11/14/2023 RESERVOIR SIMULATION Hussein Y. Ali Basrah University for Oil and Gas ď‚·Explicit Formulation LECTURE SEVEN Stability Analysis Truncation Error Consistency and Convergence 1 11/14/2023 PART 3 EXPLICIT FORMULATION INTRODUCTION • The forward-difference approximation to the flow equation results in an explicit calculation procedure for the new-time-level pressures (designated n+ 1 in the finitedifference equations). • Solving the forward-difference equation (Eq.9) for the unknown quantity , expression: yields the 2 11/14/2023 INTRODUCTION • All terms on the right side of the previous equation are known because all pressures appearing on this side are at the known (old) time level,n. • In this equation, the pressures at the new-time level can be obtained explicitly by use of these known pressures. EXAMPLE 1 • For the 1D, block-centered grid shown in Fig. below, determine the pressure distribution during the first year of production. The initial reservoir pressure is 6,000 psia. The rock and fluid properties for this problem are Δx = 1,000 ft, Δ y= 1,000 ft, Δz=75 ft, B = 1 RB/STB, c = 3.5 x 10^(-6) psi^(-1) , kx= 15 md, ø = 0 .18, µ =10 cp, and = 1 RB/STB. Use timestep sizes of Δt= 10, 15, and 30 days. Assume B acts as a constant within the pressure range of interest. 3 11/14/2023 EXAMPLE 2 • This example is identical to example 1, except that the no flow boundary on the left side of the reservoir is replaced by a constant-pressure boundary. Use the same rock and fluid properties as in example 1. Although this is a block-centered grid system, assume that the pressure in Grid block 1 is equal to the boundary pressure. se a timestep of 15 days. STABILITY ANALYSIS • Stability is a property that describes the capacity of a small error to propagate and grow with subsequent calculations. • Explicit method is only stable (errors do not propagate over time) if: Where • There are several procedures for analyzing the stability of a given finite difference approximation such as Fourier Series and Matrix methods. 4 11/14/2023 CONSISTENCY AND CONVERGENCE • Consistency: The discretization of a differential equation (e.g. by finite difference approximation) should become identical to the original PDE as the mesh size and the size of the time interval approach zero( Tr. Error should vanish). • Convergence: a numerical solution must convergent to the solution of the PDE when Δx and Δt are decreased in size. • Generally: consistency + stability = convergence TRUNCATION ERROR • Its the error that introduced when a continuous PDE replaced with a finite-difference approximation through Taylor series expansion • These errors may be introduced at a certain stage in the computations which grow uncontrollably to dominate the solution. • The problems we need to address are the magnitude of the truncation error and how can we increase the accuracy of an approximation by decreasing the magnitude of this error. 5 11/14/2023 TRUNCATION ERROR • the local truncation error or local discretization error is the deviation of the PDE from its corresponding finite-difference approximation at a given point in space and at a given instant in the time. • Where EXAMPLE.3 • Determine the local truncation error of the explicit finite-difference approximation to 6 11/14/2023 THANK YOU 7