Computational fluid dynamics in an equation-based, acausal modeling environment

Modelica;CFD

Brown, Jason ; Architecture

University:Georgia Institute of Technology

Department:Architecture

关键词: Modelica;    CFD;   

Others  :  https://smartech.gatech.edu/bitstream/1853/37247/1/brown_jason_b_201012_phd.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

The practice of building simulation is split between domains such as energy, multizone airflow, computational fluid dynamics (CFD) airflow, and controls analysis, as well as between the tools which conduct these analyses.Previous work in the integration of these analyses and tools have focused on linking existing tools, written in algorithmic programming languages, together by interfacing them using coupling mechanisms implemented in algorithmic programming languages.This thesis takes a different approach, using the equation-based, object oriented modeling language Modelica to create models in different domains and interfaces between those models within a single framework which has benefits to the modeler/analyst in terms of both representation of physical processes and flexibility in modeling systems composed of many interacting components.Specifically, the simulation of airflows within buildings has historically been compartmentalized into distinct domains such as nodal network (multizone) simulations and CFD.Such airflow simulations are also often treated independently of building energy simulations (via heat transfer) despite their interrelation.Recent work has reported on combining these types of analyses by linking pre-existing simulation software together.Here a prototype CFD package of models is built in Modelica and coupled to models of conductive heat transfer and controls.Comparisons of results of simulations so constituted to analytical solutions and benchmark data available in the literature show good agreement, indicating the technical viability of this approach.Limitations include the absence of turbulence modeling and the lack of modeling features which improve computational efficiency, such as non-uniform grids.