Events Calendar

3D Direct Numerical Simulations of Autoigntion in Turbulent Non-Premixed Flows with 1-Step and Reduc

Dr. Terese Løvas,Lecturer in Future Energy Conversion TechnologiesDepartment of Engineering,
Queen Mary University of London, UKAbstractThe autoignition of non-premixed flows is important for diesel and Homogenous Charge Compression Ignition (HCCI) engines, and it is also now a concern in the new lean premixed pre-vapourised (LPP) gas turbines. Because typically in diesel engines the ignition time is longer than an estimated turbulent timescale, the common understanding until early 1990's was that the ignition is not affected by the turbulence, but that is purely driven by the chemistry. However, it was later recognised that turbulence may affect the ignition time and the subsequent flame development significantly. Deeper knowledge of how the fluid mechanics affect autoignition will assist the design of the low-polluting HCCI engines and the new LPP gas turbines.
In the talk results from a set of 3D Direct Numerical Simulations (DNS) of autoignition in turbulent non-premixed flows will be discussed. Both a simple 1-step mechanism and a complex chemistry consisting of a 22 species n-heptane mechanism is employed to investigate spontaneous ignition timing and location. The results from simple chemistry showed that the previous findings from 2D DNS, that ignition occurred at the most reactive mixture fraction (MR) and at small values of the conditional scalar dissipation rate (N|MR), are valid also for 3D turbulent mixing fields. However, in the Negative Temperature Coefficient regime (NTC), the most reactive mixture fraction is very rich and ignition seems to occur at high values of scalar dissipation. This is not consistent with a previous conjecture that the first appearance of ignition is correlated with the low-N content of the conditional probability density function of N.
The treatment of reliable chemistry in complex flow codes are of great importance for the correct predictions of control parameters such as ignition time and flame temperatures. However, the inclusion of detailed chemistry in such complex flow codes is demanding in terms of both computational time and memory requirements. This is because the chemical reaction system is governed by a set of stiff differential equations determining the time evolution of each chemical species based on consumption and production through chemical reactions. Much effort is devoted to developing methods to eliminate the species governed by the shortest time. A method for reducing reaction mechanisms will be discussed which is based on a time-scale analysis much similar to typical model reduction techniques. This enables the set of variable that are transported in the flow codes to be significantly reduced. Also, a procedure to automatically optimize the sparsity of the Jacobian matrix governing the chemical evolution is implemented resulting in a significant computational speed-up.

Location: Stauffer Science Lecture Hall (SLH) Rm 102

Audiences: Everyone Is Invited

Contact: April Mundy