Gas pipeline capacity 2

[petroabbes1980]

Hello gents;
Does Anyone knows a methods for the calculation of a pipeline Maximum Flow capacity that it could transport, Data available: e.g 24 in, inlet P=40 barg, out P=20 Barg, lenght 100 Km?

Thanks

 

[LittleInch]

It's still a lot more than the ops 35mmscfd......

How was the"front end loaded average" worked out?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[SNORGY]

I suspect Equation #1-39 in Crane TP-410M (2011):

P'(avg) = 2/3*{[P'1^3-P'2^3]/[P'1^2-P'2^2]}

 

[zdas04]

No, the one I use (from GPSA) is:

P_avg=(2/3)(P₁+P₂-(P₁*P₂)/(P₁+P₂)

With just a bit of algebra (it is easy to find the steps if you know the end result) you can factor Crane into exactly the GPSA equation, but I find the GPSA equation to be easier to explain in class.



David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[georgeverghese]

Using the GPSA expression for average P, I get P average at 32bar abs, and at 16degC, z = 0.95 at this press from NBS tabulations for pure methane in Perry.
Including this z value in the isothermal expression, mass rate is 201 e3 kg/hr or 251.1mmscfd.

 

[SNORGY]

zdas04,

Interesting. The GPSA equation that you state is the one I ultimately used when I developed my spreadsheet tool years ago, and the one it still uses. Then I recently got a copy of the 2011 Crane TP410M and decided I'd look at it and - to be honest - came across #1-39 for the first time. It made me wonder if I had been doing it wrong all these years. As you state, a bit of algebra probably would have gotten me there. But in any event, as always, I learned something new in this thread, namely, the Hall-Yarborough correlation. I had never heard of it until today. When I built my spreadsheet tool (and several others) years ago, I had used four of the cubic equations of state in the back of GPSA to estimate Z (R-K, S-R-K, P-R and VDW) and I could toggle between whichever one I wanted. I usually use VDW to avoid having to worry about boiling point properties and acentric factors that would force me to dive into a lot more detail to accurately characterize the gas which, for what I need on a routine basis, isn't necessary for me. (I'm mechanical, not process, by background). It also aligns with my approach towards predicting J-T, which is premised on VDW. In any case, when I adjusted the input in my tool to more closely match the characterization you appear to have in your calculation, my compressibility factor (via Peng-Robinson) was Z = 0.879 and my flow rate ended up at about 247 MMSCFD; my assumed gas viscosity was 0.011 cP. So, at the end of the day, there does not appear to be much disagreement between you and georgeverghese and me.

Thanks for pointing me towards Hall-Yarborough. Never heard of it until today. Because of that, I'm just a little bit better today than I was yesterday.

 

[georgeverghese]

Snorgy, If z=0.88, flow remains at 251mmscfd using the isothermal compressible flow routine on my mathcad calc.

 

[katmar]

Should we be debating small changes in the compressibility factor (Z) when we don't know the composition of the gas and we are throwing a 95% efficiency at the problem? Nor have we defined the standard conditions.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[katmar]

ZDAS04, it is very interesting that the length of the segments affects the result. The Reynolds Number and friction factor should remain constant for the whole length of the pipe (assuming constant viscosity). If those are also varying with the segment size it may give you a pointer to the cause of the flow result variations.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[zdas04]

Katmar,
The conditions I assumed are normal for commodity pipeline gas in the U.S. (e.g., what is traded on the commodities market under the label "Henry Hub" or NYMEX Natural Gas). The pressure of this line is kind of intermediate between normal raw-gas pressures and some short-haul commodity pipeline gas so I guessed commodity. Assuming raw-gas just gives you a more complicated multi-phase flow environment that was way outside the OP's question.

For liquids, the Reynolds Number and friction factor remain constant, but for gas both density and viscosity are a function of pressure (I recalculated both for each segment) and the Reynolds Number changes considerably as you go down the pipe.

I've found that with new steel pipe the design calculations I've done over the years using a 95% efficiency factor have tended to match start-up conditions within about 2% of dP. Over time, the accumulation of liquids in the line required me to lower that number to match field conditions (which is the indicator I used to assess pigging schedules quarterly, a line that was getting worse needed to be pigged more often, a line that was above 90% could probably be pigged less often).

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[SNORGY]

georgeverghese,

I think that's simply because your model is probably better and more accurate than mine, or I might have entered something wrong.

katmar also raises a good point regarding standard conditions. I used the international standard of 288.15 K and 101.325 kPa.

Carefully re-checking my input and using better data for critical pressure and temperature for methane, my calculated Z (P-R) is now 0.92 and flow rate is 252.9 MMSCFD. Not sure what other input data was wrong on my end because I didn't save the spreadsheet with the last data I used, but now the result is getting closer to what georgeverghese is getting. I am not carrying anything in the way of an efficiency factor. My calculated friction factor is 0.0116, and I suspect that is where the slight variance comes from.

That said, I think zdas04's rationalization for using an efficiency factor is well-founded:

http://www.ewp.rpi.edu/hartford/~er...udents/Douglas/PipeFlowTutorial-Schroeder.pdf

 

[katmar]

The density of the gas will decrease along the length of the pipe, but the velocity increases in proportion to the reciprocal of the density and the product of density and velocity (basically the mass flowrate) remains constant. The decrease in density therefore has no impact on the Reynolds Number.

The data I found for the viscosity (Younglove and Ely) indicates a decrease in viscosity of about 4% in going from 40 bar to 20 bar. This would translate directly into a 4% change in Reynolds Number, but a less than 0.1% change in friction factor. This is a similar order of magnitude to the change in flowrate you calculated from the largest to smallest section length and I would suspect that the reason for the variation in your results is in the way you have averaged the viscosity for each section.

I have no quibble with applying a 95% efficiency. My point was that if we apply what is effectively a 5% safety factor we shouldn't fuss over 2% variations between different ways of calculating the flowrate. However the variation resulting from the change in section length is interesting because the same calculation method is being applied in all cases.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[zdas04]

Katmar,
I modified my program to output Reynolds Number and Fanning friction factor for the 10,000 step case. Reynolds Number changed by 4% (change is dominated by the change in compressibility since as you say density and velocity largely offset each other), resulting in a change in friction factor of 0.1%. In the 2 step case, Reynolds Number changed almost 10% and Friction Factor changed by nearly 1%.

My point is that the fewer factors that you assume to be constant, the smaller your total uncertainty is. In this example using the friction factor from step one would have resulted in no change to the outcome, that conclusion does not say that in the next example it won't lead you to selecting a different pipe size by including a redetermination of every parameter at every step. When you accept 4% uncertainty on viscosity, 2% uncertainty on friction factor, 10% uncertainty on efficiency, 20% uncertainty by using nominal instead of actual ID, etc., sooner rather than later you reach a point where you would be better off using a pipe-capacity vs. pressure chart generated at HVAC.com (I don't know if that is a real web site, but the HVAC guys seem to be the ones most content with ignoring compressibility and treating viscosity as a function of gas composition rather than composition, pressure, and temperature, mostly because their dP is consistently very low).

When I was learning this stuff we used slide rules (yep, I'm THAT old). Sensible engineers back then would pull together all of the terms that are acceptably constant for that problem (and the threshold for "constant" was pretty low) into one number and only change the few parameters that they believed actually made a difference for that step. With MathCad (or to a lesser extent Excel and calculators) redoing friction factor for every segment is no more difficult than adding a line and a function call to the program. I feel that adding that line (which admittedly rarely makes a meaningful difference to the outcome) increases my confidence in that the program will lead me to a decision that will result in a higher chance of a project performing like I predicted, which after all is the reason for the exercise.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[katmar]

If the Reynolds number is changing by more than the change in the viscosity then there is a mistake in your program. In a pipe of constant diameter the density and velocity don't "largely" offset each other. They perfectly offset each other.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"