RELEASE NOTES

SINDA/FLUINT Version 4.5

Conductor HR Values

The absolute value of the heat rate through a conductor is now available as the parameter “HR.” This output variable may be accessed in logic or expressions. Note that because an absolute value is used, there is no dependence of HR on direction: no “from node” nor “to node.”

Built-in Sonic Limits (Choking)

By default, almost all paths (including tubes) are now checked for sonic limits in their throats (AFTH), and if such limits are detected they are automatically modeled.

This new system makes the CHOKER and CHKCHP routines obsolete, and to avoid confusion their use is discouraged though still accepted for backward compatibility. Users are encouraged to delete those calls from their models, and replace them with appropriate MCH and HCH values.

Unfortunately, this new default (checking for choking everywhere and applying nonequilibrium expansions if needed) stresses many fluid property descriptions, hitting lower limits or requesting off-chart thermodynamic properties. The user may need to selectively or globally turn off such choked flow modeling (using MCH=0) if improvements to the underlying fluid properties cannot be made and if choking is deemed to be irrelevant to the problem at hand. In other words, the default was chosen as a safety measure, and it may be irrelevant or disruptive in some models.

Expanded Heat Pipe Modeling

A new routine, HEATPIPE2, has been introduced to extend the prior HEATPIPE routine to cover pipes with strong circumferential gradients: 2D heat pipe models using more than one circumferential node, non-circular heat pipes, and flat-plate vapor chamber fins. Other improvements include the ability to disregard incorrectly sized/charged VCHPs, greater flexibility of input units (HPUNITS), and a new routine (HPGLOC) is available to extract the gas front location from previous HEATPIPE or HEATPIPE2 calls.

Heat pipe users are urged to review recent FloCAD® enhancements specifically designed for heat pipe modeling inside 3D FEM/FDM thermal models.

Miscellaneous Improvements

• Many model size limits have been relaxed or even eliminated (for versions based on Fortran 90/95 compilers, which currently includes all PC versions).
• More complex syntax is now allowed for conditional operators inside of spreadsheet-like expressions. For example, “heat = ( (sub.t1 > 300) && (sub.t1 <= 400) )? on : off” meaning “if the temperature of node 1 in submodel sub is above 300 degrees but less than or equal to 400 degrees, set the register heat to on, otherwise set it to off.”
• Routines for sizing and/or simulating Peltier devices, including Thermoelectric Coolers (TECs) are now available.
• A new connector device has been added to simplify modeling of common orifices: the ORIFICE connector. Built-in correlations for sharp-edged and long orifices are available, or the user can provide their own coefficients. The resistance of the orifice is estimated, as well as the prediction of its vena contracta for the purposes of choked flow detection and modeling.
• A new connector device has been added to allow users to specify how a component behaves (pressure drop versus flowrate characteristics) using tables input as ARRAY data: the TABULAR connector. The tables can use a variety of units for pressure/head and flow. Either a single curve can be input, or a family of parametric curves based on a user-defined independent parameter. While intended to assist in the modeling of valves and other devices using vendor-supplied data, TABULAR devices represent a general-purpose means of describing a flow resistance for a variety of components.
• Much greater flexibility now exists for defining units for pump curves, enabling users to apply vendor data directly without conversions.
• To avoid any confusion, the PRVLV and BPRVLV connectors have been renamed to UPRVLV and DPRVLV connectors, respectively. The older designations are still accepted in input files, but are no longer documented to discourage their continued use.
• A new utility routine, REPATH, has been added to allow users to easily estimate the Reynold’s number for any tube or STUBE.
• Double precision is now on by default. The user can return to single precision by
  specifying SINGLEPRECISION (or simply SINGLEPR) in OPTIONS DATA.
• The values of ITHOLD and ITHLDF now default to zero: the temporary suspension of submodels during steady states (an attempt at speed savings that is not universally successful) will no longer be performed by default.
• The COMPLQ routine has been extended slightly to apply a proportioned compliance to two-phase tanks. Previously, COMPLQ only applied compliance to hard-filled liquid
  tanks. This change avoids a discontinuity at the saturated liquid line.
• A new option (YTHUSH routine) is available to eliminate fatal errors when the wye-tee (converging/diverging flow) routines such as YTKALG and YTKALI encounter condi-tions for which their correlations do not apply. Use of this option means the user accepts the burden of checking the final results, but permits both steady-state excursions (temporarily reversed flow directions, for example) as well as extrapolations of the correlations.
• A change has been made to the PID controller routines formulation: they now return values of the control variable (CV) instead of changes to the control variable (CV). This change is not backward compatible with previous models.

Significant External Program Improvements

• An Excel-based utility for reading and plotting SAVE files (all the binary files produced
  by the SAVE, RESAVE, SAVEDB etc. routines) is now available as part of the EZ-XY®
  installation.
• FloCAD® has been significantly expanded to include “pipe” methods for drawing 1D
  lines representing fluid ducts and heat pipes, and for attaching these to Thermal Desktop®
  surfaces and solids. These improvements make FLUINT modeling significantly easier for
  common pipe-flow and heatpipe problems.
• C&R Thermal Desktop® now features native finite difference solids for enhanced
  parametric modeling.

SINDA/FLUINT Version 4.4

High Speed and Compressible Flow

Although FLUINT always handled the full thermodynamic and momentum effects associated with compressible flow (including choking), it previously made the assumption of kinetic energy being negligible compared to thermal energy (enthalpy-based or advective transport). Kinetic energy transport is now included in the flow solutions, such that certain high speed flow phenomena (e.g., Fanno line flow, etc.) can be modeled.

While this enhancement is mostly transparent to the user, there are a few subtle changes and cautions required. First, all paths now have a definable flow area (AF) and throat area (AFTH), although their use is optional for many connector device types. If a path has no defined flow area, then kinetic energy cannot be taken into account and the flow will “stagnate” upstream of that path, perhaps to be accelerated again downstream. Also, the optional choking calculations (CHOKER, CHKCHL, etc.) no longer assume that the upstream lump represents stagnation, and therefore an expansion process need only be included if the throat area is less than the flow area (AFTH < AF). Finally, any lump in which velocities should be assumed zero can be marked as such using a new “LSTAT=STAG” option in lump subblocks. This designation affects not only choking and kinetic energy calculations, it will also automatically add an accelerational entrance loss (equivalent to FK=1.0) to any appropriate path (tube, STUBE, loss, etc.) attached to and flowing out of that stagnant lump.

Miscellaneous Improvements

• Previously, simultaneous (sparse matrix based) thermal solutions could be invoked on a submodel by submodel basis using MATMET=1. Submodels that are tightly coupled can now be combined into a single matrix using MATMET=2 (Section 4.4.1). Other values of MATMET retain their original meaning: MATMET=0 means iterative, and MATMET=1 means simultaneous (per submodel).
• A network connectivity checker has been added that detects isolated groups of nodes: nodes having no connection to a boundary node in steady states, and arithmetic nodes having no connection to either boundary or diffusion nodes in transients.
• Model size limits have been greatly increased.
• In an attempt to avoid confusion for new users, the built-in spell checker, controlled by the NODEBUG (Options Data) and DEBOFF and DEBON (logic blocks) options, may
  now be controlled using the SPELLON and SPELLOFF commands instead. The old options are still available but have become undocumented.
• In connector subblocks, general path inputs (FR= ..., DUP = ... , STAT = ... etc.) are now tolerated following the “DEV=” specification.
• Defaults now exist for NLOOPS, OUTPUT, and OUTPTF to avoid aborted runs. The default for EXTLIM has been reduced from 50 to 5 to avoid instabilities in models with fluid systems, heat pipes, etc.
• SINDA now tolerates nodes with zero (or absent) conductors as long as ties are present.
• The FORTRAN reserved word list has been expanded to include many f90/f95 keywords (“WHILE”, “ENDDO”). This prevents these names from being used as registers, named control data, etc.
• RESTAR, RESPAR, RESTDB, etc. now check the input record number for validity.
• A new routine, HEATPIPE, has been added to assist in simulating fixed (constant) conductance heat pipes and variable conductance heat pipes.
• CSGFAC can now be used as a CSGMIN divisor for time step control in FWDBCK (“Transient”) as well as FORWRD. In FWDBCK, use of negative DTIMEI is now taken to mean overriding of the default (DTIME=0) internal time step control, using instead a time step equal to CSGMIN/CSGFAC. Positive DTIMEI retains its original meaning as the directly specified time step.

SINDA/FLUINT Version 4.3

Reliability Engineering

A new top-level module has been added that enables analysts to treat unknowns and uncertainties statistically, estimating the chances of exceeding arbitrarily complex failure limits. Dimensions, properties, boundary conditions, usage scenarios, etc. are normally not fixed quantities, but can assume different values within a prescribed range. Margins and safety factors are normally used to deal with such uncertainties, stacking up worst-case scenarios and minimum/maximum values to produce conservative designs. The Reliability Engineering module allows the degree of over- or under-design to be quantitatively measured by predicting the chances that the design will pass muster: the reliability of the design. Combined with the Solver, it can be used to synthesize a design that meets reliability requirements up front, intelligently balancing cost against risk.

Fluid-to-Fluid Heat Transfer Ties

Heat can flow directly between lumps using "fties", a new network element. Advective transport of energy is usually always dominant over conductive transport within a fluid. However, in systems with very slow flowrates or that use highly conductive fluids (such as liquid metals), neglecting conductive transport may not be acceptable. Another use of fties is modeling thin-walled heat exchangers between two sections of the same fluid system (e.g., regenerators). Fties may be applied with a constant or user-varying conductance ("UF"), or with a constant or user-varying heat flux ("QF"), or as a companion to a tube or STUBE connector that automatically updates the ftie conductance to take into account the current fluid flow within those paths. In fact, such an "AXIAL" ftie can easily be generated along with any tube or STUBE connector.

Auxiliary routines (WETWICK, PLAWICK, CYLWICK) are available to calculate and apply ftie conductances for flows through conductive porous media.

Twinned Tanks

Whenever the assumption of perfect mixing in two-phase control volumes (equal temperatures and pressures between liquid and vapor) is inappropriate, tanks can be twinned. When twinned, the code will use one tank to model the vapor and one tank to model the liquid within the control volume, analogous to the way twinned paths may be used to solve separate momentum equations for each phase. In fact, twinned paths may be used in combination with twinned tanks for full nonhomogeneous nonequilibrium modeling. Other elements such as ties and fties may also be twinned to facilitate their application to twinned tanks. Twinned tanks use fties and ifaces between each pair, and also apply a new concept called a superpath to handle mass transfer between them. Most of these calculations and manipulations are transparent to the user when default options are used. However, in order to provide full access and control for custom solutions, the number of new variables and features is considerable. Twinned tanks make use of the NONEQ family of routines obsolete. These routines are still available, but are now undocumented. Users are encouraged to migrate to twinned tanks, which (unlike the NONEQ routines) can be used within LINE and HX macros, can be used with twinned paths, and can handle dissolution/evolution phenomena. The NONEQ routines will eventually become unavailable.

Flow-Regime Mapping Default

Because of the preponderance of options requiring flow-regime mapping in modern SINDA/FLUINT, the prior default of IPDC=1 (homogeneous pressure drop, unmapped regime) has been changed to IPDC=6 (automatic regime selection and pressure drops and other calculations based on regime). This may cause two-phase models built for prior versions to give difference results.

QDOT No Longer an Input Variable

In previous versions, a heating rate, QDOT, could be applied to a lump, or QDOT could be calculated automatically if ties were attached to it. To allow for the expansions associated with fties, and to permit users to add an additional source onto a tied lump, the prior meaning of QDOT as both an input and output variable has been split. A new input variable, "QL", now defines the heat applied to a lump in addition to heat flowing to it from nodes (via ties) and from other lumps (via fties).

QDOT retains its original meaning as the total power applied to a lump, but can no longer be used as an input variable: it is now strictly an output variable. This means that occurrences of "QDOT =" in FLOW DATA sections will now cause a preprocessor abort. They must be changed to "QL =" and the user must assure that this power is intended as an additional dissipation (if positive) if ties are present on that lump to get the same results as before. Each occurrence of QDOT in logic and expressions must be examined to see whether it is appropriate to refer to the applied power QL, or to the total resulting power QDOT.

Miscellaneous Improvements

• Thermal Desktop and RadCAD calculations can now be made dynamically from within PC (4.3L or 4.3D) SINDA/FLUINT launched from within Thermal Desktop. This major expansion enables parametric analyses, sensitivity studies, optimization, correlation, reliability estimation, and robust design activities to include variations in geometry, optical properties, thermophysical properties, insulation, contact conductance, orbital parameters, etc. The traditional separation of Thermal Math Modeling (TMM) and Geometric Math Modeling (GMM) has been eliminated. Refer to the Thermal Desktop User’s Manual for more information, or contact C&R.
• An intimate link with Engineous’ iSIGHT® software has been created. iSIGHT provides optimization, reliability engineering, and robust design within a multidisciplinary environment. Refer to the Engineous Software, Inc. website (www.engineous.com) for more information on iSIGHT, and contact C&R for information on how to exploit SINDA/FLUINT’s intimate link with iSIGHT.
• Psychrometric utilities have been added for convenience in dealing with the dew points and relative humidities in air/water mixtures (as well as other mixtures of condensibles and noncondensibles).
• PID controller simulation utilities are now available.
• By default, SINDA/FLUINT now checks array data to see if mixed integer and real data have been input within a single array. This prevents the common mistake of forgetting the decimal point for a real value. This option can be overridden (e.g., when bivariate or trivariate arrays are input) by specifying MIXARRAY in OPTIONS DATA.
• Pump and fan curves may now be non-monotonic: they may include regions with a negative slopes in the head vs. volumetric flowrate curve.
• On PC f90/f95 versions (4.3D and 4.3L) and on the Sun ULTRA 64-bit f90 version (4.3U), memory is allocated dynamically for sparse matrix solutions (which are used for all FLUINT solutions as well as any SINDA submodels using MATMET=1). This eliminates excessive allocations, aborted runs, and any need to use the NWORK or NWORKS options. No user involvement is required to use this feature.
• A replacement of the INCLUDE command, the INSERT command, has been introduced. INCLUDE will still be accepted and is documented but no longer recommended. INSERT works the same as INCLUDE, except that an INSERTed file can also contain another INSERT command: INSERT commands can be nested indefinitely.
• The use of negative signs on the multiplication factors (“F”) of SIV and SPV conductors has been historically used as a signal to use a Node A’s temperature instead of an average temperature of the two end-point nodes. New conductor options (SIVM, SIVA, etc.) have been introduced to avoid conflicts with more modern usage (expressions, FEM conversions).
• A routine, PUTPAT, has been introduced to allow paths to be relocated between lumps.
• A new routine, CHGEXP, allows the expressions defining parameters to be added or changed dynamically (during processor execution).
• A new utility, QFLOW, exists for reporting the heat flows from one or more nodes to any other collection of nodes. Nodes defining each group (“to” or “from”) may be specified individually, by submodel, or as arbitrary lists.
• Auxiliary modeling tools have been added to make modeling of wyes and tees (merging and diverging flows) more convenient, and to make manifold modeling more accurate. These tools include correlation routines (YTKALI, YTKALG) and a K-factor conversion utility (YTKONV).
• A variation of the SUMFLO routine, SUMDFLO, has been introduced. Unlike SUMFLO, SUMDFLO takes into account path duplication factors when calculating total network mass, volume, etc.
• A new routine, HTRNOD, allows any arithmetic or diffusion node to temporarily become a heater node (a type of boundary node). The action is released by RELNOD. These actions may be performed on all arithmetic and/or diffusion nodes within a submodel using HTRMOD and RELMOD.
• The HEATER (thermostatic heater) routine has been expanded to function during steady states using an internal call to HTRNOD.
• A new proportional heater routine, PHEATER, is now available.
• Multiple header blocks of the same type will no longer cause errors as long as they are syntactically identical. If encountered, such redundant blocks will be joined into a single block before preprocessing.
• Short-cut commands now exist to build all available thermal or fluid submodels into the current configuration: BUILD ALL, BUILDF ALL. Also, all current thermal models can be removed from the current configuration using “BUILD NONE”, as can all fluid submodels using “BUILDF NONE.”
• An empty (fake) “HEADER NODE DATA” is no longer required for conductor-only submodels.
• SAVE, RESAVE, and SVPART now accept a minus flag syntax meaning “save all except the following.”
• For the convenience of new users, the most commonly used steady-state routine (FASTIC) may now be referred to using the alias “CALL STEADY.” The most commonly used transient routine (FWDBCK) may now be referred to using the alias “CALL TRANSIENT.”
• Internal improvements have been made in the Solver algorithms, especially for METHO=2 in constrained problems.
• Slight nonconservation of mass and energy in transients due to unavoidable numeric truncation and similar causes has been rectified by accumulating error terms and folding them back into the solution, improving the accuracy of the predictions over several time steps. (Refer to the new RMFRAC control constant.)
• A new control constant, FRAVER, has been added to allow the FLUINT user to denote a characteristic average flowrate for a loop such that it can better decide what level of flow can be neglected as zero and what amount of change in flowrate is negligible per time step.
• A variation of USRFIL called USRFIL2 is available that does not abort if it can’t open a file. Instead, it returns an error flag.

Expressions: Miniature Logic Blocks

Spreadsheet-like registers and register-containing expressions continue to be a popular and powerful feature of the code. Expressions, which can be used to define almost any value in SINDA/FLUINT, have been tremendously expanded. First, the user can refer to processor variables such as "fred.T33" or "abszro" or "drlxcc" or "timen" in the expressions, completing the spreadsheet capabilities by allowing inputs to be declared functions of not only other inputs, but also of output (response) variables. Together with the ability to use conditional operators (IF/THEN/ELSE type logic), each expression can be as powerful as a logic block, and in fact many traditional input options and applications of logic blocks are now largely obsolete (although still supported).

Soluble Gases

Mixture properties can be extended to include the effects of gases dissolving into liquids and evolving out of them. Multiple solutes and solvents can be defined. These effects can be included during either steady-state or transient solutions. Applications include noncondensible gases in single- and two-phase thermal transport loops and vapor compression cycles, pressurant gases in liquid propulsion and in fire retardant delivery systems, and water quality assessments.

Interface Elements

A new path-like FLUINT network element, the "iface," is available for specialized fluid system modeling needs such as subdividing control volumes (including nonequilibrium control volumes or thermally stratified vessels), and for modeling springs, bellows, pistons, servo-control valves, and wicks.

Simultaneous Thermal Solutions

Simultaneous (sparse matrix solution) methods may now be applied to thermal submodels to replace or augment traditional iterative methods in both steady state and transients. Matrix methods are particularly suited to conductively dominated problems (such as translated finite element models from structural meshes) including those in which large conductors are unavoidable. By default, during each iterative pass of STDSTL, FASTIC, and FWDBCK, nodes are solved one at a time treating other nodes temporarily as boundary conditions. In problems with large conductors, with long conductive paths, and with poor initial conditions, such iterative methods can be slow or can stall (as noted by the "ENERGY STABLE BUT NOT BALANCED" caution message) despite the accelerations attempted via EXTLIM and ITERXT. As an alternative, all the nodes in a submodel can be solved simultaneously using a sparse matrix solver. Although convergence of conductance-dominated problems is normally tremendously accelerated by this approach, the cost per iteration is higher than with iterative methods and several iterations (each using a simultaneous solution in FASTIC and STDSTL) are still required to adjust for radiation, temperature-dependent conductivities and sources, and other user logic manipulations.

Miscellaneous Improvements

•  

  The propagation of spreadsheet (register and processor variable) changes is now performed automatically via internal calls to UPREG. These automatic updates can be regulated, supplemented, or even globally turned off using the NOUPDATE command in OPTIONS DATA.

• With the advent of ifaces, the stability and accuracy of nonequilibrium (NONEQ) options have been greatly improved. Also, the BELACC option has been obsoleted by the introduction of SPRING ifaces. BELACC is available but undocumented. 
  
  
• A new phase change material (PCM) simulation routine, FUSION, is now available for modeling latent energy within a diffusion node.
  
  
• An improved method has been added for predicting the density of liquid mixtures: the assumption of additive volumes has been supplanted with the modified Hankinson-Brobst-Thomson method. Therefore, the PVF parameter has been globally replaced with MF, the molar fraction (a real, floating point variable).
  
  
• Negative values of IPDC may be used to impose a user-specified flow regime (turning off the automatic selection available when IPDC=6).
  
  
• Pressure regulating values and back-pressure regulators are now available as PRVLV and BPRVLV connectors.
  
  
• NEADCC has been obsoleted. The code always adds an extra constituent equation to assure accurate tracking of trace species.
  
  
• As a convenience in switching between junctions and tanks, junctions now have volumes ("VOL"). These volumes are ignored by most program options, important exceptions being the SUMFLO routine and several gas dissolution/evolution options.
  
  
• A routine exits for estimating irrecoverable (K-factor) losses in reducers and expanders: FKCONE.
  
  
• Two new SINDA output routines are available: HRPRINT for printing heat rates through conductors, and NODTAB for tabulation-style summaries of nodal attributes including energy imbalances.
  
  
• CONSTRAINT DATA, part of the Solver, now accepts "unnamed" constraints to simplify the implementation of common constraints. For example, the user can specify "sub1.T15 <= 200.0" directly in CONSTRAINT DATA rather than first defining a constraint variable (e.g., "THEAT <= 200.0") and then updating that constraint variable later in logic blocks (e.g., THEAT = sub1.T15).
  
  
• CONTRN, NODTRN, INTCON, and INTNOD dynamic translation routines for nodes and conductors have been rewritten to execute much faster, especially for repeat calls. This enhancement is transparent to the user.
  
  
• Sparse matrix inversion underlying FLUINT solutions (and now optionally applied to SINDA solutions as well) have been expanded to include an ordering optimization which reduces both run times and memory requirements. This enhancement is transparent to the user.
  
  
• The model size limits (number of nodes, etc.) have been increased to 100000 nodes and 500000 conductors.
  
  
• A new INSERT option, analogous to the INCLUDE option, can be nested: INSERTed files can themselves include Inserted files.
  
  
• A new QFLOW routine allows energy flows between groups of nodes (including whole submodels) to be calculated.

SINDA/FLUINT Version 4.0

Thermal modelers take note! Version 4.0 introduces powerful functionality not just to FLUINT but to all aspects of SINDA usage. Understanding these improvements will change not only the way you use SINDA, but also the problems to which you apply it.

Dynamic Registers

Version 4.0 completes the integration of spreadsheet functionality into SINDA/FLUINT by enabling changes in the values of registers (spreadsheet variables) to propagate through a model even during processor execution. This enables you to make sweeping model changes with minimal effort, greatly reducing the need for user logic and rendering many traditional SINDA techniques obsolete. Parametric analyses and sensitivity studies are now quite easy to perform. The user has tremendous flexibility over when and what changes are made to the model when changes are propagated.

If you are unfamiliar with Registers, additional information is available in the release notes of SINDA/FLUINT Version 3.2, below.

Version 4.0 enables the analyst to perform parametric variations within a single run with minimal user logic through the use of dynamic registers.

In Version 4.0, the processor remembers every place that a register was used in the definition of a model, and knows how to propagate changes to registers throughout the model while the solution is proceeding.

Registers can be used in the definition of nodes, conductors, source terms, fluid system variables, and even array data. Almost every data value in SINDA/FLUINT can be defined as an algebraic expression, and in Version 4.0 sweeping model changes can be made with just a few lines of logic.

Registers themselves can be changed by providing a new expression instead of a fixed value, and registers can be updated independently from the model if desired. There are many traditional features in SINDA/FLUINT that attempt to provide such functionality, such as the use of number user constants in SIV nodes/conductors and source data. However, they do so in a relatively clumsy and confusing fashion compared to the new register-based methods.

For example, if a register named "DIAMETER" were used throughout a model to signify a key dimension, then changing that value throughout the model (conductances, inner diameters, flow areas, etc. etc.) is as easy as:

DIAMETER = 0.375
CALL UPREG $ (Note: UPREG is optional in Version 4.1)

The Solver

Not only does the user have the ability to make sweeping changes of the model while it is executing, in Version 4.0 the code itself can be tasked to perform those changes automatically, perhaps to achieve a special design goal or simulation need.

The most significant enhancement is the introduction of a new high-level solution routine, the Solver, with which you can direct SINDA/FLUINT to change selected registers as needed to achieve some modeling purpose.

Dynamic registers provide a convenient means by which the user can manipulate an entire model using a few central parameters such as key dimensions, properties, etc. SINDA/FLUINT may now be used for

1. Goal-seek. SINDA/FLUINT has traditionally served as a point design simulator. Given X, SINDA/FLUINT returns Y. Using the goal seeking capabilities, the user can easily invert the problem and have SINDA/FLUINT return X given Y. For example, the user might want the outlet temperature of a heat exchanger to be a certain value, SINDA/FLUINT will adjust the design (perhaps flow rates, dimensions, or materials, etc.) subject to arbitrarily complicated constraints, until the desired value is achieved.
   
2. Optimize. Instead of just matching a single variable, the Solver can be used to perform design optimization of an arbitrary number of design variables and constraints. For example, it can be used to find the cross section of a fin that maximizes fin efficiency while not exceeding a weight threshold. Using these features, SINDA/FLUINT can participate even in the very early stages of a design, or can be used to refine an existing design.
   
3. Correlate to Test Data. Just as a design can be optimized to suit a particular purpose, the model of that design can be optimized to best fit steady-state or transient test data. As often happens in thermal/fluid systems, key uncertainties such as contact conductances, surface optical properties, and natural convection coefficients can dominate predictions. Thus, correlation of the model against a few tests is required, making the model an intelligent extrapolator of known test data. This time- consuming task can be automated in Version 4.0, yielding best-fit models (perhaps defined by least-squares fit, or minimized maximum error, etc.) as well as insight into the magnitude of uncertain values.

The Solver works at a higher level than simply calling for a transient or steady-state run. In essence, using the Solver can program a series of SINDA/FLUINT runs as needed to automatically perform some modeling task.

In many ways, the Solver represents a convenient means of setting up a series of SINDA/FLUINT runs to perform a high-level task or to achieve a complicated analysis goal.

Traditionally, SINDA/FLUINT was a point design analysis tool, now it determines the inputs for a desired result. For example, in the past, SINDA provided the temperature of a node based on the input dimensions and surface properties. Now, SINDA will determine the dimensions and surface properties that will result in the desired temperature. Or FLUINT will determine the mass flow rate of a pump to provide the required downstream temperature in a fluid system.

Additional information is available on the Solver in the form of training notes.

Species-Specific Suction

Users can now specify which species should be extracted from tanks as needed to simulate various mass transfer processes or perhaps chemical reactions.

Miscellaneous Improvements

Other improvements include (1) new nodal convergence parameters and algorithms for faster steady-state executions, such as automatically calculated over- and under-relaxation factors; (2) an option to neglect phase change within throats for choked flow modeling; (3) modeling of trace species; (4) negative K-factors; (5) improved convergence in the STUBE connectors; (6) assistance modeling loop heat pipes (LHPs); (8) additional auxiliary routines, including monitoring heat flow through a conductor; and new function-based dynamic translation routines for thermal submodels.

SINDA/FLUINT, the popular and powerful network style thermal/fluid analyzer, has now become even more powerful. This subsection describes the new features in Version 3.2

Registers and Expressions

In past years, some users employed spreadsheets as model generators for SINDA, exploiting the ability to define a model algebraically such that a few changes would propagate consistently throughout the input file. When it was first introduced in 1992, SinapsPlus contained such spreadsheet-like expressions and user variables (registers), and these quickly became one of its most popular features.

Version 3.2 of SINDA/FLUINT now accepts registers and, equally important, complex expressions used to define data. A new Header Register Data block has been added for this purpose. For example, in past versions, to define the radial conductance in a tube from node 1 to node 2, one might input (given a conductivity of 15, a length of 10, an outer radius of 0.05, and an inner radius that is half that value):

       1,1,2,2.*3.14*15.*10./0.69

which leaves little or no self-documentation for engineers who inherit the model. Changes to the dimensions must be made manually in all places using them.

In Version 3.2, one can define up front:

HEADER REGISTER DATA
       con = 15.0
       len = 10.0
       Ro = 0.05
       Ri = 0.5*Ro

and then define the conductor indirectly:

1,1,2,2*pi*con*Len/ln(Ro/RI)

Notice the use of parentheses, of built-in constants like "pi," and of built-in functions like natural logarithms "ln( )." While the ability to use arbitrarily complex expressions is a significant improvement, the real power of registers and expressions is that they can be used throughout the input file in almost any location where SINDA/FLUINT requires a data value. This means that a key dimension (perhaps "Ro" in the above example) can be changed in one location (the Header Register Data block), and the effects will propagate consistently throughout the input file according to the relationships the user has predefined. As illustrated above, registers can themselves be defined by complex expressions that perhaps even contain references to other registers.

Registers and expressions not only greatly improve the self-documentation of the model, they facilitate long term model maintenance and the generation of parametric analyses. In the next version, the ability to perform parametric analyses will be taken a major step further with the advent of dynamic registers that can change during the solution, effecting widespread changes to the model with minimal user commands.

Condensible/Volatile Mixtures

Working fluids mixtures, previously limited to nonvolatile liquids and/or noncondensible gases, can now include a species that can either change phase or that is a real (imperfect) gas. This significant enhancement covers important applications such as psychrometrics (air/water mixtures) in heating, ventilation, and air conditioning (HVAC) systems, air/fuel systems with volatile fuels, pressurized liquid propellant and fire retardant delivery systems, as well as noncondensible gas and oil contamination in steam systems, two-phase coolant loops, and vapor compression cycles. When a substance condenses in the presence of noncondensible gases (e.g., a condensing heat exchanger) the effects of the diffusion barrier formation are automatically included along with other thermodynamic effects. In future versions, the dissolution and evolution of noncondensibles in liquid will be available in prepackaged form, as will complete nonequilibrium two-phase duct flow simulations.

Other Enhancements

Among the miscellaneous improvements made to SINDA/FLUINT is a new option for calling VARIABLES 1 within each FWDBCK time step iteration. This option not only provides higher fidelity for temperature-dependent variations and logic, it makes most logic written for steady-state analyses more reusable in transients.

Other improvements include a variable node solution order that reduces steady-state run times, and additional fluid property definitions (e.g., molecular weights, diffusion coefficients).