Card 9 - PARAM Parameters

KODE

KODE is the code PARAM (or 22)

N1 = ACCEL

N1 = ACCEL (or 3)

This option is OBSOLETE.

This option sets some of the iteration convergence accelerator parameters for Analyzer if the conjugate gradient solver is not used (Card 9 PARAM ILU OFF must be present).

If the n and T2 parameters are specified, Aitken's accelerator is used. Every n'th iteration (default = 6) the ratio r between the last two temperature changes is computed for each element i

Then, if -1 < r < 1, a new temperature is extrapolated using the formula for summing an infinite geometric series:

1/(1-r) is the acceleration multiplier. Its maximum permitted value is T2. The default for T2 is 50.

If T1 = OFF, the accelerator is disabled.

If T1 = CHEBY, the Chebyshev accelerator is used instead of Aitken's. This method is based on computing accelerator parameters from the maximum and minimum eigenvalues of the convergence matrix. It is preferable to Aitken's Method for difficult, hard-to-converge models and for models with hydraulic elements.

N1 = ADDBEAMENDS

N1 = ADDBEAMENDS (or 111) creates two zero thickness quadrilateral shell elements at the free ends of radiating beams with emissivity = 0 and absorptivity = 0 facing outward. Their purpose is to close the ends of beams, to create radiative blockages for other elements. Since beam cross-sections are assumed to be circular, and the elements fit in the circle, their areas do not fully cover the beam ends.

N1 = ALPHA

N1 = ALPHA (or 17) sets the explicit-implicit weighting factor ALPHA to T1 for transient runs with the Card 2b GRADNT = - 5 option. At each integration time step the temperature at time t + dt is computed by:

T1 must be bounded by 0 and 1. T1 = 0 defines a fully explicit and T1 = 1 a fully implicit algorithm.

If a PARAM ALPHA Card is not present, the default is ALPHA = 1.

N1 = ANADOMDEC

N1 = ANADOMDEC (or 131) activates distributed memory (MPI) parallel processing in the Analyzer module using domain decomposition approach. Each Analyzer process runs in its separate domain specific directory. Those directories are automatically created as mayampi_tmpdir1,2,.. subdirectories in the main run (solution) directory path (that path should exist on all the machines in the run, but it does not have to be on a network shared location.) Before launching the Analyzer module the domain decomposition (DOMDEC) module is run to partition the model into domains. Each domain has only a portion of the model’s full set of elements and conductances in its MODLCF file, which is written to the domain’s directory. The DOMDEC module also converts other input files of the Analyzer to their domain-specific versions that are placed in the domains’ directories. Upon the analyzer run completion the domain’s chunks of each of the solve results files (TEMPF, tmgrslt.dat, etc) are merged into the corresponding single (united) file in the main run directory. This option can be combined with parallel view factor calculations but not with PARAM DOMDEC or PARAM RADDOMDEC options. Compared to the PARAM DOMDEC option the PARAM ANADOMDEC method is applied to the Analyzer module only (GRAYB, POWER, and MEREL are run in serial) but provides more complete parallelization of that module and has no special restrictions on the models with which it can be used or on how the models are partitioned (for example, a single radiative enclosure could be split between multiple domains).

N1 = ANALYZVERSION

N1 = ANALYZVERSION (or 55) allows a different version of the analyzer module to be run. T1 is a number > 0, e.g. 6, in which case the executable tmg/exe/analyz6.exe will be run, instead of tmg/exe/analyz.exe.

N1 = ANARSLT

N1 = ANARSLT (or 136) performs the postprocessing of thermal solve results from the ANALYZER module instead of running the RSLTPOST module separately. It passes the results directly to the RSLTPOST routines instead of using the tmgrslt.dat file to read and write the results. The tmgrslt.dat file can still be written if needed.

• T1 = 0 (or blank) Performs RSLTPOST processing from the ANALYZER main execution thread rather than in a separate thread. It does not write the tmgrslt.dat file to a disk.
• T1 = -1 Deactivates the PARAM ANARSLT option so that the results are written directly to the tmgrslt.dat file, then it reads the tmgrslt.dat file and performs the postprocessing in the RSLTPOST module.
• T1 = 1 Performs the RSLTPOST processing in a separate thread concurrently with the ANALYZER computations of the main execution thread.
• T1 = 2 Writes the thermal results from memory to the tmgrslt.dat file in the main thread.
• T1 = 3 Writes the tmgrslt.dat file in the main thread and performs the RSLTPOST processing in a separate thread.
• T1 = 4 Writes the tmgrslt.dat file in a separate thread while keeping the RSLTPOST processing in the main thread.
• T1 = 5 Writes the result to the tmgrslt.dat file and performs the RSLTPOST post processing in two separate independent threads concurrently with the ANALYZER computations.
• If the PARAM ANARSLT option is used in combination with distributed-memory parallel processing in ANALYZER (with PARAM ANADOMDEC), then parallel processing will be enabled in RSLTPOST as well. The RSLTPOST parallelization in this case follows the same methodology as for PARAM MPIRSLTPOST. That is, different time steps are processed on different processes. To prevent clashes between inter-process communication of the ANALYZER and RSLTPOST threads on the same process, the RSLTPOST execution will be restricted to the main thread in each process. That is, settings T1 = 1, 3, 5 above are automatically replaced with T1 = 0, 2, 4, respectively.

N1 = AUTOCOAT

N1 = AUTOCOAT (or 47) performs automatic surface coating on free edges and surfaces of specified elements. Free edges of shells are covered with beams whose surface areas are set to the thicknesses of the shells, and free surfaces of solids are covered with zero thickness shells. The material properties of the coating elements are set to the material properties of the elements they coat.

• No new element will be created on a free surface or edge that is already covered by an element.
• The surface normals of free surfaces point outwards.
• If T1 is a group name, only the elements of T1 are surface coated. The newly created elements are assigned the group name T1.
• If T1 is a blank, all elements are surface coated.
• If T1 is OFF, then all other PARAM AUTCOAT Cards are rendered null.
• If T1 is _SOLIDS, then all solid elements (and only solid elements) will be surface coated. No group names will be associated with the surface coated elements.

N1 = AUTODELTAT

N1 = AUTODELTAT (or 134) specifies and activates all the required parameters for automatic handling of the time step size used by the solver during transient runs.

• T1 specifies the error estimate.
• T2 defines the minimal integration time step size the solver uses during transient runs. If the minimum time step overwrites the minimum time step specified in CARD 2B, a warning is issued.
• T3 defines the maximum integration time step size the solver uses during transient runs.

N1 = AXIMERGE

N1 = AXIMERGE (or 75) merges all axisymmetric elements generated for radiation calculations into the axisymmetric elements themselves for thermal coupling calculations. See Card 9 - INCLAXI Include Axisymmetric Elements Definition for more information. This option is OBSOLETE.

N1 = BACKTOBACK

N1 = BACKTOBACK (or 60) creates reverse sides element with REVNODE Cards from all back-to-back Card 5 shell elements that share the same nodes but whose surface normals point in opposite directions.

N1 = BCSUMMARY

N1 = BCSUMMARY (or 145) enables printing of tables that summarize the thermal and fluid properties of the Thermal Streams, Thermal Convecting Zones, and Thermal Voids. This table is displayed in the standard log and the verbose log files.

N1 = CAPDIST

N1 = CAPDIST (or 72)

If T1 is blank, then for the element CG method, this option redistributes capacitances and heat loads of solid elements only from the CG to the boundary elements, and eliminates the conductances from the CG to the boundaries. This is the default option.

• T1 = SUBDIV the solid element is further subdivided into subelements with capacitances assigned to the faces of the subelements. This creates a denser conductance matrix and increases the accuracy of transient solutions (but not steady-state solutions), with a corresponding increase in CPU time.
• T1 = OFF the redistribution of the capacitances from the CG for solid elements does not occur. The capacitance and the elemental heat load are lumped at the CG, not at the boundary elements, and conductances from the CG to the boundary elements are created. This was the default option in previous releases.
• T2 is a group name or element number, valid if T1 is SUBDIV. In that case, only the elements specified in T2 are subdivided.

N1 = CENTRIFUG

N1 = CENTRIFUG (or 71) specifies that a group of solid or hydraulic elements undergo rotation about an axis.

• If the elements specified are hydraulic elements, a centrifugal body force is applied to the specified hydraulic elements.
• If the elements specified are solid elements, then advection is assumed to occur between the adjacent elements. 1-way conductances are created between the adjacent solid elements.
• T1 is a group name specifying the elements. If T1 = 0, the centrifugal force applies to all hydraulic elements and advective 1-way conductances are created between all solid elements.
• T2 is the frequency of revolution multiplied by 2*Pi. For example, if the frequency of rotation is 600 rpm, then T2 would be 2*Pi*600/60 = 62.8 Rad/sec.
• The centrifugal effect on a 2-node hydraulic element is calculated by:
• ρ is the density of the fluid
• r₁, r₂ are the perpendicular distances from the ends of a 2 node hydraulic element to the axis of revolution.
• T3, T4, T5 are the i, j, k, components of a vector along the axis of revolution. If the vector points out of the paper, the direction of rotation is considered to be counterclockwise in the plane of the paper when looking at the paper.
• T6, T7, T8 are the global x, y, z coordinates of a point along the axis of revolution.
• The magnitude G1way of the 1-way advective conductance created between two adjacent solid elements is:
• C_p is the specific heat of the material of the element that lies upstream
• vol is the volume the boundary element between the adjacent solids sweeps out in unit time when rotating at T2.

N1 = CHECKPOINT

N1 = CHECKPOINT (or 49) allows the creation of a checkpoint file at the completion of the run.

• T1 is an identity number that is appended to the name of the checkpoint file. For example, if T1 = 300, the checkpoint file is called CHECKPNT300.
• T2 is a flag = 0 or 1. If T2 = 1, a command line-driven interface is automatically run with each stand-alone checkpoint run allowing the user to modify specified entities.
• A checkpoint file contains the dump of all the data in memory of the Analyzer at the completion of the run.
• If the file is renamed CHKPNT, then the Analyzer can be re-run in a standalone mode using only the tmg/exe/analyz.exe and file CHKPNT. This allows the user to create stand-alone self-running objects of complete thermal models.

N1 = CINDA

N1 = CINDA (or 65) creates a CINDA format output.

• T1 must be 0. T1 may be a blank if T2 is blank.
• T2 is an optional file name, on which the CINDA format output is written. If T2 is blank, the output is written on file FMODLF.
• T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG will attempt to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = CNVGTRA

N1 = CNVGTRA (or 37) allows the TMG CFD product to be halted during the Analyzer run if the PAUSE button is pressed.

N1 = COINCNODE

N1 = COINCNODE (or 90) allows the merging of coincident nodes if they fall within a tolerance radius of T1 of each other.

N1 = COMPILE

N1 = COMPILE (or 84) specifies that the LAHEY Fortran compiler should be used to compile user-written subroutines.

N1 = COMPRESS

N1 = COMPRESS (or 36) specifies that the properties of the hydraulic elements are calculated using compressible flow properties.

N1 = COND

N1 = COND (or 28) specifies what method is used to calculate conductive conductances. If the card is specified, the element CG method is used. If the card is omitted, the element center method is used. T2 is a flag.

• T2 = blank (default) is the same as T2 = 3.
• T2 = 1 is the old (OBSOLETE) version of element CG method, characterized by no conductances between the CG and the boundary elements.
• T2 = 2 is another OBSOLETE implementation of the element CG method.
• T2 = 3 is the up-to-date implementation of the element CG method.

N1 = CONDENS

N1 = CONDENS (or 89) is used for TMG CFD calculations only. It indicates that condensation and evaporation is modeled. T1 represents the material number for the condensate (e.g. water).

N1 = CSOLVE

N1 = CSOLVE (or 32) enables the coupled solver for TMG CFD.

• T1 is a flag determining the coupled solver option.
• T1 = 2 is the thermal only solution.
• T1 = 3 is the coupled solution, with sequential coupling option (OBSOLETE).
• T1 = 4 is the coupled solution, with concurrent coupling option.
• T2 is an optional flag that indicates the flow solver type.
• T2 = 1, the serial flow solver is used (OBSOLETE).
• T2 = 2, the parallel flow solver is used.

N1 = DOMDEC

N1 = DOMDEC (or 126) is the same as PARAM RADDOMDEC except that the domain decomposition is extended so that the non-radiative elements and conductances are divided between domains as well, instead of being duplicated in full on every domain. Instead of separate/independent linear solves in the Analyzer module for different domains, a common parallel linear solve is done. In contrast to the PARAM RADDOMDEC option, the current option is meaningfully applicable to conduction-dominated models as well.

T1 is an optional integer value that sets the level of overlap between the domains (if the field is empty, it defaults to 1). Value of 0 means that no additional domain overlap is introduced. Increasing the level of domain overlap by 1 means that each domain will be extended by a layer of out-of-domain elements that are directly connected (through non-radiative conductances) to the given domain elements. Increasing that value can improve iterative convergence of the ILU-BiCGStab linear solve, but it also increases the memory requirement and per-iteration computational cost for each domain.

N1 = DTEMPMAX

N1 = DTEMPMAX (or 86) is OBSOLETE. PARAM AUTODELTAT is the preferred option.

This option allows the integration time step to increase during transient runs.

• T1 is the maximum allowable temperature change in a given integration time step. You can specify T1 in a time-dependent table. For more information, see Card 9 - TABTYPE Table Variable Type Definition.
• T2 is the maximum allowable integration time step. T2 = 0 or blank defaults to the Card 2b TF parameter, i.e. the time of the run.
• The minimum allowable integration time step is the Card 2b DT parameter, or the integration time step calculated from a table.
• The algorithm first estimates the temperature change for each element. The integration time step value is then proportionally increased until the maximum estimated temperature change reaches T1, provided that:
• The slopes of all time-dependent boundary conditions are unchanged within the integration time step.
• Phase change does not occur on any element during the integration time step.
• The maximum estimated temperature change is ≤ T1.

N1 = DUCTHEADLOSS

N1 = DUCTHEADLOSS (or 153) specifies a duct head loss due to curvature, bends and junctions. The duct solver computes the radius of curvature along the duct, and the angle between subsequent branches of the duct at sharp bends and junctions, based on the duct geometry. From this geometrical information, the solver computes an additional head loss and applies the loss to the duct. For more information, see Head Loss Correlations for Bends and Branches.

N1 = ELECUPDATE

N1 = ELECUPDATE (or 43) specifies that if an electrical resistance matrix is specified, the properties of the matrix should be updated with a frequency of T1. Default for T1 = 10. For more information on electrical resistance matrices, see Card 9 - TABTYPE Table Variable Type Definition.

N1 = ELEMENTALBC

N1 = ELEMENTALBC (or 148) changes the scheme that evaluates boundary condition properties.

• During the solve, swirl velocity are calculated on the element CGs.
• During post processing, auxiliary variables (Trel, Tabs, Tstat, Vrel, and Swirl) on the nodes are interpolated from corresponding elemental values.

N1 = ELEMTYPE

N1 = ELEMTYPE (or 115) associates the elements of a group to a specific element type.

• T1 is a group name.
• T2 = 0
• T3 = FLUIDFACE the elements of group T1 are identified as fluid faces for CFD analysis.

N1 = ELIM1WAY

N1 = ELIM1WAY (or 117) ensures that if there is more than one 1-way conductance connected to a single element, such that the element influences the other element at the end of the 1-way conductance, then the smaller 1-way conductances are combined with the larger 1-way conductance, and the smaller 1-way conductance is replaced by a value equal to T1 times its value. T1 should be small number. This parameter can be used to eliminate false diffusion occurring with mesh misalignment when 1-way conductances are created with the PARAM CENTRIFUG option, or when a fluid velocity is specified on a PROP Card.

N1 = ENGBAL

N1 = ENGBAL (or 10) adds a system energy balance convergence criterion to steady-state Analyzer runs.

T2 may be blank, or the code FRACT.

• Without PARAM ENGBAL the default convergence criterion is that the maximum temperature difference between any two iterations must be < the Card 2b GRADNT parameter.
• With PARAM ENGBAL the system energy balance criterion must also be fulfilled. The energy balance is equal to the sum of the heat inputs into the elements, plus the heat flowing from the sources, minus the heat flowing into the sinks, minus the heat remaining in the elements connected with one-way conductances.
• T2 = blank, the energy balance must be < T1.
• T2 = FRACT, the energy balance must be < T1 times the value of heat flowing into the sink elements. With the FRACT option T1 = 0.005 generally yields acceptable results.

N1 = ESATAN

N1 = ESATAN (or 66) creates an ESATAN format output.

• T1 must be 0. T1 may be a blank if T2 is blank.
• T2 is an optional file name, on which the ESATAN format output is written. If T2 is blank, the output is written on file esatan.dat.
• T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG will attempt to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = ESATANTMG

N1 = ESATANTMG (or 108) defines the parameters for the ESATAN-TMG product. T1 is an ESATAN keyword, whose explanation may be found in the ESATAN User’s Manual. T2 may be blank, or a value.

• T1 = SOLVE, then if T2 = 1, an end-to-end ESAN-TMG solve is created, if T2 = 2, an ESATAN input file is created, and if T2 = 3, a solution is run from the ESATAN input file.
• T1 = TIMEO, then T2 is the start time value.
• T1 = TIMEND, then T2 is the end time value.
• T1 = OUTINT, then T2 is the output interval value.
• T1 = DTIMEI, then T2 is the initial time step value.
• T1 = NLOOP, then T2 is the maximum iteration count value.
• T1 = RELXCA, then T2 is the temperature convergence criterion value.
• T1 = DAMPT, then T2 is the temperature damping factor value.
• T1 = INBALA, then T2 is the absolute energy balance criterion value.
• T1 = INBALR, then T2 is the relative energy balance criterion value.
• T1 = INBNDM, then T2 is the mean energy balance criterion value.
• T1 = ARITH, then T2 is the arithmetic node factor value.
• T1 = DTPMAX, then T2 is the maximum temperature change over time step value.
• T1 = DTROCA, then T2 is the temperature error bound over time step value.
• T1 = DTMIN, then T2 is the minimum time step value.
• T1 = DTMAX, then T2 is the maximum time step value.
• T1 = TABS, then T2 is the absolute temperature offset value.
• T1 = STEFAN, then T2 is the Stefan-Boltzmann constant value.
• T1 = CVTCA, then T2 is the cyclic temperature convergence criterion value.
• T1 = CVDTCA, then T2 is the cyclic temperature ROC convergence criterion value.
• T1 = PERIOD, then T2 is the cycle period value.
• T1 = OUTPUT, then T2 is the output interval value during cyclic iterations value.
• T1 = SOLVIT, then the successive point iteration (steady state solver) is specified.
• T1 = SOLVFM, then the sparse matrix (steady state solver) is specified.
• T1 = SOLVCG, then the conjugate gradient (steady state solver) is specified.
• T1 = SLFRWD, then the forward difference (transient solver) is specified.
• T1 = SLFWBK, then the standard Crank-Nicolson (transient solver) is specified.
• T1 = SLCRNC, then the advanced Crank-Nicolson (transient solver) is specified.
• T1 = SOLCYC, then the cyclic transient solver is specified.

N1 = EXEARTH

N1 = EXEARTH (or 68) the Earth view factors and albedo factors are created not by the default method of analytical formulation and table lookup, but by explicitly modeling the section of the Earth seen by the satellite as a segment of a faceted sphere, and computing the Sun's reflections from the facets.

• T1 is the number of elements (facets) in the tangential direction.
• T2 is the number of elements in the radial direction.
• Although this method is more compute-intensive, it provides more accurate shadowing computations for the albedo calculation than with the default method.
• When specular surfaces exist on the satellite, both albedo factors and Earth view factors are ray-traced. This is not the case for the default method, where only incident solar radiation is ray-traced.

N1 = EXPRESSION

N1 = EXPRESSION (or 149) represents all angular quantities in symbolic expressions as radians. T1 is a keyword RADIANS. When it is not specified, the angular quantities are in degrees.

N1 = EXSKY

N1 = EXSKY (or 138) models the sky dome explicitly.

• T1 is the number of elements (facets) in the tangential direction.
• T2 is the number of elements in the radial direction.

N1 = FASTRANSIENT

N1 = FASTRANSIENT (or 44) This is an OBSOLETE option. Use PARAM TDIFS instead.

N1 = FEM

N1 = FEM (or 152) activates the finite element formulation.

N1 = FEMCOND

N1 = FEMCOND (or 141)

This option activates the finite element formulation to calculate heat conductance for bulk solids and shells.

N1 = FEMNODALBC

N1 = FEMNODALBC (or 150) activates the finite element treatment of convective boundary conditions, such as convecting zones, streams, and voids.

N1 = FIELDSDIRECTORY

N1 = FIELDSDIRECTORY (or 212) specifies the directory path to the field files. The software automatically writes the interpolated solid temperature on convective and immersed ducts in fields files in the specified directory path.

N1 = FSICOUPLING

N1 = FSICOUPLING (or 146) provides the coupling type for the fluid and structure interaction.

• T1 is the keyword NONE or ONEWAY.
• T1 = NONE, then there is no interaction between the fluid and the structure.
• T1 = ONEWAY, then fluid parameters, such as pressure affect the structure.
• T2 is a group name for the flow surface elements that are in contact with the structure.

N1 = FLAGPOLE

N1 = FLAGPOLE (or 70) provides conductive paths between elements that are joined with a flagpole and hinge-door topologies. By default, TMG does not provide a thermal conductive path between elements for these topologies.

The PARAM FLAGPOLE option is only applicable with the element CG method, activated by the PARAM COND NEW flag. A flagpole topology occurs when a beam element is joined to a solid or a shell at a corner only. Exceptions to flagpole topologies are:

• When the beam is non-conducting.
• The attached shell is non-conducting.
• The beam is connected with both its nodes to a shell.
• The beam is flagpole-connected to a shell but the shell has one or more conductive beams running along its edge, and one of these beams is connected to the beam.
• If a beam is attached to a combination of solids and shells at a node, it is assumed to be attached only to a single solid. If it is attached to more than one shell at a node, it is considered to be attached to only one shell.

Once the flagpole situation is identified, a connection is made by merging the 1-node lump mass boundary element to the appropriate boundary element of the other element. This is identified as follows:

• If the other element is a conductive shell, the lump mass is merged to the center of one of the boundary elements at the shell's edges.
• If the other element is a solid, then the lump mass boundary element is merged to the CG of the shell boundary element on the face of the solid.

A hinge-door topology occurs when a shell is attached to one or more solids at only two of its nodes. Exceptions to hinge-door topologies are:

• The shell is non-conducting.
• The solid has a conductive shell on its surface joining the hinge-dooring shell at its nodes.
• The shell is a boundary element to another solid, i.e. it is fully connected with all its nodes to the surface of another solid.

The hinge-door topology is addressed by identifying the beam boundary element at the nodes, and joining this to an appropriate surface boundary element of the solid.

N1 = HEIDI

N1 = HEIDI (or 25) splits every quadrilateral element into two triangular elements. T1 is an increment (e.g. 1000) such that if element N is a quadrilateral element, it is split into two triangular elements N and N + T1. The details of the split are written on file HEIDISPLIT by the DATACH module.

N1 = HYDEXPT

N1 = HYDEXPT (or 127) models each hydraulic element with an exponential temperature profile along the element length.

N1 = HYDLOOP

N1 = HYDLOOP (or 29) specifies that the maximum number of hydraulic loop iterations must not exceed T1. The default is 100.

N1 = HYDDAMP

N1 = HYDDAMP (or 42) specifies the damping parameter T1 for the hydraulic network. Default = Card 2b DT parameter.

N1 = HYBRID

N1 = HYBRID (or 22) is an OBSOLETE option which specifies that the hybrid solver should be used if a steady state run is being performed.

The hybrid solver is a combination of the substructuring option and the iterative solver. All elements connected by only linear conductances are automatically eliminated, creating a smaller model, which is then solved iteratively. The temperatures of the full model are then recovered at the end of the run.

The PARAM HYBRID option is equivalent to the PARAM SUBSTR RADNODES option. It is ignored for transient runs.

N1 = ILU

N1 = ILU (or 27) is a recommended option which activates the ILU preconditioned conjugate gradient solver. When this option is used, the default PARAM ACCEL parameter is automatically set to OFF.

The conjugate gradient solver is the default option.

• LFIL is the maximum number of terms generated for each row of the preconditioning matrix. LFIL defaults to 10.
• LFIL = -1 generates a diagonal preconditioning matrix.
• LFIL = -2 generates a unit preconditioning matrix.
• Higher LFIL values are more suitable for ill-conditioned matrices and increase the likelihood of convergence but require more CPU time. If convergence does not occur after 100 iterations within the conjugate gradient solver, the LFIL value is automatically increased by 10, until convergence is achieved.
• T2 is optional. It defines the conjugate gradient solver iteration convergence criterion.
• T3 is optional. It defines the maximum number of allowable iterations. The default is 100.
• T4 is optional. It defines the convergence norm method, either MAX or L2.
• The conjugate gradient iterations are considered converged if the norm of the residual of vector [A]{x}-{b} divided by the norm of the vector {b} is less than T2. Once the convergence criterion is satisfied, an additional convergence criterion must also be satisfied: the difference in temperatures between iterations must be less than the amount specified by PARAM TDIFS (alternately specified by GRADNT in Card 2b for steady-state runs). This additional temperature difference convergence criterion has a stabilizing effect on the outer iteration convergence. You can deactivate it using the advanced parameter GPARAM 12 1394 0 in Card 9 - GPARAM Parameter.

N1 = INITIME

N1 = INITIME (or 125) specifies the initial times at which the Analyzer module evaluates the temperature and/or water density. This parameter lets you change the default start time.

T1 is the integer value from a previous solution from which the results are used as initial conditions.

N1 = LASTTIMEONLY

N1 = LASTTIMEONLY (or 99) specifies that only a single printout is created for a transient run at the end of the run. It overrides all other options.

N1 = LOCALDAMP

N1 = LOCALDAMP (or 112) allows the specification of iteration damping parameters for selected elements only, as opposed to globally, as specified on Card 2b. T1 is an element number or group name, and T2 is the iteration damping parameter that overrides the Card 2b value for the selected elements.

N1 = MAXNODEID

N1 = MAXNODEID (or 133)

T1 is the largest label for user-defined nodes in the model.

N1 = MODCOND

N1 = MODCOND (or 56) modifies the conductances for odd-shaped elements with the element CG method.

Odd-shaped elements yield negative conductances. Large negative conductances can have two undesirable effects: convergence problems, and unrealistic temperatures on the boundaries of the element. The latter is aggravated when the boundary lies on a free surface and can convect or radiate, and the thermal gradients in the element are large.

To address this problem, the negative conductances of odd-shaped elements are automatically clipped during Analyzer runs, and a warning message is written to the verbose log file.

• T1 sets the clipping threshold for elements that have a free surface boundary. The largest negative conductance in the element is clipped such that its absolute value will not exceed T1 times the sum of the conductances of the elements it is connected to.
• T1 must be ≥ 0. T1 = 0 defaults to 1. T1 < 1 modifies more elements, T1 > 1 modifies fewer elements.
• T2 sets the clipping threshold for elements that do not have a free surface boundary. T2 must be ≥ 0. T2 = 0 or T2 = blank defaults to 10.
• T3 multiplies the default T1 and T2 values by T3.

N1 = MPIALL

N1 = MPIALL

This option activates distributed memory parallel processing for HEMIVIEW and VUFAC modules, using the message passing interface (MPI) protocol for inter-processor communication, with the purpose of reducing the CPU time of view factor computations. This Card is a substitute for PARAM MPIHEMIVIEW and PARAM MPIVUFAC Cards. The parallel run can use either a single machine with multiple processors/cores or an interconnected set of multiple machines with one or more processors on each.

• This option requires a file named mayampi.hosts, which is present in the thermal solve run directory. That file lists the host names of the machines which are used for the parallel run, one host name per line. The first line contains the name of the machine that the solve is launched from (the master machine). When multiple processes are used on any of the machines then that machine is listed multiple number of times (on multiple lines). The total number of parallel processors launched equals the number of lines in the mayampi.hosts file.
• The TMG installation directory path should be valid on all of the machines in the run, which is achieved by either placing the TMG executables in the same path on each machine local drive or installing TMG in a network shared location. The solve run directory does not have to be valid/accessible on any of the machines except the master one.
• The machines in the same parallel run should have compatible operating systems: one should neither mix Linux and Windows machines nor combine 32-bit with 64-bit machines.
• HEMIVIEW and VUFAC parallel processing uses dynamic load balancing, which allows efficient use of machines of different performance characteristics in the same run: faster machines receive more work load than the slower ones.
• For multi-host runs on Linux platform a secure shell installation (ssh command) should be available and configured to run without password prompt (that is, passwordless ssh should be enabled).
• On Windows platforms, one also needs to install MPICH2 (version 1.4.1p1) on each machine so that the MPICH2 process manager (SMPD) service is active on all the machines of the run (http://www.mpich.org/).

N1 = MPIVUFAC

N1 = MPIVUFAC (or 120) activates the distributed-memory parallel processing feature for the VUFAC module to reduce the CPU time of view factor computations. Different view factors and related quantities (e.g. solar view factors) are computed concurrently on different machines or different CPUs of the same machine.

• T1 is a mandatory integer value, and is set to the number of processes of the parallel run.
• T2 is optional. It is the name of the file that allows the specification of the machine host names, one per line. Multiple processors/cores on the same host are specified by repeating the name of the host multiple times in the file.
• If T2 is blank, then a default machine file named mayampi.hosts is first required in the run directory. The number of lines in the machine file is equal or larger than the requested number of processes T1.
• The same machine setup and machine file requirements apply as for the PARAM MPIALL option.

N1 = MPIHEMIVIEW

N1 = MPIHEMIVIEW (or 121) activates the distributed-memory parallel processing feature for the HEMIVIEW module to reduce the CPU time of view factor computations. Different view factors are computed concurrently on the graphic cards of different machines.

• T1 is a mandatory integer value, and is set to the number of processes of the parallel run.
• T2 is optional. It is the name of the file that allows the specification of the machine host names, one per line. Multiple processors/cores on the same host can be specified by repeating the name of the host multiple times in the file.
• If T2 is blank, then a default machine file named mayampi.hosts is first required in the run directory. The number of lines in the machine file should be equal or larger than the requested number of processes T1.
• The same machine setup and machine file requirements apply as for the PARAM MPIALL option.

N1 = MPIRSLTPOST

N1 = MPIRSLTPOST (or 123) activates the distributed-memory parallel processing feature for the RSLTPOST module to reduce the CPU time of the postprocessing computations. The data from different time steps of a transient run is postprocessed concurrently on different parallel processes.

• T1 is a mandatory integer value, and is set to the number of processes of the parallel run.
• T2 is optional. It is the name of the file that allows the specification of the machine host names, one per line. Multiple processors/cores on the same host is specified by repeating the name of the host multiple times in the file. If T2 is blank, then a default machine file named mayampi.hosts is first required in the run directory. The number of lines in the machine file should be equal or larger than the requested number of processes T1.
• The same machine setup and machine file requirements apply as for the PARAM MPIALL option. The current Card is best used with multiple processes on a single machine. If the same machine file is used as for PARAM MPIALL in a parallel run with multiple hosts having multiple cores each, then one can limit the RSLTPOST parallel processing to the master machine by listing all the master machine processes in the first lines of the machine file and then setting the T1 value to the number of the master machine processes.

N1 = NEGCOND

N1 = NEGCOND (or 26) allows the COND module using the element center method to write negative conductive conductances in the file MODLF, instead of merging the two connected elements together.

N1 = NEVADA

N1 = NEVADA (or 50) creates a NEVADA format output.

• T1 is equal to 0. T1 may be a blank if T2 is blank.
• T2 is an optional file name, on which the NEVADA format output is written. If T2 is blank, the output is written on file nevada.ren.
• T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG solver attempts to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = NLOOP

N1 = NLOOP (or 2)

If the T2 field is empty, this option sets the maximum number of iterations during transient runs in an integration time to T1. The default is 100. Convergence is achieved if the maximum temperature difference between two subsequent iterations is less than the value set by the PARAM TDIFS Card, with the default value of 0.001.

If the T2 field is set to STEADYSTATE, then T1 is the maximum number of iterations for the solution control STEP cards with the steady state solution type. Otherwise, the steady state equivalent of PARAM NLOOP is the Card 2b TF parameter.

N1 = NOAXISYM

N1 = NOAXISYM (or 23) defines the group T1 for non-axisymmetric elements, which allows the inclusion of non-axisymmetric elements in an axisymmetric model.

N1 = NOBUOY

N1 = NOBUOY (or 15) specifies that buoyancy pressures are not calculated for the hydraulic elements T1. T1 is an element number or a group name.

When T1 is blank, buoyancy is ignored for all elements.

N1 = NOELREC

N1 = NOELREC (or 58) is similar to PARAM NOMRECOV. When Card 8 element elimination or PARAM SUBSTR substructuring is performed, then the temperatures of the eliminated elements are not recovered. This results in a smaller file MODLCF and avoiding post-processing, in which all temperatures are required.

N1 = NOEMIT

N1 = NOEMIT (or 59) specifies that the element(s) T1 and their reverse side elements cannot emit or receive radiation in the IR spectrum, or shadow other elements in all spectral bands, during radiation calculations of the VUFAC module. T1 is a single element or a group name.

• If reverse sides are defined with REVNOM or REVNODE Cards for the element(s) T1, they are considered as non-emitting and non-shadowing. View factors are not calculated for elements T1 and their reverse sides, hence radiative couplings or radiative heat flux view factors are not calculated.
• This option is similar to PARAM NORAD. A possible use for PARAM NOEMIT is to define a set of perfectly transparent "mask" elements, which fix the ray-tracing calculation mesh for all elements behind them.

N1 = NOMRECOV

N1 = NOMRECOV (or 14)

T1 is blank or a group name.

With this option the temperatures of the merged elements of Card 7 are not recovered during Analyzer runs. By default, they are automatically recovered. When T1 is blank, none of the merged elements’ temperatures are recovered. When T1 is a group name, only the temperatures of the elements specified in the group are not recovered.

This option reduces the size of the output but should not be used when any sort of post-processing is required, in which the temperatures of all the elements are used.

N1 = NONIDEAS

N1 = NONIDEAS (or 54) identifies the nodes or elements between T2 and T3 that were not created by your CAE software.

• T1 is ELEMENT or NODE.
• T2 is the starting element or node number.
• T3 is the ending element or node number. Blank or zero defaults to T2.
• T4 is an increment. Blank or zero default to 1.

N1 = NORAD

N1 = NORAD (or 13) specifies that the front sides of element(s) T1 and the reverse sides of element(s) T2 are not considered in any radiative calculations by the VUFAC module. They cannot emit or receive any radiation or shadow any other elements in all spectral bands. It is the equivalent of specifying NORAD in Card 5a geometrical element, which is the emissivity.

• T1 and T2 are single elements, or 0, or group names.
• For the sake of backward compatibility, when T2 is blank, both the front and reverse sides of element(s) T1 are considered as non-radiating.

N1 = NOSHADOW

N1 = NOSHADOW (or 12) specifies that the element(s) T1 and their reverse sides (defined with REVNOM or REVNODE Cards) cannot shadow during radiation calculations of the VUFAC module. T1 is a single element, or a group name.

N1 = NOSOLAR

N1 = NOSOLAR (or 67) specifies that the front sides element(s) T1 and the reverse sides of element(s) T2 do not have any solar, Earth or albedo view factors calculated to them. However, they still shadow for Orbital, Sun or Earth requests. T1 and T2 are single elements, 0, or group names.

N1 = NOTCONN

N1 = NOTCONN (or 98) specifies that element(s) T1 are not connected to other elements. The temperatures of the T1 elements are not used in calculating average, Min, or Max temperatures, and all T1 elements and their nodes are flagged as PARAM NONIDEAS. T1 is an element or a group name.

N1 = ORBRAD

N1 = ORBRAD (or 139) activates the old method for computing the radiation request for all articulation times, even when no movement occurs between two articulation times. The thermal solver computes view factors in radiative enclosures and thermal couplings, which contain articulating elements.

N1 = OPPENHEIM

N1 = OPPENHEIM (or 45) specifies that radiative couplings are created with Oppenheim's Method instead of Gebhardt's Method if Card 2a M = 4. This is the recommended option.

For each radiating element, I, a corresponding Oppenheim element, I + T1, is created. A radiative coupling between I and its Oppenheim element is created with a magnitude equal to Area(I) * emissivity(I) / (1 - emissivity(I)).

The view factors VFIJ between elements I and J are transformed into radiative couplings equal to Area(I) * VFIJ between the corresponding Oppenheim elements I + T1 and J + T1.

In case of element number conflict, an unused element number is assigned to the Oppenheim element instead of I + T1.

The Card 2a RK parameter is ignored.

Oppenheim's Method has several advantages over Gebhardt's Method:

• Temperature-dependent emissivities are modeled exactly and efficiently, since only the radiative coupling between the element and its Oppenheim element needs to be updated.
• The costly matrix inversion process necessary for Gebhardt's Method is bypassed.
• The radiation matrices to be solved tend to be much smaller, resulting in faster solution time.
• The storage of the large gray body view factor matrix on file VUFF is bypassed.
• For transient runs, it is suggested that an implicit integration method should be employed, because of the presence of the many zero-capacitance Oppenheim elements solution times are costly with explicit techniques.

N1 = OPPMERGE

N1 = OPPMERGE (or 80) specifies that the Oppenheim elements of those elements that are merged should also be merged to speed up runs. T1 is OFF or blank (the default). When T1 is OFF, the Oppenheim elements are not merged. By default, the Oppenheim elements are merged.

N1 = ORIGINALSTREAM

N1 = ORIGINALSTREAM (or 147) changes the scheme that forms connections between the fluid stream elements created by the thermal solver and the wall elements selected in the two-sided thermal stream.

With this advanced parameter, the length of the fluid stream is equal to the length of the longest wall element selection of the two selections in the two-sided thermal stream. The connections between wall elements and fluid stream elements are formed by finding the closest fluid stream element to the line perpendicular to the selected wall element.

N1 = PATCH

N1 = PATCH (or 77) merges adjacent Oppenheim elements in the MEREL module into a “patch” on both the front and reverse sides of elements to reduce the temperature calculation time in the Analyzer module. The elements merged all center around a common node. Typically, 5-10 adjacent Oppenheim elements are merged. This option usually reduces the number of radiative conductances in large models by an order of magnitude, with generally only a slight reduction in accuracy.

Oppenheim elements are merged based on the following conditions:

• The elements associated with them are co-planar within a specified tolerance value.
• Their parent elements have the same emissivities.
• The topology is not that of a fence or a flagpole.

One form of inaccuracy introduced by the merging of the Oppenheim elements is a “false diffusion” type heat flow between elements within a patch, caused by the elements’ common connection to the same Oppenheim element. This is usually negligible if the adjacent elements have similar temperatures. To reduce the false diffusion effect, negative corrective radiative conductances are introduced between elements within a patch. However, no corrective conductances are introduced when they have temperature-dependent emissivities.

Any number of PARAM PATCH Cards may be used.

T1 may be:

• A numerical value (e.g. 10), which specifies the angular tolerance in degrees. If T1 is blank or = 0, T1 defaults to 15 degrees. Adjacent elements whose surface normals differ by > T1 degrees do not have their Oppenheim elements patched together.
• The code NOPATCH, in which case T2 must be an element number or a group name. This specifies the elements, which are excluded from patching.
• The code ONEPATCH, in which case T2 is a group name. This option ensures that the Oppenheim elements of the elements specified in T2 are patched together.
• The code PATCHMAX specifies a maximum area threshold criterion for patching adjacent Oppenheim elements. T2 is a group name. The smallest elemental area in T2 becomes the area threshold criterion. Elements with areas greater than this value do not have their Oppenheim elements patched together. The objective of this criterion is to prevent large elements from being patched together.
• The code PATCHSET specifies a set of elements that are only patched to each other. T2 is a group name specifying these elements. Any number of PARAM PATCH PATCHSET Cards may be present in the model.

N1 = PDMAX

N1 = PDMAX (or 21) specifies the convergence parameter for the pressure/flow solution of the hydraulic network.

• If T1 > 0, convergence occurs when the maximum total pressure difference between iterations is less than T1, and no conductance clipping is performed. Lower T1 values increase the accuracy of the solution.
• If this Card is not present or T1 = 0 or blank, T1 defaults to 1% of the maximum total pressure difference in the model.
• If T1 < 0, then convergence occurs when the maximum total pressure difference between iterations is < abs(T1) times the maximum total pressure difference in the model, and no conductance clipping is performed.

Convergence of the hydraulic network is observed with the PRINT HYDTRACE Card.

N1 = PECONV

N1 = PECONV (or 61) specifies a periodic convergence criterion for transient runs. When a transient run is cyclical with a period T2, convergence occurs if the maximum temperature difference on any element of the list between the last two cycles is less than T1. The run is then terminated.

If T3 is blank or equal to 0, only the results of the last cycle are written on file tmgrslt.dat. If T3 = 1, then all the results are written on tmgrslt.dat.

If T4 is blank or equal to 0, then the list of elements on which the convergence is checked contains all elements. If T4 is an element or a group name, then the list of elements contains only that element or elements that are in the group. You can also replace T4 by a generic entity of type CYCLIC_TEMP to check the convergence on the elements specified through the generic entity, GENERIC Card.

T2 should be ≤ the final time value specified on Card 2b TF.

N1 = PIRSUBSOLAR

N1 = PIRSUBSOLAR (or 52) specifies T1, the ratio of the IR emissive power of a planet at its subsolar point to its IR emissive power on the side opposite the subsolar point. This is used in planetary IR radiative heat flux calculations. Default for T1 is 1.

The IR radiative heat fluxes incident on an element is calculated with:

where:

• PIR is the planet emissive power in the subsolar point.
• ALB is the albedo factor.
• AL is the albedo value.
• EVF is the planet view factor.

This option is useful when the planet surface temperature is a strong function of the incident solar radiation. For example, on the Moon. This option is automatically invoked when the planet dark side IR radiation value is different from the planet sunlight side IR radiation value, on the Card 6t ORBDEF1 Card.

N1 = QUARTIC

N1 = QUARTIC (or 4) specifies that if radiative conductances are present, the Analyzer solves a quartic equation for each element at each iteration to calculate elemental temperatures. Otherwise, the radiative conductances are linearized using the current temperature values.

PARAM QUARTIC can increase stability and reduce the number of iterations needed to achieve a solution. It should be used if some of the elements radiate at high temperatures or over large temperature differences. If the temperature differences are small, however, its overhead can increase solution run time.

This option is OBSOLETE and is automatically disabled if the conjugate gradient solver is used.

N1 = RADDOMDEC

N1 = RADDOMDEC (or 124) activates the domain decomposition method of parallel processing of radiative conductances in GRAYB, POWER, MEREL, and Analyzer modules to enable solutions of large multi-enclosure radiative models that may require more memory than is available on a single machine. The original radiative model with multiple radiative enclosures will be decomposed and solved as a coupled set of submodels (domains) distributed over multiple directories on multiple machines.

The model partitioning/decomposition is done automatically, by grouping radiative enclosures into domains, in which their memory requirements are nearly equal. While the set of radiative conductances is split (without overlap) between different domains, all the non-radiative conductances of the full model are passed in their entirety to each of the domains. This makes this option best suited for predominantly radiative models.

This method is currently not supported (is automatically disabled) for models with spinning, articulation, CFD-coupled solves, or user1 subroutines.

The PARAM RADDOMDEC card requires an additional input file named tmgdomdec.setup in the main run directory (where the simulation starts). The file contains an alternating sequence of machine host names and absolute paths to domain-specific run directories, as

hostname1
 
/abs/path/dir1
 
hostname2
 
/abs/path/dir2
 
…

Here each directory path is defined locally with respect to the machine listed on the preceding line, even if that path is not directly accessible/mounted on other machines. This simplifies use of local/scratch disks for the distributed run. The main run directory should be accessible for reading and writing from every machine in the list. Note that the main run directory differs from the directory of the first (or any) domain. If the lowest level directory (e.g. “dir1” in “/abs/path/dir1”) does not exist, it is created automatically, provided the upper level directory (“/abs/path/”) exists. The paths in the list should differ between each other in the physical location they correspond to, but not necessarily in their specification strings (e.g. when on same-named local disks of different machines).

The active MPI environment is required for all machines listed in the tmgdomdec.setup, with the same requirements as for the PARAM MPIALL option.

The number of host/directory entries in the tmgdomdec.setup input file determines the maximum allowed number of domains. The actual number of domains used depends on the number and relative sizes of enclosures in the model and cannot be larger than the total number of enclosures or much larger than the total number of view factors divided by the number of view factors in the largest enclosure.

Action of this PARAM card is independent of (does not affect, is not affected by, and can be freely intermixed with) other parallel processing PARAM cards (PARAM MPIALL, PARAM MPIVUFAC, PARAM MPIHEMI, PARAM MPIRSLTPOST). While other parallel processing options are designed primarily for performance improvement, the present card aims to overcome the limitations on the model size (particularly on the number of radiative conductances) that are imposed by a single-machine memory limit. Depending on the model, the current method may not give any performance gain or could increase the overall execution time compared to if the same model is run without this PARAM on (due to a larger number of iterative updates and linear solves). But it enables execution of large models that do not fit into memory of any single machine the user may have available.

This option changes the standard execution flow in the following way. After the VUFAC module completes, the auxiliary DOMDEC module runs to partition the original model into domains and to setup run directories and input files for running GRAYB, POWER, MEREL, and Analyzer modules for each domain. Each of those modules then run in parallel via MPICH2 implementation of message passing interface (MPI), with different MPI processes running in different run directories. After completion of the Analyzer run, the solve results data are gathered in the main run directory. The postprocessing module (RSLTPOST) will proceed as for a regular undistributed run.

The DOMDEC module analyzes inter-element connectivity information (read from VUFF, MODLF, tmggeom.dat, and tmg49.dat files) and partitions the element list so that each element is assigned a single domain index. The connectivity matrix (“graph”) is constructed so that any two elements, which are connected in any of the following ways:

• Radiative or follower conductance.
• Merge card.
• Free face card.
• Different layers of the same multilayer element.
• Front and reverse side of a transparent element.

are forced to belong to the same domain, by being merged into the same element (“vertex”) of the connectivity graph. Each vertex of the connectivity graphs is assigned a weight, mainly based on the number of view factors it encapsulates. To make domains as nearly equal in their total weights (numbers of view factors) as possible within reasonable computation time, the connectivity graph is partitioned using the METIS library ["A Fast and High Quality Multilevel Scheme for Partitioning Irregular Graphs". George Karypis and Vipin Kumar, SIAM Journal on Scientific Computing, Vol. 20, No. 1, pp. 359-392, 1999]. The obtained domains are then internally sorted by their size (the first domain is the largest). If any domains could be merged so that the resulting new domain is not larger than the largest existing one, that will be done as well. The partitioning step of the DOMDEC module writes the partitioning information to file partition.dat, which lists domain index for each element number, distinguishing radiative elements by negative sign of their domain index.

The DOMDEC module is launched via MPI with different processes starting in the main run directory. After the partitioning step above completes (done only on the first process), the partitioning information is broadcasted to all other processes. Each process then reads the full VUFF and MODLF files from the main run directory and writes its own reduced (filtered) domain-specific copies of those files to the corresponding domain run directory. Other necessary TMG files (tmggeom.dat, tmg49.dat, etc) as well as partition.dat file are copied unmodified from the main directory to the distributed directories.

In spite of being launched via MPI (for easier management of distributed execution), each process of GRAYB, POWER, and MEREL modules runs essentially as if it were a regular serial run but applied to the domain-specific VUFF and MODLF files, without any internal communication between processes (no MPI calls used). For those modules, the PARAM RADDOMDEC card is only taken into account when processing QNODE cards in the MEREL module, which reads the partitioning information and excludes heat load (HTF) cards for out-of-domain radiative elements from writing to the MODLF file of its domain, to prevent double counting of those heat loads in the Analyzer module.

The main interaction between different domains occurs at the Analyzer stage (via MPI calls). For optimal convergence, all elements of the model (even elements of other domains) are solved on every domain. The main difference from the undistributed run is that all the missing radiative conductances of each out-of-domain element are replaced by one effective conductance (equal to the sum of missing radiative conductances) from the element to an effective sink element (whose temperature is set as radiative-conductance-averaged temperature of the elements that would be connected to the given element through the missing conductances). In this way, the missing radiative conductances are accounted for via corrections to the diagonal elements and the right hand sides of the linearized heat balance equation. This implementation allows a given radiative element to have radiative conductances from more than one domain. For each radiative element, its heat load will be calculated as a sum of heat load contributions from all domains. The inter-domain synchronization of temperatures and the matrix correction terms is performed after each complete linear solve for temperatures.

The Analyzer temperature solve results for all the elements of the full model (with partial exception for Oppenheim elements, see below) are written only by the first MPI process in the first domain directory. Upon completion of the Analyzer run, those files and some related ones (TEMPF, GTEMPF, PRESSF, tmgrslt.dat, tmgrslt2.dat, the report log file, QNODEF, groups.unv) are copied to the main directory.

Oppenheim elements are created independently on different domains (assigned zero domain index) and their element numbers and temperatures are meaningful only with respect to their corresponding domains (rather than the full model). The TEMPF file in the main directory includes Oppenheim elements (and their temperatures) from the first domain only.

Each distributed TMG module starts with empty the verbose log file in the domain-specific directories. After the module completes, its the verbose log file outputs in each directory are appended to the common the verbose log file in the main run directory, domain after domain, in the order of domain enumeration. The information written to the report log file stays in the domain-specific directories, with only the first domain’s the report log file being copied to the main directory. While the full updated domain-specific VUFF, MODLF, MODLCF, and tmggeom.dat files are not gathered the report log fileback in the main directory, the extra HTF cards and the view factor sums written in the distributed runs are appended to the main directory MODLF and tmggeom.dat files, respectively, for postprocessing.

N1 = RADSOLVE

N1 = RADSOLVE (or 118) calculates temperatures in the Analyzer model by solving the radiative and non-radiative matrices separately, passing boundary conditions back and forth. It is particularly effective with large radiative models, and ones where large radiative temperature differences and nonlinearities cause convergence problems. It is limited to non-articulating models using the Oppenheim’s Method.

When the results are output, the Oppenheim element temperatures are transformed into equivalent far-field radiative temperatures, with a corrective heat load for energy balance.

PARAM RADSOLVE is automatically activated with wavelength-dependent (PARAM SPECTRA Card) models.

Using this option is often more efficient because:

• The radiation matrix is solved in T4, not T, which makes it well-behaved and linear. The non-radiation matrix is always solved in T. When the two matrices are combined, they need to be solved in T, which makes the resulting matrix non-linear. Thus, the PARAM RADSOLVE option in effect eliminates radiative non-linearities, which makes it particularly effective for reducing the number of iterations when the radiation matrix is very non-linear, e.g. near cryogenic temperatures.
• PARAM RADSOLVE is very effective when the model is ill-conditioned. This is because the radiation part of the matrix is very seldom ill-conditioned, but the radiation part is usually much larger than the solid part. Ill-conditioning requires larger preconditioning matrices with higher fill-in values, which reduces the performance of the solution. Using the PARAM RADSOLVE option confines the high-fill-in preconditioner to the much smaller solid portion of the matrix, resulting in better solver performance.

N1 = REDUCE

N1 = REDUCE (or 69) creates a reduced finite-difference conductance/capacitance model between user-specified groups of elements during the Analyzer run.

T1 is a group name of elements that are condensed into a finite difference node with label T2 in the reduced model.

The reduced model is written out at the end of the run onto file INPF.reduced in INPF format.

Model creation:

• For a reduced model of n finite-difference nodes there must be n PARAM REDUCE Cards in the model, each specifying a separate group and label.
• Care should be taken that all elements of the model are specified on a PARAM REDUCE Card, none should be left out. In case some are left out, TMG assigns a single aggregate group for all left-out non-sink elements, and a separate group for each left-out SINK element.
• SINK elements of different temperatures should be specified in their own groups and should not be mixed with non-SINK elements. SINK elements of the same temperature may be grouped together.

The reduced model contains:

• TINIT Cards that specify the average temperature of the group at the printout interval.
• QNODE and XCAP Cards containing the total heat inputs and capacitances of the groups at the time of the printout. For transient runs these may be written with time-dependent tables.
• XCOND RAD Cards to model direct radiative conductance paths between the groups. Their values are calculated by computing the heat flow through the radiative conductances between the groups of elements and dividing by the . For transient runs these may be written with time-dependent tables.
• XCOND COND Cards to model direct conductive paths. Their values are computed by calculating the heat flow between the groups through all direct conductive paths and dividing by the temperature difference between them. For transient runs, these may be written with time-dependent tables.
• XCOND COND Cards to model conductance paths not included in the direct conductance and radiative paths. For transient runs these may be written with time-dependent tables.
• If T3=1, XCOND FOLLOWER conductances to recover the temperatures of the original model. If this option is specified, care should be taken that the node number T2 is different from any element label in the model.
• SINK Cards to specify the temperature of temperature boundary condition groups. For transient runs these may be written with time-dependent tables.
• NAME Cards associating the group names with the assigned labels.
• A PRINT 0 0 SUBTEMP Card to automatically create a file tmgtempe.unv to recover element temperatures.

The following example will create a three-node finite difference model between the three groups SPACE, ANTENNA, and DECK.

PARAM REDUCE SPACE 2001
PARAM REDUCE ANTENNA 1
PARAM REDUCE DECK 2

N1 = RESTART

N1 = RESTART (or 91) governs the reuse and recalculation of different thermal model parameters during restarts. The Card 2a M parameter must = 231.

T1 is a mnemonic defining the type of thermal model parameter. T1 may be:

• COND_CAP_FLUID for conductive conductances, capacitances, and hydraulic resistances calculated by the COND module, and the thermal model parameters with the mnemonics CON, CAP, and HYD on file MODLF.
• THERMAL_COUPLINGS for thermal couplings calculated by Card 6e AREA Cards of the types: CONV, COND, RAD, VIEW, NEAR, CSERIES, RSERIES, NEARA, NEARAS, NEARS, NEARAR, INTER, NEARM, NEARP, SOLAR, NEARFOL, NEARLP, ABS, XCOND, CONVLP, RADTOT, MERGE, NEARVF, NEARTOT, NEARRES, NEARA1W, RESISTA, CONV1W, INTERTO, RAD2, NEARAR2, RADTOT2, NEARAR3, INTER2, INTER2T, INTERB, INTERBTOT and the thermal model parameters with the mnemonic CNF on file MODLF.
• CONVECTION_COUPLINGS for thermal couplings calculated by Card 6e AREA Cards of the types: NEARAFU, NEARC1, NEARC4, NEARC5, NEARC12, NEARC13, FREE, NEARF, NEARC6, NEARC7, NEARC8, NEARC9, NEARC10, NEARCIN, NEARCOUT, NEARC16, NEARC19, PLATE, PLATEH, SPHERE, CYLIND, INCCHNL, CAVITY, CAVITYH, CONCYL, CONSPH, NEARC21, NEARC22, NEARC23, NEARC24, NEARC41, NEARC42, NEARC43, NEARC43, NEARC44, NEARCB1, NEARCB2, NEARCB3, NEARCB4 and the thermal model parameters with the mnemonic CNF on file MODLF.
• BLACK_BODY_VIEW_FACTORS for view factors calculated by the VUFAC and HEMIVIEW modules by requests of the types Card 6a and 6r, and the thermal model parameters with the mnemonics BVF on file VUFF. If previously calculated black body view factors were calculated with the ray-tracing VFTRACE option, and specular and transparent properties are present, they should be reused only if the specular and transparent properties have not changed from the previous run.
• RADIATIVE_CONDUCTANCES for radiative conductances calculated by the GRAYB module, and the thermal model parameters with the mnemonic RAD on file MODLF.
• ORBITAL_VIEW_FACTORS for orbital view factors, calculated with Card 6 requests of the type Card 6b, 6d, 6k, 6l, 6n, 6s, 6t, and 6u, and the thermal model parameters with the mnemonics EVF, EVR, ALB, ALR, SVF on file VUFF. If specular and transparent properties are present, previously calculated orbital view factors should be reused only if the specular and transparent properties have not changed from the previous run.
• HEAT_FLUX_VIEW_FACTORS for heat flux view factors, calculated with Card 6 requests of the type Card 6n, and the thermal model parameters with the mnemonics HVF and HVI on file VUFF. If specular and transparent properties are present, previously calculated heat flux view factors should be reused only if the specular and transparent properties have not changed from the previous run.
• ORBIT_HEAT_FLUXES for orbital heat loads calculated with the Card 2a M = 32, M = 64, and M = 128 options, and the thermal model parameters mnemonic HTF on file MODLF.

T2 is a flag governing the retention and reuse, or deletion of these parameters on files VUFF or MODLF. T2 may be the mnemonic:

• REUSE
• DELETE

If only REUSE options are present, all parameters other than the ones specified to be reused are deleted.

If only DELETE options are present, all parameters other than the ones specified to be deleted are reused. In case of a conflict between the REUSE and DELETE options, the REUSE option prevails.

T3 is a flag governing the recalculation or non-recalculation of these parameters. T3 may be the mnemonic:

• CALC
• NOCALC

N1 = RESTARTDIR

N1 = RESTARTDIR (or 91) specifies the name of the directory T1 where the restart files VUFF, MODLF, and tmggeom.dat are located if a restart is being performed.

N1 = RETRANS

N1 = RETRANS (or 36) redefines the laminar to turbulent transition Reynolds Number for duct flow to T1. The default transition Reynolds Number is 2300.

N1 = REVINC

N1 = REVINC (or 88) specifies the reverse side increment T1 for all elements whose reverse side properties are defined on MAT Cards. The reverse side of element I will be I+T1. If this Card is not present, the reverse side numbering is automatically assigned.

N1 = RUNSINDA

N1 = RUNSINDA (or 74) allows you to run any one of the SINDA versions, or ESATAN, to calculate temperatures instead of the Analyzer module. You have to customize the script tmg/com/runsinda.com or tmg/com/runsinda.cmd, and remember to specify the SINDA version with a PARAM SINDA or PARAM SINDA85 Card.

N1 = SCRMESG

N1 = SCRMESG (or 38) writes error/warning messages onto the screen for TMG CFD instead of the verbose log file.

N1 = SINDA

N1 = SINDA (or 62) creates an old COSMIC SINDA format output.

T1 must be 0. T1 may be a blank if T2 is blank.

T2 is an optional file name, on which the SINDA format output is written. If T2 is blank, the output is written on file FMODLF.

T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG will attempt to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = SINDA85

N1 = SINDA85 (or 64) creates a SINDA format output.

T1 must be 0. T1 may be a blank if T2 is blank.

T2 is an optional file name, on which the SINDA85 format output is written. If T2 is blank, the output is written on file sinda85.dat.

T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG will attempt to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = SINKBND

N1 = SINKBND (or 128) merges all boundary elements of a sink element to the original sink element. This has the effect of sinking all of those boundary elements to the sink temperature.

SOLIDACC

N1 = SOLIDACC (or 1) specifies that the conductive conductances of hexahedral or wedge-shaped solid elements that do not have unique element centers are calculated accurately by the COND module for the element center option. For the recommended element CG option this option is ignored.

A solid element will not have a unique element center if on one of its surfaces the perpendiculars from the edges do not meet at a single point. With PARAM SOLIDACC conductances are calculated by subdividing the element into tetrahedral sub-elements. The sub-element's element center closest to the solid element's CG is chosen as the element center, and all other sub-elements' element centers are eliminated with the star-delta transformation.

If a PARAM SOLIDACC Card is not present (the default option), a unique element center is assumed near the CG for odd-shaped solid elements. This yields fewer conductances, but creates a less accurate conductance matrix.

If the only solid elements present are tetrahedral, brick-shaped, or regular wedge-shaped, a PARAM SOLIDACC Card is not used.

N1 = SPECTRA

N1 = SPECTRA (or 102) specifies that radiative couplings and radiative heat fluxes are calculated using spectrum-dependent optical properties in a multispectral run, instead of the standard gray-body assumption with two (IR and solar) bands. Multispectral runs should be specified only if wavelength-dependent optical properties are present, and more accurate results are desired.

A separate radiative coupling matrix for radiative heat transfer between elements is created for each band in the IR spectrum, and separate radiative heat loads are created for each band in the total spectrum.

T1 specifies the number of bands into which the total spectrum is subdivided:

• T1 must be >1 for a multispectral run.
• If T1 = 0, a standard gray body run is performed, just as if a PARAM SPECTRA Card was not present. Radiative couplings are calculated only for the IR spectrum, and radiative heat fluxes are calculated only in the IR and solar spectra. A PARAM SPECTRA Card’s presence is necessary for gray body runs only if wavelength-dependent optical properties are present in the model to transform their properties into the equivalent solar and IR spectrum properties.
• T1 may take the form of Tn (e.g. T3), where n is a table number that specifies the wavelength breakpoints between the bands. The number of bands is then computed from the table.
• The independent variable on the TABTYPE Card must be WAVELENGTHNUMBER.
• The dependent variable must be TEMP or WAVELENGTH.
• If the dependent variable is TEMP, then an equivalent wavelength is calculated for each temperature specified in the table, using Wien’s law for a radiating black body.
• If the dependent variable on the TABTYPE Card is WAVELENGTH, then the values on the TABDATA Card must be specified in units of μ (microns), where μ=1×10^-8 m.

T2 is the constant in the equation for Planck's spectral energy distribution, value 14387.69 μK or 25897.84 μR.

T3 specifies the method by which the spectrum is subdivided when T1 > 0:

• The radiative spectrum is considered to range from 0−>λ_max and λ_max−>1000λ_max, where λ_max is defined by T3. The spectrum 0−>λ_max is subdivided according to a method defined by T3, T5, and T6.
• For T3 = MAXMINWAVE, T5 is the minimum wavelength of interest and T6 is the maximum wavelength of interest λmax, specified in μ (microns). The spectrum 0−>T5 is a single band. The spectrum T5−>λmax is subdivided into T1 bands, with breakpoints chosen to avoid bands that would not radiate significantly. The spectrum λmax−>1000λmax is a single band.
• For T3 = MAXMINTEMP, T5 and T6 are the minimum and maximum temperatures of interest. For example, these may be specified to be room and lamp temperatures for models where radiative heating with lamps is calculated. The first band 0−>λmin and the last band λmax−>1000λmax are computed so that each one of them radiates a power equal to 1/T1 using the spectral distribution for a black body at temperatures T5 and T6. The spectrum λmin−>λmax is subdivided into T1-2 bands. The breakpoints between λmin and λmax are chosen to avoid “dead bands”, i.e. bands that do not radiate significantly.
• For T3 = EQPOWER, λmax is calculated from temperature T5, and the spectrum 0−>λmax is subdivided into T1-1 bands. λmax is then chosen so that each band radiates the same power at the black body at temperature T5. T6 is ignored.
• For T3 = EQBAND, T5 is the maximum wavelength of interest λmax, specified in μ. The spectrum 0−>λmax is subdivided into T1-1 equal bands, and the spectrum λmax−>1000λmax is a single band.

T4 is a wavelength value in μ (microns) that defines the breakpoint between the solar and IR spectra. T4 has a number of uses:

• T4 permits the specification of optical properties using the standard gray body values in the IR and solar spectra. It is not necessary to specify all optical properties as wavelength-dependent. For multispectral runs, for each pair of constant IR and solar spectrum optical properties (E and ABSORPTIVITY), (IRSPEC and SPECULARITY), (IRTRANS and TRANSMISSIVITY), (IREXTINCT and SOLAREXTINCT), (IRSCATTER and SOLARSCATTER), (IRREDIFF and SOLREDIFF), TMG creates equivalent wavelength-dependent property tables for each of the bands. For example, for an absorptivity-emissivity pair of properties, the emissivity value in the table in the wavelength band λi to λi+1 where λi+1≤T4 is set to the solar absorptivity value, in the band where λi≥T4 is set to the emissivity value, and in the band where λi λi+1 it is a weighted average of the emissivity and absorptivity.
• Note: Wavelength-dependent optical properties should be defined only with IR spectrum mnemonics, not solar spectrum mnemonics. Thus, ABSORPTIVITY should not be defined as a function of wavelength, while E can be.
• For gray-body runs (T1=0) if wavelength-dependent optical properties are present, they are transformed into equivalent gray body solar and IR properties with T4.
• Radiative heat loads into elements from radiative sources (Card 6 SOURCE Cards) are calculated for all the bands in the spectrum 0−>1000λmax.
• Radiative couplings for radiative heat transfer between elements are not calculated with discrete bands in the spectrum 0−>λi where λi≤T4. In other words, minimal radiation is assumed to occur in the solar spectrum. The fraction of radiation emitted by an element is summed with the lowest radiating band. To permit radiation in all bands, you specify the ALLBAND option for T8.
• T4 is used to modify the orbital solar spectrum (solar heat input and albedo) calculations. First, an internal power vs wavelength table is created from published measured solar data. Next, this table is modified so that the power emitted in the bands with the range λi to λi+1, where λi≥T4, is set to 0, and is summed with the largest band where λi+1≤T4. Last, the table is non-dimensionalized to create a constant weighting factor for each band, where the weighting factor represents the fraction of energy emitted in that band.
• T4 is used to modify the Earth IR spectrum for orbital calculations. First, an internal power vs wavelength table is created from the specified Card 2a PIR for the Earth. Next, an equivalent Earth temperature is calculated, assuming an Earth emissivity of .612. Next, a power vs wavelength table is created for this spectrum using Planck’s Law. Next, the table is modified so that the power emitted in the bands defined by the range λi to λi+1≤T4 is set to 0, and is summed with the lowest band where λi>T4. Last, the table is non-dimensionalized to create a constant weighting factor for each band, where the weighting factor represents the fraction of energy emitted in that band.
• T4 = 0 defaults to T4=3μ.

T5: The interpretation of T5 depends on the mnemonic T3. It is λmax for EQBAND, the black body temperature for EQPOWER, the minimum temperature of interest for MAXMINTEMP, and the minimum wavelength of interest for MAXMINWAVE.

T6: The interpretation of T6 depends on the mnemonic T3. It is the maximum temperature of interest for MAXMINTEMP and the maximum wavelength of interest for MAXMINWAVE.

T7 is optional:

• If present, it must be a table specified as Tn (e.g. T5), where n is the table number in which the weighting factors for the spectrum-dependence of the sun is defined on TABTYPE and TABDATA Cards. The dependent variable on the TABTYPE Card must be PSUN, and the independent variable must be WAVELENGTH.
• The solar power magnitude in a particular band is calculated by interpolating a weighting factor from the table, and multiplying it by the PSUN value specified for that particular Card 6 request.
• If is T7 = 0, an internally generated solar power table is used.

T8 may be blank, or

• SINGLEIRSPECTRUM in which case all the radiative couplings will be merged into a single band, or
• ALLBANDS in which case all the bands radiate, including those in the solar spectrum.

N1 = SUBSTR

N1 = SUBSTR (or 5) requests the MEREL module to reduce the number of elements with substructuring, and it specifies the elements not to be eliminated.

Substructuring is a process that reduces the size of the model to be solved by performing a series of star-delta transformations before solving for temperatures. The information to recover the eliminated elements' temperatures is written on file MODLCRF.

T1 is:

• The element number or group name of element(s) not to be eliminated, or
• The code RADNODES (or -2.34), in which case none of the elements connected to a radiative conductance is eliminated, or
• 0 or blank if any element may be eliminated.

T2 is blank or may be:

• GSUM (or 33), in which case those elements whose conductance sums is > T3 are eliminated.
• RCMIN (or 31), in which case all elements with RC values < T3 are eliminated, including zero-capacitance elements. RC is the element's capacitance divided by the sum of its conductances.
• CMIN (or 32), in which case all elements with capacitance values < T3, including zero-capacitance elements, are eliminated.

T3 may be a value associated with T2 = GSUM, RCMIN, or CMIN, or is blank.

The Analyzer calculates the temperatures of the eliminated elements of the reduced model in a single iteration at each printout interval.

Several PARAM SUBSTR Cards may be used simultaneously.

The elements to be eliminated are selected as follows:

• All elements specified on Card 8 and the T2 elimination criteria are eliminated.
• Sink elements, MCV elements, phase change elements, hydraulic elements, elements referenced on INTERP or THERMST Cards, and elements connected with one-way or non-linear non-radiative conductances are automatically kept.
• The element(s) of T1 are kept, even if a T1 element is requested to be eliminated on a Card 8 or by one of the T2 elimination criteria.
• If all the elements in a steady-state run are candidates for elimination, a single PARAM SUBSTR Card with T1 = T2 = T3 = blank will eliminate all the elements.

Substructuring can reduce Analyzer run times by solving a smaller and less ill-conditioned model that has larger RC values. Substructuring is especially useful when:

• High-conductance elements are eliminated in an ill-conditioned model.
• Zero capacitance and small-RC elements that govern the integration time step are eliminated for transient analysis.

If the eliminated elements are connected to radiative conductances, the Card 2a SIGMA and TLIN parameters must be specified.

Elements referenced in user-written subroutines must be kept by specifying them on a T1 parameter.

Substructuring results are exact if the following conditions are met:

• The elements to be eliminated are joined to other elements by conductive conductances only, or by radiative conductances only, or by a mixture of the two with an exact value for TLIN.
• Steady-state analysis is used.
• The PARAM THIN option is not used.

If transient analysis is performed and non-zero capacitance elements are eliminated, the substructuring process will not be exact, since the order of the differential equations describing the model is changed. However, the error will generally be small if the eliminated element capacitances are comparatively small or their rates of change of temperature are small.

For SINDA models, only the temperatures of the reduced model will be calculated. The recovery information on file MODLCRF is not transmitted to the SINDA file on file FMODLF.

N1 = TC_CONVECT

N1 = TC_CONVECT (or 144) excludes the thermal coupling elements from the convection to environment when the thermal coupling and convection to environment boundary conditions are on the same selection.

N1 = TDEPORTHO

N1 = TDEPORTHO (or 116) specifies the criterion for updating temperature-dependent orthotropic conductances. By default, these conductances are updated only when the maximum temperature change for any temperature-dependent orthotropic element exceeds 20 degrees. Increasing this value speeds up Analyzer runs, decreasing this value increases the accuracy.

N1 = TDIFS

N1 = TDIFS (or 18) specifies the temperature iteration convergence criterion for steady-state runs, and transient runs when an implicit technique (Crank-Nicolson or backward) is used.

Convergence is achieved if the maximum temperature difference for any element between two subsequent iterations is < abs(T1). This criterion is used both in the outer (nonlinear) iteration loop as well as in the linear solver activated by PARAM ILU.

The steady state T1 default is the Card 2b GRADNT parameter, the transient default is 0.001.

T2 specifies how the convergence criterion is applied:

• If T2 is blank or ALL, T1 applies to all elements.
• If T2 is CAPONLY, T1 applies only to elements with non-zero capacitances.
• If T2 is NONOPPENHEIM, T1 applies only to non-Oppenheim elements.
• If T2 is IDEASONLY, T1 applies only to Card 5 elements.
• If T2 is an element number, T1 applies only to that element.
• If T2 is a group name, T1 applies only to that group name.
• If T2 is STEADYSTATE, T1 applies only to the solution control STEP cards with the steady state solution type.

Multiple PARAM TDIFS Cards may be used. In case of conflict, the tighter convergence criterion specified for the element is used.

N1 = TEMPCONDITION

N1 = TEMPCONDITION (or 114) ensures that at printout times no element temperature falls below the minimum SINK temperature or initial temperature. It is possible for an element temperature to fall below the minimum allowable temperature if the element is highly distorted, or because other nonlinearities occur.

N1 = TEMPDIFFTARGET

N1 = TEMPDIFFTARGET (or 119) creates an additional stopping criterion for transient runs. T1 is an element number or group name of element(s), and T2 is a temperature difference. The run is considered to be completed when the difference in average temperature of T1 between two consecutive transient timesteps is less than T2.

N1 = TEMPTARGET

N1 = TEMPTARGET (or 119) creates an additional stopping criterion for transient runs. T1 is an element number or group name of element(s), and T2 is a temperature. The run is considered to be completed when the average temperature of T1 reaches the temperature T2 either from above or below.

N1 = TIMETABLE

N1 = TIMETABLE (or 140) controls the interpolation method of time-dependent tables when the time lies outside of the table range.

• If the keyword is PERIODIC (or 0), the table values are periodic for the times larger than the last table entry. For times smaller than the first table entry, the solver uses the lower bound value.
• If the keyword is TRUNCATE (or 1), the solver uses the lower bound value for times smaller than the first table entry, and the upper bound value for times larger than the last table entry.

N1 = THERMOSTAT

N1 = THERMOSTAT (or 110) is a global flag for thermostats and is used only if Card 9 THERMST Cards are present. The following options are available:

• If T1 is the code AVGTEMP, then all the heater elements are set to the average temperature of the Thigh and Tlow values specified on the THERMST Card. This is a simple way to calculate heater power requirements. This is the default option for steady-state analysis if there are Card 9 THERMST Cards present but no PARAM THERMOSTAT Card. Note that this can result in negative heat loads on the heaters, when the heater temperatures are below those of the environment.
• If T1 is the code PROPORTIONAL, then all the thermostats are replaced by proportional controllers, with the temperature limits equal to the Thigh and Tlow values specified on the THERMST Card. Care should be taken for this option during steady-state analysis to use a low value for the Card 2b DT relaxation parameter to prevent oscillation of the solution.
• If T1 is the code Q_EQUIV, then, if a steady-state run is performed, TMG will automatically perform multiple runs to calculate through sensitivity analysis the heater powers required to keep the sensors at the average of the Thigh and Tlow temperatures specified on the THERMST Card. If there are N thermostats in the model, N runs will be performed.

T2 is the temperature convergence criterion used for convergence for the Q_EQUIV option. By default T2 is 0.1 degrees. Otherwise it is ignored.

T3 is the maximum number of iterations performed for the Q_EQUIV option. Otherwise it is ignored. T3 defaults to 10. If the model has no non-linear conductances, there should be only two iterations.

N1 = THIN

N1 = THIN (or 9) eliminates insignificant conductances in the MEREL module. A conductance is considered insignificant if:

Where:

• Gij is the conductance between elements i and j
• GSUMi is the sum of the conductances to element i
• GSUMj is the sum of the conductances to element j

Radiative conductances are linearized with the Card 2a SIGMA and TLIN parameters during the thinning out process.

T1 values above 0.01 should be used with caution; too many eliminated conductances may result in degradation of accuracy.

Thinning can considerably reduce the run times of the Analyzer module if there are too many insignificantly small conductances present.

This option also eliminates negative conductances, so it should be used with caution when the PARAM COND NEW option is in effect.

N1 = TMGVERSION

N1 = TMGVERSION (or 113) sets the defaults for a particular version number. For example, if a set of default parameters were changed in TMG version 13 from version 12, the Card PARAM TMGVERSION 12 will reset those parameters to those of version 12. The objective is to be able to perform comparison runs between two versions to evaluate the effects of a change in defaults.

N1 = TRASYS

N1 = TRASYS (or 63) creates a TRASYS format output.

T1 must be 0. T1 may be a blank if T2 is blank.

T2 is an optional file name, on which the TRASYS format output is written. If T2 is blank, the output is written on file FMODLF.

T3 and T4 are two element numbers, T4 > T3. If these are specified, TMG will attempt to create internal element numbers (e.g. boundary elements for the COND module, Oppenheim elements, etc.) within these limits.

N1 = UNITS

N1 = UNITS (or 73) specifies to TMG what units are being used.

• T1 is a numerical code, not in use at this time.
• T2 is the number of model’s length units that are in a meter. For example, if the model’s length units are in inches, T2 = 39.37
• T3 is the number of the model’s force units that are in a Newton. For example, if the model’s force units are Kgf, then T3 = 1/9.81 = 0.1019
• T4 is the model’s temperature scaling factor with respect to Centigrade. For example, if the model’s temperature units are in Fahrenheit or Rankin, T4 = 1.8
• T5 is the model’s temperature offset for absolute zero. For example, if the model’s temperature units are in Centigrade, T5 = -273.15
• T6 is the model’s time scaling units with respect to seconds. For example, if the model’s time is in hours, then T6 = 3600.

The model’s units can be accessed from a user-written subroutine with CALL TUNITS. For more information, see Card 10 - User-Written Subroutines USER1 and USERF.

N1 = UPDATEOPTICAL

N1 = UPDATEOPTICAL (or 137) accounts for the changes in the surface optical properties, caused by forming a condensation film over a surface.

T1 specifies the calculation method type of the surface optical properties. There are currently two methods supported by the solver:

• If T1=0, the first calculation method is used, where the transmissivity and specular reflectivity of the film layer are defined as a function of film thickness for irradiation and solar spectrums. The transmissivity and specular reflectivity are defined, while the software calculates the emissivity/absorptivity.
• If T1=1, the second calculation method is used, where the index of refraction and extinction coefficient of the film layer are defined as a function of wavelength. The film optical properties are automatically calculated by the thin-film analysis algorithm.

In both methods, the combined transmissivity, t, and reflectivity, r, of media 1 and 2 are calculated as follows:

T2 is the film type. If T2=0, then the film is incoherent. If T2=1, the film is coherent. The default value is 1. This option is only applicable to the second method.

T3 is the film thickness threshold for updating the optical properties.

T4 is the film thickness change rate threshold between two consecutive time steps.

Example of the first method

PARAM UPDATEOPTICAL 0 1 1.E-10 1.E-10
$
TABTYPE 40 FT_IRSPEC FILMTHICKNESS
TABDATA 40  0.00  1.000000E-07
TABDATA 40  0.00  1.500000E-07
TABDATA 40  0.00  2.000000E-07
TABTYPE 41 FT_IRTRANS FILMTHICKNESS
TABDATA 41  0.990  1.000000E-07
TABDATA 41  0.985  1.500000E-07
TABDATA 41  0.980  2.000000E-07
TABTYPE 43 FT_SPECULARITY  FILMTHICKNESS
TABDATA 43  0.00  1.000000E-07
TABDATA 43  0.00  1.500000E-07
TABDATA 43  0.00  2.000000E-07
TABTYPE 49 FT_TRANSMISSIVITY  FILMTHICKNESS
TABDATA 49  0.990  1.000000E-07
TABDATA 49  0.985  1.500000E-07
TABDATA 49  0.980  2.000000E-07

Example of the second method

PARAM UPDATEOPTICAL 1 1 1.E-10 1.E-10
$
TABTYPE 160 IREFFILM WLENGTH
TABDATA 160 1.396       0.0000002
TABDATA 160 1.373       0.000000225
TABDATA 160 1.362       0.00000025
TABTYPE 161 EXTFILM WLENGTH
TABDATA 161 6.911504    0.0000002
TABDATA 161 2.736676    0.000000225
TABDATA 161 1.683894    0.00000025

N1 = USERDLL

N1 = USERDLL (or 109) allows you to bypass the use of the required Fortran compiler for user-written subroutines. Instead, it allows you to use another Fortran compiler. A dll will be created, and automatically linked into the code.

An important limitation: You MUST have the following defined in your user-written subroutine:

DOUBLE PRECISION T, TIME, DT

Also, only TMG subroutines documented in the Reference Manual may be called.

N1 = USRCNDADD

N1 = USRCNDADD (or 51) signals to the Analyzer that the user-written subroutine USER1 is called with KODE = 5, and a set of new conductances are to be added to the Analyzer conductance matrix at run-time. For more information, see the CALL ADDCOND option in Card 10 - User-Written Subroutines USER1 and USERF.

VFADJUSTALL

N1 = VFADJUSTALL (or 92) allows the proportional adjustment of all view factors when the Card 2a KSP = 3000000 option is specified. Without this option only unshadowed view factors are adjusted. The benefit of using this option is that it increases the likelihood that after adjustment all the elements’ view factors sum to 1.

PARAM SOLIDACC
$ CONDUCTIVE CONDUCTANCES BETWEEN ODD-SHAPED
$ SOLID ELEMENTS WILL BE ACCURATELY CALCULATED
$
PARAM NLOOP 200
$ THE NLOOP PARAMETER IS SET TO 200
$
PARAM ACCEL 10 100
$ THE ACCELERATOR PARAMETERS ARE REDEFINED
$
PARAM QUARTIC 
$ THE FOURTH ROOT IS FOUND AT EACH ITERATION
$
PARAM SUBSTR KEY GSUM .1
PARAM SUBSTR 0 RCMIN .01
PARAM RECOVER
$ SUBSTRUCTURING AND RECOVERY ARE PERFORMED.
$ THE ELEMENTS ASSIGNED TO KEY ARE KEPT.
$ ELEMENTS WITH GSUM > .1 AND RCMIN < .01
$ ARE ELIMINATED.
$
PARAM ENGBAL .001
$ THE ENERGY BALANCE MUST BE < .001 AT CONVERGENCE
$
PARAM ENGBAL .01 FRACT 
$ ENERGY BALANCE MUST BE < .01 OF HEAT FLOW INTO SINKS