Home

User`s Guide - Center for Simulation of Advanced Rockets

1. lt cell kind gt is a one letter string indicating the cell kind A if the zone consists of actual cells 104 Chapter 9 Visualization V if the zone consists of virtual cells region index is a five digit string indicating the global region index Examples are shown in Fig 9 2 which represents a screen dump of part of the zone style menu in TECPLOT Zone number 1 contains the actual hexahedra of region 1 while zone number 23 contains the virtual prisms of region 2 9 2 3 2 Boundary Patch Zones The names of boundary patch zones are PAT patch index face type gt lt face kind region index where patch index is a three digit string indicating the global patch index It is important to note that this is a global patch index face type is a string indicating the face type TRI if the zone consists of triangular cells QUAD if the zone consists of quadrilateral cells face kind is a one letter string indicating the face kind A if the zone consists of actual faces V if the zone consists of virtual faces lt region index is a five digit string indicating the global region index Examples are shown in Fig 9 2 which represents a screen dump of part of the zone style window in TECPLOT Zone number 9 contains the actual quadrilateral faces on global patch number 1 in region 1 Zone number 13 contains the virtual triangular faces on global patch number 5 in region 1 Note once again that
2. SDIMENSBF Specifies the dimensionality of the boundary face stencil It can take the fol lowing values 2 use two dimensional boundary face stencil 3 use three dimensional boundary face stencil default TOLERICT Specifies the value of the tolerance used in the in cell test for locating particles and probes The tolerance is typically important only if the grid contains cell types with quadrilateral faces i e hexahedra prisms or pyramids and the quadrilateral faces are not planar Typical values should be in the range 1078 to 1079 RocfluMP will check whether the value is appropriate and if necessary increase it 5 2 8 POST Section The POST section contains the following keywords COREFLAG Specifies whether variables for vortex identification are to be written to visual ization files It can take the following values 0 Do not compute variables for vortex identification and do not write to visualization files 5 2 Input File A1 1 Compute variables for vortex identification and write to visualization files See also Sec 9 1 DISCFLAG Specifies whether variables for Schlieren images shadowgraphs and interfero grams are to be written to visualization files It can take the following values 0 Do not compute variables for Schlieren images shadowgraphs and interferograms and do not write to visualization files 1 Compute variables for Schlieren images shadowgraphs and interferograms and write to visualization f
3. iPatchGlobal INTEGER nBFaces nBTris nBQuads nBVert nBVertEst INTEGER DIMENSION POINTER bf2bg bf2c bv bvTemp INTEGER DIMENSION POINTER bf2cs bf2ct bf2v bf2vl bTri2v amp bTri2v1 bQuad2v bQuad2v1 LOGICAL movePatch smoothGrid REAL RFREAL DIMENSION POINTER gs REAL RFREAL DIMENSION POINTER bvn dXyz fc fn REAL RFREAL DIMENSION POINTER bfgwt CHARACTER CHRLEN bcName ifdef GENX INTEGER DIMENSION POINTER bFlag bcFlag REAL RFREAL DIMENSION POINTER mdotAlp pf qc qr rhofAlp tempf amp tflmAlp REAL RFREAL DIMENSION POINTER duAlp nfAlp rhofvfAlp tracf amp xyz endif TYPE t_bcvalues valMixt valTurb valSpec valPeul valRadi TYPE t_bcvalues POINTER valPlag TYPE t_tile_plag tilePlag TYPE t_buffer_plag bufferPlag END TYPE t_patch Figure 14 8 Definition of boundary data structure 154 Chapter 14 Data Structures nBVertEst is the estimated number of vertices on a boundary patch It is used only in the construction of the boundary vertex list bf2bg is an access array which maps a face of a patch to an address in the array bGradFace contained in the mixture data type t mixt It is needed because the boundary face gradients of all boundary patches are stored in a single array for convenience bf2ct is an access array which maps a given boundary face to a cell type It can only take the values CELL TYPE TET CELL TYPE HEX CELL T
4. e casename time stamp plt for unsteady flows where time stamp is a string of the form n nnnnnnnEtnn denoting time 102 Chapter 9 Visualization Table 9 1 Names and meaning of solution variables written to TECPLOT files Variable Meaning Units r Density kg m ru x component of momentum kg m s rv y component of momentum kg m s rw z component of momentum kg m s rE Total internal energy J m p Pressure Pa T Temperature K a Speed of sound m s 9 2 2 File Content 9 2 2 1 Solution Variables rflupost writes all conserved and dependent variables into the TECPLOT files In the TEC PLOT file the variables are denoted by dedicated names as listed in Table 9 1 9 2 2 2 Plotting Variables Plotting variables are labelled PVnn where nn is a two digit integer starting with 01 02 and so on The meaning of a given plotting variable is dependent on the user input for the flags DISCFLAG VORTFLAG COREFLAG and PEULFLAG Each variable is numbered consecutively depending on whether the associated flag is active 9 2 2 3 Patch Coefficient Variables 9 2 3 Zone Naming Conventions TECPLOT organizes data by zones Each zone is assigned a name and may consist of either volume or surface data Because TECPLOT allows each zone to be activated displayed or deactived not displayed separately the separation of data into zones is useful in allowing detailed and selective investigation of data For this reason the T
5. nl s 12 2 The centroid of a triangular face is computed from o 1 wez ri r2 r3 12 3 For a quadrilateral face the centroid is given by the average of the average centroids obtained by subdividing the face into two triangles 121 122 Chapter 12 Algorithms and Methods 12 1 2 Computation of Volume Properties The volume properties of interest are the volume and its centroid For a polyhedron composed of N triangular faces the volume may be computed from N 1 c i 1 The centroid of a polyhedron composed of N triangular faces may be computed from N 1 jue Gin r5 12 5 jm For polyhedra with quadrilateral faces Eqs 12 4 and 12 5 can be applied given that they only involve face properties which have already been computed It is interesting to note that Eq 12 5 expresses the volume centroid as a weighted sum of face centroids and that the weights are not guaranteed to be positive Positive weights can be guaranteed by first computing an approximate cell centroid e as the average of the vertex position vectors and then replacing r n by rf Te n in Eq 12 5 12 2 Spatial Discretization 12 2 1 Stencil Construction Explicit stencils only need to be constructed for the interpolation and gradient operators The minimum extent or size of these stencils is determined by their order of accuracy For an interpolation operator of order p on an arbitrary grid the minimum extent is given by
6. 12 50 12 3 3 3 Outflow Boundary Conditions Outflow boundary conditions discussed in this section are the ones where pressure is specified Outflow can be subsonic or supesonic subsonic outflow can be reflecting or non reflecting At a generic boundary face u has to be positive for it to be outflow un gt 0 means flow is outgoing Subsonic Outflow For subsonic outflow u lt c one characteristic corresponding to A1 is entering the domain and 4 characteristics A2 Az A4 As are leaving the domain see Fig 12 4 This implies La L3 L4 Ls would always be computed from interior solution while L need to be specified La L3 La L5 can be computed as given by equation 12 34 La A ez x Ox Ox Ov pe 3 A3 Ow Li 4 o o o Ls As 2 23 132 Chapter 12 Algorithms and Methods X ue We subsonic outflow boundary oe u gt Oand lul lt c Figure 12 4 Subsonic outflow boundary because p is imposed at outflow energy equation need not be solved at the boundary Subsonic Reflecting Outflow For reflecting subsonic outflow LODI relations for pressure can be used to express unknown wave amplitude L in terms of Ls and boundary condition as follows dp Lire 9 12 51 1 5 25 Subsonic Non Reflecting Outflow For non reflecting subsonic outflow all incoming wave amplitudes are set to zero L 0 12 52 This is perfectly non reflecting outflow However sometimes little reflec
7. These equations can be written in any orthogonal coordinate reference frame Their form remain invariant lets consider a reference frame n s r instead of z y z flow velocities would now be denoted by Un Us Ur instead of u v w which refers to velocity components in n s r coordinate directions Euler equations are now Op Opu Opts Opur 5 n ds Or a 1223 pee e quam ee quam 7 4 P i 0 12 27 Consider a boundary such that normal direction n is perpendicular to face and s r di rection are tangential to boundary Using characteristic analysis to modify the Hyperbolic terms of Euler equations 12 23 12 27 corresponding to waves propagating in the n direc 12 8 Boundary Conditions 127 tion Euler equations can be written in following form Op Opus Opu Rd cur MEE c 12 28 OpUn OpUnUs OPUnUr ae P HEEL a am 12 29 Opus Opusus Opusur Op Ot Beant pda Os Or s 12 30 opur TE purus OpUyUr _ 0p Ot Pt 9s Or ms 12 31 OpE 1 5 2 2 d OpHu OpHu 5r 5 Fu 4 Uds Dat CUR T pUL E pit dy bte or 0 12 32 The vector d which contains normal derivative terms 9 0n is given by characteristic analysis as follows OpUn La Ls L On d 2 27 5 1 NUM q Op da L Li n z n d d ins T 12 33 1 eth n y 20 ds dis nn La our d On In n s r coordinate frame L s can be written as O A On P com Li
8. and how the regions are mapped to the processors This file is only required for parallel computations If the mapping file does not exist rflupart assumes that a serial computation will be made 86 Chapter 7 Execution e A file specifying the mapping between the boundary patches used during grid genera tion and how these patches translate to the patches to be used in RocfluMP This file is only needed if a VGRIDns MESH3D or Cobalt grid file is read The file is called lt casename gt vgi lt casename gt mgi or lt casename gt cgi depending on whether the grid file is in VGRIDns MESH3D or Cobalt format 7 5 3 Output Files The following output files are written by rflupart e A grid file in RocfluMP format For parallel computations the number of grid files written out depends on the number of regions specified in the mapping file e A dimension file For parallel computations the number of dimension files written out depends on the number of regions specified in the mapping file e A version file called rflupart vrs It contains the version number and date of the executable Successive runs append to the version file 7 5 4 Batch Job Submission Guidelines Because rflupart is a serial code the partitioning of large grids can be memory intensive The memory required by rflupart is approximately independent of the number of partitions provided that the number of partitions is not too small This is because rflupart store
9. eo ec wee o o c Ro A 163 151123 Bunnge Fangs ed dus Neck Kee A 163 16 File Content and Format Specifications 167 I I 43d Pe onm aca Boedo m XR sev BR nU YR RAE eO ie Rem Xue dg 167 16 1 1 RocfluMP Grid File eo ses cra RR 167 16 1 2 CENTAUR Grid File 22x ak de 4e yk R33 93 x 3 169 16 1 3 VGRIDns Gad Flea ias ox b Hed Be Pewee wa ELGG 169 IE ol be File oir heec R a E Eam Yex cx 169 EA Fil sa ek sue woo EDS EORR E ORO RUE X A 169 15 14 MESH3D Grid Fil o we ke a Rue 433 EEL AAA 169 16 15 TEIMESM Gad Pile aca sa kaa HEE ns Ex mox 169 11 cibexte Pile 2 colors Boy we eee ee le a A 169 16 1 6 Cobalt Grid File 246 ne ee oo 9 o om o RR on 169 16 2 Flow 5olgtion File aede eaa RR xxi 170 16 3 Dimension File 0 0 00 0 ce a aa 171 16 4 Cel Mapping FA Lo ones ok SB Ee eh A eee REO ee es Res 173 16 5 Renumbering Fil sa es ras ee AAA AR A 173 16 6 Communication Lists File o a scs oc sa d dauca a ddar a cres 173 References 175 10 Contents Notation total energy per unit mass energy source from particles energy source from smoke subgrid scale energy flux vector of external volume forces momentum source from particles momentum source from smoke vector of convective fluxes vector of viscous fluxes energy flux due to radiation total stagnation enthalpy thermal conductivity coefficient mass source from particles components of unit normal vector in z y z direction static pressure heat
10. information file is written by rfluprep 5 7 Convergence File 71 The restart information file allows jobs to be restarted automatically without user in tervention This is particularly useful when running on computers which require jobs to be submitted through batch queues 5 7 Convergence File The convergence file is called lt casename gt con and is written in ASCII format For steady flows the convergence file contains the following information Column 1 The iteration number Column 2 The residual Column 3 The z component of the net force on solid walls Column 4 The y component of the net force on solid walls Column 5 The z component of the net force on solid walls Column 6 The mass flow entering the solution domain Column 7 The mass flow exiting the solution domain For unsteady flows the convergence file contains the following information Column 1 The time Column 2 The time step Column 3 The z component of the net force on solid walls Column 4 The y component of the net force on solid walls Column 5 The z component of the net force on solid walls Column 6 The mass flow entering the solution domain Column 7 The mass flow exiting the solution domain 5 8 Probe File The probe file is written by rflump if the input file contains the PROBE section and setting the variable NUMBER to an integer greater than O The name of the probe file is casename prb mmmmm where mmmmm is the number of th
11. 1 applies to patch ONTIME 1 0E6 time to start using this TBC OFFTIME 1 0E6 time to stop using this TBC ORDER 0 O piecewise constant default 1 piecewise linear NJUMPS 4 number of points at which behavior changes FRAC 0 0 fraction of input value of variable before first time TIME 0 001 first time at which behavior changes FRAC 0 1 next fraction attained constant or ramped to linear oN O 0 BF U N 66 Chapter 5 File Content and Format Specifications TIME 0 002 second time at which behavior changes FRAC 0 3 TIME 0 003 FRAC 0 6 TIME 0 004 final time at which behavior changes FRAC 1 0 final value for constant case ignored for linear case Line 2 Specifies that the variable MFRATE on an INJECT boundary is to be modified Note that this does not yet specify which injection boundary is to be modified Line 3 Specifies that variables on patch one are to be modified Line 4 Specifies the lower bound on the time window in which the values are to be modified Line 5 Specifies the upper bound on the time window in which the values are to be modified Line 6 Specifies the polynomial order of interpolation It can take the following values O For piecewise constant interpolation 1 For piecewise linear interpolation Line 7 Specifies the number of data points through which the time dependent behavior is specified Lines 9 17 Specify the behaviour of the user specified variable i
12. 29P Op Es e On x D bes um 12 34 L4 a bs M n 128 Chapter 12 Algorithms and Methods Boundary Interior Exterior Finite Volume Finite Difference Formulation Formulation Figure 12 2 Finite difference at boundary where A s are defined as i Un C Az Un As d 4 Un 12 35 As Un As Un FC Care should be given to the fact that at the interior of domain finite volume formulation is used Flux integration for interior domain uses cell averaged quantity see equations 12 36 12 37 While at the boundary flow equations are solved using Finite difference formulation and are solved for actual variables on boundary face see Fig 12 2 e f dV v nds 0 12 36 t Ja Jen y d v nds 0 12 37 dt Jao where 5 fo odV is the average value of over cell The Local One Dimensional Inviscid LODI Relations At any point on the bound ary we can obtain a LODI system by considering the system of Eqs 12 23 12 27 and ne glecting transverse terms setting s and r direction terms to zero This is one dimensional Euler equation in n direction In terms of primitive variables LODI relation can be written 12 8 Boundary Conditions 129 as 2b 5 Es La 12 38 op Ls A S 12 39 in zs E 12 40 as E 12 41 AL zo 12 42 These relations may be combined to express the time derivatives of all the other quantities of interest e g T
13. 5 1 2 Antibes Model The viscosity is computed from Tes 120K 3 T 2 Ta re Href Fret if To 190K n Tref T Sret if T lt 120K 1120 159 1 120 11 6 Boundary Conditions 119 Ts lt 120K T Href T if T lt 120K ref u T 2 120 Sree f T gt 120K pas 5 SS This formula was specified by the organizers of the Workshop on Hypersonic Flows for Reentry Problems 1 Sret is given the value 110 K and 1120 denotes the value of the viscosity at 120 K 11 5 2 Conductivity 11 6 Boundary Conditions 120 Chapter 11 Governing Equations Chapter 12 Algorithms and Methods 12 1 Geometry Definition RocfluMP uses a method originally due to Bruner 4 and improved by Wang 16 to compute geometrical properties of faces and volumes The method is particularly convenient for finite volume schemes because volume properties are expressed in terms of face properties This means that the face and volume properties can be computed in a single loop over faces 12 1 1 Computation of Face Properties The face properties of interest are the normal vector the area and its centroid For a triangular face the scaled normal vector is given by 1 n 2 rio X 93 12 1 where r r r For a quadrilateral face the normal vector is given by the average of the average normal vectors obtained by subdividing the face into two triangles The area of triangular and quadrilateral faces follows from Eq 12 1 as S
14. Control File 73 Line 3 The directory name containing the rflump output files relative to the directory from which GENx is invoked written by Roccom Line 4 The verbosity level Line 5 The checking level 74 Chapter 5 File Content and Format Specifications Chapter 6 Problem Setup This chapter gives a summary of the problem setup The objective is to assist users in setting up a run especially for coupled runs within GENx 1 Generate a grid The names and number of output files will differ depending on which grid generator is used CENTAUR Output file is hyb asc or hyb bin Rename the output files to lt casename gt hyb asc or lt casename gt hyb bin VGRIDns Output files are bc and cgosg Rename bc to casename vbc This is important because rflump uses the extension bc for its boundary condition file Rename cgosg to lt casename gt cgosg Generate a patch mapping file lt casename gt vgi as described in Sec 5 3 1 MESH3D Output file is m3d Rename to casename m3d Generate a patch mapping file lt casename gt mgi as described in Sec 5 3 2 TETMESH Output file is noboite Rename to lt casename gt noboite Gener ate a patch mapping file lt casename gt tmi as described in Sec 5 3 3 GRIDGEN Specify Cobalt as the output format Output files are bc and inp Because these extensions are used by rflump it is recommended that you re name these files as COBALT bc
15. RFLU_ModGrid F90 contains some data structures used in the conversion of exte rior grid formats to that used in RocfluMP and some data structures used in the generation of other data structures 14 4 Boundary Data Structure An overview of the boundary data structure is given in Fig 14 7 The definition of the boundary data structure is shown in Fig 14 8 The components of the user defined type for the boundary data structure are defined as follows bcType is the type of the boundary patch It is used to identify which boundary conditions is to be set on that boundary patch bcCoupled is a flag indicating whether the boundary patch is coupled to another code It can have the values BC NOT COUPLED BC NOT BURNING and BC BURNING defined in ModParameters F90 iPatchGlobal is the global index of the boundary patch For serial computations iPatchGlobal is equal to the index of the boundary patch The variable is needed to access the correct boundary condition information when reading the boundary condition file nBFaces is the total number of triangular and quadrilateral faces on a boundary patch nBTris is the number of triangular faces on a boundary patch and is read from the grid file nBQuads is the number of quadrilateral faces on a boundary patch and is read from the grid file nBVert is the number of vertices on a boundary patch 14 4 Boundary Data Structure 153 TYPE t patch INTEGER bcType bcCoupled INTEGER
16. The attributes of burning panes are defined in Table 15 5 163 164 Chapter 15 GENXx Integration Table 15 1 Definitions of volume grid pane attributes Attribute Meaning Units Type nc Coordinates m Real pconn Pane connectivity Integer T4 real Connectivity of actual tetrahedra Integer T4 virtual Connectivity of virtual tetrahedra Integer H8 real Connectivity of actual hexahedra Integer H8 virtual Connectivity of virtual hexahedra Integer W6 real Connectivity of actual prisms Integer W6 virtual Connectivity of virtual prisms Integer P5 real Connectivity of actual pyramids Integer P5 virtual Connectivity of virtual pyramids Integer Table 15 2 Definitions of surface grid pane attributes Attribute Meaning Units Type bcflag Boundary condition flag Integer patchNo Patch number Integer constr type Constraint type Integer nc Coordinates m Real t3 real Connectivity of actual triangles local numbering Integer t3 virtual Connectivity of virtual triangles local numbering Integer t3g real Connectivity of actual triangles global numbering Integer t3g virtual Connectivity of virtual triangles global numbering Integer q4 real Connectivity of actual quadrilaterals local numbering Integer q4 virtual Connectivity of virtual quadrilaterals local numbering Integer q4g real Connectivity of actual quadrilaterals global numbering Integer q4g vir
17. a general method This will in general lead to well balanced but not perfectly balanced partitions 2 Partition the grid with an imposed mapping to get perfect load balancing Note that this will work only if the cells in the grid are numbered like the cells in a structured grid and if the number of regions is a divisor of the number of cells of the unpartitioned grid Furthermore it is important to note that the ordering of the cells will have a very strong impact on the resulting partitioning see Fig 5 3 Depending on the numbering it is possible to generate non contiguous partitions and or partitions with large surface to volume ratios o 0 N O Oc BR Ww M n m o 5 2 Input File 45 5 2 10 PROBE Section The PROBE section consists of two subsections separated by the character The first sub section contains the following keyword NUMBER Specifies the number of probes Immediately following the keyword NUMBER there must be n lines where n NUMBER Each line must contain three real values which represent the r y and z coordinate of a given probe rflump attempts to find the cell whose centroid is closest to the specified coordinates The second part contains the following keywords WRITIME Offset in seconds at which data is written to probe files WRIITER Offset between iterations at which data is written to probe files OPENCLOSE Specifies whether probe files are to closed and opened after writing data It can
18. by the interpolation process TECPLOT and ENSIGHT allow the visualization of cell centered solution data 1 Use simple arithmetic averaging 2 Use k exact averaging This method interpolates polynomials of order k exactly on arbitrary grids The order k is specified by the value assigned to the keyword INTERORDER Choosing between INTERTYPE 1 and INTERTYPE 2 is largely a matter of balancing the accuracy and cost of the interpolation Simple arithmetic averaging is much faster but also less accurate than the k exact interpolation method In practice the differences between the two methods are usually negligible unless solution gradients are very large or grids are highly distorted MERGEFLAG Specifies whether the regions from a parallel computation are to be merged for postprocessing It can take the following values 0 Do not merge the regions for postprocessing Each partition will be written sepa rately to the output file Virtual cells and boundary faces will be written sepa rately for each region and patch respectively 1 Merge the regions for postprocessing Not merging regions for postprocessing can be useful for several reasons First one may want to make sure that solution contours are well behaved across partition boundaries Second when using this option in conjunction with rflupick the amount of data to be visualized can be reduced drastically by selecting only specific regions to be postpro cessed in rflupost Third it
19. can take the following values 1 Patch is interacting 2 Patch is not interacting CRECONST Specifies whether constrained reconstruction is to be used when computing face gradients for this boundary It can take the following values 0 Do not use constrained reconstruction default 5 4 Boundary Condition File 59 1 Use constrained reconstruction HFLUX Specifies the value of the heat flux W m K KIND Specifies the treatment of the boundary condition It can take the following values 0 Simple boundary condition treatment 1 Characteristic boundary coundition treatment NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers STATS Specifies whether particle impact statistics are to be gathered It can take the following values 0 Do not gather pati
20. coincide with its related patch It can take the following values 1 z coordinate direction 2 y coordinate direction 3 z coordinate direction NAME Specifies the name of the boundary PATCH Specifies the range of patches to which the data in this section is to be applied using two integers 5 4 Boundary Condition File 63 iv IA RARA VA ORENA RARA Figure 5 6 Definition of rotation angle for periodic boundaries In this example the rotation angle for the purple patch would be positive while that for the orange patch would be negative RELPATCH Specifies the patch to which the current patch is related 5 4 1 9 Slip Boundary BC_SLIPW Section The BC_SLIPW section contains the following keywords COUPLED Specifies whether patch is interacting during a computation with GENx It can take the following values O Patch is interacting 2 Patch is not interacting CRECONST Specifies whether constrained reconstruction is to be used when computing face gradients for this boundary It can take the following values 0 Do not use constrained reconstruction default 1 Use constrained reconstruction KIND Specifies the treatment of the boundary condition It can take the following values 64 Chapter 5 File Content and Format Specifications O Simple boundary condition treatment 1 Characteristic boundary coundition treatment NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of c
21. contains coefficients used in the Runge Kutta scheme trk contains coefficients used in the Runge Kutta scheme betrk contains coefficients used in the explicit multistage scheme iniVelX contains the z component of the velocity vector for the initial condition It is only used in rfluprep and written into the solution file iniVelY contains the y component of the velocity vector for the initial condition It is only used in rfluprep and written into the solution file iniVelZ contains the z component of the velocity vector for the initial condition It is only used in rfluprep and written into the solution file iniPress contains the static pressure for the initial condition It is only used in rfluprep and written into the solution file iniDens contains the density for the initial condition It is used only in rfluprep and written into the solution file unifDens contains the density value when checking RocfluMP for uniform flow preserva tion The check for uniform flow preservation is activated by compiling RocfluMP with CHECK UNIFLOW 1 unifEner contains the total internal energy value when checking RocfluMP for uniform flow preservation The check for uniform flow preservation is activated by compiling RocfluMP with CHECK UNIFLOW 1 14 5 Mixture Data Structure 159 unifMomX contains the x component of momentum when checking RocfluMP for uniform flow preservation The check for uniform flow preservation is activated by compil
22. loop over interior faces nBFaces is the total number of triangular and quadrilateral faces on all boundary patches nBQuads is the total number of quadrilateral faces on all boundary patches nBTris is the total number of triangular faces on all boundary patches nCells is the number of interior cells in the grid nCellsTot is the total number of cells in the grid i e interior and dummy cells nEdges is the number of interior edges in the grid nEdgesEst is the estimated total number of edges in the grid It is used only in the con struction of the edge list nEdgesTot is the total number of edges in the grid i e interior and dummy edges nFaces is the number of interior triangular and quadrilateral faces in the grid nFacesEst is the estimated total number of interior triangular and quadrilateral faces in the grid It is used only in the construction of the face list nFacesTot is the total number of triangular and quadrilateral faces in the grid i e interior and dummy faces nHexs is the number of interior hexahedral cells in the grid nHexsTot is the total number of hexahedral cells in the grid i e interior and dummy hex ahedral cells nPris is the number of prismatic cells in the grid nPrisTot isthe total number of prismatic cells in the grid i e interior and dummy prismatic cells nPyrs is the number of pyramidal cells in the grid nPyrsTot is the total number of pyramidal cells in the grid i e interi
23. nPrisTot nPyrs nPyrsTot amp nTets nTetsTot nVert nVertTot INTEGER DIMENSION POINTER hexFlag hex2CellGlob amp priFlag pri2CellGlob pyrFlag amp pyr2CellGlob tetFlag tet2CellGlob amp vertFlag v2c INTEGER DIMENSION POINTER cellGlob2Loc c2cs e2v e2vTemp f2c amp f2cTemp f2cs f2v f2vTemp hex2v pri2v amp pyr2v tet2v v2cInfo REAL RFREAL DIMENSION POINTER fc fn REAL RFREAL DIMENSION POINTER cgwt fgwt Grid Motion INTEGER DIMENSION POINTER degr REAL RFREAL DIMENSION POINTER gs volMin REAL RFREAL DIMENSION POINTER rhs Geometric information REAL RFREAL POINTER xyz REAL RFREAL POINTER vol cofg END TYPE t grid Figure 14 5 Definition of grid data structure 14 3 Grid Data Structure 147 The components of the user defined type t grid are defined as follows indGs is a flag used to allocate the array for the grid speeds If grid motion is active the grid speeds need to be computed and hence indGs 1 otherwise indGs 0 This allows the grid speed array gs see below to be accessed even if grid motion is not active which simplifies the code because conditional statements can be avoided The array gs will typically be accessed through a statement such as gs indGs ifc where ifc is an integer variable used in a
24. number The source term is given by m Ba T Q pleut H EAI 11 10 Beetle m pfe U d EP E where m f f EP and E represent the source terms introduced by the particle and smoke modeling The vector of external volume forces reads fe gG a 11 11 with g being the gravitational acceleration and a the acceleration of the rocket 11 2 The Geometric Conservation Law For uniform flow Eq 11 1 reduces to the Geometric Conservation Law GCL x 1 V dS 0 11 12 Ot Jo an 118 Chapter 11 Governing Equations The GCL ensures that the motion of the grid does not alter a uniform flow It must be satisfied in a discrete sense independent of the deformation of the grid and the numerical solution method The discretization of the GCL in RocfluMP is described in Sec 12 5 2 11 3 Gas Models 11 3 1 Calorically and Thermally Perfect Gas 11 3 2 Mixture of Calorically and Thermally Perfect Gases 11 3 3 Pseudo Gas 11 4 Thermodynamic Properties 11 4 1 Calorically and Thermally Perfect Gas 11 4 2 Mixture of Calorically and Thermally Perfect Gases 11 4 3 Pseudo Gas 11 5 Transport Properties 11 5 1 Viscosity 11 5 1 1 Sutherland Model The viscosity is computed from 3 H T ref Srer Href T wit T S ref where y is the dynamic viscosity Href is the reference dynamic viscosity T is the static temperature Tres is the reference temperature and Syer is the Sutherland constant 11
25. oe a EER OER ESSA MERE eR ESR SOS 118 11 5 Transport Properties nc ce cs zoo DEER nd Ea ita RR DES 118 EA S2 ke kk ea ewe OR up e ee ee Re eee 118 11541 Sutherland Model oc dca wm RR RR eR o ee eS 118 115 12 Antibes Model 2 Xs 118 IRE es cn 5 esc ccrann asp ka ied kapt er ae Ge SE eR 119 110 Boundary Condong ad do Vb are OEC PAR eee RU ES 119 12 Algorithms and Methods 121 12 1 Geometry Definition a uc om RE opu o ERR AR RARA eM 121 12 1 1 Computation of Face Properties 2222s o n ee 121 12 1 2 Computation of Volume Properties 122 EFE HARDER eco mara ck devo E ee OREO E LEO a 122 1242 1 Stencil Cometic lt s ns o RU Roy ee EERE Se EES 122 12 2 2 Interpolation Operators coco oon ee ee ee ea URS 123 12 2 3 Gradient CIO ie a Ee eee EE eRe Y ox eR Y uo 124 122 0 miad Flukes 22 2022229 o Ro kem RR Rh 125 12 24 Liw ter PunctbHS ec es eso c ore Roos 125 12347 Numerical Flox Functions ociosas ii 125 122 3 Entropy Fixes lt s besss recse e OR TRU X pD ew eS 125 1235 VICO FUS lt e o ece x amo oko e Web qo wee P den 125 12 26 Optimal LES Discrotirablon e coce pos RR ee ee hE G 125 122 51 Computation of Integrals ooa aa RS 125 12 2 6 2 Computation of Stencil Weights 125 IDA aH ne bu e hee Qe RO Y QUE Borse one eh dy 125 123 Honey LOREM ono gans Ke eon E Redi OSS a 125 12 3 1 Simple Fluid Solid Boundary Conditions 125 125 11 Slp Wall Boundar
26. of boundaries This may be reflected in the persistence of inflow at an outflow boundary despite lowering the static pressure the persistence of outflow at an inflow boundary despite increasing the stagnation pressure or recirculation regions across outflow boundaries In these cases it is necessary to reposition the relevant boundary patches These warnings will be accompanied by additional output which provides more infor mation about the precise location of the boundary faces with reverse flow Section 10 5 describes how this output can be used to visualize the cells affected by reverse flow 10 3 Explanations of Errors Absolute difference in volumes larger than specified limit rflump computes the total volume of the computational domain using two methods The first method simply sums the volumes of all the cells in a given region The second method computes the total volume from the boundary and interregion faces If the two methods disagree by more than a presently hardcoded tolerance it is likely that the interior grid in invalid and the above error message is printed Invalid quantity detected rflump detected solution variables with the value NaN This indicates a serious problem This is often the result of using a CFL number or a time step which is too large For coupled simulations invalid variables may also arise if the geometry becomes heavily deformed This error message may also indicate problems with the specification of bou
27. solver Jim Ferry Time dependent boundary conditions initial implementation of Equilibrium Eulerian method in Rocspecies Robert Fiedler Technical Program Manager CSAR Testing application of RocfluMP to fluid structure interaction problems using GENx Andreas Haselbacher Principal Research Scientist CSAR Lead developer of RocfluMP Xiangmin Jiao Research Scientist CSAR Assistance with integration of RocfluMP into GENx Fady Najjar Senior Research Scientist CSAR Lead developer of Rocpart module inte gration of Rocpart with RocfluMP testing verification and validation Adam Moody Lawrence Livermore National Laboratory Assistance with tuning of RocfluMP Manoj Parmar Graduate Student Department of Aeronautical Engineering UIUC Val idation for shock diffraction over cylinders and spheres implementation of Schlieren and Interferogram image capability in rflupost implementation of limiter functions 18 Chapter 1 Introduction Charles Shereda Lawrence Livermore National Laboratory Assistance with tuning of RocfluMP 1 4 Overview of the RocfluMP Book The RocfluMP book is divided into three parts Part I which you are currently reading is designed to provide preliminary information so that users get an overview of RocfluMP become familiar with the nomenclature conventions and restrictions of RocfluMP and can install and compile RocfluMP on their computer Part II contains information for users of Rocfl
28. source source term surface element time Cartesian velocity components velocity vector contravariant velocity vector of conservative variables Cartesian coordinates dynamic viscosity coefficient density viscous stress control volume boundary of a control volume 11 12 Contents Part I Preliminaries 13 Chapter 1 Introduction 1 1 Objective The objective of the RocfluMP book is four fold 1 To enable users to compile and install the RocfluMP source code on computer systems 2 To enable users to run the RocfluMP code for the example cases 3 To enable users to run the RocfluMP code for their own cases 4 To enable users to develop RocfluMP to suit their own needs 1 2 Overview of RocfluMP RocfluMP solves the three dimensional time dependent compressible Navier Stokes equations on moving and or deforming unstructured grids The grids may consist of arbitrary combi nations of tetrahedra hexahedra prisms and pyramids The spatial discretization is carried out using the finite volume method The inviscid fluxes are approximated by upwind schemes to allow for capturing of shock waves and contact discontinuities Steady flows can be com puted using an explicit multi stage method tuned for fast convergence Unsteady flows are computed with the fourth order accurate Runge Kutta method To solve fluid dynamics problems in which processes other than those described by the Navier Stokes equations are importan
29. take the following values 0 Do not close and open probe file after writing data 1 Close and open probe file after writing data Closing and opening the probe file after writing data can be useful because this forces write buffers to be flushed An example PROBE section for unsteady flow is given below PROBE NUMBER 4 0 001 0 000 0 499 0 001 0 000 0 499 9 999 0 000 0 499 9 999 0 000 0 499 WRITIME 5 0E 4 OPENCLOSE 1 46 Chapter 5 File Content and Format Specifications 5 2 11 REFERENCE Section The REFERENCE section contains the following keywords ABSVEL Reference velocity magnitude m s CP Reference specific heat coefficient at constant pressure J kg K DENS Reference density kg m GAMMA Reference ratio of specific heats LENGTH Reference length m PRESS Reference static pressure Pa PRLAM Reference laminar Prandtl number PRTURB Reference turbulent Prandtl number RENUM Reference Reynolds number SCNLAM Reference laminar Schmidt number SCNTURB Reference turbulent Schmidt number 5 2 12 TIMESTEP Section The TIMESTEP section contains the following keywords DTMINLIMIT The time step in seconds below which information will be printed out about the region and cell in which the minimum time step occurs The additional information can be helpful in diagnosing whether small time steps are due to unexpectedly small cells or cells of poor quality Once
30. the patch numbers are global patch numbers Global patch number 5 corresponds to local patch number 3 in region 1 9 2 3 3 Border Face Zones The names of border face zones are INT face type region index where 9 2 TECPLOT Output 105 v Zone Style Mesh Contour Vector Scatter Shade Bound Mesh Mesh Zone Show Zone Zone Zone Num Name Grp 1 HEX A 00001 2 HEX V 00001 3 PRI A 00001 4 PRI V 00001 5 C 00001247 00001 6 7 8 Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay Overlay F 00000871 PAT 000 00001 F 00000013 PAT 001 00001 INT QUAD 00001 g PAT 001 QUAD A 00001 10 PAT 001 QUAD V 00001 11 PAT 002 QUAD A 00001 12 PAT 005 TRI 00001 13 PAT 005 TRI V 00001 14 PAT 005 QUAD A 00001 15 PAT 005 QUAD V 00001 16 PAT 006 TRI 00001 17 PAT 006 TRI V 00001 18 PAT 006 QUAD A 00001 19 PAT 006 QUAD V 00001 20 HEXA 00002 21 HEX V 00002 22 PRI A 00002 23 PRI V 00002 24 INT QUAD 00002 25 PAT 001 GQUAD A 00002 26 PAT 001 QUAD V 00002 27 PAT 005 TRI A 00002 28 PAT 005 TRI V 00002 29 PAT 005 QUAD A 00002 30 PAT 005 QUAD V 00002 31 PAT 006 TRI 00002 32 PAT 006 TRI V 000
31. upper limit of the original patches does not have to be equal to the number of patches in the m3d file 5 3 3 TETMESH Patch Mapping File tmi File 5 3 4 Cobalt Patch Mapping File cgi File The cgi file serves the same purpose as the vgi and mgi files The format of the cgi file is Line 1 The number of patches after the mapping Line 2 The number of mappings hereafter referred to as nMappings The remaining nMappings lines Each line contains three integers The first two integers represent the lower and upper limits of the patches in the Cobalt file which are to be mapped to the patch indicated by the third integer The lower limit must be less than or equal to the upper limit 5 4 Boundary Condition File The boundary condition file is called lt casename gt bc Like the input file the boundary condition file is divided into sections Each section contains several lines each of which consists of a keyword and a value with the exception of the line containing the keyword PATCH on which two values are listed SECTION NAME PATCH PATCH 1 PATCH 2 KEYWORD 1 VALUE 1 KEYWORD 2 VALUE 2 52 Chapter 5 File Content and Format Specifications Each section assigns a boundary condition to either a single patch or to a range of patches The boundary condition file must be terminated by the string END The sections may be listed in any order in the boundary condition file but are listed below in alphabetical o
32. 0000 grd 00001 grd 00002 grd 00003 grd 00004 grd 00005 The parallel computation is initiated by mpirun np 5 rflump c gamm8 v 2 The precise command line may vary depending on which computer is used The results of the computation will not be shown again Instead the capability of rflupost to allow the user to visualize the partitioned geometry will be demonstrated Generate the TECPLOT file through rflupost c gamm8 s 500 v 2 Figure 8 5 depicts the five regions generated for this particular run More importantly rflupost can be used to visualize individual regions the interregion faces the virtual faces associated with particular regions and boundary patches as well as the virtual cells of any region Some of these capabilities are illustrated in Fig 8 6 8 2 GAMM Bump 97 Figure 8 5 Partitioned grid for parallel GAMM bump computation Figure 8 6 Partitioned grid for parallel GAMM bump computation showing interregion faces as well as virtual boundary faces 98 Chapter 8 Example Cases Chapter 9 Visualization This chapter describes how results obtained with rflump can be visualized The focus is on presenting information which will be helpful in visualizing results obtained specifically with rflump Consult the appropriate manuals for information on how the visualization tools themselves should be used This chapter assumes that the user has executed rflupost already and thereby pro
33. 02 33 PAT 006 QUAD A 00002 34 PAT 006 QUAD V 00002 35 HEXA 00003 36 HEX V 00003 37 PRI 00003 38 PRI V 00003 39 INT QUAD 00003 c 0 0 o oo o oo o o o o ooo oo o o o o o o o o o o Figure 9 2 Screen dump of part of zone style window in TECPLOT illustrating zone naming convention used in rflupost 106 Chapter 9 Visualization face type is a string indicating the face type TRI if the zone consists of triangular cells QUAD if the zone consists of quadrilateral cells region index is a five digit string indicating the global region index An example is shown in Fig 9 2 which represents a screen dump of part of the zone style window in TECPLOT Zone number 8 contains the quadrilateral border faces of region 1 In this particular case the zone contains only quadrilateral faces 9 2 3 4 Special Cell Zones The names of special cell zones are C cell index region index where cell index is an eight digit string indicating the cell index lt region index is a five digit string indicating the global region index Note that the names of special cell zones do not distinguish between cell types and cell kinds An example is shown in Fig 9 2 which represents a screen dump of part of the zone style window in TECPLOT Zone number 5 contains cell number 1247 of region 1 9 2 3 5 Special Face Zones The names of special cell zones are F face index PAT patch index reg
34. 1 Invocation For serial computations rflupart is invoked by typing rflupart c lt casename gt v lt verbosity gt The command line arguments read by rflupart are lt casename gt A character string used to label the input and output files lt verbosity gt An integer indicating the desired verbosity level of rflupart The verbosity level can take the following values O No output rflupart will not write any information to standard output 1 Low level of output rflupart will write some information to standard output 2 High level of output rflupart will write detailed information to standard output 7 5 2 Input Files The following input files are read by rflupart e An input file called lt casename gt inp e A grid file rflupart supports the following formats CENTAUR format The CENTAUR grid file may be in ASCII or binary format The file in ASCII format is called lt casename gt hyb asc and the file in binary format is called lt casename gt hyb bin VGRIDns format The file is called casename cgosg MESH3D format The file is called lt casename gt m3d TETMESH format The file is called lt casename gt noboite Cobalt format The file is called lt casename gt cgr GAMBIT format The file is called casename neu e A boundary condition file called casename bc e A mapping file called lt casename gt map specifying how many processors are to be used for parallel computations
35. 2 Deseriplion et Compilation Oplio s lt s o s RR Rex d II User Manual 4 Capability Descriptions Al Two Dimensional Computations gt e e s zc E Y 3 3 x oet 4 2 Periodic and Symmetry Boundary Conditions 4 3 Mass Pressure Skin Friction and Heat Transfer Coefficient Computation 44 Porte and Moment Computation o es seste e ee wee Des 4 5 Thrust and Specific Impulse Computation 3 11 13 14 14 14 17 18 19 19 19 20 21 21 21 22 22 22 23 Contents 26 Visualization of Discontinwities 2 2 2 2 See Y o RR bees 28 4 7 Visualization and Identification of Vortical Structures 28 File Content and Format Specifications 31 5 1 Filename Conventions 2 2252223 e seee RAO OR eee OA 31 Bo Wipe AA 3l 5l FLUNMOBEL Setting soccer E RARO EERE RES 32 Dons FORCES SOU ck eee ee ees be woe ok EO OX OE FOE EON X os 32 Ge FORUT Seen oe ee a oA A RR E A A OS x 33 5 24 GRIDMOTIDH SECO lt lt mes mom n RR RR PRY EE E REG 34 haa INITFLOW Seclion so eac e de ec Eo OE xoxo AO EOS ORORUE eS e 34 mob MIXTURE GE 3o AA E ERO e E Sos eek ko o 36 Bur NUMERICS Section uro Rok hom 26 do ORDER Ao oe ESE 38 E MI 40 ng PREP SED coa a o A a dnx 44 Be LO PROBE SOHO eo ok ira a Roh Ew Oe EOS Red 45 5 2 11 REFERENCE Section 2 22e o Der a we ee ee Se es 46 BOIS CEP Os uou ak on SA BS eee RO EO deb ee 46 52 13 TRANSFORM Section e ccor ons dee ea deca wh bee d REOR da 48
36. 5213 VISDMNOBEL Depom lt lt a eh ee E OR Heec Oe ees 48 ho Pad epee Fies eca do ok X A Be OE POE eee BSR Bx 49 5 3 1 VGRIDns Patch Mapping File vgiFil 49 5 3 2 MESH3D Patch Mapping File mgi File 50 5 3 3 TETMESH Patch Mapping File tmi File 51 5 3 4 Cobalt Patch Mapping File cgi File 51 of Boundary Condition FHe s se e ess e er pU cR EE Ee 51 5 41 Physical Boundary Conditions 22 5 ee ae ee ene A 52 5 4 1 1 Farfield Boundary BC_FARF Section 52 5 4 1 2 Inflow Boundary BC INFLOW BC INFLOW TOTANG Section 53 5 4 1 3 Inflow Boundary BC INFLOW VELTEMP Section 56 9 4 1 4 Injection Boundary BC INJECT Section 97 5 4 1 5 No Slip Boundary BC_NOSLIP_HFLUX Section 58 5 4 1 6 No Slip Boundary BC NOSLIP TEMP Section 60 5 4 1 7 Outflow Boundary BC OUTFLOW Section 61 5 4 1 8 Periodic Boundary BC PERIODIC Section 62 5 4 1 9 Slip Boundary BC_SLIPW Section 63 5 4 1 10 Symmetry Boundary BC SYMMETRY Section 64 5 4 1 11 Virtual Boundary BC_VIRTUAL Section 65 5 4 2 Time Dependent Boundary Conditions 65 5421 TBC PIECEWISE Seciiom o s a s ag 464 REX Xs 65 542 TEC SINUSOIDAL DOGCLIOH sa RS Remy REOR OX eni 66 O42 TRC STOCHASTIC SecboH 6 be kG oo RR GR EDD ES 68 Contents 5 02 1 TBCWHITENOISE Section iuo
37. 85 TL INR rica ena a rs a a es oe 85 Tone OP PIRE cada e A A ee eee d 85 e REI 86 7 5 4 Batch Job Submission Guidelines 86 RO TUI eo ve RUE a ORE A Ds pee ne ipeo 86 FL IGI c enc a IS A AA AA 86 e AA 88 TOS Output AA 88 164 Interactive DUE 2223 9 X sacs mos mae ex Ewe ar SE ee gd 88 Ta HODGE e bb he RR AUR RA RE ke oko dl Ro dod 90 6 Contents TO ORCA 2223 3 wok kc AA AAA RA 90 fete AME Pile coe hee eunt WORT EGER Ro Rae he Rene RR nk 90 Toa o ARIAS 90 8 Example Cases 91 E BID cp aaa a ee Oe ala E 91 AE 92 8 2 1 Dena Computatio gt e ss s se bra RS ee eo d 94 8 2 2 Parallel Computation o cisco 3x wosa baw SG Be OES Ros 95 9 Visualization 99 Bl Plubur Variables o 24 serr mier amid Aaea E RR AS E EEE ES 99 9 1 1 Visualization of Discontinuities 6 24 sea sos o o Rs 99 9 1 2 Visualization and Identification of Vortical Structures 100 9 1 3 Visualization of Eulerian Particle Fields 101 Bx CECT nasa elc ae REE EERO eee ES 101 8 2 1 File Naming Convention ss se sesede xk GRE RR RR X XA 101 ORA UE MERECE 102 922 1 Sol Vanalles lt sec oa x Gee X RR RRS RE A 102 Bax Plotting Variables lt css cc sux aes Ge ES EG 102 4223 Patel Coellicient Variables 2 46654 54 24644344 102 92 3 Zone Naming Conventions 2 2 8 55268 BRA 102 92o Volume ONES o od a m RR A ROUX 103 Dasa Bouadary P ach Zones o op sone OX Ge ae ee eee es 104 9 2 3 3 Border Face Z
38. 9519E 04 7 6913E 04 7 4307E 04 7 1702E 04 6 6490E 04 Figure 8 4 Static pressure contours for transonic GAMM bump computation With the settings contained in the files from the CVS repository rflump will take 500 steps to reach the convergence tolerance The results may be visualized with TECPLOT The postprocessor is used to interpolate the solution variables from cell centroids to vertices and to write the TECPLOT data file by typing rflupost c gamm8 s 500 v 2 The resulting data file may be read into TECPLOT by typing tecplot gamm8 00000 000500 plt The contours of static pressure are depicted in Fig 8 4 8 2 2 Parallel Computation The first step in preparing for a parallel computation is to decide how many regions should be used For this example we will use five regions and hence type rflumap c gamm8 m 1 p 5 r 5 v 2 which leads to the region mapping file 96 Chapter 8 Example Cases gamm8 map To partition the grid files the partitioner is invoked by typing rflupart c gamm8 v 2 which produces the following files gamm8 dim 00001 gamm8 dim 00002 gamm8 dim 00003 gamm8 dim 00004 gamm8 dim 00005 To initialize the solution the initializer is invoked by typing gamm8 gamm8 gamm8 gamm8 gamm8 rfluinit c gamm8 v 2 which produces the following files gamm8 flo 00001 000000 gamm8 f1lo 00002 000000 gamm8 f1lo 00003 000000 gamm8 flo 00004 000000 gamm8 f1lo 00005 00
39. BNDRY CELL KIND ACTUAL and CELL_KIND_VIRTUAL defined in ModParameters F90 tet2CellGlob contains the mapping of a given tetrahedral cell to a global cell vertFlag contains a flag indicating the kind of a given vertex It is read in from the RocfluMP grid file and can only take the values VERT KIND ACTUAL and VERT KIND VIRTUAL defined in ModParameters F90 cellGlob2Loc contains the mapping of a global cell to the local cell of a given type c2cs contains the cell stencils for each cell T his array is used in computing cell gradients and averaged variables e2v contains the two vertices defining an edge This array is used only for grid motion e2vTemp is a temporary array used to construct e2v f2c contains the two cells adjacent to a face 14 3 Grid Data Structure 149 f2cTemp is a temporary array used to construct f2c f2cs contains the face stencils for each face This array is used in computing face gradients f2v contains the vertices defining a face f2vTemp is a temporary array used to construct f2v hex2v contains the connectivity information for the hexahedral cells The vertices must be numbered as shown in Fig 14 6 a The face to vertex edge to vertex and edge to face connectivity arrays for hexahedral cells are shown in Table 14 1 pri2v contains the connectivity information for the prismatic cells The vertices must be numbered as shown in Fig 14 6 b The face to vertex edge to vertex and edge to
40. CFLU communication lists file WRITE iFile A Dimensions WRITE iFile I8 pGrid nBorders DO iBorder 1 pGrid nBorders pBorder gt pGrid borders iBorder WRITE iFile 2 18 pBorder iRegionGlobal pBorder iBorder END DO iBorder WRITE iFile A Cells DO iBorder 1 pGrid nBorders pBorder gt pGrid borders iBorder WRITE iFile 2 18 pBorder nCellsSend pBorder nCellsRecv WRITE iFile 10 I8 pBorder icgSend icl icl 1 pBorder nCellsSend WRITE iFile 10 I8 pBorderficgRecv icl icl 1 pBorder nCellsRecv END DO iBorder WRITE iFile A Vertices DO iBorder 1 pGrid nBorders pBorder gt pGrid borders iBorder WRITE iFile 3 18 pBorder nVertSend pBorder nVertRecv pBorder nVertShared WRITE iFile 10 I18 pBorder ivgSend ivl ivl 1 pBorder nVertSend WRITE iFile 10 I8 pBorderfivgRecv ivl ivl 1 pBorder nVertRecv WRITE iFile 10 I8 pBorderZivgShared ivl ivl 1 pBorder nVertShared END DO iBorder WRITE iFile A st End References 1 Abgrall R D sid ri J A Glowinski R Mallet M and P riaux J Eds Hypersonic Flows for Reentry Problems Vol III Proceedings of the INRIA GAMNI SMAI Work shop on Hypersonic Flows for Reentry Problems Part II Antibes France April 15 19 1991 Springer Verlag Barth T J and Jespersen D C The Design and Application of Upwind Schemes on Unstr
41. E 1 2 3 Nig Met Hn 12 6 In the present work filter operators are regarded as low order interpolation operators For a gradient operator of order q on an arbitrary grid the minimum extent is given by q l g 2 q 3 For both interpolation and gradient operators it may be advantageous to increase the stencil extent to include more cell centroids than necessary For filter operators this can be used to minimize the imaginary part of the transfer function The larger than necessary support necessitates the use of least squares techniques to determine the interpolated value or gradients Gradient operators are required at cell and face centroids Interpolation operators are required at cell and face centroids and at vertices Hence cell to cell face to cell and vertex to cell stencils must be constructed as depicted schematically in Fig 12 1 Nasi 12 7 12 2 Spatial Discretization 123 2 PB ER E c Figure 12 1 Schematic illustration stencils in two dimension a Cell to cell stencil b face to cell stencil and c vertex to cell stencil At present these stencils are constructed using an Octree based approach The Octree is initialized using cell centroid coordinates and queried with the locations at which the interpolation or gradient operators are to be constructed 12 2 2 Interpolation Operators The interpolation operators are constructed using a least squares approach based on a mod i
42. ECISION 1 0 RFREAL RANGE 1 0_RFREAL 167 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 41 42 43 44 45 46 47 168 Chapter 16 File Content and Format Specifications WRITE iFile A Physical time WRITE iFile E23 16 global currentTime WRITE iFile A Dimensions WRITE iFile 5 18 pGrid nVertTot pGrid nTetsTot pGrid nHexsTot amp pGrid nPrisTot pGrid nPyrsTot WRITE iFile A Coordinates DO i 1 3 WRITE iFile 5 E23 16 pGridhxyz i j j 1 pGrid nVertTot END DO i IF pGrid nTetsTot gt O THEN WRITE iFile A Tetrahedra DO i 1 4 WRITE iFile 10 18 pGrid tet2v i j j 1 pGrid nTetsTot END DO i END IF pGrid nTetsTot IF pGrid nHexsTot gt O THEN WRITE iFile A Hexahedra DO i 1 8 WRITE iFile 10 18 pGrid hex2v i j j 1 pGrid nHexsTot END DO i END IF pGrid nHexsTot IF pGrid nPrisTot gt O THEN WRITE iFile A st Prisms DO i 1 6 WRITE iFile 10 18 pGrid pri2v i j j 1 pGrid nPrisTot END DO i END IF pGrid nPrisTot IF pGrid nPyrsTot gt O THEN WRITE iFile A Pyramids DO i 1 5 WRITE iFile 10 I18 pGrid pyr2v i j j 1 pGrid nPyrsTot END DO i END IF pGrid nPyrsTot WRITE iFile A Boundaries WRITE iFile I8 pGrid nPatches DO iPatch 1 pGrid nP
43. ECK DATASTRUCT 1 Activates checking of data structures This option will print out the content of the important data structures used by RocfluMP Note that activating this option will lead to substantial screen output so it should only be activated for small cases DEBUG 1 Activates debugging compiler options If this option is not specified optimizing compiler options are chosen by default PLAG 1 Activates compilation of Rocpart This option must be specified if you wish to run computations with Lagrangian particles ROCPROF lt Path to Rocprof library Activates profiling of selected routines in RocfluMP Note that activating this option entails an overhead so it should not be used for production runs SPEC 1 Activates compilation of Rocspecies This option must be specified if you wish to run computations with chemical species and or Equilibrium Eulerian particles 24 Chapter 3 Installation and Compilation Part II User Manual 25 Chapter 4 Capability Descriptions This chapter describes the capabilities of RocfluMP and points the user to the relevant sections in Chapter 5 4 1 Two Dimensional Computations RocfluMP can compute two dimensional flows Grids for two dimensional computations may contain only a single plane of prismatic and or hexahedral cells in the z coordinate direc tion In other words each cell must have two faces on patches with z constant A virtual boundary condition as specified by the BC V
44. ECPLOT files produced by rflupost make extensive use of zones The following explains the zone naming conventions used by rflupost The zone naming convention adopted by rflupost is best explained with reference to Fig 9 1 Data written to the TECPLOT file is split into volume and boundary patch data Volume data is further split according to the cell type tetrahedron hexahedron prism or pyramid and for each cell type according to the cell kind actual or virtual cell Boundary patch 9 2 TECPLOT Output 103 Region Volume Boundary patches Borders Special objects Y Y Y Cell type Face type Face type Cell s Cell kind Face kind A Face s Zones in Tecplot file Figure 9 1 Illustration of zones written to TECPLOT files by rflupost data is further split according to the face type triangular or quadrilateral and for each face type according to the face kind actual or virtual face 9 2 3 1 Volume Zones The names of volume zones are cell type gt lt cell kind gt _ lt region index where lt cell type gt is a four letter string indicating the cell type TET if the zone consists of tetrahedral cells HEX if the zone consists of hexahedral cells PRI if the zone consists of prismatic cells PYR if the zone consists of pyramidal cells
45. IRTUAL Section must be applied to these patches Two dimensional computations are activated by the keyword DIMENS in the NUMERICS Sec tion A sample grid which could be used for two dimensional computations with RocfluMP is shown in Fig 4 1 Figure 4 1 Sample grid which can be used for two dimensional computations with RocfluMP 4 2 Periodic and Symmetry Boundary Conditions 27 4 2 Periodic and Symmetry Boundary Conditions RocfluMP can impose periodic and symmetry boundary conditions The periodic boundary condition is general and allows both linear and rotational periodicity to be imposed thus allowing channel flows or flows in turbomachinery geometries see e g Fig 4 2 to be computed The imposition of periodic and symmetry boundary conditions is described in the BC_PERIODIC and BC_SYMMETRY sections Figure 4 2 Generic turbine guide vane with leading edge cooling Such geometries can be com puted with RocfluMP using periodic boundary conditions Inset shows detail view of cooling holes at leading edge 28 Chapter 4 Capability Descriptions 4 3 Mass Pressure Skin Friction and Heat Transfer Coefficient Computation RocfluMP can compute mass pressure skin friction and heat transfer coefficients for faces on patches Definitions of these coefficients can be found in Section 12 6 4 4 Force and Moment Computation RocfluMP can compute forces and moments exerted by the fluid on the patches Definitions of the
46. Sret i Viet e Pu Siret F U S Cn vi Cre a oer E JE 12 62b ue 5 Pret V2tSrer Vest de dd Dat l Fj Wi Si C m A Cmi C i zi Cro Fo 12 62c i 5 Prot ep Sref Viet i x m Sref 12 7 Force and Moment Computation 135 Figure 12 5 Definition for positive moments where the mass coefficient C is defined by Eq 12 54 the pressure coefficient C is defined by Eq 12 55 and the skin friction coefficients Csi Cry and C5 are defined by Eqs 12 56 Non dimensional moment coefficients are defined by Mi Yi Yref Zi Zref Cm i 2 UF i UF 3 12 63a l 5 Pret V2tSret Eres Let Let M i Zi Zref Ti Tref Ou p ue 5 12 63b Pret Ve Sref Lref Lu Lost M i Li Tref Yi Yref Cu sa 2 9 2G 12 63c Pret va Stet Lrer 7 Lreg Let l The force and moment coefficients for an entire patch are simply given by the summation of the force and moment coefficients for the faces on that patch 1 ui Cr am Cos Cy ina Cai Si 12 64a 1 i Cr Cms A Cys Si 12 64b C c D ep C s 12 64c m i inzi Vfzji 2i F ref gt Vef P f gt and Ui Yref Zi Zref 1 os 12 65 Cu F Crs Lives Ev Lref i Zi Zref Ti Tref Y Cx xx D 12 65b Hh i a Lyet i Ly i Ti Tref Yi a gt 12 65 Cm Cr E ref Cr AE ref l o i 136 Chapter 12 Algorithms and Methods Chapte
47. The RocfluMP Book A Haselbacher Assistant Professor Department of Mechanical and Aerospace Engineering University of Florida 222 MAE B P O Box 116300 Gainesville FL 32608 6300 Contact information Andreas Haselbacher Assistant Professor Department of Mechanical and Aerospace Engineering University of Florida 222 MAE B P O Box 116300 Gainesville FL 32608 6300 Phone 352 392 9459 Fax 352 392 1071 Email haselbacQuf1 edu Version 1 12 0 of 10 22 06 This manual documents rfluconv version 5 4 0 of 10 22 06 rfluinit version 4 4 0 of 10 22 06 rflumap version 4 4 0 of 10 22 06 rflump version 12 4 0 of 10 22 06 rflupart version 3 4 0 of 10 22 06 rflupick version 7 4 0 of 10 22 06 rflupost version 10 4 0 of 10 22 06 Contents Notation I Preliminaries 1 Introduction RA x22 ee ie amp Kee 9 0303 Vo Wed m Eh Reb ow Ow eq y dot 1 2 Overview of RecHuMEP o se sac ra scd IR RR vo A ek mex eR 9n La CD ss oe a tk a ae hE Be aoe AS BaD A GS 1 4 Overview of the RocfluMP Book eA 2 Nomenclature Conventions and Restrictions 4l Nomenclature 22s os o ro oo eee oy s x E e e ho ae yA caw eke aK AAD ERED Oe ERG ee eS 2 5 Reun ere p ace ok Poh ek oe ee ee eee ee Ee 3 Installation and Compilation NE I 0o o PTT ALI Installation fom CVS Repository ess o sea o Rs 512 Installation itom tar pl Mie cde raros ir ga Compilation AE 3 2 1 Overview of Compilation Process lll n 3 2
48. YPE PRI and CELL TYPE PYR defined in ModParameters F90 bf2v contains the vertices defining a boundary patch face bf2v1 contains the vertices defining a boundary patch face locally numbered for each bound ary patch bTri2v contains the vertices defining a triangular boundary patch face The vertices are oriented such that the normal vector is pointing out of the computational domain bTri2vl contains the vertices defining a triangular boundary patch face locally numbered for each boundary patch The vertices are oriented such that the normal vector is pointing out of the computational domain bQuad2v contains the vertices defining a quadrilateral boundary patch face The vertices are oriented such that the normal vector is pointing out of the computational domain bQuad2v1 contains the vertices defining a quadrilateral boundary patch face locally num bered for each boundary patch The vertices are oriented such that the normal vector is pointing out of the computational domain movePatch is a logical variable indicating whether the boundary patch is moving smoothGrid isa logical variable indicating whether the boundary patch grid is to be smoothed gs contains the grid speed of each boundary patch face bvn contains the components of the unit normal vector at the boundary patch vertices dXyz contains the imposed displacement of the boundary vertices fc contains the face centroid coordinates fn contains the components of the f
49. a given cell and n is the number of fringes The indicator function can be used to plot interferograms If DISCFLAG 1 rflupost writes three plotting variables to the visualization files i e the magnitude of the density gradient the Laplacian of the density and J given by Eq 9 1 9 1 2 Visualization and Identification of Vortical Structures The computation of plotting variables which allow the visualization of vortical structures is activated by the keywords VORTFLAG and COREFLAG in the POST Section of the input file The keyword VORTFLAG activates the computation of the Cartesian components of the vorticity field Ow Ov Qu Ow dup P Qv Ou If VORTFLAG 1 rflupost writes three plotting variables to the visualization files i e wz Wy and wz The keyword COREFLAG activates the computation of variables which allow identification of vortices These variables are used in three criteria for vortex identification which are described and compared in detail by Chakraborty et al 5 The criteria are based on the velocity gradient tensor Vv and its symmetric and antisymmetric components S Vv 4 Vv 9 5 Q vIe NI Vv Vv 9 6 The criteria are 9 2 TECPLOT Output 101 1 Q Criterion Vortices are identified by regions with Q gt 0 where Q is the second invariant of Vv i e Q IQI ISI 9 7 2 A Criterion Vortices are identified by regions in which A2 lt 0 where Az is the secon
50. ace normal unit vector and the area of the face bfgwt contains the face gradient weights bcName contains the name of the boundary patch 14 5 Mixture Data Structure 155 bFlag is a flag whether a burning face has ignited or not when running RocfluMP inside GENx This is used to avoid faces which have ignited from extinguishing Used only if GENX 1 bcFlag is a flag indicating the type of interaction with other codes when running RocfluMP inside GENx It can only assume the values BC NOT COUPLED BC NOT BURNING and BC BURNING defined in ModParameters F90 Used only if GENX 1 mdotalp contains the mass flux for each boundary patch face It is allocated only for burning boundary patches Used only if GENX 1 pf contains the face pressure Used only if GENX 1 qc contains the convective heat flux It is allocated only for burning boundary patches Used only if GENX 1 qr contains the radiative heat flux It is allocated only for burning boundary patches Used only if GENX 1 rhofalp contains the fluid density for each boundary patch face Used only if GENX 1 tempf contains the fluid temperature for each boundary patch face It is allocated only for burning boundary patches Used only if GENX 1 tflmAlp contains the static temperature of the injected fluid It is allocated only for burning boundary patches Used only if GENX 1 duAlp contains the incremental displacement Used only if GENX 1 nfAlp c
51. and COBALT inp Link lt casename gt cgr to COBALT inp ln s COBALT inp lt casename gt cgr Generate a patch mapping file lt casename gt cgi as described in Sec 5 3 4 2 Generate a boundary condition file lt casename gt bc The format of the boundary condition file is described in 5 4 Pay particular attention to the following Make sure that the patch mapping files generated above are consistent with the boundary condition file i e each patch in the grid is mapped to a patch in 75 76 Chapter 6 Problem Setup the boundary condition file Changing the patch mapping file means that the preprocessor will need to be rerun For coupled simulations Make sure that each patch has the correct values for the following parameters COUPLED MVPATCH and SMGRID Incorrect values of the COUPLED parameter are likely to lead to failure of Surfdiver 3 Generate an input file lt casename gt inp Pay particular attention to the following sections Format of grid file Make sure you set the GRIDSRC keyword in the FORMATS section to the correct value Initial condition Make sure that the initial condition is correct because changing the initial condition means that the preprocessor will need to be rerun In par ticular make sure that the initial condition is sensible and compatible with the boundary conditions Reference values Make sure that the values for CP and GAMMA are correct They have a very strong effect on
52. and selective investigation of data For this reason the ENSIGHT files produced by rflupost make extensive use of parts The following explains the part naming conventions used by rflupost The part naming convention adopted by rflupost is best explained with reference to Fig 9 3 Data written to the ENSIGHT file is split into volume and boundary patch data Volume data is further split according to the cell type tetrahedron hexahedron prism or pyramid and for each cell type according to the cell kind actual or virtual cell Boundary patch data is further split according to the face type triangular or quadrilateral and for each face type according to the face kind actual or virtual face 9 3 2 1 Volume Parts The names of volume parts are VOL_ lt region index gt where lt region index gt is a five digit string indicating the global region index Examples are shown in Fig 9 4 which represents a screen dump of the data part loader window in ENSIGHT 9 3 ENSIGHT Output 109 Figure 9 4 Screen dump of data part loader window in ENSIGHT illustrating part naming con vention used in rflupost 110 Chapter 9 Visualization 9 3 2 2 Boundary Patch Parts The names of boundary patch parts are PAT patch index face type gt lt face kind region index where patch index is a three digit string indicating the global patch index It is important to note that this is a global patch index face
53. ap writes the region mapping file For coupled runs rflumap also produces the input file for Rocin 7 4 rflump 7 4 1 Invocation rflump is an MPI code and hence it s invokation is dependent on the type of machine and may also be dependent on the MPI distribution For typical MPI distributions rflump is invoked by typing mpirun np lt n gt rflump c lt casename gt v lt verbosity gt where lt n gt is the number of processes The command line arguments read by rflump are lt casename gt A character string used to label the input and output files lt verbosity gt An integer indicating the desired verbosity level of rflump The verbosity level can take the following values O No output rflump will not write any information to standard output 1 Low level of output rflump will write some information to standard output 2 High level of output rflump will write detailed information to standard output 7 4 rflump 83 7 4 2 Input Files The following input files are read by rflump e An input file called casename inp e A grid file in RocfluMP format e A flow solution file in RocfluMP format e A boundary condition file The name of the file is lt casename gt bc e A dimension file e restart information file 7 4 3 Output Files The following input files are written by rflump A grid file in RocfluMP format A flow solution file in RocfluMP format e A dimension file if the flow is unstead
54. area m REFLENGTH Specifies value of reference length m REFXCOORD Specifies value of reference z coordinate m REFYCOORD Specifies value of reference y coordinate m REFZCOORD Specifies value of reference z coordinate m 5 2 3 FORMATS Section The FORMATS section contains the following keywords GRID Specifies the format of the RocfluMP grid file It can take the following values O Grid file is in ASCIT format 1 Grid file is in binary format GRIDSRC Specifies the format of the grid file read by rfluprep It can take the following values O Grid file is in CENTAUR ASCII format 1 Grid file is in VGRIDns format 2 Grid file is in MESH3D format 3 Grid file is in TETMESH format 4 Grid file is in Cobalt format 5 Grid file is in GAMBIT format Only GAMBIT files in ASCII neutral file format are supported 10 Grid file is in CENTAUR binary format SOLUTION Specifies the format of the RocfluMP flow file It can take the following values 0 Flow file is in ASCII format 1 Flow file is in binary format 34 Chapter 5 File Content and Format Specifications 5 2 4 GRIDMOTION Section The GRIDMOTION section contains the following keywords NITER Specifies the number of smoothing iterations Only applicable if TYPE 1 or TYPE 2 SFACT Specifies the smoothing coefficient The values for the number of smoothing iter ations and the smoothing coefficient should be chosen together The recommended values are for moving
55. ary fluxes with second order accuracy RECONST Specifies the way in which the gradients computed by the reconstruction method are modified It can take the following values O The gradients are not modified 40 Chapter 5 File Content and Format Specifications 1 Weighted essentially non oscillatory WENO reconstruction in which gradients are weighted according to the inverse square of their magnitude 2 Weighted essentially non oscillatory WENO reconstruction in which gradient com ponents are weighted according to the inverse square of their magnitude 10 Gradients are modified using limiter function of Barth and Jespersen 2 11 Gradients are modified using limiter function of Venkatakrishnan 15 For flows without sharp gradients such as shock waves contact discontinuities or material interfaces or computations in which sharp gradients are resolved such as the above in inviscid flows the gradients need not be modified For flows with sharp gradients or computations in which sharp gradients are not resolved the gradients must be modified to avoid numerical instability SDIMENSC Specifies the dimensionality of the cell stencil It can take the following values 2 use two dimensional cell stencil 3 use three dimensional cell stencil default SDIMENSF Specifies the dimensionality of the face stencil It can take the following values 2 use two dimensional face stencil 3 use three dimensional face stencil default
56. atches pPatch gt pRegion patches iPatch WRITE iFile 3 18 pPatch nBTrisTot pPatch nBQuadsTot IF pPatch nBTrisTot gt O THEN DO j 1 3 WRITE iFile 10 I8 pPatch bTri2v j k k 1 pPatch nBTrisTot END DO j END IF pPatch nBTrisTot IF pPatch nBQuadsTot gt O THEN DO j 1 4 49 50 51 52 53 16 1 Grid Files 169 WRITE iFile 10 I8 pPatch bQuad2v j k k 1 pPatch nBQuadsTot END DO j END IF pPatch nBQuadsTot END DO iPatch WRITE iFile A End 16 1 2 CENTAUR Grid File 16 1 3 VGRIDns Grid Files The mapbc file produced by VGRIDns is not read 16 1 3 1 vbc File The vbc file corresponds to the bc file written by VGRIDns It is renamed because the boundary condition file of RocfluMP has the extension bc 16 1 3 2 cgosg File 16 1 4 MESH3D Grid File 16 1 5 TETMESH Grid File 16 1 5 1 noboite File The format of the noboite file is described in the TETMESH user s manual 16 1 6 Cobalt Grid File The first line of the Cobalt grid file contains three integers nDimensions nZones nBoundaryPatches where the meaning is self explanatory For use within rflump Cobalt grid files must satisfy the following restrictions nDimensions must be equal to 3 and nZones must be equal to 1 The next line contains five integers nVertices nFaces nCells nVerticesPerFaceMax nFacesPerCellMax where the last two quantities represent the maximum number of vertices defining a fac
57. be started with smaller a CFL number time step to ensure stability 3 Repeat the run with different flux function Some flux functions exhibit pathological behavior under certain conditions or do not guarantee preservation of positive definite quantities For example of the flux functions available in RocfluMP the HLLC scheme often exhibits more robustness in strong transients than the Roe scheme 4 Repeat the run with a first order scheme Failing first order accurate computations often indicate problems with problem setup 111 112 Chapter 10 Troubleshooting 5 Repeat the run with modified boundary conditions For example if problems persist during strong transients at an outflow boundary it may prove helpful to attempt a new run with supersonic or subsonic conditions 10 2 Explanations of Warnings Inflow detected at outflow boundary Outflow detected at inflow boundary One or more faces on an inflow outflow boundary were detected with outflow inflow respec tively This is often caused by the precise values of the boundary conditions imposed by the user For example too high a value of the static pressure at outflow boundaries can lead to inflow in some conditions Reverse flow on inflow outflow boundaries can be a temporary problem i e the expected flow direction becomes established as the flow develops In some cases however reverse flow on boundaries can indicate a more serious problem namely a poorly chosen location
58. becomes possible to visualize virtual cells and boundary faces which can be useful during debugging NSERVERS Specifies the number of servers when writing output files for parallel runs and visualization with ENSIGHT Only relevant if OUTFORMAT 2 OUTFORMAT Specifies the format of the visualization files It can take the following values 1 Write visualization files in TECPLOT format 2 Write visualization files in ENSIGHT format PEULFLAG Specifies whether the Lagrangian particle fields are to be converted to Eulerian particle fields and written to the visualization files It can take the following values 5 2 Input File 43 0 Do not convert Lagrangian fields to Eulerian fields 1 Convert Lagrangian fields to Eulerian fields See also Sec 9 1 PLTPATFLAG Specifies whether only the first virtual patch is to be written to output file It can take the following values O Write all patches to output file 1 Write only the first patch virtual to the output file Because virtual patches only exist for two dimensional computations PLTPATFLAG is only relevant if DIMENS 2 At present PLTPATFLAG is only relevant for TECPLOT files i e if OUTFORMAT 1 PLTTYPE Specifies the data to be written to output file It can take the following values 1 Write only the grid to the output file 2 Write the grid and the solution to output file PLTVOLFLAG Specifies whether volume data are to be written to output file It can take the followi
59. cess The compilation process for RocfluMP is automatic in the sense that the Makefiles determine the machine type and set the suitable compilation options If you intend to run on Apple IBM Linux SGI or Sun machines you do not need to modify any Makefiles If you intend to run on other machines you will need to create your own Makefile You can pattern it after the existing machine dependent Makefiles RocfluMP is compiled with MPI by default which means that you must have installed MPI on your machine before attempting to compile RocfluMP The compilation process consists of two stages The first stage is the actual computation as described below The output of the compilation process are several executables rfluconv The conversion module of RocfluMP rfluinit The initialization module of RocfluMP rflumap The region mapping module of RocfluMP rflupick The region and cell picking module of RocfluMP rflupost The postprocessing module of RocfluMP rflupart The partitioning module of RocfluMP rflump The flow solution module of RocfluMP The second stage consists of copying these executables into your HOME bin directory by typing user machine directoryl gmake RFLU 1 install 3 2 Compilation 23 3 2 2 Description of Compilation Options To compile RocfluMP type the following at the prompt user machine directoryl gmake RFLU 1 options where the currently supported options are any of the following CH
60. cle impact statistics 1 Gather paticle impact statistics Particle impact statistics are written into the patch coefficient file for visualization with TECPLOT THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch 60 Chapter 5 File Content and Format Specifications 5 4 1 6 No Slip Boundary BC NOSLIP TEMP Section The BC NOSLIP TEMP section is used to specify boundary conditions for patches with an isothermal no slip boundary condition and contains the following keywords COUPLED Specifies whether patch is interacting during a computation with GENx It can take the following values 1 Patch is interacting 2 Patch is not interacting CRECONST Specifies whether constrained reconstruction is to be used when computing face gradients for this boundary It can take the following values 0 Do not use constrained reconstruction default 1 Use constrained reconstruction KIND Specifies the treatment of the boundary condition It can take the following values O Simple boundary condition treatment 1 Characteristic boundary coundition treatment NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the f
61. d eigenvalue of S Q assuming an ordering of A gt Az gt As 3 Swirling Strength Criterion The eigenvalues of the velocity gradient tensor are as sumed to be given by A and Aer i Aci Then a vortex is identified by regions in which Ag 2 and Ae A S where and 6 are positive thresholds If COREFLAG 1 rflupost writes four plotting variables to the visualization files i e Q given by Eq 9 7 As and A and Aci from the swirling strength criterion 9 1 3 Visualization of Eulerian Particle Fields The computation of plotting variables which allow the visualization of Eulerian fields of the particle data is activated by the keyword PEULFLAG in the POST Section of the input file If PEULFLAG 1 rflupost writes eight plotting variables to the visualization files 9 2 TECPLOT Output The TECPLOT files produced by rflupost are in binary format to reduce file size and loading time The binary file are compatible across machines i e files written on little endian machines can be read on big endian machines and vice versa The files produced by rflupost make use of TECPLOT features which require version 10 or newer 9 2 1 File Naming Convention The name of the TECPLOT files produced by rflupost depends on whether a steady or unsteady flow is computed The name of the TECPLOT file is e casename iteration stamp gt plt for steady flows where iteration stamp is a six digit string denoting the iteration number
62. d NSERVERS in the POST Section 9 3 1 File Naming Convention The name of the ENSIGHT files produced by rflupost depends on whether a steady or unsteady flow is computed One variable which may be a scalar or vector is written to each file The name of an ENSIGHT file is e lt casename gt lt variable gt server index iteration stamp for steady flows where lt iteration stamp is a six digit string denoting the iteration number e lt casename gt lt variable gt lt server index time stamp for unsteady flows where time stamp is a string of the form n nnnnnnnEtnn denoting time For steady or unsteady flows variable is a character string denoting the variable contained in the file and server index is the index of the server which will read the file The meaning of the variable character string is as listed in Table 9 2 9 3 2 Part Naming Conventions ENSIGHT organizes data by parts Each part is assigned a name and may consist of either volume or surface data Because ENSIGHT allows each part to be activated displayed or deactived not displayed separately the separation of data into parts is useful in allowing 108 Chapter 9 Visualization Region Y Volume Boundary patches Y Face type Face kind Parts in Ensight file Figure 9 3 Illustration of parts written to ENSIGHT files by rflupost detailed
63. d using two integers PTOT Total pressure Pa THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 5 4 Boundary Condition File 55 Figure 5 5 Definition of angles BETAH and BETAV for inflow boundary condition Angles are positive in direction of arrows 1 Compute thrust and specific impulse for this patch TYPE Specifies whether inflow is supersonic or subsonic It can take the following values 0 Inflow is supersonic The Mach number must be specified by the keyword MACH 1 Inflow is subsonic TTOT Total temperature K Specification of Inflow Angles For planar boundary patches with known normal vector n n n n V and inflow normal to the boundary the inflow angles are given by Bo dant 5 1 Ba sin ny 5 2 For imposed velocities and flow which is not necessarily normal to the boundary the inflow angles are given by By tan 5 3 B amp sin 5 4 56 Chapter 5 File Content and Format Specifications 5 4 1 3 Inflow Boundary BC INFLOW VELTEMP Section The BC INFLOW VELTEMP section contains the following keywords KIND Specifies the treatment of the boundary condition It can take the following values 0 Simple boundary condition treatment 1 Characteristic boundary counditio
64. d variable to which TBC applies PATCH 2 2 applies to patch ONTIME 1 0E 3 time to start using this TBC OFFTIME 2 0E 3 time to stop using this TBC AMP 0 2 amplitude of sinusoid FREQ 1 0E4 frequency of sinusoid PHASE 0 0 argument of sin for t 0 Line 2 Specifies that the variable PRESS of an OUTFLOW boundary is to be modified Note that this does not yet specify which outflow boundary is to be modified Line 3 Specifies that variables on patch two are to be modified Line 4 Specifies the lower bound on the time window in which the values are to be modified Line 5 Specifies the upper bound on the time window in which the values are to be modified Line 6 Specifies the amplitude A of the sinusoidal variation Line 7 Specifies the angular frequency w of the sinusoidal variation rad s Line 8 Specifies the phase of the sinusoidal variation deg 5 4 2 3 TBC STOCHASTIC Section 5 4 2 4 TBC_WHITENOISE Section 5 4 3 Grid Motion Boundary Conditions Computations with moving boundaries require the specification of boundary conditions for the grid motion algorithm The following keywords can be specified in any of the above sections MVPATCH Specifies whether patch is moving It can take the following values O Patch is not moving 1 Patch is moving SMGRID Specifies whether grid on patch is to be smoothed It can take the following values 0 Do not smooth surface grid 1 Smooth surface grid 5 5 R
65. dition File 65 5 4 1 11 Virtual Boundary BC VIRTUAL Section The BC VIRTUAL section contains the following keywords NAME Specifies the name of the boundary PATCH Specifies the range of patches to which the data in this section is to be applied using two integers The effect of this boundary condition is that no fluxes are computed Because virtual boundaries always come in pairs and each cell in a truly two dimensional computation must have one face each on the virtual boundaries the effect of computing no fluxes on this boundary is to simulate constant properties in the z direction 5 4 2 Time Dependent Boundary Conditions Any of the physical quantities specified by the user in the above described sections can be specified to vary in time The time variations can be specified to be piecewise linear sinusoidal or stochastic using additional sections in the boundary condition file Each of these sections can be used to modify one user specified physical quantity on one boundary patch If more than one physical quantity on a given patch or physical quantities on several patches are to be modified multiple additional sections have to be included in the boundary condition file 5 4 2 1 TBC PIECEWISE Section The TBC_PIECEWISE section allows the specification of piecewise constant and piecewise linear variations Consider the following example section TBC_PIECEWISE INJECT MFRATE BC and variable to which TBC applies PATCH 1
66. ds along with their locations The location indicates whether the cell shares one or more faces with boundary patches In some cases the above information may suffice because all or a majority of cells lie on a given boundary patch indicating that the problem originates with the boundary conditions applied to this patch If the location of the cells is not immediately apparent or scattered around the computational domain it can be helpful to select individual cells for visualization using rflupick This can be useful to determine whether the problems originate in a clusters of cells of poor quality or along the interface between layers of different cell types If the source of the problem is not readily apparent from the additional output it is often helpful to place probes at the location of the cell centroids before restarting the computation or starting a new computation Taking the first three cells of above example the following PROBE may be inserted into the lt casename gt inp file PROBE NUMBER 3 0 32081289E 02 0 83698854E 01 0 12040992E 01 0 12485188E 01 0 83750698E 01 0 61473528E 03 0 35957776E 02 0 83795613E 01 0 74193030E 02 The probe files produced by repeating the run can be visualized with TECPLOT or other packages and may help in determining the cause of the problem Part III Developer and Reference Manual Chapter 11 Governing Equations 11 1 The Navier Stokes Equations RocfluMP solves the thr
67. duced one or more files for visualization The behavior of rflupost is influenced by the values assigned to the keywords in the POST Section of the input file rflupost writes output files for TECPLOT and ENSIGHT TECPLOT is best used to visu alize small to medium sized cases and to aid in debugging whereas ENSIGHT is best used to visualize large cases 9 1 Plotting Variables rflupost can be instructed to write so called plotting variables to the output files Plot ting variables are variables derived from the solution usually through non trivial additional computations in rflupost to gain insight into particular features of the flowfield 9 1 1 Visualization of Discontinuities The computation of plotting variables which allow the visualization of discontinuities is activated by the keyword DISCFLAG in the POST Section of the input file Three variables are computed to allow the visualization of discontinuities 1 The magnitude of the density gradient This variable is used to make Schlieren pictures in which highly localized regions of large density gradients indicate shock waves and contact discontinuities For an example of a Schlieren picture see Fig 4 3 99 100 Chapter 9 Visualization 2 The Laplacian of the density This variable is used to make Shadowgraphs 3 An indicator function of the normalized density pi MIN Pi 1 Le mod 2 9 1 max pi min Pi 2 where i represents the index of
68. e probe Each line of the probe file consists of seven columns which contain the following pieces of data 72 Chapter 5 File Content and Format Specifications Column 1 For steady flows the iteration number for unsteady flows the time Column 2 The density Column 3 The x velocity component Column 4 The y velocity component Column 5 The z velocity component Column 6 The static pressure Column 7 The static temperature 5 9 Mass Conservation Check File The mass conservation check file is written by rflump for unsteady flows with grid motion outside of GENx The file is called casename mass and contains the following information Column 1 The time Column 2 The mass contained in the solution domain Column 3 The mass flow entering the solution domain Column 4 The mass flow exiting the solution domain Column 5 The volume of the solution domain As implied by its name the mass conservation check file is used to check that mass is conserved during moving grid computations 5 10 Statistics File 5 11 GENx Control File When rflump is run within GENx an additional input file is required because rflump is not invoked from the command line This so called GENx control file is always called RocfluControl txt and contains the following information Line 1 The case name Line 2 The directory name containing the rflump input files relative to the directory from which GENx is invoked 5 11 GENXx
69. e 151 Face Vertices 1 1432 2 1265 3 2 3 7 6 4 3 4 8 7 5 1 5 8 4 6 5 6 T 38 Table 14 1 Face to vertex connectivity arrays for hexahedral cells Face Vertices 1 132 2 12 5 4 3 2 3 6 5 4 1 4 6 3 5 4 5 6 Table 14 2 Face to vertex connectivity arrays for prismatic cells Each cell type has not only a clearly defined numbering of its vertices but also for its edges and faces These numberings are listed in Tables 14 1 14 4 In reading these tables it is to be understood that edges and faces have an orientation This is a crucial point if the routines which construct data structures are to be understood properly Therefore the rows in these tables are to be read only from left to right Thus edge 10 of an hexahedron is pointing from vertex 2 to vertex 6 Furthermore faces are oriented such that their normal vectors are pointing out of the cell This corresponds to anti clockwise ordering of the vertices when viewing the face of a cell from the outside of that cell and to clockwise ordering when viewing the face through the cell Face Vertices 1 1 4 3 2 2 1 2 5 3 2 3 5 4 3 4 5 5 15 4 Table 14 3 Face to vertex connectivity arrays for pyramidal cells 152 Chapter 14 Data Structures Face Vertices 1 12 3 2 2 4 3 3 13 4 4 1 4 2 Table 14 4 Face to vertex connectivity arrays for tetrahedral cells Figure 14 7 Overview of boundary data structure 14 3 2 Module RFLU_ModGrid F90 The module
70. e and the maximum number of faces defining a cell For example if the grid consisted purely of tetrahedra nVerticesPerFaceMax and nFacesPerCellMax would be equal to 3 and 4 respectively On the other hand if the grid consisted of tetrahedra prisms and pyramids nVerticesPerFaceMax and nFacesPerCellMax would be equal to 4 and 5 respectively 170 Chapter 16 File Content and Format Specifications The next nVertices lines contain the x y and z coordinates for each vertex The next nFaces lines contain the face connectivity information nVerticesPerFace lt nVerticesPerFace gt vertices Celli Cell2 where Ce111 and Ce112 are the two cells which share a given face If a face lies on a boundary patch the respective cell is given by the negative patch index 16 2 Flow Solution File The flow solution file contains the conserved variables of the solution at a given iteration or time The name of the flow solution file depends on whether it is written in ASCII or binary format and whether a steady or unsteady flow is computed The name of the flow solution file in ASCII format is e lt casename gt floa_ lt mmmmm gt _ lt nnnnnn gt for steady flows e lt casename gt floa_ lt mmmmm gt _ lt n nnnnnnnE nn gt for unsteady flows where lt mmmmm gt is the region number lt nnnnnn gt is the iteration number and lt n nnnnnnnEt nn gt is the time stamp The name of the flow solution file in binary format is e lt casename gt flo_
71. e following values O Inflow is supersonic The static pressure must be specified by the keyword PRESS 1 Inflow is subsonic VELX r component of fluid velocity m s VELY y component of fluid velocity m s VELZ z component of fluid velocity m s 5 4 1 4 Injection Boundary BC INJECT Section The BC INJECT section contains the following keywords BFLAG Specifies whether this patch is burning at t 0 Only relevant for computations with GENx It can take the following values O Patch is not burning at t 0 1 Patch is burning at t 0 COUPLED Specifies whether patch is interacting during a computation with GENx It can take the following values 1 Patch is interacting 2 Patch is not interacting CRECONST Specifies whether constrained reconstruction is to be used when computing face gradients for this boundary It can take the following values 0 Do not use constrained reconstruction default 1 Use constrained reconstruction KIND Specifies the treatment of the boundary condition It can take the following values 0 Simple boundary condition treatment 1 Characteristic boundary coundition treatment MFRATE Injection mass flux kg m s NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 58 Chapter 5 File Content and Format Specifications 1 First order accuracy 2 Second order accuracy The o
72. e global cell and vertex numbers of a given region and the global cell and vertex numbers of the serial region In addition the file contains the mapping between the boundary faces of a given region and the boundary faces of the serial region The file is always in ASCII format and is called casename rnm mmmmm where lt mmmmm gt is the region index The format of the renumbering file is illustrated by the following code fragment WRITE iFile A ROCFLU renumbering file WRITE iFile A Dimensions WRITE iFile 2 18 pGrid nVertTot pGrid nCellsTot WRITE iFile A Vertices WRITE iFile 10 18 pGrid pv2sv ivg ivg 1 pGrid nVertTot WRITE iFile A Cells WRITE iFile 10 18 pGrid pc2sc icg icg 1 pGrid nCellsTot WRITE iFile A Boundary faces WRITE iFile 10 18 pGrid pbf2sbfCSR if1 if1 1 pGrid nBFacesTot WRITE iFile A End 16 6 Communication Lists File The communication lists file contains the communications lists required by each region to send data to and receive data from other regions The file is always in ASCII format and is 20 21 22 23 174 Chapter 16 File Content and Format Specifications called lt casename gt com_ lt mmmmm gt where lt mmmmm gt is the region index The format of the communication lists file is illustrated by the following code fragment WRITE iFile A RO
73. e in RocfluMP ASCII or binary format e A flow solution file in RocfluMP ASCII or binary format A dimension file e A boundary condition file e A mapping file 7 6 3 Output Files The following output files are written by rflupick e A postprocessor info file e A version file called rflupick vrs It contains the version number and date of the executable Successive runs append to the version file 7 6 4 Interactive Input After invokation rflupick presents the following screen output Picking regions manually Enter information on regions a Pick all regions s Pick some regions n Pick no regions Enter information type 7 6 rflupick 89 The input determines which regions will be postprocessed by rflupost Picking all regions Assuming that the user enters option a rflupick will loop over all regions and for each region present the following screen output the following output assumes that the region index is 00001 Picking special cells Global region 00001 Enter information on special cells b cell adjacent to boundary face c single cell cells adjacent to interior face stencil members cells meeting at vertex quit Enter information type Q 4 vn Fh This output is self explanatory Once special cells were picked rflupick will present the following screen output Picking special faces Global region 00001 Enter information on special faces b boundary face i interior face
74. e received the number of vertices for which data must be sent the number of vertices for which data must be received and the number of vertices for which data must be shared Line 28 The last line must contain the string shown otherwise reading of the file will fail OQ 0 N O Oc BR Ww M Hn gt A N n O 0 nN Oo A 16 4 Cell Mapping File 173 16 4 Cell Mapping File The cell mapping file contains the mappings between the numbering local to a given cell type and the global numbering over all cell types The file is always in ASCII format and is called casename cmp mmmmm where lt mmmmm gt is the region index The format of the cell mapping file is illustrated by the following code fragment WRITE iFile A ROCFLU cell mapping file WRITE iFile A Dimensions WRITE iFile 4 18 pGrid nTetsTot pGrid nHexsTot pGrid nPrisTot pGrid nPyrsTot WRITE iFile A Tetrahedra WRITE CiFile 10 18 pGrid tet2CellGlob icg icg 1 pGrid nTetsTot WRITE iFile A s Hexahedra WRITE iFile 10 18 pGrid hex2CellGlob icg icg 1 pGrid nHexsTot WRITE iFile A s Prisms WRITE iFile 10 18 pGrid pri2CellGlob icg icg 1 pGrid nPrisTot WRITE iFile A s Pyramids WRITE iFile 10 18 pGrid pyr2CellGlob icg icg 1 pGrid nPyrsTot WRITE iFile A st End 16 5 Renumbering File The renumbering file contains the mapping between th
75. e splitting scheme of Roe works well in general Because Roe s scheme is not positively conservative it can lead to negative densities and pressures for flows with strong transients For this reason such flows should be computed with either the HLLC solver or the AUSM scheme For shock waves with perfect or near perfect alignment on uniform hexahedral grids the Roe scheme and HLLC solver can suffer from transverse shock instability so the AUSM scheme should be used instead Vis cous flows near solid walls with strong flow to grid alignment can occasionally lead to mild forms of decoupling with the AUSM4 scheme DISSFACT Factor multiplying the dissipation terms of the flux difference splitting of Roe The dissipation may need to be reduced to capture marginally resolved flow features such as separation or vortices It is important to note however that reducing the dissipation can compromise the stability of the computation The default value of DISSFACT is 1 ENTROPY Specifies value of entropy correction coefficient Applies only to the flux difference splitting scheme of Roe ORDER The order of accuracy of the flux discretization It can take the following values 1 Compute fluxes with first order accuracy 2 Compute fluxes with second order accuracy ORDERBF The order of accuracy of the flux discretization for boundary faces It can take the following values 1 Compute boundary fluxes with first order accuracy 2 Compute bound
76. ed cell quantity in interior domain which is being solved for routine for carrying out RK 4 integration is RungeKuttaMP which calls RKInitMP and RKUpdateMP 13 1 7 Flux computation using boundary data Flux computation at the boundary faces is done making use of boundary variables and inte rior cell variables Convective flux computation using second order accurate approximation for flow variables are carried out using RFLU_NSCBC_CompSecondPatchFlux subroutine and RFLU NSCBC CompFirstPatchFlux can be used for first order approximation figure 13 1 describes NSCBC implementation in the form of a flowchart 13 2 Parallelization 140 Chapter 13 Implementation Details RFLU BXV ReadVarsWrapper RkInitMP RkinitGeneric RFLU_NSCBC_CompRhsSlipWall Ki er BC SLIPWALL BC_FARFIELD RFLU_NSCBC_CompRhsFarf PP ss ee BC OUTFLOW Sade _ RFLU NSCBC CompRhsOutflow 4 7 Figure 13 1 NSCBC implementation in Rocflu Chapter 14 Data Structures RocfluMP makes extensive use of Fortran 90 user defined types for the definition of data structures 14 1 Philosophy and Abstraction The top layer of the data structure developed for RocfluMP is depicted schematically in Fig 14 1 The top layer of the data structure consists of two main layers of abstraction 1 At the highest layer are multigrid levels constructed from the finest grid Each layer can contain an arbitrary number of regio
77. ed to as nMappings The remaining nMappings lines Each line contains three integers The first two integers represent the lower and upper limits of the patches in the VGRIDns file which are to be mapped to the patch indicated by the third integer The lower limit must be less than or equal to the upper limit The following is an example vgi file 6 4 12 12 7 7 10 10 8 9 11 11 17 22 Fe ooo oP BP WWNHON KE This file indicates that five patches will exist after merging and that eight mappings are listed For example the first mapping specifies that patches five and six are mapped to patch one and the last mapping specifies that patches 17 to 22 are mapped to patch five Note that it is possible to map several original patches to a given new patch in separate mappings That is the following example is equivalent to the one given above 50 Chapter 5 File Content and Format Specifications 10 BO 0 12 12 7 7 10 10 8 9 11 11 17 18 19 22 aoP BP 00NNKM GG It is important to note that the extrema of the lower and upper original patch indices must be equal to unity and the number of patches specified in the bc file respectively Furthermore the extrema of the new patch indices must be equal to unity and the number of patches specified on the first line respectively In practice it is usually impossible to specify the mappings without having visually inspected the grid It is therefore recommended that in a first
78. ee dimensional time dependent compressible Navier Stokes equations in integral form on moving and or deforming grids sf V d Ras Ras f dan 11 1 Ot Jo 90 90 Q The vector of the conservative variables W is W pv 11 2 The vector of convective fluxes reads pV PUV np F puV nyp 11 3 pwV t nip pHV V p FF with the face normal velocity given by V nru yb nw V 11 4 where V is the grid speed i e the grid velocity normal to the control volume face The vector of the viscous fluxes can be written as 0 Ty Tygy T NyTry T NzTrz Joc she Na ag s aT s 11 5 Tog Tza NyTzy NzTzz 1505 NOJ HNO 11 2 The Geometric Conservation Law 117 where D 8 I OT tee Chay Utes ka Ege z OT Oy D lt Ur Uta WTyz k Bee 11 6 OT O TT d 1T um pes z are the terms describing the work of the viscous stresses and the heat conduction in the fluid In the case of LES the viscous stresses read 2 Quy c bs SGS 3 8x 4 TUR ij with TRS being the subgrid stresses In the case of RANS equations the subgrid terms ESC in Eq 11 6 and 77 in Eq 11 7 are omitted Instead the dynamic viscosity and the thermal conductivity are split into a laminar and a turbulent part i e Tij 2u Sij 11 7 b npo pr 11 8 and HL LT ky kp SE m 11 9 where cp stands for the specific heat coefficient at constant pressure and Pr is the Prandtl
79. efined as q Di oT 12 57 LoetV s ref where q is the heat flux for the face 12 7 Force and Moment Computation RocfluMP computes forces and moments exerted by the fluid on the patches The force and moment on a patch are computed from the sum of the forces and moments on the faces of that patch The force on a face 7 with unit normal vector n and area 5 is F pivi ni vi Pi Pret ni Ti ni 5i 12 58 where p is the density v is the velocity vector p is the pressurem and T n is the viscous stress The force components are given by Fri Covi Dj U pi pret Nai Ti ni Si 12 59a Fyi oiv Di Vi pi Pret Nyi Ti n Si 12 59b Fyi ivi 15 Wi Pi Pref Nzi Ti nj Si 12 59c The moment about a reference location r created by the force on a face is M ri rq X Fi 12 60 where r is the position vector of the face centroid The convention for positive moments is shown in Fig 12 5 Moments around a given coordinate axis are defined to be positive if they lead to a counter clockwise rotation when looking in the negative direction of that coordinate axis The moments components are given by Mas Fai yi Vret m Bs ai Zref 12 61a Mus Faili Zref F l i Sat 12 61b Mag PONE Tref CE Fou ag ther 12 61c Non dimensional force coefficients are defined by Fyri Ui Si C p Cn Ou Chi aor 12 62a i 5 Pret V 2t
80. egion Mapping File 69 MOVEDIR Specifies in which direction s the vertices on the patch are allowed to move The direction s in which movement is allowed are indicated by integers O Vertices on patch are not allowed to be moved in any direction 1 Vertices on patch are allowed to be moved in x coordinate direction 2 Vertices on patch are allowed to be moved in y coordinate direction 4 Vertices on patch are allowed to be moved in z coordinate direction The non zero values can be combined to specify movement in planes normal to coor dinate axes or arbitrary movement 3 Vertices on patch are allowed to be moved in xy plane i e movement normal to the z coordinate direction is not allowed 5 Vertices on patch are allowed to be moved in xz plane i e movement normal to the y coordinate direction is not allowed 6 Vertices on patch are allowed to be moved in yz plane i e movement normal to the x coordinate direction is not allowed 7 Vertices on patch are allowed to be moved in xyz space 5 5 Region Mapping File The region mapping file contains the mappings between regions and processes The file is always in ASCII format and is called lt casename gt map Consider the following example region mapping file ROCFLU region mapping file Number of regions 4 Number of processes 4 Process 000001 1 1 Process 000002 1 2 Process 000003 o ON O 0 BR Ww M n 70 Chapter 5 File Content and Format Specificati
81. elevant only if KIND 1 THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch TYPE Specifies whether outflow is supersonic or subsonic It can take the following values 0 Outflow is supersonic 1 Outflow is subsonic The static pressure must be specified by the keyword PRESS 2 Outflow is mixed subsonic supersonic The static pressure must be specified by the keyword PRESS 5 4 1 8 Periodic Boundary BC_PERIODIC Section ANGLE Specifies the angle between two related periodic patches deg For linear periodicity ANGLE 0 0 For rotational periodicity the angle is that which transform the current patch to be coincident with its related patch as specified by the value assigned to the keyword RELPATCH The convention is that positive angles are defined according to the right hand rule assuming that the rotation axis as specified by the value assigned to the keyword AXIS is pointing in the respective positive coordinate direction see Fig 5 6 for an example AXIS For linear periodicity specifies the direction of the translation vector which transforms the current patch with its related patch For rotational periodicity specifies the direc tion of the axis about which the current patch is rotated to
82. emperature T OT T 1 L y 1D L L 0 12 43 8i tag zo Ls L The NSCBC Strategy for the Euler Equations Characteristics based boundary con ditions are implemented in following 3 steps 1 For each physical boundary condition imposed corresponding equation is eleminated from the system 12 23 12 27 like if p or T is imposed then there is no need to solve energy equation at boundary 2 At each boundary some of the characteristics are going out of domain and some are coming in Characteristic amplitudes L s corresponding to outgoing waves can be computed using one sided differencing Incoming wave amplitudes can not be computed using one sided differencing Using LODI relations and imposed boundary conditions these incoming wave amplitudes can be expressed in terms of outgoing wave amplitudes 3 Use the remaining conservation equations of the system 12 23 12 27 combined with the values of the L s obtained from Step 2 to compute all variables which are not imposed by boundary conditions 12 3 3 2 Inflow Boundary Conditions Inflow boundary conditions discussed in this section are the ones where velocity and tem perature are specified Inflow can be subsonic or supesonic subsonic inflow can be reflecting or non reflecting At a generic boundary face u has to be negative for it to be inflow u 0 means flow is incoming i e flow is entering the domain from outside 130 Chapter 12 Algorithms and Me
83. esidual vector due to dissipative terms of the spatial discretization cvState isa flag indicating which state is stored in the conservative state vector It can only take the values CV MIXT STATE CONS CV MIXT STATE DUVWP and CV MIXT STATE DUVWT cvVrtx contains the state vector at the vertices It is used only in rflupost after having interpolated the cell centered values to the vertices bGradFace contains the boundary face gradients It is accessed using the array bf2bg in the data type t patch gradCell contains the cell gradients gradFace contains the face gradients 162 Chapter 14 Data Structures Chapter 15 GENx Integration 15 1 Definition of Attributes 15 1 1 Grid Attributes 15 1 1 1 Volume Panes The volume grid pane attributes are defined in Table 15 1 15 1 1 2 Surface Panes The surface grid pane attributes are defined in Table 15 2 Note that multiple connectivity lists exist for each face type e g t3 real and t3g real The former is required by Roccom and its entries are used to access the coordinate array nc on each surface pane The latter is required by rflump and its entries are used to access the coordinate array nc on the volume pane 15 1 2 Solution Attributes The solution attributes are defined in Table 15 3 15 1 3 Interface Attributes 15 1 3 1 Non Interacting Panes 15 1 3 2 Non Burning Panes The attributes of non burning panes are defined in Table 15 4 15 1 3 3 Burning Panes
84. face connectivity arrays for prismatic cells are shown in Table 14 2 pyr2v contains the connectivity information for the pyramidal cells The vertices must be numbered as shown in Fig 14 6 c The face to vertex edge to vertex and edge to face connectivity arrays for pyramidal cells are shown in Table 14 3 tet2v contains the connectivity information for the tetrahedral cells The vertices must be numbered as shown in Fig 14 6 d The face to vertex edge to vertex and edge to face connectivity arrays for tetrahedral cells are shown in Table 14 4 fc contains the face centroid coordinates fn contains the components of the face normal unit vector and the area of the face cgwt contains the cell gradient weights fgwt contains the face gradient weights degr contains the degree of each vertex gs contains the grid speed of each face volMin contains the minimum volume of all cells incident to a vertex The variable is used in altering the effect of the grid smoothing algorithm to avoid inverting cells and hence negative volumes rhs contains the residual for grid smoothing xyz contains the z y and z coordinates of the vertices vol contains the volumes of the cells cofg contains the centroids of the cells 150 Chapter 14 Data Structures a c 6 4 5 _ 8 i En 2 b 4 3 1 2 d Figure 14 6 Definition of cell connectivity 14 3 Grid Data Structur
85. fect i e the molecular weight and specific heat at constant pressure are constant default The latter is given by the value of CP from the REFERENCE section The former is computed from 5 2 Input File 37 Solid wall p RVAL2 p RVAL5 u RVAL3 u 0 v 0 v 0 w 0 w 0 p RVALA p RVAL6 Inflow s RVAL1 Solid wall Solid wall Figure 5 2 Configuration and boundary conditions for cylds case the specific heat at constant pressure and the value of GAMMA from the REFERENCE section 3 The gas is considered to be a mixture of thermally and calorically perfect gases The molecular weight and specific heat at constant pressure are computed from mass fraction weighted averages of the molecular weights and specific heats at constant pressure of the mixture constituents see Section 11 3 2 This gas model is available only if species are solved i e if the Rocspecies module is activated 5 The gas is considered to be a pseudo gas The molecular weight and specific heat at constant pressure are computed from appropriately defined averages of the molec ular weights and specific heats at constant pressure of the mixture constituents see Section 11 3 3 This gas model is available only if species are solved i e if the Rocspecies module is activated 38 Chapter 5 File Content and Format Specifications 5 2 7 NUMERICS Section The NUMERICS section contains the following keywords CFL Specifies the CFL number t
86. fied Taylor series Assuming linear interpolation of a scalar variable this gives i Po V o Aros 12 8 where is the interpolated value at location 0 which may be a cell or face centroid or a vertex Assuming that Eq 12 8 is applied to the dy points in the stencil an overdetermined system of linear equations is obtained Axoi AYor 1 E Q Ax A 1 T clo NAP TR ET 12 9 f g AZod AYod 1 Pdo or Ax b 12 10 The system can be inverted using the Singular Value Decomposition SVD which gives a interpolation formula of the form do po N s 12 11 i 1 124 Chapter 12 Algorithms and Methods where the stencil weights are given by Uo Agi for 1 X i do 12 12 12 2 3 Gradient Operators The gradient operators are also constructed using a least squares approach Assuming linear gradient reconstruction of a scalar variable this gives Qi po V o Aros 12 13 where o is now the value at location 0 which may be a cell or face centroid or a vertex Assuming that Eq 12 13 is applied to the do points in the stencil an overdetermined system of linear equations is obtained Azo Ayo Ago Azo Ayo Or l _ 49 12 14 E Oy 0 Zodo AYodo Apoa Or Ax b 12 15 The system can be inverted using the Singular Value Decomposition SVD which gives a formula of the form do V o 3 woiAdoi 12 16 i 1 where the vector of stencil weights is given by wo j
87. force and moment coefficients and their computation can be found in Section 12 7 The computation of forces and moments is governed by the FORCES and the REFERENCE Sections in the input file 4 5 Thrust and Specific Impulse Computation RocfluMP can compute the thrust and specific impulse generated by the fluid on the patches The computation of thrust and specific impulse is governed by the FORCES and the REFERENCE Sections in the input file 4 6 Visualization of Discontinuities RocfluMP can compute Schlieren pictures Shadowgraphs and Interferograms to assist in the visualization of compressible flows with discontinuities The computation of Schlieren pictures Shadowgraphs and Interferograms is governed by the keyword DISCFLAG in the POST Section of the input file 4 7 Visualization and Identification of Vortical Struc tures RocfluMP can compute variables which help visualize and identify vortical structures The computation of these variables is described in Subsection 9 1 2 and is governed by keywords in the POST Section of the input file 4 7 Visualization and Identification of Vortical Structures 29 Figure 4 3 Sample Schlieren picture of flow from open ended shocktube 20 ms after rupture of diaphragm Dark regions indictate high gradients of the density and can thus be used to locate shock waves and contact discontinuities 30 Chapter 4 Capability Descriptions Chapter 5 File Content and Format Specification
88. g values O Simple boundary condition treatment 1 Characteristic boundary coundition treatment NAME Specifies the name of the boundary NSCBCK Specifies the degree of reflection at boundary The boundary is non reflecting if NSCBK 0 0 partially reflecting if NSCBK gt 0 0 and NSCBK lt 1 0 and reflecting if NSCBK71 0 This option is relevant only if KIND 1 and REFLECT 0 ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers PRESS Static pressure Pa The static pressure must only be specified if the outflow is subsonic or mixed subsonic supersonic 62 Chapter 5 File Content and Format Specifications REFLECT Specifies whether the boundary condition is reflecting or non reflecting It can take following values 0 Non reflecting boundary condition 1 Reflecting boundary condition This option is r
89. ge Kutta method derived by Wray XXX ADD REF XXX STARTITER The iteration number from which the computation is to be started Only rele vant if FLOWTYPE 0 STARTTIME The time in seconds from which the computation is to be started Only relevant if FLOWTYPE 1 TIMESTEP The maximum time step in seconds to be used in the computation Only relevant if FLOWTYPE 1 WRIITER Offset between iterations at which flow files are to be written Only relevant if FLOWTYPE 0 WRITIME Offset in time in seconds at which flow files are to be written For unsteady flows with moving grids the grid files are written also Only relevant if FLOWTYPE 1 48 Chapter 5 File Content and Format Specifications 5 2 13 TRANSFORM Section The TRANSFORM section is relevant only to rflupart It contains the following keywords ANGLE_X Angle of rotation around z axis in degrees The angle of rotation is positive in the counter clockwise direction when looking down the x axis ANGLE_Y Angle of rotation around y axis in degrees The angle of rotation is positive in the counter clockwise direction when looking down the y axis ANGLE_Z Angle of rotation around z axis in degrees The angle of rotation is positive in the counter clockwise direction when looking down the z axis FLAG Specifies whether the grid is to be scaled rotated It can take the following values 0 Do not scale and rotate the grid 1 Scale and rotate the grid SCALE_X Scaling factor for z co
90. gles ATTACK and SLIP for farfield boundary condition Angles are positive in direction of arrows PATCH Specifies the range of patches to which the data in this section is to be applied using two integers PRESS Static pressure Pa SLIP Sideslip angle deg See Fig 5 4 for the definition of SLIP TEMP Static temperature K THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch 5 4 1 2 Inflow Boundary BC INFLOW BC INFLOW TOTANG Section The BC INFLOW or BC INFLOW TOTANG section contains the following keywords BETAH Angle between velocity vector and its projection onto xz plane deg See Fig 5 5 for the definition of BETAH BETAV Angle between the projection of the velocity vector onto the zz plane and the positive x axis deg See Fig 5 5 for the definition of BETAV 54 Chapter 5 File Content and Format Specifications FIXED Specifies whether the flow is assumed to be normal to the boundary It can take the following values 0 Inflow not assumed to be normal to boundary 1 Inflow assumed to be normal to boundary Specifying the flow to be normal to the boundary can be particularly helpful if the inflow boundary represents a reservoir condition or other conditions in which the f
91. hat the boundary motion is not propagated well enough into the interior This problem can be remedied by increasing the number of smoothing interations NITER or by increasing the smoothing coefficient SFACT in the GRIDMOTION section of the input file 10 5 Locating Troublespots If problems are encountered rflump prints additional information to allow the user to deter mine precisely the locations at which problems occur For example consider the following output Printing location information Global region 00001 Cell location information Cell x coordinate y coordinate z coordinate Location 1 51 0 32081289E 02 0 83698854E 01 0 12040992E 01 Global patches 1 6 2 300 0 12485188E 01 0 83750698E 01 0 61473528E 03 Global patches 1 6 3 357 0 35957776E 02 0 83795613E 01 0 74193030E 02 Global patch 6 4 1696 0 11128962E 01 0 83783063E 01 0 56762047E 02 Global patches 1 6 5 2048 0 10507133E 01 0 83728027E 01 0 66900307E 02 Global patches 1 6 6 2066 0 87561363E 02 0 83784932E 01 0 89096344E 02 Global patches 1 6 114 Chapter 10 Troubleshooting 7 2116 0 47333590E 02 0 83785244E 01 0 97767197E 02 Global patch 6 8 2574 0 78548965E 03 0 83765772E 01 0 12463422E 01 Global patches 1 6 9 2995 0 48426506E 02 0 83799824E 01 0 90511979E 02 Global patch 6 10 6461 0 45099568E 02 0 83819846E 01 0 11576659E 01 Interior Printing location information done The output lists the cells and the coordinates of the cell centroi
92. iative heat flux W m Outgoing Face Real Tf Temperature K Outgoing Face Real bflag Burning flag Outgoing Face Integer 1 2 3 Chapter 16 File Content and Format Specifications This chapter describes the content and format of RocfluMP files which may require modi fications by a developer The content and format of RocfluMP files which requires or may require editing by a user is described in Chapter 5 The naming of most files follows the conventions outlined in Sect 5 1 16 1 Grid Files 16 1 1 RocfluMP Grid File The grid file contains the coordinates and connectivity information of the grid at a given iteration or time The name of the grid file depends on whether it is written in ASCII or binary format and whether the grid is moving The name of the grid file in ASCII format is e lt casename gt grda_ lt mmmmm gt for steady flows e lt casename gt grda_ lt mmmmm gt _ lt n nnnnnnnE nn gt for unsteady flows with moving grids where lt mmmmm gt is the region number and lt n nnnnnnnE nn gt is the time stamp The name of the grid file in binary format is e lt casename gt grd_ lt mmmmm gt for steady flows e lt casename gt grd_ lt mmmmm gt _ lt n nnnnnnnEtnn gt for unsteady flows with moving grids The format of the grid file in ASCII format is illustrated by the following code fragment WRITE iFile A ROCFLU grid file WRITE iFile A Precision and range WRITE iFile 2 18 PR
93. ies gt o o ao se ses ae cu en 125 12 3 17 Ne 3lip Wall Boundaries ius oso x Re ms 125 12 3 1 3 Injection Wall Boundaries ls 125 8 Contents 12 3 2 Simple Fluid Fluid Boundary Conditions 125 12321 InBow Boundades 23 ce tura so eae rex XS 125 12 322 Odo Boundaries cocina Ems 125 12 3 8 Non Reflecting Boundary Conditions 125 123 3 1 Overall Approach 5 25 22 oo om ox wl 126 12 3 3 2 Inflow Boundary Conditions 129 12 3 5 Outflow Boundary Conditions 5 62424 sses 131 12 3 4 Virtual Boundary Conditions lt lt 133 234 1 Periodia Boundaries 2 249 woo RR 133 1232 Symmetry BoundaB s 026 3 bee eee wee eS 133 124 Temporal Discretizati n Bo sopa do da RR OS 133 12 4 1 Rune Kulita Methods 2 2 ce Sof stredni ONU EY wee m 133 42 Computation of Time SUD es ee dox eR doa ee 3 s 133 DEB EG sc Re eRe or druide Je EEE DEES A ee ORE KR 133 To aL Grd Smoothing 205 2o xo o9 he RAPES A kom hcm Ro HES 133 12 5 2 Discrete Geometric Conservation Law 133 12 5 8 Implementation Details lt lt ec o 22 2 took n 133 12 6 Mass Pressure Skin Friction and Heat Transfer Coefficient Computation 133 127 Force and Moment Computation lt s s cz corr ue eee PL RR 134 13 Implementation Details 137 13 1 N5SCBC Implementation oops za 137 Tots Datestructire oo cocs sedo ow oe RE OE ee RA 137 15 12 Imabzation procedu
94. ile 3 1 1 Installation from CVS Repository To be able to access the CSAR CVS repository set the CVSROOT environment variable to taking the bash shell as an example export CVSROOT pserver user machine uiuc edu cvsroot and either open a new terminal or type user machine source bashrc Then type user machine cvs login and hit the Enter key at the prompt Now move into the directory where you want to install RocfluMP In the following this is assumed to be directory Then type user machine directory cvs co genx Codes RocfluidMP which will check out the source code for RocfluMP from the repository Assuming the checkout command has completed successfully you are now ready to com pile the code for serial computations and you can proceed to Sec 3 2 21 22 Chapter 3 Installation and Compilation 3 1 2 Installation from tar gz File Move into the directory where you want to install RocfluMP In the following this is assumed to be directory Move or copy the gzipped tar file assumed to be lt file gt tar gz in the following into directory Then type user machine directory gzip d lt file gt tar gz user machine directory tar xvf lt file gt tar which will unpack the source code Assuming these commands to have completed successfully you are now ready to compile the code for serial computations and you can proceed to Sec 3 2 3 2 Compilation 3 2 1 Overview of Compilation Pro
95. iles See also Sec 9 1 ERRFLAG Specifies whether errors are to be computed This can only be done if a hard coded initial solution is used See also FLAG in the INITFLOW section EXTRFLAG Specifies whether data is to be extracted from the solution This option can only be used if appropriate code to exctract data was added to the file RFLU ModExtractFlowData F90 in rflupost for the given casename If the appropriate code does not exist rflupost will not extract data write a warning and continue executing At present extracting solution data is possible for the following casenames onera c0 2d 100x50 ONERA CO case 6 skews ms2p0 skews_ms3p0 skews ms4pO Shock diffraction problem of Skews 11 12 st sodi st sod1 mp2 Sod shock tube 13 case 1 st sod2 st sod2 mp2 Sod shock tube 13 case 2 stg 2d Generic shock tube INTERORDER Specifies the polynomial order of interpolation This keyword is only relevant if INTERTYPE 2 It can take the following values 1 Use linear interpolation 2 Use quadratic interpolation 42 Chapter 5 File Content and Format Specifications INTERTYPE Specifies whether and how data is to be interpolated from the cell centers to the vertices It can take the following values 0 Do not transfer the solution from cell centers to vertices This option avoids the introduction of interpolation errors and can be useful if it is necessary to visualize small solution differences which may be smoothed out
96. ing RocfluMP with CHECK UNIFLOW 1 unifMomY contains the y component of momentum when checking RocfluMP for uniform flow preservation The check for uniform flow preservation is activated by compiling RocfluMP with CHECK UNIFLOW 1 unifMomZ contains the z component of momentum when checking RocfluMP for uniform flow preservation The check for uniform flow preservation is activated by compiling RocfluMP with CHECK UNIFLOW 1 14 5 2 Data Type t mixt The data type t mixt contains data related to the mixture and the solution of the associated transport equations The variables associated with the mixture are divided into several types 1 Conserved variables i e dependent variables for which transport equations are solved are stored in the array cv For RocfluMP the conserved variables are fp pu pv pw pE V 2 Dependent variables i e dependent variables for which no transport equations are solved are stored in the array dv For RocfluMP the dependent variables are p T c 3 Transport variables i e dependent variables such as the coefficients of viscosity and conductivity 4 Gas variables i e dependent variables such as the specific heat at constant pressure and the molar mass The dependent transport and gas variables are updated after the update of the conserved variables by calling the routine mixtureProperties F90 Because it is convenient to work with different state variables at times RocfluMP provides r
97. ion index where face index is an eight digit string indicating the face index patch index is a three digit string indicating the global patch index It is important to note that this is a global patch index The patch number is zero if the face is an interior face region index is a five digit string indicating the global region index Note that the names of special face zones do not distinguish between faces types Exam ples are shown in Fig 9 2 which represents a screen dump of part of the zone style menu in TECPLOT Zone number 5 contains cell number 1247 of region 1 Zone number 7 contains face number 13 of global patch 1 of region 1 9 3 ENSIGHT Output 107 Table 9 2 Names and meaning of variable string in ENSIGHT file name variable Meaning Units r Density kg m ry Momentum vector kg m s rE Total internal energy J m p Pressure Pa T Temperature K a Speed of sound m s 9 3 ENSIGHT Output The ENSIGHT files produced by rflupost are in binary format to reduce file size and loading time The binary file are compatible across machines i e files written on little endian machines can be read on big endian machines and vice versa The files produced by rflupost make use of ENSIGHT features which require version 8 or newer To allow visualization of large datasets rflupost can write files suitable for use with the client server version of ENSIGHT Gold The number of servers is specified by the keywor
98. j for 1 i dp 12 17 12 8 Boundary Conditions 125 12 2 4 Inviscid Fluxes 12 2 4 1 Limiter Functions 12 2 4 2 Numerical Flux Functions 12 2 4 3 Entropy Fixes 12 2 5 Viscous Fluxes 12 2 6 Optimal LES Discretization 12 2 6 1 Computation of Integrals 12 2 6 2 Computation of Stencil Weights 12 2 7 Source Terms 12 3 Boundary Conditions 12 3 1 Simple Fluid Solid Boundary Conditions 12 3 1 1 Slip Wall Boundaries 12 3 1 2 No Slip Wall Boundaries 12 3 1 3 Injection Wall Boundaries 12 3 2 Simple Fluid Fluid Boundary Conditions 12 3 2 1 Inflow Boundaries 12 3 2 2 Outflow Boundaries 12 3 3 Non Reflecting Boundary Conditions This section describes formulations to implement Navier Stokes Characteristic Boundary Conditions NSCBC for Euler and Navier Stokes equations Such implementation controls the spurious wave reflection at the computational boundaries and provide means for reflecting and non reflecting boundary conditions treatments Subject matter is this document is partly taken from 2 papers on characteristics based boundary conditions see 8 and 14 126 Chapter 12 Algorithms and Methods 12 3 3 1 Overall Approach Euler Equations are Op Opu Opv Opw 21M cA a 12 18 Opu Opuu Opuv puw Op 12 1 Ot Ox Oy Oz Ox RS Opv Opvu Opvv Opvw Op Ot Ox Oy Oz Oy 12 20 Opw Opwu Opwv Opww Op Ot Ox Oy Oz Oz dd OpE OpHu OpHv OpHw _ 5 Due 3 O 12 22
99. les are written by rfluconv e A grid file in RocfluMP ASCII or binary format e A flow solution file in RocfluMP ASCII or binary format e A surface grid file for TETMESH or YAMS e A version file called rfluconv vrs It contains the version number and date of the executable Successive runs append to the version file 7 1 4 Interactive Input The interactive input after invokation is self explanatory and will not be described here 7 2 rfluinit 7 2 1 Invocation rfluinit is invoked by typing rfluinit c lt casename gt v lt verbosity gt The command line arguments read by rfluinit are lt casename gt A character string used to label the input and output files lt verbosity gt An integer indicating the desired verbosity level of rfluinit The verbosity level can take the following values O No output rfluinit will not write any information to standard output 1 Low level of output rfluinit will write some information to standard output 2 High level of output rfluinit will write detailed information to standard output 7 3 rflumap 81 7 2 2 Input Files The following input files are read by rfluinit e An input file called lt casename gt inp e A grid file in RocfluMP format e A boundary condition file The name of the file is lt casename gt bc A dimension file A restart information file 7 2 3 Output Files The following input files are written by rfluinit e A flow solution file in rfluinit fo
100. level can consist of one or more solution regions 2 A solution region or simply region is obtained from a grid level by partitioning it for parallel processing For sequential processing the region encompasses the entire grid level 3 A grid is defined to be an arbitrary collection of grid cells or simply cells 4 A grid cell is defined to be a convex polyhedron Each cell is composed of faces 5 A face is defined to be a polygon Each face is composed of edges 6 A patch is a collection of faces on the boundary of a region on which the same boundary condition is applied 7 An edge is defined to be a straight line linking two vertices 8 A verter is defined to be a point in space A vertex belongs to at least one cell A vertex is not necessarily equivalent to a node 2 2 Conventions 1 SI Syst me International units are used in RocfluMP and all documents relating to RocfluMP 19 20 Chapter 2 Nomenclature Conventions and Restrictions 2 All coordinate systems are right handed 3 Normal vectors point out of the solution domain 2 3 Restrictions 1 Cells must be hexahedra prisms pyramids or tetrahedra 2 Cells must be conforming i e hanging edges or vertices are not allowed 3 Faces must be triangles or quadrilaterals Chapter 3 Installation and Compilation 3 1 Installation The following assumes that RocfluMP is to be installed either from the CSAR CVS repository or from a gzipped tar f
101. locity vector of the initial solution m s Only relevant if FLAG 1 VELZ The z component of the velocity vector of the initial solution m s Only relevant if FLAG 1 The meaning of the variables IVAL1 IVAL6 and RVAL1 RVAL6 is dependent on the casename and specified in the following list onera cO pipeacoust IVAL1 Specifies wave number of circumferential mode 1 rad IVAL2 Specifies wave number of axial mode IVAL3 Specifies root of derivative of Bessel function of order given by IVAL1 IVAL4 Specifies boundary conditions in axial directions O The pressure disturbance vanishes 1 The gradient of pressure disturbance vanishes RVAL1 Specifies amplitude of pressure disturbance Pa 36 Chapter 5 File Content and Format Specifications Symmetry Solid wall Outflow Injection Figure 5 1 Configuration and boundary conditions for onera_cO case 5 2 6 MIXTURE Section The MIXTURE section contains the following keywords FROZENFLAG Specifies whether mixture conservation are to be updated It can take the following values 0 Do not update mixture conservation equations 1 Update mixture conservation equations default GASMODEL Specifies the gas model The gas model determines how the molecular weight the specific heat at constant pressure and the transport coefficients are computed It can take the following values 1 The gas is considered to be thermally and calorically per
102. low velocity is very small It is important to note that the motivation for and effect of FIXED is different from specifying the flow direction through BETAH and BETAV The keyword FIXED influences the angle computed from the extrapolated velocity vector For vanishing velocity the determination of this angle becomes ill conditioned which can lead to failure of computations Instead the angle can be fixed so that the velocity no matter how small is always normal to the boundary KIND Specifies the treatment of the boundary condition It can take the following values O Simple boundary condition treatment 1 Characteristic boundary coundition treatment MACH Mach number The Mach number must be specified only if the inflow is supersonic NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applie
103. lt mmmmm gt _ lt nnnnnn gt for steady flows e lt casename gt flo_ lt mmmmm gt _ lt n nnnnnnnEtnn gt for unsteady flows The format of the flow solution file in ASCII format is illustrated by the following code fragment WRITE iFile A ROCFLU flow file WRITE iFile A Precision and range WRITE iFile 2 18 PRECISION 1 0_RFREAL RANGE 1 0_RFREAL WRITE iFile A Initial residual WRITE iFile E23 16 global resInit WRITE iFile A Physical time WRITE iFile E23 16 global currentTime WRITE iFile A Dimensions WRITE iFile 2 18 pGrid nCellsTot pRegion mixtInput nCv WRITE iFile A s amp t Mixture density WRITE iFile 5 E23 16 pCv CV MIXT DENS j j 1 pGrid nCellsTot WRITE iFile A amp Mixture x momentum WRITE iFile 5 E23 16 pCv CV_MIXT_XMOM j j 1 pGrid nCellsTot WRITE iFile A Mixture y momentum WRITE iFile 5 E23 16 pCv CV_MIXT_YMOM j j 1 pGrid nCellsTot WRITE iFile A Mixture z momentum 20 21 22 23 24 25 16 3 Dimension File 171 WRITE iFile 5 E23 16 pCv CV_MIXT_ZMOM j j 1 pGrid nCellsTot WRITE iFile A s amp Mixture total internal energy WRITE iFile 5 E23 16 pCv CV_MIXT_ENER j j 1 pGrid nCellsTot WRITE iFile A amp End 16 3 Dimension File The dimension file contains information abo
104. mponent of coordinates SCALE_Y Scaling factor for y component of coordinates SCALE_Z Scaling factor for z component of coordinates 5 2 14 VISCMODEL Section The VISCMODEL section contains the following keywords MODEL Specifies the viscosity model It can take the following values Sutherland viscosity model see Section 11 5 1 1 1 Fixed viscosity The viscosity value is given by the value assigned to the keyword VISCOSITY 2 Antibes viscosity model see Section 11 5 1 2 REFTEMP Specifies the value of the Sutherland constant K SUTHCOEF Specifies the value of the reference temperature Tef in the Sutherland and Antibes models K VISCOSITY Reference value for the dynamic viscosity kg ms o 0 Y O 0 BR Ww M KB m o 5 3 Patch Mapping Files 49 5 3 Patch Mapping Files Patch mapping files are used to define a mapping between the patches defined in the VGRIDns MESH3D TETMESH and Cobalt grid files and that desired in RocfluMP simulations The motivation for mapping patches arises because grids are usually generated with more patches than are required to impose boundary conditions for a simulation For this reason the patch mapping files are typically used to reduce the number of patches and therefore the number of boundary conditions 5 3 1 VGRIDns Patch Mapping File vgi File The format of the vgi file is Line 1 The number of patches after the mapping Line 2 The number of mappings hereafter referr
105. n time The behaviour is specified through pairs of values for the user specified value and the time at which the variation changes The user specified value of the variable MFRATE is modified by the fraction FRAC i e at any time the actual time dependent value is given by the product of MFRATE and FRAC Figure 5 7 illustrates the piecewise constant and linear variations arising from the input of the above example It is important to note that the piecewise constant and linear representations differ in their final value because the final value of FRAC is ignored for linear variations 5 4 2 2 TBC SINUSOIDAL Section The TBC_SINUSOIDAL section allows the specification of sinusoidal variations of the form a t c 1 Asin wt where c is the user specified constant value A is the amplitude if the sinusoid w is the angular frequency and is the phase Consider the following example section 5 4 Boundary Condition File 67 1 0 0 9 0 8 0 7 0 6 0 5 MF RATE 04 0 3 0 0 T a Piecewise constant 2 Piecewise linear 0 000 0 001 0 002 0 003 Time s 0 004 0 005 0 006 Figure 5 7 Illustration of piecewise constant and linear interpolations for TBC PIECEWISE time dependent boundary conditions gt A N n O 0 N Oo wc 68 Chapter 5 File Content and Format Specifications TBC_SINUSOIDAL OUTFLOW PRESS BC an
106. n treatment NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers PRESS Static pressure Pa Only read if TYPE 0 REFLECT Specifies whether the boundary condition is reflecting or non reflecting It can take following values 0 Non reflecting boundary condition 1 Reflecting boundary condition This option is relevant only if KIND 1 TEMP Static temperature K THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 5 4 Boundary Condition File 57 1 Compute thrust and specific impulse for this patch TYPE Specifies whether inflow is supersonic or subsonic It can take th
107. ndary conditions Negative positive definite quantity detected Negative positive definite solution variables such as density pressure or temperature were detected This is often the result of using a CFL number or a time step which is too large 10 4 Other Problems 113 For coupled simulations negative positive definite variables may also arise if the geometry becomes heavily deformed Negative volume s detected One or more cells with negative volumes were detected For computations without moving grids this indicates that the cell or boundary face connectivity is incorrect For computations with moving grids negative volumes indicate that either the boundary deformation is too strong given the grid motion parameters specified by the user or the grid quality is poor and deteriorates with grid motion In the latter case the grid quality needs to be improved before a new run is started Face sum greater than minimum face area rflump checks whether cells are closed by computing the sum of the face vectors and compar ing them against a presently hardcoded tolerance It the sum of the face vectors exceeds the tolerance it is likely that the cell or boundary face connectivity is incorrect 10 4 Other Problems This section lists problems and suggested remedies for problems which do not lead to warn ings or errors Cells near moving boundaries become highly stretched Highly stretched cells near moving boundaries indicate t
108. ng formats are supported TECPLOT format Tecplot Inc Bellevue WA ENSIGHT format Computational Engineering International Apex NC 16 Chapter 1 Introduction RocfluMP Visualization Figure 1 1 Overview of RocfluMP and related codes 1 3 Contributors 17 e rflupart is the partitioning module of RocfluMP It converts grid files from outside formats into binary or ASCII files in RocfluMP format and partitions the grids into regions At present the following formats are supported HYBRID format CENTAUR grid generator CentaurSoft Austin TX VGRIDns format VGRIDns grid generator Shahyar Pirzadeh NASA Langley MESH3D format MESH3D grid generator Tim Baker Princeton University TETMESH format TETMESH grid generator SIMULOG Fran Cobalt format Cobalt flow solver Cobalt Solutions LLC Springfield OH GAMBIT format GAMBIT grid generator Fluent Lebanon NH 1 3 Contributors The following people listed in alphabetical order have contributed to the development and testing of RocfluMP Mark Brandyberry Testing application of RocfluMP to fluid structure interaction prob lems using GENx Michael Campbell Research Programmer CSAR Performance analysis and tuning command line interface assistance with porting of RocfluMP to new computing platforms Heath Dewey Graduate Student Department of Aeronautical Engineering UIUC Imple mentation of Newton Krylov
109. ng values 0 Do not write volume data for postprocessing only surface data is written 1 Write volume and surface data for postprocessing The advantage of not writing volume data to the output file is to reduce the amount of data to be visualized SPECFLAG Specifies whether the postprocessor information file produced by rflupick is to be read It can take the following values 0 Do not read the postprocessor information file 1 Read the postprocessor information file Reading the postprocessor information file allows only specific regions to be postpro cessed This option is only relevant if MERGEFLAG 0 44 Chapter 5 File Content and Format Specifications 3 6 9 2 5 8 1 4 7 a b Figure 5 3 Partitioning of a quadrilateral grid into two regions using PARTMODE 2 demonstrating importance of cell numbering WRIMERGE Specifies whether merged grid and solution files are to be written See also MERGEFLAG VORTFLAG Specifies whether vorticity is to be computed and written to visualization files It can take the following values 0 Do not compute vorticity and do not write to visualization files 1 Compute vorticity and write to visualization files See also Sec 9 1 5 2 9 PREP Section The PREP section is relevant only to rflupart It contains the following keywords PARTMODE Specifies the partitioning mode It can take the following values 1 Partition the grid with
110. ns 2 At the second level are solution region A solution region is defined as the entire solution region for sequential processing applications or a single partition for parallel processing applications Each multigrid level can consist of an arbitrary number of domains Note that the multigrid levels are located atop the solution regions This means that each multigrid level is partitioned separately for parallel processing Because intra layer commu Level EN Region Figure 14 1 Overview of data structure layers 141 142 Chapter 14 Data Structures Region m Patches Global region index m Mixture Global data rm a Mixture input Grid s m Figure 14 2 Overview of region data structure nication is more important than inter level communication the possibility of optimizing each multigrid level for load balancing separately should allow for better parallel performance The types t level and t region are defined in ModDataStruct F90 14 2 Region Data Structure The region data structure contains all the information required to solve the governing equa tions in a given region A schematic overview of the region data structure is given in Fig 14 3 The region data str
111. nsverse directions Stencils are computed using RFLU BuildBF2CStencilWrapper which further calls routines for 1d 2d or 3d stencil computations After stencils are computed weights are computed using RFLU ComputeWtsF2CWrapper Once stencil and weights are available gradients are computed through RFLU ComputeGradBFacesWrapper which further uses RFLU Comput eGradBFaces 1D to compute gradient using 1d stencils and RFLU ComputeGradBFaces to compute gradients using 2d 3d stencils 13 1 5 Characteristic wave amplitude computation This is the most important step in implementing NSCBC Local One Dimensional Invis cid LODI relations are used at boundary face which uses derivatives of flow variables in 13 2 Parallelization 139 face normal direction There is need to resolve x y z components of gradients in normal and transverse directions at boundary face Time derivatives of conservative variables are expressed in terms of wave amplitudes incoming or outgoing wave and space derivatives in transverse directions This task is carried out in RFLU NSCBC CompRhs which loops over each boundary patch and depending on boundary type a specific subroutine is called to compute right hand side of system of equation at boundary e g RFLU NSCBC CompRhsOutflow for outflow RFLU_NSCBC_CompRhsFarf for farfield boundary condition and so on 13 1 6 Time integration RK 4 is used for time integration It is point quantity at boundary face unlike averag
112. o be used during computations CRECONSTC Specifies whether constrained reconstruction is to be used to compute cell gra dients It can take the following values 0 Do not use constrained reconstruction default 1 Use constrained reconstruction CRECONSTF Specifies whether constrained reconstruction is to be used to compute face gradients It can take the following values 0 Do not use constrained reconstruction default 1 Use constrained reconstruction DIMENS Specifies the dimensionality of the computation It can take the following values 2 Run two dimensional computation 3 Run three dimensional computation default The dimensionality flag is provided so that truly two dimensional computations can be performed i e a computation in which the grid contains only one cell in the z direction Three dimensional computations must contain at least three cells in all directions to be able to compute gradients If you wish to run truly two dimensional computations there must be two boundary patches which coincide with z constant planes and you should specify the boundary condition on those patches to be BC_VIRTUAL see Sect 5 4 1 11 DISCR Specifies the discretization scheme for the inviscid fluxes It can take the following values 1 The flux difference splitting scheme of Roe 10 3 The HLLC approximate Riemann solver of Batten et al 3 4 The AUSM scheme of Liou 7 5 2 Input File 39 The flux differenc
113. o4 ca eee ea eee He eRe 68 5 13 Grid Motion Boundary Conditions lt 2 2465654528465 68 ho lIegon Mappig File 222222299 ed tot pdt omadi ex xS 69 56 Betari inbarmebon File oo enoe os RG we eR Oe og ELA 70 Ar Ser eee aula Re Oe Oe os ur cep Be o 71 Bos Probe File coc hb zoo SEES RES CEE EEE HERES 71 5 9 Mass Conservation Check File T2 AULA File o creara ss rea BES aS OS LS Oe ded E dede oe E T2 5 11 GENx Control File ek wo oo e m m m A 72 6 Problem Setup 75 7 Execution 79 JI AGOI nox eee eee ee e RAM EE ROSE EAS 79 Xd RO ce ee eK E A 79 TL2 Input Fleg 222 x AS AEDS EEE DS PAAR ER o ES 80 TLS SOUR Files a nou se xa he ees dos wo E Rd 80 AE Interactive 2 e cea ee a ORE RE ESSE OEE GS 80 Wee UNG oe Gk ew dda we Ge owe HS eRe Se RS EE e ed 80 TL A oa ce n ace deo o He ch ER He mm BoB up dog 80 Tope IU DIE cia wo xe RUE RUE UE eee eres d 81 pe SO Files Lou Da te Pe bh oe GG e E elt eS GR xd 81 oe PUNO sek ee Ik So Got Ee Gwe eS SOS ROS ERS HERS OR 81 Tol PDGREDE 4 uuo Sa Peek ye amp OR hoe Ro RU Oy SOE OE ws 81 10 VA PEE ee ee Yo te tA eee BOS ee ee a eee x 82 tas PUG FES sec aerer aatan Eee Ow oo Ooo EERE ES 82 72 WM no oe ee PRE SEE Eee eee RUE eS Oe eh Ree A 82 TAL OO cs de eh aK ee EE eer eee ee ee 82 TAM Tapa Ue 22d AOE EERE SPEAKER Ea 83 T2523 SUE Fiks cu esee ee es oe AA e 83 7 4 4 Profiling and Performance Analysis Guidelines 84 To oe a ae A A EA E AE
114. of t grid mixt is the user define data type containing all the information relating to mixture It is described in Sec 14 5 2 patches is the user defined data type containing all the information relating to boundary patches See Sect 14 4 for a description of t patch global is a pointer to global data mixtInput is a user defined data type containing all the user defined input for the solution of the mixture equations It is described in Sec 14 5 1 14 3 Grid Data Structure The grid data structure contains information relating to the description of the grid An overview of the grid data structure is given in Fig 14 4 The grid data structure is defined in ModGrid F90 Some additional RocfluMP specific data is defined in RFLU ModGrid F90 which is mainly used in converting from exterior grid formats to that used by RocfluMP and in helping to construct some data structures The two modules are discussed in detail below 14 3 1 Module ModGrid F90 In understanding the grid data structure the following points are important 1 RocfluMP can operate on grids consisting of arbitrary combinations of tetrahedral hexahedral prismatic and pyramidal cells As indicated in Sec 2 1 these are referred to as instances of different cell types When running RocfluMP in parallel one also has to distinguish between actual and virtual cells so RocfluMP introduces the concept of a cell kind to distinguish between these 2 RocfluMP categori
115. ollowing values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers STATS Specifies whether particle impact statistics are to be gathered It can take the following values 5 4 Boundary Condition File 61 0 Do not gather paticle impact statistics 1 Gather paticle impact statistics Particle impact statistics are written into the patch coefficient file for visualization with TECPLOT TEMP Specifies the value of the temperature K THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch 5 4 1 7 Outflow Boundary BC OUTFLOW Section The BC OUTFLOW section contains the following keywords KIND Specifies the treatment of the boundary condition It can take the followin
116. omputing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers STATS Specifies whether particle impact statistics are to be gathered It can take the following values 0 Do not gather paticle impact statistics 1 Gather paticle impact statistics Particle impact statistics are written into the patch coefficient file for visualization with TECPLOT THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch 5 4 1 10 Symmetry Boundary BC SYMMETRY Section NAME Specifies the name of the boundary PATCH Specifies the range of patches to which the data in this section is to be applied using two integers 5 4 Boundary Con
117. ones lt 5 26 SR REX XE 104 0234 Special Cell Zones ic RS 106 0 2 3 5 Special Pace Zones lt 2 ss svi sane Yi 39k cS ed 106 93 ENSIGHT UD ec aesae a Bred pw eg Oe OW SO oe s 107 9 3 1 File Naming Convention 52 24 5444 x o RR ERR 107 93 2 Part Naming Conyentions 2 222299 46 Ree ee hee SS 107 U o 0 8 Vohlese Parts 25 x Ro o RR GR Eo aed cabs E ad 108 0222 Boundary Publ Pare eom bw oe oo A 110 10 Troubleshooting 111 10 1 General Considerations 2 6 0 6564 9393 3X RR Oe ES es 111 10 2 Explanations of Warnings lt c coe e eee o 9k EA 112 10 3 Explatations ol Errors lt lt e s sore s debo kae REOR esas es 112 104 Other Problems o dud naiai 3 o3 Rx Peel o9 X4 XR EG Y es 113 10 5 Locating Troublespots a c ce ee he kG EEO AREA sara 113 Contents 7 III Developer and Reference Manual 115 11 Governing Equations 116 11 1 The Navier Stokes Equations e e e 116 11 2 The Geometric Conservation Law llle LIT 113 Lb Molek uy o4 ic A RE ye eS Se eS BRE od 118 11 3 1 Calorically and Thermally Perfect Gas gt a 22 118 11 3 2 Mixture of Calorically and Thermally Perfect Gases 118 1L33 o ca ee ee OS ww RR ee Ee SOS OX Kok S P Rog 118 ILI Thermodynamie Properties e oso fei we vb eke XO d ee X ede x63 118 11 4 1 Calorically and Thermally Perfect Gas 118 11 4 2 Mixture of Calorically and Thermally Perfect Gases 118 LA Pee DE e he
118. ons 1 3 Process 000004 1 4 End Line 1 The first line must contain the string shown otherwise reading of the file will fail Lines 2 5 Specify the number of regions and processes Lines 6 8 Specify that process one will run one region and that that region s index is one Lines 9 17 Specify that processes two to four will run one region each and that the corre sponding region indices are two three and four respectively Line 18 The last line must contain the string shown otherwise reading of the file will fail It is instructive to consider the two additional examples shown below On the left is the region mapping file for the case in which four regions are to be run on two processes On the right is the region mapping file for the case in which four regions are to be run on a single process ROCFLU region mapping file ROCFLU region mapping file Number of regions Number of regions 4 4 Number of processes Number of processes 2 1 Process 000001 Process 000001 2 4 1 1 2 2 Process 000002 3 2 4 3 End 4 End 5 6 Restart Information File The restart information file contains the iteration numbers or time stamps at which restart files were written by rflump It is read by rflump to determine the iteration number or time stamp from which a restart should be made Restarts are always made from the iteration number or time stamp on the last line in the restart information file The initial restart
119. ontains the components of the unit face normal vector Used only if GENX 1 rhofvfAlp contains the components of the product of the fluid density times the fluid ve locity Note that this is not the same as mdotalp as the former also includes the effect of boundary motion due to deformation Used only if GENX 1 tracf contains the fluid traction for each boundary patch face Used only if GENX 1 xyz contains the x y and z coordinates of the vertices Used only if GENX 1 valMixt contains the user specified values for the enforcement of boundary conditions on the mixture 156 Chapter 14 Data Structures TYPE t mixt input INTEGER flowModel LOGICAL moveGrid externalBc INTEGER nDv nTv nGv nGrad indCp indMol REAL RFREAL prLam prTurb scnLam scnTurb turbulence modeling INTEGER turbModel species INTEGER specModel continuum particles LOGICAL conPartUsed discrete particles LOGICAL disPartUsed radiation LOGICAL radiUsed numerics INTEGER spaceDiscr spaceOrder pSwitchType INTEGER timeScheme nrkSteps ldiss 5 REAL RFREAL cfl smoocf vis2 vis4 pSwitchOmega limfac epsentr REAL RFREAL ark 5 grk 5 trk 5 betrk 5 flow initialization used within preprocessor REAL RFREAL iniVelX iniVelY iniVelZ iniPress iniDens flow initialization for uniform flow preservation check REAL RFREAL unifDens unifEner unifMomX unifMomY
120. ontrol txt file see Sec 5 11 9 You are ready to run either rflump or GENx 78 Chapter 6 Problem Setup Chapter 7 Execution This chapter contains detailed information on the command line arguments and input and output files of rfluconv rfluinit rflumap rflump rflupart rflupick and rflupost 7 1 rfluconv 7 1 1 Invocation rfluconv is invoked by typing rfluconv c lt casename gt s stamp v lt verbosity gt The following command line arguments are read by rfluconv lt casename gt A character string used to label the input and output files lt stamp gt A variable indicating the iteration or time from which the grid and solution files are to be read lt verbosity gt An integer indicating the desired verbosity level of rfluconv The verbosity level can take the following values O No output rfluconv will not write any information to standard output 1 Low level of output rfluconv will write some information to standard output 2 High level of output rfluconv will write detailed information to standard output rfluconv expects interactive user input after invokation as described in Sect 7 1 4 79 80 Chapter 7 Execution 7 1 2 Input Files The following input files are read by rfluconv A grid file in RocfluMP ASCII or binary format A flow solution file in RocfluMP ASCII or binary format e A dimension file e A boundary condition file 7 1 3 Output Files The following output fi
121. or and dummy pyra midal cells 148 Chapter 14 Data Structures nTets is the number of tetrahedral cells in the grid nTetsTot is the total number of tetrahedral cells in the grid i e interior and dummy tetrahedral cells nVert is the number of vertices in the grid nVertTot is the total number of vertices in the grid i e interior and dummy vertices hexFlag contains a flag indicating the kind of a given hexahedral cell It is read in from the RocfluMP grid file and can only take the values CELL KIND BNDRY CELL KIND ACTUAL and CELL KIND VIRTUAL defined in ModParameters F90 hex2CellGlob contains the mapping of a given hexahedral cell to a global cell priFlag contains a flag indicating the kind of a given prismatic cell It is read in from the RocfluMP grid file and can only take the values CELL KIND BNDRY CELL KIND ACTUAL and CELL_KIND_VIRTUAL defined in ModParameters F90 pri2CellGlob contains the mapping of a given prismatic cell to a global cell pyrFlag contains a flag indicating the kind of a given pyramidal cell It is read in from the RocfluMP grid file and can only take the values CELL KIND BNDRY CELL KIND ACTUAL and CELL KIND VIRTUAL defined in ModParameters F90 pyr2CellGlob contains the mapping of a given pyramidal cell to a global cell tetFlag contains a flag indicating the kind of a given tetrahedral cell It is read in from the RocfluMP grid file and can only take the values CELL KIND
122. outines to change the state of the state vector from conserved variables to two different sets of primitive variables This is advantageous when computing gradients for the viscous fluxes and printing information on the solution The possible states are as follows 1 Conserved variables given by p pu pv pw pE This is the default state and indi cated by cvState having the value CV MIXT STATE CONS The value of the integer pa rameter CV MIXT STATE CONS and the corresponding parameters for the other states is defined in ModParameters F90 2 Primitive variables given by p u v w p This state is indicated by cvState having the value CV MIXT STATE DUVWP 3 Primitive variables given by p u v w TY This state is indicated by cvState having the value CV MIXT STATE DUVWT 160 Chapter 14 Data Structures TYPE t mixt REAL RFREAL POINTER cv cvOld dv tv gv ifdef STATS REAL RFREAL POINTER tav endif REAL RFREAL POINTER rhs rhsSum diss fterm INTEGER cvState REAL RFREAL DIMENSION POINTER cvVrtx REAL RFREAL DIMENSION POINTER bGradFace gradCell gradFace END TYPE t mixt Figure 14 10 Definition of data type t_mixt Changes of the state are effected by USEing the module RFLU_ModConvertCv F90 and calling the routines RFLU ConvertCvCons2Prim pRegion cvStateFuture to convert from conserved variables to primitive
123. put i e e Deactivate printing to screen by setting the verbosity level to zero e Deactivate checking by setting the checking level to zero e Deactivate the computation of forces This is particularly important for parallel runs since the computation of forces and moments is likely to introduce a load imbalance and requires additional communication e Do not specify the extraction of data through probes This is particularly im portant for parallel runs since the writing of probe data is likely to introduce a load imbalance 3 Choose an appropriate number of regions for a given number of processors and machine configuration Many parallel computers are divided into so called nodes each of which may hold a given number of processors For example assume a given machine contains 16 processors on each node If a run is to be made on 1024 processors 64 nodes of this machine would be required In measuring outright performance of the code however it is not a good idea to use all the processors on a given node This is because the operating system will require some resources also and then slow down the computation of the region s assigned to that processor It is therefore recommended that one processor on each dataset be left unused on such machines Hence it is necessary to request more nodes or to create datasets which are multiples of N 1 where N is the number of processors per node 7 5 rflupart 85 7 5 rflupart 7 5
124. q quit Enter information type Once again the output is self explanatory Once special faces were picked rflupick will re peat the same procedure for the next region Picking some regions Assuming that the user enters option s rflupick will present the following screen output Enter global region index 1 to exit rflupick will repeat the same request until the user enters an integer smaller than unity Once some regions were picked rflupick will loop over the picked regions and present the same output as described above 90 Chapter 7 Execution 7 7 rflupost 7 7 1 Invocation For serial computations rflupost is invoked by typing rflupost c lt casename gt s stamp v lt verbosity gt The following command line arguments are read by rflupost lt casename gt A character string used to label the input and output files stamp A variable indicating the iteration or time from which the grid and solution files are to be read lt verbosity gt An integer indicating the desired verbosity level of rflupost The verbosity level can take the following values O No output rflupost will not write any information to standard output 1 Low level of output rflupost will write some information to standard output 2 High level of output rflupost will write detailed information to standard output 7 7 2 Input Files The following input files are read by rflupost e A grid file in RocfluMP format e A flow sol
125. r bolic Conservation Laws J Comp Phys 27 1 31 1978 14 Thompson K W Time Dependent Boundary Conditions for Hyperbolic Systems J Comp Phys 68 1 24 1987 15 Venkatakrishnan V Convergence to Steady State Solutions of the Euler Equations on Unstructured Grids with Limiters J Comp Phys Vol 118 No 1 pp 120 130 April 1995 16 Wang Z J Improved Formulation for Geometric Properties of Arbitrary Polyhedra AIAA J 37 10 1326 1999
126. r s oo co so goros g e Ox xe RR Eg 138 131 3 Data read write routines su ee p oe ox wood gene eee eed 138 13 1 4 Gradient computation at boundary lt lt ore 22s 138 13 1 5 Characteristic wave amplitude computation 138 18 10 Time miegraliod 2 222w9 serian ok Rx 139 13 1 7 Flux computation using boundary data 139 13 2 Pena eu ou ase mor ooh Ea ee Xo er 3 GR Reb qo do do dos 139 14 Data Structures 141 14 1 Philosophy and Abstraction oe oso ox E Aw ADEE ERR rosdi 141 14 2 Region Data Sirnetire 2 222 o ox mb uox mou wo Rob dy 142 113 Grd Data SUV oe raus on moe 9 Row eb A x A Xe boh mo doe hw o deg 144 143 1 Module ModGrid F90 ge eee eee ERROR RO RUE r 144 14 22 Module RFLU ModGrid F90 coc cee RR hee XXE x3 152 144 Boundary Date Sue co se c eo siosan RO EERE KES 152 14 5 Bene Dae DITUGUTE deo kso ee eee m xU E x 3 0x Cw ee 157 14 5 1 Data Type Cee lt s coss ook ei OA iga ki 157 1482 Data Type EM XE e suos 4444444 5428 RR aa RR Rd 159 Contents 9 15 GENx Integration 163 15 1 Definition of Attributes kn hk wh a ewe raed vedea Rew xx box 163 ISLlI Lnd AUS o oo cir as keel Be eS A ARA 163 15 L1 1 Volume Pan sB 2 x E 163 15 1 12 Surface Pares cc xES 163 15 1 3 Solution Attributes lt oa sirarsa neres bee RR Rx RR x 163 15 1 3 Interface Attributes oac soo cak eee 163 15 14 1 Mon Inberaeting Panes 44 2 4 isis RR EX es 163 ILLAS NonBurning Fanes gt
127. r 13 Implementation Details 13 1 NSCBC Implementation This document describes implementation of Navier Stokes Characterstics based Boundary Conditions in Rocflu a three dimensional time dependent compressible Navier Stokes equa tions solver on moving and or deforming unstructured grids NSCBC implementation would require following procedures e Data structure Initialization procedures Data read write routines e Gradient computation at boundary Characteristic wave amplitude computation e Time integration e Flux computation using boundary data following sections would describe each of above in detail 13 1 1 Data structure NSCBC involves solving Navier Stokes equations at boundary following variables are needed for proper implementation of NSCBC e Conservative and Dependent variables are needed at each boundary face These are updated by solving flow equations at boundary and these describes complete state of flow system at boundary at all time 137 138 Chapter 13 Implementation Details e A flag variable bcKind is needed to determine if a particular boundary is to be treated as NSCBC NSCBC routines are called if this variable bcKind BC KIND NSCBC e Boundary condition data is required at each boundary It is read from user specified boundary condition inputfile containing details of boundary type bcType bcName and other details needed for complete implementation of boundary condition e Gradient a
128. rd 00000 The initialization module is executed by typing 91 92 Chapter 8 Example Cases rfluinit c shocktube v 2 which produces the file shocktube flo 00000 0 00000E 00 The solver may be run by typing rflump c shocktube v 2 With the settings contained in the files from the CVS repository rflump will take as many time steps as needed to reach a physical time of t 1 0 107 s The results may be visualized with TECPLOT The postprocessor is used to interpolate the solution variables from cell centroids to vertices and to write the TECPLOT data file by typing rflupost c shocktube s 1 0E 3 v 2 The resulting data file may be read into TECPLOT by typing tecplot shocktube 00000 1 00000E 03 p1lt The contours of density are depicted in Fig 8 1 The CVS repository contains a TECPLOT macro file called lineplots mcr which can be used to extract line plots of density velocity and pressure along the x axis as shown in Fig 82 8 2 GAMM Bump The transonic flow over a circular arc bump in a straight walled channel is often used to test the inflow and outflow boundary conditions and the shock capturing capabilities of the spatial discretization This case will be used to illustrate both serial and parallel computations The grid used in this computation is shown in Fig 8 3 The green surface indicates the inflow boundary The outflow boundary is hidden from view in this figure The following files are required for thi
129. rdcode The hardcode depends on the casename This option can only be used if appropriate code to initialize the solution was added to the file RFLU InitFlowHardCode F90 in rfluinit for the given casename If the appropriate code does not exist rfluinit will return an error At present hardcoded initial solutions for the following casenames are provided onera c0 ONERA CO case 6 5 2 Input File 35 pipeacoust Pipe acoustics cases ringleb Ringleb flow 9 st sod1 Sod shock tube 13 case 1 st sod2 Sod shock tube 13 case 2 ssvort lt t gt lt mxn gt 1 lt p gt Supersonic free vortex lt t gt h p h and p denote hexahedral and prismatic grids respectively lt mxn gt denotes the grid resolution for both hexahedral and prismatic grids It can take the following values 20x5 40x10 80x20 160x40 320x80 lt p gt denotes the number of layers in the z coordinate direction It can take values 1 or 3 IVAL1 IVAL6 Integer variable used to set hard coded initial conditions Its meaning is dependent on the casename as specified below Only relevant if FLAG 3 PRESS The static pressure of the initial solution Pa Only relevant if FLAG 1 RVAL1 RVAL6 Real variable used to set hard coded initial conditions Its meaning is depen dent on the casename as specified below Only relevant if FLAG 3 VELX The x component of the velocity vector of the initial solution m s Only relevant if FLAG 1 VELY The y component of the ve
130. rder for simplicity 5 4 1 Physical Boundary Conditions 5 4 1 1 Farfield Boundary BC FARF Section The BC FARF section contains the following keywords ATTACK Angle of attack deg See Fig 5 4 for the definition of ATTACK CORR Specifies whether farfield point vortex correction is to be applied It can take the following values 0 Do not apply point vortex correction 1 Apply point vortex correction Note that at present the correction can only be used for two dimensional computations and has not been thoroughly tested KIND Specifies the treatment of the boundary condition It can take the following values O Simple boundary condition treatment 1 Characteristic boundary coundition treatment MACH Mach number NAME Specifies the name of the boundary ORDER Specifies the order of accuracy of computing fluxes on this boundary It can take the following values 1 First order accuracy 2 Second order accuracy The order with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries 5 4 Boundary Condition File 53 SLIP ATTACK X Figure 5 4 Definition of an
131. rder with which boundary fluxes are computed must be less or equal to the order of accuracy with which interior fluxes are computed as specified by the keyword ORDER in the NUMERICS section It can be useful to compute boundary fluxes with first order accuracy even though interior fluxes are computed with second order accuracy if strong gradients are swept across the boundary such as shock waves or jet boundaries PATCH Specifies the range of patches to which the data in this section is to be applied using two integers STATS Specifies whether particle impact statistics are to be gathered It can take the following values 0 Do not gather paticle impact statistics 1 Gather paticle impact statistics Particle impact statistics are written into the patch coefficient file for visualization with TECPLOT TEMP Injection static temperature K THRUSTFLAG Specifies whether thrust and specific impulse are computed for this patch This flag is relevant only when FLAG 1 in the FORCE section It can take the following values 0 Do not compute thrust and specific impulse for this patch 1 Compute thrust and specific impulse for this patch 5 4 1 5 No Slip Boundary BC NOSLIP HFLUX Section The BC NOSLIP HFLUX section is used to specify boundary conditions for patches with a no slip boundary condition and imposed heat fluxes and contains the following keywords COUPLED Specifies whether patch is interacting during a computation with GENx It
132. ribe each section in the input file and the associated keywords in detail For simplicity the sections are listed in alphabetical order but they may appear in any order in the input file 5 2 1 FLOWMODEL Section The FLOWMODEL section contains the following keywords MODEL Specifies which equations are to be solved It can take the following values O rflump solves the Euler equations 1 rflump solves the Navier Stokes equations MOVEGRID Specifies whether grid motion is active or not It can take the following values O Grid motion is inactive 1 Grid motion is active 5 2 2 FORCES Section The FORCES section governs the computation of forces on moments on the patches of the geometry The computation of the forces and moments and the necessary conventions are described in Section 12 7 The FORCES section contains the following keywords FLAG Specifies whether forces and moments are to be computed It can take the following values 0 Do not compute forces and moments 1 Compute forces and moments Activating the computation of forces and moments leads to the forces and moments being printed on the screen and written to files PATCHFLAG Specifies whether pressure skin friction and heat transfer coefficients are to be written to output files It can take the following values 0 Do not write coefficients to output files 1 Write coefficients to output files 5 2 Input File 33 REFAREA Specifies value of reference
133. rmat e A version file called rfluinit vrs It contains the version number and date of the executable Successive runs append to the version file 7 3 rflumap 7 3 1 Invocation rflumap is invoked by typing rflumap c lt casename gt m lt mode gt p lt nprocs gt r lt nregions gt v lt verbosity gt The following command line arguments are read by rflumap lt casename gt A character string used to label the input and output files lt mode gt An integer indicating the mode of invokation The mode can take the following values 1 Initial mode rflumap will only create a mapping file 2 Final mode rflumap will read an existing mapping file and create an input file for Rocin lt nprocs gt An integer indicating the number of processes The number of processes must be less or equal to the number of regions 82 Chapter 7 Execution lt nregions gt An integer indicating the number of regions The number of regions must be greater or equal to the number of processes lt verbosity gt An integer indicating the desired verbosity level of rflumap The verbosity level can take the following values O No output rflumap will not write any information to standard output 1 Low level of output rflumap will write some information to standard output 2 High level of output rflumap will write detailed information to standard output 7 3 2 Input Files rflumap does not read any input files 7 3 3 Output Files rflum
134. rrays at boundary These are computed and stored for each time step 13 1 2 Initialization procedures Navier Stokes equations are initial boundary value problems IBV which require an ini tial solution which is evolved in time Flow variables at NSCBC boundaries are initial ized using routine RFLU BXV InitVars which further calls specific initialization routine depending on type of boundary condition like RFLU NSCBC InitSlipWall for slip walls RFLU_NSCBC_InitOutflow for outflow boundary and so on 13 1 3 Data read write routines Boundary variables are written in seperate files than interior solution files Routines used to read write boundary solution files are RFLU_BXV_ReadVarsWrapper RFLU_BXV_WriteVarsWrapper These wrapper routines further call corresponding routines to read write binary or ASCII datafiles e g reading routines are RFLU_BXV WriteVarsASCII and RFLU BXV WriteVarsBinary similarly there are writing routines 13 1 4 Gradient computation at boundary Gradients of the primitive variables at boundary are computed using RFLU ComputeGradBFacesWrapper Gradients are computed using least square gradient reconstruction technique which need 2d 3d boundary face to cell stencils When boundary is aligned with one of axis and interior cells are lying in curvilinear direction normal to boundary face then 1d stencil can be used to compute normal derivatives of variables This is useful only when there is no flow variation in tra
135. s This chapter describes the content and format of RocfluMP files which require or may require editing by a user to prepare a computation or interpret results obtained from a computation The visualization files produced by rflupost are described in Chapter 9 The content and format of RocfluMP files which cannot or must not be modified by a user but may have to be modified by a developer are described in Chapter 16 5 1 Filename Conventions The majority of files share a user specified string the so called case name represented by casename below Many of the files whose format is described below consist of a region index and either an iteration or a time index The region index indicates the global region number with which a given file is associated A region index of zero indicates that the given file is associated with a serial or unpartitioned data set A region index of one or greater indicates that the given file is part of parallel data set 5 2 Input File The input file is called lt casename gt inp The input file is divided into sections Each section contains several lines each of which consists of a keyword and a value as shown below SECTION_NAME KEYWORD_1 VALUE_1 KEYWORD_2 VALUE_2 KEYWORD_3 VALUE_3 31 32 Chapter 5 File Content and Format Specifications Comments may be inserted after the specification of the values they are ignored by the routines reading the input file The following sections desc
136. s case gamm8 bc gamm8 inp gamm8 hyb bin The GRIDSRC variable in the FORMATS section of the gamm8 inp file indicates that the original grid file is in CENTAUR format 8 2 GAMM Bump 93 rho 8 8068E 01 6 4205E 01 4 0341E 01 1 6477E 01 Figure 8 1 Density contours for shock tube computation at t 1 0 107 s with initial conditions given by Sod s first case Sod case 1 coarse grid 300 100000 90000 250 80000 200 70000 a E T E g E 360000 Z Lx 150 L 7 8 50000 5 Kk ki gt 100 40000 30000 50 20000 015 05 0 0 05 io Lee x m Figure 8 2 Line plots for shock tube computation at t 1 0 107 s with initial conditions given by Sod s first case 94 Chapter 8 Example Cases lt gt NAAA ESI n ds Ls Ly b AL A UV y AT UV n do ADU V END ty i A AM Y N VL jn b AS ana ANG Ay VL NY A WA y Z DA V 5 CI AD A A A Grid used for GAMM bump computation Figure 8 3 8 2 1 Serial Computation By typing rflupart c gamm8 v 2 the partitioner runs and writes the following output files gamm8 dim gamm8 grd 00000 gamm8 f1lo 00000 000000 By typing v 2 rfluinit c gamm8 the partitioner runs and writes the following output files 000000 gamm8 flo 00000 The solver may be run by typing 2 rflump c gamm8 v 8 2 GAMM Bump 95 6 9096E 04 p 7
137. s only one partition at the same time as the serial grid To aid in the preparation of batch jobs Fig 7 1 summarizes the memory required to partition a grid of a given number of cells 7 6 rflupick 7 6 1 Invocation rflupick is invoked by typing rflupick c lt casename gt s lt stamp gt v lt verbosity gt The following command line arguments are read by rflupick lt casename gt A character string used to label the input and output files 7 6 rflupick 87 30 Scalability data sets Hexahedra Sphere shock diffraction Hexahedra N a NO o Memory required in Gigabytes o c a 0 5 10 15 20 25 30 35 40 Number of cells in millions Figure 7 1 Memory requirements of rflupart to partition a grid of given number of cells 88 Chapter 7 Execution stamp A variable indicating the iteration or time from which the grid and solution files are to be read lt verbosity gt An integer indicating the desired verbosity level of rflupick The verbosity level can take the following values O No output rflupick will not write any information to standard output 1 Low level of output rflupick will write some information to standard output 2 High level of output rflupick will write detailed information to standard output rflupick expects interactive user input after invokation as described in Sect 7 6 4 7 6 2 Input Files The following input files are read by rflupick e A grid fil
138. s the module simulating the evolution of chemical species and Equilibrium Eulerian particles Rocrad is the radiation module RocfluMP consists of several modules rfluconv is the conversion module of RocfluMP It converts RocfluMP solution and grid files from ASCII to binary format and vice versa and converts RocfluMP grid files into a format supported by YAMS and TETMESH rfluconv requires interactive user input rfluinit is the initialization module of RocfluMP It generates RocfluMP solution files for each region based on the information contained in the user input file rflumap is the processor mapping module of RocfluMP It generates the mapping file which is required for parallel computations It also generates the Rocin control files for computations with GENx rflumap requires interactive user input rflump is the actual solution module of RocfluMP rflupick is used in conjunction with rflupost to visualize only specific cells in the grid such as cells near boundaries or cells sharing faces or vertices For parallel computa tions rflupick can also be used to instruct rflupost to convert only specific regions for visualization This allows the visualization of nominally large cases on small machines rflupick requires interactive user input rflupost is the postprocessing module of RocfluMP It converts grid and solution files from the RocfluMP format into the formats recognized by visualization packages At present the followi
139. simplifies the code because conditional statements can be avoided prLam contains the value of the laminar Prandtl number prTurb contains the value of the turbulent Prandtl number scnLam contains the value of the laminar Schmidt number scnTurb contains the value of the turbulent Schmidt number turbModel is a flag indicating which turbulence model is used specModel is a flag indicating which gas model is used Currently it can only take the value SPEC MODEL NONE conPartUsed is a logical variable indicating whether continuum particles are used disPartUsed is a logical variable indicating whether discrete particles are used radiUsed is a logical variable indicating whether radiation modeling is used 158 Chapter 14 Data Structures spaceDiscr is a flag indicating which spatial discretization model is used It can only take the values DISCR_UPW_ROE or DISCR OPT LES spaceOrder is a flag indicating the order of accuracy of the spatial discretization Currently it can only take the value DISCR ORDER 1 nrkSteps is a flag indicating the number of steps of the explicit multistage or the Runge Kutta scheme ldiss is a flag indicating whether the dissipation terms are to be computed in a given stage of the explicit multistage scheme cfl contains the value of the CFL number epsentr contains the value of the constant in the entropy fix ark contains coefficients used in the explicit multistage and the Runge Kutta scheme grk
140. steady state pressure levels for typical rocket prob lems Changing these values influences the initial solution and means that the preprocessor will need to be rerun For coupled computations make sure that the PREP section contains the keyword SURFFLAG set to 1 if rfluprep was not compiled as part of GENx For coupled computations make sure that the units are correct If the grid was generated in units other than meters as is often the case add a TRANSFORM section with the appropriate scaling factors Inconsistent units will lead to failure of Surfdiver 4 Generate a processor mapping file if parallel runs are to be made by executing rflumap c lt casename gt m 1 p lt nprocs gt r lt nregions gt v lt verbosity gt Note that the number of regions can be larger than the number of processes See Sec 1 3 5 Run the partitioning module by executing rflupart c lt casename gt v lt verbosity gt See Sec 7 5 The partitioning module generates grid files in rflump format 6 Run the initialization module by executing TT rfluinit c lt casename gt v lt verbosity gt See Sec 7 2 The initialization module generates solution files in rflump format 7 For coupled computations you need to generate an input file for Rocin by running rflumap again rflumap c lt casename gt m 2 p lt nprocs gt r lt nregions gt v lt verbosity gt See Sec 7 3 8 For coupled computations generate the RocfluC
141. t RocfluMP is designed to interface with so called multi physics modules The multi physics modules model phenomena such as turbulence particles chemical reactions and radiation and their interaction At present the multi physics modules are under development and have not yet been integrated with RocfluMP When considering fluid flow problems with several chemical species RocfluMP may be re garded as solving transport equations for the mixture variables RocfluMP may be used to solve problems involving fluid structure interactions More specifically RocfluMP is designed to operate as a solution module inside CSAR s coupled 1 2 Overview of RocfluMP 15 rocket simulation code GENx To accommodate dynamically changing fluid domains arising from the deformation predicted by a structural simulation RocfluMP allows for moving grids The Geometric Conservation Law GCL is satisfied in a discrete sense to machine precision to avoid the introduction of spurious sources of mass momentum and energy due to grid motion The relationship of RocfluMP and the other codes is depicted schematically in Fig 1 1 A brief description of the multi physics modules follows they are described in detail in their respective manuals Rocturb is the turbulence module which implements a variety of models for Reynolds averaged Navier Stokes RANS simulations and Large Eddy Simulation LES Rocpart is the Lagrangian particle tracking module Rocspecies i
142. the grid by smoothing the boundary displacements NITER 4 and SFACT 0 25 and for moving the grid by smoothing the coordinates NITER 10 and SFACT 0 1 Only applicable if TYPE 1 or TYPE 2 TYPE Specifies the type of grid motion It can take the following values 1 Move the grid by smoothing the boundary displacements This option is only available for serial computations 2 Move the grid by smoothing the coordinates This option is only available for serial computations 3 Move the grid using the MESQUITE package This option is only available when running RocfluMP in GENx Moving the grid based on smoothing boundary displacements has the advantage that the vertices are not moved for vanishing displacements This is not true if the grid is moved by smoothing coordinates In particular non uniform or distorted grids can be strongly affected by smoothing the coordinates Therefore moving the grid by smoothing the boundary displacements is the recommended option 5 2 5 INITFLOW Section The INITFLOW section is relevant only to rfluinit It contains the following keywords DENS The density of the initial solution kg m Only relevant if FLAG 1 FLAG Specifies whether initial solution is to be generated It can take the following values 1 Generate initial solution using the values assigned to the keywords DENS VELX VELY VELZ and PRESS 2 Generate initial solution using the data contained in a file 3 Generate initial solution using a ha
143. the region and cell with the minimum time step are known rflupick can be used in conjunction with rflupost to visualize that cell Only relevant if FLOWTYPE 1 FLOWTYPE Specifies whether flow steady or unsteady It can take the following values O Flow is steady 1 Flow is unsteady 5 2 Input File 47 MAXITER The iteration number at which the computation is to be stopped A calculation is stopped if either the maximum number of iterations is reached or if the norm of the density residual has fallen below the residual tolerance Only relevant if FLOWTYPE 0 MAXTIME The time in seconds at which the computation is to be stopped Only relevant if FLOWTYPE 1 PRNITER Offset between iterations at which convergence information is printed on screen and written to the convergence file Only relevant if FLOWTYPE 0 PRNTIME Offset in time in seconds at which convergence information is printed on screen and written to the convergence file Only relevant if FLOWTYPE 1 RESTOL The tolerance for the density residual below which the computation is judged to be converged A calculation is stopped if either the maximum number of iterations is reached or if the norm of the density residual has fallen below the residual tolerance Only relevant if FLOWTYPE 0 RKSCHEME Specifies the Runge Kutta method used to integrate the discrete equations in time for unsteady flows It can take the following values 1 Classical RK4 scheme 2 Three stage Run
144. thods x Et subsonic inflow boundary Ss u lt 0and lu c Figure 12 3 Subsonic inflow boundary Subsonic Inflow For subsonic inflow u lt c 4 characteristics corresponding to Ay A2 Az A4 are entering the domain and one characteristic A5 is leaving the domain see Fig 12 3 This implies 55 would always be computed from interior solution while L4 L5 L3 L4 need to be specified L5 can be computed as given by equation 12 34 o Oun ts FE oc because Un Us Ur T are imposed at inflow momentum and energy equations are eleminated at boundary Only density need to be solved at inflow boundary Subsonic Reflecting Inflow For reflecting subsonic inflow LODI relations for 4 im posed physical quantities u Us Ur T can be used to express unknown wave amplitudes L La L3 L4 in terms of known wave amplitude L5 and boundary conditions as follows 12 8 Boundary Conditions 131 1 pc dT Y 12 4 Ig sly Dist L t 52 12 45 du la gt 12 46 p 12 46 du La 12 47 LE 12 47 12 48 Subsonic Non Reflecting Inflow For non reflecting subsonic inflow all incoming wave amplitudes are set to zero Ls Lo L L4 0 12 49 Supersonic Inflow For supersonic inflow u gt c all 5 characteristics are entering the domain No characteristic wave can travel upstream For supersonic inflow all characteristic amplitudes are set to zero L2lI 2L421 2L520
145. tion should be allowed for to simulate real conditions In such case L can be expressed as L K p Poo 12 53 Supersonic Outflow For supersonic outflow u gt c all 5 characteristics are leaving the domain No characteristic wave can travel downstream For supersonic outflow all characteristic amplitudes can be computed using relation 12 39 12 4 Temporal Discretization 133 12 3 4 Virtual Boundary Conditions 12 3 4 1 Periodic Boundaries 12 3 4 2 Symmetry Boundaries 12 4 Temporal Discretization 12 4 1 Runge Kutta Methods 12 4 2 Computation of Time Step 12 5 Grid Motion 12 5 1 Grid Smoothing 12 5 2 Discrete Geometric Conservation Law 12 5 3 Implementation Details 12 6 Mass Pressure Skin Friction and Heat Transfer Coefficient Computation RocfluMP computes mass pressure skin friction and heat transfer coefficients for faces on patches In the following the subscript denotes variables associated with a face with normal vector nj The mass flux coefficient is defined as PiVi Nj Cmi 12 54 5 Pret Vret The pressure coefficient is defined as Pi Pref C i 12 55 P E Pret Vo The skin friction coefficients are defined as Ti n C fri 12 56a id Loret Ve l Ti n Ori ee 12 56b fy 5 Pret V ria ni Cy 12 56c 2 VAN 134 Chapter 12 Algorithms and Methods where 7T n is the viscous stress acting on the face The heat transfer coefficient is d
146. try a one to one mapping is specified By inspecting the grid it is possible to merge appropriate patches by writing the proper mapping file 5 3 2 MESH3D Patch Mapping File mgi File The mgi file serves the same purpose as the vgi file The format of the mgi file is Line 1 The number of patches in the m3d file Line 2 The number of patches after the mapping Line 3 The number of mappings hereafter referred to as nMappings The remaining nMappings lines Each line contains three integers The first two integers represent the lower and upper limits of the patches in the MESH3D file which are to be mapped to the patch indicated by the third integer The lower limit must be less than or equal to the upper limit The first line contains the original number of patches because this information is not con tained the m3d file One important difference between the vgi and mgi files is that the extrema of the lower and upper limits of the original patches in the mgi file do not have to be equal to unity and the number of patches specified on the second line respectively This is because boundary faces in the m3d are grouped by arbitrary flags and not by patch numbers Hence the following is a valid mgi file o 0 N Oo wc gt A N n 5 4 Boundary Condition File 51 22 7 10 6 T 1 4 2 12 12 2 7 T 3 10 10 3 8 9 4 11 11 4 17 22 5 200 600 6 100 100 7 Note in particular the last two lines They illustrate that the
147. tual Connectivity of virtual quadrilaterals global numbering Integer 15 1 Definition of Attributes 165 Table 15 3 Definitions of solution attributes Attribute Meaning Units rhof Density kg m rhovf Momentum per unit volume kg m s rhoEf Energy per unit volume J m pf Pressure Pa TE Temperature K af Speed of sound m s Type Real Real Real Real Real Real Table 15 4 Definitions of attributes on non burning panes Attribute Meaning Units Intent du_alp Displacement m Incoming rhofvf alp Momentum flux kg m s Incoming Tb alp Boundary temperature K Incoming nf alp Normal vector Outgoing rhof alp Density kg m Outgoing pf Pressure Pa Outgoing qc Convective heat flux W m Outgoing qr Radiative heat flux W m Outgoing tf Traction Pa Outgoing Location Type Vertex Face Face Face Face Face Face Face Face Real Real Real Real Real Real Real Real Real 166 Chapter 15 GENx Integration Table 15 5 Definitions of attributes on burning panes Attribute Meaning Units Intent Location Type du alp Displacement m Incoming Vertex Real rhofvf alp Momentum kg m s Incoming Face Real mdot alp Injection mass flux kg m s Incoming Face Real Tflm alp Flame temperature K Incoming Face Real nf alp Normal vector Outgoing Face Real rhof alp Density kg m Outgoing Face Real pf Pressure Pa Outgoing Face Real tf Traction Pa Outgoing Face Real qc Convective heat flux W m Outgoing Face Real qr Rad
148. type is a string indicating the face type TRI if the zone consists of triangular cells QUAD if the zone consists of quadrilateral cells face kind is a one letter string indicating the face kind A if the zone consists of actual faces V if the zone consists of virtual faces region index is a five digit string indicating the global region index Examples are shown in Fig 9 4 which represents a screen dump of the data part loader window in ENSIGHT Chapter 10 Troubleshooting rflump classifies problems arising during execution into two categories 1 Non critical problems rflump attempts to make an automatic recovery It will print a warning message and continue with the execution The recovery attempt may entail changing input variables specified by the user 2 Critical problems rflump cannot recover without user intervention It will print an error message and stop execution 10 1 General Considerations If a problem is encountered and it proves difficult to determine the source the following suggestions may prove helpful 1 Repeat the run or restart from the last output dump with maximum verbosity level The additional output printed by activating the maximum verbosity level can indicate possible problem areas see Sec 10 5 2 Repeat the run with smaller CFL number time step Decreasing the CFL number time step can mitigate the effect of strong transients It is possible that some computations have to
149. uMP This entails a detailed description of the capabilities of RocfluMP the content and specification of the various files which either must be provided or may have to be edited by the user explanations of how to set up a problem to be solved with RocfluMP how to execute RocfluMP how to visualize results produced by RocfluMP and how to resolve problems which may arise when running RocfluMP Part III provides detailed information for developers of RocfluMP and serves as a reference manual for users It includes the precise form of the governing equations the algorithms and methods used to solve them their parallel implementation a detailed description of the data structures and the integration of RocfluMP with GENx and the content and specification of the various files which may have to changed by developers To convey information in a clear and concise manner the following graphical aids are employed Because RocfluMP continues to evolve the information contained in this book is also changing To draw attention to changes in the book so called change bars are positioned in the margins More specifically change bars indicate that either a new capability was added or that the corresponding description was changed Chapter 2 Nomenclature Conventions and Restrictions 2 1 Nomenclature The following conventions are used in this document 1 A grid level or simply level represents the entire solution domain of RocfluMP A grid
150. ucture itself is defined to be an array This gives additional flexibility in allowing several regions to be assigned to a single processor The components of the user defined data type t region are defined as follows iRegionGlobal is the global index of the local region irkStep is the index of the Runge Kutta step fieldFlagMixt is a field flag used to communicate the conserved variables for parallel cal culations using the FEM framework dt contains the timestep 14 2 Region Data Structure 143 TYPE t region INTEGER iRegionGlobal irkStep INTEGER fieldFlagMixt REAL RFREAL POINTER dt TYPE t grid grid gridOld TYPE t mixt mixt TYPE t turb turb TYPE t spec spec TYPE t radi radi TYPE t peul peul TYPE t plag POINTER plags TYPE t patch POINTER patches TYPE t global POINTER global TYPE t mixt input mixtInput TYPE t turb input turbInput TYPE t spec input specInput TYPE t peul input peulInput TYPE t plag input plagInput TYPE t radi input radilnput END TYPE t region Figure 14 3 Definition of region data structure 144 Chapter 14 Data Structures grid is the user defined data type containing all the information relating to the grid See Sect 14 3 for a description of t grid gridOld is the user defined data type containing all the information relating to the old grid when using grid motion See Sect 14 3 for a description
151. uctured Meshes AIAA Paper 89 0366 27th Aerospace Sciences Meeting Reno NV January 1989 Batten P Clarke N Lambert C and Causon D M On the Choice of Wavespeeds for the HLLC Riemann Solver SIAM J Sci Comp 18 6 1553 1570 1997 Bruner C W S Geometric Properties of Arbitrary Polyhedra in Terms of Face Geome try AIAA J 33 7 1350 1995 Chakraborty P Balachandar S and Adrian R J On the Relationships between Local Vortex Identification Schemes J Fluid Mech Vol 535 pp 189 214 2005 Ciucci A lafrati A and Schettino A Numerical Analysis of Pressure Oscillations in a Duct Test Case CO Technical Report TR 96 102 Centro Italiano Ricerche Aerospaziali September 1996 Liou M S A Sequel to AUSM AUSM J Comp Phys 129 364 382 1996 Poinsot T J and Lele S K Boundary Conditions for Direct Simulations of Compressible Viscous Flows J Comp Phys 101 104 129 1992 Ringleb F Exakte L sungen der Differentialgleichung einer adiabaten Gasstromung ZAMM 20 4 185 198 1940 Roe P L Approximate Riemann Solvers Parameter Vectors and Difference Schemes J Comp Phys 43 357 372 1981 Skews B W The Shape of a Diffracting Shock Wave J Fluid Mech 29 2 297 304 1967 175 176 References 12 Skews B W The Perturbed Region Behind a Diffracting Shock Wave J Fluid Mech 29 4 705 719 1967 13 Sod G A Survey of Several Finite Difference Methods for Systems of Nonlinear Hype
152. unifMomZ unifPres END TYPE t mixt input Figure 14 9 Definition of data type t mixt input 14 5 Mixture Data Structure 157 14 5 Mixture Data Structure 14 5 1 Data Type t_mixt_input flowModel is a flag indicating which flow model is used It can only take the values FLOW EULER or FLOW NAVST defined in ModParameters F90 moveGrid is a logical variable indicating whether the volume grid is to be moved Note that the movement of interior points does not necessarily have to be activated when boundary patches are moving nDv contains the number of dependent variables It is used to determine the size of the array dv see below nTv contains the number of transport variables It is used to determine the size of the array tv see below nGv contains the number of gas variables It is used to determine the size of the array gv see below indCp is a flag used to allocate the array for specific heat in the gas variable array If the specific heat is to vary in space indCp 1 otherwise indCp 0 This allows the gas variable array gv see below to be accessed even if the specific heat is constant which simplifies the code because conditional statements can be avoided indMol is a flag used to allocate the array for molar mass in the gas variable array If the molar mass is to vary in space indMol 1 otherwise indMol 0 This allows the gas variable array gv see below to be accessed even if the molar mass is constant which
153. ut the grid and grid related quantities for a given region It is read by the various programs to determine the sizes of arrays The file is called e lt casename gt dim_ lt mmmmm gt for steady flows and e lt casename gt dim_ lt mmmmm gt _ lt nnnnnnE nn gt for unsteady flows where lt mmmmm gt is the region index and lt nnnnnnE nn gt is the time stamp At present the time stamp of the dimension file is always zero The dimension file is always in ASCII format and consists of sections whose significance is indicated by character strings The sections can be written in any order Consider the following example dimensions file ROCFLU dimensions file Vertices 1797 4102 4922 Cells 8194 19697 23636 Tetrahedra 8194 19697 23636 Hexahedra 0 0 0 Prisms 0 0 0 Pyramids 0 0 0 Patches 3 6 1 219 468 0 0 3 360 708 0 0 5 82 155 0 0 Borders 7 4 4 988 1184 254 301 48 9 2 2904 2457 617 516 141 10 4 2121 2206 ATT 500 107 12 3 3709 3358 770 690 191 13 4 80 67 51 41 0 26 27 28 172 Chapter 16 File Content and Format Specifications 15 3 611 750 168 216 31 16 3 1477 1481 340 340 70 End Line 1 The first line must contain the string shown otherwise reading of the file will fail Line 2 Specifies the beginning of the vertex dimension section Line 3 Specifies the number of actual vertices the total number of vertices and the maximum allowed number of vertices The maximum allowed number of
154. ution file in RocfluMP format e A dimension file e A boundary condition file A file generated by rflupick detailing which regions are to be postprocessed and whether individual cells are to be visualized If this file does not exist all regions are postpro cessed 7 7 3 Output Files The following output files are written by rflupost e A file in binary TECPLOT format called lt casename gt plt e A version file called rflupost vrs It contains the version number and date of the executable Successive runs append to the version file Chapter 8 Example Cases This chapter illustrates the execution of rflumap rflupart rfluinit rflump and rflupost for several example cases For the sake of brevity the output produced by these program is not shown 8 1 Shocktube The flow in a shocktube is used to test the shock capturing capabilities of the spatial dis cretization and the accuracy of the temporal discretization The CVS repository contains coarse and fine grids for the first and second cases specified by Sod 13 The grid is not shown on account of the simple geometry The following files are required for this case shocktube bc shocktube inp shocktube hyb bin The GRIDSRC variable in the FORMATS section of the shocktube inp file indicates that the original grid file is in CENTAUR format By typing rflupart c shocktube v 2 the partitioner runs and writes the following output files shocktube dim Shocktube g
155. variables cvStateFuture must be set to either CV MIXT STATE DUVWP or CV MIXT STATE DUVWT any other value will generate an error An error will also be generated if cvState is not equal to CV MIXT STATE CONS RFLU_ConvertCvPrim2Cons pRegion cvStateFuture to convert from a primitive variable state to conserved variables cvStateFuture must be set to CV MIXT STATE CONS any other value will generate an error An error will also be generated if cvState is equal to CV MIXT STATE CONS The strict checking of cvState upon calling the conversion routines is carried out to catch programming errors where the state was changed on entering a routine but not changed back on exiting the routine Additional statements such as IF pRegionfmixt cvState CV MIXT STATE CONS THEN CALL ErrorStop global ERR CV STATE INVALID LINE END IF region ln may be placed at the top of routines to catch such errors The data defined in the data type t mixt is shown in Fig 14 10 and explained in detail below cv contains the vector of conserved variables cvOld contains the vector of old conserved variables i e from a previous timestep dv contains the vector of dependent variables 14 5 Mixture Data Structure 161 tv contains the vector of transport variables gv contains the vector of gas variables rhs contains the residual vector rhsSum contains a weighted sum of residual vectors for the Runge Kutta scheme diss contains the r
156. vertices must be equal to or greater than the total number of vertices Line 4 Specifies the beginning of the cell dimension section Line 5 Specifies the number of actual cells the total number of cells and the maximum allowed number of cells The maximum allowed number of cells must be equal to or greater than the total number of cells Lines 6 13 Specify the dimensions of tetrahedral prismatic pyramidal and hexahedral cells Line 14 Specifies the beginning of the patch dimension section Line 15 Specifies the number of patches local to the given region and the global number of patches In this case the given region contains three patches while there exist six global patches The remainder of the patch section contains one line for each local patch Lines 16 18 Specifies for each local patch its global index the number of actual triangles the total number of triangles the number of actual quadrilaterals and the total number of quadrilaterals Line 19 Specifies the beginning of the border dimension section Line 20 Specifies the number of borders in the given region The remainder of the border section contains one line for each border Lines 21 27 Specifies for each border the global index of the region with which data must be exchanged the corresponding index of the border of the region with which data must be exchanged the number of cells for which data must be sent the number of cells for which data must b
157. y and grid motion is active A restart information file A probe file e A convergence file e A mass conservation check file if grid motion is active and rflump is not run within GENx e A version file called rflump vrs It contains the version number and date of the executable Successive runs append to the version file 84 Chapter 7 Execution 7 4 4 Profiling and Performance Analysis Guidelines To profile RocfluMP it should be compiled with the Rocprof library as described in Subsec tion 3 2 2 Important routines in RocfluMP will be instrumented automatically If additional routines are to be included in the instrumentation the source code must be edited accord ingly The execution of RocfluMP is unchanged by the additional instrumentation On completing execution dedicated output files are written by the Rocprof library for analysis with the Profane tool The performance of RocfluMP is strongly affected by many factors If the goal of the analysis is to determine the outright speed of the code as a whole or selected portions of it the following guidelines should be considered 1 Create a balanced data set for parallel runs In other words the dimensions of each region should be identical or nearly identical It is important to note that this applies not only to real cells but also to virtual cells because the latter determine the amount of communication between the regions 2 Deactivate the printing and writing of out
158. zes faces according to types and kinds also A face can be of trian gular of quadrilateral type and can be of different kinds depending on whether the adjacent cells are actual or virtual ones and whether the face is on a boundary 3 Since RocfluMP is based on the cell centered method the computation of fluxes is most easily carried out by looping over faces of the grid Because boundary conditions are conveniently enforced by modifying the computation of fluxes on boundary patches the grid data structure only stores internal faces i e faces which do not lie on boundary patches 14 3 Grid Data Structure 145 Grid Num Boundaries Num Tetrahedra Num Bound Quad m Num Vertices Num Boun Tri m Hexahedra Conn Num Cells Num Hexahedra 2 m Prism Conn E A Pyramid Conn E Num Prisms m Tetrahedra Conn Num Pyramids Figure 14 4 Overview of grid data structure 146 Chapter 14 Data Structures TYPE t grid Basic grid quantities ee Sst sete INTEGER indGs nBFaces nBQuads nBTris nCells nCellsTot nEdges amp nEdgesEst nEdgesTot nFaces nFacesEst nFacesTot nHexs amp nHexsTot nPatches nPris

User`s Guide - Center for Simulation of Advanced Rockets

Contents

Download Pdf Manuals

Related Search

Related Contents