Classical risk management includes three main phases:
a hazard assessment including a hazard analysis (hazard characterization and frequency analysis) and a consequence analysis (consequence scenario and severity of consequences),
a risk assessment (risk estimation and tolerance criteria), and
a proper risk management plan through mitigation and feedback. These phases must be sequential but also iterative (Fig. 1).
The hazard assessment gathers, organizes and summarizes all data relevant to risk assessment and management. It includes qualitative and quantitative characteristics of the hazard, addresses uncertainties and provides a range of forecasts based on plausible scenarios. Forecasts may be established as far as historical data reaches back, in terms of intensity and occurrence. Such forecasts are commonly derived for extreme metocean conditions, earthquakes and slope instabilities. The hazard is characterized in terms of probability of a measurable physical parameter, exceeding a certain threshold, during a period of time. A recurrence interval (or return period) is then defined. However, if there is no past experience with a hazard, there is no basis for any forecast. Such a conventional probabilistic approach cannot be applied to non-recurrent geohazards like Gas Hydrates (GH). GH present an inherent risk with no basis in their level of predictability (i.e. no return period). The GH hazard analysis (and the possible consequences) should then be evaluated based on the degree of knowledge and the related uncertainties. The approach is qualitative or semi-quantitative, as in the case of GH a quantitative approach is generally limited by the lack of extensive and accurate data and by the spatial variability of factors and parameters.
This paper will review GH proxies used to identify critical parameters and scenarios, with particular attention paid to the limitation and the uncertainties in each of the proxies. A GH hazard-consequence risk assessment matrix is proposed, to be used as part of a global risk management process during the different phases of a deepwater E&P project (drilling and field development).
All hazard characterization relies upon the degree of understanding of the process, referred to as " the knowledge??. In the " Guide to Managing Project Risks??, the Project Management Institute (1992) has defined the well established " knowledge concept?? as follows. For the " known-knowns??, where all conditions are known and certain, decision making simplifies to an optimisation problem (if there is no uncertainty, there is no hazard). However, it may be impossible to remove all uncertainties, and there is a point where addition of more knowledge no longer reduces the uncertainty significantly. The " known-unknowns??, where we know a risk exists but we do not know how it may affect our facilities and where mitigation plans should be established. The " unknown-unknowns?? are catastrophic and unpredictable events. Uncertainty results from incomplete knowledge of a process and is related to our ability to understand, to measure and/or to describe the system. These uncertainties can be divided into three types:
- –
Type 1: Uncertainties due to the variability of the earth, random by nature and inherent to the geological process that cannot be resolved even with additional data,
- –
Type 2: Uncertainties characterized as ‘epistemic’, due to incomplete knowledge of geological process, which can be reduced through dedicated R&D projects, and in particular, model uncertainty that reflects the inability of a simulation model to represent precisely the true physical behaviour of the process,
- –
Type 3: Uncertainties due to a lack of available / accurate data, and/or poor resolution. Such uncertainties include accuracy and precision of field data (measurement errors, limited, non-representative or unavailable data, data handling errors).
