Nowadays many extremely unpleasant gases are routinely stored in very large quantities, usually in liquid form, by refrigeration and/or under pressure. For many years there has been considerable interest in understanding how such fluids behave if they escape. In this way hazards can be assessed, and risks calculated and mitigated by appropriate planning. Mathematical modelling of the fluid flow is usually separated into three phases: source terms, gas cloud dispersion, and subsequent impact of the cloud. The clouds under consideration are usually heavy, either intrinsically (e.g. in the case of chlorine), or because they are cold, or because the cloud is laden with atomized liquid droplets, or for a combination of some or all of these reasons. Unfortunately, heavy clouds tend to be the most dangerous, as they stay near the ground (in close proximity to people, plant and ignition sources) and the strong stable density gradient suppresses mixing with the air above. The state of the cloud in its early stages is governed by the mode of release, the 'source term', the theory of which is reviewed here. Various accidents to various storage systems can cause markedly different release conditions, and research has been focussed on a number of representative cases. Liquid pool evaporation and boiling is one possible source. This may be the likely outcome of a severe accident to storage of a refrigerated liquid (such as LNG or refrigerated ammonia). Cryogens boil violently, mainly by using the heat supplied from the ground onto which they fall, and form a cloud approximately at their boiling point. Less volatile liquids (which may still be very toxic) vaporize at a more sedate rate dependent upon the state of the atmosphere. A small breach in a pressurized container is likely to result in a two-phase flashing jet. In this case some of the liquid boils violently, extracting its heat of vaporization from the jet itself. The result is again a cold dense cloud containing small liquid droplets, but owing to the high speed of these jets this may form a significant distance away from the vessel. Total catastrophic failure of a vessel is considered to be relatively improbable. On the other hand, the extremely dire effects expected from such a failure mean that one has to consider the possibility, which may not be entirely remote if, for example, the vessel is weakened by flames from an accident to a neighbouring container. Significant work is currently in progress to understand this phenomenon better, and so the review here will be brief. In some cases the escaping substance may catch fire before it disperses. Source term models may also address this problem, but it is outside the scope of this review. Also of relevance to the analysis o, risks is the question of how a failure in containment comes about. It may be an accidental fracture, fatigue, or a pressure relief, e.g. caused by mixing the wrong substances accidentally in a vessel. These topics are also important, but are outside the scope of this review.
