This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada

Resource Abstract

The interpretation of predictive information can have a significant bearing upon the engineering decisions drawn from simulation and this is not always well communicated. This paper presents case studies for three fairly routine CFD based types of analysis in the oil and gas sector to demonstrate this point.



1) Projection contours for interpreting dispersion behaviour



When presenting contour plots to illustrate dispersion behaviour, one common method is to plot the concentration or temperature of the dispersing medium on a planar surface within the region of interest. However, this surface may not coincide with the maximum concentration and, therefore, may not capture the true extent of the dispersing medium – which for technical safety applications, is non conservative. An alternative method is to present envelope plots which do capture the true extent of the plume but only provide information at a single concentration.

Abercus uses projection plots to convey dispersion predictions. It is essentially a compilation of many envelope plots, so it looks almost like a conventional contour plot but it is not confined to any single plot plane and does, therefore, capture the true (maximum) extent of the dispersing medium plus concentration information in a single image.



2) Risk based approach for thermal cooldown



CFD is now used fairly routinely is for predicting thermal cooldown within subsea pipelines and equipment following the cessation of flow. The output from the CFD is typically a cooldown curve which shows how the temperature of the fluids is predicted to cool with time, and the aim of the study is to predict how long it may take until the fluid reaches a temperature threshold, at which point there may be a risk of solids forming within the fluid which presents an associated risk of blockage.

The usual CFD methodology, however, is often overly conservative and is representative of an infinitesimally small volume of fluid falling below the temperature threshold that could not credibly form a blockage within a pipe. This method has evolved simply because the information for the worst case curve is relatively easily extracted from the CFD simulation (using a minimum temperature function or similar).

This case study presents a novel method using information that is readily available from most commercial CFD codes, based upon an alternative approach that allows a small volume of fluid to fall below the temperature threshold, small enough that the risk of blockage remains negligible.

For the case presented, CFD and the novel method of interpreting the cool down data proved instrumental in saving a significant sum on the project.



3) Probabilistic methods for determining explosion design loads



The probabilistic methodology outlined in NORSOK Z-013 is routinely used for determining design explosion loads. The method requires many (often hundreds of) possible dispersion and explosion scenarios to be simulated and a probability of occurrence to be allocated to each scenario. This information is then used to compile exceedance curves for explosion load, from which the load corresponding to an allowable frequency (typically 1 in 10000 years) is determined.

However, NORSOK Z-013 does not outline in detail the precise approach. Inevitably this means that even if different practitioners had an identical underlying pool of CFD dispersion and explosion simulations, the exceedance curves derived from them will differ between practitioners.

Unfortunately, this is rarely conveyed to the wider design team. Exceedance curves are typically presented outright without any indication of the sensitivity of the exceedance curves to the assumptions used in deriving them. This case study demonstrates that by changing some of the underlying assumptions, the design explosion load could vary anywhere from 2 barg to over 5 barg.