Multiplexed Real-time Optimal Control of Overall Heating, Ventilation, and Air-Conditioning Systems
DescriptionAwareness of climate change has resulted in international pressure to reduce the energyuse of buildings. As heating, ventilating, and air-conditioning (HVAC) systems representthe largest primary energy end-user, generally consuming over 30% of the total energyused in commercial buildings, it is important to reduce the energy use of HVAC systems.Among the many tools that are used to improve the energy efficient operation of HVACsystems, real-time optimal control (RtOpt) is a very efficient option that regularlyoptimizes or resets the set-points or decision variables for local control loops regardingthe energy use. Although RtOpt of overall HVAC systems can reduce energy use morethan RtOpt of individual local loops or subsystems, it presents a large-scalemathematical programming issue, including the challenges of considering many decisionvariables and a large number of nonlinear models and complex constraints. Currently,the RtOpt of overall HVAC systems involves optimizing all decision variablessimultaneously at each decision-making time using genetic or other evolutionaryalgorithms. There are many difficulties with the conventional optimization mechanism,including a heavy online computational burden, and scenarios for which optimalsolutions may not be found. The stability and robustness of the system under overalloptimization need to be systematically analyzed.This project aims to provide a new strategy for the RtOpt of overall HVAC systems thatis more stable and robust than the conventional optimization mechanism and thatsuffers less from online computational burden, without sacrificing energy performance.To this end, we will develop a new RtOpt mechanism, namely multiplexed real-timeoptimal control (MRtOpt). Unlike the conventional mechanism, the new mechanismoptimizes only one decision variable at each decision-making time with a higheroptimization frequency. Each decision variable is optimized sequentially following apredefined order. Preliminary studies show that the new mechanism can significantlyreduce computational burden (over 98%) without sacrificing energy performance (6.7%energy saved by the conventional method vs. 6.8% by the new method). After ourencouraging preliminary results, in this project we will further develop MRtOpt strategyregarding modeling, stability, and robustness. The main tasks are to identify the mostsuitable modeling method for MRtOpt, improve the stability, and enhance therobustness. The output will be a complete MRtOpt toolbox with guidelines, which wouldhelp HVAC professionals to understand and implement the MRtOpt strategy.
|Effective start/end date||1/01/16 → 18/12/19|
- Multiplexed Optimizaiton,HVAC,stability,robustness,