Home

Mars Manual

1. 43 ad a a a o Bae Beye eae 44 5 11 Pre Set Test Simulations 0 0 00 a a 47 5 11 1 Biaxial Test e 47 5 11 2 Triaxial Test 48 5 11 3 Colorado Test 2 2 0 2 0000000000000 0084 48 6 Nodes and Particle Lists 49 6 1 Node Lists aa 49 6 1 1 Select Commands o a 000000084 50 6 1 2 Set Commands oaa 0 000000 00000088 51 6 1 3 Scale Commands aaa 0 00000 000004 51 ae Sree ea A erase ak 52 6 1 5 Other Commands aa a 000000084 52 e e os a e 53 6 1 7 Plot List Commands 0 0 0048 4 53 6 1 8 Nodal Rotations e 56 6 1 9 Extract Commands 0 000004 57 6 1 10 Insert Commands 0 0 0 0 00 000000088 57 6 1 11 Discrete Element Commands 0 4 58 6 2 MacroParticle Lists 2 a a a 58 Bei SD Beas A NR 59 ath otk et mode eH od amp eH da Ar 59 o e er ee 60 60 ele Wee A E oe es 60 7 1 1 Generate commands 0 0 0 000008 8 62 7 1 2 Select commands 00 000 000000008 65 7 1 3 Make commands 0 0 0 0 0 000000 0004 66 LLIA N tesl ssas So eee a ee we ea oe Pee eww ee es g 66 wa oe See eee e 67 a Gee eee 68 ok ENG a a Oe Seed Gn eee eae Bb bop a 68 T2 Beam Lists cc a soa wa da wR RAD RE REM ROR ER Ew wm 69 7 2 1 Time History Commands 2 2s a2 ee bee ea eee eae 70 7 2 2 Plot Commands s cc asi 000000000 eee 71 7 3 Geometric Pair Detection 0000 000 ee 74
2. if nodes are particles node thickness is the larger select one of the three contact methods DetectionDistance 0 6 cm 3 Enter contact models see previous sections ContactForce PenaltyHyst FrictionForce RollingResistance UpdateInterval 0 000010 s Debug 6 When two node lists are defined then contact conditions are detected between nodes from the first list and nodes from the seconds list but not between nodes within the same list If contacts have to be detected between nodes particles which belong to a single list then only one NodeList must be specified For example if we have two particles lists red particles and blue particles and we want to be able to detect contacts between all particles independently of color we need three contact lists NodeNodeContactList BlueRedContacts NodeList BlueParticles NodeList RedParticles NodeNodeContactList BlueOnlyContacts NodeList BlueParticles NodeNodeContactList RedOnlyContacts NodeList RedParticles There are some rules regarding the size of the nodes and particles If one of the node lists consists of spherical particles with finite radii there is no need to specify a NodeThick ness for that list However if a NodeThickness is specified MARS will take the larger value between the actual particle radius and the NodeThickness parameter If the node thickness is not specified and the node list consists of
3. 113 11 5 Stable Element Time Steps Stable time steps for each LDPM element need to be computed to ensure that the time step used in the solver is small enough to avoid local or global instabilities during the simulation The computation of the Courant stability time step for an LDPM requires the assembly of the element stiffness matrix 24x24 size and the solution of its largest eigenvalue As such this is a computationally expensive operation Unlike ductile steel materials concrete material do not experience significant plastic deformations before fracturing Thus the geometry of LDPM elements remain fairly constant during sim ulations and there is no need to recompute stable time steps The computation of the stable time steps is only done once at the beginning of the simulation Furthermore logic has been placed so that stable time steps are saved in a file named ttL ListName cts As of April 2012 when MARS initializes an LDPM list it first looks for the cts file in the current folder If it finds it it will try to use the steps computed in a previous simulation Two checks are performed to ensure that the data is usable 1 the number of elements in the list must be the same as the number of time steps in the cts file 2 the computed time steps for the first ten elements in the list must be the same as the values of the time steps in the cts file The second check is intended to account for conditions when the elastic properties o
4. 10 9 2 Ten Node Large Deformation Element The 10 node large deformation tet element formulation is similar to the small deforma tion formulation with two significant differences 1 the B matrix is computed at every step and 2 the stress tensor is rotated to account for large rotations The latter is accomplished by attaching a local reference system LRS to each of the four integration points The orientation of the LRS is updated using the spin tensor Strain rates com puted in the global reference system GRS are rotated to the LRS and used to update the stress tensor Note that the stress tensor is always defined in the local reference system The local stress tensor is then roated to the GRS and used to compute the nodal forces This formulation is invoked using the following commands TetSolidList listname 10NodeLargeDef Material materialName 11 Lattice Discrete Particle Model The Lattice Discrete Particle Model LDPM is a discrete meso mechanical model for concrete which was recently developed by Dr Cusatis and co workers at Rensselaer in collaboration with Dr Pelessone at ES3 LDPM simulates the mesostructure of concrete by a three dimensional assemblage of discrete particles whose position within the volume of interest is generated randomly according to the given aggregate size distribution A three dimensional domain tessellation based on the Delaunay tetrahedralization of the generated aggregate centers ge
5. string lbl rdr gt getShortLabel if 1bl MyMaterial return new MyMaterial nam else if 1bl MyLoadCurve return new MyLoadCurve nam 198 rdr gt error Invalid keyword for user defined object type return NULL Make file Make file OBJ user o MyFirstTetList o MySec marsU LIB libmars a 0BJ LNK o marsU L LIB OPT 0BJ lm 1z lmars MyFirstTetList o MyFirstTetList cpp MyFirstTetList h INC mars h CMP OPT MyFirstTetList cpp lt MyFirstTetList o MyFirstTetList cpp MyFirstTetList h INC mars h CMP OPT MyFirstTetList cpp lt MyHexList o MyHexList cpp MyHexList h INC mars h CMP OPT MyHexList cpp lt MyMaterial o MyMaterial cpp MyMaterial h INC mars h CMP OPT MyMaterial cpp lt MyLoadCurve o MyLoadCurve cpp MyLoadCurve h INC mars h CMP OPT MyLoadCurve cpp lt MyMaterial o MyMaterial cpp MyMaterial h INC mars h CMP OPT MyMaterial cpp lt 21 7 6 Restart Procedures If user defined objects or lists must be able to accomodate restart procedures additional methods must be defined string getTag void writeRst RstWriter void readRst RstReader Obj Model readRstUserDefinedObject obLst Model readRstUserDefinedList The method getTag must return the string u0 for user defined objects and ud for user defined lists This strings are written to the restart file and used during restart
6. vy y component of CG velocity RB bodyName vz z component of CG velocity RB bodyName wx x component of rotation rate RB bodyName wy y component of rotation rate RB bodyName wz z component of rotation rate RB bodyName cx x component of CG coordinate RB bodyName cy y component of CG coordinate RB bodyName cz z component of CG coordinate RB bodyName fx x component of force RB bodyName fy y component of force RB bodyName fz z component of force RB bodyName mx x component of moment RB bodyName my y component of moment RB bodyName mz z component of moment 131 14 Loadings 14 1 Nodal Load List Use these commands to specify forces or moments to a single node a set of nodes or a rigid body Loads vary in time according to the time history specified in the load curve NodalLoadList ListName Y 1 Enter Moments if you want to apply moments instead of forces Opt Moments 2 Enter either NodeList or RigidBody Req NodeList NodeListName RigidBody RigidBodyName 3 Enter LoadCurve Req LoadCurve LoadCurveName 4 Enter load direction Req Direction 0 1 0 5 Enter scaling factor Opt and or Distribute Scale 1 Distribute 1 6 Use one of the four lines below to specify nodes Required when NodeList is used Node 1 single node Nodes 1 4 7 9 multip
7. Wi t integral Fi t v t dt 0 5 K x t 2 0 5 K A 2 1 cos wt 2 5 F 2 K 1 2 cos wt cos 2 wt 5 F 2 K 1 cos 2 wt F 2 K 1 cos wt O O 176 0 5 F72 K sin 2 wt F 2 K 1 cos wt Ke t 0 5 M v t 7 2 0 5 M V 2 sin 2 wt 0 5 F 2 K sin 2 wt since M V 2 M A 2 w 2 M F K 2 M K F 2 K Note that the external work is equal to the sum of internal work in this case it is all in the form of elastic energy and kinetic energy W t W t Kelt The energy balance Eb Ke Wi We remains constant at 0 21 2 save plot data for later post processing While MARS makes it possible to generate pre determined plotting data through PlotLists during a simulation you may find it necessary to change some of the plotting parameters once the results of a simulation are known Rerunning a lengthy simulation with different plotting parameters is not only inconvenient but wasteful and time consuming For this reason in 2008 we introduced the capability of writing plot data at regular intervals during the simulation and reading the data back for generation of plot files viewable with Quasar or other 3 D viewers This feature is not 100 Mars Input File ControlParameters WritePlotDataFile data plt Every 0 1 ms In the command above plottable data is written to file data plt You can change this name to any name you want Postprocessing the data is also simple The idea is to use
8. Displacements imposed on plates perpend to z axis 3 Ratioes forces imposed on plates perpend to x and y axis Test TestName Triaxial TetSolidList Specimen specimen list VelocityHistory History XStressRatio 0 5 lateral over vertical YStressRatio 0 5 lateral over vertical TimeHistoryList HIST teL TestName VerticalStress teL TestName VerticalStrain teL TestName XHorizontalStress teL TestName XHorizontalStrain teL TestName YHorizontalStress teL TestName YHorizontalStrain 5 11 3 Colorado Test ReferenceSystem RSYS cartesian Model restrictions 1 The geometry is a 4 inch side cube 2 The center of the cube is at the origin Test BXCT Colorado TetSolidList CUBE X PressureHistory PTHX Y PressureHistory PTHY Z PressureHistory PTHZ 48 6 Nodes and Particle Lists 6 1 Node Lists In MARS nodes and spherical particles are used interchangably The same class in the OOL sense is used for both entitites A typical input block for specifying a node list is shown below NodeList ListName Read filename read nodelist command from file if this list is used to define a rigid body enter Density 7 8 g cm3 if this is a DP tet list enter Particles Color Red LengthUnits cm VelocityUnits cm s if velocities are entered InputFormat IXYZR index cx cy cz and rd available input formats below B boundary code UVW velocities IX
9. Name 4 ttL ListName Cells MinimumCrackOpening 0 01 mm This plotting method is available for both Quasar and Paraview 11 6 2 Plotting external facets only A variation of Cells method makes it possible to plot only the cell triangular facets that are initially visible Essentially these are the external facets of the tet mesh which are not shared by two adjacent cells The number of facets in this list does not change during the simulation I can be used for quickly assessing the progress of a simulation or for coupling with with facet contour plots and displaying the external surface in a neutral color since they are not part of the contour facet list The command for displaying external faces is is ttL ListName ExternalCellFacets This option is available for both Quasar and Paraview 11 6 3 Plotting cell outline If the option of displaying sharp edges is required the command Cell0utline creates a list of edges that are visible The initial list would include the sharp edges of the initial mesh As the Lpdm model breaks up the edges of the cells that have separated from the bulk of the model may be displyed depending on the angle of the facets sharing an edge edge 116 ttL ListName CellOutline The line lists generated using the Cell0utline command can become very large and the plot may show too many edges In these cases it is preferable to display only the initial sharp edges of the m
10. O ae pe es ete gee nee 74 7 5 Master Slave Constraints cor ra be EEG 74 7 6 Node Pair Attraction List 74 7 7 Node Pair Repulsion List lt 2i6 lt 4684 lt 4 sef44 44 Gea boo ox 75 7 8 Node Pair VanDerWaals List 75 7 9 Node Pair NanoParticle List 75 O es ee ee te ee ee a 76 7 10 1 Tabulated Forces 2 a a 76 7 10 2 VanDerWaals Forces 0 0 0 0 020000000 bee 77 E bk O Bee ee ee 77 EA 78 8 Triangular Faces and Shell Elements ep E a a idos Ge rig 8 1 1 Select commands 000000000088 8 1 2 Generate commands 0 0 0 0 0 0 0 0000008 8 8 1 3 Make commands 0 0 0 0 0000000000 8 8 8 1 4 How to make sublists 0 0 0 0 0 0 00084 8 2 Wet Triang Face List oa ee ae Se eR AER ee aa oe Hs 8 2 1 Uniform Pressure 8 2 2 Uniform Pressure Constant Face Areas batty se qs ae Bisse ae ok wo see we eg ee 8 3 Triangular DKT Suele a ek ee se ee ee SE 8 3 1 Time Histories 2 aa aa 9 1 Quadrilateral Face List 2524 3 2 3 0 24644846454 Hb 44 x 9 1 1 Select commands 0 00000 000084 9 1 2 Generate commands 0 0 0 0 0 000000008 9 1 3 Make commands 0 0 0 0 0000000000 8 9 2 Pressurized Quad Face List 6 2 6s ewe ee ee 9 2 1 Uniform Pressure 0 0 0 0 20 00 0000 0008 8 9 2 2 special Loads gis are he eee he a 4 oe bE 9 2 3 Multiple Pressu
11. TetSolidList ListName Viscous 1 Enter master list Req MasterTetList TetListName 2 Enter either load curve or damping constant Req LoadCurve CurveName Damping 0 001 1 s Prescribing the damping coefficient as a function of time using the LoadCurve option makes it possible to use this feature for computing steady state static solutions Damping can then be removed when applying dynamic loads The viscous force between nodes J and K in direction 2 is computed using the equation Fox C my mMx Ui UKi 05 yY z where C is the damping coefficient 10 8 Spatial Field Functions The purpose of this list is to provide a spatial field of a scalar variable that affects material properties at the integration points of finite elements or Ldpm elements Such variable could be temperature irradiation water level etc Typically these fields would be computed by another special purpose code For this list the spatial distribution of a variable f is specified at the nodes of a tetrahedral mesh If the field is constant during the simulation then a single record is sufficient If the field varies in time then a series of records is specified at different times This list is used in other lists For example let s assume that the mechanical properties of a material are temperature dependent and we have computed the temperature history and distribution Then the code would interpolate the temperature at
12. The first command must always be the selection of units A detail description of units is give in a separate subsection ControlParameters Unit specification should be the first input line Units English more on this below TerminationTime 1 ms CurrentTime 0 1 ms MaximumTimeStep 0 001 ms TimeStepScalingFactor 0 8 default 0 9 FragmentationTimelnterval 0 005 ms time or step cotrolled actions Monitor Frequency 10 print progress line every 10 steps more on this below PlottingDefaults 1 see below ContactDefaults 1 see below 14 DynamicRelaxationCurve DynRelax NoDynamicRelaxation stop dyn relax on restart 4 1 Unit Systems The unit system to be employed in a simulation is specified using the command Units followed by one of the unit system label SI CGS English Nano 1 SI international m Kg second 2 CGS cm gram second 3 English in pound second 4 Nano nm ng ns The unit system can be specified once typically at the top of the ControlParameters section All input quantities are converted to the selected unit system For this reason all quantities must be entered with their dimensional units For example if SI units are chosen and the user enters a time variable of 0 1 ms this variable will be automatically converted to 0 0001 s This can be very useful for material properties for example the density of a material can be found in g cm3 and it would be tricky for most
13. WetTriangFaceList ListName UniformPressureConstantFaces 1 Specify face list Req either a face list or a shell list FaceList listName ShellList listName 2 Specify either LoadCurve or Equation Req LoadCurve curveName Equation equationName 8 2 3 Multiple Pressure Histories This list is used when pressure histories are independently computed by a CFD solver at certain stations on the exposed surfaces The pressure histories are read into a Load CurveList which is referenced here This list uses the triangular faces of another list either a face list or a shell list The faces in the parent list can change node definition as it happens during fragmentation However if the parent list represents the external sur faces of a solid mesh and the number of faces increases as a result of solid fragmentation this list will only employ the initial faces TriangFaceList ListName MultiplePressureHistories 1 Specify face list Req either a face list or a shell list 85 Link FaceList listName Link ShellList listName 2 Specify PressureHistory List Req LoadCurveList PressureHistories 3 Optional AutomaticTimeOffset 1 TimeShift 345 ms 2 1 The AutomaticTime0ffset command is designed to offset the time histories so that the pressure loads start at time 0 This is done by computing the first time pressures are different than 0 i
14. ns ContactForce Hertz YoungsModulus 59 MPa PoissonsRatio 0 2 FrictionForce StaticFriction 0 3 DynamicFriction 0 2 RollingResistance Kr 1 e 4 Kt 1 e 4 Fr 0 2 Ft 0 2 VanDerWaalsForces HamakerConstant 1 61e 20 J 75 ThreshholdGap 0 4 nm IonicForces DLVO or SI 7 10 Nano Particles The nano particle interaction list employs the same contact models which are used in the contact list Please refer to the contact models section of this manual A minor difference between regular contact between particles and contact formulas for nano particles is the way nano particle penetration is calculated The acutal gap between two nano particles is calculated using the formula g d ri rT2 where d is the distance between the centers of the two particles r and rz are the radii of the two particles For contact purposes the gap can be redifined by controlling the range of the particle radii rmn rmx Essentially particle radii must be at least rmn but no more than rmx as shown in this code snippet R1 max rd1 rmn R1 min R1 rmx R2 max rd2 rmn R2 min R2 rmx gapC dst R1 R2 The negative of gapC is the inter particle penetration which when positive is used to compute contact forces The values of rmn and rmx are initialized to 0 and 1 e30 respectively They are specified via input using the commands MinContactRadius 4 nm MaxContactRadius 8 nm If the user wants
15. or single core processor Each socket contains one or more cores the socket is then labeled e g a dual core or quad core socket Each socket has its own L2 cache chip which typically is shared among its cores Core refers to a single processing unit capable of performing computations A core is the smallest unit of resource allocation Each core has its own L1 cache chip There are two reasons for distributing the computation over several processors 1 for handling large models which require extensive memory to run in core 2 for reducing execution time In the first case we need to distribute our computational model over a sufficient number of nodes so that each node can accomodated the memory requirements for its portion of the calculations In the second case we want to take advantage of the simultaneous solutions of different parts of the model which reduces global execution time time In MPI executions the user requests access to a certain number of nodes which are allocated by the system when they become available The computational model is partitioned into as many parts as the number of cores available This operation is called domain decomposition The objective of domain decomposition is to split the model in regions Each of these regions is assigned to an individual core different core involves distributing objects over various processors Currently every processor has complete exposure to the ent
16. user define load curve NA A AS ES SS a createUserDefinedList string must always be defined obLst Model createUserDefinedList string nam Reader rdr string lbl rdr gt getShortLabel if 1bl MyFirstTetList return new MyFirstTetList nam else if 1bl MySecondTetList return new MySecondTetList nam else if lbl MyHexListList return new MyHexList nam rdr gt error Invalid keyword for user defined list type return NULL _ createUserDefinedMaterial string must always be defined Obj Model createUserDefinedObject string nam Reader string 1b1 rdr gt getShortLabel if 1bl MyMaterial return new MyMaterial nam else if 1b1 MyLoadCurve return new MyLoadCurve nam rdr gt error Invalid keyword for user defined object type return NULL These classes would be accessed in the input file using the following isntructions UserDefinedList materialName MyMaterial UserDefinedList curveName MyLoadCurve UserDefinedList listName MySecondTetList Material materialName 196 Note that it is not necessary to put all the coding in a single user cpp file Indeed it is possible to create multiple files for improved readability The above example could be arranged as follows Header file for class MyFirstTetList class MyFirstTetList public ttLst
17. where the forces of the slave objects are transferred to the master objects implemented 134 in the polymorphic method reduceFrc and a stage where the velocities of the master objects are used to control the velocities of the slave objects implemented in method applyKin In the solver loop they appear in this order while t lt t_end for i 1 2 N 1 N list i gt clearNodalForces for i 1 2 N 1 N list i gt calcFrc for i N N 1 2 1 list i gt reduceFrc for i 1 2 N 1 N list i gt integrateEOM for i 1 2 N 1 N list i gt applyKin The user should note that when reducing the forces in the reduceFrc method the lists are processed in the reverse order This is very important when multiple constraints are imposed 15 1 Node Face Constraint List TrngFaceNodeBondList BNDS abbreviated notations in lt gt brackets NodeList NODS lt ndL NODS gt FaceList FACS lt tfL FACS gt Tolerance 0 45 mm constraint types 2 Master slave slave nodes may not lay on surface rotation are not constrained MasterSlaveNoRotations 3 Master slave slave nodes may not lay on surface rotation ARE constrained MasterSlaveWithRotations 7 Master slave slave nodes lay on surface rotation are not constrained SlavesOnSurface 6 Nodes slide over surface SlidingWithFriction 0 3 0 2 0 1 mm 1 Enter lines below if bonds are entered expl
18. which is used for aligning the first local axis use the keyword return to select the preferential axis 5 1 1 Cartesian RefSys ReferenceSystem RSYS cartesian 1 Define origin Opt default 0 0 0 Origin 0 2 in 0 6 in O in 2 Define orientation Req FirstDirection 1 0 0 SecondDirection 0 1 0 3 Modify Opt Translate 0 4 in 0 3 in 0 6 in X Rotate 45 4 Select Return direction Return X also available Y and Z 5 1 2 Cylindrical RefSys ReferenceSystem RSYS cylindrical 1 Define origin Opt default 0 0 0 Origin 0 2 in 0 6 in O in 2 Define orientation Req AxialDirection 1 0 0 25 RadialDirection 0 1 0 3 Modify Opt Translate 0 4 in 0 3 in 0 6 in X Rotate 15 deg Y Rotate 35 deg Z Rotate 180 deg 4 Select Return direction Return HoopDirection for shells and solids 5 1 3 Spherical RefSys ReferenceSystem RSYS spherical 1 Define origin Opt default 0 0 0 Origin 0 2 in 0 6 in O in 2 Define orientation Req AxialDirection 1 0 0 RadialDirection 0 1 0 3 Modify Opt Translate 0 4 in 0 3 in 0 6 in X Rotate 15 deg Y Rotate 35 deg Z Rotate 180 deg 4 Select Return direction Return HoopDirection for shells and solids also available AxialDirection RadialDirection 5 2 Load Curves The LoadCurve object is used to prescribe a function of the type y f x in tabulated form by entering
19. x yi pairs These tabulated functions are used in many algorithms If during interpolation the value of x falls outside of the definition range the value of y is not extrapolated Instead it is set to the first or last value in the table The input data pairs must be in ascending order of zx LoadCurve CurveName 1 Specify units Req X units time ms Y units pressure psi 2 Enter tabulated function Req ReadPairs 4 tim prss 0 000 O 0 005 100 26 0 030 O 1 000 O 3 Modify function Opt the following commands are optional X Scale 2 scale x variable X Offset 0 004 offset x variable Y Scale 2 scale y variable Y Offset 0 004 offset y variable Differentiate differentiate curve 1 zi yi yt 1 i41 Ti 1 1 The Differentiate command is used to compute a function which is the differ ential of the input function Some common functions can be internally generated in tabulated form The available functions are listed below below LoadCurve CurveName 1 Specify units Req X units time ms Y units pressure psi 2 Generate internal function yi 5 sin 45 xi for xi i 0 1 and i 0 100 Generate Sin A 5 w 45 dt 0 1 n 100 yi 5 cos 45 xi for xi i 0 1 and i 0 100 Generate Cos A 5 w 45 dt 0 1 n 100 In time history lists the current value of a time dependent function can be printed using the command 1c curveName as shown bel
20. x coordinate CG cy y coordinate CG cz z coordinate CG vx x velocity CG vy y velocity CG vz z velocity CG fx x force I fy y force I fz z force I mx x moment I my x moment I mz x moment I wx x rotation rate CG wy x rotation rate CG wz x rotation rate CG vl absolute velocity CG ke kinetic energy I px x momentum I py y momentum I pz z momentum I ax x angular momentum I ay y angular momentum 1 az z angular momentum 1 6 1 7 Plot List Commands It is possible to generate plot files for Quasar and Paraview of various node variables The input commands for the two post processors are quite different The main difference is that for Quasar files you have to preselect the information you want to plot but it can be combined with other lists For Paraview most of the nodal information is written to the file and the appearance for the plot is chosen during post processing Although it is still possible to set plot attributes for Quasar plots within a NodeList section this is no longer recommended Quasar plot attributes should be entered in the PlotList section 53 PlotList PLOT TimeInternval 0 1 ms ndL NodeListName 1 Select resolution optional default Low HighResolution MediumResolution LowResolution 2 Set or change node particle radii optional scl 0 5 scale parti
21. 1 ni 1 ni 2n1 3 ni 10 1 2 n2 1 n2 2 n2 3 n2 10 1 The indexing system follows these rules e node 5 between nodes 1 and 2 e node 6 between nodes 2 and 3 e node 7 between nodes 3 and 1 e node 8 between nodes 1 and 4 e node 9 between nodes 2 and 4 e node 10 between nodes 3 and 4 Note that the 10NodeElement keyword in the first line is used for defining a purely geometric mesh with no elasto plastic properties Ten node finite element formulations are listed below 104 10 9 1 Ten Node Small Deformation Element The 10 node small deformation tet element formulation is intended to be used for cases where the deformations are small A good description of the 10 node tet formulation is given in www colorado edu engineering CAS courses d AFEM Ch17 pdf A nice property of the T10 element is that it is not overconstrained nor underconstrained it features 10 nodes x 3 DOF node 30 DOFs in total 4 integration points each providing 4 IP x 6 constrant IP 24 constraints which leaves 6 unconstrained DOF s the six rigid body motions The formulation for small deformations save computing time by assembling the B matrix only once at the beginning of the simulation and by not rotating the stress tensors at the integration points during the simulations since rotations are assumed to be small This formulation is invoked using the following commands TetSolidList listname 10NodeSmallDef Material materialName
22. 12 of the facet withing the tet element ik 1 ik2 ik 3 are the indeces of the outside points that define facet k Ak is the actual area of facet k nk x nk y nk z is the normal to facet k not projected pAk is the projected area pk x pk y pk z is the normal to the projected face qk x qk y qk z is the first tangential direction sk x sk y sk z is the second tangential direction Number of facets on the external solid surface sharing center node nxf nf Facet connectivity and facet areas follow Index 0 refers to center node i1 1 i1 2 i1 3 A1 i2 1 i2 2 i2 3 42 Total number of edges in the cell nxe ne Edge connectivities and edge lengths follow 11 1 i1 2 Li i2 1 12 2 L2 108 where ik 1 ik2 are the indeces of edge k Lk is the length of edge k 4 The WriteFacetDataDump command is used for writing an ASCII output file contain ing all information for the 12 facets of each tetrahedral LDPM element The output data is structured as follows For each tet Tet index followed by 24 lines two for each facet n n2 cx cy cz fa nx ny nz pa Px py pz Ge qy 92 5x 8y 6z where ni index of first node on facet edge n2 index of second node on facet edge CX Cy cz coordinates of facet center fa area of facet nx ny nz normal to facet pa projected area used in calculations PX py pz normal to projected facet qx qy qz first tangential direction SX sy sz second tangential direction Note the node indeces n
23. 19 1 2 Octree The octree method creates a box that envelops the model and splits the box in 8 smaller boxes by subdividing the domain in all three directions in space It is a classic method but the number of objects in each domain may vary In this respect it may not generate a well balanced distribution of computational work More on this method at http en wikipedia org wiki Octree The octree method has been implemented in MARS but only partially tested Because ORB is so much superior to the Octree methods the Octree decomposition is not being supported 19 1 3 METIS decomposition METIS is a set of serial programs for partitioning graphs partitioning finite element meshes and producing fill reducing orderings for sparse matrices The algorithms imple 170 mented in METIS are based on the multilevel recursive bisection multilevel k way and multi constraint partitioning schemes http glaros dtc umn edu gkhome views metis The METIS decomposition works well for large meshes where the connectivities do not change during the course of a simulations When the model consists of multiple lists contact conditions special constraints it is more complicated to use METIS for assigning the work among all the cores In this respect the ORB is easier to program and execute 19 2 Treating lists In this section we discuss the MPI strategy for treating different lists Basically there are three types of lists e Lists type L1
24. 3 D shapes start penetrating each other It is possible that in some cases a node can go through a face Here are some scenarios e Nodes are small faces are thin and the maximum contact force that can be gener ated Fn max K Tnode tface 2 is not suffient for preventing crossing Remedy increase penalty stiffness 158 e The relative velocity of the objects is so large and the time step so coarse that the contact algorithm does not have a chance to work Remedy reduce time step to achieve accuracy Although it would be possible to insert logic in the code to detect these scenarios the computational cost would be considerable It make more sense for the user to be aware of these potentatial problems and take corrective actions when they occur The input commands for face contact list take the format NodeFaceContactList Name variables with the sign inherit defaults from control parameters 1 Specify lists NodeList NodeListName FaceList FaceListName 2 Enter minimum face thickness if necessary FaceThickness 0 2 cm 1 3 Enter node radius control parameters if necessary NodeThickness 0 4 cm 2 if nodes are particles node thickness is the larger of thn and node radius 5 For contact purposes nodes cannot be larger than MaxNodeRadius MaxNodeRadius 0 6 cm 6 All nodes are assigned a contact radius FixedNodeRadius independently of their actual radius FixedNo
25. 53 cm3 InjectionHistory name2 Enter the name of fluid or BulkModulus Oil K 1 5e9 Pa Water K 2 2e9 Pa Air 1 42e5 Pa BulkModulus 10e6 psi 5 8 3 Adiabatic Compression This feature was inserted in the code in the spring of 2009 The purpose of this feature is to enable the numerical simulation of the effect of air trapped inside a cavity whose volume changes during the simulation The pressure in the cavity is related to the change in volume using the adiabatic equation p t V t const where p t is the pressure V t is the volume of the cavity and y is the adiabatic index of the gas The ability to model these conditions was introduced into the MARS code through a new C class denoted as AdiabaticEquation This class is derived from the base class Equation The AdiabaticEquation class implements the very simple adiabatic equation p t Vo V t po where po and Vo are the initial value of the pressure and volume respectively The method implemented in Mars actually returns the overpressure which is defined as 35 Po t p t po The input block requires three parameters as shown below Equation name AdiabaticCompression InitialVolume 34 53 cm3 InitialPressure 14 7 psi Gamma 1 4 InvertSign The InvertSign command should be used when the surfaces point toward the cavity rather than away from it 5 9 Material Models 5 9 1 Elastic Material Model The elastic material model implem
26. 56 165 2 56 238 121 78 if mesh is internally generated enter Generate see below for generation options This command is also available MergeAllTetLists 1 3 Modify mesh optional commands ImproveMesh remesh locally to eliminate slivers FlipMesh D D X Y or Z flip mesh to make changes on the node list EditNodeList nodlist commands MergeNodes tol tol is a dimensioned length 4 Create additional lists optional Make TriangFaceList ListName generate external faces Make TetSolideList ListName generate list of selected tets Make EdgeList ListName SharpEdgeds generate edge list 5 select commands Select Unselect see below use this command to save the mesh in a separate file Write PartMeshDataFile PART mrs use this command to read nodes and elements previously saved Read PART mrs use this command to write a table where for each node we list its connectivity to other nodes number of tets etc Write ConnectivityData FileName 1 This command is used to combine meshes from all previously defined tet solid lists Currently the original nodes from the parents lists are referenced in a new node list the elements from the previous list are also referenced in the new list For this reason this command should be used in an intermediate mesh manipulation input file The combined mesh should be written to
27. ConstraintList ListName 4 SlaveNodeMasterNodesConstraint 1 SlaveNodeMasterNodeConstraints 1 SlaveRigidBodyMasterNodeConstraint 1 SlaveNodeMasterEdgeConstraint 1 SlaveRigidBodyMasterEdgeConstraint 1 NodeNodePenaltyConstraint 1 HingePenaltyConstraint 15 6 1 Slave Node Master Node Constraint This list is designed to create a series of master slave constraints that tie overlapping nodes from different lists One list is the master list and the other is the slave list A tolerance must be provided The tolerance should be a very small length such that the distance of the overlapping nodes is less than the tolerance Optionally only the translational degrees of freedom can be constrainded while the two nodes can rotate independently ContraintList ListName 4 SlaveNodeMasterNodeConstraints SlaveNodeList ListName MasterNodeList ListName Tolerance 1 mm TranslationsOnly Debug 15 6 2 Slave Rigid Body Master Node Constraint x TOBEIMPLEMENTED This constraint is designed to constrain a slave rigid body to translate with and or rotate around a master node SlaveRigidBodyMasterNodeConstraint RigidBody RigidBodyName MasterNode NodeListName cl 5 cm 2 cm 5 cm Set Translations XXX constrained translations Set Rotations 000 free rotations 141 15 6 3 Slave Node Master Nodes Constraint This constraint is designed to constrain a slave
28. MARS code Multiple parts can be generated within a single list and possibly fused together by merging the overlapping nodes using the MergeParts command The generate commands create a new list of nodes if necessary and appends the new nodes to the current list of nodes HexSolidList ListName ListType Generate Cylinder ReferenceSystem RFSY cyl axis z axis Radius 4 in EdgesOnCircle 24 must be multiple of 8 Length 8 in ElementSize 1 in Sphere ReferenceSystem RFSY cyl axis z axis Radius 4 in radius of cylinder EdgesOnCircle 40 must be multiple of 8 Translate 3 in 0 in O in rotate quad face list Q around z axis of R for A degrees Rotate Q SECT A 360 d 20 R CYLN Extrude QuadFaceList Disk Length 4 in ElementSize 0 4 in wall h 96 w 172 t 12 B 125 Block 4 ReferenceSystem RSYS Dimensions 4 in 8 in 20 in Elements 2 4 10 Move 2 in O in 4 in SuperBlock LengthUnits cm format Dir crdi elements1 crd2 elements2 crd3 X 8 4 2 42 48 0 Y 8 4 2 42 48 0 Z 8 4 2 42 48 0 MergeParts 0 001 in 12 3 Make Commands 1 The Make HexSolidList is used for generating a sub list of hex elements which have been selected The new list is of the Geometric type in other word MARS will not try to compute internal forces 2 The Make QuadFaceList is used for generating a list of all quadrilateral external surfaces of the c
29. MergeNodes 0 001 in tol 0 001 in 1 This command is used to generate a flat plate with an arbitrary polygonal outline containing any number of circular holes The polygonal outer outline is defined entering the coordinates of the vertices of the polygon This generation scheme employs the external program triangle 2 Enter one line for each hole The format is C xc yc rad where xc and yc are the coordinates of the center and rad is the radius The hole should be fully contained inside the master polygonal surface 8 1 3 Make commands The Make TriangFaceList is used for generating a sub list of triangular faces which have been previously selected The new list is of the Geometric type in other word MARS will not perform any operation on it e g computer internal forces The Make EdgeList command is used to generate a list of edges from the edges of the triangular faces This command must be terminated by a keyword for the edge selection criterion three options are available Make EdgeList ListName AllEdges Make EdgeList ListName SharpEdges Make EdgeList ListName UnmatchedEdges The option AllEdges is used to generate a list of all edges of the triangular mesh The option SharpEdges is used to generate a list of edges for which the attached faces form an angle greater than 60 degrees The option UnmatchedEdges is used to generate a list of edges which are shared by a single face note that if the surface is closed e
30. NULL obLst Model readRstUserDefinedList RstReader rst string nam return NULL createUserDefinedObject and readRstUserDefinedObject must always be defined Obj Model createUserDefinedObject string nam Reader rdr rdr gt error No user object has been defined return NULL Obj Model readRstUserDefinedObject RstReader rst string nam 1 return NULL Currently only Unix type platforms are supported The modified executable is generated using a Makefile of this type 186 CMP system dependent compiler COP compiler dependent options LNK system dependend linker LOP linker dependent options HDR folder where mars h is located LIB folder where libmars a is located marsU user o LIB libmars a LNK OPT o marsU user o L LIB lm lz lmars user o user cpp CMP I HDR COP o user o user cpp For example Note that the compiler has to be able to find the header file mars h This can be accomplished using the I HDR option in the compile line 21 7 1 User defined list As described in other sections of this documentation package lists are used for imple menting collections of homogenous objects such as finite elements contact elements constraint elements etc This subsection is designed to help users create their own ele ments and lists that operate on them Complete documentation on this subject would require extensive information At this
31. analysts to convert it to the correct dimensions in English units A list of available dimensional units is given in the next subsection 4 1 1 Dimensional Units Below is a list of implemented dimensional units and physical quantitities which are typically used for specifying input parameters of MARS structural models This list is often updated with new quantitities and units depending on what new features are inserted in MARS If the wrong units are used for an input parameter MARS errors off and prints the available units from the list below The conventions described in The Unified Code for Units of Measure are employed For SI and CGS systems kilo x 1 000 is denoted with the lower letter k as in kg or km Mega x 1 000 000 with M giga x 1 e9 with G milli x 0 001 with m micro x 1 e 6 with u pico x 1 9 with p Quantity Nondimensional Reference unit unit no label necessary 0 01 Quantity Time Reference unit s ms 0 001 micros 1e 06 s le 06 us le 06 ns 1e 09 mcs 1e 06 15 Quantity Length Reference unit m cm 0 01 mm 0 001 km 1000 in 0 0254 ft 0 3048 km 1000 nm 1e 09 mcm le 06 um 1e 06 Quantity Area Reference unit m2 cm2 0 0001 mm2 1e 06 km2 1e 06 in2 0 00064516 ft2 0 092903 Km2 1e 06 nm2 le 18 Quantity Volume Reference unit m3 cm3 1e 06 mm3 1e 09 km3 1e 09 in3 1
32. and then to create a full model which employs the mesh generated earlier In the generation phase the mesh is saved using the Write PartMeshDataFile command The data file has the following structure InsertMaterial CONC RCConcrete Dev InsertNodeList PRTCXA ReadObjects n element connectivity EOF Note that unlike other mesh datafiles the LDPM datafile includes commands for defining the material object that was used to size and distribute the LDPM particles In older versions of Mars it was possible to define a concrete material model before the TetSolidList is defined and use it in the TetSolidList in this fashion Material Concrete RCConcrete Dev gt TetSolidList ListName LDPM Material Concrete ReadFile specimen mrs In this input scenario the InsertMaterial command in file specimen mrs would have replaced the externally defined material Concrete with the material saved in the datafile The user would have been unaware of that Since May 26 2011 this is no longer possible In the current version an error trap has been placed and the material can only be defined once It is possible to change material parameters inside the TetSolidList using the EditMaterial command It is possible to change material parameters inside the TetSolidList using the EditMaterial command TetSolidList ListName LDPM ReadFile specimen mrs EditMaterial use standard input commands for RCConcrete model
33. are not translated because MARS deals with contact at the list level nodes from list A contacting faces from list B and not at the object level nodes faces Here is a partial list to be complete of features that are supported node boundary conditions solid elements shell elements beam elements load curves pressure faces The aspect ratio of Ingrid meshes can be improved by moving the internal nodes The example in Figure a at the website page http www es3inc com mechanics MARS Online HowTo htm _How_to_convert shows the mesh of a bullet generated using Ingrid Note that the aspect ratio of some element 2 elements deep from the surface can be improved by slightly moving the nodes This is done in MARS by using the Smooth command after the mesh has been imported and before it is saved The improved mesh after a series of 6 smooth commands is shown in Figure b HexSolidList H000 Rename BLLT Smooth Smooth Write PartMeshDataFile bullet mrs In the smoothing operations above the displacements of the surface nodes are constrained to ensure that the external shape does not change Thi is done by using nodal boundary conditions The Ingrid input for the model above is listed below Note how boundary conditions are used to fix the external nodes Nodes at the bottom face are free to move in the bottom plane Nodes on the plane of symmetry are free to move within the plane of symmetry The boundary conditions nee
34. are run on two processors then there may be conflicts in accessing the shared memory In any case testing on a specific problem may be the best way to figure out the optimal configuration bin bash PBS A ERDCV022215PS PBS 1 walltime 10 00 00 PBS N Ldpm4pt2 PBS q standard PBS j oe 214 PBS 1 select 4 ncpus 8 mpiprocs 2 select compute nodes requested ncpus must equal 8 mpiprocs cores node PBS 1 place scatter excl Allows PBS to use any available nodes Excludes other users from using your nodes while they are scheduled to you cd work peless 1104Jov 4pt2 export OMP_NUM_THREADS 4 mpiexec_mpt np 8 usr local u peless Mars marsM110524 B input mrs gt out032 23 5 Mars on Windows PCs PC versions of MARS are also available for computers using the MS Windows operating system However since most of the development and execution of MARS is done on Unix based computers including Apple computers which employ a Unix based operating systems we recommend that the Windows PC intended for usage has the cygwin soft ware installed As most of engineers already know cygwin provides a Unix like interface to PC users The following directives are given assuming that cygwin is available This approach makes it possible to use MARS consistently across different operating systems Note that the default installation of cygwin does not provide all the functionality that is needed You must check additional modul
35. be executed on multiple compute nodes and offers much bigger performance gains over the OpenMP version A typical script is shown below bin bash PBS A ERDCV022215PS PBS 1 walltime 10 00 00 PBS N Ldpm4pt2 PBS q standard PBS j oe PBS 1 select 4 ncpus 8 mpiprocs 8 select compute nodes requested ncpus must equal 8 mpiprocs cores node PBS 1 place scatter excl Allows PBS to use any available nodes Excludes other users from using your nodes while they are scheduled to you cd work peless 1104Jov 4pt2 mpiexec_mpt np 32 usr local u peless Mars marsM110524 B input mrs gt out032 Executing the Hybrid version of Mars Since 2009 the marsM version of mars implements the hybrid model In a hybrid execution multiple instances of the code are run on multiple compute nodes and each instance runs multiple OpenMP threads For example one can run 16 instances and each instance runs four threads This requires 16x4 64 compute cores 8 compute nodes The commands for executing this hybrid run are if you are using the C shell setenv OMP_NUM_THREADS 4 mpirun np 16 usr local u peless Mars marsM110524 B input mrs It appears that there is a slight benefit in running the hybrid version up to 4 threads When 8 threads are used there is no significant benefit This may be because as long as a processor is using its 4 cores on 4 parallel threads the four threads work very efficiently When 8 threads
36. color sheme and press the OK button 59 The following example shows the commands for generating a Paraview file which contains the exploded view of a particle list depicting the domain decomposition The parameter 1 3 controls the amount of radial motion for the domains The coordinates of the center of gravity of each domain are multiplied by this parameter A value of 1 means all particles stay where they are The larger the value of this parameter the more exploded the view looks Results are in file DomainDecomposition 000 vtu The domains may be painted using the scalar variable Domains PlotList DomainDecomposition Paraview TimeInterval 100 s ndL Particles DomainDecomposition 1 3 6 1 8 Nodal Rotations For all finite element and discrete particle formulation in MARS is currently sufficient to keep track of rotation rates Actual rotations the change in orientation of the nodes or particles are not used because most formulations employ an incremental approach for computing stresses and forces There are situations where an analyst needs to process node particle rotations To satisfy this requirement we have inserted in MARS the capability of keeping track of particle rotations by introducing a new class NodeQ derived from class Node and class Quaternion Class Node is used to instantiate all nodes and particles for most models Class Quaternion implements quaternions a mathematical construct which can be view
37. command must be terminated by a keyword for the edge selection criterion three options are available 91 Make EdgeList listName AllEdges Make EdgeList listName SharpEdges Make EdgeList listName UnmatchedEdges The option AllEdges is used to generate a list of all edges of the quadrilateral mesh The option SharpEdges is used to generate a list of edges for which the attached faces form an angle greater than 60 degrees The option UnmatchedEdges is used to generate a list of edges which are shared by a single face note that if the surface is closed e g a hollow prism all edges are matched and the generated edge list will be empty The list generated using the SharpEdges command includes unmatched edges 9 2 Pressurized Quad Face List These lists are used for applying pressure loadings on a set of quadrilateral faces Several loading types are available 1 Uniform pressure loading 2 Variable pressure loading 3 Special loads 4 Patched pressure histories The general syntax for these lists uses the format below More details are given for each specific loading tpye QuadFaceList listName loadingType 9 2 1 Uniform Pressure This list is used for prescribing a time dependent uniform pressure history over a set of quadrilateral faces The pressure can be prescribed either using a LoadCurve tabulated expression or using an Equation The latter is particularly useful w
38. currently only NodeList belongs to this type Each rank containts the full description of all objects in these lists An object in one of these lists can only be owned by a single rank but can be shared by multiple ranks An object is shared by a non owner rank when it is used in the definition of other objects that are owned by that rank For example rank 7 owns a tetrahedral element two of the four nodes used to the define the element are also owned by rank 7 but the other two nodes are owned by a different rank The latter two nodes are marked as shared by rank 7 e Lists type L2 most of finite element lists belong to this type Each rank contains a full description of the objects it owns and a reduced geometric description of all the other objects it does not own This makes it possible to perform certain operations like contact detection by all ranks at any time e Lists type L3 variable lenght lists like contact condition lists belong to this tpye Typically the objects in these lists are created during the initialization phase During the initialization phase the objects are created only in the rank that controls the domain where the objects are located located 19 3 MPI Logic This section explains the logic currently implemented for synchronizing the message passing between the processes With reference to the solver loop previously discussed the operations for synchronising nodal information at the inter dom
39. elastic contacts where some energy is dissipated This is accomplished by using different paths for loading and unloading The initial loading follows the conventional Hertz relationship Fn H p 1 5 Unloading follows a different relationship Fn C pb where b is a non dimensional parameter which is greater or equal 1 5 C FX pX b is constant pX is the penetration when unloading begins FX is the contact force when unloading begins For b 1 5 the unloading curve is the same as the loading curve and there is no hysteresis For b 2 the unloading curve is a parabolic curve with apex at the origin and intersecting point pX FX The area between the loading and unloading curve represents the energy dissipated in the loading unloading event The larger the value of b the more energy is dissipated A default value of 3 is used when no value of b is entered Reloading during the unloading phase is ruled by a special relationship Reloading follows a steeper line with the stiffness constant defined as the tangent to the unloading curve at pX Fn K p py where py is the penetration when reloading begins begins 149 Kr b C pX b 1 is the stiffness The normal force F cannot exceed the F Hp value in other words the initial loading curve acts as an upper envelope The input commands can be enter either in this format ContactModel HertzHyst E 29e6 psi pr 0 3 beta 3 or ContactModel HertzHyst YoungsMod
40. else if ttLst processCommand rdr 1b1 else if processCommand rdr 1b1 else my 188 rdr gt error Invalid keyword in tXLst readLst The second method void initialize is also polymorphic and is invoked during the initialization phase The purpose of this method is to initialize all elements void ttL_Cosserat10 initialize setStaticVariables TetCosserat10 tt TetCosserat10 ob for int i 0 i lt num i tt i gt init The third method real calcFrc is a polymorphic method invoked in the solver loop The purpose of this method is to compute internal nodal forces withing each element based on its deformation history and deformation rate This method looks like this real ttL_Cosserat10 calcFrc 4 if listIsNotActive return M_BIG obLst calcFrcInit prints debug info if necessary setStaticVariables set element static variable setComputeTimeStep to true if element computes stable time step obLst setComputeTimeStepF lag false creates arrays for storing node pointers and relative force and moment increments obLst createArrays 10 4 the first argument 10 is the number of nodes for which forces iy are computed nodes per element with 3 translational DoF s the second argument 4 is the number of nodes for which moments are computed nodes per element with 3 rotational DoF s Node ndF Node obF cast object pointers to node poin
41. g a hollow prism all edges are matched and the generated edge list will be empty The list generated using the SharpEdges command includes unmatched edges 8 1 4 How to make sublists It is often desirable to make sublists from a parent list to isolate a set of faces on which we may want to impose special conditions For example let s consider a tetrahedral mesh of a solid cylinder We want to create four lists 1 top circular surface 2 bottom circular surface 3 top and bottom surfaces 4 cylindrical surface 82 The sublists may be used to impose independent pressure conditions The commands for accomplishing this objective are shown below For convenience we assume that the cylinder is oriented in the z direction and the coordinates of the bottom and top faces are respectively 0 in and 4 in TetSolidList Cylinder Geometry Make TriangFaceList CylinderFaces TriangFaceList CylinderFaces EditNodeList Select cz gt 3 99 in Select FacesWithAll3NodesSelected Make TriangFaceList TopSurface EditNodeList Select cz lt 0 01 in Select FacesWithAll3NodesSelected Make TriangFaceList BottomSurface EditNodeList Select cz lt 0 01 in AlsoSelect cz gt 3 99 in Select FacesWithA113NodesSelected Make TriangFaceList TopBottomSurface InvertSelection Make TriangFaceList CylindricalSurface A more compact way to accomplish the same task is shown below TriangFaceList CylinderFaces Sele
42. input command is given by Weibull m0 4a1 g2 n where the optional n is used to normalize the distribution 29 5 4 2 Uniform distribution Uniform 1 d 0 2 numbers are generated evenly between 1 d 2 and 1 d 2 5 4 3 Gaussian distribution The Gaussian distribution or normal distribution is a common bell shaped distribution The probability density function is defined as PDF x 1 sqrt 2 PI s 2 expl x m 72 2 s 2 where s standard deviation m mean MARS generates random numbers which are consistent with a Gaussian distribution using the Box Muller transformation The Box Muller Transformation polar form gen erates a pair of random numbers which have a Gaussian distribution with a zero mean and a standard deviation of one See See http www taygeta com random gaussian html The input command is Gaussian s 2 m 0 3 mn O0 mx 4 where mn is a lower limit mx is un upper limit In other words random numbers less than mn or larger than mx are rejected mn and mx are optional inputs 5 4 4 Fuller distribution The Fuller Thompson equation is defined as P d u e where P percent finer than the sieve d aggregate size being considered u maximum aggregate size to be used c parameter which adjusts curve for fineness or coarseness Fuller c 0 5 1 0 4 cm u 1 2 cm x c fuller coefficient 1 lower value u upper value x examine 30 Th
43. longer added to the normal force when the particles start slipping thus simulating a failure in the bonding force In the current formulation f0 is added to the normal force when the slipping flag is false thus if slipping stops the bonding force is re established Note that the actual normal force used in the computation of the contact force is not affected The default value of f0 is 0 0 The parameter dt can be used for computing the penalty spring stiffness as an alterative to the PenaltyStiffness parameter When dt is selected the value of the penalty stiffness for each contact is computed using the equation K 0 5 ms dtxdt where ms is the minimum mass of any node particle participating in the contact 16 1 6 Tangential elasto plastic model The tangential elasto plastic model is an algorithm designed for computing tangential contact forces between two objects in the direction perpendicular to the normal at the point of contact The tangential force is computed incrementally based on the tangen tial relative motion of the two objects at the point of contact and a penalty stiffness parameter The tangential force is limited by a maximum allowable force fmx specified via input using a predictor corrector approach When the predictor exceeds fmx the 151 tangential force is scaled down to fmx and slipping between the surfaces occurs The tan gential force is computed only for surfaces with a gap smaller then a
44. makes it possible to compute the Z direction of the local z axis as a outer product Z X x U The Y direction is then computed as Y Z x X The best way to check that all beam have been initialized properly is to plot them 7 2 1 Time History Commands Currently it is possible to save the total axial force in a single beam element and material state variables at an integration point of an element In the example below we request the axial force of element 23 and the state variable number 2 one of the shear stresses of the element whose centerpoint is closes to a point with coordiates 0 1 0 5 0 3 in inches 70 TimeHistoryList Hist 4 bm Rebars 23 af axial force bm Rebars cl 0 1 in 0 5 in 0 3 in ip 1 sv 2 In the future it would be desirable to also be able to compute the maximum bending moment in an element 7 2 2 Plot Commands Plotting attributes can be specified in the element list or in the plot list BeamList PART PlotAttributes ContourVariable sv 1 NoNodalAveraging NoSmoothing HighResolution MediumResolution PlotList PLOT bmL PRT1 you may selecte one of the contour variables below ContourVariable StateVariable 1 ContourVariable Velocity ContourVariable X Velocity ContourVariable Y Velocity ContourVariable Z Velocity ContourVariable AxialStretch stresses are averaged at the nodes unless NoNodalAveraging NoSmoothing discrete colored fringe plots prescribe range a
45. ms If you are running MARS in interactive mode MARS goes interactive at the beginning and end of an execution When in interactive mode enter the character D at the prompt to generate a restart file It is also possible to force a restart file write during a batch execution To do so change to the working directory where input and output files reside cd work dir and enter the following at the prompt echo D gt stop 178 At every step MARS check if file stop is present It this file is present MARS reads its contents The character D in stop forces MARS to write a one time restart file after that MARS deletes file stop so that it is not present at the next time step MARS restart file names have the following format mars rst xxxxxx where xxxxxx is a 6 digit number which represents the time in the simulation when the file was created Typically the time is given in micro seconds For example mars rst 001456 was created when the simulation had reached a time of 1 456 milli seconds The time in micro seconds works for most simulations However in some cases the time scale is much shorter or longer For simulations at the nano scale simulations that employ the Nano unit system the time is expressed in pico seconds For simulations that exceed a second the number of digits may increase For example mars rst 2000000 represents a restart file written at 2 seconds Restarting an execution is relatively simple At the pro
46. node from one node list to a set of nodes from another list The master nodes are the nodes that lay in a sphere centered at the slave node with a given radius The slave node is tied to move with the average velocity of the master nodes SlaveNodeMasterNodesConstraint SlaveNode NodeListName 14 MasterNodeList NodeListName Radius 5 15 6 4 Slave Node Master Edge Constraint This type of constraint ties a node to a point on an edge The constraint is of the master slave type The constraint interpolates coordinates velocities and rotations of the slave node Rotation constrains can be disabled to simulate a spherical bearing Alternatively only the rotation along the edge can be free to simulate a wheel spinning on an axle MARS searches automatically for constraints the nodes of a node list that are within a given tolerance from the edges of an edge list are constrained ConstraintList ListName SlaveNodeMasterEdgeConstraint NodeList ListName EdgeList ListName Tolerance 1 mm FreeRotations AxialSpinning 15 6 5 Slave Rigid Body Master Edge Constraint This type of constraint ties the center of a rigid body to a point on an edge The constraint is of the master slave type The constraint interpolates coordinates velocities and rotations of the slave rigidbody Rotation constrains can be disabled to simulate a spherical bearing Alternatively only the rotation along the edge can be free to s
47. of faces EF i nt n n3 1 124 243 56 2 56 238 121 select commands Select Unselect see below make commands Make TriangFaceList Sublist list of selected faces Make EdgeList SharpEdges see below Make NodeList SelNodes list of nodes attached to selected faces SelectNodes attached to selected faces face orientation command 79 InvertOrientation invert orientation of all faces orient faces so that normal is toward direction nx ny nz Orient Direction nx ny nz orient faces so that normal points toward point Orient Toward 1 cm 5 cm 3 cm orient faces so that normal points away from point Orient AwayFrom 1 cm 5 cm 3 cm orient faces in same orientation as face 56 using wavefront method Orient Wavefront 56 Write PartMeshDataFile PRT1 mrs ReadFile PRT1 mrs 8 1 1 Select commands Select criteron AlsoSelect criteron Unselect criteron Reselect criteron InvertSelection criterion can take the following forms all all faces for 25 face 25 for 1 5 faces 1 through 5 for 1 100 2 faces 1 through 100 step 2 FacesWithAtLeastiNodeSelected in short in FacesWithAl13NodesSelected in short 3n FacesPointingToward 0 5 4 in short tw 0 5 4 FacesWithCrossProduct 1 0 0 gt 0 5 in short cp 1 faces such that cross product with 1 0 0 is greater than 0 5 In addition it is possible to select or unselect nodes attached to the faces that ar
48. of hexahedral elements every microsecond When the max Von Mises stress exceed and ultimate material stress Ftu th Functi e calculation stops and the code is instructed to print various information on Fi 4 46 silent real Ftu real syP real Mx logical lg Ftu 67000 syP hxL Pin MaxVonMisesStress lg syP gt Ftu if ig Mx RB X2 mx dd H H echo Final results print su print Ftu echo Mx moment at failure print Mx COCO Aaa srt eS a RS RSS a i QUIT ControlParameters ExecFunction F1 every 0 001 ms J The output at the end of the run looks like this Final results Ftu 67000 Mx moment at failure Mx 1 78388e 06 5 11 Pre Set Test Simulations This is a collection of objects for performing simulations of simple tests such as concrete biaxial triaxial or Colorado tests 5 11 1 Biaxial Test Model restrictions 1 Specimen must be a cube or a prism 2 Displacements imposed on plates perpend to z axis 3 Ratioed forces imposed on plates perpend to x axis Test TestName Biaxial 47 TetSolidList Specimen specimen list VelocityHistory History StressRatio 0 5 lateral over vertical h TimeHistoryList HIST teL TestName VerticalStress teL TestName VerticalStrain teL TestName HorizontalStress teL TestName HorizontalStrain 5 11 2 Triaxial Test Model restrictions 1 Specimen must be a cube or a prism 2
49. one of the probability functions Weibull dow 4 4 3 Cracking dmn 0 01 in efl 0 2 random fe 100 N m defaults dmn 0 efl 0 129 13 Rigid Bodies A MARS rigid body consists of a set of nodes kinematically tied together There are two main ways to select the nodes that form a rigid body 1 within a node list select all nodes or a subset of nodes and enter the node list for processing 2 enter an element list for a specific mesh As an example of the second way let s say we want to convert a solid body discretized using an hexahedral list into a rigid body we first need to define the solid using an HexSolidList specifying the density and not the material HexSolidsList HexListName Geometry Density 7 8 g cm3 NodeList NodeListName ReadObjects and then define the rigid body RigidBody RigidBodyName HexSolidList HexListName In either case the geometry and material density are used to computethe nodal masses In turn the nodal masses are used to compute the rigid body total mass and moment of intertia tensor Several types of objects can be tied together to form a rigid body All nodes used to define these objects are enslaved to the translation and rotations of the master rigid body Currently the following lists can be used for the definition of a rigid body ndL NodeList hxL HexSolidList gfL QuadFaceList qsL QuadShellList tfL TrngFaceList tsL TrngShellList bmL BeamL
50. point like nodes then their radius is zero Contact can be enforced as long as the other list has spherical nodes with finite radii It is not possible to enforce contact conditions between two nodes both of which have zero radii NodeNodeContactList GRNL 4 155 GranuleList GRNS 16 2 1 Time Histories The variables X Force Y Force and Z Force produce the sum of all contact forces in each direction The variable NumberOfContacts gives the total number of potential contacts being monitored The variable NumberOfActiveContacts gives the number of nodes that are actually in contact TimeHistoryList ListName ncL ListName X Force ncL ListName Y Force ncL ListName Z Force ncL ListName InternalWork ncL ListName NumberOfContacts ncL ListName NumberOfActiveContacts 16 3 Edge Contact List The edge edge contact makes it possible to enforce contact conditions between edges either from the same list or from two different lists For the purpose of contact an edge takes a three dimensional shape consisting of a cylinder capped by two hemispherical surfaces the shape of some commonly found pills The hemispherical surfaces are cen tered at the two nodes defining the edge The radius of the cylinder and hemispherical surfaces is either provided by the edge lists or defined via input Contact between two edges begins when the exernal surfaces of the associated 3 D shapes
51. prescribed maximum gap gmx This logic is summarized in the following pseudo code if gap lt gmx ftn K dv dt increment forces if ftn gt fmx ftn fmx ftn scale forces else ftn 0 The input commands are TangentialPlasticForce PenaltyStiffness 10 N m short notation K BoundingForce 10 nN MaximumGap 2 nm The PenaltyStiffness parameter is somewhat arbitrary Physically it represents the small local tangential deformations that take place when shear forces are present In some configurations like discrete particle models the tangential penalty stiffness may affect the overall elastic stiffness of the system This may be relevant in the small deformation range However the plastic behavior should not be affected significantly The best way to assess how sensitive results are to this parameter is to run the same case with different values of the parameter The MinNormalForce is an optional parameter intended to approximate the effects of bonding forces which oppose tangential motion even when the particles are not pressed against each other The last parameter in the line is the maximum gap for this effect to be included The minimum normal force f0 is added to the normal force when the particles are still sticking and the gap is less than the maximum gap This virtual normal force increment results in a larger maximum static friction force f0 is no longer added to the normal force
52. start penetrating each other It is possible that in some cases edges cut across each other Here are some scenarios e The edges are thin small radii and the maximum contact force that can be gen erated Fama K ri r2 is less than the contact force required to prevent crossing Remedy increase penalty stiffness e The relative velocity of the edges is so large and the time step so coarse that the contact algorithm does not have a chance to work Remedy reduce time step Although it would be possible to insert logic in the code to detect this scenarios the computational cost would be considerable It make more sense for the user to be aware The input commands for the edge edge contact list take the format 156 EdgeEdgeContactList ListName variables with the sign inherit defaults from control parameters 1 Select lists EdgeList FirstEdgeListName EdgeList SecondEdgeListName 2 Edge1Thickness 0 4 cm minimum radius for edges of list 1 1 Edge2Thickness 0 2 cm minimum radius for edges of list 2 1 3 Define detection distance DetectionDistance 0 6 cm 12 4 Define update interval UpdateInterval 0 000010 s 6 discontinued Jun 2011 MinimumDistance 0 1 cm 5 Enter contact models 3 ContactForce PenaltyHyst FrictionForce RollingResistance 6 Other commands WritePlotFile 4 Debug 6 1 The Edge1Thickness parameter is used to pr
53. style which resembles the C Java styles with long descriptive keywords Comments can be entered using the characters at any place in a line similar to the Fortran exclamation mark Entire sections can be commented out using the sequence at the beginning and the sequence at the end The C based syntax style makes it possible to take advantage of the vim vi modified editor syntax checker available on most Unix based systems including Linux The vim syntax checkers paints words in a text file with different colors to differentiate number comments keywords etc The MARS syntax checker file is based on the C file with some modifications modifications Each block of data is limited by curly brackets as shown in the previous example Within each block there may be sub blocks of data which are also contained in curly brakets For example HexSolidList SolidPart FBSinglelP NodeList Nodes EditNodeList Move 2 in O in O in J The commands inside these blocks are intended to be indented with the number of spaces proportional to the level of indentation Although MARS does not enforce indentation it is good practise to consistently indent an input file for example indent by two spaces for each level of indentation Indentation makes the input more readable 3 2 The Solver Loop The solver loop performs a sequence of tasks A summerized version of the solver loop is listed below This is specially useful for
54. that represents the level of refinement of the external mesh For a value of 1 lowest resolution the sphere is approximated with an octahedron For each increment each triangle is subdivided into four triangles making the mesh finer The level of resolution should be consistent with the internal mesh This can be visually checked using the Split command which makes it possible to split the sphere in two parts across the middle and look at its interior 11 2 4 DogBone Specimen DogBoneSpecimen Material Concrete Seed 5405 h 4 mm typical mesh length Long for defining long specimen Plot plot intermediate steps DpVolume automatically generates particle list PRTC and tet mesh given a triang face defined volume tfL VOLM Seed 23423 seed for random number generator FromExternalFaceList listName 11 3 Time Histories The LDPM elements employ the same time history commands as the regular tetrahedral element The only additional feature is the ability to plot state variables at any of the twelve facets 112 tt listName jt Facet jf StateVariable jsv where jt is the index of the Ldpm element jf is the index of the facet 1 through 12 and jsv is the index of the state variable 11 4 Using Pre Generated Meshes The typical way of performing LDPM simulations is to first generate a LDPM mesh of the specimen or component using one of the techniques described in the previous section
55. the executable have been generated can be placed in the PATH export PATH PATH TETGEN FOLDER TRIANGLE FOLDER Note in the lines above TEGEN FOLDER TRIANGLE FOLDER and BIN FOLDER must be replaced by the names of actual folders 24 1 Triangle Triangle generates exact Delaunay triangulations constrained Delaunay triangulations conforming Delaunay triangulations Voronoi diagrams and high quality triangular meshes The latter can be generated with no small or large angles and are thus suitable for finite element analysis Triangle was developed by Dr Jonathan R Shewchuck and is available at http www cs cmu edu quake triangle html Triangle is used in Mars for generating triangular meshes of flat surfaces 24 2 Tetgen TetGen is a program to generate tetrahedral meshes of any 3D polyhedral domains TetGen generates exact constrained Delaunay tetrahedralizations boundary conforming Delaunay meshes and Voronoi partitions Triangle was developed by Dr Hang Si and is available at http wias berlin de software tetgen Tetgen is used in Mars mainly for generating tetrahedral meshes for LDPM models 25 Post Processing Software 25 1 Quasar QUASAR is the graphical postprocessor for the MARS results There are two version of the QUASAR program The first version is based on the GLUT library The second version is written in JAVA Both versions have been ported to various operating systems The JAVA version is no long
56. the integration points both in time and space The scalar field mesh and the finite element mesh do not need to be the same However the scalar field mesh should contain all integration points of the finite element mesh The input commands for this list are TetSolidList listname ScalarField define tet solid mesh using standard commands such as InsertNodeList ReadObjects no tets enter list specific commands NumberOfStates n 1 TimeUnits h 2 102 ReadDataFile filename PlotDataSet 3 This list must be entered before it is used in other lists In the example below we define a history for the humidity distribution in a volume of concrete in list Humidity In the LDPM list the scalar field HumidityField is assigned to the state variable labeled Humidity Humidity TetSolidList HumidityField ScalarField TetSolidList listname Ldpm SE TetList HumidityField StateVariable Humidity f The data file is an ASCII file with the following structure t 0 f 0 1 0 2 f 0 3 0 nnd t 1 f 1 1 f 1 2 f 1 3 f 1 nnd t n f n 1 f n 2 f n 3 f n nnd The function scalars f t i must be separated by at least a space and can be written in multiple lines nnd is the number of nodes in the mesh 1 The number of states is used to read the first n datasets in the datafile If NumberOfStates is missing or n is greater than the number of states
57. the user with the List keyword is used for the domain decomposition As such the user should select the most computationally expensive list in the model In this case computational cost is defined as the number of elements in the list times the computational cost of each element The center points of the elements are then used to drive the ORB process and split the domain in a set of subdomains Since the number of subdomains doubles at each step the total number of subdomains is a power of two For this reason the number of core requested for an MPI run should also be a power of two number If a different number of cores is requested MARS will error off with a message The ORB is extremely fast and a large number of points in the order of millions can be split in multiple subdomains in less than a second on most current processors Once the domain decomposition has been completed the whole three dimensional space is subdivided in finite or semi infinite parallelepipedal regions Each region is assigned to a single core Thus the number of regions has to be equal to the number of cores been requested Although it is possible to specify FirstNode and NodeSpan these features are temporarily disabled and MARS will perform the domain decomposition based on the number of cores available The domain decomposition is used to assign the objects of most lists to the cores Some lists like PlotLists and TimeHistoryLists are excluded from this process
58. this approach is that the damping rate varies with the integration time step In MARS the scaling factor is computed as to achieve a desired velocity reduction per unit time 1 microsecond To make the approach more flexible the relaxation profile is defined as a function of time in a LoadCurve object Since the variables of a load curve require dimensional units the dynamic relaxation command must be specified father down in the input stream as shown in this example example ControlParameters Units English LoadCurve DynRelax X units time ms Y units nondimensional dyn relax operates in the first ms with velocity reduction of 1 per microsec ReadPairs 4 0 000 0 01 1 000 0 01 1 001 0 00 10 000 0 00 ControlParameters DynamicRelaxationCurve DynRelax dynamic relaxation To stop dynamic relaxation on restart use ControlParameters NoDynamicRelaxation 4 8 Gravity Gravity loads are automatically computed by specifying a gravity acceleraction The command can be placed anywhere in the input file after the ControlParameters section 23 The direction of the acceleration can be prescribed either using one of the three axis labels X Y or Z or more generically using the cosines of a generic direction ControlParameters Gravity 9 8 m s2 Z or Gravity 9 8 m s2 Direction 0 0 1 It is possible to apply the gravity load progressively in time using a time hi
59. those who intend to write user defined objects and lists while time lt terminationTime Y update MPI domains if necessary mpiDomains update reset nodal forces and moments to zero for jL 0 jL lt numLists jL list jL gt clearNodalForces compute nodal internal forces apply external forces for jL 0 jL lt numLists jL list jL gt calcFrc write time history output if necessary for jL 0 jL lt numLists jL list jL gt writeTimHistRecord reduce forces for master slave formulations 13 node loop uses inverse list order for jL numList 1 jL gt 1 jL list jL gt reduceFrc apply constraints for jL 0 jL lt numLists jL list jL gt applyConstraints write 3 D plot file records if necessary for jL 0 jL lt numLists jL list jL gt writePlotFile perform tasks from the action list for j 0 j lt numActions j action j gt exec Q integrate equations of motion for jL 0 jL lt numLists jL list jL gt integrateE0M apply kinematic conditions for master slave formulations for jL 0 jL lt numLists jL list jL gt applyKinQ update time parameters time dtl numSteps go interactive if requested checkSignal 4 Control Parameters The ControlParameters section follows the title line and is used for defining parameters that control the simulation or global parameters
60. to an adjacent face WARNING this constraint uses a mixed formulation the motion normal to the surface is imposed using master slave constraints It has not been possible to treat both formulation correctly and the internal forces in the perpendicular direction are not accounted properly This algorithm may have some errors for high strain rates 2 In MPI executions node face constraints are distributed over the processors Two options are available 1 the processor that owns the face also owns the constraint 2 the constraint is owned by the processor that owns the node The first option is the default Fixed cylindrical surface TrngFaceNodeBondList BNDS NodeList NODS lt ndL NODS gt FixedCylSurface ssm 0 0012 csm 0 00015 nsm 0 5 mm pnt 0 mm dir 0 0 1 rad 5 mm tol 0 0 mm The detection of the node face pairs is automatically done It is a good idea to check that the detection process found all the pairs This is done in interactive mode by selecting 136 the contraint list and entering P for plot at the tbL gt prompt This generates a plot file named tbL pl1t that can be viewed with Quasar 15 2 Node Tet Constraints This constraint is used to tie a set of nodes to a corresponding set of points inside a tetrahedral element mesh Each node and corresponding point share the same spatial location at time zero Both penalty and master slave formulations are available For the penalty formulation ther
61. to redirect the instantiation of object or lists to the readRstUserDefinedObject and 199 readRstUserDefinedList respectively If a single object or list are defined then these two methods are trivial Obj Model readRstUserDefinedObject RstReader rst string nam 1 return new myNewObject nam obLst Model readRstUserDefinedList RstReader rst string nam 1 return new myNewList nam If multiple objects or lists are defined in user cpp then the readRst methods must be able to instantiate the proper object or list This can be accomplished using the names of the objects or lists for example Obj Model readRstUserDefinedObject RstReader rst string nam if nam LoadHist return new myLoadCurve nam else if nam Steel return new myMaterial nam Using object list names makes the method above dependent on the input file If it is necessary to make the processure input file independent ES3 will help you write the proper coding The methods writeRst and readRst are used to save the data to the restart file and then read it into memory during the restart procedure Writing and reading must be consistent Because tags like u0 and ud are used in time history lists and plot lists to identify objects and lists the user must ensure to use these tags For example UserDefineList listName 4 PlotList plotListName udL listName 21 7 7 Time History Variable
62. up procedure Paraview provides great capabilities for changing properties and attributes of graphical components and can process very large models Unlike Quasar which combines components in a single files Paraview requires graphical components to be specified in different families of files As such the way PlotLists are specified for the two codes is different For example if we want to generate a contour plot of external facets in solid color and contour plot of crack opening for the internal facets the Quasar input would be set up using a single PlotList PlotList CrackOpening 1 Timelnterval 01 ms ttL Tile ExternalCellFacets ttL Tile FacetVariable 15 MinimumCrackOpening 0 01 mm RangeMinValue 0 01 mm RangeMaxValue 0 1 mm while the input for for Paraview must be broken up into two separate PlotLists PlotList TileExtFaces Paraview TimeInterval 01 ms ttL Tile ExternalCellFacets PlotList Contours Paraview TimeInterval 01 ms 115 ttL Tile FacetVariable 15 MinimumCrackOpening 0 01 mm RangeMinValue 0 01 mm RangeMaxValue 0 1 mm 11 6 1 Plotting cell facets The Cells plot option makes it possible to combine external and internal cell facets into a single triangular face list As discussed earlier the visibility of the internal facets is controlled by the MinimumCrackOpening parameter Decreasing this parameter makes the number of internal facet larger slowing down their rendering PlotList
63. w 10 mm m material t thickness height width x offset y offset lt Agp Hat section kx x a h la y b O x sec h m MATE t 1 h 10 W15 a4 material thickness height flange width flange width total width W a b a x offset y offset lt KSN PAB 73 7 3 Geometric Pair Detection The node pair lists consist of a set of lists that implement various types of interactions between two nodes or two particles recall that in MARS particles and nodes are used interchangeably Note that this set includes a master slave formulation which is a type of constraint list but not the inter particle contact list which is grouped with the contact lists 7 4 Geometric Pair Detection NodePairList ListName Geometry NodeList ListiName NodeList List2Name DetectionDistance 0 6 cm NodeiThickness 0 4 cm Node2Thickness 0 2 cm 7 5 Master Slave Constraints This list consists of a set of master slave constraints between pairs of nodes The con straint formulation makes it possible to release some degrees of freedom with respect to a local co rotational reference system which rotates along with the master node Each constraint is defined in a single input line The input consists of two nodes the first node is the master node the second node is the slave node The nodes can be in the same list or different lists No requirement is made on whether the nodes are overlapped If nothing else is specifie
64. z1 12 x2 y2 z2 in xn yn zn numcs m list i1 jit j12 12 j21 j22 im jmi jm2 smooth fringe fmin fmax f11 f12 f21 f22 fmi fm2 eoL Triangular Faces Triangular face lists are used for rendering surfaces consisting of triangular faces These may be generated from triangular shell meshes or external faces of tetrahedral meshes tfL LABL front color back color numpt n crd ii x1 yl z1 i2 x2 y2 z2 in xn yn zn numtf m i1 j11 j12 j13 i2 j21 j22 j23 im jm1 jm2 jm3 eoL Contour and fringe plots are possible tfL LABL numpt n crd ii x1 yl zl 12 x2 y2 z2 222 in xn yn zn numtf m i1 j11 j12 j13 12 j21 j22 j23 im jm1 jm2 jm3 smooth fringe fmin fmax fil 12 f13 21 22 23 fm1 fm2 fm3 eoL Quadrilateral Faces Quadrilateral face lists are used for rendering surfaces consisting of quadrilateral faces These may be generated from quadrilateral shell meshes or external faces of hexahedral meshes qfL LABL front color back color numpt n crd ii x1 y1 zl i2 x2 y2 z2 in xn yn zn numgf m i1 j11 j12 j13 j14 i2 j21 j22 j23 j24 im jmi jm2 jm3 jm4 eoL Contour and fringe plots are possible qfL LABL numpt n crd ii x1 y1 zl i2 x2 y2 z2 in xn yn zn 223 numgf m i1 j11 j12 j13 j14 12 j21 j22 j23 j24 im jm1 jm2 jm3 jm4 smooth fringe fmin fmax f11 f12 Tis f14 f21 f22 f23 f24 fm1 fm2 fm3 fm4 eoL Example In this complete listing of an acutal file n
65. 0 amp amp var 1001 stringstream oss oss lt lt ttL lt lt name lt lt lt lt elt lt lt internal work return oss str else return ttLst makeVariableLabel elt var and the getValue method in the user defined element class real Tet_UD getValue int var if var 1001 return wrk else return Tet getValue var 21 8 report a problem If you are experiencing problems while setting up MARS input files and you are re questing help from ES3 staff please use the following procedures for providing adequate information e Provide a brief description of the problem you are trying to solve and of the MARS model which you intend to use The first two subsections of the problems described in the MARS Example Library document should give an idea of what would be useful even though it is not necessary to go in great details Figures of the geometry can be useful e Brief description of the difficulty encountered while setting up the input files or executing the simulation screen captures can be useful 202 e If possible it would help if you can share the input files so that we at ES3 can do independent testing and debugging For clarity organize the information in an e mail in sections with section headings 1 Problem Description 2 MARS Model and Input 3 Description of Encountered Difficulties 4 Input File Location if applicable Realize that the more information you
66. 1 and n2 are also the cell indeces 5 The ScalarField command is used for specifying a time dependent spatial scalar field such as temperature humidity etc The scalar field is actually specified in a separate list which must be entered prior to the definition of the current Ldpm list For details on how to enter a scalar field list see specific sections of the manual The meshes for the scalar field and for the LDPM model can be different indeed one can use either a tetrahedral solid mesh or an hexahedral solid mesh The scalar field is interpolated at the center point of the facets of the LDPM model Furthermore the values at the facets are interpolated in time The interpolated scalar is assigned to the state variable specified in the command The state variable can be identified either by its index or its description as it appears in the material state variable table Note that the material model must be designed to accomodate the specified state variable 6 The FacetScalarField command is used for specifying a time dependent scalar filed see above directly at the facets The data is read from an input file that has the following format t f1 1 1 f1 1 2 f1 1 12 f1 2 1 f1 2 2 1 2 12 t2 12 1 1 f2 1 2 2 1 12 109 f2 2 1 f2 2 2 2 2 12 where ti is the time for the i th record fi j k is the value of the scalar field for j th element k th facet i th record The values of the scalar at the facets are
67. 2 range parameters are specified after contour or facet variable is selected and must have the units of the variable RangeMinValue 5 psi RangeMaxValue 1000 in s facets outside range are not displayed unless DisplayAboveMax facet above max are display in red DisplayBelowMin facet below min are display in blue 118 ttL Block SolidCells paraview only ttL Block Slice quasar only ttL Block DomainDecomposition 1 3 paraview only 11 6 9 Parallel processing with Paraview The MARS plot generating procedures for Quasar and Paraview are very different Quasar expects a single input file as such the contributions to each list from the various processes must be combined in MARS Paraview can read and combine files generated from different processes This eliminates the need to combine large data sets in rank zero process which could lead to memory requirement problems The Paraview files have the following name convention plotListName nnn ppp vtu where nnn is an integer rep reseting the sequential time frame and ppp is a three digit integer representing the rank of the process that generated the file In addition to these files there is an additional file name plotListName pvd that contains direction for Paraview on how to load and combine the previous files This is one of the files that shows up in the Paraview Open window and it is the one that should be selected 11 6 10 Other exampl
68. 2 The WriteStateVariableDumpEvery command is used for generating a family of ASCII files named LdpmStVarDump xxxxxx at periodic time intervals The xxxxxx string is actually the time in microseconds for example LdpmStVarDump 001240 contains all facet state variables at time 1 240 ms Each file is structured in blocks of 13 lines one block for each Ldpm tet element The first line contains the element index the following 12 lines contain the state variables for the 12 facets 3 The WriteCellDataDump command is used for writing an ASCII output file con taining all information necessary to characterize the geometry of all the Ldpm cells in 107 the model This file is printed once at the beginning of the simulation The structure of the output file is as follows For each cell Cell jc index of particle at the center of the cell Coordinates of the center particle node crd cx cy cz Total volume of cell vol V Number of outside points defining cell nxp np Local coordinates of outside points follow x y1 z1 x2 y2 z2 Number of facets internal to the solid and shared with other cells nif nf Facet connectivity and facet areas follow jit jif 11 1 11 2 11 3 A1 ni x nl y nl z pai pi x pl y pl z qi x ql y ql z sl x sl y sl z j2t j2f 12 1 12 2 12 3 A2 n2 x n2 y n2 z pA2 p2 x p2 y p2 z q2 x q2 y q2 z s2 x s2 y s2 z where jkt is the index of the tet element that contains facet k jkf is the facet index 1
69. 23 sta331 ovpr 9 33071 3 86614 190 624 C624 sta332 ovpr 10 0984 0 190 X Offset 9 54743E 04 s 23 Computing Platforms This section describes some of the specific set ups on the installation of Mars on specific computer centers It is intended to help users setting up their computational environment on these platforms 23 1 Mars on borg SCOREC The first time you use borg you need to configure the system so that it can find Mars and MPI supporting software Enter the bash shell gt bash Edit the bashre file in your main folder vi bashrc export PATH bigtmp peless Mars usr local openmpi64 latest bin PATH export LD_LIBRARY_PATH LD_LIBRARY_PATH usr local openmpi64 latest lib Insert the following two lines Activate changes source bashrc Any subsequent time you log in start the bash shell and the path is automatically updated gt bash 209 Interactive Execution To run the OpenMP version of Mars enter marsO input mrs To get the latest version of the built in manual enter marsO H gt manual 23 2 Mars on the CCNI system The first time you use ccni you need to configure the system so that it can find Mars and MPI supporting software Enter the bash shell gt bash Edit the bashre file in your main folder vi bashrc Insert the following line export PATH gpfs small MMCCpeld Mars gpfs small MMCCpeld bin PATH The MMCCpeld bin folder contains auxillary codes like tetgen
70. 63871e 05 ft3 0 0283168 km3 1e 09 nm3 1e 27 Km3 1e4 09 Quantity Mass Reference unit Kg g 0 001 1b 0 453592 1b s2 in 175 127 Ib s2 in 175 127 kg 1 ng 1e 12 nnKg le 18 mcg le 09 mg le 06 ug 1e 09 Quantity Density Reference unit Kg m3 g cm3 1000 Ib in3 27679 9 1b s2 in4 1 06869e 07 Ib s2 in4 1 06869e 07 lb ft3 16 0185 kg m3 1 nnKg nm3 1e4 09 Quantity Velocity Reference unit m s cm s 0 01 mm s 0 001 Km s 1000 in s 0 0254 ft s 0 3048 km s 1000 nm ns 1 Km hr 0 277778 mph 0 44704 Knots 0 514444 Quantity Force Reference unit N dyn 1e 05 1b 4 44822 Ibf 4 44822 kip 4448 22 nN 1e 09 Quantity Pressure Reference unit Pa psi 6894 76 ksi 6 89476e 06 lb in s 6894 76 MPa le 06 GPa le 09 dyn cm2 0 1 nPa 1e 09 16 Quantity Stress Reference unit Pa psi 6894 76 ksi 6 89476e 06 Ib in s 6894 76 MPa 1e 06 GPa le 09 dyn cm2 0 1 nPa Le 09 Quantity Energy Reference unit J erg 1e 07 ft 1bf 1 356 in lbf 0 113 nJ 1e 09 nnJ 1e 18 Quantity Stiffness Reference unit N m J m2 1 dyn cm 0 001 erg cm2 0 001 lbf in 175 127 nN nm 1 Quantity Rate Reference unit 1 s 1 ms 1000 1 ns 1e 09 Hz 1 Quantity Angle Reference unit deg rad 0 0174533 Quantity Momentum Reference unit Kg m s Kg m s 1 g cm s 1e 05 g c
71. A mesh can be simultaneoulsy translated in all three direction using the Move command Move 3 cm 0 cm 5 cm A special case of the translate command is the Align command X Align Min 4 Min Max Center In this case the code computes the mininum x coordinate and shifts the mesh in the x direction so that the minimum coordinate becomes 0 Also available are Y Align and Z Align If Max is used instead of Min then the mesh is aligned so that the maximum x coordinate is 0 If Center is used the mesh is centered around the 0 x value 6 1 5 Other Commands The Make NodeList command is designed to create a new node list that contains refer ences to the nodes selected in the current list The new list is passive in other words the list does not own the nodes and does not perform any active operation on them such as integrating the equation of motion However the node selection can be used for many purposes 1 the nodes can be used in time histories 2 the nodes can be used to set boundary conditions 52 6 1 6 Time History Commands The following line commands are intended to be used inside Time History lists to produce records of global and element variables TimeHistoryList HIST ndL NODS ke nd NODS 155 vx nd NODS cl 0 1 in 0 5 in 0 4 in vx list labels for available quantities in parenthesis value for entire list CG values at CG I volume integral cx
72. AMRC LoadCurveList is typically used in conjunction with quadrilateral or triangular face lists The mapping of load curve to face is generally done using the closest distance criterion For more information see the sections related to wet face lists LoadCurveList PRSS fmt shamrc 2 Enter location and name of data files dir cable04 TNT run2 station num 624 number of curves Read for each curve enter 1 index starting from 1 2 label 3 filename 4 station coordinates 1 C001 stal ovpr 10 0984 0 0 2 C002 sta2 ovpr 33071 3 86614 0 3 C003 sta2 ovpr 33071 3 86614 0 4 C004 sta3 ovpr 14173 7 14173 0 5 C004 sta3 ovpr 14173 7 14173 0 NN 623 C623 sta331 ovpr 9 33071 3 86614 190 624 C624 sta332 ovpr 10 0984 0 190 cgs2psi X Offset 9 54743E 04 Currently Mars input variable are always defined with their dimensions To satisfy this requirement the input will be changed to the following format LoadCurveList PRSS fmt shamrc 2 Enter location and name of data files Folder cable04 TNT run2 station NumberOfCurves 624 number of curves TimeUnits s LengthUnits cm 208 PressureUnits cgs ReadCurves for each curve enter 1 index starting from 1 2 label 3 filename 4 station coordinates 1 C001 stal ovpr 10 0984 0 0 2 C002 sta2 ovpr 9 33071 3 86614 0 3 C003 sta2 ovpr 9 33071 3 86614 0 4 C004 sta3 ovpr 7 14173 7 14173 0 5 C004 sta3 ovpr 7 14173 7 14173 0 623 C6
73. At the moment all curves are plotted using a continuous thin solid black line 25 2 1 Installation jHist requires the Java Runtime Environment JRE to be installed on your computer You can quickly check if JRE is installed by typing which java 225 If JRE is not installed google installing java and choose one of the www java com sites Installing JRE is free and simple The java class files for Hist are typically placed in a folder named MARSPATH jHist where MARSPATH is an environment variable that contains the path of the folder where Mars files are located The best way to incorporate jHist is to insert the defintion of MARSPATH and an alias in your bashrc file export MARSPATH Mars replace with actual path alias jHist java cp MARSPATH jHist jHist For PC cygwin users it is important to type the correct path the Windows path rather than the cygwin path e g alias jHist java cp c cygwin home USER Mars jHist jHist Note that a java program consists of a series of files with the class extension These files are executed through the JRE and are OS independent Thus they can be installed on any computer 25 2 2 Execution If the alias for jHist is properly set then the plots for file Hist th can be executed from the folder where the file is located using the command jHist Hist We have not been able to execute jHist and jCurv over a ssh session even when other X11 applications could
74. In February 2011 we realized that we could restructure the built in manual and generate better quality documentation Essentially using the polymorphic properties of Object Oriented C classes we created three classes that would process the same information and generate three different types of documents e The ordinary ASCII text manual already available obtained by executing Mars with the H option mars H Note that the name of the Mars executable may vary on different computer systems Check section on computing platforms e An ASCII tex file which can be processed using LaTex for generating a pdf file with a hyperlinked table of contents The LaTex file is generated using the D1 option where 1 is the lower case letter 1 e A set of html files which can be accessed using any internet browser Table of contents and hyperlinks are also available These html files are generated in the current folder using the command mars Dh The files can be accessed with any web browser 3 Code Architecture and Input Format The MARS input decks consists of a sequence of blocks called sections Each section specifies the input for an entity e g material load curve etc or a list finite elements of a component external faces of a component contact conditions etc A special block referenced as ControlParameters includes run control commands and miscellanea information The order in which these blocks of information appear in the MARS input
75. LST JohnsonCook 4 Steel 1045 39 Steel 4340 A16082 T6 Ti6A14V To plot stress strain curves for different strain rates use the Test command e g Material PLST JohnsonCook Steel 1045 Test Tensor Tension which generates a time history file named mat th 5 9 5 LDPM Concrete Material Model Material CONC RCConcrete Dev MixDesign Density 2 400 g cm3 CementContent 300 kg m3 WaterToCementRatio 0 5 AggregateToCementRatio 6 0 MinAggregate 10 0 mm MaxAggregate 16 0 mm FullerCoefficient 0 5 StaticParameters 4 NormalModulus 50000 MPa nmo DensifcationRatio 1 0 nmo Alpha 0 25 TensileStrength 3 0 MPa ftm CompressiveStrength 50 MPa fcm ShearStrengthRatio 2 5 fsm TensileCharacteristicLength 100 mm tcl SofteningExponent 0 25 ncm InitialHardeningModulusRatio 0 33 MPa IHm TransitionalStrainRatio 5 0 hts InitialFriction 0 4 mu0 AsymptoticFriction 0 0 mui TransitionalStress 50 0 MPa fts VolumetricDevitoricCoupling 0 0 beta DeviatoricStrainThresholdRatio 0 5 dk1 DeviatoricDamageParameter 5 0 dk2 StrainRateEffect 4 PseudoTimeFactor 1 0 1 40 ReferenceStrainRate 1E 6 1 s cco RateEffectParameter 5E 2 ccl 1 The PseudoTimeFactor is used to correctly integrate the constitutive equation while using a fictitiously higher rate of loading for the simulation For example assume that you have an experimental test in compression on a specimen loaded a
76. List listName MasterSlave NodeList nodeListName TetList tetListName 137 15 3 Node Hex Constraints This constraint is used to tie a set of nodes to a corresponding set of points inside an hexahedral element mesh Each node and corresponding point share the same spatial location at time zero Both penalty and master slave formulations are available The input commands for the penalty formulation are shown below NodeHexConstraintList listName Penalty NodeList listName HexList listName The stiffness constant of a constraint between node N and point P is computed using this equation E where E is the Young s modulus of the material for the hexahedral mesh and V is the volume of the element that contains point P corresponding to node N The input commands for the master slave formulation are shown below NodeHexConstraintList listName MasterSlave NodeList listName HexList listName F The mass of slave nodes N is distributed to the nodes of the master hex element containing node N using the weighting factors computed from the shape functions 15 4 Beam Tet Constraints The beam tet constraint is designed to constrain the motion of strings of beam elements to the motion of a tetrahedral mesh It is specifically intended to model rebar concrete interaction where concrete is modeled using a tetrahedral LDPM mesh The node tet constraint can be used for this purpose ho
77. List command EditNodeList 1 Select cx gt 0 SelectedNodes 7 Use line below for multiple orthogonal velocity constraints DisregardMultiple Alternate way PrescribedVelocityList ListName 4 NodeList NodeListName LoadCurve LoadCurveName Input 2 nd dx dy dz scl 54 1 0 0 2 3 133 58 1 0 0 2 5 Rigid Body PrescribedVelocityList ListName RigidBody RigidBodyName LoadCurve LoadCurveName Direction 0 1 0 In general you cannot impose multiple hard constraints on the same node For ex ample you cannot impose a prescribed velocity to a node which is part of rigid body MARS flags nodes that have hard constrained imposed on them If you intend to impose multiple prescribed orthogonal velocity conditions on the same node use the DisregardMultipleConstraints keyword keyword MPI For the purpose of MPI parallelization all processors impose velocities or ro tations rates whether they own the nodes or not This operation is computationally inexpensive and there would be no benefit in distributing it 15 Constraints Mars provides several types of constraints for modeling the interaction between differ ent parts in a model Constraints can be loosely divided into two categories penalty formulations and master slave formulations The penalty formulations are very reliable and do not create conflicts They do however only approximate the actual constraint and in some c
78. Loads 27 5 3 1 CON Web soe e Oe Eee bea eb ae o eo ARO 28 5 4 Random Number Distributions 0 0 000000 ce 29 5 4 1 Weibull distribution 0 0 0048 29 Cine hee ae eee e 30 Hee IA VUE E VU ho ge h ea 30 E oe SU ets Boe Se Be 30 AIN A ae Ace kenge ae te ee ay Ee Se ae eee os cet ete E 31 Dar BEA red eae be eS Aa 31 Dedica e y gee tee 32 paisa tras pela 32 pane ena 32 5 60 4 TimeHistorles 2 o a aa a a 33 5 7 Accelerometer ooa a a a 33 5 7 1 Attached to quadrilateral surfaces 34 5 7 2 Attached to triangular surfaces o0 oo a a 34 5 7 3 Attached to the surfaces of a beam element 34 5 7 4 TimeHistories 2 aa 34 sea a a Se es es a A 35 5 8 1 Pressurized Volume 0 0 0 0 0 20 00 00 00008 8 35 5 8 2 Mmjected Fluid e s e see fhe etn ew oe ett ot ot et ae ca E 39 A A 35 5 9 Material Models 36 5 9 1 Elastic Material Model 36 5 9 2 Elastio plastic Material Model 37 5 9 3 Rate Sensistive Elasto plastic Material Model 2 38 5 9 4 Johnson Cook Modell 39 5 9 5 LDPM Concrete Material Model 40 5 9 6 RCI Rebar Concrete Interaction Modell 41 5 9 7 LDPM Concrete Fiber Interaction Modell 42 5 9 8 Simple Cap Concrete Material Model 43 5 9 9 K amp C Concrete Material Model
79. Mars Manual ES3 August 3 2012 Contents 8 1 1 The Lattice Discrete Particle Model LDPM 9 A e a ek ok ne Ge 9 Sy aie elit th nis ee Ht Se te xi et ds do 10 O ee A ee a 10 1 5 Vehicle Response to Blast 10 1 6 Laceration of Plates and Shells 10 1 7 Steel Plates Structures 2 a 10 10 3 Code Architecture and Input Format 11 aran RRS Ree ee ee ee 13 8 2 The Solver Loop lt s ss sste dos eale d eee Seeded OP eH es 13 4 Control Parameters 14 ta He a a A ee Gre a eee 15 4 1 1 Dimensional Units a 000000000004 15 EPA soei e e E e ete ear a E a ES a SP e Ft os a e cs G 18 ee hw a yk a eea ea ee e 19 oe ee en be he bee bare eee oe 19 SETE 19 A O Ga ate ee ee ee 19 4 2 5 Flush Time Hist Files 2 002000 00000000004 19 ey ee es a A ee e 20 4 2 7 Write Plot DataFllel 20 oe Bee 6 ale oye Ges e a ra Se 20 4 3 Time Step Controll egos ao Be ae We a eT ae Ea PEW A 20 AA 21 4 5 Contact Defaults a a 22 bs a A oe A a a e 22 e as o OS hie a a ea 22 TA Sied 26H 24S Ee O4ShESREDEEEESD 6444 44 EEDA 23 de gota eH erie aia eae ee heey Eps A ae nes Ges wa ii ee a 24 25 pa a eee ge Be Be ey Be er ee te 25 oe oe hae Se a ee ee ee ata ee be 25 eh dees See Be ek See ea se Ae a e 25 O ee ve aaa ee a 26 5 2 Load QU 26 5 2 1 Load Curves Lists 0 0 0 0 0 000 000 000000848 27 5 3 Built In Blast
80. Master Edge Constraint 142 15 6 5 Slave Rigid Body Master Edge Constraint 142 eke eee eee oo eee 143 oo ee eee Bee ee ee ee 143 15 7 Particle Rebar Interaction List 0 0 00000000 eG 144 15 8 Particle Fiber Interaction List 0 00 000 ee eee 145 cana AN A Hs A es BO SE 146 16 Contacts 147 16 1 Contact Models e ara a AR Ge Wa e 147 16 1 1 Penalty contact model oo a aa 147 16 1 2 Penalty hysteresys contact model 148 16 1 3 Hertz contact modell oa aa a BAS SH 149 16 1 4 Hertz hysteresis contact modell 149 16 1 5 Stick slip friction modell 150 16 1 6 Tangential elasto plastic modell 151 16 1 7 Rolling resistance modell 000 ee ae 153 16 2 Node Node Contact List a dw ee Gea ee ee we GEES 154 16 2 1 Time Histories ee 156 16 3 Edge Contact List cuero eb ek ER PERE A ee ES 156 bgt eros a et He erin a ea ee a 158 16 4 Face Contact List lt 6 e e nee ROR RR RS er RE Bw ee ES 158 See pane hea eee eee eke 160 SA AS a eee are ee ae A 160 16 5 Rebar Contact iS mitos eee Se ew Be eh OEE 161 CA E eee ae eee Se ee ae 162 17 Interference Check 162 17 0 2 Node Hex Overlap Check o o 162 17 0 3 Node Tel Overlap Cheek ociosas wa de aes 163 17 0 4 Node Hex Overlap Check o o 163 18 Pr
81. PM define tet solid mesh using standard commands such as InsertNodeList ReadObjects no tets enter list specific commands DisableFiberFacets 1 FiberRadius 0 02 in 106 enter either an EdgeList or BeamList EdgeList FiberListName WriteStateVariableDumpEvery 0 010 ms 2 WriteCellDataDump fileName 3 WriteFacetDataDump fileName 4 ScalarField TetList listname StateVariable 17 5 ScalarField TetList listname StateVariable Temperature ScalarField HexList listname StateVariable 18 FacetScalarField 6 InputFile filename StateVariable VariableName MassScaling 7 1 The DisableFiberFacets command is used in conjunction with t ParticleFiberInteractionList to disable LDPM facets whose centers are located inside the fibers The algorith com putes the minimum distance of each facet from all edges in the edge or beam list If the distance is less than the fiber radius parameter then that facet is disabled and no longer participates in the calculation Following is an example of how to use it Obviously the beam lists in the t TetSolidList and InteractionList should be the same while the fiber radius could be different BeamList Fibers TetSolidList listname LDPM DisableFiberFacets 1 FiberRadius 0 02 in BeamList Fibers ParticleFiberInteractionList BottomHalf BeamList Fibers FiberRadius 0 02 in
82. R LEEW SE See oe 22 1 2 Gemini e read gaga da nap a a p aa a a e a a 22 2 Mechanisms List aaa a 22 2 1 Shock Strut Assembly oaa 44 2444 244 2222 Brake ues ae E gaa ae odes oe Be vee a tes ne Be Bee ee ee 22 3 Collection Listl 22 4 Unit Cell e 22 5 Load Curve Lists aa 23 Computing Platforms 23 1 Mars on bore SCOREG 60 4 0 8 es ea bee d a a 23 2 Mars on the CCNI system seas adas as ets A PAY OB 23 3 Mars on NWU QUES 2 ws eee ss mas a as o s 23 4 Mars on hpc diamond 4 64324 4404 04 Oe eee 174 174 174 177 178 179 182 185 186 187 191 192 194 195 199 200 202 203 203 203 204 204 204 205 205 205 206 207 23 5 Mars on Windows PCs 02 000 0 0002 a ee 215 215 ak es EA a a aa a a dea 216 AA A A AA A ae 216 216 AA oe E A we a A E 216 25 1 1 Quasar User Manual GLUT 32s esa sy ee be wee ew 216 25 1 2 Quasar File Format 25264 i we ee aee ba noo owe Se ex 218 o 225 bt Ban nl GG A E oe eS Be 225 ive She Sd GS ES HE ee De eo de 226 25 2 3 File format lt ad sea oeda tole ES Bae ES a A 226 o 227 25 3 1 Input file format 4 04 26 44 644 ewe Sea be ria era 227 25 3 2 Data set file formats lt a 4 qn 4d 44 4 od 46 kw E Oe ee Ee 228 1 Introduction MARS is a powerful and robust object oriented solver for simulating the mechanical response of structural systems subjected to short duration events It employs an explicit time
83. Source file for class MyFirstTetList include mars h include MyFirstTetList h void MyFirstTetList method1 Header file for class MySecondTetList class MySecondTetList public ttLst Source file for class MySecondTetList include mars h include MySecondTetList h void MySecondTetList method1 Header file for class MyHexList class MyHexList public hxLst F Source file for class MyHexList include mars h include MyHexList h void MyHexList method1 197 Header file for class MyMaterial class MyMaterialList public Material Source file for class MyMaterial include mars h include MyMaterial h void MyMaterial method1 Y i Source file for user cpp Source file for user cpp include mars h include MyFirstTetList h include MySecondTetList h include MyHexList h include MyMaterial h include MyLoadCurve h createUserDefinedList string must always be defined obLst Model createUserDefinedList string nam Reader rdr string 1bl rdr gt getShortLabel if 1bl MyFirstTetList return new MyFirstTetList nam else if 1bl MySecondTetList return new MySecondTetList nam else if lbl MyHexListList return new MyHexList nam rdr gt error Invalid keyword for user defined list type return NULL createUserDefinedMaterial string must always be defined Obj Model createUserDefinedObject string nam Reader
84. X direction LengthUnits cm required Translate 0 0 5 in global RefSys optional RebarLength 10 MaxElementSize 2 NumberOfRebars 6 optional dfl 1 RebarSpacing 4 in local Y direction EdgeList RBRS Rebar Material Steel Size 3 Generate Rebars ReferenceSystem Local 62 rebars are aligned in Local X direction LengthUnits cm required Translate 0 0 5 in global RefSys optional RebarLength 10 MaxElementSize 2 NumberOfRebars 6 optional dfl 1 RebarSpacing 4 in local Y direction BentRebars ReferenceSystem Local2 LengthUnits in MaxElementSize 1 Translate 0 0 5 Point 0 12 35 1st point Point 0 15 35 2nd point Point 0 15 30 3rd point CloseCurve Duplicate 5 10 Parallel fibers This scheme generates a system of straight parallel fibers aligned in the z direction The whole system is centered at the origin but can be later translated and reoriented The fibers are equally spaced in the x and y direction The pitch as well as the number of fibers in the x and y directions are independently defined The number of fibers in each direction should be an odd number so that there are an equal number of fibers before and after the middle fiber the fiber that crosses the origin The input commands are given below EdgeList ListName Geometry Generate ParallelFibers 4 Length 2 in X Pitch 0 25 in X Fibers 51 Y Pitch 0 25 in Y Fibers 51 T
85. YZ IBXYZB IXYZR XYZ IXYZRUVW Boundary condition code X00 fix cx free cy and cz ReadNodes 345 if InputFormat IXYZR then 5 i cx cy cz rd 1 1 4 2 4 4 5 0 12 2 1 8 2 1 4 2 0 17 else if InputFormat IBXYZB then i trn cx cy cz rot 1 00X 1 4 2 4 4 5 000 2 XXX 1 8 2 1 4 2 000 Select select nodes from list see below Set set values on selected nodes see below Scale 0 4 scale all cords by 0 4 X Scale 0 4 scale in x direction only Y Translate 0 4 in Z Rotate 45 rotate sel nodes 45 deg about z axis Move 0 3 in 0 5 in 0 3 in StructuralDamping 0 001 1 s to be implemented PlotAttributes see below ComputePackingRatio lengthUnits xmn xmx ymn ymx zmn zmx extract a node subset Extract Dogbone 1 Make NodeList SubListName 49 Write NodeDataFile PartNodes mrs CoordinateDutputFormat 10 6f ie 6 1 1 Select Commands Select commands make it possible to select a subset of nodes using various criteria The selection can be immediately used for imposing conditions or saved with a name for later processing Select criteron AlsoSelect criteron Unselect criteron Reselect criteron InvertSelection invert current node selection SaveSelection selection name criterion can take the following forms all all nodes do 25 node 25 do 1 5 node 1 through 5 do 1 100 2 nodw 1 through 100 s
86. a 12 2 Generate Commands 00000000000 2 ee 12 3 Make Commands 0000000 a ated oe Beale eee rta eee a ee 12 5 Plot Attribute Commandds 000084 12 6 Particle Generation Commands e 12 7 Fragmentation Commands 0 00005 sees 13 Rigid Bodies 13 1 Time History Omimands 5 6 0 x04 a aaa eH era ak oa Pa ee 14 Loadings 14 1 Nodal Load Listl 2 0 200 002 0000 000000000000004 14 2 Prescribed Velocities List 2 000000 e 15 Constraints 15 1 Node Face Constraint List o e 15 2 Node Tet Constraints 2 0 0000 a 15 3 Node Hex Constraints 0000 0000000000 ee 15 4 Beam Tet Constraints 0200000000 00002 ee 15 4 1 Elastic formulation 0 20020202 2000800004 15 4 2 Elastic formulation with slippage 15 4 3 Elastic formulation with slippage and volumetric effects 15 5 Beam Hex Constraints 0 0 0 00 a a 15 5 1 Elastic formulation 002020 2 000000 a 15 5 2 Elastic formulation with slippage 15 5 3 Elastic formulation with slippage and volumetric effects 123 124 125 126 127 127 128 129 130 131 132 132 133 15 6 Constraint List 140 15 6 1 Slave Node Master Node Constraint 141 15 6 2 Slave Rigid Body Master Node Constraint 141 15 6 3 Slave Node Master Nodes Constraint 142 15 6 4 Slave Node
87. ain boundaries are performed within the integrateE0M method while time lt terminationTime 4 for jL 0 jL lt numLists jL list jL gt calcFrc for jL numList 1 jL gt 1 jL 171 list jL gt reduceFrc for jL 0 jL lt numLists jL list jL gt integrateE0M apply kinematic conditions for master slave formulations for jL 0 jL lt numLists jL list jL gt applyKin Q The treatment of the boundary nodes is explained with this example Assume you have a finite element Q owned by the rank 1 process Two of its nodes 6 and 9 are also owned by rank 1 while nodes 7 and 8 are owned by rank 2 The internal forces for Q are computed in the calcFrc method The force contributions at node 7 and 8 are passed to rank 2 via an MPI message at the beginning of the integrateE0M method These contributions are added to the existing forces The equations of motion are then integrated Finally the updated coordinates and velocities for nodes 7 and 8 are passed back to rank 1 that will use them for the next time step Rank 1 Rank 2 This logic works very well for all types of list that compute internal forces including contact conditions penalty force based constraints etc The current logic is not able to handle master slave constraints This is explained with the following example Let s consider a rebar embedded in a solid tetrahedral LDPM mesh Let B be a beam element for the reba
88. al elements which are currently available is listed below e basic 4 node tetrahedral shapes e Lattice Discrete Particle Model LDPM e viscous element used to artificially dampen internal motion e rigid disconnected elements e 4 node elasto plastic element formulation e Cosserat formulation with 6 DoF s per node e explosive element which employs EoS for blast calculations e quadratic 10 node tetrahedral shapes and finite element formulations All these type are discussed in this section except for the LDPM elements which are discussed in a separate section because of the great role that this method plays in Mars The input format for tetrahedral list is given below TetSolidList ListName type where the keyword type can be one of the following Geometry this can be used to define rigid bodies Explosive this employs EOS materials Ldpm lattice discrete particle model formulation Viscous used to artificially dampen internal motion if this list is used to define a rigid body enter 1 Material properties Density 7 8 g cm3 else Material MATE You can also use InsertMaterial Mat if material was not previously defined 2 Specify mesh if mesh is explicitly defined enter NodeList Nodes or InsertNodeList Nodes 96 if node was not previously defined then enter element definition ReadObjects 345 number of elements i nt n nB n4 1 124 243
89. al strain gage is placed While physical strain gage are only placed on the surface of a part these numerical strain gage can be also placed in the interior of a part The numerical strain gage is defined using its location length and direction The main purpose of the numerical strain gage is to produce a time history record of the local strain which can be used for comparison with test data or just for better understanding of the results from the simulation The numerical strain gage computes the a total length by summing the lengths of the segment s used in the definition The length LO at time 0 is assumed to be the reference length at rest of the strain gage The linear strain eps t at time t is defined as eps t L t LO 1 where L t is the strain gage length at time t The logarithmic strain is defined as eps t In L t LO The structure of a typical strain gage input is shown below StrainGage GageName part on which the strain gage is attached to type dependent parameters More details on how to place strain gages on various types of meshes are discussed in the subsections below 31 5 6 1 Strain Gage for hexahedral solid lists Currently strain gages can only be placed on the surface of a hex solid element mesh Typical input is shown below StrainGage GageX hxL HexPartName or HexSolidList HexPartName Description Strain gage at location X5 CenterAt 0 cm 10 cm 9 cm Length 2 cm Se
90. ally forcing all nodes to have that value These bounds are used for computing the effective contact radius re of each particle using this logic rc max rad r_min rc min rc r_max Default values are min 0 Tmax Mara very large number Obviously NodeRadiusNoSmallerThan and NodeRadiusNoLargerThan should not be used when NodeRadiusExactly is used to avoid prescribing conflicting bounds 161 Similarly the radii of the beam cylinders are controlled by three optional input com mands BeamRadiusNoSmallerThan BeamRadiusNoLargerThan and BeamRadiusExactly The same logic is applied to compute and effective contact radius for each beam If the beam has already a circular shape the beam list provides a defult radius which is related to the geometric properties of the cross section for solid rebars it is the actual radius for hollow tubes it is the outside radius etc The command NoHemiSphericalCaps is optional and is used to control the 3D shape of a beam to be a plain cylinder with no hemispherical surfaces attached at the ends 16 5 1 Time History Commands The following line commands are intended to be used inside Time History lists to produce records of global list variables TimeHistoryList ListName rcL ContactListName NumberOfContacts rcL ContactListName ActiveContacts rcL ContactListName MinimumDistance rcL ContactListName MinimumGap rcL ContactListName NormalForce rcL ContactL
91. aluating different strategies for including master slave constraints in the overall MPI framework Obviously one of the requirements for any strategy is the minimization of the number of inter process mpi messages 19 4 Visualization It is possible to visualize the domain decomposition for some lists Currently this option is available for 1 node lists and 2 tet lists Details are given in the plot help sections of these lists PlotList DomainDecomposition Paraview TimeInterval 100 s ttL Tile DomainDecomposition 1 3 19 5 Examples 19 5 1 Brazilian Test The top and bottom plates and all contacts of plates with specimen are handled by rank 0 The rest of the model is split among all other processors using the ORB method Slaves nodes in contact are also assigned to rank 0 0 MpiDomainList 4 SingleNode DD1 Node O Bisect DD2 FirstNode 0 List ttL Speciment Verbose TopPlate DD1 BottomPlate DD1 Specimen DD2 PrintDetails 19 5 2 Contact between projectile and concrete slab LDPM elements are decomposed using ORB The projectile which is relatively inexpen sive is assigned to the rank 0 process Fiber and conrcrete fiber interaction constraints are distributed using the DD2 decomposition 173 MpiDomainList 4 SingleNode DDi Node O Bisect DD2 FirstNode 0 NodeSpan 32 List ttL Tile Verbose Default DD1 nd Fibers DD2 Tile DD2 FSP DD1 nc FSPTile DD2 19 5 3 Contact between particles and nanoind
92. ame 1 Refine RFN6 Refine mesh by splittying 2x2x2 select commands Select Unselect see below RemoveUnselectedElements 4 to operate on the node list EditNodeList nodelist commands use this command to save the mesh in a separate file Write PartMeshDataFile PART mrs Write ElementDataFile PRT1 mrs elements only use this command to read nodes and elements previously saved ReadFile PART mrs 12 1 Select Commands Select criterion AlsoSelect criterion Unselect criterion Reselect criteroin InvertSelection criterion can take one of the following forms all all elements do 25 element 25 do 1 5 elements 1 through 5 do 1 100 2 elements 1 through 100 step 2 vl gt 0 4 elements with volume gt 0 4 vl 0 4 elements with volume apprx 0 4 vl lt 0 4 elements with volume lt 0 4 in elements with at least one node selected 8n elements with all eight nodes selected Examples 124 Select do 1 100 select element 1 through 100 AlsoSelect do 201 300 add elements 201 300 Reselect 8n select only element with all eight nodes selected from elements selected above Hex solid lists and all derived lists include commands for selecting nodes SelectNodes UnselectNodes and ReselectNodes See examples in the NodeList section 12 2 Generate Commands These commands make it possible to generate hex meshes for simple parts within the
93. an element invert quad orientation if necessary define polygonal external outline 0 0 0 0 first node of polygon 1 0 0 0 second node of polygon BB UP Dw c close the polygon enter internal circular holes x y R s 5 min no of sides in holes optional C 0 5 0 5 0 2 3 90 3 Enter one line for each hole The format is C xc yc rad where cx and yc are the coordinates of the center and rad is the radius The hole should be fully contained inside the master polygonal surface Sphere Sphere free order R CYLN reference system r4 in radius n 3 n x 8 number of circumferential elements Duplicate Duplicate r 1 55 range x 10 time Single Element Single 4 in 6 in single element 4 in by 6 in laying in the x y plane and centered at origin Ring Ring R RefSysName reference system ring lays in local x y plane Specify either nce system D 4 in ring diameter d 2 in cross section diameter N 3 number of subs around the ring n 3 number of subs around the section 9 1 3 Make commands The Make QuadFaceList is used for generating a sub list of quadrilateral faces which have been previously selected The new list is of the Geometric type in other word MARS will not perform any operation on it e g computer internal forces The Make EdgeList command is used to generate a list of edges from the edges of the quadrilateral faces This
94. an input file that can be later used in a more complex model 97 10 1 Examples TetSolidList RigidRing Geometry 1 Density 7 8 g cm3 NodeList RigidRingNodes ReadObjects 455 TetSolidList Slab Ldpm Material Concrete Read Slab mrs 10 2 Select Commands These commands are used to select a subset of elements on which to operate Select criterion AlsoSelect criterion Unselect criterion Reselect criterion InvertSelection criterion can take one of the following forms all all elements do 25 element 25 do 1 5 elements 1 through 5 do 1 100 2 elements 1 through 100 step 2 vl gt 0 4 elements with volume gt 0 4 vl 0 4 elements with volume apprx 0 4 vl lt 0 4 elements with volume lt 0 4 in elements with at least one node selected 4n elements with all four nodes selected Examples Select do 1 100 select element 1 through 100 AlsoSelect do 201 300 add elements 201 300 Reselect 4n select only element with all four nodes selected from elements selected above Tet solid lists and all derived lists include commands for selecting nodes SelectNodes UnselectNodes and ReselectNodes See examples in the NodeList section 10 3 Generate Commands The Generate command is used for generating tetrahedral meshes of simple geometries Within the Generate subsection a single or multiple parts can be generated If multiple parts are generated some nodes of these part
95. an require signifi cant resources for large models in term of both time and memory For this reason an efficient bin sorting algorithm was developed for computing these intersections For each intersection the lengths of the fiber on both sides of the facet and the angle at which the fiber intersects the facet are also computed These parameters are saved in the facet data structure and used during the simulation for computing the incremental force effect of the fiber on the structural response of the concrete TetSolidList ListName LDPM 120 EmbeddedFibers FiberConcretelnteraction ModelName GenerateFibers FiberLength 10 mm ElementSize 2 mm Obsolete use EdgesPerFiber EdgesPerFiber 5 Specify either fiber diameter radius or area FiberDiameter 0 2 mm FiberRadius 0 1 mm FiberSectionArea 0 0314 mm2 Specify either NumberOfFibers or VolumeFraction NumberOfFibers 10000 VolumeFraction 0 02 Tortuosity 0 2 seed 345 PreferentialDirection nx ny nz scale 1 Chopped Prism 2 1 This command is used when fibers are mostly oriented in a specific direction The scale number is used to control how strong the orientation is For a scale equal to 1 there is no preferential direction For a scale equal to 10 a fiber is ten times more likely to be oriented in the given direction than to a direction orthogonal to it The best way to set this parameter is to generate fibe
96. and Side are required the seed is optional The Generate command forces MARS to write a mesh file for unit cell that has been generated The mesh file is named UnitCellMesh mrs Note that the nodes are periodic but the tetra hedrals are not This is because we have not been able to have tetgen generating periodic tet meshes when the periodic external facets are prescribed The input for the solution phase has the following format Material Concrete RCConcrete Dev 4 MixDesign 41 StaticParameters bn TetSolidList UnitCell Ldpm 4 Material Concrete ReadFile UnitCellMesh mrs UnitCell Cube Side 2 in TetList UnitCell For the solution phase the class UnitCell creates a list of constraints that tie the nodes on the surface of the unit cell Although the unit cell does not have a clean geometric 206 shape it is topologically similar to a cube There are eight vertex nodes that are tied together The vertex nodes can only rotate and their translations are set to zero There are edge nodes that are grouped in sets of four each set consists of four nodes on four parallel edges of the cube There are face nodes that grouped in sets of two each set consists of two nodes on opposite faces of the cube The internal nodes do not appear in the constraints of the unit cell In the initialization phase the masses of the nodes in each constraint are added up The summed mass is then assigned to the nodes o
97. and validated in the last few years and it has shown superior capabilities in reproducing and predicting qualitative and quantitative concrete behavior under a wide range of loading conditions 1 2 Fragmentation of Weapon Casings The MARS fragmentation algorithm for solids inserts discrete cracks within a hexahedral mesh treating failure at the structural level rather than in the material constitutive equations The initial mesh is subdivided into clusters of elements Cracks may form between clusters when a local measure of damage exceeds a local allowable Cracks can propagate and or coalesce forming fragments of various shapes and sizes The parameters of the algorithm have been calibrated for specific weapons so that generated fragment distributions match arena test data data 1 3 Nano Scale Modeling of C S H The discrete element capabilities of MARS were used to simulate the behavior of Calcium Silicate Hydrate C S H through nano scale models of C S H specimens subjected to nano indentation testing The simulation results from these models are providing new knowledge in the nanomechanical behavior of C S H and are helping in formulating con stitutive laws for higher scale simulations simulations 1 4 Parallelization The MARS software can solve extremely large analytical models which require extensive use of computer resources The large demand on computer memory and CPU time can only be satisfied by using distributed memory massivel
98. ases as a result of solid fragmentation this list will only employ the initial faces LoadCurveList PressureHistories J QuadFaceList ListName MultiplePressureHistories 1 Specify face list Req either a face list or a shell list FaceList listName ShellList listName 2 Specify PressureHistory List Req LoadCurveList PressureHistories 3 Optional AutomaticTimeOffset 1 TimeShift 345 ms 2 1 The AutomaticTime0ffset command is designed to offset the time histories so that the pressure loads start at time 0 This is done by computing the first time pressures are 93 different than 0 in any of the curves All pressure histories are moved forward in time by that amount 2 The TimeShift command moves all pressure histories in time by the specified amount 9 3 Quad Shell Lists QuadShellList NAME if this list is used to define a rigid body enter Density 7 8 g cm3 else select one of the cross sections SteelShell m MATE i 3 reference system for local element axis alignment ReferenceSystem Local default 1st direction diagonal nodes 1 3 select one of the formulation below QPH default Tsay Belytshcko Tsay formulation if this list is not generated internally enter NodeList NODS ReadObjects 345 i nt n2 n n4 n5 n6 n7 n8 1 124 243 56 165 234 312 23 126 2 56 238 121 78 56 98 126 66 else generate see below for generation optio
99. ases the forces generated are not sufficient to enforce the constraints properly Using stiff penalty parameters improves the effectiveness of the constraint but increasing the stiffness may eventually introduce local high frequency modes that require small time steps for stability When using penalty formulations the user should make sure that results are not dependent on the choice of the stiffness constants This can be done by executing two or more simulations doubling the stiffness each time If results do not change appreciably between executions then the stiffness is sufficient to enforce the constraints properly The penalty forces are computed in the polymorphic method calcFrc The master slave formulations are very rigorous in enforcing constraints However there are many pitfalls in their usage The main problems occur when multiple hard conditions are applied to the same entities These include rigid body lists prescribed velocity lists boundary conditions and multiple master slave constraints When multiple hard conditions are applied to the same nodes results can be unpredictable In most cases Mars will write a warning message when multiple conditions are applied to the same node If done properly the results can be correct It is important for the user to fully understand how constraints work and how they may interact with each other to avoid erroneous results Master slave constraints are enforced in two stages a stage
100. ater linked to make a Quicktime movie file 25 1 2 Quasar File Format The ASCII format for QUASAR input files consists of a set of optional information lines and a set of geometric entity lists Currently QUASAR is able to render five types of geometric entities 1 Spheres 2 Lines 3 Cylinders 218 4 Triangular faces 5 Quadrilateral faces A sample input file is given at the bottom of this page Spherical Entities QUASAR renders spherical entities in the model using multi faceted solids Three dif ferent resolutions are available The lower resolution which is the default is suitable for very large models Medium and high resolutions generate smoother spheres but require more system resources and may take several seconds for each update The resolution can be set in the input file with an optional command line or changed interactively for each list during execution via pop up menu The spheres are painted in light gray unless a specific color is selected using the color command line Sphere data is entered one line per sphere The first field is the sphere index followed by the three coordinates and the radius The sphere index does not have to be in sequence Thus if a subset of a longer list is used the original indeces can be used If the user wants to scale the spheres after the file has been created this can be accomplished using the optional scale command line npL LABL color clr resolution high medium
101. axis The two nodes are supposed to overlap at time 0 For pratical purposes a tolerance is specified via input and the distance between the two nodes must be less than the tolerance The penalty forces are computed incrementally using velocities The hinge direction rotates with the average rotation rate of the two nodes ConstraintList ListName 143 HingePenaltyConstraint 1 select first node Node ListName nodeIndex Ef or Node ListName cl 5 cm O cm 8 cm 2 select second node Node ListName nodeIndex or Node ListName cl 5 cm O cm 8 cm 3 Enter penalty spring stiffness Stiffness 1 4e8 dyn cm 4 Enter tolerance Tolerance 1 mm 1 5 Enter hinge axis direction HingeAxis 0 1 0 3 0 6 Enter axial motion flag optional def false AxialMotion true This constraint make sense for nodes that have rotations 15 7 Particle Rebar Interaction List The particle rebar interaction list is intended to couple rebar embedded in concrete LDPM regions The input consists of the list of nodes which are used for defining the LDPM region and the list of rebar elements The concrete rebar interaction model is entered as a Material Note that The rebar radius must be explicitely entered The constraint detection phase identifies pairs of particles and rebar finite elements that interact with each other This operation is performed once at the beginning of a sim ulat
102. be executed This would be a very desirable capability since it would eliminate the need to download history files from a supercomputer center to you local computer Unlike 3 D graphical applications that can be executed but are too slow to be practical jHist could be run very effectively from a remote server If anybody can figure out how to do this please let s us know and we will make this feature available to everybody else 25 2 3 File format Time history files have the following structure 1 title line 2 number of records n 3 1 label for first record 3 2 label for second record 3 n label for nth record 4 0 v_0 1 v_0 2 v_0 n values at time v_0 1 typically 0 4 1 v_1 1 v_1 2 v_1 n values at time v_1 1 4 m vim i v_m 2 vim n values at time v_m i The labels in section 3 of the input are used in the drop down menus for selecting the records to be displayed 226 25 3 Curv a java program for making plots from various source files jCurv is a simple utility for creating plots of multiple curves from different source files jCurv is also written in Java It is a batch program with the plot formatting commands entered in an ASCII file jCurv generates plot images that can be saved to png files for incorporation in documents Its execution is very simple At the command line enter jCurv filename where filename is the name of the input file without the required extension jcr For example i
103. because of all the check it has to perform Furthermore storing a large number of edges and nodes which are not used in the calculation wastes a significant amount of memory Since most of the FRHSC specimens so far encountered have a parallelepipedal shape and the edge list does not have to be stored the new modified procedure can solve fiber concrete interaction more effectively for prismatic geometries The new procedure consists of e Fibers are temporarily generated during the initialization phase for the sole purpose of computing fiber facet intersections e The geometry is assumed to be a parallelepipedal prism aligned with the axis this simplifies the computations of the edge locations and intersections with external surfaces The logic is MPI parallelized while all the processes generate exactly the same fibers based on the seed number each process will consider only the fibers overlapping its domain Output for the new procedure appears in the initialization phase Initializing concrete fiber interaction Number of fibers 53760 Number of edge segments 191612 Total fiber length 1 38e 3 m Fiber volume 327 69e 6 m2 Number of fiber facet intersections 526449 122 12 Hexahedral Solid Elements The HexSolidList can be used to define a set of hexahedral 8 node solid elements These elements can represent e purely geometric entities e deformable finite elements or e aset of disconnected rigid bricks The type of element i
104. can provide the easier it is for us to determine the problem and fix it or correct your input file 22 Misc 22 1 Fluid Dynamic List It is necessary to insert a fluid dynamic list when performing coupled calculations with fluid dynamic codes The purpose of this list is to specify the entities that are used in the data exchange protocol Only one fluid dynamic list can be specified and it requires no name The contents of the list vary depending on which code MARS is coupled to The FluidDynamicList is typically entered at the bottom of the input file possibly before the EOF End of File line The general syntax is FluidDynamicList 4 Following are a list of codes to which MARS has been coupled to and the commands to specify the exchange data 22 1 1 IFEM RPI The exchange protocol between MARS and IFEM establishes the following exchange of data During the initialization phase MARS sends 1 the number of mesh nodes participating in the coupling 2 their initial mass and 3 their initial position During the solution phase MARS receives 1 integration time step which is controlled by IFEM 2 velocities of the nodes and sends back 1 structural forces internal and external at the nodes Even if the selection defaults are chosen all structural nodes and no facets the FluidDynamicList must be present in the input file It this case it is entered with no internal data like this FluidDynamicList A set of trian
105. ces form an angle greater than 60 degrees This list is useful for graphics when we want to represent the outline of a solid component 100 10 5 Time History Commands The following line commands are intended to be used inside Time History lists to produce records of global and element variables TimeHistoryList HIST histories for entire list ttL PART Volume ttL PART InternalWork ttL PART DissipatedEnergy histories for single element tt PART 1 Volume element 1 only tt PART 1 StateVariable 3 For internal work we intend the work done by element internal forces which results in recoverable elastic energy and dissipated energy 10 6 Plot Attribute Commands Plotting attributes can be specified in the element list or in the plot list TetSolidList PART PlotAttributes ContourVariable sv 1 NoNodalAveraging NoSmoothing PlotList PLOT ttL PRT1 ContourVariable sv 1 NoNodalAveraging SelectedElementsOnly NoSmoothing ttL PRT2 OutlineOnly 10 7 Viscous Tets The purpose of this list is to provide some form of artificial internal damping to kill internal vibrations It is different than dynamic relaxation in the fact that a vibrating 101 moving body will stop vibrating but its average velocity is maintained This list is used in conjunction with another list consisting of deformable elements This list does not own its objects tets but uses the objects of the tet list it is connected to
106. cillations and works similar to the Beta in the friction force algorithm No damping for Beta 1 progressively more damping for larger beta s default is beta 2 The MinNormalForce is an optional parameter intended to approximate the effects of bonding forces which oppose diffferences in rotation rates even when the particles are 153 not pressed against each other The minimum normal force f0 is added to the normal force resulting in a larger value for the maximum allowable rolling moment Xr fr L Fn fO The parameter dt can be used for computing the penalty spring stiffnesss as an alterative to the RollingStiffness and T wistStiffness parameters When dt is selected the value of the penalty stiffness for each contact is computed using the equation K 0 5 rm dtxdt where rm is the minimum rotational mass moment of inertia of any node particle participating in the contact In the current version the stiffness parameters are entered with no units which makes this portion of the input unit dependent This will be corrected in future versions The MARS rolling resistance algorithm works in conjunction with the friction force algorithm to reproduce the resistive forces implied in the conventional definition of rolling resistance For example the rolling resistance Fr which is a force in the conventional definition of an ordinary car tire on sand is approximately 0 3 time its wei
107. cle size by 0 5 MinRadius 0 5 in 3 Select contour variable optional var vx plot fringe plots of x velocity vmm 0 bottom value for fringe plots 1 vmx 100 top value for fringe plots 1 4 Change displacement scale optional DisplacementScalingFactor 1000 lt dsf 1000 gt 1 The minimum and maximum values for the range are not required For each plot MARS computes the minimum and maximum of the quantity to be plotted Either one or both are overwritten if the min and or max are specified in the input By fixing the range it is possible to make animation movies where the colors are consistent across frames Plotting command for Paraview files are simpler First recall that it is preferable to have one list per plot file Second all other parameters except DisplacementScalingFactor are controlled within Paraview PlotList PlotListName Paraview TimeInterval 0 1 ms ndL NodeListName Velocity and rotation vectors of selected nodes particles are automatically saved in Par aview files The Paraview procedure for displaying particles and velocity or rotation rate vectors as arrows is described below Main Menu File gt Open Select files Particles 00 Pipeline Browser Click on Particles 00 Object Inspector Properties Apply Display particles as gray spheres Main Menu Filters gt Recent gt Gliph Object Inspector In Properties Glyph Type Sphere 54 Scale Mode
108. ct FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList TopSurface Select FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList BottomSurface AlsoSelect FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList TopBottomSurface InvertSelection Make TriangFaceList CylindricalSurface In this cases the faces of the top surface are selected by using the criterion that the cross product of their normal with vector 0 0 1 is greater than 0 5 By using appropriately 83 the node selection and face selection commands it should be possible to extract a list of faces that make up the desired surface The sublist can be used to applied pressure loads as in the example below TriangFaceList CylinderFaces Select FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList TopSurface Select FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList BottomSurface AlsoSelect FacesWithCrossProduct 0 0 1 gt 0 5 Make TriangFaceList TopBottomSurface InvertSelection Make TriangFaceList CylindricalSurface It is a good idea to first run the problem in interactive mode and at the interactive prompt select the lists and plot them to verify that the commands produced the desired surfaces 8 2 Wet Triang Face List These lists are used for applying pressure loadings on a set of triangular faces Several loading types are available 1 Uniform pressure loading 2 Variable pressure loading 3 Special loads 4 Patched press
109. d then all six degrees of freedom are tied If some DoF s need to be released then an initial local reference system need to be defined NodePairList ListName MasterSlaveConstraints ReadObjects 2 1st node is master 2nd node is slave like in master slave NodeListName 34 NodeListName 43 L RefSysName T XXX R OXX gt NodeListName 14 NodeListName 76 X 0 7 0 7 0 Y O 1 O R XOX For example to model a hinge around the local X axis it is necessary to specify a local reference system which includes the X axis as well as a Y axis which can be any arbitrary direction perpendicular to the X axis and the R OXX labels A spherical bushing can be specified by entering the label R 000 only 7 6 Node Pair Attraction List NodePairList PosNegPairs Attraction NodeList PosNodes NodeList NegNodes 74 DetectionDistance 0 6 cm NodeiThickness 0 4 cm Node2Thickness 0 2 cm AttractionConstant 45 7 7 Node Pair Repulsion List NodePairList PosNegPairs Repulsion NodeList PosNodes DetectionDistance 0 6 cm NodeiThickness 0 4 cm RepulsionConstant 45 J 7 8 Node Pair VanDerWaals List NodePairList AttractionForces VanDerWaals NodeList Particles DetectionDistance 4 cm NodeThickness 0 4 cm HamakerConstant 45 ThreshholdGap 0 4 nm 7 9 Node Pair NanoParticle List NodePairList ListName NanoParticles NodeList Particles DetectionDistance 4 cm NodeThickness 0 4 cm UpdateInterval 1
110. d a list of beam elements For the purpose of contact a node particles takes a spherical shape and a beam takes a three dimensional smooth shape already described in the edge edge contact section This type of contact can be used for some specific applications such as the interaction of loose particles with rebars The input commands for face contact list take the format RebarContactList ContactListName variables with the sign inherit defaults from control parameters 1 Specify lists NodeList NodeListName BeamList BeamListName 2 Control paricle and beam shapes NodeThickness 0 4 cm OBSOLETE NodeRadiusNoSmallerThan 0 2 cm NodeRadiusNoLargerThan 0 4 cm NodeRadiusExactly 0 3 cm BeamThickness 0 2 cm OBSOLETE BeamRadiusNoSmallerThan 0 2 cm BeamRadiusNoLargerThan 0 4 cm BeamRadiusExactly 0 3 cm NoHemiSphericalCaps 3 Set detection and update parameters DetectionDistance 0 6 cm UpdateInterval 0 000010 s 4 Enter contact models see previous sections ContactForce PenaltyHyst FrictionForce 6 RollingResistance Debug 6 The radii of the nodes are controlled by three optional input commands NodeRa diusNoSmallerThan NodeRadiusNoLargerThan and NodeRadiusExactly The first command sets a lower bound fmin for the node radii the second command sets un upper bound rmax and the third command sets both lower and upper bounds to the same value essenti
111. d from the list ttL Block FacetVariable 15 crack opening variable MinimumCrackOpening 0 01 mm do not use next four options for paraview plots RangeMinValue 0 01 mm RangeMaxValue 0 10 mm DisplayBelowMin 117 DisplayAboveMax Scale 1000 convert from m to mm The last three options should be omitted for Paraview plots since the same functionality is obtained in Paraview in the ObjectInspector gt Display gt EditColorMap form and using the threshold filter 11 6 7 Domain decomposition plots The following example shows the commands for generating a Paraview file which contains the exploded view of a tet list depicting the domain decomposition The parameter 1 3 controls the amount of radial motion for the domains The coordinates of the center of gravity of each domain are multiplied by this parameter A value of 1 means the com ponents stay where they are The larger the value of this parameter the more exploded the view looks Results are in file DomainDecomposition 000 vtu The domains may be painted using the scalar variable Domains PlotList DomainDecomposition Paraview Timelnterval 100 s ttL Tile DomainDecomposition 1 3 11 6 8 Summary of plotting options A complete list of available options is shown below PlotList Name 4 ttL Block ExternalCellFacets ttL Block Cells MinimumCrackOpening 0 01 mm ttL Block Cell0utline ttL Block Particles ttL Block FacetVariable
112. d to be modified for the actual simulation Bullet dn3d mat 1 3 ro 0 00025 e 10e6 pr 0 3 endmat start 13 55 123 5 7 9 3 17 9 12 184 0 5 6 6 di 12 12045 di 12 34 di 12045 34 sfi 1 5 12 cy 0 0 0 0 0 1 2 sfi i 12 y cy 0 0 0 0 0 1 2 sfi 1 23 sp0 0 5 2 Sfi b 32 3S y O AA O Do2 sfi 4 sp 0 0 5 2 b 001 001 001000 b 300 300 100000 b 004 004 111000 b 110 150 111000 b 110 310 111000 b 150 350 111000 end end t 0 1 cont 21 6 add and remove list during a simulation During the course of a simulation it may be required to add or remove lists For example we may need to prestress a structure and then add a dynamic load when the structure has reached a state of quasi equilibrium under the static loads Or we may want to remove a component that is no longer necessary in the simulation and continue the calculations for the component of interest This is possible using this approach We can use the Read command inside the ControlParameter section to schedule reading of an input file at a certain time see section in this manual Inside this file we can insert the commands ControlParameters Read removeBullet mrs atTime 0 2 ms The listing of file removeBullet mrs may look like this File removeBullet mrs Lists Remove HexSolidList Bullet Remove NodeList Bullet Remove NodeNodeContactList BulletTile EOF 185 The list type e g HexSolidList employ t
113. data Open plot files corresponding to edge list Press Apply button nothing shows up Select Filters Common Glyph from main menu Set GlyphType to Cylinder Set Resolution value to 30 or higher Set Radius value to desired value Leave Capping checked Set the Glyph Transform Rotate value to 0 0 90 Set Scale Mode to scalar Check Edit and set Set Scale Factor to 1 or slightly higher Press Apply button the cylinders should show up In the Display tab change Color by to Solid Color oono ARUN eere Ne O Step 8 is necessary to have the cylinder aligned with the direction of the edge This may be corrected in newer versions of Paraview and this step may no longer be necessary Steps 9 and 10 are intended to scale the height of the cylinder to the length of the edges If the edges form a curved line it is necessary to use a value greater than 1 to avoid gaps on the convex side of the line 7 2 Beam Lists The BeamList is used to define a set of beam elements with constant cross section The formulation is similar to the QPH shell element in the sense that the centerline of a beam element remains straight Shear deformations are controlled by the rotation rates of the two nodes that define a beam The beam formulation has a modular interface with the CrossSection object This makes it possible to define many
114. deRadius 0 5 cm if face list comes from shells face thickness is the larger of thf and half shell thickness 7 Do not initialize overlap ActualContactDistance 4 8 Define detection distance DetectionDistance 0 6 cm 9 Define update interval UpdateInterval 0 000010 s MinimumDistance 0 1 cm 10 Enter contact models see previous sections ContactForce PenaltyHyst FrictionForce 2 RollingResistance UpdateInterval 0 000010 s Debug 6 2 The radii of the nodes are controlled by three optional input commands NodeRadiusNoSmallerThan NodeRadiusNoLargerThan and NodeRadiusExactly The first command sets a lower 159 bound fmin for the node radii the second command sets un upper bound max and the third command sets both lower and upper bound essentially forcing all nodes to have that value These bounds are usedin the calculation to compute the effective contact radius rc for each particle using this logic max rad r_min min rc r_max rc rc 4 The ActualContactDistance command is used in rare occasions when the initial configuration is pre stressed and initial contact forces are desired In this case the gO gap offset parameter is never set 16 4 1 Time History Commands The following line commands are intended to be used inside Time History lists to produce records of global list variables TimeHistoryList HIST tcL ListName NumberOfContac
115. deck provides some flexibility The ControlParameter section must always be placed at the beginning of the input because it contains units selections The definition of new 11 entities must be done before they are referenced in other parts of the input file In this manual we use the term entity for individual objects such as material models reference systems load curves etc and lists of objects such as finite element lists contact lists constraint lists etc Each entity is given a descriptive short name which is used as identifier in the rest of the input file and during post processing Similar entities must use unique names However the program does not preclude using the same name for dissimilar entities For example the list of external faces of a list of solid elements solid component can be given the same name as the name of the solid list This is because a face list and a solid list are dissimilar A typical input file looks like this First line is always a title line ControlParameters Material Steel Elastic NodeList PartNodes Finite element definition HexSolidList PartElements FBSingleIP 4 Material Steel Nodelist PartNodes Applied loads LoadCurve NodalLoad 4 NodalLoadList Loads 4 Nodelist PartNodes LoadCurve NodalLoad Post processing TimeHistoryList History PlotList Plot EOF 12 3 1 Major Syntax Rules The current MARS format adopts a syntax
116. dles the basic 4 node tetrahedral geo metric element By doing that the new class inherits all the methods from the parent class including mesh generation methods post processing methods etc Only a few polymorphic methods need to be written for the new class The definition of the new list class is given below ass ttL_Cosserat10 public ttLst 4 public ttL_Cosserati0 ttLst O 7 ttL_Cosserat10 string nm ttLst nm initList virtual ttL_Cosserati0 void initListQ ttLst initListQ string getListSubType return Cosserat10 Object createNewObject return Object mew Tet10 Tet createNewTet return Tet new Tet10 bool processCommand Reader string void readLst Reader polymorphic void initialize polymorphic real calcFrc polymorphic void insertMidEdgeNodes list specific void examine polymorphic Fo There are a few significant methods that need to be written The first method void readLst Reader is a polymorphic method invoked during the reading phase This method reads the input file and interprets commands specific to the new list either for entering parameters or for performing tasks void ttL_Cosserat10 readLst Reader rdr rdi rdr while true rdr gt readLine string lbl rdr gt getShortLabel if 1b1 break else if obLst processCommand rdr 1b1 if rd2 NULL rdr rdi else rdr rd2
117. e and Post Processing 164 18 1 Plot ets e ss es ee ee in A a e a e 164 owe E Soe eee 165 Lee eee we eee we nee eee ee ee Se 165 19 MPI Parallel Processing 167 E E A a eta a E 169 Ae aos a as a a 169 T9 T2 OCtreel a das Set a A e e 170 La dd e dd dd a de ee E EN A a Sea e ee da ad a Se a ii a ee ee A O A ARES eea e aa a ee eee a ee ee 19 5 1 Brazilian Test we bon bg DG we BS Eee Boe Soe eS SOs 19 5 2 Contact between projectile and concrete slab 2 19 5 3 Contact between particles and nanoindenter 20 Generation of Complex Parts 21 How to Perform Specific Tasks 21 1 monitor energy 6 2 68 be ded OR ae Y ee SE Oe eo Pe ee oS 21 2 save plot data for later post processing 04 21 3 write and read restart files pe ee eee Pee we EN ae Ee 21 4 pre process input files 0 o 21 5 import Ingrid meshes 2 os ee a a ae ee 21 6 add and remove list during asimulation 21 7 create custom versions of mars o aoa e 21 7 1 User defined list 0020020000000 0004 21 7 2 User defined material 0 0 00 0 00000 eae 21 7 3 User defined mesh generator 21 7 4 User defined load curve o eee 21 7 6 Restart Procedures e ft a oy ae we es ee ee ee oe ae eee ee ee eee ee a ee eee ae 22 Misc 22 1 Flia Dynamic List ee c s s 64 044844840642 424 AA 221 1 INEM RPI o asia Sek BE
118. e selected using the commands SelectNodes and UnselectNodes Examples are given in the selection section of the node lists 8 1 2 Generate commands Several simple triangular meshes can be internally generated in MARS Following is a list of surfaces that can be generated Multiple surfaces can be generated withing the same Generate block and merged together with the MergeNodes command Generate Cylinder ReferenceSystem Local EdgesOnCircle 36 must be multiple of 6 Radius 4 in 80 Length 10 in ElementSize 1 in in axial direction Disk ReferenceSystem Local optional Radius 4 in OutsideEdges 36 must be multiple of 6 Annulus ReferenceSystem Local optional InnerRadius 4 in OuterRadius 6 in InsideEdges 24 must be multiple of 6 OutsideEdges 36 must be multiple of 6 Sphere ReferenceSystem LOCL Radius 10 in Refinement 3 Single Rectangle 110 16 index progression like in INGRID 5 5 5 5 Box Dimensions 2m2m5m Elements 4 4 10 GenericPlate 1 u in length units for this part R RSYS reference system optional L 0 15 typical length of an element i invert quad orientation if necessary P define polygonal external outline n 0 0 0 0 first node of polygon n 1 00 0 second node of polygon c close the polygon enter internal circular holes x y R s 5 min no of sides in holes optional C 0 5 0 5 0 2 fs 2 81
119. e are three ways for computing the penalty stiffness 1 constant stiffness for each constraint 2 stiffness computed based on time step and minimum node mass 3 stiffness computed using young s modulus of tet list material If the stiffness is based on a time step the following equation is used Mmin KZ At where m min is the minimum mass of the five nodes participating in the constraint and Dt is the time interval specified via input The smallest the time step the larger the penalty stiffness resulting in smaller displacements The idea is to select a time step close but slightly larger than the stable time step for the simulation If the penalty stiffnesss is based on the Young s modulis E of the material for the tetrahedral mesh the penalty stiffness is computed using the equation K EV P where V is the volume of the element that contains point P corresponding to node N The input format for the penalty formulation is NodeTetConstraintList listName Penalty NodeList nodeListName TetList tetListName choose one of the three options below ConstantStiffness 100 lbs in StiffnessBasedOnTimeStep 0 001 ms StiffnessBasedOnYoungsModulus If the master slave formulation is used the mass of slave nodes N is distributed to the nodes of the master tet element containing node N using the weighting factors computed from the shape functions The input format for a master slave formulation is NodeTetConstraint
120. e by appending th The input commands to the TimeHistoryList consist of the output time interval and a set of variables one variable per line The format for each variable is standardized and obeys the following convention convention 165 v de Description S scl O ofs D where ob is the two character tag for a list this is found in the examples at the list level in the pages for the different lists NAME is the name of the list The first line of the three lines above refers to variables for the whole model such as total kinetic energy a list of available variables is given below If the variable involves the whole list use the obL format if the variable involves an object of the list use the ob format and select the object o using one of the following methods 1 enter the index of the object e g nd NODS 45 2 find object closer to a point in space e g nd NODS cl 5 7 3 In the commands above v is used to select the variable within the list or object Optional entries are the description which will be displayed in plots and used for variable selection the scale scl and offset ofs Scale and offset are used to modify the output value according to the following expression Voutput scl Vcalc ofs The D character automatically sets the offset to the opposite of the initial value of the variable This is useful when plotting displacements in a Cartesian direction since the displacement is equal to the d
121. e deformations Default is linear 5 7 Accelerometer The Accelerometer object is designed to simulate the behavior of actual accelerometers for direct comparisons with acceleration records measured in test setups The acceleration computed by the Accelerometer object employs a corotational formulation suitable for large rotations As such it may be different than the acceleration computed at the nodes in the global fixed reference system While physical accelerometers are only placed on the surface of a part these numerical accelerometers can be also placed in the interior of a part The numerical acceleration is defined using its location length and direction The main purpose of the numerical accelerometer is to produce a time history record of the local acceleration which can be used for comparison with test data or just for better understanding of the results from the simulation Accelerations are computed as follow eS u t u t At e At where n is the corotational direction aligned with the accelerometer There are two types of numerical accelerometer 1 Those attached to a finite element component 2 Those embedded in a discrete element model cloud of points Typical input is shown below Accelerometer GageName type part on which the strain gage is attached to type dependent parameters More detailed input on how to place an accelerometer on various types of meshes are discussed in the subsections bel
122. e distribution is defined in the range of aggregate size between l and u where Y is the minimum aggregate size being considered For maximum particle density c is approximately 0 5 according to Fuller and Thompson In the example above if we consider an aggregate size of 0 8 cm then P 0 8 1 2 81 of aggregates are smaller than 0 8 cm For details see this web site http training ce washington edu wsdot modules 03_materials 03 2_body htm 5 5 Colors Colors are used in the definition of surfaces and lines There are two methods for speci fying colors The first by selecting a color from the built in list using the command Color Red Available colors are Red Green Blue Darkgray Lightgray Black Cyan Steel Rust White Yellow Orange Gold Patina Clear The other by specifying the rgb composition using the command Color rgb 0 4 0 6 0 2 where the three numbers following the rgb keyword indicate the brightness of each color red green blue in this order on a scale from 0 to 1 5 6 Strain Gages The StrainGage object is designed to simulate the behavior of actual strain gages As such it can provide a more realistic measurement of local deformations for comparison with test data than the various strain metrics from continuum mechanics The strain gage object is defined by selecting two or more material points inside or on the surface of a part The points are placed in correspondence of where the physic
123. e old_value offset Line 12 The EOF line terminates the input phase any information after the EOF line is ignored This is the sequence in which the various plot elements are displayed first the solid bands if any are painted in light gray Then the x and y axis and grid are painted in various shades of gray or black Then the curves or symbols are painted and finally the vertical bands 25 3 2 Data set file formats jCurv accepts the file format of the time history files generated by Mars from the TimeHistoryList s Furthermore it accepts two other file formats Data which may 228 be available in ASCII format can be edited to satisfy one of the three formats using an ASCII editor If you have a specific format which you would like to be incorporated please make a request to ES3 All files formats are included in the examples Format of xy files x_1 y_1 2 y_2 x_n y_n Format of seq files Title line string Title line string nRec int number of Records Record 1 label string npt int number of value for record 1 value 1 1 first value of record 1 value 1 2 second value of record 1 value 1 npt npt th value of record 1 Record 2 label string npt int number of value for record 2 value 2 1 first value of record 2 value 2 2 second value of record 2 value 2 npt npt th value of record 2 Record nRec label string npt int number of value for reco
124. ed as an extension of complex numbers Quaternions are used to describe rotations in space employing only four scalars A node list can be forced to use the NodeQ class rather than the default Node class by entering the keyword ComputeRotations in a stand alone line NodeList Particles Couput Notation i At this time node rotations can only be viewed using Paraview It is sufficient to enter a PlotList section in the input file using the following commands PlotList Particles TimeInterval 0 2 ns Paraview ndL Particles 56 Node rotations are automatically saved to the output plot file along with coordinates velocities and rotation rates The new NodeQ class was tested on a simple benchmark problem consisting of a simple supported beam loaded with a uniform load perpendicular to the beam The beam was modeled using a strip of quadrilateral shell elements Vibrations were dampened until the steady state solution was obtained The final rotation of the beam are visualized in Paraview using the steps described below Load particle file Apply Main Menu Filters gt Recent gt Gliph Set Gliph parameters in the Properties tab Scalars Rotations Vectors RotatiionVectors GlyphType Arrow Set Scale Factor set to appropriate value by trial and error X Edit Change color convention in the Display tab Edit Color Map gt Choose Preset Choose Red Yellow Green Blue bar If you want to show the particle
125. ements 8 1 Triangular Face Lists The commands for The TriangFaceList is used to define a set of triangular faces These faces can represent e purely geometric entities 78 faces subjected to pressure e membrane elements shell elements e aset of disconnected rigid triangles The type of face is specified in the first line after the list name The most basic list is the geometric type where objects consists of triangular faces defined by three nodes The commands below refer to geometric lists but can also be used in the other lists Other lists have additional commands which are described in later sections Note that the triangular face lists created by other lists e g the external faces of a tetrahedral mesh are of the geometric type TriangFaceList ListName type 4 where the keyword type can be one of the following Geometry this can be used to define plain faces Rigid this is used to define rigid triangular platelets UniformPressure surfaces subjected to pressure loads MultiplePressureHistories DktShel1 1 Specify thickness Optional Thickness 0 125 in default 0 2 Specify density for rigid body calcs Optional Density 7 8 g cm3 if thickness 0 then mass is also 0 3 Specify front and or back color Optional FrontFaceColor gray default lightgray BackFaceColor red default lightgray 4 Specify node list NodeList NODS ReadObjects 254 number
126. ensure that the cts file corresponds to the current model first it checks that the number of elements is the same then it computes the stable time step for the first element and compares it with the value in the file If either test fails Mars automatically deletes the cts file and exits with an error message prompting the user to resubmit the job If MassScaling is not used the minimum time step for the Ldpm list is used in the computation of the global integration time step 11 2 Model Generation Because of the unique requirements of LDPM models a series of special purpose of mesh generation methods are made available 110 TetSolidList listName LDPM Material materialName Generate choose one of the options LdpmBox y LdpmCylinder LdpmSphere 11 2 1 Box The LdpmBox command is used for generating an LDPM concrete parallelepipedal box This method first generates points along the edges of the box then points on the faces of the box It then randomly places a set of particles inside the box It uses tetgen to generate a tetrahedral mesh The method selects an appropriate mesh density for the exernal mesh based on expected average distance between internal particles LdpmBox Dimensions 4 in 2 in 8 in dimension in x y and z directions 1 Seed 23423 seed for random number generator ConstantSpacedParticlesOnEdges 1 The box spans from 0 to 4 in in the x direc
127. enter Particles are divided using recursive bisection method Nanointenter is handled by all nodes Contact between particles and nanoindenter are handled by RB Contact between particles are handled by RB MpiDomainList 4 SingleNode DD1 Node O Bisect DD2 FirstNode O List ndL Particles Verbose AllNodes DD3 Default DDi npL PrtInteraction DD2 U nd Particles DD2 U ndL NanInd DD3 C tcL PrtIndContacts DD2 U Update Timelnterval 0 1 ms dd DD2 nd Particles nc tc 20 Generation of Complex Parts 21 How to Perform Specific Tasks This is a collection of instructions that describe procedures for performing various taks 21 1 monitor energy In most dynamic simulation it is useful to monitor the transfer of energy across the system Most important it is essential to ensure that there is no artificial energy entering 174 into the system due to numerical instabilites or just errors in the algorithms The best way to accomplish this control in MARS is to generate time histories of work and energy for global and list variables At the global level four quantities are available TimeHistoryList ListName 4 ExternalWork InternalWork KineticEnergy EnergyBalance ExternalWork We is the work performed by external forces or pressures or reactions at nodes where the motion is prescribed Positive external work means that the nodes move in the directions of the forces and energy is entering into the
128. ents the simplest forms of constitutive equations in three main formats tensorial plane stress for shells and vectorials for beams Material ELST elastic Description Stainless Steel Density 7 8 g cm3 YoungsModulus 29e6 psi PoissonsRatio 0 3 State variables for tensorial equations XX Stress stress YY Stress stress ZZ Stress stress YZ Stress stress ZX Stress stress XY Stress stress Aor WNEFe State variables for plane stress equations XX Stress stress YY Stress stress YZ Stress stress ZX Stress stress XY Stress stress oP WN Fe 36 State variables for vectorial equations 1 Normal stress stress 2 Shear stress X stress 3 Shear stress Y stress 5 9 2 Elastio plastic Material Model Material PLST plastic Description Stainless Steel Density 7 8 g cm3 YoungsModulus 29e6 psi PoissonsRatio 0 3 YieldStress 80e3 psi HardeningModulus 200e3 psi select one of the three hardening options below IsotropicHardening KinematicHardening Beta 0 5 beta 0 istropic beta 1 kinematic hmx 8e3 psi Bounding stress for friction damping Friction damping provides hysteretical energy loss for each cycle related to cycle ampli tude and not strain rate Because it is non viscous the term friction damping is used Use a bounding stress in the order of one percent of the yield stress if damping is desired State variables for tensorial equa
129. er available on different platforms to generate the actual input file to be used as input to MARS While the sites above provide a complete overview of the preprocessor capabilities the examples below give a more tangible feel for how these capabilites can be used in the context of writing MARS input files files The first example shows how to manage two executions with a single input file The first execution performs the actual simulation and saves plot snapshots at regular time interval of DT milliseconds The second execution post processes the plot data file gen erated in the first execution and generates a sequence of Paraview plot files In addition we want to be able to change the time interval at which plot snapshots and Paraview plot files are generated The composite input file input mrs looks like this this Title line ControlParameters WritePlotDataFile data plt Every DT ms ifdef PLT PlotList Cells Paraview TimeInterval DT s ttL Tile Cells ProcessPlotData data plt endif EOF where the preprocessing directives are the lines which begin with a character The C preprocessor commands for generating the simulation input file inputS mrs are are t CC E DDT 0 001 input mrs j inputS mrs where CC is the C C compiler name which varies on different computer platforms CC works on jade g work on computers that have the gnu c compiler installed etc The resulting file inputS mrs will look like
130. er supported 25 1 1 Quasar User Manual GLUT The version of QUASAR is written in C It employs OpenGL for 3 D rendering and GLUT for a rudimentary UI It requires the user to memorize the functionality of some of the keyboard keys but it is a lot faster than the Java version for rendering large models It had been tested under Linux and under Windows cygwin 216 Viewing plot files using QUASAR The plot files generated by Mars are viewable using the quasar post processor quasar is based on the OpenGL library for three dimensional rendering The execution command is quasar input file name Currently quasar does not have a fully developed GUI interface although it uses the pop up menus provided by the GLUT library for many of its tasks To make thinks efficient control of the model is done by a combination of keyboard hits and mouse motion Repositioning the model The model can be repositioned by rotating translating and zooming the model in front of the viewer The motion is achieved by moving the mouse In earlier versions of Quasar the functionality of the mouse motions was preselected by pressing one of these four keys 1 1 e and w t turn translation on r turn rotation around two axis on e turn zooming on motion of the mouse up and down w turn spinning in the third direction on motion of the mouse left and right When manipulating the model if you are right handed you would keep the four
131. erbose 1 Use either EquilibratedArea or TiledAreas EquilibratedAreas 2 TiledAreas 24 3 ProjectedNormalAreas DisableInternalFacets 4 145 1 The Verbose command should be used only for small problems and is intended for debugging purposes When Verbose is entered Mars will print a detailed status of all the particles interacting with each of the fiber elements elements 2 The EquilibratedArea command is used for computing the interaction areas using the criteria described in the notes 3 The TiledAreas command computes interaction areas by subdividing the cylin drical surface of the fiber rebar into tiles in the axial and cylindrical direction Each area is assigned to one of the particle edge contraints The integer following the keyword indicates the number of areas in the hoop direction The number of areas in the axial directions is computed by using the equation 4 The DisableInternalFacets is an optional command for disabling the LDPM facets whose centers lay inside the cylindrical volume of the fibers These facets will not contribute any internal force na L QTR ng where R is the fiber radius L is element length and ny is the number of circumfer ential areas specified above via input Warning If an element is partially embedded in concrete Mars distributes its whole surface areas to the particles interacting with that element On the to do list there is the action item f
132. ering provides a more realistic representation of the fracturing process for two reasons material does not disappear and cracks are clearly visible The idea 114 behind this rendering is to plot visible faces from all cells of a model Since it would be computationally very expensive to plot all internal facets MARS checks the value of the crack opening and if it exceeds an input value it assumes that a large enough crack has formed and both faces of the facets are plotted All external facets are always plotted Facets edges are also plotted when the internal facets are plotted Figure 4 shows a fractured brick The relevant input for this example is listed below Currently the parameter to control plotting of cracked surfaces is specified inside the TetSolidList and is defined by the keyword CrackOpening A cell plot of an LDPM model can be requested inside a PlotList with the command ttL Brick Cells Since July 2011 the definition of the crack opening parameter can be made inside the plot lists using the command MinimumCrackOpening Notice that the level of detail in crack displaying can be controlled by varying the value of the crack opening parameter Making that value too small may make the plot too busy Most plotting capabilities are available for both Quasar and Paraview Quasar will not work reliably and efficiently for very large models but it will display a final image very quickly when it can since it does not require a long set
133. es The following example shows the commands for generating facet contour plots of the inter cell gaps for half of the model while eliminating the small fragments The elimination of the small fragments is optional The external faces of half of the tile are also generated in a separate list The external faces are supposed to be displayed in gray or any neutral color where the facets are to be displayed in color TetSolidList Tile 4 EditNodeList 4 Select cx gt 0 in SaveSelectione HalfTile Select all E PlotList ExtrnCells 4 TimeInterval 0 01 ms Paraview ttL Tile ExtCellFaces MinimumFragmentMass 1 e 6 1b s2 in LoadNodeSelection HalfTile PlotList Cntr Paraview TimeInterval 0 01 ms 119 ttL Tile FacetVariable Total crack opening RangeMinValue 0 10 mm RangeMaxValue 0 3 mm DisplayAboveMax MinimumFragmentMass 1 e 6 1b s2 in LoadNodeSelection HalfTile Scale 1000 11 7 Embedded Fibers Fibers of various types are used in construction materials to enhance durability strength to weight ration ductility energy absorption capability etc In the context of LDPM fibers concrete interaction is explicitly modeled at the micro scale level by including the mechanical effect of fiber interaction at the facets where fiber facet intersection occurs The distribution of fiber facet intersection is accomplished by generating distributions of fibers similar to actual distributions found in real specimens From a geometric poin
134. es This is compiled using the Visual Studio 2008 C compiler First place the executable marsW in a folder of your choice and either put that folder in the PATH or make an alias alias mars MARSPATH marsWvs so that marsWvs is executed when you type mars at the prompt Try it typing mars h to start an interactive help session and enter Q to exit 24 Pre Processing and Mesh Generation Software Mars interfaces with two mesh generation packages for creating tetrahedral solid meshes tetgen and two dimensional triangular meshes Triangle Both programs are freely available to researchers on the Internet Both programs are copyrighted To avoid copy right infringment ES3 is not distributing source or binaries for the two programs but we are refering users to download software from their respective websites and install them following the directions provided by the authors Both programs are executed as external modules More precisely MARS writes input files for either program executes the programs as separate threads using the command function and reads the mesh from the output files created by the programs This process is transparent to the user However it is critical to have the two programs available in the PATH This can be accomplished in two ways First symbolic links can be created in a folder already in the PATH that point to the executables 215 ln s TETGEN FOLDER tetgen BIN FOLDER tetgen Second the folders where
135. ess the user selects a different filename MARS can read the ingrido file and convert its data to MARS format This can be done in two ways The first method is the most recent and probably the most convenient because all steps are documented in input files e generate an input file that reads ingrido and saves the mesh into a MARS mesh file for example Ingrid conversion input ControlParameters Units English NodeList NODS 4 Rename BLLT HexSolidList H000 Rename BLLT Write PartMeshDataFile bullet mrs e the mesh can be easily imported into a MARS input file using the commands HexSolidList BLLT Material STEL Read bullet mrs The second method is more interactive but requires repetitive work each time a change in the mesh is necessary run mars D ingrido At the interactive prompt you can examine the INGRID generated model as if it were a MARS model When you are ready to write out the model in MARS format return to the main level mdl prompt and type the W character Select the mars output and this will generate a file name mars out which contains the ingrido data converted to MARS format 183 The mars out file should be renamed to something more descriptive and should be edited so that the names of the lists are meaningful and unique Note INGRID should be used mostly as a mesh generator Some of the features in INGRID are not translated into MARS For example contact conditions specified in INGRID
136. etch 1 decay 0 8 insert penalty spring and reduce force 2 stiff 3 do not fail 3 bonds adjacent to failed bond 3 1 Failure stretch has no dimensional units Two beam elements are disconnected at their common node when the average stretch of the two elements exceeds the local failure stretch at the node 2 The decay parameter is used to reduce the cohesive force as the two elements are pulling apart from each other thus dissipating a certain amount of energy fracture energy 3 The stiff parameter is used to prevent a string of elements from fracturing into single element fragments Square solid section sec s m MATE r 10 mmi 5p 4 m material S Side can also use A 315 mm2 section area i of int points 6 2x2 truss points 2 shears 11 3x3 truss points 2 shear p of plotting pnts 0 line 4 square Circular solid section sec o m MATE r 10mmi5pl m material r radius can also use d 20 mm diameter A 315 mm2 section area 72 i of int points 1 truss 5 beam with 3 tensile IP p of plotting pnts 0 line 1 cylinder 6 hexagonal X section 12 12 sided X section Tubular section sec O m MaterialName r 10 mm t 1 miodpil m material r radius can also use d 20 mm diameter t thichness i of int points around circumference p of plotting pnts 0 line 1 cylinder 6 hexagonal X section 12 12 sided X section Z section sec O m MATE t 1 mm h 10 mm
137. f the constraint During execution the forces and moments in each constraint are added up The summed forces and moments are then assigned to the nodes of the constraint This garantees that the nodes in the constraint move with the same velocity enforcing the conditions of periodicity 22 5 Load Curve Lists The LoadCurveList is used to defines a set of pressure time histories at certain spatial locations DYNA3D format LoadCurveList BlastLoads 1 Enter location and name of data file Directory Loads Filename pressures dat Format dyna3D shamrc csv mars 2 Enter spatial distribution type RefenceSystem RS name Distribution Plane Axisymmetric 3 Enter units X Units time s Y Units pressure psi Z Units length in 1 4 Enter curve mapping ReadObjects num curves index crx cry 1 5 6 For axysymetric load ReadObjects num curves index rad 1 O 2 de 5 Modify curves if necessary Y Scale 5 207 1 The Z Units are the location units in the table SHAMRC files This feature was inserted in Mars in 2004 in support of a study for assessing the effects of closed in blast on suspension bridge cables The pressure histories at selected locations on the surface of the cable were computed using the hydrocode SHAMRC The pressure data was provided as a set of files one file per station In a separate table each file was associated to the spatial coordinates of the station A SH
138. f the input filename is compare jcr the command would be jCurv compare jCurv creates a 3 in by 5 in plot and two actions are available The Refresh button makes it possible to refresh the plot when one or more of the source files are continuosly updated during an ongoing simulation The Make png button makes it possible to save the image to file image png in the local folder If you need to generate mulitple files you have to change the name to avoid overwriting previous images Installation is similar to jHist You need to include an alias entry in the bashrc file similarly to what was done for jHist The jCurv classes are typicall installed in the MARSPATH jCurv folder Some examples are also included They are named exam plel jcr example2 jcr etc the data sets used in the examples are named dataset1 th dataset2 th etc 25 3 1 Input file format The format of the input file is discussed by explaining the commands contained in example1 jcr one of the examples included in the installation 1 Title Example 2 XAxis lb X Axis mn 0 dl 0 1 fm 0 0 L4 3 YAxis lb Y Axis mn 2 dl 2 fm O 4 Load DSi mars dataset1 th 5 Band xr DSi 1 ybr DS1 2 ytr DS1 3 fill 6 Band xr DSi 1 ybr DS1 2 ytr DS1 3 cl black 7 Load DS2 seq dataset2 th 9 Curve xr DS2 1 yr DS2 3 cl green 8 Curve xr DS2 1 yr DS2 2 cl blue triangles 10 Load DS3 xy dataset3 th 11 Curve xr DS3 1 yr DS3 2 xs 0 1 cl red 12 EOF Although the sequence of com
139. f the material are changed during the calibration process If either condition is not satisfied then the time steps for all elements are recomputed and saved to the cts file possibly overwriting previous data This activity is documented in the output file as shown in the example below Initializing list ttL PRTC Element time steps read from file ttL PRTC cts Time steps in cts are not usable properties may have changed Stable element time steps are recomputed Min time step for stability 316 32e 9 s Time steps saved in file ttL PRTC cts Unlike previous versions of MARS where the cts file had to be explicitely deleted by the user to eliminate errors during execution the current version of MARS does all the operations automatically and the process is transparent to the user It is still a good idea to periodically check the output and see what operations are performed 11 6 LDPM Visualization There are essentially two classes of methods for visualizing LDPM models tet based methods and cell based methods Tet based methods were the first to be implemented Cell based methods require more memory for storing additional data but provide more realistic erenderingentations of the cracking process their additional computational cost is insignificant Tet based methods are no longer discussed in this manual as they generate lower quality graphical representations Cell based methods and various applications are discussed below Cell plot rend
140. ffness BasedOnTimeStep 0 002 ms 15 4 2 Elastic formulation with slippage In this formulation the beam element is assumed to be cylindrical which is a good approximation for conventional rebars A matrix of points is defined on the surface of the cylinder n points in the axial direction and m points around the circumference The corresponding points inside the tetrahedral mesh are also found Each pair of rebar surface point and corresponding tet point forms a constraint During the simulation the constraint computes the components of the relative velocity between the two points in the local reference system axial along the beam axis radial perpendicular to the beam axis and tangential hoop direction These three components are transfered to the rebar concrete interaction RCI model which is implemented as a Material class The RCI model returns the three componenets of the stress vector in the local reference system which are then used to compute forces at the beam and tet nodes BeamTetConstraintList listName RebarConcretelnteraction BeamList beamListName TetList tetListName NumberOfCircumferentialConstraintsPerBeamElement 3 NumberOfAxialConstraintsPerBeamElement 2 RebarRadius 10 mm Material RciModelName 15 4 3 Elastic formulation with slippage and volumetric effects This formulation has not been implemented at this time When implemented in addition to the relative velocity vector a volumetric expan
141. fingers of the left hand on the w e r and t keys and the right hand on the mouse You can then click the key to select what motion the mouse affects and move the model using the mouse In later versions of Quasar the functionality was modified to be consistent with the functionaly of Paraview Simple click and drag turn model around the two axis co planar with the screen Click and drag while pressing the Shift key spin model around axix perpendicular to the screen Click and drag while pressing the Control key zoom in and out Click and drag while pressing both the Shift and Control keys simultaneously translate the model Mar 2011 The combination Shift Control does not work on Windows Cygwin quasar was slightly modified so that translation can also be accom plished by pressing the Alt key on all systems Advancing frames in a family of files If mars had generated a family of files during its execution for example a series of PLOT 000 PLOT 001 PLOT 002 files created by the PlotList PLOT command you can move back and forth through the sequence pressing the lt and gt keys If your graphics card is fast enough and the model is not too complicated you can achieve a good animation effect interactively 217 Saving and loading a viewing configuration If you like the way the model is being displayed you can save the configuration parameters by pressing the Shift W write key At any time the model can be redisp
142. fter countour variables is selected use appropriate units RangeMinValue O psi RangeMaxValue 10000 psi Fragmentation Commands The beam fragmentation scheme is consistent with the fragmentation schemes employed for quadrilateral and hexahedral meshes Failed elements are not eroded or nullified 71 as in other common schemes Instead elements are disconnected at their common node when local failure criteria are satisfied This is accomplished by inserting a new node and performing a local remeshing by indroducing a discontinuity in the mesh A temporary cohesive element is introduced between the two overlapping nodes for dissipating fracture energy Currently the failure criterion at the node that connects to adjacent elements is based on the average stretch of the two elements When the average stretch exceeds an input failure value then the two elements are disconnected It is possible to introduce stochastic failure by assigning a statistical distribution to the allowable stretches at the nodes In this case the failure stretch for each node is computed at the beginning of the simulation by multiplying the input failure stretch ing a parameter obtained from the requested statistical distribution In this fashion some spots are weaker than others Add following lines after element definition Weibull 4 optional statistical distribution Cracking fail 0 03 decay 0 8 stiff 3 gt fail 0 03 failure str
143. ges to the input and visually display the resulting mesh In a twisted cable individual wires are kept from crossing each other using the edge edge con tact algorithm The node edge contact algorithm may be sufficient for some simulations saving time but being potentially unreliable BeamList CABL 4 sec o m STEL r 0 1 in Generate TwistedCable Strands 7 1 7 or 19 WiresPerStrand 7 1 7 or 19 WireRadius 0 1 in CableLength 100 in ElementLength 2 in WireRotationPitch 10 StrandRotationPitch 20 7 1 2 Select commands The edge select commands make is possible to select a subset of edges using various criteria The selection can be saved in a separate list using the MakeList command or used on a temporary basis The standard Select commands are available Select criterion AlsoSelect criterion Unselect criterion Reselect criterion InvertSelection 65 Selection criteria criterion can take any of the the following forms all all elements do 25 element 25 do 1 5 elements 1 through 5 do 1 100 2 elements 1 through 100 step 2 cp 0 0 1 gt 0 4 elemenst such that cross product with vector 0 0 1 is greater than 0 4 in elements with at least one node selected 2n elements with both nodes selected Example Select do 1 100 select element 1 through 100 AlsoSelect do 201 300 add elements 201 300 In addition three separated commands ara available to select unselect or reselec
144. ght W Fr 0 3 W In MARS this force is obtained by excercising both friction and rolling resistance models The rolling resistance algorithm applies a resistive moment to the wheel equal to Mr 0 3 Rw W This moment tends to reduce the wheel rotation rate and this in turn increase the tan gential force at the wheel to a force of Ft 0 3 W which is the rolling resistance force given by the conventional definition The static and kinetic friction coefficients must be at least 0 3 in the MARS friction model for this to work The above discussion does not take into acount the rotational inertia of the tire Note however that the MARS approach is capable or modeling not only the case of a tire rolling over sand with no apparent slippage but also the more general cases where a tire slips on sand 16 2 Node Node Contact List The node contact makes it possible to detect contacts between nodes particles from a single node list or two node lists For the purpose of contact a dimensionless node takes a spherical shape Contact begins when the external spherical surfaces of two nodes particles start penetrating each other NodeNodeContactList ContacListName 4 variables with the sign inherit defaults from control parameters 1 Define node particle lists NodeList FirstNodeListName 154 NodeList SecondNodeListName 2 Specify optional dimension controls Node1Thickness 0 4 cm Node2Thickness 0 2 cm
145. gments 3 Direction 0 1 0 5 6 2 Strain Gage for tetrahedral solid lists Strain gages can only be placed on the surface or in the interior of a tet solid element mesh Typical input is shown below StrainGage GageX TetSolidList Solid Description Strain gage at location X5 CenterAt 0 cm O cm O cm Length 2 cm Segments 3 Direction 0 1 0 5 6 3 Strain Gage for quadrilateral shell lists Strain gages can be placed on the top surface bottom surface or mid plane membrane strains of a quadirlateral shell element mesh Typical input is shown below StrainGage SGx qsL ShellPartName or QuadShellList ShellPartName Description Strain gage at location X5 CenterAt 0 cm 10 cm 9 cm TopSurface also BottomSurface 1 Length 2 cm Segments 3 Direction 0 1 0 1 If neither TopSurface or BottomSurface keyword is specified then the default is mid plane surface for computing membrane strains 32 5 6 4 TimeHistories The strain gage computes a single variable the total strain at the prescriped point over a prescribed length The only purpose of a StrainGage is to provide a measurement of local strain for time history lists Typical application show up as StrainGage Gagel F TimeHistoryList HIST sg Gagel linear S 1000 linear or logarithm 1 1 The keywords linear and logarithm are used to specify how strains are computed and it makes a difference mostly for larg
146. gular faces from lists that had been previously defined can be selected and passed to IFEM during the initialization phase phase 203 FluidDynamicList 4 tfL List1Name tfL List2Name 22 1 2 Gemini To be written 22 2 Mechanisms List 22 2 1 Shock Strut Assembly The shock strut assembly mechanism computes the forces generated by a shock absorber that connects two nodes in the model At each step the algorithm computes the distance between the nodes and their relative velocity in the direction aligned with them The stroke is computed by subtracting the distance from a stroke offset input parameter Stoke StrokeOffset Distance The value of the Stroke is used to interpolate a tabulated function ForceStroke Curve which give the hydraulic static force generated by the gas or spring in compres sion The range of the stroke is limited by maximum and minimum values If the nodes try to move closer or further away then the stroke range allows then they hit hard stops these are implemented using penalty functions with user prescribed stiffness values The viscous effect of hydraulic fluid is approximated using a linear expression DampingForce Damping RelativeVelocity To avoid wild oscillations in the damping force a moving average of the relative velocity is used Valor 0 99 x Vret mo 0 01 x Vitel The two constants 0 99 and 0 01 are currently hard coded The input commands for the ShockStrut mechan
147. hange is that updates must be synchronized For example a contact update and a MPI domain decom position update must be done at the same time For this reason the UpdateInterval command in the contact lists will be discontinued ControlParameters GlobalUpdateTimeInterval 0 1 ms 4 7 Dynamic Relaxation Dynamic relaxation is a numerical technique that uses an artificial form of damping to dissipate kinetic energy and converge to a steady state equilibrated solution if possible Dynamic relaxation can be used to compute steady state stress under static loading conditions in a system before a dynamic load is applied or find a stable configuration after a dynamic load has been applied In dynamic relaxation nodal velocities of all nodes are scaled by a factor slightly less than one There have been studies that show how to optimize the scaling factor depending on the lowest natural frequency of the system In 22 general if the scaling factor is too small the simulation overshoots the static solution possibly resulting in excessive plastic deformations or damage If the scaling factor is too large closer to 1 the convergence rate becomes very slow There is no magic formula for choosing the optimal scaling factor the user has to develop a feel for what works best using intuition and a trial and error process In some codes the scaling factor is specified via input and remains constant throughout the simulation One disadvantage of
148. he fibers can be translated and reoriented in any direction using the move and rotation commands for node coordinates For example if the fibers need to be aligned with the 63 y direction we need to rotate all nodes by 90 degree around the x axis Tthe sequence of transformations rotations and translations matters as they are done in a sequential order EdgeList ListName Geometry Generate ParallelFibers EditNodeList X Rotate 90 deg Move 0 in 0 in 1 in F When the system of parallel fibers is embedded in a LDPM solid the portions of fiber sticking out of the solid can be chopped Parallele wires This features was inserted for generating models of cable used in suspension bridges BeamList CABL sec o m STEL r 0 1 Ref CABL Generate generate a bundle of wires inside a 10 in circle Bundle L 12 1 1 d 0 2 R 10 o S CYLN Helicoidal wire BeamList HLCD 4 sec o m STEL r 0 1 Ref CYLN Generate Helicoid R 10 L 0 2 n 60 L 10 S CYLN q Random fibers EdgeList FBRS Truss Generate 64 RandomFibers Volume 6 in 6 in 6 in 6 in 0 25 in 0 25 in Length 5 cm ElementSize 1 cm Seed 134 increase Bending to make fibers more contorted Bending 0 9 NumberOfFibers 2000 Twisted Cable This generation scheme generates straight segments of twisted cable The input param eters are shown below The best way to test this generation scheme is to start from the input below ake chan
149. he same keywords that were used to create the list Be carefull to remove all lists that reference directly or indirectly on the lists removed 21 7 create custom versions of mars Since Jan 2011 MARS supports the capability of creating custom versions of the code which incorporate user defined lists and or user defined material models In May 2011 we added two methods for supporting restart procedures A user can write his her own classes in a single file for convenience we will name it user cpp Some examples are provided Some knowledge of the C language is required The user interface is very flexible and makes it possible to define multiple lists and multiple materials simultaneously The interface consists of two methods obLst Model createUserDefinedList string Reader obLst Model readRstUserDefinedList RstReader string Obj Model createUserDefinedObject string Reader Obj Model readRstUserDefinedObject RstReader string The functionality of the user defined lists and objects is achieved through the polymor phic properties inherited from the parent classes This will be explained better in future versions of the manual The basic user cpp file with no definitions is listed below include mars h createUserDefinedList and readRstUserDefineList must always be defined obLst Model createUserDefinedList string nam Reader rdr rdr gt error No user list has been defined return
150. hen the pressures depend on the change in volume affected by the motion of the faces themselves The single value time dependent pressure is multiplied by the current area of each face in the direction opposite to its normal TrngFaceList ListName UniformPressure 1 Specify face list Req either a face list or a shell list Link FaceList listName Link ShellList listName 2 Specify either LoadCurve or Equation Req LoadCurve curveName Equation equationName 92 9 2 2 Special Loads This type of wet face list is used for situations when pressures can be defined as a function of time and initial face location SpecialLoad Burst ConWep QuadFaceList ListName SpecialLoad 1 Specify face list Req either a face list or a shell list Link FaceList listName Link ShellList listName 2 Specify SpecialLoad Req SpecialLoad loadName 9 2 3 Multiple Pressure Histories This list is used when pressure histories are independently computed by a CFD solver at certain stations on the exposed surfaces The pressure histories are read into a Load CurveList which is referenced here This list uses the quad faces of another list either a face list or a shell list The faces in the parent list can change node definition as it hap pens during fragmentation However if the parent list represents the external surfaces of a solid mesh and the number of faces incre
151. here RBR1 and RBR2 are rebar slave particle lists The Make ParticleList is currently not working properly An alternative particle gen eration method is executed using the command Make ParticleList2 This method is intended to work with solid meshes where the external facets are used as external facets of the LDPM mesh The mesh density should be controlled so that the external facets are of the correct dimensions For example a LDPM sphere can be generated using the following commands HexSolidList Ball Geometry 1 Material CONC Generate Sphere Radius 150 mm EdgesOnCircle 20 Make ParticleList2 PRTC seed 173521 h More specifically MakeParticle2 follows the following logic 1 find external quadrilateral faces of the hexahedral mesh 2 split quadrilateral faces into triangular faces 3 the triangular faces are passed to tetgen which uses them when generating the tetrahedral mesh 4 generate a list of particles from largest to smallest 5 insert each particle inside the volume of the solid using a random process every point inside the volume has the same probabilty of being chosen independently of the size of the element that contains it 6 if particle does not interfere with external surfaces or previous particles fix its position 7 once all particles have been placed pass particle list to tetgen what will use them to anchor internal points of the triangular mesh 12 7 Fragmentation Commands select
152. icitly num 345 lst j jN JF 1 254 556 T Mpi directives Mpi OwnedByNodeOwner 2 135 Mpi OwnedByFaceOwner default 1 The sliding with friction constraint is designed to force a set of nodes to move over a surface The constraint perpendicular to the surface is treated using a master slave formulation The resistance to sliding within the plane is treated with a stick slip friction model When the node is sliding the friction factor fy is computed using this expression fr fet Le fey where fp is the kinematic friction factor f is the static friction factor A is a char acteristic length derived from fitting available test data and d is the cumulative slip resulting from slippage computed during the simulation The three parameters are spec ified in the command SlidingWithFriction sff kff A The node and the corresponding point on the surface stick together as long as the tangential inplane force F does not exceed the maximum friction force Fmax fpFn where Fp is the force normal to the surface When F exceeds Fmax the node starts slipping and the friction force is progressively reduced because of accumulated slippage according to the formula above A constraint can return to the stick status if tangential forces cannot maintain slipping In this case the cumulative slippage is NOT reset to zero Warning the current logic does not allow for a node to transition from one face
153. ied on the two objects The relative rotational motion is divided into two components rolling and spinning where spinning is aligned with the direction parallel to the contact normal direction and rolling is perpendicular to it MARS implemente rolling spinning resistance with an algorithm similar to the tan gential friction force algorithm For the rolling component a resisting rolling moment Mr is computed by integrating the relative rolling motion dwr of the two objects multiplied by a penalty stiffness constant Kr Mr Int Kr dwr Mr is limited by the maximum allowable rolling moment Xr which is defined as Xr fr L Fn where fr is a friction coefficient provided via input L is a characteristic length computed by the contact formulation see below and Fn is the normal contact force For nano particle interaction L is the distance between the particle centers for sphere face contacts L is the distance between the center of the ispherical particle node and the midplane of the face face The spinning component or twist is treated in a similar fashion but its parameters are defined independently RollingResistance Kr Kt Fr Ft Beta dt or RollingResistance RollingStiffness TwistStiffness RollingFrictionFactor TwistFrictionFactor dt Beta MinNormalForce 10 nN at 2 nm fO 10 N The optional parameter Beta provides a certain degree of hysteretic damping for small residual os
154. ifference between current coordinate and initial coordinate nd NODS 12 cx D displacement of node 12 in x direction The variables available for each list and the keyword to specify them are described in the various list pages under a section called Time Histories Histories HistoryList HIST TimeInterval 0 1 ms also dt 0 1 ms SkipFirstRecord use this when values at time 0 look wrong enter a list of scalar quantites Global quantities tmx X Translational momentum tmy Y Translational momentum tmz Z Translational momentum rmx X Rotational momentum rmy Y Rotational momentum rmz Z Rotational momentum ke Kinetic energy ie Internal energy iw Internal work ew External work eb Energy balance kex Kin Energy vx 166 key Kin Energy vy kez Kin Energy vz TimeStep 1 CpuTimePerStep 2 CumulativeCpuTime 3 List type quantities ndL Nodes variable options Element type quantities nd Nodes index variable options In the two exemples above nd means NodeList For specific list formats check in the history section of the list help The following options are available to all lists S s scale output by s 0 o offset output by o combined output s value o default s 1 o 0 D take difference between current and initial values U length in convert a length variable into inches Node U modifies the scaling factor s by multiplying current
155. ilable for each of the plottable lists in their sections Below are some examples PlotList listName Quasar All plot all plottable lists PlotList listName plotType hxL WALL CountourPlot StateVariable 1 18 1 1 Parallel processing with Paraview The MARS plot generating procedures for Quasar and Paraview are very different Quasar expects a single input file as such the contributions to each list from the various processes must be combined in MARS Paraview can read and combine files generated from different processes This eliminates the need to combine large data sets in rank zero process which could lead to memory requirement problems The Paraview files have the following name convention plotListName nnn ppp vtu where nnn is an integer rep reseting the sequential time frame and ppp is a three digit integer representing the rank of the process that generated the file In addition to these files there is an additional file name plotListName pvd that contains direction for Paraview on how to load and combine the previous files This is one of the files that shows up in the Paraview Open window and it is the one that should be selected 18 2 Time History Lists The input section for time history requests is typically located at the end of the input deck after all components of the model have been defined and initialized The name of the time history list is used to create the name of the output fil
156. imulate a wheel spinning on an axle MARS searches automatically for constraints the edge that contains the rigid body within a given tolerance is used as master edge ContraintList ListName SlaveRigidBodyMasterNodeConstraint RigidBody RigidBodyName EdgeList ListName Tolerance 1 mm 142 FreeRotations AxialSpinning 15 6 6 Node Node Penalty Constraint This type of constraint ties selected degrees of freedom of two nodes The two nodes are supposed to overlap at time 0 For pratical purposes a tolerance is specified via input and the distance between the two nodes must be less than the tolerance The penalty forces are computed incrementally using velocities e g F t dt F x K vzr vps dt where K is the penalty spring stiffness ConstraintList ListName 4 NodeNodePenaltyConstraint 1 select first node Node ListName nodeIndex or Node ListName cl 5 cm 0 cm 8 cm 2 select second node Node ListName nodeIndex or Node ListName cl 5 cm 0 cm 8 cm 3 Enter penalty spring stiffness Stiffness 1 4e8 dyn cm 4 Enter tolerance Tolerance 1 mm 1 5 Select one of the Translations XXX Rotations 000 This type of constraint ties selected degrees of freedom of two nodes to move together together 15 6 7 Hinge Penalty Constraint This type of constraint ties two nodes so that they can rotate around a corotational
157. in CircularCs Ro 5 125 in Ri 4 875 in or specify cross section properties explicitely CrossSection A 4in2 IX 5 33 in4 IY 0 33 in4 J 5 1 in4 gt Read mesh NodeList nodeListName ReadObjects 1 j ji j2 1 1 2 or generate mesh internally Generate see generate commands for EdgeList 67 7 1 6 Linear Elastic Beams Non Uniform Cross Section This list is essentially an extension of the previous list The only difference is that as the name implies the cross section properties may vary from beam to beam Thus the input line for each beam must include the definition of the section properties explicitely as shown in the example below below EdgeList listName LinearBeamNonUniformCs NodeList nodeListName Material materialName ReferenceSystem refSysName ReadObjects 10 j nt n2 other data 1 1 2 Lin Yo 1 0 A2 17 IY 1 8 121 8 J 3 601 3 Lcm n39 A 3 763 IY 2 813 IZ 2 813 J 5 626 4 Rect bY 2 125 in bZ 1 5 in 5 Rect b 2 125 in 6 Circ Ro 4 in 7 Cire Ro 4 in Ri 3 5 in 8 I wf 2 in tf 2 in hw 2 in tw 0 1 in gt NX O 0M 4 0 N NOOO FP WN For the first beam the cross section properties are entered explicitely In this case it is mandatory to enter the length units L in The local Y direction is entered specifying a vector V using the command Y 0 1 0 The vector V does not need to be unitary or perpendicular to X The local reference system is computed usi
158. in the datafile the program will stop when the eof is reached 2 If the TimeUnits commands is missing MARS assumes that times are given in the default time units which is typically seconds 3 The optional PlotDataSet command is used for generating a sequence of Quasar contour plots for the field variable The name of the files is the same as the name used for the list For example if the list name is Temperatures the sequence of plot files will be Temperatures 000 Temperatures 001 etc 10 9 Ten Node Tet Elements The ten node tet element is essentially a four node tetrahedral element with six additional nodes placed along the six edges of the element Typically the six nodes are placed at the midpoint of the edge However if the edges are located on a curved external surface of a 103 part the midpoint edges should be placed on the curved surface for improved geometric definition The are two ways for defining a ten node tet mesh 1 define a four node tet mesh and then insert the mid edge nodes 2 generate a ten node mesh using a commercial mesh generation package and read the mesh directly into Mars The first method employs the following input format TetSolidList listname 10NodeElement 4 Define 4 node tet mesh using standard commands InsertMidEdgeNodes The second method employs this input format TetSolidList listname 10NodeElement 4 InsertNodeList ReadObjects 45
159. in the example below The fringe keyword is followed by the range for the scalar variable to be used for painting eeL LABL color color numpt n crd ii x1 y1 zl 12 x2 y2 z2 in xn yn zn numee m list il 311 312 220 i2 j21 399 im jm1 jm2 fringe fmin fmax f11 f12 f21 f22 fm1 fm2 eoL Cylinders The reason for this type of list is for representing cylindrical surfaces of structural mem bers such as wires cables rebars etc Figure 1 shows a section of a 7x7 twisted cable consisting of 49 wires The left picture shows a wireframe representation while the right picture shows a solid rendering of the wires using cylindrical surfaces It is obvious why the solid representation is superior The format and amount of data in the plot files for the two types of lists is essentially the same The csL lists have two extra parameters radius and resolution csL LABL front color radius r resolution low medium high numpt n crd ii x1 yl zl i2 x2 y2 z2 in xn yn zn numcs m list il j11 j12 i2 j21 j22 im jm1 jm2 eoL It is also possible to paint cylindrical setctions with colors that vary continuously from blue to red to create fringe plots of physical quantities To exercise this option the fringe command line must be entered after the list block as shown in the example below The fringe keyword is followed by the range for the scalar variable to be used for painting csL LABL numpt n crd 221 ii x1 y1
160. inc com mechanics MARS Online MarsManual htm This contains several examples with figures However since it is hosted at a re mote web site that makes it difficult to keep it syncronized with the latest version 10 of MARS As such its use is more for general reference but not for specific instruc tions on how to use input commands e A blog hosted at http es3 mars blogspot com This is useful for providing a chronology of when new features are inserted in MARS e A built in manual printed to an ASCII text file directly from the Mars executable The information in the manual is consistent with the features built into the code Indeed it is relatively simple to update the documentation at the same time a new feature is added or and existing feature is modified since documentation coexists in the same source files where the algorithms are implemented implemented MARS is a fast developing solver with new features being added on a regular basis and occasional reorganization of existing material As such the built in ASCII manual was playing a critical role in providing updated documentation consistent with the version of the program from which it was printed However the format of the documentation was not practical since it did not provide an updatable table of contents Furthermore it could only be explored using a text editor like vi or Notepad which provide the basic find tools for searching specific words inside a file
161. index Lists lists Lists lst add list to general lists gt addLst ndL lists gt addLst hxL createUserDefinedList string must always be defined obLst Model createUserDefinedList string nam Reader 4 return NULL _ createUserDefined0bject string must always be defined Obj Model createUserDefinedObject string nam Reader 4 return new GenCube nam A sample input for generating a 2 in cube with 5 elements per side is given below Generation of hex element cube ControlParameters Units English UserDefinedObject Cube Side 2 in Elements 5 EOF 21 7 4 User defined load curve Below is an example of an Obj class derived from the LoadCurve class that implements a sinusoidal force function include mars h A PSS se a e a ena class SinFunction public LoadCurve private real A amplitude real w frequency public SinFunction string nam LoadCurve nam SinFunction void read Reader real interp real x return A sin w x void SinFunction read Reader rdr 4 A w 0 while true string lbl rdr gt getShortLabel 194 if 1bl else if 1b1 break else if 1bl rdr gt readLine else if 1bl Frequency w rdr gt readFrequency else if 1bl Amplitude A rdr gt readForce else rdr gt error Invalid ke
162. ing nam Reader rdr return new myMaterial nam The class definition for a new material follows the pattern below the class name myMaterial can be replaced by any other name which is not already used for defining another class In the name has already been used this will be apparent during compilation class myMaterial public Material private parameters for material model public myMaterial myMaterial string n Material n myMaterial void read Reader other methods The best way to write a new material model is to start from an existing model for ex ample the linear elastic material model which is made available by request and includes comments There are several polymorphic methods which may need to be defined de pending on the model formulation and required functionality This is also described in the sample section In the input file the new material is defined using the command UserDefinedObject and later refereced as any other material UserDefinedObject Steel input data 191 HexSolidList Part FBSingleIP Material Steel 21 7 3 User defined mesh generator Below is an example of an Obj class whose only purpose is generating a mesh by taking advantage of the Mars methods include mars h class GenCube public Obj 4 public GenCube string nam Obj nam GenCube void read Reader F void GenCube read Reader rdr rea
163. integration scheme for solving the equation of motion of large systems It implements all the capabilities and versatility of general finite element codes In addition MARS features some unique techniques such as the Lattice Discrete Particle Model LDPM and adaptive remeshing algorithms for shell and solid meshes which facilitate the solution of problems involving structural break ups fragmentation and post failure response under extreme loading conditions MARS includes standard finite element and discrete element features such as e QPH quadrilateral shell elements with physical hourglass stabilization e DKT triangular shell elements e Beam elements with various built in cross sections e 8 Node Flanagan Belytschko hexahedral elements with hourglass stabilization e Hyper elastic solid elements e Various constraint formulations including concrete rebar interaction interaction e Automatic contact algorithm for node face edge edge node edge node node con tact detection e Macro particles for simulating discrete elements with complex shapes Additionally MARS has an object oriented architecture which makes it possible to add new capabilities in an efficient and systematic fashion All entities in MARS are organized in a hierarchical framework Classes of simple entities such as edges and faces are used to derive more complex entities such as beams and shells Since February 2011 MARS incorporates an interface which makes it
164. interpolated in time The interpolation time is the time when the forces are calculated Up to 10 scalar fields can be specified 7 The MassScaling command is used to increase the mass of elements that have very small Courant time steps Typically in large meshes there are a few poorly shaped elements slivers flat elements etc that would force the entire simulations to run using small integration time step It is a common practise to increase the density of these elements because their mass is very small anyway The time steps for Ldpm elements is computed by assemblying the stiffness matrix and computing its maximum eigenvalue e The time step dt is then defined as e These operations are performed during the initialization phase A summary of these operations is printed to the output file and looks like this Min time step for stability 808 24e 9 s 510 tets were made heavier Current min time step 1000 00e 9 s Actual total mass 0 914831 kg Adjusted total mass 0 924320 kg Percent mass increase 1 03727 Note that the operations for computing the time steps are quite expensive but need to be done only once Since Oct 2011 the stable time steps for all elements are auto matically computed and saved to a binary file named using the following convention ttL listName cts The next time the same model is run Mars looks for this file and reads the previously saved time steps Since Oct 2011 Mars performs further checks to
165. ion The particle rebar interaction constraint is then applied to each pair Currently we select all particles that are suitably close to the rebar elements More specifically we do not include particles whose spherical volume overlaps the cylindrical volume oc cupied by the rebars We do include however particles when their gap from the rebar is within a range specified by the analyst using the MinimumGap and MaximumGap commands The rebar overlapping particles are connected to the other particles as if the rebar does not exist Their effect is accounted for in the bond parameters when parameter calibration is performed The constraint formulation ties a single particle to a single beam element defined by its two nodes The formulation consists of two parts a kinematics part where the local deformation state is computed from particle node velocities and rotations a force reduction part where local stresses are used to compute forces and moments at the particle and beam nodes nodes ParticleRebarInteractionList Constraints 144 NodeList Particles BeamList Rebars Material ConcRebarInt RebarRadius 0 6 cm MinimumGap 0 1 cm MaximumGap 2 0 cm EquilibratedAreas opt 15 8 Particle Fiber Interaction List The particle rebar interaction list is intended to couple rebar embedded in concrete LDPM regions The input consists of the list of nodes which are used for defining the LDPM region and the list of rebar elements The concre
166. ions Generate Prism X first 0 in X inc 2 in X rows 10 Y first 0 in Y inc 2 in Y rows 10 Z first 0 in Z inc 2 in Z rows 10 58 Cylinder Center 0 in O in O in Radius 4 in Length 8 in Sector x gt 0 y gt 0 0 lt z lt L r lt R Radius 4 in Length 8 in 2 b Read spheres from node list NodeList ListName InsertNodeList ListName Optional commands RandomRotations Opt Set vz 100 m s Opt ReadPlotFile filename 6 2 1 Flex 4 Sphere MacroParticle MacroParticleList ListName Flex4SphereParticle Parameters 4 SphereRadius TetRadius Stiffness Mom Vis 6 2 2 Rigid 4 Sphere MacroParticle Each macroparticle consists of an assembly of four spherical particles placed at the ver tices of a tetrahedron and rigidly tied together MacroParticleList ListName Rigid4SphereParticle Parameters 4 SphereRadius TetRadius 59 6 2 3 Rigid 3 Sphere MacroParticle Each macroparticle consists of an assembly of three spherical particles placed at the vertices of a triangle and rigidly tied together MacroParticleList ListName Rigid3SphereParticle Parameters 4 SphereRadius 3 mm TetRadius 2 mm 7 Edge Truss Beam Elements 7 1 Edge List Edge lists are collections of edge type objects In its simplest geometric form an edge is a line segment spatially defined by two nodes MARS implements edges in an Edge class Several st
167. ire geometry of the model This is accomplished by using the inheritance properties of OOP For example an LDPM element which is one of the most memory intensive and computationally demanding element requires almost 6 kilobytes of memory per tet element However the geometry of a tetrahedral is defined by 4 nodes 168 which require either 16 or 32 bytes depending whether we use 4 byte integers of 8 byte addresses for identifing the nodes During the reading phase each processor reads and stores all tetrahedral elements in TetSolidList During the initialization phase each processor converts the tet elements under its jurisdiction from geometrical elements to LDPM elements elements 19 1 Decomposition Schemes Currently MARS provides three schemes for performing domain decomposition 1 Orthogonal Recursive Bisection ORB 2 Octree 3 METIS libraries The ORB method has been used extensively and is the recommended method The octree method is inferior to the ORB method The METIS approach is in a development stage More information on the three approaches is given below 19 1 1 Orthogonal Recursive Bisection The orthogonal recursive bisection method surrounds the domain by a parallelepipedal box The box is first divided in the longest direction in two smaller boxes each containing the same number of objects Each of the two boxes are split into two smaller boxes and so forth until the desired number of boxes domains
168. is generated Some figure and a more extensive explanation are given at these two web sites http ww netlib org utk 1si pcwLSI text node253 html http ww2 cs mu oz au 498 notes node52 html Currently the orthogonal recursive bisection Bisect method is the preferred domain decomposition in Mars The input commands for ORB are Bisect Label List ttL ListName Req FirstNode 0 1 NodeSpan 128 must be a power of 2 1 Fixed 2 Verbose 1 If FirstNode and NodeSpan are not specified MARS will use all available proces sors for the run If NodeSpan is greater than the number of available processors MARS will automatically reduce it 169 WARNING The keywords FirstNode and NodeSpan will be changed to FirstRank and RankSpan to reflect the common practise in MPI programming to refer the processes assigned to each core as rank 0 rank 1 etc 2 Without the Fixed option the domain decomposition is updated as frequently as specified in the GlobalUpdateTimeInterval command in the ControlParameter sec tion When the domain decomposition is updated lists are affected in different ways nodes in node lists are redistributed by default elements are not migrated active con tacts are migrated across domains if necessary and new contacts are found within each domain In the next paragraphs we will explain in detail how this parallelization scheme is implemented in MARS The list selected by
169. ism are given below Mechanisms ShockStrut FirstNode NodeListName cl 0 cm O cm 10 cm SecondNode NodeListName cl 0 cm 0 cm 40 cm ForceStrokeCurve CurveName MinimumStroke 0 in MaximumStroke 16 in StrokeOffset 81 cm Stiffness 1 e6 lb cm Damping 80000 N m The units of damping are Force Velocity Since these units are currently not available in MARS assume time is given in seconds and use units for Force Length The history of significant variables can be printed out using the following commands 204 TimeHistoryList ListName 4 mm listName Stroke mm listName StaticForce mm listName DampingForce mm listName BottomTopForce mm listName TotalForce 22 2 2 Brake The brake mechanism provides a braking moment to a wheel In this formulation the wheel motion is associated to that of a reference node note that the wheel and tire can be modeled in great detail however it is necessary to associate a single node to the motion of the wheel and that node should be located on the axis of rotation The formulation requires the definition of a Runway this is a flat surface which is identified by the first face of a face list The face list can consist of triangular or quadrilateral faces The surface is used to compute the distance of the axle from the surface and the direction of the axle parallel to the surface The braking action is applied as a torque in the direc
170. ist eeL EdgeList ttL TetSolidList The rigid body initialization method automatically computes the total mass center of gravity and tensor of inertia of the rigid body based on the inertial properties of the components These values can be overwritten using the following commands RigidBody BodyName Set Mass 45 3 Kg ine 543 Kg m2 34 Kg m2 167 Kg m2 130 Boundary conditions and initial velocities imposed at the nodes of the original lists are disregarded Boundary conditions and initial velocites can be imposed on the rigid body The general input format for a rigid body is shown below RigidBody RigidBodyName 1 enter a sequence of lists of objects that will be connected as a rigid body NodeList NodeListName HexSolidList SolidListName HexSolidList SolidListName BeamList BeamListName TriangShellList ShellListName 2 set boundary conditions optional Set Translations XXX Set Rotations XXX 3 set initial velocities and or rotation rates optional Set Velocities 15 in s 0 in s O in s Set X Velocity 15 in s Set RotationRates 0 rad s 0 rad s 14 rad s Set Z RotationRates 14 4 overwrite calculated mass optional Set Mass 15 Kg 13 1 Time History Commands The following line commands are intended to be used inside TimeHistoryList s to produce records of rigid body variables TimeHistoryList Hist 4 RB bodyName vx x component of CG velocity RB bodyName
171. istName X Force rcL ContactListName Y Force rcL ContactListName Z Force 17 Interference Check This is a collection of methods designed to detect part overlapping or crossing of a part through a surface Their main purpose is to ensure that contact conditions work properly These methods do not affect the results of a computation and should be disabled when the user has gained confidence that there is no unwanted part penetration penetration 17 0 2 Node Hex Overlap Check This feature is designed to detect whether any node from a list are located inside any hex element form another list The main purpose of this is to ensure that the contact algorithms is able to prevent contacting nodes from penetrating into a solid part Checks 4 NodeHexOverlapCheck NHOC 4 NodeList NODS HexList HEXS CheckTimeInterval 0 1 ms Disable 162 Make the CheckTimelnterval smaller than the time step to perform the check at every check although this frequency is most likely execessive 17 0 3 Node Tet Overlap Check This feature is designed to detect whether any node from a list are located inside any tet element form another list The main purpose of this is to ensure that the contact algorithms is able to prevent contacting nodes from penetrating into a solid part Checks 4 NodeTetOverlapCheck NTOC 4 NodeList NODS TetList TETS CheckTimeInterval 0 1 ms Disable Make the CheckTimelnterval smaller than the time step to
172. istory MaxStepHist J 4 4 Plotting Defaults Plotting defaults are used for initializing the attributes of plottable lists PlottingDefaults TimeInterval 0 1 ms 1 DisplacementScalingFactor 1000 2 X DisplacementScalingFactor 1000 2 Y DisplacementScalingFactor 1000 2 Z DisplacementScalingFactor 1000 2 ThickShells BinaryPlotFiles 21 1 This command is used for setting a default time interval for all plot lists This value can be overwritten inside a PlotList input section 2 These commands are used to magnify displacements In most large deformation problems displacement magnification will produce meshes that are too distorted It can be useful for small deformation problems like the fracturing of concrete specimens under quasi static loadings 4 5 Contact Defaults Contact defaults are used for initializing some parameters in the contact lists Cur rently only the detection distance parameter and the node edge face thicknesses can be specified The syntax is shown below ContactDefaults 1 UpdateInterval 0 1 ms Discontinued Apr 2011 DetectionDistance 0 2 cm StaticFriction 0 3 Discontinued Apr 2011 DynamicFriction 0 2 Discontinued Apr 2011 NodeThickness 0 1 in EdgeThickness 0 2 in FaceThickness 0 1 in 4 6 Global Updates This command has been added in June 2011 It is meant to replace the updates specified in some lists such as contact lists mpi domain list etc The reason for this c
173. l edge 0 length of edge int nel 0 number of elements per side while true string lbl rdr gt getShortLabel if 1b1 else if 1b1 break else if 1bl rdr gt readLine else if 1bl Side edge rdr gt readLength else if 1bl Elements nel rdr gt getInt else rdr gt error Invalid keyword in GenCube read J int i j K int nps nel 1 int nnd nps nps nps ndLst ndL new ndLst Cube ndL gt allocate nnd real crx cry crz dc edge nel generate nodes Node nd new Node nps for i 0 i lt nps i 192 nd i new Node x nps crx 1 dc for j 0 j lt nps j nd i j new Node nps cry j dc for k 0 k lt nps k crz k dc nd i j k new Node crx cry crz ndL gt append nd i j k hxLst hxL new hxLst Cube hxL gt setNodeList ndL int nhx nel nel nel hxL gt allocate nhx Node n1 n2 n3 n4 n5 n6 n7 n8 Hex hx generate hex solid elements for i 0 i lt nel i 4 for j 0 j lt nel j for k 0 k lt nel k 4 ni ndl ill jill kl n2 nd i 1 j kl n3 nd iti j 1 k n4 nd il j 11 k n5 ndl ill j k 1 n6 nd i 1 j k 1 n7 nd i 1 j 1 k 1 n8 nd iJ j 1 k 1 hx new Hex hx gt setNodes n1 n2 n3 n4 n5 n6 n7 n8 hxL gt append hx ndL gt reindex hxL gt re
174. l elements in the list 2 These variable are computed taking the maximum value or minimum value of a quantity over all elements in the list 12 5 Plot Attribute Commands Plotting attributes can be specified in the element list or in the plot list 127 HexSolidList PART 4 PlotAttributes he ContourVariable sv 1 NoNodalAveraging NoSmoothing PlotList PLOT hxL SolidPart you may selecte one of the contour variables below ContourVariable StateVariable 1 ContourVariable Velocity ContourVariable X Velocity ContourVariable Y Velocity ContourVariable Z Velocity ContourVariable MinPrincipalStress ContourVariable MaxPrincipalStress ContourVariable VonMisesStress stresses are averaged at the nodes unless NoNodalAveraging SelectedElementsOnly NoSmoothing discrete colored fringe plots prescribe range after countour variables is selected use appropriate units RangeMinValue 0 psi RangeMaxValue 10000 psi hxL PRT2 OutlineOnly 12 6 Particle Generation Commands HexSolidList WALL Material CNCR mtCConcrete type Enter or generate FE mesh Make ParticleList PRTC 4 GapScalingFactor 0 05 lt gsf gt min inter particle gap gt 0 05 x min aggr diam df1 0 1 Seed 1543 integer seed for random no gener PlotSieveCurve Debug ConstantSpacedParticlesOnEdges 128 if rebars are present in the model enter Make ParticleList PRTC npL RBR1 npL RBR2 w
175. layed in the saved configuration by pressing the Shift R read key Viewing mesh grids outlines faces and particles When quasar starts the outline faces and particles are automatically displayed It is possible to toggle all these quantities on and off by pressing the f g o and p keys f toggle displaying of all faces in the model g toggle displaying of the mesh grid o toggle displaying of outline p toggle displaying of spherical particles and cylindrical surfaces This can be very useful when dealing with a complex model where it takes a while to repaint the frame In such cases particles and faces can be turned off and the model can be repositioned quickly using its outline Once the desired position is achieved toggle on the desired entities Saving the screen to a graphics file You can save the current quasar screen to a ppm file that can be later incorporated in a WORD document by pressing the Shift G key This generates a file named Turning model components on and off You can select what parts of the models to display by pressing the right button of the mouse This action brings a pop up menu with six entries Select Model and the rest should be intuitive Making frames for an animation Press Shift M movie or press the right button of the mouse to bring up the pop up menu and select Options Generate Movie Frames This will generate a sequence of plot ppm nnn files that can be l
176. le nodes All all nodes in the list SelectedNodes Nodes can be selected inside this block using the EditNodeList command EditNodeList Select cx gt 0 SelectedNodes 1 The Distribute command is used to distribute the load equally among the specified nodes If this command is not used each node will be loaded with the specified load This requires requires Alternate way NodalLoadList ListName NodeList NodeListName LoadCurve LoadCurveName ReadObjects 2 nd dx dy dz scl 132 an 00 H gt Roe oo oo NO N an w 14 2 Prescribed Velocities List Use these commands to specify velocities or rotatation rates to a single node a set of nodes or a rigid body Velocities vary in time according to the time history specified in the load curve PrescribedVelocityList ListName 4 1 Enter Rotations if you want to apply rotations instead of velocities Opt Rotations 2 Enter either NodeList or RigidBody Req NodeList NodeListName RigidBody RigidBodyName 3 Enter LoadCurve Req LoadCurve LoadCurveName 4 Enter load direction Req Direction 0 1 0 5 Enter scaling factor Opt Scale 1 6 Use one of the four lines below to specify nodes Required when NodeList is used Node 1 single node Nodes 1 4 7 9 multiple nodes All all nodes in the list SelectedNodes nodes can be selected inside this block using the EditNode
177. low scale s numnp n crd 11 xl yl z1 r1 i2 x2 y2 z2 r2 in xn yn zn rn eoL The spheres can be painted colors continuously varying from blue to red to create fringe plots of physical quantities To exercise this option the fringe command line must be present before the sphere data is entered as shown in the example below The fringe keyword is followed by the range for the scalar variable to be used for painting For each particle the scalar variable follows the radius npL LABL fringe fmin fmax resolution high medium low scale s numnp n crd 219 ii x1 y1 z1 r1 f1 i2 x2 y2 z2 r2 f2 in xn yn zn rn fn eoL Lines Line lists are used for a variety of purposes For example they can be used to outline the sharp edges of a solid part or the edges of shell parts They also can be used to trace the axis of beam elements for quick rendering short segments perpendicular to triangular or quadrilateral faces to indicate positive directions Lines are typically painted with a uniform solid color defined by the keyword color The default color is black eeL LABL color color numpt n crd ii x1 yl zl 12 x2 y2 z2 in xn yn zn numee m list i1 j11 j12 12 j21 j22 im jmi jm2 eoL It is also possible to paint lines with colors that vary continuously from blue to red to create fringe plots of physical quantities To exercise this option the fringe command line must be entered after the list block as shown
178. lse if this list is specified via input enter 87 NodeList NODS or InsertNodeList 1 ReadObjects 345 i ni n2 n m 1 124 243 56 165 2 56 238 121 78 selection commands optional Select Unselect see below Refine 2by2split refine current mesh into 2x2 tiles Make EdgeList ListName AllEdges Make EdgeList ListName SharpEdges Make EdgeList ListName UnmatchedEdges make FaceList of selected faces Make QuadFaceList ListName make NodeList of nodes attached to selected faces Make NodeList ListName Write PartMeshDataFile ListName use this command to read nodes and elements previously saved ReadFile PART mrs SelectNodes 1 ReselectNodes 1 UnselectNodes 1 1 The SelectNodes ReselectNodes and UnselectNodes commands are used to per form selection tasks on the nodes used in the definition of previously selected faces or all faces The node selection can then be used in the node list for various other tasks 9 1 1 Select commands The selection commands are used to select a subset of the elements in the list Select criteron AlsoSelect criteron Unselect criteron Reselect criteron InvertSelection criterion can take one of the following forms all all faces for 25 face 25 for 1 5 faces 1 through 5 for 1 100 2 faces 1 through 100 step 2 in faces that have at least 1 node selected 4n faces that have all 4 nodes selected 88 tw 0 5 4 faces that
179. m s 1e 05 lb in s 0 0115212 kg m s 1 kg m s 1 Quantity RotationRate Reference unit rad s deg s 0 0174533 rpm 0 10472 rps 6 28319 rad ns le 09 Quantity Moment Reference unit N m dyn cm 1e 07 lb in 0 112985 Ibf in 0 112985 kip in 112 985 nN nm le 18 Quantity Acceleration Reference unit m s2 cm s2 0 01 mm s2 0 001 nm ns2 1e 09 in s2 0 0254 ft s2 0 3048 17 Quantity EnergyDensity Reference unit J m3 erg cm3 0 1 Ibf ft2 47 8867 lbf in2 6895 68 nJ nm3 1 nnJ nm3 1e 09 Quantity Power Reference unit W kW 1000 MW 1e 06 erg s 1e 07 hp 745 7 ft lbf s 1 356 in lbf s 0 113 Quantity PowerDensity Reference unit W m3 kW m3 1000 MW m3 1e 06 erg s cm3 0 1 hp m3 745 7 Ibf s ft2 47 8867 1bf s in2 6895 68 Quantity Temperature Reference unit degK degC 1 degR 1 8 degF 1 8 Quantity TemperatureRate Reference unit degK s degC s 1 degR s 1 8 degF s 1 8 Quantity Temperaturelnvrs Reference unit 1 degK 1 degC 1 1 degR 0 555556 1 degF 0 555556 4 2 Actions These commands are used to schedule various types of tasks to be performed at specific times or steps during the simulation Each command is entered in a single line of input These commands are contained in the ControlParameter input block ControlParameters FlushTHFiles Every 0 1 ms GoInteractive AtTime 10 ms WriteRes
180. mands can be changed the order may affect the final look of the plot Following is an explanation of the various lines Line 1 the title line is not used in the plot but can be used as a comment line Line 2 the second line specifies the format of the x axis using the following arguments lb Title of the x axis mn value of the first tic dl increment for each tic fm format for 227 tic value number of zeros after period corresponds to the number of decimal digits The default length of the x axis is five inches which major tick at one inch intervals Line 3 the second line specifies the format of the y axis using the following arguments 1b Title of the y axis mn value of the first tic dl increment for each tic fm format for tic value number of zeros after period corresponds to the number of decimal digits The default length of the x axis is three inches which major tick at one inch intervals It is possible to make the height two inches long using the L 2 command Line 4 The Load command is used to load a data set in memory A data set consists of a set of two or more records represented as arrays of real numbers The label DS1 is a token to be used to identify the load set in following lines mars is a keyword to identify the format of the data file dataset1 th is the name of the data file Two additional file formats are available seq at line 7 and xy at line 11 Both formats are explained below Line 5 The Band command is used t
181. mid edge nodes to a regular 4 node tetrahedral mesh and convert it to a 10 node tratrahedral mesh The element formulation is implemented in method TetCosserat10 calcForces The example below shows how to retrieve the current element data and how to save the computed forces and moments at the nodes in the global list array for later processing real TetCosserat10 calcForces Node ndF real f Node ndM real m 4 retrieve nodal coordinates and velocties real cOx nd 0 gt getCx coordinate node 0 direction x real vOx nd 0 gt getVx velocity node 0 direction x real wOx nd 0 gt getwx rotation rates node 0 direction x enter detail of formulation here store nodes and forces in arrays ndF 0 nd 0 vertex nodes force array ndF 4 nM 0 midnodes force array ndM 0 nd 0 vertex nodes moment array 190 f 0 f 3 fOx ff 1 Oy f 2 fix f 4 fiy f 5 f0z forces at node 0 fiz forces at node 1 iO m0x ml 1 m0y m 2 m0z moments at node 0 return M_BIG 21 7 2 User defined material In the MARS architecture the class Material is derived from the generic class Obj which includes various types of entities such as ReferenceSystem s LoadCurve s etc The new material is integrated in the Mars architecture using the Model readUserDefinedObject O class myMaterial public Material 4 Obj Model createUserDefinedObject str
182. monNodes For the nodes to be selected they have to be used in both element lists intersection of the two sets If we had used SelectNodes in the HexSolidList then the selected nodes would include all nodes from both lists union of the two sets 6 1 2 Set Commands The Set command is used to set variables or boundary conditions on a set of nodes that was previously selected using the Select commands If no selection was previously done then the Set command will operate on all nodes NodeList ListName 4 Set TranslationsBC XXX Set RotationsBC XXX Set X Velocity 100 in s Set Y Rotation 100 1 s Set Radius 0 3 in 6 1 3 Scale Commands Scale commands make it possible change node coordinates and particle radii by multi plying them by a constant factor For example the command 51 Scale 1 4 multiplies x y z coordinates and radii of all selected nodes by the value 1 4 If different scaling factors are desired in different directions these commands may be used X Scale 1 Y Scale 1 Z Scale 1 R Scale 0 9 scale particle radii by 0 9 6 1 4 Translate Rotate Commands Translate Rotate commands are used to move and reorient a set of nodes These com mands are executed during the reading phase in the order in which they appear This is important because a rotation after a translation gives different results than the same translation after the same rotation X Translate 3 cm Z Rotate 90 deg
183. mpt type gt mars mars rst 000450 MARS will recognize the restart file load the data base and go in interactive mode In a batch file enter the command mars B mars rst 000450 gt output dat In the Mars launcher grab and drag the mars rst xxxxxx file from the left column into the command line and then press the execute button Restart files are portable across computer systems They are written in binary for mat MARS automatically detects whether they are written in little endian format and converts them if necessary It is possible to make changes to the database including adding new components to the model In this case the syntax is mars modification mrs R mars rst 000450 where file modification mrs follows the rules of conventional MARS input files An example will be provided in the future 21 4 pre process input files Since Dec 2009 MARS supports the capability of preprocessing input files using the same syntax employed for C and C source code A useful overall description of the C preprocessor can be found at at http en wikipedia org wiki C_preprocessor Gnu org provides a more comprehensive description of all capabilities of the prepro cessor at at http gcc gnu org onlinedocs cpp index htm1 Top 179 Of interest are the sections regarding conditional compilations compilations The idea is to incorporate pre processing directives in a MARS input file and use the preprocessing capabilites of the C compil
184. n 0 0 1 RadialDirection 1 0 0 86 TriangFaceList ListName DktShell ReferenceSystem CylRsS 8 3 1 Time Histories It is possible to save the histories of state variables for the DKT element Note that there is a matrix of integration points there are typically four sets of integration points over the surface of the element Each set consists of the integration points through the thickness Various combinations are possible and listed below TimeHistoryList listName tfL shellList 44 sv 1 tfL shellList 44 SIP 2 sv 1 tfL shellList 44 TIP 1 sv 1 tfL shellList 44 SIP 2 TIP 1 sv 1 Py If no surface point or thickness point is specified then the average across of all points is takes If a surface point is specified SIP 2 then the average through the thickness is computed If a thickness point is specified TIP then the average over the four surface point is computed 9 Quadrilateral Faces and Shell Elements 9 1 Quadrilateral Face List A quadrilateral face shell list consists of a collection of homogenous two dimensional 4 node elements Currently four types of sublists are available QuadFaceList listName listType list attributes FrontFaceColor gray default lightgray BackFaceColor red default lightgray if this list is internally generated enter Generate else if this list is derived from another list enter SplitTriangFaceList TFLS e
185. n any of the curves All pressure histories are moved forward in time by that amount 2 The TimeShift command moves all pressure histories in time by the specified amount Time histories variables variables 8 3 Triangular DKT Shell List This lists consists of flat triangular shell finite elements based on the discrete Kirchhoff DKT plate element formulation The input for this list include all the commands for the geometric triangular list in cluding generation selection and make commands In addition the following commands are available TriangFaceList ListName DktShell commands from geometric shell list select one of the cross sections required SteelShell m MATE i 3 specify a reference system for orienting local IP optional ReferenceSystem refSysName it is possible to run DKT with 2 or 3 IPs but 4 is better NumberGaussianPoints 4 optional If the reference system is not entered then the global reference system is used If the number of gaussian points is not entered then the four integration points are used The specification of the reference system can be useful for specific geometries For example in the case of a cylindrical shell it is desirable to compute stresses in a cylindrical coordinate system so that it would be easy to generate contour plots of the hoop or axial stresses This is accomplished using the commands below ReferenceSystem CylRS cylindrical AxialDirectio
186. n the mesh is generated in the general reference system 2 No restrictions are given on the number of elements along the circumference 3 The size of the elements in the axial direction will be equal or less than the value entered 89 Disk Disk ReferenceSystem refSysName 1 EdgesOnCircle 64 must be multiple of 8 Radius 4 in 1 Same conventions as those ised for the Cylinder The disk lays in the x y plane and is centered at the origin Rectangular surface Rectangle LengthUnits in 110 16 subd in x and y directions 5 5 coord in x and y directions 5 5 R PLAN in the x y plane with corner at the origin 0 Triangular based unstructured mesh This generation method is suitable for generating a large class of coplanar unstructured meshes The user must specify the outline of the surface Any number of internal holes are also possible Mars uses the program Triangle see Plug in section which generates a triangular mesh The output of Triangle is read back into Mars which then splits each triangle into three quadrilateral faces The mesh is generated in the X Y plane If a reference system is specified then the mesh is rotated to that reference system Note that the mesh can also be translated and rotated using the conventional commnads for node lists Triang based u in length units for this part RSYS reference system optional 0 15 typical length of
187. nders b provide a radius for edge edge contacts c compute masses for rigid body applications 2 Density is used for computing nodal masses when an edge list is used in the definition of a rigid body 3 The Write PartMeshDataFile command creates an ASCII file with node and edge information with this format commented title line InsertNodeList listName 4 LengthUnits in ReadNodes 235 nodal coordinates ReadObjects 211 edge indeces EOF The mesh data can be used as input to other files using the Read PART mrs command When generated from other lists it is possible to change some list attributes like color as shown in this example QuadFaceList ListName Make EdgeList EdgeListName SharpEdges EdgeList EdgeListName Color Red 61 7 1 1 Generate commands The generate commands are shared with beam lists Multiple parts or lines can be generated within the same block of instructions Generate StraightLine StraightLine Helicoid The overlapping nodes from these diferenct parts can be merged using the MergeNodes command Straight Line StraightLine LengthUnits cm FirstPoint 0 0 0 0 O LastPoint 20 0 0 0 you can also enter first and last node provided you entered a node list FirstNode 1 LastNode 2 NumberOfSegments 10 BeamList RBRS sec o m STEL 3 Generate Rebars ReferenceSystem RSYS rebars are aligned in RSYS
188. nerates a system of cells embedding the aggregate particles and interacting through triangular facets A mesoscale constitutive law governs 105 the interaction between adjacent cells and it simulates various features of the mesocale response including cohesive fracturing strain softening in tension strain hardening in compression material compaction due to pose collapse frictional slip rate and creep effect for dynamics etc LDPM has been extensively calibrated and validated in the last few years and it has shown superior capabilities in reproducing and predicting qualitative and quantitative concrete behavior under a wide range of loading conditions LDPM was also used successfully to the simulation of projectile penetration blast and fragmentation Figs 2a g Figures will be included in later version of the manual show a gallery of some of the results obtained using LDPM Fig 2a shows LDPM simulation result for mixed mode fracture The simulation is relevant to a four point bending test of a specimen with two notches As observed in experiments two curved fractures propagate from the notch tips towards the opposite sides of the specimens In Fig 2b a snapshot of a blast simulation is shown Fig 2c is depicts the impact of a steel rod onto a concrete block As observed in reality the number of fragments increases with the initial impact velocity and there is a transition from a failure with localized cracks left to a failure with com
189. ng these operations Z X x V and Y Z x X where x is the vector outer product For the second beam the Y direction is specified using a third node n3 In this case the Y direction is oriented from the projection point of n3 on the beam toward node n3 itself For the third beam the cross section is rectangular with 2 125 in in the local Y directions and 1 5 in in the Z direction Square cross sections can be also specified as in the fourth beam For the fifth beam the cross section is a solid cylindrical rod with radius 4 in Annular cross sections can be specified as in the sixth beam 7 1 7 Plotting options The plotting options are used for choosing three dimensional rendering shapes in Quasar Note that special shapes can be done in Paraview using gliphs objects These shapes are available Cylinders HiResCylinders and Pills Any one of these shapes can be specified in this fashion EdgeList listName Geometry 68 PlotAttributes Select only one Cylinders HiResCylinders Pills The Pills option is used to render each segment in the shape of a pill that is a cylindrical portion with two hemispherical caps The diameter of the pill is entered using the Radius command in the regular section of the input When Cylinders or HiResCylinders is used in Paraview the data is written in a special format so that the rendering is done in Paraview using glyphs These are the steps in Paraview for rendering the
190. ns optional commands Make tfL XFAC generate a list of external surfaces select commands Select Unselect see below Select Unselect see below Face Orientation commands DrientFaces Toward 0 in 0 in 7 in OrientFaces AwayFrom 0 in O in 7 in DrientFaces Direction 0 1 0 EditNodeList nodlist commands Write PartMeshDataFile PRT1 mrs Read PRT1 mrs E 94 9 3 1 Weibull distribution The Weibull distribution has been used very effectivley for characterizing probabilistic failure in materials and mechanical components The probability density function is defined as g 1 g P x Z E exp E fr a gt Tit 0 otherwise a a a where where m location parameter mu a scale parameter alpha g shape parameter gamma The cumulative distribution function is defined as _ g F z 1 exp k a The input command is given by Weibull m 0 4a 1 g2 n where the optional n is used to normalize the distribution 9 3 2 Plot commands QuadShellList PLAT PlotAttributes ThickShells To specify attributes in a PlotList file PlotList PLOT 4 qsL SHL1 thk thick plates qsL SHL2 4 ThinShells vel RangeMaxValue 1000 in s qsL SHL3 thn vel vmn 0 in s vmx 1000 in s 95 10 Tetrahedral Solid List The TetSolidList includes a series of lists which implement collections of tetrahedral shaped elements Some of the types of tetrahedr
191. nvention 067 012 2 2 31 0 3 0 6 gt 1 El E 0 0 d d d ono NNN 10 4 Make Commands The Make TetSolidList is used for generating a sub list of all tet elements which have been previosly selected The new list is of the Geometric type in other word MARS will not try to compute internal forces The Make TriangFaceList is used for generating a list of all triangular external faces of the complete mesh If the user desires to generate external faces of a portion of the mesh then the Make TetSolidList command should be done first and the face list should be generated from inside the tet sub list The Make EdgeList command is used to generate a list of edges from the edges of the tetrahedral elements This command must be terminated by a keyword for the edge selection criterion three options are available Make EdgeList ListName AllEdges Make EdgeList ListName SurfaceEdges Make EdgeList ListName SharpEdges The option AllEdges is used to generate a list of all internal and external edges of the tet mesh The other two options rely on the list of external tet faces generated using the Make TriangFaceList command If the triang face list was not generated MARS will automatically generate it The option SurfaceEdges is used to generate all external edges of the tet mesh This list is useful for edge edge contacts The option SharpEdges is used for generating a list of edges for which the attached fa
192. o depict an area limited by an upper and a lower curve Two types of bands are available 1 a solid band where the area is filled with a lighteray color 2 a striped band consisting of vertical lines connecting lower and upper curves at the given data points The arguments of the command are xr DS1 1 means use record number 1 from the DS1 dataset for the x axis value ybr DS1 2 means use record number 2 for the lower y value ytr DS1 3 means use record number 3 for upper y values fill means paint the band using a solid color rather than vertical lines Line 6 This command is similar to the command in line 5 but represents the band using vertical lines in black color Line 8 The Curve command is used draw lines or symbols to visualize time history data or experimental data points The arguments of the command are xr DS2 1 means use record number 1 from the DS2 dataset for the x axis yr DS2 3 means use record number 3 from the DS2 dataset for the y axis cl green means paint the line in green the following colors are available black blue red green cyan yellow magenta orange gray and lightgray the default color is black Line 11 In this line the command xs 0 1 means that the x coordinates are multiplied scale by the 0 1 factor Three other similar commands are available for both Band and Curve commands xo offset ys scale and yo offset These are used to scale and offset the original records using the equation new_value scal
193. o plastic Material Model Material PLST plastic Description Stainless Steel Density 7 8 g cm3 YoungsModulus 29e6 psi PoissonsRatio 0 3 YieldStress 80e3 psi HardeningModulus 200e3 psi select one of the three hardening options below IsotropicHardening KinematicHardening 38 Beta 0 5 beta 0 istropic beta 1 kinematic Di 100 1 s strain rate for intermediate point sy1 90e3 psi yield stress for intermediate point syi 100e3 psi yield stress for infinite strain rate 5 9 4 Johnson Cook Model The Johnson Cook model is purely empirical and gives the following relation for the flow stress oy p Ep T A Blep 1 Cin amp a T9 where g is the flow stress p is the equivalent plastic strain A B C n and m are material constants es is the normalized strain rate T is the normalized temperature The normalized strain rate and temperatures in the equation above are defined as T To gt and PA AAA Ep0 Tm To The input commands for specifying the material parameters are Material PLST JohnsonCook Description Stainless Steel Density 7 8 g cm3 YoungsModulus 29e6 psi PoissonsRatio 0 3 A 80e3 psi B 80e3 psi C 4 n 4 m 4 ReferencePlasticStrainRate 0 0001 1 s ReferenceTemperature 40 degC This material model is available fos solids shells beams and trusses The following materials have built in properties which were found in the literature Material P
194. odel in their current location This is accomplished by the command CellInitialOutline ttL ListName CellInitialOutline 11 6 4 Plotting particles This feature provides the capability of plotting particles as spheres It is no different than the plotting option available for standard NodeLists ttL ListName Particles This option is available for both Quasar and Paraview 11 6 5 Plotting embedded fibers If the component contains embedded fibers these can be plotted using the command Fibers ttL ListName Fibers This option is available for both Quasar and Paraview 11 6 6 Contour plotting of facet variables This plotting feature provides the ability to display state variable information at the facet level As for the Cells plot the total number of facets can be very large For this reason there are options for reducing the number of facets included in the facet list which are eventually plotted The variable to be plotted is selected using the FacetVariable keyword followed either by the variable index or the variable label These can be chosen from the list of state variables for the material being used The first level of filtering is obtained using the command MinimumCrackOpening which works in the same fashion as in the Cells plot In addition if a variable range is specified using either or both RangeMinValue and RangeMaxValue then the facets for which the variable is outside the range are remove
195. ollowing are some examples of scripts that can be performed in a function The first script converts a flat plate into a cylindrical shell Function FNC1 4 silent do not display message when executing function int nnd real cx real cy real cz real dangle dangle 3 14159 24 real angle real cs real sn nnd ndL PLAT num int i do i 1 nnd cz nd PLAT i cx cy nd PLAT i cy angle cz dangle print angle debug printing cs cos angle sn sin angle cx cs cy cy sn cy cx 1 Dex cy 1 cy print cx print cy nd PLAT i cx cx nd PLAT i cy cy nd PLAT i cz cz exec FNC1 now The second example moves nodes that are outside of a cylinder of radius RO to the surface of the cylinder Function TRIM 45 int nnd real real real real real nnd RO prin int logi do i CX cy tm rd rd rd lg if exec T fu real real lt S lt X lt XS lt S lt cx cy tm rd RO ndL TPOL num OT ley 2 t nnd i cal 1g 1 nnd nd TPOL i cx nd TPOL i cy cx cx cy cy rd tm sqrt rd rd gt RO lg I print rd print RO tm RO rd cx cx tm cy cy tm nd TPOL i cx cx nd TPOL i cy cy RIM now nctions observe spaces x y random random number between O and 1 sin x cos x abs x sqrt x The third example checks the maximum Von Mises stress in a list
196. omplete mesh If the user desires to generate external faces of a portion of the mesh then the Make HexSolidList command should be done first and the face list should be generated from inside the hex sub list 3 The Make TrianFaceList is used for generating a list of triangular surfaces by splitting the quadrilateral faces of the list generated using the Make QuadFaceList command If the quad face list was not previously generated MARS will automatically generate it The Make EdgeList command is used to generate a list of edges from the edges of the hexahedral elements This command must be terminated by a keyword for the edge selection criterion three options are available Make EdgeList ListName AllEdges Make EdgeList ListName SurfaceEdges Make EdgeList ListName SharpEdges The option AllEdges is used to generate a list of all internal and external edges of the hex mesh The other two options rely on the list of external quad faces generated using the Make QuadFaceList command If the quad face list was not generated MARS will automatically generate it The option SurfaceEdges is used to generate all external edges of the hex mesh This list is useful for edge edge contacts The option SharpEdges is used to generate a list of edges for which the attached faces form an angle greater than 126 60 degrees This list is useful for graphics when we want to represent the outline of a solid comp
197. onent 4 The option RemoveUnselectedElements is used to permanently remove unselected elements from the list This operation is done during the reading phase It should be done before Make QuadFaceList or Make EdgeList for these derived list to be consistent with the reduced hex list 12 4 Time History Commands The following line commands are intended to be used inside TimeHistoryList s to produce records of global list variables or variables associated to a single element Most of the variables are meaningful only for deformable hex elements For element formulations where the requested variables are not available the record will consists of zero values TimeHistoryList HIST histories for entire list hxL PART Volume 11 hxL PART InternalWork E1 hxL PART DissipatedEnergy 1 hxL PART ElasticEnergy 1 hxL PART HourglassWork 1 hxL PART MaxVonMisesStress 2 hxL PART MaxStateVariable 7 2 hxL PART MinStateVariable 4 2 histories for single element hx PART 1 vl volume of element 1 hx PART 1 vm Von Mises stress hx PART 1 pr pressure hx PART 1 11 first invariant hx PART 1 J2 second invariant hx PART 1 sv 3 state variable 3 hx PART 1 xCG element x coord of center of gravity hx PART 1 yCG element y coord of center of gravity hx PART 1 zCG element z coord of center of gravity 1 These variable are computed taking the integral of a quantity over al
198. or a specific contact is negative an offset value gO equal to the absolute value of the gap is applied for the rest of the simulation obviously to that contact alone gap dst R1 R2 gO g0 max gap 0 at time O 6 Update intervals are now specified in the ControlParameters section using the GlobalUpdateTimeInterval command 16 3 1 Time History Commands PlotList ListName ecL ListName NumberOfContacts ecL ListName ActiveContacts ecL ListName MinimumDistance ecL ListName MinimumGap ecL ListName NormalForce ecL ListName X Force ecL ListName Y Force ecL ListName Z Force The difference between MaximumPenetration and CurrentMaxPenetration 16 4 Face Contact List The face contact makes it possible to detect contacts between a list of nodes particles and a list of triangular faces For the purpose of contact a node particles takes a spherical shape and a triangular face takes a three dimensional smooth shape consisting of a thick plate surrouned by semi cylindrical edges and spherical corners reminding the shape of old Vicks cough drops The spherical surfaces are centered at the three nodes defining the face The radius of the edge cylinders and corner spherical surfaces is half the thickness of the plate The thickness of the plate is either provided by the face lists or defined via input Contact between a node and a face begins when the exernal surfaces of the associated
199. or computing what portion of a partially embedded fiber element is inside the concrete If fibers are entirely embedded in concrete this current formulation is adequate For single fiber pull out tests subdivide a fiber so that there are no elements that are partially embedded in other words a node of the fiber sequence should just be slightly outside the concrete surface 15 9 Bolt List BoltList BLTS abbreviated notations in lt gt brackets Material STEL Diameter 0 3 in StemLength 1 0 in BeamInStem 4 HeadThickness 0 2 in HeadDiameter 0 6 in if bolts are entered explicitly num 345 LengthUnits in lst cOx cOy cOz origin of the stem len length of the stem dcx dcy dcz direction of the stem n number of beam elements along the stem j 760 c0y c0z len dcx dcy dcz n 1 0 0 10 1 0 6 1 O O 3 146 16 Contacts 16 1 Contact Models Contact models are employed by the contact formulations e g node face node node edge edge etc to compute contact forces based on the relative velocity of the material points associated to the two objects at the point of contact Contact models have the same function as material models have in element formulations Contact model commands are embedded in the contact lists see examples 16 1 1 Penalty contact model The penalty function contact model provides the simplest method for computing the normal contact force Fn as a function of penetration p The me
200. or example for faster speed you may want to perform contact updates more frequently as well as plot dumps while reducing the overall simulation time This would require the creation of 181 multiple input files for each speed case Furthermore if the model needs to be modified the modifications have to be applied to all files This is a case where the conditional capabilities of the preprocessor come handy Look at the input file below below Title line if VO 100 define TEND 2 define DT 0 02 elif VO 200 define TEND 1 define DT 0 01 else error Velocity not accepted endif ControlParameters TerminationTime TEND ms WritePlotDataFile data plt Every DT ms NodeList 4 Set vz const VO m s ifdef PLT PlotList Cells Paraview TimeInterval DT s ttL Tile Cells ProcessPlotData data plt endif EOF The mars input files for the two speed cases can be generated using the commands CC DVO 100 input mrs gt input2 mrs or CC DV0 200 input mrs gt input2 mrs 21 5 import Ingrid meshes The INGRID program can be used to generate finite element meshes of the complete model or parts of the model Very often it is preferable to mesh individual components 182 separately and combine them later in the MARS input file In these cases the interaction between components such as bonding or contact can be defined very conveniently in the MARS input INGRID generates an output file named ingrido unl
201. ote that the first three lines are used for reference information that can be printed when displaying the image The words title steps and time are keywords title DPM grid and associated tet mesh nsteps 0 time 0 000000 tfL NAME front lightgray back null numpt 995 crd 1 1 200 0 000 6 800 2 1 200 0 000 7 100 3 1 200 0 000 7 400 4 1 200 0 000 7 700 2548 1 200 6 000 7 800 2549 1 200 6 000 8 000 numtf 1988 list 1 1 151 147 2 148 1 147 3 2 1 1987 995 761 835 1988 761 995 850 eoL eel NAME color black 224 numpt 207 crd 1 1 200 0 000 6 800 2 1 200 0 000 7 100 2549 1 200 6 000 8 000 numee 218 list 1 2 1 2 1 217 205 206 218 206 207 eoL npL GRID color lightgray numnp 1064 crd 1 0 506 2 563 7 720 0 200 2 0 055 1 495 7 545 0 200 3 0 291 4 280 5 850 0 200 4 0 827 1 526 7 474 0 200 1063 0 059 4 546 4 857 0 100 1064 0 375 3 228 5 850 0 100 eoL EOF 25 2 jHist a java post processor for Mars time history files jHist is a rather simple java program for quickly displaying time histories generated by Mars using the TimeHistoryList commands jHist defaults the x axis to the first record in the time history file which is typically the time in milliseconds and the y axis to the second record After the initial plot is displayed the records for the x and y axis can easily be changed to any other record in the file using the scroll down menu options It is possible to display multiple curves simultaneously
202. ovide a minimum edge radius for cal culating the gap between two edges The distance dst between two edges is defined as the shortest distance between any point on edge 1 and any point on edge 2 The gap is defined as gap dst R1 R2 where R1 is the radius of edge 1 and R2 is the radius of edge 2 R1 is the maximum of the actual edge radius r1 and the edge thickness T1 Ri max ri T1 The edge radius r1 is typically defined by the properties of edge list 1 Similar equations hold for R2 If the radii have been specified in the edge lists the Edge1 2Thickness command can be omitted At least one of the two lists must have a postive radius 2 The detection distance Dd for edge edge contacts is used for selecting edge edge pairs which are close enough that may come into contact In this version an edge edge pair is selected when dst Dd Ri R2 using the definitions for dst R1 and R2 from 1 4 The WritePlotFile command is used to generate a Quasar plot file that visualize the detected edge contacts at time 0 The plot contains only the edges that are used 157 in the contacts using the plot attributes color and radii defined in the respective edge lists The files is named using the convention ecL listName plt 5 Contact forces are generated when the two edges penetrate each other This occurs when the gap becomes negative To avoid contact forces at time 0 if the initial gap f
203. ow 33 5 7 1 Attached to quadrilateral surfaces Accelerometer AGO1 QuadFace 4 FaceList DuterPlate Description Accelerometer AG 01 CenterAt 0 in 10 in 9 in Tolerance 0 01 in Direction 0 0 1 5 7 2 Attached to triangular surfaces Accelerometer AGO1 TriangFace FaceList OuterPlate Description Accelerometer AG 01 CenterAt 0 in 10 in 9 in Tolerance 0 01 in Direction 0 0 1 5 7 3 Attached to the surfaces of a beam element Accelerometer AGO1 Beam BeamList OuterPlate Description Accelerometer AG 01 CenterAt 0 in 10 in 9 in Tolerance 0 01 in Direction 0 0 1 5 7 4 TimeHistories The accelerometer computes a single variable the acceleration component at the pre scriped point in the prescribed corotational direction The only purpose of an Accelerometer is to provide acceleration for time history lists Typical application show up as Accelerometer Gagel QuadFace TimeHistoryList HIST ac Gagel 34 5 8 Equations This is a collection of objects that implement global equations such as pressure volume equations etc 5 8 1 Pressurized Volume Equation name PressurizedVolume InitialVolume 34 53 cm3 InjectionHistory name2 Enter the name of fluid or BulkModulus Oil Water Air Temperature 70 degF 5 8 2 Injected Fluid Use this equation when Work still in progress LoadCurve name2 4 Specify flow rate history Equation name InjectedFluid InitialVolume 34
204. ow TimeHistoryList 1c curveName 5 2 1 Load Curves Lists It is possible to group a set of load curves that share similar characteristics in a list of load curves see specific topic in the Misc section of this manul This is the case for reading pressure histories generated by a fluid code 5 3 Built In Blast Loads The built in blast loads consists of a series of objects that are used in face lists triangular and quadrilateral faces for generating pressure loads Typically these loads are used for approximating the effects of blast Pressures are computed taking into account the location of the explosion and the position and orientation of the exposed faces 2r 5 3 1 ConWep ConWep is a collection of conventional weapons effects calculations from the equations and curves of TM 5 855 1 Design and Analysis of Hardened Structures to Conventional Weapons Effects ConWep performs a variety of conventional weapons effects calcula tions including an assortment of airblast routines fragment and projectile penetrations breach cratering and ground shock shock The equations used in this special load object are based on a report by Kingery C N and Bulmash G Airblast Parameters from TNT Spherical Air Burst and Hemispherical Surface Burst Technical Report ARBRL TR 02555 U S Army ARDC BRL Aberdeen Proving Ground MD April 1984 and other reports as follows 1 The following methods are based on the equations f
205. page length OMAN OAOTBRWNH KE 42 10 Max long side slippage length 11 Spalling length length 12 Cook Gordon crack opening length 13 Failure flag nondimensional 14 Dissipated energy energy 15 Normal crack opening length 16 Shear M crack opening length 17 Shear L crack opening length 5 9 8 Simple Cap Concrete Material Model Material PLST SimpleCap Description Simple concrete model Density 2 0 g cm3 YoungsModulus 4e6 psi PoissonsRatio 0 3 BulkModulus 0 3 ShearModulus 0 3 Fe alpha theta I failure curve Alpha 0 07 Theata 0 07 XO 3000 psi initial cap location W 0 07 void ratio select one of the blocks below 1 epv W epx D1 X X0 Di 0 001 make sure it is givne in 1 stress 2 epv W epx D2 X X0 72 D2 0 001 make sure it is given in 1 stress 2 3 epv W I X0 X1 X0 X1 40e3 psi lock up 5 9 9 K amp C Concrete Material Model The K amp C material model was inserted in Mars in 2003 and it has not been updated since In its simplest input form only three parameters are necessary to specify material properties Material Concrete KcConcreteModel Density 2 0 g cm3 PoissonsRatio 0 19 Compressivestrength 6000 psi All other parameters and curves are automatically generated If the user wants to refine material parameters to match specific data the following addtional parameters can be specified 43 LoadCurve Cur
206. perform the check at every check although this frequency is most likely execessive 17 0 4 Node Hex Overlap Check This feature is designed to detect whether any node from a list crosses any face from a triangular list The main purpose of this is to ensure that the contact algorithms is able to prevent contacting nodes from crossing through a plate or shell part Checks 4 NodeFaceCrossingCheck CheckName NodeList ListName TrianFaceList ListName FaceContactList ListName CheckTimeInterval 0 1 ms Disable Make the CheckTimelnterval smaller than the time step to perform the check at every check although this frequency is most likely execessive 163 18 Pre and Post Processing 18 1 Plot Lists The input section for plot lists is typically located at the end of the input stream before the MPI section and at the same level of the time history sections Multiple plot lists can be defined Each plot lists generates a sequence of plot files at specified time intervals Currently two plot formats are available Quasar and Paraview There are two ways for selecting plot format The first way is to specify it inside the PlotList section PlotList listName 4 Paraview or PlotList listName 4 Quasar If either keyword is omitted MARS automatically assumes the plot format is Quasar Since July 17 2011 a second way of specifying plot format is available PlotList listName Pa
207. plete fragmentation right Fig 2d shows the formation of shear bands at failure for a specimen subjected to uniaxial compression Fig 2e presents the LDPM model damage distribution and crater formation for a simulation of projectile penetration through a reinforced concrete slab The LDPM framework can also handle coupling with steel reinforcement The concrete rebar bond model is obtained through a penalty formulation that allows the simulation of nonlinear bond slipping Fig 2f shows the simulation of a dynamic pull out test leading to splitting failure Finally Fig 2g shows the LDPM capability of handling complex three dimensional fracture paths with multiple branching and crack coalescence Recently LDPM was extended to include the effect of fiber reinforcing The updated formulation named LDPM F incorporates the effect of fibers by modeling individual fibers placed within the LDPM framework according to a given fiber volume fraction The number and orientation of fibers crossing each facet is computed and the contri bution of each fiber to the facet response is formulated on the basis of a previously established micromechanical model for fiber matrix interaction Currently LDPM F is being applied to the simulation of the CORTUF concrete developed by ERDC Preliminary results clearly demonstrate the ability of LDPM to model the micromechanical phenomena characterizing CORTUF failure 11 1 Input commands TetSolidList listname LD
208. point toward 0 5 4 cp 1 0 0 gt 0 5 faces such that cross product with 1 0 0 is greater than 0 5 Four node face lists and all derived lists include commands for selecting nodes SelectNodes UnselectNodes and ReselectNodes See examples in the NodeList section 9 1 2 Generate commands The Generate sub block is used for generating simple meshes of common geometries from within Mars Multiple geometies can be generated and combined within the same block The syntax for an input block is shown below The specific inputs for each part are discussed in the next sections Generate enter one or more geometry feature Cylinder Ll 4 Disk Single MergeParts merge overlapping nodes 1 1 The MergePars command makes it possible to fuse two or more parts at the common nodes This can be useful when generating more complex meshes such as the mesh of a hollow cylindrical can with caps at both sides Attention must be paid that the nodes on the edges overlap Cylinder These commands generate a hollow cylinder Cylinder ReferenceSystem refSysName 1 EdgesOnCircle 36 2 Radius 4 in Length 10 in ElementSize 1 in in axial direction 3 1 This command employs a reference system which was previosly defined outside of the QuadFaceList section The axis of the cylinder is aligned with the local z direction and spans from z Otoz Length If the reference system is not specified the
209. possible for users to develop custom objects element formulations and lists A user can derive new classes from existing ones and then modify them by inserting new features For example a Cosserat 10 node tetrahedral element was recently developed by a student starting from the already available 10 node tetrahedral element The interface is very flexible and allows to incorporate any number of new objects material models special load histories etc and lists lists typically implement element formulations interaction mechanisms etc MARS has been successfully adopted for the simulation of a wide variety of problems which include e Modeling of cracking and failure of cement based materials materials e Response of reinforced concrete structures to blast loads and fragment impacts Cable dynamics problems e Fragmentation of ordinance casings Laceration of plates and shells 1 1 The Lattice Discrete Particle Model LDPM The Lattice Discrete Particle Model LDPM is a discrete meso mechanical model for concrete which was recently developed by Dr Cusatis and co workers at Rensselaer in collaboration with Dr Pelessone at ES3 LDPM simulates the mesostructure of concrete by a three dimensional assemblage of discrete particles whose position within the volume of interest is generated randomly according to the given aggregate size distribution A whole section of this report is dedicated to LDPM LDPM has been extensively calibrated
210. quest you need to configure the system so that it can find Mars and MPI supporting software Enter the bash shell gt bash Edit the bashre file in your main folder vi bashrc Insert the following line export PATH hpc home dpw466 Mars PATH source bashrc Any subsequent time you log in start the bash shell and the path is automatically updated gt bash Interactive Execution To run the OpenMP version of Mars enter marsO input mrs To get the latest version of the built in LaTex manual enter marsO D1 Batch Execution MPI Load the MPI environment module load mpi openmpi 1 4 3 gnu Create a script file e g job sh to control batch execution bin bash MOAB 1 nodes 4 ppn 8 MOAB 1 walltime 0 59 00 MOAB j oe MSUB A p20288 ulimit s unlimited cd yourProblemFolder mpirun marsM B input mrs gt output Then submit it using the sbatch command msub job sh Check the status using the command showq grep x where x is your username 212 23 4 Mars on hpc diamond Diamond contains 1 920 compute nodes 15 360 compute cores Each compute node contains two 2 8 GHz Intel Xeon 64 bit quad core Nehalem processors and 24 GBytes of dedicated memory The nodes are connected to each other in a HyperCube topology DDR 4X InfiniBand network Pre Processing The Intel c compiler icc is available on diamond and this is the software that should be use for pre processing a Mar
211. r distributions and compare them visually against available data The GenerateFibers command creates a new edge list automatically named Fibers For this reason it is important to avoid naming other edge lists with the word Fibers This list can be later operated on using standrard procedures For example TetSolidList ListName LDPM EmbeddedFibers GenerateFibers EdgeList Fibers Rename SteelFibers Write PartMeshDataFile fibers mrs Color red 2 The Prism option was added in June 2012 to address some performance issues for models with large number in the millions of fibers It is not a new parameter rather 121 it is a flag to activate a new generation method The default method uses the following procedure e Fibers are generated during the Reading phase and saved to an EdgeList named Fibers e The generation method employs the tetrahedral mesh for ensuring that the fiber is entirely or partially contained in the concrete e For chopped fibers the method finds the intersection of edges against the external facets of the tetrahedral mesh e In the initialization phase the fiber EdgeList is used for computing the intersection of the fibers themselves with the Ldpm facets e Fibers are not used as structural elements during the simulation The procedure above works well for medium size fiber systems less than a million When the number of fibers is in the millions the procedure is very slow
212. r example the work perfomed by the internal forces This quantity can be updated at each time step in the calcFrc method and must be saved either as an additional state variable or as a new member of the class definition If it is saved as an additional state variable then it is easily accessed through the standard commands Let s assume that we use a new member named t wrk which must explicitly appear in the class definition class Tet_UD public Tet private real wrk new member The logical command in the input file for selecting the new quantity may look like this TimeHistoryList Hist tt listName elementIndex InternalWork where InternalWork is a keyword chosen to select the new member wrk We also want to maintain the functionality already provided by the parent class The way to accomplish this objective is to redefine the parseVariableIndex method so that is assigns an 201 index of 1001 when it encounters the label InternalWork Otherwise it uses the method of the parent class A large index is recommended so that it does not overlaps with smaller indices used in the parent class int ttL_UD parseVariableIndex string 1b1 Reader rdr if 1b1 InternalWork return 1001 else ttLst parseVariablelndex 1b1 rdr Similarly the user should redefine the method for generating the variable label in the time history file string ttL_UD makeVariableLabel int elt int var if elt
213. r that spans two domains one owned by the rank 1 process and the other owned by the rank 2 process as shown schematically below Rank 1 Rank 2 8 B 9 Beam B is owned by rank 1 its nodes 8 and 9 are owned by rank 1 and rank 2 respec tively The internal forces computed for beam B at node 9 in the calcFrc method must eventually be transferred to rank 2 that ownes node 9 Currently this is done in the integrateEOM method Assume that the beam nodes are embedded in a tetrahedral mesh using master slave constraints The force reduction where forces at beam nodes are distributed to tet nodes is done in the reduceFrc method which is placed between the calcFrc and integrateE0M methods With the current logic node 9 has not yet received the force contribution from beam B Similar problems may occur when synchronizing the velocities For this scenario let s assume that the constraint that ties node 9 and a tet elements is owned by the rank 1 process The information for the tet nodes is synchronized in the integrateE0M Thus the velocities at the beam nodes are computed correctly However the velocities 172 for node 9 computed by rank 1 in the applyKin method are not shared with the other processes Indeed after the applyKin task rank 1 should communicate the velocities of node 9 to rank 2 that is the owner of node 9 and rank 2 should communicate the same info to all other ranks that use node 9 We are currently ev
214. raview or PlotList listName Quasar The typical plot list input features control commands and list specific commands The first group includes four commands PlotList listName plotType TimeInterval 0 1 ms also dt 0 1 ms 1 DeleteAllFiles in this family PLOT 2 Counter 44 reset file name counter to 44 3 NextTime 0 344 ms reset next plot write time 3 1 The TimeInterval command is used to specify the time interval used for cre ating the family of plot files Since Jul 17 2011 this command may be omitted and a default TimeInterval can be specified in the PlottingDefault subsection of the ControlParameters section For Quasar files different plot lists can use different time 164 intervals For Paraview files graphical components are saved in different plot lists as such the time interval must be the same and the default time interval command is a better way for specifying the time interval 2 The DeleteAl1Files command is used to remove all plot files previuosly gen erated For Quasar plot lists Mars executes the command rm listName For Paraview plot lists Mars executes the system commands rm listName vtu and rm listName x pvd 3 These two commands Counter and NextTime are rarely used They may be useful in restart operations The rest of the input specifies the list s to be plotted and specific plotting attributes More details are ava
215. rd nRec value nRec 1 first value of record nRec value nRec 2 second value of record nRec value nRec npt npt th value of record nRec 229
216. re SpecialLoad BRST ConWep x0 1 in 20 in O in weight 100 1b hemispherical offset 0 012 ms The offset parameter is used to offset the time of blast so that the shock wave reaches the exposed surfaces at time zero of the numerical simulation To compute the optimal offset time execute the first calculation with an offset of 0 Then look at the output and it will tell you when the shock wave reaches the surface Use that value to offset the blast This operation is done manually because if multiple explosions had to be performed in the same simulation the user must have the ability to control the relative offset 5 4 Random Number Distributions The random number generation object generates random numbers using one of the com mon distributions equations Currently four distributions are available Random number generators are used for the definition of some inputs In most cases the user can select which distribution to use since they are interchangable 5 4 1 Weibull distribution The Weibull distribution has been used very effectivley for characterizing probabilistic failure in materials and mechanical components The probability density function is defined as 5 9 1 a g P x 2 E exp E fr ES 0 otherwise a a where where m location parameter mu a scale parameter alpha g shape parameter gamma The cumulative distribution function is defined as g F x 1 exp E a The
217. re Histories a a es bbs 9 3 Quad Shell Lists 3 4 ee ae bw oe Ae Bee eee ee ke 9 3 1 Weibull distribution o 9 3 2 Plot commands 0 00 eee eee 10 Tetrahedral Solid List E Ss tases A ict sty ee ad Wats ao eats oe at ae ee 10 2 Select Commands 000000000 a 10 3 Generate Commands 00000000000 eee 10 4 Make Commands 0 000000002 ee ee eae oe oda ea Rae oe 10 6 Plot Attribute Commands 0 0 0 0 0 0 0 0 000084 T07 Viscous lets s sor ca ae Be Bea ER RA ERR A ee DES eRe EMME EE eee Oe oe ee eee 10 9 Ten Node Tet Elements e 10 9 1 Ten Node Small Deformation Element 10 9 2 Ten Node Large Deformation Element 11 Lattice Discrete Particle Model ge Be eae by CR a ee ee ae ees eine ty ines eee ee Me oe a amp ee a ko ees Soe Bee eee Gace Gee sey ae ae eae eee ee MEP A ee eR ER ES RS ES a PTA ARI a a da a ee T 11 3 Time Histories ele di a we wee ee ba ae She oe ae ae ea oe Bes ees en ee bog He Gute acta te ed hat oe ae antes eG de Bes wits ewe hae ee Nee oe dae ee eee eee as be ae bee eee hae eee ee 11 6 2 Plotting external facets only 12 22 eh pm e ee es ee oe A ee ae ee ee Cue se ee a ee ae ee ee La eo a ee ee ea Na a e ey a ria He a EU a a OA sarees denia ae e 11 6 10 Other examples loba ra a Oe o Pee REE 11 7 Embedded Fibers eo oe back oad he RR SR A Re OR 12 Hexahedral Solid Elements 12 1 Select Commands a
218. read an input file which contains regular input commands The input file may be used to add parts to the model make modifications to the model write output files etc Read FileName atTime 10 ms Read FileName atStep 10000 4 2 7 Write Plot DataFile These commands are used to schedule when to write database snapshots to the plot data file for later post processing WritePlotDatFile data plt Every 0 1 ms More detail on this procedure are given in the HowTo section 4 2 8 ExecImplicitSolver Either command is used for executing the implicit solver at step 0 or at time 0 seconds ExecImplicitSolver solverName atStep 0 ExecImplicitSolver solverName atTime 0 s 4 3 Time Step Control The time step used in the MARS explicit time integration scheme can be controlled for both stability and accuracy and is determined by the procedure described in this section First MARS computes a Courants time step for each list This time step is based on the smallest time step for each element in a list When multiple mechanisms come into play simultaneously such as contact conditions penalty constraints etc the added stiffness may require for the calculated time step to be further reduced In general it is good practice to reduce the time step when a simulation goes unstable Very often this is caused by overlapping effects If this action does not solve the instability then the instability may be caused by some other rea
219. rgy per list like the one shown below when MARS is interactive mode by typing M and w at the prompt mdl gt M n number of elements m mpi decompositions w work balance x node lists 175 t min time steps gt W List ExtWrk IntWrk KinEnr nd PRTC 0 119 tt PRTC 0 736 nd EFEP 0 006 hx EFEP 0 083 nd EFEN 0 006 hx EFEN 0 076 nd LOAP 0 059 hx LOAP 0 013 pv VEL1 1 098 tb BND1 0 000 tb BND2 0 000 Totals 1 098 0 908 0 190 Balance EW IW KE 0 0004125 Work and energy values are times e 1 J Example This example consisting of a spring SSSS5 fixed at the left and with a mass M attached at the right see schematic below is intended to illustrate the concepts above in a simple framework SSSSSS M_ lt F y A constant force F is applied at time t 0 The closed form solution to this problem is is x t A 1 cos wt v t V sin wt where w sqrt M K is the frequency K is the stiffness of the spring x t is the coor dinate of the mass x t 0 0 A F K is the amplitude of the oscillations v t is the velocity of the mass V A w is the amplitude of the velocity Thespring internal force Fi is computed as F t Ka i AK 1 cos wt F 1 cos wt The external work We internal work Wi and kinetic energy become become We t F x t F x t F A 1 cos wt F 2 K 1 cos wt
220. rom BRL Technical Report ARBRL TR 02555 for predicting positive phase parameters for either a hemispherical surface burst or spherical free air burst bool equationsAreNotValid real peakIncidentPressure psi real incidentImpulse psi ms real normallyReflectedPressure psi real normallyReflectedImpulse psi ms real timeOfArrival time of arrival of blast wave ms real duration time positive phase is done ms real shockFrontVelocity ft ms real rfp real p char tp range for given charge and pres ft real rfi real i char ti range for given charge and imp ft real rft real t char tt range for given charge and time ft 2 The following methods based on the modified Friedlander wave form are used to determine shape factor pressure history and dynamic pressure history for the positive phase real waveShapeFactor real imp real pmx real dt real pressureAtTime real tau real pmx real wsf psi real dynPrssAtTime real tau real q real wsf psi 3 Use the AFWL 70 127 curves to determine the following real reflectedPressureCoefficient real alp real pso 4 Use the TM 5 1300 procedure Figure 2 192 to find the following real reflectedImpulseAtAnAngle real alp psi ms 5 Use formula presented by W E Baker to determine the following real peakDynamicPressure real pso psi 28 The input commands for specifying this special load a
221. rs Note that the mesh part file includes the edge definition as well as the node data Sections 1 and 2 of the edge list input should not be present in the main input file file 7 1 5 Linear Elastic Beams Uniform Cross Section This list consists of a collection of linear elastic beams with uniform cross section The beam formulation is based on the Euler Bernoulli beam theory A beam element is defined using two nodes Each beam element has a corotational local reference system XYZ The X direction is aligned with the beam axis from node 1 to node 2 The second Y direction is perpendicular to X and is initialized at time zero using a ReferenceSystem object previously defined The section properties are defined in the local Y Z plane perpendicular to the beam axis The sections properties consists of the following data e cross section area A e moment of inertia about the Y axis IY e moment of inertia about the Z axis IZ and e torsion constant J The torsion constant of the section is not to be confused with the polar moment of inertia The two are identical for round shafts and concentric tubes only For other shapes J must be determined by other means The section properties are internally calculated for a few common cross sections see sample input listing below or explicitely entered EdgeList listName LinearBeam Material materialName Select one of the available cross sections RectangularCs by 3 in bz 0 5
222. ructural entities can be derived from the edge class trusses rebars beams links etc Edge lists can be created internally from other lists or input explicitly When created from other lists they are typically used in conjunction with more complex meshes consisting of triangular and quadrilateral faces or tet and hex solids Input commands for defining new lists explicitly via input are given below below EdgeList EdgeListName type where type can be one of the following Geometry Truss Rebar EdgeList Edges with no specifier defaults to Geometry 1 Specify node list NodeList NodeListName or InsertNodeList NodeListName 2 xx Read or generate edges ReadObjects 345 ff i m m2 1 124 243 2 56 238 else Generate see below for generation options 60 3 Change attributes optional Color Green Radius 0 4 mm 11 Diameter 0 8 mm 1 Density 7 8 g cm3 2 PlotAttributes 4 Cylinders 4 Other optional commands EditNodeList Make EdgeList SubListName sublist of selected edges Make NodeList SubListName nodes attached to selected edges select unselect alsoselect commands see below writing and reading external files Write PartMeshDataFile PART mrs 3 Read PART mrs 3 1 The radius or diameter are used for three main purposes a plot edges as three dimensional cyli
223. s This subsection explains how to create methods for extracting variables and printing them to a time history list for user defined lists and elements within a user defined list There are essentially five polymorphic methods that are used to accomplish this task int List parseVariableIndex string lbl Reader rdr Object List parseObject Reader rdr 200 string List makeVariableLabel int elt int var real List getValue int var real Object getValue int var The first two methods are used for reading the data from the input file The last three methods are used for providing data labels and values to the methods that write to the time history file In the great majority of cases the user wants to modify an existing class and implement a specific formulation For this reason we will explain the procedures using an example of a derived class Let s consider a new formulation for a tetrahedral element This would rely on two classes one for the tetrahedral element object and the other one for the list of the new tets These two classes are derived from the parent Tet class for inheriting the geomet rical properties and methods of a tetrahedral element and ttLst class for inheriting the methods of a tet list class Tet_UD public Tet class ttL_UD public ttLst 4 F Let s consider the case where we want to generate the time history of a parameter not currently available in the element formulation fo
224. s For b 2 the unloading curve is a parabolic curve with apex at the origin and intersecting point pX FX The area between the loading and unloading curve represents the energy dissipated in the loading unloading event The larger the value of b the more energy is dissipated Reloading during the unloading phase is ruled by a special relationship Reloading follows a steeper line with the stiffness constant defined as the tangent to the unloading curve at pX Fn Kr p pY where pY is the penetration when reloading begins begins Kr b C pX b 1 is the stiffness The normal force Fn cannot exceed Fn K p value in other words the initial loading curve acts as an upper envelope The input commands are either ContactForce PenaltyHyst Kl 1e6 N m Beta 3 or ContactForce PenaltyHyst LinearSpring 1 e6 N m Beta 3 148 16 1 3 Hertz contact model The Herz contact force computes the normal contact force Fn as a function of penetration using the well known equation Fn H p 1 5 The constant H is typically defined as a function of Young s modulus E and Poisson s ratio pr H 2 3 E 1 pr72 The input commands can be enter either in this format ContactModel Hertz E 29e6 psi pr 0 3 or ContactModel Hertz YoungsModulus 29e6 psi PoissonsRatio 0 3 ContactForce HertzTU 16 1 4 Hertz hysteresis contact model The Hertz with hysteresis contact model provides the capability of modeling non
225. s as well you must re enter Load particle file Apply Main Menu Filters gt Recent gt Gliph Set Gliph parameters in the Properties tab Scalars Rotations Vectos RotationVectors GlyphType Sphere Set node radii or scale particle radii as necessary Change color convention in the Display tab Edit Color Map gt Choose Preset Choose Red Yellow Green Blue bar 6 1 9 Extract Commands Extract Dogbone Cyl or Flat MinRad 6 cm throat MaxRad 8 cm base TotalLength 10 cm ThroatLenth 6 cm 6 1 10 Insert Commands Insert LenticularFlaw Center 5 6 nm 56 nm 64 nm 57 X Axis 1 3 4 5 2 4 Y Axis 3 4 6 7 8 4 Thickness 5 nm X Length 50 nm Y Length 20 nm Plot Insert Particles Box 10 cm 10 cm 5 cm GaussianRadiusDist Seed 345 either Number or UntilFull Number 200 UntilFull 6 1 11 Discrete Element Commands Read DEM Particle File generated using John Peter s code ControlParameters Units Nano NodeList Particles Particles Density 2 g cm3 LengthUnits nm ReadDemParticleFile Problem3_RS_Final pts 6 2 MacroParticle Lists Macroparticles are assemblies of 3 or more spherical particles tied together either using rigid constraints or spring and dampers MacroParticleList ListName Type for available Types see below Density 2 0 g cm3 1 Enter type dependent parameters see below Parameters 1 2 a Generate one of the listed distribut
226. s input file The syntax is icc E D input mrs gt inputP mrs Executing the OpenMP version of Mars The OpenMP versions of Mars are stored in the usr local u peless Mars folder Executables are named using the following convention marsO yymmdd or marsOyymmdd where yymmdd is a six digit number representing the date of the executable marsOyymmdd employs the shared memory model On diamond it can only be executed on a single compute node over 8 compute cores A typical bash script is listed below bin bash PBS A ERDCVO22215P5 PBS 1 walltime 10 00 00 PBS N Ldpm4pt2 PBS q standard PBS j oe PBS 1 select 1 ncpus 8 mpiprocs 1 select compute nodes requested ncpus must equal 8 mpiprocs cores node PBS l place scatter excl Allows PBS to use any available nodes Excludes other users from using your nodes while they are scheduled to you cd work peless 1104Jov 4pt2 export OMP_NUM_THREADS 8 usr local u peless Mars mars0110524 B input mrs gt outOMP if you are using the C shell setenv OMP_NUM_THREADS 8 usr local u peless Mars mars0110524 B input mrs Executing the MPI version of Mars The MPI versions of Mars are stored in the usr local u peless Mars folder Exe cutables are named using the following convention marsM yymmdd or marsMyymmdd where 213 yymmdd is a six digit number representing the date of the executable marsM employs the distributed memory model On diamond it can only
227. s share the same locations these nodes 98 can be merged using the t mergeParts command The general format of the Generate subsection is shown below TetSolidList NAME Generate generate one or more parts using commands discussed below merge parts when distance between nodes is lt tol mergeParts 0 001 in tol 0 001 in Cylinder The Cylinder command is used for generating a regular tetrahedral mesh of a solid cylinder or of a section of a pipe The following example generates a hollow cylinder with internal diamter of 2 in and external diameter 2 31 in The cylinder extends in the z direction from z 0 3 in to z 19 52 in with 9 elements Cylinder use ingrid convention 56 1 10 2 2 31 radii 0 3 19 52 coordinates in the axial z direction Prism The Prism command is used for generating a structured tetrahedral mesh of a paral lelepipedal shape Generate a prism with regular grid Parallelepiped use ingrid convention AA ME S EE ES e 4 4 Din Dh 2 2 Extrusion The Extrusion command is used for generating an extruded solid starting from a trian gular mesh defined in the x y plane The triangular mesh is extruded in the z direction starting from z 0 and ending at z Length with element dimension in the z direction no larger than Increment 99 Extrusion TriangularFaceList listName Length 10 in Increment 1 in Solid Disk Disk use ingrid co
228. s specified in the first line after the list name HexSolidList ListName type where the keyword type can be one of the following Geometry this can be used to define rigid bodies Rigid this is used to define separate bricks FBSingleIP Flanagan Belitshcko formulation SIP FE fomulation with 8 integration points HyperElastic formulation for rubber materials SimpleHex simplified formulation for prismatic elements Explosive this employs EOS materials 1 If this list is used to define a rigid body enter Density 7 8 g cm3 else Material MATE You can also use InsertMaterial Mat when material was not previously defined 2 If mesh is explicitly defined enter NodeList Nodes or InsertNodeList Nodes ReadObjects 345 number of elements i n n n n4 n5 n6 n7 n8 1 124 243 56 165 234 312 23 126 2 56 238 121 78 56 98 126 66 else if mesh is internally generated enter Generate see below for generation options 123 or SplitTetMesh tetMeshListName 3 optional commands Make QuadFaceList ListName 2 generate a list of external surfaces Smooth Make TriangFaceList ListName 3 generate a list of external surfaces Make particleList PRTC see below Make HexSolidList SLCT make sublist of select hexes Make EdgeList ListName AllEdges make HexSolidList of selected elements Make HexSolidList ListN
229. scalar Set Scale Factor check Edit box and enter 0 5 Radius 1 Maximum Number of Points enter number greater than num of particles Click Apply button In Display Color by Solid Color Change color if necessary Display velocity rotation vectors as colored arrows Pipeline Browser Click on Particles 00 Main Menu Filters gt Recent gt Gliph Object Inspector In Properties Vectors Velocities or RotationRates Glyph Type Arrows Scale Mode vector Set Scale Factor change if necessary Maximum Number of Points enter number greater than num of particles Click Apply button In Display Color by Velocities Magnitude Press Edit Color Map Choose Preset select one of the preset color scales Of course it is possible to paint particles according to a number of scalar quantities such as vx vy vz lvl wx wy wz Iwl etc Main Menu File gt Open Select files Particles 00 Pipeline Browser Click on Particles 00 Object Inspector Properties Apply Display particles as painted spheres Main Menu Filters gt Recent gt Gliph Object Inspector In Properties Glyph Type Sphere Scale Mode scalar Set Scale Factor check Edit box and enter 1 Radius 1 Maximum Number of Points enter number greater than num of particles Click Apply button In Display Color by choose one of the variables Click Edit Color Map button Click Choose Preset button in color scale editor window Choose one of the
230. scaling factor to conversion factor F 6 3f uses a specific format C convention 16 3f 6 fields with 3 decimal digits 10 3E exponential notation 3 decimals de Description change description 1 The time history of the time step provides useful information regarding the evolution of the stable time step during a simulation This can help when in time certain problem may occur 2 The time history of the CPU time per step samples the time it takes per cycle at the print intervals The CPU time per step is typically constant except for the step when plots are generated or contact conditions are updated Some spikes in the curve are to be expected For MPI performance evaluations one should use the prevalent value 3 The time history of the cumulative CPU time tells how much CPU time has elapsed from the beginning of the execution It is useful to detect if a considerable amount of time is spent in performing special tasks For example if the detection distance in a contact list was set very large the time required to update the contacts may show up as a visible step in the time history curve 19 MPI Parallel Processing The MPI parallelization of MARS is work in progress that begun in the spring of 2009 The MPI approach adopted in MARS takes advantage of the object oriented architecture of MARS itself itself 167 Before the advent of multi core processors the term CPU was used to identify and count individual proces
231. short notation sf DynamicFriction 0 3 short notation df StribeckVelocity 10 m s 150 dt 0 001 ms for computing stiffness Beta 3 MinNormalForce 10 nN at 2 nm fO 10 N The StribeckVelocity is an optional parameter If entered the kinetic friction force is computed using the equation below below Fkf kfc sfc kfc exp R 2 Fn where R dv Vs dv is the norm of the relative tangential slipping velocity Vs is the Stribeck velocity The parameter Beta is optional and is used to kill small residual elastic oscillations in the elastic range when the particles are sticking in a quasi motioneless equlibrium state The dampening mechanism is not viscous velocity dependent but hysteretic based on cycle amplitude For Beta 1 there is no hysteretic effect Oscillation dampening increases as Beta increases Beta must be greater or equal 1 The default value is 2 to provide some damping The MinNormalForce is an optional parameter intended to approximate the effects of bonding forces which oppose tangential motion even when the particles are not pressed against each other The last parameter in the line is the maximum gap for this effect to be included The minimum normal force f0 is added to the normal force when the particles are still sticking and the gap is less than the maximum gap This virtual normal force increment results in a larger maximum static friction force f0 is no
232. sion term is also passed to the RCI model The model returns a velocity vector as well as a pressure term which is applied to the tetrahedral element 139 BeamTetConstraintList listName RebarConcretelnteractionVE BeamList beamListName TetList tetListName NumberOfCircumferentialConstraintsPerBeamElement 3 NumberOfAxialConstraintsPerBeamElement 2 RebarRadius 10 mm Material RciModelName 15 5 Beam Hex Constraints 15 5 1 Elastic formulation BeamHexConstraintList NAME ElasticPenalty BeamList RBRS HexList CNCR Stiffness BasedOnTimeStep 0 002 ms 15 5 2 Elastic formulation with slippage BeamHexConstraintList NAME RebarConcretelnteraction BeamList RBRS HexList CNCR NumberOfCircumferentialConstraintsPerBeamElement 3 NumberOfAxialConstraintsPerBeamElement 2 RebarRadius 10 mm Material INTR 15 5 3 Elastic formulation with slippage and volumetric effects BeamHexConstraintList NAME RebarConcretelnteractionVE 4 BeamList RBRS HexList CNCR NumberOfAxialConstraintsPerBeamElement 2 RebarRadius 10 mm Material INTR 15 6 Constraint List This list includes a variety of miscellaneous constraints It is somewhat different than other lists in the sense that it can group objects that are non homogenous It was done that way because there are many possible ways constraints can be formulated and it 140 seemed that there would be a proliferation of lists if every type of constraint would be given its own list
233. sity 18 Dissipated energy density energyDensity 19 Dissipated energy density rate in tension powerDensity 20 Dissipated energy density in tension energyDensity 5 9 6 RCI Rebar Concrete Interaction Model Material RCI RebarConcreteInteraction StaticParameters 4 41 StrainRateEffects State variables for vectorial equations Axial stress stress Radial stress stress Circumf stress stress Axial slippage length Radial slippage length Circumf slippage length Max axial slippage length NOoP WN KH 5 9 7 LDPM Concrete Fiber Interaction Model This model is used exclusively for computing concrete fiber interaction at the LDPM cell factes It should not be used with any of the finite element formulations Material CFI LdpmFiber Dev ElasticModulus 210000 MPa ymf SpallingParameter 1500 MPa sps FiberStrength 1172 MPa sfu FiberStrengthDecay 0 0 fsd SnubbingParameter 0 05 fsn BondStrength 1 75 MPa tau VolumeStiffnessRatio 0 0 eta DebondingFractureEnergy 0 1 N m gdm PullOutHardening 0 05 poa CookGordonParameter 2 SpallingParameter 150 MPa State variables for vectorial equations Normal N force force Shear M force force Shear L force force Normal N strain nondimensional Shear M strain nondimensional Shear L strain nondimensional Short side slippage length Long side slippage length Max short side slip
234. son such as poor algorithm possible bug etc and the user should notify the developer MARS uses the following procedure to determine the time step used in the explicit time integration scheme First the Courants stability limit Dtc is determined list by list and the minimum is selected The Courants time step is then multiplied by a scaling factor s defined via input default value for s is 0 9 The reduced time step is compared to a maximum allowable time step Dtmax prescribed via input the minimum of the two is chosen chosen Dt s Dtc Dt min Dt Dtmax 20 The maximum time step Dtmax can be defined either as a constant throughout the simu lation or as a function of time The second method is useful when we want the simulation to start with a very small step for accuracy and not stability like in an impact problem and then relax this condition as the simulation progresses progresses For constant maximum time step the commands are ControlParameters Units English TimeStepScalingFactor 0 7 MaximumTimeStep 0 001 ms For time dependent maximum time step the commands are ControlParameters Units English LoadCurve MaxStepHist Max allowed time step starts with Dt 0 1 microsec and grows linearly to 1 microsec in the first ms of the simulation it remains constant thereafter X Units time ms Y Units time ms ReadPairs 3 0 0 0 0001 1 0 0 0010 10 0 0 0010 ControlParameters MaximumTimeStepH
235. sor units in a computer system This term can still be used to describe a physical object that sits in your machine but when discussing resource allocation for parallel computing the words CPU and processor are ambiguous and are therefore best avoided For instance documentations of cluster resource allocation schemes typically make reference to processors as meaning essentially a core as in each node has two sockets and each chip has two cores which means that each node has four equivalent processors It would then be very confusing to say that this cluster has two dual core processors on each node Instead we have a hierarchy of three levels that we deal with cores sockets and nodes Node also called host computer machine For our purposes the most important definitions of a node is that a certain amount of memory RAM is physically allocated on each node Nodes are typically separated by network connections and each node has a unique network address host name IP number Each node contains one or more sockets the node is then labeled e g a two socket or four socket node Socket refers to collection of cores with a direct pipe to memory Note that this does not necessarily refer to a physical socket chip but rather to the memory architecture of the machine which will depend on the chip vendor specifications Usually however the sockets resemble the old definition of a CPU
236. story profile The time history profiles is specified via a load curve in the same fashion of how it is done for dynamic relaxation The load curve provides a scaling factor which is multiplied by the value of the acceleration specified after the keyword Gravity In the example below the gravity is linearly increased from 0 to its final value of 9 8 m s2 in 2 milliseconds and then it is kept constant ControlParameters Units English F LoadCurve GravityHist X units time ms Y units nondimensional ReadPairs 3 O 0 00 2 1 00 100 1 00 ControlParameters Gravity 9 8 m s2 Direction 0 0 1 History GravityHist 4 9 Plugins By plug ins we mean other codes that can be executed from MARS and perform specific tasks Currently there are four plug ins e Quasar for displaying 3 d models e Xth for displaying time history plots e triangle for generating triangular meshes e tetgen for generating tetrahedral meshes 24 5 Miscellanous Objects 5 1 Reference Systems Reference systems are used for two main purposes 1 When generating parts they are used for aligning a new part in the desired orien tation 2 In finite element formulations they are used for aligning the local element axis and computing internal strains and stresses Three reference systems are available 1 cartesian 2 cylindrical and 3 spherical Some applications require the reference system to return a preferential direction
237. system Internal Work Wi is the work performed by internal stresses as a body is deforming If the strains operate in the direction of the stresses for example tensile strain increments are applied to a body already in tension then the internal work is positive For nonlinear materials internal work will result in recoverable elastic energy and dissipated energy e g due to plastic work An internal work quantity is computed for each list that consists of entities with internal forces finite elements contact elements bonding elements etc The computation of the internal work is done just before the node force increments are added to the total nodal force variable Wit Fiji Vij ts dt 2 dt where Fr are the force increments and Vz are the corresponding velocities Note that the velocities are half a step off from the forces this may result in small errors KineticEnergy Ek is the energy associated to the motion of the parts and is the summation of of where m and v are the masses and velocities of the nodes The kinetic energy is always positive EnergyBalance Eb is defined as E W Ke We The energy balance should remain constant during a calculation If a system starts in its unstressed state but with moving parts the intitial energy balance is the total kinetic energy of the moving parts at the beginning of the simulation It is possible to print a global table of external internal and kinetic ene
238. t ListName NanoParticles VanDerWaalsForces HamakerConstant 1 61e 20 J ThreshholdGap 0 4 nm 7 10 3 Plotting Options It is possible to generate contour plots of the contact stresses Contact stresses are computed as described in the online manual The command for generating contour plots in Quasar format are given below PlotList ListName npL ListName RedValue 100 psi TT BlueValue O psi StressComponent XX valid components are XX YY ZZ YZ ZX XY le If the Paraview format is chosen then there is no need to specify stress components and min max values These operations are done in Paraview The Paraview file contains eight records listed below XX component of the contact stress YY component of the contact stress ZZ component of the contact stress YZ component of the contact stress ZX component of the contact stress XY component of the contact stress Maximum contact stress oo nN on AUNE Force chain vector Note that Paraview makes it possible to plot force chain vectors writing a Paraview file sequence are PlotList ListName 4 Paraview npL ListName di 7 11 Node Pair Penalty Constraints NodePairList ListName PenaltyConstraints ForcePenaltyStiffness 1 e N m use MomentPenaltyStiffness to constrain moments MomentPenaltyStiffness 1 e Nm ReadObjects 2 List1 34 List2 43 List1 14 List2 76 8 Triangular Faces and Shell El
239. t nodes which are used in the definitions of the selected edges in the list SelectNodes UnselectNodes and ReselectNodes For example the commands below are used to select the nodes of the first forty edge elements in the list EdgeList listName Geometry 1 Select do 1 40 SelectNodes These commands are also applied to all other types of lists which are derived from the edge list 7 1 3 Make commands The Make EdgeList command is used for generating a sub list of edges which have been previously selected The new list is of the Geometric type in other word MARS will not perform any operation on it e g computer internal forces The Make NodeList command is used for generating a sub list of nodes which are used for defining the edges in the current edge list 7 1 4 Notes Currently beam elements although derived from the Edge class are treated in a sepa rate list called BeamList This may change in future versions of MARS MARS The command Write PartMeshDataFile and ReadFile can be used to transfer mesh details to a separate file and clean up the main input file For example we may import a truss edge mesh generated using a processor code and convert it to MARS in the conversion run the edge list can be saved using the command Write PartMeshDataFile PART mrs 66 In the main input file the edge data can be loaded using the commands commands EdgeList ListName type 4 Read PART m
240. t of view physical fibers can be described using few parameters fiber density length diameter and tortuosity The latter parameter is used to characterize how bent fibers are a straight fiber has no tortuosity while a fiber with many kinks is very tortuous These geometric parameters are used in MARS for generating random fibers inside a control volume which can either be the volume of the concrete part or a larger volume that contains the concrete part Each fiber is model using a sequence of one or more segments linked together Single segments are sufficient for generating straight fibers Multiple segments are necessary for generating tortuous fibers Fiber location and ori entation is random All fibers are completely contained in the control volume For cast fiber reinforced concrete parts the control volume should be equal to the volume of the concrete part For machined parts the control volume should be larger than the part The fibers that intersect the external surface of the part are treated as cut The portion of a cut fiber which lays inside the part is shorter than the length of original fiber and this affects the mechanical characteristics of the fiber facet interaction of the facets near the external surfaces In the spirit of the discrete multi scale physical character of the LDPM the occurrences of fiber facet intersections are determined by actually computing the locations where fibers cross inter cell facets This computation c
241. t 0 01 in sec In this case you have rate effect on the constitutive equation and the macroscopic behavior but the inertia effects are likely to be small or negligible compared to the internal energy a ratio of about 1000 is typically fine Then you may decide to run the simulation at for example 0 2 in sec to save computational time while not changing significantly the kinetic energy compared to the internal energy However if you do so you need to make sure that the constitutive equation feels a loading rate of 0 01 in sec otherwise you would not get the response associated with 0 01 in sec which is what you want You make the constitutive law feeling 0 01 in sec even if you run at 0 2 in sec by setting the Pseudo TimeFactor 0 2 0 01 20 20 State variables for vectorial equations 1 Normal N stress stress 2 Shear M stress stress 3 Shear L stress stress 4 Normal N strain nondimensional 5 Shear M strain nondimensional 6 Shear L strain nondimensional 7 Max normal N strain nondimensional 8 Max shear T strain nondimensional 9 Tensile strength stress 10 Post peak slope in tension stress 11 Shear L crack opening length 12 Volumetric Strain nondimensional 13 Normal N crack opening length 14 Shear M crack opening length 15 Total crack opening length 16 Facet failure flag nondimensional 17 Dissipated energy density rate powerDen
242. tartFile Every 0 1 ms 18 4 2 1 Go Interactive These commands are used to schedule when the program should go interactive This command is not executed for batch runs GoInteractive AtTime 10 ms GoInteractive AtStep 10000 GoInteractive Every 0 1 ms GoInteractive IfTimeStepLessThan 0 0001 ms 4 2 2 Print Progress Line These commands are used to schedule the printing of a line to the output file More that one command can be used PrintProgressLine Frequency 1 PrintProgressLine Frequency 10 Value vmx Values can be vmx ken default fmx wrk 4 2 3 Write Restart File These commands are used to control the frequency for writing restart files More that one command can be be used WriteRestartFile AtTime 10 ms WriteRestartFile AtStep 10000 WriteRestartFile Every 0 1 ms 4 2 4 Execute Function These commands are used to schedule the execution of a macro function More that one command can be used ExecFunction FunctionName AtTime 10 ms ExecFunction FunctionName AtStep 10000 ExecFunction FunctionName Every 0 1 ms ExecFunction FunctionName IfTimeStepLessThan 0 0001 ms 4 2 5 Flush Time Hist Files These commands are used to schedule when to flush the buffers of the time history datafiles FlushTHFiles AtTime 10 ms FlushTHFiles AtStep 10000 FlushTHFiles Every 0 1 ms 19 4 2 6 Read Input File These command is used to schedule a reading event At the requestest time or step MARS will
243. te rebar interaction model is entered as a Material Note that The rebar radius must be explicitely entered The constraint detection phase identifies pairs of particles and rebar finite elements that interact with each other This operation is performed once at the beginning of a sim ulation The particle rebar interaction constraint is then applied to each pair Currently we select all particles that are suitably close to the rebar elements More specifically we do not include particles whose spherical volume overlaps the cylindrical volume oc cupied by the rebars We do include however particles when their gap from the rebar is within a range specified by the analyst using the MinimumGap and MaximumGap commands The rebar overlapping particles are connected to the other particles as if the rebar does not exist Their effect is accounted for in the bond parameters when parameter calibration is performed The constraint formulation ties a single particle to a single beam element defined by its two nodes The formulation consists of two parts a kinematics part where the local deformation state is computed from particle node velocities and rotations a force reduction part where local stresses are used to compute forces and moments at the particle and beam nodes nodes ParticleFiberInteractionList Constraints NodeList Particles BeamList Fibers Material ConcFiberInt FiberRadius 0 6 cm MinimumGap 0 1 cm MaximumGap 2 0 cm V
244. tep 2 cx lt 0 4 in nodes with cx lt 0 4 cy 0 4 cm nodes with cy apprx 0 4 tb 00X nodes with translational BC 00X rb XXX nodes with rotational BC XXX vz gt 10 m s nodes with z velocity gt 10 cl 0 10 4 cm node closest to point 0 10 4 Examples Select do 1 100 select nodes 1 through 100 AlsoSelect do 201 300 add nodes 201 300 Reselect cx lt 0 in select nodes with negative x from nodes already selected Nodes can also be selected from within element lists Following are a few examples The first example shows how to select nodes that belong to a set of faces which point in the positive x direction These could be all the nodes of a circular face of a cylinder aligned with the x axis TriangFaceList ListName EditNodeList Unselect all Select FacesWithCrossProduct 1 0 0 gt 0 99 SelectNodes EditNodeList 50 Set Translations XXX Select all Most element lists provide the following commands SelectNodes UnselectNodes and ReselectNodes The second example shows the difference between SelectNodes and ReselectNodes in the example we want to select nodes that are shared by two element lists and save them in a new node list Refer to other commands NodeList Nodes Unselect all TriangShellList Shells SelectNodes HexSolidList Solids ReselectNodes NodeList Nodes 4 Make NodeList Com
245. ters Node ndM Node obM cast object pointer to element pointers TetCosserat10 tt TetCosserat10 lo int i ntt nlo ifdef _OPENMP pragma omp parallel private i pragma omp for endif for i 0 i lt ntt i tt i gt calcForces ndF 10 i 30 i1 ndM 4 i m 12 i element forces and forces at the nodes are added to the nodes in method procForces obLst procForces return M_BIG 189 Note that if the element formulation had only translational Dof s the method could be simplified to obLst createArrays 10 0 Node ndF Node obF TetCosserat10 tt TetCosserat10 1lo int i ntt nlo for i 0 i lt ntt i tt i gt calcForces ndF 10 i f 30xi The force and moment increments are saved in arrays f and m and later added to the nodes in method procForces This is done to avoid memory overwriting within the OpenMP parallel loop containing the calcForces element method In addition to the three polymorphic methods discussed above there are two methods that are relevant The first method bool processCommand Reader string is used to process list specific input commands such as InsertMidEdgeNodes and loooks like this bool ttL_Cosserat10 processCommand Reader rdr string 1b1 if 1bl InsertMidEdgeNodes insertMidEdgeNodes else return false return true The second method is void insertMidEdgeNodes and like the name says its purpose is to add
246. that may be used for specific tasks such as model generation of LDPM tetrahedral meshes Activate changes source bashrc Any subsequent time you log in start the bash shell and the path is automatically updated gt bash Interactive Execution To run the OpenMP version of Mars enter marsO input mrs To get the latest version of the built in LaTex manual enter marsO D1 210 Batch Execution To run the OpenMP version of Mars in batch mode create a script e g job sh bin bash echo job starting cd yourProblemFolder marsO B input mrs gt output Then submit it using the sbatch command sbatch p opterons n 1 t 120 o out job sh where p opteron is the partition n 1 is the number of compute nodes t 120 is the maximum time in minutes o out is where the output goes job sh is the input script To check status type squeue To list partitions type sinfo Batch Execution MPI First you must configure your envirnoment Information can be found at wiki ccni rpi edu index php RedHat_5_and_Modules Load the MPI environment module load mpi openmpi 1 4 gcc41 Create a script file e g job sh to control batch execution 4 bin bash echo job starting cd yourProblemFolder mpirun n 32 marsM B input mrs gt output Then submit it using the sbatch command sbatch p opterons n 32 t 120 o out job sh 211 23 3 Mars on NWU Quest The first time you use
247. the same input file used in the simulation for initializing the MARS model database However instead of running a simulation MARS reads the data frames from the plot data file at the intervals at which they were saved and generates the 3 D plot files as specified in the PlotLists You may modify the input file for generating new plot lists or change plotting parameters in existing ones The command for instructing MARS to read data from the input file is ProcessPlotDataFile and is inserted at the very end of the input file Mars Input File ControlParameters PlotList Plot1 4 update any parameter J insert new PlotLists 177 PlotList Plot2 4 ProcessPlotDataFile data plt EOF Updates 11 24 09 works for MPI runs 21 3 write and read restart files Restart files fulfill two useful functions 1 they make it possible to restart a job that was terminated for a variety of reasons 2 they make it possible to interactively post process the data in a different computer There are three ways to direct MARS on when to create restart files 1 in the input file 2 when an execution is interactive mode 3 via external command file In the input file file restart commands are written inside the ControlParameters block Three forms are available as shown in the example below More than one line can be entered ControlParameters WriteRestartFile AtTime 10 ms WriteRestartFile AtStep 10000 WriteRestartFile Every 0 1
248. this this Title line ControlParameters WritePlotDataFile data plt Every 0 001 ms y EOF Note that the string DT in the input file is replaced by the definition given in the command line by DDT The switch D indicates that what follows in this case DT is a macro with 180 an assigned value of 0 001 If no value is assigned to the macro then the preprocessor automatically assigns a value of 0 The C preprocessor commands for generating the post processing input file inputP mrs are are CC E DDT 0 001 DPLT input mrs gt inputP mrs The resulting input file looks like this Title line ControlParameters WritePlotDataFile data plt Every 0 001 ms PlotList Cells 4 Paraview Timelnterval 0 001 s ttL Tile Cells ProcessPlotData data plt EOF In the examples above the time interval parameter DT could have been defined inside the file input mrs using the define directive directive This is the title define DT 0 001 ControlParameters WritePlotDataFile data plt Every DT ms ifdef PLT PlotList Cells Paraview TimeInterval DT s ttL Tile Cells ProcessPlotData data plt endif EOF The second example is more complex but shows the potential benefits of using pre processing directives Let s assume you have a complex model and want to perform a parametric study of a projectile impacting a structure at different speeds Some of the problem parameters would change depending on the projectile speed F
249. thod requires a single parameter the stiffness Kl of the linear spring used in the equation As an alternate way of computing stiffness a time step may be entered This time step is used in conjunction with the smallest mass of the nodes participating in the contact to define the penalty spring stiffness Fn K p The input commands can be enter either in a single line ContactModel Penalty Kl 1e6 N m ContactModel Penalty dt 0 001 ms or in multiple lines ContactModel Penalty SpringStiffness 1 e6 N m ContactModel Penalty StiffnessBasedOnTimeStep0f 0 001 ms 1 1 The smaller the time step the stiffer the penalty springs are One can check the amount of maximum penetration using time history commands in the contact lists ContactForce PenaltyEP 147 16 1 2 Penalty hysteresys contact model The penalty function with hysteresis contact model provides the capability of modeling non elastic contacts where some energy is dissipated This is accomplished by using different paths for loading and unloading The initial loading follows a linear force penetration relationship Fn K p Unloading follows a different relationship Fn C pb where b is a non dimensional parameter which is greater or equal 1 C FX pX b is constant pX is the penetration when unloading begins FX is the contact force when unloading begins For b 1 the unloading curve is the same as the loading curve and there is no hysteresi
250. time we are providing a basic explanation with some examples This documentation will be expanded in the future First some knowledge of Object Oriented Architecture and Programming is required Users who are not familiar with Object Oriented Programming concepts are encouraged to google words like c classes e inheritance c polymorphism etc These terms are used here assuming the reader has some understanding of their meaning We will describe the process using an actual example In this demonstrative case we will implement a 10 node Cosserat tetrahedral element with 42 degrees of freedom 10x3 translational DoF s for the 10 nodes and 4x3 rotational DoF s for the 4 vertex nodes The class CosseratTet10 is derived from the basic class Tet whose definition can be found in file mars h Addtional members are added references to 6 mid edge nodes state variable arrays etc The definition for the new class is given below class TetCosserat10 public Tet protected Node nM 6 public TetCosserat10 virtual TetCosserati0 virtual void whatAmI cout lt lt 10 node lt lt endl void setMidEdgeNodes Node n 6 for int i 0 i lt 6 i nM i n i 187 void read Reader Node real calcForces Node real Node real Je The list class that handles a collection of the new tet elements is named ttL_Cosserat10 and is derived from the base class ttLst that han
251. tion from 0 to 2 in in the y direction and from 0 to 8 inches in the z direction 11 2 2 Cylinder The LdpmCylinder command is used for generating an LDPM concrete cylinder This method first generates an external triangular face mesh representing the surface of the cylinder It then randomly places a set of particles inside the cylinder It uses tetgen to generate a tetrahedral mesh The method select an appropriate mesh density for the exernal mesh based on expected average distance between internal particles LdpmCylinder automatically generates particle list PRTC Radius 2 in radius of the cylinder Diameter 4 in use either Radius or Diameter Length 8 in length of the cylinder GapScalingFactor 0 1 gap between particles Seed 23423 seed for random number generator 111 11 23 Sphere The LdpmSphere command is used for generating an LDPM concrete sphere This method first generates an external triangular face mesh representing the surface of the sphere It then randomly places a set of particles inside the sphere It uses tetgen to generate a tetrahedral mesh LdpmSphere automatically generates particle list PRTC Radius 2 in radius of the sphere Diameter 4 in use either Radius or Diameter Refinement 3 1 external mesh density Split generate only half sphere for mesh check Seed 23423 seed for random number generator 1 The argument of the Refinement parameter is an integer
252. tion opposite of that of rotation The time history of the braking action is prescribed using a load curve which is referenced using the BrakingHistory command Mechanisms Brakes Node NodeListName nodeIndex Runway FaceListName BrakingHistory CurveName J 22 2 3 Anti Lock Brake The anti lock brake mechanism Mechanisms AntiLockBrakes Node FrontWheels 1 Runway FaceList RampUpRate 22 3 Collection List A collection list is an assembly of lists that are grouped together for the purpose of performing the same operation on them All these operations are done at the pre or 205 post processing phase No actual computations are done during the solver loop In the input block the lists are first selected and then the operations on them are performed Typical applications of this feature are 1 Translate or rotate the entire model or part of it 22 4 Unit Cell The UnitCell class includes methods for generating a unit cell and for creating a list of constraints that enforce periodic conditions during a simulation of a unit cell There are two phases generation and simulation The input for the generation phase has the following format format Material Concrete RCConcrete Dev 4 MixDesign StaticParameters UnitCell Cube Material Concrete Side 2 in Seed 45435 seed for random number generator Generate J The parameters for Material
253. tions 1 XX Stress stress 2 YY Stress stress 3 ZZ Stress stress 4 YZ Stress stress 5 ZX Stress stress 6 XY Stress stress 7 Effective plastic strain nondimensional 8 XX Back stress stress 9 YY Back stress stress 10 YZ Back stress stress 11 ZX Back stress stress 12 XY Back stress stress 13 XX Friction stress stress 14 YY Friction stress stress 15 ZZ Friction stress stress 16 YZ Friction stress stress 17 ZX Friction stress stress 18 XY Friction stress stress 37 19 XX Strain nondimensional 20 YY Strain nondimensional 21 ZZ Strain nondimensional 22 YZ Strain nondimensional 23 ZX Strain nondimensional 24 XY Strain nondimensional 25 Energy energy State variables for plane stress equations 1 XX Stress stress 2 YY Stress stress 3 YZ Stress stress 4 ZX Stress stress 5 XY Stress stress 6 Effective plastic strain nondimensional 7 XX Back Stress stress 8 YY Back Stress stress 9 YZ Back Stress stress 10 ZX Back Stress stress 11 XY Back Stress stress 12 XX Friction Stress stress 13 YY Friction Stress stress 14 YZ Friction Stress stress 15 ZX Friction Stress stress 16 XY Friction Stress stress State variables for vectorial equations Normal stress stress Shear stress X stress Shear stress Y stress Effective plastic strain nondimensional e UNE 5 9 3 Rate Sensistive Elast
254. to set the radii of all particles to the same value the following command ContactRadius 5 nm sets rmn rmx 5 nm nm 7 10 1 Tabulated Forces When the built in equations are not adequate for representing the inter particle forces including ionic forces it may be desirable to define these forces as a tabulated function where x is the the inter particle gap and y is the inter particle force Note that these forces overlap to any other force including contact and VanDerWaals forces 76 LoadCurve CurveName inter particle force versus gap table NodePairList ListName NanoParticles TabulatedForce CurveName 7 10 2 VanDerWaals Forces The VanDerWaals force is the attractive or repulsive force between molecules or between parts of the same molecule other than those due to covalent bonds or to the electrostatic interaction of ions with one another or with neutral molecules Wikipedia The expres sion for the VanDerWaal force implemented in MARS employs the Hamaker constant hmk and a treshold gap value thg The C code of the implementation is listed below D 2 min R1 R2 R1 R2 radii of the two particles gp max wgp thg wgp gap between the two particles x gp D xl x 4 1 x2 x x 2 expression for the VanDerWaals force frc frc hmk 6 D 2 x1 x2 x1 x2 x2 2 x1 1 x1 x1x x1 The input for VanDerWaals forces is shown below NodePairLis
255. ts tcL ListName ActiveContacts tcL ListName MinimumDistance 1 tcL ListName MinimumGap tcL ListName NormalForce tcL ListName X Force tcL ListName Y Force tcL ListName Z Force 1 The MinimumDistance time history is very useful for checking whether any node has crossed a suface This is likely to happen if the distance becomes very very close to zero 16 4 2 Visualization For models of moderate size which employ contact lists it is useful to visualize the elements that can potentially come into contact at time zero Such visualization provides 1 a way to check that thicknesses of points edges and faces are prescribed correctly and 2 a way to assess that the detection distance parameter is adequate The visualization methods which were available in interactive examine mode have been improved and made available also in input file mode as new commands These new features are available in MARS versions dated April 19 2010 or later The WritePlotFile command is used to generate a Quasar plot file that visualizes node face contacts detected at time zero The plot contains only the faces and nodes that are used in the contacts using the plot attributes color and radii defined in their respective lists The file is named using the convention tcL listName plt 160 16 5 Rebar Contact List The rebar contact list makes it possible to detect contacts between a list of nodes particles an
256. types of cross section integration schemes and easily add new ones BeamList Rebars 1 RefSys is used to set beam 2nd direction ReferenceSystem RefSysName 1 69 2 select one of the cross section types sec o solid cylindrical cross section sec 0 tubular cross section set h hat section sec I sec R rectang hollow cross section sec Z Z section sec c custom section 3 1 specify mesh explicitely NodeList Nodes ReadObjects 345 number of elements i m 2 1 124 243 2 56 238 3 2 else generate mesh generate see below for mesh generation options 3 3 else read mesh from external file ReadFile filename optional Read PART mrs 4 specify plot attributes Color Rust 5 writing and reading external files Write PartMeshDataFile PART mrs 1 The selection of the reference system is very important specially for cross sections that are not axi symmetric The provided reference system is used for setting up the local co rotational reference system for each beam element during the initialization phase The local x axis is oriented in the direction of N1 to N2 where N1 and N2 are the two nodes defining the beam element The specified reference system returns a preferential direction at each point in space see section describing reference systems During beam initialization the preferential direction U is computed at the beam midpoint This
257. ulus 29e6 psi PoissonsRatio 0 3 Beta 3 16 1 5 Stick slip friction model The stick slip friction model is an algorithm designed for computing tangential contact forces between two objects perpendicular to the normal direction at the point of contact The tangential force is computed incrementally based on the tangential relative motion of the two objects at the point of contact and a penalty stiffness parameter The tangential force is limited either by the static friction force if the objects are sticking or the kinetic friction force if the objects are slipping When the objects are sticking and the tangential force exceed the static friction force the contact is assumed to convert from sticking to slipping When the objects are slipping and the tangential force becomes smaller than the kinetic friction force the contact is assumed to convert from slipping to sticking The static friction force is defined as Fsf sfc Fn Fsf sfc Fn Fkf kfc Fn where sfc is the static friction factor coefficient and Fn is the normal contact force inde pendently computed in one of the contact models described above The kinematic friction force is defined as where kfc is the kinetic friction factor coefficient kfc is typically less or equal than sfc The input commands for the friction force model are FrictionForce dt 0 01 ms sf 0 3 df 0 2 or FrictionForce PenaltyStiffness 10 N m short notation K StaticFriction 0 3
258. ure histories The general syntax for these lists uses the format below More details are given for each specific loading tpye TriangFaceList listName loadingType 8 2 1 Uniform Pressure This list is used for prescribing a time dependent uniform pressure history over a set of triangular faces The pressure can be prescribed either using a LoadCurve tabulated expression or using an Equation The latter is particularly useful when the pressures depend on the change in volume affected by the motion of the faces themselves The single value time dependent pressure is multiplied by the current area of each face in the direction opposite to its normal 84 Equation equationName Y J TriangFaceList ListName MultiplePressureHistories 1 Specify face list Req either a face list or a shell list Link FaceList listName Link ShellList listName 2 Specify PressureHistory List Req LoadCurveList PressureHistories 8 2 2 Uniform Pressure Constant Face Areas This list makes it possible to apply uniform pressure histories on a set of faces In this case the nodal forces are computed by multiplying the applied pressure to the initial surface area using the initial surface normal This approach works well for LDPM models which experience major fragmentation In some cases some of the surfaces grow very large because fragments centered at the nodes move away from each other
259. vel 4 volumetric strain versus bulk modulus curve y LoadCurve Curve2 pressure versus volumetric strain curve LoadCurve Curve3 strain rate enhancement curve Material Concrete KcConcreteModel Density 2 0 g cm3 PoissonsRatio 0 19 CompressiveStrength 6000 psi TensileStrength 600 psi BulkModCurve Curvel PressureCurve Curve2 StrainRateEnhancement Curve3 OneInch 1 2 he The user can enter partial data for example Curvel but not Curve2 and Curve3 The program will automatically generate data only for the missing data 2 The parameter OneInch was used for telling the model the units used in the calcula tions For example if the calculation employs cm as length units the value of OneInch would be 2 54 This parameters should be no longer needed since all input values are now entered with their dimensions 5 10 Functions Macros The Functions object makes it possible to write a script and perform operation on the objects of the database Variables in the database are accessed using the same convention that is used to identify variable for time history plotting Function FNC1 exec FNC1 now The Function can be also invoke by the Action scheduler Function FunctionName ControlParameters 44 ExecFunction FunctionName AtTime 10 ms ExecFunction FunctionName AtStep 10000 ExecFunction FunctionName Every 0 1 ms ExecFunction FunctionName IfTimeStepLessThan 0 0001 ms F
260. wever the spacing of the constraints is dictated by the length of the beam finite elements In the beam tet constraints we specify a number of points along each beam element and we enforce the constraint at all these points The main obvious advantage of this formulation is that we can make the spacing of the constraints smaller than the average size of tetrahedral elements In this way the forces transmitted from the rebars are distributed in a smoother fashion to the concrete A more significant advantage is the directionality provided by the beam elements This makes is possible to model slippage in the axial direction at least in the small deformation range Three formulations are available e ElasticPenalty e RebarConcretelnteraction e RebarConcretelnteractionVE volumetric effects 138 15 4 1 Elastic formulation This formulation ties a series of points along the beam element with corresponding points inside the concrete LDPM elements using penalty springs The stiffnes of the penalty spring can either be computed based on a time step or defining a stiffness per unit surface In the first case the alogorithm finds the smallest mass m of the six nodes two for the beam and 4 for the tet element participating in the constraint and computes the stiffness using the formula K m dt dt BeamTetConstraintList listName ElasticPenalty BeamList beamListName TetList tetListName NumberOfConstraintsPerBeamElement 2 Sti
261. when the particles start slipping thus simulating a failure in the bonding force In the current formulation f0 is added to the normal force when the slipping flag is false thus if slipping stops the bonding force is re established Note that the actual normal force used in the computation of the contact force is not affected The default value of f0 is 0 0 The parameter dt can be used for computing the penalty spring stiffness as an alterative to the PenaltyStiffness parameter When dt is selected the value of the penalty stiffness for each contact is computed using the equation The parameter dt can be used for computing the penalty spring stiffness as an alterative to the PenaltyStiffness parameter When dt is selected the value of the penalty stiffness for each contact is computed using the equation K 0 5 ms dtxdt 152 16 1 7 Rolling resistance model Rolling resistance is the resistance to rotational motion that occurs when two objects roll over each other Rolling resistance is caused by local deformations in one or both of the objects in contact Strictly speaking the rolling resistance is defined as a force that opposes the rolling motion of a round object over a flat surface In MARS the term rolling resistance is used in a more general way where we consider the relative rotational motion of two objects in contact The relative rotational motion is opposed by equal and opposite moments appl
262. y parallel computer systems like the ones available at the ERDC supercomputer center Over the last two years 2008 2010 MARS has been modified to implement domain decomposition and use the Message Passing Interface MPI protocol This is an on going area of research which puts MARS at the leading edge of simulation software 1 5 Vehicle Response to Blast Conversion filters in MARS translate models developed for other codes In the exam ple below the model of a Ford Taurus developed at GWU for crash simulations was employed to simulate the effect of surface charges applied on the vehicle 1 6 Laceration of Plates and Shells The MARS laceration algorithm is essentially a two dimensional version of the solid fragmentation algorithm presented earlier Cracks can develop between clusters of shell elements depending on specified local failure criteria Cracks can propagate or coalesce to form tears All material is fully accounted for as all mass is maintained and balanced 1 7 Steel Plates Structures Complex steel plate structures commonly found in civil and mechanical construction can be modeled in great detail with MARS Contact conditions between all entities in the model ensure that plates do not penetrate each other Rivet bolt or weld elements are used to hold plates and braces together 2 Available Documentation The MARS documentation consists of four different forms of documentation e An online manual hosted at http www es3
263. yword in SinFunction read J Bae Sas Sea a ae eae Saas e er createUserDefinedList string must always be defined obLst Model createUserDefinedList string nam Reader return NULL ieee EN NS Aaa a i nen createUserDefinedMaterial string must always be defined Obj Model createUserDefinedObject string nam Reader return new SinFunction nam 21 7 5 Multiple lists and objects in a single user cpp file Because the SinFunction is derived from the LoadCurve function it can be referenced by its name in every class that employs a load curve for specifying a relationship For example the load function above can be used for specifying nodal loads The versatility of the user defined interface makes it possible to introduce any number of different lists and objects simultaneously This is accomplished as shown in the example below include mars h f AAA A A I TO class MyFirstTetList public ttLst user defined tet list _ class MySecondTetList public ttLst another user defined tet list AN A E A class MyHexList public hxLst 1 user defined hex list 195 class MyMaterial public Material user defined material class MyLoadCurve public LoadCurve

Mars Manual

Contents

Download Pdf Manuals

Related Search

Related Contents