For the automotive industry, acoustic comfort is of increasing importance. The market and customer expectation make the HVAC system noise quality a question to be addressed as early as possible in the vehicle development process. On one hand, the so-called traditional sources of annoyance such as engine, road-tires contact, exhaust systems and wind-noise have been significantly reduced for most traditional combustion engine vehicles. On the other hand, considered in the past as secondary noise sources, HVAC systems become the main sources for hybrid and electric and Stop-start vehicles. Previous studies have demonstrated the ability of a CFD/CAA approach based on the Lattice Boltzmann Method (LBM) to predict HVAC system noise including real and complex ducts, registers, mixing unit and blower geometries. This LBM low dissipative numerical approach has indeed been shown to accurately capture turbulent and convective mechanisms and to propagate acoustic waves in ducted systems and in free-field. For this type of HVAC noise simulations, both the turbulent flow and the corresponding flow-induced noise contribution are captured in one simulation.The main goal of the present paper is to apply this digital solution for predicting and optimizing the flow-induced noise contributions of a production HVAC system. The validation of the numerical results is achieved by comparing the Sound Pressure Level at the passenger's ear locations to acoustic measurements performed in production vehicles. The validated method would then be used during the development as an alternative to experiment-based prototyping which is a time-consuming approach and not always easy to carry out. The validation is organized into three phases, from ducts only to full production HVAC system, in order to provide additional understanding on mechanisms involved in HVAC noise.In this paper, our focus is on flow-induced noise contribution related to the turbulent flow generated on a detailed HVAC system. Some acoustic and pressure loss measurements are performed by Tata Motors Company on a real car. Flow and noise predictions are performed using the transient, compressible and explicit CFD/CAA solver PowerFLOW 4.4 based on LBM. Noise sources are identified for each phase and new designs numerically and experimentally tested. The final result of this project is a noise reduction of 4dB on the full HVAC system.