National Technical University of Athens (NTUA)
Σχεδιασμός κατατομών οδοντωτών τροχών για τη βελτιστοποίηση του βαθμού απόδοσης και της δυναμικής τους συμπεριφοράς
Abstract
dc:descriptionThis doctoral dissertation focuses on developing methods for the optimal design of spur gear profiles to enhance their efficiency and dynamic behavior. Despite being one of the oldest and most widespread transmission systems, gear technology is evolving due to new technological developments, particularly the shift towards electrification and electric vehicles. These changes, driven also by legislative frameworks, have created new requirements for gear characteristics, primarily minimizing power losses and achieving quiet operation. As demonstrated through the literature review, modern research on gears aligns with these industry needs, emphasizing the optimization of gear characteristics through robust and reliable simulation tools. This dissertation contributes to this field by focusing on the optimal design of gear tooth profiles, which significantly influence operational characteristics. A review of relevant literature suggests that gear profile optimization falls into two major categories: optimization centered on the involute curve and optimization using alternative curves as gear profiles. The optimization of involute gears holds the greatest practical value for industrial applications, which almost exclusively use involute gears in all gear transmissions. In contrast, designing alternative profiles can significantly improve design criteria by allowing more freedom in selecting the shape of the mating profiles. To provide various approaches to profile design, enabling the designer to choose the most suitable approach depending on the application, this dissertation explores both major avenues for gear optimization. In the field of involute gears, the dissertation focuses on optimizing the parameters of their macro-geometry (e.g., number of teeth, module, face width). This approach minimizes the cost of gear manufacture or procurement while maintaining the advantages of availability and interchangeability that involute gears offer. On the other hand, in the field of non-involute gears, the dissertation develops a comprehensive methodology for the optimal design of free-form tooth flanks. Although this area has garnered less research interest, it offers the greatest freedom to designers and the most potential for improving gear performance. The primary effort in the macro-geometric optimization of involute gears was the incorporation of the dynamic response of the gears into the design criteria. Improving the dynamic behavior of the gears was achieved by minimizing the static transmission error (STE). To calculate STE, a numerical model was developed that includes tooth contact analysis and the calculation of mesh stiffness and load distribution. Although the numerical model offers high accuracy at a competitive computational cost, a surrogate model was developed in this dissertation to further reduce computational costs, which is crucial for establishing a robust optimization process. For the construction of the surogate model, neural networks, specifically Feed Forward Neural Networks (FFNN), were used to predict the entire STE curve based on the geometric and operational parameters of the gear pair. The designed neural network achieves a mean absolute percentage error of 0.395%, which is satisfactory for using the STE as a criterion for the dynamic behavior of the gears, since it results in an average error of only 0.05% in simulating the dynamic response of the gears. Additionally, it improved computation time by three orders of magnitude, reducing the time from 3 seconds to 0.001 seconds on a conventional personal computer. The developed optimization algorithm uses NSGA-II for multi-objective optimization, aiming to simultaneously reduce weight, power losses, and the levels of noise and vibrations. Through the case studies examined, valuable conclusions were drawn regarding the correlation of criteria and design variables, leading to generalized guidelines for selecting macro-geometric parameters when designing a gear transmission. Specifically, in certain applications, power losses were reduced by up to 40.8%, and vibration amplitude was reduced by up to 72.3%. Reducing the STE amplitude (peak-to-peak error) proved to be a highly effective method for improving the dynamic behavior of gears, especially in high-speed and high-load operations. It was also found that reducing the vibration amplitude requires ensuring a high contact ratio. Furthermore, during the parameter selection process of a gear stage, it was found that the best compromise between design criteria is achieved by maximizing the number of teeth, which is the controlling factor in gear optimization, and subsequently reducing the module and the width of the gears to compensate for the increase in the dimensions and weight of the transmission. Significant reductions in power losses and STE amplitude can also be achieved by simultaneously increasing the profile shift coefficients of the gears. In the field of non-involute gears, a comprehensive methodology for the optimal design of free-form tooth flanks is presented for the first time in this dissertation. The optimization of free-form tooth flanks is distinct from other optimization approaches as it does not aim to optimize the geometric characteristics of a given profile (involute or non-involute) to improve gear performance, but instead, it treats the profile geometry itself as a design variable. To effectively modify the profile geometry during the optimization process, the pinion profile is represented by a parametric B-spline curve, with its control points used as design variables. Each different pinion profile uniquely corresponds to a specific profile for the mating gear. However, calculating the conjugate geometry presents challenges, notably the need for a numerical solution to an implicit equation when following the classical analytical theory of gearing. To address these issues, a new, fully analytical and reliable approach for calculating the mating geometries was developed. A significant contribution of this dissertation is the formulation of clear, mathematically defined geometric constraints regarding the compliance of the generated profiles with the law of gearing and the avoidance of cusps and discontinuities in the mating profiles. During the research, closed-form analytical relationships were developed and presented, allowing verification of compliance with the geometric constraints from the moment the free-form curve is generated. This led to valuable conclusions regarding their permissible forms. As demonstrated by the case studies, the improvement in design criteria exceeded 50% compared to standard involute gears. Specifically, the optimized profiles achieved reductions in the mean value of relative curvature by up to 65.4%, the maximum value by up to 82.9%, the standard deviation by up to 99.9%, the average wear depth along the contact path by up to 54%, and the maximum wear depth by up to 69%. The superiority of free-form gears was further proven through experiments comparing their wear characteristics to conventional involute gears, showing an improvement in average wear depth by 44.1%. Additionally, while the reduction of the contact ratio is generally an inevitable consequence of using non-involute gears, the free-form profile optimization process can lead to designs that combine significant improvements in gear characteristics while ensuring contact ratios similar to or greater than those of corresponding involute gears. Finally, a general conclusion of the research, which applies to involute gears as well, is that to improve most key gear characteristics (power losses, wear, strength, and, to a lesser extent, STE peak-to-peak), the profiles tend to increase their slope. A direct consequence of this profile shape is the reduction in the contact ratio, highlighting the need to achieve an appropriate compromise between the degree of profile slope and the contact ratio as one of the primary objectives of gear optimization.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Kalligeros, Christos
- Καλλίγερος, Χρίστος
Subjects
dc:subject × 20- Οδοντωτοί τροχοί
- Βελτιστοποίηση
- Σχεδιασμός
- Νευρωνικά δίκτυα
- Γενετικός αλγόριθμος
- Θόρυβος
- Απόδοση
- Gears
- Optimization
- Design
- Neural networks
- Genetic algorithm
- NVH
- Efficiency
- Επιστήμη Μηχανολόγου Μηχανικού
- Επιστήμες Μηχανικού και Τεχνολογία
- Μηχανολογία
- Mechanical Engineering
- Engineering and Technology
- Mechanics
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/57301