Investigation of Mechanisms for Gravity Drainage in a Hele Shaw Cell

Abstract

Solvent recovery processes are potentially less energy and greenhouse gas intensive than thermal recovery methods because hydrocarbons are injected instead of steam. To date, an accurate model to predict oil recovery rates is lacking partly because the interplay of the mass transfer and convective mechanisms is still not well understood. In this thesis, a novel Hele-Shaw type apparatus was designed and commissioned to investigate these mechanisms in a controlled flow geometry. The apparatus consists of a Hele-Shaw cell (parallel glass plates) that can be partially filled with bitumen and rotated to set a target initial slope of the bitumen layer. Solvent is fed at the top of the bitumen at a constant volumetric flow rate. It flows along the bitumen surface, sweeps the bitumen that diffuses into this drainage layer, and is collected in sample vials. The flow rate, composition, and properties of the drained liquid are measured over time and photographs of the bitumen profile are taken periodically. Gravity drainage was measured at ambient conditions for bitumen and toluene at injection flow rates from 0.1 to 2 cm³/min, gap widths of 0.5 and 1 mm, and initial angles of inclination between 30 and 45°. The following recovery mechanisms were identified: 1) diffusion of bitumen into the drainage layer; 2) falling film flow of the drainage layer; 3) creep flow of the bitumen phase. A two-dimensional numerical model was developed where the bitumen was divided into columns, each with a solvent (drainage) layer and a bitumen layer. The drainage layer flow was modeled as a falling film. The mass transfer of bitumen into each solvent block was determined from Fick’s First Law of diffusion with an infinite acting boundary condition. The diffused bitumen was assumed to be swept immediately to the next solvent block. Creep flow of bitumen under its own weight was included to predict the bitumen profiles. After each time step, a material balance was performed and the composition and fluid properties of each block were updated. The model with a single tuning parameter matched all of the bitumen production rates with an average deviation of 7.4%.