Viscous Fluid Dynamics App for Mobile Devices Using a Remote High Performance Cluster

Classrooms and the learning process are becoming in creasingly interactive as students shift toward mobile learning platforms, yet there is a di stinct lack of engineering mobile apps. This research attempts to address this issue by developi ng and implementing an online, interactive mobile app for fluid flow, Flow HPC, which enhances the engineering student’s access to basic fluid flow information. The tool models fluid flow around any two-dimensional cross-section, and allows students to interactively experience man y fu damental aspects of fluid dynamics including viscid and inviscid flows, instantaneous drag and lift coefficients, and a visualization of velocity vectors, pressure distributions, and st reamlines about a two-dimensional object. By taking advantage of a remote, high performance c luster (HPC), the relatively low computational power of mobile devices was alleviate d. Educationally, this allows the student to access a finite element fluid flow package outside of the class and without any additional cost. This permits anyone with an internet access to solv e intensive engineering problems on a smartphone or tablet at their convenience. Since th e solution is done at a remote cluster with dozens of central processing units or CPU (hundreds of cores), the local client CPU is not relevant other than minor drawing routines. The clu ster can also accommodate hundreds of simultaneous users (estimated upper limit is 500 us ers). Flow HPC has been developed to allow the user to in teractively specify the shape of an object within a 2D flow field as well as the velocity, den sity, and dynamic viscosity of the flow. The current version of the tool includes cylindrical an d elliptical shapes. The tool has been used by students in a basic fluid dynamics course to help t hem determine drag for objects in solving flow problems. This is used as a second source for drag coefficients, the first being engineering handbooks (or textbook appendices). It has not been us d to teach computational fluid dynamics, CFD, or finite elements. Student’s positive feedbac k on using the tool for classroom discussions and assignments in a traditional fluid dynamics cla s is presented. Flow HPC was constructed as a reference tool to help students solve standard fl ui mechanic problems. The program can be downloaded at no cost from Google Play Store, Apple A p Store, and Amazon App Store. Technically, Flow HPC produces and solves a Galerki n formulation of the 2D primitive variable Navier-Stokes equations, i.e. velocity and pressure . Th Galerkin formulation produces a set of nonlinear equations. After Picard linearization, a sp rse linear equation solver, PARDISO 1 from the Intel Math Kernel Library (MKL), was wrapped in s de a Picard iteration scheme to converge on the solution. Currently, turbulence is not model e , and only low Reynolds Number (<50) are analyzed. Future plans are to include more shapes, unsteady flow, and turbulence.