**1. Introduction**

Energy use and consumption of natural resources has become a pertinent concern for current and future generations. In the U.S. alone, total energy consumption has tripled over the last 65 years from 34.6 quadrillions Btus (quads) in 1950 to 101 quads in 2019 [1]. Of the energy consumed in the U.S., non-renewable energies still represent nearly 90% of energy sources [2]. Many nations have put forth specific renewable energy targets which aim to reduce dependence on non-renewable energies and maintain a competitive edge in the global energy technology market. For example, the European Union's (EU) Renewable Energy Directive has established a goal of 20% final energy consumption from renewable sources by 2020 [3]. The U.S. Department of Energy has set a goal of having 20% electricity sourced from wind energy by the year 2030 [4]. While renewable sources are projected to grow, reductions in energy usage can work to achieve these goals as well.

Delving into the energy consumption practices in the U.S., approximately 40% of all energy goes to building operations in the commercial and residential sectors [5]. The data show that approximately 75% of the energy used in the building sector comes from fossil fuels. As a result, energy usage in buildings account for 40% of the total U.S. carbon emissions [6]. Additionally, the buildings' share of U.S. energy consumption has increased from 34% in 1980 to 40% as of 2010, and is projected to

continue in growth [7]. The commercial buildings sector accounts for half of this energy usage and is a prime target for reduction.

Examining the top commercial site energy end uses reveals that space heating (27%), lighting (14%), space cooling (10%), water heating (7%), and ventilation (6%) are responsible for 64% of energy consumption [5]. These categories can be combined more generally to refer to services required mostly when a building is occupied (space conditioning and lighting). Economically, utilities cost businesses and building owners in the commercial sector \$179.4 billion in 2010 [5]. The same five end uses listed above account for the top five cost areas, further motivating the need for energy reduction technologies.

Advanced building controls is a major area of research seeking to reduce energy usage by improving on the current practice of low-level controllers (proportional-integral or proportional-integral-derivative). In some buildings, there may be supervisory control, but frequently the components and systems operate independently and in a decentralized fashion. Because of the physically interconnected and complex nature of building systems, this uncoordinated control can often lead to inefficiencies where controllers compete with each other in achieving their desired outputs. To solve this issue, many advanced control strategies have been proposed, with Model Predictive Control (MPC) being a front-runner. A key part of MPC is the development of the objective function to be minimized. Strategies thus far have targeted energy and cost reduction, but often do not optimize occupant comfort and the associated economic impact of discomfort, accounting for occupant comfort using limits of thresholds. Thus, the energy optimal solution results in the maximum tolerated occupant discomfort. One exception included the cost of occupant discomfort as a portion of an employee's lost salary, as described in [8]. Such an objective function would enable the identification of possible energy and cost savings, serving to guide building managers and researchers as to where their efforts for increases in performance and efficiency should be focused.

This paper contributes a novel, scalable steady-state economic controller that accounts for both the cost of utilities as well as the loss of productivity due to occupant discomfort. Detailed are the development of a steady-state optimal control method based on data collected from a building on Texas A&M University's campus, the comparison of a simulated standard control implementation versus the proposed supervisory controller, and a discussion of the impact of including the cost of occupant discomfort in the control strategy. While the decision was made to focus on one building for this work, the basic structure of the algorithm is scalable and enables the optimization of larger systems/buildings that contain multiple chillers, AHUs, dozens of VAVs and hundreds of zones. Additionally, this work tackles several practical challenges that are not often addressed in current literature, including accounting for humidity as well as temperature issues as they relate to energy/comfort, having a time-varying objective function that is updated based on current operating conditions, empirical fits for component models to reflect the systems in an actual building, and switching models to capture changes in system behavior due to mechanical limits. First, a background on recent efforts in control of building energy systems is given, focusing on economic optimization. Then, information about the building and development of the steady-state control method are given, followed by the simulation methods. Simulation results are then presented, followed by a discussion of the importance of occupant comfort in building control strategies. The paper ends with a discussion of the study's outcome as well as future work.
