Failure Analysis Report
Coastal Surface Buoy Mast
Control Number: 3111-00003
Date: 2012-06-29
Prepared by: JAR
Coastal and Global Scale Nodes
Ocean Observatories Initiative
Woods Hole Oceanographic Institution
Oregon State University
Scripps Institution of Oceanography
Failure Analysis of Coastal Surface Buoy and Check of Global Surface
Buoy, Original Offset Mast Designs
Technical Report Number: JAR-TR-06-05-2012
Prepared for:
Woods Hole Oceanographic Institute
86 Water Street
Woods Hole, MA 02543-1052
JAR Classification: Client Confidential
Date Issued : June 29, 2012
Prepared by: ___________________________________
Charles J. Ritter (JAR)
Date
Checked by:
___________________________________
Michael C. Apostal Ph.D. (JAR)
Date
Jordan, Apostal, Ritter Associates, Inc.
35 Belver Avenue
North Kingstown, Rhode Island 02852
Telephone: (401)294-4589
Fax: (401)294-3826
Web Site: http://www.jar.com
References
1) Coastal CAD Solid Model,
SolidWorks 3701-00006-00001_COASTAL_SURFACE_BUOY.sldasm
2) Global CAD Solid Model,
SolidWorks 3701-00002-00001_GLOBAL_SURFACE_BUOY.sldasm
3) Assembly/Weld Drawings
3701-00002_GLOBAL_SURFACE_BUOY_ASSEMBLY_2012_01-11_RevA.pdf
3701-00003_WELL_INSERT_GLOBAL_SURFACE_BUOY_2012-01-25_RevA.pdf
3701-00004_GLOBAL_SURFACE_BUOY_DECK_SUBASSEMBLY_2012-01-10_RevA.pdf
3701-00009_FOAM_COASTAL_SURFACE_BUOY_2011-12-28_RevC.pdf
3701-00011_SURFACE_BUOY_MAST_HEAD_WELDMENT_2011-09-29_RevB.pdf
3701-00018_SURFACE_BUOY_WELL_WELDMENT_2011-12-28_RevC.pdf
3701-00019_DECK_WELDMENT_2011-12-15_RevD.pdf
3701-00025_SURFACE_BUOY_LOWER_HALO_2011-01-18_RevC.pdf
3701-00028_SURFACE_BUOY_WELL_WELDMENT_2011-12-15_RevD.pdf
3701-00034_Stanchion_Base_Weldment_2011-04-27_Rev-A.pdf
3701-00037_Stanchion_Weldment_2011-04-27_Rev-A.pdf
3701-00053_WELL_LID_ASSEMBLY_2011-12-15_RevD.pdf
3701-00165_TOWER_STIFFENER_2011-09-20_RevA.pdf
3701-00184_WELL_INSERT_ASSEMBLY_GLOBAL_SURFACE_MOORING_2012-01-30_RevA.pdf
3701-00185_GLOBAL_SURFACE_BUOY_DECK_ASSEMBLY_2012-01-25_RevA.pdf
4) Weights Spreadsheet 3301-00013_Buoy_Weight_Balance.xlsx
5) Grosenbaugh, M.,"Analysis of AST1 Data and Construction of Nonlinear Stress-Strain
Curve for AST2 Simulations _ Results of AST2 Simulations", 2010
6) Weaver, M. A.,"Determination of Weld Loads and Throat Requirements Using Finite
Element Analysis with Shell Element Models - A Comparison with Classical Analysis",
AWS Welding Journal,1999.
7) Brennen, C.E., "A Review of Added Mass and Fluid Inertial Forces", Navy Civil
Engineering Laboratory document CR82.010, 1982.
8) Missori, S., Sili, A.,"Mechanical behaviour of 6082-T6 aluminium alloy welds",
Metallurgical Science and Technology, 1999.
9) http://plastics.dupont.com/plastics/pdflit/americas/delrin/230323c.pdf
Overview
A test of the Coastal Surface Buoy ended in a structural failure with the mast carried
away and the well box at the mast step grossly failed. See Figure 1 below.
Figure 1 CPSM Buoy Well Insert Frame Damage
Damage was not confined to the mast step attachment area. Fractures are also evident at
the corner brackets that fasten the well insert frame to the deck subassembly frame. One
of the damaged stanchion/stiffener pipe pin ends was also recovered. See Figures 2 and 3
below. (The two lateral mast supports are termed stanchions, the single forward support
is termed a stiffener.)
Figure 2 Stanchion/Stiffener Pipe Pin End
Figure 3 Stanchion/Stiffener Pin End
The pin end is welded per 3701-00165 of Ref 3. The fractured pin ultimately failed at the
end of the stiffener pipe seat and welds. That failure appeared to be due to twisting,
probably as the mast came down. It is also noted that the delrin galvanic isolator failed
leading to the 301 bolt eroding and/or impacting the inner surface of the pin end. The
asymmetric pattern of erosion by the bolt and nut on the pin end - both protective coating
and base metal - indicates that the connection was loose and askew for a period of time
prior to the ultimate failure.
The condition of the pin end is suggestive of an increasing loosening of the connection
followed by damped and then unimpeded excursions due to wave action. The analyses
performed did not account for impact between pin end and bolt.
Modeling
Based on the solid models of Ref 1 and 2, models were created for both coastal and
surface buoys. These models are largely quadrilateral shell models with the float modeled
with brick solid elements. Major components are the mast with instrumentation, the deck
assembly, including the float and outer frame, and the well insert assembly. The
differences between the two models are the depth of the float and the height and weight
of the mast and instrumentation.
Figures 4 through 6 show the submodels. Figure 7 and 8 show the full assemblies.
Figure 4 Coastal Buoy Deck Assembly
Figure 5 Coastal Buoy Mast Assembly
Figure 6 Global Buoy Well Insert Assembly (common with Coastal)
Figure 7 Coastal Buoy with mooring line
Figure 8 Global Buoy (deck plates overlap float top in model display)
Based on the assembly drawings of Ref 3, the constituent assemblies were mated. In the
mast step area, areas were superimposed where the mast step upper and lower plates were
welded to the well insert frame. Only the weld lines were numerically welded together.
This allows for extraction of element edge forces and moments to post-process and
determine weld stress using standard techniques.
The model was not refined enough to include bolted joints between, for example, the well
insert and the deck assembly. These joints were handled with sufficient node to node
couples to created a mated joint without creating spurious hotspots.
Static, harmonic and transient dynamic analyses were performed. For the latter two,
damping was assumed at 10% of critical. Alpha and beta damping were used in the
transient analysis while a uniform damping ratio was used for the harmonic analysis. For
the harmonic analysis, no attempt was made to look at phasing of the loads developed
from the WHOI data tables.
Added mass effects were included based on Ref 7. For large masses, such as the fuel
bladder, mass moments of inertia were included. The case of a sphere just broaching the
surface was selected and the added mass is equal to half the water mass of the volume of
the sphere. This was selected as a reasonable fit for both vertical and lateral translational
movement.
The mooring line was modeled with a simple linear spring with stiffness derived from
Ref 5 in the cyclic force operating range.
Loads
The following waves were selected for the transient analysis and to abstract for the
harmonic analysis. These waves were at peak acceleration values within the .csv files of
1700 seconds duration. The plots show displacements based on double integration net of
gravity and inclination.
Displacement Plots for Waves 1, 2 and 3 developed from 6-Axis Accelerometer data
obtained from WHOI Web Site cgsn-omc.whoi.edu/oms
U displacement is lateral (in the plane of the two stanchions), V displacement is vertical
and W displacement is in the plane of the mast and forward stiffener.
Wave 1 - File 20110924_080000.3dmgx3.csv
2.5
Displacement (meters)
2.0
1.5
1.0
u displ
0.5
v displ
w displ
0.0
-0.5
0
2
4
6
8
10
12
14
16
-1.0
-1.5
Tim e (sec)
Wave 2 - File 20110925_050007.3dmgx3.csv
1.5
Displacement (meters)
1.0
0.5
u displ
0.0
v displ
0
2
4
6
8
-0.5
-1.0
-1.5
Tim e (sec)
10
12
14
16
w displ
Wave 3 - File 20111030_090008.3dmgx3.csv
4.0
Displacement (meters)
3.0
2.0
u displ
1.0
v displ
0.0
w displ
0
2
4
6
8
10
12
14
16
-1.0
-2.0
-3.0
Tim e (sec)
Results
Coastal Surface Buoy
Harmonic Analysis
Table R-1 below summarizes the results of a simple harmonic response analysis based on
a selected set of wave acceleration and rotation versus time data as logged in the Coastal
Surface Buoy. A modal analysis predicted three modes at approximately 0.5 hz
corresponding to mooring line response. The next set of three modes is at about 3 hz. The
wave excitation is at about 0.3hz, about an octave below the first mode sets. As a check,
static analyses were performed and similar results were obtained, indicating a lack of
modal excitation.
In areas of interest, weld lines were selected and adjoining elements were processed for
direct and shear forces plus moments along the weld edge of the elements. The three
regions selected are the well tower mount flange, the stanchion/stiffener deck mount pin
ends, termed the strut lugs in the tables and for the forward stiffener mast connection, the
U shaped fitting at the mast, termed the bridle in the tables. Using the procedures
described in Reference 5, weld throat effective stress was calculated. (These follow AWS
procedures). The adjacent elements' maximum effective stress is also tabulated as the
HAZ stress. (In actuality, the model refinement is too crude to pick up the heat affected
zone stress with any degree of confidence, but the tabulated values are indicative of local
stress risers. Also, the bridle weld calculation is suspect due to the extreme thickness of
the shell elements approximating it, although HAZ stress is good). Lastly, the maximum
force in the mast stiffener pipes is tabulated.
Figures 9 through 11 show the regions evaluated.
Figure 9 Mast Step Upper Plate and Well Frame Flange
Figure 10 Stanchion Base Weldment (Bolts to Well Housing Frame)
Figure 11 Mast Forward Stiffener Upper Connection (evaluated weld at end of u-shaped
bridle to oblong tab, which in turn is bolted to masthead fittings)
In Table R-1, Case 1 is the harmonic loading assuming no loosening of the struts. Case
2 assumes loosening of a lateral strut. Case 3 assumes loosening of the forward strut and
Case 4 assumes loosening of all struts.
Case1
Case2
Case3
Case4
Strut Lugs Stress,
MPa
HAZ
Weld
34.1
30.4
5.3
5.6
135.4
92.6
17
14.6
Bridle Stress,
MPa
HAZ
Weld
69.2
27.2
66.1
50.8
9.2
19.4
9.2
19.4
Mast Step Stress,
MPa
HAZ
Weld
11.8
35.8
9.9
43.3
118
372.9
134
493.1
Struts Force,
N
Force
6252
8169
6636
0
Table R-1 Harmonic acceleration loading results for 3 m/sec2 acceleration three axes
simultaneously - period = 22 seconds
Transient Dynamic Analysis Coastal Buoy
For three selected waves, transient dynamic analysis has been performed, using processed
data based on well accelerometer and inclination logs. The results for the Coastal Buoy
are summarized in tables R-2 through R-4 below. Only the first three cases described in
the above section are summarized.
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
47.1
57.9
100.0
89.3
50.8
69.5
Bridle Stress,
MPa
HAZ
Weld
78.6
31.1
20.4
24.6
74.5
42.4
Mast Step Stress,
MPa
HAZ
Weld
13.4
85.4
48.8
414.0
8.9
80.4
Table R-2 Transient dynamic loading results for Wave 1, Coastal Buoy
Struts Force,
N
Force
18801
18274
21885
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
13.4
17.3
26.3
24.1
10.4
14.9
Bridle Stress,
MPa
HAZ
Weld
37
5.7
6.5
4.8
3.4
6.8
Mast Step Stress,
MPa
HAZ
Weld
3.1
19.4
12.8
97.9
2.3
24.7
Struts Force,
N
Force
1812
3173
867
Table R-3 Transient dynamic loading results for Wave 2, Coastal Buoy
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
59.3
69.4
191
143.6
58.8
76.4
Bridle Stress,
MPa
HAZ
Weld
251.0
51.6
9.8
20.6
219.0
62.3
Mast Step Stress,
MPa
HAZ
Weld
18.2
114.2
92.6
692
164.5
Struts Force,
N
Force
25311
21837
21536
Table R-4 Transient dynamic loading results for Wave 3, Coastal Buoy
Transient Dynamic Analysis Global Buoy
For the Global Buoy, the same three wave loadings have been utilized to predict the
effects of a taller and heavier mast. Similar assumptions are made. As with the Coastal
Buoy, the stress predictions for some of the cases and waves exceed material strength,
indicating early low cycle failure.
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
126
92.9
350
221.6
160.0
133.6
Bridle Stress,
MPa
HAZ
Weld
439
16.7
19.5
2.2
439.0
168
Mast Step Stress,
MPa
HAZ
Weld
29.1
184.4
196.0
1457
31.1
233.4
Struts Force,
N
Force
18028
25649
26897
Table R-5 Transient dynamic loading results for Wave 1, Global Buoy
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
39.9
30.2
135.0
86.1
43.0
32.8
Bridle Stress,
MPa
HAZ
Weld
92.5
3.4
8.5
1.0
92.4
45.1
Mast Step Stress,
MPa
HAZ
Weld
7.6
39.4
73.6
542.8
4.3
45.1
Table R-6 Transient dynamic loading results for Wave 2, Global Buoy
Struts Force,
N
Force
5356
10234
5861
Case1
Case2
Case3
Strut Lugs Stress,
MPa
HAZ
Weld
144.0
103.7
490.0
303.3
142.9
102.2
Bridle Stress,
MPa
HAZ
Weld
687
24.8
25.1
2.9
691.0
25.3
Mast Step Stress,
MPa
HAZ
Weld
52.4
375.0
296.0
2222.8
69.3
472.9
Struts Force,
N
Force
25365
32623
25343
Table R-7 Transient dynamic loading results for Wave 3, Global Buoy
Observations
Both the harmonic analysis, with only in-phase acceleration, and the transient dynamic
analysis with acceleration phasing and pitch/roll/yaw effects included, highlight a number
of high stress areas. This assumes movement of the stanchion/stiffener pin end
connections. The 3701-00002_GLOBAL_SURFACE_BUOY_ASSEMBLY_2012_01-11 drawing of
Ref 3 indicates a bolt torque of 120-150 ft-lbs for the connections. Using the standard
torque/clamping force equation, this will result in a minimum clamping force of 9600 lbs
or 42720N. Ref 9 publishes static CoFr values for the delrin galvanic isolators used in
the stiffners' connections as ranging from 0.10 to 0.30. The corresponding dynamic
values range from 0.20 to 0.40. These values are for delrin/steel, delrin/delrin and
delrin/zytel and do not include the low friction PTFE filled formulation. The CoFr of
delrin on the buoy protective paint isn't known. The published values allow for slippage
even at the high end of the friction range. It is also noted that there is some assembly
clearance for all the connections, plus the forward stiffener pin end is elongated for
fitting. The approximate stiffness of the forward stiffener is 1.0 E5 (coastal) to 1.7 E5
(global) lbs/in or 1.75 E7 to 3E7 N/m. The forces predicted in Tables R-1 to R-4 do not
necessarily bottom the bolt and delrin isolator against the pin bore. Stick slip could result
in the chafing of the protective paint of the grossly failed stiffener pipe pin end as seen in
Figure 2, which would tend to unload the bolt clamping force.
Although there are hot spots in the deck mount lugs and welds (which have not failed)
and in the bridle (unknown), the highest magnitude hot spots occur in the top plate of the
mast step weld. The outer edge of the weld between the well insert H section frame and
the top plate spike well above ultimate strength for base metal.
The shear stresses in the welds of the stanchion/stiffener pin ends are on average only
18.3 MPa. A detailed tetrahedral model was developed and predict very localized high
stress but this result was judged to be an artifact of trying to model the welds. The results
aren't consistent with the clean fracture of Figures 2 and 3 of the recovered pin end.
Neither buckling of the stiffener pipes or progressive fracture of the pipe end pin fittings
seems likely. For the Global Buoy, the highest stiffener force is close enough to the
optimistic Euler prediction to have warranted a closer look.
The most likely mode of failure predicted by simulation is movement and loosening of
the stanchion/stiffener pin end connections followed by progressive failure of the mast
step top plate weld. Higher forces generated by the partial failure of the welds would lead
to a breach of the well and eventually the observed gross failure of the mast step. As the
mast is carried away, twisting the pin ends, the failure is consistent with the damage
shown in Figure 3.
There are a two alternatives to the original design. The one chosen, to be checked and
summarized in a supplemental report, is to centralize the tower with much more robust
support legs than the mast step design. If a specific application dictated, an alternative to
that is one that preserves the offset mounting. The original design requires moment
resistance over a 15cm moment arm for a 2.6M mast with instrument weight at the top
plus turbine and panel weight lower down. The Global Buoy is even taller and heavier
and based on the analysis results the moment resistance arm is too short for the mast. The
mast could be extended downwards and fasten to the outer frame to increase the moment
arm. However, this would require wholesale modification of the float and the outer frame
plus this configuration might be more difficult to assemble and maintain.