A large share of variable generation will be a common characteristic among power grids worldwide, due to efforts of reducing greenhouse gas emissions and exploiting renewable resources such as wind and sun. In this context, sources of flexibility will be key assets to compensate the inherent volatility and uncertainty of these sources and achieving the required balance between generation and demand at every time.

In this talk, we will provide a general overview of the idea of using the demand side as a source of flexibility in Smart Grids. As an application of the concept, we will focus on the Vehicle-to-Grid (V2G) paradigm in which a fleet of electric vehicles (EV) provides balancing services to the grid by using the storage capacity of the fleet during the EV idle time.

This work is part of a pilot project currently being implemented in Los Angeles, California that aims to demonstrate the concept of Vehicle-to-Grid (V2G) with an operational fleet. The objective is to allow the fleet to participate on Ancillary Services (AS) markets by using the energy storage capacity the fleet has during the Electric Vehicles (EVs) idle time. In California, performance-based payments are being introduced for AS. In the case of frequency regulation, a fast and accurate response is compensated alongside capacity payments.

We will present real-time controllers for the EV fleet that considers bidirectional-charging efficiency and we extend it to study the effect of looking ahead when implementing Model Predictive Control (MPC) schemes. Simulations show that the controller improves tracking error as compared with benchmark scheduling algorithms, as well as regulation capacity and battery cycling. Future challenges and research opportunities will also be discussed.

The large-scale integration of non-conventional renewable energy (NCRE) sources into electrical power systems poses important questions regarding the validity of traditional control, operation and long-term planning models. In particular, the inherent uncertainty and intermittency of wind and solar-based generation increase the need for flexibility in the systems; thus, they must be properly modelled in system planning models.

Traditional generation and transmission planning models make use of load duration curves, in which the demand in a year is modelled using a few non-chronological representative blocks. In presence of intermittent and hard-to-predict resources, chronological hourly behavior becomes much more relevant due to potential violation of ramping and commitment constraints; however, this effect cannot be represented by traditional models. In order to account for hourly behavior, planning models must incorporate approximate representations of dispatch and unit commitment constraints, which typically results in intractable formulations. In this talk, we discuss the use of a tight linear approximation of the unit commitment problem, obtained from the literature on Convex Hull Pricing, in a generation and transmission planning model. The proposed model is tested in reduced versions of the New England and Chilean power systems. The results show an improved performance of the proposed formulation compared to traditional planning models in terms of optimality of the solutions, and shorter convergence times compared to the model with an exact representation of commitment constraints.