Smart Cloud-Based Energy Management for Smart Buildings

Thesis Abstract

The rapid digitization of building management systems and the growing demand for energy efficiency have intensified the need for integrated, scalable solutions that leverage cloud computing to optimize energy use in smart buildings. This study investigates a Smart Cloud-Based Energy Management (SCBEM) framework designed to reduce energy consumption, enhance occupant comfort, and lower operating costs through real-time data analytics, predictive control, and scalable cloud services. The specific objectives are (1) to design an architecture for SCBEM that integrates heterogeneous building systems (HVAC, lighting, demand response, and renewable energy sources) with a cloud-based data pipeline; (2) to develop predictive models for occupancy patterns, weather-driven load, and equipment degradation using machine learning techniques; (3) to implement a cloud-enabled optimization engine that performs energy minimization under comfort and reliability constraints; (4) to evaluate the framework’s performance in a real-world setting through a quasi-experimental pilot in a university smart-building with 150 monitored zones; and (5) to assess the economic and environmental impacts via a cost-benefit and life-cycle assessment. A mixed-methods design is employed, combining quantitative data analysis with qualitative insights. The population comprises operational smart buildings within a metropolitan campus, with the primary sample consisting of one lead building (approximately 15,000 m², 6,000 daily occupants) and two auxiliary annexes for model validation. Data collection instruments include high-resolution sensor streams (temperature, humidity, CO2, occupancy, lighting, and electrical metering at 5-minute intervals), building management system logs, utility bills, and condition-based maintenance records. The cloud platform aggregates raw data, performs ETL processing, and hosts the predictive and optimization modules. Quantitative analyses involve time-series forecasting (SARIMA and LSTM for occupancy and weather-driven loads), regression-based fault detection, and a constrained multi-objective optimization formulated as a mixed-integer linear program (MILP) to minimize energy use while satisfying thermal comfort constraints (PMV/PPD) and equipment service-life considerations. Model validation employs k-fold cross-validation, back-testing against historical baselines, and sensitivity analyses. The qualitative component comprises semi-structured interviews with facilities managers to capture operational practices, adaptability, and perceived barriers, analyzed through thematic analysis. The expected findings indicate that SCBEM reduces peak building energy demand by 18–25% and total annual energy consumption by 12–20% relative to baseline operations, with improvements in occupant comfort metrics (perceived thermal comfort scores rising by 15% on standardized surveys). The predictive models are anticipated to achieve occupancy forecast accuracies within 8–12% RMSE and weather/load prediction within 6–9% MAPE, enabling the optimization engine to realize cost savings of 6–12% on annual energy expenditure and a corresponding reduction in CO2 emissions by approximately 10–15%. The study also expects to reveal practical organizational and data governance factors influencing SCBEM adoption, including data quality, latency tolerances, and the role of cloud service SLAs in maintaining system resilience during demand spikes. Contributions to knowledge include (i) a scalable, interoperable architecture for cloud-based energy management in complex building portfolios; (ii) integrated predictive-optimization methodologies that co-optimize energy use and occupant comfort under dynamic constraints; (iii) empirical evidence on the operational and environmental benefits of cloud-enabled energy management in institutional settings; and (iv) actionable insights into governance, data infrastructure, and change management necessary for deployment at scale. The study concludes that a cloud-centric, data-driven approach to energy management can deliver substantial energy savings and enhanced occupant experience without compromising reliability, provided that data quality, model interpretability, and robust cyber-physical security are ensured. Recommendations emphasize standardized data schemas, open interfaces for cross-system interoperability, staged deployment with continuous monitoring, and policy guidelines to incentivize proactive maintenance and demand-response participation.