Impacts of artificial intelligence (AI) on building energy usage : case study of an office building in Thailand | |
| Author | Kornpatchara Sangdokmai |
| Call Number | AIT Thesis no.SE-22-01 |
| Subject(s) | Building--Energy conservation--Thailand Artificial intelligence Intelligent buildings--Thailand |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering Sustainable Energy Transition |
| Publisher | Asian Institute of Technology |
| Abstract | Building energy demand in 2020 is responsible for 55 percent of the total world energy demand and continuously increase due to the increase in floor area and cooling demand. In Thailand, the building accounted for 52.8 percent of total electricity consumption in 2021. In tropical regions, heating, ventilation and air-conditioning account for 60% of total energy consumption in the commercial building and approximately 66 per cent of total electricity consumption is caused by the operation in the building, of which 30 percent is wasted due to the inefficient operation. This study aimed to identify the electricity reduction potential of the integration of AI techniques in the building and the strategies to integrate intelligence control in the building to provide good thermal comfort, responding to different weather conditions. There are three scenarios that this research conducted, including (i) scenario 1 as baseline, (ii) scenario 2 as simple energy management, and (iii) scenario 3 as advanced energy management with AI. Using building simulation modelling, this study analyzed the integration of energy management strategies with AI techniques that provide satisfying comfort and measured scenario 3 as an validation in the office building. The methodology included four main aspects, including literature review, building simulation with 3 scenarios, validation on scenario 3 and analysis to essentially study 2 main topics, i.e., (a) human comfort conditions (thermal comfort and indoor air quality), (b) electricity consumption and energy performance indication (EnPI) was used to find the energy-saving following ISO 50006. This study focuses on the office building at B.Grimm Power company which has four office buildings. This research considers the Garden Wing building that is located in Bangkok, Thailand. There are four general steps for scenario 1 as a baseline, including data collection, analysis, building energy modelling, and calibration. The building electricity consumption in the year 2021 was 101,098 kWh and the average electricity consumption was 8,425 kWh per month. The specific electricity consumption in this building was 85.23 kWh/!!/year which was lower than the building energy code (BEC) standard, i.e., 171 kWh/!!/year. The scope of this research considered building electricity consumption in October 2021 which was gathered hourly data as a baseline. It can be observed that the overall thermal comfort of the building was not acceptable by ASHSRE standard 55-2010 that the building provides the recommended comfort zone of only 13.77% (34 hours out of 247 hours) of the business time (6 a.m. to 6 p.m.) in October 2021 while indoor air quality (IAQ) was acceptable which is lower than 1,000 ppm. Then, building energy modelling is required with the acceptable calibration criteria (the mean bias error (MBE) and the coefficient of variation of the root mean square error CV(RMSE)) following ASHRAE Guideline 14. According to monthly data, the NMBE and CV(RMSE) of whole building electricity consumption in 2021 were 0.23% and 4.96% respectively while both NMBE and CV(RMSE) MBE must be ± 5% and ≤ 15%. Then, simple energy management and advanced energy management with AI are implemented with the building energy model. According to scenario 2, a simple energy management, this study conducted interruption technique, early switch on/off, and pre-cool technique to implement with air-conditioning and ventilation system. As a result, scenario 2 provided good comfort that was 209.75 hours out of 247 hours (84.92%) and good average carbon dioxide levels of between 400-1000 ppm. In terms of electricity consumption, total electricity consumption was 6,762.39 kWh and the building energy was saved by 16.5 % in October 2021. Scenario 3 as advanced energy management with AI was implemented with building energy modelling and compared with real building. There are two main strategies that AI controls, i.e., (i) the new-setpoint temperature of air-conditioning, and (ii) smart control of ventilation systems. According to the simulation result, comfort hours in the building were 243 out of 247 hours (98% of the business time). Indoor air quality (IAQ) was in the range of 400-1200 ppm of CO!. which provides 97.4% below 1000 ppm. In terms of electricity consumption, total electricity consumption was 6,611.61 kWh and the building energy saved by 18.5 % in October 2021. According to the validation result, all relative data was collected in March 2022. The comfort recommendation was 142.50 out of 299 hours (47.66 %) and slightly warm was 104.25 hours or 34.87% of the total. Moreover, the indoor air quality (IAQ) was in an acceptable range of between 400-1,400 ppm of CO!. Total electricity consumption was 11,648.60 kWh. Based on Energy performance using energy baseline (EnB) and energy performance indicators (EnPI) that are established by ISO 50006:2014, the energy can be saved by 13.41%. The problem with scenario 3 is that it would require a different building design than what is currently available. AI needs more time to collect more data and improve learning methods. Depending on the training data, processor, and system specification, the time and space constraints of machine learning models vary. |
| Year | 2022 |
| Type | Thesis |
| School | School of Environment, Resources, and Development (SERD) |
| Department | Department of Energy and Climate Change (Former title: Department of Energy, Environment, and Climate Change (DEECC)) |
| Academic Program/FoS | Sustainable Energy Transition (SE) |
| Chairperson(s) | Salam, P. Abdul |
| Examination Committee(s) | Kumar, Sivanappan;Singh, Jai Govind;Charat Mongkolsawat |
| Scholarship Donor(s) | Loom Nam Khong Pijai (Greater Mekong Subregion) Scholarships |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2022 |