Proposal : Convertion b/w Pressure Measurement References

[gratestas]

Hi,
I have thought whether conversion feature of one pressure measurement reference to another (absolute, gauge, vacuum, barometric) will fit the UnitsNet's standards and enrich it.

Regarding the matter, it is widely used feature in fluid dynamics. So, in case it is valued as worth to think about, I'd like to discuss the situation in details.

[gojanpaolo]

I believe this will be a good addition to the library. What will be your specific use cases for this feature?

For the implementation, I'm thinking this will be a new quantity class (e.g. ReferencePressure or something like that) since the calculations to convert between the different pressure references will not really fit inside the Pressure class.

If we need to convert the two quantities to each other, we can edit their CustomCode implementations. Or just write extension methods?

[angularsen]

Sounds good to me, please briefly describe:

• The motivation for this change
• Example conversions / usecases
• Pseudo code implementation or bullet points to highlight the changes needed

[angularsen]

@gratestas Waiting for your input to continue this discussion.

[gratestas]

Only for the sake of avoiding misunderstandings, I go into the details:

Pressure is a state function, for its measurement depends on environmental factors such as ambient pressure, elevation above sea level and local weather conditions. For two persons located in the different environment to speak of the same pressure, one must relate it to a reference. There are three basis reference: absolute, gauge, vacuum.

• Absolute is zero-referenced to the total vacuum.
• Gauge refers to a level of the local atmospheric pressure.
• Vacuum is the negative of the gauge.

Therefore, to obtain consistent and qualitative data, the reference measurement is crucial.

Here it is why. Tanks, pumps, compressors and turbines are sensitive to pressure changes and the choice upon the relative measurement reference.

Tanks/containers contain pressurised fluids surface of which is referenced to atmospheric pressure.

• The implication is that it is much more convenient and sufficient to work with maximum two-digits gauges. The gauge will start reading from 0 atm. Reservoirs are just the same except that it is open to the atmosphere and its surface will always be 0 atm relative to gauge.

• Another aspect to consider is the fact that during the year, the atmospheric pressure of the specific location changes enough to damage a working system.

  • So if it is desired to isolate internal process of a system from the external pressure disturbances, absolute reference is the preference.
  • However, if a certain pressure level needs to be instanteneously maintained relative to atmospheric pressure, gauge reference is better choice.

Pumps, though, have a different story. When you have a tank communicating with a pump you better go for absolute reference since the pump's performance depends on the differential pressure between the suction and discharge nozzles. Suction nozzle pulls fluid to the vacuum (negative) pressure, whereas discharge nozzle adds positive pressure to the fluid.

• When you take differential, you subtract the suction from discharge pressure. This makes life easier when both pressures are on the positive scale.
• Another point is about mathematical simplification of calculations. Without going into math, the absence of atmospheric pressure term(as the fact that it is variable too) in an equation saves you from getting too complicated differential equation and decrease cost/iteration.

Simply put, a particular system is more convenient and efficient to design relative to one reference, whereas another system would be more manageable under the other one. Such implementation will bring less confusion, less effort, more effectiveness, more power.

[gratestas]

Conversion examples as follows:

Absolute = Atmospheric + Gauge
Absolute = Atmospheric - Vacuum
Gauge = Absolute - Atmospheric
Vacuum = Atmpospheric - Absolute

Here local atmospheric pressure will be always passed parameter.

[gratestas]

I was thinking that the implementation can be viewed, at least, by two ways. One of them is to create a quantity for each type like Absolute/Gauge/Vacuum-Pressure and introduce convenient extension methods to swap between them. The other one is to create single quantity, as @gojanpaolo proposed, like ReferencePressure and then to define corresponding units projected from each of pressure units.

For instance, for PoundPerSquaredInch (psi)
(psia) -> PoundPerSquaredInchAbsolute
(psig) -> PoundPerSquaredInchGauge
(psiv) -> PoundPerSquaredInchVacuum

and the same pattern applies for remained units. In my opinion, the last one seems more reasonable. In fact, it preserves consistency with the regular convention in the industry which is good.

Pseudo code implementation example

public static ReferencePressure ToGauge( this ReferencePressure absolute, Pressure localAmbient )
        {
            return ReferencePressure.ToGauge( absolute, localAmbient);
        }

 public static ReferencePressure ToGauge( ReferencePressure reference, Pressure localAmbient )
        {
            return ReferencePressure.FromPascalGauge(reference.PascalAbsolute - localAmbient.Pascal);
        }

Problem
However, I'd like you to pay attention to the fact that if someone would like to pass input of ReferencePressure type, there is no way to set the unique rule for interchanging Absolute -> Gauge or Absolute->Vacuum since it includes all of them. At least I cannot figure it out. Maybe you will.

Instead
I thought, then, about Pressure quantity class itself and suggested that maybe all of three reference pressures can be put under it and put through conversion by extension methi. The units of Pressure will be accommodated like
Pascal
PascalAbsolute
PascalGauge
PascalVacuum
and the same for the others. Yet it will triple the list. Further, the conversion of references may be done on the level of normal conversion (I am not acquainted with the UnitsNet structure very well yet). If I am not mistaken, the only difference will be in that here there will be two pass parameters. For instance,

    var pressure = Pressure.FromPascalAbsolute(value, localAmbient);
    var convertedPressure = pressure.PascalGauge;

that gives main idea something like:

public static FromPascalAbsolute(double pressure, Pressure localAmbient)
        {
              _baseUnit = pressure ;
              _localAtmosphere = localAmbient;
            
             return new Pressure( _baseUnit, PressureUnit.PascalAbsolute)
        }

public static ToPascalGauge( )
        {
              _gauge  =  _baseUnit - _localAtmosphere;
             
             return new Pressure( _gauge, PressureUnit.PascalGauge)
        }

public static ToPascalVacuum( )
        {
              _vacuum  = localAtmosphere - _baseUnit;
             
             return new Pressure( _vacuum , PressureUnit.PascalVacuum)
        }

Again, I am not acquainted with the UnitsNet structure very well, but I will spare some time to understand it better to be able to implement it or at least to assist the implementation.

The bullet points here would be:

• the first parameter will be taken as the base reference unit and saved to the memory together with the local ambient value
• then at the level where conversion is being done conversion equations are used

[angularsen]

Awesome explanation, thanks.
I don't have enough time to digest it all right now, will have to spend more time reading it and getting familiar with the concepts.

Initial thoughts/understanding:

• What is currently missing is a way to store the reference pressure (50 PSI relative to what)
• Option A) Add PressureReference field to Pressure quantity, and when constructing pressure quantities specify the reference. Not compatible with current code generation scheme, I think it requires adding a special case. Also a breaking change unless we let the default reference value be say Gauge?
• Option B) Duplicate Pressure units to get PascalV (Vacuum) and PascalA (Absolute) variants, and either reusing existing units for Gauge or adding those too (PascalG). Compatible with code generation, but generates a lot of duplicate code.
• Option C) Add a wrapper like RefPressure

var p = new RefPressure(Pressure.FromPascals(1), PressureReference.Gauge);
Pressure pg = p.Gauge; // 1 Pa
Pressure pa = p.Absolute; // Something else
Pressure pv = p.Vacuum; // Something else

You won't get as nice syntax, but it saves a lot of code duplication.

Am I getting it right?

[angularsen]

Closing this due to inactivity, please reopen to continue the discussion.

[MrFoged]

I could also use this feature as i'm right now switching between absolute- and relative (gauge) pressure in a way that would have been better as a inbuilt feature in UnitsNet.
Hope you will look into this again some day.

When I work with pressure I always show using this notation:
BarG - (Gauge Pressure) Pressure reading relative to current atmospheric pressure.
BarA - (Absolute Pressure) Pressure reading relative to absolute vacuum.
The Vacuum pressure is a new concept for me (Been in the business of working with pressure in +5 years and have never even hear of it) Does people even use it anymore?

How should it be done?
Invent a special case for pressure or adding new units to the pressure unit - I dont know what would be easiest/best

[tmilnthorp]

Correct me if I'm wrong, but it seems to boil down to relative vs absolute pressures. Absolute is to absolute vacuum, relative is to some other pressure. That other pressure for gauge is usually 1 atm but not necessarily. What if we're on the moon?

I loathe the idea of creating a new unit for relative vs absolute. I would propose the following:

• Add a PressureType enum (Absolute and Relative)
• Add a Pressure ReferencePressure field
  • Could be absolute or relative itself!
• When creating the current pressures, assume Absolute with ReferencePressure = Pressure.Zero.
  • Relative with 1atm would be a bad assumption.
  • It would also break the current behavior with math operations
• Add a constructor to allow passing a reference temperature, and set PressureType to Relative
• Add a static Pressure.FromReference(Pressure referencePressure) method with default parameter of 1 atm
• Add ToAbsolute method
  • Absolute pressures return this
  • Relative pressures do a conversion
    • ReferencePressure.ToAbsolute could be chained n levels deep for this calculation.
• Update arithmetic operations
  • Would need a way to override arithmetic operators
    • Add protected virtual Add/Subtract/etc methods and forward arithmetic operators?

I use the term "reference" above, but Gauge could be substituted if we want.

[tmilnthorp]

Hmmm, I just tried to play with a bit of code and got stuck immediately.

• Adding a ReferencePressure member of type Pressure causes an error: Struct member 'Pressure.ReferencePressure' of type 'Pressure' causes a cycle in the struct layout

I tried deriving a new RefencePressure class from Pressure. I forgot that's also a no-go with structs.

I think for my suggestion we'd have to move to class instead of struct. Please. 😃

[angularsen]

Good design proposal above.

1. @tmilnthorp Do you still get the cyclic problem if you make the reference pressure nullable, like Pressure??

2. Is it an option to use a wrapper class type, which holds Pressure Value and Pressure ReferenceValue?

I suspect the struct vs class discussion is not getting resolved anytime soon, but yep, the arguments for class is starting to pile on by now.

[MrFoged]

That other pressure for gauge is usually 1 atm but not necessarily. What if we're on the moon?

On earth we have defined it to be exactly 1 atm or 101325 Pa so it is a constant when switching between absolute- and relative. In the real world it of cause is not constant but we dont have to worry about that.

In my experience engineers tent to use absolute pressure and people on the floor tent to use relative. Engineer wants to use a system that also work on the moon hence absolute pressure.
Workers want a system that shows '0' pressure when all the valves are open (They dont care about that the real pressure is ~1 atm)