TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractHorizontal wells are becoming a very important component in the thermal recovery of heavy oil reservoirs. The success of a cyclic steam injection project depends strongly on the selection of key parameters, such as cycle length and amount of steam injected. The numerical simulation of horizontal wells, especially under non-isothermal conditions, is computationally demanding. When optimization is combined with numerical simulation, the computing time requirement may be prohibitive and it is not guaranteed that the optimal conditions will be found.In this research, a new methodology has been developed for optimizing the cyclic steam injection process for vertical and horizontal wells. The procedure integrates oil production characterization using numerical simulation, net present value maximization through a Quasi-Newton method, and model validation/tuning. The three-stage procedure provides the optimum number and/or duration of cycles, the optimal amounts of steam to be injected in each cycle and the optimal value of the overall economic indicator.The optimization algorithm was successfully validated with published results obtained from the discrete maximum principle. The methodology was then applied to determine the optimal conditions of cyclic steam injection for a horizontal well located in Bachaquero field, Venezuela.

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractThis paper presents an analytical solution that describes transient pressure behavior of partially-penetrating wells in naturally-fractured reservoirs. The solution is obtained by combining the pseudo-steady state model for naturally fractured reservoirs with the partially-penetrating well model in homogeneous reservoirs. Verification of the analytical solution is made by comparison of pressures derived analytically with those obtained from numerical simulation. The new type curves generated from the analytical solution indicate that partial penetration and double porosity effects cause a characteristic shape of the curves at early and transition time. The type curves may be used to analyze transient well test analysis for partially-penetrating wells in naturally-fractured reservoirs at early and transition time.

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractTwo new composite, or dual-volume, wellbore models are presented. Both models describe a two-compartment wellbore coupled through a pseudosteady-state flow equation. The models differ in how the wellbore is coupled to the reservoir.The first model describes the pressure response in a well with a "leaky packer". This model may exhibit two unit-slope lines during wellbore storage, one at earliest times corresponding to the volume of the tubing, and a second one corresponding to the volume of the tubing plus the tubingcasing annulus.Gringarten noted that apparent wellbore storage coefficients for naturally fractured reservoirs are often one or two orders of magnitude higher than calculated from wellbore geometry, and attributed this effect to storage in natural fractures intersecting the wellbore. 1 The second, "nearwellbore fracture storage," model accounts for storage in the fracture system intersecting the wellbore, completion damage, and storage in the wellbore proper.Both models give pressure responses that are very similar to those exhibited by the model proposed by Fair to describe phase segregation and thermal effects. 2,3 This may explain the success of the Fair model in modeling many situations in which neither phase segregation nor thermal effects are likely to occur.The "near-wellbore fracture storage" model also gives a pressure response that is almost identical to that exhibited by a pseudosteady-state dual porosity reservoir with wellbore storage. Gringarten observed that a majority of tests in fractured reservoirs appear to exhibit pseudosteady-state rather than transient dual porosity behavior. 1 The proposed "nearwellbore fracture storage" model may account for the behavior interpreted as pseudosteady state dual porosity in many of these cases.

This paper presents the results of a full field compositional reservoir simulation study that compares the application of waterflooding and gas injection to a rich gas-condensate reservoir. Fluid (water or gas) injection is used to maintain the reservoir pressure above the dew point pressure and prevent condensate dropout in the reservoir. The simulation study shows that water injection is a viable option in producing gas-condensate reservoirs. The study shows that both water injection and gas injection result in higher recovery factors than normal depletion. Although gas injection showed higher condensate recovery factors, it may not be economical due to the required large initial investment, higher operating costs, and delay of gas sales. For the water injection to be effective, the reservoir should not be flooded completely with water. Rather, the reservoir should be blown-down before water invades the producing wells. Factors that affect the recovery of gas condensate reservoirs under water injection are highlighted. These factors include time to blow-down and imbibition relative permeability curves. Sensitivity analysis results are presented for these factors. Water injection has a major advantage over gas injection. In water injection produced gas can be immediately sold and compression cost is saved. Introduction Gas cycling has long been used to increase the recovery of condensate from gas-condensate reservoirs1–3. In this process, the rich gas is displaced with lean gas. Gas injection maintains the reservoir pressure above the dewpoint pressure of the rich gas and prevents liquid dropout in the reservoir. Lean gas also strips out some of the condensate that may develop in the reservoir. This process has proven to be efficient in increasing condensate recovery. After gas injection is stopped, the reservoir is usually produced to abandonment pressure (blowdown) to recover as much gas as possible. It is well known that the recovery factors for volumetric gas reservoirs are usually higher than those of gas reservoirs with natural water influx.4 This may be the reason that water injection, as a pressure maintenance alternative to gas injection, has not been widely used. In recent years, however, several authors5–8 have investigated water injection in gas-condensate reservoirs. They mostly discussed the theoretical aspects of the process. In this paper, we present the results of a three-dimensional compositional simulation study performed to investigate the different options in developing a rich gas-condensate reservoir. Field Background The structure of the main reservoir is an anticline cut by parallel faults, overlain by a thick shale layer. Five wells have been drilled to delineate and develop the reservoir for production. All five wells encountered the main reservoir. This sandstone contains a rich gas-condensate with an initial condensate yield of 172 STB/MMscf. Reservoir Simulation Model We constructed a three-dimensional compositional reservoir simulation model to investigate different development plans. The structure map on top of the sand and a net sand map were used to construct the reservoir simulation grid. The structure map was based on 3D seismic data calibrated with petrophysical information from the five wells. The net sand was based on petrophysical analysis of the well logs. The model had eight layers in the vertical direction. We used average porosity and permeability values for each layer. However, vertical heterogeneity existed in the model. The reservoir average porosity was around 0.16 and average permeability was around 50 md.