Date of Award

Spring 1-1-2013

Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Zhiqiang Zhai

Second Advisor

Moncef Krarti


Computational Fluid Dynamics (CFD) has been becoming an important tool in building environment and airflow modeling. Well validated CFD model can provide more detailed and informative results than other computer models and physical experiment. However, the significant computational cost of CFD has always been the main factor that restricting its broader application.

This study reviews the various means of reducing the computational cost of CFD in literature. With considering the characteristics and result accuracy requirement in indoor environment quality (IEQ) study, the most important factors that affect the computing cost are identified: grid resolution, turbulence modeling and velocity-pressure decoupling algorithm. Theoretical analysis is conducted to find out the possibilities of optimizing grid resolution and simplifying turbulence modeling. Numerical experiments using measurement data in literature is conducted to validate the hypothesis from theoretical analysis.

Transient airflow simulation is a topic that draws not as much attention in this area, but is of great importance for building indoor environment study. Building is an open system interacting with the continuously changing weather condition. Most of the airflow and heat transfer phenomenon in building is transient with time-dependent boundary condition and dynamic thermal response. This study reviews the algorithms for transient airflow simulation, identifies the potential candidates and compares the performance of different algorithms. Based on the theoretical analysis of pros and cons of each algorithm, a new algorithm called semi-Lagrangian PISO (SLPISO) is brought out. Computer code is developed to validate and test the performance of SLPISO algorithm for typical indoor airflow case. The algorithm can significantly reduce the computational intensity of PISO and improves the accuracy of FFD, but for problems with velocity and pressure closely linked, relatively small time step has be to adopted in order to get accurate transient CFD solution.