2016-2017 Funded Proposal
This research will develop multi-scale models to optimize the performance and reliability of the power grid. Power systems are evolving to a scenario with a higher penetration of renewable energy to reduce global warming and fossil fuel dependency. For example, the DOE is projecting that by 2050 up to 35% of the United States’ electricity could come from wind generation alone. Such a high penetration of renewable generation would be extremely beneficial to the environment. However, renewable generation is much more variable than traditional generation. Maintaining a reliable power grid with so much uncertainty will require the development of new mathematical models and optimization techniques (something the DOE is keenly aware of). This research will enhance UTK’s ability to do just that, and in doing so, will develop a multidisciplinary team that will be able to compete for future research funds.
The difficulty of incorporating unreliable renewable generation is felt at all scales in power-systems operation. A high penetration of renewable energy reduces the grid’s robustness. Robustness is understood as the system capacity to withstand severe perturbations while keeping the vital system variables at their acceptable levels, e.g., acceptable frequency levels after a single or multiple generator outage. Two possible strategies to keep an acceptable level of system robustness are:
- The deployment of large storage systems. This requires several storage systems at different points of the bulk power grid, but there are still many unsolved issues related to their ideal location, their optimal capacities, and their adequate controlling mechanisms.
- The provision of fast demand response. This can be obtained by enabling distribution systems to respond as a whole in real-time and provide ancillary services to the bulk power system. To this end, distribution systems need to be equipped with a real-time control system to provide regulating capabilities in short-term. These capabilities can be obtained by coordinating demand side response (adjusting buildings HVAC controllers while keeping occupant comfort), distributed storage systems, and distributed generation.
At the minute time scale, distributed generation such as residential solar panels creates voltage fluctuations due to their generated power variability. These fluctuations can affect the quality of the delivered power, and in some cases affect its reliability as these fluctuations may trip protection devices. With a higher penetration of renewable energy, conventional regulating devices would not have enough capabilities to mitigate these large fluctuations. Therefore, new controlling mechanisms must be explored and designed. Some solutions are to explore new regulating capabilities from distributed storage systems, or enable buildings to have demand response to grid signals, or design new controllers at the renewable generators themselves to smooth their power output in case of excessive variability. Solving this issue will increase distribution system resiliency and weather conditions would have a minimal negative impact on the system.
Unfortunately, the decisions made at the generation level will have a tremendous impact on the grid performance at the distribution level, and vice versa. What is needed is a multi-scale mathematical model that incorporates decision making at the planning level with decision making at the transmission level. This research will develop multi-scale optimization methods that incorporate these different time scales. Examining these models will provide better intuition into (a) the impact and relative value of different capacity expansion strategies (i.e., the value of storage systems versus demand response controls) as well as (b) the algorithmic/computational challenges that will need to be addressed. The data obtained from these results will help motive future proposals.
Work Plan: The research team will initially consist of five members: PI Pulgar, PI Ostrowski, an EECS graduate student, an ISE graduate student, and one ISE undergraduate. The graduate students will work together to develop the mathematical models and optimization techniques. The team will hold weekly meetings to discuss progress and to discuss current literature. In addition, the team will regularly meet with potential collaborators at ORNL and elsewhere (possibly TVA and EPRI).