12 April 2001 Northern California SNAME Meeting 1

Fatigue Prediction Verification of Fiberglass Hulls

Paul H. MillerDepartment of Naval Architecture and Ocean Engineering

U. S. Naval Academy

12 April 2001 Northern California SNAME Meeting 2

Why Study Fiberglass Fatigue?

• Approximately 30% of structural materials now used in the marine environment are fiberglass.

• Little long-term fatigue data exists.

• 1998 Coast Guard data shows 118 fiberglass failures resulting in 6 fatalities

12 April 2001 Northern California SNAME Meeting 3

This Project’s Goals

• Extend the standard fatigue methods used for metal vessels to composite vessels

• Verify the new method by testing coupons, panels and full-size vessels.

12 April 2001 Northern California SNAME Meeting 4

Background-Current ABS Composite Design Methods

• Semi-empirical, theory and previous vessels

• Quasi-static “head”

• Beam and isotropic plate equations

• Conservative

• 2.33 for bulkheads

• 3 for interior decks

• 4 for hull and exterior decks

Factors of Safety(Working Stress Design)

Includes fatigue and uncertainties in loads

12 April 2001 Northern California SNAME Meeting 5

Simplified Metal Ship Fatigue Design

1. Predict wave encounter ship “history”

2. Find hull pressures and accelerations using CFD for each condition

3. Find hull stresses using FEA• Wave pressure and surface elevation• Accelerations

4. Use Miner’s Rule and S/N data to get fatigue life

12 April 2001 Northern California SNAME Meeting 6

Project Overview

• Material and Application Selection

• Testing (Dry, Wet/Dry, Wet)• ASTM Coupons, Panels, Full Size• Static and Fatigue

• Analysis• Local/Global FEA• Statistical and Probabilistic

12 April 2001 Northern California SNAME Meeting 7

Material & Application Selection

• Polyester Resin (65%)• E-glass (73%)• Balsa Core (30%)• J/24 Class Sailboat

• 5000+ built• Many available locally• Builder support• Small crews

Ideally they should represent a large fraction of current applications!

Another day of research…

12 April 2001 Northern California SNAME Meeting 8

Target Structure Analysis• Hull Shell Design

• 35% of LWL aft of Fwd Perpendicular• 0 to 1’ off CL

• Determine loss of stiffness vs. stress cycle history (microcracking)

• Requires knowing load effects and test method bias

12 April 2001 Northern California SNAME Meeting 9

Loads on Target Area

• Hydrostatic

• Hydrodynamic• Slamming• Wave slap• Motion• Foil lift/drag

• Moisture

12 April 2001 Northern California SNAME Meeting 10

Quantified Material Properties• Mostly linear stress/strain

• Brittle (0.8-2.7% ultimate strain)

• Stiffness and Strength Properties Needed (ASTM tests – Wet/Dry)• Tensile• Compressive • Shear• Flex• Fatigue

E-glass Mat/Polyester Sample #1

0

5000

10000

15000

20000

0 0.05 0.1 0.15 0.2 0.25

Extension [in]

Stre

ss [

psi]

Tensile Test

12 April 2001 Northern California SNAME Meeting 11

Moisture Background and Tests

• Porous materials (up to 2% weight)• Few documented moisture failures• Test results ambiguous (Stanford vs.

UCSD)• Test methods suspect (long-term vs.

boiling)• Fickian Diffusion• Tested for 1 year

• Dry, 100% relative humidity, submerged

12 April 2001 Northern California SNAME Meeting 12

Moisture Absorption Results

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

0 33 66 99 132 165 198 231Days

Weig

ht G

ain

TNR TNW SNR SNW Fickian

1.8% weight gain for submerged

1.3% for 100% relative humidity

Equilibrium in 4 months

These results were used for coupon and vessel test preparation.

12 April 2001 Northern California SNAME Meeting 13

Finite Element Analysis

• Coupon, panel, global• Element selection

• Linear/nonlinear• Static/dynamic/quasi-static• CLT shell• Various shear deformation theories used

(Mindlin and DiScuiva)

• COSMOS/M software• Material property inputs from coupon

tests

12 April 2001 Northern California SNAME Meeting 14

Coupon Test Results

• Tensile Mod: 1.2 msi dry, -12% wet, -13% boiled

• Shear Mod: 0.56 msi dry, -11% wet, -16% boiled

• Comp Mod: 0.92 msi dry, -6% wet, -12% boiled

• Tensile Str: 11.3 ksi dry, -20% wet, -24% boiled

• Shear Str: 5.5 ksi dry, -11% wet, -22% boiled

• Comp Str: 25.3 ksi dry, -16% wet, -25% boiled

12 April 2001 Northern California SNAME Meeting 15

Coupon FEA Results

Strains were within 2%, strength within 15%

12 April 2001 Northern California SNAME Meeting 16

Fatigue Analysis for Vessels

0 )(

)(

i

i

sN

dsspfTDE

E[D] = the expected accumulated damage ratioT = the time at frequency fp(si) = the probabilistic distribution of the number of

stress cycles at stress si

N(si) = the number of cycles to failure at stress si

)()(

)cos(),()()()()(

wsswsw

wswsws UTUU

UUUfUpmppfT

12 April 2001 Northern California SNAME Meeting 17

Fatigue Testing

12 April 2001 Northern California SNAME Meeting 18

Fatigue Results – S/N Data

70%

80%

90%

100%

1 10 100 1,000 10,000 100,000 1,000,000

Stress Cycles

Pe

rce

nt

of

Ori

gin

al S

tiff

ne

ss

12.5%-Wet 12.5%-Dry 25%-Wet 25%-Dry 37.5%-Dry37.5%-Wet 50%-Dry 50%-Wet 75%-Dry 75%-Wet

Specimens failed when stiffness dropped 15-25%No stiffness loss for 12.5% of static failure load specimens25% load specimens showed gradual stiffness loss

Moisture decreased initial and final stiffness but the rate of loss was the same.

12 April 2001 Northern California SNAME Meeting 19

Panel Analysis• Responds to

USCG/SNAME studies

• Solves edge-effect problems

• Hydromat test system

• More expensive

• Correlated with FEA

12 April 2001 Northern California SNAME Meeting 20

Panel Test Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2 4 6 8 10 12 14

Pressure (psi)

Def

lect

ion

(in

)S

trai

n (

%)

Wet Defl Dry Defl Wet Strain Dry Strain

Wet vs. Dry results were similar to those from coupons; the one-sided wet specimens were marginally less stiff.

12 April 2001 Northern California SNAME Meeting 21

Panel FEA Results

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 2 4 6 8 10 12 14

Pressure (psi)

De

fle

ctio

n (

in)

Wet Deflection Dry DeflectionPanel Linear Panel NonLinPanel & Frame NonLin Panel & Frame Linear

12 April 2001 Northern California SNAME Meeting 22

Impact Testing• The newest boat had the lowest stiffness.

• Did the collision cause significant microcracking?

Yes, there was significant microcracking!

12 April 2001 Northern California SNAME Meeting 23

Global FEA

• Created from plans and boat checks• Accurately models vessel

• 8424 quad shell elements• 7940 nodes• 46728 DOF

• Load balance with accelerations

12 April 2001 Northern California SNAME Meeting 24

Full-Size Testing – Boat History

• High Mileage – J6• Daily records for 3 years• Annual records since

new• NOAA wind records for

the same period (daylight)

• Course distribution• Velocity prediction

program for speed

The Bottom Line for J6:• 11,300 hours sailing• 10,200,000 wave encounters• The “low mileage” boat had 740 hours and 600,000

waves

0

20

40

60

80

100

120

Janu

ary

Feb

ruar

y

Mar

ch

Apr

il

May

June

July

Aug

ust

Sep

tem

ber

Oct

ober

Nov

embe

r

Dec

embe

r

J6 S

ailin

g H

ou

rs

1996 1997 1998 1999 Ave

12 April 2001 Northern California SNAME Meeting 25

On-The-Water Testing- Set Up

Instrument LocationStrain Gage #1 Portside shroud chainplateStrain Gage #2 Forestay chainplateStrain Gage #3 Inside hull on centerlineStrain Gage #4 Inside hull off centerlineStrain Gage #5 Outside hull on centerlineStrain Gage #6 Outside hull off centerlineAccelerometer Bulkhead aft of strain gages

Instrument Locations for Boat Tests

Location ofsink through-hull (behindfender)

Location of straingauges (underepoxy)

12 April 2001 Northern California SNAME Meeting 26

Data Records

J6 TestImajinationTest

Wind0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

0.0016

2:29

:00

PM

2:29

:02

PM

2:29

:05

PM

2:29

:07

PM

2:29

:09

PM

2:29

:11

PM

2:29

:14

PM

2:29

:16

PM

2:29

:18

PM

2:29

:20

PM

2:29

:23

PM

2:29

:25

PM

2:29

:27

PM

2:29

:29

PM

2:29

:32

PM

2:29

:34

PM

2:29

:36

PM

2:29

:38

PM

2:29

:41

PM

2:29

:43

PM

2:29

:45

PM

2:29

:47

PM

2:29

:50

PM

2:29

:52

PM

2:29

:54

PM

2:29

:56

PM

2:29

:59

PM

Inside on CL Inside off CL

Strains

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1

0.12

2:29

:00

PM

2:29

:02

PM

2:29

:05

PM

2:29

:07

PM

2:29

:09

PM

2:29

:11

PM

2:29

:14

PM

2:29

:16

PM

2:29

:18

PM

2:29

:20

PM

2:29

:23

PM

2:29

:25

PM

2:29

:27

PM

2:29

:29

PM

2:29

:32

PM

2:29

:34

PM

2:29

:36

PM

2:29

:38

PM

2:29

:41

PM

2:29

:43

PM

2:29

:45

PM

2:29

:47

PM

2:29

:50

PM

2:29

:52

PM

2:29

:54

PM

2:29

:56

PM

2:29

:59

PM

AccelAccelerations

12 April 2001 Northern California SNAME Meeting 27

Dockside String-Test FEA

12 April 2001 Northern California SNAME Meeting 28

Slamming FEA

Inner Skin Outer Skin

Using measured accelerations and wave heights from pictures strains were 0.21% for inner and 0.17% for outer. (23% & 18% of ultimate strain)

WS=22.5 knots

12 April 2001 Northern California SNAME Meeting 29

Slamming FEA

12 April 2001 Northern California SNAME Meeting 30

Comparison of Results• Slamming (Low Mileage

Boat- “Imajination”)• Peak measured 0.136%• Ave. of measured peaks

0.117%• FEA prediction 0.125%

Imajination J6Predicted Stiff Reduction -3% -14%

Measured with Strain Gauges -4% -18%

Global "String Test" -14% -52%

With all the fatigue cycles included, the stiffness loss is:

12 April 2001 Northern California SNAME Meeting 31

The Most Useful Conclusions

• The Metal Ship Fatigue Design Process can be extended to composite vessels

• Current factors will lead to fatigue lives of 10-30 years

• Visual clues for fatigue failure are evident

• Stiffness loss may be a better method of prediction

• Good FEA accuracy requires a lot of work!

12 April 2001 Northern California SNAME Meeting 32

Thanks!

• Prof. Bob Bea

• Prof. Hari Dharan

• Prof. Alaa Mansour

• Prof. Ben Gerwick

• Mr. Steve Slaughter

• ABS

• U. S. Naval Academy

• Prof. Ron Yeung

• Gerald Bellows

• Paul Jackson

• My wife, Dawn…

Go Bears!Go Bears!