CFD Analysis of a Cyclone Separator using Discrete Phase Modelling in ANSYS Fluent.
Aim: To perform analysis on cyclone separator and calculate the separation efficiency and pressure drop.
Objectives:
1. To perform analysis for varying particle size from 1 `mum` to 5 `mum` with inlet velocity of 3 `m/s`
2. To perform analysis for varying velocity between 1 `m/s` to 5 `m/s` with particle diameter being 5 `mum`
3. Calculate separation efficiency and pressure drop for each case.
4. To understand the factors affecting efficacy of the cyclone separator.
Introduction:
Cyclone Separator:
- Cyclone separator is separation device that use the principle of inertia to remove particulate matter from the flue gases.
- It is one of the many air pollution control devices known as pre-cleaners since they generally remove large pieces of particulate matter.
- Cyclonic separation is a method of removing particulates from an air, gas or liquid stream, without the use of filters, through vortex generation.
Fig a: Sketch of the flow pattern inside a cyclone separator Fig b: Working of a cyclone separator
- Vortex effects under the influence of gravity is used to separate the particulate matter.
- A high-speed rotational effect is generated in the cylindrical / conical container which depicts the cyclonic behaviour.
- At instances to remove particulate matter from liquid hydrocyclone is used whereas for gases, a gas cyclone is employed.
- The particulate matter which is mixed in the flue gas stream possess comparatively higher density. The heavier particles experiencing cyclonic behaviour strike to the container wall, under the action of gravitational influence and fall off to the bottom where they are collected.
- The geometrical contraction of fluid flow in the conical section reduces the rotational radius of the mixed stream. This enhances to capture smaller particles while providing extra inertia assist.
- The lighter flue gas which is not affected by inertia escapes from the top opening to further treatment / atmosphere.
- The efficiency of the cyclone separator is dependent on inlet velocity, density of the particulate matter and geometrical dimensions of the set-up.
Applications of a cyclone separator:
1. Oil refineries
2. Sawmills to remove saw dust
3. Vacuum cleaners
4. Industrial water treatment
In order to design and employ a cyclone separator for the industrial use, it is important to investigate the efficacy of the cyclone separators. The design changes may vary for different applications. To evaluate separation efficiency for an intended application several mathematical models are used. These models assume several flow characteristics pertaining to the cyclonic behaviour which assists in calculating the efficacy.
The following discussed are various models based on certain assumptions and concerned flow physics.
1. Iozia and Leith model:
- This model is based on force balance.
- The model assumes that a particle carried by a vortex endures the influence of two forces; a centrifugal force and a flow resistance.
- The collection efficiency (`eta_i`) of particle diameter `dp_i` can be calculated from,
`eta_i = (1)/(1+((dp_c)/(dp_i))beta)`
2. Li and Wang model:
- This model includes particle bounce or re-entrainment and turbulent diffusion at the cyclone wall.
- A two dimensional anayltical expression of particle distribution on the cyclone is obtained.
- The resultant expression of the collection efficiency for particle of any size is given as,
`eta_i = 1-exp{-lambdatheta_1}`
3. Koch and Licht model:
- This model recognized the inherently turbulent nature of the cyclones and the distribution of gas residence times within the cyclone.
- The model described particle motion in the entry and collection regions with following assumptions,
a) The tangential velocity of a particle is equal to the tangential velocity of the gas flow, i.e. there is no slip int he tangential direction between the particle and the gas.
b) The tangential velocity is related to the radius of cyclone by `muR^n = const.`
c) A force balance and an equation on the particles collection yields the grade efficiency `eta_i`,
`eta_i = 1-exp{-2 [(Gtau_iQ)/(D^3) (n+1)]^((0.5)/(n+1))}`
4. Lapple model:
- Lapple model was developed based on force balance without considering the flow resistance.
- Lapple assumed that a particle entering the cyclone is evenly distributed across the inlet opening.
- The particle that travels from inlet half width to the wall in the cyclone is collected with 50% efficiency.
- The efficiency of collection of any size iof particle is given by,
`eta_i = (1)/(1+((dp_c)/(dp_i))^2)`
Project Methodology:
The project is intended to study the effect of particulate diameter size and gas flow velocity (discrete and continuous phase) on the efficacy of the cyclone separator. Numerical analysis is carried in ANSYS FLUENT with post-processing in CFD-post. Initially the geometry clean up is performed to reduce the inclusion of extra (unwanted) features into the meshing and solving platform. The cleaned geometry is imported in ANSYS Meshing and cartesian mesh is used to mesh the geometry. As cartesian mesh is aligned to the flow direction which certainly leads to more accurate results and better convergence in CFD solvers. The resulted mesh is finally imported in FLUENT module where numerical analysis is performed using Discrete Phase Model (DPM). The obtained results are post-processed and are discussed accordingly.
Case studies:
Case Study One: Varying particle diameter with same inlet velocity.
| Cases | Particle diameter (microns) | Particle (Discrete) and continuous fluid velocity (m/sec) |
| Case 1a | 1 | 3 |
| Case 1b | 3 | 3 |
| Case 1c | 5 | 3 |
Case Study Two: Varying inlet velocity with same particle diameter.
| Cases | Particle diameter (microns) | Particle (Discrete) and continuous fluid velocity (m/sec) |
| Case 2a | 5 | 1 |
| Case 2b | 5 | 4 |
| Case 2c | 5 | 5 |
I] Pre-processing:
A) Geometry:
The cyclone separator geometry is imported in SpaceClaim and extra features such as (extra small edges) are removed. This prevents unexpected meshing of the extra features in the domain.
1. Solid geometry:
The following geometry represents cyclone separator with one inlet and two outlets (at the top and the bottom).
2. Extracted fluid volume:
Wet volume (fluid volume) is extracted and will be used for further processing by suppressing the solid geometry.
B) Named selections:
The patches colored in red represents the faces which are subjected to subsequent boundary conditions in FLUENT solver. It helps to identify cell face zones and assignment of respective boundary conditions to it.
C) Mesh Details:
| Global element size | 7 mm |
| Maximum element size | 7 mm |
| Body fitted cartesian mesh | As per global settings |
| Face sizing | As per local settings |
| Local face mesh element size | 4.7 mm |
| Total number of nodes | 504117 |
| Total number of elements | 473504 |
Note: Same mesh is used for all case studies to expect consistent and mesh independent results. The mesh settings are selected under different trials to achieve accurate results.
1. Mesh:
2. Mesh (Sectional view):
The cartesian mesh is deployed because it produces smooth and consistent mesh elements for the swirling flow to occur.
D) Mesh quality:
1. Element quality:
2. Skewness:
3. Orthogonal quality:
II] Processing:
- Initially, the imported cartesian mesh is verified in the FLUENT solver using "Check Mesh" option.
- Steady state simulation is selected.
- Gravity is enabled in negative Y-direction.
- Viscous model: K-epsilon (RNG) by enabling Swirl Dominated Flow.
- Discrete Phase Model is selected. Following is the snapshot of the Discrete Phase Model (DPM) settings.
1. Turning on interaction with continuous phase and updating DPM sources every 10 iteration interval. Remaining settings are kept to be as default.
2. Creating particle injection
3. Setting injection properties and injecting the flow in normal direction through the face.
Note: Particle diameter and velocity magnitude may vary according to the case studies.
- Setting boundary conditions to be as follows,
Inlet: Velocity inlet = "Respective velocity as per each case set-up (m/s)"
DPM: Reflect (Here particle rebounds off the boundary with change in momentum. However the coefficicent of restitution in normal and tangential direction is by default set to one. Due to this in case the particle rebounds to the inlet boundary there is no change in momentum after reflection)
Top outlet: Pressure outlet = 0 gauge pressure (i.e. atmospheric pressure)
DPM: Escape (The particle is allowed to leave the outlet boundary with discontinuation of the trajectory calculations and particle fate is reported as "Escape")
Bottom outlet: Pressure outlet = 0 gauge pressure (i.e. atmospheric pressure)
DPM: Trap (The particle gets caught at the boundary face with termination of the trajectory calculations and particle fate is reported as "Trapped")
- Pressure-velocity coupling solution method is set to SIMPLE and discretization scheme for Turbulent Kinetic Energy (TKE) and Turbulent Dissipation Rate (TDR) is changed to Second Order Upwind. This is done to resolve TKE and TDR more accurately in the flow calculations.
- Solution Initialization: Standard Initialization is performed by computing the flow variabes from the inlet patch boundary.
- The numerical solution is time marched upto 800 iterations to achieve steady state solution.
III] Post-processing:
Case Study One: Varying particle diameter with same inlet velocity `(3 m/s)`.
1. Scaled residuals:
Case 1a:
Case 1b:
Case 1c:
2. DPM Report:
Case 1a:
Case 1b:
Case 1c:
3. Mass Flow Rate Report:
Case 1a:
Case 1b:
Case 1c:
4. Area Average Pressure Report:
Case 1a:
Case 1b:
Case 1c:
5. Pressure Contour:
Case 1a:
Case 1b:
Case 1c:
6. Pressure Contour Plane:
Case 1a:
Case 1b:
Case 1c:
7. Particle Residence Time:
Case 1a:
Case 1b:
Case 1c:
(Case Study One)
Observation table:
| Cases | Particle Diameter (microns) | Particle Velocity (m/s) | Particles Escaped | Particles Trapped | Incomplete Particles |
Efficiency `(eta_i)` |
Pressure drop (Pa) |
| Case 1a | 1 | 3 | 190 | 80 | 122 | 29.62 % | 20.7395 |
| Case 1b | 3 | 3 | 124 | 141 | 127 | 53.20 % | 20.7488 |
| Case 1c | 5 | 3 | 65 | 194 | 133 | 74.90 % | 20.7407 |
The separation effciency is given as,
`eta_i` = Trapped particles / Total number of particles
Note: The above calculated efficiency is calculated by neglecting the number of incomplete particles.
Technical discussion:
The particle diameter with 1, 3 and 5 microns with inlet velocity of 3 m/s are intended to study in this discussion. Results are tabulated in the above observation table. The smaller particles has less density as compared to the larger one and are less inertially influenced. The smaller diameter particles recirculate with zero descending angle at a certain height from the bottom. Accordingly they do not much interact with the separator walls. Due to lesser density, they tend to swirl with flue gases at the maximum axial velocity region (generally at zero radius or at the vertical mid section of the separator). The particle residence time of the smaller diameter sized particles is comparatively less. These particles have less centrifugal effects than that of gas drag force. Because of which they tend to escape with flue gases and does not get much influenced by the cyclonic effect. Whereas large sized particles possess comparatively higher inertia due to more density. The larger particles swirls and interact with separator walls as they are much dominated by the tangential velocity component / centrifugal effects. Due to this the particles slowly move down along the separator walls with a certain descending angle. Due to higher centrifugal influence they do not spin in the mid section of the separator. However these particles spend pretty much amount of time in the region of cyclone as compared to the smaller particles. As a function of time the larger particles loose momentum and inertial energy and fall off to the bottom (where they are trapped) under gravitational influence. The separation efficiency tends to increase with the larger particles as compared to smaller ones. These smaller diameter sized particles tends to escape with the gas flow and almost 1/3rd of these are collected at the bottom. In contrast to it, larger particles show good trapping separation efficiency as they separate easily from the mixed flow under cyclonic conditions. Accordingly resulted pressure drop is similar for all cases as this study evaluates the case on varying particle size with same inlet velocity.
Case Study Two: Varying inlet velocity with particle diameter being 5 `mum`.
1. Scaled Residuals:
Case 2a:
Case 2b:
Case 2c:
2. DPM Report:
Case 2a:
Case 2b:
Case 2c:
3. Mass Flow Rate Report:
Case 2a:
Case 2b:
Case 2c:
4. Area Average Pressure Report:
Case 2a:
Case 2b:
Case 2c:
5. Pressure Contour:
Case 2a:
Case 2b:
Case 2c:
6. Pressure Contour Plane:
Case 2a:
Case 2b:
Case 2c:
7. Particle Residence Time:
Case 2a:
Case 2b:
Case 2c:
(Case Study Two)
Observation Table:
| Cases | Particle Diameter (microns) | Particle velocity (m/s) | Particles Escaped | Particles Trapped | Incomplete Particles |
Efficiency `(eta_i)` |
Pressure drop (Pa) |
| Case 2a | 5 | 1 | 166 | 7 | 219 | 4.04 % | 1.7692 |
| Case 2b | 5 | 4 | 63 | 317 | 12 | 83.42 % | 38.0338 |
| Case 2c | 5 | 5 | 56 | 334 | 2 | 85.64 % | 60.5657 |
Technical discussion:
This case is intended to study the effect of gas inlet velocity on the separation efficiency with same diameter sized particles. Results are tabulated in the above observation table. The gas inlet velocity has a direct impact on the pressure drop which can be observed in the tabulated results. More the gas inlet velocity; more will be the pressure drop. The case with lesser velocity (case 2a) results in lesser pressure drop but significantly affects the separation efficiency. It is due to the fact that lesser pressure drop contributes to less acceleration of the particles to the outflow. The inertial influence on the particles will significantly reduce due to less acceleration. As we know, due to increase in acceleration the velocity increases and subsequently also the mass. And increased mass improvize inertial forces. Because of this, particles tend to swirl in the upper mid section of the separator due to less inertia and force balance due to gravity. The balanced force distribution leads to the rotation of the particles at a certain height and hinders the particle escape / trap. Accordingly the particles are more likely to escape from the top outlet due to comparatively larger outlet cross section and high pressure drop at the top outlet region. These particles also do not much interact the separator walls and swirl in the center of the walls (around the axis at lesser radial distance). Conversely in case with higher velocity (case 2b and 2c) the resulted pressure drop is much higher. Due to this, the particles possess higher acceleration to the outflow and enhances the inertial effects. The particles with higher velocity and accleration (in terms of increased mass) increases the significance of centrifugal force. These particles will swirl with higher tangential velocity component and will interact with the separator walls. The particle residence time is also very less as compared to the case with lesser velocity. Due to increased inertial effects the particles rushes to the bottom outlet under the influence of gravity. This causes more number of particles to get trap at the bottom and very less particles escapes, which certainly increases the separation efficiency.
Conclusion:
The Euler-Lagrangian model has been used to simulate the case of cyclone separator with K-epsilon (RNG) turbulence model. The separation efficiency and pressure drop has been calculated for various cases with vaying particle diameter and inlet velocity. The obtained results are promising and are able to imitate appropriate physics. The following conclusions can be drawn from the present study:
Case study one: Varying particle diameter with same inlet velocity `(3 m/s)`.
1. The smaller particles tend to escape due to less density and inertia which results in lower separation efficiency. In contrast to it, larger particles possess higher density and inertial effects due to which they get collected to the bottom under the gravitational influence.
2. The particle residence time is more in case of larger particles.
3. Pressure drop is almost same for all the cases.
Case study two: Varying inlet velocity with particle diameter being 5 `mum`.
1. The particles with lesser velocity escapes or get stucked in the re-circulation region due to less pressure difference and leads to less separation efficiency. However the particles with high velocity are accelerated to the outlet due to high pressure difference and get trapped at the bottom. This certainly increases the separation efficiency for the particles with high velocity.
2. The particle residence time is more in case of particles with lesser velocity.
3. Pressure drop increases with increase in gas inlet velocity. It increases the separation efficiency of the cyclone separator. However too much pressure drop increases the operating costs and conditions of the cyclone separator.
References:
1. Numerical study of gas-solid flow in a cyclone separator.
https://www.sciencedirect.com/science/article/pii/S0307904X06000291
2. A CFD Study on the Prediction of Cyclone Collection Efficiency.
3. Case set-up help.