A.1 Introduction

Credible events should be identified that reflect the range of loss of containment, so as to evaluate issues that might occur. A loss of containment incident on an offshore pipeline or carbon dioxide processing plant may result in the release of a gas cloud or an uncontrollable release of energy. Injury can be caused to the human body by the inhalation of the gas, low temperatures in the vicinity of the release or the effects of the physical blast. Damage to adjacent equipment, structures such as offshore rigs, and the environment should also be considered where appropriate, especially if collateral damage produces further knock-on results such as fires or explosions, release of stored material, or missiles (blast fragments). Failures may range from small, transient leaks through to large scale vessel or pipe ruptures, and the potential consequences may vary from inconsequential or reversible health effects through to fatal or serious injury. There may also be major financial repercussions.

Credible event selection:

Suitable hazard identification techniques and information of known incidents should be used to identify credible loss of containment events. These may then need to be developed for further detailed analysis.

In order to determine the risks from an accidental loss of containment of carbon dioxide being transported by pipeline or processed (e.g. for enhanced oil recovery), various failure cases for the processes involved will need to be considered. Release scenarios may then be determined and dispersion modelling carried out to evaluate the consequences of such events taking place.

Classical hazard analysis of chemical processes would attempt to cover the whole range of hazardous events to obtain a model for the total risk from the process being considered, whereas, studies to determine the worst case events need only model the consequences of the perceived largest failure cases. Either way, the use of appropriate commercial computer codes to model release cases and obtain dispersion results is undeniably the best way of carrying out the consequence calculations. If the total process risk is to be considered, the data to be processed to obtain a result will generally be so large that only computer-aided mathematical calculations will be practical. A number of commercial dispersion modelling programmes combine the risk calculations with the consequence modelling in the same software package for the addition of appropriate frequency, probability and physical data.

For worst case events (including, where required, estimation of frequency and maximum potential numbers of fatalities), manual or spreadsheet calculation from the consequence modelling is a practical alternative.

The risk here, in a general sense, is that failure to analyse the potentially large number of smaller events may result in overlooking some hazards. This might be with reference to a single person or group of individuals (such as operators or itinerant workers) much closer to the process or offshore pipeline transportation system.

Depending on the risk evaluation required, the credible events may need to cover the whole range of possible cases from relatively small continuous releases (representing undetected or irremediable leaks from the processes) through to line ruptures or catastrophic failure of vessels where large but finite inventories of hazardous material could be released.

An intermediate case that could be of interest because of the particular physical properties of carbon dioxide is a 'running crack', where an initial small failure in the pipe steadily propagates into an extensive crack running along the length of a section of line. An ultimate scenario from this type of failure might be a release which is equivalent in total flow-rate to a line break. However, it may not necessarily have the impact of the corresponding full bore pipeline rupture. Alternatively, a running crack may result in the equivalent of a full bore line rupture with CO2 flowing from both sides of the resulting rupture; i.e. upstream and downstream.

Surface effects and impingement:

Particular scenarios may need to be modelled due to project-specific characteristics. For example, where projects propose the CO2 being brought above sea level, such as on to an oil rig to facilitate enhanced oil recovery, the structure of the rig, its platforms and indoor modules would affect the dispersion of the cloud, and additional modelling may be required to understand fully the dispersion of the cloud.

In many cases, further modelling techniques such as CFD modelling should be used to evaluate the concentrations in the gaseous cloud. These well-known modelling procedures are not covered in this publication. In particular, attention should be drawn to any possible impingement near the source of the release (i.e. near the source term), which may reduce the cloud momentum and hence air entrainment into the cloud, which will increase the resultant CO2 concentration in the resultant cloud. Where any impingement occurs (for example on offshore rigs with compact layouts), additional analysis should be applied to ensure that concentrations and gas flows are modelled adequately.

Cooling structural steelwork to a temperature lower than that for which it has been designed can lead to failure if it is cooled below its fracture appearance transition temperature (FATT). This is the temperature at which 50 % of the fracture surface of a test specimen becomes brittle (the other 50 % being ductile). Thereafter the properties of the metal are dominated by brittle properties. Thus, if there is an impact load applied to the metal, instead of deforming in a plastic manner, it will fail in a brittle manner.

On an offshore platform this could be serious, as there would be reduced protection of collision from nearby vessels and waves. Impact loads applied to structural members caused by wave action are commonplace, and in harsher environments, similar to the North Sea, platforms have to be able to withstand extreme wave conditions.

Physical blast:

A rupture may result in a physical blast close to the site caused by the expansion ratio of the liquid to gas. The effect of the physical blast can be calculated using a TNT equivalent88 or similar models. The discussion of this reaction known as BLEVE is discussed in 2.3.3.

88 'TNT equivalent' is a method of quantifying the energy released in explosions. The ton (or tonne) of TNT is a unit of energy released in the detonation of one ton of TNT, approximately equal to 4,184 GJ.