Award Date

8-1-2016

Degree Type

Thesis

Degree Name

Master of Science in Engineering (MSE)

Department

Mechanical Engineering

First Committee Member

Darrell Pepper

Second Committee Member

Alexander Barzilov

Third Committee Member

William Culbreth

Fourth Committee Member

Laxmi Gewali

Number of Pages

61

Abstract

In recent years, the use of large-eddy simulation (LES) has grown into new research methods. LES is preferred when compared to Reynold’s Averaged Navier Stokes (RANS), which separates velocity components into steady and fluctuating components. While RANS is relatively easy to implement, it does not fully resolve the range of turbulence eddies and has limitations that make it inaccurate in many practical circumstances. Direct numerical simulation, DNS, fully resolves all turbulent eddies to the smallest grid scale, but requires an extremely fine grid. This makes it computationally impractical to use as the computational power required to solve even the simplest case is severely high. In LES, the large eddies are resolved while the small eddies are modelled. This can be advantageous as we lower the computational resources required for solving the flow while still maintaining accuracy.

The use of RANS as well as DNS make them highly difficult in solving issues involving combustion processes. LES models tend to be simpler and require fewer adjustments when applied to a wide range of flows. In addition to turbulence, LES can also handle species transport and chemical reactions typically found in engine configurations, and makes it a very suitable choice.

LES has many different approaches, including the popular Smagorinsky model. The main problem with the elementary Smargoinsky approach is that a model parameter, 𝐶𝑑, is assumed constant over the entire region, which is generally not true for turbulent flows. Making the model dynamic to localize the parameter, we still have problems of the parameter varying too much (as much as ten times the mean) and the eddy viscosity becoming negative, resulting in instability. The Vreman LES model, as employed in this work, performs as well if not better than the dynamic Smagorinsky model, and does not require ad hoc procedures, including issues involving clipping. In this study, the LES Vreman model with a finite element method with h-adaptation technique is verified and validated using several benchmark cases. The end result being part of an on-going effort to enhance combustion predictability and increase efficiencies within engines.

Keywords

Combustion; LES; Vreman

Disciplines

Aerospace Engineering | Mechanical Engineering

Language

English


Share

COinS