How to manage risk in energy storage

Understanding the uncertainty associated with operating variable generation systems helps manage the level of risk associated with delivering services.

Moreover, appropriate selection of the rating and charge/ discharge characteristics of an energy storage device, coupled with appropriate operational management, can improve the ability of a variable generation system to participate in certain markets.

This article explores the use of simulation and optimisation techniques for investigating the characteristics that a variable generation system with energy storage should have for given operational profiles. Two examples are considered in this article.

In the first, optimisation techniques are used to determine the rating and charge/discharge characteristics of a variable generation system to maximise revenue on the spot market.

A study of this type is based on historical data of power output and market price, and does not require detailed consideration of technology selection. 

In the second example, simulations are used to explore controlling the operational characteristics of a system that combines variable generation with energy storage, and to evaluate the affect that different energy storage interface configurations will have on grid response. Initially, a high level representation of the energy storage device is used in this type of study.

As the study progresses, more detailed representations of different technologies are included in the simulation framework.

Variable energy producer

Historical data of power output and market prices may be used to determine whether an energy storage device would improve the ability of a variable energy producer to participate in certain markets. In this article, spot market participation is considered, although the techniques discussed are applicable to ancillary markets.

Schematic of energy flow in a combined variable energy/energy storage system

Figure 1 shows a schematic of energy flow considered in the optimisation framework.

The optimisation problem is then to determine not only the charge/discharge profile of the energy storage device over a period of time, but also to determine the most appropriate energy mix at any given time for charging the energy storage device and supplying energy to the grid.

The optimisation aims to maximise revenue while ensuring that physical constraints associated with operating the energy storage device are not violated.

Outcome of optimisation over two day period.

Figure 2 shows an example of results from an optimisation for a two day period.

In line with expectation, these results show that the energy storage device is charged during periods of lower market price and discharged during periods of higher market price.

Note that compared to a real-world scenario, this example is relatively simplistic for illustrative purposes.

Operational profile of the energy storage device

Figure 3 shows the operation of the energy storage device.

In this example, storage capacity was limited to 1200 kWhr and the charge/discharge rate was limited to 200 kW.

As the amount of historic data used in the optimisation formulation increases, the risk associated with the optimisation outcome decreases. The optimisation may be formulated using linear programming, which is beneficial for reducing the computational time of larger-scale problems.

Once the rating of the energy storage device has been determined, a simulation study may be conducted to inform technology selection and the development of appropriate feedback control and supervisory control subsystems for the combined energy storage / variable energy system.

Development of the feedback control and supervisory control systems may proceed on a lower fidelity model of the energy storage device. Lower fidelity models execute faster because they omit detailed representations of power electronic devices, enabling faster simulations and faster iterations during design.

This approach works well because the bandwidth of the feedback control and supervisory control systems will be sufficiently lower than that of the power-electronic switching algorithms, meaning that inclusion of power electronics will have little effect on the RMS operation in the system simulation.

Simulation of the energy storage/wind farm system

Figure 4 shows the response of a system designed to provide firm power at the grid point-of-connection (POC).

If the energy stored is greater than 10% of capacity, then the feedback control system regulates active power at the grid POC to 0.6 per-unit.

Once the energy stored drops below 10% of capacity, then the energy storage system is charged to capacity at a fixed rate, before the POC regulation is re-engaged.

Active power (P) and reactive power (Q)

Figure 5 shows active and reactive power output for a simulation study that compares a standard 6-device insulated-gate bipolar transistor (IGBT) bridge with a 6-cell and 24-cell (per-phase) IGBT modal multilevel converter (MMC) as an interface to an energy storage system.

The system is commanded to move from 0.3 per-unit active power to 1.0 per-unit active power at 0.4 seconds, while regulating reactive power to zero.

The reference tracking capability of the feedback system is not impaired by the inclusion of different power-electronic architectures.

Terminal voltage for different converter architectures

Figure 6 shows the THD of the voltage waveform.

As expected, the THD decreases as the number of power electronic devices in the bridge architecture increases.

Detailed studies help determine the power-electronic architecture, filtering architecture, or combination of the two that is required to meet harmonic distortion requirements.

[Graham Dudgeon is Energy Industry Manager, MathWorks.] 

More information.

Send this to a friend