Colebrook Equation

Colebrook's equation is an implicit expression, thus iteration is required to calculate

the friction factor in the transition zone.

From: Flow and Heat Transfer in Geothermal Systems, 2017

Related terms:

Pipelines, Viscosity, Colebrook-White Equation, Compressibility Factor, Friction Fac-

tor, Reynolds' Number, Base Pressure, Temperature Base

View all Topics

Learn more about Colebrook Equation

Fluid Flow in Pipes

E. Shashi Menon, in Transmission Pipeline Calculations and Simulations Manual,
2015

23 Colebrook–White Equation
The Colebrook–White equation, sometimes referred to simply as the Colebrook
equation is a relationship between the friction factor and the Reynolds number,
pipe roughness, and inside diameter of pipe. The following form of the Colebrook
equation is used to calculate the friction factor in gas pipelines in turbulent flow.

(5.101)

where

f – friction factor
D – pipe inside diameter, in
e – absolute pipe roughness, in
R – Reynolds number of flow for R > 4000
Because R and f are dimensionless, as long as consistent units are used for both e and
D, the Colebrook equation is the same regardless of the units employed. Therefore,
in SI units Equation above is used with e and D expressed in mm.

It can be seen from Eqn (5.101), that to calculate the friction factor f we must use
a trial-and-error approach. It is an implicit equation in f because f appears on both
sides of the equation. We first assume a value of f (such as 0.01) and substitute it
in the right hand side of the equation. This will yield a second approximation for f
which can then be used to calculate a better value of f and so on. Generally three to
four iterations are sufficient to converge on a reasonably good value of the friction
factor.

It can be seen from the Colebrook equation, for turbulent flow in smooth pipes
the first term within the square brackets is negligible compared to the second term
because pipe roughness e is very small.

Therefore for smooth pipe flow, the friction factor equation reduces to

(5.102)

Similarly, for turbulent flow in fully rough pipes, with R being a large number,
f depends mostly on the roughness e and therefore, the friction factor equation
reduces to

(5.103)

Table 5.2 lists typical values of pipe internal roughness used to calculate the friction
factor.

Table 5.2. Pipe internal roughness

Pipe material Roughness (in) Roughness (mm)

Riveted steel 0.0354–0.354 0.9–9.0
Commercial steel/welded steel 0.0018 0.045
Cast iron 0.0102 0.26
Galvanized iron 0.0059 0.15
Asphalted cast iron 0.0047 0.12
Wrought iron 0.0018 0.045
Polyvinyl chloride, drawn tub- 0.000059 0.0015
ing, glass
Concrete 0.0118–0.118 0.3–3.0

As an example, if R = 100 million or larger and e/D = 0.0002, the friction factor from
Eqn (5.103) is

Or f = 0.0137, which correlates well with the friction factor obtained from the Moody
diagram Figure 5.9.
Example 5.14 (USCS)
A natural gas pipeline, NPS 20 with 0.500-in wall thickness transports 200 MMSCFD.
The specific gravity of gas is 0.6 and viscosity is 0.000,008 lb/ft-s. Calculate the
friction factor using the Colebrook equation. Assume absolute pipe roughness =-
 600 micro in.

The base temperature and base pressure are 60 °F and 14.7 psia, respectively.

Solution
Pipe inside diameter = 20 − 2 × 0.5 = 19.0 in

Absolute pipe roughness = 600 micro in = 0.0006 in

First we calculate the Reynolds number.

Using Eqn (5.101):

This equation will be solved by successive iteration.

Assume f = 0.01 initially and substituting above we get a better approximation as:

f = 0.0101. Repeating the iteration, we get the final value as:

f = 0.0101.

Therefore, the friction factor is 0.0101.

Example 5.15 (SI Units)

A natural gas pipeline, DN 500 with 12-mm wall thickness transports 6 Mm3/day.
The specific gravity of gas is 0.6 and viscosity is 0.00012 poise. Calculate the friction
factor using the Colebrook equation. Assume absolute pipe roughness = 0.03 mm
and the base temperature and base pressure are 15 °C and 101 kPa, respectively.

Solution
Pipe inside diameter = 500 − 2 × 12 = 476 mm.

First we calculate the Reynolds number.

The friction factor is

This equation will be solved by successive iteration.

Assume f = 0.01 initially and substituting above we get a better approximation as:

f = 0.0112. Repeating the iteration, we get the final value as:

f = 0.0112.

Therefore, the friction factor is 0.0112.

> Read full chapter

Pipeline Hydraulic Analysis

E. Shashi Menon Ph.D., P.E., in Pipeline Planning and Construction Field Manual,
2011

8.14 Colebrook–White Equation for Gas Flow

The Colebrook–White equation, discussed earlier in liquid flow, can also be used to
calculate the friction factor in gas flow. The following form of the Colebrook equation
is used to calculate the friction factor in gas pipelines in turbulent flow.

(8.51)

where, in USCS units,

f – Friction factor, dimensionless

D – Pipe inside diameter, in.
e – Absolute pipe roughness, in.
R – Reynolds number of flow, dimensionless

Because R and f are dimensionless, as long as consistent units are used for both e
and D, the Colebrook equation is the same regardless of the units used. Therefore,
in SI units, Eq. (8.51) is used with e and D expressed in mm.

It can be seen from Eq. (8.51) that, in order to calculate the friction factor f, we must
use a trial and error approach. It is an implicit equation in f since f appears on both
sides of the equation. We first assume a value of f (such as 0.01) and substitute it
in the right-hand side of the equation. This will yield a second approximation for f
that can then be used to calculate a better value of f and so on. Generally, three to
four iterations are sufficient to converge on a reasonably good value of the friction
factor.

Example 8.10 (USCS Units)

A natural gas pipeline, NPS 24 with 0.500 in. wall thickness transports 250 MMSCFD.
The specific gravity of gas is 0.6 and viscosity is 0.000008 lb/ft · s. Calculate the
friction factor using the Colebrook equation. Assume absolute pipe roughness = 600
μ-in. The base temperature and base pressure are 60°F and 14.7 psia, respectively.
Solution

First, we calculate the Reynolds number using Eq. (8.47):

Using the Colebrook–White Eq. (8.51), the friction factor is

This equation will be solved by successive iteration. First, assuming f = 0.01 and
substituting in the preceding equation, we get a better approximation as f = 0.0099.
Repeating the iteration, we get the final value as f = 0.00988.

Therefore, the friction factor is 0.00988.

Example 8.11 (SI Units)

A natural gas pipeline, DN 500 with 12 mm wall thickness transports 6 Mm3/day.

The specific gravity of gas is 0.6 and viscosity is 0.00012 Poise. Calculate the friction
factor using the Colebrook equation. Assume absolute pipe roughness = 0.03 mm
and the base temperature and base pressure are 15°C and 101 kPa, respectively.

Solution

First, we calculate the Reynolds number using Eq. (8.48)

Using Eq. (8.51), the friction factor is

This equation will be solved by successive iteration. Assume f = 0.01 initially and
substituting above, we get a better approximation as f = 0.0112. Repeating the
iteration, we get the final value as f = 0.0112.

Therefore, the friction factor is 0.0112.

> Read full chapter

Heat exchangers
G. Musgrove, ... L. Chordia, in Fundamentals and Applications of Supercritical
Carbon Dioxide (sCO₂) Based Power Cycles, 2017

8.5.4.2 Hydraulic performance

The hydraulic performance of sCO2 heat exchangers is estimated using the conven-
tional approach of estimating a friction factor or loss coefficient and multiplying by
the dynamic pressure of the fluid, as defined in Eq. (8.2). When operating away from
the critical region where the thermodynamic properties do not significantly change,
conventional correlations can be used that do not take into account thermodynamic
property variations, such as the Colebrook equation for turbulent flow friction fac-
tor (Eq. 8.3). Near the critical region, however, variations exist in thermodynamic
properties between the fluids near the wall and the bulk fluid due to different tem-
peratures. One such method is to use the property ratio method, which adjusts the
constant property friction factor by the ratio of the wall and mean fluid temperatures
taken to a power, as defined in Eq. (8.4). Similar to the thermal analysis near the
critical region, the pressure drop estimate should also be discretized along the length
of the heat exchanger to account for the variation in thermodynamic properties. To
estimate flow losses due to pipe bends and the expansions and contractions in the
manifolds, loss factors can be used from conventional pipe loss references such as
Idelchik (2007) or Miller (1990).

(8.2)

(8.3)

(8.4)

> Read full chapter

Steam–Water Circulation Design

Esa Kari Vakkilainen, in Steam Generation from Biomass, 2017

4.5.3 Friction Loss Coefficient

In general the friction loss coefficient depends on the flow conditions. The pressure
loss is different for different types of flow. The friction loss coefficient for laminar
flow is expressed by the Hagen–Poiseuille equation

(4.8)

For turbulent flow in hydraulically smooth tubes, when Re is from 3000 to 100,000,
the friction loss coefficient is expressed by the Blasius equation

(4.9)

And for turbulent flow in rough tubes the pressure loss coefficient is expressed
by, e.g., Colebrook equation (Colebrook, 1939) (Fig. 4.17).
Figure 4.17. Moody chart for the friction factor of fully developed pipe flow for
various k/d ratios as shown in right.

After Moody, L.F., 1944. Friction factors for pipe flow. Trans. ASME 66 (8), 671–684
(Moody, 1944).

(4.10)

Usually these equations are shown as the Moody chart.

The pressure loss coefficient is lower for high Reynolds numbers. On the other
hand, pressure loss is higher for high velocities and high Reynolds numbers because
dynamic pressure increases as a square of velocity. The pressure loss coefficient is
lower for large diameter tubes. Roughness, k, for drawn steel tubes is 0.0250.05 mm.
Therefore the pressure loss coefficient is usually close to 0.05 (Nikuradse, 1933).
It should be noted that there is a transition region from smooth to rough surfaces.
Flow behaves smoothly at the transition region.

Typical economizer flows in boilers have high Reynolds numbers, but superheater
flows can be in the transition region from laminar to turbulent flow. It is therefore
suggested that a combination equation be used to calculate the pressure loss. One
such equation has been developed by Churchill (1972).

(4.11)
where

> Read full chapter

FLUID FLOW
A. Kayode Coker, in Ludwig's Applied Process Design for Chemical and Petrochem-
ical Plants (Fourth Edition), Volume 1, 2007

4.15 DARCY FRICTION FACTOR, F

For laminar or viscous flow,

(4-32)

For transition and turbulent flow, use Figure 4-13 (the f in this figure only applies
for fully turbulent flow corresponding to the flat portions of the curves in Figure
4-5) with Figure 4-5, and Figures 4-14a and 4-14b as appropriate. Friction factor in
long steel pipes handling wet (saturated with water vapor) gases such as hydrogen,
carbon monoxide, carbon dioxide, nitrogen, oxygen, and similar materials should
be considered carefully, and often increased by a factor of 1.2–2.0 to account for
corrosion.
Figure 4-14a. Resistance coefficients for fittings.

(Reprinted by permission from Hydraulic Institute, Engineering Data Book, 1st ed.,
1979, Cleveland, OH.)

Copyright © 1979
Figure 4-14b. Resistance coefficients for valves and fittings.

(Reprinted by permission from Hydraulic Institute, Engineering Data Book, 1st ed.,
1979, Cleveland, OH.)

Copyright © 1979

Important Note: The Moody [3] friction factors (fD) reproduced in this text (Figure 4-5)
are consistent with the published values of references [2–4], and cannot be used with
the values presented in Perry's Handbook [1] (e.g., the Fanning friction factor, fF), as
Perry's values for fF are one-fourth times the values cited in this chapter (e.g., fF =
1/4 fD). It is essential to use f values with the corresponding formulas offered in the
appropriate text.

The Colebrook equation [9] is considered a reliable approach to determining the

friction factor, fD (Moody factor).

(4-33)

Note: The turbulent portion of the published Moody diagram is actually a plot of this
equation, which is derived to fit “sand roughness” data in pipes [10].

where

D = Pipe internal diameter

v = Liquid velocity.

Note that the term /D is the relative roughness from Figure 4-13. Equation 4-33
is implicit in fD, as it cannot be rearranged to derive fD directly and thus requires an
iterative solution (e.g., trial and error). Colebrook [9] also proposed a direct solution
equation that is reported [11] to have

(4-34)

An explicit equation for calculating the friction factor (fC) as proposed by Chen [12]
is

(4-35)

where

= epsilon, absolute pipe roughness, in. (mm)

D = pipe inside diameter, in. (mm).

The relationship between Chen friction factor and Darcy friction factor is fC = 1/4 fD.

Churchill [13] developed a single expression for the friction factor in both laminar
and turbulent flows as:

(4-36)

where

Note: Equation 4-36 gives the Fanning friction factor.

Equation 4-36 is an explicit equation, and adequately represents the Fanning friction
factor over the entire range of Reynolds numbers within the accuracy of the data
used to construct the Moody diagram, including a reasonable estimate for the
intermediate or transition region between laminar and turbulent flow.
Gregory and Fogarasi [14] have provided a detailed review of other explicit equations
for determining the friction factor, and concluded that Chen's friction factor equation
is the most satisfactory friction factor in comparison to others as shown in Table 4-4.

TABLE 4-4. Explicit Equations for Calculating the Friction Factor for Rough Pipe

> Read full chapter

Surface Engineering Concepts

David A. Simpson P.E., in Practical Onshore Gas Field Engineering, 2017

4.1.4 Friction factor

Empirical equations (such as Eq. (3.10)) for flow will have a parameter called a
“friction factor.” These factors are nondimensional and a given flow will have the
same factor in either fps or SI units. The most widely accepted relationship between
Reynolds number and relative pipe roughness in the turbulent flow regime is called
the “Colebrook equation” (Eq. (4.3)). This equation obviously requires an iterative
solution since the Moody friction factor is on both sides of the equation. Eq. (4.4) is
Poiseuille’s law or the “Hagen–Poiseuille equation” and applies to smooth pipe flow
(i.e., at low Reynolds numbers the pipe roughness becomes immaterial). Eq. (4.5)
is Churchill’s 1973 noniterative approximation of the Colebrook equation (Churchill
revised this equation in 1977 with considerable added complexity which results in
values that are very close to the 1973 equation) which provides very reasonable values
as long as you honor the limits shown in the equation statement.

(4.3)

(4.4)

(4.5)

If you have “smooth pipe flow” by the description in Eq. (4.2) then you set _D to
50×10−6 which corresponds to the last _D line that crosses the fully turbulent line
on Fig. 4.1.
Figure 4.1. Moody friction factors.

Fig. 4.1 is usually presented as log–log scales, but I find that a logarithmic y-axis
creates more confusion with interpolation than it provides in improved visibility.

Friction factors from Fig. 4.1 are used in various equations that use assumed flow
rates to estimate a pressure drop, estimate a flow rate from measured pressure drop,
or predict a flow rate from assumed pressure drop. For the last two uses, velocity is
not known at the start of the calculation so the flow equation must be iterated. You
guess a Reynolds number, solve for a friction factor, calculate a flow rate, calculate a
new Reynolds number with the new velocity, use that value to calculate a new friction
factor, and then repeat until the change from one step to the next is acceptably small.

4.1.4.1 Moody (D’Arcy) friction factor

The friction factor is a dimensionless parameter that can either be calculated using
something like Eqs. (4.3)–(4.5), or extracted from a version of Fig. 4.1.

4.1.4.2 Fanning friction factor

As researchers started moving the study of fluid flow in pipes from strictly liquid
flows to gas flows, the Moody friction factors appeared to be a bit coarse. John
Thomas Fanning (1896) looked at the stresses in gas flows and found that:

(4.6)

Many equations developed for gas flow are designed to use the Fanning friction
factor. Authors are not always clear as to which friction factor to use. A rule of thumb
that might be useful is:

• fm → Moody friction factor

• fd → Moody friction factor (the subscript “d” refers to D’Arcy’s work which
predated Moody’s work)
• ff → Fanning friction factor

• f → Assume Moody friction factor for liquids and Fanning friction factor for
gases. This can be very problematical since some researchers are too clever
for their own good and roll the ¼ factor into a constant and use f without a
subscript, but if you use Fanning you’ll fail to match field conditions by a factor
of 4. My approach in this case is that if you find an author that is too lazy to
differentiate then you should look for another source.

4.1.4.3 Average pressure

In Eq. (4.1) we need to calculate density, viscosity, and velocity. For liquids, viscosity
is a function of pressure and temperature. For gases, density, viscosity, and velocity
are all strong functions of pressure. While we can assume that a flowing pipeline is
isothermal, it is anything but isobaric; so what pressure do you use to calculate these
important parameters? Choices seem to be upstream, downstream, or average.
Average seems to make the most sense, but is it a simple average or is it somehow
weighted? Analysis of pressure drop in a pipeline is an important field of study, and
researchers have shown that the pressure drop per unit length is higher early in a
pipeline than later in the line, so an average pressure should be front-end weighted.
We do this by:

(4.7)

If we compare this to a simple average (i.e., (P1+P2)/2) you can see (Fig. 4.2) that for
short lengths of pipe it simply doesn’t matter. In this example, you don’t have a 5%
error until over 50 miles (80.5 km). Eq. (4.7) is simple enough that my approach is
to always use it and eliminate that (admittedly small) error from the body of the
calculation. In the big picture it rarely makes a material difference, but I always
wonder “is this the scenario where I would reach a different decision if I used
the right value?” You can avoid that particular nagging doubt by always using the
front-end loaded average. People who think “every line in this project is under
10 miles (16 km) so why bother?” are not wrong, and usually reach the same decision
that they would reach using the other equation. There is nothing “wrong” with
that thought process, as long as the thought process actually takes place—many
engineers have decided to use a simple average and have erased the possibility of
an alternative method from their minds, and approach a long line with the simple
average which results in density being low, velocity being high, and viscosity being
low; all by different amounts. For a gas, the end result is complex, but often results
in Reynolds number being low, which for modern pipe tends to put you in a steeper
portion of Fig. 4.1 and small changes in pressure can give you a sharp increase in
friction factor.

Figure 4.2. Comparison of average pressure methods (25 psi/mi pressure gradient).

> Read full chapter

Flow Behavior and Friction Characteris-

tics of Fluid Flow in Coiled Tubing
Zhongwei Huang, ... Subhash Shah, in Abrasive Water Jet Perforation and Mul-
ti-Stage Fracturing, 2018

5.2.2.2 Calculations for Fluid Flow in Turbulent Flow Regime

in Smooth and Rough Coiled Tubing
The pioneering work of Dean on the theoretical aspect of coiled tube flow prob-
lem has been significant. One important contribution of Dean's work was the
introduction of nondimensional parameter K, which was called Dean number. The
complexity of flow geometries and equations of fluid flow in coiled pipes attracted
the attention of theoreticians as well as engineers. Various theoretical methods have
been developed, which could be roughly grouped as analytical solutions for a small
Dean number, the numerical methods, and the boundary layer methods for high
Dean number.

The turbulent flow data analysis of Newtonian fluids in CT with roughness has been
a challenge. This is mainly because there is no published correlation on the effect of
tubing roughness on friction factor in CT. On the basis of their research work (Zhou
and Shah, 2006; Zhou, 2006), the following friction factor correlation for Newtonian
fluid flow in CT with roughness was proposed.

(5.5)

where frough is Fanning friction factor in rough CT, which is dimensionless; a/R is the
curvature ratio; a is the radius of CT in in.; R is the reel radius in in.; NRe is Reynolds
number, which is dimensionless; hr is the pipe roughness projection in in.; and d is
the pipe inner diameter, in.

The aforementioned correlation was derived based on the assumption that the
friction factor ratio in rough to smooth CT was approximately equal to the corre-
sponding ratio in ST. Although the accuracy of this assumption is yet to be verified,
the experimental data of the present study appear to indicate that Eq. (5.5) could
adequately describe the effect of tubing roughness on friction factor in CT, as
discussed later.

Certain correlations are utilized in the turbulent flow data analysis of Newtonian
fluid flow in straight rough pipes. One of these correlations was reported by Chen
(1979)

(5.6)

where hr is the tubing roughness projection and d is the tubing ID. The result of this
correlation was almost identical to Colebrook equation for the range of NRe from
4000 to 4 × 108 and hr/d ranging from 0.05 to 5 × 10−7. Since Chen equation was an
explicit equation, it has overcome the drawback of Colebrook equation, which was
implicit, and a trial-and-error method was required.

For the turbulent flow of a Newtonian fluid in smooth CT, Srinivasan correlation
(Srinivasan, Nandapurkar, & Holland, 1970) could be utilized

(5.7)

where a/R is the CT curvature ratio and a and R are the radii of tubing and reel drum,
respectively. Eq. (5.7) is valid for 0.0097 < a/R < 0.135. The Srinivasan correlation is
based on the flow tests of Newtonian fluids (water and oil) in smooth coiled pipes.

Figs. 5.10–5.13 present the plots of Fanning friction factor versus Reynolds number
for the flow data of water in 1-, 1½-, 1¾-, and 2⅜-in. ST and CT. Also on the plots
are Chen correlation of the friction factor for rough ST and Srinivasan correlation for
smooth CT. Several observations could be made and are discussed in the following
paragraphs.
Figure 5.10. Fanning friction factor versus Reynolds number of water in 1-in.
straight and coiled tubing (CT).

Figure 5.11. Fanning friction factor versus Reynolds number of water in 1½-in.

straight and coiled tubing (CT).

Figure 5.12. Fanning friction factor versus Reynolds number of water in 1¾-in.

chrome and carbon steel coiled tubing (CT).
Figure 5.13. Fanning friction factor versus Reynolds number of water in 2⅜-in.
straight and coiled tubing (ST and CT, respectively).

First, the friction factor in CT for all tubing sizes (except 1¾-in. tubing reels, which
had no corresponding straight sections) significantly exceeded the ST friction factor.
The maximum difference in friction factors could be as high as 30%. The extra
flow resistance in CT occurred, due to the secondary flow in curved flow geometry.
Furthermore, the extent of secondary flow effect was a function of curvature ratio.
This was evident by the comparison of friction factors of 500-ft and 1000-ft 1-in. CT,
as presented in Fig. 5.10. It was noted that 500-ft CT string was spooled on to a drum
of smaller diameter (48 in.) than 1000-ft CT string, which was spooled on to a 72-in.
diameter drum. The 500-ft CT reel had a higher curvature ratio (a/R = 0.0169) than
1000-ft CT reel (a/R = 0.0113). The difference in friction factors between 500- and
1000-ft CT reels could be as high as 8%. Also, for the CT reels with same curvature
ratio, the corresponding friction factors were approximately the same, as presented
in Figs. 5.11 and 5.13.

Second, Figs. 5.10 and 5.13 indicated that the friction factors in 1-in. and 2⅜-in.
CT reels were higher than Srinivasan correlation. Therefore, the tubing roughness
effect was evident in both CT sizes. To estimate the magnitude of possible tubing
roughness, the previously proposed correlation [Eq. 5.5] was applied to match the
friction factor data of 1-in. and 2⅜-in. CT. Fig. 5.14 presents 1-in. CT reel result.
It was observed that the proposed correlation was able to match the friction factor
data of 500-ft and 1000-ft CT with tubing roughness of 0.00012 and 0.00006 in.,
respectively. As indicated in Fig. 5.13, a tubing roughness of 0.00035 in. resulted in
quite good match of the friction factor data with Eq. (5.5). It was encouraging to find
out that these roughness values were quite similar to the roughness data measured
by the Tuboscope's Hommel surface roughness meter on similar CT. The measured
average roughness for a new carbon steel CT is between 0.000025 and 0.000149 in.
Figure 5.14. Friction factor versus Reynolds number of 1-in. tubing, measured and
predicted through proposed rough coiled tubing (CT) correlation.

It is important to point out that the newly proposed correlation (Eq. 5.5) not only
matched the majority of data on point-to-point basis, but also, more importantly,
provided a good match to the data trend. This has been proven for both 1-in. and
2⅜-in. tubing data sets, as presented in Figs. 5.13 and 5.14, respectively. In these
plots, it could be observed that Srinivasan correlation had a steeper slope than the
correlations of data points, which exhibited flatter slopes. The friction factor behavior
of Newtonian fluid in straight pipes was recalled, such as in Moody diagram of
the friction factor. The smooth pipe flow or “hydraulically smooth” pipe flow is
characterized by steeper slopes than the rough pipe flow, with a slope dependence
on the magnitude of Reynolds number. The flatter lines indicated higher dominance
of roughness at higher Reynolds numbers. The newly proposed correlation (Eq. 5.5)
was seen to be able to effectively adjust the slope of Srinivasan correlation to better
match the experimental data of CT with roughness.

It could be observed in Fig. 5.11 that the friction factor data of 1½-in. CT were quite
close to Srinivasan correlation. This does not necessarily mean that 1½-in. CT reels
were smooth. It was quite likely that the inner diameter of 1½-in. tubing had been
changed due to previous usage. Prior to this research project, 1½-in. CT reels had
been used extensively for testing fracturing slurries. These previous tests might have
altered the tubing inner diameter (enlarged due to slurry erosion). It would be very
tempting to have one parameter, either the tubing ID or roughness, independently
measured so that the other parameter could be estimated from the water test data
through proposed correlation in Eq. (5.5). In contrast, neither measurement was
available.

It was interesting to note that the friction factor data for 1¾-in. chrome CT were
lower than 1¾-in. carbon steel CT data. It was therefore believed that the chrome
tubing had lower roughness effects. The friction factor of the chrome tubing was
slightly lower than Srinivasan correlation. Through the new correlation (Eq. 5.5), the
roughness in 1¾-in. carbon steel CT was estimated at 0.00005 in.

> Read full chapter

Well Dynamics
David A. Simpson P.E., in Practical Onshore Gas Field Engineering, 2017

3.5.3.3 Evaporation as deliquification

If your pressure is low enough, then it is sometimes possible to evaporate all of
the liquid that flows into the well-bore. This technique works, but it requires you
to be willing/able to work under vacuum conditions and generally requires you to
remove production tubing from the well altogether. In a deep vacuum, none of the
incompressible-flow equations will have any relationship to observed data. Friction
factors based on the Colebrook equation (Chapter 4: Surface Engineering Concepts)
will not be valid, and using them will understate the friction by at least an order
of magnitude. Friction is still a strong function of velocity, but the relationship to
Reynolds number becomes very tenuous. Providing all the flow area possible (by
removing tubing) minimizes the velocity and therefore the friction.

A major concern is that evaporating the produced water will leave solids behind
that can plug the formation. There is no theory that says that this is unlikely, but
experience to date has not shown it to be a significant problem.

To get a feeling for the magnitude of inflow where this technique might be useful,
remember from Fig. 3.4 that as pressure comes down (for a given temperature),
the amount of water vapor that a gas can carry increases rapidly. At 106°F (41°C)
and 50 psia (345 kPa) 100% RH is 1096 lbm/MMSCF (17.5 gm/SCm). If you drop
the pressure to 3 psia (21 kPaa) the water content jumps to 17,980 lbm/MMSCF
(288 gm/SCm). The difference between those two data points is nearly 50 bbl/MM-
SCF (7.66 m3/SCm). There are a very large number of gas wells that make less than
50 bbl/MMSCF.

The most significant impediment of widespread use of this technique is simple fear
and superstition. This technique requires operating in vacuum conditions. As will
be discussed at length in Chapter 4, Surface Engineering Concepts, some very bad
science has led to widespread terror of oxygen-related corrosion modalities to com-
pliment widespread uneasiness that introducing oxygen could cause compressor
stations to explode. This uneasiness is unsupported in actual experience, but persists
anyway. Many jurisdictions have responded to the bad science by placing significant
regulatory impediments to operating gas wells below atmospheric pressure. Most
regulations provide processes to allow engineers to successfully make the case to
allow vacuum operations on a case-by-case basis, but the process is far from trivial
or automatic.

Colorado has regulations requiring permission to operate at a vacuum. One of my

clients elected to be the first company to attempt to acquire that permission on a
group of 22 CBM wells in La Plata County, CO (Fig. 3.27). Since the client was both
the producer and the gatherer, there were no gathering company objections to the
application. We prepared the application with the following proactive provisions:

Figure 3.27. 22 Wells on vacuum deliquification test.

• Each well on vacuum would have an on-site oxygen sensor and a slam valve
to shut the well in on high O2 (without specifying what “high” meant which
turned out to be important).
• At the aggregation points where the gas was delivered to a third party, there
would be additional O2 sensors connected to the station emergency shutdown
system.

The application was approved and equipment was ordered. Fig. 3.27 shows the
results of that test (in early 2010, both the company facilities engineer and company
production engineer retired and their replacements decided that the field needed to
go back to “tried and true” techniques and they installed tubing and pump jacks on
all of the wells, the results were disappointing). During the implementation stage
we learned:

•
 Setting the O2 sensors at a spike to 10 ppm resulted in most wells being shut-in•
most of the time during the start-up phase of the project (we reset them to
>25 ppm for 30 seconds which helped a lot).
All pressure safety valves (PSV) are designed to leak when downstream pressure•
is greater than upstream pressure. We fixed this by putting rupture disks (best)
or check valves (not as effective, but nearly) under the PSV.
Sight glass packing nearly always leaks and the leaks can be very hard to find.•
We replaced sight glass packing, but eventually just isolated the sight glasses
out of service.
Finger-tight plugs in open-ended tubing are far worse than worthless (a visual•
inspection shows the line as plugged, but the plug doesn’t really do anything).
Threaded connections often leak, and leaks can be difficult to find.

We were able to solve most of the issues during the vacuum start-up, and had them
all resolved within 6 months. I visited the field in early 2009 after 4 years of operation
and talked to the field operator (who I didn’t know) and asked him how often the O2
sensors tripped. He looked at me funny and said “do you know how to reset them?”
Meaning, of course, that in the year he had been operating the wells none of them
had ever tripped and he lived in fear of one ever tripping. I showed him how to reset
them.

> Read full chapter

ScienceDirect is Elsevier’s leading information solution for researchers.

Copyright © 2018 Elsevier B.V. or its licensors or contributors. ScienceDirect ® is a registered trademark of Elsevier B.V. Terms and conditions apply.