The 4th International Conference on Design Engineering and Science, ICDES 2017 Aachen, Germany, September 17-19, 2017 333

Study on Optimum Design Method for Small Axial

Toshiyuki HIRANO*1, Toshio OTAKA*2 and Gaku MINORIKAWA*3 *1, 2 Mechanical Engineering Course, Department of Science and Engineering, Kokushikan University 4-28-1, Setagaya, Setagaya City, Tokyo, 154-8515, JAPAN [email protected], [email protected] *3 Department of Mechanical Engineering, Faculty of Science and Engineering, Hosei University 3-7-2, Kajinocho, Koganei City, Tokyo, 184-8584, JAPAN [email protected]

Abstract plate with the fixed thickness was used for the blade In order to establish optimum design methodology of cross-sectional shape in consideration of productivity. small axial fan, response surface method as optimization The maximum camber and its location were designed to technique was applied. Small axial fan with impeller obtain enough lift. The blade number of the impeller diameter of 36 mm was designed, prototyped and was 5 and the thickness was 0.5mm. In order to examined performance characteristics by CFD and optimize the impeller, the orthogonal array table of the experiment. In the study, relationship between design factors and levels for the experimental design was made. parameters such as blade cord length, camber and blade Table 2 and Table 3 show the factors and the setting angle and performance characteristics was orthogonal array table. Although there are many design examined by multi regression analysis using obtained parameters in fan design, three parameters of the blade data. It was found that the methodology showed optimal setting angle θ on the Tip side, the camber ratio f/L and trends of design parameters visually and improvement the blade chord length L that affect the drawing of the of performance was confirmed by simulation and blades and the performance were chosen. The base fan experiment. had the blade setting angle θ =16deg, the camber of 3% Keywords: small axial fan, performance, optimum and the blade chord length L=12mm, respectively. In the design, response surface, experiment, CFD experiment, the impeller was installed in a casing with DC blushless motor (NIPPON KEIKI WORKS, 1 Introduction LF40A12). The inner diameter of the casing was 38mm In recent years, IT products such as personal and the clearance between the blade tip and the casing computer, multi-function printer, audio and visual was 1mm. equipment and so on have been widely used because of the development of information and communication technologies. Since the demand for downsizing and improvement of the processing is increasing, the packaging density of the devices is getting higher and the thermal design is getting severer. So, the forced air cooling by small axial fan is commonly used in the (a) 3D model thermal design. However, the efficiency of small axial fan is extremely low due to not only the aerodynamic viewpoint but also the restriction such as cost and productivity, compared with industrial fans. Thus, there are few reports about the design and performance prediction on small fan [1-9]. In the present study, the response surface method as an optimization technique (b) Cross sectional view of blade was applied to the small axial fan design. Fig.1 Tested impeller

2 Design of axial fan In order to apply for car navigation system and so on, Table 1 Specification of base impeller axial fan with the frame size of 40mm was usually Imp. No. Symbol Unit Specification Cambered adopted. It has the feature that the inside diameter is Airfoil section relatively large to the outside diameter of the impeller. Plate The small axial fan used in this study is shown in Camber ratio f/L [%] 6 Figure 1. The impeller has the outlet diameter of 36mm, Camber location x/L [%] 30 and the inner diameter of 23mm and the height of Number of blade Z - 5 6.2mm. Table 1 shows the main specification of the Blade thickness t [mm] 0.5 airfoil shape designed by this study and the cross Blade chord length (Tip) L [mm] 12 sectional view of the blade, respectively. A cambered Blade setting angle (Tip) θ [deg.] 16

Copyright © 2017, The Organizing Committee of the ICDES 2017 Table 2 Factor and level of tested impeller RANS (Reynolds Averaged Navier-Stokes) and the turbulent model was a standard k-ε model. The blade Factor Level efficiency of the tested fan was calculated using the Setting angle (Tip) θ [deg.] 16, 18, 20 following equation. The of the tested fan was the Camber f/L 5, 7.5, 10 sum of a moment and a viscous moment Blade chord length (Tip) C [mm] 15, 17.5, 20 on the blade surface.

Table 3 Orthogonal layout (Parameter assignment) P Q   s 100 (1) Imp. Blade Setting angle Camber Blade Chord length TN No. (Tip) θ [deg.] f/L (Tip) C [mm] 3 1 16 5 15 Ps : Static pressure[Pa], Q : flow rate[m /min], 2 16 7.5 17.5 T: Torque of fan [Nm] 3 16 10 20 4 18 5 17.5 4 Experimental apparatus and method 5 18 7.5 20 The experimental apparatus is shown in Figure 3. 6 18 10 15 This apparatus mainly consisted of a tested fan, a 7 20 5 20 chamber, a flow meter and a suction device. The wall 8 20 7.5 15 static pressure holes were installed at 40mm upstream 9 20 10 17.5 and downstream of the orifice. The tested fans were manufactured using a stereo lithography machine 3 CFD analysis (EnvisionTEC, ULTRA). The static pressure Ps, which CFD (Computational ) analysis for was the wall pressure on the front chamber, was the same flow fields as the experiment was performed measured by a small digital differential pressure gauge by using commercial software (Software Cradle (NAGANO KEIKI, GC31). The flow rate Q was SCRYU/Tetra). The calculation grids were mainly obtained by measuring the flow meter (KEYENCE, non-structural tetra mesh and partially structural FD-A50). hexagonal mesh around the blade and the wall surface. In order to obtain a performance characteristic The calculating region consisted of the inlet, the outlet curve which is called a PQ curve, firstly, the maximum and the fan as rotational region. The number of grids static pressure Psmax was measured by sealing the was about 5,800,000. The dimensions of the outlet chamber and the maximum flow rate Qmax was obtained region imitated the experimental apparatus. The entire so as the static pressure at the chamber was set to 0Pa and enlarged view of the calculation models are by adjusting the suction device. Then dividing Qmax into illustrated in Figure 2. several points, values of Ps at given Q were obtained. The of the impeller was kept at 6,800rpm by adjusting the input voltage.

(a) Overall view

Fig. 3 Schematic view of test apparatus

5. Results and Discussions Figure 4 shows the calculated PQ curves of the tested fans. In this figure, the vertical axis indicates the (b) Enlarged view of tested fan region static pressure in the chamber and the horizontal axis indicates the flow rate. The maximum static pressure Fig. 2 Overall view of computational grids and the maximum flow rate varied into 31.1-35.5Pa and 0.11-0.14m3/min by changing each parameter. The As the boundary conditions, the inlet surface was tendencies of the characteristic curves were almost set by atmospheric pressure and the outlet surface was same. regulated by given flow rates. The tested fan region was Figure 5 shows the calculated PQ and efficiency constructed by moving meshes as ALE (Arbitrary curves of fans No.1, No.2 and No.3. The vertical axes Lagrangian and Eularian) method. To obtain the show the static pressure and the blade efficiency, and maximum flow rate, static pressure at the outlet surface the horizontal axis shows the flow rate. The blade was set to 0Pa. The calculation was performed by steady setting angle of each impeller is β = 16 °, and as the impeller No. increases, the camber f/L and the chord length C increase. Comparing the analysis results, it can be seen that both the static pressure and the maximum flow rate increased as the impeller No. increased in the region where the flow rate Q was 0.03 m3 / min or greater. It is considered that the blade with a long chord length was not affected by the stall caused by the separation and the lift increased due to the generation of the pressure difference in the larger area. Figure 6 shows the calculated PQ and efficiency curves of No. 1, No. 4 and No. 7. The camber of each impeller is the same, and as the impeller No. increases, both the blade setting angle and the chord length increase. Comparing the results, both static pressure and maximum flow rate increased as the impeller No. increased. This is also considered that the performance Fig. 6 Performance and efficiency curves of tested of both the static pressure and the flow rate improved as fan (No.1, No.4 and No.7) a result of the increased amount of air pushed out by the fan due to the reaction of the increased lift force. Figure 7 shows the comparison of the PQ curves and calculated static pressure efficiency curves of No.3 and No.9, in which the largest difference on the efficiency was shown. As shown in Figure 7, the static pressure efficiency η of No.3 was larger than that of No.9 below Q=0.09m3/min. But the static pressure and maximum flow rate of No.3 were lower than that of No.3 over Q=0.09m3/min.

Fig. 7 Performance and efficiency curves of tested fan (No.3 and No.9)

Significance of the three factors was examined by F test on two cases that the maximum static pressure efficiency was set as the objective function (Optimum η) and the maximum flow rate as the objective function (Optimum Q). As a result, the confidence interval was lower than 95% in case of the Optimum η and it is considered that these three factors had little influence. On the other hand, in case of the Optimum Q, the Fig. 4 Performance curves of tested fan (No.1 – 9) confidence interval of the chord length was slightly less than 95%, but the confidence interval of the blade setting angle and chamber were more than 95%. It is thought that the influence of three factors existed. The linear regression equation was obtained by least-squares method from the relationship between the objective function and the factors in Table 4 and Table 5. These equations are indicated by equation (2) and equation (3). The expectancy of each level was estimated by using each equation as shown in Table 4. The values of the multiple correlation coefficients, the multiple contributing ratios and the contributing ratios adjusted for the degrees of freedom of correlation, which calculated the precision of the regression model from predicted value, were shown in Table 5. It is considered that there are no correlations in three factors Fig. 5 Performance and efficiency curves of tested about the Optimum η condition. On the other hand, it is fan (No.1, No.2 and No. 3) considered that there are correlations in three factors about the Optimum Q condition.

Optimum η: Table 6 Optimum value y=23.3-0.139x1-0.020x2+0.015x3 (2) Factor Optimum η Optimum Q Blade setting angle (Tip) θ Optimum Q: 16 20 [deg.]

y=0.0472+0.00289x1-0.00213x2+0.00067x3 (3) Camber f/L 10 10 Blade Chord length (Tip) C 15 20 [mm] Table 4 Orthogonal layout (expectancy) Static pressure Maximum flow rate Figure 9 shows the performance curves of each Imp. No. efficiency η [%] Q[m3/min] optimum fan and base fan, respectively. The fans of 1 21.2 0.114 Optimum η and base fan were analyzed and also tested. 2 20.0 0.112 The performance curves of analysis results and 3 21.7 0.128 experimental results in these fans were same tendency. 4 20.2 0.123 5 21.6 0.128 6 21.6 0.129 7 21.1 0.128 8 21.2 0.130 9 19.2 0.138

Table 5 Analysis of multiple correlations Optimum η Optimum Q Multiple correlation 0.284 0.978 coefficient[-] Contributing ratio[-] 0.081 0.960 Contributing ratio adjusted for the degrees -0.470 0.936

of freedom[-] (a)Base fan

The response surface was made by using the factors and the objective functions. In the study, the response surface had four dimensions because three factors were used to one function. Therefore, one factor was fixed and the three-dimensional chart with other factors as variables was made. Figure 8 shows the response surface of blade setting angle β=16deg. as an example. In this figure, the x, y and z axis indicate the blade chord length C, static pressure efficiency η and camber f/L, respectively. As the results, the optimum values of C, η and f/L in the Optimum η condition were determined as shown in Table 6. In the same way as above, the optimum values in the Optimum Q condition were also determined in Table 6. (b)Optimum η

(b)Optimum Q

Fig. 8 Response surface (β=16°) Fig.9 Performance and efficiency curves of tested fan

The maximum static pressure efficiency of Optimum η Some Parameters on Blade Elements, Transactions was increased by around 4% compared with that of base of the Japan Society of Mechanical Engineers, Vol. fan. In addition, the value of Optimum η was the 44, No. 380(1978), pp. 1301-1310. largest value in those of fans which were designed by [2] T. Ito, G. Minorikawa, A. Nagamatsu and S. Suzuki, the design of experiments. However, it should be “Experimental Research for Performance and Noise considered that the fan of Optimum η was little of Small Axial Flow Fan (Influence of Parameter of correlation between the maximum static pressure Blade),” Transaction of Japan Society Mechanical coefficient and the three factors. On the other hands, the Engineering, B, Vol. 72 (715), (2006) pp. 670-677. maximum flow rate of Optimum Q was increased by [3] T. Iwase, T. Hioki, Y. Kato, T. Tanno, O. Sekiguchi around 8% compared with that of base fan. The value of and M. Furukawa, “Influence of Interaction with Optimum Q was the third largest value in those of fans Tangential Lean Blade and Box Type Casing on which were designed by the design of experiments. Blade Passing Frequency Noise Level in Small The impeller was made on the response surface Axial-Flow Fans,” Transaction of Japan Society considered to be correlated, and since the maximum Mechanical Engineering, B, Vol. 77 (780), (2011), flow rate was improved compared with the reference fan, pp. 1620-1629. the significance of the response surface method could be [4] T. Iwase, K. Sugimura and T. Tanno, “Study on confirmed. However, the shape of the optimized Improvement of Fan Efficiency in Small Axial-Flow impeller has not yet been reached. In this study, it is Fans: 1st Report, Designing a High Efficiency Fan considered that it influenced the correlation coefficient by Using Numerical Optimization,” Transaction of and the analysis and test result in consideration of the Japan Society Mechanical Engineering, B, Vol. 75 fact that it is a parameter set in the design of the (757), (2009), pp. 1750-1756. impeller with restrictions setting the casing. [5] T. Iwase, K. Sugimura and T. Tanno, “Study on Improvement of Fan Efficiency in Small Axial-Flow 7 Conclusions Fans: 2nd Report, Influence of Tip Leakage Vortex In order to optimize the small axial fan by using on Static Pressure and Fan Efficiency,” Transaction the response surface, various types of impeller were of Japan Society Mechanical Engineering, B, Vol. 75 designed by the design of experiments and the effects of (757), (2009), pp. 1757-1762. three parameters on the performance were investigated [6] T. Sasajima and K. Kawaguchi, “Numerical Analysis by calculation and experiment. The present study of Flow around Blades in Axial Flow Small Fan,” obtained the following conclusions. Transaction of Japan Society Mechanical (1) As a result of creating the response surface, it was Engineering, B, Vol. 77 (774), (2011), pp. 255-263. possible to visually judge the optimal parameter [7] Toru Shigemitsu, Junichiro Fukutomi and Agawa trend under fan design conditions in this study. In Takuya, Internal Flow Condition of High addition, the small axial fan designed based on the Contra-Rotating Small-Sized Axial Fan, result was able to obtain high static pressure International Journal of Fluid Machinery and efficiency and maximum flow rate, and it was Systems, Vol.6, No.1, (2013), pp.25-32. possible to confirm the effectiveness of the response [8] Toru Shigemitsu, Junichiro Fukutomi and Shimizu surface method. Hiroki, Influence of Blade Row Distance on (2) Numerical analysis results and experimental results Performance and Flow Condition of Contra-Rotating of the model using the parameters of the obtained Small-Sized Axial Fan, International Journal of optimum solution show almost the same tendencies, Fluid Machinery and Systems, Vol.5, No.4, (2012), and it was found that application of the response pp.161-167. surface method to actual machines is possible. [9] Kiyoshi KAWAGUCHI, Fuminobu WATANABE and Kenji OOE, Flow around Fan Blade and Pressure Fluctuation on Blade Surfaceof Small Axial Acknowledgment Fan with Obstacle near Inlet, Journal of This work was supported by JKA and its promotion , Vol.41, No.12, (2013), pp.723-733. funds from KEIRIN RACE. (in Japanese)

References Received on April 12, 2017 [1] S. Suzuki, An Experimental Study on Noise Accepted on May 29, 2017 Reduction of Axial-Flow Fans: 1st Report, Effect of