Introduction The Weybum Unit is one of the largest medium-sour crude oil reservoirs in Canada, containing approximately 207 E6 m3 of original oil in place. The field was discovered in 1955 and has been on water flood since 1964. The Weyburn Unit includes 530 production wells and 170 water injectionwells. This includes the vertical infill wells added in the mid 1980's and over 140 horizontal wells which were added over the past seven years. The Weybum Unit has been a successful operation for PanCanadian Resources and for the many working interest owners in the Unit. However, the field is a mature water flood with high water cuts and decreasing oil production.Without some kind of tertiary recovery, it would soon become necessary to suspend or abandon large blocks of production wells in the Unit. The development of a large scale CO, flood in the Weybum Unit will be the most significant development in the SE Saskatchewan oil patchsince the discovery of the Weybum field. The Weybum Unit covers 180 km2 in southeastern Saskatchewan. The Weybum oil pool produces 25 to 34 degree API medium gravity sour crude from the Midale beds of the Mississippian Charles formation. The are twolayers in the Midale beds, the Marly zone and the Vuggy zone. The Marly is a chalky dolomite with low permeability. Below the Marly is the Vuggy zone, a highly fractured and permeable limestone. The water flood whichwas developed in the 1960's has recovered most of the oil from the Vuggy zone but much less of the oil from the MarIy zone. The production responsibilities of the Weybum field operators have been well defined for the water flood operation. Water injection rates are optimized in many parts of the field to maximize oil production. Water distribution is adjusted to maintain the target voidage replacement ratios. Pump-off controllers have been installed on many wells to improve rod pump reliability and to minimize the production well down hole pressure. The construction of a commercial scale CO 2 flood will cause a step change in the complexity of the Weyburn production operation. The operator will be managing a valuable commodity, that is the CO 2 itself, in order to maximize the production of oil. At the same time, the safe operation of the pipeline and satellites will be of utmost importance as high pressure sour gas is transported throughout the field. The field operator will become an important part of the commercial and social success of this project. The stakes are high. By improving the recovery of oil in this Unit from the 30% anticipated with conventional water flood, an incremental 21 000 000 m3 of oil will be produced over the next 20+ years. This compares to the 53 000 000 m3 of oil which have been produced in the 40 years since the field was discovered. 1 Two priorities in the execution of this project are training and safety. Training will be required to give the operating team the best understanding of the principles of CO 2 miscible flooding.

Abstract Weyburn reservoir, Located in southeast Saskatchewan and operated by PanCanadian PetroLeum Ltd., has reached its economic Limit of production by waterflooding. The reservoir is a target for tertiary CO 2 miscible flooding to enhance oil recovery and extend its production Life. A comprehensive study has been conducted by the Saskatchewan Research Council and PanCanadian to assess the suitability of the process for Weyburn. This paper presents a technical evaluation of CO 2 near miscibleinjection for Weyburn reservoir. This is based on laboratory studies conducted with three Weyburn oils collected from different regions of the reservoir. These studies included: measurement of minimum miscibility pressure for the reconstituted reservoir fluids for pure and impure CO 2 , determination of PVT properties for reservoir fluid-CO 2 mixtures. and assessment of recovery behaviour from uniquely designed Laboratory coreflood tests representing the permeability contrast of the reservoir. The studies indicated that the CO 2 minimum miscibility pressure for the Weyburn reservoir oils varied from about 11.5 to N.5 MPa. indicating suitability for CO 2 miscible flooding. PVJ data generated for the three Weyburn reservoirfluid-CO 2 mixtures showed that viscosity reduction and oil swelling by CO 2 also contributed to oil recovery. A slight manipulation of he measured PVT properties of the mixtures made it possible to obtain single property curves for the three Weyburn oils. This feature can be used to estimate the PVT behaviour for any Weyburn oil from the reservoir. Coreflood studies showed that CO 2 injected into the Vuggy zone could rise to the upper Marly zone and heLp enhance oil recovery. A separate simulation and modelling study conducted by PanCanadian Petroleum Ltd. satisfactorily matched the experimental PVT propenies. MMP values and recovery mechanism suggested by the coreflood behaviour. Introduction The total oil reserves of Saskatchewan are estimated at over 3 billion m 3 of initial oil-in-place (lOIP). The light and medium oil (LMO) reserves contribute approximately 1.6 billion m 3 IOIP. 1 Over 10% of the LMO reserves, or 178 million m 3 IOIP, are held in Weyburn reservoir. The Weyburn field is located approximately 130 km southeast of Regina in Saskatchewan and covers over 180 km 2 of production area. The field was discovered in 1955 and PanCanadian Petroleum (PCP) Ltd. is the major operator. It has over 627 producing wells and 162 water injection wells on approximately 24 ha spacing. 2 . The oil production from the Weyburn field comes from Midale beds of the Mississippian Charles Formation in the Williston Basin at a depth of 1310 to 1500 m. The field was produced by primary depletion for about 9 years until 1964. Waterflood was started thereafter using a nine-spot pattern. In 1985, to optimize the waterflood performance, both horizontal and vertical infill drilling programs were initiated. The combined oil recovery by primary and secondary production was approximately 28% IOIP by 1994. Waterflooding has almost reached its economic limit for the Weyburn reservoir. The ultimate conventional recovery is expected to be 31% IOIP, leaving a large target for tertiary recovery techniques. 3 A similar situation exists for the majority of Saskatchewan's LMO reservoir. 1,4

Abstract The objectives of this study were to identify factors controlling the performance of the horizontal well 10HZ-I 1-6-14W2 and make recommendations to improve the well's performance. A geological and reservoir model for the area around 10HZ-II was developed and simulated. A good history match of both the waterflood in the area and the horizontal well was obtained. The predictions of the performance of the well under various scenarios were generated. The recovery is improved by maintaining voidage replacement and by increased liquid production rates. A workover on the "toe" of the well could increase the oil rate initially. but the well would water-out faster, for a lower ultimate recovery. If the well had been landed at the top of the Marly zone over its entire length, the production would have been better. Introduction The Weyburn Unit is located approximately 130 km southeast of Regina, Saskatchewan (Figure I). The field covers approximately 180 km 2 and consists of 627 producing wells and 162 water injection wells. Oil is produced from the Midale Beds of the Mississippian Charles Formation, which were deposited on a shallow carbonate shelf in the Williston Basin. The reservoir is informally subdivided into an upper Marly and a lower Vuggy zone (Figure 2). The Weybum field was discovered in 1954 and was completely delineated by vertical drilling by 1960. In 1964, the field was unitized and a waterflood was implemented using inverted nine spot patterns oriented in the NE-SW direction. Between 1985 and 1992 over 150 vertical infill wells were drilled within the Unit resulting in substantial incremental oil recovery. The infill drilling program concentrated on improving the waterflood performance and vertical wells were drilled on both 24 ha and 16 ha spacings within some areas of the field. To date, only 25.8% of the 178 million stock tank m 3 of medium gravity crude oil have been produced. The remaining oil in place represents a significant large for improved recovery schemes such as horizontal drilling. In 1991, a horizontal drilling program was conducted in the Weyburn Unit. Three wells were drilled in different pans of the Unit to evaluate the potential and technical feasibility of applying horizontal technology in the field. With the number of economically attractive vertical infill targets declining, it was concluded that the future development of the Unit would be dependent on the successful application of this technology. Although promising, the results of this drilling program fell far short of what had been anticipated with only one well (10HZ-II) producing any appreciable amount of oil. The low overall productivity of these wells and the success of other wells in the area drilled by other operators led to an evaluation of the design parameters for horizontal wells, which included the development of a parametric numerical simulation study and a simulation evaluation of 10HZ-II, which is the subject of this paper The area simulated is shown in Figures 3 and 6. The reservoir properties are summarized in Table I. GENERAL GEOLOGY The Marly is a chalky lagoonal dolostone with occasional by or limestone interbeds.

Abstract One of the causes of the three fold increase in the number of well abandonmentsin Illinois since 1987 has been difficulties encountered in waterl100ding smallheterogeneous reservoirs. This paper highlights some development strategiesused in a successful waterflood project in Zeigler Field, Illinois, whichshould be applicable to other similar reservoirs. Reservoir management strategies at Zeigler Field included obtaining core andlog data from almost every well, detailed DST data, bottom-hole pressuresurveys and production/injection surveillance. These information gatheringefforts enabled the operators to locate major permeability barriers; thisallowed optimum placement of water injectors in the field and relatively highultimate oil recovery. The waterflood performance of the Zeigler Field was evaluated with anintegrated 3-D geologic and reservoir simulation model. Direct linkage of thegeologic model and reservoir simulator allowed both models to be greatlyimproved during the course of the simulation process. Results of the studyincluded: geological model of the study area evaluation of waterfloodrecovery efficiency and identification of a bypassed region, comparison ofalternative oil recovery strategies for optimal incremental production. Introduction Most of the 42,000 oil wells in the State of Illinois are stripper wells thatproduce less than 2 barrels (0.32 M 3 ) of oil per day. They areoperated by small, independent companies which have limited financialresources, and technical know-how to implement enhanced oil recovery methods.Almost 1,000 producing oil wells were abandoned and plugged in 1992 becausethey were no longer economical to operate. Since 1987, there has been a threefold increase in the plugging of producing oil wells. One of the causes forthis rapid increase in the rate of well abandonment has been the difficultiesencountered in waterflooding small heterogeneous reservoirs. This topic hasbeen discussed twice during monthly meetings of the Illinois Regional Societyof Petroleum Engineers in the spring of 1993 and the following reasons werecited as the likely causes of waterflood failures: Reservoir heterogeneity or compartmentalization. The common practice of hydraulic fracture during primary production mayhave created flow channels between injectors and producers causing early water break through. The rapid decline of pressure in small reservoirs after a relatively shortperiod of primary production can create gas channels that allow injection waterto bypass reservoir oil and cause premature water breakthrough. Adverse water-oil relative permeability effects. This paper describes the experience gained from development of the Plumfieldlease of the Zeigler Field, Illinois It illustrates the benefit of acquiringand using geological, engineering and field data for reservoir management anddemonstrates that some of the difficulties encountered in waterflooding a smallreservoir can be overcome with prudent reservoir management techniques. Theintegrated geological and engineering study carried out on this field had theobjectives of determining the reservoir architecture and developing a strategyfor improved oil recovery for typical small reservoirs in the Illinoisbasin. Zeigler Field is located in the southern half of Franklin County, in the southcentral portion of the Illinois Basin (Figure 1).

Abstract Predicted waterflood performance in the Dodsland Viking Field has not occurred. Previous study identified that the effects of hydraulic fractures must be taken into account in order to predict performance. The revised predictions are still somewhat optimistic when compared to actual production. The previous work has been extended to account for reservoir layering. Core data has been used to provide a quantitative analysis. Due to extreme permeability contrasts, it is difficult to simulate large ranges information permeability as separate layers. Hearn type relative permeability curves were therefore used. With high reservoir heterogeneity, waterflood response may consist only of a levelling of production. This prediction is consistent with actual well performance observed. The methodology used and its limitations are discussed. Additionally, the effects of well spacing and fracture extension are presented. This has important existing and work for economic implications planned waterfloods: Wells which are producing below their economic limit, awaiting predicted waterflood response, can be shut in. This was used to implement substantial reductions in operating costs. More realistic estimates of waterflood response in highly stratified reservoirs can be made, prior to implementation using core data. Introduction Outline This paper follows the layout of a study originally made for an individual unit. The methodology and results are explained under the following headings: The geology of the field is reviewed. The production performance of the field is discussed. The assumptions upon which the simulation model was based arc summarized. The results of various type sensitivities are presented. Conclusions are then made. This work is an extension of work done previously by Carlson and Andrews 1 . Although this paper has been written on a stand alone basis, further background data may be found in the previous paper. Background The Dodsland field, located in south-west Saska tchewan, was first discovered in 1953. The location of the field is shown in Figure-l. Light oil (36 degree API) production is obtained from the Cretaceous aged Viking sandstone. Delineation of the field occurred rapidly during the 1950's. Unitization and waterflooding followed in the mid 1960's. A resurgence in development occurred in the late 1970's and early 1980's due to high oil prices. Development in the mid 1980's was also fuelled by attractive government incentives. Geological Description Lithology The Viking consists of an interbedded shale and sandstone. The sandstone is a poor reservoir rock; its shallow drilling depth and large area have permitted extensive development. The lithology has been described in detail by Tooth et al 2 . The sandstone portions typically consist of sublitharenates which locally grade into siltstones of similar composition. Porosity is intergranular with significant amounts of micro-porosity associated with clay mineral assemblages. Cementing by silica, siderite, pyrite and kaolinite is poor. The more productive areas of the reservoir arc comprised of 15 to 25 foot thick sandstones with discrete shaly partings. In the less productive areas, the shale partings are dense and the pay consists of thin lenses of sandstone.

Abstract A lattice-Bollzmann automaton (LBA) is an efficient and accurate alternative to traditional methods in computational fluid dynamics such as finite differences or finite elements. In the lattice---Boltzmann approach, the non-equilibrium Bollzmann- transport equation is solved in terms of the particle distribution al each element of a lallice, and all other parameters (density, velocity, energy, etc) are derived from this spatial distribution as it evolves in time. A square LEA has been defined for the simulation of two-dimensional waterflooding involving two immiscible, incompressible fluids in an isothermal environment which incorporates pre-defined boundary conditions. combinations of porosity and permeability, and source-sink geometries. The numerical performance of the LBA has been thoroughly tested for several simple waterflooding examples which have analytical solutions. The advantages of the lattice-Boltzmann simulator include its computing efficiency, minimal numerical dispersion, insensitivity to grid orientations and easy implementation of complex reservoir geometries. Introduction During the past decade there has been a veritable revolution in our perception of the physical world. Fractals, chaos, cellular automata and neural networks, in association with fast computer graphics, have been the instigators of this revolution. Armed with today's computing power, it is possible to lest hypotheses, perform trial-and-error experiments and display complicated patterns which a decade age; were simply not practical. Some of these new concepts and techniques havebeen applied to solutions of important engineering problems and, as well, have provided new insights into our understanding of the physical processes governing these problems. One example related to petroleum reservoir engineering is multiphasetransport in porous media which is governed by physical processes on different length scales. When applied to reservoir simulations, it is important that the numerical solutions representative of the physical processes being modelled are independent of the underlying grid. With the introduction of computer tomography, one can gain insight into geological structure in the form of porosity and permeability on the laboratory scale. With present computers, it is possible to incorporate such information in numerical computation. In this paper, we develop and implement such a numerical procedure based on the lattice-Boltzmann method. The LBA is a numerical method for simulating a number of fluid and fluid-like systems that is motiva1ed by kinetic theory 1,2 . It may be viewed either as a discretization of a simplified Boltzmann equation using a symmetric lattice or as a Lagrangian finite-difference scheme for the Navier-Stokes equation. Is is the kinetic theory point of view that is emphasized here. The lattice--Boltzmann method represents the state of the fluid at a computational node using a set of real numbers which are called velocity populations and are analogous to the microscopic density function of the Boltzmann transport equation, The populations are streamed from one lattice site to another in discrete lime steps and are relaxed towards local equilibrium between every Streaming. The relaxation step or collision operation conserves mass and momentum (and energy for thermal models) just like a particle collision in kinetic theory. As a derivative of lattice-gas cellular automata 3 (which are natural parallel processors), lattice-Boltzmann methods are well-suited for simulating flow problems with complex geometries in a parallel computing environment

Abstract An extensive interwell tracer program was conducted at the Bone Creek Unit between 1984 and 1986. The objectives were to obtain insights into the flow distribution and, to infer clues regarding potential improvements to the waterflood. Detailed interpretations of tracer returns helped in refining reservoir description in terms of lateral as well as vertical heterogeneities. These additional details were independently verified by a history match of the unit's performance and, were by and large consistent with the cross-section fence diagrams of the pay zone prepared from log and core data. Based upon the composite knowledge thus obtained, a better understanding of sweep patterns around different injectors resulted. The tracer program led to certain changes in injection-production patterns. Four infil1 wells were drilled. Several additional infill locations were identified. Many parts of the unit, such as the northwestern portion, experienced poor sweep and poor oil recovery. These were caused, in part, by reservoir heterogeneities. With the improved knowledge of these heterogeneities, appropriate enhanced oil recovery schemes for the unit were explored. Similarly, a better understanding of the performance of the waterflood provided clues to its possible improvement. Introduction The Bone Creek/Ins tow pool is a Middle Jurassic, Upper Shaunavon tidal channel deposit located in the Southwestern part of Saskatchewan (Figure 1). It is situated along the Rapdan-Dollard- Gull Lake oilfield trend, an ancient shoreline sequence of sandstone deposits. The pool is approximately eight km long and three km wide, crending in a Northwest-Southeast direction (Figure 2). Relevant pool data are summarized in Table 1. The pay zone can be divided into two distinct litho-facies, a predominant calcareous sandstone with an average permeability of five millidarcies, and an interbedded quartzose sand with an average permeability of about a hundred millidarcies. Based upon the log and core data, a detailed fence diagram of the pool has been prepared (Figure 3). Because of the variable and discontinuous nature of the quartzose sand members, their continuity from one well to the next is difficult to ascertain. (The percentage of total pay interval belonging to the quartzose members is mapped in Figure 4). The pool has a low primary recovery (5%) because of the existence of a medium gravity crude oil (914 kg/m 3 or, 24 degrees API) discovered initiially at undersaturated conditions. A peripheral waterflood, initiated in 1960 is projected to recover an additional l7% oil-in place (Figures 5 and 6). The. feasibility of any improvement (workovers, altering operation, infill drilling and/or chemical flooding) depends upon a more complete knowledge of the lateral continuity and variability of quartzose sand members. Three members are considered to be the main conduits for accepting injection as well as for contributing the bulk of production. Saskoil, as the operator of the Bone Creek Unit, conducted an interwell tracer program on the pool between 1984 and 1966. Eleven injection wells received radio-active tracer slugs with the injected water (Table 2. and Figure 7).

Abstract The Southwest Midale extension of the main Midale pool was discovered in 1956 by the exploratory well Trans Empire et al Halbrile 4-25-5-12W2M which obtained oil production from the N1ississippian Midale Beds. Delineation and development drilling from 1956 to 1985, resulted in a total of 42 wells, 31 of which were cased as potential Midale oil wells. The Midale Marly member is responsible for approximately 75% of production from the thirty-one oilwells developed in the Southwest Mida1e pool. The remaining 25% is attributed to the Midale Vuggy and Frobisher Halbrite members. To date, this pool has produced in excess of 400,000m 3 medium gravity crude oil on primary production. Unique diagenetic events affecting the Marly and Vuggy reservoir lithofacies have either developed reservoir (i.e. Vuggy vadose zone solutioning, Marly dolomitization), or destroyed reservoir (i.e. Vuggy sparry calcite/anhydrite cementation, incomplete Marly dolomitization). The understanding of the depositional and diagenetic characteristics of the Midale Beds within the pool resulted in the following conclusions; (1) lateral continuity and relative homogeneity of the Marly dolomite reservoir facies (plus the general excellent quality and thickness of this reservoir in the areas of production) suggests that the Marly will be an excellent candidate for a waterflood scheme, (2) lateral discontinuity and heterogeneity of the Vuggy packstone/grainstone reservoir facies (in addition [Q general poorer quality reservoir development) suggests that the Vuggy will be prone to selective and limited waterflood applicable to localized areas, (3) negligible reservoir communication between the Marly and Vuggy members is anticipated, since the development of a laterally continuous and impermeable black, carbonaceous lime mudstone near the top of the Vuggy member acts as a vertical permeability barrier. Introduction The Southwest Midale pool in S.E. Saskatchewan, located approximately 4 miles west of the town of Midale, is a southern extension from the main Midale Field (Figure 1). The Southwest Midale pool, as defined by the Saskatchewan Oil and Gas Corporation, encompasses Section 30-5-11W2M. Sections 14–16, 22–25, and portions of Sections 35 & 36-512W2M. The majority of oil production is from the Mississippian Midale Beds (Charles Formation) with additional production obtained from the Mississippian Frobisher Beds (Mission Canyon Formation). The Southwest Midale pool was discovered in August, 1956 with the drilling and casing of the exploratory well Trans Empire et al Halbrite 4-15-S-11W1M. To date, this oilwell has produced 31,l00m 3 oil, 21,700 m 3 water in 10,290 days from the Midale Beds. Subsequent delineation and development drilling oil an 80 acre (32 ha) drainage pattern resulted in a total of 42 wells, thirty-one of which were cased for oil production. As of December 1988, this pool has produced 491,000 m 3 of medium gravity crude oil on primary production. Successful commercial waterflood schemes were implemented in the early 1960's by Shell Canada in the Midale Field and PanCanadian in the Weybum Field (Figure I). The Southwest Midale pool is considered an excellent candidate for a similar secondary recovery scheme, providing similar geological and reservoir conditions exist.

Abstract A waterflood prediction study has been developed for the Dodsland Viking field in Saskatchewan, Canada. This study demonstrates the importance of the geological description used to account for fieldwide variations in PVT properties. The methodology identifies the significance of local wellbore effects (hydraulic fracture stimulations). The increased power of SimBest II which is fully implicit and uses an improved matrix solver, 'ESPIDO", enables modelling of this problem where older technology fails (Beta 11). Finally, the results are compared against actual offset production performance and models which do not account for local wellbore effects. Introduction The Dodsland-Hoosier Viking field was discovered in 1953. The location of the field in southwestern Saskatchewan is shown in Figure 1. The field consists of a complex series of oil and gas pools, as depicted in Figure 2. Development of the field occurred rapidly during the late 1950's, followed by unitization and subsequent waterfloods. A resurgence of development occurred during the early 1980's in response to government incentives. The Kiyiu Lake Voluntary Unit No.1 was developed in 1983 and 1984. A waterflood was planned, and government approval obtained during 1985. However, the dramatic dip in oil prices in early 1986 delayed implementation of the project. A detailed re-evaluation was commissioned by the client, J.C. International Petroleum Ltd. to more closely evaluate waterflood economics under the more severe economic conditions of 1988. GEOLOGY The geology of the field was comprehensively documented in a paper by W.E. Evans 1 . Two main factors control the occurrence of oil and gas in this field: the separate linear sandstone bodies, which overlap. the structure, which is controlled by underlying solution collapse and post depositional compaction. Figure 3 shows the axes of the overlapping members and Figure 4 displays the stratigraphic relationship of the sandstone bodies. Cross sections through the center of the members (Figure 5) demonstrate how both of the above factors resulted in a system of oil and gas accumulations shown earlier (Figure 2). The lithology in the field has been studied in detail by Tooth et al 2 . The vast majority of reservoir rock consists of finely interlaminated sands, siltstones, and shales. These laminations are typically 1/2" thick. Core analysis near the study area yields porosities from 16 to 24% and permeabilities ranging from 0.1 mD to 40 mD. The average porosity and permeability are 20.7% and 6.3 mD, respectively. A basal chart conglomerate is also found in some members, but is not thought to be present in significant amounts within the study area. Overall, the Viking formation in this field is of poor reservoir quality. PRODUCTION CHARACTERISTICS Production from such a low permeability reservoir requires hydraulic fracture stimulation, and results in a characteristic production profile. Economic evaluations have been conducted for a number of companies with substantial production from this field. Reserves are determined using type curves, an example of which is shown in Figure 6. The production profile is determined by multiplying the initial production times the normalized rates in Figure 6. After two years an exponential decline is used.

Abstract Many oil reservoirs in Alberta and Saskatchewan have a bottom-water cone which leads to poor efficiency under a waterflood. A possible means of improving waterflood performance in these reservoirs would be to precede the waterfloor with a slug of a mobility control agent. This possibility was extensively investigated in this study using a laboratory flow model, with polymer as a mobility control agent. Twenty four experimental runs were conducted using polymer in various slug sizes. Core floor experiments were conducted to study the effect of permeability contrast, relative oil-water layer thickness, oil viscosity, and slug size. Oils of viscosities ranging from one to 200 mPa.s were used as different displacement tests. The blocking mechanism for polymer was investigated in detail. The flow stability and displacement pattern is discussed for polymer and the waterflood that followed. It is show that in order to have effective mobility control polymer slugs do not have to be displaced as an entity. Even though the waterflood following the polymer slug injection is highly unstable, the improvement over conventional waterflood is remarkable. However, the ultimate oil recovery depends on variables such as permeability contrast, relative oil-water layer thickness, oil viscosity, injection time interval. It is also shown that the injection of polymer as a slug is more efficient than alternating it with water. The effect of slug size is discussed and an optimum slug size is proposed. It is shown that polymer injection is particularly effeicent in cases for which waterflood performance is very poor. For instance, in the case of a thick (h b / h o =1) and very permeable (k o / k w = 0.06) bottom-water zone, about 2 % IOIP was recovered by a waterflood, whereas by the use of polymer as a mobility control agent, more than 45% IOIP was recovered. Also presented are numerical simulation results for polymer flow in an oil reservoir with a bottom-water zone. Experimental evidence suggest that polymer reduces the effective permeability to water without altering that to oil. Experimental investigations also show that the flow mechanism with polymer is dominated by effects like instability, mechanical entrapment, etc. In order to stimulate these effects the concept of 'selective resistance factor' is introduced for the waterflood following a polymer slug a model for the effective permeability to water is developed. This information was used in numerically simulating polymer blockage in the framework of a three-phase, two-dimensional reservoir simulator. Good agreement was obtained between numerical and experimental observations. Introduction Many reservoirs in Alberta and Saskatchewan contain a high water saturation zone underlying the oil reservoir. Such reservoirs show very poor performance under a conventional waterflood. There is a need to develop a technique that would improve the performance of a waterflood in such reservoirs. This problem has received attention for more than two decade and several techniques have been proposed in the literature 1–2 to control the mobility of water or gas. One of these techniques is the use of polymer solutions as a slug.

Abstract This paper is a progress report an the performance and operation of the pilot polymerflood in the Rapdan Unit. Polymer injection commenced in February, 1986 at a concentration of 1500 ppm. Oue to lower than expected viscosity with produced brine, injection was switched to lower salinity Belly River source water. This change has improved viscosity without affecting injectivity and enabled a decrease in polymer concentration to 1100 ppm. Infill drilling approximately doubled production to 80m 3 OPD in 1985. A response to the polymerflood is indicated by a decreasing water/ail ratio and increasing oil production trend. Polymer breakthrough has occurred in several wells but has remained at a very low concentration. Introduction A pilot polymerflood has been implemented in the Rapdan Unit of southwest Saskatchewan. Eleven infill well s, seven producers and four injectors were dri1led on 16 hectare spacing during 1985. There are presently five injection wells and thirteen oilwells on inverted five-spot patterns in the pilot. (see Figure 1) Injection commenced in January. 1986 with polymer being added to the injection water in February. The Rapdan Unit produces 230 API oil from the Upper Shaunavon (Jurassic) Formation at a depth of 140Om. Oil viscosity is 10cp at the formation temperature of 550 C. The lithology is primarily a dolomite cemented sandstone with a very low clay content. A net pay isopach is shown on Figure 2. The peripheral waterflood which was implemented in 1962 has been only moderately successful due to poor sweep efficiency caused by the high oil/water viscosity ratio of 20 and poor areal continuity. A study 1 was undertaken to determine the feasibility and to design a polymerflood to improve oi1 recovery in the Unit. Polyacrylamides were selected which developed high viscosity in field injection water and which would not plug the reservoir rock. Significant incremental recovery potential using polymer after waterflooding was demonstrated in corefloods. Polymer injectivity, which is critical to the Viability of polymerflooding was determined from a field test. A reservoir simulator was used to optimize polymer concentration, slug size, well spacing and pattern type and to predict flood performance. It is estimated that the polymerflood will recover an incremental 12.8% of OIP. FACILITIES A schematic of the polymer injection facilities is shown in Figure 3. The polymer emulsion product is shipped to the field in tank trucks and transferred to a 64m 3 insulated fiberglass storage tank which is heat traced. The storage tank contains a paddle mixer and is blanketed with methane to prevent separation and condensation from occurring. A pump continuously circulates the polymer emulsion through basket filters in the polymer skid unit. From this circulation line variable stroke metering pumps are manually adjusted to inject the desired polymer concentration into the high pressure water injection line. Immediately downstream is a static mixer which breaks the emulsion to allow hydration of the polymer. The polymerized water is then transported to the injection wells and to the formation through internally epoxy coated steel lines and tubulars.

Abstract A Polymer Augmented Waterflood Pilot is being implemented to test the tertiary recovery process in the Rapdan Upper Shaunavon Unit. Injection is to commence by late 1985. The Rapdan Unit has been under a peripheral waterflood since 1962. The performance of the water flood has been characterized by early water breakthrough due in large part to an oil/water viscosity ratio of 20. This paper describes the studies undertaken to determine the feasibility and to design a polymer flood to improve oil recovery in the Unit. Laboratory work was done to select the most suitable polymer and to determine polymer adsorption, effective viscosity and incremental oil recovery. Field testing was conducted to determine polymer injectivity, absorption and degradation. A reservoir simulator was used to optimize polymer concentration, slug size, well spacing and pattern type and to predict production and injection rates. Introduction The 4530 ha (11,200 acre.) Rapdan Unit is located in the southwest corner of the province of Saskatchewan approximately 30 km from the Montana border (Figure 1). The field was discovered in 1953 and unitized in 1957. The 23 °API oil is produced from the Upper Shaunavon (Jurassic) formation at a depth of 1400 m (4590 '). A peripheral waterflood was implemented in 1962 with several internal injectors added since. Extension of the field to the west occurred in the early seventies. Development is primarily on 64 ha (160 acre) well spacing and there are presently 31 active oilwells and 14 injectors in the unit. The Rapdan Unit has an estimated 16,736 10 3 m 3 (105 MM BBLS) of original oil in place (OOIP). Cumulative production is 2,412 10 3 m 3 oil (15.2 MMBLS) or 14.4% of the OOIP. Remaining recoverable reserves are estimated to be 608 10 3 m 3 (3.8 MMBBLS) for an ultimate recovery of 181 of COIP. Production in 1984 averaged 195 m oil/d (1.226 BOPD) at a 70% watercut. The peripheral waterf100d has experienced early water breakthrough at adjacent oilwells due to the high oil-water viscosity ratio. Also areas of low pressure have resulted from poor communication to peripheral injection wells. Screening of the reservoir concluded that polymer augmented water flood would be a feasible enhanced recovery process. Polymer was chosen over other EOR schemes due. co the high mobile oil saturation, the relative simplicity of the process and lower front end costs to implement. This paper presents the results of the polymer feasibility study which includes laboratory corefloods, a polymer injectivity test and reservoir simulation. RESERVOIR DESCRIPTION The Rapdan Sand (Upper Shaunavon Formation) is a shallow marine sand bar deposit which has been breached and reworked. A log type section is shown in figure 2. The lithology ranges from dolomitic quartz sandstone to sandy dolomite and includes zones of sandy oolitic limestone, dolomitic limestone and fossiliferous and oolitic dolomite. Thin, discontinuous and wavy dolomite laminations are common. A brown, coarse grained, dolomite cemented quartz sandstone has the best porosity and permeability. The highest permeabllity sand occurs with a minimum amount of calcareous cementation.