Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process (2021)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

54 Description of Parametric Study Thermal-mechanical analysis of HMIPs was performed to evaluate the effect of structural configuration on nonlinear stresses and strains occurring during the HDG process. FE models were created for these HMIPs using the commercial FE software Abaqus (Simulia, 2018). Model Geometry Models of a thin pole with thick steel plates welded at both ends (Figure 5-1) were evaluated in this study. Nineteen models of poles with round and multisided (8, 10, and 12 sides) cross- sections were analyzed. One calibration model was created to match the geometric configura- tion of an instrumented HMIP used in previous research and determine the value of the heat convection coefficient of steel embedded in a galvanizing bath that would closely match the measured temperatures and strains (Kleineck, 2011). The other 18 FE models were included in a parametric study to evaluate the effects of different parameters on strains and stresses at the pole-to-base plate connection. The base plate connection was closely inspected after galvanizing to identify cracking that may have occurred during galvanizing. The pole-to-base plate connection was modeled to include a backing ring (Figure 5-1d) and welding details (Figure 5-1b, c, and e) conforming to Wyoming DOT specifications (Wyoming DOT, 2020) because of its prevalence in current practice (Goyal et al., 2012; Kleineck, 2011; Pool, 2010; Roy et al., 2011; Stam et al., 2011). All models consisted of four parts: (a) a base plate with 12 anchor bolt holes (Figure 5-1c), (b) a pole shaft (Figure 5-1a), (c) an upper plate (Figure 5-1a), and (d) a backing ring. Both the base and top plates were circular in shape with a circular opening in the center; the dimensions of the HMIP and its parts are listed in Table 5-1. The thickness of base plate used for calibration was 3 in., but it was varied in the parametric study from 1.5 in. to 3.5 in. The outer and inner diameters for the base plate were maintained at 47 in. and 22 in., respectively. The top plate—located at the opposite end of the assembly—had the same dimensions as the base plate except for being 1 in. thick, and it was included in the model to ensure convergence and model stability. All models were for a pole with a total length of 14 ft. For multisided poles, the bend radius of the pole vertices was 1.25 in. As shown in Figure 5-1, the backing ring was connected to both the pole and the base through fillet welds, while the base plate and pole shaft were attached through a full penetration weld. The backing ring was modeled as a multisided element with the same number of sides as the pole (12 for the model shown in Figure 5-1); it had a thickness of 0.25 in. and a width of 3 in. The interactions between the parts that were welded to the plates, pole shaft, and backing ring were simulated as perfect attachment. Interactions between the pole and the backing ring and C H A P T E R 5 Computational Simulations— High Mast Illumination Poles

Computational Simulations—High Mast Illumination Poles 55   between the pole and the base plates were simulated as surface contact interactions that trans- ferred compression between the two faying surfaces. Parameters studied were pole shape (circular or multisided), number of sides, base plate thickness, and pole thickness; Table 5-2 lists details of the models evaluated in the study. Material Properties The 18 models evaluated in the parametric study were analyzed considering linear-elastic and nonlinear material properties. Results derived from models considering nonlinear material properties are presented in this chapter. Nonlinear material properties accounted for tempera- ture effects on both mechanical and thermal properties. The steel material model was repre- sentative of Grade 50 steel, with properties equivalent to those of Grade 50 steel specified in ASTM A572. Temperature-dependent properties of steel reported by Pilipenko (2001) and Perić et al. (2014) were used in this study. Material nonlinearity was modeled using isotropic hardening with von Mises yield criterion. The same thermal and mechanical properties were assumed for (a) (b) (c) (e) (d) Figure 5-1. (a) HMIP model; (b) connection details (after Stam, 2009); (c) pole-to- base plate connection; (d) backing ring; and (e) modeling of welds. 0.25 1.0 Component Dimension (in.) Base plate diameter 47.0 Base plate thickness 3.0 Access hole diameter 22.0 Pole shaft diameter 32.625 Backing ring width 3.0 Backing ring diameter 32.0 Backing ring thickness Top plate thickness Table 5-1. HMIP dimensions.

56 Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process base and weld metals as adopted by other researchers (Chang et al., 2009; Chang & Lee, 2009; Ma et al., 1995; Siddique et al., 2005; Teng et al., 2001); the properties at RT for the Grade 50 steel material assumed in these models are listed in Table 5-3. Analysis Procedure Thermal-mechanical analysis was performed to simulate temperature-induced effects of the HDG process. As HMIPs contact the galvanizing bath, a significant change in temperature is induced in the pole at the point of contact, which creates expansion in the area of the pole immersed in the bath. As the surface area of the fluid/solid interface expands, convective heat transfer between the HMIP and the sink increases. As the structure deforms, different parts may come into contact with each other, causing mechanical interactions between components at different temperatures. In the analyses used in this study, displacement and temperature fields were solved iteratively until convergence was attained, with mechanical and thermal properties of steel updated in accordance with the calculated temperature field (Harte et al., 2004; He et al., 2007) and deformed shape of the structure. Simulation of the Galvanizing Process The galvanizing process was simulated by moving the pole assembly through a static tempera- ture field consisting of two regions: air at RT of 65°F and the galvanizing bath at 830°F (shown Model # Note # of Sides Dimensions (in.) Base plate thickness Pole thickness Pole diameter 1 Baseline model 12 3.0 5/16 32-5/16 2 Pole thickness variant 12 3.0 1/2 3 Geometric variants of model 1 Round 3.0 5/16 4 8 3.0 5/16 5 10 3.0 5/16 6 Geometric variants of model 2 Round 3.0 1/2 7 8 3.0 1/2 8 10 3.0 1/2 9 Base plate thickness variants of model 1 12 1.0 5/16 10 12 1.5 5/16 11 12 2.0 5/16 12 12 2.5 5/16 13 12 3.5 5/16 14 Base plate thickness variants of model 2 12 1.0 1/2 15 12 1.5 1/2 16 12 2.0 1/2 17 12 2.5 1/2 18 12 3.5 1/2 Table 5-2. Pole models evaluated in parametric study. Young’s Modulus (Ksi) Poisson’s Ratio Density ( ⁄ ) Thermal Expansion Coefficient ( °⁄ ) Specific Heat ( °⁄ ) Thermal Conductivity ( ∙ ∙ °⁄ ) 29,000 0.35 0.28 6.6×10-6 9.20×10-2 31.74 Table 5-3. Material properties for Grade 50 steel at RT (70çF).

Computational Simulations—High Mast Illumination Poles 57   schematically in Figure 5-2). External temperature and heat convection coefficient of nodes at the fluid/solid interface are a function of pole position according to a user-defined film sub- routine as specified in Abaqus. Thermal parameters of the ambient and hot zinc environments are listed in Table 5-4. The simulation replicated the four stages of galvanizing: dipping, dwelling, extraction, and cooling. Motion of the pole through the temperature field was controlled by adjusting the amplitude of the velocity at boundary locations corresponding to crane pick points, as illus- trated in Figure 5-3. Displacement in the out-of-plane direction was restrained by fixing a node set along the bottom edge of the plates to avoid rotation of the pole assembly. For most models, the simulation extended for 900 sec, consisting of 180 sec for dipping, 300 sec for dwelling, 360 sec for extraction, and 60 sec for cooling. In a few instances the simulation extended to 3,400 sec, with all stages lasting the same as other models except for the cooling stage, which was extended to 2,560 sec to evaluate longer-term cooling effects. Figure 5-2. Four stages of the galvanizing process: dipping, dwelling, extraction, and cooling. Ambient Condition Temperature ◦C (◦F) Heat Convection Coefficient W/(m2 K) (Btu× in./sec in.2 ◦F) Molten Zinc Bath 445 (830) 1500 (5.095 × 10-4) Air 18 (65) 1000 (3.397 × 10-4) Table 5-4. Thermal parameters. Figure 5-3. Locations of applied restraints and velocity boundary conditions.

58 Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process Dipping angle and immersion/extraction speeds vary significantly in practice; 8° and 13 millimeters/sec (0.5 in./sec) were adopted for all the simulations. Element Mesh All model parts were discretized using eight-node reduced integration coupled temperature- displacement continuum elements—designated as C3D8RT in Abaqus—with hourglass effect control. Element size at the weld toe connection per AASHTO load and resistance factor design (LRFD) specifications (AASHTO, 2015) was adopted; mesh size was controlled by the shaft thickness t = 5⁄16 in. AASHTO (2015) stipulates that a maximum mesh size of t × t should be used for at least three rows of elements from the top of the welds, and at least two rows of ele- ments in the thickness direction. In addition, mesh size at the interface region should not exceed a 1:4 aspect ratio. The size of the elements at the interface region was 0.2 in. × 0.2 in.; it was the same along the length of the pole up to a distance of 4 in. from the interface. The pole thickness was discretized into two elements according to AASHTO (2015). The aspect ratio of elements at the interface region was approximately 1:1. Portions of the base plate, backing ring, and weld near the inter- face zones were also discretized with an element size of 0.2 in. × 0.2 in. Element size away from the interface region was increased gradually from 0.2 in. to 3 in. at the middle of the pole. The mesh pattern was symmetric with respect to the middle of the pole. Model Calibration Thermocouple measurements of temperature at the base plate and at the shaft of the pole assembly during galvanizing were used to calibrate the baseline FE model. The calibration was performed using the data reported by Kleineck (2011) for an instrumented pole. The purpose of the calibration was to validate the combination of mesh configuration and the value of heat convection coefficient of steel immersed in zinc that was used in the models in this study. This calibration was necessary because the calculated temperatures, stresses, and strains are sensi- tive to mesh configuration. In the calibration process, the heat convection coefficient of steel in molten zinc provided in Kleineck (2011) was adopted as a reference value. Several analyses were performed for different element sizes and the heat convection coefficients of steel in molten zinc to determine the optimal parameters for the refined mesh (details are provided in Appendix D). Results Poles do not deform in a symmetric pattern during galvanizing in spite of the uniform tem- perature; temperature-induced deformations cause significant distortions of pole assemblies, both along the length of the pole and at cross-sections perpendicular to the longitudinal axis of the pole. These distortions occur because immersion of the pole is not instantaneous, and they occur regardless of the orientation of the assembly while it is being submerged. Typical assem- blies have poles that are significantly thinner than their base plate. Thin poles absorb the heat from the galvanizing bath very rapidly and are very flexible, and the opposite is true for thick base plates. The mismatch between the pole and base plates induces significant stresses during immersion at pole-to-base plate connections, largest at the bends of multisided poles. Round shapes eliminate discontinuities in geometry along their circumference and reduce stresses caused by concentration factors. Stresses and strains were found to be larger during immersion than those that occurred during extraction. The largest strains at the pole-to-base plate connection were found to occur at the

Computational Simulations—High Mast Illumination Poles 59   level of the surface of the zinc bath, where the temperature difference between the immersed region of the pole and the region outside of the bath was the largest. Stresses were significantly lower during extraction because a much slower rate of temperature change in steel occurs as the specimen is removed from the bath than that which occurs when it is immersed. During the cooling phase, the pole shaft cools faster than the base plate, generating more stresses and strains at the welded connection, particularly at bends of multisided poles. The stresses that occur at the bends of multisided poles during extraction are much more prolonged than those occurring during the immersion, even though the maximum stresses are lower than those experienced during immersion. Data Extraction and Analysis The largest stresses and strains in HMIPs were found to consistently occur at the pole-to- base plate connection during galvanizing; therefore, this region was the focus of the compu- tational simulations. Because calculated strains and stresses at each node varied with time during the different stages of the simulation (dipping, dwelling, extraction, and cooling), it was essential to extract peak values of temperature, stress, and strain and to observe the trends of how these quantities varied with time. Analyses showed that stresses and strains peaked at different times for different locations in the HMIP assembly but typically when each location contacted the galvanizing bath. Because these strains and stresses were both time-dependent and location-dependent, it was not appropriate to choose a single time and location to compare the different models. Data extraction graphing routines were developed using MATLAB and Python to extract and visualize data (particularly strains and stresses) from various “paths” of elements within each model across the model time-history (examples provided in Appendix D). This meth- odology was used to consider the maximum stresses and strains within the entire region of the pole adjacent to the end plate, taking into account changes in location and time where the peaks occurred. Variation of Stress and Strain with Respect to Time and Location Strain metrics provide a measure of deformation in the nonlinear range of response that increases with damage. This study examined strain and stress response in HMIPs during gal- vanizing; strain results are presented in terms of the equivalent plastic strain (or accumulated plastic strain) and is designated PEEQ. PEEQ is an indicator of locations with the highest potential for crack initiation (He et al., 2018); it is used for evaluating the results for different HMIP models. Figure 5-4 shows the PEEQ versus time at a critical bend location for three poles with the same thickness and configurations but different base plate thickness (1 in. and 3 in.) and pole shape (12-sided and round). An examination of PEEQ results versus time revealed different behaviors for the three poles. The 12-sided pole with a 3-in. thick base plate experienced two large increases in strain: first when the bend location first contacted the galvanizing bath during the dipping phase, and another increase when the same bend came into contact with air during the extraction phase. The round pole with a 3-in. thick base plate experienced a smaller strain increase during the dipping phase and no inelastic deformations during the extraction phase. The 12-sided pole with a 1-in. thick plate did not experience inelastic deformations during the dipping or extraction phase. These responses indicate that the two critical times for LMAC for- mation are those when the pole region with high SCFs comes into contact with the galvanizing bath and when it exits the galvanizing bath. The first occurrence increases the plastic strain (and stress) almost instantaneously, while the second occurrence results in a more gradual increase.

60 Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process If damage accumulates as indicated by the PEEQ values, it is likely that the weldment is first affected during the dipping stage such that measurable cracks form during extraction due to the additional inelastic deformations. Also, PEEQ values indicate that reductions in base plate thickness and the use of round-shaped poles are likely to reduce damage during galvanizing. Figure 5-5 shows the maximum stress and strain values computed for the entire galvanizing process for models for five 12-sided HMIPs with a pole thickness of 5⁄16 in. and a base plate thickness varying in increments of 0.5 in., from 1.5 in. to 3.5 in. All five models indicated a yield stress in one or more of the bends. Because PEEQ is a cumulative metric, it increases over time, with the largest value occurring at the end of the simulation; these are presented in Figure 5-5c. Figure 5-4. PEEQ variation with time for three HMIP models. 0 10 20 30 40 50 60 1.5 2 2.5 3 3.5 V on M is es S tr es s (k si ) Base Plate Thickness (inch) 0 10 20 30 40 50 60 70 1.5 2 2.5 3 3.5 M ax . P ri nc ip al S tr es s ( ks i) Base Plate Thickness (inch) (a) (b) 0 500 1000 1500 2000 2500 1.5 2 2.5 3 3.5 PE E Q (µ ɛ) Base Plate Thickness (inch) (c) Figure 5-5. Stress and strain versus base plate thickness: (a) von Mises stress; (b) maximum principal stress; and (c) equivalent plastic strain (PEEQ).

Computational Simulations—High Mast Illumination Poles 61   The figure shows that the maximum PEEQ values increased nearly linearly as base plate thick- ness increased, but the stress measures (maximum principal stress and von Mises stress) were not influenced by changes in base plate thickness. Behavior of Round Poles The behavior of round poles was markedly different from that of multisided poles because of the absence of geometric discontinuities along pole circumference. In multisided poles, temperature-induced inelastic strains at the pole-to-base plate connection occurred at pole bends, with nearly zero strain occurring in regions between bends. A similar trend was observed for the stress field in multisided poles, where large stress concentrations were observed at pole bends near the pole-to-base plate interface. Stress and strain fields in round poles were found to be much more uniform along the pole circumference than those of multisided poles. The absence of bends eliminated stress concentra- tions at the pole-to-base plate interface. Localized effects were only observed in bands of high stresses near the surface of the zinc bath due to the high temperature gradient and at the base of the pole during cooling due to the temperature differential between the pole and the base plate. Thicker round poles experienced lower stresses than thinner round poles with the same base plate thickness. Effect of Number of Sides, Pole Thickness, and Base Plate Thickness on Stresses and Inelastic Strains A comparison of peak stress along the critical path for HMIP models with a 3-in. thick base plate and different numbers of sides is presented in Figure 5-6; round poles were considered as multisided poles with infinite number of sides. The figure shows that peak stress decreased with increasing number of sides and was lowest for round poles, except for the 8-sided poles that showed lower calculated stress than that for the 10- and 12-sided poles; stresses in multi- sided poles were relatively similar in magnitude but higher than those calculated for comparable round poles. It should be noted that the residual stresses induced during pole fabrication prior to galvanizing were not considered in the simulation; thus, the analysis did not capture the effects of cold-working and material hardening that would be most severe in poles with fewer bends (8-sided poles). 0 10 20 30 40 50 60 70 M ax im um P rin ci pa l S tr es s, k si Number of HMIP Sides 5/16" thick pole 1/2" thick pole 12 sides round10 sides 8 sides Figure 5-6. Effect of number of sides and pole thickness on peak stress in an HMIP with a 3-in. thick base plate.

62 Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process The effect of pole thickness and number of sides on the accumulated plastic strain was inves- tigated. The maximum PEEQ values over the galvanizing time-history were extracted from both the inner and outer surfaces. PEEQ values for poles with the same base plate thickness (3 in.) were examined to determine the influence of pole thickness and number of sides. For multi- sided poles, the maximum PEEQ values consistently occurred on the inside surface. Strains in multisided poles were largest for the 8-sided pole and decreased with increasing number of sides. Inelastic strains on the interior and exterior surfaces of ½-in. thick multisided poles were significantly lower than those for the 5�16-in. thick multisided poles. For example, the maximum PEEQ for an 8-sided, 5�16-in. thick HMIP was 2,700 micrometers (με) compared to 1,100 με for the 8-sided, ½-in. thick HMIP. The thickness effect was also apparent in round poles, where the maximum PEEQ was 1,500 με for the 5�16-in. thick pole and 500 με for the ½-in. thick pole. Thus, thicker poles exhibited smaller plastic strains, and multisided poles with a fewer number of sides exhibited greater strains during galvanizing. The effects of base plate thickness on stresses and PEEQ were examined for poles with similar thickness and number of sides but varying plate thickness. These were 5�16-in. and ½-in. thick, 12-sided poles with plate thicknesses varying from 1 in. to 3.5 in. in ½-in. increments. Strain responses are discussed here; stress responses are presented in Appendix D. Strains increased with increased base plate thickness for both pole thicknesses. For the 5�16-in. thick pole, PEEQ increased from 0 με for the thinnest base plate (indicating only elastic strains during galvanizing) to 3,000 με for the 3.5-in. thick base plate, in a nearly linear rela- tionship. For the ½-in. thick pole, PEEQ for both the 1.5-in. and 2-in. thick base plates was nearly zero at the connection, indicating that only elastic strains occurred during galvanizing. PEEQ increased almost linearly from 650 με for the 2.5-in. thick base plate to 1,450 με for the 3.5-in. thick base plate. Effect of Base Plate-to-Pole Thickness Ratio on Inelastic Strains As discussed previously, maximum inelastic strains during galvanizing increased with increasing base plate thickness and decreased with increasing pole thickness. The discussion in this section considers the plate-to-pole thickness ratio and its effect on inelastic strains and the propensity to form cracks. Figures 5-7 to 5-10 show the PEEQ values for the round, 12-sided, 10-sided, and 8-sided poles, and Figure 5-11 shows the values for all poles. These figures show that PEEQ values increased with the base plate-to-pole thickness ratio for all pole shapes, and they were lower for round poles than for any multisided pole with the same base plate-to-pole thickness ratio. Figure 5-11 shows that poles with base plate-to-pole thickness ratios of four or less expe- rienced negligible inelastic strains, while an assembly with a pole thickness of 5�16 in. and a base plate thickness of 3 in. experienced cumulative inelastic strains as high as 3,000 με. These observations suggest that reducing the base plate-to-pole thickness ratio would be an effective measure to reduce the occurrence of LMAC in HMIPs. Also, using round poles instead of multisided poles is likely to reduce the occurrence of LMAC.

Computational Simulations—High Mast Illumination Poles 63   0 500 1000 1500 2000 2500 3000 2 3 4 5 6 7 8 9 10 11 12 PE EQ , µ ε Base Plate-to-Pole Thickness Ratio 5/16 in. ROUND POLE OUTSIDE 1/2 in. ROUND POLE INSIDE 5/16 in. round pole 1/2 in. round pole Figure 5-7. Calculated inelastic strains for round poles. 0 500 1000 1500 2000 2500 3000 2 3 4 5 6 7 8 9 10 11 12 PE EQ , µ ε Base Plate-to-Pole Thickness Ratio 5/16 in. 12-SIDED POLE INSIDE 1/2 in. 12-SIDED POLE INSIDE 1/2 in. round pole 5/16 in. round pole Figure 5-8. Calculated inelastic strains for 12-sided poles. 0 500 1000 1500 2000 2500 3000 2 3 4 5 6 7 8 9 10 11 12 PE EQ , µ ε Base Plate-to-Pole Thickness Ratio 5/16 in. 10-SIDED POLE INSIDE 1/2 in. 10-SIDED POLE INSIDE 1/2 in. round pole 5/16 in. round pole Figure 5-9. Calculated inelastic strains for 10-sided poles.

64 Mitigation of Weldment Cracking in Steel Highway Structures Due to the Galvanizing Process 0 500 1000 1500 2000 2500 3000 12 12 12 12 12 12 12 12 12 12 12 8 10 12 8 10 12 R ou nd R ou nd 12 12 12 12 12 12 8 8 10 10 12 12 R ou nd R ou nd 12 12 PE EQ ,μ ε Number of Sides (above) and Base Plate-to-Pole Thickness Ratio (below) (2) (3) (4) (4.8) (5) (6)(3.2) (6.4) (7) (8) (9.6) (11.2) Inside Outside Figure 5-11. Calculated inelastic strains for all poles. 0 500 1000 1500 2000 2500 3000 2 3 4 5 6 7 8 9 10 11 12 PE EQ , µ ε Base Plate-to-Pole Thickness Ratio 5/16 in. 8-SIDED POLE INSIDE 1/2 in. 8-SIDED POLE INSIDE 1/2 in. round pole 5/16” round pole Figure 5-10. Calculated inelastic strains for 8-sided poles.