Aerodynamic wall based on industrial axial fans

07001


Introduction
Recently, due to a significant leap in the development of electronics, control automation algorithms, power storage systems and improvement of electric propulsion systems, there has been a breakthrough in the production of small-scale unmanned aerial vehicles with vertical takeoff and landing.Mainly using a quad-and multicopter schemes that have found wide application as devices for photo and video recording, for agricultural purposes and for other tasks.At the same time, interest in the development of urban passenger and cargo air transport (aero-cycles, air taxis, cargo drones, etc.) including on alternative and combined flight patterns has grown.
One of the most difficult tasks in the development of such aircraft is the study of flight aerodynamics (in particular, the transition from vertical to horizontal motion, high-speed flight and interaction with oncoming wind flows) and the development of manual and automatic piloting algorithms.Testing in the "field" conditions is difficult for many factors, so testing in wind tunnels is necessary.Testing even small vehicles under flight conditions requires wind tunnels with large working areas.Taking into account the complexity of organizing the lease of large wind tunnels, the permits for research conducting in them and the cost of a working hour, long-hours test and debugging research becomes almost impossible.
As an alternative to classical wind tunnels, the socalled pixel wind tunnels have recently been widely developed in Europe [1].They consist of an array of small but powerful server cooling fans.Such systems make it possible to obtain a flow of sufficiently high velocities over a large area.At the same time, special algorithms, by controlling the speed of individual fans, make it possible to vary the speed profile generated by the system both in space and in time.The disadvantage of such systems is the rather high price.
In this paper, using numerical and experimental methods, the possibility of creating a pixel wind tunnel (wind wall) using industrial axial fans is considered.Based on the parameters of the generated flow (more than 10 m/s) and the geometrical size of the created wind wall, the best option was the fan VO-6-300-4 0.75 kW/3000 rpm (Fig. 1).The fan diameter is 0.4 m, so when assembling a 4x4 fan array, the outlet section of the wind tunnel is approximately 2x2 m.Initially, the aerodynamic parameters of the outlet flow were experimentally determined using an anemometer and an automated coordinate device, the energy parameters of the fan were determined using current and voltage meters.The comparison of parameters with the technical characteristics of the fan declared by the manufacturer was carried.Subsequently, E3S Web of Conferences 459, 07001 (2023) https://doi.org/10.1051/e3sconf/202345907001XXXIX Siberian Thermophysical Seminar the obtained profile of the outlet flow rate was used to verify the mathematical model of a single fan.

CFD modelling
The simulation was carried out in the Ansys Fluent package using the k-ω SST turbulence model.At the first step, a single fan with full resolution of the blade system was modeled, using the sliding mesh approach.The computational mesh for the model contained about 4 million cells with local refinement in the area of flow formation.
The velocity components and turbulent characteristics obtained as results of the calculation (Fig. 2.) were stored in the outlet section of the fan.At the second stage, these data were used to form the boundary conditions for calculating the fan array.For most aerodynamic studies, it is necessary that the flow on the model be straight and uniform.In Fig. 2. it can be seen that a large swirling flow with a highly localized recirculation zone is formed at the outlet of a single fan.Such a flow is unsuitable for research; however, when the array is assembled, mutual "untwisting" of neighboring jets occurs.Thus, the flow swirl remains only at the periphery, as shown schematically in Fig. 3 for a 4x4 fan array.As the number of fans in the array increases, the relative area of the peripheral flow decreases.In addition, to suppress peripheral swirl, it is proposed to install stabilizing ribs (Fig. 4) along the perimeter at the outlet, Calculations were made and flow patterns with and without ribs are compared for variants of 3x3, 4x4 and 8x8 fan arrays.When forming an array of 3x3 fans, the peripheral flow merges with the co-rotating flow of the central fan, giving the entire flow a constant swirl.Fig. 5a shows the axial velocity in longitudinal section, and Fig. 6a-7a show the axial and tangential flow velocities in transverse section at a distance of 3 m from array exit, showing a significant flow non-uniformity.
In this case, the use of stabilizing ribs fig.5b-7b does not help to fully solve the problem of non-uniformity of the flow.A pronounced zone with a prevailing axial velocity is not observed, and the tangential component in the core of the flow reaches 20% of the axial one (Fig. 6-7).This leads to the conclusion about the inexpediency of creating arrays of 3x3 fans.However, for 4x4 fan arrays, a wide zone with a rectilinear flow is formed at the outlet.Considering the section at a distance of 3 m from the array exit, one can note the presence of a large velocity gradient; this is due to a rather large zone with peripheral swirl.In this case, the flow velocity in the core reaches 12 m/s.A swirling flow is observed along the periphery, in which the ratio of tangential to axial velocity reaches 20%.In this case, peripheral swirling leads to helical twisting of the flow core.Fig. 9a shows a noticeable rotation about the axis of symmetry.
The installation of peripheral ribs leads to a decrease of the tangential velocity in the peripheral region to 10% of the axial one, and the rotation of the core becomes insignificant (Fig. 9b, 10b).Fig. 11 shows the velocity profiles in the central section at a distance of 3 m from the array.It can be seen that the axial velocity profile has a parabolic shape.At the same time, there is a slight decrease in the axial component of the flow velocity (by 3%) in the case of using peripheral ribs.Thus, a 4x4 fan array does not allow the formation of a uniform velocity profile that is preferable for aerodynamic tests; however, such a configuration already allows a number of laboratory studies to be carried out.Fig. 12-14 shows the calculation data for an 8x8 fan array.In this case, the need to use peripheral ribs is also noted.Their use leads to the alignment of the flow core, the reduction of the zone of peripheral swirl and the reduction of the tangential velocity component by half.At the same time, as with an array of 4x4 fans, there is a slight decrease in the axial velocity component (about 2%).It should be noted that the proposed geometry of the peripheral ribs is primarily due to the simplicity of their implementation.If necessary, their optimization can lead to a greater effect of reducing the tangential flow velocity.Analyzing the velocity profiles (Fig. 15), it can be noted that in this wind tunnel configuration, a wide zone with a uniform profile is formed.At a distance of 3 m from the exit section, its area is 3x3 m2 with the size of the array 4x4 m2.Such dimensions are quite sufficient for testing the flight modes.
In the course of the work, the fundamental possibility of creating a pixel wind tunnel based on industrial axial fans is shown.It is shown that, starting from 4x4 fan arrays, a rectilinear flow zone is formed; however, the formation of a zone with a unifrom profile occurs only with an increase in the number of fans in the array.Thus, in an 8x8 fan configuration, the uniform profile area is approximately 60% of the array area.

Server fan device
Based on the results of a numerical study, it was decided to create an aerodynamic wall in the form of a 4x4 fan array.Preliminarily, to study the features of the generated flow and the possibility of using confusers to increase the speed, a model installation was created based on DELTA PFC1212DE 4.8a 120mm server fans.The size of the fan array is S = 480 x 480 mm.(Fig. 16) The horizontal velocity profiles in the central section at distances of 1Н, 3Н, 5Н and 8Н from the plane of the fans, where H is the height of one fan were measured using the TestAir AM-70 anemometer (see Fig. 17  It can be seen that at short distances the speed is highly non-uniform, the minima in speed are caused by large passive zones between the fans.At a distance of about 8H, the flow profile becomes smooth and the velocity profile looks similar to the calculated one (Fig. 11) Next, the effect of confusers on the flow characteristics was studied.The model wall was equipped with additional aerodynamic sections: a straight channel with a square section with a height of Hc = 500 mm and confusers with an outlet square section with a height of Hc = 400, 300 and 200 mm (see Fig. 18)  It can be seen that the use of a direct channel leads to the alignment of the flow and the formation of a large zone with uniform profile (Sun).Based on the velocity profiles, an estimate of the dependence of the increase in the flow velocity in the region with the uniform profile Uc/U on the decrease in the outlet section of the confuser Sin/Sout was made.An estimate of the ratio of the uniform profile behind the confuser area to the aerodynamic wall without confusers area Sun/S was also made (Table 1) It can be concluded from the table that the optimal decrease in the area of the outlet section is in the region of the value Sin/Sout = 1.5, while the speed also increases by about one and a half times.With further narrowing of the channel, the speed increases less significantlythis is because axial fans -are fans of low overpressure and with a strong narrowing of the output flow, their flow characteristics deteriorate significantly.

Industrial axial fan device
Based on the results of preliminary studies, it was decided to make an aerodynamic wall with fans VO-6-300-4 (Fig. 20).The outlet section of the implemented aerodynamic wall is 3.7 m 2 (1.92x1.92m), while the active area of the rotors is 2 m 2 -the rest of the space between the fans remains passive.Each horizontal row (4 fans) is controlled by a frequency changer EKF 4/5.5kW, 3x400 VEKTOR-100, it makes possible to vary the profile of the output flow vertically.The total power consumption is estimated at 12 kW.Using an anemometer, the velocity profiles were measured at a distance of 5H, 7.5H, and 10H downstream the aerodynamic wall.At a distance of 5H, as in the case of the server fan device, the effect of passive zones on the velocity profile is still noticeable.At a distance of 7.5H, this influence is already less noticeable; at 10H, the profile is smooth (Fig. 21).In Fig. 22. a comparison of the horizontal and vertical profiles at a distance of 7.5N is given.It can be seen that the vertical profile is not symmetrical, this is due to the strong interaction of the flow with the floor surface.In addition, as with the server fans device, it is recommended to use an additional straight channel and peripheral ribs to equalize the flow, or to place the fans higher above the floor surface.For the aerodynamic wall, a confuser was made with an outlet section of 1.5x1.5m.When using it, the speed in the core of the flow increased by 1.4 times (Fig. 23).
Fig. 24 shows a comparison of velocity profiles in a dimensionless form for server fans model and full-scale aerdynamic wall without confusers and with confusers (flow restriction coefficient S/Sc = 1.63).The normalization was carried out according to the velocity on the axis of symmetry of the wall in the variant without the confuser.The results are in good agreement, which indicates the scalability of this setup.

Fig. 24 .
Fig. 24.Comparison of dimensionless velocity profiles for server fan and industrial fan aerodynamic walls.

Table 1 .
Flow parameters for different confusers.