Speaking at a recent seminar co-hosted by France’s Ineris* organization and CMR/GexCon**. Benjamin Truchot outlined the use of computational fluid dynamics-based modeling to solve safety challenges. At issue is the use of ‘classical’ 2D ‘phenomenological’ explosion models as opposed to a full 3D/CFD approach. Both methods have their advantages and limits. 2D models may often include more domain knowledge but the 3D approach can provide more insights into specific scenarios—particularly with respect to asset configurations. Both approaches are subject to issues such as the accuracy of the underlying physical models, users’ understanding of the models and code and computational limits.
Henri Tonda described how uses a variety of tools to assess and mitigate risks in both asset design and operations. Modeling has increased in sophistication over the years. Piper Alpha was modeled as a straightforward TNT-like explosion. Today various combinations of gas dispersion, with or without fires are used to investigate interactions between potential leak points and ignition sources such as flares and vents. FPSOs and other assets are designed so that potential fires are controllable. Fire and explosion is the main hazard, causing 30% of all accidents.
Tools such as Phast, Firex and Fred are easy to use and good for multiple simulations but fail if used to model complex configurations and meteorological conditions. Gexcon’s Flacs takes the wind into account—and can give very different results. Tonda described use of 2D/3D modeling to explain why the Girassol FPSO was experiencing frequent gas detections due to tank venting. Gas was being detected on the bridge, alarms were sounding, production halted and crew mustered to safety boats. What was going on? It turned out that Girassol’s designers had used a ‘classical’ tanker design which did not anticipate very low wind speeds. Modeling with Phast concluded that gas from the vent could not reach the process deck. But 2D/3D modeling with Flacs showed the heavy gas pooling at sub 1m/s wind speeds. Various combinations of fans and ducts were modeled before Total settled on a redesigned annular, fan-driven vent which added turbulence and stopped the emergency shut downs.
Tonda concluded, ‘2D is not false but incomplete, 3D takes other factors into account like asset geometry, wind speed and deluge efficiency. 3D is key to accurate risk assessment.’ In the Q&A Tonda elaborated that in Norway, there is a push for probabilistic evaluation of explosion risk. But the problem is that risk is not generic but specific to an installation. Events such as an approaching supply boat, newly installed equipment may not be in the probability database. It is important not to extrapolate too far from generic models and perform a rain check against what it on your particular asset—such as where a crane is located. Total currently analyses over 100 potential risks in depth which is ‘both a lot and not enough.’ These focus on risks that could destroy accommodation or other ‘super critical’ cases. More from Ineris.
* National industrial environment and risk body.
** Developer of the Flacs 3D explosion modeling package.
© Oil IT Journal - all rights reserved.