Proceedings of the ASME 2011 30th International Conference on Ocean, Offshore and Arctic Engineering

OMAE2011
June 19-24, 2011, Rotterdam, The Netherlands

OMAE2011-49

EFFECTS OF TWIST ON CHAIN STRENGTH AND FATIGUE PERFORMANCE:

SMALL SCALE TEST RESULTS

Isabel Ridge Philip Smedley

Tension Technology International BP Exploration
U.K. Sunbury-on-Thames, U.K.

Roger Hobbs
Imperial College London
London, U.K.

commonly to the moorings of FPSOs, semi-submersibles and

ABSTRACT Spar production platforms. In a single mooring leg, chain is
Mooring chain can twist in response to torque generated in the typically combined with wire or fiber rope. The robustness and
adjacent elements of a mooring, or even be inadvertently weight of chain are exploited at the interface with the hull, in
installed in a twisted condition. This paper explores the effect of the touchdown ‘thrash’ zone, along the seabed, and particularly
initial twist on the static strength and fatigue life of studless at the connection of the mooring line to the anchor/pile.
chain. It reports test results for 16mm studless chain (with Between these extremes, polyester rope is often used in deep
similar proportions as the much larger chains now employed in water for its good fatigue resistance and the near-neutral
hydrocarbon production vessels) with known levels of initial buoyancy that gives a taut mooring. In intermediate water
twist subject to axial tensile loading. Twist levels of up to 24° depths, a catenary mooring is more commonly used, employing
per link were examined. wire rope between the top and bottom lengths of chain to reduce
The results were generally better than had been expected. weight.
There was little scatter in the results, giving confidence in the In recent years, there has been a growing awareness of the need
test procedures. The effect of initial twist on static strength was to avoid the generation of torque and twist in mooring
modest, with a maximum strength loss of about 4% at 18° twist components, whether wire rope, fiber rope or chain. This
per link. The effects of initial twist on fatigue life were also awareness includes the possibility of interactions between the
small. Indeed, some twist seemed to improve fatigue individual components of a mooring leg. For example, a non-
performance, perhaps because the contact between links was torque balanced wire rope may twist and induce twist and
displaced from the region of high residual (compressive and torsional moments in an adjacent torque balanced fiber rope.
tensile) stress resulting from proof loading during manufacture. Further, although the undesirability of installing components in
While the results of these small scale tests suggest that twist a twisted condition is understood, it may be difficult to remove
levels up to 24° per link were not detrimental, other practical all twist from the mooring line during installation, especially
reasons remain for continuing to minimize twist in mooring where the line has been prelaid. Consequently, the effect of
chain. Furthermore, caution should be applied in extrapolating twist in mooring lines needs to be assessed.
these results to predict the effect of twist in full scale offshore A straight chain subject to an axial load does not twist nor
mooring chains. generate any torque. However, if chain is twisted while carrying
an axial load, perhaps because adjacent elements in a mooring
INTRODUCTION put torque and turn into the chain, or subjected to axial load
Chain is an essential feature of many mooring systems, ranging when in a pre-twisted condition following installation problems,
from the yokes of tanker wishbone connections on loading it behaves in a complex manner, with the magnitude of the
buoys, to the moorings of drilling semi-submersibles, and most torque dependent both on the axial load and in a strongly non-

1 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 linear way upon the level of twist. This behavior is due to the
way in which the semi-toroidal ends of adjacent chain links
interact with each other as the chain is twisted.
At low levels of twist (≤ 3°/link), the contact between any two
links will occur on a single contact patch on the axis of the
links. Theoretical analysis has shown that, following normal
proof loading, the surface contact is elliptical in shape and
typically one quarter of the chain diameter in size [1].
Proof loading has been found to enhance fatigue endurance
under pure tensile cyclic loading due to the compressive
residual stresses associated with the contact patch [2,3].
However, high tensile residual stresses are induced immediately
surrounding the proof load contact patch which may precipitate
reduced fatigue endurance if the links are subjected to
significant sliding or out-of-plane bending action [4].
Larger levels of twist (around 3°/link to 6°/link) may thus Figure 2. Close-up of Figure 1: the wear scars caused by
result in some contact outside the proof load patch. At higher twisting under axial load can be seen.
levels of twist (>6°/link) the interlink contact will split into two
separate areas as the chain links start to ride up on each other Discussions of the problems which can arise as a result of twist
(and the chain as a whole shortens). At extreme levels of twist in chain have largely focused on the torsional moment which
the chain will lock up completely. Theoretical and experimental will be produced. Little or no attention has been given to some
estimates of the interlink contact with twist have been closely associated issues which may seriously affect the
investigated and reported by the authors [5,6]. integrity and service life of the chain:
Figures 1 and 2 show a chain carrying an axial load in a twisted • The potential to transmit bending moments into and
condition. The gap (end shortening) that has opened between between links (via the separated contact points) that
the links at the original contact can clearly be seen. Note also are much larger than the equivalent moments in a
the ‘polished’ areas on the links where the original contact patch straight chain.
has split and the links have rubbed against each other as turn
was applied in the presence of the axial load. • The effect of twist on the ultimate tensile static
strength.
• The effect of twist on the tensile fatigue performance.
Considering first the issue of bending moment, the problems
experienced with the Girassol Loading Buoy (2002) have
previously been discussed by Jean et al. [1]. They highlighted
the possibility of inducing damaging out-of-plane bending
(OPB) moments in chain links (Figure 3), largely as a result of
the contact surface deformation occurring during proof loading.
For relatively small offset, the deformed contact prevents the
links from rolling against each other and substantially increases
the effective level of friction between the links.

Figure 1. End shortening in twisted chain (27°per link)

2 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 Friction/locking patch

MOPB
MOPB

Figure 3. Out-of-plane bending (OPB) which can occur in chain

links, especially once proof loaded (after [1]).

Interlink twist (Figure 4) will enhance this effect because there Figure 5. Finite element model showing stresses above
are now two oblique contacts between links rather than the 400MPa in a highly twisted chain (29° per link) carrying a very
original single one. Indeed this possibility led to the work small axial load (1% MBL) (after [7]).
reported in references [5 to 7].
The program of work reported here set out to address the lack
of available fatigue data for the performance of twisted chain. A
substantial series of quasi-static and fatigue tests have been
undertaken on straight and twisted chain.
The work has been undertaken on relatively small (16mm bar
stock) chain. This approach was taken to allow a relatively large
number of tests to be undertaken at a reasonable cost. Since the
link geometry is similar for chains large and small, the small
scale tests should provide valuable information as to the likely
Figure 4. Schematic illustrating the increase in OPB moment
which may be obtained when a chain is in a twisted condition. behavior of the large studless mooring chains currently in
Note: The red spots indicate the positions on the inner faces of service.
the links of the contact patches and how their separation may
increase the resistance to rotation between the links. ULTIMATE TENSILE TEST PROGRAM

Chain link fatigue failures at or near the upper chain to stopper Table 1 gives basic information on the 16mm studless chain
connection have occurred on FPSOs in what are thought to be used throughout the test program. While the dimensions are
combined tension and OPB cyclic motions, but in both cases closely geometrically similar to those of offshore chain, the
damage may well have been exacerbated by twist. Wear scars linear scale factor of 1/5 to 1/10 should be borne in mind. The
associated with twist, similar to those shown in Figures 1 and 2, use of a relatively small size was driven by inevitable
have been reported in such instances. considerations of cost and time, while the use of grade 30 chain
with its modest yield strength was governed by availability and
Even if the chain is not bent while in a twisted condition, it is test machine capacity.
known [7] that the maximum principal stresses in a twisted The small scale test chain proportions of 5.85d x 3.67d are
chain will occur in areas away from the original proof loading close to full scale offshore studless mooring chain proportions
bruise (Figure 5). Thus it can be seen that it is important to be of 6.00d x 3.35d, while the proof load of 49% minimum
aware that the tensile strength and fatigue endurance of a chain specified tensile strength (MSTS) was modest compared to
may both be compromised if the chain is twisted. typical offshore studless chain proof loads (to DNV) of 70%
[2]. As reported later in this paper, in reality the proof load was
proportionally even lower at some 32.5% of measured breaking
strength while the full scale studless mooring chain would be
unlikely to have a proof load much less than 65% of the actual
breaking strength.

3 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 Figure 7 shows an example of the type of link failure which
Property Value was obtained in the series of tests conducted at 0° twist/link.
Nominal chain diameter (d) 16.0 mm The failures were at the link-link interface.
Measured chain diameter 15.98 mm
Nominal link length 93.6 mm = 5.85 d
Nominal link width 58.7 mm = 3.67 d
Grade 30
Min. specified tensile strength 96.5 kN
Proof load applied by supplier 47 kN (= 49% MSTS)
Material Carbon-manganese steel,
typically 0.15%/0.20% C,
0.7%/0.9% Mn
Heat treatment Normalized at about 860°C

Table 1. Data for 16mm chain sample.

Preliminary break load tests were conducted on chain samples

11 links long in the untwisted (0° twist/link) condition. Initially
samples were loaded at a rate of 0.1 mm/s whilst developing a
feel for how the sample would behave in the test machine. It
was found that an 11 link sample elongated so much prior to
tensile failure that by about 130 kN the 150 mm actuator had Figure 7. Sample 1604 after failure (0° twist/link).
run out of stroke.
In order to finish the test, the piston was raised and the cross- Referring to Table 2, the results obtained for 0° twist/link tests
head which has a hydraulic lock and unlock and movement was on the 11- and 9-link test samples and on the three 7-link
also moved up. Care was taken to try to maintain the link-link samples show very little scatter. It may be concluded that the
contacts. This was fairly straightforward, as by this stage the interruption to the early test to raise the crosshead did not
chain links were heavily plastically deformed. adversely affect the result.
Subsequent to these trials untwisted samples of 9- and then 7- The average value from these 5 tests was 144.47 kN, i.e. 50%
links length were tested to check that pausing the 11-link test above the manufacturer’s minimum specified tensile strength.
and unloading had not affected the result. For full scale offshore chain, measured break loads are typically
Figure 6 shows some load-strain plots for various twist levels. only 5%-10% above the manufacturer’s minimum specified
Twist does not appear to affect the character of these results. tensile strength.

Test No. Links/ Twist/ Break Mean St.dev

Test Link load
(kN) (kN) (kN)
1601-02 11 143.20
1603 9 146.05
1604 7 0°/link 144.35 144.5 1.04
1605 7 144.08
1606 7 144.65
1607 11 137.69
1608 11 142.87
1609 11 12°/link 141.84 142.8 3.58
1610 6 147.48
1611 6 144.28
1612 6 138.19
18°/link 139.3 1.61
1613 6 140.47
1614 6 141.18
24°/link 140.1 1.51
Figure 6. Load-strain plots for various twist levels. 1615 6 139.05

Table 2. Measured breaking load for samples of 16 mm chain.

4 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 The tests on the chain at higher twist angles were undertaken FATIGUE TESTS
on samples 11- and 6-links long. After the initial tests with
twists of 0°/link and 12°/link, it was decided to conduct further It was decided to run the first fatigue tests in air with a peak
tests at 18°/link and 24°/link (rather than fill the gap between load of about 70% of the break load with a view to obtaining
0°/link and 12°/link as originally specified in the test program). short but representative tests. For the 16 mm chain, 100 kN was
Figure 8 shows the failure of the link in the 24° twist/link conveniently close to this figure, at 69% of the measured break
sample. It appears similar to the failure in Figure 7. load reported above. Again for convenience, a minimum load of
20 kN was chosen, giving a load range of 80 kN and the large
nominal stress range of 199 MPa.
Following some initial testing the following parameters were
investigated:
• Five different levels of twist (0°, 6°, 12°, 18° and 24°
twist/link).
• Three different load ranges (80 kN [20-100 kN],
50 kN [20-70 kN] and 30 kN [20-50 kN]).
Consequently, all the fatigue tests had an upper load
value in excess of the suppliers’ proof load of 47kN.
• The effect on the fatigue performance of applying an
initial static load of 100 kN to the chain in its twisted
configuration.The effect on the fatigue performance of
applying an initial static load of 100 kN to the chain in
an untwisted condition.
All fatigue samples consisted of 11 links and were tested in a
100 kN tensile testing machine. Samples were loaded at a cyclic
Figure 8. Sample 1615 after tensile failure (24° twist/link). load rate of 0.5Hz.
Figure 10 shows a sample of chain in the machine ready for
The results of all the tensile tests are summarized in Figure 9. testing with a twist of 24°/link.
Again, the data appear to be highly consistent within each set of
tests.
It can be seen that for the level of twist induced in this test
series there is very little change in the breaking load. It is
interesting to note that even at extreme levels of twist the loss in
strength is insignificant, the biggest loss in any test being 4.3%
at 18° twist/link (138.19 kN) compared to the average break
load for untwisted chain (144.47 kN).

Figure 10. Sample F1622 (24°/link) in 100kN fatigue machine.

The results for the fatigue tests without an additional 100 kN

Figure 9. Breaking load (as a percentage of the average un-
proof load applied are summarized in Table 3 and Figure 11.
twisted chain strength) versus interlink twist. The results shown in Figure 11 are for fatigue life as a function
of angle of twist between the links with different load ranges.
The sets of results behave as might be expected in that the lower
the load range, the greater the endurance of the chain.

5 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 As one possibility, these results might be explained as a
Test Twist/ Load Measured Mean consequence of displacing the interlink contacts from those
No. Link Range Endurance Endurance generated during the manufacturer’s proof loading.
(kN) (Cycles) (Cycles)
1601 0° 20-100 20360 Table 4 presents the fatigue data for tests where an additional
1605 0° 20-100 7771 proof load of 100 kN (69% UTS) was applied to the specimens
1606 0° 20-100 10623 in their twisted state, and one result (#1628) where the
15845 additional proof load was applied to a specimen in an untwisted
1607 0° 20-100 15259
1608 0° 20-100 19393 state. In the final column of Table 4 the results are expressed as
1609 0° 20-100 21661 a multiple of the equivalent endurance without the additional
1610 6° 20-100 14294 14294 100 kN static load (see Table 2).
1602 12° 20-100 17426
1603 12° 20-100 14470 17169
1604 12° 20-100 19610 Test Twist/ Load Measured Av. incl.
No. Link Range Endurance 100 kN PL
1620 0° 20-70 33724
40233 (kN) (Cycles) / Av. Base
1621 0° 20-70 46741
1615 123221
1618 6° 20-70 58167 58167 0° 20-70 3.30
1626 141947
1619 12° 20-70 56041 56041 1617 120334
1624 18° 20-70 54770 54770 6° 20-70 2.20
1627 135808
1622 24° 20-70 78196 78196 1616 12° 20-70 65685 1.17
1611 0° 20-50 243560 243560 1625 18° 20-70 138341 2.53
1613 6° 20-50 124891 124891 1623 24° 20-70 126271 1.61
1612 12° 20-50 159213 159213 1614 0° 20-50 600000+ >2.46
0
1628 24 20-70 177939 2.28
Table 3. Fatigue test results (without 100 kN proof load).
Table 4. Fatigue results for samples with 100 kN initial load
applied in the twisted state (#1628 in the untwisted state).

Initially, a single test was conducted at 20-50 kN but this had to

be halted as a run-out after 600,000 cycles (≈2.5 times more
endurance that the equivalent base case test), and instead a
series of tests was conducted at the higher load range of 20-70
kN with different angles of twist. With the notable exception of
the test at 12° twist/link all the measured endurance values
following application of the additional ≈70% proof load were
relatively consistent at around 130,000 cycles.
As with the solitary 20-50 kN test, it can be seen that the pre-
load has a significant beneficial effect on the fatigue
Figure 11. Fatigue endurance versus interlink twist for three performance of the chain. As noted earlier, this effect has been
different load ranges. reported for untwisted chain [2,3] and there are, of course,
parallels with a number of other fatigue situations where a
For the highest load range (20 kN - 100 kN) the average periodic or initial overload is known to have beneficial effects,
fatigue endurance is relatively consistent between the untwisted effects thought to be associated with the generation of
and twisted chain (6° and 12° twist/link), although highest for compressive residual stresses as the overload is removed.
the case with most twist. For the 20-70 kN series, all the The single test (#1628) where the overload was applied to the
twisted chains (6° to 24° twist/link) appear to have a greater chain before twisting it produced a result significantly better
endurance than the untwisted chain. For the lowest load range than that where the twist was applied before overloading. The
(20 kN - 50 kN) the results at 6° twist/link show a decrease in plastic deformation caused by the 100 kN load was such that the
endurance, yet at 12° twist/link the endurance has increased “lift” between links on twisting was larger than that seen in the
again slightly. earlier test.

6 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme

 The fatigue results for chains with and without 100 kN preload • The chain tested is only 10% to 20% the size of chain
are summarized in Figure 12. deployed in mooring service today.
• The material grade of the test chains is inferior to that
currently employed offshore.
• The untwisted chain proof load applied by the supplier
was proportionally well below that seen in service.
• The prevention of end rotation in these tests is an
extreme form of restraint that is uncommon in service.
Further work in which cyclic end turn (free or
elastically restrained) was permitted in response to the
cyclic axial load would improve understanding in this
area.

ACKNOWLEDGEMENT

The mooring chain test work reported here was only made
possible by the support of BP Exploration Operating Company
Limited. The authors would like to express their gratitude for
the support of BP in funding and publishing this work in the
public domain.
The authors wish to emphasize that the opinions expressed
Figure 12. Influence of an initial static 100 kN load on herein are those of the authors and may not reflect the opinions
endurance for a range of 20-70 kN at different angles of twist.
of their respective Companies.

REFERENCES
DISCUSSION AND CONCLUSIONS
1. Jean, P., Goessens, K. and L’Hostis, D. Failure of chains
1. Ultimate breaking strength and fatigue tests have been
by bending on deepwater mooring systems, Offshore
performed on 16mm studless chain with similar geometry
Technology Conference, May 2005, OTC 17238, Houston.
to offshore mooring chain. The proof load applied to this
chain during manufacture was proportionately far lower 2. Shoup, G.J., Tipton, S.M., and Sorem, J.R. The
than that applied to the equivalent studless mooring chain influence of proof loading on the fatigue life of anchor
employed offshore. chain, Offshore Technology Conference, May 1992, OTC
6905, Houston.
2. The tests measuring the breaking load of the chain between
0° and 24° twist/link showed an insignificant decrease 3. Tipton, S.M. and Shoup, G.J. The effect of proof loading
(<5%) in strength compared to the untwisted chain on the fatigue behavior of open link chain, J Engineering
breaking strength. Materials and Technology, 114 (1992), 27-33.
3. The base case fatigue tests, where the maximum value of 4. Brown, M., Comley, P., Eriksen, M., Williams, I.,
the three applied load ranges used always exceeded the Smedley, P. and Bhattacharjee, S. Phase 2 Mooring
level of the suppliers’ proof load, also indicated no clear Integrity JIP – Summary of Findings, Offshore Technology
loss in fatigue endurance with twist - in some cases even Conference, May 2009, OTC 20613, Houston.
seeming to indicate an improved fatigue endurance with
5. Hobbs, R.E. and Ridge, I.M.L. Torque in mooring chain.
increased twist. It should be noted that the test conditions
Part 1: background and theory, J Strain Analysis, 40
included full torsional end fixity, while in many offshore
(2005), 703-713.
applications the generated torque might cause simultaneous
cyclic tension and twisting. 6. Ridge, I.M.L. and Hobbs, R.E. Torque in mooring chain.
Part 2: experimental investigation, J Strain Analysis, 40
4. The application of an initial overload to the twisted chain
(2005), 715-728.
(≈70% measured break load) produced a consistent
improvement to the fatigue life, an example of an effect 7. Ridge, I.M.L., Hobbs, R.E. and Fernandez, J. Predicting
familiar in other fatigue problems. The application of an the torsional response of large mooring chains, Offshore
initial overload to the untwisted chain was even more Technology Conference, May 2006, OTC 17789, Houston.
beneficial.
5. The results presented here have certain limitations:

7 Copyright © 2011 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/conferences/omae2011/69570/ on 06/08/2017 Terms of Use: http://www.asme