Parallel algorithms for numerical simulations of EHD ion-drag micropump on distributed parallel computing systems

Shakeel Ahmed, Kamboh (2014) Parallel algorithms for numerical simulations of EHD ion-drag micropump on distributed parallel computing systems. PhD thesis, Universiti Malaysia Sarawak (UNIMAS).

[img] PDF (Please get the password from ACADEMIC REPOSITORY UNIT, ext: 082-583932/ 082-583914)
Parallel Algorithms For Numerical Simulations of EHD Ion-Drag Micropump On Distributed Parallel Computing Systems.pdf
Restricted to Registered users only

Download (5MB)


The electrohydrodynamic (EHD) ion-drag micropumps are considered as active components of micro electro mechanical systems (MEMS) such as sensors and detectors. Experimental research on ion-drag micropumps has been carried out in recent years to improve their performance for different applications. But the development of new designs is challenging because of the instrumentation expenditure and microfabrication difficulty. In such circumstances, the numerical simulation of micropumps plays important role not only to understand the different working principles but also enables to model the designs with better performance. Conventionally, the numerical simulations for such devices are obtained by using the commercial simulation packages based on the Finite Element Methods (FEM). However, these simulation packages work on pre-defined built-in functions that do not provide arbitrary control on the governing equations. Since the performance of micropumps depends on the shapes and geometries of the actuator electrodes then the variations in the geometric dimensions of the electrodes require dense and fine meshes. Hence, the numerical simulations take unacceptably more execution time on sequential computers to run the simulation packages. The present study aims to formulate EHD ion-drag pumping model using Finite Difference Method (FDM) which can be easily implemented and provide more control to the user. In addition, the idea of parallelism is applied on the underlying FDM model to reduce the computational time taken for the numerical simulation. A data parallel algorithm (DPA-EHD) is designed and implemented for the EHD equations. The DPA-EHD is further modified by utilizing the pipelining parallelism to reduce the computing iterations and named as data parallel and pipelining algorithm (DPPA-EHD). To implement the parallel algorithms a distributed parallel computing laboratory using easily available low cost computers is setup. The parallel computing laboratory is configured for homogeneous systems using MATLAB Distributed Computing Server (MDCS) with Windows 7 operating system. The designed algorithms are implemented for EHD equations and their performance is evaluated by using different performance indicators. The results showed that the parallel algorithms for EHD simulations may provide 4 to 5 times more speedup over sequential algorithm for large grid sizes. This demonstrates the feasibility of using the FDM and parallel computing to reduce the computational time in the EHD model enabling to simulate the micropumps with very small dimensions of electrodes. In order to evaluate the scalability at specific data size the appropriate regression models are fitted through the measured data as functions of number of workers. The idea for selection of optimum number of workers is presented for future work in this direction. Hence, the research contributes for the reduction of computing time in the numerical simulation of EHD equations specifically and generally those solved by using FDM.

Item Type: Thesis (PhD)
Additional Information: Thesis (PhD.) -- Universiti Malaysia Sarawak , 2014.
Uncontrolled Keywords: Dissertations, Academic, Malaysia , Electrohydrodynamics , Computing systems, unimas, university, universiti, Borneo, Malaysia, Sarawak, Kuching, Samarahan, ipta, education, postgraduate,research, Universiti Malaysia Sarawak
Subjects: Q Science > QC Physics
Divisions: Academic Faculties, Institutes and Centres > Faculty of Computer Science and Information Technology
Depositing User: Karen Kornalius
Date Deposited: 24 Jun 2015 04:18
Last Modified: 17 Mar 2020 04:45

Actions (For repository members only: login required)

View Item View Item