Use this URL to cite or link to this record in EThOS:
Title: Model predictive control of the HVAC system in industrial cleanrooms for energy saving
Author: Chen, S.
Awarding Body: University of Liverpool
Current Institution: University of Liverpool
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
20-40% of the total final energy consumption, which includes both residential and commercial users, is spent on buildings, and its amount has been increasing at a rate 0.5-5% per year in developed countries. The heating, ventilation and air conditioning (HVAC) system is widely installed in commercial buildings to provide thermal comfort and acceptable indoor air quality. Cleanrooms are one of the most common applications which have a high requirement of air cleanliness. About 40% of the total energy in a commercial sector is used for heating, cooling and ventilating the buildings' environment. This thesis deals with the reduction of the energy consumption of the HVAC system in industrial cleanrooms via model predictive control (MPC). A cleanroom laboratory has been built with the HVAC system to simulate a pharmaceutical factory where the MPC is implemented. Literature reviews of MPC approaches and its applications in HVAC systems are carried out. The schematic of the cleanroom laboratory has been investigated, including the detailed specification of the HVAC hardware and software. The laboratory consists of four rooms: the entrance room and three cleanrooms constructed with different levels of air cleanliness: the change room, the small lab and the large lab. The original control of the air ventilation is implemented by proportionalintegral (PI) control installed in a building management system (BMS). The data used for further applications are collected through the object linking and embedding for process control (OPC) server connecting the HVAC hardware with the OPC Toolbox in Matlab. A black-box model of the cleanroom laboratory has been developed based on the measured data. The indoor air quality of cleanrooms is maintained by controlling the air change rate and the air pressure via the HVAC system. Modelling of the cleanroom laboratory includes single-input single-output (SISO) modelling of each PI control loop and two multi-input multi-output (MIMO) subsystems: the change room related subsystem and the small/large lab related subsystem. Three parameter estimation methods: prediction error identification method (PEM), least squares (LS) method and instrumental variable (IV) method, and three model structures including autoregressive exogenous (ARX), state space (SS) and transfer function (TF) have been investigated, respectively. The model identification is implemented by the System Identification Toolbox in Matlab. For each system model, the model structure with the best performance index has been found by comparing the prediction results with the experimental results using model validation approach. The MPC controllers are designed based on the identified models to maintain the steady air change rate and air pressure. Both SISO and MIMO MPC are investigated. The original PI controllers are replaced by the SISO MPC controllers. The SISO MPC shows a better transient performance and lowers energy consumption. MIMO MPC controllers are necessary to use in the HVAC system since the HVAC system exhibits a MIMO nature with coupled controlled variables and the interactions are not negligible. The MIMO MPC shows better control performance and lowers energy consumption than SISO MPC and PI control. The closed-loop control of the particle concentration has been built to maintain the air cleanliness in the laboratory. The particle counters are installed in the cleanroom laboratory to monitor the number of particles within a specified air volume. The particle counter based controllers have been designed and tested including the PI control, the SISO MPC and the MIMO MPC, respectively. The comparison among these control methods shows that the MIMO MPC has the best performance and consumes the least energy. The PI control and MPC have been developed in programmable logic controller (PLC) devices. A PLC based industrial personal computer (IPC) has been installed to construct a workstation panel. The Matlab based controllers have been transferred into the PLC language. The PLC based controllers have been tested controlling the airflow, air pressure and the particle concentration. The test results demonstrate that the PLC based controllers perform better and spend less energy than those in Matlab.
Supervisor: Jiang, L. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral