Large scale simulation models can aid in improving the design of spray dryers. In this work an Eulerian-Lagrangian model with coupled gas phase and droplet heat and mass transfer balances is used to study airflow dynamics, temperature and humidity profiles at different positions in the spray. The turbulent gas flow is solved using large eddy simulation (LES). A turbulent dispersion model accounts for the stochastic subgrid fluid velocity fluctuations along the droplet trajectory. The dispersed phase is treated with Lagrangian transport of droplets, and collisions between droplets which are detected with a stochastic Direct Simulation Monte Carlo (DSMC) method. The outcome of a binary collision is described by collision boundary models for water and milk concentrates. The drying of droplets is modeled by the reaction engineering approach (REA). The effect of the inlet air conditions and of droplet viscosity on the temperature and humidity distributions are analyzed. Most of the heat and mass transfer occurs in the first 10-20 cm from the nozzle where the slip velocities and temperature and humidity driving forces are higher. The droplets size increases, both in the axial and radial direction, because of the dominance of coalescence over separation in the droplet spray studied here. Because the spray domain considered in this work is relatively small, the droplet residence time is small, and consequently the amount of evaporation is still low. The droplet size distributions of milk concentrates are affected by the predominance of coalescence over separation events. The coalescence dominated regime increases when the droplet viscosity is higher.

In this work we investigate droplet-droplet collision interactions in a spray system using an Eulerian-Lagrangian model with subgrid turbulence dispersion. The effect of different droplet viscosities on the type and frequency of droplet collision is investigated, knowledge of which is essential for industrial processes such as spray drying for production of milk powder. The dispersed phase is treated with Lagrangian transport of droplets and the turbulent self-induced gas flow using large eddy simulation (LES). A stochastic Direct Simulation Monte Carlo (DSMC) method is used to detect collisions between droplets. The outcome of a binary collision is described by a collision boundary models for water and milk concentrates. A turbulence dispersion model, based on the Langevin equation, accounts for the stochastic subgrid fluid velocity fluctuations along the droplet trajectory. We compare the spray dynamics with and without droplet interactions and turbulence dispersion. For a spray with typical droplet size of 50 µm, we find that the turbulence dispersion model enhances the total collision frequencies by approximately 25%. The performance of the turbulent dispersion model is tested by investigating the rate of collisions for different milk concentrates. The evolution of size distributions inside the spray is strongly influenced by the complementary effects of collision boundary models and turbulence dispersion.

Spray drying is an important industrial process to produce powdered milk, in which concentrated milk is atomized into small droplets and dried with hot gas. The characteristics of the produced milk powder are largely affected by agglomeration, combination of dry and partially dry particles, which in turn depends on the outcome of a collision between droplets. The high total solids (TS) content and the presence of milk proteins cause a relatively high viscosity of the fed milk concentrates, which is expected to largely influence the collision outcomes of drops inside the spray. It is therefore of paramount importance to predict and control the outcomes of binary droplet collisions. Only a few studies report on droplet collisions of high viscous liquids and no work is available on droplet collisions of milk concentrates. The current study therefore aims to obtain insight into the effect of viscosity on the outcome of binary collisions between droplets of milk concentrates. To cover a wide range of viscosity values, three milk concentrates (20, 30 and 46 % TS content) are investigated. An experimental set-up is used to generate two colliding droplet streams with consistent droplet size and spacing. A high-speed camera is used to record the trajectories of the droplets. The recordings are processed by Droplet Image Analysis in MATLAB to determine the relative velocities and the impact geometries for each individual collision. The collision outcomes are presented in a regime map dependent on the dimensionless impact parameter and Weber (We) number. The Ohnesorge (Oh) number is introduced to describe the effect of viscosity from one liquid to another and is maintained constant for each regime map by using a constant droplet diameter (d∼700μm). In this work, a phenomenological model is proposed to describe the boundaries demarcating the coalescence-separation regimes. The collision dynamics and outcome of milk concentrates are compared with aqueous glycerol solutions experiments. While milk concentrates have complex chemical composition and rheology, glycerol solutions are Newtonian fluids and therefore easy to characterize. The collision morphologies of glycerol solutions and milk concentrates are similar, and the regime maps can be described by the same phenomenological model developed in this work. The regime of bouncing, however, was not observed for any of the milk concentrates.

We investigate the collision behaviour of a shear thinning non-Newtonian fluid xanthan, by binary droplet collision experiments. Droplet collisions of non-Newtonian fluids are more complex than their Newtonian counterpart as the viscosity no longer remains constant during the collision process. Despite the complex collision dynamics, we are able to present a complete regime map based on non-dimensional Weber (We) number and impact parameter (B). We compare the collision outcomes of xanthan, glycerol and a milk concentrate at similar impact conditions. These experiments reveal very rich and complex collision morphologies for shear thinning xanthan solution, strikingly different from Newtonian droplet collisions. Unlike glycerol and milk, xanthan collisions show no reflexive separation even at very high We number. Instead of breakup, we observe disc-like shapes with an oscillating behaviour of the colliding droplets. A detailed analysis reveals that this outcome is related to increased viscous energy dissipation and extensional effects.

A complete knowledge of the effect of droplet viscosity on droplet-droplet collision outcomes is essential for industrial processes such as spray drying. When droplets with dispersed solids are dried, the apparent viscosity of the dispersed phase increases by many orders of magnitude, which drastically changes the outcome of a droplet-droplet collision. However, the effect of viscosity on the droplet collision regime boundaries demarcating coalescence and reflexive and stretching separation is still not entirely understood and a general model for collision outcome boundaries is not available. In this work, the effect of viscosity on the droplet-droplet collision outcome is studied using direct numerical simulations employing the volume of fluid method. The role of viscous energy dissipation is analysed in collisions of droplets with different sizes and different physical properties. From the simulations results, a general phenomenological model depending on the capillary number (Ca, accounting for viscosity), the impact parameter (B), the Weber number (We), and the size ratio (Δ) is proposed.

An Euler-Lagrange model is presented that describes the dynamics of liquid droplets emerging from a high-pressure spray nozzle in a relatively large volume (of the order of almost a cubic meter). In the model, the gas phase is treated as continuum, solved on an Eulerian grid, and the liquid phase is treated as a dispersed phase, solved in a Lagrangian fashion, with interphase coupling through state-of-the-art drag relations obtained from direct numerical simulations. The droplets are introduced into the system at high velocities, leading to a turbulent self-induced gas flow which is solved using large eddy simulation. Despite the relatively low liquid volume fraction in the spray, the number density of droplets at the nozzle is still more than 10¹⁰m^-3, which is why we employ a highly efficient stochastic Direct Simulation Monte Carlo approach to track collisions between droplets. The droplet collision frequency is calculated on the basis of local droplet number density, droplet size and relative velocities of neighbouring droplets within a dynamically adapting searching scope, as described in Pawar et al. (2014. Chem. Eng. Sci. 105, 132-142). We use known correlations from literature to determine the outcome of a binary droplet collision, which depending on characteristic dimensionless numbers can be coalescence, bouncing or, for high velocity impacts, stretching or reflexive separation leading to formation of satellite droplets. Our simulation model is compared with droplet velocities and size distributions obtained from phase Doppler interferometry experiments on an industrial scale hollow-cone pressure swirl nozzle spray. We find semi-quantitative agreement for spray characteristics such as the axial and radial spray velocity, spray jet width, and the dependence of the droplet size distribution on position within the spray. The simulation model enables us to study the relative importance of different droplet collision events occurring in the spray volume.

Research on the dynamic flow behaviour in spray dryers has a long history. Interest in describing these flows originates from problems like roof and wall fouling. The aim of the present study is to experimentally investigate the dynamic jet behaviour and turbulent flow in a scaled-down cold flow model of a spray dryer in order to better understand and optimize spray drying units. Dynamic jet behaviour and turbulent flow features (i.e., RMS velocities) were studied by particle image velocimetry (PIV) using water as the continuous phase. To obtain more insight in the jet dynamics, we analyzed the turning point, the width and shape, and the velocity profiles of the turbulent jet at different heights and the turbulence characteristics. We found that at higher Reynolds numbers, the jet penetrates further along the downward direction with a time-averaged profile which is symmetric at the centre. In addition, we investigated the effect of the expansion ratio via proper orthogonal decomposition (POD). Outcomes of different characteristics of the dynamic jet, like steady, transient, regular, and complex precession, can be collapsed by proper scaling. These results can be used for validation of computational fluid dynamics simulations and facilitate the design (identification of jet operation boundaries) of new spray dryer configurations.