Beruflich Dokumente
Kultur Dokumente
1.
INTRODUCTION TO PLATE AND SHELL ELEMENT The Plate/Shell finite element is based on the hybrid element formulation. The element can be 3-noded (triangular) or 4-noded (quadrilateral). If all the four nodes of a quadrilateral element do not lie on one plane, it is advisable to model them as triangular elements. The thickness of the element may be different from one node to another. "Surface structures" such as walls, slabs, plates and shells may be modeled using finite elements. For convenience in generation of a finer mesh of plate/shell elements within a large area, a MESH The user may also use the element for PLANE STRESS action only (i.e. membrane/in-plane stiffness only). The ELEMENT PLANE STRESS command should be used for this purpose.
Geometry Modeling Considerations The following geometry related modeling rules should be remembered while using the plate/shell element. The program automatically generates a fictitious fifth node "O" (center node - below) at the element center. While assigning nodes to an element in the input data, it is essential that the nodes be specified either clockwise or counter clockwise (Fig. 1.9,
below). For better efficiency, similar elements should be numbered sequentially. Element aspect ratio should not be excessive. They should be on the order of 1:1, and preferably less than 4:1. Individual elements should not be distorted. Angles between two adjacent element sides should not be much larger than 90 and never larger than 180. Following load specifications are available: Joint loads at element nodes in global directions. Concentrated loads at any user specified point within the element in global or local directions. Uniform pressure on element surface in global or local directions. Partial uniform pressure on user specified portion of element surface in 2
global or local directions. Linearly varying pressure on element surface in local directions. Temperature load due to uniform increase or decrease of temperature. Temperature load due to difference in temperature between top and bottom surfaces of the element.
The distinguishing features of this finite element are: Displacement compatibility between the plane stress component of one element and the plate bending component of an adjacent element which is at an angle to the first is achieved by the elements. This compatibility requirement is usually ignored in most flat shell/plate elements. The out of plane rotational stiffness from the plane stress portion of each element is usefully incorporated and not treated as a dummy as is usually done in most commonly available commercial software. Despite the incorporation of the rotational stiffness mentioned previously, the elements satisfy the patch test absolutely. These elements are available as triangles and quadrilaterals, with corner nodes only, with each node having six degrees of freedom. These elements are the simplest forms of flat shell/plate elements possible with corner nodes only and six degrees of freedom per node. Yet solutions to sample problems converge rapidly to accurate answers even with a large mesh size. These elements may be connected to plane/space frame members with full displacement compatibility. No additional restraints/releases are required. Out of plane shear strain energy is incorporated in the formulation of the plate bending component. As a result, the elements respond to Poisson boundary conditions which are considered to be more accurate than the customary Kirchoff boundary conditions. The plate bending portion can handle thick and thin plates, thus extending the usefulness of the plate elements into a multiplicity of problems. In addition, the thickness of the plate is taken into consideration in calculating the out of plane shear. 3
The plane stress triangle behaves almost on par with the well known linear stress triangle. The triangles of most similar flat shell elements incorporate the constant stress triangle which has very slow rates of convergence. Thus the triangular shell element is very useful in problems with double curvature where the quadrilateral element may not be suitable.
Plate Element Local Coordinate System The orientation of local coordinates is determined as follows: The vector pointing from I to J is defined to be parallel to the local x- axis. The cross-product of vectors IJ and IK defines a vector parallel to the local z-axis, i.e., z = IJ x IK. The cross-product of vectors z and x defines a vector parallel to the local y- axis, i.e., y = z x x. The origin of the axes is at the center (average) of the 4 joint locations (3 joint locations for a triangle).
Following are the items included in the ELEMENT STRESS output. SQX, SQY = Shear stresses (Force/ unit len./ thk.) SX, SY, SXY = Membrane stresses (Force/unit len./ thk) MX, MY,MXY = Moments per unit width (Force x Length/length) (For Mx, the unit width is a unit distance parallel to thelocal Y axis. For My, the unit width is a unit distance parallel to the local X axis. Mx and My cause bending,while Mxy causes the element to twist out-of-plane.) SMAX, SMIN = Principal stresses in the plane of the element (Force/unit area). The 3rd principal stress is 0.0 TMAX = Maximum 2D shear stress in the plane of the element (Force/unit area) VONT, VONB = 3D Von Mises stress at the top and bottom surfaces TRESCAT, TRESCAB = Tresca stress, where TRESCA = MAX[ |(SmaxSmin)| ,
2.
ASSIGNMENT 1
Problem Statement : 1. For a given slab simply supported(pinned) on two edges, we will find the convergence on bending moment using finite element analysis with help of computer software. 2. For the same panel understanding the behavior of panel with edge beams of different sizes. Data : Slab Panel : 3m x 6m Loading : Slab : 10kN/m2 150mm
For the given problem we will start the meshing of panel from 1m x 1m and use mesh refinement for convergence. Analysis will be done using staad-prov8i software.
Results:
Element size mxm 1.00 0.50 0.30 0.25 0.20 0.15 0.10
4.00 2.00 0.00 % Error -2.00 -4.00 -6.00 -8.00 -10.00 -12.00
% error in Thickness BM m 10.29 0.15 -2.93 0.15 1.32 0.15 2.17 0.15 3.02 0.15 3.02 0.15 3.02 0.15
Convergence
18
72
200
288
450
800
1800
Exact solution = w*l^2/8 = 11.25 kN.m Conclusion : From the above results we can conclude that the bending moments in the slab converges to exact solution by mesh refinement . For the element size 0.5m x 0.5m we get minimum +positive error and for element size 0.3m x 0.3m we get minimum negative error. Refinement of mesh further does help, and the negative error goes on increasing till the element size of 0.1m x 0.1m.
PART 2 Here we will model the supporting edge with beam elements instead of pinned supports of different sizes and get the results of bending moments for both slab and beam and check with exact classical solution.
Results:
Convergence
0.00 -10.00 -20.00 % Error -30.00 -40.00 -50.00 -60.00 -70.00 Size of Beam 300x300 300x450 300x600 300x750 300x900 300x1200
300x900 BM = 69.5kN.m
300x600 BM = 55.6kN.m
Conclusion : From the above results it can be concluded that, increasing the size of edge beams the result converges to exact solution. Hence relative stiffness of beams and slab is governing criteria for convergence of the bending moments for both, beam and slab. If the sizes of supporting beams are lesser than required, there is no slabbeam effect and act together as slab elements with just edged having more stiffness. Again if the sizes of edge beams are different in sizes, the results are higher for stiffer beam and carry more bending moments than the one which is comparatively weaker.
10
3.
ASSIGNMENT 2
TWO-WAY SLABS
Problem Statement : 1. 2. 3. Comparative study of two-way slabs bending moments using IS-456 guidelines (Table 26) and finite element analysis. Convergence of results for slab with line loads. Effect on panel under consideration due to change in size of adjacent panel.
For the given problem we will start the meshing of panel from 1m x 1m and use mesh refinement for convergence. Analysis will be done using staad-prov8i software. Results
Element size mxm 1.00 0.50 0.25 0.20 0.10 No. of Elements 30 120 480 750 3000 IS-456 BM MX +ve 11.20 11.30 11.40 11.50 11.50 14.75 BM MY +ve 10.20 10.30 10.40 10.40 10.45 10.75 BM MX -ve 0.00 0.00 0.00 0.00 0.00 0.00 BM MY -ve 12.10 17.40 20.90 21.70 23.20 14.25 % error BM X+ve -31.70 -30.53 -29.39 -28.26 -28.26 % error BM Y+ve -5.39 -4.37 -3.37 -3.37 -2.87 % error BM Yve -17.77 18.10 31.82 34.33 38.58
Convergence
60.00 40.00 % Error 20.00 0.00 -20.00 -40.00 -60.00 No. of Elements 0 500 1000 1500 2000 2500 3000 3500 MX+ve My +ve MY -ve
11
12
13
Conclusion : For the given case, the mesh refinement converges the positive X and Y moments for size 0.25m x 0.25m. From the results it can be concluded that the positive moments from IS456 table is higher than results from finite elements analysis after convergence. For negative Y moments the results crosses the exact solution for element size between 1.0m x 1.0m to 0.5m x 0.5m. So for negative moments further mesh refinement add to positive error and the results of finite element analysis are more than the one calculated from IS-456, which is exactly opposite of the case with positive moments. Also on changing the size of adjacent panel by half of the original, there is decrease in negative moments at support and increase in mid span moments.
14
4.
ASSIGNMENT 3
CANTILEVER BEAM
Problem Statement : Convergence of results for cantilever beam with concentrated load at end. Data : Beam span : 3m Loading : Size : 10kN at free end 300x600mm
For the given problem we will start meshing size from 0.2m x 0.2m and further refinement will be done to get the convergence.
Results :
15
Element size No. of mxm Elements 0.2mx0.2m 45 0.15mx0.15m 80 0.10mx0.1m 180 0..05mx0.5m 720
Convergence
0.00 0 -5.00 % Error -10.00 -15.00 -20.00 No. of Elements BM 100 200 300 400 500 600 700 800
Conclusion : For the given problem we can conclude that the result for bending moments converges to exact solution for the plate size of 0.05m x 0.05m. Also the results vary for different width of beam, here we tried to converge the results for beam size of 300x600mm The results for width other than 300mm, the bending moments are different for same number of elements. Hence unlike classical solution, where we dont take the effect of width in calculation of bending moments, in finite element analysis the width of beam influences the bending moments.
16
5.
ASSIGNMENT 4
WATER TANKS
Problem Statement : Comparison of .moments, shear and hoop forces for water tanks using FEM and IS-3370. a) Rectangular tanks b) Cylindrical tanks Data : a) Rectangular tanks Tank size : Loading : Wall: 3.75m x 5.0m hieght 50kN/m 150mm
For the given problem we will use the meshing of panel 0.25m x 0.25m as we have concluded this size from above examples
17
18
Table 1 x/H 0 1/4 1/2 3/4 1 Table 1 x/H 0 1/4 1/2 3/4 1 Table 2 x/H 0 1/4 1/2 3/4 1 Table 2 x/H 0 1/4 1/2 3/4 1 Table 3 x/H 0 1/4 1/2 3/4 1 Table 3 x/H 0 1/4 1/2 3/4 1 y = 0 (Centre) MX staad %error 5.00 5.37 6.82 10.00 7.87 -27.10 12.50 11.74 -6.48 8.75 9.72 9.98 -6.25 -3.18 -96.36 y = b/4 (Quarter ) MX staad %error 1.25 0.75 -66.67 2.50 0.97 -157.73 3.75 2.47 -51.82 3.75 3.19 -17.55 -3.75 -1.85 -103.14 y = b/2 (at wall) MX staad %error -8.75 -6.57 -33.18 -13.75 -12.54 -9.65 -21.25 -19.79 -7.39 -16.25 -17.81 8.75 0.00 -1.07 100.00 y = 0 (Centre) MX staad %error 6.25 5.35 -16.82 10.00 8.11 -23.30 13.75 13.13 -4.72 15.00 9.89 -51.67 0.00 2.54 100.00 y = 0 (Centre) MY staad %error 0.00 -0.12 100.00 1.25 0.91 -36.91 6.25 5.13 -21.83 8.75 7.76 -12.76 -30.00 -20.67 -45.14 y = b/4 (Quarter ) MX staad %error 1.25 0.75 -66.67 2.50 0.89 -179.64 5.00 2.39 -109.21 5.00 3.91 -27.91 0.00 1.09 100.00 y = b/4 (Quarter ) MY staad %error 0.00 -0.09 100.00 0.00 -0.05 100.00 2.50 2.00 -25.00 3.75 4.00 6.25 -18.75 -11.91 -57.43 y = b/2 (at wall) MX staad %error -10.00 -6.42 -55.69 -16.25 -12.86 -26.38 -27.50 -21.76 -26.36 -32.50 -24.45 -32.92 0.00 -3.52 100.00 y = b/2 (at wall) MY staad %error 0.00 -0.61 100.00 -2.50 -2.51 0.40 -3.75 -3.26 -15.03 -3.75 -2.77 -35.38 0.00 -1.07 100.00 y = 0 (Centre) MX staad %error 0.00 1.02 100.00 7.50 6.68 -12.28 13.75 12.90 -6.59 13.75 14.02 1.93 0.00 2.54 100.00 y = 0 (Centre) MY staad %error 0.00 -0.15 100.00 1.25 1.00 -25.00 6.25 4.34 -44.11 12.50 14.02 10.83 0.00 3.55 100.00 y = b/4 (Quarter ) MX staad %error 0.00 0.15 100.00 2.50 1.06 -135.85 3.75 2.48 -51.27 6.25 3.93 -59.03 0.00 1.09 100.00 y = b/4 (Quarter ) MY staad %error 0.00 -0.11 100.00 0.00 -0.27 100.00 2.50 1.42 -75.56 7.50 4.71 -59.40 0.00 2.12 100.00 y = b/2 (at wall) MX staad %error 0.00 -1.23 100.00 -15.00 -10.91 -37.45 -27.50 -21.53 -27.76 -31.25 -24.42 -27.95 0.00 -3.52 100.00 y = b/2 (at wall) MY staad %error 0.00 -0.59 100.00 -3.75 -2.21 -69.68 -5.00 -3.64 -37.36 -6.25 -3.89 -60.59 0.00 -0.72 100.00 y = 0 (Centre) MY staad %error 0.00 0.49 100.00 1.25 1.65 24.24 6.25 4.80 -30.21 11.25 10.00 -12.50 0.00 3.55 100.00 y = b/4 (Quarter ) MY staad %error 0.00 0.08 100.00 0.00 0.42 100.00 2.50 1.68 -48.81 6.25 4.75 -31.58 0.00 2.12 100.00 y = b/2 (at wall) MY staad %error 0.00 -0.36 100.00 -2.50 -1.84 -35.87 -5.00 -3.59 -39.31 -6.25 -3.90 -60.46 0.00 -0.72 100.00
19
Conclusion : For the water tank wall panels, with the given panel ratio and wall thickness the results are having 50% error This error may be due to the fact that IS-3370 coefficients does not take into effect the thickness of wall panel of tank apart from the width and height.
b) Cylindrical tanks Tank size : Loading : Wall: 5m (Height) x 6m (Radius) 50kN/m 150mm
Boundary conditions : Top of tank Free and bottom hinged IS 3370 : Table 12,13 and 14.
Results
Hoop Tension IS 3370 Moment My IS 3370
7.95 37.05 66.60 97.65 131.40 168.60 206.10 230.40 229.35 119.85 42.75
-0.03 29.38 59.10 89.75 122.55 159.47 197.78 228.12 225.07 153.30 0.00
-0.01 -0.02 -0.06 -0.13 -0.18 -0.04 0.56 1.89 2.78 3.66 1.57
0.00 0.00 0.00 -0.13 -0.13 -0.12 0.03 1.53 3.29 4.17 0.00
20
Hoop Tension
mt.
mt.
IS 3370 Staad
IS 3370 Staad
21
Conclusion: For the given case, the results of staad pro and IS 3370 coefficients are nearly same with negligible, 1% error in case of hoop tension and about 13% error in moments.
22