### Abstract

The swirling flow plays a decent role in various applications. The interaction between the swirling jet flow phenomena and the CFD numerical modelling is on growing demands for many industrial/non-industrial fields such as combustion chambers and heat exchangers. The purpose of this research is to investigate the applicability of turbulence modelling typically employed in RANS equations to model a non-reacting swirling flow, and to extend the current understanding of turbulence modelling by delivering various range of numerical data to evaluate different types of turbulence models. ANSYS-CFX simulation program is used to model the flow by applying 𝑘-𝜀, RNG 𝑘-𝜀, Shear ST, LRR-IP-RS, SSG-RS and Omega-RS turbulence models at different degrees of swirl intensity. The reason for choosing 𝑘-𝜀, RNG 𝑘- 𝜀, Shear ST turbulence models is due to their common use in various published work for various kinds of flow. For instance, standard k-ε turbulence model uses for fully turbulent flows, while RNG 𝑘-𝜀 turbulence model is good for strained flows and low Reynolds number flows and Shear Stresses SST is good for transitional, free shear and compressible flows. Besides, RSMs turbulence models are excellent for swirling flow. Thus, indicate the ability of these models to perform the swirling flow at different strength is highly recommended to identify the advantage/disadvantage attached to them. For the planar jet flow, the results show that all the previous models can predict the flow behaviour and have excellent agreement with two cases of the planar jet flow, where the average error percent is 5.33% and 11% for the first and second case, respectively. The results show that all models are recommended to apply for low swirl degree with an approximate average error percent of 16%. The contours results denote a slightly non-symmetric effect due to existence of the swirl and the existence of thescrews in the geometry. However, the turbulence models were able to predict the change in the symmetric shape at each location at different level of prediction. For medium swirl degree, the study has found that the RSM models were preferred to apply because of their ability of showing the influence of swirl degree on the scalar mixing process with a 12% of average error. Additionally, the contours indicate that the non-symmetric form is higher than the first swirling number because of the increasing in the swirl strength.

However, the models show different level of computing this feature. For instance, two equations based turbulence models predict a small non-symmetric shape at the last two

locations. Though, the RMS turbulence models predict a large effect of non-symmetric structure. Remarkably, all the models have failed to predict the swirl flow at high swirl strength, apart from the LRR-IP-RS model which show a good agreement with the experiment data by an average error percent of 14.72%. The study concludes that the models based on eddy viscosity method are not ideal for high swirl flow. For instance, the two-equations based turbulence models predict a small non-symmetric shape through the computational domain because these models represent the flow like a planar flow at high velocity. Thus, the related contours of these models show high mixture fraction in all locations. Generally, all the RMS turbulence models predict a huge effect of non-symmetric structure due to the swirl force. However, Methane mass fraction is not correct specially in SSG-RS and Omega-RS turbulence model. Therefore, the RMS turbulence models can predict the high swirl effect on the symmetric structure more than the two-equations based turbulence models. On the other hand, it is also concluded that the diffusion process is dominated the scalar mixing at planar jet flow where the Reynolds number is relatively low. However, the turbulent mixing is the dominant process when the swirl occurred. Finally, the results suggest that the existence of screws have a significate effect on symmetric.

Date of Award | 2023 |
---|---|

Original language | English |

Supervisor | Helen Miao (Main Supervisor) & Artur Jaworski (Co-Supervisor) |