Abstract
Design of thermal controllers for datacenter machines has traditionally relied on heuristics and trial-and-error testing, resulting in lengthy design times and possibly suboptimal controller parameters. This disclosure describes techniques that leverage control and linear-system theory to build models that enable the software simulation of thermal controllers used in datacenters. Per the techniques, test data is captured from a real machine and used to create a reduced-order model of the machine and its thermal controller. A transfer function of the reduced-order model is obtained using linear system theory. Control theory is leveraged to determine the appropriate controller parameters for the machine according to user-specified design goals such as response time, robustness, etc.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
NA, "Data-driven Design for Thermal Management and Control in Datacenters", Technical Disclosure Commons, (December 05, 2024)
https://www.tdcommons.org/dpubs_series/7624