Field-Scale Simulation of Cyclic Solvent Injection (CSI)

Abstract

Summary Only 5-10% of the oil in Lloydminster heavy-oil reservoirs is recovered during cold heavy-oil production with sand (CHOPS). CSI is currently the most active post-CHOPS process. In CSI, a solvent mixture (e.g., methane/propane) is injected and allowed to soak into the reservoir before production begins (Fig. 1). CSI has been focused on heavy-oil recovery from post-CHOPS reservoirs that are too thin for an economic steam-based process. It has been piloted by Nexen and Husky and was a fundamental part of the CDN40 million joint implementation of vapour extraction (JIVE) solvent pilot program that ran from 2006 through 2010. This paper describes field-scale simulations of CSI performed with a comprehensive numerical model that uses "mass-transfer" rate equations to represent nonequilibrium solvent-solubility behaviour (i.e., there is a delay before the solvent reaches its equilibrium solubility in oil). The model contains mechanisms to consider foaming or to ignore it, depending on the field behaviour. It has been used to match laboratory experiments, design CSI operating strategies, and to interpret CSI field pilot results. The paper summarizes the impact on simulation predictions of post-CHOPS reservoir characterizations where the wormhole region was represented by one of the following five configurations: (1) an effective high-permeability zone, (2) a dual-permeability zone, (3) a dilated zone around the well, (4) wormholes (20-cm-diameter spokes) extending from the well without branching, and (5) wormholes extending from the well with branching from the main wormholes. The different post-CHOPS configurations lead to dramatically different reservoir access for solvent and to different predictions of CSI performance. The impacts of grid size, upscaling, solvent dissolution and exsolution rate constants, and injection strategy were examined. The assumption of instant equilibrium solubility resulted in a 23% reduction in oil production compared with when a delay in solvent dissolution and exsolution was allowed for. Increasing the gridblock size by a factor of nine reduced the predicted oil production five-fold. Assuming isothermal behaviour in the simulations decreased predicted oil production by 17%.