Linear and Convex Approach For Managing Data Centres Within Smart Grids: Incentivized Demand Response Implementation Perspective

The normal operation of a smart grid substantially relays on various types of data generated at different levels of the grids so that huge volumes of data are generated every single second within the grid. Handling theses data in terms of analysing them and creating the consequential controlling and protective decisions is usually carried out by the data centers that could result in significant energy consumption by the centers. Hence, a provider of data centers may opt for different approaches to manage its data centers in terms of, for example, the total energy consumption or operation cost. However, a significant challenge associated with such approaches is their complexity so that they conventionally rely on non-convex and non-linear techniques. In this paper, a novel formulation for managing data center energy consumption is proposed that addresses both the original problem's non-linearity and non-convexity. Furthermore, the operation of data centers must be modified to today's smart grid operational criteria. One of these specifications is an incentivized demand response strategy that the Distribution Systems Operator may employ in order to manage the grid's loading level. The novel proposed energy management methodology considers the possibility of implementation in a incentivized power market and articulates a convex and linear approach for data centers to participate in a incentivized power market. Furthermore, one of the prospective operation specifications for center installations in a modern power grid is the implementation of edge computing for the centers. The proposed framework will also reflect on this perspective as well.智能电网（smart grid）的正常运行高度依赖于电网不同层级产生的各类数据，因此电网内每秒都会产生海量数据。对这些数据开展分析并生成相应的控制与保护决策，通常由数据中心（data center）完成，而这一过程会导致数据中心产生显著的能耗。因此，数据中心运营商可能会采取多种方法来管理其数据中心，例如从总能耗或运营成本角度出发进行优化。然而，这类方法面临的一大挑战是其复杂性：传统上这类方法依赖于非凸（non-convex）且非线性（non-linear）的技术手段。 本文提出了一种用于管理数据中心能耗的新型建模方法，可同时解决原问题中的非线性与非凸性问题。此外，数据中心的运行需符合当前智能电网的运行标准。其中一项要求是配电系统运营商（Distribution Systems Operator）可采用的激励型需求响应策略，以调控电网的负载水平。所提出的新型能源管理方法考虑了在激励型电力市场中的落地可行性，并构建了一种凸线性方法，使数据中心能够参与激励型电力市场。此外，现代电网中数据中心部署的一项前瞻性运行要求，是为数据中心部署边缘计算（edge computing）。本文所提出的框架也将涵盖这一视角。