Do Heavy and Medium Oil Waterfloods Differ?

Abstract To identify the parameters which impact heavy oil waterflood success, we collected production, reservoir, and operating data for 83 western Canadian waterfloods. The waterfloods were classified as either heavy or medium, and separate multivariate analysis models were built for each set. The differences between waterflooding of heavy oils and their medium oil counterparts were substantial and revealing: In terms of operational parameters, incorporating horizontal and directional wells, both producers and injectors, was significantly important to the success of heavy oil waterfloods, but insignificant for medium oils. The two most important reservoir parameters affecting the success of waterflooding medium oils permeability and heterogeneity were insignificant for heavy oils. Introduction Waterflooding is the most common method of enhancing oil production, and is becoming increasingly important in recovering heavy oil. Of the 5201 million m3 of heavy oil in place in Alberta and Saskatchewan, 207 waterflood operations (including 8 abandoned waterfloods) recover more than 24% of that oil in place. Some of the western Canadian heavy oil waterfloods were highly successful, recovering as much as seven times the primary recovery. Other waterfloods fared less well ? eight waterfloods were abandoned in the Saskatchewan Lloydminster region. Despite its prevalence, little is known about how waterflooding heavy oils differs from waterflooding their lighter oil counterparts. There is a substantial body of work on designing, monitoring, and managing waterfloods: however, the problems specific to producing heavy oil by waterflooding are rarely addressed. Some exceptions include five case studies of heavy oil waterfloods, including Forth et al.'s statistical study identifying important parameters in the Golden Lake waterflood.1-5 Two more general studies were Smith's paper on mechanistic aspects of heavy oil waterflooding and Miller's review of the state of the art of waterflooding technology as applied to western Canadian heavy oils.6,7 Miller discussed performance prediction and problems, and offered recommendations to improve performance. In contrast to the case studies, the Saskatchewan Research Council (SRC) performed two statistical studies on a group of heavy oil waterfloods.8,9 Univariate and multivariate analysis were used to highlight the broad themes common to heavy oil waterfloods. Such an approach requires a numerical value corresponding to success, and the SRC has now tested seven such measures. Other authors have also used statistics to assess waterfloods, albeit those producing lighter oils than the western Canadian sites examined in this study. The data mining study of Weiss et al. tested the ratio of secondary production to primary production to analyze Nebraska waterfloods.10 Wu et al. unsuccessfully tried to correlate reservoir parameters with the recovery of 24 west Texas waterfloods.11 McLachlan and Ershagi equated the efficiency of waterfloods to cumulative water/oil ratio.12 Methodology The reservoir data were obtained from documents published by the two provincial regulatory bodies: Saskatchewan Industry and Resources' Reservoir Annual 2002, and Alberta's Energy and Utilities Board (EUB) 2003 Statistical Series.13,14 The production data were obtained from Accumap™.15 Many waterfloods were excluded from the study. We wanted to attribute differences in recovery to the effects of waterflooding, and so excluded operations which had previously used other enhanced oil rec