Turbulent drop breakup mechanisms in high-pressure homogenizers
Title: Turbulent drop breakup mechanisms in high-pressure homogenizers
SNIC Project: SNIC 2020/5-555
Project Type: SNIC Medium Compute
Principal Investigator: Andreas Håkansson <andreas.hakansson@food.lth.se>
Affiliation: Lunds universitet
Duration: 2020-11-01 – 2021-11-01
Classification: 21101 20306 20499
Homepage: http://www.food.lth.se/
Keywords:

Abstract

The objective of this project is to better understand the different droplet breakup mechanisms in turbulence. This will be beneficial to various applications and industries including food industry where a more efficient drop breakup in High-Pressure Homogenizers (HPH) and Rotor-Stator Mixers (RSM) could reduce energy consumption. The overall aim is to improve our fundamental understanding of turbulent drop breakup in HPHs and RSMs by quantifying breakup rates and fragment size distributions. The project includes an experimental part (breakup visualization in a scale-up model) and resolved Direct Numerical Simulation (DNS) simulations. A scale-up model of an HPH gap is designed for the experimental phase of the project with DNS simulations performed on the same HPH gap geometry. In the DNS simulations, we aim to obtain the local breakup rates as a function of the initial drop diameter and the characteristics of the turbulent field close to the drops. We will use the DNS results along with the results from the experimental phase of the project to acquire a better understanding of the drop breakup mechanisms in HPHs and develop a predictive model for breakup rates to be used by the industry. The in-house DNS code we shall use has been extensively used and validated in the group of Luca Brandt at KTH, Stockholm. Also, preliminary tests have been performed before this application to ensure the suitability of the method for our problem and to prepare the larger runs planned here, e.g. implementing a stochastic method for generating synthetic turbulent fluctuation at the inlet boundary condition. Furthermore, preliminary estimations have been carried out regarding the required resolution, domain size and the required CPU time. During the first part of the numerical work, single-phase DNS will be performed to reach statistical steady state turbulence. After analyzing the flow statistics and ensuring that the domain size and resolution are sufficient, two-phase simulations will be performed. To characterize the behavior of each drop, single droplets will be injected at the inlet with enough time delays, so that no interactions will be possible among the injected droplets.