Helicopter drivetrain noise and vibration refinement through optimization of compound planetary gear set transmission error Jose Torres Youn Park Zachary H. Wright Aero Technical Lead Head of aerospace Engineering Manager Romax Technology Ltd. Romax Technology Ltd. Romax Technology Inc. Nottingham, United Kingdom Nottingham, United Kingdom Boulder, CO, USA ABSTRACT The tools presented are focused on the development of helicopter drivetrains providing the right virtual environment to effectively implement and test the different design options. They allow evaluation of the design criteria (deflections and misalignment, durability, dynamics, etc.) and optimization of the gearbox at component and system level simultaneously. As one application of these tools, an innovative method is presented to evaluate the different combinations of planetary tooth numbers that set the phase among the planets with the objective of optimizing the dynamic behavior of a helicopter compound planetary gearbox. Virtual testing is performed to predict the dynamic excitations from factorizing and non-factorizing planetary arrangements and the response to those excitations at different locations of the transmission. A gearbox system level approach has been required and applied to fully understand the multiphysics involved that explain the important effects of something seemingly so fundamental as the gear tooth numbers on the dynamic behavior of the entire transmission. main factors should then be considered in the helicopter drivetrain noise reduction: reduction of the transmission error and improvement of the transfer path from the gears to the gearbox housing and aircraft structure. NOTATION Factorising planetary gear set: Number of teeth on ring divided by number of planets must be an integer - this implies that the number of teeth on the sun divided by the number of planets is also an integer. This paper presents the application of virtual design and testing tools to a method developed to improve the gearbox dynamic behavior according to the points mentioned above. This method plays with the combinations of gear tooth numbers of a planetary gear system to set the phase among the planets that makes the dynamic excitation generated at the gear meshes combine in such a way that their effect on the gearbox vibration is mitigated. This method has been successfully applied in different projects and practical cases by Romax Technology. Non factorising planetary gear set: Number of teeth on ring divided by number of planets is not an integer. TE: Transmission Error. INTRODUCTION Modern rotorcrafts must meet noise and vibration requirements to ensure a minimum acoustic comfort inside the cabin. In the case of helicopters, the crew and passengers are close to the main rotor gearbox, which makes the vibration generated in the drivetrain one of the most important noise sources to address. An early identification and reduction of these vibration and noise sources should then be part of the drivetrain design process in order to find the right design in terms of acoustic requirements and feedback the assessment of other design criteria accordingly. In addition, Romax Technology has developed, than 25 years, virtual design and testing tools development and optimization of helicopter These tools are presented in this paper and improve the dynamic behavior of a helicopter planetary main gearbox. over more that allow gearboxes. applied to compound Regarding sidebands and the phasing effects that different types of manufacturing and mounting errors (carrier runout, ring gear pitch error, etc.) can introduce into the gearbox system, further analysis is required. However this paper covers a small portion of the sideband analysis and focuses on the direct consequences of the different planetary arrangements. The most important source of noise in the helicopter drivetrain is the transmission error or dynamic excitation which occurs at the gear teeth when meshing. This vibration is transmitted to the gearbox housing and then to the aircraft structure through the housing mounts, inducing housing and structure borne noise which consists of tonal noise components at the frequency of the tooth mesh orders. Two Presented at the AHS 72nd Annual Forum, West Palm Beach, Florida, USA, May 17-19, 2016. Copyright © 2016 by the American Helicopter Society International, Inc. All rights reserved. 1 APPROACH AND BACKGROUND cause a moment that tilts the gear on the support bearing. The tilt is reacted by the internal components of the bearing, the carrier, and the pin. If the contact area on the tooth surface shifts from the central position due to misalignment of the gear, an additional moment is introduced that tends to restore the gear to its design-intent position (See Figure 1). For this reason, the misalignment, load sharing, and tooth contact in an epicyclic system must be solved simultaneously with the rest of the system, and cannot be considered in isolation. METHODOLOGY A helicopter main gearbox is a complex mechanical system, which includes planetary gear trains, shafts, bearings and a gearbox housing. These components interface with each other through gear meshes, bearing mountings, and other connections. As such, modifying one design parameter can alter the performance of seemingly unrelated system components. In the process of transmission design, it is imperative that one accounts for the impact of each modification on the entire system. Because of the complex interactions within a transmission, it is difficult to achieve a given set of targets using conventional transmission analysis methods. Traditionally, the gears and the housing are analyzed separately, and the influence of housing flexibility on gear mesh misalignment is obtained through a series of approximations. These approximations, for example, linear bearings and rigid housings, result in a considerable reduction in accuracy. An improved approach is required to suit the advanced state of design and manufacturing in the automotive industry. Figure 1. Restoring moment caused by shift of loaded contact area on planet gear. The approach presented here comprises a software package that allows modeling and analysis of a complete helicopter driveline in which gears, epicyclic systems, bearings, and shafts are modeled as functional analysis objects with correlation to validated tests. The software can calculate all of the gear meshing points, forces, and load distribution, and takes into account all of the assigned boundary conditions. The planetary carriers and housing are meshed in a commercial FE package, imported into the model, and coupled with the internal transmission through the bearing nodes. The transmission system model built by this approach is very compact compared to conventional finite element models and much more sophisticated and accurate than only analytical solutions. Since the gears, shafts and bearings are all defined as objects, it is much easier to develop these 3D components in the model space. In essence, the user simply needs to select the appropriate parts for the system, position them, and assign their respective attributes. Mesh Misalignment When loads are applied to the gearbox, the gears become misaligned due to deflections of the shafts, bearings, and housing. This misalignment must be included in the transmission error calculation. The mesh misalignment is found by resolving the displacements at the gear mesh across the face width along the line of action (LOA) (see Figure 2). The line of action is the imaginary line that connects the two base circles, and contact always takes place in this direction, normal to the tooth surface. Before proceeding with a noise and vibration case study, a brief review of gearbox operating attributes associated with transmission error should be given. There are several topics discussed below: tooth contact, mesh misalignment, bearing deflections, shaft deflections, housing deflections and planet load sharing. Tooth Contact Transmission error is defined as the rotational error between the input and output of a gear pair, taking into account gear ratio. It is often expressed as a linear error along the line of action. For an epicyclic system, it can be for an individual gear pair or can represent the total rotational error between any two members, such as the sun and carrier, sun and ring, or ring and carrier. The forces acting on the planet gear Figure 2. Deflections resolved along the line of action. 2 Bearing Deflections FEA packages, or FE mesh data that can be solved within the model. The stiffness of a rolling element bearing is non-linear and generally increases with applied load. The stiffness submatrix for a rolling element bearing, linking the displacements and tilts of the inner and outer races, is obtained as the slope of the force versus deflection curve near the bearing’s operating displacement. The stiffness terms are obtained from detailed bearing models, which include the contact of the rolling elements with the raceways. The non-linear effects of internal clearance and pre-load, along with centrifugal effects in high-speed bearings, are effectively modeled (see Figure 3). Planet Load Sharing For epicyclic gear trains, load sharing between planets is an important phenomenon to consider. A linear analogy is suitable to exemplify how the load sharing can be unequal between the planets. Imagine two flat plates that are separated by springs in parallel. The springs are non-linear and represent the combined shaft, bearing, and gear stiffness, and the separation of the plate from the springs represents the backlash. If the initial backlash is not equal across the plate, either due to tooth thickness tolerance or pin positional error, one of the springs will start to take up load before the others. Also, if one spring is stiffer than others, it will carry more of the load for a given displacement. See the illustration in Figure 5. Figure 3. Non-linear bearing stiffness and radial internal clearance. The total displacement of the gears includes contributions from the bearing stiffness and radial internal clearances, shaft deflections, and housing deflection. Figure 5. Load sharing between planets is analogous to a flat plate supported by springs. Shaft Deflections The uniform slender shafts are modeled using Timoshenko beam elements, and FE mesh data is used for shaft components that are more complex, such as the epicyclic carrier or the ring gear. One of the FE carriers used is illustrated in Figure 4 below. MODELING TRANSMISSION ERROR TE is the predominant excitation at the gear meshes. When considering the calculation of TE for a single gear mesh, the following must be considered: torque for the given load case, gear tooth modifications, errors, and alignment conditions (see Figure 6). Assembly Error Gear Manufacturing Error Pre-loads Loads Profile Modifications Misalignment Tooth Effects Tolerance s Tooth Flexibility TRANSMISSION ERROR Figure 4. Epicyclic carrier modeled using FE. Gear Tolerances Housing Deflections Figure 6. The variety of input parameters used to calculate TE for a single gear mesh. The outer raceway of the bearings are connected to the gearbox housing. The housing data can be either a reduced stiffness matrix, which can be created by most commercial 3 For a single gear mesh, it can be assumed that the alignment of the mesh stays constant for a constant speed and torque. RomaxDESIGNER will run a full quasi-static system deflection analysis to determine the boundary conditions of speed, torque, and misalignment required for the TE analysis (see Figure 7). transmission error of each gear mesh, including all phase information. This logic is illustrated in Figure 8. Figure 8. Planetary gear TE calculation steps. So for helicopter transmissions with epicyclic gears, the route to TE involves an iterative quasistatic simulation in which the gear contact behaviour is co-simulated with the static deflection behaviour to yield the TE – this method is called “gearbox TE” (Figure 8). Figure 7. Single gear mesh TE calculation steps. For a planetary system, the interactions between the different gear meshes within the assembly invalidate the assumption of constant boundary conditions that is used for the single mesh case. In order to look at planetary TE, a full-system quasi-static analysis is performed at each rotational position of the transmission. This full-system analysis includes the effects of: In this method to obtain the TE, a useful side effect of these static analysis methods is that the values of the non-linear component stiffnesses at the specified loading condition are automatically calculated at the same time. These values can then be used to linearise these non-linear components. Time-varying gear mesh stiffness and load point on the gear face, calculated by analyzing the gear tooth contact condition and taking into account the gear microgeometry. Time-varying misalignment due to shaft, bearing, and housing deflections. Load (torque) sharing between planets. A calculation of how the torque is shared between the planets is important; this also changes with time and is dependent on many factors, including the backlash (perhaps due to manufacturing errors) and stiffness of the gear mesh. Relative phasing of planetary gear meshes, including that between the various sunplanet meshes, the various ring-planet meshes, and the ring-planet and sun-planet meshes of a given planet. These linearised representations of non-linear components like gears, bearings, clearances etc. can then be used to build yet another model – the dynamic model. As a linear representation of the complete gearbox, this can be solved using standard eigenvalue analysis methods to get a frequency domain linear dynamic model valid for a specified torque loading condition. Fortunately all this complexity is largely hidden from the user, who simply presses a button and waits a short time. Combining the TE excitation with this frequency domain model using the gear whine analysis postprocessing tools provides the user with results familiar to any noise & vibration engineer such as vibration waterfall plots and order cuts due to gear whine excitation. A range of other tools in the post-processing user interface provide useful information to help in the identification of problem areas and provide insight into ways of improving the design to reduce noise and vibration. These include: modal energy distributions, modal response contributions, mode shapes, operating deflection shapes, frequency response functions and bearing dynamic force predictions. A key point is that the analyses are solved simultaneously in this approach. This is because it is not proper to solve the shaft-bearing system to predict the mesh misalignment and subsequently use the calculated misalignment to predict the transmission error at the gear mesh. The character of the tooth contact and the misalignment are inextricably linked. Thanks to cleverly optimised algorithms and coding combined with some one-time-only up-front calculations on 3D FE components, these accurate vibration predictions across the whole frequency range are available in a matter of seconds from the moment the engineer presses the noise & vibration analysis button. By interrogating the system analysis results in the software, it is possible to extract the time-varying misalignment and 4 Analysis Results: Excitations (e.g. transmission error or user-defined excitation) Natural frequencies, mode shapes Transfer functions, FRFs 1. SPATIAL MODEL 2. MODAL MODEL Description of drivetrain structure; mass and stiffness matrices Drivetrain described by ‘modal’ mass, stiffness and damping Figure 9. Noise and vibration analysis process and results 5 Whole system response due to excitations CASE STUDY This case study focuses on the planetary gear stages of a typical helicopter main gearbox with a compound planetary system. In particular this paper will show how the methodology explained above is applied to design, analyze and optimize the planetary gear sets to improve the dynamic behavior of the gearbox. The tools used are able to analyze the whole helicopter powertrain (see Figure 10). All the drivetrain components (bearings, gears, shafts and housings) are modelled according to the approach and methodology explained in the previous section. Figure 11. Main gearbox The main gearbox is further simplified into the model shown in Figure 12. It consists only of the compound planetary system, with two stages of gear planetary sets, the first one with 4 planets and the second one with 8 planets. This model is going to be used for an initial check of the different changes in planetary arrangements. However the design and dynamic virtual testing of the gearbox will be completed with the model show in Figure 11. Figure 10. Full helicopter drivetrain model The picture above shows the complete model of the helicopter drivetrain that is used for this case study. As the main objective is to improve the design of the planetary systems from a dynamic point of view, the main gearbox is isolated for more detail analysis. In Figure 11 this main gearbox is displayed. It contains the housing and planetary carriers specified as FE components, while the other components are defined by analytical solutions (non-linear bearing stiffness model, 6 DOF gear contact model, etc.) Figure 12. Simplified model of main gearbox This simple model is axisymmetric in terms of geometry and loading conditions and it will be used just as a reference. The loading conditions applied to this main gearbox are 6600 rpm input speed and 1 MW input power. Then 271 kW is transferred to the tail rotor while the main rotor takes 729 kW. 6 PLANETARY ARRANGEMENTS – FACTORISING & NON FACTORIZING The options when selecting the gear tooth numbers on a planetary set are at first instance to keep the planets equally or unequally spaced. For an equally spaced arrangement, if the planet gears mesh in phase their number of teeth follows a factorizing arrangement. If the planet gears are out of phase the number of teeth are set up according to a nonfactorizing arrangement. A key aspect in the planetary transmission error is how the transmission errors from the different gear meshes are combined into a compound total excitation. With the capability to calculate the total compound transmission error, the planetary gear set can be considered as a whole system that can be optimized in order to improve the dynamic behavior of the gearbox. Properties of the main gearbox planetary stages 1st stage In that sense, different parameters of the gear planetary system can be changed for that objective, one of the most important is the phase among the planets. 2nd stage Sun Planet Ring Sun Planet Ring Number 56 20 96 of teeth Factorising Mesh frequency 1458.9 (Hz) Number 55 21 97 of teeth Non Mesh factorising frequency 1447.8 (Hz) When the planet gears are meshing in a planetary system, they can be in phase (the planets start meshing at the same time and mesh simultaneously over the mesh cycle) or out of phase (the planets do not mesh simultaneously at the same meshing point over the mesh cycle). This phasing property will have a large influence on the compound total transmission error coming out of the planetary system as a dynamic excitation that is transmitted to the rest of the system. 56 20 96 537.5 55 21 97 523.9 In Figure 13 a factorising planetary gear set is displayed next to a non-factorising one. As can be seen in the factorising arrangement the mesh points of the gears are symmetric and meshing at the same roll angle simultaneously over the meshing cycle. That is not the case in the non-factorising arrangement where the gear meshes are not in phase as the planetary system rotates. The phase among the planets depends at an initial stage on the number of teeth of the planetary gear set, so by changing this parameter it is possible to highly influence the dynamic behavior of the gearbox. Factorising (equally spaced) Non factorising (equally spaced) Figure 13. Factorizing / non-factorizing planetary arrangements 7 In particular for this case study, the factorizing and nonfactorising arrangements are applied and evaluated in a virtual testing environment both in terms of total compound dynamic excitation and gearbox system response. Also 4 traces of the TE calculated at each gear mesh are dispayed. At each gear mesh there is transmission error as the planetary system rotates. As explained, the TE is a dynamic excitation generated at the gear meshes. Because this simple model does not have any additional effects on the TE, the TE signal includes only the TE harmonics. Initially the calculation of TE is performed for the simple model (figure 12) from which the first results and conclusions are obtained. In Figure 14 a top view of the second stage of the main gearbox planetary system is displayed. Figure 14. 2nd Planetary stage and excitation at each gear mesh 8 How the TE generated at each gear mesh is combined into a total compound TE is an important point that heavily depends on the phase among the planets. Both factorizing and non-factorising gear arrangements are integrated into the main gearbox model in order to evaluate the effect of each of them. Figure 15 shows for the factorizing arrangement the start and end of active profile of the planetary gear pairs when meshing. As can be seen, the sun-planet and the ring-planet gear meshes are perfectly aligned, which effectively means that they go through the same meshing point simultaneously over the meshing cycle. This means that the planets are in phase, which will determine how the previous periodic TE signals (figure 14) are going to combine into a total compound TE in the factorizing case. (Key for Figure 15 and 16) Figure 15. Phase among gear meshes of factorizing planetary arrangement Figure 16. Phase among gear meshes of non-factorizing arrangement On the other hand, Figure 16 shows the planet phasing of the non-factorising arrangement. The planet meshes are not meshing at the same roll angle at the same time, so the planets are out of phase which will also determine how the individual mesh TE signals are going to combine into a total compound TE. 9 How the TE from different individual gear meshes combines into a total compound planetary TE is an important point that will determine in the cases of factorizing and nonfactorizing arrangements which planetary design is more suitable to improve the gearbox dynamic behavior. So that, a more detail analysis in terms of the dynamic excitations generated in the factorizing and non-factorizing arrangements is required. In the non-factorising case (right side of Figure 17), since the planet gears are out of mesh, the TEs at individual gear meshes are going to generate for each planet dynamic forces that will add up in radial direction, while the dynamic excitations are going to be cancelled out in the output torsional direction. The helicopter main gearbox has two planetary stages with the ring gear fixed and directly connected to the housing and the carriers are the output of each planetary stage. With this layout the important dynamic excitations to evaluate are the resultant radial forces on the ring gear (that are then transmitted to the housing), and the total compound TE at the output of the carrier shafts, which is going to be transferred to the rest of the gearbox system through the bearings and shaft supports. Figure 17 shows for the factorizing and non-factorizing cases how the forces involved in the TE are going to combine within the planetary system. Due to the phasing effects, in the case of a factorising arrangement (left side of Figure 17), the TEs at individual gear meshes are going to generate for each planet dynamic forces that will add up the excitations in the output torsional direction, while the resultanat forces in the radial direction are going to be cancelled out. So the planetary gear set can now be evaluated as a system that generates dynamic excitations that are applied to the gearbox system. The objective is to evaluate the effects of the factorising /non-factorising planetary arrangements on the dynamic behavior of the main gearbox. Factorizing Non Factorizing Sun-Planet Forces Ring-Planet Forces Figure 17. TE Forces Factorizing / non factorizing planetary arrangements 10 Planet Resultant Forces Resultant Forces The total compound TE at the torsional output for the factorising (left) and non-factorising (right) is shown in Figure 18. COMPOUND TRANSMISSION ERROR RESULTS The factorising and non-factorising arrangements are evaluated firstly using the simple model shown in figure 12. Under the assumptions of this simple model, the only excitation is the transmission error. The resultant dynamic forces in the ring gear and the total compound TE at the carrier output are calculated. As can be seen in the factorizing arrangement the compound TE in the torsional output is much bigger that the one obtained from the non-factorizing arrangement. The single gear mesh TEs are added up in the factorizing arrangement while they are cancelled out in the non-factorising one (the FFT of each trace is shown below the trace in question). . Non Factorizing Factorizing Figure 18. Total compound transmission for factorising/non-factorising cases – simple model 11 The dynamic radial forces on the ring gear in the factorizing (left side of Figure 19) and non-factorising (right side of Figure 19) cases are also calculated. These radial forces are cancelled out in the factorising planetary arrangement and are added up in the nonfactorising case according to the results. The first harmonic of these dynamic radial forces on the ring gear happens at the planetary mesh frequency (the higher harmonics happen at multiples of the mesh frequency). Factorizing Non Factorizing Figure 19. Dynamic radial forces on ring gear for factorising (left) /non-factorising (right) arrangements – simple model 12 Transmission error results from full detail model The planetary arrangements that have been considered so far are now evaluated in the real model with all the components fully defined (the model shown in figure 11 where the housing is included and the planetary carrier and ring shafts are fully defined with their complete geometry and stiffness). This model is used now to calculate the results under realistic assumptions of geometrical complexity and loading and boundary conditions. This means that in the compound planetary system each stage now can deform in a different way under complex loads (the forces at the bevel gear mesh are non axisymmetrical, the flexibility of housing will act as common support of the ring gears ,etc.). The total compound planetary TE and resultant radial forces are calculated with this complete model for the factorising and non-factorising planetary arrangements. The calculation of these results is now performed for 4 full rotations of the main rotor so that, according to the gearbox reduction ratios, the full system rotates a full cycle (final relative positions between the planetary stages is the same as initial one). Under these conditions the system deflections of the main gearbox can be different for each angular position of the planetary stages as the system rotates, and this also has an effect on the resultant dynamic excitations. As can be seen there are several low frequency signals that make the high frequency signals to fluctuate (fluctuation as the high frequency signals move up and down, but no modulation is introduce). The high frequency ones correspond to the TE at the mesh order and its harmonics. According to these results the TEs from individual gear meshes are added up in the torsional output. Indeed comparing the amplitudes of the TE harmonics to the results obtained from the previous simple model, it is clear that the combination of the single mesh TEs occurs in the same way. The low frequency signals correspond to effects related to the carrier deflections, ring gear deflections, unequal load sharing, etc. over the rotation of the planetary system. Looking more in detail into the main gearbox system deflections, as can be seen in Figure 21, the input torque through the bevel gears creates some non-asymmetrical deflections that are transferred through the planetary systems. This load in addition to the main and tail rotor output torques cause the main gearbox to deflect in a nonaxisymmetric way. As the planetary stages rotate they are under different deflections at each angular position, which are order of excitations as can be seen in the total compound planetary TE and resultant radial forces results. In Figure 20, the total compound planetary TE at carrier output of the second stage is displayed for the factorizing arrangement. Figure 20. Total compound transmission error factorising arrangement – real model 13 However the high frequency excitations corresponding to the gear mesh TE are cancelled out among the planet gear meshes, as this is a non-factorising arrangement. The dynamic radial forces on the ring gear are also calculated using the main gearbox model with realistic set up. Figure 23 shows these results (the charts of trace have been cut off at certain point, however the FFTs contain all the information of the full periodic function). As can be seen the low frequency signals caused by the gearbox system deflections are also present in these results. The relative angular positions of the planetary stages as the whole system rotates is also related to this effect. The local deformations on the ring gear shafts occur at different relative positions between the planetary stages as they rotate, and these local deformations are transferred to the housing that support both ring gears making them linked and affected by the deflections of each other. Figure 21. Gearbox system deflections In Figure 22 the results of total compound planetary TE of the non-factorising gear arrangement are displayed. These results show the same effect related to the low frequency signals caused by the gearbox system deflections. As can be seen in figure 23, in the factorising case the radial forces on the ring gear are almost cancelled out regarding the mesh order excitation and its harmonics (amplitude of 15 N) compared to the non-factorising gear arrangement where these forces are added up (amplitude of 350 N at mesh order) in line with the conclusions obtained so far. Figure 22. Total compound transmission error non-factorising arrangement – real model 14 Factorizing arrangement Non-factorising arrangement Figure 23. Dynamic radial forces on ring gear for factorising (left) /non-factorising (right) arrangements – real model The dynamic excitations calculated for the factorizing and non-factorising planetary arrangements are transferred to the gearbox housing and from there to the rest of the helicopter frame. However none of the planetary arrangements proposed is better design by itself in terms of excitations as their suitability depends on the gearbox system and how sensitive is to each kind of excitation. 15 Sideband analysis Looking at the single gear mesh TEs more in detail, the results are as follows for the second planetary stage in the non-factorising case (see Figure 24): Ring -> planet 1 Sun -> planet 1 Figure 24. Single gear mesh TE ring->planet 1 (left) and sun->planet1 (right) of non-factorising arrangement – real model As can be seen there are very small sidebands around the mesh order that are displayed in more detail in Figure 25: 16 Ring -> planet 1 Sun -> planet 1 X= 523.9 Hz Y= 3.08um X= 523.9 Hz Y= 4.60 um X= 529.3 Hz Y=0.10 um X= 529.3 Hz Y=0.15 um X= 518.5 Hz Y= 0.10 um X= 5.4 Hz Y= 0.70 um X= 518.5 Hz Y= 0.15 um X= 5.4 Hz Y= 0.89 um Figure 25. Sideband details of single gear mesh TE ring->planet 1 (left) and sun->planet1 (right) of non-factorizing arrangement – real model The difference in frequency between the upper and lower sidebands from the mesh frequency is 5.4 Hz respectively. The frequency of the 2nd stage carrier rotation is 5.4 Hz. Therefore as the carrier rotates there are some nonaxisymmetric effects that generate an excitation at the carrier order and sidebands around the mesh order. As can be seen the unequal load sharing accords with the non-axisymmetric carrier deflections, so it is possible to conclude that the sidebands are generated by unequal load sharing among the planets. However these sidebands are very small (their amplitude is 30 times lower than the amplitude of the first TE harmonic), so they are not going to be considered further. Looking at the carrier deflections in Figure 26 indeed there are non-axisymmetric deformations in the carrier that the planets see as the carrier rotates, generating the effects mentioned above. These deflections of the carrier are in line with the gearbox system deflections mentioned above (figure 21). It is also important to analyze the unequal load sharing among the planets (Figure 27). Figure 26. 2nd stage carrier deflections - non-factorizing arrangement 17 Figure 27. Unequal load sharing results – nonfactorising arrangement real model The objective, by calculating the dynamic response on these points, is to evaluate on the one hand the vibration transmitted to the rest of the helicopter frame through the top connection and bottom housing supports and on the other hand evaluate the air borne noise radiated from the housing as the housing walls vibrate. SYSTEM DYNAMIC RESPONSE In order to evaluate how the factorizing and non-factorising planetary arrangements influence the gearbox dynamic behavior and obtain further conclusions regarding the suitability of each planetary arrangement, a more detailed analysis of the dynamic response is required. One of the key points is the transfer path of the vibration from the point of the gearbox system where the TE is generated (gear meshes) to the point where the response happens. Deeper analysis regarding how sensitive the transfer path is to torsional excitations (factorizing) or radial forces (non factorizing) is performed. According to the excitation results presented, the factorizing planetary arrangement generates much higher excitation in the output torsional direction, while the non-factorising one generates much higher radial dynamic forces. To evaluate the effect of this excitations the first harmonic of all single mesh TEs are applied into the real model shown in figure 11. Also regarding the vibration resonance at different points a more detailed analysis is carried out in terms of the dominant mode influence and the actual vibrational deflection at those points. The dynamic response is calculated at the following locations on the gearbox housing (Figure 28) for the factorising and non-factorising cases to check the differences between the responses generated in each case. Response at the top connection The response in terms of velocity at the point of the top connection is shown in Figure 29 for the factorizing arrangement on the left hand side and for the non-factorising one on the right hand side. Below these charts there is a single chart displaying both signals. The effects of each planetary arrangement on the response depends in this case on the frequency range. In particular the highest peak showing resonance in the factorizing case is at 740 Hz while the resonance peak in the non-factorising case occurs at 3100 Hz. A more detailed analysis of the dominant modes at those critical frequencies is done according to the mode shapes shown in Figures 30 and figure 31. Top connection The mode shape at 739 Hz (figure 30), which is the dominant mode driving the resonance at 740 Hz in the factorizing case, shows that the vibrational deflection travels from the planetary systems through the carrier output shafts into the main rotor shaft. Through the main rotor shaft excites the top part of the housing where the response point is located. This is a torsional mode that is more easily excited in the factorizing case since the total compound TE at the torsional carrier output is reinforced. Housing wall Housing supports On the other hand the mode shape at 3089 Hz (figure 31), which is the dominant mode driving the resonance at 3100 Hz in the non-factorizing case, shows a different dynamic excitation transfer path. Figure 28. Response nodes In this case the high response at the top connection of the housing happens due to the high vibration of the housing walls around this top section. This mode is more easily excited in the non-factorising case as the dynamic radial forces transferred from the ring gear to the housing make the housing walls vibrate, which in this case makes the top part of the housing vibrate. 18 Response at top connection – factorising/non factorising Response at top connection - factorizing 740 Hz 3100 Hz Figure 29. Response at top connection – nonfactorizing 19 Dominant mode at 739 Hz corresponding to peak at 740 Hz in the factorising arrangement Dominant mode at 3089 Hz corresponding to peak at 3100 Hz in the non-factorising arrangement peak at 3100 Hz Figure 30. Mode shape at 739 Hz factorising arrangement Figure 31. Mode shape at 3089 Hz non-factorising arrangement Response at the point located on housing wall In Figure 32 the response on the housing surface node is displayed for both factorizing and non-factorising cases. As can be seen the resonance to the dynamic excitations depends on the modes that are excited at different frequencies depending on how sensitive these modes are to torsional or radial excitations. 20 Figure 32. Dynamic response at location on housing wall – factorizing/non factorizing Response on housing supports In this case, the factorizing case shows significantly higher response than the non-factorising one, so it can be concluded that, regarding the response at the housing supports, the nonfactorising arrangement is better in term of gearbox dynamic behavior. The structure borne path by the 4 housing supports is addressed in terms of energy (Figure 33). Figure 33. Dynamic response at housing supports – factorizing/non factorizing 21 The highest peak of vibration or resonance happens at 1660 Hz. In Figure 37 the vibration (operating deflected shape) at 1660 Hz with factorizing planetary arrangement is displayed. This transfer path is sensitive to torsional excitations, so the gearbox system gets excited in this way in the factorizing case, because in the total torsional output the TE from the different gear meshes are added up. Concerning the transfer path that the excitation follows from the gear meshes to the housing supports, it is possible to see that the planetary carrier output shafts are highly excited and this vibration travels to the drive shafts going to the rotors (main rotor drive shaft going up, tail rotor one going down in the picture). From the tail rotor drive shaft the vibration is transmitted to the bottom part of the housing through the bearings supporting the shaft. This makes the housing supports located in this bottom region of the housing vibrate, as shown. Housing support ODS at highest peak at 1660 Hz in the factorising arrangement Figure 37. Operating deflected shape at 1660 Hz – factorizing arrangement 22 CONCLUSIONS REFERENCES The combination of planetary gear tooth numbers set the phase among the gear meshes and has an important effect in the gearbox dynamic behavior. By using the right tools it is possible to predict this effect, which allows the engineer to take decisions at very early stage in the development process on the optimum gear design that improves the gearbox dynamic performance. Robert H. Badgley, Thomas Chiang, “Investigation of Gearbox Design Modifications for Reducing Helicopter Gearbox Noise,” AD 742735 USAAMRDL Technical report 72-6, 1972. 1 Rajendra Singh, Teik C. Lim, “Vibration transmission through rolling element bearings in geared rotor systems,” AD-A231 325 NASA contractor report 4334, 1990. 2 The case study presented about the dynamic effects of factorising and non-factorising planetary arrangements makes clear that these planetary gear systems do not improve anything by themselves but just provide different types of excitations. These possible improvements depend indeed on the gearbox system sensitivity to the different kinds of excitations generated, so a system level analysis that also captures the influences between the different physics involved is required. David P. Fleming, “Vibration Transmission Through Bearings With Application to Gearboxes,” NASA/TM— 2007-214954, 2007. 3 4 Riza Jamaluddin, Brian K. Wilson and Edmund Stilwell, “Boundary Conditions Affecting Gear Whine of a Gearbox Housing Acting as a Structural Member,” 2009-012031 Society of Automotive Engineers, 2009. 5 Barry James and Mike Douglas, “Development of a Gear Whine Model for the Complete Transmission System,” 2002-01-0700 Society of Automotive Engineers, Inc., 2002. The tools developed by Romax Technology and presented in this paper provide the right virtual environment to design and test the different design options at the required system level (planetary gear sets, housing, etc. and the different components integrated) and multiphysics level (influence of system deflections on transmission error excitation, effect of gear planetary architecture on the helicopter main gearbox dynamics, etc.) Barry James, Dr. Owen Harris, “Predicting unequal planetary load sharing due to manufacturing errors and system deflections, with validation against test data,” 200201-0699 Society of Automotive Engineers, Inc., 2002. 6 From the analysis with the simple model with no housing and simple loading conditions, it was possible to conclude the differences in terms of dynamic excitations between the factorising and non-factorising planetary arrangements. However these results did not provide the interaction between the planetary gear sets and the rest of the system, which is the key point to analyse. Then using the same tools another model was defined with real layout and realistic boundary and loading conditions to perform virtual testing of the helicopter main gearbox in the different scenarios of factorising and non-factorising gear arrangements. This was the right virtual environment to evaluate how this gear arrangements affect the gearbox dynamic behaviour since it was possible to understand at a system level the transfer path of the excitations from the gear meshes to the response locations on the housing, providing as well the magnitude of these responses. Using the tools and methods presented the engineer can understand the helicopter gearbox behavior and improve it according to the noise and vibration design criteria and requirements. Author contact: Jose Torres jose.torres@romaxtech.com 23