This paper presents methods for evaluating constantvolume depletion (CVD) data obtained from experimental analyses of gas condensates and volatile oils. Theoretical and practical developments are supported by experimental data from a North Sea gas-condensate fluid.The three major contributions of the work are: (1) presentation of material-balance equations to calculate fluid properties from measured CVD data, (2) a simple method for calculating black-oil formation volume factors and solution GOR's using material-balance results and a separator flash program, and (3) investigation of the Peng-Robinson (PR) equation of state (EOS) as a tool for matching measured PVT data and studying vaporlliquid phase behavior during CVD.• Now with Esso E&P Norway Inc.

This paper presents results of analyzing constant volume depletion dataobtained from experimental analyses of gas condensates and volatile oils. Theoretical and practical developments are supported by analyses of experimental data from two North Sea condensate reservoirs. The three major contributions of this work are:presentation of materialbalance equations used to calculate fluid (particularly liquid) properties from measured constant volume depletion dataa simple method for calculating"black oil" formation volume factors and solution gas-oil ratios for volatilesystems using material balance results and a separator flash program, andinvestigation of the Peng-Robinson equation of state as a tool for matching measured PVT data and studying vapor-liquid equilibria phenomena during constant volume depletion. The main example presented is a rich gas condensate whose measured, calculated and simulated phase behavior are fully documented in tables and figures. Complete description of the heptanes-plus fraction is also included so that other engineers can check, modify and hopefully improve fluidcharacterization using the Peng-Robinson (or any other) equation of state. INTRODUCTION Constant volume depletion (CVD) experiments are performed on gas condensateand volatile oil fluids to simulate reservoir depletion performance and compositional variation. Resulting data can be used in a variety of reservoir engineering calculations, among the most useful being material balancecalculations, generating "black oil" PVT properties and more recently, the tuning of empirical equations of state. All of these applications are addressed in the present work.

Summary The impact on simulated waterflood performance of adopting a" discrete" and simplified flow-unit representation of reservoir heterogeneity, rather than a conventional "smooth" model based on contouring of petrophysical zonal averages, was assessed with a synthetic model of a heterogeneous barrier-bar sand body. On the basis of comparative simulations, the contour models did not generate consistently optimistic predictions of reservoir performance (compared with their flow-unitcounterparts). performance (compared with their flow-unit counterparts).Individual well behavior, however, varied significantly according to model type. Introduction Production experience from mature fields in the North Sea (and Production experience from mature fields in the North Sea (and elsewhere) has shown that reservoir simulation studies often have produced overly optimistic predictions of the timing of water produced overly optimistic predictions of the timing of water breakthrough and ultimate recovery. It is now generally agreed that this inaccuracy is more a consequence of inadequate representations of reservoir heterogeneity than fundamental inaccuracies in the differential equations used to describe fluid flow. Petrophysical reservoir description models, which serve as the basic inputs to reservoir simulators, are conventionally produced by subdividing the reservoir interval into a number of field wide layers (or zones)and contouring the petrophysical zonal averages between the wells. The interpolation can be computed or hand drawn by a geologist. Regardless of how it is done, the result is a model characterized by smooth, gradual changes in reservoir properties. Many reservoir types, however, include abrupt lateral properties. Many reservoir types, however, include abrupt lateral changes between geological facies. These abrupt changes are not captured by a contouring approach. More recently, an alternative reservoir description methodology has emerged. It involves subdividing reservoirs into a 3D mosaic of discrete volumes, or flow units, each mutually distinguishable by its geological characteristics and its ability to transmit fluids. In contrast to contoured models, flow-unit models are inherently characterized by abruptlateral changes in petrophysical properties at flow-unit boundaries. A flow-unit petrophysical properties at flow-unit boundaries. A flow-unit approach currently is finding increased use, especially in stochastic heterogeneity modeling where discrete stochastic models are used to describe large-scale reservoir heterogeneity. The purposes of this study were three-fold. 1. To assess the effects of using (discrete) flow-unit vs. (smooth)contoured reservoir descriptions on simulated reservoir behavior during waterflooding of a heterogeneous reservoir. 2. To identify situations where the contouring approach may be unreliable. 3. To gain some idea of the combined effects of reservoir heterogeneity and fluid-flow processes on reservoir performance. A synthetic and simplified geological reservoir model was constructed as the basis for experimentation. The model, which was used as a controlled laboratory, was then converted to a 3D mosaic of interlocking flow units through the assignment of appropriate petrophysical values. Permeability anisotropy within the flow units petrophysical values. Permeability an isotropy within the flow units and the degree of large-scale permeability heterogeneity between the flow units were altered systematically by varying the horizontal and vertical permeability values. In this manner, the synthetic reservoir models created ranged from moderately anisotropic and relatively homogeneous to strongly anisotropic and relatively heterogeneous. A hypothetical waterflood development scenario was constructed and subsequently applied to four areas (orelements) within the overall geological model. Reservoir performance for each of the element models was simulated in experimental pairs for each of the chosen levels of permeability heterogeneity and anisotropy; one for a flow-unitrepresentation and one for a corresponding contoured representation (based on data from four hypothetical production and injection wells). production and injection wells). The simulation results were used to evaluate the effects of simplifying heterogeneity through contouring. The flow-unit models served as standards of comparison, with the assumption that they provide more correct representations of the reservoirs' real provide more correct representations of the reservoirs' real heterogeneity. Synthetic Facies Architecture Model A regressive, barrier-bar sand-body complex, bounded above and below by transgressive marine mudstones, was selected because of the widespread occurrence of such reservoirs in the Norwegian continental shelf. Barrier-bar sand bodies are laterally and vertically continuous yet commonly contain a complex mosaic of internal facies heterogeneities (cf., the "jigsawpuzzle" variety of Ref. 13). An extensive literature review was undertaken to provide data concerning typical facies dimensions and architectures, as well as the external dimensions of barrier bars. From this review, the synthetic model was designed to cover a 25-km 2 area [5 × 5 km] 25 m thick. These dimentological framework of the model comprises five laterally continuous facies belts that migrated seaward during progradation of the barrier bar (Fig.1a). The model is progradation of the barrier bar (Fig. 1a). The model is represented by 5 × 5-m-thick grid block layers (Fig. 1b). Note that simulation layers cross cut the inclined facies belt boundaries, thereby providing an approximate chrono-stratigraphic subdivision of the reservoir. Heterogeneity within each layer is defined by 13 facies (Table 1) in the various facies belts. The 3D heterogeneity model was produced by hand drawing the position of the facies within each of produced by hand drawing the position of the facies within each of the five layers (Fig. 2). The main reservoir sand body comprises a coarsening-upward "beach-face" sequence including four facies types:an offshore transition zone, lower shore face, upper shore face, and foreshore. Tidal inlets, which cut through the beach face from the foreshore down to a few meters above the base of the lower shore face, switched positions more or less randomly as the beach face prograded. Backbarrier sandstones include floodtidal deltas and prograded. Backbarrier sandstones include flood tidal delta sand wash over fans deposited along the seaward margin of the back-barrier lagoon. The lagoonal facies define the landward termination of the main barrierbar sand body. They also provide an impermeable lateral stratigraphicseal between the barrier island and the delta plain association. Heterogeneity and Anisotropy Parameterization It was then assumed that, at a gridblock scale, each facies could be characterized (approximately) by constant petrophysical properties (which could be altered systematically for each properties (which could be altered systematically for each sensitivity run) and therefore could be regarded as a flow unit. SPERE P. 27