root/branches/maintenance/2.2.1/Modelica/Media/req_changes.txt

REQUESTED CHANGES TO MODELICA.MEDIA
0.
Add package constants for all physical data that is needed by physical
property estimation routines, and add to NASA data were available.
Initial list:
Molar Masses
Critical Temperature, Pressure and Volume/Density
Accentric factor
Dipole Moments
? Association factors

scalar for pure substance, vector otherwise

1.
Remove references to "gas" from Interfaces.PartialMixtureMedium
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Date fixed: 2005-01-14 by: Hubertus
---

2.
IdealGases.Common.SingleGas should probably be renamed to
IdealGases.Common.SingleGasNasa for uniformity reasons
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Date fixed: 2005-01-14 by: Hubertus
---

3.
Interfaces.TemplateMedium should be updated with the last changes
(add R=0 to BaseProperties, new nX and reducedX, etc.)
It should be tested by creating a new, simple medium model from scratch
just before the release of the library
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Assigned to: Francesco
Date fixed: None by: None
---

4.
References to Prandtl and Brigdman are sometimes misspelled throughout
the library
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Date fixed: 2005-01-14 by: Hubertus
---

5.
A final check of all the "restricted" WaterIF97 models should be performed,
some of them, though restricted, do not perform better than the base model,
and should then probably be removed.
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Assigned to: Francesco (keep me posted, Hubertus)
Date fixed: None by: None
---

6.
Add WaterIF97_ph models, restricted to Region 1 and to Region 2, to
allow for faster performance.
---
Date submitted: 13 Jan 2005 by: Francesco Casella
Assigned to: Francesco Casella (?)
Date fixed: None by: None
---

7.
The unit type used for the Volumetric coefficient of thermal expansion
[Bejan], a.k.a. volume expansivity or beta coefficient, is called
RelativePressureCoefficient. Although the actual unit is OK (1/K),
this has nothing to do with pressure, so the name should be changed.
----
Date submitted: 13 Jan 2005 by: Francesco Casella
Date fixed: 2005-01-14 by: Hubertus
---

8.
Function Interfaces.PartialMixtureMedium.density_derX requires the molar
mass of the mixture as an input. Although this can be useful for ideal gas
mixtures, it is probably of no use, e.g., for liquid mixtures. Therefore,
it is probably better to remove it as an input, and to add it to the
ThermodynamicState vector.
Comment: I dissagree on that issue: the ThermodynamicState vector
should be minimal if possible. But: this shows the need that a lot
more package constants are needed, molar masses, critical data and the like.

I fixed it a little differently: a new function mixtureMolaMass that
takes state as input. It works for ideal gas mixtures.

Name later changed to molarMass, since it also applies to pure media.
----
Date submitted: 13 Jan 2005 by: Francesco Casella
Assigned to: Hubertus
Testing assigned to: Francesco
Date fixed: 2005-01-14 by: Hubertus
Date fixed: 2005-02-17 by: Francesco
---

9.
1) Nasa vs. NASA: this is consistent: as part of a function or class
name nameNasa, if by itself NASA
* The spelling should be "discontinuous" and "monatomic" in the documentation of SingleGas that currently reads "A few of the data sets for monoatomic gases have a discontinous 1st derivative at 1000K, but this never caused problems so far."
[The first I saw myself - the second I detected when using spell-checking in word, and Webster confirmed it.]
* The Media-model should be SingleGas (singular) as you wrote in req_changes.txt.

Date submitted: 13 Jan 2005 by: Hans
Date fixed: 2005-01-14 by: Hubertus

10.
 Renaming of cp and cv
Date fixed: 2005-01-16 by: Hubertus
---

11.
Modelica_Media.Interfaces.PartialMixtureMedium.ThermodynamicState already contains p,T,X.
Although fairly sophisticated models would probably just need those
variables, an incompressible fluid model (e.g.) won't need the pressure
p by definition. I would thus remove those declarations, and leave to
each single medium model the task to declare its own state variables, as
it's done with media extending PartialMedium.
---
Date submitted: 17 Jan 2005 by: Francesco Casella
Assigned to: None 
Testing assigned to: None
Date fixed:  by: None
---

12. 
Modelica_Media.Interfaces.PartialMixtureMedium.BaseProperties.MM
This is surely needed for ideal gas mixtures, but maybe not for other
models (e.g. for liquid mixtures). Same cure, remove from here and add
to IdealGas.MixtureGasNasa.* models.
---
Date submitted: 17 Jan 2005 by: Francesco Casella
Assigned to: None 
Testing assigned to: None
Date fixed:  by: None
---

13.
Modelica_Media.Interfaces.PartialMixtureMedium.BaseProperties.h_component[]
Externally supplied media models (e.g. foreign C models) could supply the total 
mixture enthalpy h, but not the componentwise enthalpy. It is of course possible 
to set them to zero, but maybe it's better to remove this from the standard 
interface and to add it to gas mixture models as well.
---
Date submitted: 17 Jan 2005 by: Francesco Casella
Assigned to: None 
Testing assigned to: None
Date fixed:  by: None
---

14.
Component molar masses in current IdealGases NASA models are contained 
in the package "data" record. Unfortunately, this datum, which can be 
important when dealing with chemical reactions, is not included in any 
generic medium interfaces. I'd suggest adding it to the PartialMixtureMedium
interface.
---
Date submitted: 17 Jan 2005 by: Francesco Casella
Assigned to: None 
Testing assigned to: None
Date fixed:  by: None
---


15.
o Dimension of nX is number of substances (nX=1 for one substance)
o Changed size(substanceNames,1) everywhere to nX
o Changed p_reference to reference_p
  Changed referenceX  to reference_X
o Changed SimpleMedium to PartialSimpleMedium
o Introduced "MM_const" in PartialSimpleMedium
o Set MM_const=18.016e-3 for ConstantPropertyLiquidWater
---
Date submitted: 21 Jan 2005 by: Martin Otter
Assigned to: None 
Testing assigned to: None
Date fixed:  by: Martin Otter


16) Fixing viscosity and thermal conductivity for gases and consistent
addition of data as it becomes available:

Current situation:
The currently implemented functions for ideal gas viscosity and
thermal conductivity don't follow the standard for all other functions
that have the state record as input. They can't: they require data
that is not yet present in Modelica.Media.

In order to fix the situation, more data is needed, and if the input
convention is to be kept, the data has to be present in the form of
package constants.

The following constants are currently declared in the Media package:

In PartialMedium
    constant String mediumName "Name of the medium";
    constant String substanceNames[:]= fill("",0) "Names of the mixture substances. Set substanceNames=fill(\"\",0) if only one substance.";
    constant String extraPropertiesNames[:]= fill("",0) "Names of ... Set extraPropertiesNames=fill(\"\",0) if unused";
    constant Boolean incompressible "= true, if medium is incompressible (i.e., density = constant)";
    constant Integer nX =  if size(substanceNames, 1) > 1 then size(substanceNames, 1) else 0;
    constant MassFraction referenceX[nX] "Default composition of medium"; 
    constant Integer nC =  size(extraPropertiesNames,1) ;
    constant AbsolutePressure p_reference = 101325 "reference pressure of Medium: default 1 atmosphere";

In PartialPureSubstance:
    constant String iupacName "complete IUPAC name";
    constant String casRegistryNumber "chemical abstracts sequencing number";
    constant String chemicalFormula "Chemical formula, (brutto, nomenclature according to Hill";
    constant String structureFormula "Chemical structure formula";


The following additional constants are needed to compute viscosity and
thermal conductivity:

criticalPressure
criticalTemperature
criticalDensity (or criticalVolume)
molarMass
dipoleMoment

They could be in a record for each pure component, and a vector of
records for each mixture, parallel to the heat capacity data from
NASA. Most of the above are already defined the the record
FluidConstats for two phase media, which would instead be "redeclare
extends" from the basic one.

Except for the molar mass, none of these is available in the NASA
data. The NASA data comes with extra coefficients for viscosity in an
extra file, but for much fewer components, and complex to use in
mixtures due to separate "interaction parameters" for each pair of
gases in a mixture. I have acess to a few hundred sets of
critical data, but far less than the NASA-coefficients, and not even a
proper subset.

This highlights particular problems with medium data:
1) Data has to be obtained and used from different sources.
2) The situation changes yearly.
3) Users need a simple means to take some data from Modelica.Media,
and add their own when they have better data (subscription to a
particularly "good" property databases comes at 36 000 Euro/year

Obviously the best solution would be to completely separate the data
from the model, and query the data on demand from a database (unless
you run embedded of course). 

A solution that takes this situation into account should at least do
the following:
1) Each "data-source" that is used in Modelica.Media should have a
name (e.g. the NASA-data comes from 1 public file that is widely used,
the critical data that I have comes from a book, etc.)
2) Each medium package should have a set of Booleans (a data
structure) that lets a user query what data is available, and from
what source it comes. This data structure can grow as we get access to
more data. See a suggestion that covers the current status below.


record DataSources "data sources"
constant Boolean hasIdealGasHeatCapacity=false "true if ideal gas heat capacity is available";
constant String idealGasHeatCapacitySource="none" "data source for ideal gas heat capacity coefficients";
constant Boolean hasCriticalPressure=false "true if critical pressure is known";
constant Boolean hasCriticalTemperature=false "true if critical temperature is known";
constant Boolean hasCriticalVolume=false "true if critical volume is known";
constant String criticalDataSource="none" "source for critical data";
constant Boolean hasFundamentalEquation=false "true if a fundamental
equation";
constant String fundamentalEquationSource="none" "source of the fundamental
equation model";
constant Boolean hasLiquidHeatCapacity=false "true if liquid heat capacity is available";
constant Boolean hasSolidHeatCapacity=false "true if solid heat capacity is available";
constant Boolean hasAccurateViscosityData "true if accurate data for
a viscosity function is available";
constant Boolean hasAccurateConductivityData "true if accurate data for
a thermal conductivity function is available";
end DataSources;

For some of the NASA data, there are heat capacity coefficients for
the solid state too.  Molar Mass is always known, and could be a
separate package constant that has to be defined always.

That data record should be part of every pure medium data set, or
medium package if we don't separate it. Mixture models need a way to
query what data is available for all of the mixture constituents and
based on that, the best possible method for computing properties is
selected. In the current situation that woudl mean: only when critical
data and dipole moments of all species in a mixture are known can the
viscosity be computed. 

Now that there are a number of string parsing functions in
Modelica_Utilities, the critical data could be stored on file as a
primitive database and parsed: no changes to the medium models are
necessary as we get access to more data.

The goal is, that this could even work with external data, using a
user-supplied query function. The casRegistryNumber is the simplest
generally accepted identifier for a substance. 

I would at least like to be able to compute viscosities and thermal
conductivity for the most important technical gases, i.e. combustion
poducts, natural gas mixtures, gaseous fuels, refrigerants, ... for the 1.0
release. 

Date submitted: 27 Jan 2005 by: Hubertus


======================================================================
Martin Feb. 22 2005:


Results of Modelica_Media.Examples.Tests

- Models that are satisfactorily (they simulate, 
  static state selection and no non-linear system of equations)
    Modelica_Media.Examples.Tests.MediaTestModels.Air.SimpleAir
    Modelica_Media.Examples.Tests.MediaTestModels.IdealGases.SimpleNaturalGas
    Modelica_Media.Examples.Tests.MediaTestModels.Water.ConstantPropertyLiquidWater
    Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97OnePhase_ph
    Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_ph


- Issues to be fixed by Dynasim (Sven Erik/Hans):
  o Modelica_Media.Examples.Tests.MediaTestModels.Air.DryAirNasa
    Modelica_Media.Examples.Tests.MediaTestModels.IdealGases.Air
    Modelica_Media.Examples.Tests.MediaTestModels.Water.IdealSteam
      two non-linear systems of equations of size 1 each
      (the same reason for all media)

  o Modelica_Media.Examples.Tests.MediaTestModels.Incompressible.Glycol47
    Modelica_Media.Examples.Tests.MediaTestModels.Incompressible.Essotherm650
      two non-linear systems of equations of size 1 each
      (the same reason for both media)

  o Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_pT
      two non-linear systems of equations of size 1 each
 
  o Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_dT
      three non-linear systems of equations of size {1,2,2}

  o Modelica_Media.Examples.Tests.MediaTestModels.Air.MoistAir
    When using a function in MoistAir
        h = h_pTX(p,T,X_i);
    instead of
        h = h_component *{X_steam, X_air} + enthalpyOfLiquid(T)*X_liquid;
    Dymola does not translate
        Error for function: Modelica_Media.Air.MoistAir.h_pTX:der.
    This was already reported to Hans.
    When replacing the function by the equation, the test model
    translates and simulates. There are two non-linear systems of
    equations of size 1 each (this is understandable, since h is not
    computed via a function).


- Issues to be fixed by Hubertus and/or Jonas:
  o Modelica_Media.Examples.Tests.MediaTestModels.Incompressible.Glycol47
    Modelica_Media.Examples.Tests.MediaTestModels.Incompressible.Essotherm650
      Simulation fails (wrong initial conditions?)

    The table based media have the general problem that "u" is a function
    of "pressure". Therefore, there will be two states "p,T", two
    differential equations and two initial conditions. The flag
    "incompressible" was introduced to define whether an initial condition
    for "p" or "d" has to be provided. This means "incompressible" has to
    be set to "false", since 2 initial conditions are needed. I have
    changed the flag from "true" to "false". However, then the general
    name for the package "Incompressible" is odd.

  o Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_pT
    Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_dT
      Simulation fails. This is a bit strange, because simulation with
      Modelica_Media.Examples.Tests.MediaTestModels.Water.WaterIF97_ph
      works fine.

  o PartialMixtureMedium has "p" as ThermodynamicState.
    This is not true for incompressible mixtures.
    Probably best to remove redeclaration of ThermodynamicState