Adding renewables to the grid: Effects of Storage and Stochastic Forecasting

The electricity sector contributes to a quarter of global greenhouse emissions, and managing its evolution is a critical sustainability challenge. The context for the development and operation of electricity grids has dramatically changed in recent years. Wind and solar power have become much less expensive. Lower costs combined with increased policy action to address carbon emissions is leading to substantial shares of electricity generated by intermittent renewables. Maintaining a stable electricity supply with intermittency is a critical challenge; storage and natural gas are possible solutions. While policymakers promote storage as green grid technology, low-cost natural gas from hydrofracturing extraction raises the economic hurdle for storage. Researchers have developed complicated energy system models to help plan grids in the face of the above trends. The research in this dissertation introduces new modeling features that affect the economic and environmental outcomes of the adoption of renewable and storage technologies. First, prior models that explore the future build-out of electricity grids are nearly always deterministic, i.e., they assume that decision-makers have perfect information. Here a stochastic optimization grid expansion model is developed that presumes that expected future fluctuations, e.g. in fuel prices, influence build-out decisions. This stochastic model thus includes uncertainty and risk as core elements: Grid build-out depends on the distribution of system costs. A genetic algorithm with Monte-Carlo simulation is used for co-optimization using two objective functions: “risk-neutral,” which optimizes to minimize average system cost and “risk-averse,” which optimizes to minimize average of the top 5% of costs (also called 95% Conditional Value at Risk (CVaR)). This model is tested for the US Midwest regional grid. The results show that the risk-averse scenario does not increase mean system costs but adds significantly more wind. These results corroborate prior work showing that electricity system costs can be surprisingly inelastic to renewable adoption and further introduces quantification of how increased renewables lowers cost risk. Second, the economic and environmental performance of storage is complicated by how its introduction affects the operation of both renewable and fossil plants. In this dissertation, a model is developed that accounts for how storage operation would affect prices on the grid and in turn, the operational schedule that yields optimal revenue. Results from modeling the US Midwest region shows that this treatment of storage as a “price maker” affects results. The model indicates that storage increases carbon emissions when it enables a high emissions generator, such as a coal plant, to substitute for a cleaner plant, such as natural gas. In this case, low cost; efficient natural gas generation is relatively better than coal to realize emissions reductions with storage under economic arbitrage until renewables dominate the grid mix. Third, the operational strategies of energy storage alter the generation and profits of the other electricity generation systems. The operational effects of storage on the change in generation is investigated for all the eGRID subregions across the US based on actual historical electricity prices and the generation mix for the year 2016. Results show that storage increases the coal generation and affects the natural gas generation in the west – except in California and the Midwest regions of the US; and increases the generation of the natural gas in the eastern US regions. California, upstate New York and New England regions show an exception with an increase in natural gas generation and decrease in coal generation. The model also investigates the operational effects of storage on the profits of other generating units in California, Midwest and New York regions. Profits of other generating units are significantly affected when large capacities of storage operate as price-makers. Coal has a small increase in profits by 2% and all the other fuels continue to see a decline in profits in New York and the Midwest regions. The decrease in profits of the other generating units is because of the offset/retirements of the peaker natural gas plants that set the electricity prices. On the other hand, in California, the profits for renewables increase from the increase in electricity clearing prices set by the natural gas combined cycle plants to meet the additional demand from the storage charging