Partially Dressed Nose Landing Gear with Cavity Closed
Mehdi R. Khorrami
NASA Langley Research Center
Mail Stop 128
Hampton, VA 23681-2199
Tel: 757-864-3630
Email:
And
Thomas Van de Ven
Gulfstream Aerospace Corporation
Acoustics & Vibration Group
P.O. Box 2206, M/S R-06
Savannah, GA 31402-2206
Tel: 912-965-7509
Email:
1.0 Introduction
The following is a proposal for a benchmark problem using a 1/4-scale Partially Dressed Nose Landing Gear. This describes the context of the problem, experimental data available to validate both aerodynamic flow and acoustic results and criteria for meeting benchmark goals.
1.1 Problem Significance
A major goal of the aeroacoustic community is to mitigate the environmental noise impact of civil transports during take-off and landing operations. As a result of continued progress in engine noise research, a significant portion of aircraft approach noise is now generated by the airframe. Aircraft undercarriages are major contributors to airframe noise. For landing gears, the close proximity of many bluff bodies of diverse geometrical scales and shapes generates a highly complex and interactive unsteady flow field. The flow field (including noise sources, noise generation mechanisms, etc.) and the resulting acoustic field are poorly understood and predicted. Advances in prediction methodologies for airframe noise require high-fidelity numerical simulations in conjunction with high-quality experimental data to guide the development and validation of physics-based mathematical models.
An extensive aeroacoustic database was acquired using a physical model of a Gulfstream G550 nose landing gear. This generated database is to be used as a benchmark problem to validate the predictive capabilities of the current generation of computational aeroacoustic (CAA) methodologies as well as guide the future development of more capable CAA tools.
1.2 Experimental Configuration
The tested model is a quarter scale high-fidelity replica of a Gulfstream G550 nose landing gear that includes part of the lower fuselage section. To duplicate the angle-of-attack effect during flight, inherent in the model design is an inclination angle of 3 degrees relative to the streamwise (X) axis. The nose gear geometry to be used for benchmarking the computations is a “partially dressed” version of the fully dressed gear whereby the hydraulic and electrical lines, lighting cluster, and the steering assembly are removed and the gear cavity is closed (taped over). The aeroacoustic measurements for the partially dressed nose landing gear (PDNLG) were accomplished in multiple phases using two distinct wind tunnel configurations. Extensive aerodynamic measurements were obtained with the model installed in the closed-wall Basic Aerodynamic Research Tunnel (BART) at NASA Langley Research Center (LaRC). The corresponding acoustic and limited aerodynamic measurements were acquired in the University of Florida (UFL) anechoic open-jet facility.
The aerodynamic measurements (performed in BART) are available for M = 0.12, 0.145, and 0.166. The data comprise steady and unsteady surface pressures plus planar particle image velocimetry (PIV). Concurrent acoustic and surface pressure measurements (conducted in UFL open-jet tunnel) are available for M = 0.145, 0.166, and 0.189. In both tunnels, the flow over the shock-strut was tripped to ensure a fully turbulent flow. The other primary gear components, mostly made up from polycarbonate material, can also be assumed to have experienced turbulent flow due to the inherent surface roughness.
2. Problem Statement
Computational aeroacoustic simulations of the PDNLG are solicited. Although prediction of the acoustic field is the ultimate goal, submissions targeting only the near field unsteady flow are also welcomed and accepted. Given the differences in the model setup and tunnel configuration between the closed-wall tunnel (BART) and the open-jet facility (UFL), submissions should follow one of the two configuration tracks highlighted below. Concurrent simulation of both setups is highly desirable and encouraged. Such a dual attempt would provide a unique opportunity to fully scrutinize and understand facility dependency of the database.
2.1 Track A Configuration
The configuration of interest under this track is the installed model in BART that includes the tunnel test section. In this configuration, the model is mounted in an inverted position on the floor of the tunnel (Fig.1). The BART test section is 28” (0.7112 m) high, 44” (1.12 m) wide, and 10’ (3.048 m) long. For reference purposes, the model height from the tunnel floor to the top (bottom in aircraft coordinates) of the wheels is approximately 16.8” (0.427 m) and the wheel diameter is 5.5” (0.1397 m). The tunnel-floor wall (from the entrance to the end of the test section) and the fuselage surface should be treated as viscous boundaries. The tunnel side and ceiling walls can be assumed to be inviscid surfaces with no detrimental effects on the fidelity of the simulations. However, the decision on whether to treat the three remaining tunnel walls as inviscid or viscous surfaces is left to the individual parties.
Figure 1. Partially dressed nose landing gear model as tested in BART
Overall, simulation of the track A setup is most appropriate for benchmarking the near-field aerodynamic results as well as predicting the acoustic field using the time-resolved pressure data on PDNLG solid surfaces only. Acoustic predictions based on FW-H integration that employ data from off-body permeable surfaces become questionable due to the reflection of the low- and mid-frequency acoustic waves from the tunnel walls.
2.2 Track B Configuration
This track is concerned with acoustic and surface pressure results based on the installed model in the UFL open-jet tunnel. Simulation of the full environment for the open-jet facility that would include the incoming jet and the associated shear layers, collector, and the surrounding anechoic chamber is beyond the scope of the current solicitation. As a first step towards capturing the open-jet test environment, under track B, simulation of the PDNLG in free field (bounded by a hard wall on one side) is solicited. In this configuration, the model is mounted from a ceiling wall which theoretically extends to infinity along the transverse and aft directions (Fig. 2). It is left to the individual parties to decide how much of the ceiling is captured within the computational domain. Except for the ceiling, the computational domain contains no other tunnel walls. Once again, the ceiling wall plus the fuselage surface should be treated as viscous surfaces.
Figure 2. Partially dressed nose landing gear model as tested in UFL open-jet tunnel
Simulation of the track B setup is the more appropriate configuration for benchmarking the intermediate distance acoustic field results. Validation of the acoustic field is to be accomplished via the a) fluctuating pressure field on the gear surfaces, b) integration of the FW-H equation using collected data at off-body permeable surfaces, or c) direct simulation of the propagating acoustic waves. Attempts at acoustic comparisons utilizing any combination of the three approaches are highly desirable and encouraged. For the track B configuration, the corresponding validation of the simulated aerodynamic results is accomplished via direct comparison with the steady and unsteady surface pressure data available from the UFL open-jet tunnel tests. However, submitters are strongly encouraged to perform cross validation checks with the more extensive aerodynamic data (i.e. planar PIV dataset) available from the BART facility. Such comparisons would provide a unique opportunity to fully scrutinize and understand facility dependency of the aerodynamic database.
2.3 Provided Geometry Data
For the Track A setup, surface definition for the PDNLG including the BART test section coordinates in both STEP and IGES file formats will be provided. In this configuration, the right-handed coordinate system is fixed at the leading edge of the fuselage on the wind tunnel floor (Fig. 1) such that the x-axis corresponds to the streamwise direction, y-axis to the spanwise direction, and z-axis to the vertical (along the gear main strut) direction. In addition to the geometry file, the x-y-z coordinates for the steady and unsteady surface pressure ports as well as the locations of the PIV planes that are selected for comparison with the computational results will be provided.
For the Track B setup, surface definition for the SNLG including the ceiling wall will be provided in both STEP and IGES files formats. To remain consistent with the right-handed coordinate system of the closed-wall tunnel (track A) configuration, the y and z axes are rotated 180 degrees about the x axis (Fig. 2). Under this transformation, the coordinates for the surface pressure ports and the locations of the PIV planes remain identical between the Track B and Track A configurations.
The linear array microphones of interest (M2 through M10) are roughly 44” (1.12 m) to 60” (1.524 m) from the wheels’ axle (Fig. 3). For the purpose of acoustic comparisons, the coordinates for the individual microphones of the linear array in both flyover and sideline orientations will be furnished. For each orientation, two sets of coordinates corresponding to the physical and virtual (corrected position due to the presence of the open-jet shear layers, as shown in Figure 3) location of the microphones will be provided.
Figure 3. Linear array setup in flyover orientation in UFL open-jet tunnel
2.4 Benchmark Flow Conditions
The dataset at the M=0.166 speed regime is selected as the benchmark case for aeroacoustic simulations of both track A and B configurations. The flow Reynolds number based on the diameter of the shock strut (0.75” or 0.01905 m) and the free stream Mach number of 0.166 (185.7 ft/s or 56.6 m/s) is 73,000. The total pressure and temperature in the BART settling chamber (or the inflow plane) are 14.716 psi (101,464 Pa) and 73.9 degrees F (23.28 C) respectively, and the measured static pressure at the exit plane is 14.3936 psi (99,241 Pa).
3. Simulation Roadmap
The art of comparing simulation results with the available aeroacoustic measurements, as well as code to code comparisons, is seldom a straight forward task and typically quite challenging. In the absence of certain rules and roadmaps on how to process and report the CAA results, such comparisons have the potential of becoming the source of extra confusion rather than an insightful exercise. It is not the intention of the current solicitation to dictate the specifics of any technical approach. However, the roadmap described below attempts to 1) provide minimal constraints, and 2) achieve uniform reporting guidelines for all submitting parties.
3.1 Time Resolution and Unsteady Data Collection
The sampling frequency for surface pressure measurements is 51.2 kHz yielding a bandwidth of 20 kHz (the A-to-D converter employed has a Nyquist factor of 2.56). The measured data records are 32 seconds long containing 1,638,400 points. The data records are broken into 100 blocks of 16384 points to construct the power spectral density (PSD) at each sensor location. The resulting frequency bin width for PSD is 3.125 Hz (=100/32s). The Prms is obtained from integration of the PSD curves over the 3.125-20,000 Hz bandwidth.
While resolution of frequencies up to 16 kHz is of great interest, the submitting parties should maintain at a minimum a frequency resolution of 6-8 kHz for the unsteady flow field. In addition, the simulated data records should be of sufficient duration to allow the construction of the surface pressure PSD curves based on averaging of three independent data blocks using a frequency bin width that is equal to or smaller than 16 Hz.
3.2 Transient Data
For highly interactive and complex flows such as those encountered over landing gears, the transient flow is of significant duration and can easily be mistaken with the expected final well-established unsteady flow. Inclusion of this initial transient segment in both the sampled time records and the construction of the mean quantities irrevocably corrupt the computed results.
Submitting authors must ensure that upon the start of time-accurate simulations, at a minimum, the first 0.06 seconds of the computed data are discarded and not used for any post processing purpose. Based on free stream M=0.166, the 0.06s period is equivalent to roughly 1.4 times the convective time-scale for an eddy to travel from the wheel axle location to the exit plane of the BART test section. Alternatively, the 0.06s period can be viewed as equivalent to 24.5 times the convective time-scale for eddies to traverse one wheel diameter of 5.5” (0.1397 m). However, authors should ensure that sufficient time is allowed for their CAA tools to fully converge prior to acquiring data for post processing.
3.3 Convergence Metrics
The convergence behavior of the simulated flow field is of paramount importance and should be clearly demonstrated by every submitting author. As a global indicator, the PDNLG lift and drag behavior, without the contribution from the fuselage, should be tabulated and reported. (Maybe lift and drag on some gear subcomponents are also relevant metrics??). To clearly demonstrate convergence, the authors should extend the time duration of the initial computational record by 40%, reprocess the data, and then show the percentage of change in the selected metrics.
Demonstration of grid convergence is highly desirable but in most instances difficult to achieve. In addition to the finest resolution possible, all authors are encouraged to run their simulations at other (coarser) resolutions and report (the incurred changes??).
3.4 Other Metrics
All submissions should provide the total grid count (e.g. number of nodes, cells, etc.) as well as the cell size nearest to the gear surface at select locations on the shock strut, starboard wheel, and torque link. For track B participants, the dimension of the computational domain utilized should also be reported. Additional data regarding the computational resources that should be furnished by the authors are: a) the type and number of CPUs utilized, b) the type of data communications used, c) the time required to obtain 1000 time steps, d)??
4. Comparison with Measurements
Due to the extensive amount of data collected, benchmarking and validation of the simulations will be limited to a select subset of the aeroacoustic database. The type and locations of selected data are highlighted in the following subsections. Submitting authors should provide a full account of the comparison between their simulations and the highlighted data.
4.1 Steady Surface Pressures (Cps)
One circumferential (0 to 360 degrees) row on the starboard wheel, two transverse rows on the port wheel, five spanwise rows on the door, one streamwise and two spanwise rows on the fuselage.