Method of developing a petroleum reservoir by reservoir model reconstruction

摘要

A method of developing a petroleum reservoir according to a given development scheme, from a facies map representative of the reservoir. New measurements are taken in the reservoir to better characterize it. A geostatistical simulator is selected. A set of random numbers is then identified which are provided to the geostatistical simulator to provide a facies representation identical to the initial map by inverting a random number generation algorithm of the simulator. The map is then modified to account for the new measurements, by carrying out a geostatistical simulation constrained by the measurements, with the geostatistical simulator and the set of random numbers. Finally, the reservoir development scheme is modified by accounting for the modified map, and the reservoir is developed using the modified development scheme.

主权项

1. A method of developing a petroleum reservoir according to a development scheme, from a facies map representative of the reservoir using measurements from the reservoir allowing better characterization of the reservoir, comprising:
selecting a geostatistical simulator comprising a random number generation algorithm; identifying a set of random numbers which are provided to the geostatistical simulator to provide a facies representation identical to the map, by inverting the random number generation algorithm; modifying the facies map to account for the measurements, by carrying out a geostatistical simulation constrained by the measurements, by using the geostatistical simulator and the set of random numbers; adjusting the reservoir development scheme by accounting for the modified map; and developing the reservoir using the modified development scheme; and wherein the geostatistical simulator is a simulator using a sequential indicator simulation with a map being a grid in which each cell has a facies value V, V=(V1, . . . , Vn), and a set of random numbers, P1, . . . , Pn, is identified by: i. converting the measurements to an indictor vector Vα defined with the measurements; ii. constructing a random path by visiting all the cells of the grid; and iii. in each cell i of the path: (a) estimating an occurrence probability of each facies by an indicator kriging using as conditional data the indicator vector to which previously simulated facies V<1 being a vector with the previously simulated facies before the cell i; with V<1=(V1, . . . , Vi-1) being added; (b) normalizing the occurrence probability; (c) constructing an empirical distribution function Fv1|vαv<1 associated with the occurrence probability wherein Fv1|vαv<1 is an empirical function of Vi knowing the product VαV<1; and (d) generating a random number p1 according to a uniform law in an interval └FVi|vαv<1(Vi−1),FVi|vαV<1 (Vi)┘.