Integrated CFD - Acoustic Computational Approach to the Simulation of a Contra Rotating Open Rotor at Angle of Attack

19th AIAA/CEAS Aeroacoustics Conference May 27-29, 2013, Berlin, Germany AIAA 2013-2242 Integrated CFD - Acoustic Computational Approach to the Simulation of a Contra Rotating Open Rotor at Angle of Attack Piergiorgio Ferrante 1, Stéphane Vilmin 2 and Charles Hirsch 3 NUMECA International, Brussels, Belgium Jean Charles Bonaccorsi 4 NUMECA USA, Inc., San Francisco, CA 9410. Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Paolo di Francescantonio 5 STS - Scientific and Technical Software, Varese, Italy An innovative integrated computational approach, that includes mesh generation, CFD analysis and acoustic propagation, is presented and applied to the simulation of the F7/A7 Contra Rotating Open Rotor (CROR) tested in the NASA Lewis 9- x 15-ft Anechoic Wind Tunnel. The tonal noise source is predicted starting from the engine geometry and the CFD calculated flow properties. The simulation approach relies on the exploitation of an efficient CFD solver based on the Nonlinear-Harmonic (NLH) method. The far field noise radiation of the CROR configuration is evaluated by means of a free-field acoustic propagation tool based on the Ffowcs Williams-Hawkings (FW-H) formulation. The final comparison between the numerical results and the available measurements highlights the capability of the methodology to accurately predict the unsteady flow field and the radiated sound field in flight conditions under an angle of attack. The applicability of the simulation chain to complex calculations at affordable computational cost is also demonstrated. Nomenclature Aft = Aft AOA = Angle of attack BPF1 = Blade Passing Frequency of forward propeller BPF2 = Blade Passing Frequency of aft propeller CROR = Contra Rotating Open Rotor FAR = Federal Aviation Regulation Fwd = Forward FW-H = Ffowcs Williams-Hawkings ICAO = International Civil Aviation Organization M = Mach number NLH = Nonlinear Harmonic Ps = Static pressure RPM = Rotations per minute SPL = Sound Pressure Level Ts = Static temperature = Polar angle 1 Product Manager - Acoustics 2 Developer - CFD Group 3 President 4 Technical Director 5 Head of Software Development 1

I. Introduction Aircraft engine noise is one of the major contributors to noise emissions particularly at take-off and landing. A sound source of relevant interest is the propeller generated noise in Contra Rotating Open Rotors (CROR). In order to increase the community acceptance of aircraft transportation and ensure at the same time the satisfaction of the certification rules (FAR, ICAO) it is mandatory to develop low noise technologies and design solutions, together with the capability to optimize the power plant components adopting a multidisciplinary approach. This latter methodology requires the development and the application of CFD and CAA computational models that accurately predict the engine noise emissions in operative conditions. Predictive tools shall provide easy and fast computational grid generation, integration of sound propagation solvers with CFD codes, applicability to complex calculations at affordable computational cost. Even if several of these tools are available, their applicability is somehow limited by a series of factors that include not only the high computational cost of fully unsteady CFD and acoustic simulations at realistic Mach numbers, but also the intrinsic difficulties in properly modeling the appropriate acoustic boundary conditions in presence of flow. On CROR configurations the most common strategy for Propeller Noise estimation consists in a two-step approach, based on the exploitation of the Lighthill analogy: unsteady near-field aerodynamic flow simulation to evaluate the noise sources, coupled to a far-field acoustic propagation code [1]-[2]. In this paper an innovative integrated computational approach is applied to the simulation of the F7/A7 counterrotating propellers tested in the NASA Lewis 9- x 15-ft Anechoic Wind Tunnel [3]-[4]. The computational approach includes the automatic mesh generation of complex 3D geometries, the CFD analysis of the acoustic source domain and the computation of the sound propagation to the far-field. The approach is based on the exploitation of two main key ingredients: 1) an efficient CFD solver based on the Nonlinear Harmonic (NLH) method [5]-[6] capable to predict the unsteady flow field generated by the contra rotating blades of the CROR configuration considered, with a CPU reduction of the order of 1000, compared to full sliding grid techniques; 2) an efficient frequency domain aeroacoustic tool implementing the Ffowcs Williams-Hawkings (FW-H) formulation [7]. The paper starts with section two, illustrating the F7/A7 experimental setup and the relative aerodynamic/acoustic measurements available. Section three describes the simulation approach adopted for the computation of the acoustic sources and their propagation to the far-field. In section four the aerodynamic and the acoustic results are analyzed and compared with the experimental data available. In particular the sound level directivity is analyzed at two take-off flight conditions with and without the presence of flow angle of attack (AOA). The conclusive section provides an overall assessment of the capability of the integrated simulation tool to model successfully the whole computational chain for the application considered. II. Experimental Setup The aerodynamic and acoustic measurements available were performed in the NASA Lewis 9- x 15-ft Anechoic Wind Tunnel [3]. This test rig provides a flow speed up to Mach 0.2, suitable to reproduce take-off and approach test environment. The anechoic capability of the wind tunnel is guaranteed down to 250 Hz, ensuring to correctly measure the propellers fundamental frequencies. The acoustic measurements were performed with a polar microphone probe (Figure 1) over a side line located below the propeller (and the hypothetical aircraft), at a distance of 61 cm (24 in) from the propeller axis of rotation. The inspected polar angles ( ) range from 45 to 135 degrees, with the angles measured relatively to the center of the aft propeller disk, with respect to the rotation axis oriented in the forward direction. The rotation of the polar microphone probe around the propeller axis allowed azimuthal surveys at any side line locations. The propeller axis angle-of-attack was achieved by rotating the propeller in the horizontal plane with respect to the incident flow. A range of 16 deg was investigated. 2

Fig. 1 - Sketch of the propeller model and polar microphone probe (from ref. [3]) Two counter-rotating propellers were tested, respectively the F7/A7 and the F7/A3 configurations, the latter characterized by reduced diameter of the aft propeller. In this study only the F7/A7 configuration is considered. The design parameters are reported in Table 1. F7/A7 propeller Number of blades (fwd/aft propeller) 11 / 9 Design cruise Mach number 0.72 Nominal diameter, cm (in.) 62.2 (24.5) / 60.7 (23.9) Nominal design cruise tip speed, m/s (ft/s) 238 (780) Nominal design advance ratio 2.82 Hub-to-tip ratio 0.42 Geometric tip sweep, deg 34/31 Rotors spacing, cm (in.) 15.0 (5.9) Activity factor 150 / 150 Design power coefficient based on annulus area 4.16 A. Nonlinear Harmonic (NLH) Method Tab. 1 - Propeller design characteristics III. Simulation Setup In the NLH approach [5]-[6], the flow is decomposed into Fourier harmonics, typically associated with blade passing frequencies and their multiples, whose number is specified by the user. The unsteady flow transport equations are formulated in the frequency domain and result into the solving of transport equations for the harmonics and the mean flow. These equations are coupled. Indeed, like in the Reynolds- Averaged flow equations, time-mean products of the periodic perturbations appear in the time-mean flow conservation laws, called "deterministic stresses", representing the full nonlinear effects of the flow unsteadiness on the time-averaged flow. In this case, the closure of the model is obtained by calculating the deterministic stresses directly from the harmonic solutions, which is an important advantage as it provides an improved time averaged flow solution, incorporating all the effects of the unsteadiness. The NLH methodology has been successfully applied in the past to an integrated aero-engine including Fan, OGV (outlet guide vane) and nacelle intake in presence of ground effects and crosswind conditions [8], as well as it has been exploited on several CROR applications, in particular for a sensitivity study performed on a generic 8x8 3

puller CROR at typical take-off conditions [9]. More recently the methodology has been extended to fully account for installation effects of pylons/nacelles on the CROR aerodynamics and noise. Figure 2 shows the 11 x 9 CROR that has been used for this study. It should be noted that since the real geometry was not available it was decided to design an arbitrary geometry as much as possible equivalent to the tested one. In particular the radius and the RPM were kept the same, as well as the generated thrust. This required an iterative design process performed with FINE TM /Turbo. Obviously there is no warranty that the other parameters (chord, sweep) are the same as the real ones. Besides there is no guarantee that in the generated geometry the distance between the rotors is exactly the same as for the F7/A7 propellers. For the real configuration, in fact, this distance refers to the axial separation between the rotors pitch change axes, which cannot be clearly defined in the reconstructed geometry. Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Fig. 2 - Simulated CROR configuration The simulation is performed for a free stream Mach number of 0.2, which is representative of take-off/approach operation. Two angles of attack have been studied (0 degree and +8 degree) in order to assess the capability of the computational chain to reproduce the effects of the flow incidence on the blade aerodynamic loads and on the propagation of the sound generated. Table 2 shows the CROR characteristics that were kept unchanged during the design process, and so these characteristics are identical to the ones used during the experiments (F7/A7). Forward blade Aft blade Number of blades 11 9 Diameter (cm) 62.2 60.7 Rotational speed (RPM) -7633 7695 Tip Mach number 0.76 0.77 Tab. 2 - Simulated CROR characteristics The CFD mesh has been created with the software AutoGrid5 TM, the NUMECA International automatic hexahedral structured grid generator. The computational domain is radially extended to a radius of 0.75 m and is axially extended to a total length of 7.28 m as well. Figure 3 shows a meridional view of the mesh. The inlet bulb is meshed with a butterfly topology avoiding a singular line in the inlet section. Figure 4 shows a blade-to-blade view of the mesh at mid span. A O4H topology is used with a skin block around the blade and 4 H blocks around it. This blade-to-blade topology is extended in the far-field. Only one interblade channel per row is actually meshed to run a NLH simulation. The mesh respects standard quality criteria in terms of the maximum expansion ratio, the minimum orthogonality, and the maximum aspect ratio of cells. The mesh counts 7.6 million nodes. Figure 5 provides a view 4

of the CFD mesh on the solid surfaces. Fig. 3 - Meridional view of CFD mesh Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Fig. 4 - Blade to blade view of CFD mesh Fig. 5 - View of CFD mesh on solid surfaces Table 3 summarizes the numerical parameters that have been employed for the reference CFD computation. A simultaneous time-marching technique is used to converge to a steady-state solution of the time-mean and harmonic flow equations by means of an explicit Runge-Kutta scheme. Acceleration methods to the steady state like local time stepping (with a CFL number of 3) and multigrid (with 3 grid levels) are also used. The turbulence is modeled by the eddy-viscosity one-equation Spalart-Allmaras model, with the values of y+ not exceeding 10 on the first layer of cells above solid surfaces. The simulation with 0 degree angle of attack uses 3 harmonics and 1 perturbation per blade row (generated from the upstream blade passing or the downstream blade passing). The simulation with 8 degrees of angle of attack uses 3 harmonics and 2 perturbations per blade row. The additional perturbation is due to the incidence of the flow, which is treated as a perturbation. 5

0 deg angle of attack +8 deg angle of attack Number of harmonics 3 3 Number of perturbations per row 1 2 CFL 3 3 Full Multi Grid 3 grid levels 3 grid levels Initialization Constant by blocks Constant by blocks Processors used Intel Xeon X5650@2.67GHz Intel Xeon X5650@2.67GHz Tab. 3: Numerical parameters used for the CFD computation Table 4 shows the far-field boundary conditions used for the computations. Boundary conditions similar to the ones observed during the experiments, which are representative to take-off/approach operation, have been imposed. The blades and the hub are considered as smooth and adiabatic solid wall boundaries. Figure 6 shows the boundary conditions imposed on the solid walls. Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Boundary conditions 0 deg angle of attack +8 deg angle of attack Static pressure (Pa) 98595.03 98595.03 Static temperature (K) 285.88 285.88 Absolute Mach number 0.2 0.2 Axial speed (m/s) 67.784 67.125 Angle of attack (deg) 0 +8 Tab. 4 - Far-field boundary conditions Fig. 6 - Boundary conditions on solid surfaces The simulation with 0 degree angle of attack took about 19 hours on 12 processors for a 7.6 million grid point mesh. The simulation with 8 degrees of angle of attack took 23 hours on 12 processors on the same mesh. The CPU time increase - when running a configuration with angle of attack - is coming from the additional perturbation, which has to be resolved through an extra system of harmonic equations. B. Integrated FW-H Acoustic Solver Once that the unsteady loads have been evaluated with NLH, they can be directly used as an input to start the FW-H analysis with the integrated FW-H solver of FINE TM /Acoustics. The FW-H solver applies the Ffowcs Williams-Hawkings equation in several different forms including permeable and non-permeable formulations applicable to rotating and fixed surfaces [7]. In the case of a subsonic rotating propeller or rotor, the integration surface can be placed on the propeller/rotor surface itself, and therefore the non-permeable FW-H can be used. This formulation requires just the knowledge of the pressure time history on the surface itself, but the formulation has to consider the fact that the surface is rotating. In the case of a transonic rotating propeller or rotor the aeroacoustic sources are not only placed on the rotor surface but some sources are present also in the region around the blades. It is not possible to place the integration surface on the blade itself, but this has to be placed at some distance in such a way to enclose all the source terms. Different possibilities are 6

available in FINE TM /Acoustics since it is possible to have the surface rotating with the blades, or to have a fixed surface enclosing the rotor. Typically the last approach is the most efficient since it does not require to manage the complexities of a supersonically moving surface. The FW-H surface motion can also be included (i.e. rotation + oscillations of helicopter rotors flapping, feathering, lag ). The far-field microphones can be in relative motion with respect to the FW-H surface. The F7/A7 propeller considered in this study is characterized by subsonic conditions (relative blade tip Mach number equal to 0.76), therefore the FW-H solver is applied to a non-permeable, rotating surface, lying directly on the propeller blade surfaces. As anticipated in the previous section, the aim of the FINE TM /Turbo analysis is to reproduce the steady and unsteady pressure loads generated over the blade surfaces by two main effects, respectively (1) the relative interaction between the contra rotating propellers (Figure 7 - left) and (2) the relative interaction between the propellers with the external flow with incidence (Figure 7 - right). Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Fig. 7 - Fwd/Aft propellers interaction (left). External flow/propellers interaction (right). In order to obtain a good compromise between the simulation accuracy and the computational cost, three harmonics are used in the NLH analysis to model the first three blade passing frequencies for each row and for each interaction effect. The harmonics and the relative frequencies for each propeller are reported in Table 5. Propeller NLH harmonics H1 H2 H3 H4 H5 H6 Fwd 2299.2 Hz 4598.4 Hz 6897.6 Hz 127.2 Hz 254.4 Hz 381.6 Hz Aft 2810.1 Hz 5620.3 Hz 8430.4 Hz 128.3 Hz 256.5 Hz 384.8 Hz Tab. 5 - Harmonics considered in the propeller rotating frame These harmonics are relative to the propeller rotating frame. The subsequent transformation performed by the FW-H module, from the propeller rotating frame to the microphone fixed frame, provides the propeller tones that are effectively radiated to the far-field. Their frequency can be calculated as described in ref. [10]. The full integration of the CFD solver and the FW-H solver allows to reduce the mesh handling and to simplify the setting-up of the computational model. For example, when using a fixed permeable FW-H surface, the solver self-consistently reconstructs the CFD solution on the fixed surface from the available CFD solution on the rotating surface. The NLH simulation results (i.e. the computational mesh and the complex pressures at each harmonic) are imported in FINE TM /Acoustics for the subsequent application of the FW-H calculation. Only the solid surfaces of the propeller blades may be imported in order to limit the memory resources and speed up the computational time. The exchange file format is the standard CGNS. The imported solution is used to reconstruct the pressure time history on each node of the rotating mesh. This represents the input provided to the FW-H solver. The current FW-H implementation requires as an input a time history that is automatically extracted from the NLH harmonics. 7

Fig. 8 - Flow chart of the 4 simulations performed by FINE TM /Acoustics The analysis is split in four different computations (runs), two for the forward propeller and two for the aft propeller. Each run targets a different noise generating mechanism and is performed by considering: Alternatively the forward rotor or the aft rotor (i.e. only one rotor is considered in each simulation). A specific group of harmonics, that can be selected during the import of the NLH results in FINE TM /Acoustics. Different noise sources, i.e. thickness noise and loading noise. In each run: The harmonics are used to produce a time history (on each node of the blades surface mesh) limited to one period of the lowest imported harmonic, so that each time history contains an integer number of periods for all the harmonics of the run. The time histories are then provided as input to the FW-H solver. The acoustic solver is run to produce a time history at the far-field microphones, with a length specified by the user. An integer number of front-propeller rotations (e.g. 9 rotations) shall be covered. The resulting noise signal is inevitably non periodic (except for the run relative to the front-propeller). For this purpose the signal is windowed (Hanning windowing) before performing the Fourier transform This approach ensures to obtain, by each of the four runs performed, a periodic time signal at each microphone with the same length. Once the four signals are DFT transformed, they provide consistent spectra (i.e. with same frequency range and frequency resolutions) that can be summed (complex sum) to yield the final noise spectrum. This approach allows to compute acoustic pressure spectra and time histories either for the complete CROR, either for each of the single propellers. Furthermore, the thickness and the loading noise can be separated. The time histories were reconstructed at each microphone by covering 9 forward propeller revolutions, in order to ensure a high frequency resolution (14.1 Hz), a minimized windowing effect, and in order to sample the tonal signals up to 10 KHz at least with 10 points per period. The setup parameters are reported in Table 6. The computation of the 4 runs required took 35.2 hrs on an Intel Xeon @ 2.40GHz, 8 cores processor. The calculation at zero deg of angle-of-attack took one half of the indicated time, as only two analyses are necessary at this flow condition. 8

Propeller FW-H mesh nodes Microphones Number of cores (on Intel Xeon @ 2.40GHz) Propeller revolutions Number of points Model size Hardware Output size (spectra at microphones) Sampling freq. Freq. resolution Comp. time Memory Fwd 202 k 9 8 9 115795 Hz 8192 14.1 Hz 9.7 hrs 5 GB Aft 165 k 9 8 9 115795 Hz 8192 14.1 Hz 7.9 hrs 5 GB - Tab. 6 - FW-H solver performance Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 A. Aerodynamic analysis IV. Aerodynamic and Acoustic Analysis This section presents an analysis of the aerodynamics characteristics of the CROR. The analysis will be focused on the periodic unsteady phenomena linked to the blade passing frequencies of both rotors, which are of primary relevance for the noise generation mechanisms. The front rotor and the aft rotor produce at 0 degree angle of attack a thrust of 1568N and 1220N respectively. These values are within a 10% error margin of the experimental values. Figure 9 and Figure 10 show instantaneous solutions of the normalized entropy at 50% span and 90% span respectively. A good continuity of the wake across the rotor/rotor interface has been achieved despite a limited number of harmonics that have been used. The interaction between the wakes of the first rotor and the second rotor can be clearly observed at both radial locations. Fig. 9 - Entropy field at 50% span (instantaneous solution), 8 deg AOA (left) and 0 deg AOA (right). 9

Fig. 10 - Entropy field at 90% span (instantaneous solution), 8 deg AOA (left) and 0 deg AOA (right). The instantaneous values of static pressure can be reconstructed by cumulating the harmonics and the mean flow given by the NLH simulation (Fourier reconstruction in time). In the case with incidence, additional harmonics are associated with the flow change provoked by the relative rotation and the incidence. This is illustrated in Figures 11 and 12 for one harmonic per perturbation. In Figure 11 the harmonics on the right hand side are those induced by the incidence, and the other harmonics are provoked by the relative rotation of the adjacent rotor, being associated with the 1st BPF, i.e. passing wakes (only in the aft rotor) and potential effect. The effect of the incidence is clearly visible near the hub, indicating a variation of unsteady pressure over the whole rotor (360 deg) which can reach 20,000 Pa near the hub. Cumulating the contribution of these harmonics and the mean flow gives the instantaneous static pressure, shown here on one rotor blade in Figure 12. It can be seen that the instantaneous flow differs on the same blade if the angle of attack changes from 0 to 8 deg, illustrated here for one harmonic per perturbation. 10

Fig. 11 - Harmonics of static pressure on suction side of rotor blades, 0 deg AOA (extreme left), 8 deg AOA (center and right). Fig. 12 - Static pressure on suction side of rotor blades (instantaneous solution), 0 deg AOA (left) and 8 deg AOA (right). 11

In Figures 13 and 14 the reconstruction is done for the pressure side of the rotor blades. The incidence effect is not as strong as on the suction side (5000 Pa maximum variation of static pressure over the rotor). Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Fig. 13 - Harmonics of static pressure on pressure side of rotor blades, 0 deg AOA (extreme left), 8 deg AOA (center and right). Fig. 14 - Static pressure on pressure side of rotor blades (instantaneous solution), 0 deg AOA (left) and 8 deg AOA (right). 12

Static pressure distributions at 0 degree angle of attack and 8 degree angle of attack are very similar close to the tip but show significant differences close to hub, notably the lower half of the aft rotor. Larger variations of the static pressure distribution on the blade surfaces for the 8 degree angle of attack configuration should induce an increase of the noise level by comparison to the one of the 0 degree angle of attack configuration. B. Acoustic analysis This section presents the results obtained from the acoustic analysis. A total number of 9 far-field polar microphones are considered, located in the same positions for which experimental data are available. The microphone locations are described in Figure 15. The array is located over a side line below the propellers (and the hypothetical aircraft), at a distance of 61 cm from the axis of rotation. The inspected polar angles ( ) ranges from 50 to 130 degrees, with spacing of 10 deg, with the angles measured relatively to the center of the aft propeller disk, in the aft direction. Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 Fig. 15 - Polar reference systems Figure 16 and 17 show the numerically predicted sound pressure level (SPL) spectra obtained at the polar angles of 50, 90 and 130 degs, for the flow conditions at 0 deg and +8 deg angle of attack. The main tonal noise is produced at BPF1 (1399 Hz), BPF2 (1154 Hz) and at the relative interaction frequencies BPF1+BPF2 (2554 Hz) and 2*BPF1+BPF2 (3953 Hz). The index 1 refers to the BPF for of the forward propeller, the index 2 refers to the BPF of the aft propeller. The spectra highlight the dependence of the noise emissions on the polar directivity and show a clear increase of the SPL in flow conditions with an angle-of-attack, particularly for the blade passing frequencies. 13

Fig. 16 - Numerical predictions of the noise spectra at three polar microphones located along a 61 cm side line. Flow AOA of 0 deg (BPF1 refers to the forward propeller, BPF2 to the aft propeller). Fig. 17 - Numerical predictions of the noise spectra at three polar microphones located along a 61 cm side line. Flow AOA of +8 deg (BPF1 refers to the forward propeller, BPF2 to the aft propeller). Figures 18 and 19 show the SPL variation of BPF1 with respect to the polar angle, for the cases of 0 deg and +8 deg of angle-of-attack. At 0 deg of AOA the predicted levels exhibit a flat maximum between 70 and 90 degs. Within this angular range the experimental data are underpredicted by 5 db. At +8 deg of AOA the prediction exhibits a clear peak at 90 deg. A good agreement with the experimental data is observed, especially at the microphones extending from the aft propeller plane (90 deg) towards the aft direction, while for microphones at forward angles below 90 deg there is a general tendency to underpredict the results. Considering that the real geometry of the blades is not known, the results presented should not be strictly compared with experimental values but instead used mainly to verify that the overall computational chain can predict the expected overall behavior for what concerns the emission angles and the effect of the angle of attack, that are indeed captured by the proposed approach. A pronounced increase in the radiated noise level of the BPF1 rotor-alone tone, of approximately 10 db at 90 deg, is produced by the effect of the angle-of-attack. 14

Fig. 18 - BPF1 tone directivity along a 61 cm side line. Flow AOA of 0 deg. Comparison of numerical results vs NASA measurements. Fig. 19 - BPF1 tone directivity along a 61 cm side line. Flow AOA of +8 deg. Comparison of numerical results vs NASA measurements. Figures 20 and 21 show the SPL variation of BPF2 with respect to the polar angle, for the cases of 0 deg and +8 deg of angle-of-attack. At 0 deg of AOA the predicted levels exhibit a maximum at 90 degs. At this polar angle the experimental data are underpredicted by approximately 5 db. At +8 deg of AOA the prediction exhibits a peak shifted to 100 deg. A good agreement with the experimental data is observed at the microphones extending from 100 deg of polar angle towards the aft direction, while for microphones at forward angles below 100 deg there is a general tendency to underpredict the results. Considering that the real geometry of the blades is not known, the results presented should not be strictly compared with experimental values but instead used mainly to verify that the overall computational chain can predict the expected overall behavior for what concerns the emission angles and the effect of the angle of attack, that are indeed captured by the proposed approach. A pronounced increase in the radiated noise level of the BPF2 rotor-alone tone, of approximately 8 db at 90 deg, is produced by the presence of the angle-of-attack. 15

Fig. 20 - BPF2 tone directivity along a 61 cm side line. Flow AOA of 0 deg. Comparison of numerical results vs NASA measurements Fig. 21 - BPF2 tone directivity along a 61 cm side line. Flow AOA of +8 deg. Comparison of numerical results vs NASA measurements. The numerical results shown in Figures 18-19-20-21 are consistent with the experimental evidence that rotoralone tones typically show a maximum level near the propeller plane. Figures 22 and 23 show the SPL variation of BPF1+BPF2 with respect to the polar angle, for the cases of 0 deg and +8 deg of angle-of-attack. A general good agreement is observed between the numerical predictions and the experimental data. 16

Fig. 22 - BPF1+BPF2 tone directivity along a 61 cm side line. Flow AOA of 0 deg. Comparison of numerical results vs NASA measurements Fig. 23 - BPF1+BPF2 tone directivity along a 61 cm side line. Flow AOA of +8 deg. Comparison of numerical results vs NASA measurements. Figures 24 and 25 show the SPL variation of 2*BPF1+BPF2 with respect to the polar angle, for the cases of 0 deg and +8 deg of angle-of-attack. A general good agreement is observed between the numerical predictions and the experimental data, especially at the microphone lying in the aft propeller plane. The overprediction occuring above 100 deg might be due to a possible lack in the fidelity of the reconstructed geometry, as explained in Par. III-A. The high frequency interaction tones are, in effect, sensitive to the rotors spacing and to the blades geometry, which have been estimated on the basis of the limited geometrical information available for the F7/A7 configuration. 17

Fig. 24-2*BPF1+BPF2 tone directivity along a 61 cm side line. Flow AOA of 0 deg. Comparison of numerical results vs NASA measurements Fig. 25-2*BPF1+BPF2 tone directivity along a 61 cm side line. Flow AOA of +8 deg. Comparison of numerical results vs NASA measurements. The numerical results shown in Figures 22-23-24-25 are consistent with the experimental evidence that the levels of the interaction tones at this circumferential location are less affected by the flow angle-of-attack. V. Conclusions In this paper an innovative integrated computational approach has been applied to the simulation of the NASA counter-rotating F7/A7 propellers for which experimental acoustic data are available. The adopted simulation approach is based on the exploitation of an efficient CFD solver based on the Nonlinear Harmonic method and on an efficient frequency domain aero-acoustic tool implementing the Ffowcs Williams- Hawkings formulation. The full simulation chain (mesh generation, CFD calculation, data import into the acoustic solver, acoustic analysis and data post-processing) has been successfully exploited during the study performed. It is demonstrated that the high integration level between the CFD and the acoustic solvers provides benefits in terms of reduced mesh handling, easy model preparation, reduced computational time. 18

The accuracy of the simulation has been assessed in comparison with available NASA experimental data. For this purpose a challenging flow condition with angle-of-attack was selected, numerically reproduced and compared to the case with zero angle-of-attack. The numerical results are in global reasonable agreement with the experimental data, in particular considering that the geometry of the F7/A7 propellers was not available and, therefore, had to be reconstructed based on the best fit of the known geometrical and aerodynamic parameters. The numerical simulation performed allowed to point out the main physical effects provided by the flow angle-of-attack on the noise radiation. An important reduction of the computational time necessary for the FW-H analysis is expected from the ongoing implementation in FINE TM /Acoustics of the direct usage of the NLH harmonics for the reconstruction of the time histories. This will permit to perform the full analysis with all the rotors and all the harmonics in a single step without the need to differentiate the different rotors and harmonics, and reducing further in this way not only the computational cost but also the complexity in the model preparation. Downloaded by Paolo Di Francescantonio on May 30, 2013 http://arc.aiaa.org DOI: 10.2514/6.2013-2242 References [1] Deconinck, T., Capron, A., Hirsch, C., Ghorbaniasl, G. (2012). Prediction of near- and far-field noise generated by contrarotating open rotors, Int. J. of Aeroacoustics, Vol.11, No.2, pp. 177-196. [2] Stuermer, A., Yin, J. (2009). Low-Speed Aerodynamics and Aeroacoustics of CROR Propulsion system, 15th AIAA/CEAS Aeroacoustics Conference, AIAA 2009-3134, Miami, FL. [3] Woodward, R. P. (1992). Noise of Two High-Speed Model Counter-Rotation Propellers at Takeoff/Approach Conditions, Journal of Aircraft, Vol. 29, No. 4. [4] R. Mani, (1990). The Radiation of Sound From a Propeller at Angle of Attack, NASA Contractor Report 4264. [5] Vilmin, S., Lorrain, E., Hirsch, Ch., Swoboda, M. (2006). Unsteady flow modeling across the rotor/stator interface using the nonlinear harmonic method, ASME Turbo Expo 2006, GT2006-90210, Barcelona, Spain. [6] He, L., and Ning, W. (2005). Efficient Approach for Analysis of Unsteady Viscous Flows in Turbomachines, AIAA journal, Vol. 36, No. 11, pp. 2005-2012. [7] di Francescantonio, P. (1997). A New Boundary Integral Formulation for the Prediction of Sound Radiation, Journal of Sound and Vibration 202(4), 491-509. [8] Purwanto, A., Deconinck, T., Vilmin, S., Lorrain, E., Hirsch, C. (2011). Efficient Prediction of Nacelle Installation Effects at Take-Off Conditions, ETC-9, Paper No 316. [9] Deconinck, T., Capron, A., Barbieux, V., Hirsch, C., Ghorbaniasl, G. (2011). Sensitivity Study on Computational Parameters for the Prediction of Noise Generated by Counter-Rotating Open Rotors, AIAA 2011-2765, 17th AIAA/CEAS Aeroacoustics Conference (32nd AIAA Aeroacoustics Conference), Portland, Oregon. [10] Envia, E. (2012). Open Rotor Aeroacoustic Modelling, NASA/TM-2012-217740. 19