Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Parallel Monte Carlo Algorithms for Flowsheet Simulations based on the Population Balance Equation
This work discusses the modeling of interconnected unit operations as used for the processing and production of solid materials. The materials are thereby described as single particles which are possibly suspended in liquids (colloids) or in a gas (aerosols). The processing of the particles is based on physical mechanisms such as nucleation, coagulation, breakage, evaporation and condensation as well as particulate transport between the units. The corresponding rates of these mechanisms depend thereby on the properties of the single particles: next to the particle size, additional particle properties (like porosity, surface volume, electric charge, etc.) might have to be considered for a correct description. The solution of the resulting population balance equation poses a challenging numerical task, especially if many particle properties are modeled in a network of interconnected compartments,or units. Conventional solution strategies for the population balance equation encompass 1) the description of the particle population only by its moments, or 2) the discretization of the particle size spectrum into a finite number of sizes or size ranges. Both strategies are limited to a low number of particle properties (maximal 2–3 with a low number of sections or points). Stochastic methods, also called Monte Carlo methods, pose an attractive alternative to these approaches, enabling the modeling of several particle properties at the cost of higher computational time and an intrinsic stochastic noise of the generated result. In the scope of this work, novel strategies for the reduction of the necessary computing time are introduced by two means: 1) the application of GPUs allows the parallelization of the necessary computations and 2) the formulation of novel mathematical concepts allows for a new kind of Monte Carlo simulations which have inherent lower noise levels and guarantee therefore the same accuracy (in comparison with conventional methods) if less computational particles – and thus computational resources – ...
Parallel Monte Carlo Algorithms for Flowsheet Simulations based on the Population Balance Equation
This work discusses the modeling of interconnected unit operations as used for the processing and production of solid materials. The materials are thereby described as single particles which are possibly suspended in liquids (colloids) or in a gas (aerosols). The processing of the particles is based on physical mechanisms such as nucleation, coagulation, breakage, evaporation and condensation as well as particulate transport between the units. The corresponding rates of these mechanisms depend thereby on the properties of the single particles: next to the particle size, additional particle properties (like porosity, surface volume, electric charge, etc.) might have to be considered for a correct description. The solution of the resulting population balance equation poses a challenging numerical task, especially if many particle properties are modeled in a network of interconnected compartments,or units. Conventional solution strategies for the population balance equation encompass 1) the description of the particle population only by its moments, or 2) the discretization of the particle size spectrum into a finite number of sizes or size ranges. Both strategies are limited to a low number of particle properties (maximal 2–3 with a low number of sections or points). Stochastic methods, also called Monte Carlo methods, pose an attractive alternative to these approaches, enabling the modeling of several particle properties at the cost of higher computational time and an intrinsic stochastic noise of the generated result. In the scope of this work, novel strategies for the reduction of the necessary computing time are introduced by two means: 1) the application of GPUs allows the parallelization of the necessary computations and 2) the formulation of novel mathematical concepts allows for a new kind of Monte Carlo simulations which have inherent lower noise levels and guarantee therefore the same accuracy (in comparison with conventional methods) if less computational particles – and thus computational resources – ...
Parallel Monte Carlo Algorithms for Flowsheet Simulations based on the Population Balance Equation
Kotalczyk, Gregor (Autor:in) / Kruis, Frank Einar
03.08.2023
Hochschulschrift
Elektronische Ressource
Englisch
Calcination Flowsheet Development
| NTIS | 1978
Application of Monte Carlo Simulations
| Springer Verlag | 2021
Monte Carlo Simulations of Intermetallic Compounds
| British Library Online Contents | 1995