premium digest december 2010 inspection of gas pipelines with ultrasonic measurement techniques -...
TRANSCRIPT
-
7/27/2019 Premium Digest December 2010 Inspection of Gas Pipelines With Ultrasonic Measurement Techniques - Practical A
1/5
5 PiPelines international digest | deCeMBer 2010
integritY
Inspection of gas pipelines withultrasonic measurement techniques:
practical aspectsBy Dr Jochen Stratmann, Open Grid Europe GmbH, Pipeline Engineering Section (TLE), Essen, Germany
Pipeline inspection tools based on ultrasonic measuring techniques are requently used or the assessment o oilpipeline integrity. Due to the requirement o a liquid being present between the detection device and the pipe wall,this technique is not readily applicable to gas pipelines. Recently, the inspection o gas pipelines with ultrasonicdevices has been conducted successully by completely flling the pipeline with a liquid (or example, water) beorethe inspection. As an alternative, a short section (batch) o liquid may be run through the pipe in order to avoidcompletely flling it. The liquid is separated by pigs rom the surrounding gas in the pipeline and the measurementdevice is kept within the liquid-flled section. Batches may, however, assume considerable velocities in the event o
large slopes along the path o the pipeline, and this may inuence the probability o detection and, more practically,it constitutes a risk o pipeline damage i velocities become too high. This article ocuses on the method o calculatingthe batch velocity, predicting water batch movements in pipes as well as other practical aspects.
Ultrasonic detection devices employed by in-line
inspections (ILI) tools are used for the detection of
pipeline defects such as corrosion, grooves, laminations,
or cracks. These tools enable the detection of defects which
cannot be achieved using other techniques such as magnetic-
ux leakage (MFL), especially with respect to cracks due to
fatigue or stress corrosion. However, the contact between the
ultrasonic detector and the pipe internal wall needs to be
established through an incompressible liquid, which rules outstraightforward use of these devices in gas pipelines.
Filling the pipeline with a liquid before or while conducting
the pigging provides the most intuitive solution to this
challenge. However, typical sections of large gas pipelines may
contain volumes of several tens of thousands of cubic metres.
Any liquid employed for this use will experience a certain
degree of contamination through the contents of the pipe,
which may be a result of compressor station oil losses or the
higher hydrocarbons contained in the gas itself. The disposal or
the decontamination of the liquid may be extremely expensive,
a condition which gives the pipeline operator reason to look for
an alternative.
Additionally, it presents a considerable challenge to obtain
public and/or private permits to obtain water from rivers
or lakes. The public water supply is generally insucient
regarding capacity. To transport the water using temporary
above-ground water lines to one of the sites where pig traps
are located may also constitute a handicap. Furthermore, thedisposal of the water after its use, even after it being cleaned
using activated carbon, may not always be allowed by local
authorities.
Symbol Property Unit
a speed of propagation m/s
di internal pipe diameter m
E Youngs modulus N/m2
k compressibility of liquid l/Pa
l batch length m
ow friction coecient
ppressure dierences
(e.g. from ow friction)Pa
liquid density kg/m3
s wall thickness m
v velocity dierence m/s
Figure 1: Example of a liquid batch set-up.
Figure 2: Example of general set-up and boundary conditions (top)
and photo of mobile air compressors (bottom).
-
7/27/2019 Premium Digest December 2010 Inspection of Gas Pipelines With Ultrasonic Measurement Techniques - Practical A
2/5
6 PiPelines international digest | deCeMBer 2010
integritY
Set-up and goalsDue to the above-mentioned conditions, pigging of the gas
pipeline using short liquid batches, of only several hundred
metres length, is desirable. In this case, the batch consists of a
liquid-lled section within the pipe while the liquid is separated
from the gas by means of pigs at both ends. Figure 1 shows such
a set-up.
Depending on the pressure in front of and behind the batch
(i.e. the boundary conditions), the batch may move more or less
freely within the pipeline. In addition, the height prole of the
pipeline inuences this movement.
The set-up under consideration here is a short water batch
with a length of approximately 600 m. The gas volume behind
the batch is fed with air using multiple compressors in
continuous or intermittent operation. Therefore, the amount
of air in this section continuously increases; the pressure,
however, varies depending on the position of the batch. The
gas volume in front of the batch is vented to the atmosphere in
such a fashion that the gas pressure in this volume remains at a
constant level, while the volume continuously decreases as the
batch moves in forward direction. Figure 2 illustrates this set-
up, the questions associated with which may be subdivided into
the following aspects:
Operational
There is a need to calculate the pressure needed to drivethe batch through the pipeline considering the local height
dierences.
The time at which the batch will be expected at a certain
location within the pipeline needs to be known for general
planning and for positioning pig detection teams along the
pipeline if pig detection is done throughout the inspection at
given locations.
Measurement technique The ultrasonic device is operated at a certain detection
frequency at which the measurement is conducted. If the
velocity of the batch exceeds a certain value, for example1.5 m/s, the probability of detection for a specic size of crack
will become too low to be acceptable for the assessment of
pipeline integrity. Therefore, it becomes mandatory to nd a
means of estimating the expected velocity for each location
along the pipe. Exceeding the critical velocity with respect
to detection in a small fraction of the total pipeline may be
considered acceptable.
Pipeline integrity and safety aspects Due to the density of water, a considerable hydrostatic
pressure may act within the pipe. Naturally, the pressure in
the pipe must not exceed the design pressure. This aspect
is less relevant to short batches, but may become critical for
batches of lengths in excess of a thousand metres depending
on the height prole.
At certain conditions discussed below, the batch may
separate into two (or more) parts. This is entirely undesirable
and induces the risk of a water hammer which may occur
if two of these parts collide in the pipe at a considerable
velocity. To avoid the batch separation, a certain pressure
needs to be maintained within (and at the front) of the batch.
Also, the question arises which collision velocity leads to a
critical pressure shock in the pipeline.
At high velocities, the batch may cause a considerable
(centrifugal) force to act on bends of the pipe if the pipe is
not straight at locations characterized by high velocities,
i.e. downhill.
In summary, it appears vital to rst calculate the pressure
behind and in front of the batch as well as its velocity. From these
parameters, conclusions regarding the integrity/safety aspects
may consequently be drawn.
Pressure calculationIn the following, the batch is considered merely as an amount
of water within the pipe. The mass and volume of the individual
pigs is not taken into account, since it is considered to result in a
similar density as an equivalent section of water or less, and thus
to behave accordingly.
Three dierent characteristic pressures are discussed in this
section. These are:
The maximum pressure that occurs within the section in
which the batch is located: this pressure must not exceed the
maximum operating pressure/the design pressure of the pipe.
The minimum pressure to be maintained in front of the batch:
the reason why a certain pressure must be maintained infront of the batch is that the pressure at the highest point
within the batch (the top in Figure 3) must not become
lower than the vapour pressure of the liquid. If this is the
maximum pressure
600 m batch
minimum pressure in front of batch
pmax = 8.9 bar
pdrive,max = 23.6 bar
pmax = 23.6 bar
pfront = 10.0 bar
600 m batch
driving pressure
Figure 3: Illustration of relevant height dierences.
Figure 4: Result for the three dierent characteristic pressures for
a batch length of 600 m. The maximum pressure and the driving
pressure are calculated on the basis of a pressure in front of the
batch of 10 bar (abs). The height prole is given in each graph to
enable comparison. The location is dened as the position of the
front of the batch, see Figure 3.
-
7/27/2019 Premium Digest December 2010 Inspection of Gas Pipelines With Ultrasonic Measurement Techniques - Practical A
3/5
7 PiPelines international digest | deCeMBer 2010
integritY
case, the liquid will become gaseous at this location and,
consequently, the batch will fall apart into two separate
parts. It is vital to avoid this, as it gives rise to the danger of
a so-called water hammer if the two parts later collide in the
pipeline. Even a low velocity may generate a high-pressureshock. The pressure caused by the collision is discussed in
one of the following sections.
The pressure needed to drive the batch through the pipeline
(driving pressure, pressure behind the batch): this pressure
must be generated by one or more compressors. It depends
on the location of the batch within the height prole. This
aspect may be critical as to the duration of the pigging
process if a high pressure is needed in long sections of the
pipeline. If lling this volume at high pressure takes too long,
the pig battery capacity may become critical.
The maximum pressure in the section of the pipeline filled
by the batch exists at the very bottom of this section. It is
influenced by the hydrostatic pressure resulting from the
height profile as well as the pressure in front of the batch.
Figure 3 shows an example of a height profile which includes
an indication of the section filled with water. At this batch
location, the maximum pressure occurs at the very front and
is therefore identical to the gas pressure in front of the batch
if pig friction is disregarded. It must be ensured that this
pressure does not exceed the pipe design pressure. It must
therefore be calculated for all possible batch locations. A result
in Figure 4 shows that, for the case in hand, the maximum
pressure does not constitute a critical issue. The maximum
pressure aspect is most critical for long batches and large
height differences.The minimum pressure in front of the batch is calculated on
the basis of the following scenario: At any time or location,
the gauge pressure in front of the batch must, as a minimum,
match the hydrostatic pressure difference between the top
and front locations as indicated in Figure 3. This ensures
that the pressure at the top location never becomes lower
than 1 bar and therefore not lower than the vapour pressure
of water. This avoids the vaporisation of water at this point
at low pressure. Naturally, a certain safety margin is added
on top of this pressure. Since the height difference between
the two locations varies significantly with the position of the
batch, it must be calculated for each batch location. To ensuresafe operation of the batch run, it is easiest to maintain a
constant pressure in front of the batch as indicated above in
the section Set-up and goals. However, in some instances it
may be appropriate to vary this pressure, for example to avoid
an unnecessary operation of the compressors to create a high
driving pressure or to avoid a critical level of pressure at the
lowest point in the batch with respect to the design pressure of
the pipe. Figure 4 shows the calculation results.The driving pressure may be estimated by adding the
hydrostatic pressure dierence between front and rear of the
batch as indicated in Figure 3 to the gas pressure in front of the
batch as well as an equivalent pressure which corresponds to
the pig friction. In the set-up discussed here, this pressure must
be generated by means of air compressors as shown in Figure
2. Figure 4 shows that, due to the necessary pressure in front of
the batch of 10 bar and the pig friction, the maximum driving
pressure gets close to 25 bar even for the short batch of 600 m
length and this height prole example.
An alternative to calculate the driving pressure is to use a mass
balance of gas (air) in the volume behind the batch. Here, the
mass ow rate of the compressors used to feed the gas/air to this
volume must be considered as well as the actual, time-dependent
location of the batch. The time-dependent location of the batch
within the pipe is known after solving the problem described in
the following section.
Batch movement predictionand velocity calculation
As stated above, the presence of pigs within the water batch
is disregarded and the batch is treated as one continuous
section of water. To calculate the velocity, a balance of forces
acting on this lumped mass of water is considered.
Primarily, the movement of the batch is inuenced by thepressure behind the batch (pressure 1) and in front of the
batch (pressure 2), see Figure 5. These pressures depend on the
location of the batch within the pipe and on time (for example,
a compressor continuously feeds gas to the gas volume behind
the batch). As an example, moving the batch to the right-hand
side in Figure 5 will increase pressure 2 and decrease pressure
1. Simultaneously, a compressor feeding gas into the section
behind the batch increases pressure 1; venting gas from the
volume in front of the batch (in the direction of movement) will
decrease pressure 2.
In addition, Figure 5 displays the height prole of a gas
pipeline as an example. Due to the slopes associated with thisprole, a gravitational force acts on the batch. Therefore, this
force depends on the batch location (intuitively, downhill
means fast).
Figure 5: Forces acting on the batch. Figure 6: Example of height prole and velocity.
-
7/27/2019 Premium Digest December 2010 Inspection of Gas Pipelines With Ultrasonic Measurement Techniques - Practical A
4/5
8 PiPelines international digest | deCeMBer 2010
integritY
The force caused by friction between pig and pipe wall also
needs consideration. This force may be assumed to be constant
as it depends mainly on the geometry of pig and pipe, thematerials employed and other parameters, but not on the
location. A pig in a dual-diameter pipeline may, however, behave
dierently. Here, the friction force may be seen as constant for
each section with a constant diameter. When compared to a
pressure force, this force typically is the equivalent of 1 bar per
pig or less.
The force which has the largest inuence on the maximum
batch velocity is, however, the friction force caused by the ow
in the liquid-lled sections between the pigs. If the local ow
eld at the ends of these sections near the pigs where the ow
is reversed is disregarded (long batch), this ow has normal
turbulent pipe ow characteristics. This means that it causes a
pressure loss equivalent to such a ow. This pressure dierence
(loss) acts as a force which decelerates the batch. Equation 1
gives the relationship:
p = l/di p/2 v2 (1)
This pressure dierence is calculated using a constant ow
friction coecient of = 0.01. For other cases, this coecient is
readily obtained from textbooks on hydrodynamics.
As is typical with pipe ow, the force increases with the square
of the velocity. It is therefore most inuential at such conditions
where the velocity is high, i.e. downhill, and acts as a natural
limiter for the maximum velocity. If it is intended to reducebatch velocities, an appropriate means is to use longer batches.
This is counter-intuitive, as one would assume longer batches to
move downhill faster than short ones. Nevertheless, using this
as a remedy causes a higher hydrostatic pressure within the pipe
and creates the need for more water. The force balance results in
an ordinary dierential equation (second order) which is solved
using the numerical methods in MATLAB. The following section
shows the results.
Velocity prediction results and discussionIf applied to the given height prole example, the velocity
calculation using the method described above results in avelocity prole as shown in Figure 6. It becomes obvious that a
downhill scenario results in a high batch velocity. Due to this
height prole, the velocities predicted by the calculation are,
however, too high for the use of ultrasonic measuring devices.
Therefore, a pipeline with such a height prole must be subject
to a complete ll or a very long batch.Figure 7 shows the comparison between the calculation result
(blue line) and a measurement of the velocity of one of the pigs
from a real batch run, measured using a common odometer.
An acceptable degree of agreement between prediction and
measurement of the velocity is achieved.
At some of the locations, primarily those with high velocities,
the measurement shows higher velocity values than predicted.
These values are considered a measurement error due to the
behaviour of the odometer at high velocity for which it is not
designed. This assumption will be veried in the future by using
more advanced devices for measuring the pig velocity.
Additionally, the measurement result may be the consequence
of pig behaviour if gas is trapped within the batch causing an
oscillation with high peak velocities of the individual pigs. This
aspect will also be addressed in future simulations by solving
the coupled dierential equations.
Pipeline integrity and saety aspects
Forces in pipe bends at high velocitiesThe forces acting on pipe bends are the result of the change
of momentum of the liquid ow through the bend (centrifugal
force). The total force depends on the ow velocity and the pipe
cross-section area, but not on the radius of the bend. However,
at larger radii, the force spreads over a larger area. Here, thecentrifugal force acts in addition to the force resulting from
the pressure inside the curved pipe. Figure 8 shows the total
force from both pressure and change of momentum. It becomes
obvious that in the velocity range up to 20 m/s, pressure
forces dominate. Also, the centrifugal force at a velocity of
approximately 35 m/s has the same value (approximately
900 kN) as a pressure force resulting from a pressure of
10 bar. Due to this equivalence, centrifugal forces do not play a
signicant role, if a velocity prole according to Figure 6 with a
maximum velocity of 20 m/s is expected.
Batch collision and water hammerDiscussion of the water hammer problem is needed to assess
the consequences of a batch collision in the pipe as a what-
if scenario. At the location of the collision, the pressure is
Figure 7: Comparison between prediction and measurement.Figure 8: Total force on a pipe bend is dependent on pressure and
batch velocity.
-
7/27/2019 Premium Digest December 2010 Inspection of Gas Pipelines With Ultrasonic Measurement Techniques - Practical A
5/5
9 PiPelines international digest | deCeMBer 2010
integritY
increased immediately to a value described by the Joukowsky
equation. This pressure may well exceed the pipe design
pressure:
p = p a v (2)
This pressure shock then propagates at the speed of wave
propagation along the pipe in the backward direction until theend of each of the two batches is reached. At the end, the pressure
dierence reverses its sign (reection at open end) and the shock
front begins to propagate in the forward direction. The pressure
level in the batch may become extremely low during this process
and may well cause the liquid to vaporise if the pressure falls
below the vapour pressure.
Both the pressure increase and the velocity of propagation
of the shock wave depend on the compressibility of the water
and the elasticity of the pipe. Figure 9 shows the solution of the
Joukowsky equation for three dierent cases depending on the
relative collision velocity. If the compressibility of water is taken
into account and the pipe is regarded as entirely rigid, even low
collision velocities lead to critical pressure levels since water
reacts very stiy. Taking into account the elasticity of the pipe,
a much higher velocity is needed to create pressure levels critical
for a pipeline. It must be stressed that this pressure must be
added to the initial pressure in the pipe.
SummaryPractical aspects of relevance to gas pipeline inspection
using pigs which employ ultrasonic measurement techniques
are discussed in this article. To warrant the coupling between
measurement device and pipe wall, it appears advantageous
to move short sections of liquid (batches) through the pipe in
which the measurement device is contained. Using this approach,
the velocity of the batch in the pipeline needs to be predicted
and certain characteristic pressures in the system have to be
calculated. Consequently, key safety aspects are discussed on the
basis of the results.
A simple approach for the prediction of the velocity of liquid
batches of any length in pipelines is developed. The comparison
with measurements indicates that the approach is suitable
for the prediction of the velocity as it occurs in the eld. If theheight prole of the pipeline exhibits large slopes, the velocity
is expected to exceed a critical limit up to which the detection
of pipe faults is possible. Hence, it is suitable for pipelines with
minor height dierences only. The approach also enables the
prediction of the location at which the batch is found in the
pipeline over time.
Another vital aspect is to ensure the integrity of the batch by
maintaining a certain location-dependent pressure in front of the
batch. The collision of two water batches in a pipeline results in a
pressure level in excess of the design pressure if a certain velocity
is exceeded. This critical velocity depends on the properties of the
pipe.
In addition, due to the high density of the liquid, it must be
ensured that the maximum (hydrostatic) pressure in the pipe
does not exceed the design pressure. This aspect is important for
longer batches of several thousand metres length.
The ow of water in sharp bends of the pipe at high velocity is
found to be of minor importance when compared to the eects of
internal pressure on the pipe.
Figure 9: Pressure increase vs relative (collision) velocity, where:
k = 4.8e-10 1/Pa (compressibility of liquid)p = 988 kg/m3 (liquid density)
di = 0.9 m (internal diameter)
s = 12.6 mm (wall thickness)
E = 206,000 MN/m2 (Youngs modulus)
This paper was presented at the Evaluation, Rehabilitation
and Repair of Pipelines Conference held in Berlin, Germany,
in October 2010, and organised by Tiratsoo Technical (a
division of Great Southern Press) and Clarion Technical
Conferences, Houston.