Simulation of Aerosol Sealant Deposition
Recent efforts to improve the energy efficiency of buildings in the US have included efforts to understand and address the tightness of air ducts and building shells. Recorded field measurements tend to show that ductwork in residential and small commercial buildings leaks by approximately 20%. Ayden et al.  have shown that the majority of leaks (78-96%) occur at joints in ductwork while the remaining leaks (4-22%) occur along seams. Leaks in the HVAC system and the building envelope result in excess infiltration and exfiltration, which create an increased load on the ventilation, heating and air conditioning equipment. Naturally ventilated buildings require a certain degree of exfiltration and infiltration in order to maintain acceptable indoor air quality, however, national surveys collected by Sherman et al.  have shown that less than 10% of US homes meet ASHRAE’s airtightness standard .
Much research has been conducted on leakage in building ducts and from that research very few innovative methods of sealing those leaks have been developed. One innovative method that has been successfully implemented as a means of sealing duct leaks quickly and efficiently is aerosolized sealants. This method of sealing ducts involves blocking all the inlets and outlets of a duct system, pressurizing the ducting with a fan and then injecting an aerosolized sealant into the ducts. The aerosolized sealant then stays suspended in the air until the flow field inside the duct carries it to a leak where it is deposited. Unlike the traditional method of sealing ducts with tapes and mastics, this method of application is noninvasive and does not require knowledge of the location of the leaks. Aerosolized sealants have been shown by Modera et al.  to seal duct leaks in ducts by as much as 86%.
Although the use of aerosolized sealants has been proven effective at sealing duct leaks both experimentally and in the field, little is known about the deposition mechanism or the relationship between particle size, streamline characteristics and deposition. These phenomena have proven difficult to explore experimentally because particle deposition occurs on such a small scale and over such a brief amount of time. To overcome the limitations of laboratory instrumentation and observe what actually occurs in the immediate vicinity of the leak as particles approach it, a computational fluid dynamics (CFD) simulation of the sealing process was created.
Research has shown that the deposition of small particles suspended in a fluid can be effectively modeled using numerical methods. Ding et al.  demonstrated a lattice Boltzmann method for simulating particle dispersion and deposition in laminar flow. Agnihotri et al.  investigated three different methods of modeling particle deposition in turbulent flow and validated each with experimental data. This topic of research has encompassed a broad range of applications. Extensive research has been focused on aerosol interactions with the human respiratory system. Tian et al.  developed an effective method to numerically simulate the deposition of pharmaceutical aerosols throughout the conducting airways. Liu et al.  used numerical simulations to investigate the deposition of aerosols in the human nasal cavity. Guha  provides a review of computational methods for simulating particle motion and deposition for both laminar and turbulent flows. The physical understanding of the various transport mechanisms including Brownian and turbulent diffusion, turbophoresis, thermophoresis, inertial impaction, gravitational settling, electrical forces, and the effects of surface roughness are outlined and the numerical methods for simulating them are presented.
To facilitate a comparison between the CFD results and experimental data many of the parameters of the apparatus used by Carrié et al  in their experiment investigating the deposition of aerosol sealants in joint-type leaks with transverse flow were recreated in the CFD simulation. The geometry of the joint-type leak used is shown in Figure 1.
The experimental apparatus included a 30 cm by 30 cm square duct with a 5 cm long (transverse to the flow), 3 mm wide (parallel to the flow) slot in the wall of the duct. To create a joint-type leak, a block of aluminum with a machined pocket was bolted over the slot. The aluminum block measured 75 mm long (transverse to the flow), 51 mm wide (parallel to the flow) and 2.23 mm tall and the pocket measured 50 mm across (transverse to the flow), 25.5 mm deep (parallel to the flow) and 1.7 mm tall. The flow rate inside the duct was maintained at 42 liters per second (translating to an average velocity of 0.46667 m/s) and the duct pressure was maintained constant. Several duct pressures ranging from 100 to 400 Pa were investigated. The sealant used in the experiment was a water-based vinyl acetate polymer with a density of 982 to 1050 kg/m3, which is what is typically used to seal ducts in practice.
Carrié et al  found that the joint leak seals due to sealant deposition 1 to 2 mm inside the inlet to the joint leak. Their results included particle diameter distributions obtained from a cascade impactor, sealing profiles of flow rate through the leak versus time over the course of the sealing process and overall deposition efficiencies. Their results for pressure drops across the leak ranging from 100 to 400 Pa indicate that pressure variations do not have a significant effect on the deposition efficiency.
The aim of the CFD simulation is to provide greater resolution of information on the deposition mechanisms. In the investigation of the deposition Carrié was required to dismantle the apparatus and remove the sealant plug in order to measure the amount of sealant deposited. Consequently, the experiment had to be terminated every time that a measurement was taken, thus only single point measurements were possible. Using CFD, the investigation can focus on the change over time in the deposition locations, rates and efficiency as well as changes to the flow as the leak is sealed.
Additionally, the use of numerical methods provides the opportunity to investigate the relationship between particle diameter, streamline characteristics and deposition. Since the generation of a mono-disperse aerosol is not feasible these relationships are unclear in experimental data. Using numerical methods a single particle can be traced from its point of injection, through the flow and to the location of deposition (if the particle deposits). By specifying the diameter and injection point of each particle and determining if and where they deposit the relationships between these three parameters can be determined.