Boolean Microgrid poster.pdf
The major goal of this project is to investigate ecient, economic ways to integrate renewable energy resources.
The twin goals of maximizing the utilization of variable renewables while reducing the need for costly operating reserves necessitates ecient demand response in the smart grid. Technological developments in sensing, communication and control infrastructure have made feasible the employment of such demand response by load serving entities. It is desirable that such demand response not only be capable of smoothing out the variations in the power drawn from the grid, but that it also satisfy end users' comfort or other requirements by adhering to their thermal requirements (e.g., temperature).
Since many mechanisms envisage the use of real-time prices for demand response, it is important to understand how loads actually respond to price. One needs to understand whether and how they respond to moderate price increases or large price peaks. One also needs to understand their temporal behavior, since there may be delays involved in their response. This necessitates a dynamic system model of load responsiveness to real-time prices. Such a model will need to be developed from empirical data.
To integrate renewable energy sources into transportation, for example, electric vehicles, requires addressing of multiple coupled items including purchase, storage, pricing and scheduling. Energy will need to purchased at times that the real-time prices are low and stored. Such storage also is important in meeting peak demand without overloading the distribution system. Storage is however subject to capacity constraints as well as round-trip loss due to charging and discharging. The charging of electric vehicles will need to be optimized over time since users may have deadline constraints. Price discrimination may need to be used to separate users with tight deadlines from those with more exibility. Electric vehicle charging stations will need to perform all these coupled
optimizations.
To optimally run the power system requires situational awareness of the state of the power system. Synchrophasors that are being introduced can provide time-stamped measurements of voltages and currents, that permit determination of relative phase angles and other quantitative variables. However, they generate a large amount of streaming data, which is also high dimensional. This information will have to be rapidly assimilated in real-time to detect events. The data will also have to be made easily visualizable by human operators.
To optimally operate the overall power system, the system operator will need to maximize social welfare. This is the bene t of using power over a time interval minus the cost of generating it. Each load derives utility from consumption of energy, while each generator incurs cost, i.e., disutility, in providing it. Each load or generator may be subject to random disturbances, such wind, temperature, cost of fuel. They may also be subject to dynamic constraints such as ramping constraints, or more generally constraints on the power consumption/generation trajectory as a function of time. The optimization of the social welfare therefore gives rise to a complex, large scale, dynamic stochastic control problem with severe constraints on what information is available or provided by generators or loads to the system operator. It needs to be solved in a distributed manner through price-based mechanisms, through a bidding scheme run by the system operator. How such a scheme is to be designed is a major challenge to optimal utilization of renewable energy sources.