Simulating fluidized beds accurately using EDEM coupled with ANSYS Fluent

Numerical modelling is common-place in industry and academia, aiding understanding in many different physical systems. Approaches such as Computational Fluid Dynamics (CFD), Discrete Element Modelling (DEM), Finite Element Analysis (FEA) and Multi-body Dynamics (MBD) are used globally and their popularity is still growing, as ever-increasing computing power means previously computationally demanding simulations run in more acceptable timeframes. Although running a simulation in a reasonable time-frame is clearly important, this should not take precedence over the accuracy and applicability of the model, else the model’s main purpose of simulating a real system is lost. This post focuses on processes involving particles and fluids and the applicability of using EDEM coupled with ANSYS Fluent to simulate such systems. Specifically, we will demonstrate the ability of an EDEM-Fluent coupled simulation to model the pressure drop across a fluidized bed and compare to suitable validation data.


Fluidized bed simulation using EDEM coupled with ANSYS Fluent

Fluidized beds are common place in industry – particularly in chemical processing – and are used in a wide variety of processes. Their popularity stems from their ability to introduce high levels of contact between fluids and solids, making them ideal for processes such as freezing, cooling, heating, coating, and treating all manner of solid materials and particles.

A fluidized bed consists of a bed of particles inside a domain, with a fluid flow passing across the bed (see Figure 1). Assuming that the particles are denser than the fluid in which they are immersed, the particles’ weight will be greater than the buoyant force provided by the fluid, meaning the particles will sink and rest at the bottom of the domain. Once the fluid velocity reaches a certain value, fluidization occurs, whereby the particles are suspended by the force provided by the fluid flow. As the velocity increases, the fluid flow eventually turns turbulent and introduces eddies and velocity fluctuations within the flow. These eddies and fluctuations, and the chaotic nature of turbulence, add to the overall mixing capability of the fluidized bed.


Figure 1: A typical Fluidized bed

Fluidized beds can be difficult to simulate as they are not only dependent upon the pressure drop across the bed but also typically contain many particles, which can lead to complex interactions between the fluid and particles. Pressure drops in fluids are well suited to being modelled by CFD packages, but the presence of large numbers of particles means some of the particle models built into CFD packages are not. EDEM is built to handle particle-based applications, making it the ideal tool for handling the complex particle interactions that occur in these types of simulations. The EDEM Coupling Interface serves as a perfect platform to couple with CFD packages to allow for accurate modelling of the complex behavior occurring in fluidized beds. The EDEM-Fluent coupling is one such coupling that allows for accurate fluid calculations in Fluent and accurate particle collisions calculations in EDEM.

For any numerical simulation, whether it be CFD, DEM, FEA, MBD or another simulation type, validation is a crucial part of the numerical modelling process, as without validation data we cannot be certain that the simulation can capture the behavior of the real system. Typically, this takes the form of experimental data, but theoretical calculations are also common place, if there is a relevant and established theoretical model. For the fluidized bed simulation outlined here, we are fortunate enough to have both experimental data available and a theoretical model – based on the Ergun equation – capable of calculating the pressure drop across a fluidized bed:


where Δp is the pressure drop across the bed,  L is the height of the bed, μ is the fluid viscosity, ϵ  is the void fraction of the bed, u0  is the superficial fluid velocity, Dp is the particle diameter and ρ is the density of the fluid [1].

Back in 2013, the National Energy Technology Laboratory (NETL), in the USA, performed a series of experiments, or small-scale challenge problems (SSCP), the results of which were made available as a means of validating multiphase CFD simulations [2]. A simulation representative of ‘NETL SSCP I’ has been modelled using the EDEM-Fluent coupling, using a GPU – NVIDIA GeForce GTX 1080 – for the EDEM part of the simulation and 8 CPU cores for the Fluent part of the simulation.

Data from the NETL SSCP I experiment is provided for three different gas flow-rates / velocities, the smallest flow rate (2286.1 SLPM, or 2.19 m/sec) was simulated using the EDEM-Fluent coupling. The simulation was set to run for 10 seconds with a timestep of 1e-3 seconds on a mesh of approximately 1 million elements in Fluent. All particle data was set to be in line with that provided within the NETL SSCP I results file.


Figure 2: Pressure difference across the fluidized bed, calculated from: the EDEM-Fluent simulation, Equation 1 and NETL SSCP I experimental data.

For validation, the pressure drop across the fluidized bed is compared to both data from the NETL SSCP experiment and to a value calculated from the equation (1) for the simulation parameters. For the NETL SSCP I experiments, three pressure transducers were placed at different locations to provide fluid pressure data. The mean pressure drop between two of these locations is plotted together with a time-series of the difference between the area-averaged pressure of two planes at the same height as the two pressure transducers and the value obtained from equation (1) (see Figure 2). Once the simulation reaches a steady fluidization velocity, the pressure difference from the simulation compares favorably with the mean pressure obtained from the experiments. The mean pressure from the NETL SSCP I experiment was 690 Pa, compared to an average of 714 Pa for the simulation data. The simulation pressure data is also in good agreement with the predicted pressure drop of equation (1) of 724 Pa.

Validation data and theoretical models are not readily available for all simulations. In some cases, obtaining data is not practical and for many cases a suitable theoretical model has not yet been established. Furthermore, even if they do exist, theories and validation data are not exempt from error (far from it!) and so the user must always apply a certain amount of common sense as to when a model is considered to be ‘validated’ and what agreement is considered sufficient for their specific application. In this instance, validation data from two different sources give us confidence that the simulation provides a value for the pressure drop across the fluidized bed in agreement with this data, in turn highlighting the capability of the EDEM-Fluent coupling to simulate real-world particle-fluid systems.

To find out more about the benefits of coupling EDEM with ANSYS Fluent check our on-demand webinar: Modeling Fluid-Particle Systems using EDEM with CFD


[1] O. Akgiray & A. M. Saatci, 2001. A New Look at Filter Backwash Hydraulics. Water Science and Technology: Water Supply, Vol: 1, Issue: 2, pp 65-72

[2] Data available from the National Energy Technology Laboratory website.


Share this article