Please use this identifier to cite or link to this item:
Title: Sensitivity and uncertainty analysis, and robust optimal control strategies for air-conditioning systems with low quality measurements
Authors: Shan, Kui
Degree: Ph.D.
Issue Date: 2013
Abstract: Heating Ventilating and Air Conditioning (HVAC) systems consume significant portion of energy in buildings. The control of HVAC systems is critical to building energy efficiency and indoor environment. Nearly all the control systems require the real time operation information of buildings. Such information includes weather condition, ON/OFF status of equipment, and operation parameters (e.g. indoor air temperature, evaporator chilled water temperature, chiller power, etc.). The information is normally acquired through sensors. Some of the information may not be available in the real application due to lack of sensors, while some available information may be of low quality. Quality of measurements is considered from two aspects, i.e. the accuracy and precision. Previous researches on the optimal control of HVAC systems found in the literature are mainly based on the assumption that the information required are available and of high quality. However, in the real application, it is not the case. Therefore, it is required to research on optimal control of HVAC system considering the availability and quality of required information. This study considers the problem in two cases, i.e. the information needed is not available and the information used is of low quality. In the case of lack of information, a new strategy using a limited number of sensors is developed for demand-controlled ventilation (DCV) of multi-zone office buildings. In the case of low quality information, two kinds of work are conducted. Firstly, the impact of measurement uncertainties on the control performance is analyzed. Such analysis is conducted on DCV strategies and optimal control strategies for condenser cooling water systems. Secondly, a control strategy with enhanced robustness for condenser cooling water system is developed to handle the problem of uncertainties in measurements and models.
The conventional CO₂-based DCV strategy for multi-zone offices requires CO₂ sensors and supply air flow meters being installed in all zones. However, in many practical cases, CO₂ sensors in individual zones are not available. In order to control the outdoor air flow based on actual occupancy variations in such cases, a DCV strategy with two implementing schemes is developed for different sensor availabilities. The first scheme is used for the situation when CO₂ sensors in individual zones are not available but the air flow meters for individual zones are installed. The second scheme is used for the conditions when both CO₂ sensors and air flow meters in individual zones are not available. These two schemes use two different approaches to estimate the outdoor air flow fraction of the critical zone approximately. Both schemes only require the CO₂ sensor in the main return air to dynamically detect the total occupancy number. The developed strategy is implemented and validated in a high-rise office building in Hong Kong. The site test results show that the strategy can achieve significant energy saving while maintaining acceptable IAQ in the situations where only very limited number of sensors are available. The impacts of the measurement sensitivity and uncertainty on HVAC systems controls are analyzed in this study. The sensitivities of measurements in the control strategies are identified by using the sensitivity analysis method. Those measurements which have high sensitivities are called critical measurements. The uncertainty analysis provides information for further risk analysis. A modified exponentially weighted moving average (EWMA) filter is developed and adopted to reduce the uncertainty of the control strategies. The sensitivity and uncertainty analysis is conducted on two demand-controlled ventilation strategies for multi-zone office buildings. Improvements of the control strategies based on the analysis results are also achieved. Results show that the sensitivity and uncertainty analysis is effective in evaluating control strategies. The accuracy and uncertainty of the control strategies are significantly improved based on the analysis results. A method to enhance the robustness of optimal control strategies is developed in this study. A fuzzy approach is adopted to predict the deficiencies in models outputs. Such predicted deficiencies are then used to correct the model outputs. The method is validated in an optimal control strategy for condenser cooling water systems. The operation data of a real building system are used to validate the deficiency prediction method. A simulation platform is built to validate the enhanced strategy. Measurement uncertainties are deliberately added to the simulated system for validation tests. Test results indicate that the method is effective in predicting the deficiencies in model outputs. Significant energy savings are achieved compared with the conventional optimal control method.
Subjects: Air conditioning -- Control.
Heating -- Control.
Ventilation -- Control.
Hong Kong Polytechnic University -- Dissertations
Pages: xxv, 177 leaves : ill. ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of May 28, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.