Impact of 2-Hour Model on Peak Savings with NeuraCharge

The 4-hour peak prediction window has been the standard for most battery projects in the market. Historically, it shows the best result in meeting the peak demand in many regions. While many battery projects have been designed in this 4-hour prediction window, a 2-hour model can prove to be much more effective in meeting these demands while also minimizing cost. Here’s how our new model fares against the traditional 4-hour model.

The Limiting Factor: Battery Rating and Output 

Firstly, it’s important to understand battery rating and its constraints. “C rating” is a typical rating that comes with a battery; these ratings also come with a number beside them. For example, a battery with a 0.5 C rating will only be able to discharge half of its capacity every hour

Let’s take the example of a 2 MW / 3 MWh battery. Despite the converter being capable of handling 2 MW, the battery itself can only provide 1.5 MW of output (0.5 * 3 MWh). This means that, even though the converter has the capacity for higher power, the battery is constrained by its own energy rating.

The Facility Load: Not Always a Bottleneck

In some cases, the load from the facility behind the meter may exceed the battery’s capacity at 0.5 C. However, this doesn’t pose an issue because the system is designed to handle situations where the load goes beyond the battery’s output. The real limitation comes into play when we consider how long the battery needs to discharge to manage peak demand efficiently.

The Power of a 2-Hour Peak Prediction Model

Now, here’s where it gets interesting: if a battery can be dispatched over a 4-hour window, it can certainly handle a 2-hour window at double the power, without overworking the system. This not only maximizes battery efficiency but also optimizes the benefits of peak management, allowing for faster response times and more energy savings during peak hours.

In the last five years, Ontario’s peak hours mostly happened between 5 PM and 7 PM. With this knowledge, we can create a simple but effective algorithm. It will manage energy needs during these important hours. This can be done without needing a long, 4-hour prediction window. A 2-hour model would be sufficient to handle this peak demand, drastically improving efficiency.

Real-Time Adaptation: How Our Algorithm Works

Our new 2-hour algorithm is designed to work in real time. It starts discharging when peak demand is expected, but stops discharging if there’s no peak (even in the middle of an hour). It then switches to charging mode to prepare for future discharges, based on time limitations and the gap between charging and discharging.

For instance, if the peak hours shift, like they did this year in Ontario from 5–7 PM to 6–8 PM, our algorithm can adjust automatically. The system would stop discharging at 5:20 PM, switch to charging at 5:30 PM, and then resume discharging at 6 PM, perfectly managing the peak demand in just 2 hours.

The Bottom Line

The beauty of NeuraCharge™ 2-hour peak prediction model is its efficiency. By adjusting in real-time and only discharging when necessary, the system not only conserves battery life but also maximizes the benefits of peak management, ensuring energy is used most effectively when it’s needed the most.

In conclusion, a 2-hour peak prediction model is a more optimal approach than the traditional 4-hour window, particularly in battery storage and energy management projects. By reducing the time frame for peak dispatch, batteries can operate more efficiently, delivering maximum value while keeping energy costs low. As battery technology continues to evolve, these smarter algorithms will help power the future of energy storage and management.

Historical Data of Our 2-Hour Model Stacks Up Against 4-Hour Predictions

The following tables show past peak calls from previous years, followed by a comparison with the new 2-hour model.

2024:

2023:

2022:

2021:

2020:

2019:

Total Result:

This table shows how our new model can improve GA savings even in the 2-hour model.