A Laboratory Investigation of Foam Flow in Sandstone at Elevated Pressure

Abstract

Summary This paper describes a study made of the flow of foam through sandstone atelevated backpressures. Gamma ray attenuation was used to measure local liquidsaturations. Steady-state pressure gradients for foam were found to beessentially independent of gas velocity but nearly proportional to liquidvelocity. This flow behavior of foam is explained proportional to liquidvelocity. This flow behavior of foam is explained in terms of changes in bubbletexture. Introduction Foam is a promising fluid for mobility control in a variety of undergroundprocesses, including EOR by steam, CO2, or enriched-hydrocarbon flooding, andaquifer storage of natural gas or air. Because of its dispersed nature, foamexhibits low flow mobilities that may overcome gravity override and viscousfingering through the permeable streaks always present in underground porousmedia. porous media. Unfortunately, the flow behavior of foam is verycomplicated, with a plethora of results that are directly contradictory. Forexample, Huh and Handy reported increasing foam mobility with increasing liquidvelocity, while Nahid and Sanchez and Schechter reported the opposite. Also, Sanchez and Schechter and Treinen et al. observed hysteresis in foam mobility, whereas Huh and Handy did not. Both Huh and Handy and Sanchez and Schechtershowed that foam mobility (plotted as foam relative permeability) varies withthe liquid saturation in the core. permeability) varies with the liquidsaturation in the core. But De Vries and Wit found a constant liquid saturationin the porous medium and measured varying foam mobilities at this fixed porousmedium and measured varying foam mobilities at this fixed saturation value. Clearly, additional experiments are needed to resolve some of thesediscrepancies. Particularly needed are data in reservoir sandstones and atpressures close to reservoir conditions. Our goal is to understand foam flowbehavior under these conditions in both the steady and unsteady states. Becauseof the contradictory results noted, we desired a direct measure of the in-situliquid saturations and we wished to probe the possible separate effects of gasand liquid flow rates and hysteresis on foam flow behavior. Experiment Apparatus. Fig. 1 shows the experimental flow apparatus. A 5.1 cm-diameter,60-cm-long core of Boise sandstone was epoxy-mounted into a 316 stainless-steelcylinder designed for 20 MPa. The core porosity was 0.25, and 1 PV correspondedto 300 cm3. Swagelok O-seal pressure taps were drilled through the epoxy to thecore, and liquid-filled tubes were connected from each pressure tap to amultiplexing valve. Pressure was measured with pressure tap to a multiplexingvalve. Pressure was measured with a single Paroscientific 43KT piezoelectricquartz-crystal pressure transducer and a Scanivalve 12L7 multiplexing valve setto visit all the taps sequentially. Backpressure was maintained by a Mity-mitedome-loaded backpressure regulator. Aqueous liquid was injected at a controlledrate by a Milton Roy minipump, Model 2396. Nitrogen gas was injected at acontrolled mass flow rate through a Brooks 5850-TRP mass flow controller. Liquid saturation was measured by gamma ray densitometry, with a 47-mCi [1739MBq] (137)CS source collimated to a 0.32-cm diameter beam, with detection by a Harshaw 5.1-cm NaI(Tl) scintillation counter and an NB-15X plug-onpreamplifier. Actual counting was done on a Norland IT-5300 multichannelanalyzer with a digital gain stabilizer to compensate for drift. The intensityof the gamma ray (counts per second falling within a 662-keV peak) waspremeasured at selected stations when the core was at 0 and 100 % premeasuredat selected stations when the core was at 0 and 100 % liquid saturation. Liquidsaturation was calculated from observed beam intensity from the relationship SL = [ln(Id/I)]/[ln(Id/ Iw)], where Id and Iw are the intensity of the gamma raypremeasured at 0 and 100 % liquid saturation, respectively, and I is theintensity measured at any unknown intermediate saturation. This relationshipfollows the Beer-Lambert law. The gamma ray source and detector were mounted ona carriage that could be moved sequentially to each station by a Slow-synstepper motor and Model DPF107 motor controller. The greatest source of errorin measurement of liquid saturation is the measurement of gamma ray intensity. In counts of radioactive decay, the standard deviation of the number of eventscounted is equal to the square root of that number. Therefore, the longer thecounting period, the more precise the measurement of liquid saturation. Becauseof the precise the measurement of liquid saturation. Because of the tradeoffbetween counting time and precision in monitoring liquid saturation, it wasnecessary in transient measurements to choose between monitoring at onelocation with good time resolution or sweeping the length of the core withpoorer resolution. Generally, we counted long enough to keep the standarddeviation in liquid saturation to less than 5 saturation units (s.u.) duringtransient measurements and 2 s.u. during steady-state measurements. Pressuresweeps could be made without loss of time resolution. All measurements werecontrolled and data recorded by an HP-9000 series computer. Steady-staterelative permeabilities of the core to gas and liquid were measured in thedrainage mode, with local effective permeabilities calculated between each pairof adjacent pressure taps. permeabilities calculated between each pair ofadjacent pressure taps. Fig. 2 shows the measured relative permeabilities. Anarrow indicates the experimentally determined location of connate watersaturation, where it was not possible to gauge any liquid weepage. The foamersolution was a brine containing 18.2 g/L Na, 1.3 g/L Mg, 5.4 g/L Ca, 40.3 g/LCl, and 1 wt% active surfactant. The surfactant was one of two chemicallysimilar alkylethoxysulfates: Enordet AES 1215–9S(TM) or Steol 7N(TM). Thesurface tension, measured by the Wilhelmy plate method, was 29 mN/m; theviscosity, measured with a Contraves Model 15T rotational viscometer, was 1.2mPas. In some runs, 0.2 wt % of long-chain alcohol of average carbon number 14(Neodol 25(TM)) was added, which increased the liquid viscosity to 1.7mPas. Procedure. The core was initially vacuum-saturated with brine and Procedure. The core was initially vacuum-saturated with brine and displaced by at least 10PV of foamer solution. In all runs, liquid was injected at constant volumetricrate and gas was injected at constant mass flow rate. The progress of gas andfoam through the core was monitored by frequent pressure sweeps and bymonitoring of liquid saturation at a single station with occasional saturationsweeps. After foam had propagated through the length of the core and steadystate was reached, the gas and liquid flow rates were varied independently toreach a series of steady states. Gas flow velocity was calculated by convertingthe measured flow rate, in volume at standard conditions, to the averagepressure in the core and dividing by the core cross-sectional area. Thefractional flow of gas (also referred to as inlet foam quality) and the timeper PV were also based on the average gas pressure in the core.