This site uses cookies in order to improve your user experience and to provide content tailored specifically to your interests. Detailed information on the use of cookies on this website is provided in our Privacy Policy. You can also manage your preferences there.

By using this website, you consent to the use of cookies.

Learn more
OK

Can algorithms maximize power plant revenue?

We began thinking about this issue about two years ago. Due to tremendous changes in the energy market, operators of combined cycle power plants need to cycle their assets much more frequently. The main reason for this trend is the ever-increasing share of renewable sources in the energy mix. Because the feed-in of renewable plants depends on weather conditions, their power output fluctuates. Controllable generation technologies like combined cycle power plants need to compensate for the gaps in order to secure a stable power supply based on demand. This leads to new operating profiles with higher flexibility requirements and to increasingly complex dispatch optimization.

Closing the gap

To maximize the revenue of combined cycle power plants, energy traders are making great efforts to exploit the predicted demand profiles, prices, and weather reports. The right timing of dispatching offers significant economic advantages. But to ensure this, traders always need to know the available physical performance of the assets in their portfolio. The more accurately energy traders knows how fast the plant can reach a specific power output at a specific moment, the better they can maximize their revenues and also ensure grid stability.

Forecasting the start-up

Based on these considerations, we wanted to develop a digital solution that could forecast the potential power output of an asset for a specific time depending on defined parameters. The start-up performance of a combined cycle power plant depends primarily on the thermal state of the thick-walled components and the turbine’s capabilities under specific ambient conditions like air temperature, humidity, and pressure. So the key to accurate forecasting is an algorithm that can calculate expected start-up performance based on historical plant data and that also avoids the need for complex physical models.

Developing a digital application

In the framework of our next47 accelerator program, we developed an application call “Smart Start,” a browser-accessible application with an intuitive user interface for dispatch scenario planning. The core of Smart Start is a highly sophisticated machine-learning algorithm that evaluates the entire operational history of the power plant. Key performance parameters like start-up time, power output, and fuel consumption can be determined by factoring in the predicted weather conditions for each desired scenario.

Delivering the individually predicted start-up curve

Smart Start delivers the individually predicted start-up curve based on the desired start-up moment as input and the historical plant performance data while avoiding long processing times. It uses data from the plants’ SPPA-T3000 control system via a cloud connection. For calculation purposes, the application automatically creates a connection to the data acquisition system and starts its calculation based on the current state of the power plant. A prototype of this digital application is already successfully running at the combined cycle power plant in Herdecke, Germany.

Unleashing the asset’s full potential

Smart Start can be an important advantage for plant owners by increasing their responsiveness to volatile energy markets. The improvements include enhancement of start-up and power output prediction accuracy, increased revenues due to additional trading opportunities, and lower imbalance costs. We see Smart Start as an important tool for unleashing the full potential of an existing asset. Because it’s data-driven, the algorithm can be integrated either in plant controls or directly in dispatch optimization software.

For us, it’s another prime example of how digitalization can help improve and maximize revenues in the energy industry. What do you think? Your questions and comments are welcome!