
//php echo do_shortcode(‘[responsivevoice_button voice=”US English Male” buttontext=”Listen to Post”]’) ?>
How has energy production changed?
Ever since the first power grids were established, the flow of power has always been unidirectional, with energy flowing out of power stations, through the grid, and eventually into homes, offices, and factories. When generating power for a grid, it is essential that the amount of power being generated by all power stations equals the current demand; any difference between the two results in grid instability, which includes voltage fluctuations and frequency instability. While this is a difficult task to accomplish, it is not impossible, and grid operators use demand curves and other data to predict future demand.
However, the addition of personal renewable energy sources like small wind turbines and rooftop solar panels has fundamentally changed how power grids operate. For the first time in history, the power grid is no longer unidirectional; it has become bidirectional with consumers doubling as energy sources.
The challenge presented by this comes from the fact that individual renewable generators are not networked in the same way as mainstream power generators are. There is no way for network operators to instruct customer power sources to stop providing power, adjust their frequency, or adjust their output voltage. If these power sources start forcing power to the grid, it is up to the grid operators to detect this change and then try to accommodate the additional input power.
EVs have a problem
With the grid now bidirectional, EVs have the newly founded potential to act as both an energy source and sink.
When EVs were first designed, network operators expected that users would connect them to the grid, charge their vehicles, and then drive on. However, the inability of renewable energy sources to store excess power has introduced the need for engineers to explore energy-storage technologies.
Considering that large-scale energy-storage technologies like pumped hydro and lithium-ion are often expensive and impractical, engineers are now turning to EVs as the future of energy storage. EV operators who agree to such a scheme can potentially see massive energy savings by being paid a premium to provide power back to the grid and receiving discounts for taking energy during times of excessive generation.
Because EVs are mobile, it is extremely challenging for grid operators to predict where exactly their energy is being stored. For example, a major event that sees thousands of EVs parked in one area could prevent grid operators from taking the full amount of power that the combination of vehicles could generate due to undersized cables and other limitations in the local electrical network.
Digital twins can help
When trying to predict how a system will behave, researchers have historically turned to deterministic models. While this works for basic models, it can quickly fall apart when trying to deal with dynamic systems consisting of many variables, all of which are dependent on each other in unusual ways.
Furthermore, deterministic models are designed around generic configurations of a system, meaning inaccuracies creep in when using a singular model for multiple systems. This is especially true if changes—such as upgraded components, structural changes, software fixes, and the like—are made to the system down the line and not incorporated into the original model.
Digital twins can overcome these challenges by creating a digital copy of a system that operates in parallel to its physical twin. Real-time data is fed into the digital twin that allows it to not only predict the behavior but also to learn by comparing its output with the physical system. Thus, a digital twin can be thought of as an AI model linked to a specific system that self-improves over time, thanks to the use of a closed-loop design.
It should be noted that a digital twin will run locally to the system that it is describing to create the most accurate model possible. For example, 10 steam turbines in a power station would each have their own digital twin, and each digital twin would become unique to each turbine over time.
One major advantage of digital twins in renewable systems is that they let engineers perform numerous predictive tasks, such as expected energy generation, maintenance, and potential signs of danger. For example, a traditional mechanical danger sensor may signal a message to operators as soon as mechanical vibrations exceed some given figure. But a digital twin would use AI to identify signs that eventually lead to a serious mechanical vibration. By using prediction, operators can make allowances for downtime, identify potentially dangerous situations before they become dangerous, and even reduce the cost of repairs.
Wind farms can benefit from digital twins
To better understand how digital twins have been leveraged to help with renewable energy generation, let’s take a look at an example currently being developed by GE: wind turbines.
Wind farms consist of many wind turbines, with each turbine having its own ideal operating conditions and factors that affect those turbines specifically. If each turbine individually runs its own digital twin locally, these digital twins provide each wind turbine with predictive maintenance and anomalous operation detection. The twins can also allow for each turbine to improve energy efficiency depending on wind speed, direction, energy demand, and other environmental conditions.
Another potential benefit of digital twins in modern wind turbines is the ability to maximize the lifespan of local battery storage. As lithium-ion batteries have a limit on the number of charge/discharge cycles before degrading, such a combined energy solution must try to minimize the use of batteries. This can be achieved with the use of digital twins by learning from demand cycles and correlating this data with other environmental factors like climate conditions.
Importantly, digital twins can help model the entirety of a wind farm.
One example of why digital twins of an entire farm can be beneficial comes from the neighboring effects of wind turbines. Simply put, wind turbines at the front of a farm will create all manner of disturbances in the atmosphere, which can impact the performance of other wind turbines. Thus, digital twins can take such effects into account and make adjustments to each turbine to find optimal solutions for energy generation. For example, it is often better to run all turbines at a reduced speed than to allow some turbines to run at full speed and others stationary.