Comparison of a Daily and Sub-Daily Optimization Model for Hydropower Generation: Does Greater Resolution Result in Greater Operational Insight?

Optimization models have a long history of application in reservoir operation studies. In order to operate discretely, models assume a time-step in which release decisions are made. Designing models with long time-steps has the advantage of reducing model complexity and accommodating data availability limitations. For some applications, such as hydropower generation, a long time-step of daily duration may not be sufficient to model desired system operations. Even though most release requirements are often daily, sub-daily decisions are adopted for hydropower operation. Hydropower reservoirs commonly make releases based on energy prices, which fluctuates on a sub-daily basis. Sub-daily releases are made at the discretion of the dam owner or operator within the physical constraints of the system. This paper compares a daily and a 1 hour linear optimization model of the Deerfield watershed, located in northwest Massachusetts and southern Vermont, to examine how operational patterns change due to model time-step, i.e. alternate release strategies based on the daily or hourly fluctuation of energy prices. The Deerfield watershed contains ten significant reservoirs, including one storage reservoir and nine hydropower reservoirs. The models are compared by examining revenue and releases generated from seasonal model runs. Modelers and planners can use increased generating revenue achieved by 1 hour operation and altered release strategies as a measures of the importance and insight derived from sub-daily modeling.