Multiphase flows

The motion of dispersed objects (particles, droplets or bubbles) with a diameter larger than 10 micrometers is dominated by gravity, drag forces with the surrounding continuous medium (often air or water), and collisional contacts between the objects and with walls. Because such macroscopic objects collide inelastically, large-scale heterogeneous structures emerge. For many engineering applications it is important to be able to predict these dispersed flows through pipes, chutes, reactors and other vessels.

We are developing simulation models for macroscopic (non-Brownian) dispersed flows, where we use a multi-scale approach to cover the entire range of length scales from the particle level to the level of industrial size reactors. On the detailed level, we take into account the fundamental processes of momentum, heat and mass transfer, whereas on the larger levels we use effective (coarse-grained) interactions and correlations obtained on the detailed levels, often borrowing methods from statistical mechanics and soft matter science. Because a model is only as good as its assumptions (garbage in: garbage out), an important part of our work is the experimental validation of our models, through advanced non-invasive techniques such as Particle Image Velocimetry.

Granular flows and packings

mixing of granular particles in a rotating drum In exceptional cases, when granular particles are relatively large and the particle velocities are relatively low, the effect of the surrounding gas can be neglected. In one application, we have studied the mixing of granular materials in a rotating drum as a function of filling height and baffle size. We found that only relatively large, straight baffles perpendicular to the drum wall (67% of the drum radius) increase the mixing performance of the rotating drum.

experimental rotating table setup experimental rotating table setup We have also studied the flow of granular particles down an inclined and rotating chute. The goal of this work is to study the influence of chute rotation on the flow behaviour and segregation in granular flows, and to obtain high quality experimental data with which discrete particle model simulations can be validated. The figures to the right show the experimental setup on a rotating table, and a snapshot (top view) of our simulations. We measure the surface particle velocity field through particle image velocimetry and particel tracking velocimetry. We find that, compared to a non-rotating chute, the surface flow velocity in a rotating chute is decreased initially due to compaction against the side walls (Coriolis effect) but ultimately increased due to the increasing influence of centrifugal forces.

filling of a cylindrical tube with cylindrical particles In another line of research we study the packing of fixed bed reactors with non-spherical particles. It is highly non-trivial to take into account the collisions between highly non-spherical particles with curved surfaces and sharp edges. We have done two approaches; one along the line of the so-called GJK algorithm (followed by several other steps necessary to achieve correct forces and torques on the particles) and one along the line of the so-called Rigid Body Dynamics approach. Once the packed beds have formed, the configurations can be imported into CFD software to evaluate flow fields and heat- and mass transfer characteristics.

Fluidized beds

In a fluidized bed, an uprising gas or liquid is flowing so fast that it causes the solid particles to "float". As a result, the particle emulsion has many properties of a liquid, including the formation of flow vortices and bubbles. Fluidized beds are used frequently in chemical reactors because they allow an intense contact between the fluid (which may contain reactants) and the solid particles (which may be a growing product or contain catalysts).

contour plots of the solids volume fraction in a fluidized bed In one line of research, we are focussing on developing large scale computational models for gas-solid fluidized beds. One direction is to use stochastic Lagrangian coarse-grained multi-particle collision techniques borrowed from soft matter simulations, see the section on colloids. Another direction is to use and improve the so-called Two-Fluid Model, in which the gas and solid phases are treated as two interpenetrating continua. The latter approach requires accurate expressions for the rheology of the solid phase, for which we are using the kinetic theory of granular flow. The picture above shows contour plots of the solids volume fraction at three successive times in a cylindrical fluidized bed, obtained from these TFM simulations.

an intruder impacting onto a pre-fluidized bed of granular particles In another line of resarch, we are focussing on accurate simulation methods to measure the viscosity of the granular particle emulsion and to model the influence of intruders (which may also be baffles or stirrers). The viscoscity is obtained both through Green-Kubo relations and more directly through measurement of the stress in Couette flow and measurement of the drag on a dragged sphere. We have developed a hybrid method in which granular particles are coupled with the surrounding gas through effective drag correlations, whereas large intruders are coupled more directly through immersed boundaries. The influence of an impacting intruder is first studied for a pre-fluidized bed, in which the gas flow has slowly been turned off again. The picture to the right shows the effect of a large spherical intruder impacting into a pre-fluidized bed of much smaller granular particles.

In another line of research, we are focussing on fluidization of liquid-solid fluidized beds with the aim to adsorb certain components from the liquid. Application include expanded bed adsorption for the pharmaceutical industry and water softening reactors for drinking water production companies. For the latter, we have intense collaboration with Waternet, the drinking water company for the Amsterdam area.


Dry separation of materials through triboelectric charging

Our goal is to develop new separation technologies for dry fractionation of food products. The dry fractionation is based on electrostatic charging of powders in a gas stream when they collide with the walls of a charging tube, and subsequent separation by an external electric field (see figure above). Our aim is to accurately model the dynamics and acquired charge of the particles in the charging tube, where triboelectric charging takes place when a particle collides with the metal walls. The challenge of this project is that we need to combine triboelectric charging of the particles (about which little is known) with electrostatic interactions and gas flow. The first results are encouraging.


droplets emerging from a spray atomizer Spray drying is used to convert liquid feed materials into a dry powder form, e.g. to produce milk powder. The liquid feed is sprayed (atomized) through high pressure nozzles in a spray chamber (see picture on the left) and the resulting droplets are mixed with hot gas to evaporate the liquid content.

The main objective of this work is to develop a reliable simulation tool that can predict the droplet and particle flow, agglomeration, and size distribution for a section of a large-scale spray dryer. To be able to handle many billions of droplets and particles, we have adapted a direct simulation monte carlo method, in which the droplets and particles collide stochastically according to predictions from kinetic theory. Full two-way coupling between the droplets/particles and the gas has been implemented. Collisions between droplets and/or particles lead to different outcomes, depending on dimensionless numbers (Weber number, Ohnesorge number and impact parameter). The figure below shows the simulation results for a spray of pure droplets.

the outcome of a droplet-droplet collision depends on Weber number and impact parameter

Bubbly flows

evolution of bubbles in a bubble column Bubble column reactors are widely used as gas-liquid contactors in the chemical and energy industries. The bubble dynamics inside the column dictate the liquid phase flow. Modeling of such bubble columns is very challenging and typically empirical correlations are used even today.

The main objective of this work is to develop a stochastic Euler-Lagrange method that can predict the hydrodynamics of dense bubbly flows. For this we are adapting the direct simulation monte carlo approach to handle bubble collisions stochastically instead of deterministically, enhancing the speed of bubble collision calculations by a factor of 10 to 100 relative to more detailed discrete bubble model simulations, while still maintaining the same level of accuracy.


The research on granular flow was performed by my former PhD students Sushil Shirsath, Luuk Seelen, and Elyas Moghaddam. The research on granular viscosity and intruder impact was performed with my former PhD student Yupeng Xu. Within these projects we collaborated intensely with the Turbulence and Vortex Dynamics group (Prof. Herman Clercx) at Eindhoven University of Technology, with Prof. Detlef Lohse, Prof. Devaraj van der Meer and Dr. Martin van der Hoef of the Physics of Fluids group and with Prof. Stefan Luding and Dr. Anthony Thornton of the Multiscale Mechanics group, both at the University of Twente, and Prof. Onno Bokhove at University of Leeds. The research on fluidized beds of spherical particles was performed by my current and former PhD students Vikrant Verma and Lei Yang and was financially supported by an ERC-advanced grant of Prof. Hans Kuipers. The on-going research on fluidized beds of non-spherical particles, performed by PhD students Sathish Sanjeevi, Vinay Mahajan and Ivan Mema, and postdocs Vikrant Verma, Ahad Zarghami, Barry Fitzgerald, and Yousef Damianidis, and is financially supported by my ERC-consolidator grant and a number of NWO grants for computing time on the Dutch national supercomputer Cartesius, as well as a Prace grant for access to European supercomputers. The research on liquid-solid fluidization, performed by my PhD student Onno Kramer, is funded by Waternet. The research on triboelectric charging was performed with my PhD student Martin Korevaar, in collaboration with Dr. Maarten Schutyser of Wageningen University, and is financially supported by an STW grant. The projects on spray-drying was performed by our previous PhD student Sandip Pawar, continued by our current PhD student Giulia Finotello, and is financially supported by Tetrapak Heerenveen; we are indebted to intense collaborations with Dr. Alfred Jongsma and Dr. Fredrik Innings. A new project looking deeper into the drying of, and collisions between, complex dispersion droplets is continued by our current PhD student Stephan Sneijders, and is financially supported by NWO-TTW (Perspectief grant) and several industries, including Tetrapak, Danone-Nutricia and DSM.