Modeling battery dispatch behavior

This is a sample project modeling battery storage and dispatch behavior in the NYISO market. The goal is to understand how the example system might perform, the scale of expected profits, and how those profits might vary across the year.

Click on the "launch binder" button above to launch an interactive version of this repository and run code in a live environment.

Organization

Functions to read data and run the optimization are included in the src folder. The optimization and exploration of results are included in the Dispatch optimization notebook. The Price exploration notebook includes a few figures exploring patterns in hourly LBMP values over the course of 2017.

Project specifications

The goal here is to analyze the revenue generation from a battery storage system that is performing energy arbitrage by participating in the NYISO day ahead energy market. What are the market dynamics? I'm not trying to build a system that will actually optimize battery storage behavior going into the future, just building a quick tool to see how things have worked in the past.

System Requirements

1. The system SHALL optimize the battery storage dispatch (with an optimization time horizon of at least 1 day) for the day ahead energy market

• The battery storage’s State of Energy SHALL be continuous between optimization time horizon boundaries

2. The system SHALL accept the following as inputs for the battery storage asset:

• Max discharge power capacity (kW)
• Max charge power capacity (kW)
• Discharge energy capacity (kWh)
• AC-AC Round-trip efficiency (%)
• Maximum daily discharged throughput (kWh)

3. The system SHALL accept the following as inputs for the market revenues:

• Hourly LBMP ($/MWh)
• Zone

4. The system SHALL output the following values about a given battery storage system, for a year’s worth of data, at an hourly resolution

• Power output (kW)
• State of Energy (kWh)

5. The system SHALL output the following summary values about a given storage system:

• Total annual revenue generation ($)
• Total annual charging cost ($)
• Total annual discharged throughput (kWh)

6. The system SHALL output the following plots

• A plot that includes both hourly battery dispatch and hourly LBMP for the most profitable week
• A plot that shows the total profit for each month

Assignment

Using python, code a model that meets the Overall System Requirements and uses the following inputs and assumptions:

• Battery storage design inputs:
  • Max power capacity (both charge and discharge) = 100 kW
  • Discharge energy capacity = 200 kWh
  • AC-AC round trip efficiency = 85%
  • Maximum daily discharged throughput (kWh) = 200 kWh
• Market prices inputs
  • 2017 Hourly LBMPs
  • Zone = N.Y.C.
• Assumptions
  • The battery storage system has 100% depth of discharge capabilities
  • The battery storage system does not experience any degradation during the first year of operation
  • The battery storage system is a price taker (i.e. receives the LBMP as the market price)
  • The battery storage system charging cost and discharging revenue should both be based on the wholesale LBMPs
  • The historical LBMP data can be used directly as a proxy for price forecasts (i.e. no need to forecast future energy prices)