Optimal Moment Sets for Multivariate Direct Quadrature Method of Moments

Abstract

The direct quadrature method of moments (DQMOM) can be employed to close population balance equations (PBEs) governing a wide class of multivariate number density functions (NDFs). Such equations occur over a vast range of scientific applications, including aerosol science, kinetic theory, multiphase flows, turbulence modeling, and control theory, to name just a few. As the name implies, DQMOM uses quadrature weights and abscissas to approximate the moments of the NDF, and the number of quadrature nodes determines the accuracy of the closure. For nondegenerate univariate cases (i.e., a sufficiently smooth NDF), the N weights and N abscissas are uniquely determined by the first 2N non-negative integer moments of the NDF. Moreover, an efficient product-difference algorithm exists to compute the weights and abscissas from the moments. In contrast, for a d-dimensional NDF, a total of (1 + d)N multivariate moments are required to determine the weights and abscissas, and poor choices for the moment set can lead to nonunique abscissas and even negative weights. In this work, it is demonstrated that optimal moment sets exist for multivariate DQMOM when N ) nd quadrature nodes are employed to represent a d-dimensional NDF with n ) 1-3 and d ) 1-3. Moreover, this choice is independent of the source terms in the PBE governing the time evolution of the NDF. A multivariate Fokker-Planck equation is used to illustrate the numerical properties of the method for d ) 3 with n ) 2 and 3. The direct quadrature method of moments (DQMOM) can be employed to close population balance equations (PBEs) governing a wide class of multivariate number density functions (NDFs). Such equations occur over a vast range of scientific applications, including aerosol science, kinetic theory, multiphase flows, turbulence modeling, and control theory, to name just a few. As the name implies, DQMOM uses quadrature weights and abscissas to approximate the moments of the NDF, and the number of quadrature nodes determines the accuracy of the closure. For nondegenerate univariate cases (i.e., a sufficiently smooth NDF), the N weights and N abscissas are uniquely determined by the first 2N non-negative integer moments of the NDF. Moreover, an efficient product-difference algorithm exists to compute the weights and abscissas from the moments. In contrast, for a d-dimensional NDF, a total of (1 + d)N multivariate moments are required to determine the weights and abscissas, and poor choices for the moment set can lead to nonunique abscissas and even negative weights. In this work, it is demonstrated that optimal moment sets exist for multivariate DQMOM when N ) n d quadrature nodes are employed to represent a d-dimensional NDF with n ) 1-3 and d ) 1-3. Moreover, this choice is independent of the source terms in the PBE governing the time evolution of the NDF. A multivariate Fokker-Planck equation is used to illustrate the numerical properties of the method for d ) 3 with n ) 2 and 3. Disciplines Biological Engineering | Chemical Engineering

A Quadrature-Based Kinetic Model for Dilute Non-Isothermal Granular Flows

“…However, for some cases (e.g., infinite Knudsen number with walls at different temperatures), the tensor product formulation can be overly restrictive, leading to negative weights. In order to overcome this shortcoming, we have recently developed less restrictive moment-inversion algorithms using conditional moments [39] based on the "optimal" moments sets reported in [20], for which non-negative weights are guaranteed for any optimal set of realizable moments. Results using this conditional quadrature method of moments (CQMOM) will be reported elsewhere.…”

“…Once the fluxes G w,in and G w,out are known, the temporary quadrature weights at the wall are rescaled to satisfy the continuity constraint: 20) and the moments are recomputed from (3.3) using the updated set of weights and abscissas. Note that this procedure is equivalent to representing the equilibrium wall distribution function by…”

Numerical simulations of two-component granulation: Comparison of three methods

“…In the direct quadrature method of moments (DQMOM) the population balance is converted into a set of ordinary differential equations for selected moments (Fox, 2009;Marchisio and Fox, 2005;Marchisio et al, 2003a). …”

“…Its chief disadvantage is that it is not well-suited for systems that involve space and time gradients. (iii) Direct quadrature method of moments (DQMOM): this methodology is very efficient in one-component systems and is currently the only viable option for interfacing the PBE with fluid dynamics (Fox, 2009;Marchisio and Fox, 2005;Marchisio et al, 2003b;Marchisio et al, 2003c). Extension to two-component systems is not straightforward and requires validation against known solutions.…”

Simulation, model‐reduction, and state estimation of a two‐component coagulation process

“…Direct quadrature method of moments is an efficient technique for singlecomponent coagulation simulations. 8,9 However, Marshal et al showed this method is not as accurate as cNMC simulation for bicomponent coagulation processes.10 Yu et al employed Taylor expansions to derive a closed finite dimensional ODE system that models a single-component coagulation process.11 The accuracy of the system improves with increasing the Taylor expansion's order, which is equal to the number of moments considered in the model. In this manuscript, we exploit the idea in Yu et al's work but for bicomponent coagulation processes.…”