Ali Ecder

Prediction of rotating stall within an impeller of a centrifugal pump based on spectral analysis of pressure and velocity data

Experimental data, which was acquired in two centrifugal pumps and provided by Grundfos A/S, were analysed to determine if rotating stall could be detected from the velocity and pressure time series. The pressure data, which were uniformly acquired in time at high sample rates(10 kHz), were measured simultaneously in four adjacent di.user channels just downstream of the impeller outlet. The velocity data, which were non-uniformly sampled in time at fairly low rates(100 Hz to 3.5 kHz), were acquired either in or downstream of the impeller. Two di.erent methodologies were employed for detection of stall. The first method, which involved direct analysis of raw data, yielded qualitatively useful flow reversal information from the time series for the radial velocity. The second approach, which was based on power spectrum analysis of velocity and pressure data, could detect the onset and identify the frequency of rotating stall to a satisfactory extent in one of the two pumps. Nearly identical stall frequencies were observed in both velocity and pressure power spectra and this rotating stall phenomenon, which occurred at a very low frequency relative to the impeller speed, did not reveal any noticeable degree of sensitivity to the flow rate. In the other pump, where the available data was limited to velocity time series, the power spectrum analysis was successful in detecting stationary stall for a 6 bladed impeller but did not provide conclusive results for the existence of stall in the case of the 7 bladed impeller. Recommendations on the type of experimental data required for accurate detection of stall are provided based upon the present study.

Operator splitting techniques for the numerical analysis of natural convection heat transfer

Natural convection heat transfer in rectangular enclosures is an active research area, due to its significance for both fundamental interest and engineering applications such as thermal management of electronic components. In this present numerical study, combined conduction and natural convection have been analysed for one or two heat sources mounted on substrates in an enclosure. For one heat source, the substrate is mounted on a vertical wall, and for the case of two heat sources, the substrates are mounted vertically and horizontally. Buoyancy forces drive the fluid flow. The Picard method with a hybrid grid system is utilized to decouple between pressure and velocity components, resulting in componentwise splitting. Linear systems obtained via the Picard method have been solved by the ADI method, resulting in directional splitting. Effects of different parameters such as Prandtl number, Rayleigh number, and boundary conditions on the temperature and flow field have been analysed.

Level Set Analysis of Two-fluid Interfacial Flows

Abstract In this study, application of Level Set method in modeling incompressible immiscible two-phase flows for some benchmark problems such as rising bubble and bursting bubble at the free surface is explained. Derivations of Level Set function and con- vective terms are done using fifth order Weighted Essentially Non-Oscillatory (WENO) scheme. It is observed that Level Set method is very successful in modeling two-phase flows, especially in handling topological changes like bubble skirt pinching off and merging of bubble with the free surface.

Analysis of a bimetallic slab in non-isothermal flow

The main object of this study is to investigate a fluid—structure interaction problem using Newton—Krylov techniques. As a test problem, a bimetallic slab exposed to a non-isothermal flow field is selected. If the metals on the body have different coefficients of thermal expansion, then the structure deflects in the case of a temperature change. In this study, natural convection is the reason for the temperature gradient on the slab. Solution of incompressible Navier—Stokes equations are weakly coupled with the linearized quasi-static thermoelasticity problem. The Stream function—vorticity approach is used to analyse the governing equations of the flow problem and finite differences are used for discretization. The finite-element method is preferred to solve the deflection of the solid which is modelled with plane strain assumption. Matrix-free methodology is utilized in the Newton—Krylov solver. The inexact Newton method is employed along with preconditioned Generalized Minimum RESidual as the linear solver. Domain decomposition is utilized both to ease the solution of the deformed flow geometry and to separate both physics. Several computations are performed for various Rayleigh numbers and various results are presented.

Resolving non‐symmetry in flows via subdomain shifts

Abstract In this study, non‐symmetric flow problems are modeled by selecting subdomains and shifting them in such a way that the symmetry is recovered. As a result, the domains are made of simple grid structures and re‐generation of mesh is avoided. Three test problems with various decomposition characteristics, namely, translation, rotation and deformation are selected, and they are analyzed in different flow regimes. To study the internal flow between eccentric cylinders, two cylindrical concentric subdomains are considered, one translated relative to the other. Hence, a simple polar‐coordinates mesh can be utilized instead of generating a mesh for the solution domain between the eccentric cylinders of the original problem. External flow around a curvature tube is studied shifting the subdomain around the object in rotation, relative to the outer domain thus avoiding a re‐generation of the mesh as the angle‐of‐attack changes. A third example involves deformation of an object exposed to natural convectio...

Computational parametric analysis of rotating surface flow

Abstract The flow due to rotating surfaces in a cylindrical enclosure is commonly used in many applications such as rheometry, electronic cooling, and turbomachinery, and has drawn scientists’ attention for many years. The main objective of this study is to solve the problem with a robust and efficient code. It is achieved by using the ‘Portable, Extensible Toolkit for Scientific computation’ (PETSc), which is a computational tool for the parallel solution of scientific problems. By keeping the Newtons method as the non-linear solver, different linear solvers, preconditioning techniques, and numbers of processors are tested for the performance. In addition to computational parameters such as the performance of the linear solver, performance of the preconditioner, parallel performance, and grid dependence, the effects of some physical parameters such as the Reynolds number, aspect ratio, and altering of the rotating surface are investigated. The results indicate that the core of the circulation moves towards the stationary boundaries and the boundary layers become thinner with the increasing Reynolds number. Furthermore, decreasing the aspect ratio makes diffusion harder and lowers the maximum velocities. Some of the results are compared with data in the literature.

Thermal creep flow of rarefied gases in rectangular enclosures: a comparison between the Navier-Stokes and Burnett models

Thermal creep flow of rarefied gases in enclosures has been investigated numerically by employing Navier-Stokes (NS) and augmented Burnett equations as mathematical models. An explicit multistage time stepping scheme has been exploited to solve cell centred finite volume representations of the governing equations. We have investigated the momentum diffusion mechanism by comparing computed stress components of both models at each order and we conclude that the flow can be assumed, like a purely shear driven flow, diffusion dominated motion because of the linearly varying first order shear. The analyses show that the creep velocity could not be ignored in the continuum limit. We furthermore observed that as Kn number increases, the maximum creep velocity resulting from the Burnett model decreases in contrast to NS equations.

Multiple-Domain Analysis of Combustion of a Spherical Fuel Droplet

Multi-level approach is used to analyze numerically the diffusion flame combustion of a spherical fuel droplet. Multiple levels are achieved domain-wise, considering overlapping domains of varying mesh resolution. The resulting global domain is more refined near the fuel droplet, where large gradients of some of the problem variables exist. The sub-problems are of equal size, making the computational load for each level the same. Alternating between the sub-problems speeds up the iteration process, due to coarse grid correction principle. The non-linear set of partial differential equations is decoupled component-wise, and the resulting quasi-linear system is solved for each sub-domain using the ADI method.

Partitioning strategies for parallel domain decomposition in modelling transport phenomena

Domain decomposition is a natural form of parallel preconditioning for discretised systems of partial differential equations, and the Additive Schwarz Method (ASM) is a convenient limiting form of domain decomposition preconditioning, since it maximises concurrency for a specified number of subdomains. However, except in the case of homogeneous, isotropic, elliptically-dominated systems, few of the many algorithmic parameters required by ASM possess a theoretical basis for their selection. This paper is an attempt to fill this gap experimentally. The parameters include the magnitude of anisotropic convective and diffusive effects, the order of the upwinded discretisation, the decomposition topology, decomposition orientation (relative to the anisotropy), the subdomain overlap, the accuracy of the subdomain solutions making up the composite preconditioner. The effects of these parameters on the numerical convergence rate, execution time, and parallel efficiency on distributed-memory parallel computers and workstations clusters are reported.