You are on page 1of 5

PgDip/MSc The Energy Programme Tubing Performance Correlations

Tubing Performance Correlations

Reviews

This topic introduces the mathematical correlations utilised by many of the


commercially available software packages for analysing two phase flow in
tubing.

Content
Introduction

Many investigators have conducted research into multiphase flow on tubing. Most of the
investigative approaches have made basic assumptions that can be used to classify the
correlations derived as follows:
• Methods which do not consider slippage between phases or the use of flow
regime or pattern;
• Methods which consider slippage but not flow regimes;
• Methods that consider both flow regime and slippage.

Most of the multiphase flow correlations can be used with the general procedure. Use
will be made of the general equation:
⎛ dP ⎞ ⎛ dP ⎞ ⎛ dP ⎞ ⎛ dP ⎞
Equation 1 ⎜ ⎟ =⎜ ⎟ +⎜ ⎟ +⎜ ⎟
⎝ dD ⎠ TOTAL ⎝ dD ⎠ ELEV ⎝ dD ⎠FRICT ⎝ dD ⎠ VEL

next determine:
⎛ dP ⎞
Equation 2 ⎜ ⎟ =ρ
⎝ dD ⎠ELEV

thirdly calculate:
⎛ dP ⎞ f ⋅p ⋅ v2
Equation 3 ⎜ ⎟ = m
⎝ dD ⎠FRICT 2⋅g⋅d

and finally determine


⎛ dP ⎞ e ⋅ ∆ ⋅ rm2
Equation 4 ⎜ ⎟ = m
⎝ dD ⎠ VEL 2 ⋅ g ⋅ dD

Correlations that Consider neither Slippage nor Flow Regimes

The methods in this category include:


• Poetmann & Carpenter;
• Baxendell & Thomas;
• Tek;
• Francher & Brown;
• Hagedorn & Brown.

© The Robert Gordon University 2003 1


PgDip/MSc The Energy Programme Tubing Performance Correlations

With these methods, slippage is not considered and neither are the various flow regimes
that exist in multiphase flow. The methods are based on a fluid density calculated from
the density of the surface fluids and corrected for downhole conditions. The methods are
based on their individual approach or correlations for calculating the two phase friction
factor. The friction factor is correlated against the numerator of the Reynold’s number.

Correlations that include Slippage but not Flow Regime

The methods in this category include:


• Baker & Keep;
• Hughmark & Pressburg;
• Hagedorn & Brown.

In these methods, correlations are necessary to allow prediction of both the slippage
and the friction factor. However, since the method does not distinguish between flow
regimes, the correlations are assumed to be valid for all flow regimes. The concept of
slippage will require that the superficial velocities of the phases be correlated and the
volumetric pipe fraction for each phase is used to define the mixture properties.

Correlations that Consider both Slippage & Flow Regime

The majority of research carried out in multiphase flow lies within this category,
including:
• Duns & Ros;
• Orkiszewski;
• Aziz, Govier & Fogerasi;
• Beggs & Brill.

All the above propose a method of predicting the slippage or holdup based upon the
assumed set of flow regimes. The methods therefore widely use flow pattern maps.
Prediction of the flowing pressure loss requires correct identification of flow regime,
which prevails in the tubing.

Inclined Flow Correlations

Relatively little research has been conducted in this area, with the majority of techniques
evolving from horizontal correlations modified to account for ‘uphill’ sections of pipeline.
Amongst the published methods are:
• Beggs & Brill;
• Griffith.

The Griffith correlation only considered two flow regimes, namely slug flow and annular
flow. It does not consider the effects of pipe roughness, liquid viscosity and entrainment.
The method of Beggs & Brill is the widely accepted method for predicting inclined well
pressure drop. There are major difficulties associated with developing a general inclined
well flow correlation including:
• the inclination in the well is not constant;
• the well may have doglegging, which will impact on well slugging.

The vertical lift performance correlations available are extensive; they were developed to
account for the changes in fluid properties as a function of pressure and temperature
changes. However each correlation has been developed to satisfy the prediction of
certain specific conditions. No single correlation is available that is ideal for all wells. It is

© The Robert Gordon University 2003 2


PgDip/MSc The Energy Programme Tubing Performance Correlations

therefore very beneficial if the correlations can be short-listed based on their intended
use (ie, flow regime or oil density for example). From the short-list it is then possible to
select a correlation that most accurately matches real data eg, a well test or series of
well tests. The following paragraphs briefly describe the most common correlations.

Fancher-Brown

This correlation represents a non-slip (or homogeneous) flow equation. This means that
it will always over-predict the fraction of the pipe occupied by the gas. As a result it will
tend to under predict the pressure drop where slippage occurs. It will therefore only
accurately predict the pressure drop under high flow bubble regime conditions or single-
phase flow (eg, above the bubble point).
The importance of this correlation is that it should under predict pressure drop.
Therefore when plotted against real data it should lie to the left of the real (measured)
data.

Hagedorn and Brown

This flow correlation does not use a flow map. Instead it is based on data from a 1400 ft
test well with air, water and dead crude over a range of viscosities from 10-110 cp. It
covers bubble and slug flow and therefore should not be used for gas wells or high GOR
wells. There are some errors in large bore wells with a 35-70 degrees deviation where
oil / water slippage may occur. Despite these constraints it is frequently used and
accurate under the conditions it was designed for.

Duns and Ross


This correlation uses a flow map (ie, determines the flow regime) to determine the
hold-up and friction. It should therefore apply to all flow regimes. In reality in works best
in gas and gas condensate wells - especially in mist flow and performs poorly in oil wells
where slug flow occurs or at low rates. The loading rate prediction for gassy wells (>
6,000 scf/stb) is generally good.

Orkiszewski

This correlation is an amalgam of various correlations including Duns and Ross. It works
well over a wide range of conditions and often appears as the closest correlation to
match well test results. Unfortunately there is a discontinuity in the predictions at a
mixture velocity of 10 ft/s. This result in a jump in predicted pressures as illustrated in
the figure below, making the correlation of little practical use under these conditions. As
a result, it is commonly used for tubing sizes of up to 3 ½”.

© The Robert Gordon University 2003 3


PgDip/MSc The Energy Programme Tubing Performance Correlations

Figure 1. VLP’s for 4 ½ “ and 5 ½ “ Tubing using Orkiszewski (Prosper Output).

Beggs and Brill

This was the first comprehensive study of multiphase flow for inclined pipes4. Small
diameter pipe (up to 1.5in) was used in experiments where the flow regime was
observed. The correlation over predicts liquid hold-up in wells and pipelines however it is
in widespread use as it applies to vertical, horizontal and inclined pipes.
The main use of the Beggs and Brill correlation is for quality control. As it over predicts
the holdup, it over estimates pressure losses -typically by between 10 and 15%. This
means that actual data should lie to left of the Beggs and Brill correlation. If results lie to
the right of the correlation, possible causes are:
• inaccurate PVT data - particularly too low a liquid density;
• under recorded water cuts or over estimated gas rates;
• too low a friction factor.

Caution: arbitrarily altering any parameter in order to get a match must be avoided.
There may be a temptation to increase roughness in order to get consistency. However
this should only be done if there is physical or analytical evidence that corrosion, scale
or other deposits may be affecting the condition of the bore of the production conduit.

Ansari

Ansari is a ‘mechanistic’ model in that it predicts pressure drops based on the


mechanism ie, the flow regime. This is similar to the more sophisticated traditional
methods such as Duns and Ross. However newer mechanistic models take this further
by eliminating more empirical steps and consider the transitional regimes further, hence
it should therefore cover all flow regimes, all sizes of tubing, all deviations and
everything from heavy oil to gas. In a comparison of this model across a number of oil
and gas fields, at a variety of conditions, this correlation was the most widely applicable
– although by no means perfect, with average absolute errors of around 7%.

© The Robert Gordon University 2003 4


PgDip/MSc The Energy Programme Tubing Performance Correlations

There is still concern that this correlation (and many others) is not fully applicable for
deviated wells. In fact, these models are still only currently ‘mechanistic’ for vertical
wells, errors in predicting inclined flow will therefore be greater than that of a vertical
well. This correlation finds greatest application with low GOR, heavy or medium oil fluids
such as in the Gryphon, Foinaven and Alba fields. It is acceptable for use on higher
GOR fluids and gasses, but correlations such as Gray are often better for gas.

Gray

This correlation was developed for vertical gas condensate wells and was originally
used in the API 14B subsurface safety valve sizing program. The published limits of the
correlation are a flow velocity of 50 ft/sec, 3 1/2" tubing or less, and a condensate ratio
of less than 50 bbl/MMscf. In reality however, Gray has proven accuracy in low to
moderate CGR gas wells and is useful for predicting liquid loading. It is empirical and
uses its own PVT module for condensate and liquid prediction. It is therefore not
compatible with compositional PVT models.

Which Correlation to Use

The completion engineer must implement a strict selection process in order to choose a
realistic correlation that will allow him/her to simulate tubing performance in a realistic
manner. The selection process should be largely based on a detailed and accurate PVT
data and expected reservoir conditions. Be cautious when selecting a correlation, please
note:
• don’t use the correlation outside of its intended regime;
• once a good match has been found, use the same correlations for other wells
producing form the same field/reservoir.

Figure 2 illustrates the possible differences for the same set of data using different
correlations; note that such an error can lead to the prediction of production rates that
are unrealistic.
Figure 2. Correlation Comparison and Impact on VLP.

© The Robert Gordon University 2003 5

You might also like