Cyclone Separator Challenge

Aim

To perform analysis on cyclone separator and calculate the separation efficiency and pressure drop.

1. Introduction

Cyclone separators, also known as cyclones, are used to separate particles from a carrier gas. It employs a centrifugal force by spinning gas stream inside the separator geometry. It is widely used as pre-cleaners due to the simplicity of design, low capital investment and near maintenance free operation. Due to the limitation of cyclones to separate particles effectively in the range above 5µm, the output of the cyclones is usually fed to expensive final control devices such as bag houses or electrostatic precipitators for further filtration. Cyclones find their applications in removal of coal dust in a power plant and the use as a spray dryer or gasification reactor. It is also widely used in industrial chimneys where combustion exhaust is emitted into the atmosphere.

Complicated swirling turbulent flow in a cyclone places great demands on the numerical techniques and the turbulence models employed in modelling the cyclone pressure drop. Over time, empirical models were defined to calculate the separator efficiency of the cyclone. However, due to its reliance on experimental results and inaccuracy when subjected to different operating conditions, CFD as a tool was widely accepted. Due to low investment costs and improved accuracy, the use of CFD to determine the separator efficiency can be found in current industries. However, the results from CFD are limited due to the available processing power. More refinement of the mesh and selection of the most accurate turbulence models place huge demands on the processing capacity thereby increasing the cost expenditure. Thus, optimization of the design using CFD is still a research in progress.

The current study explores a basic computer aided design of the cyclone separator and calculates the separator efficiency using CFD models for different particles size and velocities.

2. Empirical models used to calculate the cyclone separator efficiency.

2.1Iozia and Leith Model

The empirical model is developed based on force balance. It assumes that particle carried by a vortex in the cyclone separator is under the influence of two forces: centrifugal force and flow resistance.

The collection efficiency is calculated using dependent variables- particle diameter d_pi, 50% cut size d_pc and slope parameter derived using statistical analysis β.

η=1/ [1 + (d_pc / d_pi) ^β]

2.2 Li and Wang model

This model obtains a two dimensional analytical equation for cyclone separator efficiency by including turbulent diffusion and bounce or re-entrainment of the particles at the walls of the cyclone separator. It is based on the following assumptions:

1. a) The radial particle velocity and the radial concentration profile are not constant for uncollected particles within the cyclone.
2. b) Boundary conditions at the wall are defined considering turbulent diffusion and bounce or re-entrainment
3. c) The tangential velocity is related to the radius of cyclone

2.3 Koch and Licht Model

The model recognizes the turbulent nature of cyclones resulting in uneven distribution of particle residence time within the cyclone. In additions, the model assumes the following for the particles in the entry and collection regions:

1. a) There is no slip in the tangential direction between the particle and the gas. This results in tangential velocity of the particle being equal to that of the gas flow.
2. b) The tangential velocity is related to the radius of cyclone

2.4 Lapple model

Lapple model is developed using the force balance similar to the Iozia and Leith Model. However, the Lapple model does not consider flow resistance.

The model assumes that particles that enter the cyclone are evenly distributed across the inlet opening and a particle that travels from the inlet half width to the cyclone wall is collected with 50% efficiency.

2.5 Grade Efficiency Prediction under Ambient Temperature and Pressure

Using experimental data as reference, it was found that the empirical model Li and Wang agrees better than the other models. Lapple’s model yields less accurate fitting due to the assumption that the particles entering the cyclone are evenly distributed. Koch and Licht predictions are unsatisfactory due to unjustified assumptions of uniform mixing of uncollected particles. The prediction of cyclone efficiency by Iozia and Leith logistic model is only satisfactory for cyclone diameter close to D= 0.25 m, the diameter for which the statistical analysis data is generated.

2.6 Grade Efficiency Prediction at high Temperature and Pressure

The models of Koch and Licht, and Lapple still show a reasonably good prediction under this extreme condition. Meanwhile, Li and Wang model is found to underestimate the cyclone collection efficiency under the extreme operating temperatures.

2.7 Cut-Off Size Prediction

Cut-off size is the particle diameter at which the cyclone separator efficiency reaches 50%. With experimental data as reference, the empirical models predict cut-off size inconsistently with deviation ranging from 11.9 to 33.3%.

The most accurate model to predict the cyclone separator efficiency is using Computational Fluid Dynamics which gives consistent results under all operating conditions and predicts cut-off size with deviation of 3.7% to the measured value.

3. Objective

• To perform an analysis on a given cyclone separator model by varying the particle diameter from 1 μm to 10 μm and calculate the separation efficiency in each case. Discuss the results [both velocities considered as 3m/sec.]
• To perform an analysis on a given cyclone separator model by varying the particle velocity from 1 m/sec to 5 m/sec and calculate the separation efficiency and pressure drop in each case [particle diameter size as 5 μm]

4. Solving & Modelling approach

The following are the steps involved:

1. The cyclone separator model is imported and the fluid volume is extracted
2. The extracted volume is meshed using appropriate cell size and mesh method
3. The setup includes selecting the viscous model and the discrete phase model
4. Define the appropriate parameters of DPM and define the boundary conditions
5. The setup is initialized using standard initialization and is solved for convergence of residuals
6. The results are post-processed using CFD-Post
7. Pre-processing and solver setting

5.1 Geometry

The geometry is imported into Spaceclaim and using volume extract command, the fluid volume is extracted.

Fig 1- Geometry- Solid volume of cyclone separator

Fig 2- Volume extract

The solid component is supressed for physics and the extracted volume is used to generate mesh.

5.2 Mesh

The geometry is meshed with element size of 7 mm. Considering the accuracy of results in case of structured mesh, the study uses Assembly Meshing method: Cut cell. The advantage of using cut cell approach is the reduction of cell count without compromising on the accuracy of results. The limitations of effective meshing in complex regions can be neglected considering the simplicity of the geometry.

Fig 3- Mesh

Fig 4- Mesh metrics

The inlet and the outlets are named by selecting the respective faces.

Fig 5 – Named selections

5.3 Setup

The next step involves setting up the case using the appropriate turbulence model and the boundary conditions. For the case, the steady state, pressure-based type is selected. The gravitational acceleration is enabled and is defined as 9.81 m/s² in the negative y-direction.

5.3.1 Viscous Models

For capturing the viscous effects, k-Ɛ turbulence model is selected. To capture the swirling effects in the turbulent flow, RNG model with Swirl Dominated Flow option is used.

5.3.2 DPM

The discrete phase modelling is enabled with interaction with the continuous phase. The injections are defined as surface type and the variables velocity and particle size are entered.

5.3.3 Boundary Conditions

The named selections are set with the below DPM boundary conditions:

Inlet and wall - Reflect

Outlet (Top) - Escape

Outlet (Bottom) – Trap

The corresponding fluid (air) velocity is also defined for the velocity-inlet.

5.3.4 Solution

Using standard initialization at the inlet zone, the case is run for about 1200 iterations until the residuals indicate convergence of the solution.

6. Results

6.1 Convergence

Fig 6 – Plot of residuals against iterations

It can be observed from the figure that the solution converges at about 1100 iterations.

6.2 Particle residence time

The particle residence time animation for particle size 1 µm and 10 µm are shown below:

Using the animation of the particle residence time, the flow of the discrete phase particles can be studied. The flow is uniform at the inlet where the particles are accelerated. As it enters into the cyclone, the force imbalance causes generation of vortices about the central axis of the cone. The turbulent flow captures the particles inside the cone and flow to the bottom where the particles are trapped. This allows the air flow through the low pressure outlet-top to be free of the discrete phase. However, there exist re-circulation zones inside the cyclones which affect the entrapment of the particles. Also, for smaller size of particles, the cyclone is unable to capture the particles completely and several escape through the top outlet along with the air flow. This effect is reduced when simulated using larger size of particles.

6.3 Variation of separation efficiency with particle size

This section involves varying the particle size from 1 µm to 10 µm for a constant fluid and particle velocity of 3 m/s to calculate the separation efficiency. The separation efficiency is defined as the ratio of number of trapped particles to total number of particles tracked.

Table -1 Comparison of separator efficiency for variation in particle size

It can be observed that for very small size of particles, the efficiency is very low as most of the particles escape through the top outlet. However, an increase in particle size results in improved efficiency as more and more particles are trapped at the bottom outlet/dustbin. For particle size more than 5 µm, the separator efficiency is as high as 99%

6.4 Variation of separation efficiency and pressure drop with particle velocity

The study is further extended to study the effects of varying the particle velocity in the range 1 m/s to 5 m/s. The separation efficiency and pressure drop are noted and a comparison is made.

Pressure drop = Total inlet pressure - Total Outlet Pressure (outlet top in this case).

Fig 7- Total pressure plot at inlet and outlet(top)

Fig 8 – Pressure plot over the cyclone separator

From the figure, it can be noted that as the particles flow through the inlet into the cyclone, the pressure decreases. Due to the vortex generation, there is further reduction in pressure due to an increase in rotational velocity. The pressure is further lowered at the outlet-top as the particle and air escape out of the cyclone. This pressure drop is studied for a particular particle size of 5 µm for different velocities and is tabulated below.

Table- 2 Comparison of pressure drop for different particle velocities

It can be noted that at low particle velocities, the cyclone efficiency is low as the incoming particles escape through the top outlet. However, at the higher velocities the efficiency increases and attains values of more than 95%.

The pressure drop across the inlet and outlet (top) surfaces increase with an increase in the particle velocity. This increase in pressure drop is not desirable as it can be seen as excess dissipation of energy in the system. Hence, the optimal inlet velocity is to be studied to achieve high separation efficiency with low pressure drop across the cyclone.

7. Conclusion

The study explains the empirical models used to calculate the separation efficiency of the cyclone separator. Furthermore, the study successfully simulates the flow inside the cyclone and establishes the implications of variation in particle size and velocity on the separation efficiency and pressure drop.