NOMENCLATURE
- AIP
-
Aerodynamic Interface Plane
- CDI, RDI
-
circumferential and radial distortion intensities
- D
-
AIP diameter
- DC(θ), SC(θ)
-
total pressure and swirl distortion coefficients at sector θ
- L
-
intake total length
- P t, i
-
total pressure at ring i
- q
-
dynamic pressure,
$\frac{1}{2}\rho V_a^2$
- V a , V c
-
axial and circumferential velocities on AIP
- θ, θ0
-
sector central and initial angles on AIP
- π i
-
intake total pressure ratio
1.0 INTRODUCTION
The growth in civil air transportation demand forces major hubs in the air transport network to operate beyond their capacities. The high air traffic density is a critical motive for research on future aircraft configurations and technologies. A more sustainable solution can be achieved when smaller airports within short-to-medium range are introduced into the network, instead of increasing only the capacity of hubs. In this case, such small airports require new aircraft designs to operate on short runways and to have lower noise emissions considering the proximity to population centres. In this regard, the Coordinated Research Center 880 (SFB 880) is investigating the design guidelines for future civil transport aircraft to benefit from small airport networks distributed in Europe. The project collaborators consist of Technical University of Braunschweig, University of Hannover and German Aerospace Research Center (Deutsches Zentrum für Luft- und Raumfahrt – DLR), funded by German Research Foundation (Deutsche Forschungsgemeinschaft – DFG).
The aircraft in the first research period of SFB 880 is designed as a 100 passenger single-aisle civil commuter for short range (within 2, 000 km with maximum payload)(Reference Radespiel and Heinze1). It is equipped with two turboprops and has flight mission including cruise with Ma = 0.74 at a 10.6 km altitude. In order to maintain the short take-off and landing (within 800 m), the aircraft design of the SFB 880 project uses boundary-layer control with active high-lift. Air mass flow is supplied by electric-powered small compressors embedded in the wing for suction and blowing to the Coanda flaps. The power is supplied by the turboprop engines, with the maximum shaft power estimated at ~10 MW each. Under these design specifications of the SFB 880, the focus of the Institute of Jet Propulsion and Turbomachinery (IFAS) at TU Braunschweig is on the turboprop intake design and the propulsion system integration considering the engine-airframe interaction effects.
Turboprop intakes have complex internal flow due to their default S-shaped ducts with large curvatures and covering around the propeller shaft bore. Flow in the S-duct diffuser develops secondary flows due to the pressure gradients generated at the first bend(Reference Seddon and Goldsmith2,Reference Ho3) . This flow retardation of stream-wise vorticity does not completely vanish when it reaches the second bend, but is sustained due to the radial vorticity(Reference Taylor, Whitelaw and Yianneskis4). After the second bend, the secondary flows evolve into counter-rotating vortices in the boundary layer, which are increasing the viscous loss on the outlet(Reference Britchford5). In addition to the inherent flow complexity of S-ducts, turboprop intakes have to consider the effect of the propeller downstream, which generates the boundary layer on the hub along with the favourable total pressure increase. Hancock et al(Reference Hancock, Lyman and Pennock6) reported the advantage of single scoop in terms of recovery and their simplicity to avoid hub boundary-layer effects by introducing diverter height. The same study showed that annular intakes suffer from significant loss when the propeller effect is taken into account.
Turboprop engines deliver the generated power from the turbine component to the propeller through a shaft and a gearbox. Conventional engine configuration forces the shaft of the forward-mounted propeller to pass through the S-duct intake. This connection introduces an additional wall around the shaft inside the S-duct, increasing the boundary-layer formation and adding to loss. For single scoop offset-line intakes, Little and Trimboli(Reference Little and Trimboli7) observed that the wrap-around type shaft connection provides higher pressure recovery than the plain shaft penetration type.
The influence of the propeller and nacelle is also crucial for the turboprop intake design. The effect of propellers at high subsonic cruise on the wing aerodynamics was investigated by Welge et al(Reference Welge, Neuhart and Dahlin8), showing the start of the separation by the leading edge. They proposed to install the nacelle over the wing in order to reduce the adverse effects. Secondary flow vorticies were observed by Roosenboom(Reference Roosenboom, Heider and Schröder9) in the propeller slipstream. In contrast, it was also shown by Witkowski(Reference Witkowski, Lee and Sullivan10) that the propeller downstream has improving effects on the wing lift. For the investigation of interaction effects on turboprop intakes, Sokhey(Reference Sokhey11) provided valuable results pointing out the main design parameters and important factors as the external drag and the propeller downstream flow.
Test data and design methodologies of subsonic intakes available in the literature were used for preliminary design choices of SFB 880’s turboprop S-duct. In addition to the primary design objective of uniform air mass flow delivery to the engine compressor, the intakes of the SFB 880 aircraft have to consider the operation with large core air mass flow for their high power demand. Although the S-duct intake integration of the SFB 880 can benefit from the results available in the literature, it is necessary to perform detailed aerodynamic analysis considering the high power rate and the unconventional engine installation. Müller et al(Reference Müller, Heinze, Kozŭlovic, Hepperle and Radespiel12) within SFB 880 investigated the channel wing configuration, observing that it can provide a good trade-off between aerodynamic performance, flight mechanics and noise levels in the expense of propulsive efficiency. On the other hand, their focus within the SFB 880 project was only on the wing aerodynamics and the interaction effects of the propeller from the perspective of the aircraft but without an integrated intake in the nacelle.
The objective of this study is to analyse the airframe and the propeller effects on the turboprop S-duct intake in order to observe performance with different nacelle installation types, especially on a channel wing. S-duct intakes of previous investigations of SFB 880 were integrated into a representative nacelle model and were analysed using computational fluid dynamics. The results were compared to a conventional tractor wing and over-the-wing without channel configurations. Two different types of S-duct intakes, as shaft penetration and wrap-around, were included in the comparative study. An actuator disc model was used to simulate the propeller downstream effects. The simulations were done at cruise and take-off conditions of the SFB 880 aircraft. The intake design limits were investigated by the aerodynamic performance on the Aerodynamic Interface Plane (AIP).
2.0 ANALYSIS PARAMETERS
Averaging operator in a steady state simulation analysis was weighted by the mass flow rate but it is mentioned in this paper only as the average for simplicity. Total pressure recovery was considered as the main intake performance parameter. The outlet was assumed on the AIP, excluding the gap losses between the outlet and the engine compressor face from the scope of this study. The recovery was evaluated with the ratio of average total pressure at AIP to the average total pressure at actuator disc downstream as π i in Table 1. Total pressure distortion (DC(θ)) was evaluated as defined by Seddon(Reference Seddon and Goldsmith2). This coefficient compares the average total pressure over the AIP with the average total pressure in the low pressure sector θ, normalised by the average dynamic pressure of the AIP. The swirl distortion coefficient SC(θ) was defined by the ratio of average circumferential velocity in the sector θ to the average axial velocity through AIP. Distortion coefficients were defined in the θ = 60° sector (starting from θ0 = 30°) for this study, as shown in Fig. 1. AIP was divided into five rings and each into 20 equal sectors to analyse the intensity distribution over the complete AIP (Fig. 1) for the circumferential and radial distortion intensities (CDI and RDI)(13). Data reduction was applied by averaging the CDI values and by choosing the ring maximum amongst the RDI values.
θ sector and distortion intensity rings on AIP.
Analysis parameters at AIP; averaging operators are weighted by mass flow rate
3.0 MODEL SET-UP
A representative model of single scoop, low-pinion S-duct intake was used to define the main dimensions (Fig. 2). The model was simplified by assuming the outlet plane to be on the AIP. The preliminary engine design estimated the AIP diameter as D = 700 mm for the high-power turboprop considering its flight mission with the high subsonic cruise speed and the additional electrical power requirement of the high-lift system.
Intake dimensions of analysis.
Two S-duct intakes, differed by their connection to the AIP over the shaft, were used: reference shaft penetration type and wrap-around. The shaft penetration type S-duct intake was approximated by linearly extruded curves to form an intake area ratio of A AIP /A inlet = 1.19. The wrap-around type was modelled with four spline guidelines but 60% of its shaft wall was kept linear for comparability with the shaft penetration. The radius of the first bend was kept as half of the second bend radius at r/D = 0.71 for both types. The length aspect ratio was set to L/D = 3, based on the results of the previous isolated intake analysis(Reference Atalayer, Friedrichs and Wulff14). The inlet width was kept equal to AIP diameter D with a highlight spread angle of α = 32.71°. Both were integrated into a representative nacelle body with L nacelle /D propeller = 1.6 considering the turboprop achieved by the preliminary engine design (Fig. 3). The intake's highlight proximity to propeller, and the boundary-layer diverter gap were set as t/D = 0.36 and h/D = 0.07, respectively, but their parametric analysis was excluded from the scope of this study. The highly loaded propeller is based on the design of a similar project(Reference Lenfers15), modified for the SFB 880 aircraft requirements to a diameter of D propeller = 5 m with its pitch angle to β3/4 = 57° at cruise. For this study, the propeller was modelled as an actuator disc using only a surface plane element.
Integrated and installed intake model.
The objective is to analyse the airframe and propeller influence on the intake with channel wing installation (Fig. 4(c)) in comparison to tractor (Fig. 4(a)) and over-the-wing (Fig. 4(b)). The integrated nacelle was installed on an unswept wing with a constant profile along its span (b = 5c) for all cases. DLR F15 transonic aerofoil with 12.6% maximum thickness was used for the wing (c ~ 3.7 m) in order to have comparable results with other SFB 880 studies(Reference Müller, Heinze, Kozŭlovic, Hepperle and Radespiel12).
Installation types for integrated S-duct intakes.
The actuator disc model was embedded into the channel wing by 1/6 of its diameter with a constant gap by g/D propeller = 0.02. The same height was applied on the over-the-wing. The nacelle was positioned at X n /D propeller = 0.57 upstream of the wing’s leading edge for the tractor wing.
4.0 NUMERICAL SET-UP
4.1 Installed model grids
Unstructured hybrid grids, with hexahedron and prismatic elements on the bodies and tetrahedral elements in the fluid domain, were applied on the simulation models using Centaur grid generator software (Fig. 5). The simulation domain was modelled as a cylindrical volume with the width of the wing span and 20c in radius. The S-duct wall surface at the intake end was mapped with structured hexahedron elements in order to improve the resolution in the region close to the AIP. Thirty prismatic layer elements were generated on the wall boundaries with an initial height of 0.004 mm in order to achieve y + ~ 1. Sensitivity analysis against grid size was done with a wrap-around S-duct installed on the channel wing for 14 × 106 to 20 × 106 nodes. The analysis showed maximum difference in π i reaching 0.6% (Fig. 6). Considering the relative insensitivity and the simulation runtime, grids with ~16 × 106 nodes were applied for the investigation.
Grid on channel wing tip.
Grid size sensitivity.
4.2 Fluid model, actuator disc set-up, and grid
The steady-state simulations of Reynolds averaged Navier-Stokes (RANS) equations were run using DLR’s TAU solver. Fluid was modelled in the far-field domain for cruise conditions with freestream Ma = 0.6 at 10.6 km altitude (Re = 18 × 106), instead of both turboprop and its intake’s on-design cruise condition Ma = 0.74. This choice was necessary as TAU code shows limited accuracy in solving the actuator disc surface tip with local sonic regions. As seen in Fig. 7, the local shock on the wing covers most of the downstream starting from the tip and prevented the simulation to converge. The end of take-off acceleration was used as an off-design simulation case with freestream Ma = 0.172 at sea-level.
Meridional Mach distribution; wrap-around S-duct on channel wing, at Ma = 0.74; simulation ended prematurely before meeting convergence due to developed tip shock.
The AIP of the S-duct intake was given a static pressure boundary condition coupled with an initial estimate of mass flow rate, based on the simulation flight condition at the corresponding altitude. The circular base surfaces of the computational domain were defined by symmetry conditions on the wing tips. The curved surface of the cylindrical domain was defined by the farfield freestream condition of the flow.
DLR’s TAU solver is able to handle the actuator disc assumption for a propeller on a surface domain without thickness. The surface boundary conditions for the actuator disc require pitch angle and lift-drag polar distributions along the radial direction as it is coupled with the blade element theory. Using these inputs, TAU can calculate the thrust and torque distributions at a definite revolution per minute with a steady-state slipstream. As it was given in Fig. 8, the propeller slipstream impacted an upwash to the section lift on the straight wing when the nacelle was installed at tractor position. The downwash effect was low on the section lift distribution as the actuator disc was at Y n /D prop = −0.24. The channel wing showed an increase within the span of the channel, which was amplified with the actuator disc–nacelle installation.
Section lift distribution on wing with DLR F15 profile; cruise condition.
A numerical solution for the inviscid fluxes was set up using second order central difference discretisation with artificial dissipation. The mean flow of the inviscid flux was evaluated by the central average on cell faces. The second order Roe upwind scheme was used to discretise the turbulent components of the inviscid flux, while viscous flux components were evaluated with central scheme in full gradients. The Gauss divergence theorem was used to reconstruct the gradients in order to increase the accuracy of the upwind scheme. Time stepping was regulated by CFL = 1.5 with the implicit backward Euler scheme for relaxation using a lower-upper symmetric Gauss-Seidel (LU-SGS) algorithm.
The geometry set-up of this study shared similar wings, installation positions, propeller type and spinner cone with the other studies of the SFB 880 project related to wing aerodynamics(Reference Müller, Heinze, Kozŭlovic, Hepperle and Radespiel12), in which the one-equation turbulence model Spalart-Allmaras was used for RANS simulations. On the other hand, the two-equation turbulence model k − ω shear stress transport was used for the uninstalled, isolated turboprop S-ducts in the previous analysis phase(Reference Atalayer, Friedrichs and Wulff16). The choice of a two-equation turbulence model is common practice for most of the investigations on S-ducts found in the literature(Reference Majić, Efraimsson and O’Reilly17,Reference Ruiz-Calavera, Funes-Sebastian and Perdones-Diaz18) , but Reynolds stress transport models are also advocated with strong arguments(Reference Walker, Barker, Carrotte, Bolger and Green19). However, Colin et al(Reference Colin, Aupoix, Boussugea and Chanez20) reported that computations on intakes with the Spalart-Allmaras model were capturing the flow separation similar to the two-equation models at high mass flow rates.
Comparative investigation was done with simulations using Spalart-Allmaras, k − ω (Wilcox(Reference Wilcox21) and Menter(Reference Menter, Kuntz and Langtry22) shear stress transport), and Reynolds stress transport models for the integrated S-duct intake investigation. A modification was applied on the one-equation model as negative Spalart-Allmaras(Reference Allmaras, Johnson and Spalart23) to avoid non-physical negative vorticity in low resolution regions of the grid. The comparison was done on the wrap-around type S-duct intake installed on the channel wing at cruise condition.
The results of the simulation with the Wilcox k − ω model showed only 0.3% higher (mass flow weighted) total pressure recovery than with Spalart-Allmaras, whereas the result with Menter’s shear stress transport model was 1.3% higher. The simulation with Reynolds stress transport model failed before reaching the convergence but produced total pressure recovery levels similar to the one with the Wilcox k − ω model. When the swirl coefficients SC(60°) were compared, the two-equation models predicted 30% lower than the Spalart-Allmaras. As the total pressure distributions on AIP were compared in Fig. 9, the average total pressure in the θ = 60° sector was observed lower for the Menter’s shear stress transport model. On the contrary, its total pressure distribution on the rest of the AIP was higher and more uniform than the Spalart-Allmaras results (Fig. 9(b)). It can be suggested that the secondary flow effects due to bending were more reasonably captured by the one-equation model, while acknowledging this not as a conclusive argument without an experimental validation. Considering the literature and the comparative investigation, the negative Spalart-Allmaras model was chosen for the integrated S-duct analysis of this paper.
Comparison of turbulence models by AIP flow; wrap-around S-duct installed on channel wing; cruise condition.
5.0 RESULTS
The simulations were done with three installation types: conventional tractor wing (TW), over-the-wing (OW) and channel wing (ChW). The comparison was done for two S-duct intakes differed by their shaft connection: reference shaft penetration type S-duct and wrap-around with same geometry. In addition, the simulation results of uninstalled, isolated intakes from the previous numerical investigation(Reference Atalayer, Friedrichs and Wulff16) were used for identifying the interaction effects.
5.1 Interaction effects on total pressure recovery
The intake performance results show that the lowest total pressure recovery occurred with the tractor wing type installation (Fig. 10(a)) at cruise condition. Channel wing installations experience 1.7% higher π i in average, while over-the-wing types experience 1.6% more than the tractor wing. Both over wing configurations achieved recovery levels remarkably close to their isolated performance at cruise. The wrap-around S-duct on channel wing hints even an increase over its isolated case, which requires further verification. The shaft penetration S-duct in the tractor wing configuration loses 2% π i compared to the isolated case, whereas this loss is only 1.4% for the wrap-around type. The performance improvement with the channel wing against other installations was expected, since it serves as a partial tip duct for the actuator disc. On the other hand, this advantage is observed only for the wrap-around type intake on the channel wing. When propeller interaction is considered by calculating the π i for a control volume including the farfield total pressure (Fig. 10(b)), increase in recovery is observed for both S-ducts. S-duct type comparison reveals 1.2% higher recovery in average for wrap-around intakes than shaft penetration at cruise. A sharp drop in π i is observed for all installations at the take-off condition but the wrap-around S-duct on the channel wing remains the highest (Fig. 10(a)). High mass flow rate at take-off contributes to the low performance of the S-duct in the tractor wing configuration, which becomes an advantage when the π i with farfield total pressure is used (Fig. 10(b)).
Performance comparison by installation; ChW: channel wing, OW: over-the-wing, TW: tractor wing.
The drop in total pressure recovery at the take-off condition was expected as the intakes remained in the downstream hub influence due to the radial load distribution on the actuator disc (or propeller). The large gradients caused by the hub and the captured flow shown in Figs 11(a) and 11(b) support this behaviour. Adding to the thick boundary layer, the recovery of the intakes at take-off remained significantly lower than at cruise. The total pressure difference between the downstream and the intake flow is slightly greater for the tractor wing installation than the channel wing. Therefore, the intake on the channel wing had a higher recovery, even though the total pressure distribution along the S-ducts remained similar (Figs 11(c) and 11(d)).
Meridional and axial total pressure distribution of wrap-around S-duct; take-off condition.
The actuator disc maintained most of its propulsive efficiency at cruise condition when installed as tractor wing (Fig. 12(b)). The channel wing affected the actuator disc flow by reducing the total pressure in its downstream (Fig. 12(a)). This reduction allowed the S-duct on channel wing to capture flow with higher total pressure but in larger gradients than the one on tractor wing (Figs 12(c) and 12(d)). The captured “high total pressure pockets” maintain their structure through the first bend of the S-duct when installed on the channel wing (Fig. 12(c)). On the contrary, the uniform flow captured by the S-duct on the tractor wing was overwhelmed by the bending losses, especially after the second bend at x/D = 0.21 (Fig. 12(d)).
Meridional and axial total pressure distribution of wrap-around S-duct; cruise condition.
5.2 Interaction effects on total pressure distortion
The simulation results show 15% and 24% lower total pressure distortion (DC(60°)) in average for the channel wing and over-the-wing configurations, respectively, when compared to the tractor wing (Fig. 13(a)) at cruise condition. Recalling Fig. 12(d), the increased loss due to S-duct bending concentrated in the θ = 60° sector for the tractor wing installation. It is also noticeable that the over-the-wing of wrap-around shows the lowest DC(60°) at cruise but this effect diminishes at the take-off condition. The difference of DC(60°) between the intake types is smaller at take-off than at cruise but the channel wing shows an average of 7% lower than the tractor wing, whereas for over-the-wing, this improvement remains at 5%, indicating the preferable performance of channel wing for both flight conditions. Wrap-around type intake reassures its advantage by showing an average 40% drop in DC(60°) against shaft penetration type, even though the difference with their respective isolated cases is comparably similar.
Distortion comparison by installation; ChW: channel wing, OW: over-the-wing, TW: tractor wing.
Average distortion intensity along AIP circumference results in Fig. 13(b) show that installed wrap-around S-ducts have approximately 31% drop in CDI ave compared to its shaft penetration counter-parts. When compared to their respective isolated cases at cruise, the difference observed in CDI ave is similar for both. Channel wing installations have 2% lower CDI ave , whereas over-the-wing has 3% more than the tractor wing configuration at cruise. This can be explained by the flow with high total pressure along the wrap-around S-duct reaching the AIP in Fig. 12(c). Take-off condition increases the CDI ave of only wrap-around intakes; however, they still remain lower than the shaft penetration type. The drop in CDI ave of both channel and over-the-wing when compared to the tractor wing becomes relatively similar at take-off.
Considering the distortion intensity in AIP radial direction, an increase by 2/3 is observed with wrap-around intakes, as opposed to double the shaft penetration’s RDI max when compared to their respective isolated cases (Fig. 13(c)). Comparison of installations shows that the channel wing produces higher RDI max . At this point, the data reduction technique for RDI is questioned, as comparing the maximum values renders the actual change in radial distortion. Fig. 13(d) shows the comparison of RDI on AIP rings of wrap-around and shaft penetration intakes on channel wing at cruise condition. RDI is actually amplified by 2/3 only at the tip of wrap-around, whereas it is doubled for shaft penetration on the hub. It is obvious that RDI max shifts from hub to tip when shaft penetration is replaced with the wrap-around S-duct, instead of due to the change in installation type. This shift suggests that the shaft wall friction effects of the former type are prevented. However, the tip wall of the wrap-around type still needs improvement to reduce its total pressure difference related distortion. This was observed on the total pressure distribution along the S-ducts, given in Figs 12(c) and 12(d). The largest gradients occurred at the tip walls of the wrap-around at both channel and tractor wing installations, while boundary-layer thickening after the second bend introduced large viscous loss at the lower half of the AIP.
5.3 Interaction effects on swirl
Results in Fig. 14(a) show that the channel wing installation has slightly increased swirl for both S-ducts when compared to tractor wing. Shaft penetration S-ducts have less swirl distortion at both cruise and take-off conditions. SC(60°) comparison shows that the wrap-around intake installed over-the-wing has the highest swirl, whereas smaller increases are observed with channel and tractor wing installations over their isolated cases. The unfavourable swirl distortion observed with wrap-around S-ducts was already reasoned due to the limited definition of SC(60°) parameter in the previous study(Reference Atalayer, Friedrichs and Wulff16). In order to further the swirl analysis, tangential velocity components of the two S-duct types for the channel wing installation were projected on their respective AIP. Uniform wrapping swirl patterns are observed for the wrap-around type S-duct intake (Fig. 14(b)), separated by offset swirls, as classified by Sheoran(Reference Sheoran, Bouldin and Krishnan24). The shaft penetration S-duct shows an obvious rotation of its paired swirls with large offsets, located over the analysis sector (Fig. 14(c)). It became apparent that the analysis sector of θ = 60° was insufficient to quantify the comparison. In this regard, the AIP was divided into, again, five rings and the θ = 60° analysis sector was pivoted with 10° increments. SC(60°) was calculated at each of these sectors over the complete AIP circumference.
(a) Swirl coefficient comparison by installation; ChW: channel wing, OW: over-the-wing, TW: tractor wing; (b) and (c) Tangential velocity vectors on AIP; cruise condition; channel wing.
The distributions of SC(60°) over the hub, and the tip rings of AIP (Fig. 15) compare the channel wing installation effect to the tractor wing for the two S-duct types. The wrap-around on channel wing at cruise condition (Fig. 15(a)) is showing its peak in the anti-clockwise rotation (positive) on its tip ring in the same sector as in Fig. 1 (sector initial angle θ0 = 30°). The tip also experiences the clockwise peak with the initial angle at θ0 = 90°. Its hub shows basically the same pattern but with lower amplitude. When it is installed on the tractor wing (Fig. 15(b)), only the clockwise peak (at θ0 = 90°) experiences a drop; otherwise, it keeps its distribution relatively the same.
Swirl coefficient distribution on AIP circumference; W-A: wrap-around, SP: shaft penetration.
The shaft penetration has significantly different and opposite distributions. On both installation types, its anti-clockwise peak remains at θ0 = 90° of its hub, whereas its clockwise peak is at its tip ring of the same sector. It is important to note that channel wing installation reduces the difference between its peaks by 14% compared to the tractor wing. It can be argued that replacing the nacelle on the channel wing helps only on the already worse-performing shaft penetration, whereas changing the S-duct to wrap-around effectively limits the swirl to the AIP tip and introduces a phase difference between the counter-rotating swirls.
6.0 CONCLUSIONS AND FUTURE WORK
Interaction effects on turboprop S-duct intake were investigated in terms of recovery and distortion by applying numerical flow simulations. The propeller was modelled as an actuator disc for steady-state RANS simulations. Three installation types were used as conventional tractor wing, over-the-wing and channel wing for the comparison of two different S-duct intakes at cruise and take-off conditions.
Investigation on installation types showed that channel wing installations experienced approximately 2% higher total pressure recovery than the tractor wing configurations, accompanied by lower DC(60°) and CDI ave . It can be argued that positioning the nacelle over the wing assists the intake recovery; embedding the nacelle in a channel wing further amplifies this effect. On the other hand, high RDI max levels of the channel wing installations indicated that the channel depth should be adjusted for preferable values. Wrap-around S-ducts also produced higher RDI max . This effect was analysed by distribution over AIP rings, revealing the change of maximum RDI from hub to tip for wrap-around. It was also observed that the adverse effect on RDI max due to installation remained lower for wrap-around than the shaft penetration type.
Investigation by swirl showed that SC(60°) remained high for S-duct intakes on the channel wing. SC(60°) was argued to be insufficient to quantify the complete swirl over the AIP, especially when the tangential velocity vector distribution was observed. When SC(60°) was analysed for the complete AIP, several peaks were observed. When compared to the tractor wing, the channel wing showed reduced peak difference between the two counter-rotating swirls for the shaft penetration S-duct, whereas the wrap-around channel wing experienced increase with a 90° phase shift. Shaft penetration S-ducts had their anti-clockwise rotating peak at the hub and clockwise at the tip, whereas both occurred on the tip ring for wrap-around.
The depth of the actuator disc (or propeller) embedded into the wing remains as a future parameter. The assumption of AIP on the outlet excludes the friction loss on the flow that reaches the inlet guide vane of the engine compressor. The AIP boundary condition was assumed to be uniform static pressure distribution in this study. As the compressor would have an effect on the AIP flow, this parameter needs further investigation in the future phases. The results also require verification with unsteady simulations using an actual propeller model and certainly the experimental validation for the final design of the S-duct intake design for the highly loaded turboprop.
ACKNOWLEDGEMENTS
This study is part of the Collaborative Research Center 880 (SFB 880), which is funded by the German Research Foundation (Deutsche Forschungsgemeinschaft – DFG). All simulations were run using the computational resources of North-German Supercomputing Alliance (HLRN) in Hannover and Berlin.
