressure, temperature and duration of a burning event rather than a detonation. If the rate of burn is sufficient to achieve detonation then you start to have equations that can be applied. Presumably ‘the wellbay area’ is an enclosed space so there is significant reflection and potentially Mach reflection. Do you have access to the calculations?
In terms of application to the pie this is only useful if it is perpendicular to the direction of travel of the pressure wave, otherwise there is time differential between lengths and there will be load sharing. The geometry of this would be interesting.
If you have more info I might come back for it because I’ve not worked through a gas deflagration/detonation like this before and it would be a good piece of CPD!
I just had a quick look at the summary report which was written during the initial design stage of the platform. There was no mention as to deflagration vs detonation but due to the magnitudes of the loads involved it looks like deflagration. The initial philosophy was to use the geometric model of the platform within a piece of Flame Acceleration Simulator (FLACS) software. The variables which were varied included the addition of external wind (both speed and direction), the ignition location (to provide the largest overpressure), size of gas release rate (based on a realistic assumption of hole size and internal pressures). In each geographical location around 20+ simulations were run to find the worst case event, this was judged against the loads experience on the bounding surfaces of the gas volume (in particular the designed blast walls and decks). The actual calculations are contained within the software.
The wellbay is an enclosed area, however the platform has been designed to allow for natural ventilation to reduce the effect of gas accumulation.
My understanding of blast direction wrt to pipe orientation is that the blast loading is always applied normal to pipe principle axis. This is due to the unknown of ignition locations and subsequent blast propagation direction.
The blast load studies were conducted sometime ago. The results were then used to input in the platform Safety Case (a regulatory document). All elements the platform should be designed to these levels. This is the basis upon the recent U-bolt issue and their apparent inability to withstand a blast event.
Interesting read and discussions, thanks Chris!
An interesting topic, thanks Chris