Design Modifications to Increase Fatigue Life of Fiber Ropes

Design Modifications to Increase Fatigue Life of Fiber Ropes F. Sloan*, S. Bull, R. Longerich The Cortland Companies Cortland, NY USA www.thecortlandcompanies.com

(*currently with Kuraray America Inc., Vectran Division) Abstract - This paper describes a two-phase program on cyclic bend-over-sheave (CBOS) fatigue testing of 40 mm diameter fiber ropes. A test method that simulates heavy marine CBOS applications proved to be a repeatable measure of fatigue resistance. Cycles-to-failure data are presented for ropes made from high-strength fibers of high modulus polyethylene (HMPE), liquid crystal polyester (LCP), and LCP/HMPE blends. A method for pathological examination of fatigue-damaged braided ropes is described, and a map of observed interior damage is presented. The results of the first phase of fatigue testing, coupled with observations from the post-mortem examinations, suggested possible damage mechanisms. Fatigue life in the second phase of testing was increased by more than 100%, by modifying rope designs so as to preclude or delay failure modes observed in the first phase.

I. INTRODUCTION Seismic vessels typically tow a wide array of acoustic sensors when collecting data for use by operators. The ropes used to tow the sensors often pass over large overboarding sheaves as they run from the towing winch to the array package. Premature failures of high-strength synthetic fiber ropes have been observed where the ropes cycle continuously back and forth across the sheave at high frequencies caused by wave/vessel interaction. In order to reduce the possibility of rope failure, which can cause considerable vessel downtime cost for array recovery and data loss, a test program was

initiated to improve the performance of candidate fiber ropes. The test program focused on developing cyclic bend-over-sheave (CBOS) fatigue life data on 40mm diameter fiber ropes, a common size used in the seismic industry. Another aim of the program was to establish a standard test method and performance level that might be used to qualify candidate rope designs. The study was a joint project of the fiber suppliers, the rope producer, and 3 major seismic contractors. The first phase of the study focused on fiber materials and a study of damage mechanisms, and was successful in establishing test procedures and a performance baseline. The second phase focused on improvements to rope designs suggested by analysis of the first phase results. II. MATERIALS TESTED Four fiber materials and/or blends of fibers were tested in the present study. These included two of high-modulus polyethylene (HMPE), one liquid-crystal aromatic polyester (LCP), and one blend of HMPE and LCP. All ropes were 12x12 braided construction, nominal size 40 mm diameter. Actual break strength of the ropes ranged from 110 to 140 metric tons. The ropes were coated during the manufacturing process with water-based abrasion-resistant urethane. Add-on weight in all cases was approximately 5%. III. EXPERIMENTAL During CBOS fatigue testing ropes were cycled over mild

CYCLING STROKE TEST SHEAVE

TENSIONING SHEAVE

APPLIED TENSION

TEST SPECIMEN

TEST SPECIMEN

0.95 m SBZ

1.2 m DBZ

0.95 m SBZ

Figure 1. Cyclic Bend-over-Sheave (CBOS) Fatigue Test Schematic

1

Fig. 2.

machine was cycled at a rate of 150 cycles per hour. Ropes were tested in pairs, i.e. identical ropes were loaded onto each end of the machine. The rope that failed first was taken and examined for failure mode and forensic analysis. The companion rope was taken for residual break strength testing. One rope specimen was cycled on a sheave with tread diameter 1120 mm. Once the test was started, measurements of the rope external temperature were taken using an optical laser pyrometer. The pyrometer is much simpler to use compared with embedded thermocouples, and does not interrupt the test or interfere with the test specimen. Earlier testing at TMT has indicated very good agreement between thermocouple measurement and the pyrometer measurement. Maximum rope temperatures were consistently observed in the double-bend-zone.

Over-sheave fatigue testing

steel sheaves with low-friction bearings as shown in Figs 1 and 2 using a procedure developed previously [1]. The center section (approx. 1200 mm) of each rope was subjected to two bend cycles (straight, bent, straight, bent, straight) per machine cycle, the “double bend zone” or DBZ. On either side of the DBZ was a 950 mm section (one-half a sheave circumference) subjected to only one bend cycle (straight-bent-straight) per machine cycle – the “single bend zone” or SBZ. All testing was conducted at Tension Member Technologies (TMT) in Huntington Beach, California. Sheaves used in these tests had a sheave tread diameter of 570 mm and a sheave groove diameter of 42 mm. The test Table I.

IV. RESULTS OF PHASE I TESTING The results of the first phase of testing are presented in Table I, and show that the test method is repeatable, with identical samples (or similar samples from different production runs) generally failing within 200 cycles of one another. Earlier data [1] generated under the same test conditions suggested a benchmark of 6000 bending cycles to failure (3000 machine cycles in the DBZ)., but also show that a wide range (e.g. 3500 to 6000 cycles) in fatigue life is observed across manufacturers, even for nominally similar rope sizes/designs.

Results of Phase I CBOS Fatigue Testing

Lot

Sheave

Applied

Sample

Fiber

Strength1

Diameter

Load

Sheave

Safety

ID

Type

(tonnes)

(mm)

(tonnes)

D:d

Factor

Sheave Pressure2

A-1 A-2 A-3 A-4 A-5 A-6

HMPE HMPE HMPE HMPE HMPE HMPE

134 134 134 134 134 134

570 570 1120 1120 570 570

18.2 18.2 18.2 18.2 16.4 16.4

14.25 14.25 28 28 14.25 14.25

7.36 7.36 7.36 7.36 8.17 8.17

B-1 B-2 B-3 B-4

HMPE HMPE HMPE HMPE

141 141 141 141

570 570 570 570

18.2 18.2 18.2 18.2

14.25 14.25 14.25 14.25

C-1 C-2

LCP LCP

110 110

570 570

18.2 18.2

D-1 D-2

Blend Blend

110 110

570 570

18.2 18.2

Notes:

Normalized Bending Failure

(%)

0.953 0.953 0.485 0.485 0.859 0.859

4786 4786+ 16600 n/a 5286 5286+

-41.7 --

7.75 7.75 7.75 7.75

0.906 0.906 0.906 0.906

4028 4028+ 4164+ 4164

-49.5 40.9 --

14.25 14.25

6.04 6.04

1.161 1.161

1880+ 1880

57.1 --

14.25 14.25

6.04 6.04

1.161 1.161

4576+ 4576

57.2 --

1. Actual break strength data from rope manufacturing lot. 2. Normalized Sheave Pressure = %Load/D:d 3. Break strength after fatigue cycling, relative to original Lot Strength.

2

Residual

Cycles to Strength3

-62.7

One set (one pair) of HMPE ropes was tested first, failing near 4800 bending cycles. Equilibrium external temperature was 50 ºC. Two sets from a second HMPE were also tested. Failure occurred at 4100+/-80 cycles for the two ropes that parted, again indicating low scatter in the test. The equilibrium temperature of the HMPE ropes was 50-60 oC. One set of LCP ropes was tested. Failure occurred at 1900 bending cycles, within 100 cycles of a previous LCP sample tested in an earlier program. Observed rope temperatures of 80ºC were much higher for LCP than for HMPE. One blended LCP/HMPE rope was tested, with failure at 4590 cycles. This was a major improvement over LCP acting alone (although no improvement over HMPE alone). This indicates that the HMPE fibers help the LCP, perhaps by providing lubricity and protection against self-abrasion. Equilibrium temperatures (65ºC) were much cooler than LCP and close to the HMPE result. One HMPE rope (A-3) was tested at the same load but on a sheave with twice the diameter of the standard sheave. This rope broke at 16,600 cycles, compared with approximately 4800 for the standard sheave. This confirms the strong effect of sheave size on rope fatigue life. A reduced load (samples A-5 and A-6) also produced an increase in fatigue life, as expected. IV. ROPE PATHOLOGY A. 12-Strand Rope Structure Consider a line of visible right-lay strands (or picks) lying along the surface of a 12-strand rope. The cross section in Fig. 3A represents a position along the rope where the center of one of these right lay picks is at the 12 (clock) position. Each parallelogram-shaped block represents one of the twelve twisted strands (in a 12x12 braid each block represents a 12-strand braided sub-rope). Taking another cross section one half of a strand pick along the rope would now reveal the cross section shown in Fig. 3B. The 12 position now lies between two adjacent right-lay surface strands. Moving an additional one half of a strand pick down the rope would show again the cross section of Fig. 3A, and so on, although the position of individual strands changes with each cross section. Using this technique the path of an individual strand can be “tracked” as it works its way along the rope. Figs 4A and 4B show the two cross sections of the rope, as before. However, now an alphabetical sequence of letters (lower-case in place of 12

12 10

R

L

8

10

R L

L R R

2

L

R L

L 6

R

L

R 4

8

R

R

L R

L

2

L L R R 4 L 6

Fig. 3A Fig. 3B Fig. 3. Sequential Cross Sections of 12-Strand Rope

3

the R’s and upper-case in place of the L’s), indicates the progressive path of the strands along the length of the rope (e.g., a-b-c-… etc). The left-lay strands progress in a manner similar to the right-lay but in the opposite direction (e.g. A-B-C…). 12

12 10

a

D

c

F k 8

i

g J

H

2

B

10

L e

4

6

8

b 2 A C j G K d h f 4 I

E

l

6

Fig. 4A Fig. 4B Fig. 4. Strand Position Progression in 12-Strand Rope

B. General Damage Observations Inter-strand abrasion damage began several rope diameters prior to entering the SBZ, presumably as a result of internal strand movement needed to accommodate the load redistribution during bending in the adjacent SBZ section. Abrasion damage was readily apparent and occurred primarily between right-lay and left-lay sub-ropes as they moved relative to one another, as illustrated in Fig. 5. Abrasion damage at these crossing points created concentrations of broken filaments (fuzzing) with occasional severed fiber bundles. No damage was observed within the 12-strand sub-ropes.

Fig. 5. In all rope types, abrasion was observed primarily at the intersections between sub-ropes. Little damage is observed within sub-ropes. In an attempt to better understand the severity and location of sub-rope interactions, rope samples one meter in length were removed from the single-bend-zone at the conclusion of fatigue cycling. Sub-rope segments that appeared on the surface of the rope were then marked using the ”surface” letters of Fig. 4 (A, I, E, and a, i, e.). Ropes were next carefully separated into sub-ropes and the sub-ropes examined individually for abrading surfaces. The alphabetic reference marks on the sub-ropes

stress excursions as the 2-6-10 strands. All of the specimens examined in these experiments showed evidence of sheave groove loading centered at the 1 o’clock position, an indication of a natural self-centering load distribution mechanism. The most significant surface damage was noted where the rope interacted with the sheave sidewalls, a result of abrasion by the sheave. The associated fuzzing may have also resulted from damaged fibrils migrating into the portion of the sheave not contacted directly by the rope. Abrasion from the sheave was mild compared with strand crossover abrasion. A significant amount of hardened polymer dust particles were found in all ropes during inspection. As these particles appeared much harder than the properties of the normally cured coating, it may be an indication that the coatings have undergone a post-curing reaction leading to hardening and embrittlement. The high pressures and temperatures observed during these tests could easily cause post-curing reactions in typical polyurethane systems.

were then used to help identify specific damage locations. The damage interaction map of Fig. 6 quantifies and maps the damage observed during the examination of several sub-rope samples. This was an admittedly subjective process, but the results seem self-consistent and reasonable. Abrasion damage was found to be concentrated at the F-k-D-a / f-K-d-A and C-l-E-j-G / c-L-e-J-g interfaces, and was roughly symmetric with respect to the sheave surface. Although the areas identified do not correspond to the areas of highest pressure (these would presumably occur near the sheave surface), they do seem to correspond qualitatively to areas of high relative motion of sub-ropes during bending. In certain areas (C-l or L-c) damage was created on one surface but not the other, perhaps caused by high motion of one surface while the other remained stationary. A more complete theoretical analysis is likely to show that abrasion damage correlates well to areas where lateral pressure and inter-strand motion are at high levels simultaneously. This may occur at tangent points where the sub-ropes are released from the frictional hold of the sides of the sheave. Two distinct load/stress cases can occur. In the first case the bottom of the sheave groove lies at the 12 position, and the maximum sheave pressure is exerted on a single line of picks. For that condition the loading on the 4-8-12 strands will be different from the 2-6-10 strands. The second load case occurs when the bottom of the sheave groove lies at the 1 o’clock position, thereby load sharing the maximum stresses between two adjacent pick rows (one right lay at 12, one left lay at 2). In this second load case, the 4-8-12 strands will see the same A

B +

a Position on Clockwise (Right-hand lay) Sub-Rope

C

d

HMPE samples tended to fail gradually, with little noise, but with ample prior warning to the test operator in the form of extensive elongation prior to failure. This suggests that creep is contributing to the damage mechanism in HMPE, particularly given the high interior rope temperatures generated within the DBZ. HMPE fibers are known to be sensitive to heat, i.e. their creep rate increases and breaking strength decreases with increasing temperature.

Position on Counter-clockwise (Left-hand lay) Sub-Rope D E F G H I J ***** 12 +++++

10

b c

C. Material-Specific Damage Observations

***** ++++

D F k i

8

g h

g J

H

e

j

****** ++

k

****

***

Fig. 6.

*******

+++

*** **** ++

** + +++++

****** ++

** ++ +++++

+

+++++ +++ +++++ +++ +

**

12

* +++++

*** ++ ******* +++++

* +++++

+++

l

Sheave

+++

4

Note: Each ' * ' indicates damage found on left-lay sub-rope locations A to L. Each '+' indicates damage on right-lay locations a to l. Sub-ropes locations that do not interact are shaded. Damage from the sheave surface was minimal. Diagrams show sheave surface and areas of maximum observed fiber damage.

I

Sheave

B c L

L

Sheave Surface

2

6

e f

a

K

******

* +++++

***

10

8

b 2 A C j G K d h f 4 I 6 E

l

**

Over-Sheave Fatigue Damage Interaction Map for 12-Strand Braided Rope

4

**

+++

HMPE ropes showed extensive regions of fused sub-ropes, i.e. sub-ropes permanently “frozen” into a fixed shape. These hardened regions are formed as the result of plastic flow that can occur in both the fiber and coating materials. (This phenomenon is often observed in high load areas of HMPE ropes, for example at the back of spliced eyes, but has no discernible effect on static rope strength.) It is unclear whether these “frozen” areas would have also been fused at the high temperatures of the test, or whether the materials would have been more flexible. This also suggests that HMPE ropes may experience additional damage during intermittent CBOS fatigue, from forced reorientation of hardened areas within the rope structure. The LCP sample failed abruptly during the testing, with a loud report, and with little warning to the test operator. The failed sample was highly distorted by recoil effects, also suggesting a sudden failure. Close examination of the sample revealed extensive abrasion and cutting of sub-ropes by adjacent sub-ropes. No strand fusing was observed. The blend sample failed during the night and the failure mode was not recorded. However there was no evidence of recoil in the broken ends of the rope, suggesting a gradual failure mode in contrast to 100% LCP.

jackets on individual sub-ropes, to delay the onset of direct strand-to-strand interaction. The added weight, cost, and size often make such jackets difficult to justify. The fatigue process appeared to be accelerated by abrasive particles of coating material broken away from fused sliding surfaces during cyclic bending. Urethane rope coatings tend to post-cure under high pressures and/or temperatures, possibly becoming hard and brittle under CBOS conditions. For the second phase of testing, a modified non-post-curing polyurethane coating was used for all specimens. The HMPE ropes seemed to fare best with respect to internal abrasion. This may be due to the low friction and high abrasion resistance of the fiber. However, these ropes showed signs of creep elongation during the test, particularly when heated by the dynamics of the test. By contrast, LCP ropes showed relatively more rapid abrasion, but did not show evidence of creep even at high temperatures. Blending HMPE fibers with low-creep fibers (e.g. LCP or aramid) in small diameter ropes has been explored by a variety of rope makers as a way to reduce creep in the HMPE. The phase I experiments suggest that such a blend also increases CBOS fatigue life, by forming a synergistic combination of creep resistance and internal abrasion resistance. Thus additional blend samples were tested in the second phase of testing.

VI. DESIGN MODIFICATIONS FOR PHASE II TESTS The primary damage modes identified in the first phase of testing were (1) strand-to-strand (or subrope-to-subrope) abrasion, (2) abrasion by internal hardened particles, and (3) creep elongation. Three rope designs were submitted for the next phase of testing based on targeted approaches to avoiding or delaying these failure modes -- using (1) modified twist/pick counts and/or strand jackets, (2) non-hardening polyurethanes, and (3) blending HMPE with low-creep LCP fibers. The primary damage mechanism observed during these tests was strand-to-strand abrasion. Primary twist levels and/or strand or subrope pick lengths were modified slightly in an attempt to better optimize crossover angles or pressures. However, the most direct and obvious design change placed Table II. Lot 1

VII. RESULTS OF PHASE II TESTING Table II shows the substantial increase in CBOS fatigue life achieved by using the rope design modifications discussed above. Compared to the Phase I and previous baseline of 6000 bending cycles to failure, the Phase II designs outperformed the Phase I ropes by 30-100%, in some cases more than doubling the fatigue life. The data is also shown in Fig. 7. All ropes in the second phase of testing utilized a slightly different twist design and an improved non-hardening urethane coating. The HMPE ropes (F1-F4) achieved an average cycles-to-failure of greater than 8000 cycles, compared to 4800 cycles from the Phase I testing. Temperature measurements indicated an equilibrium external temperature of 65+ ºC.

Results of Phase II CBOS Fatigue Testing Sheave

Applied

Normalized

Bending

Residual

Cycles to

Strength3

Failure

(%)

Sample

Fiber

Strength

Diameter

Load

Sheave

Safety

ID

Type

(tonnes)

(mm)

(tonnes)

D:d

Factor

Sheave Pressure2

F-1 F-2 F-3 F-4

HMPE HMPE HMPE HMPE

142 142 142 142

570 570 570 570

18.2 18.2 18.2 18.2

14.3 14.3 14.3 14.3

7.80 7.80 7.80 7.80

0.899 0.899 0.899 0.899

7814 7814+ 8564 3360+

-80.6 -92.2

G-1 G-2

LCP/HMPE LCP/HMPE

130 130

570 570

18.2 18.2

14.3 14.3

7.14 7.14

0.982 0.982

11708 12302

H-1 H-2

HMPE/Jackets HMPE/Jackets

144 144

740 740

18.2 18.2

14.8 14.8

7.91 7.91

0.854 0.854

11924 not tested

------

Notes: See footnotes Table I.

5

1.4

12000

1.2

10000

1.0

NSP= (%Load)/(D:d)

Bending Cycles-to-Failure

14000

8000 6000 4000 2000

0.8 HMPE A LCP Blend HMPE B New Coating Coated Blend Strand Jackets

0.6 0.4 0.2

0

Baseline Coating

Blend

Jackets

0.0 1000

Fig. 7. Improvement in Fatigue Life as a result of Design Modifications The LCP/HMPE blended ropes (G1-G2) achieved an average cycles-to-failure of greater than 12000 cycles, compared with 4600 cycles achieved in Phase I testing. Equilibrium temperatures exceeded 70 ºC. One HMPE sample (H-1) was tested with “strand jacketing” (braided fiber jacketing of the individual 12-strand sub-ropes). This specimen also achieved almost 12000 cycles-to-failure, indicating a 50% improvement in life due to the presence of the jackets; this is discussed in more detail below. Equilibrium temperatures observed were near 65 ºC. VIII. DISCUSSION

increase in creep rate, so creep elongation was also measured in HMPE and blend samples during testing. The results are shown in Fig. 10. It should also be noted that all creep elongation was occurring over the relatively short length of the SBZ/DBZ where internal heating was at a maximum. The combination of HMPE and LCP fibers more than doubled the fatigue life compared with the baseline, presumably by hindering any heat-induced creep elongation of the HMPE while the HMPE provided additional lubrication and 80

70

o

Temperature Asymptote, C

2T T / d 2 % Load = ∝ D⋅d D/d D:d

100000

Bending Cycles-to-Failure Fig.8. CBOS Fatigue Data from Phase I and Phase II Testing

Fatigue data from the present study is plotted in Fig. 8 as cycles-to-failure against normalized sheave pressure or NSP,

NSP =

10000

that adjusts the results for sheave size and rope strength (T is rope tension, D is sheave diameter, d is rope diameter, %Load is tension as a percent of breaking load). A further discussion of the NSP can be found elsewhere [2]. The significant effect of sheave diameter (D:d) and/or safety factor (%Load) on rope life is clear. The use of advanced coatings and modified twist levels increased the fatigue life by more than 2000 cycles or 30% compared with the previously established baseline performance. Post-test examination showed no evidence of loose hardened particles or fused regions within the rope. Steady-state internal temperatures were measured on one HMPE sample (remaining from Phase I study) over a range of bending cycle rates. A thermocouple was inserted into the rope body for internal measurements. Results are shown graphically in Fig. 9. Of the high performance fibers, HMPE is most affected by high temperature environments, most often as a localized

6

60

50

40

Nominal Cycle Rate

30

20 0

100

200

300

400

500

Bending Cycles per Hour (double-bend zone)

Fig. 9.

Effect of Cycle Rate on Rope Temperature

support for the LCP. The blend design achieved a much greater fatigue life even though the absolute strength was slightly lower than the standard designs. The jacketed-strand rope also more than doubled the fatigue life compared with the baseline. Strand jacketing delays the onset of direct strand-to-strand abrasion while providing some additional protection against exterior damage. Strand jacketing reduces the convenience of complete visual access, but does allow inspection of strand interfaces. The reparability feature of the 12x12 construction is not significantly affected by the addition of the strand jacket. By comparison, an overall jacket (braided over the outside of the entire rope) would not prevent strand interactions, and would greatly impede inspection and repair of the rope. The presence of the jacket increased the rope’s diameter by 25% so that proportionally larger sheaves were required to maintain a comparable nominal sheave pressure (NSP). Strand jacketing also reduces the strength of a rope in this size range by approximately 35% compared to the same size rope with no jacket. Thus if jacketed and unjacketed ropes of equivalent diameter were compared under conditions of equivalent safety factors (assuming equivalent D:d ratios), the jacketed-strand rope would be the preferred choice. However, if jacketed and unjacketed ropes of equivalent size were compared on the basis of equivalent service loads, then the unjacketed rope would far outperform the jacketed. This comparison does not take into account any benefit from improved external abrasion resistance provided by the jacket, or the drawback of increased cost – particularly for the HMPE/LCP blended fiber jacket used in the present study. IX. CONCLUSIONS This study presents the results of a cyclic bend-over-sheave (CBOS) fatigue test program funded by a seismic industry JIP. A repeatable testing format is detailed which can be used to compare new rope materials and/or improved constructions.

A method for detailed pathological examinations of braided rope is also presented. The primary damage mechanism in 12x12-strand braided rope appears to be self-abrasion damage concentrated internally at sub-rope crossovers. Ropes tested in the first phase of this study did not meet a previously established benchmark of 6000 bending cycles-to-failure. Analysis of failure modes in the first phase of testing led to the development of several construction innovations involving twist ratios, coating materials, fiber blends, and strand jackets. Modified rope designs tested in the second phase of testing exceeded previous best cycles-to-failure performance by a minimum of 33%, and in some cases doubled previous best performance. The program goal of 10,000 bending cycles-to-failure was exceeded by two of the three candidates tested. One candidate that surpassed the program goal was made from a blend of HMPE and LCP fibers, and achieved in excess of 12,000 bending cycles to failure. Acknowledgments The authors would like to gratefully acknowledge the seismic companies PGS, Western Geco, and CGG for helping to fund this test program. In particular, we would like to thank Mr. Paul Utvik of PGS, who suggested the test method and provided data from previous studies for comparison. Thanks also to the fiber suppliers Honeywell, DSM, and Kuraray for providing fibers for the test program. Spectra®, Dyneema®, and Vectran® are trademarks of Honeywell, DSM, and Kuraray, respectively. References [1] P. Utvik, “Lifetime of Ropes over Sheaves,” internal report, PGS Exploration AS, Lysaker, Norway, 1999. [2] F. Sloan, R. Nye, and T. Liggett, “Improving Bend-over-Sheave Fatigue in Fiber Ropes,” Oceans 2003, Marine Technology Society, pp. 1234-1237, Sept. 2003.

Crosshead Displacment, inches

16 14 12

HMPE

10 8

Blend

6 4 2 0 0.00

10.00

20.00

30.00

Test Duration, Hours

Fig. 10.

Reduction in Heat-Induced Creep Elongation using Fiber Blends

7

40.00