Coastal Engineering 42 2001. 313335 www.elsevier.

comrlocatercoastaleng

Onset of scour below pipelines and self-burial

B.M. Sumer ) , C. Truelsen, T. Sichmann, J. Fredse
Department of Hydrodynamics and Water Resources (ISVA), Technical Uniersity of Denmark, 2800 Lyngby, Denmark Received 3 April 2000; received in revised form 19 October 2000; accepted 8 November 2000

Abstract This paper summarizes the results of an experimental study on the onset of scour below and self-burial of pipelines in currentsrwaves. Pressure was measured on the surface of a slightly buried pipe at two points, one at the upstream side and the other at the downstream side of the pipe, both in the sand bed. The latter enabled the pressure gradient which drives a seepage flow underneath the pipe. to be calculated. The results indicated that the excessive seepage flow and the resulting piping are the major factor to cause the onset of scour below the pipeline. The onset of scour occurred always locally but not along the length of the pipeline as a two-dimensional process.. The critical condition corresponding to the onset of scour was determined both in the case of currents and in the case of waves. Once the scour breaks out, it will propagate along the length of the pipeline, scour holes being interrupted with stretches of soil span shoulders. supporting the pipeline. As the span shoulder gets shorter and shorter, more and more weight of the pipeline is exerted on the soil. In this process, a critical point is reached where the bearing capacity of the soil is exceeded general shear failure.. At this point, the pipe begins to sink at the span shoulder self-burial.. It was found that the self-burial depth is governed mainly by the KeuleganCarpenter number. The time scale of the self-burial process, on the other hand, is governed by the KeuleganCarpenter number and the Shields parameter. Diagrams are given for the self-burial depth and the time scale of the self-burial process. q 2001 Elsevier Science B.V. All rights reserved.
Keywords: Currents; Onset of scour; Pipeline; Scour; Self-burial; Waves

1. Introduction If the initial embedment of a pipeline laid on a seabed is not very large, and the flow induced by currentsrwaves. is sufficiently strong, the bed may be washed away underneath the pipe, the onset of scour it may be noted that the bed may not be washed away underneath the pipe, and yet some slight scour may occur at the pipeline. In this paper, however, the term Aonset of scourB will be used for
)

Corresponding author. Fax: q 45-45-932860. E-mail address: sumer@isva.dtu.dk B.M. Sumer..

the case when the bed is washed away underneath the pipe.. The onset of scour is basically related to the seepage flow in the sand beneath the pipeline, which is driven by the pressure difference between the upstream and downstream sides of the pipe. The critical conditions for the onset of scour have been studied by Mao 1986., Chiew 1990., Sumer and Fredse 1991. and Klomp et al. 1995.. Mao 1986. has described the role of vortices that form in front and at the rear of the pipe. He has also discussed the seepage flow underneath the pipe in relation to the onset of scour. The latter has been further elaborated by Chiew 1990.. The latter author

0378-3839r01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 3 8 3 9 0 0 . 0 0 0 6 6 - 1

314

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

has also linked the onset of scour to the process of piping. Although the previous work has given a considerable insight into the process of onset of scour, the precise impact of the above mentioned pressure difference on the soil behaviour was not fully described. Another problem regarding the onset of scour concerns the critical condition defining the onset of scour. Sumer and Fredse 1991. conducted experiments to determine the critical conditions in the case of waves, and expressed it in terms of two parameters, namely the KeuleganCarpenter number, KC, and the initial embedment-to-diameter ratio, erD. Klomp et al. 1995. later extended the Sumer and Fredse 1991. study to the case of combined waves and current. However, no study is yet available investigating this matter for the case of currents, a common case which is of large practical importance, considering pipelinesrcables laid on a riverrsea strait bed. The first part of the present study addresses the two issues mentioned in the preceding paragraphs, namely: 1. the mechanism of the onset of scour, and the role of the pressure gradient in the latter process in wavesrcurrents; and 2. the critical condition for the onset of scour incurrents. Once the scouring commences, it will propagate along the length of the pipeline, as sketched in Fig. 1. A three-dimensional scour pattern emerges in which the scour holes are interrupted by stretches of

soil, called span shoulders, where the pipe obtains its support, section A-A in Fig. 1a. Various modes of self-burial of the pipe may occur, depending on the flow, the soil, and the pipe stiffness: 1. Scour, sagging, backfilling and eventual selfburial of the pipeline between span shoulders. 2. The soil supporting the pipeline may fail due to liquefaction, leading to the self-burial of the pipeline. 3. The self-burial of the pipeline occurs at span shoulders due to the so-called general shear failure. The first case has been investigated by Fredse et al. 1988.. Various accounts of the spreading process the spreading of scour along the length of the pipeline. have been given in Leeuwenstein et al. 1985., Bernetti et al. 1990., and Hansen et al. 1991, 1995.. The second case has been investigated by Sumer et al. 1999., and various quantities such as the sinking depth, the time scale, the influence of the pipes specific gravity, the influence of wave characteristics, etc. have been discussed. It was demonstrated that a pipe initially sitting on the bed could sink to a depth of 22.5 D in a soil confined with an impermeable base below. The third case, i.e. the self-burial of pipelines at span shoulders, has been investigated by Sumer and Fredse 1994.. However, in this latter work, the flow environment was limited only to steady current. In the second part of the present study, attention is concentrated on the self-burial of pipelines at span shoulders in waves. It turns out that the self-burial depth is a function of KC, and the variation of the self-burial depth with KC is the same as that for the scour depth for a fixed pipeline.

2. Experimental set-up Two kinds of experiments were conducted: 1. The experiments related to the onset of scour; and 2. those related to the sinking of pipeline at span shoulders.

Fig. 1. General scour picture around a pipeline.

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

315

2.1. Experiments related to onset of scour Two kinds of experiments were carried out: 1. current experiments; and 2. wave experiments. The current experiments were conducted in an open flume, 2 m in width, 0.5 m in depth and 23 m in length. The water depth was maintained at 0.30 m. A 5.5-m-long and 0.10-m-deep sand-bed section was established in the flume, protected at two ends by sections of crushed stones 3.5 cm in size. with 1:10 slope. The upstream end of the sand-bed section was 12 m from the inlet section of the flume. The test section itself was 3 m from the upstream end of the sand-bed section. A vertical, guiding wall made of plywood. divided the working section into two parts along the length of the sand-bed section so that one section had the same width as the length of the test pipe in the wave flume, namely 0.6 m. Two pipe sizes were used in the experiments, diameter D s 10 and 5 cm. The pipe was rigidly fixed in this section to the side wall of the flume at one end, and to the guiding wall at the other. The pipe surface was smooth except in one test where the surface of the 10-cm diameter pipe was roughened, as will be detailed later. The junction between the pipe and the side wall may be a critical section for the onset of scour. This is because a half horseshoe vortex may form due to the separation of the boundary layer on the side wall, and cause scour. To avoid this, the upstream part of the junction between the pipe and the side wall was filled with sand with an extent of about one diameter, both along the length of the pipe and along the side wall., moulded in the form of a streamlined surface. This prevented the scour at the two ends of the pipe. The wave experiments were carried out in a flume, 0.6 m in width, 0.8 m in depth and 26.5 m in length. The water depth was maintained constant at 0.33 m. Monochromatic waves were produced by a pistontype wave generator. Similar to the current experiments, a 0.10-m deep sand section was established in the flume, 3-m-long, protected at the ends by crushed stones. The offshore end of the sand section was 11 m from the wave generator. The test section was halfway through the length of the sand section. The pipe was rigidly fixed to the two side walls of the flume. A wave absorber at the onshore side of the wave flume was used to minimize the reflection. The

flow velocity was measured by a bi-directional micropropeller. The pipe was equipped with two pressure tappings, 5 mm in diameter and covered with 40 mm nylon filters, 328 apart, as sketched in Fig. 2. They were connected to pressure transducers. The pressures were recorded automatically at a sampling frequency of 30 Hz. The length of recording was O 10 s. for the current experiments, and O 30 s. for the wave experiments, corresponding to the length of time from the start of the flow to the instant when the scouring commences. The purpose of the pressure measurement was to obtain the pressure gradient that causes the seepage flow. at the instant of the onset of scour, as will be detailed later in the paper. In these measurements, the pipe was slightly buried with a burial depth of e s 0.64 cm, as sketched in Fig. 2. Two kinds of sand were used in the experiments: d 50 s 0.18 mm and geometric standard deviation sg s d 84 rd16 .1r2 s 1.2, and d 50 s 1.25 mm and sg s 1.2. In the wave experiments, flow visualization tests were also made. For this, a laser sheet of light scanned the experimental section vertically, and the flow was made visible with the sand itself. The test conditions regarding these experiments are given in Table 1.

Fig. 2. Set-up for the pressure measurement.

316

Table 1 Test conditions for the onset of scour Test number Flow depth, h cm . Burial depth, erD Pipe diameter, D cm . Wave period, Tw s. U or Um , cmrs. Bed friction velocity Uf or Ufm , cmrs. Sand size, d 50 mm . Shields parameter, u Pipe Reynolds number, Re 0 5.0 = 10 4 U 2 .r g s y 1 . 1 y n . D . or 2 Um .r g s y 1 . 1 y n . D . 0 0.30 Keulegan Carpenter number, KC Remarks

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

(a ) Steady current experiments smooth pipe O1.1 30 0.064 10 O1.12

Increased gradually from 0 up to about 50 22 26 28 33 56.5 72.5 27 37 41 54 58 66 70

0 2.5

0.18

0 0.20

O2 O3 O4 O5 O6 O7 O8 O9 O10 O11 O12 O13 O14

30 30 30 30 30 30 30 30 30 30 30 30 30

0.01 0.02 0.03 0.05 0.10 0.15 0.04 0.07 0.10 0.08 0.10 0.125 0.15

10 10 10 10 10 10 5 5 5 10 10 10 10

1.0 1.2 1.3 1.5 2.6 3.3 1.3 1.8 2.0 3.2 3.4 3.9 4.2

0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 1.25 1.25 1.25 1.25

0.03 0.05 0.06 0.08 0.23 0.38 0.06 0.12 0.14 0.05 0.06 0.08 0.09

2.2 = 10 4 2.6 = 10 4 2.8 = 10 4 3.3 = 10 4 5.7 = 10 4 7.3 = 10 4 1.4 = 10 4 1.9 = 10 4 2.0 = 10 4 5.4 = 10 4 5.8 = 10 4 6.6 = 10 4 7.0 = 10 4

0.06 0.09 0.10 0.14 0.42 0.69 0.17 0.35 0.44 0.38 0.44 0.57 0.64

Pressure measurements at the critical conditions for the onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour

(b) Steady current experiments. Rough pipe k s r D s 6.0 = 10 y 2 O15 30 0.10 10.0 76

3.4

0.18

0.40

6.0 = 10 4

0.72

Critical condition for onset of scour

(c ) Wae experiments O16 33 0.064

10

45

2.4

0.18

0.19

4.5 = 10 4

0.27

18

O17 O18 O19 O20 O21 O22 O23 O24 O25 O26 O27 O28 O29 O30 O31

33 33 33 33 33 33 33 33 33 33 33 33 33 33 33

0.07 0.01 0.02 0.10 0.03 0.01 0.07 0.05 0.04 0.10 0.10 0.10 0.10 0.15 0.15

10 10 10 10 10 10 10 10 10 10 5 5 5 5 5

1.8 5 4.8 5.2 4 7.4 3.2 4.3 6.9 6.3 2.6 4.4 4.8 4.3 3.6

12.0 9.0 10.0 14.1 20.0 13.2 30.5 26.0 21.6 25.0 13.0 21.0 25.0 25.3 40.4

1.5 1.0 1.1 1.2 1.1 1.1 2.1 1.8 1.4 1.6 1.4 1.6 1.7 1.7 2.3

0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18 0.18

0.08 0.04 0.04 0.05 0.04 0.04 0.15 0.11 0.07 0.09 0.07 0.09 0.10 0.11 0.18

1.2 = 10 4 9.0 = 10 3 1.0 = 10 4 1.4 = 10 4 1.0 = 10 4 1.3 = 10 4 3.0 = 10 4 2.6 = 10 4 2.1 = 10 4 2.5 = 10 4 6.5 = 10 3 1.1 = 10 4 1.3 = 10 4 1.3 x 10 4 2.0 = 10 4

0.02 0.01 0.01 0.03 0.05 0.02 0.12 0.09 0.06 0.08 0.04 0.17 0.12 0.17 0.43

2.2 4.5 4.8 7.3 8.0 9.8 9.8 11.0 14.9 15.6 6.8 19.0 24.0 21.8 29.0

Pressure measurements at the critical conditons for the onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour Critical condition for onset of scour

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335 317

318

Table 2 Test conditions for self-burial of pipelines. Sand size was d 50 s 0.18 mm and angle of internal friction was w s 438 Test number Flow depth, h cm. Pipe diameter, D cm. Wave period, Tw s. Wave height, H cm. U or Um cmrs. Bed friction velocity Uf or Ufm cmrs. 1.7 Shields parameter, u 0.09 Pipe Reynolds number, Re 4.5=10 4 2.0=10 4 2.5=10 4 6.5=10 3 9.1=10 3 1.0=10 4 1.2=10 4 1.4=10 4 1.3=10 4 1.6=10 4 1.6=10 4 1.6=10 4 1.6=10 4 1.6=10 4 1.4=10 4 1.5=10 4 1.6=10 4 1.9=10 4 2.1=10 4 2.1=10 4 2.4=10 4 2.5=10 4 2.4=10 4 4.4=10 3 6.0=10 3 8.0=10 3 1.0=10 4 KeuleganCarpenter number, KC Pipe specific gravity sp

(a) Steady current experiments S1 25 10 (b) Wae experiments S2 33 S3 33 S4 33 S5 33 S6 33 S7 33 S8 33 S9 33 S10a 33 S10b 33 S10c 33 S10d 33 S10e 33 S11 33 S12 33 S13 33 S14 33 S15 33 S16 33 S17 33 S18 33 S19 33 S20 33 S21 33 S22 33 S23 33

45

3 B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

10 10 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 2 2 2 2

1.5 1.5 2 2 2 2 2 2.5 2.5 2.5 2.5 2.5 2.5 3 3 3 3 3 3 3 3 3.5 3 3 3 3

10.0 11.7 6.4 6.9 10.1 10.3 11.6 12.1 14.6 14.6 14.6 14.6 14.6 10.5 12.1 12.6 14.3 15.2 15.7 16.3 17.2 18.5 8.2 9.3 13.6 16.6

19.6 25.3 13.0 18.2 20.6 23.1 28.0 26.5 31.3 31.3 31.3 31.3 31.3 27.1 30.5 32.0 37.2 41.3 42.3 47.1 50.0 48.1 22.2 30.0 40.1 50.2

2.0 2.3 1.5 1.8 1.9 2.0 2.2 2.0 2.2 2.2 2.2 2.2 2.2 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.5 1.8 2.1 2.4 2.7

0.14 0.18 0.08 0.11 0.13 0.14 0.17 0.14 0.17 0.17 0.17 0.17 0.17 0.14 0.15 0.16 0.19 0.21 0.21 0.23 0.25 0.22 0.11 0.15 0.20 0.25

3 4 6 7 8 9 11 12 16 16 16 16 16 16 18 19 22 25 25 28 30 34 22 45 60 75

3 3 3 3 3 3 3 3 1.25 2 3 4 6 3 3 3 3 3 3 3 3 3 3 3 3 3

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

319

In Table 1, U is the undisturbed flow velocity at the top of the pipe in the case of the steady current, Um is the maximum value of the undisturbed, orbital velocity of water particles at the bed, Uf is the undisturbed friction velocity in the case of the steady current, and Ufm is the maximum value of the undisturbed friction velocity in the case of waves, u is the Shields parameter defined by

us

Uf2 g s y 1 . d 50

2.1 .

in which s is the specific gravity of sand grains, g is the acceleration due to gravity. In the case of the waves, Uf is replaced by Ufm . Also, in Table 1, Re is the Reynolds number Re s UD

2.2 .

the velocity U is replaced by Um for the case of the waves., and KC is the KeuleganCarpenter number KC s Um Tw D

2.3 .

in which Tw is the wave period. Furthermore, k s in Table 1 is Nikuradses equivalent sand roughness. The porosity of the sand used in the tests was n s 0.53 for d 50 s 0.18 mm sand and n s 0.47 for d 50 s 1.25 mm sand. Finally, as seen from Table 1, almost all the tests correspond to the live-bed conditions, i.e. u ) ucr . 2.2. Experiments related to sinking of pipeline at span shoulders Two kinds of experiments were carried out: 1. current experiments; and 2. wave experiments. The current experiments were conducted in the same current flume as that for the onset-of-scour experiments with the same test setup but without the splitter wall, yielding a width of the test section of 2 m. The mean water depth was 0.25 m. An aluminium, 10 cm diameter pipe with hydraulically smooth surface was used in the current experiment. It was 1.98 m in length, with a 1-cm gap between the pipe and the side wall at each end. These gaps were designed on purpose, to enable the scour process to start at the two ends of the pipe, and

propagate towards the center of the pipe Fig. 13a., similar to that in Sumer and Fredse 1994.. This enabled controlled tests. The pipes were mounted to a vertical frame. The frame itself was mounted to another frame with frictionless supports, enabling the pipe to move freely in the vertical direction. The details of the system can be found in Sumer and Fredse 1994.. The vertical displacement of the pipe was measured with the aid of a potentiometer. The wave experiments were carried out in the same wave flume as in the onset-of-scour experiments. The test setup was the same as that in the current experiments described in the preceding paragraph. The water depth was 0.33 m. Three kinds of pipes were used in the tests, with diameters D s 2, 5 and 10 cm, all measuring 59.4 cm in length, allowing 0.3 cm gap at the ends. The gap was large enough for the scour to start at the two ends of the pipe in the same way as in the current experiments. The time development of the scour along the length of the pipe was recorded with the aid of two video cameras, one viewing the entire length of the span shoulder in plan view, and the other viewing the process a close-up view. at the junction between the pipe and the scoured bed Fig. 13a.. Experiments were done for different values of the specific gravity of the pipe. This was achieved by applying additional weights to the frame to which the pipe was mounted. The flow velocity was measured by a bi-directional micropropeller, similar to the onsetof-scour experiments. The same sand was used in these tests as for the onset-of-scour tests. The angle of internal friction of the sand was measured in a vacuum tri-axial test and was found to be w s 438. Test conditions for these tests are summarized in Table 2.

3. Mechanism of onset of scour 3.1. Seepage flow and piping underneath the pipe When a pipeline is laid on a sediment bed, and subject to a current Fig. 3., the pressure difference between the upstream and the downstream of the pipe will induce a seepage flow underneath the pipe.

320

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

Fig. 3. Seepage flow underneath the pipe cf. Fig. 2..

When the current velocity is increased, a critical point is reached where the discharge of the seepage flow will be increased more rapidly than the driving pressure difference dictates, and simultaneously, the surface of the sand at the immediate downstream of the pipe will rise, and eventually a mixture of sand and water will break through the space underneath the pipe. This process is called piping, and is well-

known in soil mechanics in conjunction with the so-called piping failures at hydraulic structures such as dams, cofferdams, etc. Terzaghi, 1948.. Mao 1986., Chiew 1990. and Sumer and Fredse 1991. considered the piping as the main mechanism responsible for the onset of scour below pipelines. Piping or quicksand. conditions occur Fig. 3. in a cohesionless granular material when the pressure gradient E prg .rE x exceeds the floatation gradient s y 1.1 y n.: E p G s y 1. 1 y n . 3.1 . Ex g in which E p .rE x . is the pressure gradient driving the seepage flow just underneath the pipe Fig. 3., s is the specific gravity of sand grains s s gsrg ., gs is the specific weight of sand grains, g is the specific weight of water, and n is the porosity at the moment when these two quantities in Eq. 3.1. are equal, the soil element at the exit has lost its internal shear strength.. The previous investigators Mao, 1986; Chiew, 1990; Sumer and Fredse, 1991. emphasized, however, that the vortices that form in front of the pipe and in the lee-wake may contribute to the process of the onset of scour. We shall return to this point later in the paper.

3.2. Current case Fig. 4 shows the time series of the pressure gradient E prg .rE x in the case of the steady cur-

Fig. 4. Time series of the pressure gradient just underneath the pipe that drives the seepage flow. Current case. Test O1.6.

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

321

Fig. 5. Sequence of piping process captured by video. Time instants correspond to those in Fig. 4.

rent. In this test, the flow velocity is increased gradually until the critical point where the mixture of sand and water breaks through underneath the pipe. is reached. The pressure gradient is calculated from the two pressure time series recorded at points A and B Fig. 2.: E Ex g

pA y p B

gAB

3.2 .

It may be noted that the pressure distribution along the surface of the pipe in the soil measured in a separate test, by rotating the pipe at small increments. showed that the pressure distribution, when

plotted as a function of the distance x Fig. 3., is linear, revealing the way in which the pressure gradient is calculated Eq. 3.2... In the test, the junction between the downstream side of the pipe and the bed was videotaped simultaneously with the pressure measurements with a mini underwater camera Fig. 5a.. To enable the onset of scour to occur precisely at the section of the pressure measurements, the bed was loosened by removing the sand at this section, and then replacing it in a very gentle way. In addition to that, a small channel on the bed 15 cm long, and with decreasing depth, from 3 to 0 mm. was established, as sketched in Fig. 6 note that the figure is not to scale; the dimensions of the channel are grossly exaggerated.. With this arrangement, the onset of scour occurred precisely at the same section where the pressure measurements were made. This arrangement enabled us to relate the measured pressure gradient to the videotaped onset of scour. The obtained picture from the video recording is displayed in Fig. 5b and c cf. the time instants in Figs. 4 and 5.. From Figs. 4 and 5, the following deductions can be made. 1. There are two stages in the process of piping, leading to the onset of scour Fig. 5b and c.. As the pressure gradient increases with increasing velocity., a point is reached where the surface of the sand at the immediate downstream of the pipe begins to rise Fig. 5b., consistent with the description of the piping process described in conjunction with dams in Terzaghi 1948. it may be noted that the video recording showed clearly that this change in the bed level was not in the form of piling-up of the sand due to the lee-wake vortex, but rather in the form of rise of the bed en masse..

Fig. 6. Small channel 15-cm long, and with decreasing depth from 3 to 0 mm. on the bed, enabling the onset of scour to occur precisely at the section of the pressure measurements. The figure is not to scale; the dimensions of the channel are grossly exaggerated.

322

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

2. This stage continues for some period of time about 5 s, Figs. 4 and 5bc., and is subsequently followed by the process where a mixture of sand and water breaks through Fig. 5c.. The instant when the surface downstream starts the rise marks the instant when the pressure gradient exceeds the floatation gradient. Subsequently, grains are progressively removed and a breakthrough develops. The process will depend on the porosity, internal friction, and length of flow path the longer the path, the longer it takes for the breakthrough to develop.. 3. The onset of scour never occurred concurrently along the length of the pipe in a two-dimensional fashion, rather it always occurred locally, as illustrated in Fig. 7. If the bed were absolutely homogenous, the piping would occur over the fulllength simultaneously. The local occurrence is associated with the weakest point. 4. Fig. 4 shows that, for the piping condition to occur, the pressure gradient E prg .rE x has to reach the value equal to 1 y n. s y 1. cf. Eq. 3.1... Note that 1. the porosity value used here is n s 0.53, and it was determined for the sand in the loosest condi-

tion, consistent with the condition experienced in the actual tests; 2. a total of 12 experiments were made, and the mean value of the pressure gradient E prg .rE x was found to be 0.74 with a standard deviation 0.14; the slight variation of the pressure gradient from one experiment to the other may be attributed to the turbulent wake behind the pipeline; and 3. as seen the mean value of the pressure gradient E prg .rE x, namely 0.74, is slightly smaller than the floatation gradient 1 y n. s y 1. s 0.77. As described in the preceding paragraphs, the bed was loosened at the measurement section, and consequently, the floatation gradient was probably less than 0.77 with a porosity larger than the measured value 0.53, consistent with the expectation that the piping occurs when the pressure gradient just exceeds the floatation gradient. However, we were unable to measure what exactly the porosity was for the loosened sand at the measurement section. To facilitate comparison, a supplementary test was conducted somewhat similar to the staticpressure-gradient test of Chiew, 1990.. In this test, the pipe was placed in the same flume on the sand bed

Fig. 7. Onset of scour.

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

323

with the same burial depth, namely e s 0.64 cm corresponding to erD s 0.064. The sand bed was prepared in exactly the same way as in the current experiment, corresponding to the loosest condition. Then, both sides of the pipe were filled with water halfway through the pipe height see the small sketch in Fig. 8.. Subsequently, the water level at Side A of the pipe Fig. 8. was gradually increased, and the pressures at Points A and B were continuously recorded. Fig. 8 displays the time series of the pressure gradient E prg .rE x obtained from these records. In the test, it was observed that 1. the piping occurred in much the same way as in the current experiment, and 2. for the piping to occur, the pressure gradient E prg .rE x has to reach the value 1 y n. s y 1. as seen in Fig. 8, cf. Figs. 4 and 8., revealing the criterion in Eq. 3.1.. Comparison of this result and that in the case of the current experiment indicates that, in the current case Fig. 4., although the pressure-gradient force is apparently the major mechanism, there is also an additional mechanism which contributes to the onset of scour. This mechanism may be related to the vortices mentioned earlier see the small box in Fig. 4.. These vortices do not exist in the supplementary test, however, they do exist in the current test, and therefore they may have contributed to the piping

process exhibited in Fig. 4. However, no clear explanation has been found as to how these vortices contribute to the onset of scour. Finally, it should be noted that visual observations made in the current test showed that the vortices generated at the downstream and upstream parts of the pipe see the small box in Fig. 4. did not undermine the pipe prior to the onset of scour which would otherwise lead to a slight reduction in the length of the streamline of the seepage flow, presumably resulting in larger pressure-gradient forces.. 3.3. Wae case Fig. 9 shows the time series of the pressure gradient E prg .rE x in the wave case. The experimental setup was precisely the same as that in the current case. For the onset of scour to occur, the wave height had to be selected very large, and this led to a highly asymmetric wave asymmetric between the crest and the trough. Fig. 9.. As seen from the figure, the onset of scour takes place in the crest half period; clearly, the pressure gradient in the trough half period is not large enough to cause piping. As seen from Fig. 9, the onset of scour occurs when the pressure gradient E prg .rE x reaches the

Fig. 8. Time series of the pressure gradient just underneath the pipe that drives the seepage flow. No current. The seepage flow is caused by the rising water level at side A.

324

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

Fig. 9. Time series of the pressure gradient just underneath the pipe that drives the seepage flow. Wave case. Test O16.

value s y 1.1 y n., and even exceeds it. This result is different from that obtained for the current case Fig. 4.. This difference may be attributed to the time over which the sand is exposed to the critical pressure-gradient force. In the case of the current, this time is quite large, namely in the order of magnitude of 5 s Fig. 4., the mixture of sand and water breaks through only after O 5 s. upon the application of the critical pressure-gradient force in the case of the supplementary test mentioned in connection with the current test, this time is also large, even larger than that experienced in the case of the current, Fig. 8.. By contrast, in the case of the waves, the pressure gradient necessary for the onset of scour is available only for a very short period of time O 0.5 s.. for each crest half period, Fig. 9.. It seems that apparently this small exposure to the critical pressure gradient is not long enough for the piping to occur. It is only when the pressure gradient is increased further, and after some number of exposures, the piping takes place, resulting in the onset of scour. It may be added that the breakthrough is a progressive process; each wave loosens some grains on the exit side. Simultaneous measurements of the surface elevation h not shown in Fig. 9 to keep the figure relatively simple. and the pressure gradient

E prg .rE x showed that there is a phase difference between h and E prg .rE x: the pressure gradient Fig. 9. lags about 20258 behind the surface elevation. Fig. 10 shows a sequence of flow pictures from the laser-sheet flow visualization study for the

Fig. 10. Sequence of flow pictures over one period of wave. Test O16. h is the surface elevation. A new vortex, vortex N, forms frame 3. after vortex M is washed over the pipeline frame 2..

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

325

same wave conditions as in Fig. 9. over one wave period. The instant where the onset of scour occurs coincides almost with the passage of the wave crest frame 1 in Fig. 10. where the flow is in the direction of wave propagation and the lee-wake with vortex M. is well established. This observation reconciles with the flow pattern in the case of the steady-current see the small sketch in Fig. 4.. Although the flow picture is similar when the trough is passing frame 3 in Fig. 10., the onset of scour did not occur in this half period because the magnitude of the pressure gradient was not large enough, as has been pointed out in conjunction with Fig. 9.

4. Criterion for the onset of scour 4.1. In steady current The criterion for the onset of scour Eq. 3.1.. can be written in the following non-dimensional form: Onset of scour occurs if

Ep)

U2

E x ) gD 1 y n . s y 1 .

qR

G1
cr

4.1 .

in which p p) s , rU 2

x)s

x D

4.2 .

r is the water density, U is the undisturbed flow velocity at the top of the pipeline the top velocity rather than the center-line velocity is adopted here, considering the cases where the pipeline may be buried with erD larger than 0.5., and g is the acceleration due to gravity. The term R is a small, non-dimensional term, and is included here to represent the effects other than the pressure-gradient force mainly the effect of the vortices forming in front of the pipe and, particularly, in the lee-wake discussed in the preceding paragraphs.. Both E p ) .rE x ) . and R are essentially a function of the burial-depth-to-diameter ratio, erD. Therefore, the criterion for the onset of scour can be written in the following form.
U2 gD 1 y n . s y 1 .
cr

Gf

/
D

4.3 .

where the function f erD . is to be determined from experiments. It may be noted that f is actually a function of not only the gap-to-diameter ratio, erD, but also the pipe Reynolds number, Re s UDrn , and the relative roughness k srD in which n is the kinematic viscosity and k s is the surface roughness of the pipe. However, it is expected that the influence of these latter parameters will not be very significant, if there is no significant change in the flow regime, i.e. if the flow around the pipe does not change from the subcritical regime to the supercritical regime, or from the supercritical regime to the transcritical regime see, e.g. Sumer and Fredse, 1997, Chapter 1., as will be demonstrated later in the section. Also, it may be mentioned that cohesionless granular material is considered in the present analysis. Otherwise, soil properties including permeability. will also influence the onset of scour clearly, in the case when the permeability 0, the breakthrough will never occur.. The focus in this subsection will be on the variation with erD. The data obtained in the present investigation for the onset of scour is plotted in the form of Eq. 4.3. in Fig. 11. The procedure used in the tests is as follows. 1. Level off the bed. 2. Place the pipe on the sand bed gently without pressing it., and fix it to the side walls. Then, fill the two sides of the pipe with sand up to the level corresponding to the burial depth, e, to be studied in the test. Make sure that the bed at the two sides of the pipe is more or less horizontal. 3. Increase the flow velocity in small increments until the onset of scour occurs. Identify this critical velocity at this point keep the time of increasing the flow velocity as short as possible to ensure that no buildup of sand at the downstream side of the pipe occurs; the latter would obviously change the picture regarding the seepage flow and the piping, and therefore not desired.. 4. Repeat the exercise in items 13 for the next burial depth to be investigated. Fig. 11 shows that the larger the burial depth, the higher the critical velocity for the onset of scour, as expected. This is because, as the burial depth increases, the pressure gradient will be decreased, therefore relatively higher velocities will be required to cause piping.

326

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

Fig. 11. Critical condition for the onset of scour. Current case. Tests O1O14.

Fig. 11 further shows that 1. the results for two different pipe diameters, namely D s 5 and 10 cm, seem to coincide, when plotted in terms of the non-dimensional quantities in Fig. 11, and 2. likewise, the results for two different sand sizes, namely d 50 s 0.18 and 1.25 mm, seem to collapse on a single curve, revealing that the results are unaffected by the sand size. Note that the experiments for coarser sand were not conducted for burial depths smaller than erD s 0.08. This is because, for such small values of erD, the sand will no longer act as a continuous medium. Therefore, the results will not make sense.. The data in Fig. 11 can be represented by the following empirical expression 2 Ucr e 0.5 s 0.025 exp 9 4.4 . gD 1 y n . s y 1 . D

in which Ucr is the critical undisturbed flow velocity measured at the level of the top of the pipeline. for the onset of scour. Finally, it may be noted that the time required for the flow to remove the grains and open a AbreachB will be appreciably longer for larger diameter pipes than for those used in the present study. 4.2. Effect of change in flow regime To see the effect of change in the flow regime around the pipeline, the 10 cm diameter pipe was

coated with 0.3 cm diameter and 0.3 cm height cylinder-shaped plastic grains. The burial depth tested in this experiment was erD s 0.1. The grains were glued in a densely packed manner. to the cylinder, and the roughness height measured from the base pipe surface to the top of the roughness elements. was 0.3 cm, or alternatively the Nikuradses equivalent sand roughness k s + 2 = 0.3 s 0.6 cm, giving a relative roughness of k srD s 6 = 10y2 . To keep the boundary condition in the sand the same as in the case of the smooth pipe, the holes between the roughness elements were filled with plastic for the portion of the pipe that stays in the sand bed. The only difference between the rough-pipe test and the smooth-pipe test is that, in the smooth-pipe case, the flow around the pipe was in the subcritical regime Re s 5.7 = 10 4 ., whereas, in the rough-pipe case, it was in the transcritical regime Re s 6 = 10 4 , k srD s 6 = 10y2 . Sumer and Fredse, 1997.. The result of this experiment is compared with its smooth-pipe counterpart in Table 3. As seen, the critical value of the parameter U 2r gD1 y n. s y 1.. is now a factor of 1.7 larger than that for the smooth-pipe case. This is because, in the case of the rough pipe, the flow is in the transcritical regime, therefore the pressure gradient will be smaller due to the relatively larger wake pressure, see, e.g. Sumer and Fredse, 1997, p. 41., and hence, relatively larger velocities will be re-

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335 Table 3 Comparison of critical condition for the onset of scour for two different flow regimes. Burial depth, erD s 0.1 Test number Pipe Critical velocity for the onset of scour, Ucr cmrs. 56.5 76.0 Pipe Reynolds number at the critical velocity, Re s Ucr Drn 5.7 = 10 4 6.0 = 10 4 Pipe roughness, k srD 6 = 10y2 Flow regime
2 Ucr r gD s y 1. 1 y n..

327

O6 O15

Smooth Rough

Subcritical Transcritical

0.42 0.72

quired for the onset of scour. This result suggests that, for extremely large pipelines with smooth surface Re ) O 10 5 .., or for mediumrlarge size pipelines with very large roughness Re ) O 10 4 ., k srD ) O 10y2 .. where the flow regime is transcritical, see, e.g. Sumer and Fredse, 1997., the critical curve for the onset of scour displayed in Fig. 11 may be shifted upwards so that the critical value

of the parameter U 2r gD1 y n. s y 1.. would be a factor of 1.52 larger than depicted in Fig. 11. 4.3. In waes In the case of waves, the criterion given in Eq. 4.3. can be adopted provided that the following will be followed.

Fig. 12. Critical condition for the onset of scour. Tests O16O31. Steady-current result is taken from Fig. 11.

328

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

1. U is replaced by Um , the maximum value of the orbital velocity of water particles at the bed. 2. There will be an additional parameter regarding the function f, namely f erD, KC .. This is because, in this case, the terms E p ) .rE x ) . and R in Eq. 4.1. are also governed by the Keulegan Carpenter number KC see, e.g. Sumer and Fredse, 1997.. The dependence of the onset of scour on KC has been discussed by Sumer and Fredse 1991.. In the latter study, the variation of the critical burial depth for the onset of scour with KC was obtained; however, the role of the parameter U 2r gD1 y n. s y 1.. was not recognized. The data obtained by Sumer and Fredse 1991. has been recast, and plotted in Fig. 12 together with the present data. It is seen that both parameters, 2 namely KC and Um r gD1 y n. s y 1.., are equally significant. For a given value of KC, the critical value of the 2 parameter Um r gD1 y n. s y 1.. increases with increasing erD. This can be explained in the same way as in the case of the steady current. Likewise, for a given value of erD, the critical value of the 2 parameter Um r gD1 y n. s y 1.. increases with increasing KC. This is because the pressure gradient decreases with increasing KC cf. the pressure diagram given in Sumer and Fredse, 1991, Fig. 3, and that in Bearman and Zdravkovich, 1978, Fig. 1., therefore larger and larger velocities will be needed for the onset of scour, meaning that the critical value 2 of Um r gD1 y n. s y 1.. will increase with increasing KC. Fig. 12 indicates that, as the KeuleganCarpenter 2 number increases, the critical value of Um r gD1 y n. s y 1.. approaches that in the case of the steady current. For example, for erD s O 0.05., the critical 2 value of Um r gD1 y n. s y 1.. approaches that for the steady current for KC ) O 20.. This is linked to the fact that the pressure gradient in the case of the waves approaches that experienced in the case of the steady current. The present study focuses on the variations with respect to KC and erD. The variation with the number of waves or the time. required for the piping to occur has not been studied. Similar to the case of steady current, the time required for the piping to develop will be appreciably longer for

larger diameter pipes than for those used in the present study.

5. Self-burial of pipeline at span shoulder After the scour breaks out underneath the pipe, it propagates along the length of the pipeline, as sketched in Fig. 1. A three-dimensional scour pattern emerges in which the scour holes are interrupted by stretches of soil, called span shoulders, where the pipe obtains its support, section A-A in Fig. 1a. As has been mentioned in Section 1, various cases may occur, depending on the flow, the soil, and the pipe stiffness. The present study focuses on the self-burial of pipelines at span shoulders. As indicated in Section 1, attention will be concentrated on the case of waves. 5.1. Mechanism of self-burial The process of self-burial occurs as follows. The scour begins to propagate along the length of the pipeline after the onset of the process Hansen et al., 1991.. As the process continues, the length of the free span will be larger and larger at the expense of the span shoulder. Therefore, more and more weight of the pipe will be exerted on the soil over a shorter and shorter length of the span shoulder Fig. 13a.. This process may reach such levels that the bearing capacity of the soil is exceeded, and therefore the soil fails. The failure occurs by sliding in the two outward directions, as indicated in Fig. 13b. This type of failure is known as a general shear failure in soil mechanics Terzaghi, 1948.. Clearly, as the scour continues, the bearing capacity of the soil will be exceeded constantly due to the continuous reduction of the bearing area, leading to the permanent sinking of the pipe. The process will stop, only when the pipe sinks to such depths that it will be protected against the scour. When the scour stops, obviously the constant failure of the soil will stop, and consequently the sinking of the pipe will come to an end. As implied in the preceding paragraph, the scour at the two ends of the span shoulder Fig. 13a. is the key mechanism for the process of pipe sinking. The scour process itself is governed mainly by the KeuleganCarpenter number Sumer and Fredse, 1994,

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

329

Fig. 13. Definition sketch. Sinking of pipeline at span shoulder.

1999.. This is essentially linked to the lee-wake, precisely in the same way as in the case of two-dimensional scour below a pipeline; the higher the KeuleganCarpenter number, the longer the lee-wake that forms behind the pipe in each half period of the motion., the larger the scour see Sumer and Fredse, 1990, for a detailed discussion.. This suggests that, first of all, the sinking depth the self-burial depth., e, normalized by the pipe diameter D, should be a function of the KeuleganCarpenter number KC; and secondly erD should increase with increasing KC. The following sections will basically discuss this issue. 5.2. Self-burial depth Fig. 14 presents the results of a self-burial test made in the present study. Here KC was 16 Table 2.. There are three stages in the process.

1. During the first 40 s, the scour spreads along the length of the pipe until the length of the bearing area is reduced to l s 40 cm Fig. 14b.. 2. At this point, the pipe begins to sink in the sand Fig. 14a. due to the general shear failure, as described in the preceding paragraphs. This stage continues for about a little more than 400 s. 3. Subsequently, the space between the pipe and the scour hole is gradually backfilled with sand, and the length of the span shoulder begins to increase by virtue of the backfilling process. This stage takes place from 480 to 540 s Fig. 14b.. With the completion of this stage, the pipe is eventually buried to a depth of 2.0 cm erD s 0.39.. The process occurs in precisely the same manner as described by Sumer and Fredse 1994. for the case of the steady current. Fig. 15 displays the data regarding the equilibrium self-burial depth obtained in the present experi-

330

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

Fig. 14. a. Time series of sinking of the pipe and b. that of the length of span shoulder. Test S10c.

ments. The existing steady current data from the present work and from Stansby and Starr 1991. is also included note that only the data of Stansby and Starrs work with u ) ucr live-bed. is included.. First of all, Fig. 15 shows clearly that the selfburial depth is a function of the KeuleganCarpenter number, KC. It increases with increasing KC, as expected see the discussion in the previous Section 5.2.. Secondly, the influence of the pipes specific gravity, s, on the end results is insignificant see the data point for KC s 16, the cross, and also the legend in Fig. 15., for this KC number, five tests were conducted with five different values of s in the range 1.256, and it was found that the sinking depth was practically unchanged. Sumer and Fredse 1994. reached the same conclusion in the case of the steady current. The sinking is uninfluenced, because the key element in the process is the scour; when the scour stops, the sinking will also stop, as

discussed in the preceding subsection. Since the pipes specific gravity is not an influencing factor for the scour, it will therefore not affect the self-burial depth. Although not included here, the present wave results compare well with those of Stansby and Starr 1991.. However, this is for KC larger than about 20. For KC Q 20, the self-burial depths of Stansby and Starr begin to assume smaller and smaller values with decreasing KC for KC Q 20. This may be attributed to the small pipe size in Stansby and Starrs experiments; for KC numbers smaller than about 20, the self-burial depth measured by the latter authors is e s O 0.20.3 cm., and clearly, such small values of e may be subject to considerable uncertainty. Fig. 16 compares the present self-burial data with the data of Sumer and Fredse 1990. for the scour depth below a fixed pipeline with an originally zero

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

331

Fig. 15. Self-burial depth vs. KC. Tests S1S23. Live bed. u ) ucr .

clearance between the pipe and the bed.; see the legend in Fig. 16. As seen, these two sets of data agree quite well. This is an interesting result. It can be explained as follows. Now, as the sand at the span shoulder fails progressively, the pipe sinks in the sand, and, at the same time, it falls in the scour holes at the two sides of the span shoulder Fig. 13a.. The scour process comes to an end when the pipe reaches the bottom of these scour holes Fig. 17b.. At this moment, the scour depth, S Fig. 17b. will be fairly close to that obtained for a fixed pipe originally in contact with the bed Fig. 17a., Fredse et al. 1988.. This implies that the self-burial depth at the span shoulder should be the same as the latter, as revealed by Fig. 16.

Although not tested directly due to the experimental constraints., the preceding results imply that:

1. a pipeline may be self-buried completely for KC larger than O 100.; and 2. the self-burial depth of pipelines may reach values as high as erD , 3, for very large KC numbers such as O 1000., representing the tidal flow conditions.

The following empirical expression, given originally by Sumer and Fredse 1990. for the equilibrium scour depth below a fixed pipeline with an initial zero gap between the pipe and the bed., can be

332

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

Fig. 16. Self-burial depth, and scour depth vs. KC. Live bed. u ) ucr .

used to assess the self-burial depth of pipelines at span shoulders for KC Q 100 Fig. 16.: e s 0.1'KC . 5.1 . D Caution must be exercised when extrapolating the preceding equation for KC larger than 100, as it has not been tested for the self-burial of pipelines for such large KC numbers. Finally, it may be noted that the process of selfburial studied here is due to general shear failure of the soil. However, the self-burial may occur due to other processes as well, as mentioned in Section 1. Sumer et al. 1999. report self-burial of pipelines in the laboratory to depths as much as 22.5 D, due to liquefaction of soil caused by the buildup of pore pressure. Sakai et al. 1994. report block subsidence in the laboratory, due partly to the so-called momentary liquefaction where the soil is, over a short period of time, liquefied during the passage of wave troughs, see Sakai et al., 1992., and partly to the

oscillatory flow action. Raudkivi 1976, p. 365. reports a pipeline settling in the surf zone in the field. 4 m under the bed level, due to liquefactionrfluidization of bed under waves.

Fig. 17. Definition sketch.

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

333

5.3. Time scale of self-burial process The sinking of the pipe develops towards the equilibrium stage through a transitional period Fig. 14a.. The time scale of the process may be defined by the following equation Ts 1 e
`

parameter, u . Here, T ) is the normalized time scale defined by T s

)

g s y 1. d 3 D2

1r2

5.3 .

H0

e y et . d t

5.2 .

see the definition sketch in Fig. 18.. Here, e is the equilibrium sinking depth and e t is the sinking depth at time t. The time scale T can be interpreted as the time over which a substantial amount of self-burial takes place. The data regarding the time scale is plotted in Fig. 18 as a function of two parameters, namely the KeuleganCarpenter number, KC, and the Shields

Fredse et al., 1992., considering that the time scale of the self-burial process is similar to that of the scour process that induces the self-burial. Fredse et al. 1992. gives the background regarding the nondimensional time scale T ) . This non-dimensional quantity can easily be obtained by normalizing the equation of sediment continuity. 1. As seen from Fig. 18, the time scale decreases with increasing u . This is because the larger the value of u , the higher the sediment transport in the scour process, the faster the sinking of the pipe in the sand; therefore, the time scale should increase with

Fig. 18. Time scale of sinking process as a function of KC and u ) ucr .

334

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

increasing u . On the other hand, the figure shows that the larger the KeuleganCarpenter number, the larger the time scale. This is because the larger the value of KC, the larger the scour depth, therefore the larger the volume of the sand undergoing scour, hence the larger the time scale. 2. Although there is only one data point for the case of the steady current, apparently the time scale of the self-burial process in the case of the steady current seems to be quite close to that for the case of waves with KC s 24. This may be linked to the self-burial depth. Fig. 16 suggests that the self-burial depth for the two cases namely, the steady-current case and the case of KC s 24. are rather close. Therefore, the time scale of the self-burial process should also be close to each other.

critical point is reached where the bearing capacity of the soil is exceeded general shear failure.. At this point, the pipe begins to sink at the span shoulder. 6. The ultimate equilibrium. self-burial depth is a function of KC. The larger the value of KC, the larger the self-burial depth of the pipeline. 7. Eq. 5.1. may be used to assess the self-burial depth for a given wave climate. In the case of the steady current, on the other hand, the self-burial depth is 5080% of the pipe diameter. 8. The time scale of the self-burial process is dependent of two parameters, namely the KeuleganCarpenter number and the Shields parameter, and it may be assessed, using Fig. 18. List of symbols D Pipe diameter Sand size d 50 e Embedment burial. depth of the pipeline for the onset of scour tests.; otherwise, equilibrium self-burial depth of pipeline et Self-burial depth sinking depth. at time t g Acceleration due to gravity h Water depth KC KeuleganCarpenter number, Um Tw rD n Porosity of sand p Pressure Non-dimensional pressure, pr r U 2 . p) Re Reynolds number, UDrn , or Um Drn s Specific gravity of sand, rsrr Specific gravity of pipeline, rprr sp t Time T Time scale of self-burial process Nondimensional time scale of self-burial T) process, w g s y 1. d 3 x1r2 TrD 2 Wave period Tw U Undisturbed flow velocity at the top of the pipeline Maximum value of the orbital velocity of Um water particles at the bed Friction velocity in the case of currents Uf Maximum value of the friction velocity in Ufm the case of waves x distance along the surface of the pipeline in the sand Fig. 3. ) Non-dimensional distance, xrD x g Specific weight of water Specific weight of sand grains gs

6. Conclusions 1. The pressure difference between the upstream and downstream sides of a pipeline drives a seepage flow underneath the pipe. When this seepage flow becomes excessive, piping occurs; a mixture of water and sand breaks through underneath the pipe, resulting in the onset of scour below the pipeline. 2. The previously mentioned pressure difference together with other effects of secondary importance such as vortices forming in front of and at the lee-wake of the pipe. are the agitating forces for the piping process. 3. In the case of the steady current the critical condition for the onset of scour is determined by two parameters, the burial depth of the pipe, erD, and a parameter involving the flow velocity, U 2r gD1 y n. s y 1... The critical condition can be assessed for a given pipeline and a given current climate from Fig. 11. 4. In the case of the waves, there is an additional parameter, namely the KeuleganCarpenter number, KC. The critical condition in this case can be determined, using Fig. 12. 5. Once the scour breaks out, it will propagate along the length of the pipeline. Scour holes are interrupted with stretches of soil span shoulders. supporting the pipeline. As the span shoulders get shorter and shorter, more and more weight of the pipeline is exerted on the soil. In this process, a

B.M. Sumer et al.r Coastal Engineering 42 (2001) 313335

335

h u n w

Surface elevation Shields parameter Kinematic viscosity Angle of internal friction of sand

Acknowledgements This study was partially funded by the Commission of the European Communities, Directorate-General XII for Science, Contract No. MAS3-CT97-097, Scour Around Coastal Structures SCARCOST., and by the 5-year 19992004. Framework Programme AComputational HydrodynamicsB of the Danish Technical Research Academy, STVF.

References
Bearman, P.W., Zdravkovich, M.M., 1978. Flow around a circular cylinder near a plane boundary. J. Fluid Mech. 109, 3348. Bernetti, R., Bruschi, R., Valentini, V., Venturi, M., 1990. Pipelines placed on erodible seabeds. Proc. 9th International Conference on Offshore Mechanics and Arctic Engineering Conference, ASME, Houston, TX, vol. V, pp. 155164. Chiew, Y.-M., 1990. Mechanics of local scour around submarine pipelines. J. Hydraul. Eng., Am. Soc. Civ. Eng. 116 4., 515529. Fredse, J., Hansen, E.A., Mao, Y., Sumer, B.M., 1988. Three-dimensional scour below pipelines. J. Offshore Mech. Arctic Eng., Am. Soc. Mech. Eng. 110, 373379. Fredse, J., Sumer, B.M., Arnskov, M.M., 1992. Time scale for wavercurrent scour below pipelines. Int. J. Offshore Polar Eng. 2 1., 1317. Hansen, E.A., Staub, C., Fredse, J., Sumer, B.M., 1991. Timedevelopment of scour induced free spans of pipelines. Proc. 10th Offshore Mechanics and Arctic Engineering Conference, ASME, Stavanger, Norway. Pipeline Technology, vol. 5, pp. 2531. Hansen, E.A., Klomp, W.H.G., Smed, P.F., Bijker, R., Bryndum, M.B., 1995. Free span development and self-lowering of

pipelinesrcables. Proc. 14th Offshore Mechanics and Arctic Engineering Conference, ASME, Copenhagen, Denmark. Pipeline Technology, vol. 5, pp. 409417. Klomp, W.H.G., Hansen, E.A., Chen, Z., Bijker, R., Bryndum, M.B., 1995. Pipeline seabed interaction, free span development. Proc. 5th Int. Offshore and Polar Eng. Conf., The Hague, Netherlands, June 1116, vol. II, pp. 117122. Leeuwenstein, W., Bijker, E.A., Peerbolte, E.B., Wind, H.G., 1985. The natural self-burial of submarine pipelines. Proc. 4th International Conf. on Behavior of Offshore Structures BOSS., vol. 2. Elsevier, pp. 717728. Mao, Y., 1986. The interaction between a pipeline and an erodiblebed. Series Paper 39, Tech. Univ. of Denmark, ISVA, in partial fulfillment of the requirement for the degree of Doctor of Philosophy. Raudkivi, A.J., 1976. Loosed Boundary Hydraulics. 2nd edn. Pergamon. Sakai, T., Hatanaka, K., Mase, H., 1992. Wave-induced effective stress in seabed and its momentary liquefaction. J. Waterw., Port Coastal Ocean Eng., Am. Soc. Civ. Eng. 118 2., 202 206, See also Discusions and Closure in vol. 119, No. 6, pp. 692697. Sakai, T., Gotoh, H., Yamamoto, T., 1994. Block subsidence under pressure and flow. Proceedings of the 24th Conference on Coastal Engineering ICCE 94., 2328 October, Kobe, Japan. pp. 15411552. Stansby, P.K., Starr, P., 1991. On a horizontal cylinder resting on a sand bed under waves and current. Int. J. Offshore Polar Eng. 2 4., 262266. Sumer, B.M., Fredse, J., 1990. Scour below pipelines in waves. J. Waterw., Port Coastal Ocean Eng., Am. Soc. Civ. Eng. 116 3., 307323. Sumer, B.M., Fredse, J., 1991. Onset of scour below a pipeline exposed to waves. Int. J. Offshore Polar Eng. 1 3., 189194. Sumer, B.M., Fredse, J., 1994. Self-burial of pipelines at span shoulders. Int. J. Offshore Polar Eng. 4 1., 3035. Sumer, B.M., Fredse, J., 1997. Hydrodynamics Around Cylindrical Structures. World Scientific, xviiiq530 pp. Sumer, B.M., Fredse, J., 1999. Wave Scour Around Structures. In: Liu, P.L.-F. Ed.., Advances in Coastal and Ocean Engineering, vol. 4. World Scientific, Chapter, 51 pp. Sumer, B.M., Fredse, J., Christensen, S., Lind, M.T., 1999. SinkingrFloatation of pipelines and other objects in liquefied soil under waves. Coastal Eng. 38 2., 5390, October. Terzaghi, K., 1948. Theoretical Soil Mechanics. Wiley, New York.