Experimental and numerical investigation on particle deposition in a compact heat exchanger

Accepted Manuscript Research Paper Experimental and Numerical Investigation on Particle Deposition in a Compact Heat Exchanger S. Baghdar Hosseini, R. Haghighi Khoshkhoo, M. Javadi PII: DOI: Reference:

S1359-4311(16)34376-9 http://dx.doi.org/10.1016/j.applthermaleng.2016.12.110 ATE 9730

To appear in:

Applied Thermal Engineering

Received Date: Revised Date: Accepted Date:

28 August 2016 20 November 2016 26 December 2016

Please cite this article as: S. Baghdar Hosseini, R. Haghighi Khoshkhoo, M. Javadi, Experimental and Numerical Investigation on Particle Deposition in a Compact Heat Exchanger, Applied Thermal Engineering (2016), doi: http:// dx.doi.org/10.1016/j.applthermaleng.2016.12.110

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Experimental and Numerical Investigation on Particle Deposition in a Compact Heat Exchanger S. Baghdar Hosseini1, R. Haghighi Khoshkhoo2, *, M. Javadi3 1,2 3

Department of Mechanical and Energy Engineering, Shahid Beheshti University, Tehran – Iran

Department of Mechanical Engineering, Quchan University of Advanced Technologies, Quchan-Iran

* Corresponding author

E-mail: [email protected]

Telephone No: (+98) 912 564 49 02

Abstract In this study the effect of particle size on deposition in compact heat exchanger was investigated experimentally and numerically. An experimental setup was designed to visualize particle deposition and measure pressure drop across the exchanger. Numerical study was performed on five fin channels. The flow was modeled by solving Reynolds-Averaged Navier-Stokes (RANS) equations, and particle motions were simulated by discrete particle model (DPM) with UDF to model deposition. Experimental study was performed for particle size over a range from 1 μm to 4 mm and numerical investigation were done for particle size from 1 μm to 100 μm which were placed in A1 particle group of experiment. Experimental results show enhancement of particle deposition; besides, pressure drop rises with increase of particle size. Numerical study also demonstrates that particle deposition increases with increase of particle size up to 50 μm. Studies show that velocity increase pressure drop and can promote or hinder particle deposition. Keywords Compact heat exchanger, CFD analysis, DPM, Experimental investigation, Particle deposition, Pressure drop Nomenclature Cc

Cunningham correction

Up

Particle velocity

d

Diameter

V

Volume

dij

the deformation tensor

vrel

relative velocity of particle-fluid

X

Position

gravitational acceleration k

turbulent kinetic energy

Greek Symbols

ks

sliding ratio

δij

Kronecker delta

KB

Boltzmann constant

ε

turbulent dissipation

p

Pressure

ζi

zero-mean, unit-varianceindependent Gaussian random numbers

1

Pb

turbulence kinetic energy generation due to buoyancy

θ

Impact angle

Pk

turbulence kinetic energy generation due to the mean velocity gradients

μ

Dynamic Viscosity

r0

contact radius

ν

Kinematic viscosity

S

fluid to particle density ratio

ρ

Density

Sk

user-defined source term for k

τp

particle relaxation time

Sɛ

user-defined source term for ε

Ω

Specific dissipation rate

Sn, ij

Spectral intensity

Subscripts

t

time

f

Fluid

T

Temperature

p

Particle

u

Fluid velocity

t

Turbulent

1. Introduction Air-side particulate fouling is known as accumulation of solid particles which are suspended in the air onto the heat transfer surface [1]. Particles settle on surfaces due to various mechanisms such as gravity and turbulent diffusion. Dust deposition on air cooled condensers such as car radiator, unburned fuels or ashes deposition on boiler tubes etc. are examples of particulate fouling. Particulate deposition form a layer on surfaces which may influence pressure field of stream and heat transfer process and consequently reduce performance of compact heat exchangers (CHEs). So, cleaning CHEs over time is essential to remain its efficiency unchanged but it is a time consuming and costly process. Understanding particulate fouling process, the mechanisms and parameters involve it can help us to relieve its adverse effects. Many experiments have been performed to understand air-side fouling process [2-12] and some numerical studies have shown the effect of air-side fouling on heat exchangers performance. Numerical studies employs computer for calculations and implementing mathematical model for a physical system [13]. Currently various approaches are introduced to model fluid flows, such as finite difference method (FDM), finite volume method (FVM), finite element method (FEM), control volume based finite element method (CVFEM), Lattice Boltzmann method (LBM) simulation, molecular dynamics simulation, and direct simulation Monte Carlo. The first five methods are widely employed approach in computational fluid dynamics (CFD) [14]. However application of FDM can make some difficulties when geometries are complex [15]. The FEM schemes can be intricate for solving conservative equations; however, the nonstandard FEMs have low computational efficiency [16]. FVM can be implemented in complex physics as it is more accurate and less time consuming compared to other methods; however application of FVM is difficult to cases with complicated moving boundaries [17]. CVFEM method are developed in order to solve fluid flow and heat transfer

2

problems especially in complicated geometries; however, it is very difficult to extend the model to 3 dimensions [18]. LBM is developed to solve compressible flow of ideal gases, it can also theoretically simulate the compressible Navier-Stokes equations. LBM can be modified with the Chapman-Enskog expansion in order to simulate incompressible flow for low Mach numbers (Ma < 0.15) [19-21]. LBM uses regular square grids and makes it difficult to simulate curved boundaries [22]. Chamkha and Rashad [23] studied the flow of a nanofluid around a non-isothermal wedge using FDM and they considered the Brownian movement and the thermophoresis effects. They concluded that the presence of the Brownian motion and the thermophoresis effects caused the local Nusselt number to decrease and the Sherwood number to increase. Selimefendigil and Oztop [24] applied finite element method to investigate MHD mixed convection of nanofluid filled partially heated triangular enclosure with a rotating adiabatic. Sheikholeslami et al. [25-27] used Control volume based finite element method to study the effects of a magnetic field on natural convection in different enclosures filled with nanofluids; besides, they examined constant temperature and heat flux boundary condition for Al 2O3– water nanofluid filled enclosure [28–31]. Sheikholeslami et al. [32-36] applied Lattice Boltzmann method to study nanofluid hydrothermal behaviors in an enclosure with curve boundaries. Sheikholeslami et al. [37, 38] studied enhancement of turbulent heat transfer in heat exchangers by means of FVM and they used Renormalization group (RNG) k–ε model to simulate turbulence flow. Haung et al. [39] developed a numerical model using FVM to describe the main deposition processes of ash particles with a predetermined chemical composition. They coupled their deposition model to the FLUENT CFD package and used a Lagrangian method to model two-phase flow. They also used a Rosin-Rammler particle size distribution. Their results showed that the denser particles were more inclined to deposit than less dense particles. They also showed that boiler geometry act as a controlling factor in the slagging process. Han et al. [40] also used finite volume method to study the effect of exchanger geometry on fouling rate, heat transfer and hydrodynamics performance. They developed a numerical model based on FLUENT software and extended of User Defined Functions to predict flue-ash particle deposition rate on a tube bundle and showed that particles accumulated primarily in the flow stagnation region, recirculation region, the vortex separation and reattachment regions. Wacławiak et al. [41] introduced a practical numerical approach for prediction of particulate fouling in pulverized coal-fired boiler. They presented results of 2D modeling of powdery, mediumtemperature deposit formation on superheater tubes. The simulations were carried out with use of FLUENT code by means of UDFs. They compared their results with real measurements in boiler and showed good agreements. The understandings of recent studies show that air-side fouling process is very complicated and depends on many factors such as particle type and particle size as well as exchanger geometry. Particles have different approach in facing exchangers according to their size; they either pass through exchangers or block among them and create nucleation points of deposition. In order to predict the behavior of gas-solid flows various types of CFD models are available. In basics, two approaches are in use which their name are the Eulerian–Eulerian method for granular

3

flows (two-fluid model-TFM) and the Eulerian–Lagrangian method. Under the Eulerian Lagrangian approach, the Lagrangian discrete phase model (DPM), dense discrete phase model incorporated with the kinetic theory of granular flow (DDPM), and discrete element method (DEM) are available. In these models, the way of treating particle-particle interactions and the numerical method used to solve the equations are different. Each model has advantages and limitation, so suitable model should be chosen according to the factors that are in priority [42]. For instance, the Eulerian-Eulerian method requires fewer computational resources compared to Eulerian-Lagrangian approaches [43]. However, this approach has major limitations in considering variations of particle properties, for example, wide particle size distribution, density diversification and sphericity consideration. In contrast, the Eulerian-Lagrangian method can provide analysis of flows with a wide range of particle types, sizes, shapes and velocities but at the cost of computation [44]. Among Eulerian-Lagrangian methods, DPM is the most promising one as it is less CPU expensive, and has no stability and convergence problem [45]; however, DPM are not suitable for dense fluid-particle flows because of the restriction on the volume fraction of the discrete phase. The objective of this study is to investigate the effect of particle size on deposition. The effect of flow velocity on particle deposition and the mechanism of particle deposition for different particle size are also studied and discussed. 2. Experimental Procedure The experimental setup is illustrated in Fig. 1. It consists of a circular inlet with a diameter of 0.5 m and 1 m length which is attached to a laminarization chamber with a diameter of 1.5 m. The test rig includes a wind tunnel with 40 cm×40 cm cross-section that is made of plexiglass which allows visualization of the particulate fouling process. An industrial vehicle radiator is located at the end of this section and its characteristics are mentioned in table 1. In order to prevent air leakage, all parts of the test rig is insulated. The experimental setup is made of steel and it is connected to the earth in order to eliminate possible static electricity which can enhance the accumulation of particles. Air flow is entered into the test rig by a blower and is exited through a circular channel with a diameter of 0.5 m and 3 m length. The blower is coupled to DC motor of 6 kW with adjustable rotational speed over the range from 50 to 1660 rpm. Air velocity can be changed from 1 m/s to 5 m/s and is measured by means Pitot tube which is located near the exchanger. Pressure tapping is drilled into the tunnel floor upstream and downstream of the exchanger and is connected to a differential pressure measuring instrument in order to measure pressure gradient before and after fouling. The specification of measurement instruments is shown in table 2. In each experiment, a group of particles is weighted by a lab digital balance and then are introduced into the test rig from tunnel roof just over 1 m upstream of the radiator. Introduced particles are wood shavings which are sorted and classified into six different sizes by a sieve shaker with various sieve net size. Table 3 show particle size distribution in which the upper limit of each group represents the size of sieve net. During experiments, particles is

4

introduced into the test rig in discrete lots instead of continuously because of technical difficulties and cost of such a setup. Hence, a total mass M of particles are divided into n definite group then they are injected into the wind tunnel. Numerous experiments were carried out for different rates of particle introduction using accurate and calibrated devices and show that the particles either passed through the exchanger, were blocked on the exchanger surface, or rebounded to the tunnel floor upon impact. Each of these components is weighed and compared to the amount introduced. Photographs of the surface is also taken for analysis. The rate of deposition is calculated by dividing the mass of deposited particle to the total mass. Observation showed that gravitation affect the passage of particles from the injection point; as a result, a small zone on the above side of the radiator remains clear. In order to eliminate the consequent errors from this dead zone, it is insulated. So all results of experimental study refer to the “active” or fouled area of the exchanger surface [46]. In order to ensure the accuracy of the results, each experiment is repeated three to five times and the uncertainty are calculated according to the following equations [47]. (1) (2) (3)

Where

is the arithmetic mean value of measured data, n is the number of measurements,  is the

standard deviation, and m is the mean standard deviation. Data analysis showed uncertainty less than 5%.

3. Numerical Methods and Procedure In the following section physical model and computational domain is first provided, then boundary conditions and the basic governing equations of continuous phase flow as well as turbulence model is presented. Finally, the discrete phase model which is used to solve particle trajectories is introduced. 3.1. Physical Model A 3D schematic of five fin channels of CHE is illustrated in Fig. 2. As it is seen computational domain is extended upstream and downstream for a distance of five times the fin base of channels in order to ensure the uniformity of flow at inlet and after fin channels. Total number of grids on computational domain is about 66.5×104. Simulations with finer grids show that the quality of prediction is not improved by enhancing the number of cells used. The air flow enters from upstream with velocity over a range from 1 m/s to 5 m/s and particles introduce to the computational domain having various diameter of 1 μm, 10 μm, 30 μm, 40 μm, 45 μm, 50 μm, 60 μm, and 100 μm. The computational domain has six boundaries: inlet, outlet, wall surfaces (top and bottom) and periodic

5

boundary surfaces (left and right). Air flow is entered to the computational domain from inlet boundary with uniform velocity and the turbulent intensity (I) of 5% and particles are injected normal to the inlet boundary condition surface during certain time. At the fins and wall surfaces (bottom and top), no slip conditions for the velocity is assumed. 3.2. Fluid phase governing equations In this work CFD modeling is performed using ANSYS Fluent [48] and Reynolds-Averaged NavierStokes (RANS) equations are solved in three-dimensional domain using second-order discretization scheme. In order to apply suitable turbulence model, the effect of using different turbulence model on velocity and pressure is studied. It can be seen from following Fig. 3 and Fig. 4 that all models show good results and each of them can be used. The plotted results of Fig. 3 and Fig. 4 are acquired from a line in the middle of numerical domain, which is placed exactly between two fins. Turbulence is modeled with standard k-epsilon model with standard wall functions because it has been proved to be able to predict the average velocity profiles, pressure drop, and shear stresses successfully for the multi-regime flow field [49, 50]. The standard wall function is known to be effective and robust for 30< Y+<300 and when using this model Y+ value should ideally be above 15 to avoid erroneous modelling in the buffer layer and the laminar sub-layer. The range of Y+ for different velocity is shown in table 4. It shows that Y+ in all cases are acceptable for standard k-epsilon model. However, the effect of Y+ on velocity is studied. As it is depicted in following Fig. 5, with implementation of standard wall function, Y+ has negligible effects on velocity. A pressure based unsteady solver is used for the calculation and pressure-velocity coupling is done by SIMPLE scheme [51]. Air properties in CFD modeling are assumed to be constant and incompressible. Equations (4a) and (4b) represent RANS equations governing the conservation of mass and momentum. (4a) (4b) .

Turbulent kinematic viscosity, νT, is given in equation (5), in which cμ is a constant with an accepted value of 0.09, and fμ is a function of RT=k/νω, as defined in equation (3). (5) (6) K-epsilon turbulence model equations are given in Eqs. (7a) and (7b) in which YM shows the contribution of the fluctuating dilatation in compressible turbulence to the overall dissipation rate, σk

6

and σɛ represent the turbulent Prandtl numbers for k and epsilon and are 1 and 1.3 respectively. C1ɛ, C2ɛ and C3ɛ are constant and have value of 1.44, 1.92, -0.33 respectively [49]. (7a) (7b)

3.3. Particle phase governing equations In CFD modeling an Eulerian approach is used to solve the continuous phase flow field. Particle phase can be solved either on an Eulerian basis or a discrete Lagrangian approach. It is well understood that inertia is one of the major mechanisms on transport and deposition of particles and its effect on deposition rate cannot be modeled on Eulerian basis [45, 52]. So, Lagrangian Discrete Phase Method (DPM) is adopted for particle phase. The general form of the trajectory equations for a single particle with constant mass is as presented in equations (8a) and (8b). (8a) (8b) In which

is position vectors of particle. The right-hand side of equation (8a) is gravity–buoyancy,

drag, Saffman lift, Brownian and virtual mass forces per unit mass of the particle, respectively. In drag force

represents unit vector in the direction of fluid velocity relative to the particle. Saffman

lift force becomes considerable for micro-particle in high velocity gradient area [53] and its general expression is presented in equation (9); in which K represents a constant with an accepted value of 2.594 [52]. (9) Brownian motion can be accounted for sub-micron particle and is modeled as a Gaussian white noise process with spectral intensity given in equation (10a) and (10b). However, its effect is considerable in laminar flow and it can be neglected in contrast with other forces effects on particle deposition. Amplitudes of the Brownian force components can be written in the form of equation (11) [54]. (10a) (10b)

(11)

7

Virtual mass force expression is represented in equation (12), in which Vp is particle volume. (12) In this study the effect of Saffman lift force, virtual mass force and pressure gradient force on the particle trajectories are investigated and the influence of Brownian force and other forces are neglected. Influence of instantaneous turbulent velocity fluctuations on particle trajectories can be predicted using stochastic tracking (random walk model). These values are derived from a Gaussian probability distribution. In this study Discrete Random Walk (DRW) model or eddy lifetime model is used to account turbulent velocity fluctuations [55]. ANSYS Fluent program cannot stand alone to simulate fouling process on heat transfer surfaces; therefore some user defined subroutines which are necessary is implemented as follow [48]: 

Computation of the deposition mass rate



Number of particles impact to the heating surfaces



Computation of the density of the fouling layer



Computation of the total rate and layer thickness

3.4. Deposition Mechanism Fig. 6 shows the process of particle impact with velocity Up and angle θ on the heating surface. When a particle is brought to the wall to contact it, a force balance is made as particle deposition criterion. The forces that are worked on a particle in contact with the wall are shown in Fig. 6. Three ratios for normal, tangential rolling and sliding are computed for each particle according to the following equations [56]. If one or more of the ratios is larger than unity, the particle will rebound; otherwise it will stick to the surface.

(13)

(14)

(15) Where FL,eff is the sum of lift force in a shear flow, Saffman lift force, the pressure gradient force etc. and Fad is adhesive force (Van de Waals). In above equation the value of ks is 0.3 [56].

8

4. Results and Discussion 4.1. Experimental Results In this study effect of particle size on deposition process was studied experimentally and numerically. As it is already mentioned, particles were introduced in discrete lots rather than continuously in experiments. For example, as it is shown in Fig. 7, 35 gr of an A3 particle group once was injected in five groups of 7 gr (5×7) and another time was injected in seven groups of 5 gr (7×5). In order to distinguish these two distinct lots, they have been named N1 and N2 respectively. In Fig. 7 the effect of mass particle and velocity on pressure drop for N1 and N2 are demonstrated. The comparison showed that the rate of introduction had little or no effect on the fouling process, and the mass of foulant on the exchanger does not change from one case to the next. It also depicts that pressure drop increase with increase of flow velocity over a range from 3 m/s to 5 m/s. The measurements also showed that pressure drop was grown with addition of mass particle and this can be concluded that particles were deposited. The effect of particle size on pressure drop is shown in Fig. 8 and Fig. 9 for velocity of 4 m/s and 5 m/s respectively. They show that A1 particles had no effect on pressure drop and they all passed through the exchanger and had very little deposition on CHE surfaces. Pressure drop was increased with increase of particle size from A2 to A6 and showed that the majority of bigger particles were deposited on the surfaces and led to increase of pressure drop. Both Figs. shows that the main variation took place between A2 and A4. The results also depicted that the deposition of A1 particles was increased with increase of velocity from 4 m/s to 5 m/s. Deposition pattern of various particle size are demonstrated in Fig. 10. It can be seen that A1 particles were passed through CHE without any deposition while a portion of A2 particles were settled on front of CHE at the first edges. A3 and A4 particles were surprisingly deposited on CHE surfaces; they also were penetrated into the fin channels. A5 and A6 particles which are the biggest particles in this study were impacted to the front face of the CHE and blocked the fin channels; furthermore, a large portion of these group were fell down because of their weight and effect of gravity force. So, A3 particle group were the critical category, in which the most deposition were took place. 4.2. Numerical Results Experimental results shows that particles larger than group A3 cannot pass through fin channels and settle down the exchanger surfaces, they just impact on front of exchanger and increase pressure drop during experiments. Furthermore, after switching off test rig and when there is no drag force, these large particles fell down to the tunnel floor and did not deposited on the surfaces. Considerable portion of large particles in A5 and A6 group also fell down on the tunnel floor during the experiments because of gravity force effects. As a result deposition of particles over range of 1μm to 100 μm are studied numerically. In order to validate the reliability of the numerical method, comparison among experimental and numerical results are depicted in Fig. 11 and shows pressure drop for different velocity without

9

particle injection. As it can be seen, numerical and experimental results are in good agreement and this agreement demonstrated that numerical model is reliable. Fig. 12 demonstrate pressure drop due to mass particles increase in numerical and experimental study. It shows that in both study pressure drop follow same trend and it increases with increase of injected mass particles. Fig. 13 shows velocity and vorticity contour of flow for inlet velocity of 5 m/s, without any particle injection through fin channels. Velocity contour shows an increase of velocity through fin channels and a decrease through edges of each channel. This shows a tendency to particle deposit. It can be seen from vorticity contour that air was flowed through edges of channels and created turbulence and shedding behind edges, so forming vortices. These vortices restrained to the side of the channels. Fig. 14 depicts deposit ratio on various edges of fin channel for different particle diameter at velocity of 3 m/s. As it is seen, deposit ratio for smallest particles was negligible because they tended to follow air flow and this reduced the chance to touch the heating surface and consequently declined the rate of fouling. However, with increase in particle diameter, the increased inertia leaded to more impacts and promoted the possibility of deposition. Largest particles with increased mass tended to fall in front of channels and thus their deposition ratio was small. As it is seen in Fig. 15, the effect of flow velocity on particle deposition with a range between 1 m/s and 5 m/s was studies for dp= 50μm. It shows that the differences on deposit ratio for each edge were not considerable within the same velocity. It can be observed that particle deposition on the first edge was higher than the other edges because of the particle residence time and stagnation region near the first edge. Furthermore, particle deposition on the last edge was also remarkable in contrast with other edges, it can be explained that rebounded particles from other edges had more chance to deposit on this recirculating region. Fig. 16 depicts the effect of flow velocity on deposition of particles with different sizes. Increase of velocity and drag force leaded small particle to move away flow streamline and impacted to the heat exchanger surface. For particles with diameter over range of 10 μm to 45 μm, flow velocity raises improved deposit ratio; however, velocity increase reduced deposition of bigger particles of 50 μm, 60 μm and 100 μm because drag force dominated inertia force. It can be understood that small particles cannot settle on a heating surface unless inertia force enhance their movement toward surface. Big particle has potential to reach the surface and deposit on but velocity increase will enhance drag force and decrease deposition ratio. The biggest particle cannot reach surfaces and fall down because of gravity dominance and velocity increase will help them to deposit due to inertia increase. Also, it can be seen that deposition ratio were increased over a range from 1 μm to 50 μm and then decreased for particle size of 100 μm. So, particle with diameter of 50 μm was the critical particle in which the most deposition occurred. Deposition patterns of various particle sizes are presented in Fig. 17. It can be seen that the motion of small particles with low stokes number were tightly coupled to the fluid motion. In other word, particles dispersal were as the same as the fluid dispersal. So, small particles that were moved through

10

fin channels had little chance to deposit. Particles with diameter over range of 10 μm to 40 μm had more chance to deposit. Bigger particles with larger Stokes number were not influenced by the fluid and were passed through the flow without much deflection in its initial trajectory, so they had much more chance to deposit on the surfaces. Besides, particles with 100 μm could settle on the first edge and block the channels.

5. Conclusion Experimental and numerical study on particle deposition was performed on the compact heat exchanger. The effect of particle size was studied. In this paper influence of velocity ranging from 1 m/s to 5 m/s on deposition were discussed. In experimental study, introduced particles were classified into six poly disperse group range from 1 μm to 4 mm while in numerical study mono disperse particles were injected with diameters over range of 1 μm to 100 μm and were placed in A1 particle group of experiment. The following conclusions are obtained. 1- Most of the particles are deposit on the front of CHE and also on the first and last edges of fin channels. 2- Injecting more particle increases pressure drop of all particle group. Besides, increasing particle size lead to more pressure drop but deposition does not follow this trend. That is, deposition enhances up to a critical size and after that it decreases. Because bigger and heavier particles cannot pass through and they settle on the front part of CHE. In experiments, these particles fall from the front surface of CHE when the flow stops; besides a portion of bigger particles fall down before they reach CHE. 3- Increase of flow velocity lead to an increase in pressure drop; furthermore, deposition of small particle enhances with increase of flow velocity because they move away from streamlines and have more chance to impact CHE surfaces. But, this enhancement, increase the drag force of bigger particles and consequently decreases their deposition.

11

References [1] P. Venturini, D. Borello, K. Hanjali, F. Rispoli, “Modeling of Particles Deposition in an Environment Relevant to Solid Fuel Boilers”, Applied Thermal Engineering, 2012, 49: 131-138. [2] J. A. Siegel, “Particulate Fouling of HVAC Heat Exchangers”, PhD Thesis, University of California, Berkeley, 2003. [3] R. Haghighi-Khoshkhoo, F. M. J. McCluskey, “Air-Side Fouling of Compact Heat Exchangers for Discrete Particle Size Ranges”, Heat Transfer Engineering, 2007, 28(1): 58–64. [4] M. S. Waring, J. A. Siegel, “Particle loading rates for HVAC filters, heat exchangers, and ducts” Indoor Air Journal, 2008, 18(3): 209-224. [5] I. H. Bell, E. A. Groll, H. König, “Experimental analysis of the effects of particulate fouling on heat exchanger heat transfer and air-side pressure drop for a hybrid dry cooler”, Heat Transfer Engineering Journal, 2011, 32(3): 264 271. [6] Y. T. Shi, M. Gao, G. H. Tang, F. Z. Sun, W. Q. Tao, “Experimental research of CFB ash deposition on helical finned tubes”, Applied Thermal Engineering, 2012, 37: 420-429. [7] S. D. Oduro, “Assessing the Effect of Blockage of Dirt on Engine Radiator in the Engine Cooling System”, International Journal of Automotive Engineering, 2012, 2 (3): 163-171. [8] C. Paz, E. Suárez, M. Concheiro, J. Porteiro, “Experimental study of soot particle fouling on ribbed plates: Applicability of the critical local wall shear stress criterion”, Experimental Thermal and Fluid Science, 2013, 44(1): 364-373. [9] T. G. Walmsley, M. R. W. Walmsley, M. J. Atkins, J. R. Neale, “Fouling and pressure drop analysis of milk powder deposition on the front of parallel fins”, Advanced Powder Technology, 2013, 24(4): 780-785. [10] F. Zhan, J. Tang, G. Ding, D. Zhuang, “Experimental investigation on particle deposition characteristics of wavy fin-and-tube heat exchangers”, Applied Thermal Engineering, 2016, 99: 10391047. [11] Y. Gong, G. Yu, Q. Guo, Z. Zhou, F. Wang, Y. Liu, “Experimental study of the particle deposition characteristics in an entrained flow gasifier”, Chemical Engineering Science, 2015, 138: 291-302. [12] D. Patiño, B. Crespo, J. Porteiro, J. L. Míguez, “Experimental analysis of fouling rates in two small-scale domestic boilers”, Applied Thermal Engineering, 2016, 100: 849-860. [13] F. S. Mirhashemi and S. H. Hashemabadi, “Experimental and CFD study of wall effects on orderly stacked cylindrical particles heat transfer in a tube channel,” International Communications in Heat and Mass Transfer, 2012, 39(3): 449–455. [14] Z. Zhang, X. Zhang, “Direct simulation of low-re flow around a square cylinder by numerical manifold method for Navier-Stokes equations,” Journal of Applied Mathematics, 2012, Article ID465972, 14 pages.

12

[15] J. Nordstrom, J. Gong, E. van derWeide, and M. Svard, “A stable and conservative high order multi-block method for the compressible Navier-Stokes equations,” Journal of Computational Physics, 2009, 228 (24): 9020–9035. [16] Z. Chen, L. Zhang, “A stabilized mixed finite element method for single-phase compressible flow,” Journal of Applied Mathematics, 2011, Article ID 129724. [17] S. Boivin, F. Cayre, J. Herard, “A finite volume method to solve the Navier-Stokes equations for incompressible flows on unstructured meshes,” International Journal of Thermal Sciences, 2000, 39(8): 806–821. [18] B. R. Baliga and S. V. Patankar, “A New Finite-Element Formulation for Convection-Diffusion Problems”, Numerical Heat Transfer, 1980, 3: 393-409. [19] X. Liu, H. Liu, Y. Liu, “Simulation of magnetorheological fluids based on Lattice Boltzmann method with double meshes,” Journal of Applied Mathematics, 2012, Article ID 567208, 16 pages. [20] A. A.Mohamad, “Lattice Boltzmann Method: Fundamentals and Engineering Applications with Computer Codes”, Springer, 2011. [21] Y. Shi, T. S. Zhao, Z. L. Guo, “Lattice Boltzmann method for incompressible flows with large pressure gradients,” Physical Review E, 2006, 73 (2), Article ID026704. [22] A. Karimipour, A. H. Nezhad, A. D’Orazio, E. Shirani, “The effects of inclination angle and prandtl number on the mixed convection in the inclined lid driven cavity using lattice Boltzmann method,” Journal of Theoretical and Applied Mechanics, 2013, 51 (2): 447–462. [23] AJ Chamkha, AM. Rashad, “MHD forced convection flow of a nanofluid adjacent to a nonisothermal wedge”. Computational Thermal Science, 2014, 6: 27–39. [24] F. Selimefendigil, HF. Öztop, “MHD mixed convection of nanofluid filled partially heated triangular enclosure with a rotating adiabatic”, Journal of Taiwan Institute Chemical Engineering, 2014, 45: 2150–62. [25] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, “Effect of a magnetic field on natural convection in an inclined half-annulus enclosure filled with Cu–water nanofluid using CVFEM”, Advanced Powder Technology, 2013, 24: 980–91. [26] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, “MHD natural convection in a nanofluid filled inclined enclosure with sinusoidal wall using CVFEM”, Neural Computing and Applications, 2014, 24: 873–82. [27] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, S. M. Seyyedi “Natural convection of nanofluids in an enclosure between a circular and a sinusoidal cylinder in the presence of magnetic field”, International communications in heat and mass transfer, 2012, 39: 1435–43. [28] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, “Heat flux boundary condition for nanofluid filled enclosure in presence of magnetic field”, Journal of Molecular Liquids, 2014, 193: 174–84.

13

[29] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, “Magnetic field effect on nanofluid flow and heat transfer using KKL model”, Journal of Taiwan Institute Chemical Engineering, 2014, 45: 795–807. [30] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, S. Soleimani, “Thermal management for free convection of nanofluid using two phase model”, Journal of Molecular Liquids, 2014, 194: 179–87. [31] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, P. Rana, S. Soleimani, “Magneto hydrodynamic free convection of Al2O3–water nanofluid considering thermophoresis and Brownian motion effects”, Computers & Fluids, 2014, 94: 147–60. [32] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, “Investigation of nanofluid flow and heat transfer in presence of magnetic field using KKL model”, Arabian Journal for Science and Engineering, 2014, 39(6): 5007–16. [33] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, “Natural convection in a nanofluid filled concentric annulus between an outer square cylinder and an inner elliptic cylinder”, Iranian Journal of Science and Technology Transactions (B) of Mechanical Engineering, 2013, 20(4): 1–13. [34] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, “Magnetic field effects on natural convection around a horizontal circular cylinder inside a square enclosure filled with nanofluid”, International Communications in Heat and Mass Transfer, 2012, 39: 978. [35] M. Sheikholeslami, M. Gorji-Bandpy, S. M. Seyyedi, D. D. Ganji, H. B. Rokni, S. Soleimani, “Application of LBM in simulation of natural convection in a nanofluid filled square cavity with curve boundaries”, Powder Technology, 2013, 247: 87–94. [36] M. Sheikholeslami, M. Gorji-Bandpy, D. D. Ganji, “Free convection of nanofluid filled enclosure using lattice Boltzmann method (LBM)”, Applied Mathematics and Mechanics, 2013, 34(7): 1–15. [37] M. Sheikholeslami, D.D. Ganji, “Heat transfer improvement in a double pipe heat exchanger by means of perforated turbulators”, Energy Conversion and Management, 2016, 127: 112–123. [38] M. Sheikholeslami, D.D. Ganji, “Turbulent heat transfer enhancement in an air-to-water heat exchanger”, Journal of Process Mechanical Engineering, 2016, 0(0) 1–14. [39] L. Y. Huang, J. S. Norman, M. Pourkashanian, A. Williams, “Prediction of Ash Deposition on Super Heater Tubes from Pulverized Coal Combustion”, Fuel, 1996, 75 (3): 271-279. [40] H. Han, Y. L. He, W.Q. Tao, Y. S. Li, “A Parameter Study of Tube Bundle Heat Exchangers for Fouling Rate Reduction”, International Journal of Heat and Mass Transfer, 2014, 72: 210–221. [41] K. Wacławiak, S. Kalisz, “A Practical Numerical Approach for Prediction of Particulate Fouling in PC Boilers”, Fuel, 2012, 97: 38-48. [42] E.V.P.J. Manjula, W.K. Hiromi Ariyaratne, Chandana Ratnayake, Morten C. Melaaen, “A Review of CFD Modelling Studies on Pneumatic Conveying and Challenges in Modelling Offshore Drill Cuttings Transport”, Powder Technology, 2016, doi: 10.1016/j.powtec.2016.10.026.

14

[43] X. Chen, J. Wang, “A comparison of two-fluid model, dense discrete particle model and CFDDEM method for modeling impinging gas–solid flows”, Powder Technology, 2014, 254: 94-102. [44] H. Lu, X. Guo, W. Zhao, X. Gong, J. Lu, “Experimental and CPFD Numerical Study on Hopper Discharge”, Industrial & Engineering Chemistry Research, 2014, 53: 12160-12169. [45] P. Kosinski, A. C. Hoffmann, “Modeling of dust lifting using the Lagrangian approach”, International Journal of Multiphase Flow, 2005, 31: 1097-1115. [46] R. H. Khoshkhoo, “Experimental Investigation into the Air-Side Fouling of Compact Heat Exchangers”, PhD Thesis, Joseph Fourier de Grenoble University, 2004. [47] D. Montgomery, “Experimental Design for Product and Process Design and Development”, Journal of the Royal Statistical Society, 1999, 48: 159-177. [48] ANSYS Inc. ANSYS Academic Research, Release 16; 2014. [49] Y. Shin Khor, Q. Xiao, “CFD simulations of the effects of fouling and antifouling”, Ocean Engineering, 2011, 38: 1065–1079. [50] Lin Tian, Goodarz Ahmadi, “Particle deposition in turbulent duct flows-comparisons of different model predictions”, Aerosol Science, 2007, 38: 377 – 397. [51] Partankar SV: Numerical heat transfer and fluid flow. Hemisphere, Washington, DC, 1980. [52] P.W. Longest, C. Kleinstreuer, “Comparison of blood particle deposition models for non-parallel flow domains”, Journal of Bio Mechanic, 2003, 36: 421–430. [53] A. Li, G. Ahmadi, Dispersion and deposition of spherical particles from point sources in a turbulent channel flow”, Journal of Aerosol Science Technology, 1992, 16: 209–226. [54] A. Li and G. Ahmadi. "Dispersion and Deposition of Spherical Particles from Point Sources in a Turbulent Channel Flow". Aerosol Science and Technology. 1992, 16: 209–226. [55] J. Tomeczek, W. Krzysztof, “Two-dimensional modeling of deposits formation on platen superheaters in pulverized coal boilers”, Fuel, 2009, 88: 1466–1471. [56] L. Shi and D. Bayless. “Comparison of boundary conditions for predicting the collection efficiency of cyclones”, Powder Technology, 2007, 173: 29–37.

15

Table Captions Table 1- Characteristics of a compact heat exchanger Table2- List of measurement instruments and their specifications Table 3- Particle size distribution Table 4- Y+ values for different values of velocity

Figure Captions Fig. 1- (a) Experimental setup; (b) test section and measurement instruments schematic Fig. 2 – A schematic of computational domain Fig. 3- Comparison of different turbulence model on pressure Fig. 4- Comparison of different turbulence model on velocity Fig. 5- the effect of Y+ on velocity Fig. 6- Schematic of particle impact Fig. 7- The effect of distinct particle injection on pressure drop measurement Fig. 8- The effect of particle size on pressure drop for velocity 4 m/s Fig. 9- The effect of particle size on pressure drop for velocity 5 m/s Fig. 10- Deposition pattern for various particle sizes Fig. 11- comparison among experimental and numerical results Fig. 12- Comparison of pressure drop due to mass particle increase between numerical and experimental results Fig. 13- Contour of velocity and vorticity Fig. 14- Deposit ratio on ten edges for different particle size Fig. 15- Effect of flow velocity on deposit ratio on different edges for dp= 50μm Fig. 16- Effect of flow velocity on total deposit ratio Fig. 17- Deposition patterns of various particle sizes

16

(a)

(b) Fig. 1- (a) Experimental setup; (b) test section and measurement instruments schematic

17

Fig. 2 – A schematic of computational domain

Fig. 3- Comparison of different turbulence model on pressure

18

Fig. 4- Comparison of different turbulence model on velocity

Fig. 5- the effect of Y+ on velocity

19

Fig. 6- Schematic of particle impact

Fig. 7- The effect of distinct particle injection on pressure drop measurement

20

Fig. 8- The effect of particle size on pressure drop for velocity 4 m/s

Fig. 9- The effect of particle size on pressure drop for velocity 5 m/s

21

A1

A2

A3

A4

A5

A6

Fig. 10- Deposition pattern for various particle sizes

Fig. 11- comparison among experimental and numerical results

22

Fig. 12- Comparison of pressure drop due to mass particle increase between numerical and experimental results

23

a. Velocity contour (m/s)

b. Vorticity contour (s-1) Fig. 13- Contour of velocity and vorticity

Fig. 14- Deposit ratio on ten edges for different particle size

24

Fig. 15- Effect of flow velocity on deposit ratio on different edges for dp= 50μm

Fig. 16- Effect of flow velocity on total deposit ratio

25

1 μm

10 μm

30 μm

40 μm

45 μm

50 μm

60 μm

100 μm Fig. 17- Deposition patterns of various particle sizes

26

Table 1- Characteristics of a compact heat exchanger Fin density per 100 (mm)

44

Fin type

Fin louver

Geometric fin shape

V shape

Fin base (mm)

3

Fin summit

1,6

Fin height (mm)

18,5

Fin thickness

0,15

Tube direction

horizontal

Orientation of the largest dimension of a fin

vertical

Actual dimensions of the heat exchanger (cm)

64×33

Heat exchanger depth (cm)

3

Table2- List of measurement instruments and their specifications Instrument

Range

Resolution

Accuracy (%)

Testo 510 Differential pressure measuring instrument

0-104 Pa

1 Pa

±3

0-10 Pa

0.25 Pa

±1

Up to 1200 gr

0.001 gr

±0.01

3

Pitot tube Lab digital Balance FX-1200IN

Table 3- Particle size distribution A1=[1μm < dp< 0,8 mm] A2=[0,8 mm < dp< 1,25 mm] A3=[1,25 mm < dp< 1,85 mm] A4=[1,85 mm < dp< 2 mm] A5=[2 mm < dp< 3,15 mm] A6=[3,15 mm < dp< 4 mm]

Table 4- Y+ values for different values of velocity

Velocity (m/s) 1 2 3 4 5

Y+ 32.63 32.85 32.97 33.13 33.2

27

Highlights 

The effect of particle size and injected mass on particle deposition is investigated.



The influence of velocity on particle deposition is studied.



Particle deposition occurs mostly on the first and last edges of fin channels.



Increase of particle size can enhance deposition ratio in fin channels.



Increase of air velocity can promote particle deposition or restrain it.

28