Académique Documents
Professionnel Documents
Culture Documents
ADINA R & D, Inc. 71 Elton Avenue Watertown, MA 02472 USA tel. (617) 926-5199 telefax (617) 926-0238 www.adina.com
Notices ADINA R & D, Inc. owns both this software program system and its documentation. Both the program system and the documentation are copyrighted with all rights reserved by ADINA R & D, Inc. The information contained in this document is subject to change without notice. ADINA R & D, Inc. makes no warranty whatsoever, expressed or implied that the Program and its documentation including any modifications and updates are free from errors and defects. In no event shall ADINA R & D, Inc. become liable to the User or any party for any loss, including but not limited to, loss of time, money or goodwill, which may arise from the use of the Program and its documentation including any modifications and updates. Trademarks ADINA is a registered trademark of K.J. Bathe / ADINA R & D, Inc. All other product names are trademarks or registered trademarks of their respective owners. Copyright Notice 8 ADINA R & D, Inc. 1995 - 2010 June 2010 printing Printed in the USA
Table of contents
Table of contents
Introduction..............................................................................................................................1 A.1 Thick cylinder under internal pressure.......................................................................7 A.2 Axisymmetric shell with one end fixed under internal pressure ................................9 A.3 Simply supported circular plate under pressure loading ..........................................12 A.4 Circular cylindrical shell under pressure load (shell element) .................................14 A.5 Circular cylindrical shell under line load (shell elements).......................................16 A.6 Circular cylindrical shell under line load (plate elements) ......................................18 A.7 Cantilever beam under tip loads, Hermitian beam element .....................................19 A.8 Cantilever beam under tip loads, iso-beam element ................................................21 A.9 Pinched circular ring, iso-beam elements ................................................................22 A.10 Pinched circular ring, Hermitian beam elements .....................................................24 A.11 Curved beam under out-of-plane load, iso-beam elements......................................25 A.12 Curved beam under out-of-plane load, Hermitian beam elements...........................27 A.13 Cantilever truss structure under concentrated load using skew systems..................28 A.14 Cantilever under distributed load using skew systems (plane stress) ......................30 A.15 Cantilever under distributed load using skew systems (plane strain) ......................32 A.16 Cantilever under distributed load using skew systems (3-D solids) ........................33 A.17 Cantilever under distributed load using skew systems (Hermitian beam) ...............34 A.18 Cantilever under distributed load using skew systems (iso-beam) ..........................35 A.19 Cantilever under distributed load using skew systems (plate) .................................36 A.20 Cantilever under distributed load using skew systems (shell) .................................37 A.21 Planar truss...............................................................................................................38 A.22 Tapered cantilever under tip load ............................................................................39 A.24 Folded cantilever plate under tip line load...............................................................43 A.25 Clamped square plate under pressure load...............................................................45 A.26 Clamped hyperbolic paraboloid under uniform pressure loading ............................47 A.27 Simply supported square plate under uniform pressure ...........................................49 A.28 Simply supported square plate under twisting .........................................................51 A.29 Rhombic plate cantilever under pressure load .........................................................53 A.30 Scordelis-Lo cylindrical roof benchmark problem, two cubic shell elements .........55 A.31 Scordelis-Lo cylindrical roof benchmark problem, plate elements .........................57 A.32 Pinched cylindrical shell, quadratic shell elements..................................................58 A.33 Pinched cylindrical shell, plate elements .................................................................60 A.34 Analysis of an assemblage of concentric fluid-filled cylinders, displacement-based fluid elements...........................................................................................................61 A.35 Cantilever beam, cubic shell element with top and bottom nodes ...........................63 A.36 Cantilever beam, cubic shell element with constraints ............................................65 A.37 Cantilever beam, cubic shell element with midsurface nodes..................................66 A.38 Cantilever beam, cubic iso-beam .............................................................................67 A.39 Orthotropic plate under pressure load (16-node shell).............................................68 A.40 Cyclically symmetric frame structure ......................................................................70 A.41 Beam on elastic foundation......................................................................................72
iii
Table of contents
A.42 A.44 A.45 A.46 A.47 A.48 A.49 A.50 A.51 A.52 A.54 A.55 A.57 A.58 A.59 A.62 A.63 A.64 A.66 A.67 A.68 A.69 A.71 A.72 A.73 A.74 A.75 A.76 A.80 A.104 A.110 A.111 A.113 A.114 A.115 A.116 A.117 A.118 A.120 A.121 A.122 A.123
Perforated tension strip ............................................................................................74 Fundamental frequency of a cantilever (plane stress)..............................................76 Fundamental frequency of a cantilever (plane strain)..............................................78 Fundamental frequency of a cantilever (3-D solids)................................................79 Fundamental frequency of a cantilever (Hermitian beam).......................................80 Fundamental frequency of a cantilever (iso-beam)..................................................81 Fundamental frequency of a cantilever (plate).........................................................82 Fundamental frequency of a cantilever (shell).........................................................83 Fundamental frequency of a simply supported plate, shell elements.......................84 Wave propagation in a rod.......................................................................................86 Cantilever subjected to ground motion ....................................................................89 Cylindrical tube under step loading .........................................................................91 Thick-walled curved beam under end load ..............................................................93 Beam subjected to a traveling load ..........................................................................95 Frequencies of a triangular cantilever......................................................................97 Analysis of orthotropic plate, patch of 4-node plane stress and shell elements .......99 Pure bending of orthotropic plate, patch of two 8-node distorted shell elements ..101 Analysis of concentric fluid-filled cylinders, potential-based fluid elements ........103 Fundamental frequency of a cantilever (4-node shell)...........................................107 Scordelis-Lo cylindrical roof benchmark problem, 4-node shell elements............108 Analysis of a flanged elbow...................................................................................109 Analysis of a rotating tubular shaft ........................................................................111 Frequency analysis of a tall water column with a free surface ..............................113 Transient analysis of a water column under compression, potential-based and displacement-based elements.................................................................................115 Transient analysis of a piston-container-spring system .........................................117 Analysis of a cylinder oscillating in a water cavity ...............................................119 Analysis of a breathing sphere in an infinite fluid .................................................121 Transient analysis of an axisymmetric plate in water ............................................123 Response spectrum analysis of a simply supported beam (beam elements) ..........125 Clamped spherical shell under internal pressure....................................................128 Frequency analysis of a cooling tower using cyclic symmetry..............................130 A pentagon subjected to various loadings using cyclic symmetry.........................133 Multilayered simply-supported square plate under uniform pressure (4-node shell) ...............................................................................................................................134 Gravity waves and acoustic waves of water in a tank............................................136 Electromagnetic interaction between two conductors (beam) ...............................138 Harmonic and random vibration analysis of a simply supported beam .................140 Axisymmetric crack in a bar ..................................................................................143 Analysis of a cracked weld attachment..................................................................145 Failure of a composite shell (4-node shell element) ..............................................147 Plate subjected to acoustic noise............................................................................149 Floor response spectrum analysis of an inclined building, response spectrum load ...............................................................................................................................152 Floor response spectrum analysis of a building with sloped roof, time history load
iv
Table of contents
A.124 B.1 B.2 B.5 B.6 B.7 B.8 B.9 B.10 B.15 B.16 B.17 B.18 B.19 B.20 B.21 B.22 B.23 B.28 B.32 B.33 B.34 B.35 B.36 B.38 B.39 B.40 B.41 B.42 B.43 B.44 B.45 B.46 B.48 B.49 B.50 B.51 B.52 B.53 B.54 B.55
...............................................................................................................................155 Semi-elliptical crack in a half space, 3-D analysis ................................................159 Elastic-plastic analysis of a truss structure ............................................................163 Large amplitude oscillations of a simple pendulum ..............................................165 Cantilever beam in creep .......................................................................................167 Pipe whip analysis, direct integration solution ......................................................169 Pipe whip analysis, mode superposition solution ..................................................171 Uniaxial cycling loading solution for von Mises plasticity, isotropic bilinear hardening ...............................................................................................................172 Uniaxial cycling loading solution for von Mises plasticity, bilinear kinematic hardening with material rupture.............................................................................173 Cable under gravity load........................................................................................174 Cyclic creep analysis of a thick walled cylinder....................................................176 Large displacement analysis of a simply-supported plate......................................178 Thermo-elastic analysis of a cantilever beam ........................................................180 Analysis of an underground opening .....................................................................182 Static analysis of a reinforced concrete beam ........................................................184 Dynamic analysis of a reinforced concrete beam ..................................................187 Large deflection analysis of a shallow arch ...........................................................188 Eight-story building subjected to impact load using substructures........................189 Elastic-plastic analysis of a sand specimen under compression ............................191 Elastic-plastic bending of a pipe, with axial loading and internal pressure ...........193 Excavation of a tunnel ...........................................................................................195 Cable/frame structure with cable failure ................................................................197 Elastic-plastic cantilever plate under end moment.................................................199 Cracking in a steel-lined concrete cylinder............................................................201 Buckling of an elastic-plastic spherical cap ...........................................................203 Thermal stress analysis of a cantilever beam, part 2..............................................207 Cantilever beam in creep (2-D plane stress, iso-beam), variable time step ...........208 Analysis of snap through of an arch structure........................................................209 Snap back of a bar structure...................................................................................211 Nonlinear response of a shallow spherical shell ....................................................212 Linearized buckling analysis of a stiffened plate...................................................214 Large displacement analysis of a stiffened plate with initial imperfections ..........216 Linearized buckling analysis of a circular arch......................................................218 Linearized buckling analysis of a column..............................................................220 Large displacement analysis of a cantilever subjected to deformation dependent pressure loading .....................................................................................................224 Elastic-plastic analysis of a thick-walled cylinder .................................................226 Thermal buckling of a simply supported rectangular plate....................................228 Large deflection analysis of a cable structure ........................................................230 Large displacement analysis of a thin cantilever subjected to end moment...........232 Elasto-plastic analysis of a rectangular shaft subjected to a twisting moment.......234 Elasto-plastic analysis with multilinear material model, isotropic hardening........236 Elasto-plastic analysis with multilinear material model, kinematic hardening......238
Table of contents
B.57 B.58 B.59 B.62 B.65 B.66 B.69 B.70 B.71 B.72 B.73 B.74 B.75 B.76 B.82 B.99 B.106 B.113 B.114 B.115 B.116 B.117 B.118 B.119 B.120 B.121 B.122 B.123 B.124 B.127 B.128 B.130 F.1 F.2 F.4 F.5 F.7 F.9 F.10 F.12 F.13 F.28 F.33
Analysis of Hertz plane strain contact problem .....................................................239 Rubber ring pressed between two plates, no friction .............................................241 Dynamic analysis of frictional sliding of a point mass ..........................................243 Analysis of Hertz dynamic contact problem..........................................................245 Large displacement analysis of a spherical shell (LDC method)...........................247 Torsion of a squat rubber cylinder .........................................................................249 Pipe elbow under internal pressure with interaction effects ..................................251 Elasto-plastic uniaxial large strain solution, isotropic hardening...........................253 Elasto-plastic uniaxial large strain solution, kinematic hardening.........................255 Contact analysis of two concentric cylinders.........................................................256 Analysis of a single DOF elasto-plastic spring ......................................................258 Creep solution using creep law no. 3 .....................................................................260 Cantilever beam in creep (2-D plane stress, iso-beam), time step blocks..............261 Large displacement analysis of a thin cantilever subjected to end moment, ATS method ...................................................................................................................262 Analysis of an orthotropic thermoelastic brick ......................................................263 Torsion of a squat rubber cylinder (Ogden material).............................................264 Belleville spring.....................................................................................................265 3-D plane stress elements with wrinkling ..............................................................266 Multilayered elasto-plastic cantilever (8-node shell) .............................................268 Pre-stressed cylinder modeled with initial strains..................................................270 A slice of a Sandia pressure vessel ........................................................................272 Thermo-elasto-plastic analysis with multilinear material model, isotropic hardening ...............................................................................................................................275 Dynamic impact of two dissimilar elastic bars (2-D model) .................................277 Viscoplastic analysis of a thick-walled cylinder....................................................279 A thin rod impacts a rigid surface..........................................................................281 Frequency analysis of a rotating blade...................................................................283 Analysis of a centre cracked plate with quadratic temperature distribution ..........285 Fracture analysis of an elastic-plastic CT-specimen ..............................................287 Crack propagation in a 2-D specimen....................................................................289 Thermal stress analysis of a cantilever (4-node shell), part 2 ................................291 Displacement controlled crack propagation, 2-D analysis.....................................292 Towing of a submerged sphere ..............................................................................294 General Couette flow (2-D and 3-D elements) ......................................................299 Flow in a pipe (axisymmetric elements) ................................................................301 Flow in an annulus (3-D elements) ........................................................................303 Laminar flow between rotating cylinders ..............................................................305 Non-Newtonian flow between two parallel plates (2-D and 3-D elements) ..........307 No flow test............................................................................................................309 Flow in a converging channel (Hamel problem)....................................................311 Square wall-driven cavity problem ........................................................................313 Natural convection in a square cavity ....................................................................315 Flow in an annulus (3-D elements) B transient analysis.........................................318 Sinusoidal oscillation of a flat plate supporting a fluid..........................................319
vi
Table of contents
F.35 F.37 F.39 F.41 F.48 F.52 F.62 F.64 F.66 F.67 F.69 F.76 F.77 F.78 F.80 F.81 F.82 F.83 F.84 F.85 F.86 F.87 F.88 T.1 T.2 T.3 T.4 T.5 T.6 T.7 T.8 T.9 T.10 T.11 T.12 T.51 T.52 T.53 T.54 T.55
Heat generated in laminar flow between two rotating cylinders (2-D elements)...321 Free-convection flow between two vertical plates (2-D elements)........................323 Viscous dissipation in pipe flow (axisymmetric elements)....................................325 Natural convection in a porous medium ................................................................326 Conjugate heat transfer in a pipe heated on finite length.......................................328 Flow past a cylinder, Re = 20, 40 ..........................................................................331 Fluid flow over a cantilever in a channel ...............................................................333 Round die-swell problem, Re = 23.7, Ca = 3.568 ..................................................335 Draining of a circular tank by an opening..............................................................337 Analysis of two merging streams with moving boundaries ...................................339 Natural convection in a square cavity (2-D 4-node elements) ...............................340 Fluid flow over a tower in a wind tunnel ...............................................................341 Laminar flow over a backward-facing step............................................................343 Turbulent flow over a backward-facing step .........................................................345 A compressible fluid passing a flat plate (2-D elements) ......................................347 A compressible fluid passing a flat plate (3-D elements) ......................................349 Shock wave of inviscid compressible fluid in a channel .......................................350 Analysis of a free-falling parachute .......................................................................352 Radiation exchange between two concentric cylinders..........................................354 Radiation exchange between parallel square plates, transient analysis..................356 Analysis of two merging streams with mass transfer (NMASS=2) .......................358 Pressure drop in a pipe (plus mass transfer)...........................................................360 Slightly compressible test case 1 (load type: displacement) ..................................362 Frequencies of a water-filled acoustic cavity.........................................................367 Transient heat conduction in a semi-infinite solid .................................................369 Steady-state heat conduction in a square column ..................................................371 Steady-state heat conduction in a long hollow cylinder.........................................373 Thermal stress analysis of a cantilever beam (part 1) ............................................374 Thermal eigenvalues and mode shapes for a conduction problem.........................375 Torsional rigidity of a square shaft ........................................................................377 Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using Euler backward method) .......................................................379 Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using trapezoidal rule) ....................................................................381 Semi-infinite region subjected to constant heat flux on the boundary (Euler forward method) ..................................................................................................................382 Analysis of a semi-infinite solid subjected to internal heat generation (trapezoidal rule)........................................................................................................................384 Analysis of a semi-infinite solid subjected to internal heat generation (Euler forward method) ..................................................................................................................386 Transient temperature analysis of a slab ................................................................387 Transient temperature analysis of a space shuttle ..................................................389 Solidification of a semi-infinite slab of liquid .......................................................391 Solidification of a corner region ............................................................................393 Change in electric potential due to crack growth in a specimen ............................395
vii
Table of contents
Analysis of one-dimensional steady state seepage flow ........................................397 Unconfined flow through a rectangular dam .........................................................399 Unconfined seepage through a gravity dam...........................................................401 Analysis of a semi-infinite region subjected to a constant surface temperature undergoing two phase changes ..............................................................................402 T.60 Parallel plates, 3-D radiosity solutions ..................................................................404 T.62 Radiation exchange of two concentric cylinders ...................................................406 T.63 Unconfined seepage through a gravity dam, 3-D elements and constraint equations ...............................................................................................................................408 T.65 Unconfined flow through a rectangular dam, user-supplied model .......................411 T.66 Condensation/convection analysis of an axisymmetric structure...........................412 T.67 Radiant interchange between circular disks, axisymmetric elements ....................414 T.68 Thermal stress analysis of a cantilever (4-node shell), part 1 ................................416 T.69 Solidification of a corner region, shell elements....................................................417 T.70 Solidification of a semi-infinite slab of liquid, ATS method .................................418 T.71 Radiant interchange of adjoining plates with opening angle 60...........................419 T.72 Radiation analysis of a cask with annular regions .................................................421 T.73 Analysis of the piezoelectric actuation of an aluminum structure .........................424 T.74 Consolidation analysis (plane strain) with ADINA-TMC .....................................426 T.75 Radiation analysis of a cask with annular regions and a shield .............................428 Index......................................................................................................................................431
viii
Introduction
Introduction
The objective of the ADINA verification manual is to present solutions that verify and demonstrate the usage of the ADINA system programs (ADINA, ADINA-T, ADINA-F, ADINA-FSI, ADINA-TMC and the AUI). Since one of the aims is to compare the analysis results with analytical and experimental solutions, relatively small problems are solved. Each verification problem is assigned an abbreviation of the form letter.number. The letter denotes the type of problem as follows: A = ADINA linear problem, B = ADINA nonlinear problem, F = ADINA-F problem, T = ADINA-T problem. Generally problems that test newer features of the ADINA system are assigned higher numbers. There are intentional gaps in the problem numbering. Each verification problem is presented in the same order: problem description, finite element model, problem solution, user hints, references and explanatory figures and tables. We intend to update the ADINA verification manual with additional problems as we continue our work on the ADINA system. If you have any suggestions regarding the problems presented in this manual, or have suggestions for additional problems, we would be glad to hear from you.
Introduction
For detailed instructions about running verification problems, see the Installation Notes provided with the ADINA system.
cp eij
Specific heat at constant pressure Strains, i=1,2,3 or x,y,z Tractions, i=1,2,3 or x,y,z Acceleration of gravity Convection coefficient Radiation coefficient Thermal conductivity Pressure Heat flux, i=1,2,3 or x,y,z Velocity, i=1,2,3 or u,v,w Cartesian coordinates, i=1,2,3 or x,y,z Thermal diffusivity= k / c p Coefficient of volume expansion Temperature Fluid viscosity
Kinematic viscosity= / Density Surface tension coefficient Stresses, i=1,2,3 or x,y,z Vorticity, frequency Biot number Brinkman number Capillary number Eckert number Grashof number Nusselt number Peclet number Prandtl number Rayleigh number Reynolds number
fi
g
ij
Bi Br
h hr k p qi
vi
Ca
Ec Gr Nu Pe Pr Ra Re
xi
Introduction
Additional documentation
At the time of printing of this manual, the following documents are available with the ADINA System: Installation Notes Describes the installation of the ADINA System on your computer. ADINA User Interface Command Reference Manual Volume I: ADINA Solids & Structures Model Definition, Report ARD 102, June 2010 Volume II: ADINA Heat Transfer Model Definition, Report ARD 10-3, June 2010 Volume III: ADINA CFD Model Definition, Report ARD 10-4, June 2010 Volume IV: Display Processing, Report ARD 10-5, June 2010 These documents describe the AUI command language. You use the AUI command language to write batch files for the AUI. ADINA User Interface Primer, Report ARD 10-6, June 2010 Tutorial for the ADINA User Interface, presenting a sequence of worked examples which progressively instruct you how to effectively use the AUI. Theory and Modeling Guide Volume I: ADINA Solids & Structures, Report ARD 10-7, June 2010 Volume II: ADINA Heat Transfer, Report ARD 10-8, June 2010 Volume III: ADINA CFD & FSI, Report ARD 10-9, June 2010 Provides a concise summary and guide for the theoretical basis of the analysis programs ADINA, ADINA-T, ADINA-F, ADINA-FSI and ADINA-TMC. The manuals also provide references to other publications which contain further information, but the detail contained in the manuals is usually sufficient for effective understanding and use of the programs. ADINA Verification Manual, Report ARD 10-10, June 2010 This manual. TRANSOR for PATRAN Users Guide, Report ARD 10-14, June 2010 Describes the interface between the ADINA System and MSC.Patran. The ADINA Preference, which allows you to perform pre-/post-processing and analysis within the Patran environment, is described. TRANSOR for I-DEAS Users Guide, Report ARD 10-15, June 2010 Describes the interface between the ADINA System and UGS I-deas. The fully integrated TRANSOR graphical interface is described, including the input of additional data not fully described in the I-deas database.
Introduction
ADINA System 8.7 Release Notes, June 2010 Provides a description of the new and modified features of the ADINA System 8.7. You will also find the following book useful: K.J. Bathe, Finite Element Procedures, Prentice Hall, Englewood Cliffs, NJ, 1996 Provides theoretical background to many of the solution techniques used in the ADINA System.
A.1
Objective To verify the stress and displacement variation of the axisymmetric element and the application of axisymmetric pressure loading. Physical problem Fig. A.1 shows the cylinder to be analyzed. The cylinder is assumed to be guided so that no axial displacement can occur and it is acted upon by the internal pressure. Finite element model The model consists of eight 8-node elements. Solution results The theoretical stress solution is given in [1] p.60:
y =
a 2 p b2 1 2 b a2 y
2
a2 p , x = 2 b a2
b2 1 + 2 y
, z = ( x + y )
The displacement in the radial direction can be calculated from the circumferential strain: y = y x = x ( y + z ) E
The ADINA numerical solution obtained is as follows: Radial displacements: y-coord. (m) Theory (m) 0.5 4.766710-5 1.0 3.033310-5 Stresses: y-coord. (m) 0.5 1.0
User hint $ By symmetry, no variation of stresses can occur in the axial (Z) direction. A linear displacement assumption in the axial direction would therefore be sufficient and the higher order displacement assumption in the axial direction cannot improve the results. Reference [1] Timoshenko, S. and Goodier, J.N., Theory of Elasticity, Second Edition, McGrawHill, 1951.
Figure A.1
A.2 Axisymmetric shell with one end fixed under internal pressure
A.2
Objective To verify the axisymmetric element under distributed loading when used for axisymmetric shell bending problems. Physical problem Fig. A.2 shows the cylinder to be analyzed. The radius to thickness ratio is 20 so thin shell theory is applicable. The cylinder is fixed at one end and it is loaded by internal pressure. The cylinder is long so that the effects of bending at one end do not affect the other end. Finite element model Forty-five 8-node axisymmetric elements with 33 integration are used to describe the bending behavior. Since most of the bending is concentrated towards the fixed end a finer mesh is used for that part. The loading due to the z-component of the internal pressure is simulated by an axial pressure load. Solution results The behavior of the cylindrical shell under edge loading and internal pressure is described for example in [1] p. 140. Using
n4 =
3 (1 2 ) h a
2 2
M0 =
Pa 2n 2 Pa n
C , 2 C , 2
C=
a a h 2
D=
12 (1 2 )
Eh3
T0 =
Pa =
Pi C2
where E = Young's modulus, = Poisson's ratio, a = mean radius, h = thickness, Pi h = internal pressure applied at radius a , we obtain the radial displacement and 2 axial bending moment as
w=
nz C (1 e ( cos nz + sin nz ) ) , ( 4n D ) 2 Pa
4
T M = 0 + M o sin nz + M 0 cos nz e nz n
Note that the factor C is introduced to calculate equivalent loads acting at the midsurface. The stresses in the axial and circumferential directions are
zz =
Pa a 12 M ( a y ) , + 2h h3
xx =
wE + zz a
We obtain the following results from ADINA: Radial displacements: Z-coord. (m) 0.025 0.050 0.075 0.100 0.200 0.300 0.400 0.900
Radial displacement (10-6m) Theory ADINA 2.91 3.74 9.51 10.73 17.41 18.72 25.08 26.31 43.21 43.72 45.53 45.81 44.29 44.61 43.75 44.11
The ADINA results are evaluated at the nodes on the mean surface of the cylinder. Stresses at inner Gauss points: Z-coord. (m) xx (MPa) Theory ADINA .0014 9.91 8.82 .0063 9.22 8.69 .0111 8.68 8.66 .0139 8.42 8.51 .0188 8.05 8.16 .0236 7.80 7.87 .1188 13.3 13.8 .8972 20.0 20.4
zz (MPa) Theory ADINA 33.0 32.2 30.5 29.8 28.1 27.5 26.8 26.1 24.6 24.1 22.6 22.1 5.05 5.14 9.76 9.76
10
A.2 Axisymmetric shell with one end fixed under internal pressure
User hints $ The 8-node element describes the bending behavior quite well. The 4-node element, which has a linear assumption on the displacement behavior in the radial and the axial direction, would not be satisfactory. $ The portion of most interest in edge bending of an axisymmetric cylindrical shell extends approximately the length 2 ah in the axial direction. This portion should, therefore, be modeled with a finer mesh. $ In thin shell theory the pressure loading is applied at the midsurface of the shell. For 2D elements, however, the pressure loading is applied at the element boundaries. A correction factor C is, therefore, introduced in the formulas obtained using thin shell theory in order to calculate equivalent loads acting at the midsurface. $ This example could also be solved using the axisymmetric shell element. Reference [1] Kraus, H. Thin Elastic Shells, John Wiley & Sons, 1967.
Figure A.2
11
A.3
Objective To verify the axisymmetric element under distributed loading when used for plate bending problems. Physical problem Fig. A.3 shows the circular plate to be analyzed. It is a thin plate since the diameter to thickness ratio is 20. The boundary of the plate is simply supported. Finite element model Ten 8-node axisymmetric elements are used. Solution results The theoretical solution is given in [1] p.56 as follows: Center deflection: ( 5 + ) pa 4 = 0.1739 102 m, wmax = 64 (1 + ) D Moments: p M y = (3 + ) ( a2 y 2 ) , 16 Stresses at X=0 and Y=0: 6M y N yy = 2 = 123.75 106 2 , h m The following results are obtained: Center deflection (m) Theory ADINA -3 -1.73910 -1.75210-3
D=
12 (1 2 )
Eh3
Mx =
p 2 ( a ( 3 + ) y 2 (1 + 3 ) ) 16
xx =
6M x N = 123.75 106 2 h2 m
12
xx ( N m 2 )
Theory 123.75106 User hint ADINA 124.02106
yy ( N m 2 )
Theory 123.75106 ADINA 124.02106
$ The 8- or 9-node element should be used in order to model the bending behavior. Reference [1] Timoshenko, S., Woinowsky-Krieger, S., Theory of Plates and Shells, Second Edition, McGraw-Hill, 1959.
Figure A.3
13
A.4
Objective To verify the membrane behavior of the curved thin shell element under distributed loading. Physical problem The pressure vessel shown in Fig. A.4 is loaded by internal pressure. A section of the shell subjected only to membrane action is to be analyzed. Finite element model We use one 8-node shell element extending 15 degrees and skew systems oriented in the radial and circumferential directions for ease of applying the boundary conditions. The axial loading is simulated by concentrated forces with magnitudes equivalent to a consistent line load. Solution results The radial deformation and the stresses are pa 2 pa pa rad = hoop = , w= 1 , Eh 2 2h h The ADINA numerical results are as follows:
w (103 m )
Theory 0.5041 User hints ADINA 0.5041
rad ( N m 2 )
Theory 10.0 ADINA 10.0
hoop ( N m 2 )
Theory 20.0 ADINA 20.1
$ The quadratic shell element is chosen in order to approximate the circular shell. This approximation leads to some very small variation in the displacement and stress results since the curvature of the element varies slightly over the element.
14
$ The rotational DOF systems at the boundary nodes are explicitly specified to refer to the applicable skew systems.
Figure A.4
15
A.5
Objective To verify the bending behavior of the curved thin shell element. Physical problem Fig. A.5 shows the circular cylindrical shell to be analyzed. It is acted upon by equal and opposite line loads. The solution is sought for the part of the shell which is not disturbed by end effects. Finite element model By symmetry, only a 90 portion of the cylinder needs to be modeled. All 4 boundary faces have symmetry boundary conditions. In addition, the displacements in the xdirection are zero. Solution results The theoretical solution is given in [1] p. 381 for a thin ring under equal and opposite forces. A correction to account for the plane strain condition in the axial direction gives 3 3 1 1 2 bPr 2 1 bPr z = (1 2 ) , y = (1 2 ) 2 2 4 EI 2 EI where z is the radial displacement at the application of the load and y is the radial displacement 90 from the load (along the Y-axis). The theoretical bending moment is bPr 2 M = cos 2 The following results are obtained:
z ( mm ) , = 90
Theory -1.625 ADINA -1.625
y ( mm ) , = 0
Theory 1.492 ADINA 1.492
16
( degrees )
0 90 User hint
$ The rotations at all nodes are referred to the local V1, V2 axes. At the boundary nodes, the V1 direction is parallel to the X-axis. In general, it is recommended that the reference system for rotations be set to either a global system or a skew system for those midsurface nodes where the rotational directions need to be known. Reference [1] Timoshenko, S., Strength of Materials, Part I, Elementary Theory and Problems, Third Edition, D. Van Nostrand, 1955.
Figure A.5
17
A.6
Objective To verify the bending behavior of the plate element when applied to a curved shell structure. Physical problem The cylindrical shell to be analyzed is the same as in Example A.5. Finite element model Forty-five plate elements are used to model a 90 portion of the cylinder. Solution results The theoretical solution is the same as for Example A.5. The following results are obtained with ADINA:
z ( mm ) , = 90
Theory -1.625 ADINA -1.627
y ( mm ) , = 0
Theory 1.492 ADINA 1.492
N m Bending moment, , = 90 m
Theory -255.0 ADINA -275.9 (on side x=0 of model) -251.3 (on symmetry plane x=0.25)
User hint $ The membrane action in the plate element corresponds to a constant strain assumption. Hence the membrane forces are constant over each element. Reference [1] Timoshenko, S., Strength of Materials, Part I, Elementary Theory and Problems, Third Edition, D. Van Nostrand, 1955.
18
A.7
Objective To verify the three-dimensional action of the straight Hermitian beam element under end loads. Physical problem The cantilever beam shown in Fig. A.7 is loaded by a transverse end load and moment. Finite element model The Hermitian beam describes the behavior of the cantilever using only 1 element. The torsional constant J is set to J = 0.229 0.02 0.013 = 4.58 10-9 m4 as given in [1] p. 53. The shear factor used for the calculation of the effective shear 5 area is . 6 Solution results The theoretical solution for the end displacement and for the end rotations b due to bending and t due to torsion is [2]
The following solution is obtained with ADINA: Displacements (mm): Theory x z 5.0003 2.5008
19
Rotations (radians 10-3): Theory x 3.7500 References [1] [2] y 2.8384 z -7.5000 x 3.7500
Oden, J.T., Ripperger, E.A., Mechanics of Elastic Structures, Second Edition, McGraw-Hill, 1981. Roark, R.J., Formulas for Stress and Strain, Fourth Edition, McGraw-Hill, 1965.
Figure A.7
20
A.8
Objective To verify the three-dimensional action of the straight isoparametric beam under end loads. Physical problem Same as in Example A.7. Finite element model One 4-node element is used to model the cantilever. Solution results The theoretical solution is the same as given in Example A.7. The following results are obtained: Displacements (mm): Theory x z 5.0003 2.5008 Rotations (radians 10-3): Theory x 3.7500 User hint $ The 4-node isoparametric beam has cubic assumptions for all displacements along the beam and can in this case describe the bending behavior exactly. y 2.8384 z -7.5000 x 3.7500
21
A.9
Objective To verify the curved isoparametric beam element under bending. Physical problem A circular ring of square cross section is subjected to equal and opposite concentrated forces, see Fig. A.9. The thickness of the ring and the mean radius are the same as for the shell in Examples A.5 and A.6. Finite element model By symmetry, only a quarter of the ring needs to be modeled. Four cubic iso-beam elements are used. Solution results The theoretical solution is given in [1] p. 380 as follows:
1 2 Pr 3 1 2 1 Pr 3 Pr z = , y = , M = 2 4 EI 2 2 EI
where z is the radial displacement at the application of the load, y is the radial displacement 90 from the load (along the y-axis) and M is the bending moment at the application of the load. The following results are obtained: z (mm) Theory ADINA -1.785 -1.786 M (N-m) Theory -5.09 ADINA -5.09 y (mm) Theory ADINA 1.639 1.640
22
User hints $ The ADINA results are obtained using three point integration along the centroidal axis of each element. The four point integration results give about 4% less displacements. $ This structure is in a state of plane stress while the structures in Examples A.5 and A.6 were in a state of plane strain. As for these previous examples the corresponding 8-node 2-D solid element could also effectively be employed. Reference [1] Timoshenko, S., Strength of Materials, Part I, Elementary Theory and Problems, Third Edition, D. Van Nostrand, 1955.
Figure A.9
23
A.10
Objective To verify the Hermitian beam element when applied to a curved structure. Physical problem Same as Example A.9. Finite element model Twelve Hermitian beam elements are used to model a quarter of the ring. Solution results The theoretical solution is the same as for Example A.9. The following ADINA results are obtained: Displacements: z (mm) Theory ADINA -1.785 -1.780
Theory 1.639
Bending moment at the point of load application: M (N-m) Theory ADINA -5.09 -5.09
24
A.11
Objective To verify the three-dimensional action of curved iso-beam elements including torsion. Physical problem Fig. A.11 shows the curved beam to be analyzed. The beam is relatively slender and fixed at one end. The other end is loaded by a concentrated force in the out-of-plane direction. Finite element model Six 3-node iso-beam elements are used to approximate the 90-degree circular bend. Solution results The theoretical tip displacement in the direction of the force, torsional moment Mtor and bending moment Mb are ([1], p. 412):
x =
PR 3 EI s 3 + 2, EI S 4 C 4
M tor = PR (1 sin ) ,
M b = PR cos
where C = 0.141 a 4 G . A comparison with results from ADINA yields: x (m) Theory 0.05 (degrees) 0 30 45 60 90 ADINA 0.05 Mtor (N-m) Theory 100.0 50.0 29.3 13.4 0.0 ADINA 99.9 50.0 29.3 13.4 0.0 Mb (N-m) Theory 100.0 86.6 70.7 50.0 0.0 ADINA 100.1 86.6 70.7 50.0 0.0
25
User hints $ Torsional effects are very important in this example. The iso-beam element contains special displacement assumptions for torsional behavior and the element performs very well. $ The constant curvature of the circular bend is only approximated by the 3-node (parabolic) isoparametric beam elements. Since six elements are used for the bend the solution still agrees very well with the exact solution. Reference [1] Timoshenko, S., Strength of Materials, Part I, Elementary Theory and Problems, Third Edition, D. Van Nostrand, 1955.
Figure A.11
26
A.12
Objective To verify the three-dimensional action of the Hermitian beam element, including torsion, when applied to a curved structure. Physical problem Same as Example A.11. Finite element model Twelve Hermitian beam elements are used. The torsional constant J is set to
27
A.13
Objective To verify the truss element when used with skew coordinate systems. Physical problem A cantilever truss structure under concentrated end load is considered as shown in Fig. A.13. Finite element model The truss structure is modeled with 2-node truss elements. The structure is inclined in the global coordinate system so it is effective to employ skew coordinate systems. Solution results The displacement of the point of load application in the direction of the cantilever is: 9 iPl 2 = = 1.125 103 m i =1 hAE and the theoretical internal forces are Element 1 11 21 Internal element force (N) -10000.00 9000.00 1414.21
ADINA gives the exact solution for this problem. User hint $ The skew coordinate system used for each node is oriented in the principal directions of the cantilever. The input concentrated force and all nodal results are then referred to this coordinate system.
28
A.13 Cantilever truss structure under concentrated load using skew systems
Figure A.13
29
A.14
Objective To verify the plane stress element subjected to distributed loading and when employing skew coordinate systems. Physical problem A cantilever beam of rectangular cross-section is subjected to a triangular distributed load, see Fig. A.14. Finite element model Ten 8-node finite elements are employed. The upper face of the elements is acted upon by a linearly varying pressure load. The model is inclined in the global coordinate system. Solution results Using beam theory the theoretical solution for the end displacement is:
11 PL3 5 PL 60 EI 6 AS G
The following ADINA results are obtained: End displacement (mm) Theory ADINA -1.113 -1.110 Axial stress at fixed end (N/m2) Theory ADINA 6 40.0010 39.77106 The axial stress is evaluated in the element closest to the fixed end at the top surface stress point.
30
A.14 Cantilever under distributed load using skew systems (plane stress)
User hint $ The exact displacement solution according to beam theory for a linearly varying distributed load contains the 5th power of the coordinate. Several 8-node (parabolic) elements are therefore used to obtain a good approximation to the analytical solution.
Figure A.14
31
A.15
Objective To verify the 2-D plane strain element subjected to distributed loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Same as in Example A.14 except that plane strain elements are used. To obtain the same theoretical solution as for the plane stress case of Example A.14 the following material data is used:
E* =
1 + 2
(1 + )
* =
= 0.230769 1+
where E and are the Young's modulus and Poisson's ratio used in Example A.14. Solution results The same results as in Example A.14 are obtained. User hint $ Note that the thickness of the plane strain element is always equal to unity. The pressure is, therefore, here acting on a 20 times wider cantilever than in Example A.14, giving a factor of 20 times larger reaction forces than for Example A.14.
32
A.16 Cantilever under distributed load using skew systems (3-D solids)
A.16
Objective To verify the 3-D solid element subjected to pressure loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Ten 20-node 3-D solid elements are employed. Solution results The ADINA numerical solution gives an end displacement of -1.111 mm and an axial stress at the fixed end of 39.8106 N/m2. User hint $ The midside nodes in the transverse direction of the beam are included in order to model the anticlastic curvature. For a description of the phenomenon of anticlastic bending, see for example [1] p. 175. Reference [1] Oden, J.T. and Ripperger, E.A., Mechanics of Elastic Structures, Second Edition, McGraw-Hill, 1981.
33
A.17
Objective To verify the Hermitian beam element subjected to distributed loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Ten Hermitian beam elements are employed. Solution results The ADINA numerical solution gives an end displacement of -1.110 mm and an axial stress at the fixed end of 40106 N/m2. User hint $ The Hermitian beam does not predict the exact displacement in this example since its displacement assumption along the beam is a cubic polynomial while the exact displacement varies with the 5th power of the axial coordinate. However, 10 Hermitian beam elements give a very good solution to this example problem.
34
A.18 Cantilever under distributed load using skew systems (iso- beam)
A.18
Objective To verify the isoparametric beam element subjected to distributed loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Ten 3-node iso-beam elements are employed. Solution results The ADINA numerical solution gives an end displacement of -1.109 mm and an axial stress at the fixed end of 40106 N/m2. User hint $ Note that the predicted displacements agree well with the analytical solution even though 3-node iso-beam elements were used (with a parabolic displacement assumption in each beam element).
35
A.19
Objective To verify the plate element subjected to distributed loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Twenty plate elements are employed. Solution results The ADINA numerical solution gives an end displacement of -1.101 mm. The axial stresses at the elements attached to the fixed end are 38.67106 N/m2 and 40.52106 N/m2.
36
A.20
Objective To verify the shell element when subjected to distributed loading and when employing skew coordinate systems. Physical problem Same as in Example A.14. Finite element model Five 9-node shell elements are employed. All nodes have five degrees of freedom, three translations (referenced to a skew coordinate system) and two rotations (referenced to a local shell mid-surface coordinate system). Solution results The ADINA numerical solution gives an end displacement of -1.110 mm and an axial stress at the fixed end of 40.08106 N/m2. User hints $ All shell director vectors are in this example generated automatically by ADINA. Since in this example V1 is parallel to the skew a-axis and V2 is parallel to the skew b-axis, we can conveniently express the boundary conditions for rotations without explicitly specifying global (or skew) directions for the shell rotations. In general it is, however, recommended to specify global (skew) directions for all shell nodes with boundary conditions for rotations. $ In this example, the local coordinate system for stresses is oriented in the principal directions of the beam, since all shell director vectors are perpendicular to the midsurface. The stresses are therefore conveniently requested to be given in the local system.
37
A.21
Planar truss
Objective To verify the truss element when applied to a two-dimensional structure. Physical problem The planar truss structure in Fig. A.21 is considered. Finite element model Nine 2-node truss elements are sufficient to model this problem. Solution results The theoretical displacement solution is given in [1] p. 257. ADINA gives the exact theoretical solution. Nodal displacements (10-3 feet): Node y-dir. z-dir. Node 1 0 0 4 2 1.48 -4.38 5 3 2.96 -2.50 6 Reference [1] Tuma, J.J. and Munshi, R.K., Theory and Problems of Advanced Structural Analysis, Schaum's Outline Series, McGraw-Hill.
Figure A.21
38
A.22
Objective To verify the shell element with variable thickness when used as a transition element connected to a 3-D solid element. Physical problem A cantilever of variable thickness and subjected to a moment load at the end, see Fig. A.22, is considered. Finite element model The model consists of two 8-node shell elements and two 3-D solid elements. The shell elements use a cubic displacement assumption in the direction of the cantilever. Solution results Beam theory gives the following displacement solution along the cantilever:
( x) =
The ADINA numerical solution is as follows: End displacement (inch) Theory ADINA -0.02468 -0.02462 User hints $ Poisson's ratio is set to zero in order to simulate the plane stress condition associated with beam theory. $ The material law for the shell element is such that the stress in the thickness direction is zero. The stiffness in the thickness direction of the transition shell element is, therefore, negligible at the transition nodes. However, the adjoining 3-D element provides stiffness in the thickness direction. $ The shell rotations (and moments) at the cantilever tip nodes are referred to the
39
global coordinate system. The applied moments at the end nodes are then ensured to be acting about the X-direction. $ Note that shell element midsurface nodes have no rotational stiffness about the direction vector. For the cantilever tip nodes this means that the rotational stiffness about the Z-direction is zero, so this degree of freedom must be deleted.
Figure A.22
40
41
Figure A.23
42
A.24
Objective To verify the use of transition elements to model an intersection between shells. Physical problem A folded roof structure is subjected to a line load at its tip and fixed at the other end, see Fig. A.24. Finite element model Four cubic shell elements are used to model the folded roof structure. The intersection between the two plane parts of the roof is modeled with transition shell elements. Solution results The results compare well with the solution obtained in [1] p. 136 for a 2-D fine mesh of the same structure. User hints $ The shell element stresses may be requested to refer either to the global X-Y-Z coordinate system or to the local r s t system at each integration point. Note that this local system for stresses need not be the same as the local r - s- t system, which is used in the numerical integration of the stiffness matrix. For the non-transition elements in this example the two local systems r - s- t and r s t are the same since the shell normal vector is orthogonal to the midsurface of the element and since the elements are flat and rectangular. For the transition elements, however, the two systems are different. Reference [1] Bathe, K.J. and Ho, L.W., "Some Results in the Analysis of Thin Shell Structures", in Nonlinear Finite Element Analysis in Structural Mechanics, Wunderlich, W., et al. (eds.), Springer Verlag, 1981.
43
Figure A.24
44
A.25
Objective To verify the behavior of the cubic shell element subjected to a pressure load. Physical problem A thin plate with clamped edges under a pressure load as shown in Fig. A.25 is considered. The load is acting in the negative Z-direction. Finite element model Because of symmetry conditions only a quarter of the plate is considered. Nine 16node shell elements are used for the finite element model. The model allows no deformation in the X- and Y-directions and no rotation about the Z-axis. Solution results The theoretical solution for this problem is given in [1] p. 197 and the expression for the central deflection is given as
w = 0.00126
pa 4 , D
D=
Eh3 12(1 2 )
where p = pressure load and a = side length. The expression for the top surface stress at points C and D are:
6 pa 2 , h2
xxD = 0.0513
6 pa 2 h2
The following numerical solutions have been obtained using ADINA: w (m) Theroy -6.8810-4 ADINA -6.9110-4 xxC (N/m2) Theory ADINA 6 -4.4910 -4.49106 xxD (N/m2) Theory ADINA 6 9.9710 9.87106
45
User hint $ Note that the 16-node element provide a reliable and accurate solution. If the internal nodes are not used (i.e., the 12-node shell element is employed), the mesh is much too stiff, i.e., the elements lock. Reference [1] Timoshenko, S.P. and Woinowsky-Krieger, S., Theory of Plates and Shells, Second Edition, McGraw-Hill, 1959.
Y A C
region discretized
B D
E = 2.1 . 1011 N/m 2 n = 0.3 a a = 1.8 m X h = 0.01 m (thickness) p = 1.0 . 103 N/m 2 (pressure)
a
Figure A.25
46
A.26
Objective To verify the plate element when used to model a curved structure subjected to pressure loading. Physical problem The hyperbolic paraboloid shown in Fig. A.26 is considered. The equation of the c c hyperbolic paraboloid is Z = 2 X Y or Z = 2 (Y 2 X 2 ) . a a Finite element model Using symmetry conditions only one quarter of the shell (ABC, Fig. A.26) is modeled with 36 plate elements. All elements are subjected to a uniform surface pressure. Solution results Reasonable agreement with the solutions in [1] and [2] is obtained, but considerable stress resultant jumps occur. If detailed stress results are required then a finer mesh should be used. References [1] [2] Connor, J.J. and Brebbia, C., "Stiffness Matrix for Shallow Rectangular Shell Elements" Proc. ASCE, Vol. 93, No. EM5, Oct. 1967, pp. 43-65. Batoz, J.L., "Analyse Non Lineaire des Coques Minces Elastiques de Formes Arbitraires Par Elements Triangulaires Courbes", Doctoral Thesis, Dept. of Civil Eng., Universite Laval, Quebec, 1977.
47
All edges are clamped a = 50 cm c = 10 cm h = 0.8 cm E = 2.85 . 104 kg/cm 2 n = 0.4 q = 0.1 kg/cm 2
a c
h
a A D
X
B
a a
Figure A.26
48
A.27
Objective To verify the bending behavior of the plate element subjected to uniform pressure loading. Physical problem A square plate with simply supported edges under uniform pressure loading as shown in Fig. A.27 is considered. Finite element model Because of the symmetry conditions only one quarter of the plate is considered. The finite element model consists of 64 plate elements. The model allows no X and Y translations or Z-rotation. Solution results The theoretical stress solution for this problem is given in [1], article no. 30. The expression for the deflection at the center of the plate is given as
w = 0.00406
pa 4 , D
D=
12 (1 2 )
Eh3
where p = pressure load and a = side length of the plate. The bending moments per unit length at the center of the plate are
M x = M y = 0.0479 pa 2
The following results are obtained using ADINA: w (m) Theory 1.68910-4 ADINA 1.67510-4 Mx (N-m/m) Theory ADINA 30.7 30.5 My (N-m/m) Theory ADINA 30.7 31.0
49
Reference [1] Timoshenko, S.P. and Woinowsky-Krieger, S., Theory of Plates and Shells, Second Edition, McGraw-Hill, 1959.
region discretized
Figure A.27
50
A.28
Objective To verify the twisting behavior of the plate element. Physical problem A square plate which is supported at three of its corners is shown in Fig. A.28. A concentrated load acting in the negative Z-direction is applied at the free corner. Finite element model An irregular mesh with eight plate elements is used. Solution results The theoretical solution for this problem is a constant twist [1]. The central deflection is 0.0624 in., the deflection under the applied load is 0.2496 in and the bending and twisting moments in the whole model are Mx = My = 0, Mxy = 2.5 lb-in/in. ADINA gives the exact theoretical solution. User hint $ This example may be regarded as a patch test problem since the theoretical solution is a constant twist. The ADINA solution obtained for the irregular mesh is exactly equal to the theoretical value and the plate element has, therefore, the ability to represent stresses due to a constant twist. Reference [1] Batoz, J.L., Bathe, K.J. and Ho, L.W., "A Study of Three-Node Triangular Plate Bending Elements", Int. J. Num. Meth. in Eng., Vol. 15, pp. 1771-1812, 1980.
51
Y Pz 4 E = 1.0 . 10 psi n = 0.3 a = 8.0 in. h = 1.0 in. (thickness) Pz = 5.0 lb. X
Figure A.28
52
A.29
Objective To verify the behavior of the plate element subjected to a uniform load. Physical problem A thin rhombic cantilevered plate under pressure load is shown in Fig. A.29. The uniform pressure load acts in the negative Z-direction. The distribution of the transverse deflection of the plate is to be determined. Finite element model A relatively coarse mesh of 32 plate elements is used. Solution results The transverse displacements at six locations, see Fig. A.29, predicted in the analysis are given in the table below. Experimental values [1] are also shown in the table for comparison. A further discussion of this example problem can be found in [2]. Transverse displacement (in): Location A ADINA 0.304 Experimental[1] 0.297 References [1] Clough, R.W. and Toucher, J.L., "Finite Element Stiffness Matrices for Analysis of Plate Bending", Proc. Conf. on Matrix Meth. in Struct. Mech., WPAFB, Ohio, 1965, pp. 515-545. Batoz, J.L., Bathe, K.J. and Ho, L.W., "A Study of Three-Node Triangular Plate Bending Elements", Int. J. Num. Meth. in Eng., Vol. 15, pp. 1771-1812, 1980.
B 0.199 0.204
C 0.113 0.121
D 0.121 0.129
E 0.056 0.056
F 0.023 0.022
[2]
53
Y F
A D
E
q a
X E = 10.5 . 10 n = 0.3 6
a q h p
psi
Figure A.29
54
A.30 Scordelis-Lo cylindrical roof benchmark problem, two cubic shell elements
A.30
Objective To verify the cubic isoparametric shell element when subjected to gravity loading. Physical problem A cylindrical shell roof subjected to gravity loading is considered, see Fig. A.30. The shell roof is supported on diaphragms at the ends and it is free along the longitudinal sides. Finite element model Due to symmetry, only one quarter of the cylindrical shell roof needs to be considered. The shaded region in Fig. A.30 is modeled using two cubic isoparametric shell elements. The symmetry boundary conditions are specified as follows: on side BC, u x = y = z = 0, along side CD, u y = x = z = 0. The nodes corresponding to the diaphragm side are fixed for translation in the Y- and Z-directions. Solution results This example problem has been used extensively as a benchmark problem for shell elements. The analytical shallow shell solution generally quoted for the vertical deflection at the center of the free edge (point B in Fig A.30) is -3.703 inches [1] although some authors use -3.696 inches. A deep shell exact analytical solution quoted is -3.53 inches. The following results are obtained with ADINA: uz (in) rr (psi) point B point C point B -3.605 1344.36 -61.80 User hints $ Note that only two elements can be used to model this example problem resulting in good agreement with theoretical results for displacements. $ Since the shell curvature is smooth, all midsurface nodes not on symmetry boundaries have rotations referenced to the V1, V2 directions. Midsurface nodes on symmetry boundaries have rotations referenced to the global coordinate system to
55
allow the deletion of selected rotational degrees of freedom. $ Note that ADINA automatically handles the specification of midsurface vectors that are used by the shell elements. Reference [1] Scordelis, A.C. and Lo, K.S., "Computer Analysis of Cylindrical Shells", J. Amer. Concr. Inst., Vol. 61, pp. 539-560, 1964.
Z C B region discretized
j R diaphragm X
E = 3.0 . 10 6 psi R = 300 in. n = 0.0 L = 600 in. h = 3.0 in. (thickness) f = 40 Shell weight = 90.0 lb/ft 2
Figure A.30
56
A.31
Objective To verify the plate element when applied to a curved shell structure subjected to gravity loading. Physical problem Same as Example A.30. Finite element model As for the previous example only one quarter of the structure needs to be modeled. A 12124 mesh of plate elements is used. Solution results The vertical deflection at point B of Fig. A.30 is predicted to be -3.521 inches. User hint $ Note that even though a large number of plate elements are used, the displacement solution for point B is still not as good as the solution obtained in Example A.30, where only two isoparametric shell elements are used. One reason is that the plate element is flat and the cylindrical shell roof is therefore approximated by straight segments. Another reason is that the membrane action of the plate element is the same as for a constant strain triangle. The membrane forces are constant over each element, which limits the capability of the plate element to describe structures in which the membrane forces vary significantly.
57
A.32
Objective To verify the membrane and bending behavior of the quadratic shell element when applied to a curved structure. Physical problem The thin cylindrical shell structure shown in Fig. A.32 is analyzed for its static response. The cylinder is freely supported at its ends and is loaded by two centrally located and diametrically opposed concentrated forces. Finite element model Using the double symmetry of the structure and the load, only one eighth of the cylinder is analyzed. One hundred 8-node shell elements are employed. The mesh is refined around the point of load application. Solution results A comparison between analytical and calculated solutions for some displacement and stress distributions using different mesh topologies is presented in [1]. The theoretical solution is given in [2]. The following results are obtained using ADINA: uz (in) Theory -1.642 References [1] [2] Bathe, K.J. and Ho, L.W., "A Simple and Effective Element for Analysis of General Shell Structures", J. Comp. and Struct., Vol. 13, pp. 673-681, 1981. Lindberg, G.M., Olson, M.D. and Cowper, E.R., "New Developments in the Finite Element Analysis of Shells", National Research Council of Canada, Quarterly Bulletin of the Division of Mechanical Engineering and The National Aeronautical Establishment, Vol. 4, pp. 1-38, 1969. point C ADINA -1.661
58
P C
Figure A.32
59
A.33
Objective To verify the membrane and bending behavior of the plate element when applied to a curved structure. Physical problem Same as Example A.32. Finite element model As for Example A.32 only one eighth of the cylinder needs to be considered. The finite element model consists of 576 plate elements. Solution results ADINA gives a displacement at point C of -1.619 in. User hint $ As discussed in Examples A.30 and A.31, a larger number of elements is required in general when modeling a curved shell structure with plate elements.
60
A.34
Objective To verify the behavior of the two-dimensional axisymmetric displacement-based fluid element in a fluid-structure static analysis. Physical problem Five concentric fluid-filled cylinders are analyzed for an axial load applied to the stiff end cap as shown in Fig. A.34. Finite element model Because of symmetry only one half of the assemblage needs to be considered in the axisymmetric model. The finite element model consists of 100 4-node axisymmetric structural elements and 100 4-node axisymmetric fluid elements. In order to simulate the slip boundary conditions, fluid and structural elements are modeled with separate nodes while the nodal displacements in the Y-direction for adjacent structural and fluid element nodes are constrained to be equal. The stiff cap is simulated using rigid links. Solution results An experimental solution for this problem is presented in [1] with pressure values and standard deviations. Cylinder 1 2 3 4 5 Pressure (KPa) Experimental ADINA 51.1 55.7 .4 42.8 45.2 .6 33.3 34.0 .5 23.5 25.3 .5 12.7 14.1 .4
61
References [1] Munro, M. and Piekarski, K., "Stress Induced Radial Pressure Gradients in LiquidFilled Multiple Concentric Cylinders", J. of Appl. Mech., Vol. 44, No. 2, pp. 218-221, 1977.
Z
Stiff cap
25.250
Line of symmetry
3.213 3.373 4.610 4.775 6.050 6.203 6.655 6.820 7.925 8.090
Cylinders (6351-T6 aluminum): E = 6.895 . 1010 N/m2 n = 0.33 Fluid (Light hydraulic oil): k = 1.64 . 109 N/m2 3 r = 910 kg/m
Y (cm)
Figure A.34
62
A.35 Cantilever beam, cubic shell element with top and bottom nodes
A.35
Cantilever beam, cubic shell element with top and bottom nodes
Objective To verify the isoparametric shell element with particular reference to the behavior when top and bottom nodes are employed. Physical problem A cantilever beam with rectangular cross-section subjected to a concentrated force at the free end is shown in Fig. A.35. Finite element model The beam is modeled using one 16-node cubic shell element with top and bottom nodes. The element is cubic in the Y direction and linear in the X direction. The boundary conditions are such that no deformation in the X-direction is allowed. Solution results Using beam theory the theoretical expression for the tip deflection is
PL3 PL , 3EI AS G
5 AS = bh 6
The theoretical tip deflection is -8.048 mm and the theoretical bending stress at the integration point closest to the fixed end is 128.9106 N/m2. The ADINA model gives exactly these values (the deflection for ADINA is the average deflection for the top and bottom nodes at the free end). User hints $ Examples A.35 - A.38 are included to clarify the behavior of the shell element and compare with the corresponding results of the iso-beam element. A more effective way to solve the physical problem would be to employ the 2-node Hermitian beam element.
63
h Z b X
Figure A.35
64
A.36
Objective To verify the isoparametric shell element with particular reference to the behavior when top and bottom nodes are employed. Physical problem Same as in Example A.35. Finite element model The beam is modeled using one 16-node shell element with top and bottom nodes. The element is cubic in the Y direction and linear in the X direction. No deformations in the X-direction are permitted and the top and bottom nodes at the free end of the beam are constrained to have the same vertical displacements. Solution results The theoretical tip deflection is -8.048 mm, the theoretical bending stress at the integration point closest to the fixed end is 128.9106 N/m2 and the theoretical shear stress along the whole model at Z=2.88710-2 m is -4.0106 N/m2. The ADINA model gives exactly these values. User hint $ By constraining the vertical displacements to be the same for the top and bottom end nodes, the shell behaves as if midsurface nodes were used for this example. The total number of degrees of freedom for a pair of top and bottom nodes is 6 and the constraint reduces the number to 5. These 5 degrees of freedom may be thought of as 3 translations and 2 rotations as for a midsurface node.
65
A.37
Objective To verify the isoparametric shell element when midsurface nodes are employed and to compare with the results of Examples A.35 and A.36. Physical problem Same as in Example A.35. Finite element model The cantilever beam is modeled with one 8-node shell element with midsurface nodes. The element is cubic in the Y direction and linear in the X direction. Solution results The same results as in Example A.36 are obtained. User hint $ The one shell element model gives exactly the analytical results because the shell element contains the cubic transverse displacement variation and the Timoshenko beam theory shear stress assumption.
66
A.38
Objective To verify the behavior of the cubic iso-beam element and to provide a comparison for Examples A.35 - A.37. Physical problem Same as in Example A.35. Finite element model The finite element model consists of one 4-node isoparametric beam element. Solution results The theoretical tip deflection is -8.048 mm and the ADINA model gives -8.040 mm. The theoretical bending stress at the integration point closest to the fixed end is 183.4106 N/m2; the ADINA model gives exactly this value. User hints $ Note that no shear correction factor can be specified for the iso-beam element. The theoretical solution using a shear correction factor equal to unity is -8.040 mm, which agrees exactly with the ADINA result obtained. $ Apart from the effects of the shear correction factor, the results of the iso-beam agree with the results of the shell element models used in Examples A.36 and A.37, since the formulation of these two isoparametric elements is the same for bending action in the single plane considered.
67
A.39
Objective To verify the bending and twisting behavior of the cubic shell element when the material is anisotropic and the shell is subjected to a uniform pressure load. Physical problem A simply supported square plate under uniform loading, as shown in Fig. A.39, is considered. Finite element model Because of symmetry only one quarter of the plate needs to be considered. The mesh consists of 16 cubic shell elements. An orthotropic linear elastic material model is used with the principal material axes a and b coinciding with the global coordinate axes X and Y, respectively. Solution results The theoretical solution for this problem is discussed in [1], chap. 11. The expression for the deflection w in the Z-direction takes the form of a double trigonometric series,
w = amn sin
m n
m x a
sin
n y b
m = 1, 3, 5, ...,
n = 1, 3, 5, ... ,
and the expression for the coefficients amn is given on page 371 in [1]. From this expression the bending and twisting moments in the plate are easily derived using eq. 212 in [1]. The following results are obtained with ADINA: Vertical deflection, w(m): Location C E F
68
Stress (N/m2): Location Stress component C (el. 16) x y E (el. 15) x y E (el. 14) x y
* The theoretical results include terms up to m = n = 29. User hint $ Note that a finer mesh is necessary when modeling this simply supported plate when an orthotropic material is used than when an isotropic material is used. This is evident from the large gradient of the y-stress occurring along the line A-C, which in turn is due to the very small E-modulus in the Y-direction. Reference [1] Timoshenko, S.P. and Woinowsky-Krieger, S., Theory of Plates and Shells, Second Edition, McGraw-Hill, 1959.
Y,b A F 2c C E
region discretized B c = 0.03 m h = 0.015 m (thickness) 3 p = 1.0 . 10 N/m2 (pressure) D X,a 10 Eaa = 1.38 . 10 9 E bb = 1.15 . 10 8 Eab = 5.31 . 10 9 Gab = 1.17 . 10 N/m2 N/m2 N/m2 N/m 2
2c
Figure A.39
69
A.40
Objective To verify the use of constraint equations when applied to enforce conditions of cyclic symmetry. Physical problem The frame structure considered and the applied loads are shown in Fig. A.40. Finite element model Since the frame structure and the applied loading display cyclic symmetry, only onequarter of the structure needs to be considered. Four 2-node beam elements are used to model one-quarter of the frame as shown, together with the following displacement constraints: v17 = w15, w17 = -v15, 17 = 15
As a demonstration the whole structural model with sixteen 2-node beam elements is also analyzed. Solution results The solutions obtained from the four-element model and the sixteen-element model are identical. User hint $ The cyclic symmetry option of ADINA can be used to automatically generate the proper constraint equations.
70
P P b
w17
71
A.41
Objective To verify the capability to model a beam supported by an elastic foundation. Physical problem A beam resting on an elastic foundation is subjected to a concentrated transverse load at mid-span, as shown in Fig. A.41. Finite element model Thirteen Hermitian beam elements are used to model one-half of the beam. Twonode truss elements are used to represent the lumped stiffness of the elastic foundation. Solution results The analytical solution for the beam on elastic foundation is found in [1]. The following results are obtained with ADINA: Deflection at mid-span (in) Theory -1.634810-3 User hints $ The differential equation for the beam on an elastic foundation is of the same form as the differential equation for an axisymmetric cylinder, see Example A.2. The finite element representation is, however, accomplished here by beam elements and truss elements. $ It is important to have a fine enough mesh close to the applied load in this example since the significant displacements occur close to the load and decrease rapidly with increasing distance from the applied loads. $ The stiffness of the truss elements is calculated based on lumping the foundation stiffness, so that half of the foundation stiffness under each beam element is represented in the truss elements at each of the two end nodes. If an isoparametric beam element with, say, 3 nodes were used, a consistent lumping of the stiffness ADINA -1.634810-3 Bending moment at mid-span (lb-in) Theory ADINA 1.53104 1.49104
72
would be recommended so that, for each beam element, 1/6th of the foundation stiffness is attributed to the end nodes, and 4/6th to the midside node. $ Note that a small positive (upward) displacement occurs for a few nodes. If no tension can develop in the foundation, a nonlinear analysis could be carried out using the non-linear elastic material model for the truss elements. The foundation material would then be modeled to be elastic in compression but with zero stiffness in tension. Reference [1] Timoshenko, S.P., Strength of Materials, Part II, Third Edition, D. Van Nostrand Comp., 1958.
Y P
region discretized
h b
L 2
Beam properties:
73
A.42
Objective To verify the performance of the two-dimensional plane stress element when employed to model stress concentrations around holes. Physical problem The rectangular perforated strip shown in Fig. A.42 is subjected to a uniform tension at two opposite sides. Finite element model Using symmetry considerations, one-quarter of the strip is modeled with thirty 8-node plane stress elements. Solution results The analytical solution for the stress zz at locations C and D in Fig. A.42 is given in [1]. ADINA gives the following results: Stress zz (N/mm2) location C Theory ADINA 107.5 108.8 Reference [1] Timoshenko, S.P. and Goodier, J.N., Theory of Elasticity, Third Edition, McGrawHill, 1970, pp. 94-95.
Theory 18.75
74
Figure A.42
75
A.44
Objective To verify the dynamic behavior of the two-dimensional plane stress element in frequency analysis when employing skew coordinate systems. Physical problem A cantilever beam of rectangular cross-section as shown in Fig. A.44 is considered. The cantilever is the same as previously analyzed for a static load in Example A.14. Finite element model The cantilever beam is modeled using ten 8-node plane stress elements with a consistent mass matrix. Since the model is inclined in the global coordinate system, a skew coordinate system is used. The subspace iteration method is used for the frequency calculations. Solution results The lowest natural frequency of a cantilever beam is given in [1], p. 108 as:
f =
2 2
EI m
2L
=1.87510407
where m = mass/unit length. In this formula rotary inertia and shear deformations are not considered. ADINA gives the following results: Frequency (Hz) Theory ADINA 81.80 81.19 User hints $ For a pinned-pinned beam [1] p. 181 gives the following expression for the ratio between the frequency including rotary inertia and shear deformations and the frequency including flexural effects only:
76
(f) (f)
where K =
= 1
I 1 + E 2L A KG
2 2
for a rectangular section. Assuming the effects of rotary 12 + 11 inertia and shear deformation to be of the same order for a cantilever, we obtain the following estimate: ( f ) rot . + shear = 80.43 Hz. The ADINA solution for a consistent mass matrix gives a higher value for the lowest natural frequency than this analytical estimate. $ Of course, the determinant search method can also be used and it gives the same frequency results. Reference [1] Blevins, R.D., Formulas for Natural Frequency and Mode Shape, Van Nostrand Reinhold Company, 1979.
10 (1 + )
Z h
L q
Y b
L b h q
Figure A.44
77
A.45
Objective To verify the dynamic behavior of the 2-D plane strain element in frequency analysis and when employing skew coordinate systems. Physical problem Same as in Example A.44. Finite element model Same as in Fig. A.44 except that plane strain elements and a lumped mass distribution are used. The determinant search method is used for the frequency solution. To obtain the same theoretical solution as for the plane stress case of Example A.44 the following material data is used:
E* = 1 + 2 E = 1.89349 10
11
(1 + )
N , * = = 0.230769 2 m 1+
where E and v are the Young's modulus and Poisson's ratio used in Example A.44. Solution results ADINA gives a fundamental frequency of 80.76 Hz. User hint $ An analysis using the consistent mass discretization gives a frequency of 81.19 Hz which is exactly equal to the result obtained in Example A.44 (which should be the case).
78
A.46
Objective To verify the dynamic behavior of the three-dimensional solid element in frequency analysis when employing skew systems. Physical problem Same as in Example A.44. Finite element model Ten 20-node 3-D solid elements are used in the model. A consistent mass discretization and the determinant search method of frequency analysis are used. Solution results ADINA gives a fundamental frequency of 81.16 Hz for motion in the Y-Z plane. This is the second mode of the model. User hint $ Of course the subspace iteration method can also be used and it gives the same results.
79
A.47
Objective To verify the dynamic behavior of the Hermitian beam element in frequency analysis when employing skew systems. Physical problem Same as Example A.44. Finite element model The inclined cantilever is modeled using ten Hermitian beam elements. The element stiffness matrices are calculated in closed form and the mass is represented in a lumped mass matrix. The subspace iteration method of analysis is used. Solution results ADINA gives a fundamental frequency of 80.94 Hz for motion in the Y-Z plane. This is the second mode of the model. User hint $ Employing a consistent mass matrix in this finite element analysis gives a fundamental frequency of 81.15 Hz. Note that the consistent mass matrix for the Hermitian beam element does not include contributions due to shear deformation.
80
A.48
Objective To verify the dynamic behavior of the isoparametric beam element in frequency analysis when employing skew systems. Physical problem Same as in Example A.44. Finite element model The included cantilever beam is modeled using ten parabolic isoparametric beam elements. The element stiffness matrices are evaluated using 244 Gauss integration and the mass is represented in a lumped mass matrix. The determinant search method of frequency analysis is used. Solution results ADINA gives a fundamental frequency of 81.30 Hz for motion in the Y-Z plane. This is the second mode of the model. User hints $ Employing a consistent mass matrix in the finite element analysis gives a fundamental frequency of 81.24 Hz. $ Note that the shear factor for the iso-beam element is always equal to 1.0 and the user has no control over it.
81
A.49
Objective To verify the dynamic behavior of the plate element in frequency analysis when employing skew systems. Physical problem Same as in Example A.44. Finite element model The finite element model consists of twenty 3-node plate elements. A lumped mass matrix is employed in the analysis, i.e. one-third of the element mass is attributed to each element translational degree of freedom. The displacements are constrained to be zero in the Z-direction. The subspace iteration method of frequency analysis is used. Solution results ADINA gives the fundamental frequency of 81.16 Hz. User hint $ A more refined finite element model (40 plate elements) and a lumped mass matrix gives a result of 81.61 Hz while using a consistent mass matrix for this refined model gives a result of 81.76 Hz.
82
A.50
Objective To verify the dynamic behavior of the shell element in frequency analysis when employing skew systems. Physical problem Same as in Example A.44. Finite element model Five 9-node shell elements are used to model the cantilever beam. The element stiffness matrix is calculated using a shear factor of 5/6. A consistent mass discretization is used. The subspace iteration method of analysis is employed in the frequency solution. Solution results ADINA gives a fundamental frequency of 81.17 Hz for motion in the Y-Z plane. This is the second mode of the model. User hints $ Employing a lumped mass matrix in the finite element analysis gives a fundamental frequency of 80.59 Hz.
83
f1 =
Eh3 2 a 12 h (1 2 )
The insertion of the numerical values gives f1 = 12.00 Hz and the ADINA solution is f1 = 11.98 Hz. User hints $ Note that full integration (default) is necessary when a consistent mass matrix is used since reduced integration (222) gives a spurious mode of very small frequency for this example. $ A consistent mass matrix gives 12.03 Hz for the fundamental frequency. Reference [1] Blevins, R.D., Formulas for Natural Frequency and Mode Shape, Van Nostrand Reinhold Company, 1979.
84
Region discretized
Figure A.51
85
A.52
Objective To verify the dynamic behavior of the truss element in wave propagation analysis. Physical problem A uniform bar free at both ends is considered, see Fig. A.52. An axial step force is applied at one end of the bar at time 0. Finite element model The bar is modeled using ten 2-node truss elements. The explicit central difference method is chosen for the time integration and a lumped mass matrix is employed. The time step t is selected according to
t =
Le = Le c E
where Le = length of one truss element, c = wave velocity in the material, = mass density and E = Young's modulus. Since the step load is applied at time 0 it is necessary to specify the corresponding acceleration using
a1 =
P m1
where m1 = mass at node 1, P = applied force and a1 = acceleration at node 1. Solution results The theoretical solution for this problem is presented for example in [1].
86
The following results are obtained with ADINA: Longitudinal stress in bar, (N/m2) Theory ADINA 0.0 0.0 5 -1.010 -1.0105
The wave front can be seen in the ADINA solution to advance one element length per specified time step, which is in agreement with the theoretical solution. User hints $ In this example with equal lengths of the truss elements the exact theoretical solution is obtained by selecting the time step to be equal to the critical time step. The exact solution would not be obtained using a smaller time step or if unequal element lengths are used. $ Note that the load vector input by time functions at the start of the solution is not used in ADINA. Instead the load vector 0 R at time 0 is assumed to be 0 && & R = M 0 U + C 0 U +0F
& && where 0 U and 0 U are the initial velocity and acceleration vectors, M and C are the
mass and damping matrices and 0 F is a vector with the nodal equivalent forces corresponding to the initial displacements. Therefore, the initial acceleration is specified as described above to simulate the applied force at time 0. References [1] Zukas, J.A., Nicholas, T., Swift, H.F., Greszczuk, L.B. and Curran, D.R., Impact Dynamics, John Wiley & Sons, 1982.
Z P X Y
a = 0.1 m L = 1.0 m a a
P = 1000 N at time 0
Figure A.52
87
$ In this model, the mode shape corresponding to the smallest period of the finite element assemblage is not the one-dimensional compression that is analogous to the truss element model. Periods that are smaller that this period exist, and if the model is run for enough time steps, round-off error will cause the solution to become unstable. The smaller periods can be removed by employing constraint equations as in Example A.70.
88
A.54
Objective To verify the dynamic behavior of the beam element and the solution procedure for mass proportional loading due to ground acceleration. Physical problem A water tower subjected to ground acceleration, as shown in Fig. A.54, is considered. The time history of the ground motion is also shown in Fig. A.54. Finite element model One Hermitian beam element is used to model the tower and the water tank is modeled as a concentrated mass. The transverse stiffness is set equal to the spring constant k for the tower. Thus 3EI = k = 2.7 106 lbf/ft L3 and we select (for ease of computation) I = 1 ft4, L = 10 ft, E = 9108 lbf/ft2. The Newmark method of time integration is chosen with =0.25 and =0.50. The time step size is 0.005 sec. Solution results The table below shows the calculated results for some time steps in the analysis together with the analytical solution and results reported in [1]. Time (s) 0.010 0.020 0.030 0.040 0.050 0.075 Analytical 0.000214 0.00169 0.00551 0.0113 0.0174 0.0255 Transverse displacement, u(t) (ft) Ref. [1] ADINA 0.0002 0.000239 0.0017 0.00173 0.0055 0.00554 0.0114 0.0113 0.0176 0.0174 0.0254
89
User hints $ Note that a ground acceleration in the negative X-direction corresponds to an applied mass proportional loading in the positive X-direction. $ Note also that the X-Y-Z system is moving with the ground and we obtain a response measured in this moving system. Therefore, to obtain, for example, the absolute acceleration response (which is measured in a fixed system) it is necessary to add the ground acceleration time history. Reference [1] Clough, R.W. and Penzien, J., Dynamics of Structures, McGraw-Hill, 1975, pp. 102105.
u(t) M
k a(t) Y a0
X a(t)
Figure A.54
90
A.55
Objective To verify the dynamic behavior of the two-dimensional axisymmetric element and the use of direct time integration. Physical problem A cylindrical tube initially at rest is subjected to a uniform circumferential line load at its midspan at time t=0+, as shown in Fig. A.55. Finite element model Using symmetry considerations only one-half of the tube is modeled with sixteen 8node axisymmetric elements. The trapezoidal rule (the Newmark method with =0.25 and =0.50) is employed in the analysis to obtain the step-by-step dynamic response. A step size of 10-5 sec. is used and the solution response is evaluated for 70 steps in the analysis. A consistent mass matrix is used. Solution results The theoretical solution for this problem is presented in [1]. The computed displacements agree well with the theoretical solution. User hints $ The trapezoidal rule is unconditionally stable, and the time step size is therefore selected based on accuracy considerations only. Note that a time step for use of the central difference method would be considerably smaller. $ The numerical value of the density must be entered in consistent units (in this case lbf-sec2/in4). $ In direct time integration, all modes of the finite element model (including poorly modeled high frequency modes) can contribute to the solution. In this solution, the high frequency oscillations seen in the velocity and acceleration time histories are most probably inaccurate.
91
Reference [1] Reismann, H. and Padlog, J., "Forced Axisymmetric Motions of Cylindrical Shells", Journal of The Franklin Institute, Vol. 284, No. 5, Nov. 1967.
Z
L/2
P h
L/2
L = 18.0 in D = 6.0 in h = 0.3 in E = 3.0 10 7 psi n = 0.3 r = 3.66 10-2 lb-sec2/in4 P = 1000 lb/in at time 0
Figure A.55
92
A.57
Objective To verify the behavior of the cubic isoparametric beam element when employed to model a thick-walled curved beam. Physical problem A thick-walled beam clamped at one end and subjected to a concentrated tip load, as shown in Fig. A.57, is considered. Finite element model Two 4-node isoparametric beam elements are used to model the beam. Solution results The following results are obtained: Tip deflection, w, (in) Theory [1] ADINA -3 -5.3610 -5.3110-3 At the integration points closest to the fixed end, ADINA gives Radius (in) 10.208 10.990 12.010 12.792 Reference [1] Timoshenko, S.P. and Goodier, J.N., Theory of Elasticity, 3rd Ed., McGraw-Hill, 1970, pp. 83-88. Axial stress (psi) Theory [1] -73.42 -26.79 24.50 58.61
93
R
Z
Figure A.57
94
A.58
Objective To verify the mode superposition analysis procedure and the force arrival time option when applied to a model employing Hermitian beam elements. Physical problem The simply supported beam shown in Fig. A.58 is subjected to a constant magnitude force traveling across its span at a constant velocity. Finite element model The model consists of twenty equally spaced Hermitian beam elements with a lumped mass distribution. The dynamic response of the beam is calculated using the mode superposition analysis with one mode. To model the traveling load the time function and arrival time options are used in ADINA. Solution results The analytical solution for a one mode approximation is given in [1]. The following results are obtained: Time (s) 0.04 0.08 0.12 Reference [1] Biggs, J.M., Introduction to Structural Dynamics, McGraw-Hill, 1964, pp. 315-318. Mid-span deflection, uy (in) Theory ADINA -0.0152 -0.0160 -0.1080 -0.1071 -0.3103 -0.3053
95
Y v v. t P
X L
Time variation of load for each node: F(t) L = 480 in 1 EI = 2.0 1010 lb-in 2 n = 0.0 r = 0.1 lbf-sec2/in4 P L3 = 1 48 EI v = 600 in/sec
0.04
0.08
t [s]
Figure A.58
96
A.59
Objective To verify the dynamic behavior of the plate element in frequency analysis. Physical problem The triangular cantilever plate shown in Fig. A.59 is analyzed for its free vibration response. Finite element model The finite element model consists of seventy-two 3-node plate elements. A consistent mass distribution is employed in the analysis and the natural frequencies are calculated using the subspace iteration method. Solution results In the table below, the six lowest natural frequencies predicted by the finite element model are shown. The analysis results using ADINA are also compared with the predictions given by a higher precision element and experimental results [1]. Mode Natural frequencies (Hz) Experimental Higher results precision element 34.5 36.6 136.0 139.3 190.0 194.1 325.0 333.8 441.0 455.4 578.0 593.2
ADINA
1 2 3 4 5 6 Reference [1]
Cowper, G.R., Kosko, E., Lindberg, G.M. and Olson, M.D., "Static and Dynamic Applications of a High-Precision Triangular Plate Bending Element", AIAA Journal, Vol. 7, No. 10, Oct. 1969, pp. 1957-1965.
97
Figure A.59
98
A.62 Analysis of orthotropic plate, patch of 4-node plane stress and shell elements
A.62
Analysis of orthotropic plate, patch of 4-node plane stress and shell elements
Objective To verify the orthotropic material model in the patch test of 4-node plane stress and 4node shell elements. Physical problem The orthotropic plate is loaded in-plane by biaxial tension and shear, as shown in Fig. A.62. Finite element model Two patches of five 4-node shell elements and plane stress elements are used. Material constants for the shell element correspond to those of the plane stress element. Solution results The theoretical solution is given in [1]. The displacements of point B are as follows: u (m) Theory 2.0307710-6 ADINA 2.0305110-6 Theory 2.7531410-6 v (m) ADINA 2.7534610-6
A constant stress state of x = y = xy = 1 corresponding to the applied loading is obtained in all elements. The solution for strains is the same as the analytical solution. Hence the patch test is passed. User hints $ For the shell element additional loading conditions would have to be considered to prove that the element passes all patch tests. $ The patch test is also passed if a materially-nonlinear-only analysis is performed.
99
Reference [1] Jones, R.M., Mechanics of Composite Materials, Scripta Book Company, 1975.
Z Fy F z S S S B Fz Fy S
Figure A.62
100
A.63 Pure bending of orthotropic plate, patch of two 8-node distorted shell elements
A.63
Pure bending of orthotropic plate, patch of two 8-node distorted shell elements
Objective To verify the orthotropic material model in the patch test of the 8-node (MITC8) shell element. Physical problem The orthotropic plate shown in Fig. A.63 is subjected to pure bending conditions. Finite element model Two 8-node shell elements are used and consistent bending moments are applied at the nodes to produce constant bending within the elements. Solution results The following results are obtained: Point A B C Vertical displacement Theory [1] ADINA 0.07875 0.07875 0.07510 0.07511 -0.09405 -0.09405
A constant stress field corresponding to pure bending is obtained. User hints $ The patch test is passed for the 8-node shell element, provided the element sides are straight. Reference [1] Jones, R.M., Mechanics of Composite Materials, Scripta Book Company, 1975.
101
Thickness = 0.1 m B L b My a = 10 m = 30
My L
M y = 6 N-m Eaa = 2.17391 10 7 N/m 2 Eab = 4.34783 106 N/m 2 1.08696 107 N/m 2 8 10 5
N/m 2
E bb = Gab = Gt =
8 10 5 N/m2
Figure A.63
102
A.64
Objective To verify the behavior of the two-dimensional axisymmetric potential-based fluid element in a fluid-structure static analysis with multiple fluid regions. Physical problem Same as Example A.34. Finite element model Because of symmetry only one half of the assemblage needs to be considered in the axisymmetric model. The finite element model consists of fifty 8-node axisymmetric structural elements and fifty 8-node axisymmetric potential-based fluid elements. The stiff cap is modeled using rigid links. The potential-based fluid elements are divided into five element groups, one group for each fluid region. Fluid-structure interface elements are used to couple the solid elements and the fluid elements (these interface elements are automatically generated by the AUI). Solution results The following results are obtained: Cylinder 1 2 3 4 5 User hints $ It is necessary to enter the density of the fluid even for static analysis. Pressure (KPa) Experimental [1] ADINA 51.3 55.7.4 42.9 45.2.6 33.4 34.0.5 23.5 25.3.5 12.6 14.1.4
103
Reference [1] Munro, M. and Piekarski, K., "Stress Induced Radial Pressure Gradients in LiquidFilled Multiple Concentric Cylinders", J. of Appl. Mech., Vol. 44, No. 2, pp. 218-221, 1977.
104
n m = c + , m, n = 0,1, 2,3,..., c = a b
The following results are obtained with ADINA: Natural frequency (rad/sec) Theory ADINA 9132 9166 15220 15277 17749 17846 User hints
$ Note that there is one rigid body mode in the solution. $ This problem can also be solved using ADINA-T, see Example T.1.
105
Reference [1] Blevins, R.D., Formulas for Natural Frequency and Mode Shape, Van Nostrand Reinhold, 1979, pp. 337-341.
a
a = 12 in b = 20 in k = 3.16 . 10 5 psi (bulk modulus) r = 9.35 . 10-5 lb-sec2/in4 (density)
Figure A.65
106
A.66
Objective To verify the dynamic behavior of the 4-node shell element in frequency analysis when employing skew systems. Physical problem Same as in Example A.44. Finite element model Ten 4-node shell elements are used to model the cantilever beam. The subspace iteration method of analysis and a consistent mass discretization are employed in the frequency analysis. Solution results ADINA gives a fundamental frequency of 81.33 Hz for motion in the Y-Z plane. This is the second mode of the model. User hints $ Employing a lumped mass matrix in the finite element analysis yields a fundamental frequency of 81.17 Hz. $ Both the first and second mode natural frequencies are in good agreement with the analytical solution. $ If incompatible modes are not used in the shell elements, the second mode natural frequency is in good agreement with the analytical solution, but the first mode natural frequency is considerably higher than the analytical solution. This is because the first mode is modeled using the in-plane membrane action of the shell (linear displacement interpolations), which is too stiff when modeling in-plane bending, whereas the second mode is modeled using the out-of-plane bending action of the shell.
107
A.67
Objective To verify the 4-node shell element for use in modeling cylindrical shell structures. Physical problem Same as Example A.30. Finite element model As in Example A.30 only one-quarter of the structure needs to be modeled. 144 shell elements are used. Solution results The vertical deflection at point B of Fig. A.30 is predicted to be -3.626 inches. User hints $ Using the 4-node shell elements, the effect of mesh refinement on the obtained solution for the vertical displacement at point B is shown in ref. [1]. Reference [1] Dvorkin, E.N., Bathe, K.J. "A continuum mechanics based four-node shell element for general nonlinear analysis", Engineering Computations, Vol. 1, March 1984, pp. 79-88.
108
A.68
Objective To verify the pipe element for modeling the in-plane bending behavior of a flanged elbow. Physical problem Fig. A.68 shows the flanged elbow problem considered. Finite element model Only one half of the pipe elbow (from A to C, see Fig. A.68) is modeled because the ovalization boundary conditions are symmetric about the mid-point C and the elbow is subjected to a constant bending moment. Six 4-node pipe elements are employed. The boundary conditions are $ At point A: $ At point C: Solution results The flexibility factor for the pipe is given by f = point B. ADINA gives the following result: Flexibility factor, f Experiment [1] ADINA 1.80 2.06 Reference [1] Whatham, J.F., "In-Plane Bending of Flanged Elbows", Proceedings Metal Structures Conference, The Institution of Engineers, Perth, Australia, Nov. 30 - Dec. 1, 1978. Fixed displacement, rotation, in-plane ovalization and warping. The zero-slope-of-pipe-skin constraint is applied. The zero-slope-of-pipe-skin constraint is applied.
109
d A C q R 45 g B 2a
M Z Y X
Figure A.68
110
A.69
Objective To verify the option of consistent mass centrifugal loading in the analysis of rotating axisymmetric structures. Physical problem A long hollow cylindrical shaft rotating with an angular velocity of 10000 rad/sec is considered (see Fig. A.69). The cylinder is assumed to be guided so that no axial displacement can occur. Finite element model Due to the end boundary conditions and the applied loading, the zz-component of strain is zero along the entire length of the shaft and therefore only a unit length of the shaft is discretized. Eight 8-node axisymmetric elements are used to model the shaft in the radial direction. Consistent mass centrifugal loading is employed. Solution results The analytical solution is given in [1]:
xx = yy =
r 2 r 2 1 + 2 2 1 3 2 2 2 r r0 + r12 + 0 2i r 8 1 3 2 r 2r 2 1 3 2 2 2 r0 + r12 + 0 2i r 2 , zz = ( xx + yy ) 8 1 r
The displacement in the radial direction can be calculated from the circumferential strain:
= r x =
r xx ( yy + zz ) E
111
ADINA gives the following results: Y-coord (in) 0.16 0.25 Y-coord (in) 0.16 0.25 User hints $ The numerical results are not as good when lumped mass centrifugal loading is employed. Reference [1] Wang, C.T., Applied Elasticity, McGraw-Hill, 1953. Radial displacement (10-5 in) Theory ADINA 2.0912 2.0912 1.7054 1.7053 xx (psi) Theory 1470 767.4 ADINA 1471 767.6 yy (psi) Theory 0. 0. ADINA 2.8 0.7 zz (psi) Theory 490.1 255.8 ADINA 491.2 256.1
E = 1.0 107 psi ro w ri Y Z n = 0.33333 r = 2.54 10-4 lb-sec2/in4 w = 10000 rad/sec ro = 0.25 in r i = 0.16 in
Figure A.69
112
A.71
Objective To calculate the natural frequencies of a water column with a free surface using the two-dimensional potential-based fluid element. Physical problem The two-dimensional rectangular cavity, shown in Fig. A.71, is filled with water with a free surface. The water is analyzed for its natural frequencies. Finite element model Four 8-node planar potential-based fluid elements are used to model the water in the cavity. No structural elements nor fluid-structure elements are used. All nodes have velocity potential degrees of freedom except the nodes on the free surface where the velocity potential degrees of freedom are deleted. The modal stresses are calculated in ADINA so that the AUI can display the modal pressures. Solution results The theoretical solution for the natural frequencies is [1]
where c is the wave velocity. This solution does not include gravity effects in the surface. ADINA gives the following results: Mode 1 2 3 Natural frequency (rad/sec) Theory ADINA 4566 4567 13698 13924 22830 25517
113
User hints $ If a more accurate modeling of the free surface is desired with the inclusion of gravity effects, free surface elements can be included as in Example A.114. $ There are no rigid-body modes in this frequency analysis because some of the velocity potential degrees of freedom are deleted. Reference [1] Olson, L.G. and Bathe, K.J., "A Study of Displacement-Based Fluid Finite Elements for Calculating Frequencies of Fluid and Fluid-Structure Systems", Nucl. Engng. Des., Vol. 76, 1983, pp. 137-151.
a b k r
= = = =
Water
Figure A.71
114
A.72
Transient analysis of a water column under compression, potential-based and displacement-based elements
Objective To verify the behavior of the two-dimensional potential-based fluid element for transient analysis. Physical problem Fig. A.72 shows the physical problem in which a column of air is compressed by a rigid, massless piston. Finite element model The model contains two unconnected regions, each one modeling the same water column using different fluid elements. One region consists of four-node potentialbased fluid finite elements. The rigid piston is modeled by prescribed displacements applied directly to the potential-based fluid region. The AUI automatically generates a fluid-structure interface element on the corresponding boundary. The other region (with a mesh identical to the potential-based mesh) consists of fournode displacement-based fluid elements (included for comparison), with the prescribed displacements applied directly onto the fluid. The trapezoidal rule is used in the time integration with a time step increment of 610-4 sec. Solution results The results for the two models compare well.
115
Prescribed displacement U
1m
1m
Time
116
A.73
Objective To verify the behavior of three-dimensional potential-based fluid elements in transient fluid-structure interaction analysis. Physical problem A stiff, massive plate sits on top of a column of water with a spring connecting the plate to the base of the container; see Fig. A.73. An initial positive displacement (moving away from the base) is applied to the plate, and a corresponding decrease in pressure to the water. The transient solution response is to be determined. Finite element model Two three-dimensional solid elements with a large stiffness are used to model the plate, and a single two-node truss element represents the spring. Four threedimensional potential-based fluid elements model the fluid. The AUI automatically generates fluid-structure interface elements between the fluid and the solid. The trapezoidal rule with a time step increment of 1.510-3 sec is used in the solution response calculation. An initial displacement of 0.2 m is applied to the geometry volume used to model the plate. The corresponding initial pressure in the fluid is -4.2107 Pa (fluid pressure is positive in compression). This initial pressure corresponds to an initial potential & velocity through the formula p = . Therefore an initial potential velocity of 2 2 42000 m /s is applied to the geometry volume used to model the fluid. Solution results The theoretical solution for the lowest natural frequency of this problem is given in [1]:
L k + cA cot c = 0, c = m
where c is the wave velocity and A is the surface area of the plate. Substituting numerical values gives a natural frequency of 211.3 rad/sec.
117
The ADINA model is observed to vibrate with a frequency close to that predicted by the theoretical solution. Reference [1] Olson, L.G. and Bathe, K.J., "A Study of Displacement-Based Fluid Finite Elements for Calculating Frequencies of Fluid and Fluid-Structure Systems", Nucl. Engng. Des., Vol. 76, 1983, pp. 137-151.
Plate (rigid) 1000 kg mass
L = 10 m b = 1 m
Figure A.73
118
A.74
Objective To verify the two-dimensional potential-based fluid element for added mass effects in frequency analysis. Physical problem A rigid cylinder is attached to a massless spring and submerged in water, as shown in Fig. A.74. The frequency of vibration of the cylinder is to be determined. Finite element model Due to symmetry only half of the domain is analyzed. The cylinder is modeled as a rigid structure, while the spring is modeled as a two-node truss. Eight 8-node potential-based fluid finite elements are employed to model the fluid region. The AUI automatically generates fluid-structure interface elements to connect the fluid and structural regions. The nodes on the interface elements are constrained by the node with the concentrated mass. Modal stresses are calculated by ADINA so that the AUI can display the modal pressures. Solution results The added mass for the cylinder is [1]
m = a 2
and therefore the natural frequency of the system is
k m + m
119
The following results are obtained with ADINA: Natural frequency (rad/sec) Theory ADINA 4.44 4.532 2276. 2278.
$ The use of the displacement-based fluid element will produce poor results in this type of fluid-structure analysis [2]. $ The actual physical problem may be that the cylinder vibrates in an infinite fluid region. In this case, a frequency solution cannot directly be performed, but the transient response can be calculated, see Examples A.75 and A.76. $ The acoustic frequency of the box enclosing the fluid varies with the size of the box. Note that these acoustic frequencies are here much higher than the frequency of the cylinder. References [1] [2] Harris, C.M. ed., Shock and Vibration Handbook, Third Edition, McGraw-Hill, 1988, Chapter 29. Olson, L.G. and Bathe, K.J., "A Study of Displacement-Based Fluid Finite Elements for Calculating Frequencies of Fluid and Fluid-Structure Systems", Nucl. Engng. Des., Vol. 76, 1983, pp. 137-151.
region discretized
2m
Figure A.74
120
A.75
Objective To verify the axisymmetric potential-based infinite fluid elements. Physical problem A sphere of stiffness k and mass m that "breathes" (moves in a purely radial direction) in an infinite fluid medium is analyzed, see Fig. A.75. The motion starts when a pressure pulse is applied to the interior of the sphere at time zero. The transient response of the sphere is to be determined. Finite element model Symmetry considerations allow the modeling of only one half of the domain. The sphere is modeled using axisymmetric shell elements. Near the sphere axisymmetric potential-based fluid finite elements are used, and at the outer edge of this zone a potential-interface of type infinite (spherical) is used. At time zero, a step in pressure is applied to the inside of the sphere. The trapezoidal rule is used for the time integration with a time step increment of 210-4 sec. Solution results The theoretical solution for this problem is given in [1]. The ADINA model gives almost the exact solution. User hint $ The thickness of the sphere is neglected, i.e. the pressure is applied to the midsurface of the axisymmetric shell elements and the fluid is attached to the midsurface. Reference [1] Olson, L.G. and Bathe, K.J., "An Infinite Element for Analysis of Transient FluidStructure Interactions", Eng. Comp., Vol. 2, pp. 319-329, 1985.
121
r = 10 3 k = 10 8
a=1
Sphere E=5 108 n=0 t=0.1 r=10000 Equivalent sphere properties: m=4000 p k=4 108 p
Figure A.75
122
A.76
Objective To verify the axisymmetric potential-based infinite fluid element. Physical problem A circular plate is mounted (clamped) into an infinite wall and exposed on one side to water and the other side to vacuum, see Fig. A.76. The transient response of the plate is to be determined. Finite element model The plate is represented by six eight-node axisymmetric solid elements. The water near the plate is modeled using axisymmetric potential-based fluid finite elements and water further from the plate is modeled using a potential-interface of type infinite (spherical). The plate is given an initial displacement corresponding to its first mode in vacuum and then released. The trapezoidal rule of time integration is employed to calculate the transient response. Solution results An approximate analytical solution is given by Lamb in [1]. The finite element and approximate analytical results show good agreement. References [1] Lamb, H. "On the Vibrations of an Elastic Plate in Contact with Water", Proceedings of the Royal Society (London), Series A, Vol. 98, pp. 205-216, 1921.
123
(Vacuum) Circular iron plate clamped into wall: Infinite, rigid wall
Figure A.76
124
A.80
Objective To verify the response spectrum analysis methods for beam elements. Physical problem A simply supported beam is exposed to an earthquake load and a gravity load in the Z-direction as shown in Fig. A.80. The earthquake load is defined by the given acceleration response spectrum. The gravity load acting in the negative Z-direction is 9.81 m/s2. It is desired to estimate the maximum response of the beam using the response spectrum method. Finite element model The model consists of eighteen 2-node beam elements. A consistent mass matrix is used and the subspace iteration method is used for the frequency analysis. In the response spectrum calculations, two response spectrum definitions and one response combination are considered. Response spectrum definition RS1 is an SRSS combination of the modal displacements. Response spectrum definition RS2 is also an SRSS combination of modal displacements but also includes the residual displacements. The response combination is a superposition of response spectrum RS2 upon the displacements due to gravity loading. The input for this problem is divided between the pre- and post-processing phases of the analysis as follows: Pre-processing: $ $ $ $ Material properties, problem geometry, finite elements and boundary conditions One load step corresponding to the static load case A frequency analysis with a consistent mass matrix Response spectrum calculations and residual (static correction) calculations are to be performed.
125
Post-processing: $ $ $ The damping factors of each of the modes The earthquake response spectrum The response spectrum and response combination definitions.
Solution results The calculated results are in good agreement with analytical results. Some selected results obtained for spatial combination number 1 are as follows: Theory -0.3591 64.19 -58861. 19684. ADINA -0.3575 63.76 -58746. 19067.
Displacement at X=5 m Acceleration at X=5 m Moment at X=5 m Reaction at X=0 m User hint
$ The option of calculating residual quantities can be important when a significant portion of the response comes from high frequency modes, namely those modes which are not explicitly included in the response spectrum solutions.
126
u(X,t) Z
ug(t)
E = I = A = L = r =
EI , A , r
L
All modes have a damping factor of 1%. Earthquake response spectrum: Frequency 0.5% 2.0% (Hz) damping damping 33 9.81 9.81 9 48.66 34.74 2.5 58.37 41.69 0.25 7.22 5.64
127
Membrane force (per unit length) Bending moment (per unit length)
User hint $ The ADINA membrane force and bending moment are computed by dividing the y reaction and x-moment reaction by the radius of the shell (reactions are computed by ADINA per radian of structure). Reference [1] Timoshenko, S. P. and Woinowsky-Krieger S., Theory of Plates and Shells, 2nd Edition, McGraw-Hill, 1959, p. 552.
128
1.0
50.0
Figure A.104
129
16 ( x 5 ) 225
y2 + z2 =1 9
Finite element model Taking advantage of the geometric symmetry of the whole structure, only one cyclic part of the whole structure needs to be modeled. The following types of elements are used in the cyclic part: Shell: Iso-beam: Beam: 3-D: Plate: Ten 4-node elements to model the tower. One element to model the stiffener on the lower side of the tower. Two elements to model the supporting feet. Two elements to model the foundation. Eight elements to model the service platform on the upper edge of the tower. Truss: Two elements to model the supporting arm for the service platform. Pipe: One element on the lower edge to model the supply piping system. 3-D plane stress: Two elements on the top of the foundation to model the cover of the foundation. Solution results To verify the results obtained when using cyclic symmetry, the same problem was analyzed by modeling the whole structure without using cyclic symmetry. All the results, including the natural frequencies, displacements, stresses and reactions, are identical with the cyclic symmetry analysis. For example, the first four frequencies obtained from both analyses are
130
Mode number 1 2 3 4
For this problem, the use of cyclic symmetry drastically reduces the amount of input required for the model. User hints $ Two beam elements are used on each side of the cyclic part to model the supporting foot, which happens to be on the cyclic symmetric inner boundary lines. The full section properties of the beam are entered and ADINA correctly accounts for the reduction of the stiffness (and mass) for those elements lying completely on the inner boundaries. The same is true for the truss elements. $ Note that the shell nodes on the cyclic symmetric inner boundary lines must be 6 DOF nodes. $ When comparing mode shapes from the cyclic symmetry model to those from the whole model, note the following. For mode shapes corresponding to distinct eigenvalues, all of the components of the two results are proportional. For mode shapes corresponding to repeated eigenvalues, the mode shape from one analysis can be constructed as a linear combination of mode shapes from the other analysis.
131
Tower
Supporting feet
Foundation
Figure A.110
132
Figure A.111
133
A.113 Multilayered simply-supported square plate under uniform pressure (4-node shell)
Objective To verify the computation of the stresses in the thick multilayered 4-node shell element when using the orthotropic material model. Physical problem A simply-supported square plate is subjected to a uniform transverse pressure load (Fig. A.113). The lay-up is made of three layers, all with the same orthotropic material properties. The material axes are different in each layer and are oriented at 0, 90 and 0 degrees from the x axis, respectively, for layers 1, 2 and 3. The layers have the same thickness. Finite element model A mesh of 66 4-node shell elements is used to model one quarter of the plate. For each layer, the integration order through the thickness is set to 5 Newton-Cotes points to obtain an accurate profile of the transverse shear stresses. Solution results The normal displacement at the center of the plate and the transverse shear stress distributions at points A and B are compared with values obtained by superposition (1313 terms) of the analytical solutions given by Pagano for sinusoidal pressure loads [1]. The normal displacement results are Theory 1.2710-5 ADINA 1.2810-5
Good agreement with the solution of Pagano is observed. Reference [1] Pagano, N.J., "Exact solutions for rectangular bidirectional composites and sandwich plates", J. Comp. Mat., 4, pp. 20-34, 1970.
134
A.113 Multilayered simply-suppored square plate under uniform pressure (4 node shell)
region discretized B b l A Y X a
Pressure = 10
-4
20 Ea = 25.0 10 4 Eb = 1.0 10 4 Gab = 0.5 10 4 nab = 0.01 Gac = 0.5 10 4 nac = 0.01 Gbc = 0.2 10 4 nbc = 0.25 All layers have equal thickness. Ply orientations b: 0/90/0
Figure A.113
20
135
g h ( kL ) tanh ( kL ) , L L
( kL ) = , 2,...
Acoustic waves:
=c
i ( 2 j 1) , i = 0,1,..., + 2h L
j = 1, 2,...
where ishe natural frequency (rad/s) g is the gravitational acceleration, L is the tank width, h is the height of water in the tank and c is the wavespeed ( c =
).
136
Mode number 1 2 3 4 5 6 7 8
Natural frequency (rad/sec) Theory ADINA 0.0 1.010-5 1.188 1.188 1.751 1.753 2.149 2.157 2.481 2.509 2.774 2.920 3.039 3.208 3.283 3.577
Mode number 9 10 11 12 13 14 15 16
Natural frequency (rad/sec) Theory ADINA 3.509 3.973 3.722 4.329 3.923 4.487 232.0 232.0 328.2 328.2 518.9 519.5 696.1 696.5 733.8 734.1
Modes 12 to 16 are acoustic wave modes that have much higher frequencies than the gravity wave modes. User hints $ Only eleven gravity wave modes can be calculated by this finite element model because there are only eleven nodes on the free surface. $ If the gravity wave modes are not desired, but the acoustic wave modes are, the velocity potential degrees of freedom on the free surface should be deleted as shown in Example A.71. Reference [1] Sarpkaya, T., Isaacson, M., Mechanics of Wave Forces on Offshore Structures, Van Nostrand Reinhold, Co., 1981.
20
Figure A.114
137
F21 =
where F21 = force on conductor 1 due to interaction with conductor 2, i1(t) = current in conductor 1, i2(t) = current in conductor 2, dr1 = differential vector element of conductor 1, dr2 = differential vector element of conductor 2, r21 = vector from the middle of differential vector element dr1 to the middle of differential vector element
138
Location A B C D
A Y 5.0 C
1.0
B
30 D X
Figure A.115
139
140
Solution results Harmonic loading: Response HARMONIC_MAXIMUM gives the maximum values of the beam response for a loading frequency of 40 Hz. Some of the maximum values are Z-displacement at center of beam Z-reaction at support Z-acceleration at center of beam Moment at center of beam Random loading: Response RANDOM gives the RMS values of the beam response due to the random loading. Some of the RMS values are Z-displacement at center of beam Z-reaction at support Z-acceleration at center of beam Moment at center of beam : : : : 7.35918104 m 3.75600103 N 1.90746102 m/s2 2.16056103 N-m
2.0 1011N/m2 8.33333 10 -6 m 4 0.01 m 2 10 m 7800 kg/m3
: : : :
u(X,t) Z
ug(t)
E = I = A = L = r =
EI , A , r
L
Figure A.116(a)
141
PSD of load amplitude (g2/Hz) 0.014 1.0 1.0 0.00012 Figure A.116(b)
142
K I = 0.475 K 0 , K 0 =
P b 2
b , P = p R 2
Numerical bounds of K I can be obtained from the value of the energy release rate given by the finite element analysis, considering that K I =
(1 2 )
EG
using plane
strain assumptions and K I = EG using plane stress assumptions. These relations provide lower and upper estimates of K I . ADINA gives the following results, using virtual shift number 4: KI (107 N-m-3/2) ADINA ADINA (plane stress) (plane strain) 3.267 3.425
Theory 3.368
143
The maximum difference between the reference value and the ADINA values is 3 %. User hints $ Note that, for symmetry reasons, the total energy release rate G is twice the value given by the finite element model. $ The value of the energy release rate given by the first virtual shift is not accurate since low order elements are used to mesh the crack tip area. Triangular 6-node elements with 1/4 point midside nodes should be used to obtain a better representation of the stress/strain singularity, if the result from the virtual shift including only the crack tip node is to be considered. Reference [1] Y. Murakami, Stress Intensity Factor Handbook, Pergamon Press, 1987.
p
b 2R
Figure A.117
144
K I = 1.317 K 0 , K 0 = p a
so, for this geometry, K I = 1.651 107 N-m -3/2 . K I is calculated from the ADINA energy release rate G using the relation K I = ADINA calculates G = 1207
(1 2 )
EG
N-m using virtual shift 6, and from this G, m K I = 1.657 107 N-m -3/2 . The comparison is very good.
Reference [1] I.J. Smith, "The effect of geometry changes upon the predicted fatigue strength of welded joints." Proc. 3rd Int. Conf. Num. Meth. in Fract. Mech., Swansea, U.K., 1984.
145
p b = 0.05 m a/b = 0.10 L/b = 12.0 t/b = 1.0 w/b = 0.5 a = 45 E = 2.0710 11 N/m2 n = 0.3 p = 1.0 108 N/m2 a t
a L
Figure A.118
146
M 1.0
1.0
Figure A.120(a)
147
Three shell materials: Material 1: Material 2: E=2.01011, =0.3 Ea=1.01011 ab =0.3 Gab=7.69231010 E= 2.0109, =0.3 Eb=2.01011 bc =0.3 Gbc=7.69231010 Ec=1.01011 ac =0.3 Gcc=3.81010
Layer 1 uses material 1 Layer 2 uses material 3 Layer 3 uses material 1 Shell model 2: Layer 1 uses material 2 Layer 2 uses material 3 Layer 3 uses material 2 The failure stress for both shell models is 3.0108. Figure A.120(b)
148
149
frequencies FMIN to FMAX) and PSL is the pressure spectrum level (the sound pressure level contained within a band of frequencies 1 Hz wide). Each pressure spectrum level value is assigned to the frequency at the center of the frequency band. Solution results Of the first 10 mode shapes, only modes 1 and 6 have nonzero modal participation factors with respect to a constant pressure load. This is because of the symmetry of the problem and its loading. The RMS displacements, accelerations and bending moments in the plate are calculated as a modal-combination. The maximum obtained values over the plate are Maximum RMS displacement: 3.41559 10-4 m Maximum RMS acceleration: 7.92526 10-2 m Maximum RMS bending moment-rr: 1.40203 N-m/m Maximum RMS bending moment-ss: 1.40203 N-m/m Maximum RMS bending moment-rs: 0.43733 N-m/m User hint $ We could also have used symmetry boundary conditions and modeled only onequarter of the plate. Reference [1] Seto, W. W., Schaum's Outline of Theory and Problems of Acoustics, McGraw-Hill Book Company, 1971.
150
0.2 m
Thickness = 0.001 m E = 7.1 1010 N/m 2 n = 0.334 r = 2700 kg/m3 Edges of plate are clamped
Noise characteristics: Frequency interval (Hz) 100.000 to 125.990 125.990 to 158.740 158.740 to 200.000 200.000 to 251.980 251.980 to 317.480 317.480 to 400.000 400.000 to 503.790 503.790 to 634.960 634.960 to 800.000 800.000 to 1007.930 1007.930 to 1269.920 1269.920 to 1600.000
Pressure band level (db) 114.0 120.0 126.0 132.0 138.0 139.0 135.0 131.0 127.0 123.0 119.0 115.0
(the frequency intervals are the 1/3 octave bands between 100 Hz and 1600 Hz). The reference pressure is the standard value of 2.0 10-5 N/m2 (corresponding to air). Figure A.121
151
A.122 Floor response spectrum analysis of an inclined building, response spectrum load
Objective To demonstrate floor response spectrum analysis when ground motion response spectra are specified. Physical problem A simplified model of an inclined building is shown in Fig. A.122. A machine that is very light is modeled as a single degree of freedom (SDOF) system and vibrates in the horizontal direction as shown. The response of the machine as a function of the machine natural frequency is to be determined. The building plus machine system is subjected to an earthquake with horizontal ground motion and vertical ground motion specified by response spectra. These response spectra are given in the input data to this problem. Finite element model The building is modeled using six Hermitian beam elements. Notice that the machine is not included in the finite element model. The first 6 natural frequencies and mode shapes of the building are calculated using the subspace iteration method. The user input for this problem is divided between the pre- and post-processing phases of the analysis as follows: Pre-processing: Building model geometry, number of natural frequencies to calculate, material properties, boundary conditions. ADINA is requested to perform response spectrum calculations. Post-processing: Modal damping factors, earthquake response spectra, machine natural frequency range and damping factor functions. The SRSS method is used to combine modal responses and spatial responses. The building natural frequencies are re-entered into a frequency table to specify that floor response spectrum values are calculated for machine frequencies equal to building natural frequencies. Also the peak broadening factor for each building natural frequency is specified in the frequency table.
152
A.122 Floor response spectrum analysis of an inclined building, response spectrum load
Solution results The results are presented as floor response spectrum curves. Results are presented for the two loads acting separately and then for a combined load case in which the x direction loads are scaled by 0.9 and y direction loads are scaled by 1.5. The combined load case results are presented as unsmoothed curves and as smoothed curves with peaks broadened. Also presented are the combined load case results for different damping factor functions for the machine. To provide a comparison, a finite element model of the building plus a machine with natural frequency of 4 Hz was analyzed using the conventional response spectrum method. The machine was modeled using a spring element and a lumped mass of 0.001 kg. In the comparison, only the earthquake load in the x direction is considered: Model/analysis Building only, floor response spectrum, 6 modes (this analysis) Building only, floor response spectrum, 12 modes Building plus machine, response spectrum, 6 building modes Building plus machine, response spectrum, 12 building modes Acceleration (m/s2) 15.8488 15.8925 15.8925 15.8925
The analysis of building plus machine, 12 building modes is taken to be the "exact" solution since all of the building modes are considered. When the floor response spectrum analysis contains all of the modes of the model, the machine damping factor is small and the machine natural frequency is not close to any building natural frequency, then identical results to the conventional response spectrum method are obtained. When the floor response spectrum analysis does not contain all of the modes, then a small error is present due to neglecting the remaining modes. User hints $ We deliberately incline the building so that horizontal ground motions cause responses in the vertical direction and vice versa. Thus the physical problem is somewhat artificial. $ The acceleration computed in this floor response spectrum analysis is an absolute acceleration, but the displacement and velocity are relative to the ground motion.
153
$ It is important to obtain floor response spectrum responses for machine natural frequencies equal to the building natural frequencies in order to estimate the maximum machine response. $ Not all of the building natural frequencies correspond to peaks in the floor response spectra. Only those building natural frequencies that correspond to peaks should have nonzero peak broadening factors in the frequency table. $ When the ground motion is specified by time histories, floor response spectra can be calculated, see Example A.123.
A
10 m
Building girders:
10 m
60
E = 2 1011 N/m2 r = 7800 kg/m2 A = 0.01 m 2 Iss = Itt = 8.333 10 6 m 4 -5 m 4 Irr = 1.470 10 The machine is not included in the finite element model.
10 m
Building mode damping factors: Mode Damping ratio Mode (in percent) 1 1.0 4 2 1.5 5 3 2.0 6
Figure A.122
154
A.123 Floor response spectrum analysis of a building with sloped roof, time history load
A.123 Floor response spectrum analysis of a building with sloped roof, time history load
Objective To demonstrate floor response spectrum analysis when ground motions are specified by time histories. Physical problem A simplified model of a building with a sloped roof is shown in Fig. A.123(a). Notice that the building is oriented at an angle to the X-axis. A machine that is very light is connected to point A. The machine is modeled as a single degree of freedom (SDOF) system and vibrates in the X coordinate direction. The building plus machine system is subjected to three ground motions, each ground motion acting in a coordinate direction. Time histories of these ground motions are given in Fig. A.123(b). The maximum response of the machine as a function of the machine natural frequency is to be determined. The machine natural frequency can vary from 0.5 to 5 Hz and the machine damping ratio is 2.0%. Finite element model The building is modeled using 58 Hermitian beam elements. Notice that the machine is not included in the finite element model. A consistent mass matrix is chosen and the Newmark method with = 0.25, = 0.5 is used for the time integration. For each ground motion direction, an ADINA linear dynamic analysis is performed. A time step of 0.01 seconds is chosen to accurately integrate frequencies of up to 5 Hz. The response is obtained for the duration of the ground motion (1.5 seconds). These ADINA analyses are combined into a single ADINA-PLOT database file. This permits ADINA-PLOT to combine the floor response spectra from the different ground motions.
155
Solution results The calculated floor response spectra show the maximum absolute acceleration of the machine as a function of the machine natural frequency, for the ground motions acting separately and for a SRSS combination of the ground motions. The results for a machine with natural frequency of 1.5 Hz are as follows: Loading X ground motion Y ground motion Z ground motion SRSS combination of X, Y and Z ground motions Maximum absolute acceleration (ft/sec2) 22.9194 16.2062 0.0510 28.0703
These results can be confirmed by adding a spring element to the model at point A, with natural frequency of 1.5 Hz and damping ratio of 2%, vibrating in the X direction. When the augmented model is subjected to the same ground motions, the maximum absolute accelerations of the spring node are the same as in the above table. User hints $ When it is not required to combine the results for different ground motion directions, then each ADINA analysis can be loaded into a separate ADINA-PLOT database. It is necessary to load the ADINA analyses into the same ADINA-PLOT database only when the results for different ground motion directions are to be combined using ADINA-PLOT. $ When combining the ADINA analysis into a single database, it is important that the number of time functions in each analysis be the same. However, the time functions used within each analysis can be different. In this case, a dummy function is added to the ADINA analysis for ground motion in the Z direction.
156
A.123 Floor response spectrum analysis of a building with sloped roof, time history load
Z t Z 12 ft 12 ft 12 ft A A t t
12 ft X 25 ft Y Y X 30 X Vertical columns: E = 30 106 psi n = 0.3 A = 19.70 in 2 Irr = 360.4 in 4 Iss = 271.8 in4 Itt = 88.6 in 4 r = 0.28 lbm/in3 (r along beam axis, s parallel to X) Other girders: E = 30 106 psi n = 0.3 A = 10.22 in 2 Irr = 154.3 in 4 Iss = 145.8 in 4 Itt = 8.5 in4 r = 0.28 lbm/in3 (r along beam axis, t as indicated in figure) 25 ft Y
Figure A.123(a)
157
fx1 f x2
fz1
1.5
Figure A.123(b)
158
159
Z Y a X
a = 0.254 m c = 0.635 m Material properties E = 2.07 1011 N/m2 n = 0.3 Tension s = 6.89 10 5 N/m 2 a
f X
Figure A.124
160
161
162
B.1
Objective To verify the geometric and material nonlinear behavior of the truss element. Physical problem A three-member lattice structure is subjected to a concentrated load as shown in Fig. B.1. The material of the structure is assumed to be elastic-perfectly-plastic. The load is applied in 5 equal increments. Finite element model Each member is represented by a 2-node elastic-plastic truss element. The large displacement, small strain kinematic formulation is used with stiffness reformation and BFGS equilibrium iterations at every step. Solution results The ADINA numerical results obtained are as shown in the table below. Due to symmetry, the force in truss 1 is the same as in truss 3. Time 1 2 3 4 5 * Plastic Truss 1 Force (kips) 26.12 52.23 78.35 100.00* 100.00* Truss 2 Force (kips) 52.23 104.45 156.66 215.17 300.00*
The sum of the internal truss forces in the Y-direction at time 5 is, since the displacement in the Y-direction at point A is calculated to be 0.327614 inch,
2 100
5.327614 5 + 5.3276142
2
163
User hints $ The large displacement, small strain kinematic formulation must be used for a physically realistic modeling of the structure. If a small displacement, small strain kinematic formulation is employed, the structure is predicted to become unstable once truss 1 and truss 3 become plastic since at that state there is no stiffness in the X-translational degree-of-freedom at point A. Using the large displacement, small strain kinematic formulation, the analysis proceeds beyond this load point, because the forces in all the members provide a geometric stiffness to the structural model. $ The truss element is formulated for small strain conditions only. Therefore, the area of the truss element is assumed to remain constant during the analysis.
Y P A
a
sy
3sy X
sY
a
2
Bar areas = 1.0 ft a = 5.0 ft 5 2 E = 2.0 10 kips/ft ET = 0.0 2 s y = 100.0 kips/ft Py = 441.42 kips (limit load considering small displacements only) P = 1.01 Py = 445.8342 kips
Figure B.1
164
B.2
Objective To verify the dynamic behavior of the truss element in large displacement analysis. Physical problem The simple pendulum shown in Fig. B.2 is released from a horizontal position at time t = 0. The response for several periods of oscillation is to be determined. Finite element model The pendulum is idealized as a truss element (large displacement, small strain formulation) with a concentrated mass at its free end. The trapezoidal rule is employed to obtain the step-by-step response with a step size of 0.025 seconds. Stiffness reformation and BFGS equilibrium iterations are performed at each step of the solution. Solution results The maximum axial force in the truss is predicted by ADINA to be 293.78 N which can be compared to the theoretical value 3Mg = 294.00 N . User hints $ It is important in a dynamic analysis to have tight enough tolerances. If, for example, in this analysis ETOL is set to 0.001 and RTOL is not used, then considerable energy is lost in the analysis, see [1] p. 827. $ Note that the load vector input by time functions at the start of the solution (t = 0) is not used in ADINA. In this example, an initial acceleration is therefore prescribed at time t = 0. Reference [1] Bathe, K.J., Finite Element Procedures, Prentice-Hall, 1996.
165
Y
l
X q L M g
L = 3.0443 m 5 EA = 1.0 10 N
166
B.5
Objective To verify the 2-D plane stress and isoparametric beam elements for materialnonlinear-only analysis using a creep material model. Physical problem The cantilever beam shown in Fig. B.5 is subjected to a constant bending moment at the tip. The material of the beam is assumed to obey the uniaxial creep law
eC = 1016 3t m/m
where is measured in KPa and t in hours. Finite element model Two separate models of the cantilever are included, one using a plane stress element and the other using an isoparametric beam element. For the analysis the integration parameter = 1 and three time steps with t = 200 are used to reach the stationary state. The ADINA restart option is employed to compute the creep response starting from the elastic solution. Full Newton iterations without line searches are used for the equilibrium iterations. Solution results The numerically obtained results for the plane stress element are compared with the analytical solution [1]. The differences in the results are due to the relatively small number of integration points in the z-direction used in the finite element analysis. Similar results are obtained for the iso-beam element. User hint $ The solution algorithm displays solution stability even with large time steps. However, the time step size has to be small enough to obtain an accurate solution. Reference [1] Penny, R.K. and Mariott, D.L., Design for Creep, McGraw-Hill, London, 1971.
167
Z 4.0 m
0.4 m
0.4 m
Figure B.5
168
B.6
Objective To verify the dynamic behavior of the beam element in elastic-plastic analysis. Physical problem The cantilever pipe analyzed in ref. [1] and shown in Fig. B.6 is considered. The pipe is initially at rest and is suddenly subjected to a constant tip load, causing it to impinge onto a restraint. Finite element model The pipe is modeled using six Hermitian beam elements and the pipe material is assumed to be elastic-perfectly-plastic as shown in Fig. B.6. The restraint, under the assumption that it is subjected to a monotonically increasing compressive load, is represented by a nonlinear elastic truss element with the stress-strain relation given in Fig. B.6. The truss element contributes no stiffness until a strain of -0.125 (corresponding to a gap of 3 in.) is reached. For the step-by-step direct integration, the trapezoidal rule (Newmark method) is used and the response is calculated using the BFGS iteration for seven time steps with a time step of 0.002 sec. A lumped mass matrix assumption is employed in the analysis. Solution results The pipe impinges onto the restraint at time 0.008 sec. User hints $ The characteristics of this problem are that the pipe is very flexible and impinges suddenly onto a relatively stiff stop, resulting into a high nonlinearity. The BFGS equilibrium iterations are very effective for this analysis, and converge with even a relatively large time step. $ In this analysis only 7 time steps are used for the dynamic response prediction. Comparing the response predictions to the response calculated with a smaller time step [1,2], it is noted that the solution given is sufficiently accurate.
169
$ The restraint can also be modeled using the gap option in the truss element. The gap element can only resist compression. Note, however, that the gap element is elasto-plastic whereas the restraint model in this example is nonlinear elastic. References [1] Ma, S.N. and Bathe, K.J., "On Finite Element Analysis of Pipe Whip Problems," J. Nuclear Engineering and Design, Vol. 37, pp. 413-430, 1976. [2] Bathe, K.J. and Cimento, A.P., "Some Practical Procedures for the Solution of Nonlinear Finite Element Equations," J. Comp. Meth. App. Mech. Eng., Vol. 22, pp. 59-85, 1980.
a diameter d restraint Do = 30.0 in Di = 27.75 in L = 360.0 in Pipe material: s sy E e Restraint material: s -.125 E e sy a = 3.0 in b = 21.0 in d = 5.75 in b
Di
Do
P = 6.57 10 lb at time 0
slug/in
Figure B.6
170
B.7
Objective To verify the use of mode superposition in nonlinear analysis. Physical problem Same as in Example B.6. Finite element model Same as in Example B.6 except that mode superposition is used to obtain the dynamic response. Two modes are employed in the mode superposition analysis and modified Newton equilibrium iterations are performed in the response calculation. The number of time steps and the step size are the same as in Example B.6. Solution results The results are similar to those obtained in Example B.6. User hints $ The mode superposition solution is very effective when only a few modes need to be used to represent the response to sufficient accuracy. The pipe whip example solved here is such a case. $ Note that the mode shapes of the linear structure (corresponding to time 0) are used in the mode superposition solution,
0 2 K 1 = 1 M 1 , 0
K 2 = 2 M 2 2
2
dependent coefficients evaluated in the solution. In essence, therefore, a twodegree-of-freedom finite element model is considered with the displacement assumptions 1 and 2 but all nonlinearities in the pipe and the stop are included in the analysis.
171
B.8
Uniaxial cycling loading solution for von Mises plasticity, isotropic bilinear hardening
Objective To verify the finite element solution when a bilinear von Mises elastic-plastic material with isotropic hardening is subjected to uniaxial cycling loading conditions. Physical problem The bar shown in Fig. B.8 is subjected to a cyclic load. Finite element model Six finite element models are used, each employing one of the following element types: truss, isoparametric beam, pipe, plane stress, 3-D solid and shell. Solution results The numerical results are identical to the analytical solution. The same response is obtained using any of the element types listed. User hints $ The results are exact for any number of time steps used in each of the three loading intervals since the loading is radial.
A = 0.01 m 2 L = 1.0 m 11 E = 2.0 10 N/m2 sy = 4.0 10 8 N/m 2 ET = 2.0 1010 N/m 2
L P s sy E e ET A
Figure B.8
172
B9. Uniaxial cycling loading solution for von Mises plasticity, bilinear kinematic hardening with material rupture
B.9
Uniaxial cycling loading solution for von Mises plasticity, bilinear kinematic hardening with material rupture
Objective To verify the finite element solution when a von Mises elastic-plastic material with kinematic hardening is subjected to uniaxial cycling loading conditions. Also, to demonstrate the ADINA solution when rupture of the material occurs. Physical problem Same as Example B.8, except that the material can rupture at eAP = 0.007 . Finite element model Six finite element models are used, each employing one of the following element types: truss, iso-beam, pipe, plane stress, 3-D solid and shell. In the models employing pipe, plane stress and shell elements, two elements of the specified type are used, one without rupture (cross-sectional area of 0.009) and one with rupture (cross-sectional area of 0.001). These elements are connected to the same nodes, so that the effective area of the assemblage before rupture is 0.01. In the other models, a single element of the specified type is used without rupture. Solution results Before the rupture of the material is reached, the solution results are the same for each element. These results are exactly equal to the analytical solution. After rupture occurs (at step 14) the elements coupled to those with rupture continue to carry the load and the stresses in these elements are therefore increased. User hint $ When using the rupture condition in the material model, it is important to use small load increments (as in this example) because large load increments can cause large unbalanced loads during the equilibrium iterations within a load step. In this case the iteration path can lead to an unrealistically early occurrence of the rupture condition.
173
B.10
Objective To verify the large displacement behavior of the truss element and demonstrate the modeling of cables. Physical problem A cable is stretched between a ground anchor point and a tower attach point, see Fig. B.10. The cable has a cluster of 6 insulators attached to a point in the middle portion and 3 insulators attached to the lower portion. The weight of each insulator is 510 lbs. The initial prestress of the cable is 7520 lbs and the self weight is 0.106667 lbs/in. Finite element model The finite element model consists of twelve 2-node truss elements. The insulators are attached as concentrated masses. The material model for the truss element is linear elastic and the large displacement, small strain kinematic formulation is used. In the numerical solution, the BFGS method of equilibrium iteration is used with stiffness reformation for every load step. The gravity loading is applied in 5 equal load steps. Solution results We note that the cable stiffens as the gravity load is increased. For an analytical verification we select the following results as typical: Node 9 6 10 Element 7 8 Displacements (inch) y z 215.745 -248.196 242.152 -275.898 199.365 -226.345 Truss internal force (lbs) 22843.5 24975.0
Using the above displacement solution the following internal truss forces can be calculated theoretically:
174
175
B.15
Objective To verify the creep material model when 2-dimensional axisymmetric and 3dimensional elements are employed. Physical problem Figure B.15(a) shows the thick-walled cylinder which is to be analyzed for cyclic internal pressure. Plane strain conditions are assumed. The material of the cylinder is assumed to obey a uniaxial creep law of the form
176
P r1 r2 Y
Plane strain conditions E = 2.0 10 7 lb/in2 n = 0.3 . -18 e C = 6.4 10 . s 4 4 . t in/in Temperature 100 F r1 = 0.16 in r2 = 0.25 in
-365
Figure B.15
177
B.16
Objective To verify the shell element for large displacement analysis. Physical problem A simply supported square plate is subjected to a uniform pressure load, see Fig. B.16. The plate material is linear elastic. Finite element model Because of symmetry conditions only 1/4th of the plate needs to be modeled. One 16node shell element is used. Constraint equations are used to model uniform in-plane edge displacements. In the numerical solution, the loading is applied in 22 load steps up to the final load parameter K = 500. For every load step the stiffness is reformed and the BFGS method of equilibrium iteration is employed. Solution results The computed displacement response agrees very closely with the solution given by Levy [1]. User hint $ The large displacement response of shells generally involves severe nonlinearities because the structure, initially only in bending action, is subjected to increasing membrane action with increasing load. This results in a stiffening effect. The BFGS method is generally quite effective for these types of problems. If difficulties are encountered in the convergence of the equilibrium iterations, then the more powerful full Newton iteration method with line searches can be employed. Reference [1] Levy, S., "Bending of Rectangular Plates with Large Deflections," Technical Notes, National Advisory Committee for Aeronautics, No. 846, 1942.
178
Region discretized
a E n h a q 10 psi 0.1 0.12 in (plate thickness) 24 in uniform applied pressure per unit area qa4 K = Eh 4 All edges are simply supported
Figure B.16
= = = = =
179
B.17
Objective To verify the thermo-elastic material model when used with 3-D solid elements. Physical problem A cantilever beam is subjected to a linearly varying temperature distribution in the Zdirection, see Fig. B.17. No other loads are applied to the beam. Finite element model Since the problem is anti-symmetric with respect to the neutral axis of the beam, only the portion above the neutral axis is modeled. Three 20-node 3-D solid elements are used. Solution results The ADINA analysis is carried out for material nonlinearities only. The analytical solution using the theory given, for example, in [1] is w = -0.0054 inch and excellent agreement can be observed. User hints $ The thermo-elastic material model allows Young's modulus E, Poisson's ratio v and the mean coefficient of thermal expansion to vary as a function of temperature. The material model is considered nonlinear since the resulting stiffness matrix, in general, is dependent on the applied temperature loading. $ Since the theoretical displacement solution varies quadratically in the X- and Ydirections, the 20-node element can describe the displacement field exactly. Therefore, since the beam is free to bend in both the X- and Y-directions, the stresses are zero everywhere. Reference [1] Boley, B.A. and Weiner, J.H., Theory of Thermal Stresses, John Wiley and Sons, pp. 307-314, 1960.
180
1 in
neutral axis
1 in
Figure B.17
181
B.18
Objective To demonstrate the use of the curve description model in a simplified analysis of an underground opening. Physical problem A tunnel through a rock mass, see Fig. B.18, is considered. The walls of the tunnel are reinforced with rock bolts. The rock material is assumed to have constant Young's modulus and Poisson's ratio. Finite element model The rock is modeled using 8-node plane strain elements and the rock bolts by 2-node truss elements. The pressure due to the overburdened rock material is simulated by a pressure load. The curve description model is employed to describe the rock material. The tensile failure option is used with material density = 0.12 kip/ft3 and shear reduction factor = 0.00099. Both the tensile stress and the shear stress are therefore reduced to zero for a crack, i.e. when any principal tensile stress exceeds the in-situ gravity stress at an integration point. A materially-nonlinear-only analysis is carried out in eleven steps using full Newton iterations with line searches. The final overburden pressure reached in the analysis is 30 kip/ft2. Solution results Significant cracking occurs in the rock underneath the tunnel at the final overburden pressure. User hints $ When cracks form in a finite element model using the curve description material model the stiffness generally changes drastically. Depending on the in-situ gravity pressure the change in stresses associated with the formation of a crack may also be significant. These effects generally result in a considerable redistribution of stresses. It is, therefore important to take small enough load steps whenever cracks are forming in order to avoid solution difficulties.
182
$ Note that the in-situ gravity pressure pi at the nodes is calculated using pi = z i where zi is the nodal z-coordinate. The stress corresponding to this gravity pressure is not included in the calculated stresses which are output. The gravity pressure is, however, taken into account when the bulk and shear moduli are determined and in the criteria for cracking.
Z Y P kip/ft
2
z = -60 ft
16 ft Steel bolt 4 ft E = 4.23 10 4 kip/ft 2 n = 0.0 3 g = 120 lb/ft Esteel = 4.32 10 kip/ft 2 Asteel = 0.0021 ft 2 18 ft
7
6 ft
12 ft
32 ft
4 ft
12 ft
12 ft
Figure B.18
183
B.19
Objective To verify the concrete material model in plane stress analysis, and to demonstrate the modeling of reinforced concrete. Physical problem A simply supported reinforced concrete beam subjected to two symmetrically applied concentrated loads as shown in Fig. B.19 is considered. The steel reinforcement area Ast = 0.62 in2 is used in this analysis. Finite element model Using symmetry conditions only one-half of the structure needs to be considered. Fifteen 8-node plane stress elements are used to model the concrete of the beam and five 3-node truss elements are used to model the reinforcement. The material of the steel bars is modeled as elastic-plastic with strain hardening. The solution is obtained in 35 equal steps using Newton iterations with line searches. Solution results The numerical solution is compared with the response predicted by Suidan and Schnobrich [1] and reasonable agreement is observed. User hints $ The solution of concrete problems can be difficult, due to the sudden nonlinearities that can take place as a result of cracking and crushing of the material. The overall structural nonlinearities are more pronounced when only small amounts of steel reinforcements are used in the structure. Significant stress redistributions can take place in the concrete for each increment in load applied (note that depending on the structure, the stress redistribution may well propagate the cracking of the material and lead to total structural collapse). In this case small incremental load steps may be required due to difficulty in equilibrium iteration convergence.
184
Reference [1] Suidan, M. and Schnobrich, W., "Finite Element Analysis of Reinforced Concrete," ASCE, J. Struct. Div., Vol. 99, No. ST10, Oct. 1973.
P _ 2 50" Region discretized 36" P _ 2 50" 6" 12"
2 Ast = 2.00 in2 0.62 in
Figure B.19(a) Steel Young's modulus Poisson's ratio Initial yield stress Strain hardening modulus Density 30000 ksi 0.2 44 ksi 300 ksi 7.33910-4 lbf-s2/in4
1"
Concrete Uniaxial parameters: Initial tangent modulus Poisson's ratio Uniaxial cut-off tensile strength Uniaxial maximum compressive stress (SIGMAC) Compressive strain at SIGMAC Uniaxial ultimate compressive stress Uniaxial ultimate compressive strain Figure B.19(b), part 1
6100 ksi 0.2 0.458 ksi -3.74 ksi -0.002 in/in -3.225 ksi -0.003 in/in
185
Triaxial compressive failure curves: Principal stress ratios SP1(I) SP3(I,1) (at SP2=SP1) SP3(I,2) (at SP2=BETA$SP3) SP3(I,3) (at SP2=SP3) 1 0.00 1.00 1.25 1.20
Curve number, I 2 3 4 5 0.25 0.50 0.75 1.00 1.35 1.75 2.15 2.50 1.70 2.10 2.55 2.95 1.60 2.00 2.40 2.80
Other parameters: Stress ratio for failure surface input (BETA) Constant for definition of critical strains (C1) Constant for definition of critical strains (C2) Constant used to define the tensile strains corresponding to zero stress (XSI) Stiffness reduction factor Shear reduction factor Density Figure B.19(b), part 2
186
B.20
Objective To demonstrate dynamic analysis of a reinforced concrete model. Physical problem Same as in Example B.19 except that the reinforcement in the beam is increased to Ast = 2.0 in2 and that the concentrated load is applied as a step load at time t = 0+ with magnitude P = 13.5 kips. Finite element model The finite element model is the same as in Example B.19 except that the cross section area of the trusses modeling the reinforcement of the beam is increased to Ast = 2.0 in2. The dynamic response is evaluated using the trapezoidal rule (the Newmark method with = 0.25 and = 0.50) with a time step t = 0.001 sec. A lumped mass matrix is used in the analysis and equilibrium iterations using the full Newton method with line searches are employed at each time step. The ATS method is used to reduce the time step when convergence is not achieved. Solution results The beam vibrates about a steady-state configuration in which the lower half of the beam is in tension. Significant cracking occurs in the lower half of the beam. Most of the cracks remain open throughout the vibration, but some of the cracks near the neutral axis close when the beam moves to the top of its motion and reopen when the beam moves to the bottom of its motion. User hint $ It is generally important in a nonlinear dynamic analysis to iterate for dynamic equilibrium. If no iterations are performed errors may arise, which, in a dynamic analysis, accumulate.
187
B.21
Objective To verify the large displacement behavior of the isoparametric beam element and to demonstrate the restart option in static analysis. Physical problem A shallow arch is subjected to a concentrated apex load, as shown in Fig. B.21. Finite element model Using symmetry conditions, four 4-node iso-beam elements are used to model one half of the structure. The apex load is applied in sixteen equal load steps and the restart option is employed after the first eight steps. Stiffness reformations and BFGS equilibrium iterations are employed for each load step. Solution results Good agreement with ref. [1] is obtained for the predicted displacements. Reference [1] Mallet, R.H. and Berke, L., "Automated Method for the Large Deflection and Instability Analysis of Three-Dimensional Truss and Frame Assemblies," AFFDLTR-66-102, 1966.
h b R
188
B.22
Objective To demonstrate the capability of performing nonlinear dynamic analysis using substructures. Physical problem The eight-story building shown in Fig. B.22 is analyzed for its dynamic response when subjected to an impact load at the top. Finite element model The division of the building into substructures is indicated in Fig. B.22. The building is modeled with beam elements in the master structure and the substructures. The foundation is modeled with elastic truss elements in the vertical direction and with elastic-plastic truss elements in the horizontal direction. The dynamic response is evaluated using the trapezoidal rule (the Newmark method with = 0.25 and = 0.50) with a time step t = 0.02 sec. A lumped mass matrix discretization is used in the analysis. Stiffness reformation and BFGS equilibrium iterations are employed in every solution time step. Solution results The same results are obtained when the building is modeled without substructures.
189
P 20
Substructure 1, reuse 2
96 ft
Substructure 1, reuse 1
15 ft
Figure B.22
190
B.23
Objective To verify the Drucker-Prager material model when used for the axisymmetric element. Physical problem A uniaxial strain test for McCormick Ranch Sand is considered, see Fig. B.23. Finite element model Four 4-node axisymmetric elements are used to model the sand specimen. The top nodes in the model are subjected to prescribed displacements. Twenty prescribed displacement steps are used with stiffness reformation and BFGS equilibrium iterations at every step. Solution results The calculated solution is in reasonable agreement with the experimental data given in [1]. User hints $ For the present example one 4-node element is also satisfactory, since there is no radial displacement variation and only a linear displacement variation in the vertical (Z) direction. Reference [1] DiMaggio, F.L. and Sandler, I.S., "Material Model for Granular Soils," J. Eng. Mech. Div. ASCE 97 (EMS), pp. 935-949, 1971.
191
Prescribed displacement
Region discretized
D E = 100.0 ksi n = 0.25 Yield function parameter a = 0.05 Yield function parameter k = 0.1 ksi Cap hardening parameter W = -0.066 Cap hardening parameter D = -0.78 ksi -1 Tension cut-off limit T = 0.01 a Initial cap position 0I 1 = 0.0 ksi D = 1.0 in H = 0.5 in
Figure B.23
192
B28. Elastic-plastic bending of a pipe, with axial loading and internal pressure
B.28
Objective To verify the elastic-plastic response of the pipe element subjected to bending and axial loading and to internal pressure. Physical problem A straight, simply supported pipe is loaded laterally and axially as shown in Fig. B.28. The axial loading corresponds to a given thermal loading [1], [2]. The pipe is also subjected to a constant internal pressure. Finite element model Eight pipe beam-type elements are used to model the entire structure. The multilinear elastic-plastic material model with isotropic hardening conditions is used. Full Newton iterations with line searches are employed. To take into account the bending deformations due to the axial force Fy, a large displacement analysis is performed. The lateral force is increased according to the following table in order to follow the distinct experimental points [2]. Lateral load P 4 5 6 357 376 401
Step P (kips)
1 101
2 198
3 299
7 425
8 435
9 453
10 460
Solution results The moment-curvature relation is computed for the test section of the pipe and compared with experimental results. The curvature is computed as = B/ LAB where B is the rotation at point B and LAB is the distance between points A and B. Very good agreement between the computed and experimental results is obtained.
193
User hint $ The solution effort is much affected by the number of integration stations used and it should be noted that the pipe beam-type element is the most effective since it requires only one integration point along the pipe length. References [1] [2] Bounkamp, J.G. and Stephen, R.M., "Large Diameter Pipe Under Combined Loading," Transp. Eng. J. ASCE, V. 99, No. TE3, pp. 521-536, 1973. Price, P. St. J. and Kim, H.O., Mechanical Stress Analysis of Buried Pipeline, Northern Engineering Services Company, Ltd., Calgary, 1975.
93" P _ 4 24" 6" P _ 4 132" 6" 24" P _ 4 P _ 4 93"
B Fy
Internal pressure = 917 psi 6 Axial force Fy = 2.52 10 lb Lateral force P = 1000 lb
Figure B.28
194
B.32
Excavation of a tunnel
Objective To verify the 2-D plane strain element when employing the birth/death option in analyzing the construction of a tunnel. Physical problem The construction of a tunnel shown in Fig. B.32 requires the creation of the tunnel opening by removal of material and the erection of props at the center line. The displacements and stresses during the different phases of the construction are to be determined. Finite element model Eight 4-node plane strain elements are used to model one half of the underground region. The element death option is active for the element modeling the tunnel region, and the element birth option is active for the truss element modeling the props. The following step-by-step solution is performed to simulate the different phases of construction: Step 1: Step 2: All elements except the truss are active and a uniform loading of q = 80 kips/ft2 is applied at the top surface. The element modeling the tunnel region becomes inactive to simulate the creation of the opening. The truss remains inactive and the applied loading is the same as in step 1. The truss element becomes active to simulate the addition of the props. The applied loading is the same as in step 1. The applied loading is increased to 120 kips/ft2 and the response of the completed tunnel is obtained.
Step 3: Step 4:
Stiffness reformation is performed at every load step. Solution results No comparisons of these finite element results with experimental or other computed results are given, but the solution reported here satisfies some qualities that should be observed; namely, a redistribution of stresses due to the excavation of the tunnel (step 2), no change in stresses due to the addition of the props only (step 3), and the tensile
195
stress developed in the roof due to the additional loading is relatively small when the props are active (step 4). User hints $ Note that the only nonlinearity in this solution is due to the removal/addition of the elements representing the tunnel and the props. Hence, since in this example a new stiffness matrix is calculated at each step corresponding to the conditions at the end of each step, this matrix does not change during the solution step and a linear system is, therefore, considered in each step. Therefore the equilibrium iterations converge immediately in this solution. $ Note that the truss element representing the props has as its initial length the distance between the element end nodes in the deformed configuration at time 2.0. Hence, the element is born stress free. This holds for all elements that are born in a solution.
a Prop Opening r
2r
a = 20.0 ft r = 24.0 ft
196
B.33
Objective To verify the nonlinear elastic truss element when used to simulate cable failure in a cable/frame structure. Physical problem The cable/frame structure shown in Fig. B.33 is subjected to a concentrated downward load at the midspan. The cable is assumed to break at the joint when a strain of 3% is reached. The behavior of the structure before and after the failure of the cable is to be analyzed. Finite element model The frame is modeled with four linear elastic 2-node Hermitian beam elements and the cable is modeled using a nonlinear 2-node truss element with the stress-strain relationship shown in Fig. B.33. The load is applied in ten equal load steps. Stiffness reformation and equilibrium iterations using the BFGS method are used at each solution step in the analysis. Solution results The cable breaks at an applied load of 1100 kips. The structure becomes much softer at this load and then begins to stiffen for higher loads. User hints $ The nonlinear elastic material can be used to model material failures as shown in this example. Of course, general stress/strain reversals are then not possible. The tension failure strain may only be reached from below. $ Note that since the magnitude of the maximum deflection after the cable failure is large, in practice a large displacement analysis including elastoplasticity is more appropriate to predict the structural behavior.
197
a = 20 ft b = 12 ft Y b
X P a P = 1500 kips Frame: 6 E = 4.3210 kips/ft 2 n = 0.28 -2 4 I = 2.77810 ft A = 0.3333 ft 2 Cable: 6 2 E = 4.3210 kips/ft 2 -3 A = 6.944 10 ft a
E = 0
0.03
Figure B.33
198
B.34
Objective To verify the 2-D plane stress element when employed in an elastic-plastic analysis. Physical problem The cantilever plate shown in Fig. B.34 is subjected to a bending moment at its free end. The bending moment is increased until most of the plate cross-section is plastic and then unloaded. The residual stresses in the plate are to be determined. Finite element model The cantilever plate is modeled using one 8-node plane stress element which is subjected to two equal and opposite forces at its end. The material of the plate is assumed to obey an elastic-plastic isotropic hardening material law. Three solution steps are employed to calculate the finite element response and the variation of the applied loading is shown in Fig. B.34. In the analysis, the full Newton equilibrium iteration method with line searches is employed and the finite element matrix is calculated using 4 4 Gauss integration. Solution results The residual stress field is as follows: Residual stress yy (N/mm2) Beam theory ADINA -133.6 -155.7 225.9 210.4
$ Note that a high enough integration order in the thickness direction of the cantilever plate is required to predict the spread of the plasticity through the thickness.
199
Z thickness 20 mm
h L
Time
Figure B.34
200
B.35
Objective To demonstrate the use of the concrete material model in a two-dimensional analysis. Physical problem The steel-lined concrete cylinder shown in Fig. B.35 is subjected to a uniform internal pressure. The pressure at which the concrete section is completely cracked is to be determined. Finite element model The cylinder is modeled using six 8-node axisymmetric elements. The solution response is evaluated in eighteen load steps using the load curve shown in Fig. B.35 and BFGS equilibrium iterations. Solution results In the finite element solution, cracks in the hoop direction first occur at the integration points close to the inner surface of the concrete section at a pressure of 700 psi. The cracks propagate in the radial direction during the load history and the concrete section is completely cracked at a pressure of 1300 psi. The steel lining remains elastic throughout the entire solution.
Steel Steel Concrete a = b = h c = d = h = d
in in in in in
a Time
Figure B.35(a)
201
Material Parameters Steel Young's modulus Poisson's ratio Initial yield stress Strain hardening modulus Concrete Uniaxial parameters: Initial tangent modulus Poisson's ratio Uniaxial cut-off tensile strength Uniaxial maximum compressive stress (SIGMAC) Compressive strain at SIGMAC Uniaxial ultimate compressive stress Uniaxial ultimate compressive strain Triaxial compressive failure curves: Principal stress ratios SP1(I) SP3(I,1) (at SP2=SP1) SP3(I,2) (at SP2=BETA$SP3) SP3(I,3) (at SP2=SP3) 1 0.00 1.00 1.30 1.25 2 0.25 1.40 1.50 1.45 30,000 ksi 0.2 44 ksi 300 ksi
6100 ksi 0.2 0.458 ksi -3.74 ksi -0.002 in/in -3.225 ksi -0.003 in/in
Curve number, I 3 4 5 0.50 0.75 1.00 1.70 2.20 2.50 2.00 2.30 2.70 1.95 2.25 2.65
Other control parameters: Stress ratio for failure surface input (BETA) Constant for definition of critical strains (C1) Constant for definition of critical strains (C2) Constant used to define the tensile strains corresponding to zero stress (XSI) Stiffness reduction factor Shear reduction factor Figure B.35(b)
202
B.36
Objective To verify the geometric and materially nonlinear behavior of the axisymmetric element when subjected to uniform pressure in dynamic analysis. Physical problem The spherical cap shown in Fig. B.36 is at rest and is suddenly subjected to a constant and uniform pressure over its top surface. The same problem is also considered in [1] and [2]. Finite element model Because of symmetry conditions, only one radian of the cap is modeled using ten 8node axisymmetric elements. A skew coordinate system is used to model the boundary conditions at the support as shown in Fig. B.36. The solution response is calculated using the Newmark method ( = 0.55, = 0.276) with 50 time steps and a step size of 10-5 sec. The equilibrium iterations are performed using the full Newton method without line searches. The large displacement, small strain kinematic assumptions are used and the material is described as elastic-plastic with isotropic hardening. The pressure load is modeled as deformation-dependent. Solution results Good agreement with the solution in [2] is observed. User hint $ In nonlinear dynamic analysis it is generally important to iterate in each time step, although only relatively few iterations may be required for convergence. References [1] Ishizaki, T. and Bathe, K.J., "On Finite Element Large Displacement and ElasticPlastic Dynamic Analysis of Shell Structures," J. Comp. & Struct., Vol. 12, pp. 309318, 1980. Nagarajan, S. and Popov, E.P., "Nonlinear Dynamic Analysis of Axisymmetric Shells," Int. J. Num. Meth. Eng., Vol. 9, pp. 535-550, 1975.
[2]
203
a P b
w0
P (psi)
600
Time a
a
a = 26.67 h = 0.41 in R = 22.27 in E = 1.05 10 psi n = 0.3 4 sy = 2.40 10 psi ET = 2.10 10 psi -4 r = 2.54 10 lbf-s2/in 4
5 7
Figure B.36
204
205
Solution results From equilibrium considerations, the theoretical solution is Tim e 0.0 1.0 2.0 3.0 Time 0.0 1.0 2.0 3.0 End displacement Bar 1 Strain 0.0 0.041667 0.166667 0.211538 Stress 0.0 50.0 200.0 253.846 Bar 3 Strain Inactive Inactive 0.0 0.038462 0.0 46.1538 Stress
TBIRTH = 2.2 s, birth option, el. no. 3 TDEATH = 1.8 s, death option, el. no. 2
Figure B.37
206
B.38
Objective To demonstrate the capability of performing a temperature and thermal stress analysis using the ADINA-T and ADINA programs. Physical problem Same as in Example B.17. Finite element model The cantilever beam is modeled using six 20-node 3-D solid elements. In the temperature analysis using ADINA-T (Example T.5) the temperature distribution through the thickness is calculated and the resulting nodal point temperatures are written to the temperature file. A thermal stress analysis is performed using the ADINA program by reading the nodal point temperature data from the temperature file. In the ADINA analysis, a thermo-elastic material model is used with the same data as in Example B.17. Solution results Practically the same numerical solution is obtained by ADINA for this finite element model as for the model described in Example B.17.
207
B.39
Cantilever beam in creep (2-D plane stress, iso-beam), variable time step
Objective To demonstrate creep analysis with variable time step in ADINA by using the restart option. Physical problem Same as in Example B.5. Finite element model Same as in Example B.5, except that the time steps chosen are as follows: Number of steps 1 5 5 3 t 0.001 20.0 40.0 100.0 Comments The elastic solution is computed.
The restart option is used to change the time step. Solution results Practically the same numerical results are obtained in this analysis as in Example B.5. The numerical solutions for bending stress at time t = 600 hrs compare with the analytical stationary state solution as follows: distance from the neutral axis (m) 2-D element Iso-beam element 0.2000 0.2000 bending stress (KPa) analytical solution 4375. 4375. ADINA 4506. 4559.
(The results for the 2-D element are obtained by extrapolation from the integration point results.)
208
B.40
Objective To perform a postbuckling analysis of an arch structure using the load- displacement control method. Physical problem An assemblage of two identical bars is subjected to an apex load, see Fig. B.40. Finite element model From symmetry, the arch structure is modeled using one 2-node truss element. Solution results The analytical solution for the load-deflection relationship is ([1], p 495):
1 P = 2 EA 1 + 2 1 2 sin15o + L L
o sin15 L
An excellent comparison with the analytical solution is observed. User hint $ If a static solution is attempted by using either the modified Newton iteration, full Newton iteration or BFGS method, the displacement response can be traced only to = 1.1. Beyond this region the program execution is terminated because the arch stiffness matrix is not positive definite. Reference [1] Bathe, K.J., Finite Element Procedures, Prentice-Hall, 1996.
209
15
210
B.41
Objective To verify the load-displacement control method for post collapse analysis of structures. Physical problem The bar structure shown in Fig. B.41 is considered. A concentrated force, P, is applied to the structure at its left end and the structural large displacement response is analyzed. Finite element model The bar structure is modeled using three 2-node truss elements and the large displacement, small strain kinematic formulation is employed. Solution results An excellent comparison with the solution given in [1] is observed. Note that the displacements at points 1 and 2 are equal when the applied load is zero. Reference [1] G. Powell and J. Simons, "Improved iteration strategy for nonlinear structural analysis", TRR. Lab., Rep. 1049, 1982.
E = 3 106 1 Area = _ 6 A P 5
DA
Area = 1.0 5
DB
5
Figure B.41
211
B.42
Objective To analyze the large displacement response of a shell structure using the loaddisplacement control method. Physical problem A spherical shell is subjected to an apex load, see Fig. B.42. In an X-Y plane view, the structure is square (2a 2a) in shape. All four edges are hinged. Finite element model From symmetry, only one quarter of the shell structure is discretized using nine 4node shell elements. Symmetrical boundary conditions are applied at nodes along sides AB and AD using the six degrees of freedom per node option. The five shell degrees of freedom per node option is used for all other nodes. Also, for nodes along sides BC and CD, all displacement degrees of freedom are deleted to model the hinge boundary condition. Solution results The ADINA numerical results for the applied force apex displacement relationship are in good agreement with the solution presented in refs. [1] and [2]. User hints $ Note that one 16-node isoparametric shell element can also be used instead of nine 4-node shell elements to discretize one quarter of the shell structure. The stiffness matrix obtained has a larger bandwidth than the stiffness matrix for the 4-node shell element model. The 16-node shell element model also gives answers which are in good agreement with the response given in refs. [1] and [2]. $ Due to symmetry, the analysis can also be performed by only considering 1/8th of the total shell structure (e.g., region ABC in Fig. B.42). References [1] Leicester, R.H., "Finite Deformations of Shallow Shells," Proceedings of American Society of Civil Engineers, Vol. 94, (EM6), pp. 1409-1423, 1968.
212
[2]
Horrigmoe, G., "Finite Element Instability Analysis of Free-Form Shells," Report 772, Division of Structural Mechanics, The Norwegian Institute of Technology, University of Trondheim, Norway, 1977.
Z X Y
Apex load P B
region discretized
C A D 2a
2a
Radius = 2540 Thickness = 99.45 a = 784.90 E = 68.95 n = 0.30
Figure B.42
213
B.43
Objective To evaluate the linearized buckling load and corresponding mode shape for a stiffened plate subjected to longitudinal compression. Physical problem A simply supported plate stiffened by a longitudinal rib is considered, see Fig. B.43. The rib is located symmetrically with respect to the plate. The plate-stiffener assembly is subjected to a longitudinal compressive total force P = 29160. This force is increased until buckling of the plate occurs. Finite element model The buckling mode is expected to be symmetric [1]. Thus, one quarter of the plate is modeled using nine 4-node shell elements and three 2-node isoparametric beam elements. For shell nodes along the symmetry axes AB and AD, the six degrees of freedom per node option is employed. At all other nodes, five shell degrees of freedom are considered per node. Symmetrical boundary conditions are applied along sides AB and AD. Along sides BC and CD, the z-displacement degree of freedom is deleted. The magnitude of the applied load is, from symmetry conditions, half of the total load. Solution results The ADINA program performs a linearized buckling analysis as follows: $ For the given total load, P = 29160, the displacement response of the stiffened plate is calculated. The structure is in a state of uniaxial compression. $ Using the deformed geometry corresponding to the external load P = 29160, an eigenvalue problem is solved to obtain the critical load Pcrit = 36322. Hence, Pcrit cr = = 1257 cross-sectional area of the stiffened plate From ref. [1], the analytical solution is cr = 1220 .
214
User hints $ If the linearized buckling solution is performed with the plate subjected to a smaller total initial compressive load, P = 2916 nearly, the same buckling load is predicted, i.e., Pcrit = 36338. $ Since the plate geometry is flat, the linearized buckling analysis can also be performed by considering six degrees of freedom per shell node and deleting the zrotational degree of freedom at all nodes except those on side AD. $ If there is no a priori knowledge of the buckling mode shape, the full structure must be discretized to include the possibility of a non-symmetric buckling mode shape. $ To calculate the concentrated loads to be applied to the stiffened plate nodes, it is necessary to add the load needed for the beam to the loads needed for the unstiffened plate. Reference [1] Timoshenko, S.P. and Gere, J.M., Theory of Elastic Stability, 2nd edition, McGrawHill, 1961.
C D
L
A
0.50
Figure B.43
215
B.44
Objective To perform a large displacement analysis of a plate structure with initial imperfections using the load-displacement control method. Physical problem A simply supported stiffened plate subjected to a longitudinal compressive force is considered (see Example B.43). The initial imperfection in the stiffened plate geometry is chosen to be:
216
Solution results The numerical solution shows that the lateral displacement at the center of the plate increases rapidly when cr (cr = 1220.1, see Example B.43). User hints $ As the specified initial imperfection is small, the bending moments generated in the plate due to the applied compressive force are small and thus the plate is primarily in a state of uniaxial compression until > 0.80cr. $ If a large initial imperfection is specified, the lateral displacement at the center of the plate increases more gradually with increases in the applied compressive load. $ The shell element normal vectors are generated internally for the perfect geometry and subsequently modified automatically by ADINA to account for the initial imperfections.
217
B.45
Objective To evaluate the first and second in-plane buckling mode shapes and corresponding linearized buckling loads for a circular arch subjected to a uniform pressure loading. Physical problem A circular arch with fixed ends is considered, see Fig. B.45. The arch is subjected to a distributed load, p = 1.0/unit length. An analysis is performed to calculate estimates of the applied distributed load intensities corresponding to the first and second inplane buckling modes of the arch. Finite element model The first buckling mode is expected to be skew-symmetric [1]. The full arch is modeled using twenty 2-node isoparametric beam elements. All out-of-plane motions are suppressed. Solution results The ADINA program performs a linearized buckling analysis as follows: $ For the given distributed load, p = 1.0/unit length, the displacement response of the circular arch is calculated. $ Using the deformed geometry corresponding to the prescribed external loading, an eigenvalue problem is solved to obtain the following values of the linearized buckling distributed load intensities: Buckling mode number 1 2 Intensity of applied distributed load Theory [1] 83 ADINA 86 129
218
User hints $ If a linearized buckling analysis is performed for the circular arch subjected to uniform pressure and all buckling modes (i.e., in-plane and out-of-plane) are permitted, the lowest two modes correspond to: Buckling mode number Intensity of applied distributed load 1 (out-of-plane) 40 2 (in-plane) 85
$ Even though the arch structure has an axis of symmetry, the full arch is discretized because a mode shape of interest is skew-symmetric. Reference [1] Timoshenko, S.P. and Gere, J.M., Theory of Elastic Stability, 2nd edition, McGrawHill, 1961.
P H a
R a2 R ~ 2H = 10.0 _ h h h b
Figure B.45
219
B.46
Objective To use the option of linearized buckling analysis to obtain the lowest Euler buckling load for a column. Physical problem The column shown in Fig. B.46 is considered. Finite element model The column is modeled using two 2-node Hermitian beam elements. Solution results The theoretical Euler buckling load is Pcr = The following results are obtained: Pcr Theory 51.40 User hint $ To predict the higher buckling loads accurately, it is necessary to employ a finer finite element discretization. ADINA 51.70
2 EI . L2
220
Load P Cross-section b h
E L h b L Y
= = = =
Figure B.46
221
B.47
Objective To verify the 4-node shell element subjected to a temperature gradient through its thickness. Physical problem A circular simply supported plate with a uniform temperature gradient through its thickness is considered, see Fig. B.47. The temperature gradient is the same at all points within the plate. Finite element model Due to symmetry, only a quarter of the plate is discretized. Four degenerated 4-node shell elements are used to discretize the region A-O-B in Fig. B.47. At each shell d = 20.0 where is the node, the value of the temperature gradient specified is dx temperature. The reference temperature for the entire plate is REF = 0.0. Solution results The analytical solution for the central deflection W0 is given by [1]
W0 =
6R2 M T Et 3
3
222
Reference [1] Boley, B.A. and Weiner, J.H., Theory of Thermal Stresses, John Wiley and Sons, p. 402, 1960.
Region discretized
X Z
A O Y t
Figure B.47
223
B.48
Objective To verify the 2-node isoparametric beam element subjected to deformation dependent loading. Physical problem A cantilever is subjected to a distributed deformation dependent pressure loading, see Fig. B.48. The cantilever experiences large in-plane bending displacements due to the applied loading. Finite element model The cantilever is modeled using six 2-node isoparametric beam elements. The solution response is traced in two steps using full Newton iterations without line searches. Solution results The obtained numerical results are compared with the solution given in refs. [1] and [2] and good agreement is observed. User hints $ The load-deflection relationship of the model shows that the cantilever structure stiffens considerably with increasing tip displacement. Thus, full Newton iterations are best used in the solution of this problem. $ The auxiliary node must be located so that it always remains on the same side of each element in the structure throughout the analysis. Otherwise the applied pressure loading will switch direction in each element for which the auxiliary node switches sides. References [1] Bathe, K.J., Ramm, E. and Wilson, E.L., "Finite Element Formulations for Large Deformation Dynamic Analysis," Int. J. Numer. Meth. Eng., Vol. 9, pp. 353-386, 1973.
224
B48. Large displacement analysis of a cantilever subjected to deformation dependent pressure loading
[2]
Holden, J.T., "On the Finite Deflections of Thin Beams," Int. J. Solids Struct. Vol. 8, pp. 1051-1055, 1972.
P lb/in
Cross-section h
Figure B.48
225
B.49
Objective To verify the load-displacement control method for solution of materially nonlinear problems. Physical problem A thick-walled long cylinder is subjected to internal pressure, see Fig. B.49. The cylinder material is elastic-perfectly plastic. Finite element model Four 8-node axisymmetric elements are employed to discretize a unit length of the cylinder. The axial strain is zero throughout the cylinder. The elastic-plastic response is traced using the load-displacement control (LDC) algorithm. Solution results A good comparison between the numerical results and the solution given in ref. [1] is observed. User hint $ When using an equilibrium iteration method with prescribed load levels, very small load steps need to be used near the collapse load. However, the LDC algorithm computes the complete load-displacement response automatically without the necessity of specifying the time functions for the variation of the applied external load. Reference [1] Prager, W. and Hodge, P.G., Theory of Perfectly Plastic Solids, Dover Publications, New York, 1968.
226
Internal pressure P
r1 r2
Figure B.49
227
B.50
Objective To verify the linearized buckling analysis option when temperature effects are included. Physical problem A rectangular flat plate with a hinged boundary is considered, see Fig. B.50. The initial plate temperature is zero degrees and the temperature is uniform through the plate thickness. The in-plane thermal stresses are generated within the plate by changing the plate temperature. An estimate of the lowest temperature at which the generated thermal stresses will cause plate buckling is to be calculated. Finite element model Since the buckling mode is expected to be symmetric, only a quarter of the plate is modeled using four 8-node isoparametric shell elements. The material constitutive law is thermoelastic and the material elastic constants do not vary with temperature. Solution results The analytical solution for this problem is given in ref. [1] as
cr =
a 2 2 D 1 Eh3 (1 ) 1 + 2 2 + REF , D = Eh 12 (1 2 ) b a
where REF = reference temperature at which the plate is stress-free. The following results are obtained: Critical temperature cr (F) Theory ADINA 102.8 101.4
228
User hint $ The eigenvalue solution is performed using the subspace iteration method. A tight convergence tolerance (SSTOL = 1.0 10-12) is used to obtain the eigenvalue of interest accurately. Reference [1] Boley, B.A. and Wiener, J.H., Theory of Thermal Stresses, John Wiley and Sons, 1960.
Translational displacements are zero along the boundary Region discretized
= 0F q REF h = 2.5 in (plate thickness) 6 E = 2.1 10 psi n = 0.25 a = 2.0 10-5 in/in/F a = 10 in b = 50 in
Figure B.50
229
B.51
Objective To verify the BFGS method for large displacement analysis of cables. Physical problem Figure B.51 shows the cable problem considered. The cable is fixed at end A and is free to move into the Y-direction at end B. The deformed cable configuration is to be calculated when the end B is displaced by 14.4 meters along the Y-direction. Finite element model The cable is modeled using 10 truss elements and it is subjected to a uniformly distributed load simulating the cable self weight. The large displacement, small strain kinematic formulation is employed in this analysis. The Y displacement at end B is prescribed using the time function given in Fig. B.51. Solution results A good comparison with the experimental results referred to in [1] is observed. User hints $ In the unstressed configuration, the straight cable has no stiffness in the direction normal to its longitudinal axis. Thus to obtain a positive definite stiffness matrix in the first time step, nonzero initial conditions for displacements at the cable nodes are specified such that tensile stresses are introduced in the cable. These initial conditions agree with the prescribed displacement time function. $ In this problem, after the self weight forces are applied to the cable nodes, the external force vector for the cable structure does not change for subsequent time steps. This makes the energy convergence criterion more difficult to satisfy than the force convergence criterion (as the denominator of the energy criterion is a small quantity). Therefore, for this case, the solution convergence is measured using the force convergence criterion.
230
Reference [1] Bathe, K.J. and Cimento, A.P., "Some Practical Procedures for the Solution of Nonlinear Finite Element Equations," Computer Methods in Applied Mechanics and Engineering, Vol. 22, pp. 57-85, 1980.
cable A = 0.0005498 m 7 2 E = 1.746 10 N/m F
UB B
F 15 m F F A F F F F F F
F = 0.264 N
Y 28.94 m
68
Figure B.51
231
B.52
Objective To verify the full Newton iteration method with line searches for large displacement analysis. Physical problem A thin cantilever beam is subjected to an end moment, see Fig. B.52. Finite element model The structure is modeled using ten 2-node isoparametric beam elements. Solution results The theoretical solution for the end rotation and the translational displacements at end B is given by
B =
sin B WB 1 cos B ML VB , = 1 , = EI L B L B
The following results are obtained: M 196.4 10-4 392.7 10-4 B (radians) Theory ADINA 0.79 0.79 1.58 1.58 VB/L Theory ADINA 0.10 0.10 0.36 0.37 WB/L Theory ADINA 0.37 0.37 0.63 0.64
A good comparison between the numerical and analytical solutions is observed. User hint $ The solution response is traced in four equal load steps. The full Newton iteration method with line searches is effectively used in the solution of this problem.
232
0.1
M E = 3 10 b = 0.1 (thickness)
4
10
Figure B.52
233
B.53
Objective To verify the isoparametric beam element when subjected to a uniform twisting moment. Physical problem A cantilever structure with a square cross-section as shown in Fig. B.53 is considered. The beam has a bilinear constitutive law with isotropic hardening behavior. The bilinear material model properties are chosen to be a close approximation to the Ramberg-Osgood material relationship employed in ref. [1]. A torque is applied to the structure at its free end. Finite element model The cantilever is modeled using one 2-node isoparametric beam element. The solution is obtained using three load steps with full Newton iterations with line searches. Solution results The numerical results obtained are compared with the solution of ref. [1] and good agreement is observed. User hint $ The solution can also be obtained very effectively using BFGS iterations or full Newton iterations without line searches. Reference [1] Greenberg, H.J., Dorn, W.S. and Wetherell, E.H., "A Comparison of Flow and Deformation Theories in Plastic Torsion of a Square Cylinder," in Plasticity, Proc. 2nd Symp. on Naval Structural Mechanics (E.H. Lee and P.S. Symonds, Eds.) Pergamon Press, New York, 1960.
234
) ]
a=1 L = 10
Figure B.53
235
B.54
Objective To verify the multilinear material model for the truss, pipe, 2-D solid, 3-D solid, shell and iso-beam elements. Physical problem A bar structure is subjected to cyclic uniaxial loading, see Fig. B.54. The material behavior is modeled by a multilinear stress-strain relationship with isotropic hardening. Finite element model Six separate models of the same structure shown in Fig. B.54 are considered as follows: Model 1: The bar structure is discretized using one 2-node truss element. Model 2: The bar structure is modeled using one pipe element. Model 3: The bar structure is modeled using one 4-node 2-D solid (plane stress) element. At nodes 5 and 7, the z-displacement degree of freedom is kept active to permit a nonzero zz-component of strain. Model 4: The bar structure is modeled using one 8-node 3-D solid. Nodes 9 to 12 remain in the y-z plane, while nodes 13 to 16 are free to move into the xdirection to provide plane stress conditions. Model 5: The bar structure is modeled using one 4-node shell element. At nodes 17 and 19, the z-displacement degree of freedom is kept active to permit a nonzero zz-component of strain. Model 6: The bar structure is modeled using one 2-node iso-beam element. The stiffness matrix of the elements is reformed at each time step and the full Newton iteration method without line searches is used to obtain the solution.
236
Solution results The numerical solution obtained yields an identical load-deformation relationship for each of the models. The numerical solution agrees with the analytical solution. User hints $ Since v = 0.0 for the 2-D solid element, the zz-component of strain is zero in the elastic region. Subsequently the zz-component of strain is nonzero because the plastic strains are considered to be incompressible. Thus, if the z-displacement were constrained at nodes 5 and 7, the 2-D solid element would not give the same answers as the truss, pipe and iso-beam elements. Similar arguments hold for 3-D solid element and the shell element. $ Since the loading is radial the solutions are exact for any number of time steps in each loading regime. In practice, the loading is generally complex and smaller load steps should be used to accurately calculate the plastic response.
u
P L
A = 1.0 (cross sectional area) L = 1.0 2000 Stress 1500 1000 500
5
Strain
Figure B.54
237
B.55
Objective To verify the multilinear material model for the truss, pipe, 2-D solid, 3-D solid, shell and iso-beam elements. Physical problem Same as Example B.54, but the material behavior is modeled by a multilinear stressstrain relationship with kinematic hardening. Finite element model Same as in Example B.54, except for the hardening assumption and the time function. Solution results The numerical solution obtained yields an identical load-deformation relationship for all models. The numerical solution agrees with the analytical solution. User hint $ The same solution can be obtained using only 4 steps because the loading is radial (proportional). In practical analyses involving more complex stress states, the load step chosen typically needs to be smaller to permit an accurate calculation of the plastic response.
238
B.57
Objective To employ the 2-D contact surfaces to analyze plane strain contact conditions. Physical problem An elastic cylinder contacts a flat rigid surface, see Fig. B.57. The radius of the contact region is much smaller than the cylinder radius. The contact surfaces are assumed to be frictionless. Finite element model The elastic cylinder is chosen to be the contactor body. From symmetry conditions, only one quarter of the cylinder needs to be modeled. The mesh is refined near the region of anticipated contact. Because the deformations are small, linear plane strain elements are employed. A 2-D plane strain contactor surface is defined from A to B. The flat rigid surface is chosen to be the target body. It is modeled using 2 nodes, each with all its degrees of freedom deleted. The load is applied by constraining all of the Z-displacements of the top surface of the mesh to be equal and applying the load to one of the nodes. Also used is a soft truss connecting the cylinder to ground. This truss ensures that the model is stable when not in contact. Only a negligible amount of load applied to the structure is carried by the truss. Solution results The numerical results obtained for contact tractions are compared with the classical Hertz solutions [1] and reasonable agreement between the solutions is observed. Reference [1] Timoshenko, S.P. and Goodier, J.N., Theory of Elasticity, McGraw-Hill, pp. 409-420, 1970.
239
Y E = 30000 n = 0.25 R = 10 A B
240
B.58
Objective To verify the mixed u/p formulation in conjunction with frictionless contact conditions for modeling axisymmetric rubber problems. Physical problem A rubber ring is pressed between two rigid plates, see Fig. B.58. The force-deflection curve, stress distributions and contact traction distributions are to be obtained. Frictionless contact is assumed. Finite element model Due to symmetry, an axisymmetric analysis is performed. Only one-half of the ring cross-section needs to be discretized. Constraint equations are used along the top of the discretized region to apply uniform z-displacements. The z-displacement of the single unconstrained node on this line is prescribed and the force applied to the model is measured using the corresponding z-reaction. A compressible Mooney-Rivlin material description is used to model the rubber. The bulk modulus is chosen to set the ratio of bulk modulus to shear modulus equal to 1500 (which is equivalent to a Poisson's ratio of .4997 in a linear elastic analysis). Because of the almost incompressible material description, the mixed u/p formulation is used with nine-node elements. Full Newton iterations without line searches are used in the equilibrium iterations. A contactor surface is placed along the bottom of the ring and a rigid target surface is specified using two nodes which have all degrees of freedom removed. The coefficient of friction is set to zero in this analysis. Solution results The ring expands radially as the ring is compressed. Reasonable agreement between the force-deflection solution in [1] and the ADINA results is observed.
241
User hint $ The displacement-based formulation cannot be used to solve this problem because the elements will lock. Reference [1] Lindley, P.B., "Compression Characteristics of Laterally-Unrestrained Rubber ORings," J. Instn. Rubb. Ind., Vol. 4, pp. 209-213, 1967.
Mooney-Rivlin constants: C 1 = 0.293 MPa C 2 = 0.177 MPa k = 1410 MPa
Prescribed displacement
C L
0.012 m
Region discretized
0.04865 m
Figure B.58
242
B.59
Objective To verify the use of 3-D contact surfaces to model frictional sliding. Physical problem A point mass is attached to two identical springs which are anchored to a flat rigid surface, see Fig. B.59. The coefficient of friction between the mass and the rigid surface is 0.15. Both springs have the same linear force-deflection relationship. The point mass is given an initial velocity of -1.0 in the X direction. The movement of the mass takes place along the X direction and the motion is resisted by the developed frictional force (when the mass is moving along the positive X direction, the frictional force acting on the mass equals 0.30 in the negative X direction). An analysis is performed to evaluate the effect of friction on the vibrations of the massspring system. Finite element model Target Surface The flat rigid surface is chosen to be the target surface and is modeled using four nodes each with no degrees of freedom. Contactor surface The mass-spring system is chosen to be the contactor body. An auxiliary node (fixed in space) is used to define the contactor surface consisting of two contactor segments. The auxiliary node lies outside the region of the rigid target surface and thus is not in contact throughout the analysis. The direction of the normal vector at the point mass is obtained from the geometry of the contactor segments. Solution results The obtained solution is such that at any given time, the work done by the frictional force equals the difference between the initial energy and the current energy of the system. Hence the maximum amplitude of vibration decreases linearly in the successive oscillations.
243
10
10
m = 0.2
Figure B.59
244
B.62
Objective To verify the use of 3-D contact surfaces for analysis of dynamic contact. Physical problem Figure B.62 shows the problem considered. A sphere moving with a uniform initial velocity impacts a flat, rigid surface. Finite element model Due to symmetry of the sphere geometry, only a thin wedge from the sphere continuum is discretized. The wedge is bounded by its two semicircular sides and the enclosed wedge angle is 1 degree. The wedge is discretized using eight node 3-D solid elements. The contactor surface is defined over the wedge boundary. The target surface is defined to be the flat rigid surface, which is modeled by specifying target nodes with no degrees of freedom. Solution results The obtained numerical solution for the normal contact tractions compares well with the approximate analytical solution [1]. User hint $ This problem can also be solved using 2-D axisymmetric elements. Reference [1] Timoshenko, S.P. and Goodier, J.N., Theory of Elasticity, McGraw-Hill, 1970.
245
Region discretized
R Z Y
E n r R
= = = =
Figure B.62
246
B.65
Objective To verify the 2-D solid axisymmetric element in large displacement analysis using the load-displacement control (LDC) method. Physical problem A spherical shell is subjected to a concentrated apex load, see Fig. B.65. Finite element model Ten 8-node axisymmetric elements are used. The nodes at the apex can slide in the vertical (Z) direction while the nodes at the other end are fixed. The LDC method is used to automatically select the load magnitudes. In the first step, the Z-displacement at the apex of the shell is prescribed to be 0.01 in. The maximum apex displacement of interest is -0.175 in. Solution results Good agreement with the solutions by Stricklin [1] and Mescall [2] is observed. User hint $ The axisymmetric element extends 1 radian in the circumferential direction. The total apex load of 100 lbs applied to the actual shell is therefore represented by a load of 100/2 applied to the model. References [1] Stricklin, J.A., "Geometrically Nonlinear Static and Dynamic Analysis of Shells of Revolution," High Speed Computing of Elastic Structures, Procedures of the Symposium of IUTAM, University of Liege, August, 1970. Mescall, J.F., "Large Deflections of Spherical Shells Under Concentrated Loads," J. App. Mech., Vol. 32, pp. 936-938, 1965. -
[2]
247
F h w q
R
Region discretized H
Figure B.65
248
B.66
Objective To verify the 3-D mixed u/p formulation element using a rubber material description. Physical problem The solid rubber cylinder shown in Fig. B.66 is fixed at one end and is subjected to a prescribed rotation at the other end. The cylinder is constrained in the axial direction. Finite element model The cylinder is modeled using eight collapsed 27-node elements that use the mixed u/p formulation. The rubber is modeled using the compressible Mooney-Rivlin material description, with the bulk modulus chosen to set the bulk to shear modulus ratio equal to 5000. On the far end of the cylinder, all of the nodes are fixed. On the near end, the angular displacement of the center node is prescribed and nonlinear rigid links placed between this node and the other nodes on the near end cause the entire cross-section to rotate by this amount. Solution results The analytical solutions for the moment M and compressive axial force N are given by Rivlin [1] (assuming total incompressibility) as
M = ( C1 + C2 ) r 4 , N = ( C1 + 2C2 ) r 4 2 L L
where r is the cylinder radius, L is the cylinder length and is the total angular displacement. Good agreement is observed between the analytical solution and the ADINA results. User hint $ The Ogden material can also be employed in this example, see Example B.99. Reference [1] Rivlin, R.S., "Large Elastic Deformations of Isotropic Materials III. Some Simple Problems in Cylindrical Polar Coordinates," Phil. Trans. A.240, pp. 509-525 (1948).
249
N M
Mooney-Rivlin material:
C1 = 3.0 10 Pa 5 C2 = 1.5 10 Pa 9 k = 4.5 10 Pa
5
Figure B.66
250
B.69
Objective To verify the solution procedure for computation of the response of a pipe structure subjected to bending when stiffening effects due to internal pressure are important. Physical problem The pipe structure is shown in Fig. B.69 is loaded by the forces F and the internal pressure. Finite element model Three pipe elements are used to model the structure. Only the section of the pipe between points B and C needs to be considered due to symmetry. Solution results The ovalization of the pipe at the symmetry point decreases as the internal pressure increases. The comparison between computed and experimental results of refs. [1] and [2] is shown in the table below. The flexibility factor is defined as
f =
4a 3 E D MR
where a is the mean radius of the pipe cross-section, M is the bending moment, D is the rotation around the x axis at point D and the other quantities are shown in Fig. B.69. Internal pressure (psi) 100. 200. User hints $ Ovalization effects are important in this problem, so ovalization degrees of freedom are employed. Flexibility factor Experiment 7.75 4.75 ADINA 7.52 4.26
251
$ The internal pressure acts against ovalization and can significantly reduce the flexibility of the pipe structure. The effect of internal pressure on flexibility is analyzed in refs. [1] and [2]. References [1] [2] Kafka, P.G. and Dunn, M.B., "Stiffness of Curved Circular Tubes with Internal Pressure," ASME Journal of Applied Mechanics, Vol. 78, pp. 247-254, 1956. Rodabaugh, E.C. and George, H.H., "Effect of Internal Pressure on Flexibility and Stress-Intensification Factors of Curved Pipe or Welding Elbows," ASME Journal of Applied Mechanics, Vol. 79, pp. 939-948.
F C d
L D Z D B 45 R Y
Figure B.69
252
B.70
Objective To verify the ADINA large strain elastic-plastic algorithm in 2-D and 3-D solid elements. Physical problem The material is stretched uniaxially by applying a load as shown in Fig. B.70. Finite element model One 2-D plane stress and one 3-D solid element are used. The first load step corresponds to the start of yielding, then the load is increased in the next 20 steps; then follows an elastic unloading in steps 22 and 23 and reverse plastic deformations in steps 24 to 33. The full Newton method without line searches is employed in the analysis. Solution results Both elements give the analytical solution. User hints $ The stresses are computed including the effect of change of the element dimensions. As can be observed, the stresses balance the total axial force applied at any time. $ The solution is exact for any number of time (load) steps used when the loading is radial (proportional) as in this example. However, convergence in the solution of the equilibrium equations must be achieved. $ When reversing the direction of load application, it can be helpful to take a small step first and then larger steps to complete the load reversal.
253
1.0
1.0
5450
Figure B.70
254
B.71
Objective To verify the ADINA large strain elastic-plastic algorithm in 2-D and 3-D solid elements in the case of kinematic hardening conditions. Physical problem Same as Example B.70, except that the material is uniaxially loaded as shown in Fig. B.71 and the large strain elastic-plastic material description uses kinematic hardening. Finite element model Same as Example B.70, except that kinematic hardening is used instead of isotropic hardening. Solution results Both elements give the analytical solution. User hints $ See the user hints for Example B.70.
deformationindependent pressure load py 5500 5490 3500 2000 1700 1 21 22 26 31 Time
Figure B.71
255
B.72
Objective To verify the use of contact surfaces when the surfaces initially overlap. Physical problem Two long concentric cylinders are press-fit into each other, see Fig. B.72. The contact pressure and stress distributions are to be determined. Plane strain conditions along the cylinder axis are assumed. Finite element model Eight 8-node 2-D solid elements are employed along with two contact surfaces to model a section of the structure. The elements adjacent to the contact surfaces partially overlap. No loads are applied to the model. Solution results The model deforms to eliminate the overlap. The analytical solution for the contact pressure can be found using the displacement solution for a thick-walled cylinder [1]:
p=
The following results are obtained: p (N/mm2) Theory ADINA 68.5 68.4 Reference [1] Reismann, H. and Pawlik, P., Elasticity: Theory and Applications, John Wiley and Sons, 1980.
256
Region discretized
E1 n1
E2 n2
10 mm
C L
Figure B.72
257
B.73
Objective To check the static to dynamic restart option in ADINA. Physical problem Fig. B.73 shows the spring and mass considered. The material constitutive relation is elastic-plastic. Finite element model The spring is represented as a truss element. The mass is one-half of the truss element mass. The total static response is reached in one load step and the subsequent dynamic response is calculated using t = 0.00025. Solution results The following results are obtained: Theory 0.2725 0.7725 225.7 2252.0 ADINA 0.2725 0.77232 224.4 2244.0
Static displacement Maximum displacement Frequency of vibration prior to reaching the maximum response Frequency of vibration after reaching the maximum response User hint
$ The dynamic response is in this example only due to the externally applied increase in load applied after the static load application; hence due to the load of 500 lbs. This load causes further plastic deformations from time = 1.0 to 1.014 after which an elastic oscillation occurs.
258
R 10
Figure B.73
259
B.74
Objective To verify the isoparametric beam, pipe, shell, 2-D solid and 3-D solid elements in the analysis of steady state shear creep. Physical problem A cantilever is loaded in pure shear, see Fig. B.74. Finite element model The structure is modeled by one element. In the isoparametric beam, pipe and shell elements, all rotational degrees of freedom are restrained to eliminate any normal stresses due to bending. The time integration of the creep response is performed using the Euler backward method ( = 1). Solution results The numerical solution is in excellent agreement with the analytical solution which is
F E, n
A = 0.01 m2 E = 2.07 10 5 MPa n = 0.3 F = 1 MN 0.8 e C = 4.1 10-11 s 3.15 (t + 0.5 t )
Figure B.74
260
B75. Cantilever beam in creep (2-D plane stress, iso-beam), time step blocks
B.75
Cantilever beam in creep (2-D plane stress, iso-beam), time step blocks
Objective To demonstrate creep analysis with variable time step in ADINA by using time step blocks. Physical problem Same as Example B.39. Finite element model Same as Example B.39, except that time step blocks are used to change the time step and restarts are not employed. Solution results The same numerical results are obtained in this analysis as in Example B.39.
261
B.76
Large displacement analysis of a thin cantilever subjected to end moment, ATS method
Objective To verify the automatic time stepping (ATS) method for large displacement analysis. Physical problem Same as Example B.52. Finite element model Same as Example B.52, except that the ATS method is used. The entire load is applied in one load step. Solution results Same as Example B.52.
262
B.82
Objective To verify the orthotropic thermoelastic material model when used with 3-D elements and skew systems. Physical problem An orthotropic thermoelastic brick is heated to a state of uniform temperature, see Fig. B.82. The material constants are defined in the skew system a-b-c. Finite element model A patch of two 8-node 3-D solid elements is used. The boundary conditions are chosen to allow the block to expand freely. Solution results The finite element analysis solution is stress-free, which is correct. The total strains and thermal strains predicted by the model are
TH TH TH eaa = eaa = 1.0 106 , ebb = ebb = 2.0 106 , ecc = ecc = 3.0 106
La
Figure B.82
263
B.99
Objective To verify the 3-D mixed u/p formulation element using the Ogden material description. Physical problem Same as Example B.66, except that the Ogden material is employed in this example. Finite element model Same as Example B.66 except that the rubber is modeled using the compressible Ogden material description, with the bulk modulus chosen to set the bulk to shear modulus ratio nearly equal to 5000. The material constants are 1 = 1.75105 Pa 2 = -7.5104 Pa 3 = 2.5104 Pa Solution results The analytical solutions for the moment and compressive axial force can be derived assuming total incompressibility following reference [1]. The ADINA solutions agree well with the analytical solutions. Reference [1] Ogden, R.W., Non-linear Elastic Deformations, Ellis Horwood Limited, p 294, 1984. 1 = 3.5 = 4.5109 Pa 2 = -2.4 3 = 4.5
= 4.5 109 Pa
264
Figure B.106
265
ADINA gives exactly these results. In step 9, the deformations were chosen so as to introduce compressive strains within element 1. Therefore this element wrinkles. User hint $ If no initial strains are introduced into the model, the initial stiffness matrix is singular. Reference [1] Bathe, K.J., Finite Element Procedures, Prentice-Hall, 1996, Chapter 6.
266
b a
b a 10
Z
DX DY Y
l
25
25
Fabric material:
Ea = 1540 Eb = 1540 Ec = 1540 Gab = 770 nab = 0 nac = 0 nbc = 0 thickness = 1.0
-5
Figure B.113
267
268
20
5 M
1 Layer 3 Layer 2 Layer 1 Layers have equal thickness. Material properties: Layers 1, 3: Elastic-plastic material E = 2.5 10 n = 0.3 sy = 7.0 4 E T = 2.5 10
5
Figure B.114
269
1 =
bp b 2 + a 2 bp c 2 + b 2 , 2 = + 2 2 E b a E c2 b2 E ( b 2 a 2 )( c 2 b 2 ) 2b3 ( c 2 a 2 )
Hence p =
270
Good agreement with the stress-free boundary conditions and with the analytical contact pressure is observed. User hints $ Note that the same results are obtained by subjecting the external cylinder to thermal loading with
where = coefficient of thermal expansion and = temperature increase. $ Within a stress-free element, a positive uniform initial strain field corresponds to a shrinkage of the element. This is because the strain that enters into the stress-strain law is calculated as the sum of the total (mechanical) strain and the initial strain.
C L
= 7.55 10-4
a b c
a b c d
= = = =
5 10 15 4
E = 2.1 10 n = 0.0
Figure B.115
271
272
5.969 m
Figure B.116(a)
273
Material Parameters Concrete: Uniaxial parameters: Initial tangent modulus Poisson's ratio Uniaxial cut-off tensile strength Uniaxial maximum compressive stress (SIGMAC) Compressive strain at SIGMAC Uniaxial ultimate compressive stress Uniaxial ultimate compressive strain
24800 MN/m2 0.15 4.1 MN/m2 -41 MN/m2 -0.00186 m/m -30 MN/m2 -0.0032 m/m
Triaxial compressive failure curves: Principal stress ratios SP1(I) SP3(I,1) (at SP2=SP1) SP3(I,2) (at SP2=BETASP3) SP3(I,3) (at SP2=SP3) 1 0.0 1.00 1.20 1.20 2 0.1 1.62 2.25 2.06
Curve number, I 3 4 0.2 0.3 2.10 2.50 2.75 3.01 2.32 2.55
Other parameters: Stress ratio for failure surface input (BETA) Constant for definition of critical strains (C1) Constant for definition of critical strains (C2) Constant used to define the tensile strains corresponding to zero stress (XSI) Stiffness reduction factor Shear reduction factor Density Figure B.116(b)
274
275
Solution results The numerical solution obtained yields an identical load-deformation relationship for each of the models. The numerical results agree with the analytical solution. User hints $ The temperature for the first three steps is = 100 and the yield curve for this temperature is obtained by the interpolation between the input yield curves at = 0 and = 200. $ The accumulated effective plastic strain e P in steps 4 to 10 is the same as in Example B.54 since the stress-strain curves shown in Fig. B.54 and Fig. B.117 for = 200 are identical and the loading is radial to the yield surface.
u A P
A = 1.0 L = 1.0
Stress
q=0
q = 0: q = 200
E = 10 n = 0.3 -5 a = 10
6
ET2 = 2.0 10
5
E T1 = 5.0 10
q = 200: E = 10 n = 0.0-4 a = 10
0.001
0.004
Figure B.117
276
277
Z Y
Frictionless contact
Intitially at rest
10 E = 1 10 n = 0.0 r = 4.0
5
Figure B.118
278
&VP eij =
Sij 3 1 2 y
where Sij are the deviatoric stress components, is the effective stress, defined as
3 Sij Sij 2
279
Also, y is the yield stress and is the fluidity parameter. The viscoplastic strain rates are set equal to zero when y The solution for the first time step (t = 1 10-4) corresponds to the elastic stress state. The steady state is reached after 5 steps of t = 0.5. The results of the two models are very close. User hints $ The steady-state solution can also be obtained using the plasticity model and perfect plasticity conditions.
Z
r1 Internal pressure P r2
Figure B.119
280
281
Vo
E = 1.562510 MPa n = 0.25 r = 7500 kg/m 3 d = 0.1 V0 = 5 m/s L=5m Circular cross-section
Figure B.120
282
1 2 3 4 User hints
$ It is important to perform a nonlinear analysis and model the centrifugal loading as deformation-dependent to obtain an accurate solution. $ Incompatible modes are used in the 3-D elements (the default). If incompatible modes are not used, the natural frequencies are much higher, because 8-node 3-D elements are too stiff in bending.
283
25
10
w = 3000 rpm
c
Clamped 60
1042.5
Figure B.121
284
0 = 2 c
2 4 2 y = 10 y
with = 0 C at the centre of the plate and = 100 C at the free edge. Five different virtual material shifts are used to check the J-integral shift independence. Solution results The stress intensity factor KI for this problem is given in reference [1] as
K I = 1.0 K 0 , K 0 =
E 0 b 2 a 2 a c2 3 2
A value of KI is derived from ADINA using the relation K I = EJ where J is twice the value of the J-integral obtained in ADINA, since only half of the material around the crack is taken into account in the finite element model.
285
The following results are obtained using virtual shift 2: KI (107 N-m-3/2) Theory ADINA 1.626 1.632
User hint $ The data for temperature loads can be read from a file where they can be previously generated using a simple Fortran program. Reference [1] T.K. Hellen and F. Cesari, "On the solution of the centre cracked plate with a quadratic thermal gradient", Eng. Fract. Mech., vol. 12, pp 469-478, 1979.
q y
b = 0.10 m h/b = 2.5 a/b = 0.10 4 q = 10 y 2 11 2 E = 2.07 10 N/m n = 0.3 -5 a = 1.35 10 m/m/C
a b
Figure B.122
286
287
C D
Dimensions: W = 50.35 a = 27.713 B = 25.0 D = 12.5 C = 62.5 d = 60.0 H = 25.5 V = 3.5 All lengths in mm.
V H
a W
Figure B.123(a)
Multilinear elastic-plastic stress-strain curve: E = 2.11011 N/m2 v = 0.3 e (log strain) 0.0010 0.0050 0.0125 0.0760 0.2800 (108 N/m2) (true stress) 2.128 2.593 3.125 4.388 4.920 Figure B.123(b)
288
289
Region discretized
a W
Resistance curve
10.
20.
30.
Figure B.124
290
291
292
D/2
a0
Control curve Multilinear plastic stress-strain curve Log strain 0.5 Crack length (mm) 2.0
Figure B.128
293
294
Water surface
Prescribed displacement
Towing cable
Towed sphere Y
X (1) (2)
(1) Initial geometry (2) Steady-state equilibrium under gravity and hydrodynamic forces
Length of cable = 300 ft Diameter of cable = 0.03 ft Young's modulus of cable = 3.0 108 psf Density of cable = 300 lb/ft3 Drag coefficient of cable = 1.5 Inertia coefficient of cable = 0.0 Figure B.130 Diameter of sphere = 1.0 ft Weight of sphere = 700 lb Drag coefficient of sphere = 0.5 Inertia coefficient of sphere = 0.0 Density of seawater = 1.987 slugs/ft3
295
296
ADINA-F problems
297
ADINA-F problems
298
F.1
Objective To verify the behavior of the two- and three-dimensional quadratic fluid flow elements. Physical problem The problem of the general Couette flow is considered, see Fig. F.1. The bottom surface is fixed and the top surface is moving at a constant velocity. There is also a pressure gradient applied in the flow direction. The flow is assumed to be a fully developed laminar flow and the velocity profile is to be found. Finite element model This problem is solved using two separate two- and three-dimensional models in the same analysis. In each model three quadratic elements are used normal to the flow direction. A steady-state flow condition is assumed. Solution results The governing equation for this problem is
d 2 v dp = dz 2 dy
v=
1 dp z 2 hz Vz dp h V + , yz = z + dy 2 2 h dy 2 h
The finite element and analytical solutions for the velocity and shear stress distributions are identical. User hints $ Note that here the tractions normal to the vertical boundaries are equal to the pressure. ADINA-F applies the normal tractions (and hence the pressure).
299
ADINA-F problems
$ The pressure gradient is imposed by applying a uniform pressure (normal traction) of 5.0 at the left face of the finite element model and no pressure at the right face. dp This gives the desired gradient of = 1 . Note that the same solution for dy velocity would be obtained if, instead, pressures of 5.0 + p0 and p0 were applied to the left and right faces, respectively, for any value of p0. $ Note that the mathematical traction (MT) formulation is used. This formulation is useful when modeling the artificial finite boundary of an infinite domain. References [1] [2] Schlichting, H., Boundary-Layer Theory, 7th edition, McGraw-Hill Company, New York, 1979. Potter, M.C. and Foss, J.F., Fluid Mechanics, The Ronald Press Company, 1975.
V
dp dy
dp dy V h r m
= -1 = 60. = 3. = 1. = 0.01
Figure F.1
300
F.2
Objective To verify the behavior of the 4-node axisymmetric FCBI elements. Physical problem The problem of a fully developed flow in a pipe is considered, see Fig. F.2. A pressure gradient is applied in the axial direction, and the velocity profile is to be found. Finite element model The finite element model consists of 4-node axisymmetric FCBI elements. The pressure gradient is imposed by applying a uniform pressure of 10 on the left face and no pressure on the right face. A steady-state flow condition is assumed. Solution results The governing equation for this problem is
1 d dw 1 dp (here r y) r = r dr dr dz
and the analytical solution is
w=
1 dp 2 2 r ( r r0 ) , rz = 2 dp 4 dz dz
The velocity profile and the streamlines obtained from the finite element analysis correspond to the analytical solution. User hints $ An axisymmetric element models an angle of one radian in the circumferential direction. $ Note that the pressure on the vertical boundaries of the model is equal to the normal traction on these boundaries, and in ADINA-F normal tractions are applied.
301
ADINA-F problems
$ Note that the mathematical traction (MT) formulation is used. This formulation is useful when modeling the artificial finite boundary of an infinite fluid domain. Reference [1] Potter, M.C. and Foss, J.F., Fluid Mechanics, The Ronald Press Company, 1975.
dp dz w(r)
r0 z
dp = -1 dz r0 = 2 r = 1.0 m = 0.2
Figure F.2
302
F.4
Objective To verify the behavior of the 8-node three-dimensional FCBI elements. Physical problem The problem of a fully developed flow in an annulus is considered, see Fig. F.4. A pressure gradient is applied in the flow direction, and the velocity profile in the annulus is to be found. Finite element model This finite element model consists of 8-node FCBI elements. A sector of 1 degree is represented, and a steady-state flow condition is assumed. Solution results The governing equation for this problem is
1 d du 1 dp r = r dr dr dx
and the analytical solution is
2 2 1 dp 2 2 ( r2 r1 ) r u= ( r r2 ) + r ln r 4 dx ln 1 2 r2 2 2 ( r2 r1 ) 1 1 dp 2r + rx = 4 dx r1 r ln r2
The velocity profile and stress distributions from the finite element analysis compare well with the analytical solutions.
303
ADINA-F problems
Reference [1] Potter, M.C. and Foss, J.F., Fluid Mechanics, The Ronald Press Company, 1975.
dp dx
r2 r1
x
u(r)
y,r
Figure F.4
304
F.5
Objective To verify the behavior of the 4-node two-dimensional FCBI elements. Physical problem The problem of a fully developed flow between concentrically rotating cylinders is considered, as shown in Fig. F.5. Body force effects are ignored. Finite element model The finite element model consists of 4-node 2-D FCBI elements. The inner and outer rotating cylinders are modeled with an angular velocity boundary condition. The pressure at one point on the inner cylinder is set to zero. Solution results The governing equations for this problem are
d 2 ( v ) dr 2
d v dr r
= 0,
2 dp v = , dr r
v =
( r
2 2 2
1r12 )
r22 r12
r+
2 2 p ( 2 r2 1r1 ) = 2 2(r2 r2 ) 2 1
(r
2 1
)+
(r
2 2
2 2 1
r ln r1
r r ( 2 1 ) 1 1 2 2 2 r1 2 ( r22 r12 ) r
4 4 1 2 2
=2
( 2 r22 1r12 )
(r
2 2
r12 )
= constant
305
ADINA-F problems
The velocity, shear stress, pressure and vorticity obtained from the finite element analysis compare well with the analytical solutions. Reference [1] Potter, M.C. and Foss, J.F., Fluid Mechanics, The Ronald Press Company, 1975.
vq W2 r1 = 1 r2 = 2 W1 = 1 W2 = 2 r =1 m=1
W1 r2 r r1
Figure F.5
306
F7. Non-Newtonian flow between two parallel plates (2-D and 3-D elements)
F.7
Non-Newtonian flow between two parallel plates (2-D and 3-D elements)
Objective To verify the fluid power law material model for the quadratic 2-D and 3-D fluid elements. Physical problem The laminar steady flow of a non-Newtonian fluid between two parallel plates is considered, see Fig. F.7. A pressure gradient is applied in the flow direction. The velocity profile is to be found. Finite element model This problem is solved using two separate two-and three-dimensional models in the same analysis. Due to symmetry, only one half of the fluid flow is considered. The finite element model consists of 9-node 2-D elements and 27-node 3-D elements, with five elements in the z direction. The power law fluid model is used to represent the non-Newtonian fluid. Solution results The general governing equations for this problem are
dp yz = , yz = 2 e yz z dy
where A ( D ) , D = 2 e yz and e yz =
n
1
307
ADINA-F problems
User hint $ The boundary conditions are applied as in Examples F.1 and F.2. Note that here again the normal tractions on the left and right vertical faces are equal to the pressure. Reference [1] Crochet, M.J., Davies, A.R. and Walters, K., Numerical Simulation of Non-Newtonian Flow, Elsevier, New York and Amsterdam, 1983.
dp dy z v(z)
Figure F.7
308
F.9
No flow test
Objective To verify the behavior of the 4-node 2-D FCBI elements when subjected to gravity loading. Physical problem The fluid in the domain shown in Fig. F.9 is subjected to gravity loading. Zero velocities are imposed on the fluid boundaries, and the solution should of course give zero velocities everywhere and a hydrostatic pressure distribution. Finite element model The finite element model consists of 72 4-node 2-D FCBI elements. The pressure is assumed to be zero at one node on the top surface (z = 0) and gravity loading is applied. Solution results The analytical solution of this problem is p = gz , v = 0 everywhere. The model gives the predicted pressure variation which is equal to the analytical solution. User hints $ This is a very simple but instructive problem. The analysis tests whether the gravity loading option is properly implemented and whether elements can predict the pressures in the domain accurately [1]. $ Note that in this analysis, the pressure at one node must be prescribed. This is because the normal velocity conditions on the complete boundary of the domain are prescribed. Reference [1] Fortin, M. and Fortin, A., "Experiments with Several Elements for Viscous Incompressible Flows," Int. J. for Numerical Methods in Fluids, Vol. 5, pp. 911-928, 1985.
309
ADINA-F problems
z y 2 10 g 20
20
Figure F.9
310
F.10
Objective To demonstrate the modeling of an accelerated flow in a converging channel using 4node 2-D FCBI elements. Physical problem Fig. F.10 shows the problem considered. For the ideal Hamel problem, the fluid flows in the space between two infinite plates intersecting at an angle. The velocity profile is self-similar at all radii and each fluid particle moves along radial lines towards the sink. The Reynolds number for this problem is defined by
Re =
Vr r
where is the wedge half angle, r is the radial coordinate and Vr is the velocity along the centerline at r. The case of Re = 61 is studied in this problem. Finite element model In the finite element analysis only a finite length of the wedge geometry can be modeled, and only one half of the fluid flow domain is considered due to symmetry conditions. The boundary conditions are that the exact velocity is applied on the upstream surface and no slip conditions are prescribed at the plate walls. Solution results The analytical solution of the Hamel problem is found in ref. [1]. The velocity along the centerline and the wall pressure compare well with the analytical solution. Note that the radial mesh is needed to obtain an accurate finite element solution. Reference [1] Batchelor, G.K., An Introduction to Fluid Dynamics, Cambridge University Press, 1981.
311
ADINA-F problems
r = 4.0
Region discretized
30
no-slip boundary
r=1 m = 0.04
r = 0.25
y
Figure F.10
312
F.12
Objective To demonstrate the analysis of a square wall-driven cavity. Physical problem Fig. F.12 shows the problem of a square wall-driven cavity. In this analysis the tangential velocity is imposed along the upper boundary while the normal velocity on that boundary is zero. The flow is characterized by the Reynolds number
Re =
U top L
where L = width of the wall and Utop = uniform tangential velocity applied on the top surface. The steady state analyses for Re =100, 400 and 1000 are performed. Finite element model A mesh of 4-node 2-D FCBI elements is employed. For the case of Re = 1000, two load steps are used to obtain a converged solution. Solution results The results are in good agreement with those of ref. [1]. User hint $ Note that in this analysis the pressure at one node must be prescribed. This is because the normal velocity conditions on the complete boundary of the domain are prescribed. Here, we select the pressure at point A to be equal to zero. Reference [1] Fortin, M. and Thomasset, F., "Mixed Finite Element Methods for Incompressible Flow Problems," J. of Comp. Physics, Vol. 30, No. 1, pp. 113-145, 1979.
313
ADINA-F problems
Utop
Figure F.12
314
F.13
Objective To demonstrate the capability of solving strongly temperature-dependent and coupled flow problems. Physical problem The problem of natural convection in a square cavity is to be analyzed, see Fig F.13. The fluid in the cavity is subjected to gravity and the vertical walls are at different temperatures resulting in natural convection. This problem is characterized by the Rayleigh number and the Prandtl number
Ra =
g L3 , Pr =
where = temperature difference between the hot and cold walls and L = depth of the cavity. The problem is solved for the Rayleigh numbers of 103, 104, 105 and 106, each time with Pr = 0.71 and assuming steady-state conditions. Finite element model The finite element analysis uses the Boussinesq approximation, which assumes that the fluid density varies with the temperature but the fluid flow is governed by the incompressible flow equations. The finite element mesh consists of a 12 12 mesh of 9-node 2-D elements of gradually varying sizes. The smaller sized elements are used near the wall to account for the velocity and thermal boundary layer effects. For the case of Ra = 106, five load steps are used to obtain a converged solution. Only one solution step is required for the other cases. In the input to ADINA-F, we use the above equations in a modified form so that the density input gives the Rayleigh number. Solution results The governing equations of the mathematical model are non-dimensionalized by c 2 as scale factors for length, velocity, pressure and choosing L, , 2 and L L h c temperature. Then the nondimensionalized equations are [1]
u = 0, u u = p + ( Ra Pr ) e + Pr 2 u, u = 2
315
ADINA-F problems
Nu =
hL qL q = = k k k ( L )
and is equivalent to a dimensionless heat flux. Here, the heat flux is calculated at element integration points nearest to the hot wall. The results of maximum horizontal velocity, maximum vertical velocity and maximum and minimum local Nusselt number show good agreement with the solutions of Davis [1]. User hint $ Since boundary velocities are prescribed in this analysis, the pressure is set equal to zero at point A. Reference [1] De Vahl Davis, G., "Natural Convection of Air in a Square Cavity: A Bench Mark Numerical Solution," Int. J. for Numerical Methods in Fluids, Vol. 3, pp. 249-264, 1983.
316
v = w = 0
qc v = w = 0
qh v = w = 0
v = w = 0
0.5
Figure F.13
317
ADINA-F problems
F.28
Objective To verify the use of the time integration method in transient analysis using 8-node 3-D FCBI elements. Physical problem The same problem as described in Example F.4 is solved. However, a transient flow condition is considered here. Finite element model The finite element model is the same as in Example F.4. The Euler backward integration method with unequal time step increments is used in the transient analysis. The time step increment t = 1 is for the first three steps, 2 for the fourth step, 2.5 for the fifth to seventh steps and 37.5 for the last step. Solution results The steady-state solution is given in Example F.4. During the analysis, the finite element solution tends towards the steady-state solution.
318
F.33
Objective To verify the use of the restart option and the assignment of initial conditions in transient analysis using 4-node 2-D FCBI elements. Physical problem An infinite flat plate is supporting a fluid and oscillating sinusoidally, see Fig. F.33. The velocity profile of the fluid is to be determined. Finite element model The finite element model consists of forty 4-node 2-D FCBI elements. The Euler 2 is used for a total backward time integration method with a time increment t = 80 of 80 time steps. Two consecutive runs are made. The first run consists of 40 time steps which correspond to the time span 0 to . The second run using the restart option consists of another 40 time steps which correspond to the time span from to 2. Initial velocities obtained from the analytical solution are assigned to all nodes for the first run. Solution results The governing equation of this problem is [1]
v 2 v = 2 t z
subject to the boundary conditions
v ( 0, t ) = V cos t , v ( z = , t ) = 0
The solution is given in ref. [1] as
v = V cos t z exp 2 z 2
Good agreement between the finite element solution and the analytical solution is
319
ADINA-F problems
observed. Reference [1] Potter, M.C. and Foss, J.F., Fluid Mechanics, The Ronald Press Company, New York, p. 289, 1975.
Fluid
Figure F.33
320
F35. Heat generated in laminar flow between two rotating cylinders (2-D elements)
F.35
Heat generated in laminar flow between two rotating cylinders (2D elements)
Objective To verify the behavior of the 4-node axisymmetric FCBI elements, in particular the calculation of viscous dissipation. Physical problem The problem of a fully developed flow between two concentrically rotating cylinders is considered. The same problem is also considered in Example F.5 but without viscous dissipation. Here 1 = 0, 2 = 1 and k = 0.2, corresponding to a Brinkman number of 5, see below. Finite element model The finite element model is the same as in Example F.5, except that viscous dissipation is considered. Solution results The governing equation and the solution for velocity, shear stress, pressure and vorticity are given in Example F.5. The governing equation for the temperature arising from dissipation is [1]:
k d d dv v r + = 0 r dr dr dr r
with boundary conditions
= 1 at r = r1 , = 2 at r = r2
The solution is (ref. [1], but note that there are errors in [1] regarding the formulas of velocity and temperature)
r (1 2 1 ) 1 = Br 2 2 1 ( r2 r1 )
4 2
2 2 1 r1 1 r1 2 2 r r
r r ln ln r r1 + 1 r2 ln r2 ln r1 r1
321
ADINA-F problems
where Br =
profile compares well with the analytical solution. Reference [1] White, F.M., Viscous Fluid Flow, McGraw-Hill Book Company, New York, p. 117, 1974.
322
F.37
Objective To verify the behavior of the 4-node two-dimensional FCBI element in ADINA-F when subjected to a buoyancy force. Physical problem A fluid between two vertical plates of different temperatures rises near the hot plate and falls near the cold plate due to buoyancy effects, see Fig. F.37. The flow field is to be considered in steady state conditions. Finite element model The finite element model consists of twenty 4-node 2-D FCBI elements. Only one layer of elements is used in the vertical direction. The temperature is prescribed at the nodes along the two vertical plates. Solution results The governing equations are (the body force term g is not included) [1]
g ( m ) +
d 2w d 2 = 0, k 2 = 0 dy 2 dy
where m =
= 0 at y = -h,
w* =
Gr * ( y y*3 ) , * = y* 6 hw , 2 y* = m y , * = h 1 m
= Grashof number
where w* =
Gr =
g 2 h3 ( 1 m )
323
ADINA-F problems
Note that the solution is independent of k and cp. Very good comparisons of velocity and temperature are shown between the ADINAF and the analytical solution. User hint $ Note that when the body force g (which induces zero velocity) is deleted the pressure output from ADINA-F is the dynamic pressure, which is equal to the total pressure minus the hydrostatic pressure. Reference [1] White, F.M., Viscous Fluid Flow, McGraw Hill Book Company, New York, p. 115, 1974.
g
Z h=1 g=1 r=1 m=1 cp = 1 k=1 b=1
q = q2
q = q1
Figure F.37
324
F.39
Objective To verify the behavior of the 4-node axisymmetric FCBI elements, in particular the capability to include viscous dissipation. Physical problem Same as Example F.2, except that viscous dissipation is included. The temperature at the pipe wall is prescribed to be zero. Finite element model The finite element model consists of 4-node 2-D axisymmetric FCBI elements. A steady state flow condition is assumed. Solution results The governing equation and the solution for the velocity and shear stress are given in Example F.2. The governing equation for the temperature arising from dissipation is [1]
1 d d dw r = r dr dr dz
(here r y)
The velocity w is calculated as shown in Example F.2 and is used to solve for the temperature,
= w +
w2 k
r4 1 4 R
R 2 ( dp dx ) , w= 8
where w is the pipe wall temperature and w is the mean velocity. The temperature profile obtained from ADINA-F compares very well with the analytical solution. Reference [1] White, F.M., Viscous Fluid Flow, McGraw-Hill Book Company, New York, p. 130, 1974.
325
ADINA-F problems
F.41
Objective To demonstrate the capability of solving natural convection problems in porous media. Physical problem The problem of natural convection in a rectangular enclosure with a porous medium is to be analyzed, see Fig. F.41. Constant heat is generated in the medium. The fluid in the enclosure is subjected to gravity. The top and bottom walls are insulated while the two vertical walls have a constant temperature of zero. The fluid is free to flow in the tangential direction along the boundary of the enclosure while it cannot penetrate the boundary of the enclosure. Darcy's law and the Boussinesq approximation are assumed to be valid for the flow. A steady-state condition is assumed. The solution for this problem is governed by the Rayleigh number, see below. The case of Ra = 100 is considered here. Finite element model Due to symmetry, only the left half of the enclosure is modeled. A mesh of 10 40 equally sized 4-node FCBI elements is used. The symmetry boundary condition is applied along the vertical center line. Solution results The governing continuum mechanics equations are
vi = p,i + gi (1 ( 0 ) ) i % c p vi ,i = k ,i + q B vi ,i = 0,
( )
,i
where gi is the gravitational acceleration component in the i-direction, 0 is the reference temperature, i is the permeability in the i-direction, is the porosity, % k = k + (1 ) k is the effective thermal conductivity with k and ks being the thermal
s
conductivities of the fluid and solid respectively and qB is the uniform volumetric heat g q B L3 . generation rate. The Rayleigh number is Ra = % k
326
Good agreement between the finite element solution and the solution of [1] is observed along the vertical centerline of the enclosure. Reference [1] Haajizadeh, M., Ozque, A.F. and Tien, C.L., "Natural Convection in a Vertical Porous Enclosure with Internal Heat Generation," Int. J. Heat Mass Transfer, Vol. 27, No. 10, pp. 1893-1902, 1984. 1.0
w=q=0
qB
2.0
v=0 q = q0 g
v=0 q=0
v=0 q = q0
Figure F.41
327
ADINA-F problems
F.48
Objective To verify the 4-node FCBI axisymmetric elements in modeling conjugate heat transfer problems. Physical problem A section of pipe of infinite length is shown in Fig. F.48. The flow in the pipe is assumed to be described by the Hagen-Poiseuille parabolic velocity distribution. The external surface of the pipe is heated over a finite length L with constant heat flux . The unheated part of the external surface is assumed to be adiabatic. Dissipation due to friction is neglected. Finite element model The finite element mesh consists of 4-node FCBI elements. The region to be discretized includes the heated region and its extension to 5R both upstream and downstream. The length of the extension is selected such that the heat flux is close to zero at the upstream and downstream boundaries of the mesh in both the radial and the axial directions. The normal heat flux is prescribed at the external surface in the heated region while it is set to zero on all the other external boundaries. The reference temperature is set to zero at the origin. A constant pressure gradient is applied along the axial direction. The degrees of freedom of velocity and pressure in the solid are deleted. Solution results The heat transfer equation for the fluid is
c p w
where
= kf z
1 2 r + 2 (here r y) r r r z
2 r 2 dp R w = Wmax 1 , Wmax = dz 4 R
328
= 0 at r = 0, r
= 0 as z , z
= 0 as z r
1 s r r r r
2 s + 2 =0 z
= s , k f
= ks s r r
should be satisfied. For a fully developed pipe flow with the heat flux imposed along the complete pipe section which extends to infinity both upstream and downstream the analytical solution is (see ref. [1])
= 0 +
4 qw ( R + t ) Wmax R 2 c p
z+
( R + t ) qw r
kf
4 1 r 7 R 4 R 24
k f
4 qw ( R + t ) k f 2r r 3 = = constant, k f = ( R + t ) qw 2 4 2 z r Wmax R c p R R
329
ADINA-F problems
For a pipe with heat flux imposed along a finite length, the closed form solution is not known. However, when L is large, a fully developed thermal flow is observed in the heated section and the solutions given above are approximately correct. Good agreement is observed between the finite element solution and the analytical solution. Note that the analytical solution for the temperature is calculated with 0 = 0.49077 which comes from the ADINA-F solution. The heat flux at the mesh boundaries is zero as should be expected. Reference [1] Lienhard, J.H., A Heat Transfer Textbook, second edition, Prentice-Hall, Inc., Englewood Cliffs, NJ, 1987.
L = 10.0 t = 0.1 z y,r t
R
q=0
-
k t = 1.0 ks = 1.0
r = 1.0 R = 1.0
cp = 5.0 q=0
C L
q= 0
qw = 0 5.0
Figure F.48
330
F.52
Objective To demonstrate the analysis of flow past a cylinder. Physical problem The flow field around a cylinder is considered, see Fig. F.52, for two values of 2VR and V is the velocity of Reynolds number, Re = 20 and Re = 40, where Re = the uniform incoming flow. Finite element model A fine mesh of 3/4-node mixed FCBI elements is used. A steady-state analysis for Re = 20 and 40 is performed. Solution results The pressure acting on the cylinder is normalized as
p* =
p p0 1 V 2 2
where p0 is the pressure at infinity (the pressure at the most upstream point). The vortex length and the pressure on the cylinder are compared with experimental results [1] and good agreement is observed. User hint $ Since the shear stresses along the outlet boundary are unknown the mathematical traction formulation is employed. Reference [1] Batchelor, G.K., An Introduction to Fluid Dynamics, Cambridge University Press, pg. 259, reprinted 1981.
331
ADINA-F problems
Slip wall
R=1 v = 1.0
Figure F.52
332
F.62
Objective To demonstrate the use of ADINA-FSI in the solution of a one-way coupling of fluidstructure interaction problem. Physical problem A cantilever in a channel is subjected to the pressure of the wind blowing in the channel, see Fig. F.62. The fluid field around the cantilever as well as the deflections and stresses of the cantilever itself are to be solved. It is assumed that the deflection of the cantilever does not change the fluid field. Therefore, this is a one-way coupling of fluid-structure interaction problem. Steady state solutions corresponding to two pressure levels at the inlet are requested. Finite element model Since this is a one-way coupling of fluid-structure interaction problem, first the ADINA-F model is solved to obtain the flow field and then the ADINA model is solved to obtain the deformed shape of the cantilever. The finite element mesh for the ADINA-F model consists of 4-node FCBI elements. No-slip boundary conditions are applied on the wall of the channel and on the surface of the cantilever. The cantilever itself is not included in the mesh. Normal tractions are applied at the inlet boundary of the channel and the flow field is assumed to be unchanged at the downstream boundary by using the mathematical traction formulation. The iterative method of solution is used and reaction forces are calculated. Solution results The cantilever bends under the drag forces from the fluid. The fluid flow obeys the no-slip condition corresponding to the original configuration of the cantilever. User hint $ When the deformation in the cantilever is assumed to change the flow field, then the analysis can be performed as a two-way coupling of fluid-structure interaction problem. See Example F.76 for an example of this type of analysis.
333
ADINA-F problems
dp dy h
w dp -2 dy = -10 d = 40 h = 15 w= 1
Figure F.62
334
F.64
Objective To verify the behavior of the 4-node axisymmetric FCBI elements when a free surface is modeled. Physical problem The extrusion of a viscous jet from a round die is considered, see Fig. F.64. The behavior and shape of the free surface are of particular interest. Gravity effects are neglected in this example. The capillary number is Ca =
U = 3.568 , where is the surface tension and U is the Ur0 mean velocity in the die. The Reynolds number is Re = = 23.7 , where r0 is the radius of the die.
Finite element model The finite element model consists of 4-node FCBI elements. The no-slip condition is imposed on the wall. The flow is assumed to be a fully-developed Poiseuille flow at the in-flow boundary. The stress-free condition is imposed at the out-flow boundary except at the node on the free surface where surface tension exists. The free surface boundary condition is automatically imposed by the program on the elements with free surface. Solution results Fairly good agreement between the free surface profile from the finite element solution and the results from [1] is observed. User hints $ The physical traction formulation is used since the physical traction (the surface tension) is applied on the free surface. $ The finite element boundary at the downstream location has to be sufficiently far enough away to ensure a stress free boundary condition (i.e., a uniform velocity profile is reached).
335
ADINA-F problems
Reference [1] Gear, R.L., Keentok, M., Milthorpe, J.F. and Tanner, R.I., "The shape of low Reynolds number jets", Phys. Fluids, 26, pg. 7, 1983.
U= 3 4
r0
Figure F.64
336
F.66
Objective To verify the behavior of the 4-node axisymmetric FCBI elements with free surface in a transient analysis. Physical problem A water tank with an initial flat water surface is drained by an opening as shown in Fig. F.66. The flow rate is assumed to be constant. The flow field and the free surface shape are to be determined. Finite element model The finite element model consists of 4-node FCBI elements. The boundary conditions on the wall are such that the fluid can slip on but not penetrate the tank wall. The velocity at the opening is prescribed so that the constant flow rate is maintained. The iterative method of solution is used in this example. Solution results Reasonable agreement between the finite element solution and the solution from [1] is observed. Reference [1] Nitikitpaiboon, C. and Bathe, K. J., "An arbitrary Lagrangian-Eulerian velocity potential formulation for fluid-structure interaction", Computer & Structures, Vol. 47, No. 4/5, pp. 871-891, 1993.
337
ADINA-F problems
r = 1.0 m a = 0.2 m h = 0.35 m r = 1000 kg/m3 m = 10-3 kg/m sec g = 9.8 m/sec 2 flow rate = 0.31 m3/sec
Figure F.66
338
F.67
Objective To demonstrate the use of ADINA-F in a problem with multiple moving boundaries. Physical problem Two fluids with different viscosities flowing in two inclined ducts merge into a single duct whose upper wall terminates at a downstream location as shown in Fig. F.67. The fluids are assumed to be immiscible. The interface between the two fluids and the shape of the free surface which begins at the lip of the upper duct wall are to be determined. Surface tension effects are ignored in this example. Finite element model The initial mesh has a flat fluid-fluid interface and a flat free surface. The moving direction of the moving boundaries is in the Z-direction. Solution results Although no analytical or other solution is given, the solution looks reasonable.
free surface
m1 = 5 m2 = 4 r1 = 1 r2 = 1
Figure F.67
339
ADINA-F problems
F.69
Objective To verify the behavior of the 2-D 4-node FCBI elements. Physical problem Same problem as Example F.13, except that only Ra = 106 is considered. Finite element model The finite element model consists of 576 2-D 4-node FCBI elements. Two load steps are used to obtain a converged solution. Solution results The results are very close to those obtained using the 9-node elements. User hint $ The statistics regarding the use of two different element types for the same physical problem are: 4-node FCBI elements (this analysis) 625 576 2307 2 9-node quadratic elements (in Example F.13) 625 144 1851 5
340
F.76
Objective To demonstrate the use of ADINA-FSI to analyze a fully coupled fluid-structure problem. Physical problem A tower is tested in a wind tunnel with two levels of wind pressure, see Fig. F.76. The deformation of the tower is so large that its influence on the fluid field cannot be neglected. Steady-state solutions are to be determined. Finite element model The finite element mesh for the ADINA-F model consists of 4-node FCBI elements. No-slip boundary conditions are applied on the wall of the wind tunnel and on the surface of the tower. The no-slip boundary conditions applied on the wall of the wind tunnel are specified by applying a wall boundary condition, whereas the no-slip boundary conditions applied on the surface of the tower are specified by applying an FSI boundary condition. Normal tractions are specified at the inlet and the velocity derivatives at the outlet are set to zero by using the mathematical traction formulation. The finite element mesh for the ADINA model consists of 4-node plane strain elements. The mesh is subjected to pressures from the flow field calculated by the ADINA-F model. Solution results The tower bends under the drag forces from the fluid. The fluid flow obeys the noslip condition corresponding to the deformed configuration of the tower. User hint $ Since the vorticity and stress are not smooth, use of a finer mesh might be appropriate.
341
ADINA-F problems
z y
dp dy h
m,r
Figure F.76
342
F.77
Objective To demonstrate the use of the 2-D 4-node FCBI elements in problems with laminar flow. Physical problem The laminar flow in a channel with a backward-facing step is considered, see Fig. F.77. Two cases are studied. One is a flow with a Reynolds number of 200 (the length scale is based on the step height) and the other with a Reynolds number of 600. Finite element model The finite element mesh consists of 568 4-node FCBI elements. The no-slip boundary condition is imposed at the wall by deleting the velocity degrees of freedom. Solution results A recirculating region is observed downstream of the step. The length of the recirculating region increases as the Reynolds number is increased. User hint $ Enough of the channel downstream of the step must be included in the model so that the recirculating region is entirely contained in the model.
343
ADINA-F problems
8 r = 1 m = 1.43 10-5
Figure F.77
24
344
F.78
Objective To verify the k- turbulence model. Physical problem The same problem of Example F.77 is considered, except that only the case of the Reynolds number of 6.97 104 is studied with the assumption of the k- turbulence model. The material constants and the values of k and at the inlet are listed in Fig. F.78. Finite element model The finite element model is finer than the one used in Example F.77. It consists of 1500 4-node FCBI elements. Also the k and degrees of freedom are active and the no-slip boundary is specified explicitly as a wall boundary condition. Solution results A comparison of k and velocity against experimental results [1] shows that the calculated velocities compare quite well with the experimental results, but the calculated values of k are significantly lower that the measured values. This is deemed to be due to various effects, including the assumptions in the k- model. Reference [1] Kim, J., Kline, S.J. and Johnston, J.P., "Investigation of a Reattaching Turbulent Shear Layer: Flow Over a Backward-Facing Step", Vol. 102, Sept. 1980, Transactions of the ASME.
345
ADINA-F problems
24
Figure F.78
346
F.80
Objective To verify the behavior of the viscous compressible 2-D fluid flow elements. Physical problem A viscous compressible fluid flows past a semi-infinite long flat plate, see Fig. F.80. The plate is assumed to be adiabatic. The Prandtl number is 1, the Mach number is 0.4 and the Reynolds number is 1000. Finite element model The 2-D compressible fluid elements are used in the model. The no-slip and adiabatic boundary conditions are applied on the plate. The external condition with specified velocity, density and temperature is applied to the upstream and top boundaries. Along the downstream boundary, the flow condition is assumed to be uniform and therefore the (free) outlet condition is applied. Finally the symmetric condition is applied to the bottom upstream of the plate. Solution results The comparison of the finite element solution with the analytical solution [1] is good. User hints $ The initial conditions for the pressure and the temperature should always be specified for the compressible fluid calculation. $ In the compressible fluid calculation the following two conditions should be always satisfied to obtained a physically sound problem: The pressure and the temperature should always be positive. The coefficient cp should be always larger than cv.
$ In ADINA-F the normal traction and not the pressure should be specified as the boundary condition, except at the upstream boundary of the external flow, where the pressure as well as the velocities and the temperature should all be prescribed. $ The primary variables in the compressible fluid are velocities, pressure and temperature.
347
ADINA-F problems
$ The density, the local sound speed, the internal energy, the enthalpy and the local Mach number can be calculated from the primary variables:
cp p , c= ( c p cv ) , cv ( c p cv ) u c
e = cv ,
e + p = c p , M =
Reference [1]
Schlichting, H., Boundary-Layer Theory, 7th edition, McGraw-Hill Company, New York, pp. 333, 1979.
External condition: q = 4.464 u=1 r=1
Region discretized
Outlet condition
Plate Symmetric condition m = 1 10-3 cp = 3.5 cv = 2.5 k = 3.5 10-3 No-slip adiabatic condition
Figure F.80
348
F.81
Objective To verify the behavior of the viscous compressible fluid flow 3-D elements. Physical problem Same as Example F.80. Finite element model The 3-D compressible fluid elements are used in the model. The same boundary conditions and initial conditions as described in Example F.80 are applied on the 3-D model. The iterative solution method is employed. Solution results The comparison of the finite element solution with the analytical solution [1] is good. Reference [1] Schlichting, H., Boundary-Layer Theory, 7th edition, McGraw-Hill Company, New York, pp. 333, 1979.
349
ADINA-F problems
F.82
Objective To verify the behavior of the 2-D compressible fluid flow elements with the shock wave capture capability in a transient analysis. Physical problem An infinitely long channel contains an inviscid compressible fluid, see Fig. F.82. Initially there is no flow in the channel. Due to the existence of the pressure jump a flow is induced. A transient analysis is performed to determine the development of the flow. Finite element model The problem is essentially one-dimensional and only one element is used in the direction perpendicular to the flow. Only a finite length of the channel is included. The boundary condition on the wall of the channel is that the fluid can not penetrate but can slip on the wall. The two open ends of the channel are also modeled as if they were walls. This is because during the analysis no flow is induced near the open ends. Sixty time steps with a time increment of 0.015 are used in the transient analysis. Solution results The comparison of the finite element solution with the analytical solution is good. User hint
$
To simulate the inviscid compressible fluid flow specify a zero viscosity coefficient, a zero thermal conductivity coefficient and a slip but nonpenetrable boundary condition on the wall.
350
Figure F.82
351
ADINA-F problems
F.83
Objective To demonstrate the use of ADINA-FSI to solve a fluid-structure interaction problem. Physical problem A parachute is freely falling in the air as shown in Fig. F.83. The shape of the parachute as well as the surrounding fluid field is to be predicted at the steady state condition. Finite element model A transient analysis of the fluid-structure interaction for the free-falling parachute is performed until steady state is reached. Axisymmetric conditions are assumed. The finite element mesh for the ADINA-F model consists of mixed 3/4-node FCBI elements. No-slip boundary conditions are applied on the parachute. The inlet velocities are specified along the bottom line of the fluid domain. The finite element mesh for the ADINA model consists of axisymmetric shell elements and truss elements. The only load acting on the parachute is from the fluid flow and is calculated by the ADINA-F model. One end node of the truss element is fixed while the other end node connected to the parachute is free. At the axisymmetric centerline the node is fixed in the y-direction. Solution results During the analysis, the parachute forms a rounded shape and circulation is developed in the surrounding air. There is a noticeable pressure difference between the upper and lower face of the parachute.
352
Figure F.83
353
ADINA-F problems
F.84
Objective To demonstrate the usage of the specular-diffuse-transparent radiosity boundary condition with 2D FCBI elements. Physical problem The domain includes two concentric cylinders, see Fig. F84. Initially, the inner cylinder has a higher temperature, while the outer one has a lower temperature. Note that the temperature at the inner surface of the outer cylinder is prescribed. The temperature at the inner cylinder will decrease due to radiation exchange between the cylinders. Finite element model Two-dimensional FCBI elements which are treated as solid are used in this model. The heat transfer option is turned on. The specular-diffusive-radiation boundary condition is assigned at both the outer surface of the inner cylinder and the inner surface of the outer cylinder. A high uniform temperature is prescribed at the inner cylinder as an initial condition. A low temperature(not changing with time) is prescribed at the inner surface of the outer cylinder as a boundary condition. A transient analysis is performed. Solution results The time history of the temperature at the outer surface of the inner cylinder shows the effect of the radiation exchange. The result looks reasonable.
354
Low Temperature
High Temperature
Figure F.84
355
ADINA-F problems
F.85
Objective To demonstrate the usage of the specular-diffuse-transparent radiosity boundary condition and radiation boundary condition with 3D FCBI elements. Physical problem Two parallel square plates have different temperature initially, see Fig. F.85. The temperature history at the two plates due to the radiation exchange is to be found. Finite element model Three-dimensional FCBI elements which are treated as solid elements with heat transfer are used in the finite element model. The specular-diffuse-transparent radiosity boundary condition is prescribed at the inner surfaces of the two plates. The radiation boundary condition is prescribed at the outer surfaces of the two plates. Initially, a uniform, high temperature is prescribed at the upper plate, while a low temperature is prescribed at the lower plate. A transient analysis is performed. Solution results The time history of the temperature at the two plates shows the effects of the radiation exchange. The results look reasonable.
356
High temperature
Low temperature
Figure F.85
357
ADINA-F problems
F.86
Objective To demonstrate the mass transfer capabilities of ADINA-F for a problem involving two species and fluid. Physical problem Two species flowing in two inclined ducts merge into a single duct, see Fig. F.86. The diffusion of the two species is to be determined. Finite element model 4-node planar FCBI elements are used. There are two species (plus the fluid) in this model. At both the upper inlet and lower inlet, the mass ratio of two species are specified separately. A parabolic flow velocity is assigned at both the upper inlet and the lower inlet using a spatial function. Skew systems are used at the two inlets to enforce that the flow direction is the normal direction of the inlet surface. The effect of the species on the fluid is neglected. The flow domain is divided into three parts. The mass diffusive coefficient inside each part is assumed to be constant. A steady analysis is performed. Solution results Although no analytical or other solution is given, the solution looks reasonable.
358
Figure F.86
359
ADINA-F problems
F.87
Objective To verify the behavior of the 4-node axisymmetric FCBI elements, especially with mass transfer. Physical problem The problem of a fully developed incompressible flow in a pipe is considered with mass transfer, see Fig. F.87. A parabolic velocity is applied in the axial direction, and a species is injected at r = 0.5R (R is the radius of the pipe, r is the radial position of the injection). The pressure drop and the species diffusion along the axial direction are to be found. Finite element model 4-node axisymmetric FCBI elements are used to simplify the model from 3D to 2D. There are two materials in this model, one is the fluid, another one is the species. The effect of the species on the fluid is neglected. A parabolic velocity distribution is prescribed at the inlet using two spatial functions. The species injection is modeled by applying a zero mass ratio to the two lines at the inlet and a unit mass ratio to the middle point at the inlet. A steady analysis is performed. Solution results The pressure drops linearly in the axial direction. This is the same as the analytical solution. The diffusion of the species in the axial direction is observed.
360
v= w=0
v=0
v = 0, w = 1-y , =
0 otherwise
1 y = R/2
Figure F.87
361
ADINA-F problems
F.88
Objective To demonstrate the capability for solving slightly compressible flow problems. Physical problem The fluid inside a cylinder is compressed at the top end, see Fig. F.88. The bottom end and the side wall are fixed. The top end moves down two-thirds the length of the cylinder uniformly between time 0.0 and time 1.0. Then it stays at the final position. The flow is assumed to be laminar and axisymmetric. Finite element model This problem is solved using the two-dimensional slightly compressible flow model. In the model, axisymmetric FCBI elements are used to simplify the model from 3D to 2D. The velocity at the bottom is set to be zero and the y-velocity at the center line is fixed as a symmetric boundary. The side wall is treated as a nonslip wall. The top wall is treated as a moving wall, in which the displacement is prescribed using a time function. The fluid bulk modulus ( ) is specified as a material constant to model the compressibility of the fluid. A transient analysis is performed. Solution results The finite element and analytical solutions for the time history of the pressure inside the cylinder are identical. A uniform pressure and a vortex are found inside the cylinder at the final time step. User hint $ In general, in order to obtain an accurate transient solution of the acoustic wave c t 1 ~ 10 , where motion, the time step length should be chosen such that x
c = and is the density of the fluid. However, in this problem, there are
two main time scales, one from the acoustic wave and the other from the moving piston. Due to the fact that the acoustic wave is much faster than the piston, these two scales differ significantly. Since the major concern in this case is to simulate the piston, the time step length is chosen to resolve the piston moving process. c t This time step length 8200 is much larger than the one that is needed to x resolve the acoustic wave. Therefore, the simulation is more like a quasi-steady
362
case for the acoustic wave. One can see that the pressure inside the cylinder is almost uniform at every time step.
v=0
v=w=0
Figure F.88
nonslip wall
363
364
ADINA-T problems
365
ADINA-T problems
366
T.1
Objective To verify the two-dimensional conduction element when used for frequency analysis. Physical problem The acoustic cavity considered for analysis is shown in Fig. T.1. The cavity is bounded by rigid walls and filled with water which is assumed to be inviscid. The natural frequencies of the cavity are to be determined. The wave equation governing the motion of the fluid inside the cavity is given by [1]:
2 2 1 2 + = , c= y 2 z 2 c 2 t 2 = 0 at the boundary n
where is the velocity potential, c is the velocity of sound in water, is the bulk modulus and is the density of water. The natural frequencies of the cavity can be obtained by performing an eigenvalue analysis of a planar heat flow problem in which the conductivity k and specific heat c correspond to and respectively. Finite element model Four 8-node planar conduction elements are used to model the cavity. The degrees of freedom of all the nodal points are left free corresponding to the boundary conditions of the cavity. In the finite element frequency analysis, a consistent heat capacity assumption is used. Solution results The lowest four frequencies predicted by the finite element model are shown in the table below. Analytical solutions evaluated using formulas given in [2] are also shown for comparison.
367
ADINA-T problems
n m * = c + , m, n = 0,1, 2,3,... a b
User hints $ Since only the gradient of the potential is zero at the boundaries while the potential is free, there is one zero frequency (rigid body) mode in this problem. $ This problem can also be solved using ADINA, see Example A.65. References [1] [2] Landau, L.D. and Lifshitz, E.M., Fluid Mechanics, translated by Sykes, J.B. and Reid, W.H., Pergamon Press, pp. 245-246, 1959. Blevins, R.D., Formulas for Natural Frequency and Mode Shape, Van Nostrand Reinhold, pp. 337-341, 1979.
Rigid boundary
Water
b
a = 12 in b = 20 in
Figure T.1
368
T.2
Objective To verify the behavior of the 1-D conduction element in a transient analysis. Physical problem A uniform infinite solid is initially at zero temperature and suddenly exposed to a constant uniform surface heat flow input at time t = 0+, see Fig. T.2. Finite element model Fifteen equal length 1-D conduction elements are used. A lumped heat capacity representation and the Euler backward method for the time integration are employed. Solution results The calculated temperature distributions are compared with the analytical solution [1] and good agreement is observed. User hints $ Note that the same problem can also be analyzed using the 2-D and 3-D elements, but since only one-dimensional heat flow shall be predicted, it is most appropriate to employ the 1-D elements in the idealization. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, 1959.
369
ADINA-T problems
Region discretized
Area = 1 in2
3.0"
Figure T.2
370
T.3
Objective To verify the behavior of the two-dimensional planar conduction element in a steadystate temperature analysis. Physical problem A square column is subjected to 100 F on one side and to 0 F on the other sides, see Fig. T.3. Finite erlement model Using symmetry conditions, only one-half of the column is modeled. The model consists of eight 8-node planar conduction elements and the conductivity of the material is assumed to be temperature independent. Solution results The calculated temperature distribution along the centerline of the column is compared with an analytical solution [1] and good agreement is observed. User hints $ Nodes with temperatures of zero degrees can be deleted as degrees of freedom. If nodal temperatures are specified to be different from zero, prescribed temperatures must be used. $ The mesh used in this analysis gives a reasonably accurate temperature solution along the centerline. However, high temperature gradients are present at the lower left-hand corner of the column. To accurately model this region, a finer mesh should be used. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, 1959.
371
ADINA-T problems
4 in
100F
0F
Region discretized
0
4 in
Figure T.3
372
T.4
Objective To verify the behavior of the axisymmetric conduction element when subjected to a uniform surface temperature. Physical problem A long hollow cylinder is subjected to a uniform inside surface temperature of 100 F and a uniform outside surface temperature of 0 F, see Fig. T.4. The cylinder is assumed to be insulated to prevent axial heat flow. Finite element model The model consists of ten 4-node axisymmetric conduction elements and the material is assumed to have constant isotropic conductivity and constant specific heat. Solution results The solution results are compared with the analytical solution [1] and good agreement is observed. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, 1959.
Z
Region discretized
q = qi
h
q = qo
ri = 1.0 in q i = 100 F ro = 2.0 in qo = 0 F h = 0.1 in k = 1.0 Btu/in-sec-F
ri
ro Y
Figure T.4
373
ADINA-T problems
T.5
Objective To demonstrate the capability of performing a temperature and thermal stress analysis using the ADINA-T and ADINA programs. Physical problem Same as in Example B.17. Finite element model The cantilever beam is modeled using six 20-node three-dimensional solid elements. In the temperature analysis using ADINA-T the temperature distribution through the thickness is calculated and the resulting nodal point temperatures are written to the temperature file. A thermal stress analysis is then performed using ADINA in Example B.38. The element conductivity matrices in ADINA-T are calculated using 2 2 2 Gauss integration. Solution results The correct temperature and heat flux results are predicted by ADINA-T. User hint $ The temperature solution times written to the temperature file by ADINA-T need not correspond to the times for which the stresses are calculated using ADINA. ADINA automatically establishes the nodal temperature values at the required ADINA time points by linearly interpolating the temperatures written by ADINA-T to the temperature file at successive ADINA-T time points.
374
T.6
Objective To verify the transient behavior of the 1-D conduction element in thermal eigenvalue analysis. Physical problem A uniform infinite liquid is considered as shown in Fig. T.6. The conductivity, specific heat and density of the liquid are assumed constant. The thermal eigenvalues and mode shapes are to be determined for a column of the liquid. Finite element model The finite element model consists of eight equally spaced 1-D 2-node conduction elements. A lumped diagonal heat capacity matrix is employed in the analysis. Solution results The thermal eigenvalues are
n k n = , n = 0,1, 2,... L c
with associated mode shapes
2 nx exp ( n t ) cos cL L
The following results are obtained: Mode 0 1 2 Theory 0 0.6662 2.665 ADINA-T 0 0.6577 2.531 at node 1 Theory ADINA-T 0.5 0.5 0.7071 0.7071 -0.7071 -0.7071
375
ADINA-T problems
1
Insulated
Figure T.6
376
T.7
Objective To verify the behavior of the two-dimensional conduction element when applied to a torsional problem. Physical problem The square shaft shown in Fig. T.7 is analyzed for its elastic torsional behavior. The torsional rigidity of the shaft, defined as the ratio of the applied torque to the angle of twist per unit length, is to be determined. The governing equation for the elastic torsional behavior is given in [1], art. 104:
2 2 + + 2G. = 0 y 2 z 2
where is a stress function, G is the material shear modulus and is the angle of twist per unit length. This equation is analogous to the heat transfer equation governing planar heat flow:
B ky + kz +q = 0 y y z z
where ky and kz are the thermal conductivities corresponding to the principal axes y and z respectively, is the temperature in the body and qB is the internal heat generation. Finite element model Four 8-node planar conduction elements with unit conductivity are used in the finite element model. In the analysis, qB is set equal to 2 and the resulting nodal point temperatures i correspond to i for the torsional problem. The condition of a G stress-free boundary is imposed by setting the temperatures of the boundary nodal points to zero.
377
ADINA-T problems
Solution results The following results are obtained in the finite element analysis:
9 = 12 = 14 = 17 = 1.94118 G, 13 = 2.11765 G
Following the procedures given in Example 7.3 in ref. [2], the torsional rigidity is predicted to be 35.8 G. The theoretical solution, given in [1] art. 109, is 36.0 G. User hints $ Due to symmetry conditions it would be sufficient to model only one quarter of the cross-section of the shaft. $ The shear stresses can easily be obtained from the calculated heat fluxes within the elements, see [2] p. 664. References [1] [2] Timoshenko, S.P. and Goodier, J.N., Theory of Elasticity, 3rd Edition, McGraw-Hill, 1970. Bathe, K.J., Finite Element Procedures, Prentice-Hall, pp. 642-670, 1996.
4"
T
4"
a
Figure T.7
378
T8. Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using Euler backward method)
T.8
Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using Euler backward method)
Objective To verify the 2-D conduction element for transient analysis using the Euler backward method. Physical problem Fig. T.8 shows the 90 semi-infinite corner considered. The two sides AB and BC are subjected to a prescribed temperature s = 50 and the initial temperature within the semi-infinite domain is 0 = 0 . The heat capacity of the material is constant. A transient analysis is performed to calculate the temperature distribution within the semi-infinite domain at different values of time. Finite element model The domain is discretized using a 10 10 mesh of 4-node 2-D conduction elements. The conduction matrix is evaluated using a consistent heat capacity matrix. The time step is t = 0.016. Solution results Good agreement between the calculated temperatures and the analytical solution [1] is observed. User hint $ A larger value of time step t can also be employed for analysis because the Euler backward time integration scheme is unconditionally stable. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, 1959.
379
ADINA-T problems
k = 35.0 c = 100.0 A
Region discretized
qs = 50 0.75 y
qo = 0
x qs = 50 0.75
Figure T.8
380
T9. Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using trapezoidal rule)
T.9
Transient analysis of a semi-infinite corner subjected to prescribed temperature on the boundary (using trapezoidal rule)
Objective To verify the 2-D conduction element for transient analysis using the trapezoidal rule (Euler method, = 0.5). Physical problem Same as Example T.8. Finite element model Same as Example T.8. Solution results Good agreement with the analytical solution is observed. User hint $ A larger value of time step t can also be employed for analysis because the trapezoidal rule of time integration is unconditionally stable.
381
ADINA-T problems
T.10
Semi-infinite region subjected to constant heat flux on the boundary (Euler forward method)
Objective To verify the one-dimensional conduction element in transient analysis using Euler forward integration. Physical problem Fig. T.10 shows the semi-infinite domain considered. The boundary at x = 0 is subjected to a constant heat flux. An analysis is performed to calculate the temperature distribution within the domain at different values of time. Finite element model The finite element model consists of eight 1-D conduction elements. A lumped specific heat matrix is employed and the time step is t = 0.015. Solution results Reasonable comparison with the analytical solution [1] is observed considering the coarseness of the mesh. User hints $ The Euler forward method is conditionally stable. The time step t is chosen such that
( x )
4
where x = shortest element length and = thermal diffusivity = k/c. In this problem x = 0.250 and = 1.0. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, 1959.
382
T10. Semi-infinite region subjected to constant heat flux on the boundary (Euler forward method)
k = 1 c = 1 qs = 1.0
A = 1.0 q0 = 0
qs
Region discretized
x
2
Figure T.10
383
ADINA-T problems
T.11
Objective To verify the 1-D conduction element with the option of internal heat generation. Physical problem Fig. T.11 shows the semi-infinite domain considered. The domain surface at X = 0 is insulated. The internal heat is only generated in the region 0 X B. A transient analysis is performed to calculate the temperatures within the domain at different values of time. The material conductivity and specific heat are constant over the entire temperature range of interest. Finite element model The finite element model consists of 16 1-D conduction elements. The heat generation region, 0 X B, is discretized using two elements. The time integration is performed using the trapezoidal rule ( = 0.5). The time step used is t = 0.0016 and a lumped specific heat matrix is used. Solution results A good comparison with the analytical temperature solution of ref. [1] is observed. Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, p.80, 1959.
384
T11. Analysis of a semi-infinite region subjected to internal heat generation (trapezoidal rule)
q B = 1500
A = 1.0
dq =0 dx
Region discretized
Figure T.11
385
ADINA-T problems
T.12
Analysis of a semi-infinite solid subjected to internal heat generation (Euler forward method)
Objective To verify the 1-D conduction element with the option of internal heat generation and with the time integration performed using the Euler forward method. Physical problem Same as Example T.11. Finite element model Same as Example T.11. The time step employed is t = 0.0004. Solution results A good comparison with the analytical solution is observed. User hints $ The Euler forward method is conditionally stable. The time step t is chosen such that
t =
( x )
4
where x = shortest element length and = thermal diffusivity, In this example, x = 0.02 and = 0.25.
386
T.51
Objective To verify the behavior of the 1-D conduction element in ADINA-T when subjected to simultaneous boundary convection and radiation in a transient analysis. Physical problem The slab shown in Fig. T.51 is considered. The slab is initially at a uniform temperature i and at time t = 0+ the slab surfaces are exposed to convection and radiation. The time history of the temperature at the surface and at the center of the slab is to be determined. The surface temperature and the temperature at the center of the slab depend upon two parameters: the Biot number Bi and the radiation parameter . In this analysis Bi = 4 and = 4. Finite element model Because of symmetry conditions only one-half of the slab is modeled. The finite element model consists of twenty equally spaced 1-D 2-node conduction elements. The thermal conductivity and the heat capacity of the material are assumed to be constant. A diagonal heat capacity matrix is used. The transient temperature response is evaluated using the Euler backward method in the step-by-step analysis. Conductivity reformations and heat flow equilibrium iterations using the modified Newton-Raphson method are performed at each time step. In order to obtain the required accuracy in the ADINA-T analysis, the time step size is changed during the response prediction, as shown in the table below.
t* =
t L2
t * =
387
ADINA-T problems
Solution results The ADINA-T solutions are compared with solutions obtained by Haji-Sheikh and Sparrow [1] who used a probability method. Good agreement is observed. User hint $ The same problem has also been solved with ADINA-F, see Example F.26. Reference [1] Haji-Sheikh, A. and Sparrow, E.M., "The Solution of Heat Conduction Problems by Probability Methods," Trans. ASME, J. Heat Transfer, Vol. 39, pp. 121-131, 1967.
2.0
Region discretized
Conductivity Specific heat Thermal diffusivity Convection coefficient Shape factor Emissivity constant Boltzmann's constant Radiation parameter Biot number Figure T.51
k = 0.01 Btu/Fhrin cp = 0.01 Btu/Fslug = k/cp = 1.0 in2/hr h = 0.04 Btu/Fhrin2 f = 1.0 = 1.0 (black body) = 1.1895810-11 Btu/in2hrR4 3 f 1 L = =4 k hL Bi = =4 k
388
T.52
Objective To verify the behavior of the 1-D conduction element using nonlinear material models in a transient analysis. Physical problem The space shuttle orbiter thermal protection system shown in Fig. T.52 is to be analyzed. The protection system is composed of different nonlinear materials and the thermophysical properties of these materials are given in the input data for this problem. The uniform initial temperature is 322.22 K. At time t = 0+ a step heat flow input is imposed on the surface of the thermal protection system and maintained for 100 sec. after which no heat flow input is imposed. Simultaneously, the surface of the protection system is exposed to radiation with a sink of absolute zero temperature. It is desired to predict the transient surface temperature of the protection system. Finite element model The finite element model consists of nineteen 1-D conduction elements and one boundary radiation element. A lumped heat capacity matrix is used. Conductivity reformations and heat flow equilibrium iterations using the modified NewtonRaphson method are performed at each step of the analysis. In order to obtain the required accuracy in the ADINA-T analysis, the time step size is changed during the response prediction. The table below gives the time step size selections using the -family time integration method with = 0.7. t (sec) 0 t 15 15 < t 100 100 < t 115 115 < t 140 Solution results This problem was also solved by Williams and Curry who used the finite difference method [1] and excellent agreement is observed. t (sec) 2.5 5.0 2.5 5.0 Number of time steps 6 17 6 5
389
ADINA-T problems
References [1] Williams, S.D. and Curry, D.M., "An Implicit-Iterative Solution of the Heat Conduction Equation with a Radiation Boundary Condition," Int.J. Num. Meth. Eng., Vol. 11, pp. 1605-1620, 1977.
RTV-560
Felt
50
Figure T.52
390
T.53
Objective To verify the 1-D conduction element in ADINA-T when employed in transient conduction with phase changes. Physical problem A uniform infinite slab of liquid is initially at zero temperature, see Fig. T.53. Suddenly the temperature of the surface of the liquid is reduced to -45 F and maintained constant. The freezing front position and the temperature in the slab are to be predicted. Finite element model The model consists of thirty-two 2-node equally spaced 1-D elements. The conductivity, specific heat, and density of the liquid and solid phases are assumed equal and constant. A lumped heat capacity matrix is employed. The Euler backward method of time integration with a time step of 0.1 sec. is used in the finite element analysis. Equilibrium iterations are employed at every time step and the iteration convergence tolerance is 0.0001. Solution results Reasonable agreement between the finite element solution and the analytical solution [1] is observed. Reference [1] Morgan, K., Lewis, R.W. and Zienkiewicz, O.C, "An Improved Algorithm for Heat Conduction Problems with Phase Change," Int. J. Num. Meth. Eng., Vol. 12, pp. 1191-1195, 1978.
391
ADINA-T problems
1 in 0 qs = -45F
qi = 0 F x
4 in Material properties: k = 1.08 Btu/in-sec-F rc = 1.0 Btu/in3 -F rL = 70.26 Btu/in3 (latent heat capacity) qf = -1F (freezing temperature)
Figure T.53
392
T.54
Objective To verify the 2-D conduction element in ADINA-T when employed in transient heat conduction. with phase changes. Physical problem The corner of a uniform infinitely long container carrying a liquid with initial temperature and freezing temperature f is considered, see Fig. T.54. At time t = 0, the temperature of the container surface is reduced to a temperature lower than f and is maintained constant. The solution for the temperature and the position of the phase transition region is to be determined. Finite element model A 10 10 mesh of four-node planar linear conduction elements is used to model the corner of the liquid. A lumped heat capacity assumption must be used in the model since phase change effects are included. The Euler backward method of time integration with a time step of 0.01 sec. is employed in the step-by-step solution. Equilibrium iterations are performed at each time step using the modified NewtonRaphson method and the standard tolerance 0.001 is used. Solution results Reasonable agreement between the finite element solution and the analytical solution [1] is observed. Reference [1] Budhia, H. and Krieth, F., "Heat Transfer with Melting or Freezing in a Wedge", Int. J. Heat Mass Trans., Vol. 16, 195-211, 1973.
393
ADINA-T problems
q = 0 n
Region discretized
Fluid
qs
q = 0 n
Container
qs = -1 1
Figure T.54
394
T.55
Objective To verify the two-dimensional planar conduction element when applied to an electric conduction problem. Physical problem The V-notched bending specimen shown in Fig. T.55 is analyzed for a crack propagating downward from the notch. This specimen has experimentally been analyzed (Ritchie [1]) by passing an electric current through the specimen and measuring the voltage across the crack. The following analogies are used in this analysis: temperature is analogous to voltage (field potential), heat flux is analogous to current. Finite element model By symmetry, only one-half of the specimen is modeled. The model consists of 8node isoparametric planar conduction elements, some of which are degenerated into 6-node triangles. Along the line of crack propagation, convection elements with a large convection coefficient are used to apply the voltage boundary condition of 0 volts. These convection elements are removed using the element death option to simulate crack growth. In this way, the voltages for several crack lengths are obtained in one ADINA-T run. Solution results The ratio Vc a Vc is plotted as a function of crack length for the ADINA-T results and Ritchie's experimental data. Good agreement is observed. Reference [1] Ritchie, R.O. and Bathe, K.J., "On the Calibration of the Electrical Potential Technique for Monitoring Crack Growth Using Finite Element Methods," Int. J. of Fracture, Vol. 15, No. 1, pp. 47-55, February 1979.
395
ADINA-T problems
45 5.2 in
c
Propagating crack
L/2 L = 100 in w = 20 in c = 5 in
Line of symmetry
Figure T.55
396
T.56
Objective To verify the seepage model for analysis of unidirectional seepage flow. Physical problem Three layers, each of different permeability, are placed on top of each other, see Fig. T.56. At the top face (Face A), the total seepage head is prescribed to equal 20.0 and at the bottom face (Face B), the total seepage head is prescribed to equal 8.0. Finite element model Two separate models are considered: Model 1: The 2-D conduction elements are used to discretize the domain of interest. Each layer is represented by one 4-node element. Model 2: The 3-D conduction elements are used to discretize the domain of interest. Each layer is represented by one 8-node element. Solution results The ADINA-T numerical solution obtained for seepage head matches exactly with the analytical solution which is obtained from Darcy's law:
d =q dz
where k = permeability, q = rate of seepage flow and = total seepage head. In this case z =8.0 = 18.29, z = 4.0 = 14.86 , and q = 0.857.
397
ADINA-T problems
4.0 Layer 1 k1 = 2.0 Layer 2 k2 = 1.0 z y Face B total head (prescribed) = 8.0 Layer 3 k3 = 0.5
4.0
4.0
Figure T.56
398
T.57
Objective To perform an unconfined flow seepage analysis using 2-D conduction elements. Physical problem An analysis of steady-state free surface seepage through the rectangular dam shown in Fig. T.57 is considered. The upstream and downstream water levels are maintained at sixteen and zero feet respectively, and isotropic conditions with a constant permeability are assumed. Finite element model The rectangular dam is discretized using an 8 8 mesh of 2-D conduction elements. 3 3 Gauss integration is employed. Solution results Good agreement is observed between the calculated free surface and the results of [1,2]. Reference [1] [2] Herbert, R., "Time Variant Ground Water Flow by Resistance Network Analogies," J. of Hydrol., Vol. 6, pp. 237-264, 1968. France, P.W., Parekh, C.J., Peters, J.C. and Taylor C., "Numerical Analysis of Free Surface Seepage Problems," Proceedings of the American Society of Civil Engineers, J. Irrigation and Drainage Division, Vol. 97, pp. 165-179, 1971.
399
ADINA-T problems
16 ft
Water level
Permeability ky = kz = 1 ft/hr Specific weight g = 62.3 lb/ft3 Water
16 ft
Dam impermeable
Figure T.57
400
T.58
Objective To perform an unconfined flow seepage analysis using 2-D conduction elements. Physical problem The steady-state free surface seepage through the dam shown in Fig. T.58 is considered. The dam material is isotropic. Finite element model The mesh consists of 90 4-node finite elements. The permeability matrix for the domain of interest is evaluated using 3 3 Gauss integration. Solution results Good agreement is observed between the calculated free surface seepage line and the solution from ref. [1]. User hint $ In the starting iteration vector for the seepage head, (0) , each entry not prescribed by the boundary conditions is taken to be equal to 4.0, which corresponds to the maximum value of the seepage head in this problem. Reference [1] Harr, M.E., Groundwater and Seepage, McGraw-Hill, New York, 1962.
Region discretized Isotropic material: ky = kz = 2.5 ft/hr
45
Water level
4 ft
45 Impervious
12.64 ft 16 ft
Figure T.58
401
ADINA-T problems
T.59
Analysis of a semi-infinite region subjected to a constant surface temperature undergoing two phase changes
Objective To verify the one-dimensional conduction element for analysis of phase changes. Physical problem The semi-infinite region shown in Fig. T.59 is subjected to a prescribed surface temperature that is lower than the initial temperature of the region. The material includes two phase changes. Finite element model The finite element model consists of forty 1-D conduction elements. The temperature degree of freedom at node 1 is deleted to impose the zero surface temperature. The element conduction matrix is evaluated using one point integration and a lumped specific heat matrix is employed. The time integration is performed using the Euler backward method and the time step is t = 0.04. Solution results Good agreement between the calculated temperature distribution and the analytical solution of ref. [1] is observed. User hints $ The material conductivity and specific heat are assumed to remain constant in the entire range of temperatures from 0 to 100. As a result the semi-infinite domain is discretized using linear one-dimensional conduction elements with constant material properties. $ In practice, if the conductivity and specific heat of the material vary with temperature, then the corresponding finite element model can be obtained using a nonlinear material description.
402
T59. Analysis of a semi-infinite region subjected to a constant surface temperature undergoing two phase changes
Reference [1] Carslaw, H.S. and Jaeger, J.C., Conduction of Heat in Solids, 2nd ed., Oxford University Press, Inc., N.Y, p. 290, 1959
A = 1.0
qo = 100 qs = 0
First phase change: qf = 50 Dqf = 0 Second phase change: qf = 90 Dq f = 0
8.0
Region discretized
k = 1.0 c = 1.0
Figure T.59
403
ADINA-T problems
T.60
Objective To verify the behavior of radiation surface elements, including low order and high order elements, in modeling the radiation exchange between diffuse-grey reflecting surfaces. Physical problem Radiation between two parallel plates is considered, see Fig. T.60. The radiation surfaces are assumed to be grey and to have the diffuse reflection property. The radiosity distributions on the plates are to be calculated. Finite element model Three separate models are considered in the same analysis: one pair of four 4-node surface elements, one pair of four 9-node surface elements and one pair of one 16node surface elements. Gauss integration orders of 2, 3 and 4 are used in each model to calculate the radiosity and heat flux. No conduction elements are needed for the steady-state analysis. Solution results The nodal radiosities along the plate diagonal are compared with those given by Sparrow et al.[1]. The result demonstrates that the higher order element gives the better solution; indeed just one 16-node element gives a reasonable result. User hints $ When the distance between two plates becomes smaller, more elements should be used for the radiosity calculations. $ If only the radiosity solution is sought, all the nodal temperatures must be prescribed. $ In the model preparation, the radiation surfaces are attached to shell conduction elements. These elements, which would be used in analyses in which conduction effects are important, are deleted in this analysis before the end of the model preparation.
404
Reference [1] Sparrow, E. M. and Haji-Sheikh, A., "A Generalized Variational Method for Calculating Radiant Interchange Between Surfaces", Journal of Heat Transfer, Vol. 87, pp. 103-109, Feb. 1965.
L _ 2
plate 1
q1 = 66.534K e1 = 0.9
Figure T.60
405
ADINA-T problems
T.62
Objective To verify the behavior of the diffuse-grey radiation element in two-dimensional transient analysis. Physical problem Two infinitely long cylinders are considered, see Fig. T.62. The temperature of the outer cylinder is kept constant at 530 R (70 F) and the initial temperature of the inner cylinder is 660 R (200 F). The temperature of the inner cylinder changes due to the radiation exchange between the two cylinder surfaces. Finite element model The whole geometry (cylinder cross sections) needs to be included in the finite element model to obtain a thorough visibility checking. The inner cylinder body is represented by 10 (radial) 80 (circumferential) planar conduction elements. The surfaces of the inner and outer cylinders are represented by 80 and 200 planar grey-diffuse radiation elements. Since the material properties (conductivity, heat capacity and emissivity) are constant, the matrices are not reformed in the step-by-step analysis. The nonlinearity of this problem lies in the radiation exchange between cylinder surfaces and so equilibrium iterations must be performed at each step. The Euler backward time integration method with two time blocks (t=0.1 for the first 100 time steps and t=0.5 for another 20 steps) is used. Solution results The calculated result predicts that the temperature of the inner cylinder approaches the temperature of the outer cylinder, as anticipated. The conductivity effect in the cooling process is also observed in the calculated solution.
406
e = 0.5
e = 0.7
0.25 ft
0.5 ft
s (Boltzmann's constant) = 1.73 10
Figure T.62
-8
407
ADINA-T problems
T.63
Unconfined seepage through a gravity dam, 3-D elements and constraint equations
Objective To verify the 3-D conduction elements in seepage analysis and to verify the constraint equations. Physical problem Same as in Example T.58. Finite element model The mesh consists of 90 8-node conduction elements, with the same two-dimensional layout as in Example T.58. The nodes in the X=-1 plane are constrained (thermal coupling) to the corresponding nodes in the X=0 plane. Therefore, two-dimensional solutions should be obtained. Solution results Same as Example T.58.
408
409
ADINA-T problems
L L
Figure T.64
410
T.65
Objective To verify the usage of the user-supplied material model in ADINA-T. Physical problem Same as in Example T.57. Finite element model The same finite element mesh as in Example T.57 is used. However, the user-supplied material is employed to predict the two-dimensional seepage flow. The user subroutine TUSER2 which provides the permeability coefficient and calculates the seepage is provided along with the input to this problem. TUSER2 needs to be compiled and linked with the rest of the ADINA-T program. Solution results The results are identical to the solutions obtained in Example T.57.
411
ADINA-T problems
T.66
Objective To demonstrate the capability of the condensation effect in the boundary convection elements. Physical problem An axisymmetric structure is considered, see Fig. T.66. The cylinder end A-B and its side A-D are subjected to condensation/convection conditions. The initial temperature of the structure is 20E C. The temperature distributions of the structure due to condensation/convection effects are to be calculated. Finite element model The structure is represented by 10 10 axisymmetric elements and each condensation/convection surface is modeled by ten axisymmetric convection elements. Euler backward time integration is performed with the selection of three time blocks, t=1 for the first 60 steps, t=2 for the next 30 steps and t=5 for the next 36 steps. Whether the condensation or convection takes place is calculated based on the boundary conditions and material properties. Solution results The solution compares well with the one in ref. [1]. Reference [1] H. Mynott, Siemens Energieerzeugung KWU, Germany, in private communications.
412
100 mm
esat (C) = -84.0306 + 183.835 p0.1514 (bar), p 2 = -11.770 + 118.495 p0.2216 (bar), p > 2 374.15C
B PB P P P P P
Figure T.66
413
ADINA-T problems
T.67
Objective To verify the behavior of axisymmetric radiation elements in modeling the radiation exchange between parallel circular disks. Physical problem The radiation between two parallel circular disks is considered, see Fig. T.67. The temperature of one disk is kept at 100 C and the temperature of the second disk is kept at 0 C. The radiation surfaces are assumed to be grey and assumed to have the diffuse reflection property. The heat flux distributions on the plates are to be calculated. Finite element model Ten two-node axisymmetric elements represent each circular disk. A Gauss integration order of 3 for the radial direction is used in each element. There are no conduction elements included in this steady-state analysis. Solution results The calculated local heat flux solution compares well with the one given by Sparrow et al. [1]. User hints $ In this analysis the radiosity is calculated analytically along the circumferential direction. The radiosity can also be calculated numerically, but then at least 24 Gauss points should be used. $ If the distance between the two disks becomes larger, then fewer elements can be used for the radiosity and heat flux calculations. $ If only the radiosity solution is sought, all the nodal temperatures must be prescribed.
414
Reference [1] Sparrow, E. M. and Gregg, J. L., "Radiant Interchange Between Circular Disks Having Arbitrarily Different Temperatures", Journal of Heat Transfer, pp. 494-502, Nov. 1961.
Figure T.67
415
ADINA-T problems
T.68
Objective To demonstrate the capability of using 4-node shell elements in thermal and stress analysis. Physical problem Same as in Example B.17. Finite element model The cantilever beam is modeled using three 4-node shell elements. In the temperature analysis using ADINA-T the temperature distribution through the thickness is calculated and the resulting nodal point temperatures and temperature gradients are written to the temperature and temperature gradients files. A thermal stress analysis is then performed using ADINA in Example B.127. A steady-state analysis with the input of three time blocks is performed. Solution results The correct temperature and heat flux results are predicted by ADINA-T.
416
T.69
Objective To demonstrate the capability of using shell conduction elements in phase change analysis. Physical problem Same as in Example T.54. Finite element model A 10 10 mesh of 4-node shell conduction elements is used to model the corner of the liquid. A lumped heat capacity must be used since phase change effects are included. The Euler backward method of time integration with a time step of 0.01 is employed in the step-by-step solution. Equilibrium iterations are performed at each time step using the modified Newton-Raphson method and the default tolerance 0.001 is used. Solution results The results are identical to those obtained in Example T.54.
417
ADINA-T problems
T.70
Objective To demonstrate the capability of the ATS method in nonlinear transient analysis. Physical problem Same as in Example T.53. Finite element model The same finite element model as in Example T.53 is used. Here, the maximum number of equilibrium iterations is limited to 20 and then the ATS method is activated. Solution results The results are very close to those obtained in Example T.53.
418
T.71
Objective To verify the behavior of radiation planar elements in modeling the adjoining plates problem. Physical problem Radiation heat transfer occurs between two adjoining plates with an angle of 60, see Fig. T.71. Planar conditions are considered. The radiation surfaces are assumed to be grey and to have the diffuse reflection property. The local heat fluxes are to be calculated. Finite element model Ten planar radiation elements represent each plate and a Gauss integration order of 10 is selected for each element. Since only a steady-state solution is sought, the conduction elements are not included here. However, the temperatures of both plates are prescribed at 100 C. Solution results The calculated local heat flux distributions are close to those given in ref. [1]. User hints $ Using an emissivity of 0.1 is a severe case for this analysis, therefore 10 Gauss points are used for the calculation of the radiosity matrix and heat flux. When a higher value of emissivity is used (eg. 0.7), a lower integration order can be employed. Reference [1] Sparrow, E. M. et al. "Analysis, Results, and Interpretation for Radiation Between Some Simply-Arranged Gray Surfaces", Journal of Heat Transfer, Vol. 83, pp. 207-214, May, 1961.
419
ADINA-T problems
60
L
Figure T.71
420
T.72
Objective To verify the behavior of the diffuse-grey radiation element in two-dimensional transient analysis. Physical problem The physical problem, shown in Fig. T.72, consists of several different annular regions. Region I contains a volumetric heat source, region II represents a monolithic cask wall, region III is a voided neutron shield, and region IV is a thermal radiation shield. Regions II and IV exchange heat solely by thermal radiation. Heat is also exchanged with surrounding environment by thermal radiation. All surfaces and the environment are assumed black. The analysis consists of two solution runs. First, a steady-state solution, with radiant heat loss to a 54.4 C environment, is sought. Second, using the steady-state solution for the initial conditions, the environment is subjected to a step increase to 800 C and is kept at this temperature for 30 minutes. Then, the environmental temperature is lowered to 54.4 C and the analysis is continued for 60 minutes (cool-down period). Finite element model Regions I and II are each represented by 5 (radial) 40 (circumferential) planar conduction elements. Region IV is represented by 1 (radial) 80 (circumferential) planar conduction elements. To model the radiation between the surfaces of region II and IV, 40 and 80 planar grey-diffuse radiation elements are used on each opposing surface. On the outer surface of region IV, 80 boundary radiation elements with the property of pure emission are used to model the heat exchange with the environment. A lumped heat capacity matrix is formed for the transient analysis and full Newton iterations are used throughout the solution. Euler backward time integration is used and the time steps blocks for the two solution runs are as follows:
421
ADINA-T problems
Restart
Time increment (sec.) 1000. 2000. 5000. 0.25 4.5 5.0 12.5 25.0 50.0 10.0 100.0
Solution results The calculated temperatures at marked locations T3, T8 and T4 are given in the following table. Time (sec.) 0 1800 5400 T3 ADINA-T 136.19 687.5 202.87 T8 ADINA-T 203.67 374.29 297.37 T4 ADINA-T 213.45 262.32 312.31
The results compare well with those given in ref. [1]. User hint $ In the finite element analysis, an emissivity coefficient of 0.999 (instead of 1.0) should be used for the black surfaces. Reference [1] Sandia National Laboratories, "Sample Problems Manual for Benchmarking of Cask Analysis Codes", Report no. Sand88-0190.TTC-0780.UC-71, Feb. 1988 and Sand880380.TTC-0783.UC-71, Aug. 1988.
422
Region 1
T4
T8 T3
Radius, r (cm) Density, (kg/m3) Specific heat, cp (cal/gmC) Conductivity, k (kw/mC) Heat source, Q (kw/m3)
P P B B P P
423
ADINA-T problems
T.73
Objective To demonstrate the use of ADINA-TMC, here applied to the solution of a piezoelectric problem. Physical problem Experiments performed on an aluminum cantilever by Anderson and Crawley [1] are considered here, see Fig. T.73. Two piezoceramic actuators are bonded to each of the upper and lower surfaces of the beam. The beam specimen is statically deformed by applying an electric field in a direction normal to the middle plane of the actuators. The actuators on the top of the beam are subjected to a field equal in value but opposite in sign to the actuators on the bottom, to produce bending in the Y-Z plane. The experiments showed a nonlinear constitutive relation in the piezoelectric coupling for such an electrical field. Finite element model In the finite element analysis, 2-D 9-node elements are used for both the aluminum beam and the piezoelectric actuators. A perfect bond is assumed between the beam and the actuators. The nonlinear constitutive relation reported in the experiments [1] is used in this analysis,
d13 =
4 Beyy A 1 + 1 + 2 A2
P P P P B B
where A = 2.54 107 mm/volt, B = 1.94 1010 mm2/volt2 and d13 is a coupling coefficient dependent on the strain component eyy. The other coefficients are considered to be constants.
B B
Since this is a fully coupled problem, ADINA-TMC is used. The total number of iterations for this problem is 4 for an applied voltage of 100 volts. In the ADINA finite element model, the user-supplied subroutine CUSR2P is used for this problem, and may be modified by the user (this subroutine is provided along with ADINA). A convergence tolerance of 0.001 for the displacements is used.
424
In the ADINA-T finite element model, the subroutine TUSR2P is used, and may be modified by the user (this subroutine is provided along with ADINA-T). A convergence tolerance of 0.001 for the voltage is used. Solution results The strain component at point P calculated from the finite element analysis is 6.03 10-5 which compares well with the value reported in [1].
P P
Reference [1] Anderson, E.H. and Crawley, E.F., "Piezoceramic actuation of one- and twodimensional structures." Space Systems Laboratory SSL 5-89, Massachusetts Institute of Technology, Cambridge, MA 02139, May 1989.
51 mm
Aluminum beam
Y
267.1 mm
Piezoceramics
P
Strain gages
+V
63.5 mm
25.4 mm
X Beam thickness = 1.59 mm
Figure T.73
425
ADINA-T problems
T.74
Objective To present a soil consolidation analysis using ADINA-TMC. Physical problem A column of soil confined laterally so that no lateral expansion can occur is submitted to a uniform pressure at the top, see Fig T.74(a). Water can only escape at the upper surface of the column (which is supposed to be very porous). Both the displacement of the soil downward and the water pressure depend only on the altitude z and the time t. The soil skeleton is elastic. The strains and displacements are assumed to be small. Finite element model ADINA model: A 110 mesh of 2-D plane strain elements is used. ADINA-T model: A 110 mesh of 2-D planar conduction elements is used. The water pore pressure is zero at the top of the column. The water flow is zero on the lateral and bottom faces of the column. Solution results ADINA model: The settlement of the soil is measured at the top of the column. The settlement time history obtained with ADINA-TMC compares very well with the solution given in [1]. ADINA-T model: The profile of the water pore pressure along the column is compared with the results given in [1] and excellent agreement is observed. User hints $ Modifications, if any, can be made in subroutines CUSR2C (ADINA model) and TUSR2C (ADINA-T model). $ It is preferred to solve soil consolidation problems using the pore pressure material model of ADINA (and not using ADINA-TMC).
426
Reference [1] Biot, M.A., "General Theory of Three-dimensional Consolidation", J. Applied Physics, Vol. 12, February 1941.
surface
Z Y
a = 1.0 (consolidation parameter 1, soil volumetric strain) b = 0.001 (consolidation parameter 2, pore pressure) k (permeability) = 1.0 in/day
Figure T.74
427
ADINA-T problems
T.75
Objective To verify the behavior of the diffuse-grey radiation element in two-dimensional transient analysis. Physical problem This problem is similar to Example T.72, except that a shield is added, see Fig. T.75. In addition to the heat exchanges being described in T.72, the radiant heat transfer between the cask and the shield and between the shield and the ambient are also considered. The analysis consists of two solution runs. First, a steady-state solution, with radiant heat loss to a 54.4 C environment, is sought. Second, using the steady-state solution for the initial conditions, the environment is subjected to a step increase to 800 C and is kept at this temperature for 30 minutes. Then, the environmental temperature is lowered to 54.4 C and the analysis is continued for 60 minutes (cool-down period). Finite element model Regions I and II are each represented by 5 (radial) 40 (circumferential) planar conduction elements. Region IV is represented by 1 (radial) 80 (circumferential) planar conduction elements. The plate is modeled by 20 (along the width direction) 2 (along the thickness direction) planar conduction elements. To model the radiation between the surfaces of region II and IV, 40 and 80 planar grey-diffuse radiation elements are used on each opposing surface. The environment is represented by a closed surface with prescribed temperatures. The heat interchanges are modeled by 80, 44 and 80 planar grey-diffuse radiation elements on the outer surface of casket, on the plate surface and on the environment. A lumped heat capacity matrix is formed for the transient analysis and modified Newton iterations are used throughout the solution. Euler backward time integration is used and the time step blocks for the two solution runs are as follows:
428
Time increment (sec.) 500.0 0.25 4.5 5.0 12.5 25.0 0.25 4.5 10.0 25.
Solution results The calculated temperatures at the marked locations T1, T3 and T6 are given in the following table. Time (sec.) 0 1800 5400 T1 ADINA-T 87.08 762.98 206.86 T3 ADINA-T 143.25 658.83 243.87 T6 ADINA-T 135.18 686.93 199.60
The results compare well with those given in ref. [1]. User hint $ In the finite element analysis, an emissivity coefficient of 0.999 (instead of 1.0) should be used for the black surfaces. Reference [1] Sandia National Laboratories, "Sample Problems Manual for Benchmarking of Cask Analysis Codes", Report no. Sand88-0190.TTC-0780.UC-71, Feb. 1988 and Sand880380.TTC-0783.UC-71, Aug. 1988.
429
ADINA-T problems
T6
Region 4
Region 3 Region 2
Region 1
T4
T8 T3
T1
Shield characteristics Width, W 109.2 cm Thickness, 2.54 cm Distance from cask, D 30.48 cm Density, 7832.8 kg/m3 Specific heat, cp 0.113 cal/gmC Conductivity, k 0.045 kw/mC
P P B B
Figure T.75
430
Index
Index
1 16-node shell elements, 45, 65, 178, 212 1-D conduction elements, 369, 375, 382, 384, 386, 387, 389, 391, 402 2 2-D conduction elements, 379, 381, 393, 397, 399, 401 2-D contact surfaces, 239 2-D fluid elements, 299, 307, 315, 339, 347, 350 2-D potential-based fluid elements, 103, 105, 113, 115, 119, 121, 123, 136 2-D solid elements, 23, 237, 256, 270 3 3-D conduction elements, 397, 408 3-D contact surfaces, 243, 245 3-D displacement-based fluid elements, 88 3-D fluid elements, 307, 349 3-D potential-based fluid elements, 117 3-D solid elements, 33, 39, 79, 176, 180, 207, 237, 245, 253, 255, 260, 263, 275, 279, 283 4 4-node shell elements, 99, 107, 108, 134, 147, 212, 214, 216, 222, 236, 275, 416 8 8-node shell elements, 14, 39, 58, 66, 101, 149, 268 A Acoustic cavity, 105, 367 Addition of elements, 205 Annulus, 303, 318 Arrival time option, 95 ATS method, 187, 262, 418 Axisymmetric conduction elements, 373 Axisymmetric elements, 7, 9, 12, 91, 111, 143, 176, 191, 201, 203, 226, 245, 247, 281, 301, 325, 328, 412, 414 B Backward-facing step, 343, 345 Bending action, 67, 107, 178 Boussinesq approximation, 315, 326 Buckling analysis, 21421, 220, 228 C Cables, 174, 197, 230, 294, 295 Cask, 421, 422, 428, 429 Central difference method, 86, 88, 91, 277, 281 Centrifugal loads, 133, 283 Circular plate, 12, 123, 222 Compressible fluid, 34751, 362 Concentrated loads, 28, 51, 163, 184, 187, 215, 247 Concentric cylinders, 61, 103, 256, 270, 354, 406 Concrete material, 184, 187, 201, 272 Condensation, 412 Conjugate heat transfer, 328 Constant isotropic conductivity, 373 Constant specific heat, 373 Constraint equations, 70, 88, 119, 178, 241, 408 Contact problem, 239, 245 Contactor surfaces, 239, 241, 243, 245 Convection, 2, 315, 316, 323, 326, 327, 340, 387, 388, 395, 412, 413 Convection coefficient, 2, 388, 395 Couette flow, 299 Crack propagation analysis, 287, 289, 292 Creep material, 167, 176, 208, 260, 261 Curve description model, 182 Cyclic symmetry analysis, 70, 130, 131, 133
431
Index
D Dam, 399, 401, 408, 411 Darcy's law, 326, 397 Determinant search method, 77, 78, 79, 81 Die-swell problem, 335 Diffuse-grey radiation, 406, 409, 421, 428 Drag forces, 333, 341 Drucker-Prager material, 191 E Elastic foundation, 72 Elastic-plastic analysis, 163, 169, 170, 172, 173, 184, 189, 193, 199, 203, 226, 234, 236, 238, 253, 255, 258, 268, 275, 287, 288, 292 Electrical potential, 395 Electromagnetic interaction, 138 Element birth, 195 Element death, 195, 395 Emissivity coefficient, 422, 429 Explicit time integration, 294 External flow, 347 F Fabric analysis, 266 FCBI elements, 301, 303, 305, 309, 311, 313, 318, 319, 321, 323, 325, 326, 328, 331, 333, 335, 337, 340, 341, 343, 345, 352, 354, 356, 358, 360, 362 Fluid-fluid interface, 339 Fluid-structure interaction analysis, 117, 333, 337, 352 Fracture mechanics, 159 Free-convection, 323 Freezing, 391, 393 Frequency analysis, 76, 7884, 97, 105, 107, 113, 119, 125, 130, 136, 283, 367 Full Newton iterations with line searches, 182, 193, 234 Full Newton iterations without line searches, 167, 224, 234, 241 Fully coupled problem, 341, 352, 424 Fully developed flow, 301, 303, 305, 321
G Gravity load, 125, 174 Ground motions, 89, 140, 152, 154, 155, 156 H Hamel problem, 311 Heat conduction, 369, 371, 373, 388, 390, 391, 393 Heat generation, 327, 377, 384, 386 Hermitian beam elements, 19, 24, 27, 34, 63, 72, 80, 89, 95, 138, 152, 155, 169, 197, 220 Hydrodynamic forces, 294 I Imperfections, 216, 217 Inertia forces, 294 Initial conditions, 88, 230, 319, 347, 349, 421, 428 Initial strains, 266, 270, 326 Internal heat generation, 327, 377, 384, 386 Iso-beam elements, 21, 22, 25, 26, 35, 63, 67, 81, 167, 188, 208, 236, 238 J J-integral, 159, 285, 287, 289, 292, 305 L Laminar flow, 299, 321, 343 Large displacement, large strain formulation, 287 Large displacement, small strain formulation, 163, 164, 174, 203, 211, 230, 265, 266, 289 LDC method, 247 Linearized buckling analysis, 214, 215, 216, 218, 219, 220, 228 Long cylinder, 226, 279
432
Index
M Mass transfer, 35861 Materially-nonlinear-only formulation, 99, 182 Mathematical traction (MT) formulation, 300, 302, 331, 333, 341 Membrane action, 14, 18, 57, 107, 178 Merging streams, 339, 358 Mesh refinement, 108 Mode superposition analysis, 95, 171 Mooney-Rivlin material, 241, 249 Moving boundaries, 339 N Natural convection, 315, 316, 326, 327, 340 Newmark method, 89, 91, 155, 169, 187, 189, 203, 277 No flow test, 309 Nodal equivalent forces, 87 Noise, 149, 151 Nonlinear elastic material, 169, 170, 197, 272 Non-Newtonian flow, 307, 308 O Ogden material, 249, 264 One-way coupling of fluid-structure interaction analysis, 333 Orthotropic material, 68, 69, 99, 101, 134, 147, 263 Orthotropic plate, 68, 99, 101 P Parachute, 352 Phase change, 391, 393, 417 Physical traction formulation, 335 Piezoelectric analysis, 424 Pipe elbow, 109, 251 Pipe elements, 109, 193, 236, 251, 275 Pipe flow, 325, 329 Pipe whip analysis, 169, 171 Planar conduction elements, 367, 371,
377, 395, 406, 421, 428 Plane strain elements, 32, 78, 182, 195, 239, 266, 341, 426 Plane stress elements, 30, 74, 76, 99, 167, 184, 199, 266 Plate elements, 18, 36, 47, 49, 51, 53, 57, 60, 82, 97 Porous medium, 326 Postbuckling analysis, 209 Prescribed displacements, 115, 133, 191, 230, 266, 292 Prescribed temperatures, 371, 379, 381, 428 Pressure vessel, 14, 272 R Radiation, 2, 354, 356, 38790, 404, 406, 409, 414, 419, 421, 428 Radiation exchange, 354, 356, 404, 406, 414 Radiation interchange analysis, 404, 414, 415, 419 Radiosity, 356, 404, 414, 419 Random vibration analysis, 140, 149 Reinforced concrete, 184, 185, 187, 272 Residual stresses, 199 Response spectrum method, 12527, 152 56 Restart option, 167, 188, 208, 258, 294, 319 Rhombic plate, 53 Rigid links, 41, 61, 103, 249 Rotating cylinders, 305, 321 Rubber analysis, 241, 242, 249, 264 S Seepage analysis, 397, 399, 401, 408, 411 Shell conduction elements, 401, 404, 417 Shell structures, 18, 43, 57, 58, 60, 108, 203, 212 Shock wave, 350 Skew systems, 14, 15, 17, 28, 30, 32, 33 37, 7983, 107, 176, 263, 358 Square cavity, 315, 316, 340
433
Index
Square plates, 45, 49, 51, 68, 134, 147, 178, 409 Steady-state temperature analysis, 371 Stiffener, 41, 130, 214 Stress intensity factor, 14345, 285 Subspace iteration method, 76, 79, 80, 82, 83, 84, 97, 107, 125, 140, 149, 152, 229 Substructures, 189 T Target surfaces, 241, 243, 245 Temperature gradients, 222, 291, 371, 416 Temperature loads, 286 Thermal flow, 328 Thermal stress analysis, 207, 291, 374, 416 Thermo-elastic material, 180, 207, 222, 275, 285, 291 Thermo-elastic-plastic material, 275 Thick cylinder, 7 Thick-walled cylinder, 176, 226, 256, 279 Torsion, 19, 25, 234, 249, 264 Torsional rigidity, 377, 378 Transient flow, 318 Transient heat conduction, 367, 393 Transient temperature analysis, 387, 389 Transition elements, 39, 43 Trapezoidal rule, 91, 115, 117, 121, 123,
165, 169, 187, 189, 381, 384 Traveling load, 95 Truss elements, 28, 38, 72, 73, 8688, 117, 131, 16365, 169, 170, 174, 175, 182, 184, 189, 19597, 209, 211, 230, 236, 258, 272, 352 Turbulence, 345 Turbulent flow, 345 Two-way coupling of fluid-structure interaction analysis, 333, 341 U User-supplied material, 279, 411 V Variable time step, 208, 261 Viscoplastic material, 279, 280 Viscous dissipation, 321, 325 Von Mises plasticity, 172, 173 W Wall-driven cavity, 313 Wave equation, 367 Wave propagation, 86, 88 Wind tunnel, 341
434