Load Management and Demand Response in Small and Medium Data Centers
Year: 2018
Type: dissertation
Abstract: Data centers are the backbone of a growing number of activities in modern economies. However, the large increase of digital content, big data, e-commerce, and Internet traffic is also making data centers one of the fastest-growing users of electricity. The total energy consumption of data centers corresponded to almost 1.5% of the global electricity consumption and has an approximated annual growth rate of 4.3%. Therefore, it is very important to increase the energy efficiency in data centers with actions such as power usage management, server consolidation, energy efficient components and systems, as well as demand response programs and renewable energy sources. Small and medium data centers account for more than 50% of the total electricity consumption in this sector. In fact, surveys indicate that this data center profile waste more energy than larger facilities. Nevertheless, existing studies tend to be focused on the energy related issues for large data centers rather than small and medium data centers. Thus, this work aims to present how an intensive energy consumer, such as small and medium data centers, can become more efficient from the energy point of view and how they can take advantage of demand response programs to decrease costs and to cooperate with the grid to ensure higher reliability and sustainable development goals. For this purpose, a set of actions have been taken in pursuit of this objective. Firstly, a meticulous state-of-the-art literature review of data centers energy efficiency and demand response perspectives was performed, providing fundamental information about the complex and technological world of small and medium data centers, in order to support the knowledge of their flexible load. An energy efficiency survey directed at Brazil, Portugal and the United States was also conducted, showing an alarming reality regarding energy efficiency practices within small and medium data centers. In the same context, an impact and potential assessment on energy efficiency towards small and medium data centers was performed allowing to observe the main advantages, disadvantages and applications in three important methodologies. A design of load management strategies and opportunities for demand response was developed considering a framework with two layers encompassing an energy efficiency methodology, the management of flexibility and the deployment of demand response scenarios with dynamic tariffs or an incentive-based contract. Alongside, the small and medium data centers load mathematical modeling and specification of optimization processes were provided within the same scope with two algorithms implemented with different approaches; one focused on small and medium data centers operators, and other dedicated to distribution system operators. At the data center level, a mixed-integer linear programming optimization technique is used to control flexible loads (considering ICT workload, cooling and UPS), in order to reduce the cost function in a price-based outline or to match the load reduction requirements in an incentive-based outline. Concerning the distribution system operator, a random-rotation and fairness algorithm selects the data centers to be used in each demand response event. Such algorithm deals with small and medium data centers in an aggregated and equitable way by drawing on their joint and flexible loads in contractual terms, as if they were one large data center. Finally, the optimization results were firstly demonstrated in a small and medium data centers perspective, as well as their respective scenarios, in which it has been proven the algorithm operation and reliability. The potential for cost savings in demand response was also proved, being achieved with the considered incentives in the simulation process, savings of 1.33% for small and 5.15% for medium data centers in the incentive approach, and 0.21% for small and 0.68% medium data centers in the dynamic tariff approach. Subsequently, the same premises were utilized with focus on the distribution system operator point of view, enabling to predict specific contractual policies that can be adopted in this type of relationship through the best and the worst scenarios simulations. The scenario with a preponderance of incentives stimulates the adoption of demand response programs applied to small and medium data centers, nevertheless the unchanged and penalty cases enable to forecast an unfavorable scenario for this sort of demand response program. In this context, on one hand, the one single day analysis showed a variation in the reduction potential between 18.02% and 91.16%, while the highest value in the penalty profile was 356 € and in the incentive profile 10,580 €. The 22 business day analyses presented a fluctuation in the reduction potential in the range of 13.12% and 71.38%, whilst the highest value in the penalty profile was 1,200 € and in the incentive profile 84,700 €.
Author: Thiago Lara Vasques
Institution: University of Coimbra
Language: English
Topic: Smart Grid Energy Management
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2902314265
