Yuntianhua Develops the First Large Model for Coal Gasification RTO to Drive a Green and Intelligent Chemical Industry

In Yuntianhua's central coal chemical control room in the plateau of southwestern China, operators traditionally needed to monitor the numbers on dashboards and adjust the oxygen-to-coal (O/C) ratio every half hour based on their professional experience. This decades-old practice was essential to the operational stability of ammonia synthesis units.

Today, however, challenges caused by high-temperature and high-pressure production environments, such as gasifier temperature fluctuations, slag layer thickening, and rising coal consumption, are being addressed by data-based AI models.

Developing the world's first large model for coal gasification RTO

Gasifiers are the "heart" of the coal chemical industry, but these massive chemical reactors present numerous challenges.

Gasifier stability is affected by many factors, and maintaining precise control of gasifiers remains a pressing industry challenge. Traditional mechanistic models rely on complex assumptions and frequently fail in highly-nonlinear reactions, so key parameters such as gasifier temperature and slag viscosity can only be estimated manually. In complex operating conditions, delays in manual decision-making can trigger chain reactions that significantly hinder production efficiency and quality. Furthermore, a 1% difference in the energy conversion rate per metric ton of coal means a revenue loss of tens of millions of yuan for units that have an annual processing capacity of millions of metric tons.

In 2024, Yunnan Yuntianhua Information Technology Co., Ltd., Yunnan Yuntianhua Dawei Ammonia Co., Ltd. (Yuntianhua Dawei Ammonia for short), and Huawei jointly released the world's first AI large model for coal gasification real-time optimization (RTO). The RTO model trained on Huawei's Pangu foundation model can accurately simulate and predict key gasifier operating parameters like gasifier temperature, slag layer thickness, and slag viscosity. This ensures the continual optimization and stability of the production process.

This project combined AI algorithms, advanced process control (APC), and RTO technologies. It first preprocessed and labeled the historical data of key processes in the gasifier, and then used Huawei's Pangu foundation model to train the RTO model and create scenario-specific business models. During production, the RTO system uses these business models and real-time data to constantly calculate the optimal parameters for the gasifier. The APC system then makes prompt adjustments accordingly to ensure precise control over production parameters.

After 18 months of hard work, the project team managed to overcome major challenges like soft measurement of gasifier temperature and control of strongly-coupled variables. They developed 16 scenario-specific AI models that can intelligently control coal gasification units in complex operating conditions in just a few clicks.

Making production intelligent to ensure every piece of coal is fully used

The intelligent RTO system has been running safely for over a year since entering operation on June 28, 2024. It has set new industry records for several key indicators, facilitating the intelligent transformation of the entire coal gasification production process.

First, operating methods have undergone a groundbreaking shift. The automation rate of coal gasification units has surged to 96.88%, meaning operators no longer need to constantly keep an eye on dashboards and manually adjust parameters based on their experience. This reduction in manual operations has allowed production control to enter a new phase of automated system decision-making.

Second, equipment operation is now much safer and more stable. Coal gasification units operate in high-temperature and high-pressure environments with significant safety risks. By enabling real-time monitoring of production conditions and prediction of production parameters, the RTO model can effectively prevent equipment failures and production accidents caused by abnormal parameter fluctuations.

Third, efficiency and emissions reduction have been enhanced throughout the process. Operating data shows that within individual units, the average coal consumption per metric ton of steam in the boiler has decreased by 2.08%. This saves 6,602 metric tons of standard coal annually. The average ammonia consumption per urea unit has decreased by 1.51%, which means a 3,162 metric ton annual reduction in carbon emissions. These breakthroughs make the entire production process more energy-efficient.

Yuntianhua Dawei Ammonia has used digital and intelligent technologies to become low-carbon, improve quality, and increase efficiency. In the first half of 2025, the company's semi-annual yields of synthetic ammonia, urea, and high-pressure steam all surpassed previous records. Furthermore, while production was increased, energy consumption was reduced. By improving energy efficiency, the company used 8,500 metric tons less standard coal and emitted 22,300 metric tons less carbon dioxide in H1 2025, generating direct benefits of nearly CNY10 million (≈USD1.4 million).

In addition, the project team uses real-time operating data to continue training and iterating the eight models that support gasifier units, increasing model availability by over 50%. The team has also established a standardized O&M system to ensure the long-term operational stability of the intelligent system. Now, the project is expanding to Yuntianhua's other subsidiaries and influencing other companies in the industry by offering replicable best practices for intelligent transformation.

Moving forward, Yuntianhua will continue working with Huawei to deepen its intelligent transformation through innovative technologies. This will help lay the groundwork for the green and quality development of the entire coal chemical industry.