MUESTRA

The literature is rich with two phase studies addressing the flow behaviour for low viscosity liquids and gases. However, very limited published work has been done to address high viscosity multiphase flow behaviour. The usage of relative terms (low and high) for viscosity in these studies were associated with values of 1 to 50 cP for low viscosity and anything beyond 100 cP is considered as high viscosity. A summary of the experimental research studies on gas-liquid flows found is given at the end of the section (Table 2-3).

Weisman et al. (1979) studied experimentally the effect of fluid properties and pipe diameter in horizontal pipes. The testing fluids consisted of air-glycerol water solutions with viscosities of 75 to 150 cP. They concluded liquid viscosity has minor effect on flow pattern transition boundaries.

Taitel and Dukler (1987) investigated the effect of pipe length on flow pattern transition boundaries for high viscosity liquids. The liquid (glycerine/water mixture) viscosity ranged from 90 to 165 cP. They concluded that pipe length can play considerable effect of transition boundaries for high viscosity liquids.

Andritsos et al. (1989) have done research into the effect of liquid viscosity on the initiation of gas-liquid slug flow in horizontal configuration. They proposed a mechanism for viscous liquids that slugs arise from small wavelength Kelvin-Helmhotz (KH) waves, which agreed with experimental results. However the new mechanism is applicable to liquids beyond 20 cP only.

Barnea (1991) proposed a combined model of viscous and inviscid KH stability analysis to determine the transition between slug and annular flows. It was shown that the combined model gave good results for different experimental data. As the results were compared to Taitel and Dukler (1976) model, which is the first mechanistic model to predict flow pattern transitions for horizontal and near horizontal gas-liquid flow, it was also shown that Taitel and Dukler (1976) model was valid for different liquid viscosities up to 100 cP.

Barnea and Taitel (1993) investigated the inviscid and viscous KH stability criteria for stratified flow. It was shown that results for low liquid viscosities were different for each analysis; on the other hand, they were similar for high liquid viscosities.

Colmenares et al. (2001) considered pressure drop models for horizontal slug flow for viscous oils. Based on their experimental results, the slug flow pattern enlarged as the oil viscosity increased; also by evaluating slug models, they concluded that the Barnea and Taitel (1993) was best to be used for high viscosity oils, and based on this, a modified model was developed for 480 cP.

They concluded that slug length decreased, frequency and liquid film height increased as the liquid viscosity increased.

McNeil and Stuart (2003) experimentally investigated the high viscous liquid phase effect on two phase flow in vertical pipe. The liquid viscosities were ranged from 1 and 550 cP using water and glycerine solution. Annular flow was mostly observed in their study. It was concluded that correlations of entrained liquid fraction and interfacial friction factor for low viscous fluid are inapplicable to highly viscous range. They also developed a new correlation for the interfacial friction factor based on their experimental findings for high viscous fluid.

Gokcal et al. (2006) investigated two phase gas-oil flow in horizontal pipe. The experiments were conducted using viscous oil Citgo Sentry 220 and air. Holdup, flow regime maps, and pressure drop measurements were presented for 4 oil viscosities (181, 257, 378, and 587 cP). As most flow regimes observed was slug flow in addition to stratified wavy and annular flows, it was reported that the frequency of elongated bubbles increased with increasing viscosity while the bubble length decreased. Also, the presented data showed no significant effect of high oil viscosity on the location of transition boundaries. Finally, after comparing the experimental data for pressure gradient and liquid hold up against Zhang et al. (2003) unified and Xiao et al. (1990) mechanistic models,

viscosity oils and new models should be developed to predict flow patterns, pressure losses, and liquid holdup more accurately.

Al-Safran et al. (2011) studied the slug flow characteristics for high viscosity liquid in horizontal pipe. Based on their experimental results, using viscous mineral oil, the slug front under high liquid viscosity condition was less turbulent due to low Reynolds number with top boundary layer moving faster than the slug body; bubble nose at the back of slug was long and accelerated by the wake of entrained gas pockets which leads to short slug; and liquid film height was large and aerated. A theoretical analysis with proposed empirical slug length model was developed and compared with existing models (Brill et al., 1980; Scott et al., 1981; Norris, 1982; Barnea and Brauner, 1985; Dukler et al., 1985) against experimental data. Although the Al-Safran et al. (2011) model had the best performance, the authors recommended more independent data was needed for further comparison and verifications.

Matsubara and Naito (2011) considered flow patterns identification using aqueous solution of polysaccharide thickener flowed concurrently with air through a test section, which had an ID of 20 mm and 19 m in length. Viscosity of these water solutions varied from 1 to 11000 mPa s with Newtonian viscosity.

The results obtained showed that the flow patterns strongly depend on the liquid viscosity, and were not in agreement with previous work by Weisman et al.

(1979). Also Taitel and Dukler (1976) model could not predict the flow pattern transition. Therefore different approach is needed to cover flow prediction in highly viscous fluid.