6 replies to this topic

[Steve McGahey]

Hi all,

I've been doing some pipeline modelling in PipeSim recently, and I've found a really puzzling holdup-vs-flow curve, which runs counter to every other holdup-vs-flow curve I've seen before. I'm wondering if anyone else here has
(1) seen something similar,
(2) an idea of if it is realistic
(3) why such an odd curve occurs

In general, I'd expect my curves to look like the first one in this uploaded file - large holdup at low flowrates, and low holdup at low flowrates, following an asymptotic-like decay.

In the problematic case (which is a gas-condensate case with little liquid and low gas flowrates), I'm seeing the second sort of curve, which is low at the start, high in the middle, and low at the end.

The odd thing is that using three standard mechanistic correlations (Olga, Shell and Ainsley), I'm getting the second curve, but if I use the default Pipesim correlation (Beggs and Brill revised), I get the first curve. Three out of four correlations suggest the second curve is right, but I'm a little cautious about going with the "unusual" behaviour.

Can anyone here shed any light on this odd behaviour? Nearest I can figure out, I've somehow exceeded the validity range of the mechanistic correlations, or else there's a mechanism other than deposition and entrainment going on which kills off the holdup at low flowrates.

Any thoughts would be appreciated.

Thanks,
- Steve.

Attached Files

•  Odd_Holdup.jpg   12.48KB   99 downloads

[Steve McGahey]

I've resolved it now - and think I'll write the reasoning and solution up properly before posting it... the clue, for anyone else out there, is to watch the flow-regime in your riser - it'll change everything!

[joerd]

Interesting - so Beggs&Brill apparently doesn't properly predict the flow regime / slugging in the riser?

Interesting - so Beggs&Brill apparently doesn't properly predict the flow regime / slugging in the riser?

hi,
i checked your hold up vs flow fig. can i ask u something what type of fluid are u talking about and did u take the various friction factors into considerations
i am not sure how this can help u but plzz do check this out

adi

[Steve McGahey]

Interesting - so Beggs&Brill apparently doesn't properly predict the flow regime / slugging in the riser?

hi,
i checked your hold up vs flow fig. can i ask u something what type of fluid are u talking about and did u take the various friction factors into considerations
i am not sure how this can help u but plzz do check this out

adi

Hi Joerd,

That's right - Beggs and Brill doesn't predict this behaviour - but that's for a few good reasons
- it's an empirical correlation
- it's optimised for the normal part of the flow regime.
- it's based on air/water systems, not something as nearly phase-continuous as gas/condensate/aqueous systems.
- it's based on small diameter pipes

When you make a move to things like the Shell, Olga or other mechanistic correlations, you see this behaviour at very low flows and low condensate-gas ratios (eg when there is little condensate).

Hey Adi,

It is a (fairly light) hydrocarbon gas-condensate system. Can you explain what you mean when you refer to the "various" friction factors? I've let PipeSim calculate the frictional losses based on the roughness of the material, and I believe that's all I could do.

- Steve

[Radionise]

I was about to shed some lights on this odd hold-up profile but I guess Steve had aptly summarised the points leading to these anomalies in the post above. And I can stress that his last two points on pipe diameter and the difference between air-water mixture with real hydrocarbon two-phase mixture are the main reasons that causes the Beggs and Brills to fail at predicting the liquid hold-up in large riser correctly.

Some research on two-phase flow have been extensively carried out at Univ. of Nottingham. One of which was on the transition of flow pattern in risers with large diameter. As the gas fraction increases, most flow pattern maps would predict transition from bubbly to churn and subsequently to slug flow before changing to annular flow. These flow pattern maps were developed based on observations and experiments in smaller pipe diameters.

However, with larger diameter, it was observed that for the flow pattern changed from bubbly to churn and directly to annular, yes, no slug flow. Hence, riser diameter does influence the flow pattern and hence liquid hold-up prediction.

Some research papers have been published on this recently but I don't exactly know in which journals.

[Steve McGahey]

Thanks for the info on the work that has been done on this... I'm quite interested to see some experimental results in the area. If you happen to remember something about these research papers, or the names of the people involved, I would be quite appreciative.

Cheers,
- Steve.

I was about to shed some lights on this odd hold-up profile but I guess Steve had aptly summarised the points leading to these anomalies in the post above. And I can stress that his last two points on pipe diameter and the difference between air-water mixture with real hydrocarbon two-phase mixture are the main reasons that causes the Beggs and Brills to fail at predicting the liquid hold-up in large riser correctly.

Some research on two-phase flow have been extensively carried out at Univ. of Nottingham. One of which was on the transition of flow pattern in risers with large diameter. As the gas fraction increases, most flow pattern maps would predict transition from bubbly to churn and subsequently to slug flow before changing to annular flow. These flow pattern maps were developed based on observations and experiments in smaller pipe diameters.

However, with larger diameter, it was observed that for the flow pattern changed from bubbly to churn and directly to annular, yes, no slug flow. Hence, riser diameter does influence the flow pattern and hence liquid hold-up prediction.

Some research papers have been published on this recently but I don't exactly know in which journals.