## Presentation on theme: "Liquid Loading Current Status, New Models and Unresolved Questions"— Presentation transcript:

Mohan Kelkar and Shu Luo The University of Tulsa

Model Formulation Model Validation Program Demonstration Summary

What is liquid loading? Minimum pressure drop in the tubing is reached
The liquid drops cannot be entrained by the gas phase (Turner et al.) The liquid film cannot be entrained by the gas phase (Zhang et al., Barnea) The answers from different definitions are not the same

As gas flow rate increases 𝑑( ∆𝑝 𝑔 ) 𝑑( 𝑉 𝑔 ) and 𝑑( ∆𝑝 𝑓 ) 𝑑( 𝑉 𝑔 ) At low velocities 𝑑( ∆𝑝 𝑔 ) 𝑑( 𝑉 𝑔 ) decreases faster than increase in 𝑑( ∆𝑝 𝑓 ) 𝑑( 𝑉 𝑔 ) When two gradients are equal, minimum occurs

Definition Based on Mechanisms
Two potential mechanisms of transition from annular to slug flow Droplet reversal Film Reversal Models are either based on droplet reversal (Turner) or film reversal (Barnea) 2 mechanisms: droplet and film reversal Droplet model determine liquid loading when droplet falls back Film model determine liquid loading when liquid film falls back

Literature Data Air-water data are available
The data reported is restricted to 2” pipe Very limited data are available in pipes with diameters other than 2” No data are available for other fluids

Generalized Conclusions (2” pipe)
Minimum pressure drop for air-water flow occurs at about 21 m/s The liquid film reversal starts at around 15 m/s The dimensionless gas velocity is in the range of 1.0 to 1.1 at minimum point 𝑢 𝐺 ∗ = 𝑢 𝐺 𝜌 𝐺 1/2 𝑔𝑑 𝜌 𝐿 − 𝜌 𝐺 1/2

Liquid Film Reversal Westende et al., 2007

flow counter current with the gas stream
Liquid Film Reversal At 15 m/s, liquid starts to flow counter current with the gas stream Westende et al., 2007

Liquid Film Reversal Minimum is at 20 m/s (blue line)
Residual pressure reaches a zero value at lower velocity Zabaras et al., 1986

Entrained Liquid Fraction
Alamu, 2012

For vertical pipe OLGA = 12 m/s Exptl = 14 m/s Belfroid et al., 2013

Our Data

Air-Water Flow Skopich and Ajani conducted experiments in 2” and 4” pipes The results observed are different based on film reversal and minimum pressure drop – consistent with literature However, the experimental results are very different for 2” versus 4” pipe

Calculation Procedure
Total pressure drop is measured and gradient is calculated Holdup is measured and gravitational gradient is calculated Subtracting gravitational pressure gradient from total pressure gradient to get frictional pressure gradient By dividing the incremental pressure gradient by incremental gas velocity, changes in gravitational and frictional gradients with respect to gas velocity are calculated.

dPG vs. dPF Air-Water, 2 inch, vsl=0.01 m/s
Minimum

Total dp/dz Air-Water, 2 inch, vsl=0.01 m/s
Film Reversal

dP/dz)G vs. dP/dz)F Air-Water, 2 inch, vsl=0.01 m/s
dp/dz)F is zero

dPT - dPG Air-Water, 2 inch, vsl=0.01 m/s
Transition at 16 m/s

Pressure at Bottom Air-Water, 2 inch, vsl=0.01 m/s
Pressure build up No pressure build up

dP/dz)G vs. dP/dz)F Data from Netherlands (2 inch)
dp/dz)F is zero

What should we expect for 3” or 4” pipeline?
𝑢 𝐺 ∗ = 𝑢 𝐺 𝜌 𝐺 1/2 𝑔𝑑 𝜌 𝐿 − 𝜌 𝐺 1/2 Based on the above equation, the minimum should shift to right as diameter increases If the above equation is correct, the ratio of uG/√d at unstable point should be constant

dPG vs. dPF Air-Water, 4 inch, vsl=0.01 m/s
Minimum

Total dp/dz Air-Water, 4 inch, vsl=0.01 m/s
Film Reversal

dP/dz)G vs. dP/dz)F TUFFP (3 inch, vsl=0.1 m/s)
dp/dz)F is zero

dP/dz)G vs. dP/dz)F Air-Water, 4 inch, vsl=0.01 m/s
dp/dz)F is zero Film reversal

Why diameter impacts? Film thickness?
Skopich et al., SPE

Liquid loading starts when liquid film reversal occurs We adopt the model of film reversal to predict inception of liquid loading The reason for this adoption, as we will show later, is because we are able to better predict liquid loading for field data using this methodology.

Background Turner’s Equation
The inception of liquid loading is related to the minimum gas velocity to lift the largest liquid droplet in the gas stream. Turner et al.’s Equation: This equation is adjusted upward by approximately 20 percent from his original equation in order to match his data. 𝑣 𝐺,𝑇 = 𝜎 𝜌 𝐿 − 𝜌 𝐺 𝜌 𝐺 Turner proposed two models of removal of liquid. One is liquid film move along the walls of pipe, another is liquid droplets entrained in the gas core. And he found the liquid droplet is the dominating mechanism of liquid removal. Then he developed the equation to calculate terminal velocity. Comparing with field data, he suggested that a upward 20% adjustment of the equation will best fit the data.

Background Drawbacks with Turner’s Equation

Approach Film Model Two film models are investigated to predict liquid loading: Zhang et al.’s model(2003) is developed based on slug dynamics. Barnea’s model(1986) predicts the transition from annular to slug flow by analyzing interfacial shear stress change in the liquid film.

Approach Barnea’s Model
Constructing force balance for annular flow and predict the transition from annular to slug flow by analyzing interfacial shear stress changes. The combined momentum equation: Interfacial shear stress with Wallis correlation: 𝜏 𝐼 𝑆 𝐼 1 𝐴 𝐿 𝐴 𝐺 − 𝜏 𝐿 𝑆 𝐿 𝐴 𝐿 − 𝜌 𝐿 − 𝜌 𝐺 𝑔 sin 𝜃 =0 𝜏 𝐼 = 1 2 𝑓 𝐼 𝜌 𝐺 𝑣 𝑆𝐺 2 (1−2𝛿) 4 Schematic of Annular Flow

Approach Barnea’s Model
Solid curves represent Interfacial shear stress from combined momentum equation Broken curves represent Interfacial shear stress from Wallis correlation Intersection of solid and broken curves yields a steady state solution of film thickness and gas velocity at transition boundary Another transition mechanism is liquid blocking of the gas core. Two mechanisms for blockage of core Instability of annular flow that prevents a stable annular configuration Liquid film gets large enough to create spontaneous blockage For low liquid rates, transition occurs due to the first mechanism, whereas, for high liquid rates, transition occurs due to second mechanism Both mechanisms are checked for transition to slug flow Transition

Model Formulation In inclined wells, the film thickness is expected to vary with radial angle Vertical Well Inclined Well

Original Barnea’s Model at Different Inclination Angles

Non-uniform Film Thickness Model

Non-uniform Film Thickness Model
Let A1=A2, we can find this relationship. If film thickness reaches maximum at 30 degree inclination angle 𝛿 𝑐 = 1 2 [𝛿 0,𝜃 +𝛿 𝜋,𝜃 ]

Non-uniform Film Thickness Model
We will use the following film thickness equation in the new model: 𝑭𝒐𝒓 𝟎≤𝜽≤𝟑𝟎 𝒅𝒆𝒈𝒓𝒆𝒆 𝑭𝒐𝒓 𝜽>𝟑𝟎 𝒅𝒆𝒈𝒓𝒆𝒆 𝜹 𝜱,𝜽 = 𝜽 𝟑𝟎 𝒔𝒊𝒏 𝜱−𝟗𝟎 +𝟏 𝜹 𝒄 𝜹 𝜱,𝜽 = 𝒔𝒊𝒏 𝜱−𝟗𝟎 +𝟏 𝜹 𝒄

Non-uniform Film Thickness Model
Only maximum film thickness will be used in the model because thickest film will be the first to fall back if liquid loading starts. Find critical film thickness δT by differentiating momentum equation. δT equals to maximum film thickness δ(π,30). 𝛿 𝑐 = 1 2 [0+𝛿 𝜋,30 ]= 𝛿 𝑇

Non-uniform Film Thickness Model

Other Film Shape

Interfacial Friction Factor
Critical gas velocity calculated by Barnea’s model is conservative compared to other methods. Fore et al. showed that Wallis correlation is reasonable for small values of film thickness and is not suitable for larger film thickness liquid film. A new correlation is used in the new model : 𝑓 𝐼 = 𝑅𝑒 𝐺 ℎ 𝐷 −0.0015

Turner’s Data 106 gas wells are reported in his paper, all of the gas wells are vertical wells. 37 wells are loaded up and 53 wells are unloaded. 16 wells are reported questionable in the paper. Current flow rate and liquid loading status of gas well are reported.

Turner’s Model Results Turner’s Data
Vg < Vg,c Vg > Vg,c This graph shows a plot of boundary velocity calculated by the method vs. actual gas velocity at well head conditions. If boundary velocity is greater than observed velocity, slug flow is present; if less, annular flow is present. The prediction is quite reasonable with 90 % success.

Barnea’s Model Results Turner’s Data
This graph shows a plot of boundary velocity calculated by the method vs. actual gas velocity at well head conditions. If boundary velocity is greater than observed velocity, slug flow is present; if less, annular flow is present. The prediction is quite reasonable with 90 % success.

New Model Results Turner’s Data
This graph shows a plot of boundary velocity calculated by the method vs. actual gas velocity at well head conditions. If boundary velocity is greater than observed velocity, slug flow is present; if less, annular flow is present. The prediction is quite reasonable with 90 % success.

Coleman’s Data 56 gas wells are reported, all of the wells are also vertical wells. These wells produce at low reservoir pressure and at well head pressures below 500 psi. Coleman reported gas velocity after they observed liquid loading in gas wells.

Turner’s Model Results Coleman’s Data
Coleman examined the data for wells which were only at critical conditions. Which means every one of these wells were loading. As can be seen, the prediction of the new method is quite good in correctly predicting that all the wells are loading.

Barnea’s Model Results Coleman’s Data
Coleman examined the data for wells which were only at critical conditions. Which means every one of these wells were loading. As can be seen, the prediction of the new method is quite good in correctly predicting that all the wells are loading.

New Model Results Coleman’s Data
Coleman examined the data for wells which were only at critical conditions. Which means every one of these wells were loading. As can be seen, the prediction of the new method is quite good in correctly predicting that all the wells are loading.

Veeken’s Data Veeken reported offshore wells with larger tubing size.
67 wells, which include both vertical and inclined wells, are presented. Similar to Coleman’s data, critical gas rate was reported. Liquid rate were not reported in the paper. We assumed a water rate of 5 STB/MMSCF. Liquid rate will not affect

Turner’s Model Results Veeken’s Data
Bad prediction may be pipe diameter and inclination angle

Barnea’s Model Results Veeken’s Data
Well deviation angle : Veekan’s data also only shows wells which are producing at critical conditions. Clearly, the method is not working as well and in many wells, the model is predicting annular flow.

New Model Results Veeken’s Data
Well deviation angle : Veekan’s data also only shows wells which are producing at critical conditions. Clearly, the method is not working as well and in many wells, the model is predicting annular flow.

Chevron Data Production data:
Monthly gas production rate Monthly water and oil production rate 82 wells have enough information to analyze liquid loading Two tubing sizes: and inch Get average gas and liquid production rate when cap string is installed from service history. Assume liquid loading occurred at this point.

Production Data

Turner’s Model Results Chevron Data

New Model Results Chevron Data

ConocoPhillips Data Daily production data and casing and tubing pressure data are available Select 62 wells including 7 off-shore wells Two tubing size: and inch Determine liquid loading by casing and tubing pressure divergence.

ConocoPhillips Field Data

Turner’s Model Results ConocoPhillips Data

New Model Results ConocoPhillips Data

Future Improvements Better interfacial fi correlation

Improvements Liquid Entrainment Collection of 5” data
Impact on the inception of liquid loading Collection of 5” data Pressure drop inspection for larger diameter pipes Incorporation of foam data in model

Summary Liquid film reversal is the most appropriate model for defining liquid loading The effect of diameter on liquid loading is significant and is related to square root of diameter The film reversal can be detected either by observation of film or residual pressure drop

Thank You! Questions…