Refuelling with compressible flows 1

[hydrogen_joe]

The attached graph plots real data from a h2 refueller that is heavily reduced by a needle valve. The system cascade fills based on the differential between a large store to a smaller vehicle tank. The needle valve gives a system resistance co ‘K’ of approx. 5500 depending on opening.
Data is plotted against theoretical compressible mass flow, assuming both adiabatic and isothermal flows. Due to the high flow rates, it is my belief that adiabatic is more realistic. See formula’s below, both from CRANE TP410, isothermal adjusted as per guidance. I manually choke the flow based on the upstream pressure (which drops during the fill, so difficult to consider it stagnant).

For adiabatic - Could anyone suggest why the higher the pressure differential the more the data deviates from the theoretical prediction, is there an additional adjustment factor? Darcy eq.
For isothermal – the cutoff to choked is more dramatic as expansion factor is not considered, considering Y smooths off the curve but shifts the curve down. Is there another adjustment factor? (1-28)

Lastly, does anyone know of more appropriate ways of tackling this problem?

 

[danschwind]

I have not given much thought, but if I had to guess where the problem is my first hyphotesis would be your valve's K value.

Are you able to match the real data flowrates with an equivalent Cv value that is lower than the 100% needle valve Cv? If that is the case, I reckon you might be considering the a wrong coefficient in your calculations.

Daniel
Rio de Janeiro - Brazil

 

[1503-44]

What do you get using this method?

https://www.emerson.com/documents/a...zing-standardized-method-fisher-en-140724.pdf

A black swan to a turkey is a white swan to the butcher ... and to Boeing.

 

[hydrogen_joe]

@danschwind - the K term is mainly dominated by the needle valve and yes I have a graph plotting 'turns open' against percentage of full cv for the needle valve. Considering a more open needle valve to match the needle valve is possible, but it doesn't match up with data from every fill. I currently convert Cv to K for the needle valve and components using its nominal bore, then add K values together to get a total.

@1503-44 - I will have a look into this.

 

[Latexman]

When the model does not predict reality, use reality to make the model.

This puts a lot of focus on the accuracy of the mass flow instrument(s), so double check those.

Good Luck,
Latexman

 

[danschwind]

Joe,

How much confidence you have on that graph? Are you able to validate the Cv value x % Open for that valve through an experiment?

And I fully agree with Latexman: double check the mass flowrate instrument you are using, just to be sure.

Daniel
Rio de Janeiro - Brazil