Advanced control of vapor compression cycle for large and transient heat flux removal

Yang, Zehao
Thumbnail Image
Other Contributors
Wen, John T.
Bequette, B. Wayne
Mishra, Sandipan
Peles, Yoav
Issue Date
Mechanical engineering
Terms of Use
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute, Troy, NY. Copyright of original work retained by author.
Full Citation
The rapid development of high-power electronics results in high-heat-flux heat removal issues. Current cooling techniques with single-phase flow have low heat transfer coefficient and require large mass flow rate to meet the cooling needs, which results in the low energy efficiency of the cooling system. Vapor compression cycles (VCCs) that implement two-phase flow cooling and boiling heat transfer stand out as a promising solution candidate for this cooling issue. However, these cooling systems are highly susceptible to critical heat flux (CHF) during large transient heat loads and are nonlinear and heavily cross-coupled. This dissertation presents the designs of advanced controllers for VCCs subject to large transient disturbances. The main objective of these controllers is CHF avoidance. Different control strategies are used based on the knowledge of the heat-load disturbances. For unpredictable disturbances, a gain-scheduling controller with local robust controllers are used to address nonlinearity and cross-coupling inside the VCC; a static optimal feedforward controller is added to the gain-scheduling control to improve the disturbance rejection performance. For predictable disturbances, predictive controllers (model predictive control (MPC) and quasi-static predictive control) are used to preemptively lead the VCC to more advantageous operating points by improving the cooling capability of the VCC. The controller designs are validated using nonlinear simulation and experimental validation on a VCC testbed, and both results show clear improvements of the closed-loop over the open-loop operations. Predictive controllers show better disturbance rejection performance than feedback-feedforward controllers due to disturbance prediction and preparation.
December 2016
School of Engineering
Dept. of Mechanical, Aerospace, and Nuclear Engineering
Rensselaer Polytechnic Institute, Troy, NY
Rensselaer Theses and Dissertations Online Collection
Restricted to current Rensselaer faculty, staff and students. Access inquiries may be directed to the Rensselaer Libraries.