28.02.2013 Views

Handbook of Turbomachinery Second Edition Revised - Ventech!

Handbook of Turbomachinery Second Edition Revised - Ventech!

Handbook of Turbomachinery Second Edition Revised - Ventech!

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

To satisfy these requirements, it is desirable to utilize the geometric<br />

modeling capabilities <strong>of</strong> a computer-aided design (CAD) system, if possible.<br />

These systems <strong>of</strong>fer the ability to efficiently model complex shapes, using<br />

representations such as Beziers and NURBS (Non-Uniform Rational B-<br />

Splines) to precisely define surfaces. Use <strong>of</strong> such surface-description devices<br />

allows for a relatively automated scheme for the description and<br />

modification <strong>of</strong> component geometry. CAD tools also permit easy access<br />

by different user groups via the concept <strong>of</strong> neutral files, typically written in a<br />

standard IGES format. This facilitates the concept <strong>of</strong> a master model, used<br />

by both design and manufacturing functional organizations within a<br />

concurrent engineering environment.<br />

Grid Generation. Once the geometric model <strong>of</strong> the turbomachinery<br />

component has been established, the next step in the process <strong>of</strong><br />

communicating this configuration to the CFD analysis program is to<br />

define the computational grid within the physical domain.<br />

The boundaries <strong>of</strong> this region are typically defined by the flow-path<br />

surfaces (end walls, blades, etc.) and by the periodic boundaries between<br />

blade passages, where appropriate. Inlet and exit boundaries are established<br />

at points upstream and downstream, where the necessary flow conditions<br />

are assumed to be known. Within this region, a three-dimensional<br />

computational grid is applied, such that the governing equations will be<br />

solved at every point on the grid, or within every cell formed by the grid.<br />

The grid imposed on the physical domain must conform to the boundaries<br />

<strong>of</strong> that domain and must provide adequate resolution in all areas <strong>of</strong> the flow<br />

field to permit accurate prediction <strong>of</strong> the flow behavior.<br />

Computational grids may be classified by their type <strong>of</strong> construction. A<br />

structured grid has a fixed number <strong>of</strong> points in each <strong>of</strong> three directions, and<br />

the grid lines are arranged according to some scheme that is related to the<br />

geometric configuration <strong>of</strong> the physical domain. Common types <strong>of</strong><br />

structured grids include H-grids, C-grids, and O-grids. In contrast,<br />

unstructured grids are not oriented in predetermined directions, nor do<br />

they have a fixed number <strong>of</strong> points in any particular direction. Unstructured<br />

grids typically utilize tetrahedral elements, or volumes, as opposed to the<br />

hexahedral elements <strong>of</strong> structured grids. Because the unstructured grid is not<br />

restricted in its formation, it is well suited to complex geometric regions,<br />

where structured grids would be difficult to construct.<br />

Grid generation for structured grids may be performed using either<br />

algebraic methods or differential equation methods. In the algebraic<br />

approach, grid coordinates are interpolated between boundary surfaces<br />

using algebraic functions. In contrast, the differential equation method uses<br />

partial differential equations, such as Poisson’s equation, to determine the<br />

Copyright © 2003 Marcel Dekker, Inc.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!