Deep-sea mining transportation elbow area abrasion prediction method considering ore grain morphology
By constructing an irregular three-dimensional mineral particle morphology database and fluid-structure interaction simulation, combined with sensor monitoring and machine learning optimization, the problem of accuracy in wear prediction in deep-sea mining was solved, enabling refined pipeline design and predictive maintenance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG LABORATORY OF SOUTHERN OCEAN SCIENCE AND ENGINEERING (GUANGZHOU)
- Filing Date
- 2026-02-12
- Publication Date
- 2026-05-08
AI Technical Summary
Existing technologies in deep-sea mining have failed to fully consider the impact of mineral particle morphology on wear, resulting in significant discrepancies between wear predictions and actual conditions. This makes it difficult to accurately identify high-risk areas and hinders the refined design and predictive maintenance of pipelines.
An irregular three-dimensional mineral particle morphology database was constructed. Fluid-structure interaction simulation was performed by combining smooth particle hydrodynamics and the discrete element method to obtain collision information between mineral particles and pipe walls. Wear depth was calculated by improving the wear calculation equation. A sensor monitoring system was deployed, and machine learning was used for adaptive optimization of the model.
It enables accurate quantification of collision and scraping behavior of irregular mineral particles, improves the accuracy of wear prediction and the adaptability of the model, and supports predictive maintenance and safety assurance of pipelines.
Smart Images
Figure CN121997685A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of deep-sea mining technology, and in particular relates to a method for predicting wear in deep-sea mining transport bends that takes into account the morphology of mineral particles. Background Technology
[0002] In deep-sea mining operations, mined minerals are typically transported to surface platforms via long-distance pipelines in a solid-liquid two-phase flow manner. Current research on pipeline wear prediction largely relies on the assumption of idealized spherical particles and employs classic wear models (such as the Archard model or its derivatives) for estimation. These methods simplify mineral particles into regular spheres and predict the wear rate and distribution of the pipe wall material by calculating parameters such as the collision frequency, normal force, and sliding distance between the particles and the pipe wall. Due to their simplicity and high computational efficiency, these models have found some application in preliminary engineering design and wear trend analysis, and provide a reference for pipeline material selection and structural layout.
[0003] However, the wear prediction methods based on the assumption of spherical particles have significant limitations. In actual deep-sea mining, minerals, after being crushed, often exhibit irregular, multi-faceted, flaky, or rod-shaped forms. Their trajectories, rotational characteristics, contact areas with the pipe wall, and collision energy differ significantly from those of spherical particles. Especially in transport bends, the rapid changes in flow direction easily generate complex flow structures such as secondary flows and vortex shedding. Irregular mineral particles are more prone to violent collisions and scraping under centrifugal force, leading to intensified local wear. Existing models fail to fully consider the impact of particle morphology on collision behavior and do not effectively incorporate the real-world conditions of high-pressure, low-temperature environments and high-speed, high-concentration turbulent flows in the deep sea. This results in significant discrepancies between predicted and actual wear conditions, making it difficult to accurately identify high-risk wear areas and hindering the development of refined pipeline design and predictive maintenance strategies. Summary of the Invention
[0004] To address the aforementioned technical problems, this invention proposes a method for predicting wear in deep-sea mining transport bends that considers mineral particle morphology, thereby resolving the issues present in the prior art.
[0005] To achieve the above objectives, this invention provides a method for predicting wear in deep-sea mining transport bends considering ore particle morphology, comprising: Step 1: Construct an irregular three-dimensional mineral particle morphology database and a high-precision three-dimensional geometric model of the deep-sea transport bend; Step 2: Using a numerical method that couples smooth particle fluid dynamics with the discrete element method, simulate the transient solid-liquid two-phase interaction process between the mineral particle group in the irregular three-dimensional mineral particle morphology database and the seawater medium in the curved pipe channel, and obtain the collision information between the mineral particles and the pipe wall. Step 3: Based on the improved wear calculation equation that introduces a morphological factor, calculate the cumulative wear depth at each location on the pipe wall according to the collision information, and generate a wear distribution map to locate high-risk areas; Step 4: Deploy a sensor monitoring system in the critical area of the bend to collect real-time data on wall thickness reduction and impact vibration; use machine learning algorithms to compare the real-time monitoring data with the prediction results of Step 3, dynamically and inversely correct the morphological factor, and achieve adaptive optimization of the wear prediction model; predict the wear development trend based on the corrected model, assess the remaining life of the pipeline, and establish a wear management database.
[0006] Preferably, in step 1, the specific method for constructing the irregular three-dimensional mineral particle morphology database is as follows: obtain the three-dimensional geometric data of real deep-sea mineral particles through three-dimensional laser scanning or CT scanning image reconstruction technology, and discretize the mineral particles into DEM units to calculate their volume, surface area, centroid and inertial tensor.
[0007] Preferably, in step 1, when establishing a high-precision three-dimensional geometric model of the deep-sea transport bend, the initial velocities of particles and fluids, the inlet velocity boundary conditions of the pipe, the outlet pressure boundary conditions of the pipe, and the no-slip boundary conditions of the pipe wall are set.
[0008] Preferably, in step 2, the fluid motion is solved using the smooth particle fluid dynamics method, including the continuity equation and the momentum equation; the trajectory of each irregular mineral particle under the action of fluid drag force, pressure gradient force, gravity, buoyancy and contact force is solved using the discrete element method; and the force and motion interaction between the fluid domain and the particles is realized through a two-way coupling mechanism.
[0009] Preferably, in the discrete element method, the contact force model between the mineral particle and the wall includes normal contact force and tangential contact force, the expressions of which are as follows: ; ; in, For normal contact force, For normal spring stiffness, The normal damping coefficient is... For overlapping volume, For relative normal velocity, n To contact the normal, For tangential contact force, For tangential spring stiffness, Relative tangential velocity, The tangential damping coefficient is... dt For time step, L T This represents the tangential spring displacement.
[0010] Preferably, in step 3, all the collision data output in step 2 are used as input, the wear amount of a single collision is calculated using the improved wear calculation equation, and the wear contribution of all collisions is accumulated on the inner wall surface of the bend through numerical integration to generate a wear depth distribution cloud map. The improved wear calculation equation is as follows: ; in, V For wear volume, Where W is the morphological factor and W is the initial wear constant. For the collision normal force, This represents the tangential sliding distance.
[0011] Preferably, in step 4, the sensor monitoring system includes an ultrasonic thickness sensor and a vibration acceleration sensor deployed on the outer arch wall of the bend.
[0012] Preferably, in step 4, the specific method of dynamically correcting the morphology factor using a machine learning algorithm is as follows: using the collision feature parameters obtained in step 2 and the predicted wear data calculated in step 3 as input features, and using the actual wear data obtained by the sensor in real time as the training target, the morphology factor is automatically corrected through an iterative optimization algorithm.
[0013] Preferably, in step 4, based on the corrected wear prediction model, numerical simulations are performed for different mining conditions to dynamically predict the spatiotemporal evolution of pipe wall wear, and the remaining service life of the pipeline is evaluated based on the preset critical wear thickness.
[0014] Preferably, the wear management database established in step 4 is used to store time-series monitoring data, simulation parameters and model calibration records, and based on the database, data mining or prediction algorithms are used to perform retrospective analysis and long-term prediction of the wear trend of the pipeline throughout its entire life cycle.
[0015] Compared with the prior art, the present invention has the following advantages and technical effects: This invention, through the technical solutions in steps 1-3, uses real irregular mineral particle morphology for high-fidelity fluid-structure interaction simulation and embeds factors characterizing the influence of morphology into the physical wear model, fundamentally overcoming the model bias caused by the traditional spherical particle assumption, making the quantification of irregular mineral particle collision and scratching behavior and the resulting wear prediction more accurate.
[0016] This invention acquires actual wear data through the sensor network in step 4, and uses this data to drive the online adaptive optimization of key model parameters. This enables the model to automatically adjust to changes in working parameters such as slurry concentration, flow rate, and particle morphology during actual deep-sea mining, thereby ensuring the effectiveness and accuracy of the prediction model in long-term operation.
[0017] This invention achieves closed-loop intelligent management and decision support from prediction to maintenance through two technical features: "assessing the remaining life of pipelines" and "establishing a wear management database." Based on a calibrated high-precision model for life assessment, and combined with a database integrating historical data, simulation records, and monitoring information, a full lifecycle management system is formed, encompassing simulation prediction, online monitoring, model calibration, life assessment, and early warning decision-making. This provides a systematic and quantitative basis for predictive maintenance, optimized operation, and safety assurance of pipelines. Attached Figure Description
[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a method for predicting wear in deep-sea mining and transportation bends that takes into account the morphology of mineral particles, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the initial modeling of the transport bend in an embodiment of the present invention; Figure 3 This is a schematic diagram of the solid velocity field during the transport of mineral particles in the bend of the pipe according to an embodiment of the present invention. Figure 4 This is a schematic diagram of the main wear area of the transport bend in an embodiment of the present invention. Detailed Implementation
[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0020] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.
[0021] Example 1 like Figure 1 As shown, this embodiment provides a method for predicting wear in deep-sea mining transport bends that considers mineral particle morphology, including: Step 1: Construct an irregular three-dimensional mineral particle morphology database and a high-precision three-dimensional geometric model of the deep-sea transport bend; Furthermore, in step 1, the specific method for constructing the irregular three-dimensional mineral particle morphology database is as follows: three-dimensional geometric data of real deep-sea mineral particles are obtained through three-dimensional laser scanning or CT scanning image reconstruction technology, and the mineral particles are discretized into DEM units to calculate their volume, surface area, centroid and inertial tensor.
[0022] Furthermore, in step 1, when establishing a high-precision three-dimensional geometric model of the deep-sea transport bend, the initial velocities of particles and fluids, the inlet velocity boundary conditions of the pipe, the outlet pressure boundary conditions of the pipe, and the no-slip boundary conditions of the pipe wall are set.
[0023] Specifically, geometric data of real deep-sea ore samples are obtained through 3D laser scanning technology, or a 3D ore particle database with irregular polyhedral shapes is constructed using a random shape generation algorithm; simultaneously, a high-precision 3D geometric model of the deep-sea transport pipe is established, such as... Figure 2 As shown.
[0024] Using image reconstruction techniques based on 3D laser scanning or CT scanning, we obtain real 3D geometric data of typical deep-sea mineral particles (such as polymetallic nodules and cobalt-rich crusts). The irregularly shaped mineral particles are discretized into a set of points, lines, and surfaces in the form of DEM units, and their volume, surface area, centroid, and inertial tensor are calculated. A high-precision geometric model of the three-dimensional deep-sea transport bend was established using a parametric modeling method. The geometric entity was generated in the three-dimensional modeling software by defining the centerline trajectory and diameter of the bend, which was then used for subsequent fluid-structure interaction simulation analysis. The pipe was placed vertically, and initial velocities of particles and fluid, inlet velocity boundary conditions, outlet pressure boundary conditions, and no-slip boundary conditions of the pipe wall were set.
[0025] Step 2: Using a numerical method that couples smooth particle fluid dynamics with the discrete element method, simulate the transient solid-liquid two-phase interaction process between the mineral particle group in the irregular three-dimensional mineral particle morphology database and the seawater medium in the curved pipe channel, and obtain the collision information between the mineral particles and the pipe wall. Furthermore, in step 2, the fluid motion is solved using the smooth particle fluid dynamics method, including the continuity equation and the momentum equation; the trajectory of each irregular mineral particle under the action of fluid drag force, pressure gradient force, gravity, buoyancy and contact force is solved using the discrete element method; and the force and motion interaction between the fluid domain and the particles is realized through a two-way coupling mechanism.
[0026] Specifically, a high-fidelity numerical simulation of solid-liquid two-phase flow was conducted. A coupled numerical method combining smooth particle hydrodynamics (SPH) and the discrete element method (DEM) was employed to simulate the interaction between irregularly shaped mineral particle groups and seawater in a complex curved pipe channel. The model precisely sets the parameters for the high-pressure and low-temperature deep-sea environment and configures corresponding flow velocity and concentration boundary conditions, such as… Figure 3 As shown, the simulation of the high-speed, high-concentration turbulent flow state in the bend region is caused by secondary flow, particle collisions, and vortex shedding.
[0027] Using data derived from the constructed 3D deep-sea transport bend model, fluid motion control equations were built. The fluid domain was discretized into a series of smooth particles using the SPH method, including continuity and momentum equations. The expression for the continuity equation is as follows: (1) in, For particle density, For particle mass, For particles i and j The spacing, Given a smooth length, the second term on the right-hand side of the above equation is a density diffusion term, used to stabilize the density field during high-frequency oscillations. for abbreviation, v i and v j Particles i and j The speed. for -The SPH model coefficient is recommended to be 0.1 in most numerical simulations. For numerical speed of sound, The item can be written as: (2) in, and This represents the total density of the particles and the still water density. It is a particle j and particles i Total density difference. It is a particle i and j The spacing. It is a particle i and j The difference in still water density between them can be written as: The discrete form of the momentum equation is: in, and They represent particles respectively i and particles j Pressure Represents gravitational acceleration. Indicates kinematic viscosity. It is the artificial viscosity coefficient. The first and second terms in equation (3) represent linear viscous pressure and subparticle-scale stress, respectively. SPS stress tensor express and The shear stress component in the direction.
[0028] In the solid-state solution section, the DEM method is used to perform kinematic and dynamic solutions for each irregular mineral particle, calculating its trajectory under the action of fluid drag force, pressure gradient force, gravity, buoyancy, and mutual contact forces. In the discrete element method, the contact force model between the mineral particle and the wall includes normal contact force and tangential contact force, the expressions of which are as follows: (4) (5) in, For normal contact force, It is the normal spring stiffness. It is the normal damping coefficient, which can be solved by the Young's modulus and the coefficient of restitution of the material. It is the overlapping volume. It is the relative normal velocity. n The normal line represents the point of contact. For tangential contact force, It is the tangential spring stiffness (usually) ), Represents relative tangential velocity. Represents the tangential damping coefficient. dt Represents the time step. L T It is the displacement of the tangential spring from its equilibrium position.
[0029] The contact force and torque acting on the DEM particle are: (6) (7) in, R c This refers to the location of the contact point between DEM particles. R 0 represents the current centroid position of the DEM particles.
[0030] Through a two-way coupling mechanism, the fluid forces calculated by SPH are applied to the DEM particles, while the motion of the DEM particles is fed back to the SPH fluid domain, achieving a fully coupled transient simulation of solid-liquid two-phase flow. Based on the calculated fluid-solid contact forces, solid particles are simulated in the DEM module. k The motion of a single DEM particle includes translational and rotational motion. The translational and rotational motions of a single DEM particle can be solved using Newton's second law: (8) (9) in, and Representing DEM particles respectively k The mass and inertia matrices. and These are translational velocity and angular velocity, respectively. and Indicates DEM particles k The forces and torques applied, superscript s and f These represent the force or torque obtained from the solid phase and the fluid phase, respectively.
[0031] Step 3: Based on the improved wear calculation equation that introduces a morphological factor, calculate the cumulative wear depth at each location on the pipe wall according to the collision information, and generate a wear distribution map to locate high-risk areas; Furthermore, in step 3, all the collision data output in step 2 are used as input, the wear amount of a single collision is calculated using the improved wear calculation equation, and the wear contribution of all collisions is accumulated on the inner wall surface of the bend through numerical integration to generate a wear depth distribution cloud map. Specifically, based on the existing wear model, an improved wear calculation equation is established. Based on the simulation results of step 2, the collision information between each mineral particle and the wall surface is extracted, the cumulative wear depth at each location on the pipe wall is calculated through numerical integration, and intuitive wear cloud maps and contour maps are generated to accurately locate high-risk wear areas.
[0032] The improved wear calculation equation is as follows: (10) in, V This indicates the volume of material removed due to wear. It is a morphological factor. Indicates the normal contact force. It is the tangential sliding distance of the particle along the pipe wall. W It is the initial wear constant.
[0033] Using all the collision data output in step 2 as input, the improved model is used to calculate the material wear caused by each collision. Through numerical integration, the wear contribution of all collisions is accumulated across the entire inner wall surface of the bend. The accumulated wear results are mapped onto the bend's geometric model to generate a high-resolution cloud map of the pipe wall wear depth distribution, such as... Figure 4 As shown, the system visually identifies the most severely worn "hot spots" and outputs the maximum wear depth and average wear rate at key locations, accurately locating high-risk wear areas.
[0034] Step 4: Deploy a sensor monitoring system in the critical area of the bend to collect real-time data on wall thickness reduction and impact vibration; use machine learning algorithms to compare the real-time monitoring data with the prediction results of Step 3, dynamically and inversely correct the morphological factor, and achieve adaptive optimization of the wear prediction model; predict the wear development trend based on the corrected model, assess the remaining life of the pipeline, and establish a wear management database.
[0035] Furthermore, in step 4, the sensor monitoring system includes an ultrasonic thickness sensor and a vibration acceleration sensor deployed on the outer arch wall of the bend.
[0036] Furthermore, in step 4, the specific method of dynamically correcting the morphology factor using machine learning algorithms is as follows: using the collision feature parameters obtained from the simulation in step 2 and the predicted wear data calculated in step 3 as input features, and using the actual wear data obtained from real-time monitoring by the sensor as the training target, the morphology factor is automatically corrected through iterative optimization algorithms.
[0037] Furthermore, in step 4, based on the corrected wear prediction model, numerical simulations are performed for different mining conditions to dynamically predict the spatiotemporal evolution of pipe wall wear, and the remaining service life of the pipeline is evaluated based on the preset critical wear thickness.
[0038] Furthermore, the wear management database established in step 4 is used to store time-series monitoring data, simulation parameters and model calibration records, and based on this database, data mining or prediction algorithms are used to perform retrospective analysis and long-term prediction of the wear trend of the pipeline throughout its entire life cycle.
[0039] Specifically, the steps in step 4 are as follows: Step 401: Deploy a sensor monitoring system in the critical wear area of the bend to collect real-time data on pipe wall thickness reduction and vibration impact signals; Step 402: Using machine learning algorithms, such as neural networks, construct a nonlinear mapping correction model between monitoring data and wear prediction model parameters: using the collision feature parameters (velocity, angle, position) obtained in step 2 and the predicted wear data calculated in step 3 as input features, and the actual wear rate or thickness change data obtained by real-time monitoring by sensors as training labels, the morphology factor α, which characterizes the influence of mineral particle morphology in the wear prediction model, is automatically corrected through iterative optimization using the backpropagation algorithm, so as to realize the dynamic adaptive update of model parameters; Step 403: Based on the above-corrected wear prediction model, numerical simulations are performed for different mining conditions to dynamically predict the spatiotemporal evolution of pipe wall wear, and the remaining service life of the pipeline is evaluated based on the critical wear threshold. Step 404: Establish a wear history database to store time-series monitoring data, simulation parameters, and model calibration records; based on this database, use time series analysis or deep learning algorithms to perform retrospective analysis and long-term prediction of the wear trend throughout the pipeline's entire life cycle.
[0040] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for predicting wear in deep-sea mining transport bends considering ore particle morphology, characterized in that, Includes the following steps: Step 1: Construct an irregular three-dimensional mineral particle morphology database and a high-precision three-dimensional geometric model of the deep-sea transport bend; Step 2: Using a numerical method that couples smooth particle fluid dynamics with the discrete element method, simulate the transient solid-liquid two-phase interaction process between the mineral particle group in the irregular three-dimensional mineral particle morphology database and the seawater medium in the curved pipe channel, and obtain the collision information between the mineral particles and the pipe wall. Step 3: Based on the improved wear calculation equation that introduces a morphological factor, calculate the cumulative wear depth at each location on the pipe wall according to the collision information, and generate a wear distribution map to locate high-risk areas; Step 4: Deploy a sensor monitoring system in the critical area of the bend to collect real-time data on wall thickness reduction and impact vibration; use machine learning algorithms to compare the real-time monitoring data with the prediction results of Step 3, dynamically and inversely correct the morphological factor, and achieve adaptive optimization of the wear prediction model; predict the wear development trend based on the corrected model, assess the remaining life of the pipeline, and establish a wear management database.
2. The method according to claim 1, characterized in that, In step 1, the specific method for constructing the irregular three-dimensional mineral particle morphology database is as follows: obtain the three-dimensional geometric data of real deep-sea mineral particles through three-dimensional laser scanning or CT scanning image reconstruction technology, and discretize the mineral particles into DEM units to calculate their volume, surface area, centroid and inertial tensor.
3. The method according to claim 1, characterized in that, In step 1, when establishing a high-precision three-dimensional geometric model of the deep-sea transport bend, the initial velocities of particles and fluids, the inlet velocity boundary conditions of the pipe, the outlet pressure boundary conditions of the pipe, and the no-slip boundary conditions of the pipe wall are set.
4. The method according to claim 1, characterized in that, In step 2, the fluid motion is solved using the smooth particle fluid dynamics method, including the continuity equation and the momentum equation; the discrete element method is used to solve the trajectory of each irregular mineral particle under the action of fluid drag force, pressure gradient force, gravity, buoyancy and contact force; and the force and motion interaction between the fluid domain and the particles is realized through a two-way coupling mechanism.
5. The method according to claim 4, characterized in that, In the discrete element method, the contact force model between the mineral particles and the wall surface includes normal contact force and tangential contact force, the expressions of which are as follows: ； ； in, For normal contact force, For normal spring stiffness, The normal damping coefficient is... For overlapping volume, For relative normal velocity, n To contact the normal, For tangential contact force, For tangential spring stiffness, Relative tangential velocity, The tangential damping coefficient is... dt For time step, L T This represents the tangential spring displacement.
6. The method according to claim 1, characterized in that, In step 3, all the collision data output in step 2 are used as input, the wear amount of a single collision is calculated using the improved wear calculation equation, and the wear contribution of all collisions is accumulated on the inner wall surface of the bend through numerical integration to generate a wear depth distribution cloud map. The improved wear calculation equation is as follows: ； in, V For wear volume, Where W is the morphological factor and W is the initial wear constant. For the collision normal force, This represents the tangential sliding distance.
7. The method according to claim 1, characterized in that, In step 4, the sensor monitoring system includes an ultrasonic thickness sensor and a vibration acceleration sensor deployed on the outer arch wall of the bend.
8. The method according to claim 1, characterized in that, In step 4, the specific method of dynamically correcting the morphology factor using machine learning algorithms is as follows: the collision feature parameters obtained from the simulation in step 2 and the predicted wear data calculated in step 3 are used as input features, and the actual wear data obtained from real-time monitoring by the sensor is used as the training target. The morphology factor is automatically corrected through iterative optimization algorithms.
9. The method according to claim 1, characterized in that, In step 4, based on the corrected wear prediction model, numerical simulations are performed for different mining conditions to dynamically predict the spatiotemporal evolution of pipe wall wear, and the remaining service life of the pipeline is evaluated based on the preset critical wear thickness.
10. The method according to claim 1, characterized in that, The wear management database established in step 4 is used to store time-series monitoring data, simulation parameters and model calibration records, and to perform retrospective analysis and long-term prediction of the wear trend of the pipeline throughout its entire life cycle based on data mining or prediction algorithms.