The main objective of the thesis is to propose and test a way to model flow development downstream a choke valve.
Experiments were performed in the SINTEF medium scale flow loop in order to study oil-water flow development along the pipe. The test section was a horizontal, 10.91cm diameter, 220m long pipe with three 180◦ bends and inlet mixing valve at the beginning of the section. There were four 0.5m long transparent sections with video recordings, three traversing gamma densitometers and three droplet size distribution measurements (Canty Inflow Particle sizer) along the test section. The experiments in this setup were done in two campaigns with different oils. In June 2018, a mixture of Exssol D60 (density 789 kg=m3 and viscosity 1.43cp) and tap water with Span 83 surfactant added to the oil was used. The mixture velocities tested were 1, 1.5 and 2m/s with water cuts between 0-75%. The second campaign in March 2019, tap water and a mineral oil blend of Primol 352 and Exxsol D80 (density 850 kg=m3 and viscosity 35cp) was used with Span 80 surfactant added to the oil. The mixture velocities tested were 0.5, 1 and 1.5m/s for water cuts between 0-90%.
Flow patterns are identified by using the video recordings and phase fraction profiles from the gamma densitometers. The identified flow patterns include Stratified mixed (SM), Homogeneous oil continuous dispersion (Do-H), Inhomogeneous oil continuous dispersion (Do-I),Inhomogeneous water continuous dispersion (Dw-I), Oil continuous dispersion with a dense layer of water droplets (Do-DP), Water continuous dispersion with a dense layer of oil droplets (Dw-DP) and Oil continuous dispersion with a dense layer of water droplets and a water layer (Do-DP + w). The flow patterns identified are classified in flow pattern maps as functions of either mixture velocity and water cut or superficial velocities of oil and water. The measured pressure gradients are related to the flow patterns. The phase inversion point is observed at a mixture velocity 1.5m/s with 25% water cut with a peak of 322 Pa/m pressure gradient. The effect of pressure drop across the inlet mixing valve to the flow development along the pipe is analyzed in terms of pressure gradient and local dispersion factor. For both campaigns, the increase in pressure drop across the mixing valve leads to increase in the pressure gradient.
For 2018 campaign, the pressure gradient along the pipe begins by increasing due to settling of droplets before it starts decreasing due to formation of free water layer. For 2019 campaign, the pressure gradient trend was a gentle decrease along the test section because the flow was dispersed throughout the test section. For experiments with the same water cut and pressure drop across the mixing valve, higher flow rate causes a delayed separation of phases along the pipe. Clear flow development and lower average pressure gradients are seen for the 2018 campaign compared to 2019 campaign; this is associated with higher oil viscosity in the 2019 campaign.A dynamic model from Schumann (2016) is modified and used to predict the flow development along the pipe. The model consists of three sub-models named valve model for determining the initial droplet size, phase distribution model for determining the heights and areas of the oil, dispersion and water layer, and the pressure gradient model for calculating the pressure gradient along the entire test section. In this model, the mixture is assumed to be initially homogeneous dispersed with a single droplet size and the phase distribution considers two main mechanisms: droplet settling and coalescence. The results from the model compared to the experimental results show a satisfactory agreement in the interfacial tension positions and the droplet sizes. The pressure gradient model does not give the same trend as in the experimental values.