PreprintArticleVersion 1This version is not peer-reviewed
Application of the Lagrangian Meshfree Approach to Modelling of Batch Crystallisation: Part II—An Efficient Solution of Integrated CFD and Population Balance Equations
Version 1
: Received: 1 November 2016 / Approved: 2 November 2016 / Online: 2 November 2016 (05:31:48 CET)
How to cite:
Nikolić, D.; Frawley, P. Application of the Lagrangian Meshfree Approach to Modelling of Batch Crystallisation: Part II—An Efficient Solution of Integrated CFD and Population Balance Equations. Preprints2016, 2016110012 (doi: 10.20944/preprints201611.0012.v1).
Nikolić, D.; Frawley, P. Application of the Lagrangian Meshfree Approach to Modelling of Batch Crystallisation: Part II—An Efficient Solution of Integrated CFD and Population Balance Equations. Preprints 2016, 2016110012 (doi: 10.20944/preprints201611.0012.v1).
Cite as:
Nikolić, D.; Frawley, P. Application of the Lagrangian Meshfree Approach to Modelling of Batch Crystallisation: Part II—An Efficient Solution of Integrated CFD and Population Balance Equations. Preprints2016, 2016110012 (doi: 10.20944/preprints201611.0012.v1).
Nikolić, D.; Frawley, P. Application of the Lagrangian Meshfree Approach to Modelling of Batch Crystallisation: Part II—An Efficient Solution of Integrated CFD and Population Balance Equations. Preprints 2016, 2016110012 (doi: 10.20944/preprints201611.0012.v1).
Abstract
The second article in the series presents the application of the Smoothed Particle Hydrodynamics (SPH) method to modelling of batch crystallisation in stirred tanks. A methodology to integrate the population balance equations (PBE) in parallel and independently from the Navier-Stokes equations is demonstrated. The benefits of the proposed methodology in terms of computational requirements, accuracy and availability of the crystal size distribution are discussed. The specific formulation of the SPH equations where the resulting system of ordinary differential equations is solved using the weighted contributions rather than numerically by solving a linear system of equations allows for massive parallelisation and a very loose coupling of the population balance and the fluid dynamics. It has been demonstrated, that the population balance equations can be solved on a Shared Memory Architecture (SMA) system using the OpenMP interface while the fluid dynamics equations being computed independently on a General Purpose Graphics Processing Unit (GPGPU) using the NVidia CUDA technology. This way, a significant portion of the computational overhead due to the large number of additional transport equations resulting from the discretisation of the population balance was removed: the SPH simulation coupled with 200 population balance equations was only 40% slower compared to SPH-only simulation.Two methods for the solution of population balance equations that preserve full crystal size distribution were implemented: discretised population balance (DPB) and method of characteristics (MOCH). The DPB equations are solved using the high-resolution finite-volume method with flux limiter and the effect of a large number of different flux limiters have been investigated. Both methods were validated using the case studies from the literature where an analytical solution can be derived.The developed models were applied to a numerical solution of coupled computational fluid dynamics and population balance equations to model a batch crystallization process. The effect of the hydrodynamics on the local temperature/supersaturation and the resulting crystal size distribution was captured and compared to the ideal mixing case. The simulation results from the DPB and MOCH methods were compared in terms of computational requirements and accuracy and MOCH selected as computationally more efficient and accurate.
Subject Areas
CFD; SPH; population balance; NVidia CUDA; OpenMP
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Dragan Nikolić