Abstract In heterogeneous reservoirs like Hassi Messaoud, the exact interwell location of each reservoir sandstone body can not be in most cases easily located, and the potential of drilling unsuccessful production wells is usually present. In this situation a reservoir characterization model with usable uncertainties is needed. Conventional modeling techniques fail to quantify heterogeneities in lateral direction, however horiontal well data appear to compensate for this lack of information, providing some knowledge about variability of reservoir parameters, using data recorded along the horizontal drainhole of horizontal wells, hence providing adequate measurements of reservoir properties in the interwell locations. In this study a petrophysical model was generated for the north east area of Hassi Messaoud field comprising four horizontal wells. The generated models of porosity permeability and shale distribution were in agreement with Hassi Messaoud braided fluvial depositional model. The facies lateral variations have confirmed the nonuniform depletion throughout this sector of the field, which has led to its zonation. Once the reservoir characterization was complete, the generated model was validated using a full-field model of North East area, that involves history match of each well individually. Introduction For each target reservoir, many geostatistical realizations can be generated by integrating seismic, sedimentology, geology, and petrophisical data with their respective uncertainties captured by these models. Traditionally, reservoir description has been interpolated at the inter well locations, using simple algorithms, which fails to capture the true geological complexity and therefore results in ineffective prediction. To address this failure in the traditional route, one needs to consider using tools which allow the user to represent more accurately the range of plausible geological cases. Stochastic modeling techniques are used to construct detailed geological description using structural information, diagenetic and depositional models, reservoir stratigraphy, log and core data, and fault and fracture information. In most cases, adequate measurements of reservoir properties are not available to evaluate inter well variability at small scales. Information about lateral variations of reservoir parameters comes from horizontal wells or seismic data. Log data from horizontal wells have been used to improve inter well reservoir characterization, identify fractures and lateral variation of facies. Stochastic reservoir modeling is becoming commonly used tool to describe reservoir heterogeneities. It involves the generation of images of the reservoir lithofacies and rock properties that ideally, would honor all available data (Core measurements, well logs, seismic and geological interpretations, analog outcrops, well test interpretations, etc). Certain information, like production data or effective properties derived from well tests, can not be easily incorporated into the reservoir model. Almost always, a stochastic reservoir modeling exercise will involve a hybrid technique combining the best features of a number of available algorithms. Simulated annealing is an algorithm initially developed for the solution of combinatoin optimization problems. Generally, conventional modeling techniques fail to capture the true geological complexity of the reservoir in the lateral direction due to the lack of data in unsampled zones. In most field cases, only vertical well data representing a small fraction of the reservoir are available to describe the spatial distribution of reservoir properties. Horizontal wells, however, have emerged to quantify heterogeneities in the lateral direction because of their extended reach.

Abstract Infill drilling is a common approach to improve oil recovery in the Permian Basin. The economical success of such projects depends on the guessed reservoir characterization. In this paper, the impact of infill drilling locations on oil recovery is discussed. The major characteristic of this study compared to previous work is the reservoir description used to select the infill locations. The reservoir was totally characterized by using an automatic history matching algorithm that solves an inverse problem using the simulated annealing method (SAM). The major advantages of this approach are: it can estimate a large number of reservoir parameters and it is able to integrate various disciplines, such as geology, petrophysics, and reservoir engineering. The parameters estimated from a recent study for the pilot area of the East Vacuum Grayburg San Andres Unit (EVGSAU) were discussed for designing infill and waterflood projects. The oil saturation map obtained at the end of waterflood period helped in locating interwell areas with a high saturation of mobile, unswept oil as potential infill drilling locations. By combining the oil saturation map with other parameters estimated by the inverse model, isolated infill locations were suggested in the pilot area. The incremental oil recovery of various schemes of infill and waterflood are compared. It appears that infill drilling locations selected with scientific criteria do improve the oil recovery of a very heterogeneous reservoir. Introduction For carbonate reservoirs such as the ones located in the Permian Basin (West Texas and New Mexico), obtaining a reservoir description for simulation is challenging because of the heterogeneous nature of the reservoirs and the scarcity of available field and laboratory data. On the other hand, the more a reservoir deviates from ideal behavior, which is normally the result of heterogeneity, the greater the opportunity for incremental recovery by infill drilling. However, implementing an infill drilling project without a careful reservoir description may lead to an economic failure. Permian carbonate reservoirs are stratified and have areal discontinuities. Thus, heterogeneity causes the primary and secondary production to reach the economic limit at unexpectedly early times. To accelerate production or to produce incremental oil, a reliable reservoir description is required. The importance of reservoir description in infill drilling projects has been discussed and stressed by previous authors. Today, it is commonly accepted that the success of any infill and/or waterflood projects depends entirely on the reservoir description. P. 741^

Abstract Current computer technology has given us the ability to generate realizations of spatial reservoir properties as never before. However, achieving the goal of generating reservoir descriptions which are consistent with all of the available information continues to be a difficult and time-consuming process. Especially, the inability to constrain probabilistic-based reservoir descriptions using (dynamic) production data is becoming recognized as one of the primary short-comings of existing techniques. This paper describes a procedure to identify, quantify and incorporate within a conditional simulation procedure the spatial reservoir characteristics which dominate well performance. Primary production data are used to determine controlling spatial characteristics. The primary production parameters are combined with waterflood constraints and the method of simulated annealing to produce realistic reservoir descriptions––both in terms of their spatial characteristics and simulated well performance. Although excellent matches of well performance are also obtained using these "indirect performance constraints" while excluding the variogram constraint, it is shown that the resulting alternative reservoir descriptions poorly reproduce actual spatial reservoir properties. Thus, it is concluded that spatial correlation structures cannot be extracted from performance data. The robustness and flexibility of the approach is demonstrated using a two-dimensional areal, field scale reservoir study. The proposed technique can be used to greatly reduce the uncertainty of predicting future reservoir performance. Introduction A probabilistic modeling approach is suitable for translating geologic uncertainties to uncertainties in future reservoir performance. Integration of all available data into the reservoir description minimizes the range of uncertainty. The concept of minimizing uncertainty by including additional conditional simulation constraints is illustrated in Figure 1. Stochastic modeling is used to generate several equally probable reservoir descriptions. These reservoir descriptions are then flow simulated to determine the range of possible outcomes. Frequency distributions of an example flow simulation output parameter, ultimate recovery, is depicted in Figure 1. If an insufficient number of relevant constraints have been imposed on the system, it is very likely that the actual ultimate recovery represents an unlikely occurrence. This situation is represented by the two frequency distributions labelled "conventional constraints." Note that depending on the inadequacies of the imposed constraints, the resulting distributions can be unrealistically broad or narrow. However, if additional constraints can be imposed, each having a significant influence on fluid flow characteristics, then the corresponding frequency distribution of outcome parameters should be greatly reduced. This is illustrated in Figure 1 by the distribution labelled "relevant constraints." Not only should the spread in possible outcomes be reduced but, as indicated by this second scenario, the actual value of ultimate recovery should fall well within the bounds of the revised frequency distribution. A more informed reservoir management decision can be made given the reduced range in possible outcomes. Thus, it becomes important to identify the spatial reservoir features which have the greatest impact on reservoir performance. Once these features have been identified, a method is required to quantify these characteristics and incorporate them as constraints within a probabilistic method. This basic approach was used previously for investigating permeability heterogeneity constraints for a five-spot waterflood pattern. The work presented here, supplemental to the previous investigation, focuses on primary production and integration of both primary- and secondary-based constraints into one conditional simulation method. P. 749^

SPE Members Abstract This paper presents a new and practical methodology to identify the flow units in a mappable geological formation. This approach uses Interwell Flow Capacity Index (IFCI) to describe the flow unit. IFCI is defined as the ratio of actual flow rates (production or injection) in any two wells in a producing unit, with the higher fluid flow rate of the two wells in the denominator. The presented approach uses core, log and production-injection data to describe the variations in reservoir connectivity. The flow units are identified and described by the use of IFCI. A flow unit is identified by the similarities between IFCI's calculated by using petrophysical and production-injection data. A Colombian (South America) sandstone, La Cira Field "C1C unit", is used to illustrate the application of IFCI to identify and describe flow units. Introduction Understanding the variations in reservoir connectivity is the key to successful implementation of improved oil recovery programs. The reservoir connectivity modeling is the basis for defining a flow unit model. A flow unit model incorporates all pertinent detailed geological. petrophysical and engineering informations available. This model provides a reservoir description that retains the complexities of reservoir architecture and variation in reservoir parameters. The flow unit model is useful in predicting production performance of secondary and tertiary recovery processes. A flow unit is defined as a lateral and vertical continuity for flow of fluids through a genetic or geological unit. Flow units are related to genetic units distribution, but do not necessarily coincide with facies boundaries. Therefore, flow units may not be areally contiguous. This paper presents a new and practical methodology to identify and describe flow units within mappable geological formations. This methodology utilizes core, log and production-injection data to describe the variations in reservoir connectivity. A flow unit is described by the use of Interwell Flow Capacity Index (IFCI). A Colombian (South America) sandstone, La Cira Field "C Zone", is used to illustrate the application of the IFCI. The C zone is a low to moderate permeability, fluvial sandstone reservoir. This zone produces from the Mugrosa Formation of Eocene-Oligocene age. An area of about 1750 acres was selected for detailed study. The study area has 70 producers and 29 injectors. In the C zone 16 major genetic units are distinguished on the basis of grain size, texture, sedimentary structures and location. P. 883^

Ravnaas, R.D.,* and Strickland, R.F.,* Cawley, Gillespie and Assocs.; Lake, L.W.,* U. of Texas; Yang, A.P.,* Texaco E and P Technology Div.; Malik, Mohammad, U. of Texas; Prezbindowski, D.R., Prezbindowski, D.R., International Petrology Research; and Mairs, Tom, Consulting Geologist Abstract A three dimensional conditional simulation model, integrating a comprehensive data base coupled with engineering, geological, and petrological studies, has significantly enhanced reservoir management capabilities in the Schneider (Buda) Field. Presented are the results of a series of conditional simulations of the Schneider (Buda) Field. The Schneider (Buda) Field is developed in a Buda reef complex. Porosity types change significantly between reservoir facies. A three dimensional conditional simulation generated using only well control points did not adequately characterize the reservoir. An intermediate two dimensional conditional simulation using additional assumptions was then generated to refine and control the final three dimensional conditional simulation. The resulting three dimensional conditional simulation delineates permeability distributions that better: represent the calculated heterogeneity, match the available core data, and model the geologic characterization, including the defined facies. The resulting distributions have been applied to a three dimensional fluid flow simulation to better predict fluid movement through the reservoir. Introduction The Schneider (Buda) Field is located in Wood County, Texas, approximately 100 miles east of Dallas near the town of Hainesville (Figure 1). Production is from Cretaceous reefal Buda Limestone discovered under the overhang of the Hainesville salt dome (Figure 2). Bounded by faults to the east and west, the updip limit is the salt dome, and the downdip limit is a tar mat/water contact. Originally a depletion drive black oil reservoir, it was undersaturated at discovery. The field was discovered by the Dan Peacock No. 1 well flowing 1,008 BOPD with a 1,000 psi flowing tubing pressure from perforations at 9,400 feet on March 4, 1988 (Figure 3). The field was unitized and a water injection project initiated in October, 1990, with three producing wells and one injection well. Production/Injection History Production/Injection History Production began in the Schneider (Buda) Field in March, 1988, with the discovery well, the Peacock No. 1. Four wells have been drilled in the field with the discovery well converted to injection in October, 1990. As of August 31, 1991, cumulative production from the field was 1,286 MSTB with recent production rates averaging a total of 1,500 BOPD for the current three producing wells (Figure 4). P. 313

SPE Members Conditional simulation is a geostatistical approach to generate multiple equiprobable distributions of a property with specified univariate and spatial attributes. These attributes are derived from sample data or inferred from external information. This paper evaluates the effectiveness of conditional simulation to describe distributions of porosity in a carbonate environment. The reservoir studied produces from a dolomitized formation of complex lithology. The conditional simulation methods tested are the simulated annealing simulation (SAS) and the sequential indicator simulation (SIS). A validation procedure is introduced to compare observed and simulated procedure is introduced to compare observed and simulated porosity distributions and to quantify the quality of porosity distributions and to quantify the quality of two- and three-dimensional conditional simulations. Results of three- dimensional validations indicate that SAS yields a good overall reproduction of observed porosity distributions. Comparisons of equiprobable two-dimensional distributions suggest that the quality of the porosity descriptions for SAS are slightly better than for SIS. These results show that conditional simulation renders descriptions which capture variations of a property due to inherent geologic complexities. The major advantages of conditional simulation over traditional methods are: it incorporates in a distribution of a property different degrees of variability for different directions, and it accounts for uncertainties in the description. Introduction Spatial distributions of reservoir properties, in particular, porosity and permeability, play an important role in porosity and permeability, play an important role in predictions of hydrocarbon reserves and fluid flow performance. predictions of hydrocarbon reserves and fluid flow performance. Logs and cores of vertical and horizontal wells, geologic models and outcrop observations and seismic data provide evidence of the irregular distribution of reservoir properties. Depending on factors such as depositional environment, the nature of the distribution of a property is often characterized by spatial variability which is manifested as anisotropy with different degrees of variation along different directions for different scales. Despite this complex nature of properties, the traditional approach has been to represent distributions by oversimplified models (e.g., layer cake models of constant properties). In many cases, such as the carbonate reservoir properties). In many cases, such as the carbonate reservoir considered in this paper, a layer cake model is not consistent with geologic conceptions of the environment. As an alternative to these simplistic and often subjective models, this paper investigates the robustness of conditional simulation methods to describe the complex variability of reservoir properties. properties. Conditional simulation can be defined as a methodology to generate equi-probable distributions of a property constrained by: conditioning data and a set of statistical parameters. Conditioning data consist of sample data of a parameters. Conditioning data consist of sample data of a property observed at wells, such as porosity or permeability property observed at wells, such as porosity or permeability derived from logs or cores. P. 329

Abstract The investigation of carbonate-ramp deposits of the upper San Andres Formation that crop out along the Algerita Escarpment, New Mexico, focused on characterizing the complex heterogeneity associated with depositional and diagenetic processes at the interwell scale. Detailed geologic mapping processes at the interwell scale. Detailed geologic mapping revealed a series of upward-shallowing parasequences, 10 to 40 ft thick, that form the geologic framework for the reservoir model. Parasequence boundaries are typically marked by tight mudstone/wackestone beds that display variable degrees of lateral continuity, ranging from several hundred feet to more than 2,500 ft and are important as potential flow barriers. Detailed permeability measurements (using both mechanical field permeability measurements (using both mechanical field permeameter and core plugs) of the grainstone rock-fabric unit in permeameter and core plugs) of the grainstone rock-fabric unit in parasequence 1 were made at scales of 1 inch, and 1, 5, 25, parasequence 1 were made at scales of 1 inch, and 1, 5, 25, and 100 ft. Within parasequence 1, distinct variability of petrophysical characteristics occurs at scales well below those of petrophysical characteristics occurs at scales well below those of interwell spacing (660 to 1,330 ft). Geostatistical analyses of permeability measurements indicate varying permeability permeability measurements indicate varying permeability correlation at the different measurement scales. In all cases, however, approximately half of the permeability variability (variance) is due to locally random heterogeneity (nugget effect). Conditional simulation of permeability in parasequence 1 indicates apparent randomness owing to the large nugget effect. Numerical waterflood simulations using conditional permeability realizations were compared with numerical simulations using geometric-mean permeability distribution and gave similar results. Simulations using small-scale heterogeneity and associated capillary-pressure phenomena will increase sweep efficiency compared with simulations using a facies-averaged permeability distribution with single capillary-pressure curves. permeability distribution with single capillary-pressure curves. Introduction The investigation of carbonate-ramp deposits of the upper San Andres Formation that crop out along the Algerita Escarpment, New Mexico, is a research element of ongoing geologic and petrophysical studies conducted at the Bureau of Economic petrophysical studies conducted at the Bureau of Economic Geology's Reservoir Characterization Research Laboratory (RCRL). The primary goal of the investigation is to develop an integrated strategy involving geological, petrophysical, geostatistical, and reservoir-simulation studies that can be used to better predict flow characteristics in analogous subsurface reservoirs. predict flow characteristics in analogous subsurface reservoirs. Geologic investigations and detailed measurements of petrophysical parameters on continuous outcrop were used to petrophysical parameters on continuous outcrop were used to determine not only the vertical distribution of the data but also their lateral distribution, which is typically lacking in subsurface studies. To characterize the complex heterogeneity associated with depositional and diagenetic processes at the interwell scale, detailed permeability data were collected within the overall geologic framework from the outcrop at Lawyer Canyon, Algerita Escarpment, New Mexico (fig. 1). Geologic mapping showed a series of upward-shallowing parasequences (10 to 40 ft thick and several thousand feet long). Parasequence boundaries are typically marked by tight mudstone/wackestone beds that display variable degrees of lateral continuity ranging from several hundred feet to more than 2,500 ft and are potentially important as flow barriers (fig. 2). Within these parasequences, distinct variability of facies and petrophysical characteristics is present at scales well below those of interwell spacing typical for their subsurface counterparts (660 to 1,330 ft). Pore types and permeability- porosity relationships can also be specific to porosity relationships can also be specific to individual parasequences. P. 287