Acta Polytechnica https://doi.org/10.14311/AP.2024.64.0398 Acta Polytechnica 64(5):398–406, 2024 © 2024 The Author(s). Licensed under a CC-BY 4.0 licence Published by the Czech Technical University in Prague THE EFFECT OF THE INLET CHAMBER AND DIFFUSER ON THE REVERSE AERODYNAMIC PERFORMANCE OF AN AXIAL FLOW FAN Václav Cyrusa,b,∗, Jan Cyrusa a AHT Energetika, Štětínská 15, 180 00 Prague, Czech Republic b Czech Technical University in Prague, Faculty of Mechanical Engineering, Department of Energy, Jugoslávských partyzánů 3, 160 00 Prague, Czech Republic ∗ corresponding author: cyrus.aht@iol.cz Abstract. Flow reversal in axial fans is usually carried out by changing the direction of rotation of the fan and rotating the stator vanes. The total pressure losses of selected inlet fan chambers were determined using CFD data for the normal and reverse flow conditions. These losses are 3.5 to 7.2 times greater for the normal flow increase due to the flow reversal in three chambers with the hub/tip ratio of 0.45 to 0.63. The operating points of two complete fans made up of the inlet chamber, the diffuser, and the axial stages A and B with different aerodynamic load were obtained using one-dimensional engineering calculation method. It uses the measured performance data relating to both fan stages. The calculated volumetric flow rate ratio for reverse and normal operation is 0.68 and 0.55, respectively, for the complete axial fan with the stages A and B. The acquired results are in accordance with the customers’ requirements. Keywords: Axial flow fan, reversing, inlet chamber, performance, simulation. 1. Introduction Change of the flow direction is often required in emer- gency situations, in road tunnels, coal mine ventilation, and in a number of chemical plant production lines where the axial flow fans are used. They consist of the inlet chamber (IC), the rotor blade row (R), the inlet guide vane (IGV), the stator vane rows (S), and the outlet diffuser (D); see to Figure 1. Reverse flow is most commonly achieved by changing the direction of rotation of the fan and rotating the stator vanes. The required reverse volume flow rate Qrev is typ- ically at least 50 % to 60 % of the design operating condition Qn. The stage blading (IGV, rotor and stator rows) efficiency ηST,max should be higher than ηST,max = 89 % to 90 % for standard fan operation. The fan efficiency ηrev is usually not specified for the emergency situations. There are only a few works in the literature, namely by Brusilovskij [1],Wallis [2] and Dunn et al. [3], which give brief information on the special reverse axial flow fans. The contribution to the understanding of the flow mechanism in the blade system, based on the investigation data relating to three stages of the reverse axial flow fans, can be found in the works of Cyrus et al. [4], [5]. Our paper presents design study of the complete reverse axial flow fan (IC+IGV+R+S+D) where the directional change of fan rotation and turning of the stator vanes occurs. If the standard axial fan stage is used in the reverse axial fan design then the modified stator vanes must be used in order to meet the condi- tion of the stator vanes being able to turn in one go. In paper by Cyrus et al. [6], the design of new stator vanes for the reverse fans is carried out using the flow simulation results. Abdolmaleki et al. [7] analysed the aerodynamic performance of two-dimensional cascades with ellip- tical profiles for isolated rotor blade rows of an axial flow fan. The results were obtained using CFD sim- ulations. Flow reversal is achieved by changing the direction of the fan rotation. The new concept of the axial flow fan stage is described in the publication by Krasyuk et al. [8]. It consists of rotor blades and inlet guide vane rows. During fan reversal, the rotation of the rotor blade row is stopped. The rotation of the inlet guide vane row has then started. This procedure helped to increase the efficiency of the fan during flow reversal. During civil aircraft landings, the thrust of the turbofan engine can be reduced by rotating the rotor blades of the axial flow fan. The core multistage axial compressor operates unchanged. This approach can produce a reverse flow in the axial flow fan as described in design studies such as [9], [10]. They show the calculated flow fields at selected fan operating points. Some suggestions for a new approach to the aerodynamic design of fan blades are also given. The new results of the last cited papers cannot be applied in our proposed investigations. In fact, the effort is concentrated on the complete axial fan, where the flow reversal is achieved by changing the direction of rotation of the fan and rotating the stator blades. In order to determine the complete reverse axial fan aerodynamic performance, the inlet chamber pres- sure losses data for the normal and reverse fan flow directions are required. Our focus is primarily on the 398 https://doi.org/10.14311/AP.2024.64.0398 https://creativecommons.org/licenses/by/4.0/ https://www.cvut.cz/en vol. 64 no. 5/2024 The effect of the inlet chamber and diffuser on the reverse . . . Figure 1. Axial flow fan (Stage A: Dt = 600 mm, Dh = 330 mm). Design parameters Stage φD ψD DFR,m ν zR zS υR,m υS,m ( s c ) S,m ( s c ) S,h φrev φ ∗ φrev φ ∗∗ [°] [°] A 0.40 0.30 0.36 0.55 12 15 19.5 45.0 1.36 0.86 0.77 0.81 B 0.35 0.42 0.48 0.63 16 17 24.6 51.1 1.20 0.56 0.61 - ∗ IGV+R+S configuration. ∗∗ R+S configuration. Table 1. Fan stages design parameters. inlet chamber, where the inlet flow velocity vector is perpendicular to the fan axis (Figure 1). New re- sults were obtained using the commercial CFD code Numeca [11] for the inlet chamber with rectangular inlet sides with the ratio of b a = 1.32; see Figure 1. Three hub/tip ratios: ν = 0.45, 0.55, and 0.63 were investigated. The last part of our paper presents the evaluation of the key parameters, namely the fan reversibility, the flow rate ratio Qrev Q . It has been determined for two complete fans with the axial flow stages A, B [4] and [5]. The fan performance maps of the specified fan stages practically cover the customers’ requirements in the mining and chemical industries. Our calculation is based on the simple one-dimensional engineering calculation method using the theoretical and test stage performance curves. The total pressure losses of the inlet chamber, diffuser, and confuser for standard and reverse fan operation have been applied. The results obtained will allow the fan designer to assess the effect of individual components of the complete fan on its aerodynamic performance during standard and reverse fan operation. 2. Axial flow stages for reverse fans Our paper analyses the data relating to the two axial fan flow stages, denoted A and B, designed at AHT Energetika Ltd. with the aid of in-house design code designated for the axial flow compressor stages. The key fan stage parameters, namely the flow coefficient φ, the pressure coefficient ψ, and the efficiency η are defined as follows: φ = Q Aut , (1) ψ = 2∆pT ρu2 t , (2) η = Q ∆pT P . (3) Selected key aerodynamic and geometric parameters are given in Table 1. A is the flow area A = πD2 t 1−ν2 4 ; ut is the peripheral velocity; ∆pT is the total pressure increment in the fan stage ∆pT = pT 3 − pT 0; ρ is the gas density; Dt is the external diameter; ν is the stage hub/tip ratio ν = Dh Dt and P is the fan stage work input P = Mkω. Angular frequency is denoted 399 Václav Cyrus, Jan Cyrus Acta Polytechnica by Greek letter ω. The test rig was driven by a DC motor with a swinging stator in order to measure the torque moment Mk by weighing. Equation (3) can be derived from the definition of the isentropic efficiency of a turbo-compressor, e.g. [12]. Here, the compressibility of the air is taken into account. Stages A and B have moderate aerodynamic load, which is suitable for designing a larger number of ventilation fans. The blade geometry is given at the mean radius rm (rm = 0.5(rt + rh)). The stage pa- rameters φ and ψ are defined by the Equations (1) and (2), respectively. The symbol z denotes the num- ber of blades. The cascade profile camber υ is given by the following relationship: υ = βL1 − βL2, where βL1 and βL2 are the blade angles (Figure 2) that the entry and exit tangents to the camber line make with the axial direction. DFR,m is the design diffusion fac- tor of the rotor blades on the mean radius, originally formulated by Lieblein [13]. The cascades of stages A and B have the rotor diffusion factor at the mean radius of DFR,m = 0.36 and 0.48, respectively. The external diameters of the tested fan stages were the same: Dt = 600 mm (Figure 1). The rotor and stator profiles for all stages were NACA 65 series with reinforced trailing edge and circular camber lines. The profile coordinates can be found, for example, in [2], [12]. The stage B had curved plate stator vanes. The blades of all tested axial stages were designed assuming the uniform span- wise work distribution. Near the rotor blade axis, the ratio of tip clearance and chord length sr ct was 0.008. The tests were carried out on fan stages A and B at revolutions of 2 200 and 1 500 rpm, respectively. The inlet flow Mach number lower than Ma1 = w1 a < 0.35. a is sonic velocity. The air compressibility can be neglected. The Reynolds number values Re = ut cR,m ν ranged from 300 000 to 450 000, where ν is the kine- matic viscosity of the air and cR,m is the chord length of the rotor blade at midspan. Detailed flow field investigations in planes 0, 1, 2 and 3 (Figure 1) were carried out using 5-hole conical pressure probes with a sensor diameter of 2.5 mm. The pressure probes were located at 6 to 11 radial positions and the stator vanes pitch at 15 to 20 circumferential points. Thermocouples were fitted in planes 0 and 2 at the midspan. This measurement technique was used for the stage A. In the case of the stage B, the pressure probes were only in planes 0 and 3. The flow parameters acquired from the pressure probes and thermocouple data were averaged over the vane pitch. The fan stage performance curves’ points were calculated using the averages of the flow properties in the inlet and outlet planes of the tested stage. The absolute measurement uncertainty of the stage efficiency was calculated as ±0.9 % to ±1.1 % at the design point. Measurement errors of the following properties were considered in the uncertainty analysis: pressure ±5 Pa, temperature ±0.3 K, flow angle ±1°, torque ±0.08 Nm. Figure 2. Axial flow fan stage diagrams with velocity triangles: standard flow. 3. Flow description in the fan stage blades during reverse operation Figures 2 and 3 show 2D basic fan stage cascades to- gether with the velocity triangles valid for the standard and reverse runs. Firstly, we consider the operating condition for the rotor blade row cascades during the fan stage reversal [4]. By changing the rotor blade row direction of a ro- tation, the compressor-type cascade becomes the turbine-type cascade. For the compressor work, dur- ing the rotor blades reverse operation, the relationship is valid for the relative flow angles (Figure 3): β′ 2 > β′ 1. (4) By introducing the incidence angle i′R and the flow deviation δ′ R inequality (4), it changes to: βL2,R + i′R > βL1,R − δ′ R, (5) where βL1,R and βL2,R are the rotor blade angles. Simplifying, we arrive at: i′R > kυR, υR = βL1,R − βL2,R, (6) where the constant k can be within the range of k = 0.60 to 0.85 for the standard cascade parame- ters [4]. The condition (6) shows that the required 400 vol. 64 no. 5/2024 The effect of the inlet chamber and diffuser on the reverse . . . Figure 3. Axial flow fan stage diagrams with velocity triangles: reverse flow. rotor row i′R incidence angle, essential for achieving the compressor cascade regime, increases with the blade profile camber υR. During the flow reversal, the stator vanes are rearranged in the opposite sense of the stagger angle γS in relation to the original position. This step follows from the velocity triangle in plane 2 (Figure 3). In order to achieve the same relative flow angle β′ 2, and therefore the same incidence angle of rotor row i′R, the axial velocity component w̃′ a2 would have to be significantly reduced, as shown by the ve- locity triangle. The original setting of the stator blade has been retained. However, this is not desired, as it increases the gap between the performance lines for the standard and reverse operation [1], [4], [5]. 4. Performance of fan stages A and B (IGV+R+S) during standard and reverse operation Comparisons of the fan stage A measured pressure co- efficient ψ, the efficiency η, and the flow coefficient φ for the standard and reverse operation are shown in Figures 4 and 5. The stator vanes were set to the stagger angle γS = −10° at flow reversal. The resis- tance graph shown in Figure 4 intersects the design point D. The ratio of the flow coefficients was found to be φrev φn = 0.77. For the fan stage B, this ratio is lower φrev φn = 0.62 owing to the higher aerodynamic loading of the rotor blade elements; see to Table 1. It is apparent that in the fan stage, the reverse flow parameter φrev φn (Qrev Q ) decreases with the increasing aerodynamic loading of the rotor blade elements. It relates to the increased blade profile camber υR. Figure 4. Relationship between the pressure coef- ficient and the flow coefficient for stage A and the complete fan during the standard and reverse flow. Figure 5. Relationship between the efficiency and the flow coefficient for stage A and the complete fan during the standard and reverse flow. 401 Václav Cyrus, Jan Cyrus Acta Polytechnica Figure 6. Inlet chambers for fan hub/tip ratios: ν = 0.45, 0.55 and 0.63; Dh = 270 mm, 330 mm and 378 mm. 5. Inlet chamber pressure losses during the normal and reverse flow 5.1. CFD simulations of the inlet chamber A detailed flow analysis of the complete axial fan was carried out using CFD flow simulations [11] and test results, [5], [14] over a wide operating range at standard fan operation. The 3D flow in the outlet diffuser and inlet chamber was also studied, particu- larly at off-design conditions. The results obtained are compatible with published data, e.g. [15], [12]. Data relating to the inlet chamber reverse flow pres- sure losses are not available in the relevant literature. For this reason, the study was carried out on one type of chamber in the range of hub/tip ratios ν = 0.45 to 0.70. The relative geometric parameter of the inlet rectangular area was: b a = 1.32 (Figure 1). The com- parison of three flow channels of the inlet chamber with different hubs is shown in Figure 6. The hub/tip ratio values were ν = 0.55, 0.63, and 0.45. The first two hub variants are formed by parts of a hemisphere, the last one by a part of an ellipsoid. CFD simulations were performed using the CFD commercial code Numeca [11] in the inlet chamber with the rectangular inlet duct L Dt = 1.67 (Figure 1, relative distance of the planes) for the standard and reverse flow directions. The commercial code solved the full Reynolds-averaged Navier Stokes equations for the structure multi-block arbitrary grid topologies. Hexadron cells were used. The explicit time march- ing 4 steps Runge-Kutta procedure with the implicit residuum smoothing was used. A cell centred second order finite volume discretisation was applied. The cell width, at walls, was kept in the range of 0.03 to 0.07 mm in all cases to ensure that the wall y+ did not exceed 5. The air was modelled as a perfect gas. A no-slip boundary condition was imposed on the Figure 7. Flow velocity contours in the fan inlet chamber’s vertical symmetry plane during the stan- dard flow (ν = 0.55). Figure 8. Flow velocity contours in the fan inlet chamber’s vertical symmetry plane during the reverse flow (ν = 0.55). walls. The Spallart-Almaras turbulence model was applied. To take advantage of the lateral symmetry of the domain, only half of the chamber section was modelled, see Figures 7, 8, and 9. The position of the meridional symmetry plane µ can be seen in Figures 1 and 9. It is defined by the straight lines p and q. By imposing the symmetry condition on the surface of the plane µ, zero shear is obtained. 402 vol. 64 no. 5/2024 The effect of the inlet chamber and diffuser on the reverse . . . The computational grid consisted of 2.5 million cells, distributed over three main sections of the domain: the rectangular inlet duct, the box and the annular section. The number of cells was chosen according to rules published in the literature, e.g. [11], based on CFD results. These data were obtained for the cell number in the range of 1.0 to 3.5 million. The flow turbulent and laminar viscosity ratio νT νL = 6.5 was used in the turbulence model at the inlet chamber’s inlet plane. This is appropriate for the turbulent flow behind the fan blades. The convergence was evaluated through a compu- tation of residuals, which, at the end of calculations, were of the order of 10−6. The inlet flow conditions were modelled in the CFD simulations according to the Stage A test results. For the normal fan operation, the inlet flow angle in plane 00 was zero: α0 = 0°. The spanwise distri- bution of total pressure pT and flow angle α0 in the chamber inlet plane 1 was used for the flow rever- sal. These data refer to experimental results obtained for operating points in the range of flow coefficient φrev = 0.22 to 0.41 (Figure 4). The spanwise distribu- tions of the angles are in the range of values α0 = −7° to +7° for the relative blade height z h = 0 to 0.8. 5.2. CFD results and their analysis The comparison of the total pressure losses given by the loss coefficients at the standard ζIC and the reverse flow directions ζIC,rev is shown in Table 2 for three hub/tip ratios: (ν = 0.45, 0.55, and 0.63). Definitions of loss coefficients ζIC are as follows: ζIC = pT 00 − pT 0 q1 standard flow, ζIC,rev = pT 0 − pT 00 q1 reverse flow, q1 = 0.5ρ(Q A )2. (7) Table 2 shows that the loss coefficient ζIC,rev in- creases with increasing hub/tip ratio ν. It is re- lated to the increasing area ratio A00 A0 in the range of A00 A0 = 1.9 to 2.5. This tendency causes the growth of the separated flow areas on the surfaces of the inlet chamber flow channels. At the lowest value of ratio ν = 0.45, the loss coefficient ζIC,rev is significantly lower ζIC,rev = 0.265 than in the cases of hub/tip ratios ν = 0.55 and 0.63: ζIC,rev = 0.45 and 0.501, respectively. This can be explained by lower flow dif- fusion in the inlet chamber at flow reversal. Inlet flow angle α0 = 0° was considered in comparisons. The effect of the inlet flow angle of the chamber α0 on total pressure loss coefficient ζIC,rev was investi- gated for three constant spanwise distributions at three typical values of α0 = 0°, -7°, +7°. It is appar- ent that the differences are not significant. Figures 7 and 8 compare the contours of the constant flow veloc- ity of the inlet chamber symmetry plane µ during the normal and reverse flow. The computed flow patterns Figure 9. 3D flow streamlines in the half of fan inlet chamber and the inlet piping during the reverse flow (ν = 0.55). ν A00 A0 ζIC,rev α0 [°] ζIC,rev ζIC 0.45 1.90 0.246 0 3.5 0.55 2.16 0.467 0 6.8 0.500 0 7.2 0.63 2.5 0.452 −7 0.492 7 Table 2. Pressure losses in the inlet chamber. are valid for the hub/tip ratio of ν = 0.55 and a flow angle of α0 = 0°. The Reynolds number Re = 124 200 (Re = w1 h ν ) was for the normal flow in our fan; h is the annulus height, h = 0.5(Dt −Dh). The inlet flow conditions refer to the design point of fan stage A (φn = 0.40). The flow is separated on the hub and the outer annular surface of the inlet chamber due to relatively high flow diffusion. The area ratio A00 A0 is 2.16. The vortex structure of the separated flow is shed and trans- ported in the direction of the flow on spiral streamlines as shown in Figure 9 (α = 0°). Here, the complex 3D flow in one half of the inlet chamber is displayed. The streamlines’ colour is derived from flow veloc- ity values. In the outlet plane 00 with a rectangular shape, there are two counter-rotating vortices. In their centre (core), the streamlines with a relatively low velocity (V < 9 m s−1) are concentrated. Figure 9 shows only one vortex due to the flow symmetry to the vertical meridional plane µ. The division of the vortex into two branches takes place at the chamber bottom. 403 Václav Cyrus, Jan Cyrus Acta Polytechnica The relatively high total pressure losses of the inlet chamber were evaluated, as can be found in Table 2. The flow energy losses in turbomachinery chambers and diffusers are often divided into friction and mix- ing losses. In our case of the reverse fan operation, the total pressure losses were mainly caused by flow mixing. This phenomenon occurs after the sudden enlargement of the flow area in the inlet chamber during flow reversal. This is a standard example of thermodynamically irreversible process. The conser- vation of mass and momentum allows the rise in static pressure and the fall in total pressure. The total pres- sure friction losses are apparently lower, as can be seen from Figures 8 and 9. The flow around chamber surfaces often exhibits separation. However, during a normal fan operation, the flow is without significant separation on channel surfaces (Figure 7). The mixing energy losses are decreased compared to the case of the fan during flow reversal. 6. The effect of the inlet chamber and the diffuser pressure losses on the overall fan aerodynamic performance during the flow reversal The complete aerodynamic performance of the axial flow fan is calculated using a one-dimensional flow model. The results, shown in the previous paragraph, relating to the inlet chamber and the diffuser total pressure losses [15], [14] are applied. Our objective is to determine the relationship between the pressure coefficient ψF and the flow coefficient φ during the standard and reverse operation for two fans with the stage blading A and B. This is required to determine the flow rates of the fan operating points on the resis- tance curve of the connected piping system. The same resistance curve is assumed for both flow directions. The following equation is derived by using the ax- ial flow fan total flow pressure increment ∆pT,F , the stage total pressure increment ∆pT,ST, and the to- tal pressure losses in the fan inlet ∆pT,IC and outlet ∆pT,D fan sections: ∆pT,F = ∆pT,ST − (∆pT,IC + ∆pT,D). (8) By using the definitions of coefficients φ, ψ, and η: (1), (2) and (3), respectively, along with Equa- tion (8), the following relationship can be obtained: ψF ψST = ηF ηST = 1 − (ζIC + ζD) φ 2 ψST . (9) This equation shows that the fan pressure coeffi- cient ψF and efficiency ηF decrease due to the energy losses in the inlet chamber and the diffuser (ζIC + ζD) with the increasing flow coefficient φ and the decreas- ing pressure coefficient ψST. The equation reflects the normal and reverse flow. Definitions of loss coefficients are as follows: ζD = pT 3 − pT 4 q1 , ζD,rev = ζcon = pT 4 − pT 3 q1 , q1 = q0 = q3 = 0.5ρ(Q A )2. (10) It should be added that the axial fan diffuser acts as a confusor during the reverse flow. The performance of the stage blading (IGV+R+S) during the flow reversal cannot be reliably determined using CFD results due to the origin of large regions of separated flow on the blades’ surfaces. This applies to our fan stages A and B. Therefore, we rely on the stage test performance. As mentioned earlier in this paper, the comparison of the dependences of the pressure coefficient ψ and the efficiency η on the flow coefficient φ measured in stage A for normal and reverse flow runs are shown in Figures 4 and 5. The resistance curve shown in the mentioned Figure 4 intersects the performance curve valid for the normal flow at the point with the coordinates: φn = 0.415, ψST,n = 0.31. During the flow reversal, we get the point with the following coordinates: φrev = 0.318, ψST,rev = 0.19. Then, the flow coefficients’ ratio is φrev φn = 0.77. The pressure coefficient valid for the reverse flow in the complete fan ψF,rev is determined in accordance with the Equation (9). The following relationship is obtained after substitution: ψF,rev = ψST,rev − (ζIC,rev + ζcon)φ2 rev. (11) Using the data given in Table 2, namely ζIC,rev = 0.467 and in Idelchik’s work [16], namely ζcon = 0.04 in Equation (11), the reverse characteris- tic of the fan is determined (brown crosses). It crosses the resistance curve at the point R with coordinates: φF,rev = 0.27, ψF,rev = 0.139 (Figure 4). Again, the complete fan characteristics for the nor- mal flow were determined using Equation (9). Af- ter substituting the inlet chamber loss coefficient ζIC = 0.09 and the diffuser loss coefficient ζD = 0.05, φn = 0.41 and ψST,n = 0.31, we obtain ψF = 0.286. The diffuser loss coefficient was determined according to works of Japikse et al. [15] and Cyrus et al. [14]. The desired flow coefficient ratio of φrev,F φF = 0.68 was achieved based on the calculated results. The complete fan efficiency values can be found in Fig- ure 5: ηF = 0.891 (φ = 0.41) and ηF,rev = 0.295 (φrev = 0.27). For the fan stage blading B, which has a higher aerodynamic load, the ratio φrev φn is lower, i.e. 0.62; see Table 3. The flow coefficient ratio φrev,F φF valid for the complete fan B is only φrev,F φF = 0.55. The same calculation method was used as for the fan A, as described above. In order to increase the fan flow ratio φrev φn , it is appropriate to reduce the inlet chamber pressure losses 404 vol. 64 no. 5/2024 The effect of the inlet chamber and diffuser on the reverse . . . Stage Fan Stage (IGV+R+S) (IC+IGV+R+S+D) φrev φn φF,rev φn A 0.77 0.68 B 0.62 0.55 Table 3. Ratio of flow rates at reverse and standard operation for stage and fan (A and B). at flow reversal. The new design concept of the inlet chamber could consist of an extension of its axial length and a reduction of the sides of the rectangle ratio b a . 7. Conclusion This paper summarises the main findings of the aero- dynamic phenomena that occur during flow rever- sal, when the direction of rotation of the axial fan is changed and the stator vanes of the fan are rotated. Our objective was to determine the characteristics of the complete fan, consisting of the inlet chamber, inlet guide vanes, rotor and stator blade rows, and a diffuser, during normal and reverse operation. We also needed to obtain the inlet chamber pressure loss data from the flow simulation results for typical fan hub/tip ratios. The aerodynamic performance of two selected axial flow fans with stages A and B subjected to different aerodynamic loads was then calculated using a one-dimensional flow model. The experimen- tal aerodynamic performance of both stages was used in our study due to an inaccurate CFD performance prediction at flow reversal. The main results of our study are as follows: (1.) Tests have shown that stages A and B can be used for the flow reversal as the minimum ratio of flow rates is φrev φST = 0.77 and 0.62, respectively. (2.) The inlet chamber pressure losses increase at flow reversal was investigated for three hub/tip ratio values : ν = 0.45, 0.55, and 0.63 on the basis of CFD data. As the hub/tip ratio was increased the area ratio increased in the range of A00 A0 = 1.9 to 2.5. This tendency causes the growth of the separated flow areas on the surfaces of the inlet chamber flow channels. At the lowest value of ratio ν = 0.45, the loss coefficient ζIC,rev is significantly lower ζIC,rev = 0.265 than in the cases of hub/tip ratios ν = 0.55 and 0.63: ζIC,rev = 0.450 and 0.501, respectively. This can be explained by lower flow diffusion in the inlet chamber at flow reversal. (3.) Prediction of the complete axial fan performance is carried out with the use of one-dimensional flow calculation. The inlet chamber and the outlet dif- fuser pressure losses have a significant impact on the complete reverse fan performance. For the fan A, for example, the fan flow coefficient ratio is re- duced from φST,rev φST,n = 0.77 valid for the stage to φF,rev φF,n = 0.68 for the complete fan. In terms of efficiency, there will be a drop from ηST,rev = 0.420 to ηF,rev = 0.295. In the case of the fan with the stage B, the ratio is lower φF,rev φF,n = 0.55. These obtained results meet customer requirements of φF,rev φF,n > 0.50–0.60. (4.) The results of our study will allow the fan designer to assess the effect of individual components of the complete fan on its aerodynamic performance dur- ing the standard and reverse fan operation. A fur- ther refinement of the calculation method is pos- sible by using the test data of the complete fan model. List of symbols A Flow area [m2] c Blade chord [m] c Absolute flow velocity [m s−1] D Diameter [m] DF Diffusion factor DF = 1 − cos α1 cos α2 + 0.5 s c cosα1(tanα1 − tanα2) h Height [m] i Incidence angle [°] Ma Mach number Mk Torque [Nm] P Fan input [W] p Pressure [Pa] Q Volume flow rate [m3 s−1] r Radius [m] s Blade spacing [m] u Peripheral velocity [m s−1] w Flow velocity [m s−1] z Blades number α Absolute flow angle [°] β Relative flow angle [°] βL Blade angle [°] γ Blade stagger angle [°] ζ Loss coefficient η Efficiency υ Camber angle [°] µ Plane of symmetry ν Hub/tip ratio ρ Density [kg m−3] φ Flow coefficient ψ Pressure coefficient ω Angular frequency [s−1] Indexes D Design D Diffuser F Fan h Hub IC Inlet casing IGV Inlet guide vanes m Mean n Normal 405 Václav Cyrus, Jan Cyrus Acta Polytechnica rev Reverse R Rotor S Stator ST Stage t Tip T Total 1 Inlet 2 Outlet 0, 1, 2, 3 Planes of stage ()′ Reverse Acknowledgements Our study was carried out for ZVVZ Machinery and was supported by a grant from the Czech Technology Agency (TACR, TA04020228) and a grant from the Czech Ministry of Trade and Industry (TIP FR-TI1/347). The authors were pleased to receive these grants and would like to express their gratitude. References [1] I. J. Brusilovskij. Aerodinamiceskij rascot osevych ventilatorov [In Russian; Aerodynamic design of axial fans]. Mashinostrojenije, Moscow, Russia, 1986. [2] R. A. Wallis. Axial flow fans & ducts. John Wiley & Sons, New York, USA, 1983. [3] M. F. Dunn, F. S. Kendorski, S. Bhattacharya, et al. Reverse performance characteristics of main mine fans. Mining Science and Technology 1(1):59–68, 1983. https://doi.org/10.1016/S0167-9031(83)90132-9 [4] V. Cyrus. Axial fan at reversal flow. In Proceedings of the American Society of Mechanical Engineers Turbo Expo 2004, vol. 5, pp. 437–446. 2004. https://doi.org/10.1115/GT2004-53446 [5] V. Cyrus, J. Pelnar, J. Cyrus. Reversing of axial flow fans for ventilation. In Proceedings of the American Society of Mechanical Engineers Turbo Expo 2011, vol. 4, pp. 471–482. 2011. https://doi.org/10.1115/GT2011-46062 [6] V. Cyrus, J. Cyrus. Design study to axial flow fan stator blade row used in reverse ventilation. In Proceedings of International Conference FAN 2018, p. 30. 2018. [7] M. Abdolmaleki, H. Afshin, B. Farhanieh. Performance analysis of elliptic-profile airfoil cascade for designing reversible axial flow fans. AIAA Journal 57(4):1492– 1501, 2019. https://doi.org/10.2514/1.J057843 [8] M. J. Krasyuk, P. V. Kosygh. Reversivnyj osevoj ventiljator, aerodinamiceskaja schema ventiljatora, sposob reversirovanija, vrascajuscijsa napravljajuscij apparat [In Russian; Aerodynamic design of axial fan with guide vane for reverse air flow]. Fyziko-techniceskije problemy razrabotky poleznych iskopaemych 59(3):85–95, 2023. https://doi.org/10.15372/FTPRPI20230309 [9] T. S. Williams, C. A. Hall. Reverse thrust aerodynamics of variable pitch fans. Journal of Turbomachinery 141(8):081008, 2019. https://doi.org/10.1115/1.4043139 [10] T. S. Williams, C. A. Hall, M. Wilson. Variable pitch fan aerodynamic design for reverse thrust operation. Journal of the Global Power and Propulsion Society 7:127–139, 2023. https://doi.org/10.33737/jgpps/160096 [11] Numeca International. FINETM Turbo Numeca code, 2017. [12] N. A. Cumpsty. Compressor aerodynamics. Krieger publishing Company, Florida, USA, 2004. [13] S. Lieblein. The Aerodynamic Design of Axial Flow Compressor, chap. VI. Experimental flow in two-dimensional cascades. NASA, 1966. Reprinted NASA SP 36 in 1965, originally NACA RME 56B03. [14] V. Cyrus, J. Cyrus, P. Wurst, P. Panek. Aerodynamic performance of advanced axial flow fan for power industry within its operational range. In Proceedings of the American Society of Mechanical Engineers Turbo Expo 2014, vol. 1A, p. V01AT10A006. 2014. https://doi.org/10.1115/GT2014-25339 [15] D. Japikse, N. C. Baines. Diffuser design technology. Concepts ETI, Vermont, USA, 1998. [16] I. E. Idelchik. Handbook of hydraulic resistance. Jaico Publishing House, Delhi, India, 3rd edn., 2003. 406 https://doi.org/10.1016/S0167-9031(83)90132-9 https://doi.org/10.1115/GT2004-53446 https://doi.org/10.1115/GT2011-46062 https://doi.org/10.2514/1.J057843 https://doi.org/10.15372/FTPRPI20230309 https://doi.org/10.1115/1.4043139 https://doi.org/10.33737/jgpps/160096 https://doi.org/10.1115/GT2014-25339 Acta Polytechnica 64(5):398–406, 2024 1 Introduction 2 Axial flow stages for reverse fans 3 Flow description in the fan stage blades during reverse operation 4 Performance of fan stages A and B (IGV+R+S) during standard and reverse operation 5 Inlet chamber pressure losses during the normal and reverse flow 5.1 CFD simulations of the inlet chamber 5.2 CFD results and their analysis 6 The effect of the inlet chamber and the diffuser pressure losses on the overall fan aerodynamic performance during the flow reversal 7 Conclusion List of symbols Acknowledgements References