Influence of scenario choices when performing CFD simulations for explosion risk analyses: Focus on dispersion

• The simulations were performed for a large number of locations over the entire FPSO process area.
• There was the uncertainty in having only few dispersion simulations representing all possible scenarios.
• The adopted approach provided realistic risks including a safety margin and within reasonable computational efforts.
• Given a case where the design loads are overestimated, considerable design costs will be incurred.
• Increasing the number of simulations would be a way to reduce uncertainties and costs in the design.

Explosion risk analyses (ERAs) are widely used as means for deriving the dimensioning accidental loads (DALs) for design of offshore topside facilities. ERAs can predict explosion loads in detail, including overpressures, differential pressure, and drag loads (FABIG TN-08). Loads with returning frequencies of 1E-04 per year are commonly adopted as DALs and are effectively incorporated in standards and legislation such as NORSOK Z013 (NORSOK, 2010) and ISO 19901-3 (ISO, 2010). Detailed guidelines for how to perform these analyses are described in the Norwegian standard NORSOK Z013. Computational fluid dynamic (CFD) simulations form a key part of these analyses and how these simulations are performed, influence the accuracy and uncertainty of the predicted loads. Dispersion simulations are a particular important part of these studies as they both define the size of the clouds and give basis for detailed ignition probability modeling. This paper explores how explosion risk and the dimensioning accidental loads are affected by dispersion scenario choices and address the associated uncertainty. This research work is conducted as a part of an explosion risk analysis for a large process area of an FPSO. This study involved extensive CFD efforts with more than 3000 CFD simulations minimizing the use of simplifications. Subsequent steps involved testing various means of simplification and also an assessment of how the choices made by individual analysts could influence the results.