Development and applications of models and algorithms for model-predictive control in buildings

Donghun Kim, Purdue University


In the past three decades, there has been great interest in modeling approaches for characterizing a building's thermal dynamics. The set of mathematical models may be categorized into two groups, namely simulation models and controller models. The former is typically based on detailed physical building descriptions and provides a simulation platform for retrofit analysis, auditing energy usages and evaluating control performances. On the other hand, the latter is typically constructed from measurements and is used for demand response using building thermal storage, for optimally operating the building using predictions of future loads and for monitoring building energy performance. New modeling approaches for the two types of building models are introduced in this thesis. For simulation purposes, 1) a reduced-order model for large size multi-zone buildings and 2) a reduced-order CFD coupled model that can capture spatial variations of thermal comfort were developed and their unique applications are demonstrated. For control purposes, system identifications under the presence of unmeasured disturbances and closed loop operation was studied and guidelines for obtaining a better model under these conditions are presented. The guidelines were applied for developing a practical control algorithm to coordinate multiple rooftop units (RTUs) for small/medium commercial buildings, in which very few advanced algorithms are available. The algorithm was designed to minimize sensor requirements and implementation costs in order to achieve successful market penetration.




Braun, Purdue University.

Subject Area

Architectural|Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server