Forward Co-located BESS Capture Curves

To create a forward hourly price, an additive decomposition methodology is applied to decompose the fundamental seasonality and volatility properties of the spot electricity price for each time step into three components:

Deterministic component: This represents the seasonality trend (including weekly and daily pattern) of the electricity prices.

Stochastic spike component: This represents the price spikes due to unexpected drastic changes to the electricity supply and demand (e.g. unexpected maintenance of a nuclear power plant).

Stochastic residual component: This represents the residuals after removing the deterministic component and the stochastic spike component from the spot electricity price. This part could be a result of short-term changes in supply-demand relationships due to, for example, weather changes.

The figure below shows an example of the additive decomposition methodology:

The statistical characteristics of these three components are first analyzed using historical data and then estimated for future years based on the projected supply and demand curves for the market zone considered. The methodologies used to estimate these three components are briefly described in the following subsections.

The deterministic component is estimated using the hourly price forward curve (HPFC) multiplied by the monthly baseload price. It is assumed that the HPFC structure is mainly influenced by the residual demand level, which is equal to the demand minus the renewable generation. The cannibalization effect brought on by increasing renewable penetrations is structured into the deterministic component for future hourly prices. The impacts of the renewable penetration on the residual demand and finally the price structure are quantified using the historical data. Then for any future year, the HPFC is built based on the quantified relationship and the projected residual demand level changes of the corresponding year.

The estimation of the stochastic spike component is further broken down into the estimation of the spike probability and the spikes size. For each time step, the probabilities of having positive and negative spikes are estimated using two conditional probability functions given the renewable market share (i.e. the ratio of the renewable generation to the demand of the market zone). Then the sizes of positive and negative spikes are sampled from two probability distribution functions constructed based on historical data.

For the stochastic residual component, an autoregressive model is applied which is based on the strong time-correlation identified for the stochastic residual components between different time steps.

As the names of these three components suggest, scenarios of the forward hourly price are generated as a result of the stochastic process included in the estimations of stochastic spike and residual components.

A stochastic optimization model is built to incorporate the uncertainties in the forward hourly prices by generating a set of forward hourly prices where for each a given Co-located BESS configuration is optimized against it. The result is a range of forward Co-located BESS capture factors which are then averaged separately for each case (High & Low) and are displayed as the ratio of realized effective renewable energy price of the Co-located BESS system.