Marine Structures 58 (2018) 254–277

Contents lists available at ScienceDirect

Marine Structures
journal homepage: www.elsevier.com/locate/marstruc

Testing of jacket pile sleeve grouted connections exposed to

T
variable axial loads
Atle Johansena,∗, Gunnar Sollanda, Andreas Lervika, Martin Strandea, Terje Nybøb
a
DNV GL AS, Norway
b
Statoil ASA, Norway

AR TI CLE I NF O AB S T R A CT

Keywords: Modern offshore jacket structures are designed with piles that are carrying variable axial loads,
Pile sleeve grouted connections large shear forces and bending moments. Even if the geometry of the grouted connection is quite
Shear keys simple the load transfer mechanism is complex. It involves friction, contact stresses, brittle
Axial capacity material, abrasion, cyclic degradation, etc.
Dynamic loading
Current design recommendations were developed based on relatively small scale test and with
Jacket structures
only a few tests performed with cyclic loads that include change of load direction. In order to
study the capacity of grouted connections with shear keys for alternating axial load, four (4)
segment specimens were tested. The specimens were full scale representation of a segment of a
part of the annulus of an axially loaded pile sleeve connection.
The load-carrying axial capacity under cyclic loading was found.

1. Introduction

Jacket structures are used for exploration of oil and gas resources as well as support for generators and substations for offshore
wind farms. The foundations of jacket structures are usually made with piles that are connected to the jacket by a grouted connection
with shear keys. The piles of modern jacket structures are considerably more exposed to alternating axial loading and bending and
shear loading than what was the case when design recommendations as implemented in ISO 19902 [6] and NORSOK N-004 rev 2 [8]
were developed. The reason for this increase is related to the change from inclined (battered) to vertical piles, larger water depths
(hence larger ratio of horizontal against vertical loads) and increase in pile dimensions and steel material strength. However, research
results on the capacity of grouted connection for this type of loads are scarce. This is pointed out by Lotsberg & Solland [1] and Dallyn
et al. [5]. In this study a laboratory test program to investigate the capacity of grouted pile sleeve connections against variable axial
loading was undertaken. Typical normal strength neat cement grout was used. The tests were carried out on specimen representing
segments of the real connection. The segment represented full scale shear keys and grout thickness. The test specimens were designed
with a representative radial stiffness similar to that of grouted connections of piles with diameter in the order of 2 m–2.5 m. The test
specimen design is similar to the design described by Lotsberg et al. [2–4]. The design of the specimens and the test program was
made to ascertain the same load and deflection pattern as in the full scale connection. The focus of the test program was to determine
the capacity for alternating and reversed axial loads as the axial load in typical grouted pile-sleeve connections is dominated by
environmental loads primarily from waves.

∗
Corresponding author.
E-mail address: atle.johansen@dnvgl.com (A. Johansen).

https://doi.org/10.1016/j.marstruc.2017.11.005
Received 11 November 2016; Received in revised form 29 August 2017; Accepted 17 November 2017
0951-8339/ © 2017 Elsevier Ltd. All rights reserved.
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 1. Typical jacket structure with pile sleeve connection detail.

2. Problem description

The typical grouted pile to sleeve connection of an offshore jacket platform is schematically shown in Fig. 1. Piles are driven
through the pile sleeves and when the target penetration is achieved the piles are connected to the sleeves by grouting the annulus.
The piles are subjected to variable axial loads, large shear forces and bending moments primarily from environmental loads like wind
and waves. These loads are cyclic in nature and may load the pile foundation from all directions.
The axial load in the jacket leg are usually transferred to the pile by a grouted connection with shear keys.
The axial force from the sleeve will be transferred as shear force through the grouted connection. The shear keys transfer a large
part of the load through compressive struts between shear key pairs on each side of the grout as indicated in Fig. 2.
As a small relative displacement between pile and sleeve is needed to activate the shear keys, the bottom shear keys will initially
take more load than the shear keys further up in the connection. Due to elastic deformations of the steel parts, typically a relative
displacement between pile and sleeve in the order of a few millimetres is needed at the lower shear keys to activate the entire
connection. As the load alternates, this will over time create local damage to the grout close to the lower shear keys. It is expected that
this type of local damage will progress upwards in the connection over time as the damage makes the lower shear keys less effective
and the shear keys further up will take more of the load.
From experience grouted connections perform well for alternating loads that do not change sign. Less data exists on connections
where the axial loads change sign. Typically alternating reversed load occur only in storm situations for modern jacket structures.
This implies high load levels and relatively few load cycles. The test programme reflects this.
In order to investigate the capacity of pile sleeve connection exposed to alternating reversed axial load a test program was
executed as described in the following chapters.

3. Test specimens

3.1. Design principles

The tests were carried out on specimens representing a segment of the grout annulus and denoted box specimens. For practical
reasons the tests were made as a double sided connection with two grout volumes being tested simultaneously. The specimens were
designed using the same principles as described by Lotsberg et al. (2013) [2] and [4]. Due to concern about scale effects and
contribution to resistance from surface irregularity, Lotsberg et al. decided to design box specimens that represented a similar radial
stiffness as that of large scale cylindrical grouted connections. Steel plates are providing resistance against lateral deformations which

255
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 2. Force flow in grouted pile sleeve connection subjected to alternating axial load (P).

corresponds to circumferential deformations in a real connection.

The principle for radial flexibility is shown in Fig. 3. The aim is to select the properties of the box specimen so the load situation
for the grout is representative for the real connection. The radial stiffness of the pile and the sleeve needs to be considered.
For a contact pressure p between steel and grout the radial deflection of the pile and the sleeve is derived as

Rp2 Rs2
δp + s = p+ p
Etp Ets (1)

or

δp + s Rp2 Rs2
= +
p Etp Ets (2)

where

Rp = Radius of pile
tp = thickness of pile
Rs = Radius of sleeve
ts = thickness of sleeve
E = Young's modulus of steel
p = radial contact pressure between steel and grout

The “radial” deflection of the outer box plate in Fig. 3 is derived as

256
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 3. Box specimen for simulation of behaviour of large diameter connection.

δBox δ /2
= measured
p F /A (3)

where

F = lateral force on spring plates

A = area of grouted section (area including shear keys)
p = radial contact pressure between steel and grout
δmeasured = measured total deflection of the two plates for jacking force F.

The specimens have two volumes with grout, one on each side of the centre plate. Each of these volumes represents the grout in
one connection. The spring plates on each side of the centre plate represent the radial stiffness of the pile and the transition piece.
The geometry of the spring plates was determined requiring the same deflection in the boxes as in the large diameter connections.
This was first assessed by hand calculation, then by detailed ABAQUS analysis and finally by measurements of actual stiffness by use
of jacking devices in the box specimens and measurement of displacement as function of actuator load. The measured stiffness was
used together with equation (2).

3.2. Test specimen geometries

The lateral stiffness of the box specimens were designed to represent certain pile-sleeve geometries. The strength is assumed to be

Table 1
Proposed pile-sleeve dimensions corresponding to measured deflections.

Test 1 2 3 4

Pile diameter dp (mm) 2134 2134 2438 2438

Sleeve diameter ds (mm) 2390 2380 2756 2756
Grout diameter dg (mm) 2207 2207 2557 2557
Pile thickness (mm) 100 100 80 80
Sleeve thickness (mm) 55 50 40 40
Grout thickness (mm) 73 73 119 119
Measured box stiffness (mm/MPa) 0.175 0.183 0.312 0.312
Radial stiffness (mm/MPa) 0.178 0.189 0.315 0.315

257
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 4. Test rig.

a function of the radial stiffness as given in equation (2). Assuming an infinite connection of two concentric pipes the dimensions of
pipes with same radial stiffness as in the test specimens are determined and shown in Table 1.
The box specimen geometry used for tests number 1 & 2 was selected to achieve a specimen that is representative for a typical
large pile to sleeve connection in a jacket structure (with similar radial stiffness). As the same geometry was previously tested with
high performance grout by Lotsberg et al. [4], comparison for similar tests with different grout quality could be made.
The purpose of box specimen number 3 & 4 was to achieve a geometry that is typical for modern jacket pile to sleeve connection
with respect to shear key spacing and radial stiffness.
Full heights of shear keys equal 12 mm was used in these specimens. These shear keys were fabricated from steel rods which
implied lower fabrication tolerances with respect to height than that resulting when using weld beads.

3.3. Grout quality

The four box specimens were grouted with a normal strength neat cement grout (consisting of only water and cement). A specific
gravity of 1.98 kg/dm3 was targeted as per material specification for the targeted jacket. The characteristic cube strength of this
material is ∼60 MPa.

4. Testing

4.1. The test rig

The tests were performed in a Schenk 7500 kN dynamic uniaxial test machine, see Fig. 4. The interface between the test machine
and the box specimen was designed specifically for this test program. The connections between the box specimen and the test
machine are considered fixed-fixed.

258
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 5. Grout mixing and pumping process - principle sketch.

4.2. Grout material

The four (4) box specimens were grouted with a neat cement grout (consisting of only water and cement). The grout constituent
materials were Norcem Portland EN 197-1-CEM I 52.5N and artificial salt water mixed according ISO15711. A specific gravity of
1.98 kg/dm3 was targeted during grout mixing in the laboratory. The measured specific gravity ranged from 1.98 to 2.01 kg/dm3.
To represent the offshore grouting operation the grout was pumped into water filled box specimens through inlets at the lower
part of the specimens. A schematic sketch of the grouting process is shown in Fig. 5.
The specimens for material testing were cast during grouting of the box specimens, using the same batch of grout used to cast each
box specimen. In addition to 28 days specimens for reference properties, the specimens were cured for more than 28 days and until
the actual testing started for each individual box.

4.3. Instrumentation

All the four (4) box specimens were instrumented as shown in Fig. 6:

• Two (2) displacement transducers installed on both sides of the centre plate for measurement of the relative displacement between
centre plates and side wall plates. Note that sliding (failure) may occur along the interface between the grout and the centre plate
and/or along the interface between the grout and the side wall plates.
• Three (3) or five (5) displacement transducers installed on both sides of the box between each “spring” plates for derivation of the
equivalent radial stiffness of the grouted connection, and an assessment of the equivalent radial contact pressure between the steel
and grout.

259
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 6. Instrumentation and channel names (Sg = strain gauge, Disp = displacement transducer, Ea = east, We = west, So = south, No = north).

260
A. Johansen et al. Marine Structures 58 (2018) 254–277

Table 2
Test matrix.

Test No. Test specimen Load/displacement control

Test 1 Same box design as Box 2 in tests described by Lotsberg et al. [4]. Load control 1:1 (fully reversed)
Test 2 Same box design as Box 2 in tests described by Lotsberg et al. [4]. Load control 1:2 (half reversed)
Test 3 Box designed to simulate large targeted jacket platform. Load control 1:2 (half reversed)
Test 4 Box designed to simulate large targeted jacket platform. Displacement control 1:2 (half reversed)

All tests were carried out in air. The dynamic testing was carried out at a frequency of 0.2 Hz.

• Two (2) strain gauges were attached to each side of the centre plate and two (2) strain gauges were attached to the 30 mm thick
box support plates to measure axial strains (stresses) in the steel plates to check magnitude of the axial test load applied and
centricity of the test load application. The strain gauges were located on both sides of the box specimen as shown as coloured dots
in Fig. 6.

In addition the following data were recorded:

• Measurement of the axial load and displacement exerted by the dynamic actuator using the integrated load cell and displacement
sensor.
• Indicative temperature measurements of the test specimen during dynamic load cycling, done by heat flux camera at start and
ending of each dynamic loading increment.

4.4. Test matrix

Table 2 shows the tests performed. The purpose of the tests is briefly described below:

• Test 1: For comparison with fatigue S-N curve from Lotsberg et al. [4] but now with normal strength grout.
• Test 2: For assessment of load ratio effects on fatigue capacity (relative to S-N curve from Lotsberg et al. [4])
• Test 3: Assess capacity of targeted jacket connection geometry tested with targeted load ratio (similar tensile loading relative to
the compressive loading during storm as for targeted jacket)
• Test 4: Assess capacity of targeted jacket connection geometry tested with targeted load ratio (similar tensile loading relative to
the compressive loading during storm as for targeted jacket, in displacement control)

5. Test results

5.1. Material test results

The box tests were complemented by a large number of grout material tests. Table 3 shows the performed material testing to
document the hardened properties of the neat cement grout applied in the box specimens. In addition to the testing outlined in
Table 3, samples documenting the specific gravity of the mixed grout were taken frequently during mixing to ensure that the desired
grout specific density was achieved. Based on the test results the grout properties are considered representative of the installed grout
in the targeted jacket platform.
The specimens for material testing were cast at different stages during grouting of the box specimens, using the same batch of
grout used to cast each box specimen. The cubes were cured in fresh water. The material tests were performed at Aalborg University,
Department of Civil Engineering.
Test results for flexural and compressive strength of 40 × 40 × 160 mm prisms are given in Table 4. The average flexural strength
is 5.1 MPa for the reference specimens tested after 28 days of curing submerged in water at 20 °C. Compared to the high performance
grout used by Lotsberg et al. [4], this value is approximately 33%.

Table 3
Testing program of hardened properties for neat cement grout.

Type of test Standard No. of specimens for each test No. of specimens for reference Total no. of specimens
boxa strengthb

Compressive strength 75 mm cubes NS-EN 12390-3 6 6 30

Compressive strength 40 mm cubesc NS-EN 196-1 12 12 60
Flexural strength 40 × 40 × 160 mm prisms NS-EN 196-1 6 6 30

a
The specified test specimens for each test box specimen was tested at commencement of the dynamic testing.
b
The specimens for the reference strength was tested at an age of 28 days.
c
Compressive strength testing on 40 mm cubes carried out on each broken half from the flexural strength testing.

261
A. Johansen et al. Marine Structures 58 (2018) 254–277

Table 4
Flexural and compressive strength of prisms.

Test No. Flexural Strength (MPa) Compressive Strength (MPa)

a
28 Day reference tests 5.1 70.3b
Test 1 (43 days) 4.9c 80.5d
Test 2 (61 days) 4.2c 82.0d
Test 3 (77/90 days) 5.0c 87.7d
Test 4 (124 days) 4.8c 79.7d

a
Mean value based on nine (9) individual test results.
b
Mean value based on eighteen (18) individual test results.
c
Mean value based on six (6) individual test results.
d
Mean value based on twelve (12) individual test results.

Table 5
Compressive strength of cubes.

Test No. Compressive Strength (MPa)

28 Day reference tests 73.6a

Test 1 (43 days) 82.6b
Test 2 (61 days) 87.4a
Test 3 (77/90 days) 86.0c
Test 4 (125 days) 89.7a

a
Mean value based on six (6) individual test results.
b
Mean value based on five (5) individual test results.
c
Mean value based on eight (8) individual test results.

Test results for compressive strength of 75 mm cubes are given in Table 5. The average compressive strength is 73.6 MPa for the
reference specimens tested after 28 days of curing submerged in water at 20 °C. In the grouting report from installation of the targeted
jacket, the average compressive strength tested on 75 mm cubes after 28 days of curing was derived as 74.5 MPa which is comparable
with the obtained test results.
In addition to the tests described in Table 3 four (4) 30 × 410 mm tubes were filled with grout for assessment of the autogenous
shrinkage development at 20 °C in accordance to ASTM C1698, see Fig. 7. The average autogenous shrinkage derived after 111 days
of curing is 1.849 mm/m. Compared to typical high performance grouts this value is in the order of four to ten times larger.
A number of 150 × 300 mm cylinders cast intended for testing of compressive strength and static modulus of elasticity were
observed to be cracked upon arrival at Aalborg University. The crack patterns were observed as distinct horizontal planes perpen-
dicular to the axis of the cylinders. When the remaining cylinders, apparently still intact, were taken out from the curing container for
test preparations, distinct vertical and horizontal cracks propagated for all remaining cylinders. As a result; no 150 × 300 mm
cylinders were available for testing of the compressive strength and static modulus of elasticity.
The reason for the cracked 150 × 300 mm cylinders is considered to be a combination of heat of hydration and autogenous
shrinkage. Compared to typical high performance grouts developed the recent years for the offshore wind industry, the neat cement
grout used for this project contains no fine/coarse aggregates, additives or admixtures. This in combination to a relative low water to
cement ratio results in a high temperature gradient for such large test specimens caused by heat of hydration during initial curing.
Additionally, as shown above, the applied neat cement grout is more prone to autogenous shrinkage compared to typical high
performance grouts. Most likely weak zones/small cracks developed in the test specimens during initial curing and then at a later

Fig. 7. Autogenous shrinkage development beyond 8.3 h of age.

262
A. Johansen et al. Marine Structures 58 (2018) 254–277

Table 6
Load levels and number of cycles – Test 1.

Load level Load (kN) Number of cycles Total increase in relative displacement (mm) Increase in relative displacement per cycle (mm)

1 ± 400 1973 0.23 0.00012

2 ± 540 1946 0.31 0.00016
3 ± 680 1997 0.70 0.00035
4 ± 820 284 3.27 0.01151

stage the autogenous shrinkage caused the observed crack patterns.

5.2. Test 1

Test 1 was performed in load control with a compression to tension ratio of −1.0. The load was applied in sequences of ap-
proximately 2000 load cycles on increasing load levels, see Table 6. The development in relative vertical displacement between the
centre plate and the side wall plates is shown in Fig. 8. As seen, the development was moderate for the first two load levels, but
increased after 1500 cycles at load level 3. For load level 4 the increase was dramatic, indicating that the connection failed in the last
part of load level 3.
Fig. 9 shows the relative vertical displacement during the initial load cycles and during the last load cycles of the test. Fig. 10
shows the relative lateral displacement during the initial load cycles and during the last load cycles of the test.
Fig. 11 and Fig. 12 shows the damages to the grout after the test. The damages were severe, likely due to the additional load cycles
after initial failure indication during load cycle 3. Two different failure modes are seen; trough thickness cracks between shear key
pairs and local damage close to the shear keys.

5.3. Test 2

Test 2 was performed in load control with a compression to tension ratio of −2.0. The dynamic load was applied in sequences of
2000 load cycles on increasing load levels, see Table 7. The development in relative vertical displacement between the centre plate
and the side wall plates is shown in Fig. 13. The development was moderate for the first four load levels, but increased rapidly after
1000 cycles at load level five. The test was stopped after 74 cycles at load level six as the development in relative displacement
indicated that the connection had failed.
Fig. 14 shows the relative vertical displacement during the initial load cycles and during the last load cycles of the test. Fig. 15

Fig. 8. Relative vertical displacement range (max-min) vs. Number of cycles - Test 1.

263
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 9. Load vs. Relative vertical displacement during first and final load level - Test 1.

Fig. 10. Load vs. Lateral displacement during first and final load level - Test 1.

264
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 11. Condition of grout after Test 1–Surface facing centre plate.

Fig. 12. Shear key on box specimen centre plate after Test 1.

shows the relative lateral displacement during the initial load cycles and during the last load cycles of the test.
Fig. 16 and Fig. 17 shows the damages to the grout after the test. Also for this test two different failure modes are seen; trough
thickness cracks between shear key pairs and local damage close to the shear keys.

265
A. Johansen et al. Marine Structures 58 (2018) 254–277

Table 7
Load levels and number of cycles – Test 2.

Load level Load (kN) Number of cycles Total increase in relative displacement (mm) Increase in relative displacement per cycle (mm)

1 −400/+200 1973 0.02 0.00001

2 −540/+270 1976 0.11 0.00006
3 −680/+340 2057 0.11 0.00005
4 −820/+410 1969 0.25 0.00013
5 −960/+480 1279 1.07 0.00084
6 −1100/+550 74 0.56 0.00752

Fig. 13. Relative displacement range (max-min) vs. Number of cycles - Test 2.

5.4. Test 3

The box specimen was grouted in two steps due to failure of the grout pump during grouting. The transition planes between the
two grout steps were located 675 mm and 945 mm from the bottom of the box specimen. A stinger was used to perform the second
round of grouting. The same failure mechanism was seen during Test 3 as in all other tests. No grout failure was observed in the
construction joint between the two grout sequences.
Test 3 was performed in load control with a compression to tension ratio of −2.0. The dynamic load was applied in sequences of
100 load cycles on increasing load levels, see Table 8. The development in relative vertical displacement between the centre plate and
the side wall plates is shown in Fig. 18. A gradual increase in relative displacement per load cycle was seen for load level 16.
Fig. 19 shows the relative vertical displacement during the initial load cycles and during the last load cycles of the test. Fig. 20
shows the relative lateral displacement during the initial load cycles and during the last load cycles of the test.
Fig. 21 and Fig. 22 shows the damages to the grout after the test.

5.5. Test 4

Test 4 was performed in displacement control with a load compression to tension ratio of −2.0. The dynamic load was applied in
sequences of approx. 1000 load cycles on increasing displacement levels, see Table 9.
Fig. 23 shows the maximum and minimum load through Test 4. A distinct load drop can be observed after each test stop several
times through the test. See Fig. 23. In Table 10 this load drop is eliminated by removing the first 40 cycles after each test stop. This is
done to give a better presentation of the actual stiffness of the grout connection.
Fig. 24 shows the relative vertical displacement during the initial load cycles and during the last load cycles of the test. Fig. 25

266
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 14. Load vs. Relative vertical displacement during first and final load level - Test 2.

Fig. 15. Load vs. Lateral displacement during first and final load level - Test 2.

267
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 16. Close-up of grout surface towards box centre plate.

Fig. 17. Shear-keys on centre plate after test 2.

Table 8
Load levels and number of cycles – Test 3.

Load level Load (kN) Number of cycles Total increase in relative displacement (mm) Increase in relative displacement per cycle (mm)

1 +300/-600 100 0.10 0.00105

2 +350/-700 100 0.04 0.00038
3 +400/-800 100 0.04 0.00039
4 +450/-900 100 0.02 0.00025
5 +500/-1000 100 0.03 0.00027
6 +550/-1100 100 0.02 0.00023
7 +600/-1200 100 0.02 0.00025
8 +650/-1300 100 0.03 0.00028
9 +700/-1400 100 0.05 0.00048
10 +750/-1500 100 0.03 0.00031
11 +800/-1600 100 0.04 0.00037
12 +850/-1700 100 0.03 0.00031
13 +900/-1800 100 0.05 0.00052
14 +950/-1900 100 0.10 0.00097
15 +1000/-2000 100 0.08 0.00084
16 +1050/-2100 100 0.37 0.00372

268
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 18. Relative displacement range (max-min) vs. Number of cycles - Test 3.

Fig. 19. Load vs. Relative vertical displacement during first and final load level - Test 3.

269
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 20. Load vs. Lateral displacement during first and final load level - Test 3.

Fig. 21. Close-up of surface facing centre plate - shear-key 1 and 2 from top.

shows the relative lateral displacement during the initial load cycles and during the last load cycles of the test.
Fig. 26 and Fig. 27 shows the damages to the grout after the test.
Initial settlement of connection after test stop is not included.

6. Observations from the tests

All tests showed a characteristic hysteresis loop with a constant or near constant part with increased stiffness towards the
maximum load. All tests showed an increasing load level needed to move the specimen beyond the neutral position. The load level to
start the movement was typically 100 kN for the first few load cycles increasing to 200 kN at the end of the test.
The stiffness at the peak load on unloading was in all cases larger than when on-loading; possibly due to static friction effects. The
stiffness at the maximum load level was approximately constant for different load levels.

270
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 22. Condition of second shear-key from top - Test 3.

Table 9
Displacement levels and number of cycles – Test 4.

Displacement level Applied dynamic displacement (compression) Applied dynamic displacement (tension) Number of cycles

1 Max. displacement as found during static load to Max. displacement as found during static load to 961
−700 kN +350 kN
2 Max. displacement as found during static load to Max. displacement as found during static load to 1000
−1100 kN +550 kN
3 Max. displacement as found during static load to Max. displacement as found during static load to 960
−1500 kN +750 kN
4 Max. displacement as found during static load to Max. displacement as found during static load to 997
−1700 kN +850 kN
5 Max. displacement as found during static load to Max. displacement as found during static load to 999
−1900 kN +950 kN
6 Max. displacement as found during static load to Max. displacement as found during static load to 999
−2100 kN +1050 kN

All tests showed an increasing deflection (load control) or reduced load (deflection control) for each cycle for all load levels. All
tests in load control were made by gradually increasing the load until the deflection per cycle was rapidly increasing.
The tests with unsymmetrical loading (test 2 and test 3) showed a drift towards the maximum loaded direction. This was not the
case for test 4 as it was tested in deformation control but the decay in loads was larger for the half cycle with the largest deflections. It
may be noted that also test 1 drifted even if this test was symmetrically loaded.

7. Evaluation

7.1. General

The key data for the 4 tests are shown in Table 11 below. The evaluation of the test results are made by studying the following
aspects:

• Lateral (radial) stiffness

• Static strength
• Axial stiffness
• Fatigue strength
The grout strength is taken as the mean cube test values at the time of testing.

7.2. Static strength

Fatigue strength can be presented in a traditional SN-diagram with the fatigue strength presented as a function of the static
strength. Then the fatigue strength is assumed to vary similar to the static strength from the various strength parameters as grout

271
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 23. Max/min load vs. Number of cycles - Test 4 (δ = displacement range).

Table 10
Displacement levels and reduction in load – Test 4.

Displacement level Displacement range (mm) Number of cycles Total reduction in load range (kN)* Reduction in load per cycle (kN)

1 2.0 961 25.8 0.03

2 2.5 1000 26.7 0.03
3 3.8 960 209.7 0.22
4 4.7 997 196.1 0.20
5 5.7 999 358.5 0.36
6 7.0 999 335.1 0.34

strength, shear key geometry etc. It is therefore of interest to compare the static strength formulations in API RP2A [7] and NORSOK
N-004 [9] (NORSOK N-004 recommendations are nearly the same as in ISO 19902 [6]).
The box specimens were designed to have stiffness properties that represent the stiffness of full scale pile-sleeve connections. The
ratio of the maximum cyclic load during the testing over the static is presented in Fig. 28. The figure also show the ratio if API is used
to calculate the static capacity. In NORSOK the mean cube grout strength are used and in API the mean cube strength is multiplied
with an assumed safety factor of 2.6 in order to be comparable to NORSOK. In the Commentary to API RP2A the safety factor is said to
vary between 2 and 16.6 with a mean safety factor of 4.8. As there also should be a bias between the characteristic value and the
mean value in NORSOK a factor of 2.6 is used in the comparison. It could also be noted that it is not quite clear to us if the grout
strength parameter in the API grouted connection capacity formula should be determined as cube strength or the effective com-
pressive strength. In the calculation the mean cube strength is used.
It can also be noted that the radial stiffness is not a parameter in API.
It can be seen from the figure that the maximum loads in the fatigue tests are for all tests significantly less than the static load
meaning that in case of cyclic loading that change the directions a fatigue check will be necessary.

7.3. Fatigue strength

7.3.1. General
The fatigue strength can be presented as a function of the static strength. This is done in Ref. [4]. In this paper the fatigue capacity
is presented as the capacity for the maximum cyclic stress with an equivalent number of cycles as the tests are made with cycles of
gradually increasing load levels. The equivalent number of cycles is determined by assuming that the damage at each level can be
added (Miner rule) and that the shape of the S-N diagram is known. The fatigue stresses is presented as half the cyclic stress relative to

272
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 24. Load vs. Relative vertical displacement during first and final load level - Test 4.

Fig. 25. Load vs. Lateral displacement during first and final load level – Test 4.

273
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 26. 45° crack patterns between first and second shear-key from top on centre plate.

Fig. 27. Condition of grout on lower shear-key on centre plate.

Table 11
Key data for the tests considered in the evaluation.

Test 1 2 3 4

Number of shear key pairs 3 3 4 4

Pile diameter dp (mm) 2134 2134 2438 2438
Sleeve diameter ds (mm) 2390 2380 2756 2756
Grout diameter dg (mm) 2207 2207 2557 2557
Pile thickness (mm) 100 100 80 80
Sleeve thickness (mm) 55 50 40 40
Grout thickness (mm) 73 73 119 119
Grout strength (MPa) 83 87 86 90
Grout E-modulus (MPa) 16000 16000 16000 16000
Width of specimen (mm) 398 398 400 400
Height of shear key (mm) 12 12 12 12
Distance between shear keys (mm) 200 200 325 325
Radial stiffness (N/mm) 454545 434783 416667 416667
Maximum cyclic load (kN) 820 1100 2100 2100
Minimum load/maximum load −1 −0,5 −0,5 −0,5
Load application Load control Load control Load control Displacement control

274
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 28. Comparison of maximum cyclic load during testing and static capacity according to NORSOK and API.

the static strength.

For three of the tests carried out the dynamic load ratio (maximum stress divided with minimum stress) is different from −1,
which was used in Ref. [4]. It is therefore of importance to decide which definition of the dynamic stresses that should be used in the
evaluation of the results.
The static strength in NORSOK (and ISO) is a function of the radial stiffness. It is therefore necessary to consider the radial
stiffness of the specimen when the fatigue capacity is presented in this manner.
In order to evaluate the results from the tests we need to consider the following issues:

• How to evaluate the capacity when the tests are made with cycles at different levels of load?
• How to select the best reference stress for the dynamic stresses?
• How to present the capacity? Relative to static strength or not?
7.3.2. Evaluation of fatigue test results with cycles at different load levels
The test procedure for the tests was set up to start with moderate load levels that were increased after a number of cycles
(typically a few thousand). By carefully monitoring the test in this way it was possible to secure that the tests provide results in the
interesting area even if it is difficult to predict the number of cycles to failure. However, when this method is used, it is not possible to
present the result directly into an S-N-diagram. In Ref. [4] this was overcome by calculating an equivalent number of cycles to failure
by using Miner-rule and the assumption of a known S-N-curve.
In the load controlled tests, a steady increase in the permanent displacement was observed for all load levels. This indicates that
the fatigue capacity will be reached for all the tested load levels. In order to estimate a fatigue limit for all the tested load levels, the
number of cycles needed to reach a limit of permanent displacement was calculated. This limit was set to as 4 mm total displacement
as at this displacement the capacity is reached for the connections at the highest load level. One may argue that the connection would
carry loads at a larger displacement for lower load levels, but as the grouted pile-sleeve connections in jacket structures usually are
designed as a long connection with a large number of shear keys it is necessary to limit the displacement in order to be sure that the
entire length is effective. Hence 4 mm is used as a reasonably failure limit.
By simply extrapolating the average increased displacement for each cycle until a relative displacement of 4 mm is reached, a
fatigue limit for each of the load levels used during the test can be established. See Fig. 29.
For Test 4 which was run in displacement control, the evaluations will be somewhat different. In order to extrapolate the number
of cycles that will represent the fatigue limit of the various load levels tested, it is necessary to transform the load loss per cycle to a
displacement. As for the load controlled tests, it is assumed that failure is reached if the accumulated displacement exceeds 4 mm. The
displacement is found by calculating the displacement needed to restore the load to its previous level using the stiffness that can be
found from the response as the load is increased from one level to a higher level.
By comparing three different methods of assessing the fatigue load namely minimum load, maximum load and half load range, the
following was observed:
Use of the minimum (compression) load gave the largest scatter for the four tests.
Use of maximum (tension) load gave similar capacity for different number of cycles in the three cases with −0.5 load ratio (Test
2, Test 3 and Test 4). But Test 1 that was tested with load ratio −1 gave considerable less capacity.

275
A. Johansen et al. Marine Structures 58 (2018) 254–277

Fig. 29. Fatigue results presented as ratio of the half load range over NORSOK static capacity from the normal grout strength tests presented both as extrapolated
damages and as equivalent cycles.

Use of the half load range shows the minimum scatter compared with a SN-curve in the shape similar to what is shown.
It was observed in the tests that the permanent displacement growth was largest on the most loaded side. Consequently it is reason
to assume that ignoring the minimum load (compression) will not give the most consistent results for different load ratios. That
means that only considering the tensile stress is not recommended. From this it was concluded that using the half load range as the
load parameter seems to be the best choice.
The test results are for comparison plotted against the design SN-curve presented by Lotsberg et al. in Ref. [4] and given in
Equation (4) that is assumed to be valid for structures in water.
y = 0.675 − 0.125 log N for N ≤ 103 cycles
y = 0.510 − 0.070 log N for 10 < N < 105 cycles
3

y = 0.260 − 0.020 log N for N ≥ 105 cycles (4)

7.3.3. Summary of the evaluations

The inspection of the specimens after the tests showed that the grout specimens experienced crushing of the grout at the shear
key. This was not seen to the same degree in similar tests for high strength grout [4].
From the extrapolated data points shown in Fig. 29 it seems that the data indicate an SN-curve that is steeper than the S-N curve
from Ref. [4]. It is probably more in line with the steepness of the first segment of the SN-curve. It may also be argued that the
observed steady increase in permanent displacement that occurred at all load levels will indicate that there will not be a less steep
curve on lower load levels. Instead one may assume, that as long as the connection is sliding for each cycle, the cyclic load will add to
the damage.

8. Conclusions

The cyclic capacity of axially loaded pile sleeve grouted connections has been investigated. Experiments were carried out by four
(4) segment tests. The conclusions that can be drawn from the results obtained are summarised as:

⁃ The cyclic capacity is significantly less than the static strength

⁃ The load ratio is of importance
⁃ Use of half load range as design load parameter gives the least scatter of the three parameters tested
⁃ The specimens made from normal strength grout showed ability to carry load at larger deflections than what has been found for

276
A. Johansen et al. Marine Structures 58 (2018) 254–277

high strength grout

9. Proposal for future testing and analyses

⁃ In ISO 19902 [6] it is stated that loads in the opposite direction (tension) below the plain pipe capacity will not contribute to the
fatigue damage of the connection. This should be demonstrated by box testing as the present basis for this assumption is based on
small scale test specimens.
⁃ Variable amplitude testing should be made as the present tests are done by gradually increasing the load and the behaviour of the
connection for later smaller cycles is not known.
⁃ The behaviour of long connections should be analysed with numerical methods using the data from the box tests to simulate the
behaviour of long joints.
⁃ The detrimental effect of variable bending moment on the axial capacity should be estimated from non-linear FE-analyses of
actual joint geometries.
⁃ Box tests should be carried out for other geometrical parameters as shear key height, shear key spacing, grout thickness, grout
strength and the shear key position on sleeve relative to the pile.
⁃ More tests should be carried out for other load ratios to confirm the adequacy of using the same SN-curve based on half load range
load levels for other load conditions.

Acknowledgements

We would like to thank Statoil Petroleum AS for funding this study and allowing us to publish the results in open channels. We
would like to thank DNV GL AS for funding the article writing.

References

[1] Lotsberg Inge, Solland Gunnar. Assessment of capacity of grouted connections in piled jacket structures. Proceedings of the ASME 2013, 32nd international
conference on ocean, offshore and arctic engineering, OMAE2013 June 9-14. 2013. Nantes, France.
[2] Lotsberg Inge. Structural mechanics for design of grouted connections in monopile wind turbine structures. Mar Struct 2013;32:113–35.
[3] Lotsberg I, Serednicki A, Cramer E, Bertnes H. Behaviour of grouted connections of monopile structures at ultimate and cyclic limit states. Struct Eng February
2013:51–7.
[4] Lotsberg I, Serednicki A, Oerleans R, Bertnes H, Lervik A. Capacity of cylindrical shaped grouted connections with shear keys in offshore structures reported from
a joint industry project. Struct Eng January 2013:42–8.
[5] Dallyn P, El-Hamalawi A, Palmeri A, Knight R. Experimental testing of grouted connections for offshore substructures: a critical review. Structures 2015https://
doi.org/10.1016/j.istruc.2015.03.005.
[6] ISO 19902. Petroleum and natural gas industries - fixed steel offshore structures. 1st ed. 2007-12-01.
[7] API RP2A WSD. Recommended practice for planning, designing and constructing offshore platforms – working stress design. 21st ed. August 2007.
[8] NORSOK Standard N-004. Design of steel structures, revision 2. October 2004.
[9] NORSOK Standard N-004. Design of steel structures, revision 3. February 2013.

277