Third SPE Comparative Solution Project: Gas Cycling of Retrograde Condensate Reservoirs

Abstract

Third SPE Comparative Solution Project: Gas Cycling of Retrograde Project: Gas Cycling of Retrograde Condensate Reservoirs Summary Nine companies participated in this artificial modeling study of gas cycling in a rich retrograde-gas-condensate reservoir. Surface oil rate predictions differ in the early years of cycling but agree better late in cycling. The amount of condensate precipitated near the production well and its rate of evaporation varied widely among participants. The explanation appears to be in K-value techniques used. Precomputed tables for K values produced rapid and thorough removal of condensate during later years of cycling. Equation-of-state (EOS) methods produced a stabilized condensate saturation sufficient to flow liquid during the greater part of cycling, and the condensate never completely revaporized. We do not know which prediction is more nearly correct because our PVT data did not cover the range of compositions that exists in this area of the reservoir model Introduction SPE conducted two earlier solution projects, both designed to measure the state-of-the-art simulation capability for challenging and timely modeling problems. The first project involved a three-layer black-oil simulation with project involved a three-layer black-oil simulation with gas injection into the top layer. Both constant and variable bubblepoint pressure assumptions were used. Model predictions were in fair agreement. No simulator predictions were in fair agreement. No simulator performance data (run times, timestep size, etc.) were given. performance data (run times, timestep size, etc.) were given. Seven companies participated in the project. The second project was a study of water and gas coning with a radial project was a study of water and gas coning with a radial grid and 15 layers. Authors of the project felt that unusual well rate variations and a high assumed solution GOR contributed to the difficulty of the problem. Some significant discrepancies in oil rate and pressure were obtained. Eleven companies joined in the project. For the third comparative solution project, the Committee for the Numerical Simulation Symposium sought a compositional modeling problem. Numerical comparisons of the PVT data match were considered important. Speed of the simulators was not to be of major interest. The problem we designed is the outcome of this fairly general request. Some features of interest in current production practice of pressure maintenance by gas injection production practice of pressure maintenance by gas injection are included. The results confirm the well-known trade-off between the timing of gas sales and the amount of condensate recovered. Several features of interest in a more complete examination of production from gas-condensate reservoirs are ignored. These include the effects of nearwell liquid saturation buildup on well productivity and of water encroachment and water production on hydrocarbon productivity. We did not address the role of numerical dispersion. In addition, the surface process is simplified and not representative of economical liquid recovery in typical offshore operations. We simplified the surface process to attract a larger number of participants because not all companies had facilities for simulating gas plant processing with gas recycling in their plant processing with gas recycling in their compositional simulators. Nine companies responded to the invitation for participation. Table 1 is a list of the participants in this project. participation. Table 1 is a list of the participants in this project. Participant responses were well prepared and required a Participant responses were well prepared and required a minimum of discussion. We invited all the companies to use as many components as necessary for the accurate match of the PVT data and for the simulation of gas cycling. Companies were asked to give components actually used in the reservoir model, how these components were characterized, and the match to the PVT data obtained with the components. We first outline the problem specifications, including sufficient data for others who may wish to try the problem. The pertinent PVT data are given. We show each problem. The pertinent PVT data are given. We show each participant's components, the properties of these participant's components, the properties of these components, and the basic PVT match obtained. In many cases, EOS methods were used exclusively, but in others, a combination of methods was applied. The results of the reservoir simulation are given and comparisons are shown between companies for both cycling-strategy cases. Finally, some facts regarding simulator performance are given, although this information was voluntary. Problem Statement Problem Statement The two major parts to a compositional model study are the PVT data and the reservoir grid. For the PVT data, participants were supplied with a companion set of fluid participants were supplied with a companion set of fluid analysis reports. The specification of the reservoir model is given in Tables 2 and 3 and the grid is shown in Fig. Note that the grid is 9 × 9 × 4 and symmetrical, indicating that it would be possible to simulate half the indicated grid. Most participants chose to model the full grid. Note also that the layers are homogeneous and of constant porosity, but that permeability and thickness vary among porosity, but that permeability and thickness vary among layers. JPT p. 981