Cvs_US and Cvs_UK in BasicControlValve and SpecificValveType

[RaphaelGebhart]

Deprecation of Cvs_US and Cvs_UK in ThermofluidStream

Background

In ThermofluidStream.FlowControl.BasicControlValve and ThermofluidStream.FlowControl.SpecificValveType, the main flow coefficient parameter is Kvs (SI-compliant). There are also optional legacy parameters:

• Cvs_US - US customary units (gpm at 1 psi)
• Cvs_UK - UK imperial units (UK gallons per minute at 1 psi)

Issues

1. Non-SI units: Both Cvs_US and Cvs_UK are not SI-compliant, which conflicts with the rest of the Modelica standard library.
2. Conversion required:
   • Kv ≈ 0.865 · Cv_US
   • Kv ≈ 0.72 · C_uk
     This adds unnecessary complexity and potential for user error.
3. Legacy usage: Cvs_UK is a historical unit rarely used today. Cvs_US is also less relevant for SI-based libraries and global applications.

Recommendation

• Deprecate both Cvs_US and Cvs_UK for v1.3. The user can set them but this will create a warning, noting that they will be removed in v2.0, encouraging the users to switch to Kvs for SI-consistent modeling.

This approach simplifies the API, ensures consistency, and aligns with Modelica and IEC/EN standards.

[RaphaelGebhart]

Do you approve? @nieweber @fvanderlinden @CorentinLepais @dzimmer

[MartinOtter]

Please, take into account the docu about valve parametrization from @casella in Modelica.Fluid.UsersGuide.ComponentDefinition.ValveCharacteristics.
His conclusion is (please read the full documentation):

The pragmatic approach used in Modelica.Fluid.ControlValves is to accept the fact that m3/h and USG/min are not the real units of Cv and Kv, so we can't use the general unit conversion mechanism, put them just as mnemonic labels in the comment, use non-dimensional coefficients in the interface, and then define properly dimensioned unit conversion within the model

Please, also look at valve catalogs how Cv and Kv are defined there - and make it not unnecessarily error prone for the end-user (best that an end-user can directly use catalog values).

[RaphaelGebhart]

Thank you for the remark. I briefly checked the literature, and in the datasheets I reviewed, Kv is commonly given as unitless. Therefore, I am in favor of the approach presented in Modelica.Fluid.UsersGuide.ComponentDefinition.ValveCharacteristics:

• Kv is unitless (a number).

• Kv is defined based on the mixed “Zahlenwertgleichung - Größengleichung” as stated in the User Guide (where e_q is the unit of q and e_dp is the unit of dp): ***

   q = Kv * sqrt(dp / (rho / rho0)),   with e_q = m3/h, e_dp = bar
  

• Compute the effective area Av in m2 based on Kv and use it for the "Größengleichung" as stated in the User Guide (the TFS requires the inverse version).

   dp = m_flow^2/rho/Av^2
  

• In the other case calculate Av from m_flow_ref, dp_ref, rho_ref as:

   Av = m_flow_ref/sqrt(rho_ref*dp_ref)
  

• Datasheets that give Kv in m3/h are incorrect

One alternative that would allow unit conversion is to convert the above “Zahlenwertgleichung - Größengleichung” into a real "Größengleichung" by introducing a reference pressure drop, similar to what is done in the TFS library:

    q = Kv * sqrt(dp / dp0 / (rho / rho0))

In this case q, Kv, dp, dp0, rho, rho0 are physical quantities not numbers and Kv = q(dp = dp0). Personally, I prefer this second version. However, it seems that the “Zahlenwertgleichung - Größengleichung” is the one used in practice.

@casella, do you suggest keeping separate Cvs_UK and Cvs_US values, or should we only use a single Cvs?
Based on my quick datasheet search I would still use Kvs, Cvs instead of Kv, Cv.

***Remark: It is really weird, because rho and rho0 are physical quantities but q, Kv, dp are numbers.

[casella]

I usually refer to the IEC 60534-2.1 norm. This says:

with

There are units for pressures and flows, but not for the flow coefficient $C$ nor for the constant $N_1$

Assuming for simplicity that the coefficient $F_r = 1$ and $F_p = 1$ (they're non-dimensional) the ur-formula in good old SI units, which is derived from Bernoulli's law, would be:

$Q = A_v \sqrt {\frac{\Delta p}{\rho}}$

where $Q$ is a volume flow in m3/s, $A_v$ is an area in m2, $\Delta p$ is a pressure in Pa and $\rho$ is the density in kg/m3. The engineering law in US units is instead

$Q = C_v \sqrt {\frac{\Delta p}{\rho / \rho_0}}$

which uses the relative density $\rho/\rho_0$ instead of the absolute one, so it is actually a different law. This also implies that $A_v$ and $C_v$ are actually different quantites (besides the choice of units), so it's not just a choice of units, but it's more compicated.

In the case of the SI version, if you don't care about being rigorous with units because there is a magic coefficient $N_1$ that fixes everything, since $\rho_0 = 1000$ kg/m3, the division by $1/\rho_0$ corresponds to the multiplication by 1000, so it actually converts the kPa into Pa, so $N_1$ must just convert m3/s into m3/h by multiplying by 3600. So you can still use the US-law with the relative density, as long as you use the correct $N_1$ magic factor. Urgh.

The whole thing is a rather bloody mess, as usual when non-SI units are employed.

In Modelica.Fluid, I decided to write the physical law in a way that is dimensionally consistent, so if $A_v$ is an area, then you need to use the absolute density, not the relative one. Then, the conversion from $C_v$ and $K_v$ to $A_v$ are just done by magic numerical coefficients.

[casella]

BTW, the IEC 60534 norm has $K_v$ and $A_v$, not $K_{vs}$ and $A_{vs}$, and there is no UK vs. US units

[casella]

I guess the TFS code refers some VDI/VDE norms, perhaps?