Hp-adaptive multiphysics simulations of nanoelectromechanical systems
The proposed research project focuses on the optimization of Nanoelectromechanical systems (NEMS) devices, i.e., MEMS devices shrunk to nanometer scale. NEMS are nano-scale devices that consist of miniaturized electrical and mechanical apparatuses such as actuators, beams, sensors, pumps, resonators, and motors. Due to the low power consumption, fast response time, and low mass, NEMS achieve extreme measurement sensitivity and have thus vast potential in many engineering and industrial applications. The design and analysis of any electromechanical system start with understanding the devices’ thermo-electro-mechanical behaviour and their interactions with external forces. When the critical dimension shrinks to the nanometer scale, as in NEMS, the material defects and surface effects significantly impact their performance. To adequately simulate a NEMS device, including damage and fracture at the atomic level that can lead to undesired behaviour, a multiscale multiphysics approach is required. The ultimate goal of the proposal is to develop a reliable, accurate and robust numerical tool that can be extended to capture essential phenomena on the atomic level; however, still computationally efficient enough for practical simulations. First, the project focuses on the challenges of simulating their behaviour, the need for a unified numerical approach, and proposes a meshless numerical methodology to address the problem at hand. The project aims to develop an hp-adaptive multiphysics simulations, enabling the study of NEMS devices in areas such as the effects of material heterogeneity, defects, interfaces, and temperature on NEMS devices. Although general, the proposed methodology will be developed on a use case from piezoelectric NEMS devices implemented as composite plates in nano-scale dimensions. Further, the project focuses on linking numerical simulations with algorithmic geometry modelling tailored to the NEMS. Geometric modelling can result in very complex geometries which in general require manual assistance to properly mesh. Node positioning required by meshless approaches is preferred to meshing, since it can be fully automated even for complex 3-dimensional domains. The results of algorithmic modelling can be coupled with node positioning to form input for the numerical simulation approach. Finally, the project merges computer simulation approach and algorithmic geometry modelling with a stochastic optimization approach to form an optimization framework for NEMS. Both the geometry of NEMS and their properties can be optimized with such a framework. Such an optimization approach is computationally expensive but can benefit greatly from parallelism and is scalable on modern computer systems.
P-Lab team
Funding
