Source Terms

Background

This section is concerned with characterisation of the initial outflow of fluid following a loss of containment (discharge rate and exit condition), as well as the spread and evaporation of liquid pools and the near-source dispersion of jet releases.

Proper characterisation of the source term is an important element of hazard assessment. However it can be an onerous task due to the complex nature of releases which occur on offshore installations, ie high pressure, multi-phase, multi-component releases with depressurisation occurring through complex networks of pipework and vessels with the added complication of blowdown and/or isolation.

Hydrocarbon release statistics (OTO 99 079 and Pratt, 2001) show that gas leaks are the most common type of release (56% of all releases) with oil releases (16%) and non-process releases (11%) also significant. Pipework is the most significant leak source (61% of all leaks).

Current position

Discharge from vessels, pipelines and equipment

The prediction of single and two-phase release rates was reviewed in Phase I of the Joint Industry Project (JIP) on Blast and Fire Engineering for Topsides Structures (BFETS), work package G1(c). This study concluded that, whilst the idealised cases of single phase and single component, two-phase releases had been studied in detail, there was little information on the behaviour of the type of hydrocarbon mixtures likely to be found on offshore structures. The report (OTI 92 587 [PDF 2mb]) suggested that a modest experimental programme of releases involving representative hydrocarbon mixtures would be beneficial in understanding the flow regime for such releases and establishing a dataset upon which to further develop existing modelling techniques. The experimental programme was not been undertaken in subsequent phases of the JIP, therefore this uncertainty remains. Catastrophic releases (eg due to rupture of a vessel) are also a significant area of uncertainty, although are not a primary concern for offshore installations.

Near-source dispersion of high-momentum jet releases

Gas jet releases

Air entrainment into gas jets has been extensively studied and various models exist to predict the dispersion of unobstructed jets in the atmosphere (see below). Experimental work has been undertaken in the field of impinging releases but little of direct relevance to offshore installations (complex arrays of obstacles). The work of most relevance is that recently undertaken as part of the JIP on 'Gas Build-up from High Pressure Natural Gas Releases in Naturally-Ventilated Offshore Modules' (Cleaver et al, 1998, 1999), which included both free and impinging gas releases. Phase I of the BFETS JIP identified a lack of data for the under-expanded region of high pressure jet releases. This does not appear to have been addressed in subsequent phases of the JIP and therefore remains an uncertainty.

Flashing liquid releases

Various experimental programmes have been conducted since Phase I of the BFETS JIP to better characterise the release and near-field dispersion of flashing liquid jets and provide data for model development and validation, eg Barton and Moodie (1992), Allen (1996, 1998, 2000) and AIChE (1999). These experiments have included impinging releases, which represent the area of greatest outstanding uncertainty (concerning rain-out, ice formation and re-evaporation). Further experimental and modelling work is planned and in progress in this area.

Liquid pool spread and evaporation

Recent HSE-funded experimental work (Cleaver et al, 2001) has been undertaken to resolve uncertainties in the spreading behaviour of liquid spills over horizontal surfaces and to investigate pool spread within bunded areas of different shapes, including the potential for bund overtopping.

Investigation of the characteristics of spreading of unconfined oil pool fires on steel decks is underway.

[back to top]

Modelling capabilities

Discharge from vessels, pipelines and equipment

A range of simple models are available to predict discharge rates from single vessels or pipework, covering single phase flow (gas or liquid), two-phase flow and pumped releases, as documented in references such as CMPT (1999) and Lees (1996). Improvements have been made, in particular, to the models and guidance now available for two-phase flow prediction.

Most of the above models treat multi-component mixtures as pseudo single component fluids with 'average' properties. Models are not yet available which incorporate rigorous multi-component thermodynamics, although an HSE-funded research project is currently in progress to develop such models (Topalis, 1999).

For the blowdown of, or accidental release from, pipelines, simple models are available, but sophisticated codes have also been developed to account for the complex thermodynamics, fluid mechanics and heat transfer processes occurring in multi-component, multi-phase flash and depressurisation (Richardson and Saville, 1995 and Magherefteh et al, 1999). The former model is equally applicable to depressurisation of networks of vessels and pipework.

Near-source dispersion of high-momentum jet releases

Gas jet releases

Various integral models are available to predict the dispersion of free (unobstructed) jets in the atmosphere, eg JINX (Advantica), AEROPLUME (Shell) and TECJET (DNV). Simple models have also recently been developed under HSE funding for obstructed jets (Lewis, 1998 and Cooper, 2001), although their validity for application to the highly congested environment of many offshore installations is uncertain.

Flashing liquid releases

As for gas jet releases, various integral models are available for free (unobstructed) two-phase jet dispersion, eg EJECT (HSE/AEA Technology) and RELEASE (AIChE, 1999), which make the simple assumption of homogeneous equilibrium flow. CFD has also successfully been applied to two-phase jet dispersion to allow for non-homogeneous, non-equilibrium conditions (Kelsey, 2000). Recent reviews of two-phase release and near-source dispersion have been undertaken on behalf of HSE (Ramsdale and Tickle, 2000 and Witlox and Bowen, 2002). Work is on-going by HSL to adapt the EJECT model to cover impinging two-phase jets.

Liquid pool spread and evaporation

Various integral models are available to cover evaporating as well as boiling pools, bunded and unbunded releases, instantaneous or continuous releases and spills on water as well as land. These models include GASP (HSE/AEA Technology), LSMS (CERC), LPOOL (Shell/Exxon). It is not known whether any or all of these models have been modified to take account of the most recent research undertaken on liquid spread over horizontal surfaces (referred to above).

[back to top]

Industry practice

Simple correlations or integral models form the basis of most evaluations of source terms in QRA studies of offshore installations, although the details of such modelling are rarely presented in Safety Cases. The key issues which have been identified are as follows:

  • Use of release rate ranges. Offshore QRA studies, out of necessity, categorise releases into defined ranges in order to simplify the consequence modelling. This categorisation may be on the basis of mass release rate or hole size and may be different for gas or oil releases. The choice of release rate ranges is arbitrary but, as noted in CMPT (1999), the QRA results can be sensitive to this choice.
  • Calculation of release rates. Most QRA studies appear to assume constant (ie time invariant) release rates. In reality the release rate will vary with time depending on such factors as depletion of the source inventory, successful operation of emergency isolation and/or blowdown and changes in the phase of the release. Well-validated models exist to simulate the transient release from a pipeline or network of connected vessels (eg Richardson and Saville, 1995) but are not commonly used in hazard assessment.
  • Selection of representative set of release scenarios. The selection of the representative release scenarios for an installation is a vital step in QRA, yet is often not explained in sufficient detail in Safety Cases. Release scenarios should be 'representative' in respect of the magnitude of the release, its duration, location, orientation, the type of fluid released and the coverage of the various areas on an installation from which a major accident could emanate. Deficiencies in the selection of representative release scenarios have been found in recent Safety Cases. CMPT (1999) provides guidance on selection of failure cases but notes that this is arbitrary. With the increasing use of risk-based approaches to fire and explosion resistant design and assessment (eg BP, 2001), the adequacy of the representation of the full range of potential release scenarios on an installation is becoming increasingly important.
  • Treatment of multi-component thermodynamics. As noted above, most hazard assessment calculations are undertaken on the basis of 'pseudo' single component behaviour. Codes such as those of Richardson and Saville (1995) and Magherefteh et al (1998) for analysis of blowdown (or accidental release from) networks of vessels and pipelines, which include rigorous multi-component thermodynamics, are not generally used.
  • Appreciation of uncertainties in source term. Simplifications are a necessary part of source term evaluation for offshore installations due to the complex nature of the hydrocarbon releases which can occur. However, there appears to be no systematic appraisal by dutyholders of the uncertainties associated with the various modelling simplifications and their overall impact on the hazard assessment. Without this, it is difficult make judgements as to appropriateness of the assumptions made. An appreciation of the relevant uncertainties is becoming more important with the increasing focus on 'realistic' gas cloud volumes, the evaluation of which may be critically dependent on the correct representation of the source characteristics.

[back to top]

Strategy development issues

Near-source gas jet dispersion

Impinging releases

  • In the congested environment of offshore installations impingement of gas jets on adjacent equipment or structures and interaction with background ventilation is likely and will be a significant factor determining gas dispersion. Simplified treatments of impinging gas jets have recently been developed but lack validation.
  • CFD has been applied to large scale impinging gas jet releases as part of the recent JIP on 'Gas Build-up from High Pressure Natural Gas Releases in Naturally-Ventilated Offshore Modules' but the detailed findings of this work are not yet available.

Under-expanded region

There is also a lack of understanding of the behaviour of the near source (under-expanded) region of high pressure gas releases, which is presently overcome by defining a 'pseudo' source.

Two-phase jet dispersion

  • Two-phase releases represent a significant, although not the predominant, fraction of offshore hydrocarbon releases (9%). A common source of error is in the assessment of two-phase flashing flow rates. This may introduce non-conservatisms if two-phase releases are assumed where liquid releases would in reality occur or gas releases are assumed where two-phase release may in fact occur. Current integral models of two-phase jet releases make simplifying assumptions (ie homogeneous equilibrium) and may be insufficiently validated.
  • Numerical models have been successfully applied to two-phase jet dispersion but require further validation. Such models provide the basis for evaluation of flow features of practical interest, eg impingement and rain-out, which cannot currently be addressed (development work underway on former).

Multi-component fluid releases

  • Phase I of the JIP suggested a modest experimental programme to characterise the release behaviour of the type of hydrocarbon mixtures found on offshore installations. This experimental programme appears not to have been performed in subsequent phases of the JIP.
  • Most assessments of source terms in offshore QRA studies are based on pseudo single component behaviour which may be subject to significant uncertainty.
  • More complex codes for vessel and pipeline blowdown (including accidental release) include rigorous multi-component thermodynamics, but do not appear to be used in hazard assessment work.
  • Research is currently underway to incorporate multi-component thermodynamics in commonly used hazard assessment software.

Gas lift hazards

  • Downhole annulus gas inventories may be of the order of several tonnes and represent a significant topsides fire and explosion hazard as they are outside the scope of protection of the blowdown system.
  • Widely-varying assumptions are made about the nature of releases associated with gas lift systems and may even omit consideration of the potential for release of the downhole gas inventory.

Selection of representative set of release scenarios

  • The selection of release scenarios is a vital step in QRA and is becoming increasingly important given the use of risk-based approaches to fire and explosion resistant design and assessment.

Uncertainties in source term evaluation

  • Evaluation of source terms in hazard assessment studies of offshore installations invariably involves making simplifications and approximations.
  • Examples include:
    • use of pre-defined release rate ranges (various approaches adopted);
    • assumption of constant release rates;
    • representation of multi-component releases as pseudo single component (as discussed above);
    • omission of consideration of impinging releases; and
    • simplified treatment of two-phase releases.
  • An appreciation of the uncertainties in source term evaluation is becoming more important with the increasing focus on 'realistic' gas cloud volumes, the evaluation of which may be critically dependent on correct representation of the source characteristics.
  • Advances continue to be made in the evaluation of source terms, covering such aspects as the modelling of impinging gas jet releases (simple models have now been developed), incorporation of multi-component thermodynamics into simple hazard assessment software (HSE-funded project in progress) and the modelling of two-phase releases (increasing guidance available for hazard assessment work).
Back to top
Updated 2023-08-03