Enabling better thermal management in multi-zone indoor spaces
Loading...
Authors
Tariq, Zaid Bin
Issue Date
2019-05
Type
Electronic thesis
Thesis
Thesis
Language
ENG
Keywords
Computer Systems engineering
Alternative Title
Abstract
In multi-zone indoor spaces with thermal dependencies between zones and limited heating/cooling sources, choosing optimal zonal thermostat or heat input settings is often not enough to attain desired zonal temperatures. In the first part of the thesis, the problem of determining how the thermal flow in-between the different zones of the indoor space, or between the zones and the ambient, should be adapted towards improving the overall thermal environment in the multi-zone space. Despite the complex structure of the problem, this thesis shows how the optimal zonal heat transfer adaptation problem -- which involves determining the optimal zonal heat input settings as well -- can be formulated and solved efficiently. The work also determines conditions under which the problem has a convex structure. Simulations on a 6-zone indoor space model show that zonal heat transfer adaptation can result in significant improvement of zonal temperatures and/or energy costs, particularly when not all zones are equipped with heat sources. The second part of the thesis provides experimental evaluation of a learning-based data prediction model for predictable thermal control of a shared indoor space. Data from a smart conference room experimental test bed is used for learning the black-box model which is then incorporated into finite horizon control problem for controlling heating and cooling inputs of the indoor space. The results show that the data-driven predictive controller can maintain temperatures closer to that desired by occupants than thermostat based non-predictive control implemented by the building thermal management systems.
Description
May 2019
School of Engineering
School of Engineering
Full Citation
Publisher
Rensselaer Polytechnic Institute, Troy, NY