| Abstract
This report reviews published experimental and theoretical investigations of particle deposition from turbulent flows and considers the applicability of this body of work to the specific case of particle deposition from flows in the ducts of heating, ventilating and air conditioning (HVAC) systems. Particle deposition can detrimentally affect the performance of HVAC systems and it influences the exposure of building occupants to a variety of air pollutants.
The first section of this report describes the types of HVAC systems under consideration and discusses the components, materials and operating parameters commonly found in these systems. The second section reviews published experimental investigations of particle deposition rates from turbulent flows and considers the ramifications of the experimental evidence with respect to HVAC ducts. The third section considers the structure of turbulent airflows in ventilation ducts with a particular emphasis on turbulence investigations that have been used as a basis for particle deposition models. The final section reviews published literature on predicting particle deposition rates from turbulent flows.
A large quantity of experimental data regarding particle deposition from turbulent flows has been collected using a range of techniques of varying quality. Nearly all of these data have been collected from straight tubes or ducts with a fully developed turbulent flow profile and the data are widely scattered. Most of the data of acceptable quality have been collected from tubes or ducts with hydraulic diameters much smaller than ducts in typical HVAC systems. Particle deposition from turbulent flow with a developing flow profile has not been systematically investigated and only two investigations of particle deposition from turbulent flow through bends have been published. Developing turbulent flow profiles and bends are common in HVAC ducts.
Owing to the large number of investigations into particle deposition from turbulent flow, much is known; however, the direct applicability to the case of particles in HVAC ducts is limited. Particle size, turbulence intensity and the roughness and orientation (horizontal or vertical) of the deposition surface are the parameters that control particle deposition rates and all of these factors are likely to be pertinent in HVAC ducts. Particle diameters of concern in HVAC ducts range from about 0.003 to 30 ¦Ìm and deposition rates are known to vary strongly in this range. Friction velocities in HVAC ducts are
iii
likely to be in the range 0.1-1 m/s and variations of turbulence intensities in this range are likely to influence deposition rates. Both microscale surface roughness (from less than 1 micron up to hundreds of microns) and macroscale roughness (about 1 mm and larger) have been demonstrated to enhance particle deposition relative to the case of a smooth surface. Microscale roughness intrinsic to the duct material, or due to corrosion or previous deposition of particles, and macroscale roughness from thermal insulation, joints between duct sections and debris are all potentially important in HVAC ducts. The floors of horizontal ducts are likely to experience higher rates of particle deposition than vertical duct walls or horizontal duct ceilings owing to the influence of gravity on large particles. With respect to deposition in HVAC ducts, the overall experimental data set is most lacking in information regarding the influence of developing turbulent flow profiles, microscale roughness and duct bends on particle deposition.
Particle deposition from turbulent flow depends on the nature of the turbulent flow field. Interactions between particles and air turbulence frequently determine particle deposition rates. Properties of turbulent flow that have been incorporated into particle deposition models include the fluctuating velocity component normal to the wall and the eddy viscosity. Coherent structures in near-wall turbulence such as low-speed axial streaks, near-wall streamwise vortices, bursts and downsweeps have been theorized to be important for the deposition of certain sized particles. A representative fraction of experiments and numerical simulations investigating these turbulent properties and structures are described. Turbulence near both smooth and rough walls is addressed; however, investigations into turbulence near smooth walls are more substantial and consistent in their findings. These descriptions of turbulent flow provide a foundation for understanding the theory behind particle deposition models and the results of numerical simulations of particle deposition from turbulent flows.
Four broad methods of predicting particle deposition rates are found in the literature: empirical equations, Eulerian models, sublayer models and Lagrangian simulations. These methods usually require information about the particle size and density, as well as the air speed and dimensions of the duct containing the flow. Deposition rates are most commonly reported in the form of the dimensionless deposition velocity, Vd+, versus the dimensionless relaxation time, ¦Ó+, a measure of particle inertia.
iv
Empirical equations are simple best fits to experimental data and are easy to apply. They can be combined and applied to a broad range of particle sizes and they can also be
applied to both vertical and horizontal deposition surfaces. They are of limited value for rough deposition surfaces because of sparse experimental data. Empirical equations are the only type of predictive method that has been developed for particle deposition from turbulent flow in bends. While offering little fundamental understanding, certain empirical equations can be applied to aspects of particle deposition in HVAC ducts with reasonable confidence.
Eulerian models include gradient diffusion models, free-flight models and turbophoretic models, all of which are quickly solvable with current computing power. Gradient diffusion models, when applied with reasonable assumptions, are unlikely to perform well over the full range of particle sizes for deposition in HVAC ducts. Some free-flight models successfully predict particle deposition rates through a combination of theory and empiricism. Free-flight models as presented in the literature are solely applicable to vertical deposition surfaces, though they could be adapted to horizontal surfaces. Some free-flight models achieve moderate success predicting deposition to rough surfaces by a simple modification to the method of predicting deposition to smooth surfaces. Recommended free-flight models may yield reasonable predictions in HVAC ducts, but there is little advantage to these models over the recommended empirical equations. Turbophoretic models are a significant improvement over gradient diffusion and free-flight models and are the models that are most appropriate for application to HVAC ducts. These models are applicable to vertical and horizontal surfaces over the entire range of particle sizes. Turbophoretic models are able to account for a broader range of particle transport mechanisms than other Eulerian models. They achieve good agreement with the trends and magnitudes of the experimental data with only a small amount of empiricism. The same simple method of accounting for surface roughness is used in turbophoretic models as in free-flight models.
Sublayer models are semi-Lagrangian models that can be solved rapidly with current computing power to give reasonable predictions of particle deposition to vertical and horizontal surfaces. To account for surface roughness, some sublayer models apply a similar method as the simple method used in free-flight models. Some of these models achieve reasonable agreement with the magnitudes of experimentally observed deposition rates and recommended sublayer models may be applied to the case of deposition in HVAC ducts with reasonable expectations about their performance.
v
Lagrangian simulations of particle deposition have included those conducted in relatively simple modeled turbulent flows to highly detailed flows generated by large eddy simulation (LES) and direct numerical simulation (DNS). Lagrangian simulations are often considered ¡®numerical experiments' because the results are for discrete particle sizes and specific to the numerically simulated flow conditions, much like physical experiments. The results of Lagrangian simulations are valuable for informing expectations about particle deposition in HVAC ducts; however, the high level of computational power required by these simulations makes them unsuitable at present for predicting particle deposition rates under the variety of conditions found in buildings. These simulations provide vast amounts of information on the forces acting on particles, particle velocities and particle deposition rates and offer insight into the factors that contribute to experimental uncertainty. The results of Lagrangian simulations of particle deposition generally agree with the trends and magnitude observed in experiments for both vertical and horizontal surfaces. The single reported Lagrangian simulation to consider roughness at the deposition surface resulted in the same trends as observed in experiments and also in Eulerian models using similar methods.
Turbophoretic models are the best models for application to the case of particle deposition in HVAC ducts. Turbophoretic models offer accuracy similar to other models when predicting particle deposition rates; however, they offer greater versatility in application and are based on physically realistic assumptions. Turbophoretic models have two main limitations when being applied to HVAC ducts. First, the models assume a fully developed turbulent flow profile which is not appropriate throughout HVAC systems. Second, the method of accounting for surface roughness is somewhat simplistic and corroborated by only a single data set. While models do capture the broad trends seen in experiments, they can deviate markedly from observations. In general, existing modeling approaches and empirical data are not sufficient to reliably predict particle deposition in HVAC ducts. Lastly, obtaining accurate input information for predicting particle deposition rates is a concern regardless of the type of predictive method used. | 