Master Thesis / Research Project: Methodological Validation of Server Thermal Characterization Across Cooling Strategies

Background

In a data center, the electrical energy supplied is almost completely

converted into heat and released into the environment. Currently, in

most cases this waste heat source is not utilized. In order to gain a

better understanding of the quantity and quality of the waste heat

generated, a test bench for servers was set up at the IER. A central

component of the method is a fully automated and remotely

controlled test bench with different sensors. Benchmarks are used to

specifically load the server components.

Scope of the thesis

In previous work, a server test bench was constructed, set up and

optimized. In addition, the individual components of the

measurement setup and concept were put into operation and tested.

All subsystems are fully automated and can be controlled remotely

using Python. In the course of this work, a validation of the

developed method will be carried out on another server and

processor. This is to check the extent to which the results already

obtained are transferable. For this purpose, the previous automation

and measurement approach must be adapted to the new server and

CPU.

For this purpose, the following questions are to be addressed:

• Can the test results also be reproduced on another server?

• What influence does a variation in the cooling strategy have on the quantity and quality of waste heat?

• Analyzing and processing the measurement data to create a waste heat model.