International Journal of Fluid Machinery and Systems DOI: http://dx.doi.org/10.5293/IJFMS.2018.11.4.

412
Vol. 11, No. 4, October-December 2018 ISSN (Online): 1882-9554

Original Paper

Progressive Orthogonal Optimization Design Method for High

Efficiency Axial-flow Blower

Chuang Li, Jianrun Zhang and Wenbin Yu

School of Mechanical Engineering, Southeast University

2 Southeast University Road, Nanjing, 211189, P.R. China, [email protected]

Abstract

To solve the problem of excessive design parameters of flow channel during the comprehensive optimization of the
axial-flow blower's total pressure efficiency and flux, a progressive optimization design method which is based on the
importance of structural parameters is proposed. The accurate numerical simulation model of flow channel is built, verified
and updated by experiment results. According to the model, progressive orthogonal optimization design is developed as
following steps: the first is axial fan design, then the orthogonal design of the structure parameters of entire flow channel
is further advanced. This approach is applied to the optimization design of an axial-flow blower in this paper, and the
results show that the uniformity of the flow field in the flow channel is significantly increased and the vertexes spreading
in the flow channel are obviously reduced, the flux is increased from 0.145m3/s to 0.171m3/s, and the efficiency increased
from 45.47% to 58.13%, it indicates that the method can meet the design requirements well while being efficient.

Keywords: Axial-flow blower; High efficiency; Overall optimization; Progressive orthogonal

1. Introduction
Compared with centrifugal fan, axial-flow fan has advantages of compact, low cost [1], large flux and easy installation [2], so it is
widely used in electric tools, air conditioners, air compressors and other fields. As people are increasingly concerned about energy saving
and environmental protection, the improvement of efficiency has received increasing attention from designers. Traditional design
methods verify the feasibility of the design through experiments, which takes time and effort. With the development of computational
fluid dynamics (CFD) and computational aeroacoustics (CAA), flow field and sound field can be accurately analyzed and predicted
accurately by numerical simulation [3].
The flow field of the blower is related to many factors due to its complex structure, researchers have conducted extensive research on
various aspects. Taku et al. [4] studied the influence of blade number on flow field and aerodynamic noise. The work of A. Asgarshamsi
et al. [5] indicates that a reasonable bend and sweep design can have a certain effect on the flux and efficiency of axial-flow turbines. Jin
et al. [6] found that the application of splitter blades in a centrifugal fan leads to significant improvement in the overall fan performance.
T. Shigemitsu et al. [7] proposed the design of axial flow fan in contra-rotating rotors, the superiority of the form was verified by
simulation. Ye et al. [8] found that design blades top as trough type can obviously increase the efficiency, but the flow and noise levels
are deteriorated simultaneously. Yang et al. [9] investigated the effect of guide vane on flow flied inside the separators. Zhang [10] found
that a reasonable reduction of the axial clearance between the fan and the back guide vane can promotes the performance of the R+S
axial-flow fan.
Although many scholars have analyzed the design of the axial flow blower, the findings are mostly directed at the individual structural
components, there are relatively few studies on the effect of matching between components on fan performance [11], and no scholars
have conducted the whole machine. Fan performance is related to component matching, and the individual optimization of the parts is
difficult to directly apply. The optimization design of whole machine can well consider the mutual influence between components, find
the optimal design scheme within a given design domain, and the results can be directly applied to the engineering design.
To realize the structure optimization of the whole machine, applying intelligent algorithms (such as genetic algorithm (GA) [12] and
Particle Swarm Optimization (PSO) [13]) seems to be an effective way, however, CFD calculation is needed during the execution of the
algorithm, which makes the time costs unacceptable. The establishment of surrogate model (such as artificial neural networks (ANN))
can greatly reduce the time costs, but it is still too high for engineering applications. Another problem with these two practices is that it
is difficult to handle the discrete variant of the airfoil. Due to these circumstances, a large number of studies are currently conducted
through orthogonal experiments. However, in the optimization process of the whole machine, traditional orthogonal experiments also
faces the problem of excessive time costs because of excessive design variables. Therefore, progressive orthogonal optimization design
Received May 29 2018; revised June 20 2018; accepted for publication August 11 2018: Review conducted by Minsuk Choi.
(Paper number O18022K)
Corresponding author: Jianrun Zhang, Professor, [email protected]

412
(POOD) method was proposed. It not only solves the problem of being difficult to apply to engineering due to long time-consuming, but
also makes it very convenient to handle the variable of the airfoil.
Key design parameters of flow channel were selected first and POOD was proposed because of the excessive design parameters. The
method firstly analyzes the effect of each parameters on blower performance, the preliminary orthogonal optimization analysis is
developed based on the selection, and then the POOD is developed for other factors based on the preliminary optimization results, a
complete process to optimize the overall structure of axial-flow blower by POOD has been achieved in this way.

2. POOD method for axial-flow blower

The flow channel contains axial fan, guide vane, diversion cone and other structures, and design parameters of axial fan and guide vane
are numerous. Since the matching relationship has a significant effect on performance of the blower, the matching relationship of
components could not be ignored. If the blower is optimized by traditional orthogonal experimental method, 64 repeated trials are required
because of the 10 structural parameters selected, which is not conducive to rapid design and cost savings.
This paper adopts a method of progressive orthogonality to divide the orthogonal optimization design of the axial-flow blower structure
into two steps, as shown in Fig. 1.

Fig. 1 Progressive orthogonal design process

One of the important tasks before progressive orthogonal optimization is to establish a numerical simulation model of the axial-flow
blower to ensure the reliability of the entire calculation process. Another important job is the selection of orthogonal factors. Based on
the relevant theories that have been established, previous research results and simulation results, key structural parameters have been
selected as design variables. Then the preliminary orthogonal optimization design is implemented for the core components of the flow
channel: axial fan, the orthogonal design of the entire flow channel is further advanced on this basis, and the optimal design of the blower
is completed through a two-step progressive orthogonal design.
2.1 Structure analysis and numerical simulation model establishment
2.1.1 Structure analysis
Conceptual model of the blower is based on a company’s product, its flux and efficiency cannot meet the company's requirements due
to unreasonable design, necessary simplification and geometric clean-up were carried out, and the flow channel of conceptual model is
shown in Fig. 2, the motor drives the axial fan to rotate during work and the rest of the components are stationary. During the rotation of
the axial fan, energy is transmitted to the gas through the interaction between the blade and the air flow, which drives the gas movement,
as the driving component of the field, it has a great influence on the flux and efficiency.

Collector Guide vane Safety barrier Outlet tube

Axial fan Motor Diversion cone

Fig. 2 Model of the conceptual model’s flow channel
Velocity of gas flowing from rear edge of fan can be regarded as synthesis of axial velocity and convolute velocity. If a reasonable
rectification design is not performed, the rotational kinetic energy corresponding to convolution velocity is generally consumed in vortices
and friction losses. Guide vanes and diversion cones can convert the energy into pressure energy and increase the efficiency and flux of
the blower. Therefore, guide vanes and diversion cones should also be emphasized.

413
2.1.2 Establishment of simulation model and results analysis
Main parameters of conceptual model are shown in Table 1.
Table 1 Main parameters of the conceptual model
Parameters Value Parameters Value
D: Channel diameter(mm) 88 NF: Blades number 12
D1: Inlet hydraulic diameter(mm) 133 Axial θF: Installation angle(°) 65
D2: Outlet diameter(mm) 70.5 fan βF: Deflection angle(°) 35
n: Motor speed(rpm) 17000 r: Hub ratio 0.51

According to actual working condition, simulation settings are shown in Table 2. All the calculations were performed in ANSYS Fluent,
the influence of gravity and actual wall conditions were ignored during calculation. The flow of air during the operation of the blower
can be considered as an incompressible flow, so a pressure-based solver was employed. Since the blower inlet and outlet are directly
connected to the atmosphere, it is reasonable to set as pressure inlet and pressure outlet. The rotation speed of the fan area was set as
17000rpm according to standard working conditions, and its rotation was achieved by sliding mesh. Standard k-ε model was selected as
turbulence model because it already guarantees enough calculation accuracy. To speed up the convergence of the numerical iterative
process, the SIMPLEC algorithm was employed. To improve the accuracy of the simulation, pressure, momentum, density and turbulent
kinetic energy were discarded in second-order upwind. Simulation results are shown in Table 3.
Table 2 Simulation settings
Parameters Settings
Solver Pressure-Based; absolute; Steady
Inlet Pressure inlet, D1=133mm, P1=1atm
Outlet Pressure outlet, D2=70.5mm, P2=1atm
Rotation area Multi-reference model(MRF), n=17000rpm
Viscous Model Standard k-ε
Solution Methods SIMPLEC
Table 3 Simulation results
Q: T: PtF: Total PS: Shaft PP: Total η: Total pressure
Flux(m3/s) Torque(N∙m) pressure(Pa) power (W) Pressure power(W) efficiency (%)
0.145 0.154 857.11 274.16 124.67 45.47
Note: PtF represents the total pressure of the blower, i.e. the difference of total pressure between inlet and outlet.
The velocity field distribution is shown in Fig. 3, it can be found that the position with highest velocity is concentrated in the vicinity
of the axial fan, the area where the vortex forms obvious is also here, it can be determined that the optimal design of the axial fan plays
an important role in improving the performance of the blower.

(a) Velocity streamline diagram

(b) Partial velocity vector diagram

Fig. 3 Velocity streamline diagram

414
2.1.3 Verification of simulation model
To verify the accuracy of the simulation model, the flux and total pressure of the blower was measured using a pitot tube, the experiment
was designed according to the IOS 3966[14].
(a) Flux
Experimental device and measurement point distribution are shown in Fig. 4.

1-Axial-flow blower 2-Transitional tube 3-Wind tunnel (Numbers at both ends of the
4-Bracket 5-EY-200A Digital Micro Pressure Gauge diameter indicate the number
6- pitot tube 7-Experiment bench of points on this line, in order)
(a) Experimental setup (b) Velocity measurement points
Fig. 4 Measurement of flux
Gas velocity at each point are shown in Table 4, the relationship between the measured velocity and Dis is shown in Fig. 5. It can be
found that the velocity is basically symmetrical about the center of the tube, and the velocity of the measuring point near the center of
the tube is relatively high, which is consistent with the theory of fluid mechanics. The biggest difference in the velocity at the same
distance in the measurement are point 19 and point 24, the relative error is
𝑣19 − 𝑣24
𝑒𝑣 = × 100% = 5.36% (1)
𝑣19
This error was mainly caused by the inability to ensure that the pipeline was completely leveled during installation.
The flux
∑24𝑖=1 𝑣𝑖 𝐷2
𝑄𝑒𝑥𝑝 = ∙𝜋∙ = 0.137𝑚3 /𝑠 (2)
24 4
Simulation error
|𝑄 − 𝑄𝑒𝑥𝑝 |
𝑒= × 100% = 5.84% (3)
𝑄𝑒𝑥𝑝
Table 4 Velocity measurement results
Points i vi (m/s) Points i vi (m/s) Points i vi (m/s) Points i vi (m/s) Dis/mm
Diameter 1 Diameter 2 Diameter 3 Diameter 4
1 38.61 7 38.11 13 38.31 19 37.47 30.42
2 41.32 8 41.56 14 42.03 20 41.13 23.72
3 44.33 9 43.13 15 43.27 21 44.07 8.22
4 43.82 10 43.31 16 42.94 22 43.86 8.22
5 41.64 11 41.92 17 41.59 23 42.03 23.72
6 38.79 12 39.43 18 38.17 24 39.48 30.42
Note: “Dis” in the rightmost column of the table indicates the distance between the measurement points and the tube center.

Fig. 5 Measured velocity on each diameter

(b) Total pressure PtF
The blower inlet is close to open, the inlet total pressure can be considered as 0 (relative to atmospheric pressure), PtF equals the outlet
total pressure Pout, and Pout is the sum of dynamic pressure and the static pressure, i.e.
𝑃𝑡𝐹 = 𝑃𝑜𝑢𝑡 = 𝑃𝑑𝐹 + 𝑃𝑠𝐹 (4)
PdF is the dynamic pressure of outlet and PsF is the static pressure of outlet. The pressure measurement section was selected near the
outlet according to standard, as shown in Fig. 6. The distance between the measuring section and the outlet is 2mm, the transition tube
and the wind tunnel are no longer required now.

415
 (a) Experimental setup (b) Static pressure measurement points
Fig. 6 Measurement of outlet static pressure
The dynamic pressure of the section can be calculated based on the flux measurement results, as Qexp=0.137m3/s, outlet diameter
D2=70.5mm and air density ρ is approximately 1.205 at 20℃, it can be calculated that
𝜌 𝜌 𝑄𝑒𝑥𝑝
𝑃𝑑𝐹 = 𝑣 2 = ( )2 = 742.10𝑃𝑎 (5)
2 2 𝜋(𝐷2 /2)2
The distribution of the static pressure measurement points is shown in Fig. 6(b). The static pressure hole of pitot tube is connected to
the positive electrode of the digital micromanometer (since the outlet static pressure is positive), the negative electrode is connected to
the atmosphere, and the reading is the static pressure at the measuring point (relative to the atmospheric pressure). The measurement
results are shown in Table 5.
Table 5 Static pressure measurement results
Points i 1 2 3 4
PsFi/Pa 142 165 191 158
The outlet static pressure
∑4𝑖=1 𝑃𝑠𝐹𝑖
𝑃𝑠𝐹 = = 164𝑃𝑎 (6)
4
Thus
𝑃𝑡𝐹_𝑒𝑥𝑝 = 𝑃𝑜𝑢𝑡 = 𝑃𝑑𝐹 + 𝑃𝑠𝐹 = 906.10𝑃𝑎 (7)
Simulation error
|𝑃𝑡𝐹 − 𝑃𝑡𝐹_𝑒𝑥𝑝 |
𝑒𝑝 = × 100% = 5.41% (8)
𝑃𝑡𝐹_𝑒𝑥𝑝
The error of flux and pressure are acceptable and indicated that the established simulation model is credible.
2.2 Selection of factors
Axial fan, guide vane and diversion cone are components that have a major influence on the flow in the flow channel, this paper
optimizes the design of these components and their matching relationships. Fan and guide vane are airfoil blades, there are too many
design parameters and the amount of direct orthogonal experiments is too large. The blower performance is very sensitive to the
parameters of the fan, and many studies have shown the feasibility and reliability of the fan design in the wind tunnel, therefore, this
paper proposed to first optimize the axial fan design in the wind tunnel, then progressively optimize the whole machine based on the fan
optimization results. Fan parameters have been separated from other parameters of the flow channel, the optimization process only
requires 16 wind tunnel calculations and 32 whole machine calculations, and the structure of wind tunnel is simple, which can greatly
improve the design efficiency and reduce the design cost.
2.2.1 Axial fan
Design parameters of axial fan mainly include installation angle, deflection angle, sweep angle, blade thickness, hub ratio and number
of blades. In the blower of this paper, the hub ratio r cannot be optimized due to the limited conditions; the results of the literature [10],
[15] indicate that the blade's bending design has a significant influence on the aerodynamic noise, but the improvement of the flux and
efficiency is not obvious, thus the sweep angle is not considered. The study in [16] shows that proper reduction of the blade thickness
can improve the efficiency of the fan, but at the same time, the aerodynamic noise level will increase, and reducing the blade thickness
will reduce the blade strength, so the optimize of thickness is not considered. In summary, airfoil AF, installation angle θF, deflection
angle βF, and number of blades NF are selected as design variables of axial fan.
2.2.2 Others
Velocity direction of the air flow from the fan will form an angle with the direction of the pipe axis, if the follow-up guide vane or other
rectification structure is improperly designed, the air flow may hit the surface of the pipe and even back flow, which greatly reduces the
flux and efficiency. Therefore, the match between the guide vane and the fan has a significant effect on flux and efficiency.

（a）BE3259B （b）NACA M19

（c）E63 （d）KUPFER
Fig. 7 Four guide vane airfoils

416
 Airfoil blades have better aerodynamic performance than equal thickness blades, thus the airfoil AG are selected to optimize the guide
vane. Different airfoil performances differ greatly, so the airfoil should be a design variable and four airfoils selected (Fig. 7). The design
of the guide vane is mainly to matching the angle with the fan and achieve the purpose of rectification, therefore, the design of installation
angle θG and deflection angle βG of the guide vane is very important. In addition, the influence of the number of blades NG cannot be
ignored.
The function of the diversion cone is to guide the uniform flow of the axial gas flow from the guide vanes into the entire pipeline. If
the length is too short, the cross section of the axial gas flow will change too violently, problems such as turbulence and vortices may
occur, while too long will increase the friction and impact probability of flow and wall, resulting in energy loss, so length L is selected as
the design variable of the diversion cone.
The axial clearance d between the guide vane and the axial fan (as shown in Fig. 8) is also a very important design variable, since the
velocity of the gas flowing from the axial fan is non-uniform, too large clearance will increase the loss along the stroke while too small
will aggravate the flow field fluctuations. From the simulation results of the conceptual model, the flow conditions in this part is very
complicated, the resulting loss of flux and efficiency cannot be ignored, so the axial clearance d is chosen as a design variable.

Fig. 8 Axial clearance

In summary, the selected design factors are shown in Table 6.
Table 6 Progressive orthogonal design factors
First Second
Airfoil AF Airfoil AG
Installation angle θF Installation angle θG
Axial fan Guide vane
Deflection angle βF Deflection angle βG
Blades number NF Blades number NG
Diversion cone Length L
Structure parameter Axial clearance d

3. Orthogonal optimization design of axial fan

3.1 Wind tunnel simulation of conceptual model fan
The calculation model of the wind tunnel is shown in Fig. 9. The tunnel diameter D=88 mm, which is the same as the internal diameter
of the flow channel, the length of the inlet section L1=200 mm, and the length of outlet section L2=500 mm.

Fig. 9 Wind tunnel calculation model Fig. 10 The original fan model

Model of original is shown in Fig.10, the simulation results are shown in Table 7.
Table 7 Wind tunnel calculation results of original fan
Q (m3/s) T (N·m) PtF (Pa) PS (W) PP (W) η (%)
0.145 0.122 521.39 216.41 75.64 34.95

It can be found that the efficiency calculated in the wind tunnel is lower than the actual structure, the reason is that there is no rectifier
structure in wind tunnel, the rotational kinetic energy of the fan outflow cannot be effectively used. Since the efficiency analysis in the
wind tunnel cannot reflect the actual situation, only the flux will be analyzed in the wind tunnel calculation.
3.2 Airfoil primary selection
Airfoils can be classified into four types according to geometric shapes, as shown in Fig. 11.

(a) Symmetrical (b) Plano-convex

(c)Concave-convex (d)Biconvex
Fig. 11 Four types of airfoil
Before orthogonal experiments, several suitable airfoils need to be selected from a large airfoil library as the level of airfoil in orthogonal
experiments, only airfoil was changed and the simulation results are shown in Table 8.

417
 Table 8 Simulation results of airfoil primary selection
Airfoil section Airfoil Type Q(m3/s)
Original Concave-convex 0.145
CLARK Y Plano-convex 0.139
NACA 0012 Symmetrical 0.089
AH 79-100C Concave-convex 0.153
NACA 23012 Biconvex 0.125
AH 79-100B Concave-convex 0.152
GO-398B Plano-convex 0.143
GO-7955 Plano-convex 0.142
N-10 Plano-convex 0.144
NACA65-209 Symmetrical 0.125
NACA6409 Concave-convex 0.149
RAF33 Biconvex 0.133
It can be found that the biconvex and symmetrical airfoil flow has a significant gap compared with the plano-convex and concave-
convex airfoils, and the concave-convex airfoil exhibits the best performance, AH79-100B, AH79-100C, N10, and NACA 6409 airfoils
with the highest flux were selected as the fan airfoils in orthogonal experiments.
3.3 Orthogonal optimization design and result analysis
The orthogonal experimental design of the axial fan is based on the fan's airfoil, number of blades, installation angle and deflection
angle, the levels of various factors are shown in Table 9.
Table 9 Factor level table of orthogonal experiment
AF θF(°) βF(°) NF
1 AH79-100B 55 25 11
2 AH79-100C 60 30 12
3 N10 65 35 13
4 NACA6409 70 40 14
Orthogonal table and calculation results are shown in Table 10.
Table 10 Orthogonal experiment table of axial fan
No. AF θF βF NF Q(m3/s)
1 1 1 1 1 0.149
2 1 2 2 2 0.151
3 1 3 3 3 0.154
4 1 4 4 4 0.155
5 2 1 2 3 0.139
6 2 2 1 4 0.158
7 2 3 4 1 0.147
8 2 4 3 2 0.157
9 3 1 3 4 0.122
10 3 2 4 3 0.129
11 3 3 1 2 0.154
12 3 4 2 1 0.117
13 4 1 4 2 0.119
14 4 2 3 1 0.139
15 4 3 2 4 0.157
16 4 4 1 3 0.148
K1 0.608 0.528 0.609 0.552
K2 0.601 0.577 0.564 0.581
K3 0.522 0.611 0.573 0.569
K4 0.563 0.577 0.548 0.592
k1 0.152 0.132 0.152 0.138
k2 0.150 0.144 0.141 0.145
k3 0.131 0.153 0.143 0.142
k4 0.141 0.144 0.137 0.148
R 0.021 0.021 0.015 0.010
Note: K represents the sum of experimental data of a factor at a certain level. For example, for airfoil AF, K1=0.608,
it means that the sum of all experimental results with the airfoil level 1(experiment No.1, 2,3,4) is 0.608;
k1 is the mean value of K1, and R is the range difference of the columns k1, k2, k3, and k4.

418
 The influence of various factors on the flux Q is shown in Fig.12. In the relationship between the airfoil AF and the flux Q, it can be
found AH79-100B performed well, and N10 (plano-convex airfoil) is much smaller than that of the three kinds of Concave-convex
airfoils, combined with the results of airfoil primary selection, it is indicated that the concave-convex airfoil is an airfoil suitable for the
blower of the present invention. The flux increases with the installation angle θF first and then decreases, θF=65° showed the best
performance, but the flux does not show regularity with the change of deflection angle βF. The flux is relatively large when the number
of blades NF=12&14 and 14 is the optimal level. NF=14 is already a relatively large number of blades since the fan diameter is 86mm, if
NF continue to increase, the flux will not increase significantly or even decrease due to the increase in resistance., so NF=14 was selected.
It can be seen that the airfoil AF and installation angle θF are the main factors affecting the flux from the range chart.

(a) Diagram of each factor and flux (b) Range chart

Fig. 12 Effect of various factors on Flux Q
According to the results, the fan airfoil selected AH79-100B, NF=14, θF=65°and βF=25°, the optimal fan’s simulation results are shown
in Table 11, the flux and power have been significantly improved, the flux has increased by 11.72%, but the efficiency has decreased
slightly, this is increase number of blades, resulting in increased resistance and increased energy consumption.
Table 11 Simulation results of optimal axial fan
fan Q(m3/s) PS(W) PP(W) η (%) Q increase (%)
Origin 0.145 216.41 75.64 34.95 0
Optimized 0.162 343.59 116.51 33.91 11.72

4. POOD of basic parameters

An axial fan with excellent performance was obtained through the previous optimization, further orthogonal optimization design of
other design parameters in the flow channel was developed according to the optimized fan to obtain a high-efficiency, high-flux axial-
flow blower through the two-step POOD.
4.1 Orthogonal experimental design and result analysis
Further orthogonal optimization design is based on the optimized fan, considering all selected factors except axial fan’s, as shown in
Table 12.
Table 12 orthogonal design factors and levels of overall structure
AG NG θG(°) βG(°) d(mm) L(mm)
1 BE3259B 5 76 3 5.6 74.5
2 NACA M19 9 79 5 8.6 84.5
3 E63 11 82 7 11.6 94.5
4 KUPFER 13 85 9 14.6 104.5

Orthogonal experimental design and simulation results are shown in Table 13.
Table 13 Orthogonal experimental results of overall structure
No. AG NG θG(°) βG(°) d(mm) L(mm) Q(m3/s) η (%)
1 1 1 1 1 1 1 0.169 57.06
2 1 2 2 2 2 2 0.170 57.56
3 1 3 3 3 3 3 0.167 55.48
4 1 4 4 4 4 4 0.163 52.74
5 2 1 1 2 2 3 0.158 49.08
6 2 2 2 1 1 4 0.164 53.04
7 2 3 3 4 4 1 0.165 53.28
8 2 4 4 3 3 2 0.162 50.82
9 3 1 2 3 4 1 0.154 45.72
10 3 2 1 4 3 2 0.165 52.79
11 3 3 4 1 2 3 0.152 44.21
12 3 4 3 2 1 4 0.160 49.72
No. AG NG θG(°) βG(°) d(mm) L(mm) Q(m3/s) η (%)

419
 13 4 1 2 4 3 3 0.162 50.79
14 4 2 1 3 4 4 0.170 57.29
15 4 3 4 2 1 1 0.160 50.66
16 4 4 3 1 2 2 0.166 55.00
17 1 1 4 1 4 2 0.156 47.19
18 1 2 3 2 3 1 0.167 55.34
19 1 3 2 3 2 4 0.170 57.69
20 1 4 1 4 1 3 0.171 58.76
21 2 1 4 2 3 4 0.147 40.55
22 2 2 3 1 4 3 0.163 51.83
23 2 3 2 4 1 2 0.166 53.98
24 2 4 1 3 2 1 0.167 55.49
25 3 1 3 3 1 2 0.149 42.03
26 3 2 4 4 2 1 0.141 38.47
27 3 3 1 1 3 4 0.168 56.32
28 3 4 2 2 4 3 0.165 53.36
29 4 1 3 4 2 4 0.154 45.15
30 4 2 4 3 1 3 0.157 48.14
31 4 3 1 2 4 2 0.170 57.17
32 4 4 2 1 3 1 0.168 56.55

Refer to Table 9 for data processing, the influence of various factors on flux and efficiency is shown in Fig.13.
The correlation between flux Q and efficiency η can be clearly seen in various diagrams, so the comprehensive analysis of η and Q can
be simplified as efficiency analysis. The selected airfoils are all concave-convex airfoils, and efficiency has been significantly improved,
indicated that concave-convex airfoil is suitable for guide vane. As the number of blades NG increases, the efficiency gradually increases,
but the speed is getting smaller and smaller, although continue increasing may further increase the efficiency, the effect is not obvious,
and too many vanes will lead to the increasing of resistance, and reducing the efficiency, NG=13 can also meet the relatively prime
conditions with the number of axial fan blades, effectively preventing the occurrence of resonance. As the installation angle θG increases,
the efficiency gradually decreases, and the decreasing speed becomes faster and faster, while deflection angle maximizes efficiency at
βG=5°. In Experiment No. 31, θG=76° and βG=5°, the velocity vector diagram is shown in Fig. 14, it can be seen that the distribution of
the flow field in the fan and guide vane area is much more uniform and less vortex formation compared with The conceptual model,
which can reduce energy loss and increase efficiency. When the guide cone length L=94.5mm, the maximum efficiency is obtained,
which means that L=94.5mm can better ensure the uniformity of the flow field.
The installation angle is the most significant factor affecting the efficiency range, followed by the guide vane airfoil and the number of
blades. Axial fan’s outflow direction is certain because it has not been changed, and the sensitivity of the efficiency to the installation
angle reflects the importance of matching the guide vane installation angle with the fan outflow angle, and also shows the rationality of
factor selection.

(a) Diagram of each factors’ effect on flux and efficiency

(The flux (m3/s) is on the left side of the vertical axis and efficiency (%) on the right)

420
 (b) Range chart of flux and efficiency
Fig. 13 Influence of various factors on Q and η

Fig.14 Axisymmetric section velocity vector in experiment 31

Based on the foregoing results, the optimal combination is shown in Table 14.
Table 14 Optimal Combination of Progressive Orthogonal Optimization Design
Airfoil Blades number Installation angle Deflection angle
Axial fan
AH79-100B 14 65° 25°
Guide vane BE3259B 13 76° 5°
Length of diversion cone 94.5 mm
Axial clearance 8.6 mm

Simulation results of the optimal combination are shown in Table 15.

Table 15 Simulation results of the optimal combination
Q(m3/s) PS(W) PtF(Pa) PP(W) η(%) Q Increase (%) η increase(%)
Conceptual model 0.145 274.16 857.11 124.67 45.47
Optimized 0.171 347.39 1182.16 201.95 58.13 17.93 12.66

It can be seen that the flow of blower has increased by 17.93% and the efficiency has also increased by 12.66% after optimization, it
can be found from the data of two orthogonal experiments that once the axial fan is determined, the performance of blower has been
generally improved, guide vane and diversion cone have an impact on performance, but they are much smaller compared to axial fans,
this also shows that it is reasonable to select the axial fans as the first step of progressive orthogonal optimization.
4.2 Experimental verification
To ensure the accuracy of the optimization results, the optimized model was measured experimentally, see Section 2.1.3 for the
experimental design, Table 16 and Fig. 15 shows the flux measurement results and Table 17 shows the pressure measurement results.
Table 16 Optimized model flux test results
Points i vi （m/s） Points i vi （m/s） Points i vi （m/s） Points i vi （m/s） Dis/mm
Diameter 1 Diameter 2 Diameter 3 Diameter 4
1 44.13 7 43.87 13 44.39 19 43.52 30.42
2 47.94 8 47.72 14 47.27 20 47.14 23.72
3 50.13 9 50.17 15 49.06 21 50.76 8.22
4 49.86 10 48.67 16 48.93 22 49.81 8.22
5 47.25 11 46.53 17 46.91 23 47.03 23.72
6 45.13 12 43.84 18 44.07 24 44.15 30.42
ev(%) 4.29
Q（m3/s） 0.156
Simulation error e（%） 8.97

421
 Fig. 15 Measured velocity on each diameter
Table 17 Optimized model pressure test results
Points i 1 2 3 4
PsFi/Pa 257 278 302 284
PsF/Pa 280.25
PdF/Pa 962.21
PtF/Pa 1242.46
ep/% 4.85

The results show that flux of the blower has been greatly improved after the progressive orthogonal optimization, the simulation error
is acceptable and shows that the optimization of the axial-flow blower through simulation is of practical significance.

5. Conclusion
(1) Progressive orthogonal optimization method was proposed to optimize the overall structure of axial-flow fan’s flow channel. The
preliminary orthogonal optimization design is developed for axial fan first, cause it is the core components of the flow channel, the
orthogonal design of the entire flow channel is further advanced on this basis, problems of excessive workload caused by excessive
design parameters in the flow channel have been solved in this way and it has great reference value for practical applications.
(2) The results of the progressive orthogonal optimization show that the selected design variables have a great influence on the flux
and efficiency of the axial-flow blower, and the influence of the axial fan related parameters is relatively larger. The uniformity of the
flow field in the flow channel was significantly increased after optimization, the flux of the blower was increased from 0.145 m3/s to
0.171 m3/s, the total pressure efficiency was increased from 45.47% to 58.13%, and the experimental results indicated the feasibility of
using the progressive orthogonal optimization method to optimize the axial-flow blower.

Acknowledgments
The authors kindly acknowledge Science and Technology Support Plan of Jiangsu Province (No. BE2014133) and Science and
Technology Achievement Transformation Project (No. 21071213) for the financial support.

Nomenclature
AF Axial fan airfoil Pout Total pressure of outlet [Pa]
AG Guide vane airfoil PP Total pressure power [W]
D Channel diameter [mm] PS Shaft power [W]
Dis Distance of measurement point and tube center [mm] PsF Static pressure of outlet [Pa]
D1 Inlet hydraulic diameter [mm] PtF Simulated total pressure of the blower [Pa]
D2 Outlet diameter [mm] PtF_exp Experimental total pressure of the blower [Pa]
d Axial clearance [mm] Q Flux obtained through simulation [m3/s]
e Simulation error of flux [%] Qexp Flux obtained through experiment [m3/s]
ep Simulation error of total pressure [%] r Hub ratio
ev Measurement error of points with the same Dis [%] T Torque [N∙m]
L Diversion cone length [mm] vi Measurement points velocities [m/s]
L1 Wind tunnel inlet section length [mm] η Total pressure efficiency [%]
L2 Wind tunnel outlet section length [mm] βF Axial fan deflection angle [°]
n Motor speed [rpm] βG Guide vane deflection angle [°]
NF Guide vane blades number θF Axial fan installation angle [°]
NG Axial fan blades number θG Guide vane installation angle [°]
PdF Dynamic pressure of outlet [Pa] ρ Air density [kg/m3]

422
 References
[1]Masi, M., Castegnaro, S., and Lazzaretto, A., 2016, “A Criterion for the Preliminary Design of High-Efficiency Tube-Axial Fans,”
ASME Turbo Expo, Seoul, South Korea, GT2016-56960.
[2]Luo, L., Wang, Q., Yuan, M. X., et al., 2017, “Structure analysis and optimization of the axial-flow fan based on fluid-structure
interaction,” Journal of Jiangsu University of Science and Technology (Natural Science Edition), Vol. 31, No. 02, pp. 61-65+94.
[3]Tae, L., Wan, J., and Gaku, M., 2018, “Prediction of the Noise Reduction by Shroud Shape and the Characteristics of Flow-
induced Noise Generated in an Axial Flow Fan,” International Journal of Fluid Machinery and Systems, Vol. 11, No. 2, pp.181-190.
[4]Taku, I., Tetsushi, K., and Masato, F., 2017, “Study on Influence of Blade Number on Aerodynamic Noise of Half-ducted
Propeller Fans for Packaged Air-conditioners,” International Journal of Fluid Machinery and Systems, Vol. 10, No. 4, pp. 318-327.
[5]Asgarshamsi, A., Hajilouy-Benisi, A., Assempour, A., et al.,2014, “Multi-point Optimization of Lean and Sweep Angles for
Stator and Rotor Blades of an Axial Turbine,” Iranian Journal of Science & Technology Transactions of Mechanical Engineering,
Vol. 229, No. 5, pp. 1-13.
[6]Jin, K., Kyung, C., Kwang, Kim., et al., 2012, “Numerical Investigation on Aerodynamic Performance of a Centrifugal Fan with
Splitter Blades,” International Journal of Fluid Machinery and Systems, Vol. 5, No. 4, pp.168-173.
[7]Shigemitsu, T., Fukutomi, J., and Okabe, Y., 2010, “Performance and Flow Condition of Small-Sized Axial Fan and Adoption
of Contra-Rotating Rotors,” Journal of Thermal Science, Vol. 19, No. 01, pp. 1-6.
[8]Ye, X., Li, P., Li, C., et al., 2015, “Numerical investigation of blade tip grooving effect on performance and dynamics of an axial
flow fan,” Energy, Vol. 82, pp. 556-569.
[9]Yang, F., Liu, A. L., and Guo, X. Y., 2017, “Numerical Simulation on the Performance of Axial Vane Type Gas-Liquid Separator
with Different Guide Vane Structure,” International Journal of Fluid Machinery and Systems, Vol. 10, No. 1, pp. 86-98.
[10]Zhang, S. Z., 2011, “Study on influence of axial gaps on performance of axial flow fans based on CFD,” Mining & Processing
Equipment, Vol. 39, No. 07, pp. 33-36.
[11]Yin, C., Hu, J., Tu, B. F., et al., 2014, “Numerical investigation of effect of outlet hub geometry on performance of large scale
axial fan,” Journal of aerospace power, Vol. 29, No. 05, pp. 1170-1176.
[12]Xiong, J., Cui, H., and Li, J. Y., 2016, “Optimization Design of Anti-Dust Accumulation Centrifugal Fan Using Genetic
Algorithm,” ASME International Mechanical Engineering Congress & Exposition Proceedings, Phoenix, USA, IMECE2016-67575.
[13]Si, J. Z., Sun, G., and Meng, D. H., 2013, “an optimization method for offshore wind turbine blades based on PSO,” ACTA
AERODYNAMICA SINICA, Vol. 31, No. 4, pp. 498-502.
[14]Standards, B., 1977, “Measurement of fluid flow in closed conduits; velocity area method using pitot static tubes,” International
Standard Iso, iso 3966-1977(e).
[15]Hurault, J., Kouidri, S., Bakir, F., et al., 2010, “Experimental and numerical study of the sweep effect on three-dimensional
flow downstream of axial flow fans,” Flow Measurement and Instrumentation, Vol. 21, No. 2, pp.155-165.
[16]Sarraf, C., Nouri, H., Ravelet, F., et al., 2011, “Experimental study of blade thickness effects on the overall and local
performances of a Controlled Vortex Designed axial-flow fan,” Experimental Thermal & Fluid Science, Vol. 35, No. 4, pp. 684-
693.

423