×

Application of the nonuniform fast Fourier transform to the direct numerical simulation of two-way coupled particle laden flows. (English) Zbl 1416.76074

Hernández, S. (ed.) et al., Advances in fluid mechanics XII. Proceedings of the 12th international conference on advances in fluid mechanics, AFM, Ljubljana, Slovenia, July 10–12, 2018. Southampton: WIT Press. WIT Trans. Eng. Sci. 120, 237-248 (2019).
Summary: We present the application of the Nonuniform Fast Fourier Transform (NUFFT) to the pseudo-spectral Eulerian-Lagrangian direct numerical simulation of particle-laden flows. In the two-way coupling regime, when the particle feedback on the flow is taken into account, a spectral method requires not only the interpolation of the flow fields at particle positions, but also the Fourier representation of the particle back-reaction on the flow fields on a regular grid. Even though the direct B-spline interpolation is a well-established tool, to the best of our knowledge the reverse projection scheme has never been used, replaced by less accurate linear reverse interpolation or Gaussian regularization. We propose to compute the particle momentum and temperature feedback on the flow by means of the forward NUFFT, while the backward NUFFT is used to perform the B-spline interpolation. Since the backward and forward transformations are symmetric and the (non local) convolution computed in physical space is removed in Fourier space, this procedure satisfies all constraints for a consistent interpolation scheme, and allows an efficient implementation of high-order interpolations. The resulting method is applied to the direct numerical simulation of a forced and isotropic turbulent flow with different particle Stokes numbers in the two-way coupling regime. A marked multifractal scaling is observed in the particle statistics, which implies that the feedback from the particles on the fields is far from being analytic and therefore only high-order methods, like the one here proposed, can provide an accurate representation.
For the entire collection see [Zbl 1410.76011].

MSC:

76F65 Direct numerical and large eddy simulation of turbulence
65T50 Numerical methods for discrete and fast Fourier transforms
76T20 Suspensions
PDFBibTeX XMLCite
Full Text: Link

References:

[1] Zonta, F., Marchioli, C. & Soldati, A., Direct numerical simulation of turbulent heat transfer modulation in micro-dispersed channel flow. Acta Mech,195(1-4), pp. 305- 326, 2008. · Zbl 1144.76025
[2] Ireland, P.J., Bragg, A.D. & Collins, L.R., The effect of Reynolds number on inertial particle dynamics in isotropic turbulence. Part 1. Simulations without gravitational effects. J Fluid Mech,796, pp. 617-658, 2016. · Zbl 1462.76077
[3] Toschi, F. & Bodenschatz, E., Lagrangian properties of particles in turbulence. Annu Rev Fluid Mech,41(1), pp. 375-404, 2009. · Zbl 1157.76020
[4] Falkovich, G. & Pumir, A., Sling effect in collisions of water droplets in turbulent clouds. J Atmos Sci,64(12), pp. 4497-4505, 2007.
[5] Elghobashi, S., Particle-laden turbulent flows: direct simulation and closure models. Appl Sci Res,48(3), pp. 301-314, 1991. · Zbl 0729.76609
[6] Yeung, P. & Pope, S., An algorithm for tracking fluid particles in numerical simulations of homogeneous turbulence. J Comp Phys,79(2), pp. 373-416, 1988. · Zbl 0655.76042
[7] Hinsberg, van, M., Thije Boonkkamp, ten, J., Toschi, F. & Clercx, H., On the efficiency and accuracy of interpolation methods for spectral codes. SIAM J Sci Comput,34(4), pp. B479-B498, 2012. · Zbl 1250.65165
[8] Gualtieri, P., Picano, F., Sardina, G. & Casciola, C.M., Exact regularized point particle method for multiphase flows in the two-way coupling regime. J Fluid Mech,773, pp. 520-561, 2015. · Zbl 1331.76123
[9] Eggers, H., Knopp, T. & Potts, D., Field inhomogeneity correction based on gridding reconstruction for magnetic resonance imaging. IEEE T Med Imaging,26(3), pp. 374- 384, 2007.
[10] Toukmaji, A.Y. & Board, J.A., Ewald summation techniques in perspective: a survey. Comp Phys Comm,95(2), pp. 73-92, 1996. · Zbl 0923.65090
[11] Pippig, M. & Potts, D., Parallel three-dimensional nonequispaced fast fourier transforms and their application to particle simulation. SIAM J Sci Comput,35(4), pp. C411-C437, 2013. · Zbl 1362.65141
[12] Dutt, A. & Rokhlin, V., Fast fourier transforms for nonequispaced data. SIAM J Sci Comput,14(6), pp. 1368-1393, 1993. · Zbl 0791.65108
[13] Greengard, L. & Lee, J., Accelerating the nonuniform fast fourier transform. SIAM Review,46(3), pp. 443-454, 2004. · Zbl 1064.65156
[14] Sundaram, S. & Collins, L.R., Numerical considerations in simulating a turbulent suspension of finite-volume particles. J Comp Phys,124(2), pp. 337-350, 1996. · Zbl 0849.76054
[15] Kumar, B., Schumacher, J. & Shaw, R.A., Lagrangian mixing dynamics at the cloudy– clear air interface. J Atmos Sci,71(7), pp. 2564-2580, 2014.
[16] Zamansky, R., Coletti, F., Massot, M. & Mani, A., Turbulent thermal convection driven by heated inertial particles. J Fluid Mech,809, pp. 390-437, 2016. · Zbl 1383.76517
[17] Pan, Y. & Banerjee, S., Numerical simulation of particle interactions with wall turbulence. Phys Fluids,8(10), pp. 2733-2755, 1996.
[18] Maxey, M.R. & Riley, J.J., Equation of motion for a small rigid sphere in a nonuniform flow. Phys Fluids,26(4), pp. 883-889, 1983. · Zbl 0538.76031
[19] Canuto, C., Hussaini, M.Y., Quarteroni, A. & Zang, T.A., Spectral Methods in Fluid Mechanics, Springer, 1988. · Zbl 0658.76001
[20] Hochbruck, M. & Ostermann, A., Exponential integrators. Acta Numer,19, pp. 209- 286, 2010. · Zbl 1242.65109
[21] Beylkin, G., On the fast fourier transform of functions with singularities. Appl Computat Harmonic A,2(4), pp. 363-381, 1995. · Zbl 0838.65142
[22] Pekurovsky, D., P3DFFT: a framework for parallel computations of Fourier transforms in three dimensions. SIAM J Sci Comput,34(4), pp. C192-C209, 2012. · Zbl 1253.65205
[23] van Hinsberg, M.A.T., Boonkkamp, J.H.M.t.T., Toschi, F. & Clercx, H.J.H., Optimal interpolation schemes for particle tracking in turbulence. Phys Rev E,87, p. 043307, 2013.
[24] Bragg, A.D. & Collins, L.R., New insights from comparing statistical theories for inertial particles in turbulence: I. Spatial distribution of particles. New J Phys,16(5), p. 055013, 2014.
[25] -[5]. The ratio, c/h, increases the rate of convergence by decreasing h, the average fill distance between pairs of points (h-refinement) or increasing the average value of the shape parameter, c, (c-refinement). The basic question is for a given application is whether hrefinement or c-refinement is more cost effective to achieve the target accuracy with the least computational effort. The answer to this question is not definitive but depends upon the nature of the application and the available resources. In general, as system of equations becomes larger, the higher is the probability that rounding-errors will severely contaminate the solution. A measure of this severity is the condition number, (A), of a system of equations relative to the inverse machine epsilon, 1/M. Whenever (A) > 1/M the solution is worthless. While it is impossible to construct an ideal Platonic computer with infinite precision, infinite memory capacity that operates infinitely fast, there are some procedures that can be adopted to control the damage caused by the accumulation of round-off errors on a finite precision computer. Increasing the number of digits of precision requires both sacrificing memory and increasing the processing time per unit operation. This sacrifice is worthwhile if the total processing time to achieve target accuracy is smaller than the total processing time with single or double precision. 2 TREATING ILL-CONDITING
[26] -[19]. Assume the problem is continuous, and the computational domain is arbitrarily decomposed as a union of smaller subdomains containing an equal number of data centers (approximately). Just as with finite element methods, the larger the number of centers within a subdomain, the larger will the condition number be, either with CS-RBFs or GS-RBFs. The IE or PDE in each subdomain has its “boundary” and interior conditions. In each subdomain, the set of equations is solved on separate independent processors in parallel. Then the solutions in each subdomain are iteratively blended to enforce continuity of the function and its normal and tangential derivatives. Typically, only 2-3 iterations are required to obtain satisfactory blending convergence. However, the parallelism is somewhat different if compact or global support is used; this will be discussed later. If the problem being solved is fairly homogeneous spatially, then increasing the shape parameters or c-refinement while increasing the arithmetic precision is vastly more efficient since the total number of data centers can be reduced by many orders of magnitude [20]. However, some problems may have so much fine scale structure that h-refinement, rather than c-refinement, is required and slower converging compactly supported RBF methods may be required for a given available computer. Common sense is required regarding how to approach a given problem using the available resources. Very promising is the development of global optimization methods for solving systems of equations [21]-[28]. This is a very broad field of research and development that has the potential to solve large systems of both linear and nonlinear equations without the need for LU methods that exacerbates ill-conditioning. Global optimization can be either search procedures, stochastic procedures, or a hybrid mixture of both. 3 RECOMMENDED STRATEGIES
[27] Kansa, E.J., Multiquadrics: A scattered data approximation scheme with applications to computational fluid dynamics I: Surface approximations and partial derivative estimates. Comput. Math. Applic., 19(8/9), pp. 127-145, 1990. · Zbl 0692.76003
[28] Kansa, E.J., Multiquadrics: A scattered data approximation scheme with applications to computational fluid dynamics II: Solutions to parabolic, hyperbolic, and elliptic partial differential equations. Comput. Math. Applic., 19(8/9), pp. 147-161, 1990. · Zbl 0850.76048
[29] Madych, W.R. & Nelson, S.A., Multivariate interpolation and conditionally positive definite functions ii. Math. Comp., 54, pp. 211-230, 1990. · Zbl 0859.41004
[30] Madych, W.R., Miscellaneous error bounds for multiquadric and related interpolators. Com-put. Math. Appl.,24(12), pp. 121-138, 1992. · Zbl 0766.41003
[31] Madych, W.R. & Nelson, S.A., Bounds on multivariate polynomials and exponential error estimates for multiquadric interpolation. J. Approx. Theo., 60, pp. 94-114, 1992. · Zbl 0764.41003
[32] Tikhonov, A.N., On the stability of inverse problems. Dokl.Akad.i Nauk SSSR, 39(5), pp. 195-198, 1943.
[33] Tikhonov, A.N. & Arsenin, V.Y., Solution of Ill-posed Problems, Winston & Sons: Washington, 1977. · Zbl 0354.65028
[34] Tikhonov, N., Goncharsky, A.V., Stepanov, V.V. & Yagola, A.G., Numerical Methods for the Solution of Ill-Posed Problems, Kluwer Academic Publishers, 1995. · Zbl 0831.65059
[35] Advanpix LLC., Multiprecision computing toolbox for MATLAB, Yokohama, Japan, 2008-2017. www.advanpix.com.
[36] Grcar, J.F., Operator coefficient methods for linear equations. Technical Report SAND89-8691, Sandia National Labs, 1989.
[37] Fasshauer, G.E., Hickernell, F.J. & Wozniakowski, H., Rate of convergence and tractability of the radial function approximation problem. SIAM J. Numer. Anal., 50(1), pp. 247-271, 2012. · Zbl 1243.65025
[38] Fasshauer, G.E. & McCourt, M.J., Stable evaluation of Gaussian RBF interpellants. SIAM J. Scient. Comput., 34(2), pp. A737-A762, 2012.
[39] Hon, Y.-C. & Wu, Z., Additive Schwarz domain decomposition with radial basis approximation. Int. J.Appl. Math. Stat.,4, pp. 81-98, 2002. · Zbl 1051.65121
[40] Li, J. & Hon, Y.-C., Domain decomposition for radial basis meshless methods. Num. Meth. PDEs,20, pp. 450-462, 2004. · Zbl 1048.65124
[41] Ingber, M.A., Chen, C.-S. & Tanski, A., A mesh free approach using radial basis functions and parallel domain decomposition for solving three-dimensional diffusion equations. Int. J. Num. Meth. Eng., 60, pp. 2183-2201, 2004. · Zbl 1178.76276
[42] Duan, Y., Tang, P.F., Huang, T.Z. & Lai, S.J., Coupling projection domain decomposition method and Kansa’s method in electrostatic problems. Comput. Phys. Commun., 180, pp. 200-214, 2009. · Zbl 1198.78013
[43] Beatson, R.K., Light, W.A. & Billings, S., Fast solution of the radial basis function interpolation equations: Domain decomposition methods. SIAM J. Sci. Comput., 22, pp. 1717-1740, 2006. · Zbl 0982.65015
[44] Hernandez-Rosales, A. & Power, H., Non-overlapping domain decomposition algorithm for the Hermite radial basis function meshless collocation approach:
[45] Cavoretto, R., De Rossi, A., Donatelli, M. & Serra-Capizzano, S., Spectral analysis and preconditioning techniques for radial basis function collocation matrices. Numer. Lin. Alg. Appl., 19, pp. 31-52, 2012. · Zbl 1274.65308
[46] Galperin, E.A. & Kansa, E.J., Application of global optimization and radial basis functions to numerical solutions of weakly singular Volterra integral equations. Comput. Math. Appl., 43(3-5), pp. 491-499, 2002. · Zbl 0999.65150
[47] Skalna, I., A global optimization method for solving parametric linear systems whose input data are rational functions of interval parameters, Lecture Notes in Computer Science, 6068, Springer: Berlin, Heidelberg. ISBN: 978-3-642-14402-8.
[48] Mafteiu-Scai, L.O. & Mafteiu-Scai, E.J., Solving linear systems of equations using a memetic algorithm. Int. J. Comput. Applic., 58, pp. 16-22, 2012. · Zbl 1349.65171
[49] Taheri, S. & Mammadov, M., Solving systems of nonlinear equations using a globally convergent optimization algorithm. Global J. Tech. & Optim.,3, pp. 132-138, 2012.
[50] Kaeck, J.E., Constrained global optimization with radial basis functions. Research Report MdH-IMa-2004, Department of Mathematics and Physics, Malardalen University, Sweden.
[51] Bjorkman, M. & Holmstroem, K., Global optimization of costly nonconvex functions using radial basis functions. Optim. & Eng., 1, pp. 373-397, 2000. · Zbl 1035.90061
[52] Gutmann, H.M., A radial basis function method for global optimization. J. Global Optim.,19, pp. 201-227, 2001. · Zbl 0972.90055
[53] Le Thi, H.A., Vaz, A.I.F. & Vicente, L.N., Optimizing radial basis functions by DC programming and its use in direct search for global derivative-free optimization. TOP, 20, pp. 190-214, 2012. · Zbl 1267.90110
[54] Jakobsson, P., Patriksson, M., Rudholm, J. & Wojciechowski, A., A method for simulation based optimization using radial basis functions. Opt. & Eng.,11, pp. 501- 532, 2010. · Zbl 1243.65068
[55] Higham, N.J., Multi-precision world. SIAM News, pp. 2-3, Oct. 2017.
[56] The GNU
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.