Energy pricing is an important issue to consumers, utilities, and politicians alike. The ability to estimate future electricity costs can help all of these parties form a more informed plan for the future. In order to manage energy costs, electricity production should mirror demand or have the ability to be stored for later use. Electricity storage technology, however, is currently infeasible for large-scale applications in most places as it is expensive, still in development, and the infrastructure is not in place. With this in mind, we will focus on addressing the energy pricing problem with the former approach, fitting electricity production rates to predicted demand. To illustrate the importance of this project, consider the costs of surplus electricity. In 2013, the Canadian province of Ontario incurred losses over $1 billion to export its surplus to neighboring provinces and U.S. States. The government had contracts to purchase electricity for about 8.5 cents per kilowatt hour and was forced to sell it for 3 cents per kilowatt hour - less than half of that purchase price (CTV News Kitchener, 2014). A more accurate method of predicting demand could have saved Canadian taxpayers much of this cost.
The unit commitment problem involves finding the least-cost combination of power plants to have running in order to meet the electrical load. Outside of the constraint of meeting the demand, the main constraint involves the ramping time of different power plants, or how long it takes the plant to turn on. Smaller plants can typically ramp in about an hour, while large coal and nuclear plants require up to half a day and several days respectively. Additionally, some plants are not meant to be cycled frequently, as turning them on and off too often damages them. We created a binary programming model
incorporating these constraints to solve the unit commitment model of the various plants available for use.
We made the decision to test our unit commitment model on Kansas City Power and Light (KCPL), an electric utility serving 800,000 customers in 47 counties in Missouri and Kansas. The company operates 9 generating stations and 10 peaking plants. We chose KCPL because its monthly electricity demand history was publicly available through the Energy Information Administration (EIA).
The unit commitment problem involves finding the least-cost combination of power plants to have running in order to meet the electrical load. Outside of the constraint of meeting the demand, the main constraint involves the ramping time of different power plants, or how long it takes the plant to turn on. Smaller plants can typically ramp in about an hour, while large coal and nuclear plants require up to half a day and several days respectively. Additionally, some plants are not meant to be cycled frequently, as turning them on and off too often damages them. We created a binary programming model
incorporating these constraints to solve the unit commitment model of the various plants available for use.
We made the decision to test our unit commitment model on Kansas City Power and Light (KCPL), an electric utility serving 800,000 customers in 47 counties in Missouri and Kansas. The company operates 9 generating stations and 10 peaking plants. We chose KCPL because its monthly electricity demand history was publicly available through the Energy Information Administration (EIA).