On the use of computational fluid dynamics for predicting natural displacement ventilation flows through a large enclosure
One of the major barriers to the adoption of passive engineering strategies in buildings, such as the provision for ventilation by natural means, is the limitation of the predictive techniques currently available for their design. At present there are three generic predictive methods for buoyancy-drivenn atural ventilation flows: simple analyticalm odelling,e xperimentawl ater-baseds cale-modetle stinga nd computational fluid dynamics( CFD). In addition,t here is a shortageo f experimentadl ata from real buildingsf or the validationo f such predictivem ethods. This work was concerned with current and emerging methods for predicting buoyancy-driven natural displacement ventilation flows within buildings. There were two main objectives for this research; to conduct a thorough experimental study on the natural ventilation flow through a full-scale enclosure representative of a real building with air as the fluid medium in order to provide benchmark data for model validation and to use this benchmark data to identify the preferred method for predicting detailed airflow patterns and thermal stratification for natural displacement ventilation flows within buildings. A single benchmark case that has received much attention in the past 15 years was identified for the experimental program: the natural displacement ventilation flow through an enclosure with low-level and high-level openings, driven by a point source of buoyancy at floor level. A simple mathematical model was proposed to describe this flow, which was validated experimentally using the small-scale water-based salt-bath technique (Linden et. al., 1990). More recently, another small-scale water-based technique has been developed and used to verify the mathematical models (Chen et. al., 2001). The simple models have also been validated numerically using the CFD approach (Cook, 1998). Despite the widespread interest in this class of ventilation flow, there had not yet been any experimental validation work reported to the authors knowledge using a full-scale air-based enclosure. To address this, a full-scale air-enclosure was constructed as part of this work and the natural displacement ventilation flow through the space investigated for a number of heat sources and for a range of opening configurations. In particular, the temperature stratification established within the enclosure and the displacement flow rates through the space were monitored and are presented. The rate of heat transfer through the walls of the enclosure and the surface temperatures of the walls were not recorded. In terms of its geometrical size, the full-scale experimental enclosure was representative of an occupied space within a real building. Due to budgetary constraints, however, it was constructed from chipboard sheet material rather than more traditional building materials, so that the thermal properties of the walls were not necessarily representative of a real building. Nonetheless, the experimental data presented does form a valuable set of benchmark data for a natural ly-d riven ventilation flow with air as the fluid medium that does not suffer from geometrical scaling problems. It was found that the simple analytical models that have been proposed to date and the water-based scale-modelling do not compare favourably with the data from the experimental study. It is thought that this is because the analytical models and the water-based scale-models effectively assume that the only significant transport mechanism within the space is convection, so that the mechanisms of diffusion and thermal radiation are neglected. Realistic predictions for this type of ventilation flow can be achieved using the CFD technique, which is not affected by scaling restrictions and can be easily extended to model additional physical processes including turbulent transport and thermal radiative transfer. This approach does, however, require further development before it can be used routinely, particularly with respect to the prediction of rates of heat transfer at solid walls. Reasonable agreement with the experimental benchmark data from the full-scale enclosure was observed only when the thermal radiation model was incorporated within the CFD-model. Improved agreement was observed when the radiation absorption characteristics of air due to the content of water vapour in the atmosphere were properly represented. The choice of which turbulence closure should be employed was found to be of secondary importance. It would, therefore, appear that thermal radiative transfer is an important transport mechanism within enclosures with air as the fluid medium. It is concluded that the CFD-technique has the potential to accurately predict the detailed airflow patterns and thermal stratification for buoyancy-driven natural ventilation flows within buildings where simpler analytical models or water-based experimental methods have limitations. A FV-radiation model should be incorporated into the CFD-model, and the absorption coefficient ic should be in the range 0.1 Orn" < ic< 0.1 5m". If possible, the rate of heat transfer at the walls of an enclosure should be prescribed in advance, as further work is required before this information can be realistically determined as part of a CIFID-simulation.