You are on page 1of 20

OTC-26026-MS

Experience with Interface Shear Box Testing for Axial Pipe-Soil Interaction
Assessment
Z.J. Westgate, Fugro, Houston, USA
D.J. White, University of Southampton (UK), University of Western Australia (Australia)

Copyright 2018, Offshore Technology Conference

This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 30 April - 3 May 2018.

This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents of
the paper have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.

Abstract

This paper presents experience and best practice for a new geotechnical laboratory test protocol – the low stress interface shear
box (ISB) test – for determining soil properties relevant to pipeline-seabed friction. The paper is underpinned by a major
database (>250 tests) that demonstrates the protocol and shows general variations in the key parameters that may be useful for
early design purposes. By accurately quantifying shear resistance along the pipe-soil interface under low normal stresses
imposed by subsea pipelines, ISB tests allow design ranges in axial friction to be narrowed and tailored to specific pipeline
conditions. These improved geotechnical inputs to pipe-soil interaction can alleviate unnecessary axial expansion, walking or
buckling mitigation, unlocking cost savings otherwise unavailable without accurate geotechnical parameter characterization.

A large database is presented of recent industry experience with low normal stress interface shear testing using a modified
direct shear box device. This device, while still considered novel, is emerging as the new industry standard for axial pipe-soil
interaction testing, gaining wider adoption than the tilt table and torsional shear devices. The test database comprises several
soft clays from various deep water hydrocarbon-producing geographical regions, and several types of pipeline coatings.

The database populates a theoretical framework for axial pipe-soil interaction with new data, illustrating general trends for key
parameters such as normal stress, shearing duration, interface roughness, and changing pipeline weight, each of which is
shown to vary the axial resistance by a factor of two or more. The shear resistance can also change by a factor of two or more
due to consolidation or swelling during and between individual cycles of movement, associated with pipeline operations. This
database and the populated theoretical framework can guide pipeline and geotechnical engineering by providing a basis for
initial estimates of axial friction, and an approach for improving these estimates via focused site-specific testing.

The new approach has been applied through >250 ISB tests for >16 oil and gas projects conducted in the laboratories of Fugro
and the University of Western Australia, to deliver better pipe-soil interaction data than is available via other means. This
paper shares this experience to support new projects in two ways: Firstly, the large high quality database provides a
significantly improved basis for estimates of geotechnical soil properties for subsea pipeline design where site-specific data is
not yet available. Secondly, we provide guidance on the planning, execution and interpretation of low stress interface shear
tests, to allow best practices to be adopted more widely across industry.

Introduction

This paper presents experience and best practice guidance for a new geotechnical laboratory test protocol – the low stress
interface shear box test – for determining soil properties relevant to axial pipeline-seabed friction for the design of high
pressure, high temperature (HPHT) pipelines, focusing on soft fine-grained soils in deep water. The experience is illustrated
through presentation of a large database of test data. General trends emerge to quantify the effects for key governing factors
2 OTC-28671-MS

influencing soil-interface shear resistance. These trends can be used to guide early assessment of axial pipe-soil shear
resistance, and highlight conditions that warrant site-specific testing and further analysis.

Pipeline Global Movements: Axial Expansion, Walking and Buckling. HPHT pipelines expand and contract due to
changes in internal temperature and pressure. These movements are opposed by the axial resistance between the pipe and the
seabed. Excessive compressive force may lead to buckling of the pipeline, depending on the soil resistance and the pipe
weight. Once a buckle forms, the axial force drops significantly as pipe feeds axially into the buckle. Excessive feed-in – due
to low axial resistance – leads to high bending strains within the buckled section, especially when these are coupled with high
lateral soil resistance.

Thermal cycles of expansion and contraction can lead to the accumulation of axial movement, termed pipeline walking (Carr
et al. 2006). The walking arises from asymmetry in the startup and shutdown processes, or from other asymmetry in the
pipeline such as a seabed slope or variations in contents density. Walking is not a limit state for the pipeline itself, but can lead
to failure at the mid-line or end connections. To mitigate against excessive walking, rock-dumping (limited to shallower water)
or large hold-back anchors may be required. This mitigation can be very costly for large flowline developments, and increases
safety hazards through additional offshore activities.

The expansion, buckling and walking behavior is influenced by the axial pipe-soil interaction (PSI) behavior. ISB testing
assists the characterization of axial PSI by simulating the behavior at the pipe-soil interface.

Design Approach for Axial Pipe-Soil Interaction Assessment. The soil input parameters required for axial PSI assessment
are associated with a force-displacement model, linking the shear resistance (per unit pipe length), T, normalized by the pipe
weight W (i.e. a ‘friction factor’ T/W) to the mobilization distance x (Figure 1). This model is akin to a ‘t-z’ model for pile
axial movement. The key parameters for soft clay conditions is the steady or residual value reached at large axial pipe
displacements, and any initial breakout peak is of less importance.

The residual resistance can be mobilized in undrained or drained conditions depending on the consolidation characteristics of
each soil zone/unit through which the pipeline crosses, the pipeline dimensions, and time periods associated with the event
being considered. Axial pipeline movements may take place over short or long periods relative to the drainage process, so
often both undrained and drained conditions need to be considered for typical soft clays. In sandy conditions, the undrained
strength parameters may not be applicable due to the fast draining nature of the soil; however, sands with significant fine-
grained soil (i.e. clay and silt) may behave undrained or partially drained for typical axial pipeline movements.

Undrained axial movements generate excess pore pressure, which dissipates during the intervening period when the pipe is
stationary. Over time the axial resistance will transition towards the drained value due to successive generation and dissipation
of pore pressure, although this effect can be suppressed for smooth interfaces that do not cause soil-soil failure. The change in
strength, referred to as consolidation hardening, may occur within a single movement event (e.g. Randolph et al. 2012) or
through cycles of movement (e.g. Yan et al. 2014).

Figure 1. Axial force-displacement model


OTC-26026-MS 3

The parameters that influence axial pipe-soil interaction are:

• Soil properties, including strength and consolidation coefficient, which vary with soil type
• Normal stress level, relating to the bearing pressure imposed by the pipeline onto the seabed
• Overloading history, or changes in the bearing pressure due to pipeline contents changes through commissioning and
operation (which can create an overconsolidation of the soil at the pipeline-seabed interface)
• Pipeline coating roughness (which can be considered relative to the soil grain size)
• Duration of axial movement during startup or shutdown events (which affects the level of drainage that can occur
during the movement)
• Duration between operating events (i.e. startup and shutdown, which affects the level of drainage that can occur
between movements)
• Pipeline embedment, via the ‘wedging’ factor that enhances the normal contact force to exceed the pipe weight.

A conceptual model for capturing each of these key parameters for pipeline design on soft clays has been previously proposed
by White et al. (2012), developed from a wide range of laboratory and model testing on marine clays (Hill et al. 2012). The
model applies to soft clays where the strength of the clay in contact with the pipe is controlled by consolidation under the
weight of the pipe. The key elements to the model are illustrated on Figure 2. The strength properties of the soil in contact
with the pipe are linked to the axial resistance via separate expressions for drained and undrained conditions, which are now
accepted as the basis for design expressions (White et al. 2017):

Tres,d
Drained conditions:   tan  res (1)
W
s   W' 
m

 R nc OCR m   u ,int, res   max  which simplifies to Tres,u  ps u ,int, res OCR 
Tres,u m
Initial undrained condition: (2)
W  ' no  nc  W' 

where: su,int,res Undrained residual interface strength


res Residual interface friction angle
OCR The ratio between the maximum previous pipe weight and the current value, OCR = Wmax/W
p Pipe-soil contact length around the pipe perimeter (i.e. varying with pipe embedment)
 Wedging factor, representing the ratio N/W where N is the normal force between pipe and soil per unit
length (see White & Randolph 2007).  rises from 1 at zero embedment to 1.27 at an embedment of half the
diameter, based on analytical simplifications that match well with numerical simulations (Yan et al. 2016)

The drained resistance is based on Coulomb friction, adjusted for wedging. The undrained resistance is based on the undrained
interface strength acting on the perimeter of contact with the seabed, with this strength being derived from the normal stress
applied by the pipeline. The strength parameters res, Rnc and m (or su,int,res for a given stress history) can be derived directly
from ISB test results that mimic the relevant normal stresses. The geometric parameters p and  are selected from estimates of
the as-laid embedment.

The left-hand side of Figure 2 illustrates Equation 2 with the undrained residual resistance, Tres,u (normalized by the pipe
weight) being a function of OCR. Wmax is typically controlled by the flooded pipeline weight for hydrotesting prior to
operation. Two trendlines show the generic difference between a smooth and a rough interface surface.

The lower undrained limit is tied to the normally-consolidated case (OCR = 1). As a normally-consolidated soil is sheared
along the pipeline, consolidation causes the resistance to rise towards the drained value (center figure). If the soil is initially in
an overconsolidated state, the undrained resistance may be higher than the drained limit. In this case, the undrained resistance
drops towards the drained limit as the soil along the interface swells during shearing. Cycles of shearing cause the undrained
value to converge towards the drained limit, for normally-consolidated or overconsolidated conditions (although for smooth
interfaces this effect can be suppressed). Finally, the shear resistance is also a function of the normal stress as shown on the
right-hand side of Figure 2, due to the curvature of the Mohr-|Coulomb failure envelop that is evident in tests at low stresses
(e.g. Najjar et al. 2007, White et al. 2012).

Use of this approach to capture site-specific effects on axial resistance allows pipeline designers to move away from the
conventional (low) values of axial pipe-soil friction, e.g. 0.2 for clay as recommended in DNV Recommended Practice F109
(2007), and derive ranges of axial friction for specific events or time periods in the operating life of a pipeline. This approach
has recently been adopted by the new DNV Recommended Practice F-114 (2017), but was previously introduced to industry
through the SAFEBUCK JIP (Atkins 2015, White 2014).
4 OTC-28671-MS

Undrained assessment Transitional behaviour Drained assessment

Undrained limit, Residual axial friction, Drained limit,


(Tres,u/W) (Tres/W) (Tres,d/W)
0.9
Drained limit
Effect of OCR 0.8
Transition to critical state Effect of stress
on su-int,res
0.7 due to cycles of sliding
Higher stress or0.7 level on res
lower roughness
and consolidation
0.6 0.6
0.5 0.5
0.4 0.4
OCR=1
0.3 0.3
0.2 Undrained limit Transition to drained 0.2
conditions due to consolidation Lower pipe
0.1 during axial movement 0.1 roughness
Lower pipe roughness
0 0
1 0.01 0.1 1 10 100 0
Pipe-induced over-
Consolidation time, t/t50 Normal pipe-soil
consolidation ratio,
(or elapsed time of pipe movement) stress W/p
OCR = Wmax/W

Figure 2. Conceptual model for axial residual pipe-soil resistance (White 2014, Atkins 2015, DNV 2017)

Project PSI Workflow. The level to which site-specific conditions need to be quantified for a given project are driven by the
potential to unlock benefits by narrowing the design range in axial friction. A typical workflow through key project stages is
illustrated on Figure 3. During concept stage, there is typically no site-specific soil data and therefore axial friction factors are
estimated based on regional experience, and a cautiously-wide range is required covering both undrained and drained
conditions. These values are then used in pipeline engineering to assess susceptibility to excessive axial expansion, walking or
other design issues. Typically, a site investigation to support front end engineering design (FEED) would then allow for
laboratory testing on site-specific soils using project-specific pipeline parameters (such as interface type and pipeline weight,
and therefore normal stress). On some projects, a third iteration has been performed, using model testing or additional site
investigation data, depending on the level of risk and value opportunity to the project. The model testing could be performed in
the laboratory using recovered soil samples (e.g. Smith & White 2014), or could be performed at the seabed, in situ, using a
tool such as the Fugro SMARTPIPE (Ballard et al. 2013).

Figure 3 compares pathways for two projects in which PSI has different criticality. The cost-PSI uncertainty trade-off for
Project A has a step benefit at a particular uncertainty threshold. This could represent, for example, the potential to eliminate
one or more pipeline anchors or buckle initiation structures, if the PSI resistance can be bracketed sufficiently tightly. This
scenario illustrates a case with impetus to pursue advanced PSI characterization – e.g. model testing – in order to narrow the
design range of PSI parameters: the cost of this additional characterization is worthwhile given the anticipated cost benefit
from the simplified project architecture. Project B illustrates a case in which the PSI parameter uncertainty has less influence
on the pipeline design and therefore the more advanced characterization is not considered worthwhile. These two scenarios
illustrate that there is no single best scope for PSI characterization. Instead, the appropriate scope should be determined based
on the potential change to the project architecture and risk that would be unlocked by refinement of the PSI parameters.

In particular, box core or gravity core sampling followed by low stress interface shear box testing is now becoming accepted as
the baseline for FEED level design on HTHP pipeline projects. Increasingly, this type of testing is being performed during
concept stage engineering since the sampling can be piggybacked onto a geophysical survey. The testing program can explore
a range of pipe parameters – such as coating type – as well as the other governing parameters affecting axial resistance listed
earlier. Project-specific model testing for axial PSI is relatively rare, but has underpinned the development and validation of
the simple theoretical models used to link interface properties and pipeline friction factors (such as the work conducted in the
SAFEBUCK JIP, White (2014), and the projects using the Fugro SMARTPIPE, e.g. Ballard et al. (2013). Model testing is
more common for lateral PSI, where the complexity of soil berm effects can lead to unacceptable uncertainty without project-
specific model test data.
OTC-26026-MS 5

SI Extra SI, lab or Design

e.g. axial friction


estimates

PSI parameter,
model testing
Cost of pipeline system, $

Project B
Project A Upper
Step change (PSI is Best
in mitigation Low
requirement
critical to
design)

SI

e.g. axial friction


PSI parameter,
Project A Project B
(PSI less
critical to
PSI parameter range, e.g.
design)
axial friction

Time
Concept FEED Detailed Operation

Figure 3. Example axial PSI workflow through project design stages

Background

Previous Approaches to Interface Shear Testing. Historically, laboratory testing for soil-structure interface behavior has
focused on normal stress conditions relevant to pile foundations. This involves normal stress levels that are one to two orders
of magnitude greater than the stress imposed from a pipeline onto the seabed. These early studies highlighted the key
governing factors influencing soil-structure shear resistance. Laboratory devices to measure interface shear resistance have
evolved over the past several decades, as summarized in Table 1. This overview is limited to laboratory element testing
devices, which are all subject to complex boundary effects and uncertainty when extrapolating to larger soil mass. Model
testing – both in situ using specialized tools and ex situ in a 1g model setting or a geotechnical centrifuge – has proven useful
to measure axial pipe-soil interface shear resistance directly, although at a higher cost.

Modified Low Stress Interface Shear Testing for Pipeline Design. In most of the devices in Table 1 (all except the Cam-
Tor), the shear test is not globally undrained since there is no membrane surrounding the specimen. However, in soft clays all
of the tests can be conducted sufficiently fast that the measured resistance is unaffected by drainage. For typical soft clays with
cv values around 1 m2/yr, a displacement rate of 0.1 mm/s is sufficient to limit drainage in the soil, based on the solutions
proposed by Gibson and Henkel (1954), whereas a displacement rate of 0.001 mm/s is sufficient to allow almost full
dissipation of the pore pressures within the specimen (see also Boukpeti & White, 2017).

The main challenge with low stress shearing for pipeline design applications is limiting the system friction, arising from the
shear ring-interface sliding resistance or other sources of resistance associated with the limitations of a mechanical device.
Since a component of the system friction is usually independent of the applied normal stress, the lower the applied normal
stress the greater the percentage of the total measured resistance that is due to system friction. Tilt table tests eliminate system
friction completely but are subject to other limitations as described in Table 1, namely the inability to control the displacement
rate. Examples of modified direct interface shear boxes that limit the amount of system friction are shown in Figure 4,
including a schematic of the general test setup. The two devices used for the majority of the test database presented in this
paper comprised a ShearTrac-II-DSS system developed by Geocomp Corporation of Houston (Figure 4a) and a Humboldt
5560 (Figure 4b).

A range of interface surfaces can be used to quantify the effect of pipeline coating type and roughness. Ideally, project-
specific pipeline coating samples are used, although this may not be possible when the design and the contractor are not
finalized. Generic ‘smooth’ and ‘rough’ samples can be substituted (and their roughness measured pre- and post-testing) to
allow for later interpolation of the results to the roughness of the actual coating surface. Figure 5 shows examples of a clay
specimen on the interface after being removed from the circular shear ring. The rough surface shows that the failure plane was
predominantly through the clay, while the smooth surface shows suggests the failure plane was along the interface. The critical
roughness for this transition in shearing mechanism is commonly reported to be around 10 microns for clay soils (Tsubakihara
& Kishida 1993), although this value is likely to vary with particle size, as has long been recognized for sands (Uesugi &
Kishida 1986).
6 OTC-28671-MS

Table 1. Element Testing Approaches to Measuring Interface Shear Resistance


Mechanical
Device Example Image Advantages Limitations References
Description
Direct Interface surface Controlled Short horizontal
Potyondy 1961,
Interface sliding below a displacement rate; displacement range; no
Tsubakihara and
Shear Box fixed square or simple apparatus; pore pressure
Kishida 1993
(ISB) circular specimen wide experience. measurement

Bishop et al. 1971,


Interface surface Variation in strain rate
Unlimited Bromhead 1979,
rotating under radially across
Ring Shear displacement Yoshimi and
fixed annular specimen; no pore
achievable Kishida 1981, Stark
specimen pressure measurement
and Eid 1993

Circular specimen
Short horizontal Bolton et al. 2007,
sliding above a Controlled
displacement range; no Kuo 2011,
Cam-Shear fixed interface displacement rate;
pore pressure Ganesan et al.
surface (similar to simple apparatus
measurement 2013
ISB)

Uncontrolled
displacement rate; Pedersen et al.
Thin specimen Simple, apparatus,
no pore pressure 2003, Najjar et al.
Tilt Table sliding along titled machine friction
measurement; cyclic 2003, Najjar et al.
interface surface eliminated
capability depends on 2007
set-up.

Less machine
friction than ring
shear and smaller
Circular interface
specimen (can use
rotating under Radial variation in Kuo et al. 2015, De
Cam-Tor intact samples);
fixed circular displacement rate Brier et al. 2016
globally undrained
specimen
testing possible
(membrane
surrounds sample)

Test Database

The database presented in this paper represents testing performed over the last 6 years for 16 projects from 8 different
hydrocarbon-producing regions, and includes over 250 individual tests on clay soils. The interface surfaces used in the testing
comprise a range of plastics, rubber, sandpaper, and concrete, with average roughness (R a), ranging from 0.2 microns to more
than 80 microns. Some of the interface material was donated by pipeline coating suppliers and then cut from a sample pipeline
section and mounted onto a flat interface plate. Other interfaces used were produced by attaching generic materials such as
sandpaper to the interface plate. An overview of the test database is presented in Table 2.

Table 2. Overview of Test Database


Gulf of Mexico, South America, North Atlantic, East Africa,
Geographic regions:
West Africa, Caspian Sea, Mediterranean Sea, Indian Ocean
Acetal, Polypropylene, Polyethylene,
Interface surfaces:
Sintered fusion-bonded epoxy, Rubber, Sandpaper, Concrete
Average roughness, Ra: 0.2 micron to more than 80 microns
Size of database: Over 250 tests across 16 projects in 8 different regions
OTC-26026-MS 7

(a) (b)

(c)

Figure 4. Modified direct interface shear box devices at (a) the University of Western Australia (also at the Fugro Soil Laboratory in
Perth), (b) the Fugro Soil Laboratory in Houston, and (c) a schematic illustration of the test setup.

(a) (b)

Figure 5. Example clay specimen after being shear across (a) rough coating surface and (b) smooth coating surface.
8 OTC-28671-MS

Soil Types. Soil types within the overall database include siliceous and carbonate materials. While the focus of this paper is
on soft clay, several tests were also performed on a range of silty, sandy, gravelly and shelly materials. Tests for which the
portion of coarse-grained material was greater than about 10% were excluded from the presented database. A summary of the
soil properties for the clays included in the presented database is provided in Table 3. All of the samples were remoulded at
either the in situ moisture content or a moisture content close to the liquid limit during preparation of the test specimens.

Table 3. Soil Properties within Presented Database

Soil Property Values Included in Database

In situ undrained shear


Generally less than 5 kPa within upper 0.5 m of seabed
strength, su

In situ water content, w 22% to 310%

Liquid Limit, LL 45% to 200%

Plasticity Index, PI 30% to 152%

Carbonate content, CaCO3 Up to 90%, but generally less than 10%

Up to 275 m2/yr, but generally less than 1 m2/yr, including in situ and derived from the
Coefficient of consolidation, cv
consolidation stage of ISB test

Clay fraction Generally greater than 50% (with silty material)

Typical Results

Example cyclic ISB test results show typical resistance and deformation responses for a normally-consolidated clay sheared on
a smooth interface (Figure 6), an overconsolidated clay sheared on a smooth interface (Figure 7), and a normally-consolidated
clay sheared on a rough interface (Figure 8). The data in these figures is not corrected for system friction. All three tests were
performed using the following procedure:

1. Remold the soil at the in situ water content to form a thin paste (generally close to the liquid limit for deep water soft
clays).

2. Pre-consolidate the sample under a light pressure (~1 kPa) and then trim and transfer the sample to the specimen ring.

3. Consolidate the specimen under the target normal stress, no for a minimum of 24 hours (with an intermediate stage
at the target maximum normal stress, n,max if an over-consolidated sample is required), monitoring settlement to
confirm consolidation is complete after each stage.

4. Perform 1 cycle pair of ‘fast’ shearing between displacement limits of +/- 5 mm. A rate of 0.1 mm/s was adopted for
‘fast’ shearing.

5. Allow the specimen to consolidate for 30 minutes (the settlement of the platen can be used to confirm that dissipation
of pore pressure generated in the previous cycle pair is complete after this duration).

6. Repeat steps 4 and 5 until a total of 18 fast shearing cycles (or 9 fast shearing cycle pairs) has been completed.

7. Perform 1 cycle pair of slow shearing between displacement limits of +/- 5 mm. A rate of 0.001 mm/s was adopted
for ‘slow’ shearing.
OTC-26026-MS 9

1 -0.2

Vertical displacement (mm)


Shear stress ratio, t/'no

0
0.5
0.2
0 0.4
0.6
-0.5
0.8
-1 1
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Horizontal displacement (mm) Horizontal displacement (mm)

(a) (b)

Figure 6. Example results for normally-consolidated clay on smooth interface: (a) shear stress ratio versus horizontal displacement
and (b) vertical displacement versus horizontal displacement (gray lines are fast shearing and black lines are slow shearing)

1 Vertical displacement (mm) -0.04


-0.02
Shear stress ratio, t/'no

0.5 0
0.02
0
0.04

-0.5 0.06
0.08
-1 0.1
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Horizontal displacement (mm) Horizontal displacement (mm)
(a) (b)

Figure 7. Example results for overconsolidated clay on smooth interface: (a) shear stress ratio versus horizontal displacement and (b)
vertical displacement versus horizontal displacement (gray lines are fast shearing and black lines are slow shearing)

1 -1
Shear stress ratio, t/'on

Vertical displacement (mm)

0
0.5
1
0 2
3
-0.5
4
-1 5
-6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6
Horizontal displacement (mm) Horizontal displacement (mm)
(a) (b)

Figure 8. Example results for normally-consolidated clay on rough interface: (a) shear stress ratio versus horizontal displacement
and (b) vertical displacement versus horizontal displacement (gray lines are fast shearing and black lines are slow shearing)
10 OTC-28671-MS

Key aspects of the behavior in Figures 6-8 are:

• For all three specimens, there is an initial peak in resistance that quickly drops to a steady residual value over the first
millimeter of shearing.

• For the normally-consolidated clay sheared on a smooth interface (Figure 6), the undrained residual resistance (gray
lines, Figure 6a) is less than the drained resistance. Minimal vertical displacement occurs during fast shearing (gray
lines, Figure 6b), confirming undrained conditions. Settlement between cycles is evident, confirming pore pressure is
generated and dissipates. Vertical displacement accompanies the slow cycles (black lines, Figure 6b), but does not
occur afterwards, confirming that the slow cycles were drained. The shear resistance during the fast cycles (gray
lines, Figure 6a) gradually converges towards the drained limit during the slow cycles (black lines, Figure 6a).

• For the overconsolidated clay sheared on a smooth interface (Figure 7), the undrained residual resistance (gray lines,
Figure 7a) is greater than the drained resistance, and the vertical displacement during the initial fast shearing (gray
lines, Figure 7b) exhibits specimen dilation (shown as negative displacement values). The shear resistance during the
fast cycles (gray lines, Figure 7a) gradually reduces towards the shear resistance observed in the final slow cycles
(black lines, Figure 7a).

For the normally-consolidated clay sheared on a rough interface (Figure 8), the response is similar to the normally-
consolidated clay sheared on a smooth interface (Figure 6). However, the drained resistance (black lines, Figure 8a) is
significantly greater than the initial undrained residual resistance (gray lines, Figure 8a), and the specimen settlement is greater
than for the smooth interface.Performing the test in the manner described above provides the two key design parameters for
axial PSI – the initial undrained strength ratio (su,res,int/no) and the final drained stress ratio (tan res). The intervening fast
cycle pairs allows assessment of the change in undrained resistance due to consolidation hardening of a normally-consolidated
clay or swelling of an overconsolidated clay.

Smooth, OCR = 1 Smooth, OCR = 3 Rough, OCR = 1 Smooth, OCR = 1 Smooth, OCR = 3 Rough, OCR = 1

1 0
Vertical displacement (mm)
Shear stress ratio, t/'no

0.8
0.04
0.6
0.08
0.4
0.12
0.2

0 0.16
0 4 8 12 16 20 0 4 8 12 16 20
Cycle number Cycle number

(a) (b)

Figure 9. Example results versus cycle number showing (a) shear stress ratio (closed symbols are even cycle numbers showing the
steady state resistance; fast shearing is shown in black, slow shearing in gray) and (b) vertical displacement during each
consolidation period (i.e., after each even-numbered cycle)

The variation in shear resistance with cycle number shows this consolidation effect (Figure 9a). Each marker in Figure 9a
shows the average of the forward and backward resistance during a cycle, with system friction removed, with a consolidation
period allowed after each pair of cycles. The consolidation hardening – expressed as the ratio of drained to initial undrained
resistance – is greater for the rough interface (triangles) compared to the smooth interface (diamonds). The resistance rises
from the undrained strength ratio, su,res,int/no = 0.4, to a drained friction of tan res = 0.8, i.e. a doubling of strength and
therefore axial resistance. For the smooth interface, the normally-consolidated clay (diamonds) hardens from su,res,int/no 
0.33 to a tan res  0.42. The overconsolidated clay (circles) converges from su,res,int/no  0.58 to tan res  0.46. For all three
cases, the final undrained resistance (cycle 18, closed black symbols) is close to the drained resistance (cycle 20, closed gray
symbols). These are examples of the hardening and swelling behavior, but within the database other soil types do not harden or
swell as fast as these cases, particularly on a smooth interface surface.
OTC-26026-MS 11

Figure 9b shows the vertical displacement of the top load platen over each 30-minute rest period. The settlement (shown here
as positive contraction) for the two normally-consolidated specimens reduces as the specimen contracts further with each
cycle. The smooth surface test at OCR = 1 exhibits significantly less consolidation than the rough surface test, due to the
thicker shear zone, while the smooth test at OCR = 3 exhibits very little consolidation due to its initial dilatancy (Figure 7b).

Key Trends in the Database

The initial normally-consolidated undrained strength ratio (su,res,int/no)nc and the final drained strength ratio (tan res) were
interpreted from each of the shear box tests in the database. Figure 10a shows the distribution of (su,res,int/no)nc values (170 in
total) from the initial pair of cycles on normally-consolidated specimens. Figure 10b shows the distribution of tan res values
(216 in total) from the final pair of cycles on all specimens – both normally-consolidated and overconsolidated.

Low and high estimates of axial resistance are required for pipeline design since different design cases, such as axial walking
and lateral buckle initiation, are governed by different values. While the mean and standard deviation values are provided in
Figure 10 for each parameter, estimates associated with a probability of occurrence are generally preferred by pipeline
engineers since this allows for consistent handling of uncertainty in probabilistic structural reliability analysis. The 10 th, 50th
and 90th percentiles have been calculated directly from the data for each of the undrained and drained parameters, and are also
shown on Figure 10. Other percentiles can be calculated from these values if needed. The percentile ranges for each parameter
are quite wide, mainly due to the inclusion of all surface roughness values.

4 2.5
3.5 Pop. = 170 Pop. = 216
Mean = 0.36 2 Mean = 0.61
3 Std. dev. = 0.18 Std. dev. = 0.25
Frequency density

Frequency density

LE (P10) = 0.21 LE (P10) = 0.34


2.5
BE (P50) = 0.34 1.5 BE (P50) = 0.53
2 HE (P90) = 0.55 HE (P90) = 0.96

1.5 1

1
0.5
0.5
0 0
0.1 0.3 0.5 0.7 0.9 1.1 1.3 0.1 0.3 0.5 0.7 0.9 1.1 1.3
Undrained shear strength ratio, Drained shear strength ratio,
(su,res,int/'no) nc tan res
(a) (b)

Figure 10. Distributions of (a) normally-consolidated undrained residual shear strength ratio (su,int,res/’no)nc and (b) drained residual
shear strength ratio (tan res)

Effect of Interface Roughness. To assess the effect of the interface surface average roughness (Ra), the undrained strength
ratio (su,res,int/no)nc and the final drained strength ratio (tan res) were each plotted as a function of roughness, presented on
Figure 11a and Figure 11b respectively. Normalizing the average roughness by the mean particle size (D 50) could provide a
more complete assessment, but D50 values were generally not measured for the clay soils in the database. Figure 11a and 11b
both show that the undrained and drained strengths increase with interface surface roughness, with a critical roughness
defining a transition in strength broadly around values of Ra = 2 to 10 microns. This is consistent with other studies of clay-
interface shearing on a range of materials including steel, glass, plastic and epoxy (Tsubakihara and Kishida 1993, Lemos and
Vaughan 2000, Najjar et al. 2007).

For the purposes of illustration in this paper, the database is divided into ‘smooth’ and ‘rough’ tests, where all surface
roughness values 2 microns or less were classified as ‘smooth’ and all surface roughness values 10 microns or more were
classified as ‘rough’. There were no tests performed on surfaces with Ra values between 2 and 10 microns.
12 OTC-28671-MS

1.4 1.4
Undrained strength ratio, (su,res,int/'no) nc

Drained strength ratio, tan res


1.2 1.2

1 1.0

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0.0
0.1 1 10 100 0.1 1 10 100
Average roughness, Ra (micron) Average roughness, Ra (micron)

(a) (b)

Figure 11. Shear strength ratio versus average surface roughness (Ra) for (a) fast displacement ‘undrained’ conditions and (b) slow
displacement ‘drained’ conditions

Effect of Normal Stress. The drained residual shear resistance is known to be a function of the applied normal stress (e.g.
Skempton 1985, Stark & Eid 1994, Najjar et al. 2007). The SAFEBUCK guidance has adopted a power law fit to quantify the
effect of normal stress, using the following formula:

t = a 'nob (3)

where a and b are parameters fitted to the drained shear resistance measured at different normal stress levels. The undrained
resistance is also influenced by the initial normal stress. A power law has been fitted to the trends of undrained and drained
shear resistance with initial normal stress (Figures 12a and 12b), using a constant value of b = 0.9 for all cases, which is the
best fit to the curvature of the strength envelopes based on the full database. The a values, denoted as au for undrained
strength and as ad for drained strength, both in units of kPa (1-b), differ by a factor of 2 between the undrained and drained
strength, which is consistent with the difference between drained and undrained shear strengths from constitutive models such
as Modified cam clay. Best fit (P50), and 10th (P10) and 90th (P90) percentiles have been calculated. The resulting undrained and
drained strength parameters from these power law trend lines range from (su,res,int/no)nc = 0.20 (P10) to 0.54 (P90) and from tan
res = 0.33 (P10) to 1.03 (P90), over the tested normal stress range from no = 2 to 16 kPa.

The a and b parameters presented in Figure 12 offer a further improvement, linked to normal stress, compared to the wider
range of (su,res,int/no)nc and tan res shown in Figure 10. However, the range in a values for both undrained and drained cases
still reflect the full range of interface materials used and their associated range in roughness. Separating the database into
smooth and rough cases (defined using a critical roughness of R a = 2-10 micron as described above) results in stress- and
interface-specific ranges of the a parameter for each case.

Figure 13a and 13b show the undrained shear strength for the smooth and rough interface surfaces, respectively, while Figure
14a and 14b show the drained shear strength for the smooth and rough interface surfaces, respectively. The b values are again
common across all cases, with a constant value of 0.9. Here, the resulting undrained strength parameters range from
(su,res,int/no)nc = 0.17 to 0.45 for smooth surfaces and from (su,res,int/no)nc = 0.27 to 0.65 for rough surfaces. The resulting
drained strength parameters range from tan res = 0.30 to 0.57 for smooth surfaces and from tan res = 0.53 to 1.17 for rough
surfaces.

These stress- and roughness-specific values of a are useful for selecting parameters for preliminary PSI calculations. However,
it is important to recognize that actual design values may lie outside of these ranges, i.e. the P 90 for a given site could lie
outside the P90 of the full database. For example, the highest strengths tend to be associated with angular carbonate sediments,
and for one of the soil types within the database the best estimate of tan res is 1.1, i.e. equal to the database P90. As discussed
earlier in this paper, projects should use appropriate levels of site-specific characterization based on the level of risk and cost
reduction opportunity. Design ranges from project-specific interface shear box testing using the relevant interface surface
roughness are invariably much narrower than the database ranges presented here.
OTC-26026-MS 13

16 16
Undrained shear strength, tu (kPa)

Drained shear stregth, td (kPa)


LE (P10) | BE (P50) | HE (P90) LE (P10) | BE (P50) | HE (P90)
14 au: 0.27 0.38 0.58 14 ad: 0.43 0.65 1.10
b: 0.9 for all cases b: 0.9 for all cases
12 Pop. 170 data points (all surfaces) 12 Pop. 216 data points (all surfaces)

10 10
8 8
6 6
4 4
2 2
0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Initial normal stress, 'no (kPa) Initial normal stress, 'no (kPa)
(a) (b)
Figure 12. Trend lines fit to the database for (a) undrained shear strength and (b) drained shear strength

16 16
Undrained shear strength, tu (kPa)

LE (P10) | BE (P50) | HE (P90) LE (P10) | BE (P50) | HE (P90)


Undrained shear strength, tu (kPa) 14
14 au: 0.22 0.35 0.48 au: 0.35 0.50 0.70
b: 0.9 for all cases b: 0.9 for all cases
12 Pop. 104 data points (smooth surfaces) 12 Pop. 66 data points (rough surfaces)

10 10
8 8
6 6
4 4
2 2
0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Initial normal stress, 'no (kPa) Initial normal stress, 'no (kPa)
(a) (b)

Figure 13. Undrained shear strength versus initial normal effective stress for (a) smooth surfaces and (b) rough surfaces

16 16
LE (P10) | BE (P50) | HE (P90)
Drained shear strength, td (kPa)

Drained shear strength, td (kPa)

LE (P10) | BE (P50) | HE (P90)


14 ad: 0.39 0.48 0.61 14 ad: 0.70 0.95 1.25
b: 0.9 for all cases b: 0.9 for all cases
12 Pop. 122 data points (smooth surfaces) 12 Pop. 92 data points (rough
surfaces)
10 10
8 8
6 6
4 4
2 2
0 0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Initial normal stress, 'no (kPa) Initial normal stress, 'no (kPa)
(a) (b)

Figure 14. Drained shear strength versus initial normal effective stress for (a) smooth surfaces and (b) rough surfaces
14 OTC-28671-MS

Effect of Overloading History. The effect of overloading history on the soil can be interpreted from ISB results using the
SHANSEP approach as follows, which is consistent with Equation (2) for undrained axial pipe-soil resistance, and is presented
using the same notation:

(su,res,int/no)oc = (su,res,int/no)nc OCRm = Rnc OCRm (4)

where the OCR value is taken as the pipeline flooded weight divided by the operating weight, and m is a fitting parameter,
which is equivalent in Cam clay terminology to the plastic volumetric strain ratio. The m parameter was evaluated directly for
several of the clay and interface surface materials within the database. This was done by performing multiple tests on the
same sample and the same interface at different OCR values. Power law trend lines, based on SAFEBUCK guidance (Atkins
2015, White 2014, DNV 2017), were fitted to each data set as shown on Figure 15a.

The undrained strength ratio is lower for the smooth cases (open symbols) compared to the rough cases (closed symbols), but
there is no systematic difference between the back-calculated smooth and rough m parameter values. The mean, standard
deviation, and 10th, 50th and 90th percentiles have been calculated directly from the full dataset of 32 values, and are shown on
Figure 15b. The range is quite wide, partly because any small experimental variations in the measured strength have a large
effect on the fitted value of m. However, the P50 is consistent with values back-analyzed from model testing donated to the
SAFEBUCK JIP (White 2014, Atkins 2015, DNV 2017)). The range in m is wider than deduced from conventional undrained
strength tests in the early work of Ladd and Foott (1974), but is comparable to the scatter in the larger database of Mayne
(1980), reproduced in Muir Wood (1990).

1
Undrained strength ratio, su,res,int/'no

3 Mean = 0.70
open symbols: smooth surfaces
Std. dev. = 0.91
closed symbols: rough surfaces
LE (P10) = 0.19
0.8 2.5
BE (P50) = 0.51
Frequency Density

HE (P90) = 0.97
2 Pop. = 32
0.6
1.5
0.4
1

0.2 0.5

0 0
0.5 1 1.5 2 2.5 3 3.5 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
OCR m parameter
OCR
(a) (b)

Figure 15. Effect of overconsolidation: (a) gain in strength from over-consolidation and (b) back-calculated range of m parameter

Effect of Clay Type. Previous studies have investigated the effect of clay type and composition on low stress shear resistance,
including liquid limit (Stark and Eid 1994, Najjar et al. 2007, Eid et al. 2015), plasticity index (Najjar et al. 2007, Eid et al.
2015), clay fraction (Lemos and Vaughan 2000, Eid et al. 2015), and carbonate content (Mesri et al. 1975). One of the key
conclusions from previous studies on low stress interface shear testing is that existing correlations between shear strength and
soil properties derived from tests at high normal stress (typically n > 50 kPa) do not readily extrapolate to low stress
conditions. However, attempts have been made to link index properties to low stress strength parameters, albeit using
relatively small datasets or combined datasets from various types of equipment and studies.

The test database presented in this paper has been used to test for a correlation between interface shear strength and plasticity
index (PI). An equivalent correlation for clay undrained strength at higher stresses was set out by Skempton (1954) and
remains widely quoted. Meanwhile, for interface strength, Ramsey et al. (1998) and Jardine & Chow (2005) gave similar
correlations for clay-steel drained interface strength, intended to assist design work when ring shear test data is unavailable
(but Atterberg limits are).
Figure 16a compares the present database of undrained strength ratio versus PI, and Figure 16b presents the drained strength
ratio versus PI. There is no trend with PI for either the undrained or drained strengths, for the smooth surfaces (closed
OTC-26026-MS 15

symbols) or the rough surfaces (gray symbols). Figure 16c expands the comparison by combining the drained interface friction
angle (res) values from Figure 16b with data from several previous studies listed in Table 4. This additional data indicates a
possible trend of reducing friction angle with increasing plasticity index for soils with plasticity index below ~50%. For the
collated dataset, across a wider range of PI, there is no trend.

The database for higher stress levels intended for pile design from Jardine et al (2005) is shown on Figure 16d, using the same
axes as Figure 16c to allow visual comparison. The pile design database sites within a corner of this plot, which reflects two
factors. Firstly, it demonstrates that pile design studies have focused on lower plasticity clays, such as found in the North Sea,
compared to the higher plasticity deep water clays found at the mudline in the frontier development regions of the recently-
developed ISB database. Secondly, it highlights the higher residual friction angles found at the lower stresses of the ISB
database compared to the higher stresses of pile design database. The trend within the pile design database for reducing
friction angle with increasing PI around PI < 50% mirrors that detected in the low stress studies (Figure 16c). However, the
wider ISB dataset shows that this trend is not applicable in the geotechnical conditions that tend to be found in current pipeline
projects.

1.4 1.4
gray symbols: rough surfaces gray symbols: rough surfaces
Undrained strength ratio, (su,int,res/'no) nc

Drained strength ratio, tan res


black symbols: smooth surfaces black symbols: smooth surfaces
1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
Plasticity Index (%) Plasticity index (%)
(a) (b)

70 70
white symbols: soil-soil white symbols: soil-soil
Drained interface friction angle, res
Drained interface friction angle, res

gray symbols: rough surfaces gray symbols: rough surfaces


60 60 black symbols: smooth surfaces
black symbols: smooth surfaces

50 50

40 40

30 30

20 20

10 10
circles: Fugro database circles: Fugro database
triangles: others (4 sets - see text for description) triangles: others (4 sets - see text for description)
0 0
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140
Plasticity index (%) Plasticity index (%)
(c) (d)
Figure 16. Effect of clay type: (a) undrained strength ratio versus plasticity index for all normally-consolidated specimens, (b) drained
strength ratio versus plasticity index for all specimens, normally- and overconsolidated, (c) drained interface friction angle versus
plasticity index for all tests including those from previous studies and (d) comparison with higher stress database for pile design
(Jardine et al. 2005).
16 OTC-28671-MS

Table 3. Previous Studies of Low Stress Shear Testing

Reference Soil Type Plasticity Index Shear Device Interface Surface Normal Stress
Lehane and
Bothkennar clay 55 to 120% Ring shear Soil-soil only 2 to 8 kPa
Jardine 1992
Pedersen et al. Soil-soil and soil-acrylic
Kaolinite 22% Tilt table 1 to 8 kPa
2003 (smooth)
Soil-soil, soil-epoxy (smooth
and rough), and soil-
Gulf of Mexico clay 52.5 to 82.5% Tilt table 1.7 to 5.8 kPa
Najjar et al. polyurethane (smooth and
2007 rough)

Kaolinite 23% Tilt table Soil-soil only 1.8 and 3.5 kPa

Fraser River silt 3%

Kaolinite 22%
Soil-soil and smooth epoxy
Modified ring and steel
Eid et al. 2015 Gulf of Doha clay 19 to 56% 3 and 6 kPa
shear

Nile Delta clay 12 to 49%

Gray silt 17% Soil-soil and smooth steel

Guidance for Planning, Executing and Interpreting Low Stress Interface Shear Box Tests

Test Program Design. To develop a fit-for-purpose program of interface shear box tests for a given project, an axial
resistance model should be developed for each soil unit or zone. The model should quantify the following key aspects
governing axial PSI, which allows design estimates of axial resistance parameters to be determined for the relevant range of
pipeline types:

• Effect of normal stress. The test program should span normal stresses that cover the full range of empty through
flooded pipeline submerged bearing pressures (W/D), and extend to slightly higher stresses (i.e. pipe weight divided
nominal diameter). This higher extent is required because the normal stress in the field condition may be higher if the
pipeline is only partially-embedded (i.e. up to a maximum of W/p). As observed for the test database, most of the
tests have been performed using a normal stress range of 2 to 8 kPa. This is typically sufficient for most single-wall
pipelines; however, for heavy pipe-in-pipe lines and seabeds where pipeline embedment is limited, higher normal
stress values may be required. A wide range of normal stress helps to capture the stress effect curvature associated
with the power law fitting parameters a and b.

• Effect of overloading history. For some projects, particularly regions where oil (rather than gas) is produced, the
difference between the flooded and operating pipeline weights is minimal, e.g. OCR values less than 1.1. In this case
tests at higher OCR levels may not be warranted. For such low values of OCR the uncertainty in the m parameter
may have no practical effect on the estimated axial resistance. For light oil, gas, or thin-walled lines, the OCR effect
is generally important, and for gas lines the OCR caused by flooding may exceed 10. For these cases, it may useful to
perform tests at three values of OCR (e.g. 1, 2, 3, or 1, 3, 10) on the same sample material and the same interface, to
give a more precise assessment of the m parameter for the required range of OCR.

• Effect of pipeline coating roughness. At the concept stage of a project, the pipeline coating material is often not
known, and the manufacturing method and roughness may not be finalized until after the installation contractor is
selected. The pipeline coating process for typical deep water pipelines (that do not require any concrete coating for
secondary stabilization measures) begins with layers of polypropylene or polyethylene insulation. Normal
application of the coating results in a relatively smooth finish, so performing tests using a smooth (Ra < ~2 microns)
plastic coating is often a reasonable base case for testing. However, projects for which a roughened coating is
anticipated (e.g. for pipelay tensioning purposes), or where concrete coatings will be required, may consider a rough
interface (Ra > ~10 microns) material as the base case for testing. In most cases it is justified to perform tests on the
OTC-26026-MS 17

opposite roughness to the base case. This may highlight the opportunity to deliberately alter the axial resistance via a
change in coating choice, for example to indicate whether a deliberately roughened coating would raise the friction
sufficient to partially or wholly mitigate the possibility of pipeline walking. Tests using a variation in coating
roughness will also indicate the potential change in axial resistance that could be imposed by a change in the coating
approach. It can be more cost-effective to roughen the pipeline coating to reduce excessive axial expansion or
walking compared to other mitigation options.

The minimum data required to develop site-specific trends for the basic axial PSI parameters can be obtained by performing
the above sensitivity cases via tests that follow the procedure illustrated in Figures 6-8 (i.e. at least two pairs of cycles at fast
and slow displacement rates, to obtain undrained and drained parameters). Variations to the test procedure can be considered
for advanced PSI parameters associated with the rate of evolution of the shear resistance due to consolidation or swelling,
either through continuous hardening (during cycles) and episodic hardening (between cycles). Continuous hardening occurs
due to drainage during the movement, whereas episodic hardening takes place between cycles of movement. Further details are
provided in Randolph et al (2012), Yan et al. (2014), White et al. (2015), and Boukpeti and White (2017).

For continuous hardening, tests are performed at intermediate displacement rates (or a wider range in displacements). Tests
performed at intermediate displacement rates allow for more precise assessment of model parameters relating to continuous
consolidation hardening, e.g. the time for 50% dissipation in the soil, t50 (Figure 2, center plot). For episodic hardening, tests
performed using a total of 20 cycles (18 fast and 2 slow) is often sufficient to reach the fully hardening state; however, more
cycles may be required, particularly for highly plastic clays sheared against a smooth interface. Methods for scaling from the
drainage rates in ISB tests to drainage around a model pipeline have been established by comparison of ISB tests and large-
scale model tests (White 2014).

The number of tests performed per soil unit or zone should be selected depending on the range of interfaces, stress levels and
OCRs. An example program of 10 tests (plus system friction tests on each surface in the absence of soil) is shown in Table 5.
The additional fast cycles performed in a subset of tests provide supplementary data on episodic hardening rates, which can be
useful if the rate of hardening, as well as the initial undrained and final (drained) axial resistance limits influence the design.

Table 5. Typical Test Program for a Single Soil Unit for PSI Parameter Development
Test Consolidation Normal Surface Number of Cycles
Test Description Design Parameters Obtained
No. Stress Stress Roughness Fast Slow
1 X X smooth 2 2 Low normal stress su,res,int/’no and tan res, or
2 Y Y smooth 18 2 Base case undrained and drained a and b
3 Z Z smooth 2 2 High normal stress parameters for smooth surface
4 2Y Y smooth 18 2 Low OCR m parameter for smooth
5 3Y Y smooth 18 2 High OCR surface (with test 2)
6 X X rough 2 2 Low normal stress su,res,int/’no and tan res, or
7 Y Y rough 18 2 Base case undrained and drained a and b
8 Z Z rough 2 2 High normal stress parameters for rough surface
9 2Y Y rough 18 2 Low OCR m parameter for rough surface
10 3Y Y rough 18 2 High OCR (with test 7)

Test Procedure. Best practice testing procedures have evolved over the past several years as this type of testing has become
more familiar to pipeline and geotechnical engineers, and laboratories have refined equipment and software. The basic
procedures are described earlier in the paper, and most conventional direct shear box testing protocols also apply to low stress
ISB tests including load cell calibration and sensitivity, specimen trimming, and use of porous load platen as described in
ASTM D3080 (ASTM 2011). However, substitution of the lower half of the shear box with a flat interface surface changes
some of these protocols, such as no requirement for a specimen area correction as described for large scale interface shear
testing of geosynthetics in ASTM D5321 (ASTM 2011). While no ASTM standard yet exists specifically for low stress ISB
tests for assessing pipeline axial friction, some additional guidance is provided below:

• Sample selection. For sites with similar surface soil conditions across the pipeline routes, it may be appropriate to
assign several cores to a single soil zone type, and combine these cores to create a single batch sample. The batch
sample approach, versus performing sets of tests on sub-samples from individual cores, reduces the number of tests to
be performed. It also aids the parameter fitting process since any slight differences between individual core properties
do not add to scatter within the data. This averaging is acceptable since small local variations in axial resistance on
the pipeline are averaged in the build-up of effective force along the pipe length anyway.
18 OTC-28671-MS

• Sample preparation. Due to the pipelaying process, surface soils are highly disturbed from the cyclic motion of the
pipeline, driven by the seastate (Westgate et al. 2010). Water entrainment is also possible, where the water content of
the soil is increase due to the moving pipeline. Because of this, performing tests on intact soil is often not necessary
and instead samples are initially remolded to a water content value at or slightly above the in situ value. This moisture
content is usually close to the liquid limit for typical deep water soft clays.

• Layered soils. For sites where a thin surficial layer of slurry, biogenic material, sand or shells is present at the
mudline, separate test sets on each soil layer may be warranted, particularly if the expected pipe embedment range
extends into both layers. Different mixtures of each layer (e.g. 25% shells, 75% clay, and vice versa) can be used to
assess the influence of the mixture proportions that may result from the dynamic pipelaying process. This can be
important for axial PSI since the sand or shell content can dominate the frictional response of the clay mixture;
conversely a low permeability slurry or biogenic layer can govern the drainage characteristics of the clay mixture,
delaying the consolidation hardening process.

• Consolidation. Due to the very soft nature of remolded clay at or close to the liquid limit, it can be necessary to pre-
consolidate the sample outside of the shear ring under a nominal low stress level to create a workable sample. This
stress level should be below half of the target normal stress level for the test. After the sample is transferred into the
shear ring, it is useful to monitor the sample settlement during consolidation phase, which may be performed in
multiple increments of stress. The settlement response can provide an estimate of the coefficient of consolidation of
the soil, which is useful in various aspects of PSI analysis. For the consolidation period between cycles, 30 minutes is
commonly used, but this could be reduced for more permeable materials or increased for very impermeable materials.
Settlement of the top load platen during the consolidation stages can be monitored to assess whether the dissipation is
complete.

• Displacement limits. A limitation of the ISB test is the short displacement limits allowed, typically 10 to 20 mm for
monotonic shearing and 5 to 10 mm for cyclic shearing. The limits used in the majority of tests performed in the
database were 5 mm. To ensure a steady residual value is achieved during each shearing stage, a protocol of
performing cycles in pairs was adopted, with a total of 40 mm of shearing per cycle pair. This distance has been
shown by others, using ring shear testing, to be generally sufficient to achieve a steady residual resistance (e.g. Lemos
and Vaughan 2000, Eid et al. 2015). However, for rough surfaces particularly when the clay includes a portion of
coarse-grained material, higher displacement limits may be required to achieve a steady residual value. A check on
whether a steady resistance has been reached can be made by comparing the resistance at the end of all four half
cycles within each cycle pair.

• Soil-soil tests. To confirm whether a given rough interface is fully rough, supplementary soil-soil shear (i.e. direct
shear) tests can be performed. In projects where a roughened pipeline coating is planned, but the type of material or
roughening method is unknown, soil-soil tests could be performed to provide the upper bound to the interface shear
resistance (e.g. replacing tests 6-10 in Table 5).

• System friction tests. Different interface surfaces exhibit different levels of system friction from the shear ring
sliding over the interface surface. System friction tests with no soil should be performed for each surface used, both
prior to starting the test program and following completion of the program to assess any changes in friction due to
abrasion. Smooth plastic surfaces may become scratched due to coarse-grained material, or rough plastic or concrete
surfaces may wear down, changing the average roughness of the surface materials (e.g. Eid et al. 2015). Application
of a lubricant to the inner walls of the shear ring minimizes the additional shear ring – interface friction caused by
downdrag effects from the soil specimen being loaded. Measuring system friction in both forward and backward
directions allows assessment of any surface preparation effects, and is required for the interpretation of measured
shear stress values described below.

Test Interpretation. Interpretation of the interface shear box test is further described in Boukpeti and White (2017). The
guidance below provides additional insights from the project experience collated in this paper.

• Residual resistance. The residual resistance should be selected from portions of the tests where a stable value is
observed and be based on averaging of a sufficient number of data points to eliminate noise in the data, particularly at
very low stress levels when load cell noise could be significant. For example, using the test data presented in Figure
7a, the shear resistance was averaged over horizontal displacement ranges from approximately 3 mm to 5 mm in both
forward and backward directions. The absolute values of shear resistance in each direction are then compared and
adjusted to remove any asymmetry due to any initial load cell zeroing. Although load cell zeroing should be
performed before a test, the symmetry of the cycles is useful as a check that can be used to update the zero reference.
OTC-26026-MS 19

• Peak Resistance. A peak in shear resistance can occur during a first undrained movement, and a smaller peak is
sometimes observed during subsequent undrained movements following each rest period. The peak resistance may
be estimated from shear box tests, but is generally not relied on in design due to the low likelihood that the peak will
be mobilized everywhere simultaneously along an axially expanding pipeline.

• Cyclic Axial Resistance. The rate of increase in shear resistance with cycles from the initial undrained to final
drained conditions is measured directly in ISB tests. It can also be predicted from critical state parameters, or these
parameters can be back-fitted to ISB test results (Boukpeti & White 2017). Also, scaling approaches exist to estimate
the different levels of consolidation between cycles of movement around a pipeline compared to an ISB tests,
allowing for the different scales and also for different levels of pore pressure dissipation between cycles (White et al.
2015, Low et al. 2017). In some projects, this approach has been used to quantify the modest amount of pipeline
walking that would accumulate before hardening occurred, bringing walking to a halt.

• Mobilization Displacement. In pipeline structural modelling, the axial response is usually modelled using a bilinear
elastic – perfectly plastic model, which requires specification of the mobilization displacement, xres (Figure 1). If a
breakout peak is modelled, there are two mobilization distances, xbrk and xres. For assessments of pipe walking, a low
value of xres creates a higher rate of walking (Hill et al. 2012). To be conservative, a bilinear fit to the non-linear
response should be a tangent fit to the initial part of the axial force-displacement response, which represents the
elastic recoverable part (as indicated in Figure 1). In other situations, a higher value of xres can be more onerous, and
the bilinear fit should be a secant fit to the displacement when the shear resistance is fully mobilized. Since the
mobilization displacement is known to be at least partly linked to the pipeline diameter, element testing is of limited
use for selection of mobilization displacement. In the absence of project-specific assessments, the mobilization
displacements can be selected using the SAFEBUCK model test database (White 2014, Atkins 2015) or equivalent
advice for pile-soil interaction (e.g. API RP2GEO 2011).

Summary

This paper presents experience and best practice for a new geotechnical laboratory test protocol for determining soil properties
relevant to pipeline-seabed friction. A major database of >250 low normal stress interface shear box tests was presented,
comprising marine clay from 16 projects across 8 different hydrocarbon-producing geographical regions, and several types of
pipeline coatings. The database populates a theoretical framework for axial pipe-soil interaction with new data, illustrating
general trends for key parameters such as normal stress, shearing duration, interface roughness, and changing pipeline weight,
each of which is shown to vary the axial resistance by a factor of two or more. The observed trends may be useful for
preliminary calculations of axial pipe-soil interaction following best practice industry approaches, and provide useful insights
into development of site-specific testing programs. While the basic trends have been identified from the database, more
investigation is needed to develop correlations between undrained and drained shear strength and other influential parameters,
including soil index properties. No correlation between shear strength and plasticity index is evident.

References

API (2011). RP 2GEO: Recommended Practice for Geotechnical Engineering.


ASTM (2011). Annual Book of ASTM Standards, ASTM International, West Conshohocken, PA.
Atkins (2015). SAFEBUCK JIP - Safe Design of Pipelines with Lateral Buckling: Design Guideline, Report No. 5087471 / 01 / C, dated 15
January 2015. (Confidential to JIP participants)
Ballard J.-C., de Brier C., Stassen K. and Jewell R. (2013). Observations of pipe-soil response from in-situ measurements. Offshore
Technology Conference, Houston, USA. Paper OTC 24154.
Bishop, A.W., Green, G.E., Garga, V.K., Andresen, A., and Brown, J.D. (1971). A new ring shear apparatus and its application to the
measurement of residual strength. Géotechnique, 21(4): 273–328.
Bolton M.D., Ganesan S.A. & White D.J. (2007). SAFEBUCK Phase II: Axial pipe-soil interaction testing using the Cam-shear device.
Report to Boreas Consultants (SAFEBUCK JIP), ref. SC-CUTS-0609-R01. 56pp. (Confidential to JIP Participants).
Boukpeti, N. & White, D.J. (2017). Interface shear box tests for assessing axial pipe-soil resistance, Geotechnique, Vol. 67, No. 1, pp. 18-30.
Bromhead, E.N. (1979). A simple ring shear apparatus. Ground Engineering, 12(5):40–44.
Carr, M., Sinclair, F. Bruton, D. (2006). Pipeline walking – understanding the field layout challenges, and analytical solutions developed for
the SAFEBUCK JIP. Offshore Technology Conference. OTC-17945.
De Brier, C., Ballard, J-C., & Colliard, D. (2016). On the added value of advanced interface shear testing for pipeline walking mitigation,
Proc. Offshore Pipeline Technology Conf., Amsterdam, Feb. 24, 25.
DNV (2011), On-bottom stability design of submarine pipelines, Oslo, Norway. DNVGL-RP-F109. November 2011
DNV (2017) Pipe-soil interaction for submarine pipelines. Recommended Practice F-114. DNVGL. May 2017.
Eid, H.T., Amarasinghe, R.S., Rabie, K.H. & Wijewickreme, D., (2015). Residual shear strength of fine-grained soils and soil-solid
interfaces at low effective normal stresses, Canadian Geotechnical Journal, 52: 198-210.
20 OTC-28671-MS

Ganesan, S., Kuo, M., & Bolton M.D. (2013). Influences on pipeline interface friction measured in direct shear tests. ASTM Geotechnical
Testing Journal 37(1) 14pp.
Gibson, R., and Henkel, D. (1954). Influence of duration of tests at constant rate of strain on measured drained strength. Géotechnique, 4(1):
6–15.
Hill A., White D.J., Bruton D.A.S., Langford T., Meyer V., Jewell R. & Ballard J-C. (2012). A new framework for axial pipe-soil interaction
illustrated by a range of marine clay datasets, Proc. Int. Conf. on Offshore Site Investigation and Geotechnics. SUT, London.
Jardine R.J., Chow F.C., Overy R.F. & Standing J.R. (2005) ICP design methods for driven piles in sands and clays. Thomas Telford.
Kuo, M. (2011). Deep ocean clay crusts: Behaviour and biological origin. PhD Thesis, University of Cambridge. Dated February 2011.
Kuo M. Y-H., Vincent C.M., Bolton M.D., Hill A. and Rattley M. (2015). A new torsional shear device for pipeline interface shear testing.
Proc. of the 3rd International Symposium on Frontiers in Offshore Geotechnics, Oslo, Norway, 10-12 June 2015.
Ladd CC and Foott R. (1974). New design procedure for stability of soft clays. Journal of Geotechnical Engineering 100: 763–786.
Lehane, B.M., and Jardine, R.J. (1992). Residual strength characteristics of Bothkennar clay. Géotechnique, 42(2): 363–367.
Lemos, L.J.L., and Vaughan, P.R. (2000). Clay-interface shear resistance. Géotechnique, 50(1): 55–64.
Low, H.E., Ramm, M., Bransby, M.F., White, D.J. & Westgate, Z.W. (2017). Effect of through-life changes in soil strength and axial pipe-
seabed resistance for HPHT pipeline design. Proc. Int. Conf. on Offshore Site Investigation and Geotechnics. SUT, London
Mayne, P. (1980). Cam-clay predictions of undrained shear strength, ASCE Journal of Geotechnical Engineering Division, 106 (GT 11):
1219-1242.
Mesri, G., Rokhsar, A., and Bohor, B.F. (1975). Composition and compressibility of typical samples of Mexico City clay. Géotechnique,
25(3): 527–554.
Muir Wood D. (1991). Soil Behaviour and Critical State Soil Mechanics. Cambridge, UK: Cam-bridge University Press, 488pp.
Najjar, S.S., Gilbert, R.B., Liedtke, E.A., and McCarron, B. (2003). Tilt table test for interface shear resistance between flowlines and soils.
In Proceedings of the 22nd International Conference on Offshore Mechanics and Arctic Engineering-OMAE, Vol. 3, ASME, Cancun,
Mexico, pp. 859–866. doi:10.1115/OMAE2003-37499.
Najjar, SS., Gilbert, R.B., Liedtke, E.A., McCarron, B. & Young, A.G. (2007). Residual shear strength for interfaces between pipelines and
clays at low effective normal stresses. ASCE Journal of Geotechnical and Geoenvironmental Engineering, vol 133, no. 6, pp. 695-706.
Pedersen, R.C., Olsen, R.E. and Rausch, A.F. (2003). Shear and interface strength of clay at very low effective stresses. ASTM Geotechnical
Testing J., 26(1):71-783.
Potyondy, J.G. (1961): Skin friction between various soils and construction materials. Geotechnique 11(4): 339-353.
Ramsey, N., Jardine, R.J., Lehane, B.M., and Ridley, A.M. (1998). A review of soil-steel interface testing with the ring-shear apparatus. In
Proceedings of the Offshore Site Investigation and Behaviour Conference, Society for Underwater Technology (SUT), London, UK,
pp. 237–258.
Randolph, M.F., White, D.J. & Yan, Y. (2012). Modelling the axial soil resistance on deepwater pipelines. Géotechnique, 62(9):837-846
Skempton A.W. (1985) Residual strength of clays in landslides, folded strata and the laboratory. Geotechnique 35(1):3-18.
Stark, T.D., and Eid, H.T. 1993. Modified Bromhead ring shear apparatus, Geotechnical Testing Journal, ASTM, 16(1): 100–107.
doi:10.1520/GTJ10272J.
Stark T.D. & Eid H.T. (1994) Drained residual strength of cohesive soils J. Geotech. Engng. 120(5):856-871.
Smith, V.B. & White D.J. (2014). Volumetric hardening in pipe-soil interaction. Proc. Offshore Technology Conference, Asia. Kuala
Lumpur. Paper OTC-24856-MS
Tsubakihara Y. and Kishida H. (1993). Frictional behaviour between normally-consolidated clay and steel by two types of direct shear
apparatus. Soils and Foundations. (2):1-13
Uesugi, M. & Kishida, H. (1986): Frictional resistance at yield between dry sand and mild steel. Soils and Foundations 26(4):139-149.
Westgate, Z.J., Randolph, M.F., White D.J. & Li, S. (2010). The influence of seastate on as-laid pipeline embedment: a case study, Applied
Ocean Research, Vol. 32, No. 3, pp 321-331.
White DJ, Campbell M, Boylan N and Bransby MF. (2012). A new framework for axial pipe-soil resistance: illustrated by shearbox tests on
carbonate soils. In: Proc. 8th Intl Conf on Offshore Site Investigation and Geotechnics. London: Society for Underwater Technology.
White, D.J. & Randolph, M.F. (2007). Seabed characterisation and models for pipeline-soil interaction, Journal of Offshore and Polar
Engineering, 17(3): 193-204.
White D.J. (2014). Axial pipe-soil interaction framework. Report to Atkins and the SAFEBUCK Joint Industry Project Phase III/GEO.
UWA Geo Report 13684v2 176 pp.
White, DJ, Clukey EC, Randolph MF, Boylan NP, Bransby MF, Zakeri A, Hill AJ, Jaeck C. (2017). The state of knowledge of pipe-soil
interaction for on-bottom pipeline design. OTC 27623, Proc. Offshore Technology Conference, Houston.
Yan, Y., White D.J. & Randolph M.F. (2014). Cyclic consolidation and axial friction on seabed pipelines Géotechnique Letters 4:165–169
Yan Y., White D.J. & Randolph M.F. (2016). Elastoplastic consolidation solutions for scaling from shallow penetrometers to pipelines.
Canadian Geotechnical Journal 54:881–895.
Yoshimi, Y. and Kishida, T., (1981). “A Ring Torsion Apparatus for Evaluating Friction Between Soil and Metal Surfaces”, Geotechnical
Testing Journal, GTJODJ, Vol. 4, No. 4, December, pp. 145-152.

You might also like