Multiphysics Thermal-Acoustic Modeling of a Server in a Data Center

Modern day data centers contain multiple racks which enclose numerous servers within them to perform the intended data processing functionality. These racks usually have high system level heat flux and hence an air-cooling system is deployed using fans as a thermal management solution. The fans operate at high speeds and therefore generate a large sound pressure in the data center. The objective of this work is to propose a multi-physics thermal-acoustic modeling capability within ANSYS-Fluent to understand the environmental conditions and sound pressure level (SPL) distribution caused by servers in a data center. The model proposed can solve the acoustics and air flow physics simultaneously within the data center. The acoustic modeling in ANSYS-Fluent is validated against previous work from the authors that used ANSYS-Mechanical for a scenario of a single speaker within an IT rack with open doors. The maximum sound pressure level obtained from ANSYS-Fluent and ANSYS-Mechanical is shown to be within 10 dB, and hence a good validation between both models was established. Next, a thermal-acoustic transient model was developed in ANSYS-Fluent to understand the temperature and noise level distribution inside a server with two speakers representing a noise source. The server domain was exposed to uniform volumetric heating along with uniform velocity profile to mimic the heating and forced convection within a real server. The two speakers were positioned at the front and rear of server to mimic noise sources cause by the fans. A User Defined Function (UDF) was developed to dynamically excite the speaker surface mesh in a sinusoidal fashion to emanate sound pressure waves from the surface of speaker into the rack domain. Later, the Ffowcs Williams and Hawkings (FWH) acoustics model was used to perform the fast Fourier transformation to generate the final noise distribution at steady state. The SPL was observed on walls directly besides the speakers and a thermal gradient was established within the server. A high-fidelity multi-physics model was successfully developed in ANSYS-Fluent to capture the acoustics and thermal effects simultaneously in a server.