Author: Lai, Chung Nok Title: Characterize long term CO₂ concentrations in an office environment (data analysis & computational work) Advisors: Wong, Ling Tim (BSE) Degree: M.Eng. Year: 2021 Subject: Indoor air qualityCarbon dioxide -- Environmental aspectsAir conditioningIndoor air pollutionHong Kong Polytechnic University -- Dissertations Department: Department of Building Services Engineering Pages: 67 pages : color illustrations Language: English Abstract: Indoor Air Quality has drawn more attention in the society, and global CO2 problem has become extremely series. The indoor CO2 concentration level has been pointed out that several problems are derived from this air parameter. Health problem and working performance are the two predominant benefits gained from dealing with high CO2 concentration in a indoor environment. In view of this, computational analysis seems to be necessary when it comes to enhance the performance of a HVAC system. There are many parameters potentially included in time series modelling, however, sorely CO2 concentration changes over time in the office are covered as the parameter. Box-Jenkins methodology has been adopted as the time series analysis, in which an iterative three stages modelling is involved. Model identification, parameter estimation and statistical model checking are employed so as to conduct corresponding tests on stationarity, seasonality, best fit coefficients for ARIMA, and the stationary univariate. Further explaining the modelling approach, a number of tests have been carried out prior to ARIMA modelling. Identifying the parameters by comparing the ACF and PACF plots, stationarity tests by ADF test, Q-Q plots are examples that ensure the assumption behind the ANOVA tests and the modelling premises. To minimize the work size, locations 4,11,14 are chosen as the most representative locations under comparison in ANOVA tests. The corresponding relationship of 17 locations are in comparison according to the statistical significance. In order to study the data adequacy, using longer time intervals are also one of the objectives to be achieved. Data ranging from 1 to 3 weeks has been employed and computed for residual analysis such that the impact of time duration can be clearly observed. Since the work is confined by several limitations, there are more computational work and further research expected to be conducted as follow-up work after this research. It is ideally anticipated that electricity consumption will be covered or other environmental impacts could be interpreted from the study of time series models. Rights: All rights reserved Access: restricted access

Files in This Item:
File Description SizeFormat
5729.pdfFor All Users (off-campus access for PolyU Staff & Students only)2.03 MBAdobe PDF

Please use this identifier to cite or link to this item: `https://theses.lib.polyu.edu.hk/handle/200/11253`