Building HVAC system with multi-level model predictive control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The deployment of model predictive control (MPC) systems in HVAC industry is hindered by the complexity of managing large numbers of building zones, leading to impractical single optimization problems that are difficult to solve in real time, especially in campus-wide implementations with hundreds of zones.
Innovation Solution
A distributed MPC system is implemented, featuring a high-level MPC that generates optimal indoor subsystem load profiles and low-level indoor MPCs that optimize setpoints for indoor VRF units, decomposing the overall problem into manageable sub-problems to reduce computational complexity and enable real-time solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single combined control system is used for campus-wide implementations with hundreds of zones, then comprehensive control coverage is achieved, but the optimization problem becomes too large to solve in real time
Solution Approach 1:
The control system is divided into a high-level MPC that solves a reduced optimization problem for overall load management and multiple low-level MPCs that solve local optimization problems for individual zones or equipment. This segmentation allows the large campus-wide control problem to be decomposed into smaller, computationally manageable sub-problems that can be solved in real-time while maintaining comprehensive control coverage across all zones.
2Loss of energy
If traditional MPC is applied to large numbers of building zones, then optimal energy cost control is achieved, but computational complexity becomes impractical for real-time deployment
Solution Approach 1:
The optimization problem is segmented into two levels: a high-level problem that determines optimal load profiles and setpoint trajectories for groups of zones, and low-level problems that determine specific equipment setpoints. This reduces the dimensionality of the optimization problem at each level, making it computationally tractable for real-time deployment while still achieving energy cost optimization.
Solution Approach 2:
The control architecture introduces a temporal dimension by solving optimization problems over a prediction horizon, allowing the system to anticipate future energy prices and loads. This predictive capability enables load shifting strategies that reduce peak energy costs while maintaining comfort, achieving optimization without requiring excessively complex real-time computation.
Data Source
AI summary
A heating, ventilation, or air conditioning (HVAC) system for a building includes indoor subsystems, a high-level controller, and low-level controllers. Each indoor subsystem includes one or more indoor units configured to provide heating or cooling to one or more building spaces. The high-level controller generates a plurality of indoor subsystem energy targets, each indoor subsystem energy target corresponding to one of the plurality of indoor subsystems and generated based on a thermal capacitance of one or more building spaces to which heating or cooling is provided by the corresponding indoor subsystem. Each low-level indoor controller corresponds to one of the indoor subsystems and generates indoor setpoints for the one or more indoor units of the corresponding indoor subsystem using the indoor subsystem energy target for the corresponding indoor subsystem and operates the one or more indoor units of the corresponding indoor subsystem using the indoor setpoints.