Approaches to Enable Demand Response by Industrial Loads for Ancillary Services Provision

Demand response has gained significant attention in recent years as it demonstrates potentials to enhance the power system’s operational flexibility in a cost-effective way. Industrial loads such as aluminum smelters, steel manufacturers, cement plants, and air separation units demonstrate advantages in supporting power system operation through demand response programs, because of their intensive power consumption, already existing advanced monitoring and control infrastructure, and the strong economic incentive due to the high energy costs. In this thesis, we study the three aforementioned manufacturing processes each with its own capabilities and constraints. We provide approaches to efficiently integrate each of these types of manufacturing processes as demand response resources. Aluminum smelting is an energy-intensive electrolytic process that is widely used to produce aluminum. The electricity cost thereby constitutes a significant portion of the total operation cost. At the same time, the smelting process is able to change its power consumption both accurately and quickly by controlling the pots’ DC voltage, without affecting the production quality. Hence, an aluminum smelter has both the motivation and the ability to participate in demand response. First, we focus on determining the optimal regulation capacity that such a manufacturing plant should provide to maximize the combined profit from producing aluminum and providing regulation. The approach is based on stochastic programming and the stochastic variable is the regulation signal sent to the smelter. Next, we focus on determining the optimal bidding strategy in the day-ahead energy and spinning reserve markets for an aluminum smelter. By bidding into the electricity market, the smelter provides flexibility to the power system operator and gets compensation which reduces the overall electricity cost. Finally, we focus on industrial loads which provide both energy and regulation service and study the optimal bidding strategy to maximize their revenues in the day-ahead markets. The approach is based on stochastic programming with a set of possible price curves as input, which represents the possible price scenarios for the next day. We also use a recently proposed method called the Multiple Quantile Graphical Model to represent the price distributions and use Gibbs sampling to sample the price curves. Electric arc furnaces (EAFs) in steel manufacturing consume a large amount of electric energy, and the energy cost constitutes a significant proportion of the total costs of producing steel. However, a steel plant can take advantage of time-based electricity prices by optimally arranging energy-consuming activities to avoid peak hours. Besides, the EAFs’ power rate can be adjusted by switching transformers’ taps, which offers additional flexibility for arranging energy consumption and minimizing the cost of electricity. We first propose scheduling methods that incorporate the EAFs’ flexibilities to reduce the electricity cost. However, the scheduling of steel plants is very