Abstract This paper presents a cost effective, quantitative methodology that has been successfully demonstrated to provide improved reservoir characterization and prediction of permeability, production and injection behavior during primary and enhanced recovery operations. The method is based on identifying intervals' of unique pore geometry (rock type). This approach applies high level image analysis techniques of core material to quantitatively identify various pore geometries. When combined with more traditional petrophysical measurements on conventional core samples, various pore geometries (rock types) can be recognized from conventional wireline logs in non-cored wells or intervals. This allows the calculation of rock type and a superior estimation of permeability and saturation. Based on geological input, the reservoirs can then be divided into hydrodynamically continuous layers (flow units) and grid blocks for simulation. An overview of the results obtained in a complex carbonate and a sandstone reservoir are presented. When combined with production data, the improved characterization and predictability of performance obtained using this unique technique has provided a means of targeting the highest quality development drilling locations, improving pattern design, rapidly recognizing conformance and formation damage problems, identifying bypassed pay intervals, and improving assessments of present and future value. Introduction This paper presents a technique for improved prediction of permeability and flow unit distribution in two different types of reservoirs. The studies focus on the use and integration of pore geometrical data and wireline log data to predict permeability and define hydraulic flow units in complex reservoirs. The two studies referenced in this paper are funded by the US Department of Energy as part of the Class II and Class III Oil Programs for shallow shelf carbonate (SSC) reservoirs and slope/basin clastic (SBC) reservoirs. One objective of the program is to demonstrate advanced reservoir characterization tools that will result in a significant increase of reserves. SHALLOW SHELF CARBONATE RESERVOIR Background SSC reservoirs in the USA originally contained > 68 BBO (about one-seventh of all the oil discovered in the Lower 48 States). Recovery efficiency is low; some 20 BBO have been produced and current technology may only yield an additional 4 BBO. 1 The problem of low recovery efficiency in SSC reservoirs is not restricted to the USA - it is a worldwide phenomenon. SSC reservoirs share a number of common characteristics, including: A high degree of areal and vertical heterogeneity, relatively low porosity and relatively low permeability Reservoir compartmentalization, resulting in poor vertical and lateral continuity of the reservoir flow units and poor sweep efficiency Poor balancing of rates of injection and production, and early water breakthrough in certain areas of the reservoir. This indicates poor pressure and fluid communication and limited re-pressuring Porosity and saturation as determined from analysis of wireline logs do not accurately reflect reservoir quality and performance Many injection and production wells are not optimally completed with regard to placement of perforations, and the stimulation treatment can be inadequate for optimal production and injection practices The North Robertson Clearfork Unit exhibits all of these characteristics.

Abstract Aqueous and hydrocarbon phase traps can occur in porous media when water- or oil-based fluids come into contact with a formation which exhibits a "subirreducible" initial liquid saturation of the phase of interest. This commonly occurs with water-based fluids in many low permeability desiccated gas bearing formations and in depleted conditions in rich gas retrograde condensate reservoirs. This paper documents how phase traps are induced by direct displacement, counter current imbibition or depletion effects, and presents techniques for diagnosing whether a reservoir is a candidate for an aqueous phase trapping problem. Techniques to minimize problems with aqueous and hydrocarbon phase trapping are reviewed, followed by discussion of methods to reduce or remove the effect of existing phase traps, such as increased drawdown, alteration of 1FT, alteration of pore geometry or direct removal methods. A brief discussion of laboratory techniques used to screen the optimum process for selection are also presented.. Introduction The phenomena of permanent entrainment of extraneous or insitu generated aqueous or hydrocarbon based liquids in porous media has been documented in the literature as a mechanism for significant permeability impairment in low permeability intercrystalline sandstone and carbonate formations 1–5 . Phase traps normally occur when a water or hydrocarbon based fluid is either forced or imbibed into porous media with a subirreducible water or hydrocarbon saturation (ie. at a saturation less than the irreducible liquid saturation given the geometry, wettability and capillary mechanics of the system under consideration). Subirreducible hydrocarbon saturations are common in rich gas retrograde systems existing in a sub-dew point condition, or in mature gas fields which may have migrated into previously oil saturated strata. Subirreducible water saturations in low permeability gas reservoirs are also quite common. The mechanism for the establishment of a sub-irreducible water saturation in a low permeability gas-bearing reservoir is the subject of some controversy. The dominant mechanism is thought to be desiccation motivated by a large regional migration of gas under conditions of increasing temperature and pressure through a given reservoir area over a long period of geological time. The basic mechanism of a phase trap is created by the relative permeability effect associated with an increase in the immobile water or hydrocarbon saturation. This phenomena is illustrated on a pore scale in Figure I and from a mechanistic point of view as Figure 2. The severity of the phase trap is strongly influenced by: The magnitude of the difference between the "initial" and final trapped "irreducible" liquid saturation which exists in the porous media. The greater this difference, the larger the adverse relative permeability effect and the greater the potential reduction in permeability. The configuration of the gas or oil phase relativ permeability curves at low liquid saturation levels. The more adverse the configuration of these curves (ie. the more convex the relative permeability curve), the more significant the reduction in permeability for a given increase in trapped liquid saturation. The depth of invasion of the trapped phase. The greate the volume of fluid lost and deeper the invasion, the more difficult and slow the mobilization of this fluid becomes.

Abstract The implementation is given for an efficient streamline method that is regularly employed where minimal modeling effort is desired or where the problem is not practical for traditional finite-difference simulation. Two streamline methods are described with full details given for one method. The method detailed uses a finite-difference 2D stream junction, open or closed boundaries, fractured or unfractured wells streamline tracking by contouring, volumetric analysis by geometry, and single-phase resistance by geometry. These steps are contrasted with an alternative method. Results from both streamline methods are shown for 2 cases: a five-spot for comparison with an analytical solution and a skewed pattern element to illustrate a minimal modeling application. Results of one method are also given for case the Windalia sand waterflood. The Windalia sand is a tight heterogeneous sandstone developed on 20 to 40 acre spacing with 600+ wells. The 30 year waterflood history is characterized by realignment from 9spot to line drive patterns, hydraulically fractured producers, infill drilling, and stress induced fracturing at the injectors. The Windalia sand case illustrates a problem too complex for traditional finite-difference simulation. 2. CATEGORIZATION OF STREAMLINE METHODS Two categories of streamline methods are summarized in Table 1. Method A and Method B can each be separated into five steps. This categorization both gives an overview of the options and serves as a common vocabulary. This table could be split much finer by listing variations within each broad category listed here. The steps of Method A are described in detail in this paper and contrasted briefly with the corresponding steps of Method B. The five steps are sequential with each step adding a new capability to the analysis: Step 2 allows a visual picture of the path lines, Step 3 allows for single-phase displacement calculations, and Step 4 forms the basis fot multiphase displacement calculations done in Step 5. 3. REVIEW OF PREVIOUS WORK Stream tubes were first introduced by Higgins and Leighton 1 in 1962 for waterflooding using Method A of Table 1. They divided a five-spot along streamlines and mapped a Buckley-Leverett function 2 onto the streamlines treating them as I-D conduits. LeBlanc and Caudle 3 alternatively used Method B of Table I preferring to calculate pressure over stream function. Martin and Wegner 4 used Method A with a finite-difference stream function. Other semi-analytical techniques have been introduced to generate the pressure field for streamline tracking in "sectionally homogenous" 5 permeability fields. One of these 6 was presented with the rationale that "methods based on finite-difference techniques are difficult to implement, because they require substantial post-processing for front tracking and cannot handle singularities near the well." However, as shown in Table I, the initial calculation of pressure or stream function is independent of the post processing which occurs in Parts 2 through 5. Additionally, a finite-difference calculation of stream function can easily handle singularities near wells and fractures as will be shown in this paper. Recently there has been renewed interest in time-of-flight Methods 7,8 which are classified here as Method B. These have revived primarily because they provide an inexpensive solution for a true 3D problem.

Abstract Pressure derivatives have been shown to be more sensitive to disturbances in the reservoir than pressure signals; resulting in more detail on derivative graphs than is apparent on pressure graphs. The semilog pressure derivative is widely used in well test analysis. One reason for its popularity is that, for radial systems, the response appears as a horizontal line during the infinite-acting radial flow period, resulting in easier identification. However, when the semilog pressure derivative is applied to flow geometries other than radial, the responses are not horizontal; making identification of flow regimes more difficult. Thus, a generalized pressure derivative is necessary to simplify the identification of flow regimes in any flow geometry. In this study, a generalized pressure derivative is defined and used to identify the various flow regimes for composite systems in radial, elliptical, linear and spherical geometries. This generalized pressure derivative is of the power law type, alld is characterized by a different exponent for each of the flow geometries. Using well test data from analytical solutions for radial, elliptical, linear and spherical composite reservoirs, a graph of the generalized pressure derivative versus time, for any of the flow geometries appears as a horizontal line during the primary flow regime characteristic of that geometry. Design and analysis equations, based on the generalized pressure derivative, are presented for well testing of composite reservoirs in various flow geometries. Reservoir parameters estimated using these equations will add to the degree of confidence in the estimated parameters based on pressure analysis. The generalized pressure derivative is also used to investigate differences and similarities among the four flow systems. Results from this study confirm that for radial and elliptical systems, the long term pressure derivative behavior is influenced only by the mobility ratio between the inner and outer regions of the composite system. For linear and spherical systems, however, long term derivative behavior is governed by both the mobility ratio and the storativity ratio. This finding has a significant impact on the development of type curves for either manual or automated type curve matching for the various flow geometries. Introduction Pressure derivatives have been shown to be more sensitive to disturbances in the reservoir than pressure signals. This results in greater detail on a derivative graph than is apparent on a pressure graph. Pressure derivatives were first introduced by Tiab and Kumar (1) , who presented the derivative of pressure with respect to time. Later, Bourdet et al. (2) introduced the semi-log pressure derivative, defined as the derivative of the well pressure with respect to the natural logarithm of time. The semilog pressure derivative response appears as a horizontal line during the infinite-acting radial flow period, resulting in an easy identification of the radial flow regime. As a result the semilog pressure derivative is widely used in well test analysis of not only homogeneous, but also composite reservoirs. To analyze well tests for thermal recovery projects, reservoirs have been idealized as composite reservoirs.

Abstract The process ofinjecti1lg water above the fracture opening and closing pressure (P foc ) has enjoyed liberal use and abuse. Since injection above (P foc ) is rarely designed for, it has received little attention from a modelling perspective. Recent interest, particularly in the North Sea in the design and implementation of pressure induced fracturing, has resulted in a series of numerical simulation studies that investigate this process in detail. The work has identified both pros and cons for pressure induced fracturing. Water flood projects can benefit from increased injectivity improved frontal contact area and increased producer to injector ratios, all of which can lead to improved economics. Potential pitfalls include reduced areal sweep efficiency, loss of injected fluids to up/downhole thief zones and wellbore failure due to altered stress states. This paper describes the techniques used to analyse lab data and field tests that can be used to help design and evaluate pressure induced fracturing. Both numerical modelling and field examples will be used to demonstrate the factors controlling fracture geometry, growth rates likelihood of length stabilization and more. Introduction Injection of fluid under fracturing conditions occurs when the bottomhole pressure exceeds the fracturing opening and closing pressure. This situation has been traditionally avoided for fear of opening a fracture that could result in either a loss of injected fluids to another formation or generation of a fracture which, due to excessive length, could result in a loss of sweep efficiency and hydrocarbon recovery. The deliberate operation of injectors above P foc is rare and the modelling of the consequences even less common. Injection of low viscosity fluids, such as water, differs dramatically from traditional high viscosity fracture completion fluids, due to higher leak-off rates encountered in the case of water injection. Recent work in the North Sea has led to a new understanding of waterflood pressure induced fracturing and modelling techniques that can be used to predict fracture growth rates and geometries. Knowledge of such behavior can be used to the economic benefit of the operator by reducing the number of injectors required and by increasing recovery via contact of additional reserves. Due to the many parameters controlling fracture propagation, the evaluation of the problem is complex. A systematic evaluation of field tests, combined with numerical modeling can be used to quantify dominating parameters. This paper describes the nature of the problem and procedures which can be used to design for waterflooding above P foc Nature of the problem Semantics Since the fracturing industry is fraught with conflicting terminology the following brief review is warranted: Equations (available in full paper) The fracture propagation pressure (FPP) is the bottomhole pressure at the well under fracturing conditions. The minimum horizontal stress (σ H ) is the initial confining stress exerted on a vertical fracture. The orientation of σ H detennines the fracture orientatlon (perpendicular to σ H ). Values for σ H range typically between 14.7 and 21.5 kPa/m of depth (0.65 and 0.95 psi/ft) for vertical fractures.

Abstract A capillary equilibrium equation has been derived for porous rock. The basisfor the derivation is thermodynamic equilibrium, accounting for hydrostaticequilibrium phase equilibrium and minimization of Gibbs free energy. Theequation provides a relationship between oil saturation, pore size and heightin the reservoir, relative to the oil-water contact. In the lateral directionthe saturation in one pore call be related to another of a different size.Given the pore size distribution, the average saturation in a rock sample canbe found at an elevation above the oil water contact. Furthermore, the equationprovides insights into the laboratory imbibition and drainage pressurecurves. The equation is developed for a water wet, oil-water reservoir. A similar derivation could be carried out for differentsystems. Introduction Virgin oil saturations within a reservoir increase rapidly with height from thebase of the oil column. This rapid change is the transition zone, which in mostreservoirs is less than 6 metres thick. The transition zone is followed byconstant water (and therefore constant oil) saturation with height, called theirreducible water saturation (S wirr ). Capillary pressure defines thewater (and oil) saturation versus height, lso referred to as the initialequilibrium state apillary pressure is also an important element of flow, causing crossflow between beds, crossflow between fractures and the matrix, mitigating the formation of lingers and other factors affecting the ature ofthe flow. Capillary pressure is not a fundamental potential of nature and as such must bethe sum of other fundamental potentials. If we consider these other potentialsto be pressure potentials, it would be possible to carry out a force (pressure)balance to etermine the initial equilibrium state_ However, the pressurebalance becomes very complex, when trying o balance pressure in pores ofdifferent sizes, shapes, oil saturations and at different heights. On the otherhand if we work in the realm of energy and entropy the problem becomestractable. The total energy entropy balance about a pore is the methodologyused here to derive capillary pressure for a water-wet A derived capillary equilibrium equation, based on an energy-entropy balanceabout a pore is provided in detail in Appendix A. This equation in the form ofthe water pressure is: s w gh = Equation (1) Available in full reading text To understand the various terms in the equation, five elements need to bedefined: I) Adsorption Potential, H, represents the interaction between thewater and the rock. The adsorption potential arises from the unbalancedelectrostatic charge density at the surface of the solid. Water's polar natureallows the dissipation of this electrostatic energy by aligning with the porewall. The dissipating of the surface energy releases heat, called the heat ofadsorption. The effect of the heat of adsorption on the water phase is to alignthe water molecules into a more ordered state (a decrease in entropy). Themeasured heat of adsorption of pure water onto a ilica gel or a glass surfaceis approximately 0.070 N/m at room temperature 1 .

Abstract During the 1992 facilities turnaround of the Countess Upper Mannville "UU" Pool pressure buildup data was obtained from two vertical wells and one horizontal well. The relatively high permeability of the sand and the proximity of the horizontal well to the top of structure predetermined a lack of early time radial flow response from the horizontal well buildup analysis. The analysis of vertical well buildup response in an offsetting vertical well and information acquired from log and 3-D seismic enabled the generation of a preliminary reservoir model from which the horizontal interpretation could be initiated. This model was refined until a reasonable match was obtained from the horizontal buildup data. Good agreement was achieved between the vertical and horizontal analyses. By utilizing this offsetting data set in the horizontal analysis, a better quality, higher confidence interpretation was achieved. Benefits and limitations of this interpretation technique are discussed. Introduction Pressure transient analysis techniques have progressed significantly in the past fifteen years due to improvements in diagnostic methods through use of the pressure derivative, through rigorous integration of specialized analysis techniques to calculate total interpretation model behavior, through the improvements in pressure gauge accuracy and resolution, and through the increases in available computer power. The limitations of manual techniques such as semi-log methods or type curve matching have been transcended by state-of-the-art integrated computerized analysis methods which have greatly reduced the uncertainty f pressure transient analysis results and the man-lime required to achieve expert level solutions. In the 1990's, the analysis of most classical configurations of wellbore, reservoir and simple outer boundary solutions as been reduced to a simple process for well test analysis specialists. However, coincident to these well test analysis advances, additional research has produced theoretical solutions for an increasing number of complex well and reservoir geometries found in the field. The increasing complexity of pressure transient solutions for composite odels on certain applications produces more solution parameters than unique diagnostic regions on which to tune the variables on an independent basis. Breakthroughs in horizontal well drilling and completion has promoted the successful application of this well orientation to both virgin and mature reserves in North America and throughout the world. The increased productivity from the large horizontal well section within the reservoir bas permitted the exploitation of reserves previously considered to be beyond economic reach. In addition, horizontal wells have been proven to improve weep efficiencies in secondary recovery processes in mature pools. The pressure transient behavior of the horizontal well geometry has been determined by various authors such as Daviau 1 , Goode (2,3) , and Kuchuk (4,5,6) , among others. The application of this theoretical approach to horizontal wells in multi-well pools is the subject of this paper. Anatomy of the pressure transient behavior of an ideal horizontal well The log-log plot of the pressure and pressure derivative profile from an ideal horizontal well producing a single phase fluid from an infinite-acting homogeneous reservoir is set out in Figure la. The transient behavior can be sub-divided into five regions.

Abstract This paper develops a discrete fracture approach to flow analysis of dual-porosity systems. The well-established methods of analyzing dual-porosity flow in fractures deal primarily with matrix blocks having idealized shapes, such as slabs, blocks, and spheres. While these geometric simplifications allow analytical solution of difficult now problems, they also neglect the geometric complexities of real fracture systems. The discrete-fracture, dual-porosity approach presented here is an enhancement of discrete fracture network model which associates a volume of storative material with each fracture surface. The flow interactions between the storative volume and the fracture may be either steady-state or transient. Two applications are presented. The first uses a single horizontal fracture as a test case. The second employs a small number of vertical fractures to investigate local effects on drawdown behavior. Introduction The analysis of flow in fractures has followed two broad approaches over the past several years. On one hand, there has been the discrete fracture approach which has attempted to study the flow in individually-identified fracture conduits. Among these have been the approaches of Long, et al: (1982), and Dershowitz (1984) which model flow in discrete fracture networks generated by stochastic simulators. On the other hand, there have been numerous methods which idealize the fracture systems as porous continua with special attributes such as being an anisotropic continuum (Hsieh, 1983) or heterogeneous (stochastic) continuum (e.g., Smith and Freeze, 1979). In these methods it is assumed that the fracture system behaves as a continuum at scales larger than a representative elementary volume. As Moench (1984) points out, the theory of dual-porosity flow is a continuum approach in which there are two representative elementary volumes (REV's). One REV represents the fracture system, the other the matrix. As in all continuum approaches, the details of the fracture-system geometry are lumped into the properties of the continuum. Potentially important details on interconnection among fractures and between boundaries are lost. This paper presents the extension of dual porosity concepts from continuum-flow analysis to discrete-fracture flow analysis. The paper begins with a discussion of dual-porosity continuum methods, concentrating on the description of flow between the matrix blocks and the fracture continuum. A dual-porosity model of discrete fracture flow is developed by associating a generic matrix block with each fracture segment in the discrete-fracture network. The paper concludes with presentation of two discrete-fracture, dual-porosity simulations. The first simulation uses a single horizontal fracture as a test of the model against a analytical solutions. The second uses a single horizontal fracture and several vertical fractures. Dual-Porosity Continuum Approaches Among the continuum methods, the dual porosity approach has received considerable attention for applications in rock which contains both fractures and intergranular permeability (Warren and Root, 1963, Moench, 1984; Barker, 1985; Streltsova, 1988). The dual porosity approach idealizes the fracture-flow system as a porous reservoir with homogeneous properties. The rock matrix appears as blocks embedded within this porous aquifer. The blocks have a lower conductively and a higher storage than the fracture continuum.

Abstract Rod and tubing failures due to wear are persistent problems in many rod pumped wells. Historically, guidelines for deviations while drilling have been issued, and followed, and still, some wells have rod on tubing wear problems. Clearly, the criteria governing rod on tubing wear is not described adequately by maximum deviation guidelines for drilling. In order that rod on tubing wear problem; can be properly addressed, avoided where possible, and ameliorated when not avoidable, the factors contributing to rod on tubing wear have to be identified and analyzed for significance. This paper addresses the problem by identifying the variables, numerically quantifying their contribution to rod on tubing wear where possible, and identify those variables which are not quantified and require additional research. A method for combining the quantifiable variables into formulae useful for analysis is presented. Variables in rod design are investigated and evaluated using these numerical methods and, in some cases, bench marked against measured field conditions. Historical Methods As a preventative step in minimizing rod on tubing wear, drilling contracts almost always include provisions for wellbore surveying. Drilling contractors are required to take measurements of the direction of the borehole at specified intervals and if deviations exceed some specified minimum, remedial action is usually required at the expense of the drilling contractor. It has been common practice to use these measurements taken at the time of drilling to calculate dog leg severity (DLS), expressed as degrees per 30 meters. DLS calculations can be based on many relationships, but the most common relationship is the minimum curvature method, wherein adjacent survey points are assumed to be on the circumference of a circle and DLS quantities are the degrees of arc projected by the survey points, and scaled to an arc length of 30 meters. It has also been common practice to use the DLS calculations to estimate the location and seriousness of anticipated rod on tubing wear problems. At locations of the highest DLS, the most severe wear problems might be expected. Unfortunately, rod breaks and tubing leaks due to wear often occur at other locations in the wellbore. Many operators have taken more detailed wellbore surveys after wear problems have occurred only to learn that the surveys taken while drilling do not adequately describe the geometry of the wellbore and many severe doglegs were skipped. Figure 1 shows DLS profiles for a deviated Alberta well based on surveys taken while drilling and a survey taken later at 10 meter intervals. Because the data used for Figure 1 is taken from a deviated well survey points taken while drilling are relatively numerous. In vertical wells, the number of survey points taken while drilling would be expected to be fewer, and consequently, more doglegs would be missed. In many cases, though, even the detailed survey fails to identify the reason for rod on tubing wear. Clearly, other significant factors are involved in addition to DLS.

Abstract Often we are forced out of ignorance to use valve and wellhead seals which are over designed or overly expensive for a given environment, particularly in an environment of hot production fluids. A common industry approach has been to use the seals Which will withstand the worst thermal conditions possible in both the inside and outside environment. A combined program of testing and mathematical modeling has been designed to predict system thermal response and permit seal optimization based on environment, geometry and conductivity. Testing Tests were run on several designs of 2-1/16 " 5000 PSI alloy steel gate valves in an attempt to find means of calculating internal temperatures for equipment in steam injection or other hot production fluid service. Valves tested are designated valve A, valve B, and valve C, each having (of course) the same flange and bore size, but different geometry. The valves were set in an enclosed still air environment with thermocouples attached, and connected to a hot oil circulation tank. (See Figure 1) The valves were equipped with thermocouples as shown in Figure 2. Information was collected and recorded by a computer system capable of sampling, recording, and displaying each data point at two second intervals. Tests were run for a minimum of three and a half hours to establish steady-state conditions. Two tests were run on valve A; one test each was run on valves B and C. Results are shown graphically in Figures 3-6. The results are as expected, with all temperatures in the valve bodies cooler than the flow temperature. The worst possible case for heat removal from a valve in such service; i.e., the case in which the internal temperatures are the highest, is the case in which it is sitting in totally still air. Thus in any other conceivable instance the internal temperatures at any location would be lower relative to the flow. Concurrently with the valve testing, additional tests were done on a wellhead body for API fire resistance modeling. The wellhead tests differed from the valve tests in that the transient response of the system in fire conditions was modeled, but the basic equations (without the time dependency) can still be considered accurate. This work, as shown in OTC Paper 90-6479 (1), demonstrated that despite irregularities on the surface, the exterior convection coefficient on a wellhead body can be treated as a constant over the surface. It also showed that the hollow wellhead body can be reasonably approximated as a solid cylinder for computational purposes. Analysis of Test Results: In order to predict the temperature of any point in a body, knowledge of various physical parameters is required. These parameters include geometry, conductivity of the materials, boundary temperatures, and the convection coefficients. Geometry and materials, being functions of design and API requirements, may be taken as "given" for any particular valve. Similarly, the external and internal temperatures are "given" for any particular problem. The convection coefficient must be determined.

Abstract The equations originally developed by Nolte and Lee for the interpretations of a pump-in/shut-in test were modified and successfully used in the pump-in/flow-back tests. A new graphical technique has been developed for physical interpretation of the fracture closure pressure from a pump-in/ flow back test. The methodology developed in this paper has been successfully used in the field for the determination of the needed fracture parameters such as, fracture closure pressure, fluid loss properties, and fracture geometry from mini- or micro-fracturing tests. Introduction Mini- or Micro-fracturing tests such as pump-in/shut-in and pump-in/flow-back tests are commonly conducted for on-site evaluation of fracture parameters. Generally, the former is used to determine the fluid loss of the formation and the fracture geometry, while the latter is normally used to determine the fracture closure pressure. Conventionally, in a pump-in/shut-in test, a log-log plot of the pressure difference versus shut-in time is usually used to identify the existence or of linear or bi-linear flow. In which an one-half slope normally indicates the existence of linear flow and a quarter slope for bi-linear flow. In the case of linear flow, a log-log plot of the bottom hole pressure versus square root time is then made and the fracture closure pressure is determined from the point where the curve deviates from the straight line. Similarly, the fracture closure pressure for a bi-linear flow can also be determined from a log-log plot of the bottom hole pressures versus one fourth-root of the shut-in time. However, due to the inconclusive nature of the transient pressure decline during shut-in, the log-log plot sometimes may exhibit multiple straight lines with inconsistent slopes, which makes proper identification of the flow region impossible. Moreover, it normally requires excessively long period of time to reach the point of deflection where the closure pressure can be determined, especially in the case of high fluid efficiency. A fluid efficiency of 95 percent, for instance, it requires a shut-in time of approximately 240 folds that of pump-in time to reach the point of inflection. The pump-in/flow-back test is normally conducted separately or in concurrent with the pump-in/shut-in test. In this test, the well is allowed to flow back at a low, constant rate through a surface choke after shut-in. The flow-back process will normally enhance the rate of pressure decline, causing a continuous change of the curvature on a pressure versus flow-back time plot. There are four techniques which are normally applied to interpret the closure pressure from a Cartesian plot of bottom-hole pressure versus cumulative time including shut-in. Unfortunately, their differences are normally in the order of hundreds of psi, which add substantial uncertainties in the field applications. In this paper, the equations originally developed by Nolte and Lee for pump-in/shut-in test were modified and adopted for the pump-in/flow-back tests. A new graphical technique was also developed to resolve the difficulties involved in the conventional methods in determining the fracture closure pressure.

Abstract By substituting net pressure values for the fluid mechanics of typical fracture growth calculations, numerical models for calculating fracture growth from these pressures, using any of three popular fracture width equations, have been developed. One model uses expected net pressure behavior, ideally obtained from a similar treatment in an offset well, to predict fracture growth for design calculations. A second model uses collected pressures and injection or flow-back rates to calculate fracture dimensions over time, allowing fracture behavior to be monitored during a treatment or analyzed afterward. The governing equations for these models were examined under specific conditions, leading to guidelines for interpreting fracturing pressures under a variety of conditions. And, by considering fluid mechanics, as well as fracture mechanics and a volume balance, expected pressure behaviors for power-law fluids in all three fracture geometries have been determined. Introduction Normal hydraulic fracturing treatment design calculations combine fracture mechanics, fluid mechanics, and a volume balance to predict fracture growth with time. Fracture mechanics relate fracture width to pressure and fracture length, height, or radius; fluid mechanics relate pressure to injection rate, width, and length or radius and the volume balance relates fracture volume to injection and fluid-loss rates. As pointed out by Shlypaborsky et. al., 1,2 pressures obtained during fracturing treatments do not always agree with pressures predicted by fracture design models. They listed five factors that had the potential for causing this disagreement: (1) high perforation friction pressure, (2) high friction pressure in the fracture, (3) the generation of multiple parallel fractures, (4) higher actual fracture toughness values than measured in the lab, and (5) a non-penetrating region near the fracture tip. To isolate the cause of the disagreement, they measured overpressure, the difference between down-hole instantaneous shut-in pressure and the least principal stress, and thus eliminated the three friction-related effects. The overpressure, the result of one or both of the remaining two factors, was then used to determine an apparent fracture toughness. The apparent fracture toughness was subsequently used in a geometry model that considered fracture toughness in its solution to the fracture mechanics portion of the problem. To allow all four of the factors that may occur within the fracture to be compensated for and to provide more flexibility, models were developed that substitute net pressure – – i.e., the difference between bottom-hole treating pressure and least principal stress – –for fluid mechanics calculations. By substituting either a given net (excess) pressure value, a correlation between net pressure and time, or a set of net pressure values for the net pressures determined through fluid mechanics relationships, fracture geometry models that can be used in fracturing treatment design, monitoring and analysis have been developed. The models allow calculations to be made for radial (penny-shaped) fracture geometry and for geometries based on Khristianovic-Zheltov 3 and Sneddon 4 width equations for constant height fractures. By considering the variation of injection rate and pressure with time, the version of the model developed for treatment monitoring and analysis can be used to calculate fracture behavior during shut-in and flow-back as well as during injection.

Abstract Fluids used in hydraulic fracturing are required to have several characteristics: high viscosity and low leak-off into the formation during the injection, good proppant carrying properties, rapid breakdown during the closure. and easy clean-up during the post stimulation phase, Fluids with complex chemical formulations has been developed to meet those specifications: linear and crosslinked polymeric fluids, energized fluids (foams), gelled acids .. Those fluids depart from newtonian descriptions, and have a history-dependent rheology (to take into account the effects of temperature and shear dependency, with eventually a time-dependency due to the cross-linking process). The rheology of the fracturing fluids is an essential component in the design of the stimulation treatments, and fluid flow equations dictate the lateral and vertical pressure distribution in the fracture. Algebraic techniques have been used to provide relations, between the fracture parameters (length, width, excess pressure or injection rate) and the fluid rheological description for power law fluids 1. Introduction This paper uses a numerical technique for the simulation of three-dimensional propagation of a fracture, and considers the effects of deviation from power-law on the final fracture c:haraderistics. Simpler models (P3DH2) have been developed in the literature, and the previous model provides a tool to validate them. We compare here the vertical growth rates predicted by the fully three-dimensional model with the ones used by a discrete P3DH formulation. Many fracturing fluids have been shown to deviate from the classical power law representation, and we examine the effects of using some of the more complex rheological models on fracture geometry and proppant placement in the fracture. The paper finally addresses the problem of the influence of fracture roughness on treatment parameters, and fracturing pressure decline analysis. FORMULATION AND EQUATIONS Assuming an isotropic elastic behaviour of the formations, And that the linear fracture mechanics assumption hold, a variational Formulation for the general three-dimensional fracture propagation Equations can be formulated3 by coupling the three sets of equations. Elasticity The formulation is based on a finite element formulation of a general boundary integral equation, and has the advantage wrt the classical finite element systems of reducing the dimension of the problem by one: Equations available in full paper.

Abstract This paper describes some of the theoretical aspects which have been considered in developing a coupled flow/deformation finite element package capable of modelling the nonlinear oil sand characteristics. gas evolution, and the nonlinear fluid and flow behavior under the boundary conditions imposed during the recovery process. It is shown that the sand production is governed by the changes in permeability and soil properties which develop around the wellbore. The scheme employed in forming the wellbore controls the fluid pressure gradient close to the well and hence the reservoir stability. It is shown that the volume of evolved gases is responsible for most of the fluid flow under primary conditions. The model provides means of testing the influence of various recovery schemes which can be employed in controlling the sand and fluid production. Introduction Formation of wellbores into the deep seated oil sand reservoirs leads to significant changes in the stress state and pore pressure profile around the cavity. The shear stresses developed and the pore pressure gradients established are sufficient to cause instability over an appreciable distance beyond the stress relieved zone. The instability gives rise to sand production which is believed to be the major reason for (a) poor performance in the subsequent recovery projects (e.g. steam injection), (b) severe channelling in water floods, and (c) direct communication between offset wells. Furthermore, sand production alters the geometry around the hole which affects the fluid production during the primary stage. Recovery predictions require a knowledge of soil movements, changes in material properties and geometry. An understanding of the problem would assist in making operations changes, or employing new schemes, which could improve oil production. Analytical solutions to analyze sand stresses around a wellbore have been attempted but under Some simplifying assumptions. Risnes et al. (1) have presented elasto-plastic solutions and proposed a stability criterion for the problem by assuming steady state flow conditions, isotropic and homogeneous soil, fully saturated fluid, constant material properties and geometry. The assumptions employed, however, conceal features which may play an important role in the overall performance. For instance, the fluid pressure gradient near the wellbore would initially be well in excess of the long term steady state gradient which may result in non-predictable early instability. Eventually steady state conditions will be approached but by this time the effective geometry of the well and the properties of the formation will be considerably different from the original. Moreover, in the solutions provided by Risnes et al. no allowance is made for any gas exsolution. This is a serious limitation since the main cause of fluid flow under transient conditions is attributed to the volume of gas which evolves when the confining pressure is reduced below the bubble pressure (Byrne and Vaziri (2) ) This paper describes some of the theoretical aspects which have been considered in developing a finite element package capable of modelling the nonlinear soil and fluid characteristics under realistic boundary conditions.

Abstract An analytical method based on the finite element variation principle and fracture mechanics is proposed to assess the in-situ fracture of oil sands formations. The concept on which the method is based involves the optimization of: (1) the wellbore geometry with "fracture starters", and (2) the stress fields ahead of the crack fronts produced by the optimized fracture driving forces originating from the fracture starters. The primary function of the fracture starters is to control the direction of the initial crack propagation, while the driving forces will create desirable stress fields ahead of the propagating crack, and thus control the subsequent propagation paths. While there is no explicit fracture criteria available for the in-situ oil sands at the present time, two models for determining oil sands fracture are proposed. Numerical results from a case study will provide engineers with a comprehensive description of the fracture pattern in the oil sands formation with given geometries of the wellbore and the fracture starter, as well as the loading history. It will be demonstrated that extensive horizontal fracture is possible with particular combinations of initial geometry and loading conditions. Introduction The enormous amounts of deep oil sands reserves make a suitable in-situ recovery technique extremely attractive. There are however, two major problems associated with this method [1,2]: These are: (1) the lack of communication between wells, due to low effective permeability of the deposit; (2) the high viscosity of the in-place bitumen. The methods which are currently used to overcome the first problem include [3 – 5]: inducing hydraulic fracture by hot steam or cold gelled water; using natural communication channels, e.g., a high permeability water zone; using existing solid permeability, if it exists. In some areas where a zone having a high water saturation underlies the oil sands formation, it is possible to heat the oil zone by heat conduction from the steam flooded water zone without fracturing the formation. This approach, however, requires a thin water zone underlying a relatively thick oil zone. Such a condition exists only in a few areas among the vast oil sands deposits in Canada. For the majority of Canadian oil sands in-situ projects which have been carried out or have been planned for, a type of hydraulic fracture process has been included as a necessary first step for inter-well communication which facilitates, mainly through heat conduction, a more efficient mobilization and displacement of the oil. However, in most cases it induces a limited fracture system through which fluids could flow. Consequently, in some cases, sweep efficiency tended to be very low [2,6]. Many in-situ processes such as fire flooding, in-situ combustion, in-situ hydrogenation, and the use of an aqueous emulsifying fluid have been considered as potential recovery tools of viscous fluids. However, such processes necessitate a prior extensive fracture of the sands in order to provide the necessary distribution of the active agent, a condition which is not provided by the classical hydrofracture processes.

Abstract The application of enhanced oil recovery schemes requires a more complete understanding of reservoirs than that accepted in primary and secondary production practices. Developments in every discipline involved require increased efforts from specialists with different backgroands to express the implications of their respective findings in plain terms. In this paper, a comparison of three reservoirs is used to illustrate some geometric characters which might be overlooked in reservoir modelling due to the difficulty in expressing them in numbers. Three Mannville (Lower Cretaceous) sand reservoirs from east central Alberta were selected by the Petroleum Recovery Institute to examine the feasibility of specific EOR processes: Chauvin South Sparky A & B (Polymer), Provost – Upper Mannville B (Steam), and David-Lloydminster A (Polymer), In each case, the project started with a geological study of the reservoir. The Reservoir Geometry section of each pool study is presented in this paper. Then a comparison is made to emphasize the differences in types of inter-well communication that can be anticipated for these cases. Introduction Geological reservoir studies are becoming increasingly sophisticated, with progress mainly in the fields of reservoir petrology (through the improvement of laboratory techniques) and reservoir geometry (through the investigation of depositional trends). Reservoir and production engineering are relying increasingly on computer science. For reference to these specialized technologies, new terminologies are being created and communications between specialists of disciplines as different from one another as geology and numerical simulation require that these specialists increase their efforts to express, as clearly as possible, the implications of their respective findings. The description of the geometry of a pool is based on a partly hypothetical interpretation of the geology using scattered well data, It is necessary for a mathematician designing a model of that pool to understand both physical concepts and interpretive uncertainties, so that uncertain parameters can be realistically adjusted in the course of history matching, To illustrate this point, a comparison of three reservoirs described in qualitative terms is used to point out some geometric characters which might otherwise be overlooked in reservoir modelling. The reservoirs were selected by the Petroleum Recovery Institute 1 to examine the feasibility of specific EOR processes: Chauvin South – Sparky A and B (Polymer flooding), Provost – Upper Mannville B (Steam stimulation). and David-Lloydminster A (Polymer flooding). These reservoirs are all located in east central Alberta (Figure 1) and oil is produced from sands belonging to the Hannville Group (Lower Cretaceous) (Figure 2). This paper is based on n more complete study or the pools described in a report of the Petroleum Recovery Institute 2 . The logs pictured in the illustrations are SP-Resistivity logs, except in cross-sections K-K' (David pool). In order to save space the logs have been juxtaposed in the cross-sections, therefore, the distance between wells is not to scale. A. CHAUVIN SOUTH-SPARKY A & B The following description refers to the portion of the Chauvin South oilfield designated as "Sparky A and B Pool" by tile Alberta Energy Resources Conservation Board more specifically to the portion covered by Projects A-12.3l (Unit 1) and A-1317 (Unit 2).

Abstract Detailed reservoir description studies usually precede the selection of an enhanced recovery process and the establishment of operational procedures for a particular field. Considerable attention sometimes must be paid to the rock framework because some recovery processes are very sensitive to changes in physical properties. Reservoir geology studies have several objectives: determining the number and the vertical and lateral limits of reservoir units, establishing the thickness distribution of reservoir units and the occurrence of "shales" in these units. developing the three-dimensional distribution of porosity permeability, and hydrocarbon saturation properties for each unit. locating areas where various reservoir units are in vertical pressure communication, and providing information on the storage capacity and permeability capacity of the aquifer. Using a pilot test site as an example. a four-step procedure for geologic studies is presented. The four steps are essentially functional in nature: rock description, determination of reservoir continuity and geometry. evaluation of reservoir quality, and integration of framework data. Sources for data include cuttings and cores, well logs, core plugs, seismic surveys, well tests, and pressure-production records. Experience gained from outcrop and subsurface studies often guides the data analysis. This approach to detailed reservoir description is designed to determine the relative influence of depositional and post-depositional processes an reservoir characteristics. In the case of sandstones, for example. we first need to know whether the reservoir units are fluvial, deltaic, near shore or turbidite deposits. Subsequently, we must as certain how compaction, faulting (and fracturing), cementation, and leaching have modified the depositional character. The accuracy of the description commonly is evaluated by computer modeling, pilot studies or both. As in the earlier steps, the geologist, geophysicist, and engineer must work closely together to provide the best possible description. INTRODUCTION Many candidates for enhanced recovery operations require careful geologic study to ascertain the physical framework that contains residual hydrocarbons. Accurate knowledge of this physical framework not only permits firm calculations of "remaining reserves" but provides critical information that will aid the selection of the most appropriate recovery scheme. Because stratigraphic properties are usually the most common framework factor needing to be determined this paper will deal mostly with methods for developing reservoir geometry, continuity, and quality. Although discussion is limited to sandstone reservoirs, the methods presented are applicable to carbonates, with some modification required where secondary porosity is important. Fractured reservoirs mayor may not be suitable for analysis according to the procedures described herein. The main topic of this paper 1 is the presentation of a four-step procedure for reservoir geology analysis. The procedure will be described in general and the various steps will be demonstrated by means of (I case study. Before proceeding to the main topic, brief mention will be made of the principal factors that influence reservoir geometry, continuity, and quality. This will provide a context for the description procedure.

Abstract The solution of the wave equation applied to sucker rod pumping makes possible the prediction of a dynamometer card of considerable accuracy, for all beam pumping geometries, aver any application. Development of the net crankshaft torque history of the predicted dynagraph, along with the accompanying cyclic load factor and surface efficiency also permits the forecasting of an optimum size electric motor prime mover for the particular unit geometry and application in question. Using the performance curves of the optimum size motor selected, along with the torque history. and utility company rate schedule involved – accurate electric energy and demand costs can also be forecast. Determination of the optimum size electric prime mover and the accurate prediction of electrical costs for various beam pumping combinations enables the operator to select the mast efficient and economical beam pumping system. Introduction The impact of the energy crisis on much of the free world has recently changed the status of fuel conservation, from desirable, to imperative. Because a great proportion of the world's oil is produced by artificial lift, it becomes increasingly important to expend minimal energy in producing this vital commodity. Since large numbers of wells are produced by electrically driven, beam pumping units, it is more necessary than ever that the prime mover be properly sized, and unit geometry and pumping mode optimized – not only to reduce costs, but to maximumly conserve primary energy as well. It is of further importance to accurately predictelectrical demand and energy costs for various combinations of prime mover, unit geometry, downhole hardware, and pumping mode – so that the most efficient and economical combination can be selected. The development of these two techniques: optimum prime mover selection, and accurate prediction of electrical costs – is based on the ability to forecast a particular dynamometer card (for the proposed application) which closely matches the actual measured dynagraph, eventually taken in the field. PREDICTING A DYNAMOMETER CARD PROFILE Forecasting an accurate dynamometer card shape is based on the solution of the wave equation applied to sucker rod pumping. This predictive method was developed independently by the Sucker Rod Research Institute (SRI) and Dr. S. G. Gibbs of the WABLA Corporation, and is outlined in Gibbs' paper, "PREDICTING THE BEHAVIOR OF SUCKER ROD PUMPING SYSTEMS"(Ref, I). Gibbs' powerful technique makes possible the prediction of an accurate dynamometer card profile for different pump unit geometries under all conditions of operation. The same mathematical model is used for all geometries and pumping modes, and requires only the insertion of the actual kinematic linkage dimensions of the particular pumping unit geometry in question. Fig. 1 shows a typical beam pumping application applied to five different geometries. The Gibbs technique not only predicts accurate torsional and structural loads, and pump displacements, but also forecasts a precise dynamometer card profile for each of the different geometries considered. Because arious pumping unit geometries develop significantly different torsional and structural loads and pump displacements, this type of information is important in selecting the proper geometry, pumping mode, nd prime mover size, for any application.