Etapa de pre-investigación

Currently, there are a number of calculation methods available to designers for smoke

1 Introduction

control of atrium buildings. And there are also many researches on smoke movement in atrium building. Different geometrical arrangements of fire inside a building could lead to different entrainment, and hence different plume expressions.

The amount of air entrained into the plume will depend on the configuration of the plume produced. Milke identified five configurations of smoke plume which may exist within atrium buildings, these are:

1) Axisymmetric plume

An axisymmetric plume is generally expected from a fire located near the centre of an atrium floor. This type of plume is typically remote from any walls and air is entrained around all sides of the plume. Entrainment of air will occur over the full height of the plume until it reaches the interface with a smoke layer which may have formed above.

A classical analysis of axisymmetric plumes has been carried out by Morton, Taylor and Turner ^[21].

2) Wall plume

A plume which is generated from a fire against a wall is known as a wall plume.

Zukoski^[22] developed a wall plume entrainment correlation based on “mirror symmetry”. Work by Poreh and Garrad has highlighted that further research on wall plume entrainment is desirable.

3) Corner plume

A plume which is generated from a fire located in the corner of a room, where the walls form a 90^o angle, is known as a corner plume. Zukoski ^[23] treated the corner plumes in a similar manner to a wall plume with the use of “mirror symmetry” for plume entrainment. Again, work by Poreh and Garrad has demonstrated that further research is desirable for corner plume entrainment.

4) Spill plume

A spill plume is a vertically rising plume resulting from an initially horizontally moving smoke layer which then subsequently rises at a spill edge (e.g. at an opening onto an atrium space). This type of plume is the major focus of this work as is described in detail in the following section 6.

5) Window plume (door plume)

A window plume is a plume which flows from a window (or doorway) into an atrium space. Typically, window plumes are generated from post-flashover fires. An entrainment correlation was developed by Heskestad, by comparing the air entrainment for a window plume with that of an axisymmetric plume. The window plume entrainment correlation is given by Klote and Milke ^[24].

Full-scale burning tests were performed to derive the plume expressions empirically.

However, widespread use of this approach is not economically feasible and experimental data are limited. With the rapid development of computational fluid dynamics (CFD), it is now possible to assess the plume equations, and CFD might be regarded as a useful tool in solving some plume flows in buildings. There had been earlier works on assessing the temperature and mass flux formulae for some expressions on axisymmetric plumes and balcony spill plumes with CFD packages.

However, there is less research on the effect of fire location on plume entrainment, balcony plume entrainment and smoke movement in large space with the sloping floor.

There are three specific objectives which this report aims to address, these objectives are described below:

1) Plume entrainment in a large space with different fire location

2) Balcony plume in a large space building with communicating compartment 3) Smoke movement in a large space with sloping floor

Computational Fluid Dynamics and scale model experiments are two possible methods for the determination of smoke movement in the large building.

2 Governing Equations and Large Eddy Simulation

2 Governing Equations and Large Eddy Simulation

Smoke and fire movement in a building is usually a turbulent flow with significant density variation due to large temperature gradients. In the field of fire protection engineering, zone models are frequently used for study of fire hazards or design of protections systems. In recent years, however, computational fluid dynamics (CFD) has shown great value as a tool to study smoke and fire movement. In the CFD approach a limited number of assumptions are made and a high-speed digital computer are used to solve the resulting governing fluid dynamic and heat transfer equations.

CFD is divided into three types:

(1) Direct numerical simulation (DNS);

(2) Reynolds averaged Navier-Stokes equation (RANS);

(3) Large eddy simulation (LES).

DNS requires the grid resolution to be as fine as a viscously determined scale. Using DNS with simple geometry and a low Reynolds (Re) number flow provides very valuable information for verifying or improving turbulence models. However, the number of DNS grid points required for the resolution of all scales increases approximately as Re. This creates difficulty in handling high Re numbers with strong buoyant flows, such as fire and smoke movement problems in realistic conditions.

Large-eddy-simulation (LES) for solving the fluid dynamic equations of 3D elliptic, reacting flow.

LES is considered somewhere between DNS and k-e turbulence model computations.

The basic idea behind the LES technique is that eddies that account for most of the mixing is large enough to be calculated with reasonable accuracy from the equations of fluid dynamics. The small-scale eddy motion can be crudely accounted for. The LES approach systematically captures more and more of the dynamic range contained in the Navier-Stokes equations as the spatial and temporal resolution is improved. This approach to the field modeling of fire phenomena emphasizes high enough spatial and temporal resolution with an efficient flow solving technique.

In a LES calculation where the grid is not fine enough to resolve the diffusion of fuel and oxygen, an adjusted mixture fraction-based combustion model is used. The

large-scale transport of combustion products can be simulated directly, but combustion processes occurring at small length and time scales are represented in an approximate manner.