The accurate computation to predict the heat transfer and flow characteristics in an IRS system is challenging because of the complex designs of thermo-fluid dynamics, such as the entrainment of cold atmospheric air and mixing with the hot exhaust from the gas turbine housed in naval or cargo ships. This study considers the IRS device with four cylindrical funnels stacked one above the other. This study aims to find the suitable turbulence model(s) for numerical modeling of the IRS device. The Reynolds number based on the nozzle diameter is varied between 3527 and 7231 (laboratory-scale) and to (real-scale) as per the availability of the experimental data. Eight turbulence models, viz., standard , RNG (with standard wall function and enhanced wall functions), realizable , low Reynold number (LRN) Launder-Sharma , low Reynold number (LRN) Yang-Shi , standard , SST, are used to model the complex entrainment and mixing process inside the IRS device. The LRN-LS model estimates the thermos-fluid dynamics quite accurately for the lower Reynolds number range (laboratory-scale IRS), and SST model predicts accurately for higher Reynolds number values (real-scale IRS). We believe that the present study would help further the numerical modeling of IRS devices.