Abstract:
The helicopter is a prime example of a nonlinear multi-body dynamic system that is subjected to numerous forces and motions to which the system must react. When a helicopter, with a conventionally articulated rotor head, is resting on the ground with its rotor spinning, a condition called ground resonance can develop. Ground resonance is a specific self-excited oscillation of the helicopter and is caused by the interaction between the main rotor blades and the fuselage structure. Inertia forces of the blades perform an out-of-phase lagging motion, which reacts with the elastic landing gear of the helicopter. For certain values of the main rotor angular velocity, the frequency of these inertia forces coincides with a natural vibration frequency of the fuselage structure. If this occurs, the inertia forces of the lagging blades produce oscillations of the fuselage, which then further excite the lagging motion of the blades. This interaction is responsible for an instability of conventionally articulated main rotor helicopters, which is called ground resonance.<br><br> The ground resonance phenomenon is investigated by means of a classical analytical approach in which the ground resonance equations are derived from Euler-Bernoulli beam theory and verified with results in literature. These equations are required to discuss ground resonance stability in further detail and determine the specific regions in which the phenomenon occurs. These results are incorporated in a simplified numerical model using an elastic multiple-body dynamics analysis program called DYMORE to simulate the South African Rooivalk Combat Support Helicopter. DYMORE is a program that offers nonlinear multi-body dynamic analysis code, using the finite element method, which was specifically developed for helicopter modelling. The complexity of helicopter modelling generally requires large amounts of computing power to ensure reasonable processing time. In order to prevent excessive computational time, the numerical model will be simplified in terms of aerodynamic and structural aspects. The scope of the numerical investigation is, therefore, limited to the ground resonance phenomenon without the effect of aerodynamic forces and representing the fuselage as multi-body beam structures of specified stiffness.<br><br> The DYMORE analysis is used to investigate various circumstances in which battle damage from a single point failure (Vibration Isolation System inactive) to a multiple-point failure (Vibration Isolation System inactive, no tire damping and no shock absorber damping) may give rise to the ground resonance phenomenon. Both static and dynamic analyses are done on various components of the helicopter model to define operational conditions in which ground resonance occurs. The conditions are determined to be that the aircraft is operating at 5600 kg, main rotor speed 187 rpm, no Vibration Isolation System and no tire damping. Including no shock absorber damping aggravates the situation even further.<br><br> To model the fuselage more accurately and to reduce computational time, the Rooivalk model is redesigned in MSC ADAMS. This software package is a family of interactive motion simulation software developed to analyse the complex behaviour of mechanical assemblies. One of the sub- modules, ADAMS AIRCRAFT, can be used to build a complete, parameterized model of a new aircraft. The model templates provided by the software are based on fixed-wing aircraft and, therefore, new design templates for a helicopter airframe configuration are created. These templates are stored as subsystems that are subsequently combined into a full aircraft assembly.<br><br> A static analysis of the fuselage structure is performed to determine the uncoupled fuselage modes of vibration, which, when combined with the main rotor blade lag frequency, indicate the possible regions of ground resonance. The flexible beam elements that are required to model the blade are not available in the MSC ADAMS software package and need to be developed in a finite element model such as MSC NASTRAN and imported into ADAMS AIRCRAFT as modal neutral files. As the flexible beam elements cannot be developed or changed within the ADAMS AIRCRAFT software package, the main rotor blades are, initially, assumed to be rigid. They are modelled by nine beam elements, which are defined in terms of their mass and mass moments of inertia in a local axis system attached to the element. The local axis systems are defined by construction frames that also define the blade twist of the main rotor blades. The beam elements are joined by fixed joints, which make the blades rigid.<br><br> The dynamic analyses performed on the full Rooivalk model with rigid blades subsequently show that the ground resonance phenomenon occurs between 275 rpm and 320 rpm, at an aircraft mass of 6258 kg, with the tire and shock absorber damping reduced to 0.001% of their fully operational value and the lead-lag damping in a single blade reduced by 20%. At 292 rpm rotor speed, the phenomenon is most pronounced. Further analyses, in which the lead-lag hinge on one of the main rotor blades is disturbed by a point torque resulting in a 0.5 rad/sec change in velocity of the blade, support the initial analyses.<br><br> Although the operational conditions for ground resonance, as predicted by the MSC ADAMS model, compare favourably with the aircraft manufacturer's prediction in terms of rotor speed (275 rpm) and main rotor blade lead-lag frequency (2.548 Hz or 16 rad/sec), an attempt is made to increase the accuracy of the blade modelling. This is done by replacing the fixed joints connecting the nine rigid beam elements with bushings. These bushings are, essentially, revolute joints for which stiffness characteristics can be specified. The stiffness characteristics, as specified for the original DYMORE blade model and the real rotor blade, are incorporated at the appropriate points of the blades where the nine beam elements join. This makes the blade semi-rigid with the same stiffness characteristics as the real blade defined at 0.777 m intervals. Various attempts at simulating the new rotor model prove to be unsuccessful as the simulation fails at balance simulation times of 6.9 seconds or less. The rotor blades deform unrealistically as MSC ADAMS assumes linear deformations. In order to simulate the nonlinear behaviour of the flexible rotor blade, it must be modelled with short flexible beams that must be developed in a separate finite element model. Highly accurate data regarding the blade construction and its material is required, which was not available at the time of this research project.