STRAIN ANALYSIS OF HDPE BUTT FUSION JOINT IN SIDE BEND TEST Meng Xiangli (Shawn), McElroy Manufacturing, Inc., Tulsa, OK Abstract The standard practice ASTM F3183-16 (F3183) has been developed for conducting a three point side bend test to evaluate the ductility of a polyethylene butt fusion joint, but our understanding of the bending strains that are applied across the specimen during the test is far less than desired since there are no test values provided by the practice. In this paper we perform the strain analysis through grid method to investigate the strain level applied to the specimen. The present paper provides the results of strain levels at different bending angles for various selected combinations of loading nose radius and specimen thickness and then compares them with the estimated strain from pure bending theory. The experimental results show the strains increase with the bending angle and are higher than the estimations when the bending angle is larger than 60. In addition, we conduct the side bend tests in three different temperature conditions to address the effect of temperature on the strain levels applied to the specimen at a given bending angle. The test results show that the higher strains are applied to the specimens at a lower temperature. In general, the present work will be a good resource to assist in the employment of the standard practice F3183 and advance our understanding of the process of the side bend test. Introduction In general, the three point bending test is a classical experiment for evaluating the ductility of a material. Recently, ASTM International has established the standard practice ASTM F3183-16 (F3183) for conducting a guided three point side bend test to evaluate the ductility of butt fusion joints in polyethylene pipe having a wall thickness 25.4mm (1.00 in.) or greater. The plastic welding industry has developed the apparatus which applies a lateral (side) bending strain across a specimen. The applied strain needs to exceed the maximum elastic strain limit of the specimen so as to have the ductility of the fusion joint properly evaluated. Therefore, understanding the strain level applied to the specimen is essential to get a full assessment of the side bending process of the fusion joint. In the case of a high density polyethylene (HDPE) fusion joint, the average of the elastic strain limit from tension tests is around 9.5% depending on the parent material [1]. In fact, the strain level applied to the specimen in the side bend test exceeds the measurement range of the common strain gages, which leads us to investigate the methods of experimental strain measurement. The grid method of strain analysis is one of the most effective techniques known to experimental stress analysis. However, there are a couple of common difficulties associated with the employment of the grid method for measuring strain. First, the strain being measured is usually quite small and thus the displacement readings can t be made with sufficient precision to keep the accuracy of the strain determination within reasonable limits. Also, the definition of the grid lines on the images is usually poor, so that considerable errors are introduced in the displacement readings. However, these two difficulties can be avoided in the cases involving large deformations that impose a strain level of the order of 5 percent or greater [2]. In the polyethylene fusion joint side bend test, the strains applied to the specimen are large enough to avoid those difficulties; hence the grid method can be effectively utilized to measure the strain in this experimental study. In this study, to begin with, we perform the side bend test by using the loading nose radius and side bend test specimen thickness in accordance with standard F3183 to find the strain values. Then, we conduct the test by using different combinations of the load nose radius and specimen thickness to explore the relationship between the strain levels and R/t at different bending angles. Finally, we conduct the tests in different temperature conditions to investigate the effect of the temperature on the strains applied to the specimen. Generally, the present paper provides the strain measurement results and gives the interpretations of the data. The aim of this paper is to assist in the implementation of the standard practice F3183 and advance our understanding of the bending process in the test. Definition Statement of definition and theory Definitions below are in accordance with standard practice F3183. Loading Nose: A bar located equidistant between and opposite to rotatable supports and having a cylindrical forward surface. The loading nose is extended at a uniform rate of displacement in between the rotatable supports to bend the side bend test specimen. See Figure 1. SPE ANTEC Anaheim 2017 / 2218
The following definitions are specific to this paper: Bend Angle: The included space between the surfaces of the side bend specimen having contact with rotatable supports that is formed by the deflection of the side bend specimen when the loading nose extends the side bend test specimen (Figure 1). Note that Bend Angle is defined differently in this paper and standard practice F3183. Density of the Grid: The number of grid lines per unit length. Pitch of the Grid, p: The center to center distance between the grid lines on the unstrained specimen. Figure 1. Illustration of side bend test [3] Rotatable Supports: Two cylindrical bars spaced equidistant from and parallel to the loading nose that turn freely on their central longitudinal axis and support the side bend test specimen. Bend Test Coupon: A transverse section of butt fused polyethylene pipe extending from the pipe outside wall to the pipe inside wall and having approximately equal lengths of pipe on each side of a centrally located butt fusion joint. The side bend test specimen is produced from the bend test coupon. Side Bend Test Specimen: A transverse section of the wall of butt fusion joined pipe that is machined (planed) from a bend test coupon. Pure Bend Theory When a member is subjected to pure bending moment, the axial lines are transformed into circular arcs with a common center and the transverse sections remain plane. In the bending process there exist axial lines that do not extend or contract. The location in the cross section where the axial lines do not extend or contract is called neutral axis (Figure 3). The axial strain in a line element is defined as: Ԑ = - y/ρ (1) Where y is the distance from the line element to the neutral axis and ρ is the radius of the neutral axis. R/t: A dimensionless number representing the ratio of the loading nose radius, R, in mm (or inches) to the measured thickness, t, in mm (or inches) of the side bend test specimen. Combined Fusion Bead Zone: A transverse throughwall section of the side bend specimen that is bounded by imaginary planes that extend across the pipe wall from the inner and outer fusion bead surfaces of Pipe A and Pipe B fusion beads. Butt fusion joints typically produce beads that extend (roll) over the pipe ends both inside and outside of the joint (Figure 2). Figure 3. Deformation of member in pure bending [4] In the side bend test (Figure 1), if we assume that the plane of the neutral axis is at the mid-point of the specimen thickness, based on pure bend theory, the maximum absolute strain at the outer surface of the specimen can be written by: Ԑ m = (t/2) / (t/2+r) = 1 / [1+2(R/t)] (2) Where t is the thickness of the specimen and R is the radius of the loading nose. Figure 2. Combined fusion bead zone [3] The equation (2) shows the bending strain at the outer surface of the specimen is a function of R/t and irrelevant to bending angle. Table 1 presents the estimated strain for SPE ANTEC Anaheim 2017 / 2219
selected combinations of loading nose radius and side bend test specimen thickness. Table 1. Strain estimations based on pure bending theory Loading Nose Radius, R mm (inch) Specimen Thickness, t mm (inch) Estimated Maximum Strain, ε (%) R/t 6.4 (0.25) 6.4 (0.25) 33.0 1.0 6.4 (0.25) 12.7 (0.50) 50.0 0.5 12.7 (0.50) 6.4 (0.25) 20.0 2.0 12.7 (0.50) 12.7 (0.50) 33.0 1.0 19.1 (0.75) 6.4 (0.25) 14.2 3.0 19.1 (0.75) 12.7 (0.50) 25.2 1.5 25.4 (1.00) 6.4 (0.25) 11.3 4.0 25.4 (1.00) 12.7 (0.50) 20.0 2.0 However, the pure bending is more of an assumption than reality. In the case of three point bending, most of the load is bending, but there is stress localization at the specimen mid-point where the loading nose is in contact with the specimen. Also, there exists friction between the load nose and the top surface of the specimen. The friction force counteracts the compressive force produced by the bending and thus hinders the top layer of the specimen from contracting freely, which violates the assumption made in the pure bend theory that each layer of the member is free to expand or contract. More importantly, the assumption that the neutral axis is at the mid-point of the specimen may not be valid for this test. For many plastics including polyethylene, the compressive stiffness is higher than the tensile stiffness. In other words, a small net cross-sectional area of the specimen in compression is stiff enough to resist a large cross-sectional area under tension and therefore the neutral axis shifts up toward the loading nose. Consequently, if we assume the neutral axis is at the mid-point of the specimen, we will under-predict the tensile strain at the outer surface of the specimen. In summary, it can be reasoned that the specimen subjected to three point side bend test will have the peak strain at the outer surface of the specimen mid-point with reduced strain elsewhere and the peak strain is more than likely higher than the estimated strain from pure bend theory. Figure 4. Side bend test specimen preparation sequence [3] Grid method requires placement of a grid on the surface of the specimen which is attained by laser engraving. During the process of laser engraving, one important procedural step is to optimize the power and speed settings for the laser cutting to gain a good contrast and yet a minimum depth of the grid so that a good definition of the grid lines on the images and unaltered material properties are achieved. We start the power settings with a small value and apply a grid on a piece of HDPE material, and then increase the power value slightly to optimize the settings. Upon the completion of the power and speed settings, we visually locate the dot of light from a red laser pointer at the center of the combined fusion bead zone to engrave the grid properly at the center of the fusion joint. In this study, we use the Epilog model 13000 laser system and CorelDraw software to engrave grids. The grid we used is shown in Figure 5a. Material and Methods We follow the preparation procedures in the standard F3183 and prepare the side bend test specimens from the bend test coupon pairs that are cut from a sample PE 4710 pipe butt fusion joint (Figure 4). It is important to note that equal amounts are removed from each side of the bend test coupon to achieve the side bend test specimen with a uniform thickness of t ± 0.5mm (0.02 in.). Figure 5a. Drawing of the rectangular grid SPE ANTEC Anaheim 2017 / 2220
The grid consists of two series of orthogonal parallel lines over the surface of the fusion joint zone. The horizontal lines are marked with numbers from 1 to 21 and the vertical lines are marked with letters from A to K. The Figure 5b presents a specimen with an engraved grid centered at the combined fusion bead zone. Figure 6. Equipment setup for side bent test Testing Procedure Figure 5b. Grid engraved at the center of the combined fusion bead zone. Test Setup We follow the testing procedures in standard practice F3183 to conduct the side bend test. First, we position the side bend test specimen at the center of the rotatable supports and slowly lower the loading nose to just contact the surface of the specimen. Then, we visually align the centerline of the grid to the centerline indicator marks on both ends of the loading nose (Refer to Figure 1) and take a picture of the grid of the unstrained specimen. Next, we simultaneously start the timer and engage the actuator to extend the loading nose at a steady and constant movement rate of 76.2 ± 25.4 mm/min (3.0 ± 1.0 in./min) to bend the specimen to a desired angle. Figure 7a shows a specimen that is bent to 90. Finally, we stop the actuator and the timer and then quickly adjust the focus of digital camera and take a picture of the deformed grid in situ. The tests require a high resolution digital camera and proper lighting to carefully photograph the grid before and after loading the specimens to obtain the images which will show the deformation of the grid. The basic equipment setup is shown in Figure 6. In order to minimize the heat generated by the light source and keep the temperature of the specimen within a desired range constantly, we use a fluorescent bulb for the lighting and keep the light off when the picture is not being taken. Figure 7a. Specimen bent to 90 indicated by bend angle label SPE ANTEC Anaheim 2017 / 2221
Next, we restart the timer and extend the loading nose to bend the specimen to the next desired angle and then take another picture of the deformed grid. In this manner, we perform the tests and take images of the deformed grid when the specimen is bent to different desired angles, respectively. The figure 7b shows an image of the deformed grid when the specimen is at a bending angle of 90. known length 25.4mm (1.00 in.). Go to menu Analyze - Set Scale to set the known distance of the line in the Know Distance Field. Set the units of the measurement. Now the image is calibrated and ready to be used for taking measurements. Care shall be taken to insure the selected line is from the right (left) edge of the grid line to the right (left) edge of next grid line to minimize the displacement reading errors. Once the software is validated, we draw a line from the right (left) edge of the line E to the right (left) edge of the line G on the image of deformed specimen and use the function Analyze - Measure or press Ctrl + M to get the result. Then, we repeat the procedures to measure every distance between line E and line G from line 1 to line 21 and then find the average of them, which is the new length, L. Finally, we substitute the value of L and L into equation (3) to get the normal maximum strain of the specimen. Results and Discussion Figure 7b. Specimen at a bend angle of 90 Strain Measurement In this paper, we use the engineering normal strain to perform the strain analysis which is defined as below, ε = (L L)/L*100 = L/L*100 (3) ε = Normal Strain, % L = Original Length, mm (inch) L = New Length, mm (inch) L = Change in Length, mm (inch) As the peak strain produced in the three point side bend test is at the specimen mid-point, we perform the strain analysis at the two grid spacing areas (E-F-G) in the middle of the specimen to find the maximum strain during the test. The out-plane motion at the specimen mid-point is minimal and negligible, which eliminates the drawback of the grid method that the images can only show the in-plane displacement of the specimen. Therefore, the original length would be the distance between the line E and G which is equal to two pitch lengths of the grid, namely L= 2.54mm (0.10in.). ImageJ is scientific image measurement software, and it is used to measure new length L. The operating procedures are as follows: Open the software ImageJ and go to File to open the image being measured. Use the Line Selection tool to draw a selection line of a known length. In such case, the selection line is the centerline of the grid (Line F excluding the alignment marks) with a This section presents the side bend test results as well as discusses the results of the experimental work carried out to demonstrate the strain level applied to the specimens. Standard F3183 Bend Test According to the standard practice F3183, we use the load nose with a radius of 12.7mm (0.50in.) to bend the specimen that has a thickness of 6.4mm (0.25in.) to a bend angle of 90 in air at a temperature of 65-75 F. The results based on a sample size of 15 specimens are presented in Table 2. Table 2. Standard F3183 bend test result for a sample size of 15 Load Nose Radius, R mm (Inch) Specimen Thickness, t mm (Inch) Bend Angle Strain, Ԑ Standard Deviation, σ Estimated Maximum Strain, Ɛ 12.7 (0.5) 6.4 (0.25) 90 26.2% 3.2% 20.0% 2 In this particular case, the experimental results show the actual strain level has a large variation, but still the results can be considered close to the estimated strain from pure bend theory. Bend Tests Using Varied R and t We conduct the side bend tests by using different combinations of the load nose radius and the specimen thickness in air at a temperature of 65-75 F to investigate the relationship between the strains applied to the specimen and R/t for a given bending angle. The results are shown in Table 3. R/t SPE ANTEC Anaheim 2017 / 2222
Table 3: The strain levels at certain bending angles for a sample size of 15. Load Nose Radius, R mm (Inch) Specimen Thickness, t mm (Inch) Bend Angle Strain, Ԑ Standard Deviation, σ Estimated Maximum Strain, Ɛ 12.7 (0.5) 12.7 (0.5) 30 21.7% 2.6% 33.0% 1 12.7 (0.5) 12.7 (0.5) 60 41.2% 3.1% 33.0% 1 12.7 (0.5) 12.7 (0.5) 90 59.3% 1.2% 33.0% 1 12.7 (0.5) 12.7 (0.5) 120 66.2% 3.2% 33.0% 1 12.7 (0.5) 6.4 (0.25) 30 6.0% 1.1% 20.0% 2 12.7 (0.5) 6.4 (0.25) 60 17.0% 1.8% 20.0% 2 12.7 (0.5) 6.4 (0.25) 90 26.2% 3.2% 20.0% 2 12.7 (0.5) 6.4 (0.25) 120 31.1% 4.6% 20.0% 2 25.4 (1.0) 12.7 (0.5) 30 19.8% 3.0% 20.0% 2 25.4 (1.0) 12.7 (0.5) 60 32.4% 1.0% 20.0% 2 25.4 (1.0) 12.7 (0.5) 90 38.4% 0.9% 20.0% 2 25.4 (1.0) 12.7 (0.5) 120 N/A N/A 20.0% 2 25.4 (1.0) 6.4 (0.25) 30 5.4% 0.8% 11.3% 4 25.4 (1.0) 6.4 (0.25) 60 13.8% 1.2% 11.3% 4 25.4 (1.0) 6.4 (0.25) 90 15.8% 0.8% 11.3% 4 25.4 (1.0) 6.4 (0.25) 120 17.2% 0.8% 11.3% 4 As we expected, the experimental results contradict the estimated strain from pure bend theory. First, the study shows the strains applied to the specimen increase with the bending angle. Second, the strains don t appear to be a function of R/t since the strain is shown to be different for a given R/t of 2. Finally, the strain is higher than the estimations from pure bend theory when the bending angle is larger than 60. It is interesting to note that a larger loading nose radius imposes a smaller strain given a specimen thickness and a bending angle. On the other hand, a thicker specimen is applied a larger strain at a given load nose radius and bending angle. Bend Tests in Different Temperature Conditions It s known that the strength of the polyethylene is sensitive to the temperature and therefore we conduct side bend tests in air at a temperature of 20-30 F and 115-125 F to explore the effect of the temperature on the strain level applied to the specimen. We use the same specimen thickness of 12.7mm (0.50in.) and the same load nose radius of 12.7mm (0.50in.) to perform the tests, based on the approach of one-factor-at-a-time. The strain measurements in three different temperature zones are presented in Table 4. Table 4: The Strain level at certain temperature conditions for a sample size of 15 Temperature, F 20-30 65-75 115-125 Bend Angle Standard Strain, Ɛ Deviation, σ Strain, Ɛ Standard Deviation, σ Strain, Ɛ Standard Deviation, σ 30 28.4% 2.0% 21.7% 2.6% 19.0% 2.1% 60 55.6% 2.8% 41.2% 3.1% 33.8% 3.2% 90 75.7% 1.5% 59.3% 1.2% 47.2% 2.2% 120 84.3% 2.9% 66.2% 3.2% 55.6% 3.4% R/t For a given bending angle, the experimental test results show the strain levels applied to the specimen vary with the testing temperature conditions and a higher strain level is observed on the lower temperature specimen. The observation may be explained as follows: the stiffness of polyethylene increases with the decreasing temperature [5], i.e. the lower temperature, the higher stiffness. As a consequence, a larger force is required to apply the same amount of displacement on a lower temperature specimen, leading to a larger imposed strain. Limitations and Validation Although the grid method of strain analysis is an economical and effective technique, there are some unavoidable errors introduced during the process of the strain measurements. First, the displacement reading errors can t be avoided completely. The distances between the grid lines on the images of the unstrained samples are measured for every single test to investigate the level of the reading errors. The results show the strain percent error is within the order of 0 ± 0.5 percent which is very small and negligible. Second, the centerline of the grid was used to set scale based on its known length of 25.4mm (1.00 in.), but it is possible that the centerline is distorted due to Poisson s effect, to some extent, it may affect the accuracy of the results of the strain level. The Poisson s Ratio is defined as: ѵ = - Lateral Strain / Axial Strain ѵ = - Ԑ y / Ԑ x (4) The relation (4) shows the specimen will undergo a lateral compressive strain (- Ԑ y ) when the axial tensile strain (Ԑ x ) is applied to the specimen by the bending. In other words, the centerline will shrink in the process of bending due to Poisson s effect. When the software is scaled by the centerline based on its original length of 1 inch (25.4mm) while the actual length of the centerline becomes shorter because of the deformation. Consequently, the strain measurements will be higher than their actual values. In order to explore the possible errors introduced in the process of setting scale using the centerline of the grid. A 25.4mm (1.00 in.) precise ruler is attached to the specimen during the test and used to set the scale of the software (Figure 8). SPE ANTEC Anaheim 2017 / 2223
3. ASTM F3183-16, Standard Practice for Guided Side Bend Evaluation of Polyethylene Pipe Butt Fusion Joint, ASTM International, West Conshohocken, PA, 2016, www.astm.org. 4. B. McGinty, Beam Bending, Web. 2013. www.continuummechanics.org/beambending.html. 5. M.H. Es-Saheb, The Temperature Effects On High Density Polyethylene (HDPE) Pipes, Eng. Sci, 8, 47-60 (1996). Acknowledgments Figure 8. A specimen with an attached precise ruler that is used for scaling the software The strain measurements are determined when the measurement software is scaled by the centerline and the 1 physical scale, respectively. The results confirm that the strain measurements are lower when the software is scaled by the precise ruler and the difference of these two strain values increases with bend angle. However, the strain percent difference is still relatively small when we compare it to the overall strain in most cases. In general, we can perform the strain analysis without adding the precise ruler to the specimen in order to avoid the unnecessary complexity if a strain percent error of less than 5% is acceptable. Conclusions This paper provides the results of the strain analysis of the HDPE fusion joint in different side bend test configurations, which will be a useful source of reference for the selection of the loading nose radius, specimen thickness and bend angle in utilizing standard practice F3183. Additionally, we observe higher strain levels on the specimen at lower temperature for a given bending angle, however, the variation range of the strain is not very large when considering the tests are conducted at the most extreme temperature conditions, which could lead to the refinement of temperature requirements for the side bend test. Overall, this experimental study provides insights into the process of the fusion joint side bend test and thus advances our understanding and will hopefully serve to inspire further work in the field in the future. This paper was supported by McElroy Manufacturing, Inc.. I would like to thank my advisor Professor Dr. Keith Good from Oklahoma State University and my colleagues Jason Lawrence, Brandon Jackman and Matt Porter from McElroy Manufacturing, Inc. who provided expertise that greatly assisted this experimental research. Also I am grateful to Jim Johnston from McElroy Manufacturing, Inc. for providing the insightful suggestions to improve the manuscript. Side Bend Apparatus Appendix The side bend fixture securely holds all of the essential parts and the side bend test specimen in a stable configuration while the practice is conducted. The testing fixture shall provide for accurate visual alignment of the side bend test specimen relative to the centerline of the loading nose, and shall provide visual determination of side bend test specimen bend angle. The testing fixture shall be constructed such that full and continuous contact of the side bend test specimen with the loading nose is maintained as the test is performed. The essential parts are as follows: rotatable supports, movable member, loading nose, actuator and time device. The schematic and a typical side bend test fixture are shown in Figures 9 and 10. References 1. ASTM D638-14, Standard Test Method for Tensile Properties of Plastic, ASTM International, West Conshohocken, PA, 2014, www.astm.org. 2. J.W. Dally, W.F. Riley, Experimental Stress Analysis, McGraw-Hill, Inc., New York, pp. 154-155 (1991). Figure 9. Schematic of side bend test apparatus [3] SPE ANTEC Anaheim 2017 / 2224
The settings used for laser engraving in 50 watt Epilog model 13000 laser system is presented as reference. Figure 10. Guided Side Bend Tester Image Credit: McElroy Manufacturing, Inc. Laser settings SPE ANTEC Anaheim 2017 / 2225