CFD-based Computations of Flexible Helicopter Blades for Stability Analysis

As a collaborative effort among government aerospace research laboratories, an advanced version of a widely used computational fluid dynamics code, OVERFLOW2.1z, was recently released. This latest version includes additions to model multiple flexible, rotating blades. This paper describes how the OVERFLOW code is applied to improve the accuracy of airload computations from the linear lifting line theory that uses displacements from the beam model. In the case used, data transfers required at every revolution are managed through a Unix-based script that runs jobs on large supercluster computers. Results are demonstrated for the 4-bladed UH-60A helicopter, deviations of computed data from flight data are evaluated, and Fourier analysis post-processings suitable for aeroelastic stability computations are performed. Use of airload data for flutter speed computations needed for stability analysis is demonstrated for a typical section of a blade. Introduction Accurate aeroelastic computations of helicopter rotor blades involve use of high-fidelity fluids and structures models. The flows are often dominated by shocks waves, bladevortex interactions and flow separation, and need the use of 3-D Navier-Stokes equations [1]. The primary aeroelastic characteristics of a helicopter rotor blade without accounting for multi-body dynamics can be modeled using beam modes [2]. Several 3-D Navier-Stokes based computational fluid dynamics (CFD) codes are in use today. OVERFLOW, one of the popular CFD codes for rotorcraft applications, has been extensively applied for rigid configurations to-date [3,4]. Of recent OVERFLOW is applied for aeroelasticity by coupling with computational structural dynamics (CSD) methods [2]. FUN3D [5] is another advanced CFD code based on unstructured grid methodology for rotorcraft applications. OVERFLOW uses overset structured grids to model the flow field. Efforts are in progress to add advanced aeroelastic capability to OVERFLOW. Recently, a beam finite-element-based structures [6] was added for isolated blades with a single grid and was demonstrated for cases that do not need trimming. In collaboration with U.S Army engineers [7], NASA added the multi-blockdynamic-deforming grid capability to the latest version of the code, OVERFLOW 2.1z, to compute accurate airloads using the prescribed aeroelastic motions of multiple blades for steady flight [8]. In this effort, OVERFLOW 2.1z solutions are applied to correct the airloads computed from linear aerodynamics based comprehensive analysis (CA) code CAMRAD [9] to improve the accuracy of aeroelastic responses. CAMRAD, which is similar to UMARC [10] and RCAS [11], computes the airloads using lifting-line theory [12], utilizing the displacements from the beam model of the rotor blade. -----------------------------------------------------------------------------* Sr. Scientist, Applied Modeling and Simulation Branch, AIAA Associate Fellow This paper describes computations made using OVERFLOW 2.1z. The prescribed structural deformation data for each revolution is computed using CAMRAD. A Unix script is employed to facilitate the data transfer between OVERFLOW and CAMRAD. Results are demonstrated for a UH-60A helicopter [13] blade. The quality of the results is assessed by computing deviations from measured flight-test data. Fourier analysis is employed to compare data with flight tests. Use of computed data for computing flutter boundary is demonstrated for a typical section of a blade. Approach Accurate computations of airloads for the UH-60A rotorcraft in forward flight require trim solutions [14]. The current state-of-the art to compute trim is based on lifting-line solutions tuned with measured thrust forces [7]. Trim solutions exclusively using CFD loads have yet to be developed. In this effort, trim parameters are computed using CAMRAD, which solves the harmonic Hamiltonian equations to give solutions only at the end of each revolution. On the other hand, OVERFLOW is based on a time marching scheme. In order to utilize the trim solutions from CAMRAD, airloads from OVERFLOW are computed at the end of every revolution and applied to correct the airloads of CAMRAD, which in turn are used for computing the aeroelastic displacements. The structural displacements are computed using the beam finite-element solver in CAMRAD. This approach, known as loose coupling (LC), is described step-bystep in the next paragraph. First, a solution in the form of blade displacement data (known as the motion file for OVERFLOW) is obtained from CAMRAD using flight parameters. The linear liftingline theory, along with free wake model, is used to compute this initial estimate of motion data from the beam model in CAMRAD. Assuming steady forward flight, this fullrevolution motion data, defined to be the same for all blades, is used as a prescribed motion for OVERFLOW. Using the required time step, computations are made for one revolution and aerodynamic forces are computed. These CFD-based aerodynamic loads are used to correct aerodynamic forces in CAMRAD, and a new motion data file with superimposed trim corrections is computed. The new motion data file is used as a prescribed motion in OVERFLOW to compute corrected aerodynamic forces. The CAMRAD/ OVERFLOW computations and data corrections are repeated until the results are converged. Convergence of results is established first by increasing the number of OVERFLOW/CAMRAD iterations and then by increasing the number of time steps per revolution in OVERFLOW. Figure 1 shows a flow diagram of the OVERFLOW/CAMRAD data exchange process. A Unix shell script [15] is used to facilitate the data exchanges between OVERFLOW and CAMRAD. OVERFLOW is run on Pleiades supercomputer at the NASA advanced supercomputing center facility at Ames Research Center [16] using portable batch system (PBS) [17] with Message Passing Interface (MPI) [18]. CAMRAD is run on a front end Linux node. Fig 1. OVERFLOW/CAMRAD data exchange process.