The recent trend in automotive automatic transmissions is towards increased number of forward gear ratios. Stemmed primarily from the power transmission efficiency concerns, eight or more forward gear ratios are becoming the norm, presenting a level of design complexity that cannot be handled without a systematic and automated set of mathematical tools. Two major tasks must be undertaken by transmission engineers in designing an automatic transmission are (i) identifying gear kinematic configurations combined with clutching schemes that can provide all of the desired forward and reverse speed ratios, and (ii) designing the planetary gear sets forming the selected gear train configuration. In view of these two critical design tasks, a family of automated design tools for the design of the planetary gear trains of N-speed automotive automatic transmissions is developed in this study. First of these design tools is a comprehensive kinematic configuration search algorithm, which identifies all transmission gear concepts delivering required gear ratios with defined number of gear decks, including specific connections of gear and carrier components, input, stationary and output members as well as the shift schedules according to smooth shifting rules. A ranking scheme is applied to the results of this search algorithm to rank order the identified candidate designs based on user-defined attributes such as number of friction elements used.
The design concepts identified by the search algorithm must be checked whether they can be assembled in a transmission. This requires the representative graph of the collapsed (all connections superimposed) cross-section of the transmission to be planar. A novel planarity check tool is developed using the graph theory and several specialized algorithms of vertex addition. This tool checks all the viable design concepts identified by the search algorithm for planarity, retaining only the ones that are planar (i.e. possible to assemble). Furthermore, the same planarity check algorithms are manipulated to automate the embedding of the graphs to guide the definition of the corresponding the transmission cross-section.
The rest of this dissertation focuses on development of a methodology for an automated design of each of the planetary gear trains forming the gear train concept identified. For this purpose, an automated design search method is developed for single- and double-planet planetary gear sets. This methodology starts a search with the definition of ranges of numbers of teeth of gears according to desired values of the ring-to-sun gear ratio, the maximum radial size of the gear set, and system-level requirements regarding planet spacing and phasing. A large number of gear set design candidates, each consisting of a matching set sun-planet, planet-planet (for double-planet sets only) and planet-ring designs, are identified within the design domain and checked for any interferences. Gear meshes of viable candidate designs are analyzed for their load distributions to determine the performance metrics of each candidate in terms of their durability, noise and efficiency outcomes. A small sub-set of the candidate designs that meet user-defined performance metrics is identified to arrive at designs that balance all of the performance requirements well.