An Interactive Record of Research Results
Narayanan Komerath (Click here to email )
School of Aerospace Engineering,
Georgia Institute of Technology, Atlanta, GA 30332-1050


Project Initiation In the early 1980s, steady pressure coefficients along the top surface of the simplest fuselage models, immersed in a rotor wake, could not be predicted to within an order of magnitude accuracy. The review article by Sheridan and Smith cited various aspects of aerodynamic interactions between the rotor wake and the other components of a rotorcraft. When the Army Research Office initiated Centers of Excellence in Rotorcraft Technology at Georgia Tech, U. Maryland and R.P.I., Professors Jim Hubbartt and Howard McMahon and research staffer Narayanan Komerath at Georgia Tech began a systematic study of the interaction problem using a simple 2-bladed teetering rotor, and a circular cylinder airframe with a hemispherical nose, mounted in the John J. Harper 7' x 9' wind tunnel. This CD documents the results from that project. The work is attributed below to the PhD and MS candidates who formed the research team.

Baseline Data & Prediction Methods Wall effects in the facility were shown to be negligible for advance ratios above 0.6. Brand(88.1) measured the time-averaged static pressure on the airframe surface, and the periodic part of the unsteady pressure. Liou measured the velocity field above and below the rotor, and close to the top of the airframe. Mavris(89) used a combination of Scully's Free Wake Code and Clark & Maskew's AMI VSAERO panel code to construct a computational alogorithm for the rotor/cylinder interaction, named GTRAIC. Our team showed that the dominant unsteady pressure signature along the top of the cylinder could be accurately modeled as that due to the passage of the pressure field of the airfoil section of the blade: this was called the "Blade Passage Effect", and was successfully modeled using a 2-D airfoil code, applied piecewise to the rotor blade. We developed a method to document the trajectory of the tip vortex as it approached the cylinder and distorted in its vicinity, using the seed particle deficit in the vortex core as it went through a thin, strobed laser sheet (Brand88.2). Brand(90) and Liou(90) showed that the other major features in the pressure were due to collision of the tip vortex with the surface. We also found that in coupling the free wake code with the fuselage panel code, the energy addition at the rotor had to be included. Initially, Mavris (89) did this using time-averaged annular segments of the rotor disk, imposing the stagnation pressures instantly on the airframe surface.

By 1988, we were able to compute the unsteady and thus the time-averaged, pressure distribution along the top of the cylinder, to within 5% (Komerath et al 1988). Along the sides of the cylinder, however, we encountered major errors. When the time-averaged energy addition was replaced by a more rigorous unsteady Bernoulli equation formulation, the errors actually increased. We found (Komerath 91) that using the measured velocity distribution along the sides, instead of those computed from the singularity distribution, yielded better, but not accurate, values of the pressure distribution on the Retreating Blade Side of the cylinder. It was clear that the velocity distribution along the sides was very different from that predicted by potential flow methods which did not properly resolve the tip vortex collision.

Fuselage Flow Separation In this interaction test case, we did not find any large region of time-averaged flow separation, which is a problem on most rotorcraft, with their large hubs and bulging airframes. Kim studied the effect of massive flow separation on the physics of the wake/cylinder interaction, by removing the aft portion of the cylinder from its sting mount, thus leaving a large backward facing step and a massively separated recirculation region. He captured the interaction of the wake and the recirculation region using laser sheet visualization. The shear layer was found to roll up into the tip vortex, and be destroyed and re-created once for every vortex interaction. As complex as it appears, this flow problem can be modeled by a vortex sheet simulating the shear layer, interacting with the tip vortex. Kim found, using laser velocimetry, that the tip vortex itself was largely unaffected by the shear layer interaction. We argued that this was a general result: In low-speed flight, any such separation shear layer would have far less vorticity than the tip vortex coming off the rotor. As the advance ratio increased, the shear layer would get stronger, but the wake would not interact with the separated flow as much. Hence it appears that flow separation on helicopter fuselages has little effect on the rotor wake, and hence can be computed without much iteration with the wake.

Structure of the Near Wake The distribution of bound circulation along the radius of a rotating blade of simple geometry is very different from that on a fixed wing. In the rotor case, the circulation rises to a peak just inboard of the tip, then falls sharply, even going negative before recovering on some modern blade tips. From the Prandtl model of the finite wing, the strength of any filament in the trailing vortex system is equal and opposite to the value of the spanwise gradient of the bound circulation at the radial station from where the filament is trailed. Thus the vorticity in the tip vortex is opposite in sign from that in the inboard vortex sheet. The classical model of the wake of a rotor as given by Gray shows the vortex sheet and the tip vortex convecting downwards from the rotor disk, with the inboard vortex sheet moving down faster than the tip vortex. Kim showed that at the edge of the inboard sheet from our two-bladed rotor, the sheet begins to roll up into a discrete structure. The roll-up is greatly accentuated by interaction with the tip vortex from the previous blade. Within 180 degrees of vortex age, the edge of the inboard sheet rolls up into a structure with nearly 50% of the strength of the tip vortex, with opposite sense of vorticity. This feature is not observed in the wake of a single-bladed rotor, but has been observed clearly in the wake of 2-bladed rotors, 3-bladed proprotors and 4-bladed rotors of various scales. It is clear that this feature must be included in Free Wake computations to obtain accurate results for near wake dynamics.

Vortex-Cylinder Interaction ( Click here to view the instantaneous pressure distribution and tip-vortex geometry during  the interaction )

When the vortex reaches the airframe, the interaction process depends on the sense of the vorticity. The interaction process is summarized in

Kim, J.M., and Komerath, N.M., "Summary of the Interaction of a Rotor Wake and a Circular Cylinder". AIAA Journal Vol. 33, No. 3, March 1995, p. 470-478.