Elsevier

Powder Technology

Volume 382, April 2021, Pages 70-84
Powder Technology

Combining 10 meta-heuristic algorithms, CFD, DOE, MGGP and PROMETHEE II for optimizing Stairmand cyclone separator

https://doi.org/10.1016/j.powtec.2020.12.056Get rights and content

Highlights

  • To improve Stairmand cyclone performance, CFD, DOE, MGGP and ten meta-heuristic algorithms were combined.

  • 17 cyclones with different dimensions based on DOE are simulated by CFD.

  • Two models including the collection efficiency and pressure drop are obtained by MGGP.

  • The obtained models are optimized by ten meta-heuristic algorithms to achieve the cyclones with high performance.

  • MOPSO algorithm generated more optimal designs, which dominated the Stairmand design.

Abstract

Gas cyclone separators have been widely used in different industries. In this study, to find the best geometrical ratios of Stairmand cyclone separator, computational fluid dynamics (CFD), design of experiments (DOE), multi-gene genetic programming (MGGP), and ten meta-heuristic algorithms were combined. Six geometrical dimensions of the gas cyclone separator including inlet height and width, vortex finder length and its diameter, cylinder height and cone-tip diameter were optimized. The obtained models from MGGP were optimized by ten meta-heuristic algorithms and non-dominated Pareto fronts were analyzed using six unary and binary metrics and PROMETHEE II as a decision making method. According to the optimization results, multi-objective Particle Swarm Optimization (MOPSO) showed the best performance and generated more preferred designs than Stairmand design compared to other algorithms. These preferred designs increased the collection efficiency within 0.36 to 6% and decreased the pressure drop within 3.3 to 27.5% compared to the Stairmand.

Introduction

Low cost and simple design of gas cyclone separators are primary advantages to make them widespread for removing particles from a gas stream. In addition, the cyclone can be adapted for using in extreme operating conditions such as high pressure and temperature. Cyclone performance is highly dependent on its configuration and dimensions. The main parameters of cyclone performance are pressure drop and particle collection efficiency with the aim of increasing the efficiency and decreasing the pressure drop. Three main sources of pressure loss are cyclone inlet (inlet loss), cyclone body (friction loss) and vortex finder (outlet loss). The existence of a swirling gas flow through vortex finder is the main source of pressure loss. By increasing body and conical heights, the most kinetic energy stored in the tangential component is lost due to dissipation (friction loss) and it is a reason for decreasing the outlet loss so that a decrease in overall pressure loss within the cyclone is observed [1,2]. Gas rotation in the cyclone causes centrifugal force and makes denser particles suspended in the gas move toward the cyclone wall. By increasing the inlet velocity to saltation value, the collection efficiency also increases [3]. There are two inner and outer vortexes in a cyclone. There is a downward motion close to the wall and movement direction changes upward in a distance from the axis. The axial and tangential velocities inside the cyclone are important, tangential velocity is the factor that affects the centrifugal force exerted on the particle and axial velocity indicates the presence of two upward and downward streams inside the cyclone [4]. Iozia and Leith [5] considered the effect of cyclone dimensions on gas flow pattern and cyclone efficiency and based on the gas flow pattern within the cyclone, they predicted particle cut diameter.

The types of inlet like axial [6] or symmetrical spiral [7] affect cyclone performance. Su et al. [8] studied different square cyclone separator inlets. Their results showed that the inlet configuration influences the pressure, velocity and turbulent kinetic energy distribution. The highest collection efficiency was achieved by single normal inlet, whereas double declining inlets had the minimum pressure drop.

The pressure drop and the collection efficiency are the most important operating parameters of a cyclone and should be optimized simultaneously. Many studies have been done to investigate the effect of geometrical dimensions on these parameters [4,5,9]. Demir et al. [1] studied the effect of cylindrical and conical heights of a cyclone on pressure drop. Experimental results showed that by increasing cylindrical and conical heights, the pressure drop decreases; however, according to the results the best heights for cylindrical and conical portions were about 1.5D and 2.5D, respectively. Sun and Yoon [10] considered a multi-objective optimization using genetic algorithm (GA) and CFD simulations to maximize the efficiency and minimize the pressure drop. They studied the impact of several cyclone geometries on its performance and according to their results, increasing the inlet height and width, vortex finder diameter and surface roughness led to reduce significantly both the efficiency and pressure drop, while increasing cylindrical height and cone diameter caused to reduce the pressure drop and increase the efficiency slightly. The first four effective factors were considered to define the fitness function by using central composite design in the response surface methodology (RSM) and for optimizing these four factors the second-generation non-dominated sorting genetic algorithm (NSGA-II) was applied.

Elsayed et al. [11,12] optimized the cyclone geometries to decrease the Euler number and cut-off diameter instead of decreasing pressure drop and increasing collection efficiency. In this investigation, the RSM method and Radial Basis Function Neural Networks (RBFNNs) were employed as well as NSGA-II algorithm. The effects of four geometrical dimensions including the inlet height, inlet width, vortex finder diameter and total height on cyclone performance were investigated. According to their CFD results, cut-off diameter and Euler number in new design decreased about 60.1 and 53.5%, respectively than in Stairmand design. An another study [12] based on Iozia and Leith model and NSGA-II algorithm was performed to determine the effect of seven geometrical parameters on the cyclone performance using two RBFNNs, which were trained by experimental data available in literature. The results revealed that Euler number and cut-off diameter decreased 3.48 and 19.6% in new design compared in the Stairmand. Sun et al. [13] optimized the pressure drop and the cut-off diameter and considered the effect of 10 geometrical factors by CFD simulations and RSM. Their results showed that the pressure drop and cut-off diameter were reduced 20.7 and 75.38%, respectively in new design compared in the Stairmand. Sgrott et al. [14] studied the effect of seven geometrical factors on the performance of the cyclone using COMPLEX (Constrained simplex) algorithm and CFD simulations. Their results showed that the collection efficiency increased and the pressure drop decreased significantly in comparison with those of the Stairmand and Lapple cyclones. Singh et al. [15] studied the effect of seven design variables on Euler and Stokes numbers of cyclone. Three surrogate models including Kriging, radial basis functions and support vector regression were coupled with the Evolutionary multi objective optimization (EMO) algorithm and the obtained results were compared with NSGA-II as traditional and well-known algorithm. The results indicated the Pareto set designs found by EMO outperformed the designs found by NSGA-II while EMO needed significantly fewer calculations. Kumar et al. [16] focused on vortex finder in a cyclone and by reshaping this section and considering five parameters including three diameters at different axial locations and the distances between these locations, enhanced the performance of the cyclone. In this investigation, both RSM and GA techniques were used. According to the results, the pressure drop decreased and collection efficiency increased by about 14.2 and 1.7% respectively compared to the reference design; in addition, genetic algorithm provided better solutions than RSM.

Safikhani [17] investigated three geometrical dimensions to optimize the design of the Karagoz cyclone by employing the CFD simulations and ANNs (GMDH type) accompanied by NSGA-II algorithm. According to one of the obtained design, the pressure drop and collection efficiency were determined about 356.73 Pa and 80.64%, respectively. These results were significantly better than the results obtained previously. Fathizadeh et al. [3] studied on the optimization of the vortex-finder length in a high pressure gas cyclone to improve its collection efficiency. The model of Lieth and Licht was applied to determine the cyclone collection efficiency and pressure drop. Their results indicated that based on CFD simulations, both pressure drop and collection efficiency increased in new design compared to Stairmand. Kashani et al. [18] optimized three dimensions of a cyclone in a cement plant preheater system including vortex-finder length, cylindrical height and cone tip diameter to increase the cyclone efficiency and decrease the pressure drop. In this study, CFD simulations, multi-gene genetic programming (MGGP) (the computer program GPTIPS toolbox in MATLAB) and multi-objective genetic algorithm were employed to obtain the Pareto optimal solutions. According to results of CFD simulation of one point in Pareto solutions, the efficiency was increased by about 13.4% while the pressure drop of designed cyclone was decreased by about 2.2%. Although various studies have been conducted to find better cyclone geometrics, none of them compared different optimization algorithms using different metrics. In this study, the effect of geometrical dimensions on the performance of Stairmand cyclone was investigated and ten meta-heuristic algorithms were applied and compared to find the best cyclone dimensions.

This study was aimed to investigate the effect of geometrical dimensions on the performance of Stairmand gas separator cyclone. The collection efficiency and pressure drop of several cyclones with different geometrics were determined using design of experiments and CFD simulations. Two equations for the efficiency and pressure drop were derived using MGGP. The equations were optimized by ten meta-heuristic algorithms and the results of algorithms were compared by several metrics and the best one was chosen to find the optimum design for Stairmand cyclone.

Section snippets

Stairmand cyclone

The geometrical ratios that presented by Stairmand in 1951 [19] is still in use. These ratios for high efficiency Stairmand design are given in Table 1. They consist of vortex finder length (S) and its diameter (Dx), inlet height (a) and width (b), cylindrical height (h), conical height (hc), and the cone-tip diameter (Bc). A schematic diagram of Stairmand cyclone is illustrated in Fig. 1. Most optimization researches had been done with the purpose of finding a higher cyclone efficiency than

Methodology

The CFD simulations of Stairmand cyclone were run by Ansys Fluent 19.2. For validation of CFD, the results were compared to the experimental data of Hoekstra [23]. A Stairmand cyclone with 0.29 m diameter were considered, the inlet air volume flow rate, the viscosity, density and operating pressure were 0.1374 m3/s, 1.81×10−5 Pa.s, 1.204 kg/m3 and 101,325 Pa, respectively.

Results and discussion

First, the CFD results of the cyclone and DOE are given in this section. Then, ten optimization algorithms are compared to optimize the dimensions of the cyclone and choose the best algorithm.

Conclusions

In this study, in order to find geometrical ratios that improve the performance of Stairmand cyclone, CFD simulations and DOE accompanied by multi-gene genetic programming, and 10 meta-heuristic algorithms were employed. First, by using the results of CFD simulations, the effect of six geometrical dimensions on cyclone performance was investigated based on DOE and then to optimize geometrical ratios, two models for pressure drop and collection efficiency were obtained using MGGP. The obtained

A. Supplementary data

Supplementary material

Declaration of Competing Interest

None.

Acknowledgments

The authors would like to express their appreciation to the management of computer center of Chemical Engineering Department, Shahid Bahonar University of Kerman, Kerman, Iran for supporting this work.

References (56)

  • O.L. Sgrott et al.

    Cyclone optimization by COMPLEX method and CFD simulation

    Powder Technol.

    (2015)
  • X.S. Yang et al.

    Multi-objective flower algorithm for optimization

  • M. Hamdy et al.

    A performance comparison of multi-objective optimization algorithms for solving nearly-zero-energy-building design problems

    Energy Build.

    (2016)
  • S. Demir et al.

    Effects of cylindrical and conical heights on pressure and velocity fields in cyclones

    Powder Technol.

    (2016)
  • D. Kahrimanovic et al.

    Minimizing pressure drop in cyclone separators - measurements and numerical simulations

  • N. Fathizadeh et al.

    Design and simulation of high pressure cyclones for a gas city gate station using semi-empirical models, genetic algorithm and computational fluid dynamics

    J. Nat. Gas Sci. Eng.

    (2015)
  • K. Elsayed et al.

    The effect of cyclone vortex finder dimensions on the flow pattern and performance using LES

    Comput. Fluids.

    (2013)
  • D.L. Iozia et al.

    Effect of cyclone dimensions on gas flow pattern and collection efficiency

    Aerosol Sci. Technol.

    (1989)
  • Z. Xiong et al.

    Development of a cyclone separator with high efficiency and low pressure drop in axial inlet cyclones

    Powder Technol.

    (2014)
  • B. Zhao et al.

    Development of a symmetrical spiral inlet to improve cyclone separator performance

    Powder Technol.

    (2004)
  • Y. Su et al.

    Numerical simulation of effect of inlet configuration on square cyclone separator performance

    Powder Technol.

    (2011)
  • K. Elsayed et al.

    Modeling, analysis and optimization of aircyclones using artificial neural network, response surface methodology and CFD simulation approaches

    Powder Technol.

    (2011)
  • X. Sun et al.

    Multi-objective optimization of a gas cyclone separator using genetic algorithm and computational fluid dynamics

    Powder Technol.

    (2018)
  • K. Elsayed et al.

    CFD modeling and multi-objective optimization of cyclone geometry using desirability function, artificial neural networks and genetic algorithms

    Appl. Math. Model.

    (2013)
  • K. Elsayed et al.

    Modeling and Pareto optimization of gas cyclone separator performance using RBF type artificial neural networks and genetic algorithms

    Powder Technol.

    (2012)
  • X. Sun et al.

    Multi-objective optimization of a Stairmand cyclone separator using response surface methodology and computational fluid dynamics

    Powder Technol.

    (2017)
  • P. Singh et al.

    Shape optimization of a cyclone separator using multi-objective surrogate-based optimization

    Appl. Math. Model.

    (2016)
  • V. Kumar et al.

    Multi-objective shape optimization of vortex finders in cyclone separators using response surface methodology and genetic algorithms

    Sep. Purif. Technol.

    (2019)
  • H. Safikhani

    Modeling and multi-objective Pareto optimization of new cyclone separators using CFD, ANNs and NSGA II algorithm

    Adv. Powder Technol.

    (2016)
  • E. Kashani et al.

    CFD simulation of the preheater cyclone of a cement plant and the optimization of its performance sing a combination of the design of experiment and multi-gene genetic programming

    Powder Technol.

    (2018)
  • C. Stairmand

    The design and performance of cyclone seperators

    Trans. Instn. Chem. Engrs.

    (1951)
  • F. Inc.

    Fluent 6.3 - User’s guide

    (2006)
  • W.D. Griffiths et al.

    Computational fluid dynamics (CFD) and empirical modelling of the performance of a number of cyclone samplers

    J. Aerosol Sci.

    (1996)
  • S.K. Shukla et al.

    The effect of modeling of velocity fluctuations on prediction of collection efficiency of cyclone separators

    Appl. Math. Model.

    (2013)
  • A.J. Hoekstra

    Gas flow field and collection efficiency of cyclone separators, TU delft, Ph. D. Thesis

    Delft Univ. Technol.

    (2000)
  • M. Carmona et al.

    A numerical study of the flow in a cyclone separator using the K - e reliazible turbulence model

    V Eur. Conf. Comput. Fluid Dyn. ECCOMAS CFD

    (2010)
  • T.G. Chuah et al.

    A CFD study of the effect of cone dimensions on sampling aerocyclones performance and hydrodynamics

    Powder Technol.

    (2006)
  • A. Kȩpa

    The efficiency improvement of a large-diameter cyclone - the CFD calculations

    Sep. Purif. Technol.

    (2013)
  • Cited by (11)

    • Improvement of the cyclone separator performance by the wedge-shaped roof: A multi-objective optimization study

      2023, Chemical Engineering Science
      Citation Excerpt :

      Therefore, many researchers have carried out multi-objective optimization studies on the cyclone separator to find the structure with the best performance. Some researchers have used the design of experiments (DOE) to sample the design variables, and then used data processing methods such as the response surface methodology or neural network to obtain predictive models of optimization objectives, and finally used genetic algorithms to carry out the multi-objective optimization to generate Pareto front points, which can be selected to obtain the cyclone separator with the optimal structural size for the performance (Brar and Elsayed, 2018; Brar and Elsayed, 2017; Misiulia et al., 2017; Elsayed et al., 2020; Kashani et al., 2018; Izadi et al., 2021; Park et al., 2020). Sun et al. (Sun and Yoon, 2018; Sun et al., 2017) used screening experiments to select a few important factors for several structural parameters of the cyclone separator, obtained computational models using the response surface methodology based on the central composite design, and carried out the multi-objective optimization of the pressure drop and collection efficiency using genetic algorithms and response surface optimization methods.

    • Surrogate modeling method for multi-objective optimization of the inlet channel and the basin of a gravitational water vortex hydraulic turbine

      2023, Applied Energy
      Citation Excerpt :

      One part of the data is used for training the algorithm and the remaining one, called the validation sample, is used for calculating the risk of the algorithm to estimate the value of the validation samples [61]. HV measures the quality of non-dominated solutions by calculating the area of the region faced by the non-dominated solutions and enclosed above by a reference point [27,62]. Based on the quality of the model, the designer decides if further samples are needed.

    View all citing articles on Scopus
    1

    Ahad Izadi and Elham Kashani shared the first authorship and contributed equally to this paper.

    View full text