Wind and solar power have transformed the global energy landscape, offering clean and inexhaustible sources of electricity. But power levels from these sources varies over time. As Northern Europe pushes towards carbon neutrality, managing this variability has become one of the biggest technical puzzles of our time.
While battery storage and grid interconnectivity play a growing role, a surprising hero has emerged in the quest for grid stability: ultra-supercritical (USC) thermal power units. Research from the North China Electric Power University in Beijing [40.1°N, 116.3°E] presents a novel way to make these advanced coal-fired plants far more flexible — turning them into key allies in balancing the fluctuations of renewable energy.
Rethinking Thermal Power in the Renewable Era
Traditionally, coal-fired power plants have been seen as the problem, not the solution. Their rigid, slow response to demand changes made them poor partners for renewables. But modern ultra-supercritical (USC) units, which operate at extreme temperatures and pressures, are proving far more adaptable than their predecessors.
With efficiencies reaching 46%, USC plants are the most advanced coal-fired units in existence, emitting significantly less CO₂ per unit of electricity. However, they still struggle to ramp up and down quickly enough to respond to fluctuations in wind and solar output. Improving their flexibility would allow them to provide a backup role, filling in supply gaps without the inefficiencies and emissions of older power stations.
The Data-Driven Breakthrough: Smarter Load Management
The study proposes a radical shift: using advanced artificial intelligence and machine learning to optimise how USC units handle fluctuations in the grid. Traditional control systems struggle with the complexity of fast-changing loads, but new approaches using Bidirectional Test-Time Training (BiTTT) networks and Improved Temporal Convolutional Networks (ITCNs) offer a breakthrough.
These AI-driven models:
- Predict renewable generation fluctuations with greater accuracy
- Adjust USC power output dynamically in response to demand shifts
- Reduce fuel waste and emissions by optimising plant performance in real time
By combining historical operational data with real-time grid conditions, the system can fine-tune power generation to support renewables rather than compete with them.
Cleaner, More Flexible Coal — A Transition Technology?
For a region like Northern Europe, where coal is being phased out, this research presents an interesting dilemma. Should flexible coal-fired plants be used as a temporary solution to stabilise the grid while battery and hydrogen storage scale up?
The reality is that many parts of the world — including Germany, Poland, and the Netherlands — still rely on coal for backup power. If these plants are going to operate in the short term, making them more efficient and more responsive to renewables is crucial.
The research suggests that by integrating AI-driven modelling, USC plants could play a role similar to gas turbines — providing rapid-response power when renewables fall short, while consuming far less fuel than traditional coal plants.
A Grid Fit for 100% Renewables
Ultimately, the goal remains a fully renewable energy system, where wind, solar, hydro, and storage technologies replace fossil fuels entirely. But as we navigate the transition, intelligent power balancing will be essential.
This study presents a pragmatic approach — rather than waiting for perfect storage solutions, we can use AI-driven optimisation to make existing grid infrastructure work smarter, not harder.
As Northern Europe continues its march towards a carbon-free grid, innovations like these could bridge the gap between fossil fuels and a fully renewable future — ensuring that clean energy, when available, is always put to maximum use.
Source
Guolian Hou, Zeyu Liu, Data-driven modelling for ultra-supercritical unit based on bidirectional test-time training and improved temporal convolutional network, 2025-02-26
Boˆreal 