A decentralized algorithm for optimal distribution in HVAC systems

• A typical optimal group operation problem named optimal distribution problem is concluded and defined.
• A novel idea of decentralized control method is presented to solve the optimal distribution problem.
• Devices in chiller plant become smart devices and could collaborate with each other in decentralized system.
• Simulation results demonstrate the decentralized control algorithm to be near optimal.
• Decentralized hardware platform is developed and tested by applying the decentralized control algorithm.

This paper introduces a novel decentralized method as a solution to a typical problem in HVAC systems, the optimal distribution problem, which can be described as the optimal control of equipment groups such as the pump and chiller groups, with the goal of attaining optimum allocation in response to a given demand under actual constraints. In this decentralized control system, each unit is fitted with a decentralized controller such that it becomes a smart unit, such as a smart pump or a smart chiller, which can communicate with the other units and thus operate collaboratively to satisfy the control requirements and be energy efficient. The concept of decentralized control is introduced first, then a decentralized optimal control algorithm, which could be applied to the group operation of both pumps and chillers, is developed and validated through the simulation of different cases, and further demonstrated through application to actual hardware. Compared with the traditional centralized control method, this decentralized control method is much more flexible and scalable.