2D MODELING OF THE DEFLECTION OF A SIMPLY SUPPORTED BEAM UNDER POINT OR DISTRIBUTED LOADS YIN-YU CHEN MANE4240 – INTRODUCTION TO FINITE ELEMENT ANALYSIS PROFESSOR ERNESTO GUTIERREZ-MIRAVETE APRIL 28, 2014 Page |1 TABLE OF CONTENTS REFERENCES ........................................................................................................................ 2 ABSTRACT ........................................................................................................................... 3 INTRODUCTION .................................................................................................................... 3 FORMULATION AND SOLUTION ............................................................................................ 5 RESULTS AND DISCUSSION................................................................................................... 8 CONCLUSIONS .................................................................................................................... 13 Page |2 REFERENCES (a) M1 Abrams; Wikipedia: http://en.wikipedia.org/wiki/M1_Abrams; 2014 (b) Mechanics of Materials – 2nd Edition; Craig, Roy; 2000 (c) Theory of Elasticity: Two Dimensional Problems; Timoshenko, Stephen; June 1970 (d) ANSYS Tips: ANSYS Quick Start – Belcan; Dufour, Paul; http://www.ux.uis.no/~hirpa/KdB/FEM/ANSYS_quick_start.pdf; December 2003 Page |3 ABSTRACT This report details the analysis of a simply supported elastic beam subject to a point or uniformly distributed load and the beam’s maximum deflection under these loads. The analytical method through the use of hand calculations and MAPLE determined the maximum deflection of an 8m x 0.6m x 0.1m steel beam subjected to the force, due to half the weight of an M1 Abrams tank, as a point load at the center of the beam and as a uniformly distributed load across the beam. A modeling/numerical method through the use of COMSOL also obtained the maximum deflection of the beam for both cases. Results of this study show that in the numerical COMSOL method, using triangle cells for the mesh, presents a 9% error in comparison to the analytical method. For additional comparison, an ANSYS beam element model, using quadrilateral cells for the mesh, was created and analyzed, which presented only a 0.03% error in comparison to the analytical method. From this study, it may be concluded that the maximum deflection of a simply supported elastic beam subject to point or distributed loads may be achieved using either the modeling/numerical or analytical methods. However, the shape of the cells for the mesh is a major factor in the accuracy of the maximum beam deflection results. A brief slide presentation summarizing the study and results is attached in Appendix A. INTRODUCTION The system under this study is the maximum deflection of a simply supported elastic beam subject to point or distributed loads. The specific scenario considered in this study is that of an M1 Abrams tank equally supported by two simply supported steel beams that are the length and width of the tank’s tracks. In this scenario, there is a land mine flush with the ground under the center of each beam. This makes the maximum deflection of the beam under the load of the M1 Abrams tank of specific interest to determine the required height of the beam from the ground in order to avoid setting off the land mine, as presented in Figure 1. Figure 1 – Sketch of Side View of M1 Abrams Tank on Beam over Land Mine Page |4 As provided in Reference (a), the properties of the M1 Abrams Tank include a weight of 61326kg, hull/track length of 7.93m and a track width of 0.635m. In this study, it is assumed that the weight of the tank is evenly distributed, and that each beam supports half the weight of the tank. Therefore, the study requires analysis of only one beam with half the weight of the tank. Simplifying the study by rounding some of the specifications of the tank, the properties of the tank and beam used in this study are presented in Tables 1 and 2 respectively below: Table 1 - M1 Abrams Tank Properties M1 Abrams Tank Description Value Unit Force 300000 N Hull/Track Length 8 m Track Width 0.6 m Table 2 - Beam Properties Beam Description Value Unit Length 8 m Width 0.6 m Thickness 0.1 m Young's Modulus (E) 2.00E+11 Pa Poisson's Ratio (n) 0.3 Density (r ) 7800 kg/m3 Of additional interest is the difference in maximum deflection of the beam when assuming all of the force due to the weight the tank as a point load at the center of the beam versus a uniformly distributed load across the beam. These two cases are presented in Figures 2 and 3 respectively. Figure 2 – Simply Supported Beam with Point Load at the Center Figure 3 – Simply Supported Beam with Uniformly Distributed Load Page |5 FORMULATION AND SOLUTION Two methods were utilized for solution of the maximum deflection of the beam in this study. The analytical method through the use of hand calculations and MAPLE determined the maximum deflection of the beam subjected to the force, due to half of the weight of the M1 Abrams tank, as a point load at the center of the beam and as a uniformly distributed load across the beam. A modeling/numerical method through the use of COMSOL also obtained the maximum deflection of the beam for both cases. The other program used to calculate and tabulate the results for this study was Microsoft Excel 2010. ANALYTICAL METHOD: For the analytical method, the system was setup starting with the Moment-Curvature Equation, provided in Reference (b) and expanded upon in Reference (c), is represented in Equation 1 below: π2 π¦ πΈπΌ ππ₯ 2 = π Eq. 1 Where “E” is Young’s Modulus and “I” is the Moment of Inertia. To determine the Moment of Inertia of the rectangular cross section of this beam, Figure 4 and Equation 2 below are utilized, where in this study b=0.6m and h=0.1m. Figure 4 – Moment of Inertia of a Rectangular Cross Section of a Beam πΌ= πβ3 Eq. 2 12 For the case of a simply supported beam with a point load at the center (Figure 2), the bending moment at any position “x” up to the center of the beam is given by Equation 3 below: π= ππ₯ Eq. 3 2 Equation 4 below is the result of substituting Equation 3 into Equation 1. Integrating Equation 4 with respect to x once and twice result in Equations 5 and 6 respectively: π2 π¦ πΈπΌ ππ₯ 2 = ππ₯ 2 Eq. 4 Page |6 ππ₯ 2 ππ¦ πΈπΌ ππ₯ = πΈπΌπ¦ = ππ₯ 3 12 4 + π1 + π1 π₯ + π2 Eq. 5 Eq. 6 The constants of integration, c1 and c2, may be determined from the following boundary conditions: a) Due to no deflection at the end of the beam: At x=0, y=0 b) Tangent to the beam is horizontal at the center of the beam: At x=L/2, dy/dx=0 Therefore, substituting boundary condition “b” into Equation 5 results in Equation 7 below: πΏ π( )2 2 πΈπΌ(0) = π1 = − 4 + π1 ππΏ2 Eq. 7 Eq. 8 16 Thus, the constant c1 is solved in Equation 8. Substituting Equation 8 and boundary condition “a” into Equation 6 results in Equations 9 and 10 below to determine the constant c2: πΈπΌ(0) = π2 Eq. 9 π2 = 0 Eq. 10 Substituting the solved constants of integration, c1 and c2, into Equation 6 results in Equation 11 below: πΈπΌπ¦ = ππ₯ 3 12 − ππΏ2 16 π₯ Eq. 11 Substituting x=L/2 into Equation 11 provides the standard equation of the maximum deflection at the center of the beam due to a Point load, as presented in Equation 12 below and verified in Appendix E2 of Reference (b). ππΏ3 πΏπππ₯ = π¦ = − 48πΈπΌ Eq. 12 For the case of a simply supported beam with a uniformly distributed load (Figure 3), the bending moment at any position “x” is given by Equation 13 below: π= π€πΏπ₯ 2 − π€π₯ 2 2 Eq. 13 Equation 14 below is the result of substituting Equation 13 into Equation 1. Integrating Equation 14 with respect to x once and twice result in Equations 15 and 16 respectively. Page |7 π2 π¦ πΈπΌ ππ₯ 2 = ππ¦ πΈπΌ ππ₯ = πΈπΌπ¦ = π€πΏπ₯ 2 4 π€πΏπ₯ 3 12 π€πΏπ₯ 2 − − π€π₯ 2 − π€π₯ 3 6 π€π₯ 4 Eq. 14 2 + π 1 Eq. 15 + π 1 π₯ + π 2 24 Eq. 16 The constants of integration, s1 and s2, may again be determined from the following boundary conditions: a) Due to no deflection at the end of the beam: At x=0, y=0 b) Tangent to the beam is horizontal at the center of the beam: At x=L/2, dy/dx=0 Therefore, substituting boundary condition “b” into Equation 15 results in Equation 17 below: πΈπΌ(0) = π 1 = − πΏ 2 π€πΏ( )2 4 − πΏ 2 π€( )3 6 + π 1 Eq. 17 π€πΏ3 Eq. 18 24 Thus, the constant s1 is solved in Equation 18. Substituting Equation 18 and boundary condition “a” into Equation 16 results in Equations 19 and 20 below to determine the constant s2: πΈπΌ(0) = π 2 Eq. 19 π 2 = 0 Eq. 20 Substituting the solved constants of integration, s1 and s2, into Equation 16 results in Equation 21 below: πΈπΌπ¦ = π€πΏπ₯ 3 12 − π€π₯ 4 24 − π€πΏ3 π₯ 24 Eq. 21 Substituting x=L/2 into Equation 21 provides the standard equation of the maximum deflection at the center of the beam due to a uniformly distributed load, as presented in Equation 22 below and verified in Appendix E2 of Reference (b). 5π€πΏ4 πΏπππ₯ = π¦ = − 384πΈπΌ Eq. 22 These equations and required parameters were input into a MAPLE worksheet solution and also calculated by hand. Both the MAPLE worksheet solution and hand calculations have been provided in Appendix B. Page |8 MODELING/NUMERICAL METHOD USING COMSOL: For the modeling/numerical method, the system was setup in COMSOL using the 2D Structural Mechanics, Solid Mechanics and Stationary presets. Rectangular geometry was used to build the length (8m) and thickness (0.1m) of the beam, with the width (0.6m) specified as the “Thickness” in the “Solid Mechanics” tab. A material (steel) was then created and assigned to the rectangle, with the critical properties of Young’s Modulus, Poisson’s Ratio and Density set to the values in Table 2. Next, prescribed displacements of 0m were set for both x and y directions on the bottom left corner of the rectangle, and the y direction on the bottom right corner of the rectangle. The purpose of this is to set the boundary conditions of a simply supported beam, as previously presented in Figures 2 and 3. For the case of the point load at the center of the beam, a point was created at the center of the top boundary of the rectangle under the “Geometry” tab. The purpose of this is to allow for an application point for the point load of 300000N. This was prescribed using the point load feature under the “Solid Mechanics” tab, and was designated as negative to reflect the downward direction. For the case of the uniformly distributed load across the beam, the force per unit length must be determined. The 300000N is first divided by the square area of length x width of the beam, which is 4.8m2. This results in a force per unit area of 62500N/m2, which is a value that would be useful in a 3D model. To determine the force per unit length, the force per unit area is multiplied by the width of the beam (0.6m), which results in 37500N/m. This force per unit length was prescribed using the boundary load feature under the “Solid Mechanics” tab on the top boundary of the rectangle, and was designated as negative to reflect the downward direction. For both cases, the mesh is addressed under the “Mesh” tab, and the sequence type was left as Physics-controlled. The default element size provided by COMSOL is the Normal element size. However, a total of five different element sizes were analyzed for both cases as part of the mesh extension study. Along with the Normal element size, two finer meshes (Finer and Extremely Fine) and two coarser meshes (Coarser and Extremely Coarse) were also analyzed. After each mesh was built, the study was computed, solution determined and results output. In addition to the default von Mises stress plot, a 1D plot was created to capture the displacement of the beam across the length of the beam. RESULTS AND DISCUSSION The results of the modeling/numerical method using COMSOL to solve for the maximum deflection of the beam when assuming all of the force of the tank as a point load at the center of the beam and for a uniformly distributed load across the beam are summarized in Tables 3 and 4 respectively. Page |9 Table 3: Simply Supported Beam with Point Load at the Center Degrees Max Min of Triangular Edge Vertex Element Element Freedom Size Size Solved x ο€max (m) ο€ max (m) ο€ max (cm) Mesh Nodes Elements Elements Elements Extremely Fine 511 414 208 5 0.08 1.60E-04 2074 4.0 -0.29131 -29.131 Finer 349 120 118 5 0.296 0.001 718 4.0 -0.29125 -29.125 Normal 349 120 118 5 0.536 0.0024 718 4.0 -0.29125 -29.125 Coarser 349 120 118 5 1.04 0.048 718 4.0 -0.29124 -29.124 Extremely Coarse 313 103 105 5 2.64 0.4 624 4.0 -0.29123 -29.123 Table 4: Simply Supported Beam with Uniformly Distributed Load Degrees Max Min of Triangular Edge Vertex Element Element Freedom Size Size Solved x ο€max (m) Mesh Nodes Elements Elements Elements Extremely Fine 511 414 208 4 0.08 1.60E-04 2074 4.0 Finer 349 120 118 4 0.296 0.001 718 4.0 Normal 349 120 118 4 0.536 0.0024 718 4.0 Coarser 349 120 118 4 1.04 0.048 718 4.0 Extremely Coarse 313 103 105 4 2.64 0.4 624 4.0 ο€ max (m) ο€ max (cm) -0.18206 -18.206 -0.18202 -18.202 -0.18202 -18.202 -0.18203 -18.203 -0.18202 -18.202 The modeling/numerical method shows that for both cases, the Nodes and Degrees of Freedom solved at any specific mesh size level remained the same. The only difference is that the case with the point load at the center has an extra vertex element due to the point created for the application of the load. As expected, the maximum and minimum element size decrease while the number of nodes, elements and degrees of freedom solved increase as one moves from the extremely coarse mesh to the extremely fine mesh. It is noted that while the maximum and minimum element size decreased between the coarser, normal and finer meshes, the number of nodes (and likewise elements and degrees of freedom) remained the same. An example of the COMSOL Normal mesh of the beam is presented below in Figure 5. In addition, plots of the maximum deflection (ο€max) for the COMSOL Normal mesh for both cases have been provided in Figures 6 and 7. Results show that the maximum deflection of the beam was always located in the center of the beam at x=4m. The COMSOL mesh extension study shows that the difference between the extremely coarse mesh and extremely fine mesh is 0.027% and 0.022% for the point load at the center and uniformly distributed load respectively. With the COMSOL Normal mesh, the maximum deflection of the beam was -0.291m and 0.182m for the point load at the center and uniformly distributed load respectively. P a g e | 10 Figure 5 – COMSOL Normal Mesh of Simply Supported Beam Figure 6 – COMSOL Maximum Deflection (ο€max) of Simply Supported Beam with Point Load at the Center P a g e | 11 Figure 7 – COMSOL Maximum Deflection (ο€max) of Simply Supported Beam with Uniformly Distributed Load The analytical results of the maximum deflection of the beam in both cases between the hand calculations and MAPLE worksheet were identical (-0.32m and -0.20m for the point load at the center of the beam and uniformly distributed load across the beam respectively), as provided in Appendix B. A comparison of the results of the analytical method and modeling/numerical method to solve for this study are summarized in Table 5 below. Table 5: Comparison of COMSOL Modeling/Numerical and Analytical Method Results Analytical Mesh Nodes Total Elements ο€ max (m) Result ο€ max (m) % Error COMSOL Normal Mesh Point Load 349 243 -0.29125 -0.32 -8.985% COMSOL Normal Mesh Uniformly Distributed Load 349 242 -0.18202 -0.20 -8.988% While the percentage of error is near identical between the two cases, there is approximately a 9% error between the modeling/numerical method and analytical method. In both cases for the modeling/numerical results, the magnitude of maximum deflection of the beam is lower than the analytical result. For comparison, a 1 meter long P a g e | 12 beam with a 0.1m x 0.1m rectangular cross section was also modeled in COMSOL, and both cases produced a 5% error from the analytical method using an extremely fine mesh. This appears to stipulate that the COMSOL modeling/numerical method may be more accurate for a proportionally smaller length to cross section ratio of the beam. For additional comparison, an ANSYS beam element model was created and analyzed, with the assistance of the tutorial provided in Reference (d), for the case of a simply supported beam with a uniformly distributed load. The results of this ANSYS beam element model analysis are provided in Table 6. A plot of the beam’s deflection and mesh with a similar number of nodes compared to the COMSOL normal mesh is provided in Figure 8. Through the mesh extension validation in the ANSYS model, the results show percentage of error from the analytical result to be on an order of 50 to 200 times smaller than that of the COMSOL results. With the 5 element sizes meshed in the ANSYS model, the largest magnitude of error was 0.154%, with a majority of the magnitudes of error at 0.038%. It is observed that while both the ANSYS and COMSOL meshes are in a structured grid, the cells of the ANSYS mesh is quadrilateral while the cells of the COMSOL mesh are triangle. It was also observed that when an odd number of elements resulted due to the input element size, the result would fluctuate a minuscule amount from the converging maximum beam deflection value of 0.200077m. Table 6: Comparison of ANSYS Modeling/Numerical and Analytical Method Results Analytical Result Element Size Nodes Total Elements ο€ max (m) ο€ max (m) % Error 0.05 481 160 -0.20008 -0.20 0.038% 0.075 322 107 -0.20006 -0.20 0.028% 0.1 241 80 -0.20008 -0.20 0.038% 0.33 76 25 -0.19969 -0.20 -0.154% 1 25 8 -0.20008 -0.20 0.038% P a g e | 13 Figure 8 – ANSYS Maximum Deflection (ο€max) of Simply Supported Beam with Uniformly Distributed Load CONCLUSIONS From this study, it may be concluded that the maximum deflection of a simply supported elastic beam subject to point or distributed loads may be achieved using either the modeling/numerical or analytical methods. An ANSYS model using a quadrilateral cell mesh was within 0.028% of the analytical result. However, it appears that the shape of the cells for the mesh is a major factor in the accuracy of the maximum beam deflection results. This was apparent when the COMSOL models using a triangle cell mesh reported results with a 9% error from the analytical solutions. Therefore, when using the modeling/numerical approach with beams and deflection, a quadrilateral cell mesh may offer the most accurate solution. In this scenario of an M1 Abrams tank equally supported by two simply supported steel beams with a land mine flush with the ground under the center of each beam, uniformly distributing the load across the beam gives a more accurate representation of the maximum deflection of the beam. The beam requires a minimum height of 0.2m from the ground for the tank to avoid setting off the land mine. However, using a point load to determine the maximum deflection of the beam would be a more conservative approach P a g e | 14 to determine the minimum height of the beam from the ground to avoid setting off the land mine. It is noted that if the results from the COMSOL analysis of the uniformly distributed load across the beam were used without a factor of safety greater than 1.1 for the height of the beam from the ground, the maximum deflection due to the tank would set off the land mine. This highlights the necessity for verifying the reliability of the approximate solution by comparing the results to a theoretical/exact solution, a different modeling approach or a mesh extension validation.