Executive Summary : | Respiratory diseases like COVID-19 spread through coughing, sneezing, talking, and breathing, which can be effectively modelled as turbulent jets or puffs laden with muco-salivary droplets. However, speech flows generated during conversations between two people that cause asymptomatic airborne transmission have remained largely unexplored due to the difficulties involved in modeling realistic speech patterns. Recent studies have used simplified speech patterns to simulate aerosol transport during conversations in quiescent air to quantify the risk of infection. The proposed work builds on previous insights by incorporating effects of speech loudness and background ventilation on the interaction of speech jets during two-people conversations. The study will use direct numerical simulation (DNS) of these flows, solving the Boussinesq Navier-Stokes equations using a second-order accurate finite-volume code MEGHA-5 in cuboidal domains. Three levels of loudness will be used, and corresponding air flow rates will be specified at an elliptic orifice modeling the mouth opening. Staggered speech profiles will be used for simulating conversation cases, and different combinations of speech loudness will be specified for the two speakers. Four different background ventilation patterns will be introduced to understand how the interaction of speech flows is affected by ambient air flows. Over 25 simulations will be carried out, and maps of inflection probability will be generated for different parameter settings. This study represents a unique approach to understanding disease transmission due to speech flows and the implications for incorporating the risk of infection during conversations in epidemiological models. |