Conference Year
July 2018
Keywords
Optimization, Model Predictive Control, Central Energy Facility, Electrical Storage
Abstract
On large campuses, energy facilities are used to serve the heating and cooling needs of all the buildings, while utilizing cost savings strategies to manage operational cost. Strategies range from shifting loads to participating in utility programs that offer payouts. Among available strategies are central plant optimization, electrical energy storage, participation in utility demand response programs, and manipulating the temperature setpoints in the campus buildings. However, simultaneously optimizing all of the central plant assets, temperature setpoints and participation in utility programs can be a daunting task even for a powerful computer if the desire is real time control. These strategies may be implemented separately across several optimization systems without a coordinating algorithm. Due to system interactions, decentralized control may be far from optimal and worse yet may try to use the same asset for different goals. In this work, a hierarchal optimization system has been created to coordinate the optimization of the central plant, the battery, participation in demand response programs, and temperature setpoints. In the hierarchal controller, the high level coordinator determines the load allocations across the campus or facility. The coordinator also determines the participation in utility incentive programs. It is shown that these incentive programs can be grouped into reservation programs and price adjustment programs. The second tier of control is split into 3 portions: control of the central energy facility, control of the battery system, and control of the temperature setpoints. The second tier is responsible for converting load allocations into central plant temperature setpoints and flows, battery charge and discharge setpoints, and temperature setpoints, which are delivered to the Building Automation System for execution. It is shown that the whole system can be coordinated by representing the second tier controllers with a smaller set of data that can be used by the coordinating controller. The central plant optimizer must supply an operational domain which constrains how each group of equipment can operate. The high level controller uses this information to send down loadings for each resource a group of equipment in the plant produces or consumes. For battery storage, the coordinating controller uses a simple integrator model of the battery and is responsible for providing a demand target and the amount of participation in any incentive programs. Finally, to perform temperature setpoint optimization a dynamic model of the zone is provided to the coordinating controller. This information is used to determine load allocations for groups of zones. The hierarchal control strategy is successful at optimizing the entire energy facility fast enough to allow the algorithms to control the energy facility, building setpoints, and program bids in real-time.