Abstract
During the solution of fully implicit reservoir simulation time steps, it is often observed that the computed Newton updates may be very sparse, considering computer precision. This sparsity can be as high as 95% and can vary largely from one iteration to the next. In recent works, a mathematically sound framework was developed to predict the sparsity pattern before the full linear system is solved. The theory is restricted to general, scalar nonlinear advection-diffusion-reaction problems in multidimensional and heterogeneous settings. This theory had been applied to reduce the size of the linear systems that were computed during sequential implicit time steps for two-phase flow. The results confirmed that the linearization computations and the linear solution processes may be localized by as much as 95% while retaining the exact Newton convergence behavior and final solution. Inspired by the great success of that methodology, this work develops algorithmic extensions in order to devise localization algorithms for fully implicit coupled multicomponent problems. We propose, apply, and test a novel algorithm to resolve a system of hyperbolic equations obtained from an Equation of State–based compositional simulator. When applied to various fully implicit flow and multicomponent transport problems, involving six thermodynamic species, on the SPE 10 geological model, the observed reduction in computational effort ranges between six and 49-fold depending on the level of locality present in the system. We apply this algorithm to several injection and depletion scenarios with and without gravity and capillarity in order to investigate the adaptivity and robustness of the proposed method to the underlying heterogeneity and complexity. We demonstrate that the algorithm enables efficient and robust full-resolution fully implicit simulation without the errors introduced by adaptive discretization methods or the stability concerns of semi-implicit approaches.
Similar content being viewed by others
References
Allgower, E., Böhmer, K., Potra, F., Rheinboldt, W.: A mesh-independence principle for operator equations and their discretizations. Soc. Indus. Appl. Math. Numer. Anal. 23(1), 160–169 (1986)
Axelsson, O.: Iterative Solution Methods. Cambridge University Press (1994)
Aziz, K., Settari, A.: Petroleum Reservoir Simulation. Elsevier Applied Science (1979)
Berger, M., Colella, P.: Local adaptive mesh refinement for shock hydrodynamics. J. Comput. Phys. 82 (1), 64–84 (1989)
Cai, X., Keyes, D., Marcinkowski, L.: Non-linear additive schwarz preconditioners and application in computational fluid dynamics. Int. J. Numer. Methods Fluids 40(12), 1463–1470 (2002)
Cao, H., Tchelepi, H., Wallis, J., Yardumian, H.: Parallel scalable unstructured cpr-type linear solver for reservoir simulation. In: SPE Annual Technical Conference and Exhibition (2005)
Chen, K.: Matrix Preconditioning Techniques and Applications. Cambridge University Press (2005)
Chen, Z., Huan, G., Ma, Y.: Computational Methods for Multiphase Flows in Porous Media. Computational Science and Engineering Society of Industrial Applied Mathematics (2006)
Christie, M., Blunt, M.: Tenth spe comparative solution project: A comparison of upscaling techniques. Soc. Petroleum Eng. Reservoir Evalu. Eng. 4(4), 308–317 (2001)
Coats, K.: A note on IMPES and some IMPES-based simulation models. Soc. Petroleum Eng. 5, 245–251 (2000)
Dawson, C., Klíe, H., Wheeler, M., Woodward, C.: A parallel, implicit, cell-centered method for two-phase flow with a preconditioned Newton–Krylov solver. Comput. Geosci. 1(3–4), 215–249 (1997)
Deuflhard, P.: Newton Methods for Nonlinear Problems: Affine Invariance and Adaptive Algorithms, vol. 35. Springer (2011)
Engstler, C., Lubich, C.: Multirate extrapolation methods for differential equations with different time scales. Computing 58(2), 173–185 (1997)
Gear, C., Wells, D.: Multirate linear multistep methods. BIT Numer. Math. 24(4), 484–502 (1984)
Gerritsen, M., Lambers, J.: Integration of local–global upscaling and grid adaptivity for simulation of subsurface flow in heterogeneous formations. Comput. Geosci. 12(2), 193–208 (2008)
Hornung, R., Trangenstein, J.: Adaptive mesh refinement and multilevel iteration for flow in porous media. J. Comput. Phys. 136(2), 522–545 (1997)
Hwang, F., Cai, X.: A parallel nonlinear additive Schwarz preconditioned inexact newton algorithm for incompressible Navier–Stokes equations. J. Comput. Phys. 204(2), 666–691 (2005)
Kelley, C., Sachs, E.: Mesh independence of newton-like methods for infinite dimensional problems. J. Integral Equ. Appl. 3(4), 549–573 (1991)
Killough, J.E., Kossack, C.A.: Fifth comparative solution project: Evaluation of miscible flood simulators. In: SPE Symposium on Reservoir Simulation (1987)
Lacroix, S., Vassilevski, Y., Wheeler, M.: Decoupling preconditioners in the implicit parallel accurate reservoir simulator (ipars). Numer. Linear Algebra Appl. 8(8), 537–549 (2003)
Moncorgé, A., Tchelepi, H., Jenny, P.: Sequential fully implicit formulation for compositional simulation using natural variables. Journal of Computational Physics (To appear). arXiv:1804.09751 (2018)
Møyner, O., Tchelepi, H.: A mass-conservative sequential implicit multiscale method for isothermal equation of state compositional problems. SPE Journal (To appear) (2018)
Müller, S., Stiriba, Y.: Fully adaptive multiscale schemes for conservation laws employing locally varying time stepping. J. Sci. Comput. 30(3), 493–531 (2007)
Sheth, S., Younis, R.: Localized computation of newton updates in fully-implicit two-phase flow simulation. Procedia Comput. Sci. 80, 1392–1403 (2016)
Sheth, S., Younis, R.: Localized linear systems in sequential implicit simulation of two-phase flow and transport. Soc. Petroleum Eng. J. 22(05), 1542–1569 (2017)
Sheth, S., Younis, R.: Localized solvers for general full-resolution implicit reservoir simulation. In: SPE Reservoir Simulation Conference (2017)
Skogestad, J., Keilegavlen, E., Nordbotten, J.: Domain decomposition strategies for nonlinear flow problems in porous media. J. Comput. Phys. 234, 439–451 (2012)
Sun, S., Keyes, D., Liu, L.: Fully implicit two-phase reservoir simulation with the additive schwarz preconditioned inexact newton method. In: Society of Petroleum Engineers Reservoir Characterisation and Simulation Conference and Exhibition (2013)
Trangenstein, J.: Multi-scale iterative techniques and adaptive mesh refinement for flow in porous media. Adv. Water Resour. 25(8), 1175–1213 (2002)
Wallis, J., Kendall, R., Little, T., et al.: Constrained residual acceleration of conjugate residual methods. In: SPE-13536-MS Inproceedings of the SPE Reservoir Simulation Symposium (1985)
Younis, R.: A sharp analytical bound on the spatiotemporal locality in general two-phase flow and transport phenomena. Procedia Comput. Sci. 18, 473–480 (2013)
Younis, R., Tchelepi, H., Aziz, K.: Adaptively localized continuation-newton method–nonlinear solvers that converge all the time. SPE J. 15(02), 526–544 (2010)
Acknowledgements
The authors would like to thank TOTAL management for permission to publish this work. The authors also acknowledge the members of the Future Reservoir Simulation Systems and Technology program for their partial financial support.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Mathematical Development Including Capillarity
Appendix: Mathematical Development Including Capillarity
Starting with Eq. 6 and assuming pressure of the first phase to be the primary variable, we obtain
where Pcp1 = Pp − P1 is the capillary pressure between phase p and the primary phase 1. Introducing notation
Equation 1 is reduced to
Following the derivation in the previous section, the infinite-dimensional iteration, in three (Cartesian) spatial dimensions, can be written as
Expanding the last term and grouping the second, third, and fourth term gives
Expanding the last term once again gives
The final equation grouped by the derivatives of \(\delta _{\infty ,d}\) is written as
Substituting
for all c,d ∈{1,...,nc} in Eq. 10, gives
Assuming that the residual can be projected onto a discrete Banach space as shown in Eq. 17, Eq. 44 can be written out for each control volume as
The boundary conditions used to solve the continuous Newton iteration in this case are (1) \(\delta _{\infty , {\Omega }} = \delta _{\Omega } \forall x\in {\Omega }\), given by the second sub-equation in Eq. 45 and (2) \(\delta _{\infty , {\Omega }}\) is bounded at \(x = \infty \). Similar to the derivation presented in the previous section, the next step towards solving (45) is to homogenize the variable coefficients. After the first step of homogenization, we obtain
The coefficient of \(\delta ^{*}_{\Omega }\) in the second part of Eq. 46 is not homogenized as it is constant and evaluated at the centroid of the control volume (see assumption 3 in the next section). Finally, radially symmetric solutions, \(\tilde \delta ^{*}_{\infty , {\Omega }}(r)\), can be obtained by homogenizing the coefficient matrices containing vector entries. Homogenizing the coefficient of \(\frac {\partial \delta ^{*}_{\infty ,{\Omega }}}{\partial x_{i}}\) in Eq. 46 and writing
we get
The solution of this equation depends on the sign of \({4 \tilde A \tilde D + \tilde C^{2}}\). Real and distinct roots are obtained for \({4 \tilde A \tilde D + \tilde C^{2}} > 0\). Cases when \({4 \tilde A \tilde D + \tilde C^{2}} < 0\), complex conjugates are obtained and \({4 \tilde A \tilde D + \tilde C^{2}} = 0\) results in repeated real roots. In this work, it is assumed that \({4 \tilde A \tilde D + \tilde C^{2}} > 0\). This condition is enforced by taking the norm of the eigenvalues while computing the local coefficient matrices. Further research is required to study the behaviour of the solutions obtained when this condition is not satisfied or when the local matrices have complex eigenvalues. The solution then becomes
Finally, following the derivation in the previous section, the conservative estimate, \(\hat \delta ^{*}_{\infty , {\Omega }}\), such that \(\left \Vert {\hat \delta ^{*}_{\infty , {\Omega }}}\right \Vert \ge \left \Vert {\tilde \delta ^{*}_{\infty , {\Omega }}}\right \Vert \), is given by
where \(\lambda _{\min \limits }\) is the norm of the minimum eigenvalue of \(\frac {\sqrt {4 \tilde A \tilde D + \tilde C^{2}} + \tilde C}{2\tilde D}\) in Eq. 51. The radius of influence is calculated by Eq. 25.
Rights and permissions
About this article
Cite this article
Sheth, S., Moncorgé, A. & Younis, R. Localized linear systems for fully implicit simulation of multiphase multicomponent flow in porous media. Comput Geosci 24, 743–759 (2020). https://doi.org/10.1007/s10596-019-09840-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10596-019-09840-9