VALIDATION OF OPENFOAM MODELS WITH MAGNETIC RESONANCE IMAGING AND NUCLEAR GAMA DENSITOMETER MEASUREMENTS

In the oil & gas industry custody transfer application, the transaction of crude oil between buyers and sellers as well as the determination of government taxation is based on internationally accepted quality measurement (or sampling) and quantity measurement (flow metering) standards. In liquid flow sampling, the composition of oil-water two-phase flow is unknown as a priori and the automatic quantification of the amount of water (or water cut) in crude oil is paramount importance to correct the flow metering that has to be measured with an accuracy/linearity of less than ±0.15%. The sampling process has five stages and the most crucial part is the mixing process to create a representative sampling to quantify the water-cut with the highest possible accuracy to correct a meter. This is one of the most important measurement tasks. Because, sampling is not just another function of measurement but it is the heart and soul of the profit figure – affecting government taxation as well as our energy bills and weekly shopping too. However, due to the difficulty of achieving homogeneous mixing to a scale of the required accuracy, sampling has been the “elephant in the room”. To circumvent this critical problem, we tackled it using a full cycle approach by building world class flow loop facilities, innovative diagnostic techniques, advanced design optimisation tools and the best in class mixing system –namely, the SmartMix® device with a Liquid Jet In Cross Flow (LJICF) configuration [1,2]. One of the diagnostic tools we utilised to characterise and optimise the mixing process in the prototype SmartMix® device is magnetic resonance imaging (MRI) in a project conducted in collaboration with the University of Cambridge Magnetic Resonance Research Centre (MRRC). However, this experiment was conducted using a 2.5” diameter flow loop and the production SmartMix® device is only characterised and validated using a 10” diameter flow loop [3]. Although a 10” diameter flow loop can realistically be considered an industrial scale, typical crude transmission lines usually have nominal pipe diameters of 48”and above. Unfortunately, there are no known testing or characterising facilities or accredited calibration laboratories in the world to mimic flows in such large pipes. Therefore, computational fluid dynamics (CFD) is one of the few, if not the only one, alternative methods that could be utilised as a design guide or design tool in such critical applications. However, recent studies [4] showed that predicting dispersed oil-water flows, which are a key and desired outcome in quality measurement or sampling applications, using well known commercial codes such as ANSYS FLUENT being a challenge. Therefore, we dedicated to put our effort on a tool that we can extend, modify and validate in order to capture this critical application. To that end, the multiphase models in the Open Source CFD Toolbox OpenFOAM® [5] became prime targets. We considered the twoLiquidMixingFoam (2LM) model and compared it with an experimental measurement that uses gamma densitometer. The 2LM prove to be deficient at low flow velocities due to artificial mixing or mixing. We then modified the 2LM model (called TPD model) by using an implicit TVD scheme both to reduce the computational cost and improve its accuracy. The model was compared both with data from the gamma densitometer measurement and a magnetic resonance imaging (MRI) data we generated for an advanced mixing application that employs a complex liquid jet in cross flow (LJICF) configuration. The TPD model showed excellent agreement both with the densitometer and MRIT data.