화학공학소재연구정보센터
Journal of Canadian Petroleum Technology, Vol.45, No.2, 41-50, 2006
Field-scale compositional simulation of a CO2 flood in the fractured Midale field
The Midale Unit in southeastern Saskatchewan has been wider waterflood operations for 40 years, and Substantial quantities of remaining oil make the Unit a good target for implementation of an EOR process. Although naturally fractured, the Midale Unit proved to be a good candidate for a miscible CO2 flood. Encouraging results from a demonstration CO2 flood in a small part of the Midale Unit (+/- 2,000 acres) prompted an extensive study to evaluate the potential of the full-field CO2 flood. The study involved analytical modelling and several compositional simulation models, ranging from a small, element-of-symmetry model to the comprehensive, field-scale model. This paper describes the compositional simulation conducted on a large 3D model containing 40,000 grid blocks. The 204-well model covered 20 inverted nine-spot waterflood patterns and the CO2 demonstration flood area including both vertical and horizontal wells, comprising approximately 25% of the total Unit. The reservoir fluid was characterized using a 9-component Peng-Robinson equation-of-state model. Pattern-based permeability anisotropy was a key parameter used to accurately match individual well response to the waterflood and the CO2 flood. A very good history match was obtained on both the field and the well scale. The tuned model was used to predict future reservoir performance under various operating and development scenarios. Various CO2 injection strategies were considered, including continuous CO2 flood, Water-Alternating-Gas (WAG), and hybrid WAG. Under each strategy a detailed sensitivity modelling was performed investigating the optimal CO2 volumes, injectivity, WAG ratio, and the benefits of pressure support and limiting the gas production. The study showed that incremental recoveries of 11 - 15% OOIP can be expected by implementing the optimized CO2 flood scenario. Results of the compositional model were then used to mile the streamline simulator for the full-field model. A full-field development plan was delivered in four months.