ELASTIC-PLASTIC TEARING WITH REMESHING AND MAPPING

Mark James Kansas State University This document describes some of the new elastic-plastic tearing capabilities implemented in F !"#$%&'( These new capabilities are located within the M)%*F+&%,F*", T,! *"- and M)%*F+&M)., # !#K( The new functionality is not fully implemented and only works for a single edge crack( * strongly urge you to work the e/ample problem0 since the order of the steps0 and part of the e/planation is embodied there(

Example 1 3 M(T) wi ! Reme"!i#$

This e/ample problem describes some of the basic features of F !"#$%&' for elastic-plastic tearing with remeshing and state variable mapping( The problem described here is for an analysis of a 12 M3T4 3middlecrack tension4 specimen( This is a good introductory e/ample because the problem si5e is relatively small and because there are other numerical as well as e/perimental results readily available for comparison( Figure 6 below shows the mesh for the e/ample( 7e assume symmetry along the vertical center-line of the specimen0 but model both the top and bottom halves( The problem will be run in displacement control( The crack will be introduced on the left edge as an edge crack midway between the top and bottom0 and symmetry boundary conditions on the left edge of the model will enforce the middle-crack behavior(

Problem Set-up
Start the F !"#$%&' program and read in the mt6(inp mesh file( -o to the 8 ,-8 )#,SS menu and verify that the problem type is plane stress( "e/t set the bottom edge to have + fi/ity and the left edge to have 9 fi/ity( -o to the M!T, *!' option and change material one to .)" M*S,S( Some of the material properties are already set correctly( Set the :! %,"*"- , property to the file option0 and read in the data from the mat$;$<(t/t file( Set the +*,'% ST ,SS to =; ksi0 and set the 8 )> T+8, to 8lane Stress( Set the critical crack tip opening displacement0 #T)%c0 to ;(;;1=2( The #T)%c *"*T value should be the same as #T)%c( "e/t we will describe the applied displacement boundary condition on the top edge of the model( -o to the !88'% %*S8 menu option in the 8 ,-8 )#,SS menu( Select ,%-, %*S80 then + %*S8( -ive a value of ;(;;62 for the displacement value0 then apply the displacement on the top edge of the model( eturn to the main menu and save a restart file using the 7 *T, F*', option( .erify the boundary conditions by running a linear elastic analysis( Use the !"!'+S*S&'*",! & %* ,#T ST*FF option( The number of e?uations should be 6=@A0 and the solution time is almost negligible on most of todayBs faster machines( eturn to the main menu and proceed to 8)ST-8 )#,SS0 and view the %,F) M,% M,S:( "e/t use ")%, *"F) and verify that the + displacement on the top edge is ;(;;62( #onfirm that the stress in the problem is uniform in the #)"T)U option by viewing the S*- + component( The + stress should be about 6(A1 ksi(

Introducing the Crack

The crack we want is an edge crack( -o to the M)%*F+&",7 # !#K&")"-#):,S*.,&,%-, # !#K option( *ntroduce the crack at the node on the left edge of the model that is mid-way between the top and bottom of the model( The initial $a&7 for the e/periment we want to model is 6&1( This means

%raft ;6&$$&6<

-6-

that the crack tip will ultimately be at coordinates 3;(=0;(;4C however0 it is beneficial to grade the mesh density along the crack face from the edge of the model to the crack tip( 7ith this in mind0 set the initial crack tip at 3;($=0;(;40 and use 1 element subdivisions on the crack face( !ccept the default mesh0 then return and enter the M)., # !#K&ST!"% M,T: menu option( ,/tend the crack to 3;(<@0;(;4 using @ element subdivisions on the crack face( Finally0 reenter the M)., # !#K&ST!"% M,T: menu and e/tend the crack a final time to 3;(=0;(;4 using two element subdivisions on the crack face( Figure $ below shows the mesh in the region of the initial half inch crack(

Figure 6D *nitial mesh for 12 middle-crack tension analysis( There are a couple of important aspects to introducing the crack in this manner( First0 the mesh is graded from the edge of the model0 reducing the mesh si5e and number of degrees of freedom0 and in this case0 without compromising the solution0 since thereBs no significant Eaction2 behind the crack tip( Second0 the material properties specify that we measure the #T)% fracture criterion at ;(;<2 behind the crack tip( The final crack e/tension step grew the crack ;(;<2 with two elements( The default minimum number of crack e/tension elements in F !"#$%&' is two elements0 so the fracture criterion will always be measured with no fewer than two elements behind the crack tip( *t certainly would have been possible to use more elements along the crack face for that last ;(;<2 of e/tension( ,/perience has shown that for many applications with $;$<-T1 aluminum0 as we are using here0 a ;(;$2 element si5e is ade?uate( )ne point to considerD the element si5e needs to be fi/ed during crack growth so that the fracture criterion can be sampled at a fi/ed distance behind the crack tip as the crack propagates(

%raft ;6&$$&6<

-$-

Figure $D *nitial mesh with the initial crack for the 12 M3T4 problem(

Automatic Tearing
This section describes the control parameters that are necessary for the "ewtonBs method solver and for the propagation algorithm( First0 we need to initiali5e the tearing data structures for the crack( ,nter the M)%*F+&%,F*", T,! *"-&T*8 menu( The crack tip will automatically be initiali5ed for tearing0 and highlighted bo/es will be drawn at each node on the crack face( :ighlighted bo/es indicate that the nodes are tied for un5ipping( For this internal crack0 no un5ipping will be used( Use *"*T S!7#UT to untie the crack face nodes by setting the initial sawcut length to $$ nodes 3only the crack tip node will be highlighted4( "ow go to the M)%*F+&M)., # !#K&!UT) T,! *"- menu( *n the #)"T )' 8! !M menu change the following analysis control parametersD 'oad Sub-Steps 6;; Ma/ *terations 1;; !ppl %isp Factor 1= eturn to the !UT) T,! *"- menu and confirm the following tearing control parametersD ,' S*F, ;(;$2 # K *"# $ ,' >!#K $ ST,8S = U"F*8 ")

%raft ;6&$$&6<

-1-

# *T

#)%

,nter the M)"*T) menu and use the !%% ")%, option to define monitor data for the upper left node in the model 3this will monitor this particular node as the solution progresses for each load step4( "e/t set the F*', "!M, to something like mt1.mon( This is the output file that will contain the monitor data(

Running the Analysis

eturn to the !UT) T,! *"- menu and hit the 8 )8!-!T, menu to start the analysis( The analysis routines will increment the far field displacement while monitoring the crack tip fracture parameters( 7hen the critical value is reached0 the remeshing routines will be invoked to e/tend the crack by the specified amount( Then the analysis will resume0 and the process will repeat for the specified number of steps 3in this case0 five steps4( Figure 1 below shows the load crack e/tension data e/tracted from the mt1.mon output file( This data was e/tracted using two scripts designed for post-processing the mon file(

3" M(T) Crack Growth

4.6 4.5 4.4 4.3 Load (kip) 4.2 4.1 4 3.9 3.8 3.7 0 0.05 0.1 0.15 0.2 0.25 Crack Extension (in)

Figure 1D 'oad crack e/tension data from mt6(mon(

Some Hints
*tBs obvious 3or should be4 that the density of the mesh where you initiate a crack needs to be about the same si5e as the elements that you introduce( *n the e/ample above0 we grade the element si5e along the crack face so that this will be true( The elements near the free edge0 but along the crack face are larger0 but still well formed0 while the elements at the tip transition nicely(

%raft ;6&$$&6<

-<-

7hatBs sometimes not so obvious is that the mesh that youBre propagating into must also be about the same si5e 3or slightly larger0 if you want4 than the step size that youBre taking with the program( This is crucial to obtain the most accurate results0 especially with elastic-plastic analysis0 where there are no Especial2 crack tip elements to capture crack tip behavior(

Me#% I em De"&'ip i(#"

M)%*F+ The following items are new to0 or have been modified in0 the M)%*F+ menu( M)%*F+&M)., # !#K There have been several changes to the M)., # !#K menu0 all related to propagating cracks using the mapping functionality( T,! )", This is a convenience button that performsD S!., M!8 %!T! ST!"% M,T: %) M!88*"!UT) T,! *"This is the elastic-plastic e?uivalent of !UT)M!T*# 3which is for linear elastic4( S!., M!8 %!T! Saves away all of the mapping data for the current mesh( This data can be used to manually restore the state to a new mesh after remeshing has occurred( %) M!88*"8erform the mapping using the previously stored data using the S!., M!8 %!T! option( M)%*F+&%,F*", T,! *"The %,F*", T,! *"- menu contains options for initiali5ing the tearing data structures for various kinds of cracks( T*8 *nitiali5es the tearing data structures for traditional crack tips embedded in the mesh( S+MM,T + *nitiali5es the tearing data structures for symmetry cracks that lie along the 9 a/is( *"T, F!#, *nitiali5es the tearing data structures for cracks that lie in a non-linear interface(

%raft ;6&$$&6<

-=-

#',! !'' #lears all of the internal tearing data structures( ,' ST,8S Sets the number of load steps over which the forces on a tied crack face are released( The default is one load stepC this is satisfactory for simple single cracks( :owever0 when multi-site damage is involved0 a typical value is ten( The forces need to be released over a number of load steps because the release and growth of one crack may cause other cracks to grow( >y releasing the loads over a number of steps0 the plasticity at all crack tips can evolve in an orderly manner0 and the reaction and history for each tip can be recorded( "UM >!#K The number of nodes back from the crack tip where the critical value of #T)% is measured( This should always be an even number 3donBt measure at mid-side nodes4( M)., # !#K&!UT) T,! *"This is the elastic-plastic e?uivalent of !UT)M!T*# 3which is for linear elastic4( The control parameters are specific to elastic-plastic analysis and controlling the elastic-plastic tearing( #)"T )' 8! !M This menu defines the load control parameters for the "ewtonBs solver( These options are described in the manual( ,' S*F, This is the characteristic element si5e along the crack face( For tearing in aluminum a typical value might be ;(;$ inches( # K *"# This is the number of elements that the crack will be e/tended at each step( The total increment at each step is ,'GS*F, H # KG*"#( ! typical value is two( This value should always be greater than 6( #) , :T The height of the plane strain core0 e/pressed in term of ,' S*F,( The total core height is ,'GS*F, H #) ,G:T( This option is currently not active. ,' >!#K The number of elements behind the crack tip to measure the fracture criterion( This is e/actly "UMG>!#K&$ 3from the %,F*", T,! *"- menu4( The fracture criterion must always be measured at a corner node in the mesh( ST,8S The number of crack propagation steps( 8S# M!T

%raft ;6&$$&6<

-@-

The material used for the plane strain core( This option is currently not active. U"F*8 Un5ip nodes between crack tips( !fter a crack is e/tended0 and new surface area is created0 some new nodes e/ist along the crack face from the location of the old crack tip0 to the location of the new crack tip( The U"F*8 option controls how these nodes are handled( 7hen U"F*8 is +,S0 these nodes are tied and se?uentially released as the load is increased( 7hen U"F*8 is ")0 these nodes are not tied( *f the crack e/tension 3as described above as ,'GS*F, H # KG*"#4 is the same distance as where the fracture criterion is measured0 then the results will be the same independent of whether U"F*8 is +,S or ")( *f the crack increment is larger than the distance to the node where the fracture criterion is measured0 then the U"F*8D +,S option should be used( # *T This is the fracture criterion( #)% is the Sutton&Fashang criterion( %* This is the direction prediction criterion( %*S8'!+ These are options that control the display as the automatic propagation progresses( M)"*T) This is described in the manual( This allows stresses0 displacements0 and crack tip progress to be monitored as the solution progresses( The data for the monitor are written to a file( 8 )8!-!T, %o the propagation based on the above defined parameters( %,F*", T,! *"-&T*8 These options define the initial tearing data for a crack tip( *"*T S!7#UT *nitiali5e data for a new tip( %*FF, ,"T T*8 Switch to another tip to update parameters( %,F*", MS% Toggle nodes on the crack face to create MS% cracks(

%raft ;6&$$&6<

-A-