Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 Porto, Portugal, 30 June - 2 July 2014 A. Cunha, E. Caetano, P. Ribeiro, G. Müller (eds.) ISSN: 2311-9020; ISBN: 978-972-752-165-4 An Integrated Ice-Shedding Model of Electric Transmission Lines with Consideration of Ice Adhesive/Cohesive Failure Ghyslaine McClure 1, Kunpeng Ji2, Xiaoming Rui 2 Department of Civil Engineering and Applied Mechanics, Faculty of Engineering, McGill University, Montreal, Canada 2 Department of Power Machinery Engineering, School of Energy Power, Mechanical Engineering, North China Electric Power University, 102206, Beijing, China email:, ghyslaine.mcclure@mcgill.ca; jkp2135@gmail.com; rxm@ncepu.edu.cn 1 ABSTRACT: This research is an attempt to propose an integrated accreted ice failure model for iced overhead line conductors that will lead to more realistic nonlinear dynamic analysis of the ice-shedding phenomenon of transmission lines, by taking into account the adhesive / cohesive strength of ice deposits. Ice shedding induced by sudden mechanical forces is understood to be a two-stage process. First, the continuous ice deposits along the conductor span are broken into smaller separate ice chunks and fragments (ice fracture failure), and then these fragments detach from the cables and fall off due to insufficient cohesive strength within the ice or adhesive strength at the icecable interface (ice detachment failure). Two recent successive studies have developed computational models using ice deposit failure criteria based on the maximum effective plastic strain and the maximum bending stress. These models have yielded reasonably accurate results in predicting cable tensions and mid-span displacements, by comparing their numerical results with experimental data from tests carried out on a 4m reduced-scale model span with varying cable diameters and ice thickness, following sudden mechanical shock loads. However, there is still about 20% disparity in ice fracture rates between the computational results and experimental data, and it is deemed necessary to refine the ice failure model to introduce the effects of adhesive / cohesive forces. Therefore, as the first step towards the development of an integrated two-tier ice shedding criterion, the authors have improved the previous FE models, in terms of mesh size, load types and locations, material models and so on, to provide a better description of the experiment results. Then, the refined FE models of the reduced-scale span tests are used to check the newly proposed ice adhesive /cohesive failure criterion. The idea of this criterion is to simply compare the inertia forces acting on the fractured ice segments, and the ice adhesive strength or cohesive strength. The process is done automatically by a subroutine interacting with the nonlinear dynamic analysis commercial software ADINA. Although there is no satisfactory model to calculate the adhesive and cohesive strengths of glaze ice - especially for atmospheric ice, only several representative pairs of values are selected. Validation is underway to ascertain that the proposed two-tier glaze ice shedding criterion provides a more realistic description of the ice-shedding phenomenon of transmission lines than in the previous studies. KEY WORDS: Nonlinear dynamics; Shock loads; Computational models; Ice failure. 1 INTRODUCTION Atmospheric icing is one of the major threats to the security of overhead electric transmission lines in cold regions. These threats can be classified into two categories: electric ones, (such as flashovers, outages), and mechanical ones (such as galloping, overload icing, uneven icing, ice shedding, etc.). During the Great Ice Storm of January 1998 in North America, the losses on the Hydro Québec transmission grid alone included: 600 steel tower collapses and 100 damaged towers, and 16 000 line components failure (poles, cross-arms, hardware, cables) in the distribution network [1]. An even more destructive ice storm hit South China in 2008, which resulted in more than 140 000 collapsed towers on 10~110 kV lines and 1 500 towers on the transmission grid (above 220 kV) operated by the State Grid company [2]. After these events, a large number of anti-icing (referring to methods which are used before or during the early stage of icing to prevent the accumulation of ice on cables) and deicing techniques (referring to methods which are employed during or after icing to remove accreted ice from cables) have been proposed and tested by researchers and engineers from all over the world. A technical brochure edited by CIGRE Working Group B2.29 [3] gives a comprehensive review on the operational and potential anti-icing and de-icing methods, which can be divided into four categories: passive methods, active coating methods, mechanical methods and thermal methods. Among these methods, mechanical methods shows clear advantages in de-icing ground wires and short sections of strategic lines as timely and fast intervention, because they are easy to apply and cost effective. A portable mechanical de-icing device, DAC (De-icer Actuated by Cartridge), proposed by Hydro Quebec, is the focus of this research. DAC takes advantage of the brittle characteristics of glaze ice at high stain rates (>10-3/s). So, when the shock wave travels along the span, it can break the ice accreted on cables into small fragments by releasing the energy of shock waves. Details of this device can be found in [4]. A reduced-scale single span physical model was tested in the CIGELE laboratory at UQAC, Chicoutimi, Canada, to test the effect of this device and to evaluate its impact on supports and the jump heights of midpoint, which may give useful information on the design and optimization of DAC[5]. 3731 Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 Two recent successive studies have developed computational models using ice deposit failure based on the maximum effective plastic strain and the maximum bending stress, to model the “unzipping” process when shock loads were applied to iced cables [5-8]. These models have yielded reasonably accurate results in predicting cable tensions and mid-span displacements, by comparing their numerical results with experimental data from tests carried out on the 4m reduced-scale model span with varying cable diameters and ice thickness. However, there is still about 20% disparity in ice fracture rates between the computational results and experimental data, and it is necessary to refine the ice failure model to introduce the effects of adhesive/ cohesive forces. This research is an attempt to propose an integrated accreted ice failure model that will lead to more realistic nonlinear dynamic analysis of the ice shedding phenomenon of transmission lines, by taking into account the adhesive / cohesive strength of the ice deposits. 2 Four scenarios are studied in the tests, though only the scenario with 1 mm of equivalent radial ice thickness on a cable with the diameter of 4.1 mm will be presented in this study. Since the round cross section cannot be used for the iso-beam element, the accreted ice is modeled using an equivalent iso-beam with rectangular section with the same area and second moment of area as the idealized round cross section: The resulting dimensions are a width of 6.36 mm and height of 2.52 mm. The impact force used for this scenario is shown in Figure 2. SENSITIVITY STUDY Ice shedding induced by sudden mechanical forces is understood to be a two-stage process. First, the continuous ice deposits along the conductor span are broken into smaller separate ice chunks and fragments (ice fracture failure), and then these fragments detach from the cables and fall off due to insufficient cohesive strength within the ice or adhesive strength at the ice-cable interface (ice detachment failure). So, as the first step, a series of sensitivity study will be done on the basis of previous studies. Figure 2. Characteristics of the impact force [5] Two cable material models (Figure 3) are used: one is the theoretical tension-only model (MAT1) with a constant Young’s modulus of 172.4 GPa, and the other one (MAT2) is a multi-linear tension-only model whose values are got from static tensile test of the cable specimen. General modeling assumptions 2.1 Since previous research using the nonlinear dynamic analysis commercial software ADINA [6, 8-11] has yielded good simulation results compared with experiments, , it is also used in this study using similar modeling assumptions and computational methods. The iced cables were modeled by paralleling ice elements and cable elements which share the same end nodes, as shown in Fig. 1. [5, 6]. The conductor is modeled as a mesh of 3-D two-node iso-parametric truss elements with tension-only material properties, and the ice accretion is modeled with a parallel mesh of general 3-D isobeam elements, with bilinear plastic material model. The total Lagrangian formulation is used for this large displacement but small strains problem in ADINA [12, 13]. Aerodynamic damping is neglected, and an equivalent viscous damping of 2% critical is used to model the structural damping of the iced conductor, by using a nonlinear axial dashpot element in parallel to each cable element. More details about the selection of damping constant are discussed by Roshan Fekr et al.[9] and McClure and Lapointe[11]. a) b) Figure 1. FE model of the 4 m single span tests [5] Figure 1 a) shows the whole FE model , and b) shows the diagram of the iced cable FE model at the element level. 3732 a) MAT1 b) MAT2 Figure 3. Material models of the cable [5] The ice material model has a Young’s modulus of 10 GPa, Poisson’s ratio 0.33, initial yield stress 2 MPa, and maximum allowable effective plastic strain 10-10 (for the maximum normal stress failure criterion) or maximum allowable effective plastic strain 9.756×10-5 (for the maximum normal strain failure criterion). 2.2 Mesh size analysis The motivation for a mesh size analysis stems from the following: 1) It is essential to achieve convergence and stability when the number of elements are large to achieve convergence and stability. This was not examined in previous studies as the meshes were relatively coarse with only 20 to 30+ cable elements per span; 2) the experiments showed that the ice accretion remaining on the cable was broken into small fragments with an average length of around 5 mm, which means that fracture did occur within every ice element and between its integration points, thus leading to a rate of ice shedding (RIS) of 100% for the mesh size of 25 (element length of 160 mm); 3) the impact pulse was applied at the Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 midpoint of the span in these tests, and the measured displacement values are also for the midpoint, but there was a 80 mm drift from the mid span when the shock loads were applied at node 13 in the 25-element model. The mesh size sensitivity analysis is conducted from 25 to 4000 cable elements per span. The ice element number changes to accommodate the fact that there is a 30 cm length of cable with no ice, at the far ends of the spans owing to the limitation of the spraying system[5]. observed when the 1000 elements FE model was used, as shown in Figure 5. a) Figure 4. Comparisons of cable tensions with different mesh densities As is shown in Figure 4, the computational results of tension of the left end cable element begin to converge until no less than 400 elements were used. The comparison of mid span displacements showed the same trend, although not shown here due to space limitations. Clearly, 25 cable elements (21 ice elements) are not sufficient for this research, and the mesh with 1000 cable elements (850 ice elements) will be used throughout this study. It is also seen in Figure 4 that the finer mesh models better predict the trend after the first peak, and that the finer the mesh model is, the first peak occurs earlier in time. Besides, the time step is set to 0.05ms for the first 2000 steps (100 ms) and then to 0.25 ms for the remainder. The reasons are: 1) the maximum peak of the shock load arrives at approximately 0.20-0.25ms, so if the time step is set to be 0.25ms as in previous studies, the shock load will be applied to the system from 0 to maximum in a single step, which may cause computational instability of the nonlinear model; 2) after the duration of shock load, a larger time step can help to accelerate the computation and save time. 2.3 Material characteristics In previous studies, different cable material models were used to get results of cable tension and mid span displacement separately in the same scenario, that is the MAT1 cable material model with greater tensional rigidity (EA= 2,275,680 N) was used to generate the time history of midspan displacement, and the MAT2 material model with less tensional rigidity (EA= 346,500 N) was used in its initial linear section (0-105MPa) to obtain the time history of cable tension [5, 7, 8]. The flexible cable model can result in a decrease of 52% of the maximum cable tension compared with the rigid one [5]. By doing this, the simulation results can better agree with the experimental data. This was also b) Figure 5. Comparisons of different FE models a) cable tension at left support b) Displacement at mid point However, the numerical models still overestimate the maximum value of both the cable tension and mid-span displacement. This discrepancy may be explained as follows: 1) The amount of ice detached from the cable as predicted by the FE simulation (rate of ice shedding R.I.S =100%) and that observed in the test (R.I.S =8%) are very different. This is partially validated by assuming a portion of 8% ice elements near the midpoint to “undergo element death” in ADINA when the shock load peak arrives (t=1.05025s), and setting the maximum allowable strain of ice to be an unrealistic large value (e.g. 1.0 ). Therefore, only these ice elements are removed at the early stage of the simulation, while the others will remain on the cable throughout the analysis. As a result, the computational maximum cable tension decreases to the level of values measured in the test, and the overall displacement tendency seems in accordance with test data, as shown in Figure 5. Also, the ice shapes in the tests are irregular, which is not considered in the present study. 2) The real tensile rigidity (EA) of a stranded cable is not constant but varies with time and cable deformation, which cannot be considered in the present FE model. 3) The bending rigidity (EI) of the cable is totally neglected in the FE model by using truss elements. However, the bending rigidity of the cable is about 4.42 ( with MAT1 model, EI=2.3913 m2 ·N) or 0.67 ( with MAT2 model, EI=0.3641 m2 · N ) times the bending rigidity of ice (EI=0.5409 m2 ·N). This simplification may be useful and reasonable in simulating the global motion of cables (e.g. galloping and Aeolian vibration), but not for localized shock 3733 Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 loading which involves dramatic local cable bending motion and initiates wave propagation after the shock load. This model refinement is still ongoing. 4) The flexibility of supports (assumed perfectly fixed in the FE model) may also play an important role in the disparity of computational and experimental results. Besides, in the attempt to get more accurate numerical simulation of the tests, another way of applying the shock load is also simulated in the FE model. It divides the original force into three equal lumped forces distributed on a length of about 1 cm (i.e. on nodes 500, 501 and 502 of the model), which is equal to the diameter of the piston end. As shown in Figure 5 (with the legend ‘1000E_MAT2_1/3 load’), the results are similar to those obtained in previous model and the refinement is not deemed necessary. 3 PRELIMINARY STUDY OF ADHESIVE/COHESIVE ICE FAILURE CRITERIA Since the experiments showed that large amounts of ice were sticking on the cable after the shock loads, it becomes necessary to consider the de-icing process as a two-stage process. First, the continuous ice deposits along the conductor span are broken into smaller separate ice chunks and fragments (ice fracture failure), and then these fragments detach from the cables and fall off due to insufficient cohesive strength within the ice or insufficient adhesive strength at the ice-cable interface (ice detachment failure). 3.1 General concept of adhesive/cohesive ice failure criterion The main and simple concept of this criterion is to compare the inertia forces acting on the fractured ice segments, and the ice adhesive or cohesive stress resultants. the length of the ice fragment, the diameter of cable, τ cohesive a is the acceleration, Dcable is tice is the ice thickness, τ adhesive and are the adhesive and cohesive strengths of glaze ice. The first difficulty for the implementation of this failure criterion is to select realistic adhesive and cohesive strength values for atmospheric glaze ice. Although the physical and mechanical properties of ice have been studied for decades, a theoretical model to calculate the adhesive and cohesive strengths is still lacking, and experimental values reported in literature are both scarce and with large variability as ice is a highly complex natural material. The measured values vary considerably with many factors, such as ice deposit types, temperature, nature and texture of substrate, wind speed, test methods, and so on [14, 15]. In spite of these difficulties, one can get some general conclusions: 1) the adhesive strength values obtained in tensile tests are at least 15 times larger than that in shear tests [16, 17]; 2) the adhesive strength of ice-metal interface is larger than the cohesive strength of ice, contrary to polymeric materials [14, 18]; 3) the adhesive strength tested with high strain rates or with dynamic test methods is significantly less than that with low strain rates achieved in static or quasi-static tests; and 4) brittle fracture happens when ice is subjected to high tensile stress, and the adhesive strength is temperature independent in this instance, while ductile failure occurs when tensile stresses remain below a critical value, and the adhesive strength increases linearly when temperature decreases below 0℃ [14]. Several tests results are summarized in a recent study published in 2012, which shows the adhesive shear strength of ice-Aluminum and ice-Stainless steel interfaces varying between 0.002 to 1.96 MPa [19]. On the basis of these published test results, four pairs of adhesive and cohesive strength values were chosen for this research (with dcable = 4.1 mm, tice = 1mm), and the critical vertical accelerations are calculated as shown in Table 1. The calculated accelerations are very large because the mass of ice per unit cable length is very small; heavier deposits are easier to shed. Table 1. Examples of Adhesion/Cohesion Strengths and Critical Ice-shedding Accelerations Figure 6. Schematic diagram of ice detachment failure criteria No. As shown in Figure 6, if the inertia force of an ice segment is larger than its adhesive and cohesive stress resultants, the ice segment will detach from the cable. This can be described by Equations (1)-(4). (1) Finertia ≥ Fadhesive + Fcohesive πρ la ( Dcable + 2tice ) − Dcable 2 8 Fadhesive = Dcablel τ adhesive Fcohesive = 2ticel τ cohesive F= inertia 2 (2) (3) (4) Finertia , Fadhesive and Fcohesive are inertia ,adhesive and cohesive force separately. ρ is the density of glaze ice, l is where 3734 1 2 3 4 3.2 Adhesion Strength (MPa) Cohesion Strength ( MPa ) Critical Acceleration 0.005 0.01 0.05 0.25 0.004 0.008 0.04 0.20 4.0 7.9 40 198 ( 103m/s2 ) Acceleration distribution along the span Before applying the adhesive/cohesive ice failure criterion into the FE model, the maximum acceleration values of the whole model calculated in each time step of analysis are checked. The results show that the absolute maximum acceleration occurs first at the mid-span where the shock load is applied, and then moves to the two ends, as time increases. After reaching the ends, the absolute maximum values go Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 from the ends to the midpoint. Besides, the absolute maximum values decrease from the midpoint to each end before arrive the ends as time increases, as shown in Figure 7. It is worth noting that according to the ice fracture criterion, all the ice elements “have died” at 1.05125s, when the acceleration was 4,000 m/s2, and the absolute maximum acceleration occurs near node 480 and 520. A in the test and simulation, the changing cable tensile rigidity, and the neglecting of cable bending stiffness. The general idea of the newly proposed ice detachment failure criterion is presented, and several representative pairs of adhesive and cohesive values are selected to calculate the critical acceleration needed to shed off the broken ice fragments induced by the shock wave. The analysis of maximum acceleration values at each time step and each node, and the effective shedding ice (8%) make it possible to estimate the critical acceleration and the adhesive and cohesive strengths of ice in the test, which in turn validates the ice detachment failure criterion. The effort of integrating the user-supplied subroutine containing the proposed ice detachment failure criterion into ADINA is underway. ACKNOWLEDGMENTS a) This work was supported by the Fundamental Research Funds for the Central Universities of China (No. 12QX10) and the China Scholarship Fund. The authors also acknowledge Dr. T. Kálmán for providing of experimental data. The first author also acknowledges financial support from the Natural Sciences and Engineering Research Council of Canada. REFERENCES 1. 2. b) Figure 7. Maximum/Minimum acceleration values with respect to time a) Long time duration b) Detail of A So considering 8% ice shedding (i.e. 68 ice elements dead) actuated by the proposed ice detachment failure criterion, the critical acceleration should occur around nodes 467 and 535, which is predicted to be approximately 3,000 m/s2.With this critical value, the ice adhesive strength of this test can be estimated with Equations (1)-(4) and is approximately 0.004 MPa, which falls into the reasonable range shown in Table 1 (assuming that τ adhesive =1.25 τ cohesive ). This validates the feasibility of the proposed ice detachment failure criterion for the case under study. This process is to be programmed in a subroutine (written by the second author) which will interact with the nonlinear dynamic analysis commercial software ADINA. This implementation is still ongoing. 4 3. 4. 5. 6. 7. CONCLUSION This research is an attempt to better understand and simulate the conductor ice shedding phenomenon induced by shock waves. The 1000 cable elements FE model is proved to be adequate and able to simulate the nonlinear dynamic response of the physical reduced-scale tests. The discrepancies between the numerical results and experimental data are believed to be “reasonable” before considering ice detachment failure, and are analyzed in terms of the different amounts of ice shedding 8. McCLURE G, JOHNS KC, KNOLL F, PICHETTE G: Lessons From The Ice Storm Of 1998 Improving The Structural Features Of HydroQuébec's Power Grid. In; the 10th International Workshop on Atmospheric Icing of Structures. 2002 Hu Y: Analysis and Countermeasures Discussion for Large Area Icing Accident on Power Grid. High Voltage Engineering 2008, 34:215-219. CIGRE: Systems for prediction and monitoring of ice shedding, anti-icing and de-icing for power line conductors and ground wires. (B2.29 WG ed., vol. 438; 2010. Leblond A, Lamarche B, Bouchard D, Panaroni B, Hamel M: Development of a portable de-icing device for overhead ground wires. In the 11th International Workshop on Atmospheric Icing of Structures. pp. 399-404. Montreal; 2005:399-404. Kálmán T: Dynamic Behavior of Iced Cables Subjected to Mechanical Shocks. Université du Québec à Chicoutimi (Canada); 2007. Kalman T, Farzaneh M, McClure G: Numerical analysis of the dynamic effects of shock-loadinduced ice shedding on overhead ground wires. Computers & Structures 2007, 85:375-384. Mirshafiei F, McClure G, Kalman T, Farzaneh M: Improved ice shedding modelling of iced cables: A comparison with experimental data. In Annual Conference of the Canadian Society for Civil Engineering 2010; Winnipeg, Canada. 2010: 978987. Mirshafiei F: Modelling the dynamic response of overhead line conductors subjected to shockinduced ice shedding. Master's thesis, McGill University, Department of Civil Engineering and Applied Mechanics 2010. 3735 Proceedings of the 9th International Conference on Structural Dynamics, EURODYN 2014 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 3736 Fekr MR, McClure G: Numerical modelling of the dynamic response of ice-shedding on electrical transmission lines. In 7th International Workshop on Atmospheric Icing of Structures, 3-7 June 1996; Netherlands. Elsevier; 1998: 1-11. Jamaleddine A, McClure G, Rousselet J, Beauchemin R: Simulation of ice-shedding on electrical transmission lines using ADINA. Computers & Structures 1993, 47:523-536. McClure G, Lapointe M: Modeling the structural dynamic response of overhead transmission lines. Computers & Structures 2003, 81:825-834. Bathe KJ: Finite Element Procedures. Upper Saddle River,NJ, USA: Prentice Hall; 2006. IARD Inc.: ADINA Theory and Modeling Guide. Watertown, MA, USA.; December 2011. Raraty L, Tabor D: The adhesion and strength properties of ice. Proceedings of the Royal Society of London Series A Mathematical and Physical Sciences 1958, 245:184-201. Druez J, Nguyeh DD, Lavoie Y: Mechanical properties of atmospheric ice. Cold Regions Science and Technology 1986, 13:67–74. Jellinek HHG: Adhesive properties of ice. Journal of colloid science 1959, 14:268-280. Jellinek HHG: Ice Adhesion. Canadian Journal of Physics 1962, 40. Andrews E, Lockington N: The cohesive and adhesive strength of ice. Journal of materials science 1983, 18:1455-1465. Fortin G, Perron J: Ice Adhesion Models to Predict Shear Stress at Shedding Journal of Adhesion Science and Technology 2012, 26:523-553.
© Copyright 2024