Scientists from China University of Petroleum developed an intellectual system for controlling one of the most sensitive processes in underground coal gasification – water inflow. The technology allows for forecasting in advance the volume of water, which will be in-flowing into the underground gas generator, assessing the safety of the area selected for gasification and selecting the optimal mode for supplying water to assure the maximum hydrogen output.
Underground gasification means converting coal into combustion gas directly in the subsoil: oxygen and steam are supplied to the reservoir through the wells, high-temperature combustion is initiated and the received gas is brought to the surface. Water plays a double role in this process. On one hand, it is the necessary agent for chemical reactions resulting to hydrogen generation. On the other hand – its excessive inflow from aquifers can cool the reaction zone, degrade the composition of gas, increase CO₂ emissions and even terminate the process. Until now, there was no integrated system combining forecast of the water inflow, risk assessment and operational control of the process in the real-time mode.
The system proposed by Chinese scientists consists of three interconnected elements. The first one is the water inflow forecasting model based on artificial intelligence. The scientists created a hybrid architecture integrating the capabilities of convolution neural networks (CNN), bidirectional recursive networks (BiLSTM), attention capturing mechanism (Attention) and support vector machines (SVM). Such combination allows for simultaneous account for spatial characteristics of the rock, the timewise dynamics of the process and for working with limited data. The model was trained using the assembly of geological and hydrodynamic data from the pilot section in Santanhu coal field in Xinjiang. The accuracy turned out to be record-high – determination coefficient R² reached 0.95.
The second element of the system is updated criteria of assessing the area safety. It is important to account not only for dynamic water inflow forecast, but also the geological protection of the reservoir against external waters. It is known that for safe gasification the coal bed top needs to be isolated by at least 36-42 meters of dense rock, and the coal bed base needs to be separated from the aquifer horizon by a layer of 20-30 meters. These are the guiding reference points allowing for assessing the risk of water breaking through into the reaction zone. Combining such statistical parameters with the water inflow forecast the researchers developed the risk maps dividing the territory into zones with low, medium and high hazards. In the pilot section the analysis showed that the expected water inflow remains below the critical level – approximately 1 m³ of water per one ton of coal. This confirms that this section is good for underground gasification.
The third element is the optimization model for calculating the volume of water to be pumped in. The researchers derived a balanced equation taking into account both the need for water and the inflow from the reservoir, losses during combustion of coal and inflow by wells. This may be the basis for selecting the pumping mode for each stage of gasification process. Thus, for the pilot section (discharge pressure of 11 MPa, evolutionary pressure of 9 MPa) the model showed that at the initial stage no additional water supply is required, while as at the stable phase about 120 m³ of water are required per day to achieve the optimal gas composition and maximize hydrogen output.
The hydrodynamic analysis made in the course of the research allowed for identifying the key factors affecting the water inflow. The most important of them are the stage of gasification, the temperature in the reaction zone and pressure differential. The scientists also discovered that at the initial stage of the process it is feasible to apply soft relief of pressure in the producing well to minimize water inflow and assure stable ignition.
In future the researchers intend to add economic performance metrics into the system and to form a major industrial database together with the companies engaged in underground gasification.
