Three-dimensional dynamic simulation of motion characteristics of metal particle in DC GIL Authors: Feng Bin, Zhensheng Tan, Jixiang Feng, Deguang Wu, Qiuqin Sun Published: 2025 ABSTRACT AbstractDynamic behavior simulation of metal particles in gas-insulated metal-enclosed transmission lines (GIL) is crucial for predicting their migration process and distribution pattern. However, existing two-dimensional (2D) simulations of particle motion characteristics fail to accurately reflect real-world particle behavior due to the limit of spatial dimensionality. To address these issues, this study develops a three-dimensional (3D) simulation model comprising an electrostatic field analysis module and a particle tracking module, grounded in dynamic contact theory and utilizing random reflection angle. Our simulation systematically investigates the motion characteristics of spherical metal particles between electrodes and around basin-type insulators, as well as the influence of key parameters. The simulation results demonstrate that an increase in reflection angle, which determines the coupling effect of radial and axial components of particle velocity, leads to the significant expansion of radial distribution range and maximum axial displacement. Besides, particles, whether far from or near the concave surface of basin-type insulator, repeatedly collide between the high-voltage (HV) conductor and the shell upon being lifted, followed by migration toward the convex side of insulator; particles near the convex surface exhibit surface-climbing behavior along the profile of insulator. Furthermore, particle vitality has a positive correlation with applied voltage but an inverse relationship with particle size; collision frequency is predominantly influenced by applied voltage and material properties, along with little dependence on particle size. These findings can provide a robust theoretical foundation for predicting metal particle behavior in GIL systems.Keywords: GIL, Metal particle, Reflection angle, 3D simulation, Particle vitalitySubject terms: Engineering, Materials science, Physics Introduction With the development of clean energy such as offshore wind power, desert photovoltaic systems, and high-altitude thermal energy applications, direct current (DC) gas-insulated metal-enclosed transmission lines (GIL) have become increasingly prominent due to their flexible layout, high capacity, and robust environmental adaptability1,2. However, during manufacturing, installation, or operation, physical wear will generate metal particles. When these particles become charged, they may collide with electrodes, thereby reducing the breakdown voltage of gas gaps and potentially triggering surface discharges in proximity to basin-type insulators3,4. According to statistics, insulation failures caused by metal particle motion account for up to 20% in GIL equipment5. Compared to experimental research6–8, simulation research on particle motion provides several distinct advantages like reduced costs, more comprehensive data, enhanced visualization of intermediate processes, and increased flexibility. Therefore, the simulation of metal particle motion in DC GIL provides more significant insights for predicting particle behaviour and preventing particle hazards. Previous research9–13 has extensively explored the motion characteristics of metal particles in gas-insulated equipment using two-dimensional(2D) modelling. For instance, Lv et al.14 investigated the metal particle motion in 1100 kV gas-insulated switchgear (GIS) by establishing a 2D simulation model, specifically analysing the movement patterns of spherical metal particles under power frequency voltage. Han et al.15 employed finite element analysis to design a 2D lift force model for particles with various cross-sectional shapes. Wiener et al.16 studied the impact of different insulating gas mixtures on the motion behaviour of metal particles under alternating current. Metwally et al.17 utilized a fourth-order Runge-Kutta method to investigate the effects of initial particle position and radius of particles on motion trajectories. Dai et al.18 employed the coupled equations to obtain the axial motion range, lifting voltage, and breakdown characteristics of linear metal particles within a 2D GIL model. Wu et al.19 researched the effects of particle materials, applied voltage, and shielding electrodes on axial motion characteristics based on charging mechanisms and dynamic equations. Zhang et al.20 numerically analysed the influence of random reflection angles, gravity inclination, and particle size on particle motion. Nevertheless, existing 2D simulations of metal particle motion can only analyse either the radial or axial motion trajectories and distribution. In reality, the radial and axial motions of metal particles exhibit nonlinear coupling effects, which cannot be achieved using 2D simulations. In this work, we developed a three-dimensional (3D) motion simulation model that explicitly accounts for the coupling effects between the radial and axial particle motions, and discuss the motion characteristics of these particles between electrodes and around basin-type insulators, as well as the influence of key factors on their behaviour. Our findings provide the solid theoretical foundation for predicting metal particle behavior in GIL and the substantial guidance for enhancing insulation reliability. Kinetic model of metal particles DC GIL model and force analysis of metal particles A simplified GIL model is established, as shown in Fig. 1. Its coaxial structure comprises a high-voltage (HV) conductor with the outer radius R1 = 20 mm and the shell with inner radius R2 = 60 mm. The ratio of R1 to R2 meets the requirements of insulation. The cavity, whose length L is set to 160 mm, have both ends terminated by basin-type insulators with a thickness of 20 mm and a tilt angle of 45°. Other parameters are listed in Table 1. 3D GIL model and analysis of force on metal particles. Table 1Parameters of GIL model.ParameterValueVacuum permittivity ε08.854*10− 12 F/mGravity acceleration g9.8 m/s2Relative permittivity of insulating gas εr1.002Coefficient of viscosity η2118.2*10− 6 Pa·sRelative dielectric constant of insulator εb4.5Temperature293.15 KSF6 gas pressure0.4 MPa Parameters of GIL model. Both buoyancy force Ff and viscous drag force Fres need to be considered since SF6 pressure is significantly greater than the standard atmospheric pressure. The dielectrophoretic (DEP) force22 refers to the force exerted on metal particles in non-uniform electric fields, driving their migration along the electric field gradient as a result of field-induced polarization. The electric force, which dominates over gravity in DC GIL, is expressed as13 where k denotes an undetermined correction coefficient that makes up for calculation mistakes in electric field strength at dielectric interfaces when the method of images is applied. When particles contact with HV conductor or shell, k = 0.832; in other condition, k is equal to 1. As experimentally validated in23, when k = 0.832, the relative error between the predicted and experimentally measured values of electric field force is below 3% at the contact point. The particle charge is equivalent to the surface integral of the electric displacement vector D over the surface of metal particle24. Therefore, the particle charge after collisions with HV conductor or shell is expressed as25 where represents the particle radius, and UDC is the applied DC voltage.Main stresses imposed on metal particles in the GIL are summarized in Table 2. Since, , and are dependent on particle shape, these formulas are applicable only to spherical particles. Table 2Force on metal particle.ForceDirectionExpressionElectric Force+RFq = k|q−E|-RFq = k|q+E|Gravity-zG = 4π3ρpg/3Buoyancy+zFf = 4π3ρgg/3Viscous Drag Forces-vFres =DEP Force-RFDEP = ρp and ρg denote the densities of metal particles and SF6, respectively; v represents particle velocity, and R is the radial distance from particle centre to HV conductor. Force on metal particle. ρp and ρg denote the densities of metal particles and SF6, respectively; v represents particle velocity, and R is the radial distance from particle centre to HV conductor. Reflection velocity after collision Based on inelastic collision, after the particle collides with electrode, its normal coefficient of restitution ke, which depends on particle material, is defined as where and are the normal components of the incident velocities and , respectively. When the particle collides with the electrode surface, its velocity decreases and its reflection angle varies randomly. Figure 2 shows the random reflection range of particle after the collision. Here, the random reflection range is represented by an orange-colored conical area and the light green plane denotes the incidence plane. The collision reflection process is described as follows: Random reflection range. XOY is the collision plane, is symmetric with about the Z-axis. Firstly, incidence velocity of particle is given by where and are the normal and tangential components of , respectively. The random reflection angle, which reflects the influence of surface roughness20, can be described as where α is the pre-collision incident angle, φ represents the maximum angular deviation, and σ∈ [−1,1] means a random factor. When σ = 1, the angle between the line of reflection of metal particles and the surface normal line reaches its maximum value, and at this time, it is the maximum reflection angle. Then, for α ≠ 0, particle velocity after collision is denoted as where manifests the reflection velocity, and and are the normal and tangential components of , respectively. Lastly, 3D reflection velocity is obtained through rotating about by rotation angle δ : where denotes the Rodrigues rotation matrix20, N represents the unit vector along the axis of rotation, R(N) is the cross-product matrix of , corresponds to the rotation angle, and refers to the identity matrix. Simulation of particle motion This section employs the electrostatics module and particle tracing module from the multi-physics coupled simulation software to solve the electric field distribution and particle trajectories in the cavity of GIL, respectively. The simulation flowchart is shown in Fig. 3, as detailed as follows: Flow chart of particle motion simulation. (i)3D model construction. Import a pre-designed 2D GIL model into the geometry module, and subsequently transform it into a 3D GIL model via rotation. 3D model construction. Import a pre-designed 2D GIL model into the geometry module, and subsequently transform it into a 3D GIL model via rotation. (ii)Parameter configuration. Define the critical parameter, including HV conductor voltage UDC = 40 kV, gas properties, particle radius = 0.15 mm. The solver configuration adopts the adaptive time-stepping scheme and the total duration t = 0.5 s. Parameter configuration. Define the critical parameter, including HV conductor voltage UDC = 40 kV, gas properties, particle radius = 0.15 mm. The solver configuration adopts the adaptive time-stepping scheme and the total duration t = 0.5 s. (iii)Mesh refinement. This study performs mesh generation independently for the electrode, insulator, shell, and gas domain in GIL. Specifically, boundary layer meshing is applied to the shell, curvature-based refinement combined with boundary layer meshing is used on the insulator surface, tetrahedral elements are employed for the gas domain, and adaptive meshing is implemented in the vicinity of the electrode. When a particle collides with an electrode or insulator, it experiences an instantaneous change in force accompanied by electron transfer. This phenomenon necessitates the refinement of mesh density on the surfaces of electrodes and insulators26. The mesh generation result is illustrated in Fig. 4. Mesh refinement. This study performs mesh generation independently for the electrode, insulator, shell, and gas domain in GIL. Specifically, boundary layer meshing is applied to the shell, curvature-based refinement combined with boundary layer meshing is used on the insulator surface, tetrahedral elements are employed for the gas domain, and adaptive meshing is implemented in the vicinity of the electrode. When a particle collides with an electrode or insulator, it experiences an instantaneous change in force accompanied by electron transfer. This phenomenon necessitates the refinement of mesh density on the surfaces of electrodes and insulators26. The mesh generation result is illustrated in Fig. 4. (iv)Electrostatic field computation. Set the ground and applied voltages, and then utilize GMRES iterative method to solve the steady-state electric field equation. Electrostatic field computation. Set the ground and applied voltages, and then utilize GMRES iterative method to solve the steady-state electric field equation. (v)Particle motion simulation. The particle motion equations are solved by the implicit Runge-Kutta method. Particle motion simulation. The particle motion equations are solved by the implicit Runge-Kutta method. Mesh refinement. Electric field in GIL The electric field distribution, as shown in Fig. 5, indicates a significant electric field concentration near the interface between the HV conductor and the basin-type insulator, with a field strength of approximately 4 × 103 V/m. This is due to the great differences in the relative permittivity of aluminum, epoxy resin, and SF6. As illustrated in Fig. 6(a), the radial component of electric field strength on the convex side (AB) of insulator diminishes near the shell, as the electric field tends to orient perpendicularly to the conductor surface. On the concave side (CD), the radial field strength rises as z increases owing to the influence exerted by HV conductor. In the region between the concentric cylinders (EF), the radial field strength exhibits a gradual decrease with an increase in Z-position. As depicted in Fig. 6(b), the axial field strength is roughly one-tenth that of the radial field strength. The axial field direction reverses at approximately z = −30 mm for the convex side and z = −50 mm for the concave side due to the alterations in the insulator curvature and boundary conditions. In the region between the concentric cylinders, the axial electric field strength is nearly constant and approaches zero. Electric field distribution in GIL. Electric field. (a) Radial component, (b) Axial component. Influence of reflection angle on particle distribution The random reflection angle is an important factor affecting the range of motion of particles. As derived from () and (), upon collision with electrodes or insulators, the reflection velocity of particles is redistributed between radial and axial directions according to the reflection angle. Consequently, φ serves as a critical factor determining the coupling effect between the radial and axial velocity components of particles. The 3D distribution of particles under different φ is demonstrated in Fig. 7, where the number of particles is set to 500. As φ increases, the radial dispersion of particles broadens. Specifically, these particles near the HV conductor exhibit lower velocities in compared to those adjacent to the shell. From Fig. 8, the following observations can be made: (i) At φ = 5°, all particles are confined within ± 20°. (ii) At φ = 10°, 81.8% of particles remain within ± 20°, demonstrating well agreement with experimental results reported in20. (iii) At φ = 30 °, only 20.8% of particles are distributed within + 20°. This study uses particle activity27 to characterize the extent of particle dispersion in the radial plane. The results indicate that an increase in the reflection angle greatly expands the particle activity. 3D distribution of particles under different φ. (a)5°, (b)10°, (c)15°, (d) 20°, (e) 25°, (f) 30°. Radial distribution of particles. Motion characteristics of particle in coaxial cylinder The particle motion in coaxial cylinder is defined as the motion in which the particle does not collide with insulator within the simulation time. Radial and axial motion characteristics Figure 9 depicts the projections of motion trajectory of single particle from multiple perspectives. Notably, the radial projection indicates that the particle primarily resides within the lower half of the cavity. The axial projection reveals that metal particle is propelled towards the convex surface of insulator due to the influence of axial electric field forces. Figure 10 illustrates the particle velocity in the Z-direction (Vz), which increases before collision but decreases afterward due to energy loss caused by collision. Figure 11 shows the instantaneous axial positions of particle. The velocity direction reverses at 0.08 s,0.21 s,0.27 s,0.32 s, and 0.34 s due to the collisions. The collision time interval is governed by the radial component of particle velocity. A greater radial velocity component leads to a shorter collision time interval. Projections of motion trajectory of single particle. (a) Radial projection, (b) Axial projection. Particle velocity in the Z-direction. Instantaneous axial position of particle. 3D motion trajectory Figure 12 depicts the variation in charge of particle. When stationary on the shell, particle acquires negative charges through electrostatic induction. Conversely, when colliding with the HV conductor, particle carries positive charges. The charge state of the particle remains unchanged until it encounters another collision with either the HV electrode or the enclosure. Figure 13 illustrates the electric field force on particle during its motion. The maximum and minimum values of Fq mean that the particle collides with the HV conductor and the shell, respectively. The charge polarity and motion direction of particle undergo changes at this time. Due to the distortion of electric field near the HV conductor, the electric force exerted on the particles is significantly enhanced; conversely, it decreases near the shell where the electric field intensity is low. Figure 14 shows the trajectory of particle between electrodes. In the case of upward electric field force surpassing gravity force, the particle is lifted towards the HV electrode. It subsequently descends towards the shell under the influence of gravity and electric field forces, ultimately migrating to the convex surface of insulator and experiencing repeated collisions between the electrodes. The results indicate that metal particles exhibit directional migration toward the convex surface of insulator under the influence of the resultant force. Consequently, the convex surface of insulator becomes a high-risk area for the accumulation and collision of particles. Particle charge. Electric field force. 3D motion trajectory. Influence of release position on axial motion characteristics To investigate the lifting-off phenomenon of metal particle, the initial position of metal particle is set on the shell directly beneath HV conductor. For the same release position, we conduct 10 repeated simulations. Table 3 presents the mean of final position at 0.5 s, as well as the standard deviation (SD) and offset in x-direction, in the case of five different release positions. The results indicate that all particles migrate to the convex surface of insulator regardless of the release position. Consequently, the release position has no impact on the axial motion characteristics of particles. Table 3Influence of release position.Release position (mm)Mean of final position (mm)SD in x-directionOffset in x-direction (mm)(120, 0, −60)(94, −10.2, −53.35)5.426(140, 0, −60)(113.9, −38, −39.8)4.226.1(160, 0, −60)(126, −18.8, −53.8)3.834(180, 0, −60)(148, −26.8, −47.9)3.932(200, 0, −60)(179.2, −27.1, −49.5)3.530.8 Influence of release position. Motion characteristics of particles near the basin-type insulator The particle motion near insulator is defined as the movement of particle located directly beneath the convex surface of insulator. The electric field near insulator exhibits significant variation, necessitating an in-depth investigation into the motion characteristics of metal particles in this region. Unlike the reference condition, the initial position of particle is set at (80, 0, −60) mm with a simulation duration of 0.25 s. Figure 15 illustrates the particle motion trajectory near insulator. After being lifted by the electric field force, the particle collides the insulator surface, subsequently climbing along it towards the HV conductor. When reaching the HV conductor, it acquires opposite charges, then falls back to the shell under the combined influence of gravity and electric field forces, ultimately colliding between the electrodes. The first collision times on the HV conductor and the shell are recorded as 0.055 s and 0.073 s, respectively. Trajectory of single particle near insulator. Influences of key parameters Applied voltage Figure 16 illustrates the radial motion trajectory projections of single metal particle in GIL under different voltages. Figures 17 and 18 depict the influence of applied voltage on the instantaneous axial position and velocity in the Z-direction, respectively. As the applied voltage increases, the range of radial motion expands, the maximum radial velocity exhibits nonlinear growth, and the frequency of radial velocity direction change intensifies. These phenomena arise from the synergistic effects of multi-physics coupling and collision dynamics: (i) DEP force-driven radial acceleration. Particle motion is governed by the net force: where . Consequently, higher voltages enhance DEP forces, amplifying radial acceleration. It results in the expansion of trajectory projection areas in Fig. 16 and increases collision frequencies in Fig. 18. (ii) Collision induced tangential velocity modulation. According to (), the axial displacement increments has positive correlation with . (iii) Multi-physics balance limiting velocity. Buoyancy partially counteracts gravity, restricting the maximum axial displacement in Fig. 17. Furthermore, the growth rate of Fres exceeds that of DEP force at 100 kV, causing slope-reducing of velocity curve in Fig. 18. Radial motion trajectory projections of particle for different voltages. UDC = 10 kV, 40 kV, 70 kV, 100 kV, respectively, from left to right. Instantaneous axial position versus time for different applied voltages. Speed of particle versus time for different applied voltages. Particle radius Figure 19 shows the motion trajectories of metal particle with different radius. The maximum axial displacements of particles with radius of 0.05 mm, 0.15 mm and 0.25 mm are 24 mm, 15 mm and 10 mm (see Fig. 20), respectively. As the particle radius increases, the radial motion range decreases, and the maximum axial displacement exhibits nonlinear attenuation. The above phenomenon is attributed to the synergistic effect between electric force-dominated motion and collision reflection mechanisms: (i) Electric force-driven radial motion. From Fq ∝ and m ∝ , it can be deduced that radial acceleration has a negative correlation with particle size. Consequently, smaller particles have higher radial accelerations, resulting in broader radial motion ranges. (ii) Collision-induced axial displacement enhancement. increases for smaller particles due to their higher reflection efficiency according to (). It leads to a greater axial displacement increments as observed in Fig. 20. Radial motion trajectory projections of particle for different radius. r = 0.05 mm, 0.15 mm, 0.25 mm, respectively, from left to right. Axial displacement versus time for different radius. Particle material Aluminum particles are generated through mechanical wear or arc ablation of aluminum conductors, iron particles originate from the wear of ferromagnetic components such as bolts, and copper particles are primarily attributed to the HV conductor and electromagnetic shielding components1. Table 4 shows the collision frequency of particles made from different materials. Figure 21 compares Fqz (i.e., the radial component of Fq) and G exerted on metal particles. Figure 22 depicts the time-dependent characteristics of Vz. Aluminum particles exhibit high-frequency collision behavior with periodic peaks of velocity due to their low density and permittivity. On the contrary, copper and iron particles display low-frequency motion mode. Our simulation results exhibit strong consistency with the experimental data reported in28, and the accuracy can be rigorously validated by quantitatively assessing the relative error (RE). These errors primarily arise from the fact that the value of η and ke are set to constants in the simulation, as well as the limited accuracy of numerical computations. Table 4Comparison of simulation results with experimental data.Materialρp (kg/m3) Collision frequency (Hz)RE (%)Simulation resultsExperimental data [28]Aluminum27000.5521.42209.1Iron78000.7415.6816.86.7Copper85000.7919.3717.113.2 Comparison of simulation results with experimental data. Fqz versus time for different materials. GFe = −1.077 × 10− 6 N, GAl = −3.7 × 10− 7 N, GCu = −1.17 × 10− 6 N. Vz versus time for different materials. Conclusions As a key parameter for three-dimensional distribution of metal particles in GIL, the reflection angle increases from 5° to 30°, resulting in radial divergence enhancement (proportion of particles within ± 20° are decreased from 100% to 28.8%) and axial displacement amplification (the displacement of peak particles concentration is increased from 4.4 mm to 35.8 mm). As a key parameter for three-dimensional distribution of metal particles in GIL, the reflection angle increases from 5° to 30°, resulting in radial divergence enhancement (proportion of particles within ± 20° are decreased from 100% to 28.8%) and axial displacement amplification (the displacement of peak particles concentration is increased from 4.4 mm to 35.8 mm). The electric field near the convex side of insulator is significantly enhanced, which causes the particles to migrate towards the convex side. Concave geometry induces a reversal in electric field gradient that drives particle migration toward shell. 3D simulation results reveal that metal particles in GIL ultimately gather around the convex side of insulator after repeated collision-migration cycles and exhibit a surface-climbing behavior along the profile of insulator. The electric field near the convex side of insulator is significantly enhanced, which causes the particles to migrate towards the convex side. Concave geometry induces a reversal in electric field gradient that drives particle migration toward shell. 3D simulation results reveal that metal particles in GIL ultimately gather around the convex side of insulator after repeated collision-migration cycles and exhibit a surface-climbing behavior along the profile of insulator. Applied voltage has positive correlation with the radial motion range as well as the maximum axial displacement of particles, while significantly increasing collision frequency of particles between electrodes. Besides, particle vitality, which is less affected by the material, decreases as particle size increases. Moreover, the collision frequency is ranked as follows: aluminum spheres > copper spheres > iron spheres. Applied voltage has positive correlation with the radial motion range as well as the maximum axial displacement of particles, while significantly increasing collision frequency of particles between electrodes. Besides, particle vitality, which is less affected by the material, decreases as particle size increases. Moreover, the collision frequency is ranked as follows: aluminum spheres > copper spheres > iron spheres. In the future, we will experimentally investigate the motion behavior of metal particles within DC GIL to validate the particle trajectories and aggregation zones predicted by the 3D simulations, and the design of particle traps to suppress the hazards of metal particles based on our results. In the future, we will experimentally investigate the motion behavior of metal particles within DC GIL to validate the particle trajectories and aggregation zones predicted by the 3D simulations, and the design of particle traps to suppress the hazards of metal particles based on our results. Author contributions Conceptualization, writing – review & editing and funding acquisition, Feng Bin. Writing – original draft, Zhensheng Tan. Software and methodology, Jixiang Feng. Visualization and validation, Deguang Wu. Writing –review & editing, investigation and formal analysis, Qiuqin Sun. All authors have read and agree to the published version of the manuscript. Funding This work was jointly supported by the National Natural Science Foundation of China (No. 52307157) and the Natural Science Foundation of Hunan Province, China (No. 2025JJ90178). Data availability Data is provided within the manuscript or supplementary information files. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Data is provided within the manuscript or supplementary information files.