Towards a Concurrent MultidisciplinaryDesign Optimization of Mechatronic Systems
Time: Thu 2021-09-02 13.00
Subject area: Machine Design
Doctoral student: Fariba Rahimi , Maskinkonstruktion (Inst.), KTH Royal Institute of Technology
Opponent: Johan Ölvander, Linköpings universitet
Supervisor: Martin Grimheden, Mekatronik, Maskinkonstruktion (Avd.), Maskinkonstruktion, KTH Royal Institute of Technology
Multidisciplinary design optimization of mechatronic systems is a cumbersome task that considers integration of several engineering domains simultaneously. Hence, a holistic method should treat these engineering domains concurrently in the development phase and result in a solution for the system that is optimum considering several disciplines. Conventional design approaches for multi-disciplinary systems often consider each domain separately and integrate them at the end stage of the design. In these methods, the interactions and couplings between parameters from different disciplines can get lost and if any error arises at later design phase, it might lead to back-tracking and debugging, and hence, be time and cost consuming.
Therefore, a method that includes concurrent design of mechanics, electronics, control, considering the impact of embedded control implementation (on physical design and performance) which results in an integrated system is of noteworthy importance. This dissertation summarizes research by the author regarding ideas and suggestions for an integrated multi-criteria design method. The purpose of this research is to enable an early-phase design that takes into account three domains (physical design, control design, and embedded control implementation) simultaneously. Therefore objectives, specifications and constraints from each domain are taken into account. The efficiency feature is enhanced by the use of an early-phase design method which reduces time and cost consuming debugging, and removes the necessity to have iterative design loops in later design phases.
The method develops two types of components: physical and control. Physical components are basic mechanical/electrical elements which include three types of models: physical dimension, static properties, and dynamic behaviour models. Control components include control methods and dynamic performance constraints. The concept of a mechatronic system under design in the supporting software toolbox is configured using the components library. A multi-criteria optimization method is employed in a system level which yields an optimal solution for the system in terms of size, implementation cost, hardware cost and control performance. Using this system level optimization, there is no need to partition the problem or to integrate several optimization loops in the method.
Four design cases are used to enable some features of the software toolbox and investigate capability of the method to handle multi-DOF nonlinear systems; and to highlight correlation between engineering domains and broaden the coverage of disciplines. The feasibility of the method is evaluated by variations of design tests for the design cases. Accordingly, mechanical and control components are studied, developed and integrated into the IDIOM (Integrated Design Optimization of Mechatronic Systems) software toolbox. Since the model of each component is treated separately in the design and modeling stage, any system configuration that uses the available components can be handled by the method. The contribution of the thesis can be summarized as follows:
- Multidisciplinary design method and investigation of couplings and correlations between engineering domains
- Models and co-design methods to include nonlinear multi-degree of freedom mechatronic systems
- Extended method to cover key aspects in discrete time systems, and key factors in embedded control implementation
The goal of this thesis is to improve system development efficiency by integrating engineering domains in an early design phase. Accordingly, the method in this thesis is a fundamental move in evaluation of mechatronic systems design which assists in better system development and analysis. However, there is no single `best' approach for the design of mechatronic systems; the presented method in this thesis facilitates an efficient simultaneous integrated design optimization and has a broader coverage of engineering domains. The results achieved by the method ensure an optimum system solution in regards to the different involved engineering domains.