Simulation Test System for CO₂ Sequestration in Coal Seams under Multi-Field Coupling Conditions

Abstract

This application discloses a simulation test system for carbon dioxide coal seam sealing under multi-field coupling conditions. The system includes: a sample preparation and pretreatment module; a multi-field coupling environment simulation chamber module; a fluid injection and circulation control module, which injects supercritical carbon dioxide, simulated formation water, or multi-component gas into the multi-field coupling environment simulation chamber module and controls fluid circulation; an in-situ real-time multi-parameter monitoring module, which monitors the dynamic evolution data of the physical, chemical, and mechanical parameters of the coal and rock samples and the fluid within the chamber in real time; and a data integration, analysis, and intelligent control module. The geochemical sensor, acoustic emission array, distributed optical fiber, and online gas analyzer integrated in this application realize comprehensive, in-situ, and real-time monitoring from microscopic chemical reaction rates and internal micro-fracture events to macroscopic temperature / strain field spatial distribution and the evolution of the composition of the produced fluid, providing data support for revealing the correlation between the microscopic mechanisms and macroscopic responses under multi-field coupling.

Description

Technical Field

[0001] This application belongs to the field of geological engineering experimental technology, specifically involving a simulation test system for carbon dioxide coal seam sealing under multi-field coupling conditions. Background Technology

[0002] The effectiveness and long-term safety of carbon dioxide geological storage, especially in deep, unminable coal seams, are profoundly influenced by the complex coupling of multiple physical fields, including temperature, stress, seepage, and geochemical fields within the reservoir environment. Therefore, constructing a simulation system at the laboratory scale that can realistically reproduce the multi-field coupling conditions underground is crucial for revealing the storage mechanism and assessing its potential and risks.

[0003] Currently, there are some experimental setups available for studying CO2-coal seam interactions, but they generally suffer from the following limitations:

[0004] The simulation environment is too simple or the coupling is insufficient: Most devices focus on simulating a single field (such as seepage) or a simple coupled field (such as temperature-pressure), making it difficult to simultaneously, independently and accurately control the three-dimensional non-isobaric stress field, temperature field and pore fluid pressure field in the real strata, resulting in significant deviations between the experimental conditions and the real multi-field coupling state underground.

[0005] The sample characterization is disconnected from the experimental process: the detailed characterization of coal and rock samples before the experiment (such as mineral composition and pore structure) lacks an effective correlation with the state monitoring during the experiment. The initial heterogeneity of the samples was not accurately quantified and used as input conditions for the simulation, which reduced the interpretability and repeatability of the experimental results.

[0006] Insitu and real-time monitoring methods are lacking: Existing systems are mostly limited to monitoring macroscopic parameters (such as overall pressure and flow rate), and lack the ability to monitor key parameters such as microscopic chemical reactions inside the sample during the experiment (such as changes in ion concentration and pH value), micro-fracture dynamics, and spatial distribution of temperature / strain. This results in the understanding of the mechanism of multi-field coupling remaining superficial.

[0007] Data isolation and rigid control: Data generated by each subsystem is isolated from each other, lacking fusion analysis within a unified spatiotemporal framework. System control relies heavily on preset fixed programs or simple PID feedback, failing to adaptively optimize and adjust according to the real-time state of the experimental process, making it difficult to simulate complex nonlinear geological processes or achieve intelligent experimental path optimization.

[0008] Therefore, there is an urgent need for a comprehensive simulation test system that can achieve high-fidelity multi-field coupled environment simulation, full-cycle digital sample characterization, integrated in-situ multi-parameter real-time monitoring, and intelligent closed-loop control based on data and models, in order to fill the gap in existing technologies and provide a powerful tool for basic research and technical evaluation of CO2 coal seam sealing. Summary of the Invention

[0009] This application provides a simulation test system for carbon dioxide coal seam sealing under multi-field coupling conditions, which aims to solve the problems of existing technologies having a single simulation environment or insufficient coupling, and the disconnect between sample characterization and experimental process.

[0010] A simulation test system for carbon dioxide coal seam sealing under multi-field coupling conditions, the system comprising:

[0011] The sample preparation and preprocessing module is used to prepare standardized coal and rock samples and generate an initial state holographic data package and a three-dimensional digital core model of the coal and rock samples.

[0012] The multi-field coupling environment simulation chamber module is communicatively connected to the sample preparation and pretreatment module. It is used to receive the three-dimensional digital core model or related parameters and load the coal and rock sample inside it to simulate the coupling environment of temperature field, three-dimensional stress field and pore fluid pressure field.

[0013] The fluid injection and circulation control module is connected to the multi-field coupled environment simulation chamber module via a high-pressure pipeline. It is used to inject supercritical carbon dioxide, simulated formation water, or multi-component gas into the multi-field coupled environment simulation chamber module and control the fluid circulation.

[0014] The in-situ real-time multi-parameter monitoring module is integrated or connected to the multi-field coupled environment simulation chamber module to monitor the dynamic evolution data of the physical, chemical and mechanical parameters of the coal and rock sample and the fluid in the chamber in real time.

[0015] The data integration, analysis, and intelligent control module is communicatively connected to the sample preparation and pretreatment module, the multi-field coupled environment simulation chamber module, the fluid injection and circulation control module, and the in-situ real-time multi-parameter monitoring module, respectively. It is used to synchronously collect, fuse, and store data from each module, construct and run a digital twin model synchronized with the physical experiment based on the three-dimensional digital core model, and generate control commands based on the data analysis results to perform adaptive closed-loop control on the multi-field coupled environment simulation chamber module and / or the fluid injection and circulation control module.

[0016] Optionally, the sample preparation and preprocessing module includes a CNC precision core drilling and cutting unit, a sequential polishing unit, a fully automated mineral analysis subsystem, and a high-resolution micro-computed tomography subsystem; the module generates and outputs the initial state holographic data package and the three-dimensional digital core model to the data integration, analysis, and intelligent control module.

[0017] Optionally, the multi-field coupled environment simulation chamber module includes a cylindrical or cubic high-pressure reactor made of high-strength corrosion-resistant alloy, and an integrated triaxial loading unit, a circumferential heating and temperature control unit, a fluid injection and pore pressure control unit, and a built-in sensor array; the triaxial loading unit includes a confining pressure loading unit that applies lateral pressure to the coal sample through a flexible heat-shrink tubing or rubber sleeve, the confining pressure medium being high-purity silicone oil or water, and the confining pressure chamber and the pore system of the coal sample are physically isolated through the flexible heat-shrink tubing or rubber sleeve; the module can independently or collaboratively control axial stress, stresses in two mutually perpendicular horizontal directions, temperature, and pore pressure.

[0018] Optionally, the multi-field coupled environment simulation chamber module adopts a built-in true triaxial loading configuration, including a first axial loading unit for applying axial stress, and two independently controlled horizontal loading units for applying two principal stresses perpendicular to each other in the horizontal direction.

[0019] Optionally, the fluid injection and circulation control module includes a supercritical carbon dioxide supply subsystem, a simulated formation water preparation and injection subsystem, a multi-component gas supply subsystem, a fluid circulation and produced fluid treatment subsystem, and a central control and data acquisition unit; the central control and data acquisition unit is used to coordinate the work of each subsystem and collect process data and upload it to the data integration, analysis and intelligent control module.

[0020] Optionally, the system also includes a hydrate suppression unit, which is integrated into the fluid injection and circulation control module. This unit includes physical insulation and active heating devices for high-risk pipe sections, as well as chemical inhibitor injection devices, and performs intelligent linkage control by monitoring pipe wall temperature and pressure difference at key nodes.

[0021] Optionally, the central control and data acquisition unit runs a dedicated control algorithm. This algorithm adopts a hierarchical architecture, including a state machine that manages the sequential control of the macroscopic experimental process stages, and a model predictive controller embedded in the state machine for dynamic process predictive control of key process variables. The dedicated control algorithm receives optimized control commands generated by the data integration, analysis, and intelligent control module, adaptively adjusts the fluid injection parameters, and feeds back the execution results to the data integration, analysis, and intelligent control module in real time to form a closed-loop control.

[0022] Optionally, the in-situ real-time multi-parameter monitoring module includes a geochemical in-situ monitoring unit, an acoustic emission and microseismic monitoring unit, a distributed optical fiber sensing unit, and an online gas composition analysis unit; the monitoring data generated by each unit are uploaded to the data integration, analysis, and intelligent control module after time synchronization.

[0023] Optionally, the data integration, analysis, and intelligent control module includes a physical experiment-numerical simulation real-time interactive kernel. The kernel periodically assimilates real-time observation data from the in-situ real-time multi-parameter monitoring module into the digital twin model to dynamically correct the model parameters and make the state of the digital twin model approximate the current state of the physical experiment.

[0024] Optionally, the data integration, analysis and intelligent control module further includes an artificial intelligence-based adaptive closed-loop control unit. The unit generates optimized control commands based on the state of the digital twin model after being corrected by the real-time interactive kernel and historical data, and sends them to the underlying actuators of the multi-field coupled environment simulation chamber module and / or the fluid injection and circulation control module.

[0025] Compared with the prior art, this application has at least the following beneficial effects:

[0026] This application utilizes a built-in true triaxial loading configuration and an independently controlled circumferential heating and pore pressure system. This system can accurately and independently reproduce the three-dimensional non-uniform pressure stress state of underground reservoirs. The system incorporates temperature gradient field and pore fluid pressure field, enabling high-precision coordinated and decoupled control of multiple physical fields, which greatly improves the similarity between the experimental environment and real geological conditions.

[0027] This application uses the "initial state holographic data package" and "three-dimensional digital core model" obtained during the sample preparation stage as the digital starting point of the experiment. Through the data integration module, it realizes real-time interaction and online assimilation between physical experiments and numerical simulations, enabling the "digital twin" model to dynamically track and predict the state of physical experiments, greatly improving the transparency and interpretability of the experiments, and also providing a powerful numerical analysis tool for mechanism research.

[0028] The geochemical sensor, acoustic emission array, distributed optical fiber, and online gas analyzer integrated in this application enable comprehensive, in-situ, and real-time monitoring of everything from microscopic chemical reaction rates and internal micro-fracture events to macroscopic temperature / strain field spatial distribution and the evolution of the composition of the produced fluid. This provides data support for revealing the correlation between microscopic mechanisms and macroscopic responses under multi-field coupling. Attached Figure Description

[0029] Figure 1This is a schematic diagram of the module connection of a carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions provided in one embodiment of this application.

[0030] Figure 2 This is a schematic diagram of the key verification dimensions in the experiment.

[0031] Figure 3 The results show the comparison of the accuracy of multi-field coupled control.

[0032] Figure 4 This is the result of verifying in-situ monitoring capabilities.

[0033] Figure 5 For the predictive performance of digital twins Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.

[0035] The carbon dioxide coal seam sealing simulation test system provided in this application includes:

[0036] The sample preparation and pretreatment module prepares standardized coal and rock samples with defined initial states for subsequent multi-field coupled experiments and performs quantitative characterization of their in-situ fine structure. This module is an integrated preparation and analysis workstation, which includes at least: a CNC precision core drilling and cutting unit in an inert gas environment, a sequential polishing unit with coarse polishing and argon ion beam fine polishing functions, a fully automated mineral analysis (AM) subsystem, and a high-resolution micro-computed tomography (μ-CT) subsystem.

[0037] The CNC precision core drilling and cutting unit can non-destructively drill and cut standard cylindrical or cubic samples from large coal and rock columns under inert gas protection. The dimensional tolerance is controlled within ±0.1mm, and the end face parallelism error is less than 0.02mm, so as to meet the sealing and loading requirements of the subsequent high-pressure chamber.

[0038] The sequential polishing unit first uses a diamond suspension to mechanically polish the sample observation surface to a mirror finish. Then, it bombards the surface with an energy-adjustable wide-beam argon ion polisher to remove the surface strain layer and amorphous layer generated during mechanical polishing, ultimately obtaining an atomically smooth, damage-free, clean observation surface. The argon ion polishing parameters, including ion energy, incident angle, and bombardment time, can be adaptively optimized based on the hardness and brittleness of the coal sample.

[0039] The fully automated mineral analysis (AM) subsystem comprises an integrated field emission scanning electron microscope (FE-SEM) and an energy dispersive X-ray spectrometer (EDS). Before analysis, the system performs a low-resolution rapid scan of multiple predetermined areas on the polished surface of the sample according to a preset program, generating a panoramic backscattered electron (BSE) image. Subsequently, based on this panoramic image, the system automatically selects representative regions containing different mineral phases and pore structures, switches to high-resolution mode, and performs surface scanning analysis, simultaneously acquiring the BSE signal and EDS spectrum of each pixel. Through a built-in mineral phase identification algorithm, the energy dispersive spectral data is analyzed in real time into specific mineral phases, ultimately generating a high-resolution mineral phase distribution map, elemental distribution map, and corresponding quantitative statistical data, including the area percentage, particle size distribution, and spatial correlation information of each mineral phase.

[0040] The high-resolution micro-computed tomography (μ-CT) subsystem possesses a certain spatial resolution. This subsystem performs three-dimensional scans of the same coal sample before and after AM analysis. Precise sample stage positioning and coordinate registration techniques ensure spatial consistency between the two scans. After reconstruction, a three-dimensional pore structure model and a framework density distribution model of the coal sample can be obtained. By fusing and spatially registering this data with the two-dimensional mineral distribution map obtained by the AM subsystem, a three-dimensional "digital core" model containing mineral phase information, pore network information, and organic matter framework information can be constructed.

[0041] The module's operation and data output are as follows: First, the drilling and cutting unit prepares a sample conforming to geometric specifications. Then, the sample is transferred to the sequential polishing unit for processing. After polishing, the sample is placed in the sample chamber of the AM subsystem, where qualitative and quantitative analysis of surface two-dimensional minerals and porosity is performed under vacuum. Next, the same sample, without changing its orientation, is transferred to the sample stage of the μ-CT subsystem for three-dimensional structural scanning. The two-dimensional quantitative data (mineral content, porosity, fractal dimension, etc.) generated by the AM subsystem, the three-dimensional structural data (porosity, pore throat distribution, connectivity, etc.) generated by the μ-CT subsystem, and the "digital core" model generated by their fusion together constitute the coal sample's "initial state holographic data package." This data package is encrypted and transmitted through a standard data interface and stored in a central database, serving as the sole benchmark for the evolution of all physical parameters in subsequent multi-field coupled experiments and the initial boundary conditions for numerical simulation.

[0042] As a further technical detail, the scanning area selection logic of the AM subsystem can be intelligently guided based on the porosity development area or density anomaly area identified by μ-CT pre-scanning, achieving "three-dimensional positioning and two-dimensional fine analysis". The energy range of the argon ion polisher is preferably 2-8keV to ensure effective removal of the damaged layer while minimizing disturbance to the organic macromolecular structure in the coal. The construction of the "digital core" model adopts a voxel-level fusion algorithm based on FIB-SEM data and μ-CT data. This algorithm can correct the beam hardening effect and some volume effects in CT scanning, improving the model accuracy. These specific process parameters and algorithms are necessary to realize the module functions, and their optimization selection is a routine experiment that can be performed by those skilled in the art based on the sample characteristics, and will not be elaborated here.

[0043] The multi-field coupled environment simulation chamber module is characterized by its core pressure boundary and loading environment, which constitutes the entire experimental system. This is a multifunctional integrated high-pressure chamber capable of independently and collaboratively applying and controlling the temperature field, three-dimensional stress field, and pore fluid pressure field. The main body of the module is a cylindrical high-pressure reactor made of a high-strength, corrosion-resistant alloy (e.g., Inconel 718 or an equivalent high-temperature alloy).

[0044] The reactor is designed to operate at a pressure of no less than 60 MPa and a temperature of no less than 180°C. Its vessel body has a double-layer structure: the inner layer is the main chamber that bears pressure and comes into contact with the fluid, while the outer layer is an insulation and protective sleeve. A metal self-tightening seal structure (such as a wedge gasket seal or a triangular gasket seal) is used between the vessel lid and the vessel body, and hydraulic pre-tightening bolts are used to ensure uniform application and reliable maintenance of the sealing force.

[0045] This module integrates the following units:

[0046] Triaxial loading unit: Used to apply independent axial pressure and confining pressure to the coal sample inside the chamber. The triaxial loading unit is driven by an external independent hydraulic servo control system;

[0047] Axial loading unit: Located at the top or bottom of the reactor, it transmits force to a pressurizing piston at the top of the coal sample via a high-pressure dynamic sealing plunger that penetrates the reactor lid. This unit is equipped with a high-precision force sensor and displacement sensor (LVDT) to monitor and control axial load and displacement in real time, enabling stress-controlled or strain-controlled loading modes.

[0048] Confining pressure loading unit: Uniform lateral pressure is applied to the coal sample via a flexible heat-shrink tubing or rubber sleeve surrounding the sample. The confining pressure medium is high-purity silicone oil or water, supplied and pressure-controlled by an external servo booster pump. The confining pressure chamber is physically isolated from the coal sample's pore system by the sleeve wall.

[0049] Circumferential heating and temperature control unit: used for precise control of the temperature field inside the chamber. Multiple layers of independently zoned resistance heating tape are tightly wound around the outer wall of the reactor, and the exterior is covered with composite insulation material. Each heating zone is driven by an independent temperature control module (using a PID control algorithm). Thermocouple temperature measuring points are arranged in the sample area, on the reactor wall, and between the heating tape layers, forming a multi-point temperature measurement and feedback network to ensure that the temperature uniformity deviation in the sample area is within ±2℃ (at a setpoint of 150℃). The circumferential heating and temperature control unit can achieve linear programmed heating and isothermal control from room temperature to 150℃.

[0050] Fluid injection and pore pressure control unit: used to simulate the formation fluid environment and injection process, including:

[0051] The injection system includes at least two high-pressure fluid channels, one for supercritical carbon dioxide and the other for formation water. The channel inlets are located on the vessel lid and connected to the reactor via a micro-servo plunger pump, enabling constant flow or constant pressure injection. The injection lines are equipped with a preheating section to ensure the fluid reaches the set temperature before entering the sample.

[0052] The pore pressure control and drainage gas production channel has a fluid outlet at the base located at the lower end of the coal sample, connected to a back pressure regulator. By precisely controlling the pressure of this back pressure regulator, the pore pressure inside the coal sample can be independently set and maintained to simulate different reservoir pressure conditions. This channel is also used for the collection and metering of produced water and gas.

[0053] Built-in sensor array: Miniaturized, high-temperature and high-pressure resistant sensors are integrated in key locations within the chamber to provide real-time feedback of environmental parameters. These include, but are not limited to: pore pressure sensors embedded in the sample base and pressurized piston, armored thermocouples arranged around the sample, and miniature strain gauges (leading out via special leads) for monitoring circumferential deformation.

[0054] All the drive mechanisms (hydraulic pumps, servo motors, heating power supplies) and sensor signals of the aforementioned units are connected to an external data acquisition and control unit via a multi-channel high-voltage electrical connector that penetrates the vessel body. This control unit, based on an industrial programmable logic controller (PLC) and a host computer software platform, receives real-time temperature, pressure, force, and displacement signals from the sensors. Based on a preset experimental plan or instructions from the system's intelligent control module, it performs closed-loop regulation of each actuator to control temperature (T), axial pressure (T), and displacement (F). ), confining pressure ( ), pore pressure ( Independent or coupled control of the injection flow rate (Q);

[0055] Furthermore, in order to simulate the stress state in a real stratum where the three principal stresses are unequal ( The multi-field coupled environment simulation chamber module adopts a built-in true triaxial loading configuration. The core of this configuration lies in its ability to provide axial pressure (…). In addition to the first axial loading unit, two independently controlled horizontal loading units are integrated inside the reactor, which are used to apply two mutually perpendicular horizontal principal stresses. and );

[0056] The specific implementation is as follows: The main cavity of the reactor is a cubic or cuboid structure to accommodate six loading surfaces. The coal sample to be tested is processed into a cube and placed in the center of the cavity. Each of its six faces is equipped with an independently movable loading plate. Two loading plates located on the upper and lower surfaces of the coal sample are connected to a first axial loading piston and a bottom support piston that penetrate the reactor lid and bottom. These two work together to apply and control axial stress. );

[0057] The key feature is that four horizontal loading holes are respectively opened on the four side walls of the reactor. A second horizontal loading unit is installed in two opposite holes located on the same axis; an independent third horizontal loading unit is installed in two opposite holes located on the other vertical axis. Each horizontal loading unit includes:

[0058] Horizontal loading piston: made of high-strength alloy, with its front end in contact with the loading plate and its rear end connected to the drive mechanism;

[0059] Independent hydraulic servo drive system: Located outside the reactor, it is connected to the rear end of the piston via a high-pressure oil pipe, providing independent thrust and control for the piston;

[0060] High-pressure dynamic seal assembly: Installed at the loading hole, it ensures that the high-pressure fluid medium (such as silicone oil, used to transmit confining pressure) inside the reactor does not leak during the reciprocating motion of the piston. This seal typically employs a multi-stage combination seal, such as a combination structure of "O-ring + Step seal + dust ring";

[0061] Built-in force and displacement sensors: The force sensor is integrated inside the piston or on the back of the loading plate, and the displacement sensor (such as LVDT) measures the piston displacement. The signal is led out from a dedicated electrical penetrator through armored wires.

[0062] With the above structure, the loading systems in the three directions are completely independent, each driven by one of three external servo controllers. Operators or the central control unit can independently set and adjust the stress values ​​or displacement rates in the three directions σ1, σ2, and σ3 in real time, thereby accurately reproducing the true three-dimensional non-uniform compressive stress state of the underground rock mass. Flexible gaskets or roller designs are typically placed between the loading plate and the coal sample to reduce friction and ensure uniform stress application.

[0063] As those skilled in the art will recognize, the challenges in implementing this true triaxial loading configuration lie in ensuring the long-term reliability of multiple high-pressure dynamic seals, the compact arrangement of multiple actuators within a limited space, and eliminating mutual interference (coupling) between loading directions. Therefore, the diameter and stroke of the loading piston need to be precisely calculated and optimized to balance the loading force and spatial constraints. The independent hydraulic servo drive system must possess high rigidity and rapid response characteristics to suppress pressure fluctuations caused by coal sample deformation or rupture. These engineering details are standard optimization considerations in specific mechanical design and control system integration.

[0064] Furthermore, the specific implementation steps of the PID parameter self-tuning strategy are as follows: After the system is initially put into operation or after a major change in operating conditions, the self-tuning program is initiated. This program first drives the temperature control system to perform a predefined test action, typically by causing a moderate step increase in system temperature from a stable value under constant heating power, or by causing a step decrease through a built-in small auxiliary cooling unit (if equipped), completely recording the temperature response curve over time. Based on this response curve, a system identification algorithm is used, for example by calculating the tangent to the response curve or using the area method, to estimate the key dynamic parameters of the controlled object (i.e., the reactor and sample assembly), including equivalent lag time, time constant, and steady-state gain. Subsequently, according to the pre-set control objectives (such as "no overshoot," "fast response," etc.), and based on empirical rules such as the classic Ziegler-Nichols tuning formula or the Cohen-Coon tuning formula, the initial recommended values ​​for the proportional gain (Kp), integral time (Ti), and derivative time (Td) are automatically calculated. These calculated parameters will be automatically loaded into the PID controller to complete the initial tuning. For higher precision requirements, fine-tuning can be performed online within a small range based on the actual control performance deviation during closed-loop operation, using optimization algorithms such as gradient descent.

[0065] To address the endothermic or exothermic effects of geochemical reactions (such as mineral dissolution and precipitation) during the experiment, the temperature control system employs a feedforward-feedback composite control structure. While maintaining basic stability in the feedback loop (i.e., the aforementioned PID control), a dynamic feedforward compensation channel is added. The core of this channel is a thermal disturbance observer or a simplified empirical correlation. Its input signal is a process variable that can indirectly or directly reflect the reaction intensity, such as: the instantaneous carbon dioxide injection rate, the rate of change of pressure inside the reactor, or specific gaseous products detected by an online gas analyzer (e.g.,...). The system calculates the rate of change in the concentration of the sample. Based on a preset or learned "reaction rate-heat effect" correlation model (which maps the aforementioned process variables to equivalent heat flow disturbances), the system estimates in real time the magnitude of the heat power disturbance generated by the current sample reaction. Subsequently, this estimated value is converted into a feedforward control variable, which is directly added to the output of the PID controller. For example, when an exothermic reaction is detected, the feedforward channel outputs a negative power correction command to reduce the heating power in advance to offset the temperature rise caused by the sample's own exothermic reaction; conversely, for an endothermic reaction, the heating power is increased. The magnitude of this feedforward compensation variable can be adaptively corrected according to the deviation between the actual temperature and the set value to improve its accuracy.

[0066] As those skilled in the art will recognize, the amplitude of the test actions in the self-tuning program must be selected within the limits of system safety and process allowance to avoid irreversible damage to the experimental samples. The establishment of the "reaction rate-thermal effect" correlation model can be obtained through preliminary calibration experiments, such as measuring the adiabatic temperature rise data of a specific reaction under different temperature and pressure conditions. The superposition ratio coefficient of the feedforward compensation and feedback control needs to be adjusted during actual operation to achieve the best compensation effect. These all fall within the scope of routine engineering debugging and optimization performed during specific implementation based on the characteristics of the controlled object and the control accuracy requirements.

[0067] The fluid injection and circulation control module, independent of the environmental simulation chamber, is connected to the chamber via high-pressure pipelines. It can simulate the fluid injection, displacement, and long-term circulation processes of underground reservoirs. Its core function is to achieve high-precision, programmable control injection of supercritical carbon dioxide, simulated formation water, and multi-component gases, with closed-loop circulation capability.

[0068] This module consists of the following four core subsystems and a central control unit:

[0069] Supercritical carbon dioxide (ScCO2) supply subsystem: This subsystem provides a stable and controllable CO2 fluid. It includes a liquid CO2 storage tank, a cryogenic cooling unit, a high-pressure plunger booster pump, and multi-stage preheating coils. The CO2 is first pressurized to a set pressure (e.g., above 20 MPa) by the booster pump, then flows through a series of precisely temperature-controlled preheating coils, gradually heating it to above its critical temperature (31.1°C) to ensure it is converted to a homogeneous supercritical state before entering the main injection line. A pulsation damper is installed between the pump outlet and the preheater to smooth flow fluctuations.

[0070] Simulated Formation Water Preparation and Injection Subsystem: Used to prepare simulated formation water with target ionic composition and mineralization, and to perform deoxygenation and isothermal transportation. This system includes: a high-purity water storage tank and multiple precision metering pumps (for injection). The system includes a plasma standard solution, a static mixer, a vacuum degassing device, and a constant-temperature storage tank. The prepared formation water is delivered by a separate, corrosion-resistant, high-precision servo plunger pump (ceramic plunger). This pump can operate in constant flow or constant pressure mode, with a flow rate ranging from μL / min to mL / min.

[0071] Multi-component gas supply subsystem: Used to provide gases such as CH4 and N2 to simulate the original reservoir gas environment or to conduct gas mixing and injection experiments. This system typically includes multiple high-pressure gas cylinders, precision pressure reducing valves, high-precision mass flow controllers (MFCs), and a gas mixing chamber. The flow rate of each gas is independently controlled by the MFC, and after being uniformly mixed in the mixing chamber according to a set ratio, they are delivered through a common injection pipeline.

[0072] Fluid Circulation and Produced Fluid Treatment Subsystem: This subsystem enables the module to operate in a closed-loop manner. It mainly consists of a corrosion-resistant circulation pump, a high-efficiency multi-port switching valve assembly, a back pressure regulator, and an online solid filter. Through valve switching, the following processes can be achieved: a) guiding the produced fluid (water + gas) from the environmental simulation chamber outlet back to the injection pump front end for re-injection, simulating formation water circulation; b) guiding the fluid to the online analyzer or sampling port; c) guiding the fluid to the back pressure regulator for metering. The back pressure regulator is used to precisely control the system back pressure, simulating reservoir pressure.

[0073] Central Control and Data Acquisition Unit: Coordinates and controls the pumps, valves, heaters, and MFC of all the above subsystems. It receives sensor signals from various pipelines, including: quartz pressure sensors (range 0-60MPa, accuracy ±0.05%FS), Coriolis mass flow meters (simultaneously measuring density and mass flow rate), and platinum resistance thermometers (Pt100). All pressure, flow, and temperature data, along with the pump stroke status of the servo pumps, MFC setpoints and readings, are uploaded in real time to the next-level data integration and analysis control module via industrial fieldbus (such as Modbus TCP / IP or Profibus).

[0074] Furthermore, regarding the risk that supercritical carbon dioxide (ScCO2) and simulated formation water may form solid hydrates under high pressure and low temperature conditions in the fluid injection and circulation control module, potentially leading to blockage of pipelines, valves, or filters, this system implements a multi-pronged suppression strategy combining proactive prevention and dynamic intervention. This strategy integrates two technical means: physical insulation heating and chemical inhibitor injection, and coordinates and manages them through intelligent control logic based on process parameter feedback.

[0075] Firstly, regarding physical insulation and active heating, high-risk sections of the fluid piping system are given priority protection. These high-risk sections include: the pipeline from the ScCO2 booster pump outlet to the preheater (where the fluid temperature may be lowest), the pipeline downstream of all fluid mixing nodes (such as T- or Y-joints), and the throttling area before the back pressure regulator. Two levels of protection are implemented for these sections:

[0076] Full thermal insulation: All high-pressure pipelines transporting ScCO2, formation water and mixed fluids are wrapped with flexible composite insulation material with a thickness of not less than 50mm, and its thermal conductivity is less than 0.05W / (m•K);

[0077] Key point: Active electric heat tracing. In the aforementioned high-risk pipe sections, self-regulating electric heat tracing tape or mineral-insulated heating cables are tightly wrapped inside the insulation layer. The heat tracing system is divided into several independent temperature control zones, each driven by a separate temperature control module. Using feedback signals from temperature measuring points on the pipe wall surface of that zone (measured using armored thermocouples), a PID algorithm controls the pipe wall temperature to be above a safety margin value higher than the ScCO2-hydrate formation temperature under the current pipe pressure. This margin value is typically set to 10-15℃. The outlet temperature setting of the preheating coil system must also ensure that the ScCO2 temperature is far from the hydrate formation zone before entering pipe sections that may come into contact with water.

[0078] Secondly, a chemical inhibitor injection unit is established as a supplement to physical heating and as an emergency measure. This unit includes one or more inhibitor storage tanks (store methanol, ethylene glycol, or novel low-dose hydrate inhibitors), a high-precision micro-injection pump, and a static mixer. The injection point of the micro-injection pump is located upstream of the formation water injection pipeline and the ScCO2 injection pipeline before they merge. Inhibitor injection rate ( The value is not constant, but dynamically adjusted according to a correlation algorithm. The input variables of this algorithm include at least: the real-time injection flow rate of formation water. Real-time injection flow of ScCO2 ( ), and the current system pressure (P). That is: The functional relationship f is predetermined based on the minimum required concentration of the target inhibitor in the aqueous phase (e.g., 20% methanol by weight) and the hydrate formation phase equilibrium curve. The central control unit calculates and adjusts the injection pump settings in real time.

[0079] Finally, intelligent monitoring and coordinated control strategies form a closed loop for risk prevention. The system continuously monitors two key indicators:

[0080] The difference (ΔT) between the pipe wall temperature of each high-risk pipe section and the hydrate formation temperature corresponding to its pressure.

[0081] The pressure difference (ΔP) across the filter and critical valves;

[0082] When any ΔT falls below a safety threshold (e.g., 5°C), the control system will automatically increase the heat tracing power setting for the corresponding temperature control zone. When ΔP rises abnormally and exceeds the preset alarm limit, the system will determine that initial hydrate blockage may exist and immediately execute the following actions: a) Increase the heat tracing power of the relevant pipe section to the maximum; b) Instantly increase the chemical inhibitor injection rate proportionally for "impact injection"; c) Selectively and temporarily reduce system pressure or temporarily interrupt ScCO2 injection. After ΔP returns to normal, the system will gradually adjust the heat tracing and inhibitor injection back to the preventative operation level.

[0083] As those skilled in the art will recognize, specific parameters such as the safety temperature margin, the minimum effective concentration of the inhibitor, and the rate and duration of the "shock injection" need to be determined and optimized through preliminary testing or phase equilibrium calculations, based on the specific composition of the fluid used, the experimental pressure and temperature conditions, and the type of inhibitor selected. The setting of the differential pressure alarm limit needs to consider normal pressure loss under clean system conditions. These all fall within the scope of conventional engineering determinations in specific process applications.

[0084] Furthermore, the central control unit contains a dedicated control algorithm for coordinating the sequential operation of multiple servo pumps, booster pumps, multiple multi-way switching valves, and back pressure regulators within the fluid injection and circulation control module, and ensuring a smooth transition of system pressure and flow during process switching or setpoint changes. This algorithm is characterized by employing a hierarchical architecture, combining a sequential control state machine based on deterministic logic with feedforward-feedback predictive control based on a dynamic model to achieve high-precision and highly robust automatic control.

[0085] The core of this control algorithm consists of two layers:

[0086] The first layer is the sequential control state machine, which manages the macroscopic experimental process stages (such as "initialization," "ScCO2 constant flow injection," "formation water circulation," "well shut-in balancing," and "production stage"). Each state predefines the target setpoints, enabling states, and interlocking conditions for all actuators (pumps, valves, heaters) within that stage. Transitions between states are controlled by explicit triggering conditions, including operator commands, timer timeouts, or critical parameters (such as cumulative injection volume and gas-oil ratio) reaching thresholds. For example, when switching from the "ScCO2 injection" state to the "formation water circulation" state, the state machine executes the following sequence: 1) Stop the ScCO2 booster pump; 2) Switch the multi-way valve to the circulation pipeline; 3) Delay and wait for the pipeline pressure to stabilize; 4) Start the formation water circulation pump to a low flow rate; 5) After the pressure rebalances, gradually increase the circulation pump flow rate to the target value. This state machine ensures that all equipment starts, stops, and switches in a safe and reasonable sequence, avoiding pressure shocks or fluid mixing.

[0087] The second layer: Dynamic Process Predictive Control (MPC). This layer is embedded within each stage of the state machine requiring precise pressure / flow control, responsible for rapidly and smoothly adjusting key process variables at a micro-timescale. The algorithm is based on a simplified, real-time updated Multiple-Input Multiple-Output (MIMO) process model. Model input variables (MV) include: stroke frequency or speed of each servo pump, and the opening degree of the back pressure regulator. Model output variables (CV) include: injection point pressure (…). ), mainstream traffic ( ), cyclic back pressure ( The model also defines confounding variables (DV), such as changes in flow resistance caused by variations in sample permeability;

[0088] In each control cycle (e.g., 100 milliseconds), the algorithm performs the following steps:

[0089] Model prediction: Based on the current process status, the current position of the actuator, and the trajectory of setpoint changes over a future period, the internal model is used to predict the changing trend of each CV over multiple sampling periods in the future.

[0090] Rolling optimization: This involves solving an optimization problem whose objective function aims to minimize the deviation between the predicted future CV value and the desired setpoint, while penalizing drastic changes in the actuator (MV). The optimization process must adhere to a series of hard and soft constraints, such as: maximum and minimum speeds of each pump, safe upper and lower pressure limits, and upper limits for flow rate changes.

[0091] Implement initial control: Output the first control action obtained from the optimization calculation (i.e., the optimal adjustment amount for each MV in the next cycle) to the corresponding actuator. In the next control cycle, repeat the above process, and perform a new round of prediction and optimization based on the latest actual measurement values;

[0092] This MPC layer is particularly useful during transitions between process changes or setpoint changes. For example, when the state machine command switches from constant pressure injection of one fluid to constant flow injection of another, the MPC controller acts in advance: before shutting down the current pump, it begins to fine-tune the back pressure regulator or the pump to be started based on model predictions to proactively offset the predicted pressure fluctuations caused by valve switching and pump start-up and shutdown. This achieves a "seamless" and smooth transition of pressure and flow, rather than waiting for deviations to occur before correction (as with traditional PID).

[0093] As will be known to those skilled in the art, the process model can be established by obtaining initial parameters through system identification methods (such as step response testing), and continuously adaptively corrected based on actual input and output data during operation. The optimization problem can be solved using an efficient quadratic programming (QP) solver. Data interaction between the state machine and the MPC layer is achieved through shared memory or event signals. These specific implementation techniques all fall within the scope of conventional engineering applications in the field of advanced process control.

[0094] The in-situ real-time multi-parameter monitoring module is used to capture dynamic evolution data of the physical, chemical, and mechanical properties of a coal-water-gas-chemical multiphase system in real time without disturbance during high-temperature and high-pressure experiments. This module achieves full-scale in-situ monitoring from microscopic chemical reactions to macroscopic fracture behavior through dedicated sensors resistant to harsh environments and high-fidelity signal transmission and conditioning circuitry.

[0095] This module consists of the following four functional units:

[0096] Geochemical In-situ Monitoring Unit: Used for direct monitoring of the chemical state evolution of pore fluids in contact with the sample. The core of this unit is the integration of a miniaturized, modular electrochemical sensor probe into the sample chamber base or upper pressure head of the reaction vessel. The probe tip is encapsulated with a high-temperature, high-pressure resistant solid reference electrode, a pH glass electrode, a redox potential (Eh) electrode, and a electrode for targeting key ions (such as...). The system employs an ion-selective electrode (ISE). The probe body is housed in a Hastelloy shell, and the electrode leads are insulated and armored with magnesium oxide, extending from a high-voltage through-wire for the reactor. The signal is connected to an external high-impedance, low-drift signal conditioner for amplification and analog-to-digital conversion. To overcome electrode potential drift under high temperature and pressure, the system is equipped with an automatic online calibration function: a micro-injection pump periodically injects standard buffer solution or ion standard solution of known concentration into the fluid, and the measured value is automatically corrected based on the sensor response.

[0097] Acoustic emission and microseismic monitoring unit: This unit is used to capture the dynamic process of microcrack initiation, propagation, and penetration within coal and rock samples under stress and chemical action. It comprises at least six broadband acoustic emission sensors (frequency range 50kHz-1MHz), uniformly arranged at specific locations on the outer wall of the reactor using a high-temperature coupling agent. The sensor array layout is optimized to cover the entire sample area and allow for subsequent acoustic emission source localization analysis. All sensor signals are preamplified and continuously acquired and processed by an external multi-channel acoustic emission acquisition system. This system employs a real-time triggering algorithm based on thresholds and the short-time average / long-time average ratio to extract characteristic parameters for each acoustic emission event, including arrival time, amplitude, energy, count, and rise time. By using the time difference of the same event received by different sensors, combined with a known model of sound wave propagation velocity in the reactor and sample, the spatial coordinates of the acoustic emission event source can be calculated and displayed in three dimensions in real time.

[0098] Distributed fiber optic sensing unit: Used to acquire continuous spatial distribution information of temperature and strain fields in the sample and near-field region during experiments. This unit uses high-temperature resistant polyimide-coated single-mode optical fiber as the sensing medium. During sample preparation, the optical fiber is pre-embedded in a similar material module simulating the geological formation in a specific path (such as a spiral or mesh pattern), and this module is in close contact with the coal sample; or the optical fiber is directly attached to the surface of the coal sample and then encapsulated together. Another optical fiber is tightly wound or attached to the inner wall of the reactor to monitor the boundary thermal field. The fiber optic lead is connected to an external distributed fiber optic sensing demodulator (based on optical frequency domain reflection OFDR or Raman / Brillouin scattering principles). This instrument can demodulate the temperature (accuracy ±0.5℃) and / or strain (accuracy ±0.5℃) at millimeter-level spatial resolution along the length of the optical fiber in real time. The temperature / strain value is used to generate a temperature / strain distribution curve along the fiber path, and a spatiotemporal distribution cloud map is formed as it evolves over time.

[0099] Online Gas Composition Analysis Unit: Used for real-time, continuous composition and isotope analysis of gases produced from the reactor. The core of this unit is a multi-channel high-pressure gas sampling and pretreatment system and an online mass spectrometer or miniature gas chromatograph. The high-pressure produced gas is first reduced to atmospheric pressure by a back pressure regulator, and then enters a multi-port valve group. The valve group can: a) directly guide the gas to vent; b) introduce the gas into a constant-temperature, constant-volume sample loop; c) inject the gas from the sample loop into the analyzer in a pulsed manner. The online mass spectrometer can monitor the partial pressure and isotope ratios (e.g., δ¹⁴) of multiple gases (such as CO₂, CH₄, N₂, O₂, H₂S) per second. 13 (C-CO2); gas chromatography offers higher quantitative accuracy. Analytical results are timestamped with corresponding flow rate and cumulative yield data.

[0100] All the massive time-series data generated by the above units are synchronized with high precision (synchronization accuracy better than 1 millisecond) through a unified time base server and packaged and uploaded to the central database to provide a foundation for multi-field coupling correlation analysis.

[0101] The data integration, analysis and intelligent control module is responsible for synchronously aggregating, fusion and analyzing the massive, heterogeneous, high-concurrency time-series data generated by the entire system, and based on this, driving a "digital twin" model that runs in parallel with the actual physical experiment, ultimately realizing intelligent prediction and adaptive optimization control of the experimental process.

[0102] This module consists of three parts: a hardware server cluster, a dedicated software platform, and control logic algorithms. Its core functions and workflow are as follows:

[0103] The module synchronously aggregates and fuses heterogeneous data from multiple sources. All data streams from the aforementioned modules—including T, P, σ, and displacement from the environmental simulation chamber, pressure, flow, and composition from the fluid module, and chemical, acoustic, fiber optic, and gas data from the monitoring module—are converged to this module via high-speed industrial Ethernet. The module incorporates a high-precision time-base server to assign a unified timestamp to all data, achieving a synchronization accuracy better than 1 millisecond, ensuring precise correlation between different physical processes on the timeline. Data is written in real-time to a structured time-series database. This database defines a unified metadata framework for different data types (such as scalars, arrays, and images) and establishes indexes, supporting millisecond-level historical and real-time data correlation queries.

[0104] The "Physics Experiment - Numerical Simulation" real-time interactive kernel is based on a commercial or self-developed multiphysics simulation platform kernel (such as COMSOL Multiphysics, OpenFOAM) and is used for secondary development to construct a parameterized three-dimensional numerical model corresponding to the current physics experiment. The model includes the reaction vessel geometry, the initial structure of the sample (derived from the "digital core" model of the sample module), the constitutive relations of materials, fluid flow, heat conduction, and the governing equations for chemical reactions.

[0105] In reality, the interaction is achieved through an online data assimilation algorithm. During the experimental run, the simulation engine does not run independently, but rather performs an "assimilation step" periodically (e.g., every minute):

[0106] State update: The key real data monitored from the physical experiment at the current moment (such as the pressure difference between the upper and lower ends of the sample, the local strain measured by the distributed optical fiber, and the CO2 / CH4 ratio of the produced gas) are used as observations and input into the assimilation algorithm (such as ensemble Kalman filter EnKF or four-dimensional variational assimilation 4D-Var).

[0107] Parameter correction: The assimilation algorithm automatically adjusts previously uncertain or time-evolving key parameters in the numerical model (e.g., the model's permeability field, chemical reaction rate constant, and equivalent stiffness of the fracture network) by comparing the differences between the state predicted by the numerical model and the actual observed state. The corrected model parameters are updated so that the state of the numerical model approximates the current state of the real physical experiment as closely as possible.

[0108] Prediction and Validation: Using the calibrated model, short-term future predictions are made (such as seepage pressure distribution or deformation trends in the next hour). These predictions are compared with subsequent actual monitoring data, forming a "prediction-validation" closed loop to continuously evaluate and improve the model's fidelity. This process enables the digital model to dynamically track and "digitally twin" the physical experimental conditions.

[0109] Based on artificial intelligence, the adaptive closed-loop control module integrates intelligent decision-making and control units, building upon real-time assimilation and prediction. This unit contains a trained machine learning model (such as a deep reinforcement learning network), whose inputs are fused and feature-extracted real-time and historical data (including raw monitoring data and assimilated model state parameters), and whose output is an optimized control command set for the environmental simulation chamber and fluid injection module.

[0110] Its control logic manifests on two levels:

[0111] Strategic adjustments: Long-term strategies are dynamically adjusted based on the achievement of macroscopic experimental objectives. For example, if model predictions indicate severe CO2 fingering at the current injection rate, the AI ​​controller may decide to switch to a slug displacement strategy of "injecting a segment of CO2 followed by a segment of water," and automatically generate the set sequence of pressure and flow rates for each stage under this strategy.

[0112] Real-time fine-tuning: Feedforward compensation and setpoint fine-tuning are performed on the servo control system. For example, when both acoustic emission monitoring and the model indicate that the sample is about to enter the accelerated fracture stage, the AI ​​controller can fine-tune the axial compression loading rate or pore pressure in advance to control the fracture process; or dynamically adjust the heating power of different zones according to the temperature gradient changes monitored by the fiber optic cable to maintain a uniform thermal field.

[0113] After all control commands are verified for safety, they are sent to the corresponding underlying PLCs for execution, thus forming a complete adaptive closed loop of "perception (monitoring) - cognition (model assimilation and analysis) - decision-making (AI optimization) - execution (control) - re-perception", enabling the entire experimental system to autonomously optimize or safely execute complex nonlinear experimental paths.

[0114] In one embodiment, a system performance verification experiment is provided, using standard samples (n=5) from the No. 3 coal seam in a mining area in Shanxi Province, to compare the performance of this system with that of the commercial equipment GCTSRR-1000 under the same working conditions.

[0115] The comparative experiment design is as follows:

[0116] Control group: Using commonly used industry equipment (such as GCTSRR-1000 triaxial system + independent temperature control / gas injection module).

[0117] Experimental group: The system proposed in this application (integrating multi-field coupling + in-situ monitoring + digital twin);

[0118] Sample: Standard coal sample (Φ50×100mm) taken from the same mining area and with the same preparation process;

[0119] Operating conditions: Simulated deep coal seam conditions (confining pressure 25MPa, temperature 55℃, CO2 injection pressure 11.5MPa).

[0120] Key verification dimensions such as Figure 2 As shown;

[0121] After the experiment, the data are as follows: Figures 3-5 As shown;

[0122] Based on the experimental results, it is shown that the proposed scheme achieves high-precision and high-synchronization multi-physics field collaborative control; the sensor integration scheme significantly improves the ability to capture and analyze weak signals; and the real-time data assimilation mechanism greatly improves the reliability of model prediction.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A simulation test system for carbon dioxide coal seam sealing under multi-field coupling conditions, characterized in that, The system includes: The sample preparation and preprocessing module is used to prepare standardized coal and rock samples and generate an initial state holographic data package and a three-dimensional digital core model of the coal and rock samples. The multi-field coupling environment simulation chamber module is communicatively connected to the sample preparation and pretreatment module. It is used to receive the three-dimensional digital core model or related parameters and load the coal and rock sample inside it to simulate the coupling environment of temperature field, three-dimensional stress field and pore fluid pressure field. The fluid injection and circulation control module is connected to the multi-field coupled environment simulation chamber module via a high-pressure pipeline. It is used to inject supercritical carbon dioxide, simulated formation water, or multi-component gas into the multi-field coupled environment simulation chamber module and control the fluid circulation. The in-situ real-time multi-parameter monitoring module is integrated or connected to the multi-field coupled environment simulation chamber module to monitor the dynamic evolution data of the physical, chemical and mechanical parameters of the coal and rock sample and the fluid in the chamber in real time. The data integration, analysis, and intelligent control module is communicatively connected to the sample preparation and pretreatment module, the multi-field coupled environment simulation chamber module, the fluid injection and circulation control module, and the in-situ real-time multi-parameter monitoring module, respectively. It is used to synchronously collect, fuse, and store data from each module, construct and run a digital twin model synchronized with the physical experiment based on the three-dimensional digital core model, and generate control commands based on the data analysis results to perform adaptive closed-loop control on the multi-field coupled environment simulation chamber module and / or the fluid injection and circulation control module.

2. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 1, characterized in that, The sample preparation and preprocessing module includes a CNC precision core drilling and cutting unit, a sequential polishing unit, a fully automated mineral analysis subsystem, and a high-resolution micro-computed tomography subsystem; the module generates and outputs the initial state holographic data package and the three-dimensional digital core model to the data integration, analysis, and intelligent control module.

3. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 1, characterized in that, The multi-field coupled environment simulation chamber module includes a cylindrical or cubic high-pressure reactor made of high-strength corrosion-resistant alloy, and an integrated triaxial loading unit, a circumferential heating and temperature control unit, a fluid injection and pore pressure control unit, and a built-in sensor array. The triaxial loading unit includes a confining pressure loading unit that applies lateral pressure to the coal sample through a flexible heat-shrink tubing or rubber sleeve. The confining pressure medium is high-purity silicone oil or water, and the confining pressure chamber is physically isolated from the pore system of the coal sample through the flexible heat-shrink tubing or rubber sleeve. The module can independently or collaboratively control axial stress, stress in two mutually perpendicular horizontal directions, temperature, and pore pressure.

4. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 3, characterized in that, The multi-field coupled environment simulation chamber module adopts a built-in true triaxial loading configuration, including a first axial loading unit for applying axial stress, and two independently controlled horizontal loading units for applying two principal stresses in mutually perpendicular horizontal directions.

5. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 1, characterized in that, The fluid injection and circulation control module includes a supercritical carbon dioxide supply subsystem, a simulated formation water preparation and injection subsystem, a multi-component gas supply subsystem, a fluid circulation and produced liquid treatment subsystem, and a central control and data acquisition unit. The central control and data acquisition unit is used to coordinate the operation of each subsystem and collect process data to upload to the data integration, analysis and intelligent control module.

6. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 5, characterized in that, The system also includes a hydrate suppression unit, which is integrated into the fluid injection and circulation control module. This unit includes physical insulation and active heating devices for high-risk pipe sections, as well as chemical inhibitor injection devices, and performs intelligent linkage control by monitoring pipe wall temperature and pressure difference at key nodes.

7. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 5, characterized in that, The central control and data acquisition unit runs a dedicated control algorithm with a hierarchical architecture, including a state machine that manages the sequential control of the macroscopic experimental process stages, and a model predictive controller embedded in the state machine for dynamic process predictive control of key process variables. The dedicated control algorithm receives optimized control commands generated by the data integration, analysis and intelligent control module, adaptively adjusts the fluid injection parameters, and feeds back the execution results to the data integration, analysis and intelligent control module in real time to form a closed-loop control.

8. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 1, characterized in that, The in-situ real-time multi-parameter monitoring module includes a geochemical in-situ monitoring unit, an acoustic emission and microseismic monitoring unit, a distributed optical fiber sensing unit, and an online gas composition analysis unit; the monitoring data generated by each unit are uploaded to the data integration, analysis, and intelligent control module after being synchronized in time.

9. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 1, characterized in that, The data integration, analysis, and intelligent control module includes a real-time interactive kernel for physical experiments and numerical simulations. The kernel periodically assimilates real-time observation data from the in-situ real-time multi-parameter monitoring module into the digital twin model to dynamically correct the model parameters and make the state of the digital twin model approximate the current state of the physical experiment.

10. The carbon dioxide coal seam sealing simulation test system under multi-field coupling conditions according to claim 9, characterized in that, The data integration, analysis and intelligent control module also includes an artificial intelligence-based adaptive closed-loop control unit. The unit generates optimized control commands based on the state of the digital twin model after being corrected by the real-time interactive kernel and historical data, and sends them to the underlying actuators of the multi-field coupled environment simulation chamber module and / or the fluid injection and circulation control module.

Images