Engineers in Private Practice

by
Charles R. Morin, PE
Chairman, Prin. Engr. and Co-founder
Engineering Systems, Inc.
Chicago, Colorado Springs, Houston,
St. Louis, New Orleans, Ft Myers and
Atlanta
www.esi-website.com

Failure Analysis: Why is it

Important?
• Product Design and Development
• Product or System Performance
Improvement
• Forensic (legal and insurance)

1
MILLER PARK CRANE
COLLAPSE
Milwaukee, Wisconsin
July, 1999

Root Cause Investigation

Lessons “Learned”
and Safety Implications

2
 Presentation Outline
• Initial On-Scene Documentation
• Evidence Committee
• Metallurgical Examinations
• Wind Studies/Drag Loadings
• Structural Analysis (FEA)
• Foundation/Soils
• Dynamic Analyses
• Root Cause(s) Determination

ESI Project Team

C. R. Morin, P.E., Proj. Mgr., Metallurgical
D. A. Wojnowski, Ph.D., P.E., Civil & Met.
P. A. Lenzini, P.E., Geotechnical
Cleon Biter, Consulting Meteorologist
R. C. Winn, Ph.D., P. E., Wind Loads and Dynam.
J. A. Wilkinson, P.E., Mechanical Eng.
M. T. Kenner, P.E., FEA
S. J. Smith & A. V. Kenner, Graphics and Animations

3
 Evidence Committee
• Representatives of interested parties
including; OSHA, States Attorney,
General Contractor, Subs, Union, MHIA
& Lampson
• Site Documentation and Metallurgical
Examinations as per Joint Protocols

Initial On-Scene Efforts

• Scene Documentation; aerial, stereo and
video
• Survey; mapping and elevations
• Identification and logging of parts
• Selection of samples for testing/analysis

4
5
6
 Metallurgical Examinations
• King pin failure mode, ~ pure shear
• King pin machined from 12 inch diameter, 4340
steel forging; YS - 103ksi, UTS - 130ksi,
USS - 89ksi, CVN 25-28ft-lbs @84 degreesF.
• Comparisons to engineering drawings and
specifications
• Non-standard soft brass spacer from lower
connection of king pin assembly.

7
8
9
 Wind Studies
• Steady (Mean) and Gust Wind Speeds
• Wind Measurements from Surrounding Reporting
Stations
• Gradient Speed at Altitude
• Assessment of Terrain and Roughness Effects
• Wind Conditions at Site - 19mph, Gusting to 31mph
• 19 mph> 7 kip drag; 31 mph (3 sec.) Gust > 20 kip
• Wind Alone Could Not Account for Failure
• Lampson (and Industry) Uses Steady Wind for Load
Chart, Per Expert, 2% Side Load in Service is Safe.

10
 Structural Analysis
• Stress Analysis of King Pin Connection by
Conventional and FEA Analyses
• Detailed Elastic/Plastic Analyses of King Pin
Assembly Using Contact and Friction
• FEA of King Pin Assembly for “Equivalent Spring”
Used in Global FEA Model
• Wind Loads Applied at Centroids of Segments
• Determined P-Delta Effects for Entire Crane on a
Rigid Level Base & Rigid Inclined Base

11
12
13
 Foundation/Soils
• Geotech Est. Differential Settlement of the Crawler
• Crane was about 0.7% Out of Level, (3” Differential
Settlement due to Non-Uniform Soil Stiffness)
• Manufacturer Called for < 0.2%, (3/4” in 40’)
• Manufacturer Provided Geotechnical Consultant
with the Track Pressure for the Project (Uniform
Pressure Assumed!)
• MHIA Used Ground Pressure Limit to Design
Concrete Runways to Support the Crane
• Manufacturer Insisted Track Pressure was Always
Uniform!

14
 Dynamics Analyses
• Load Swing - Static Wind Drag Force, Rigid Crane
• Load Swing - 3 Sec. and 10 Sec. Gusts, Rigid Crane
• Compound Dynamic Model, Finite Difference
Technique for Load and OTM vs. Time
• Parameters Studied by Bounding Analyses:
– Wind Gusts and Durations
– Crane and Foundation Stiffness
– Spacer Yielding (backlash)
– “ENDO” Correction

15
16
17
 Conclusions
Root Cause found to be a combination of:
• Load swing from stopping crane (plus
“ENDO”)
• Mean wind loading within rated limits; Gust
loading equivalent to ~2% side load
• Lack of sufficient rigidity; Flexible crane
• Soft soil
• King pin split retainer with a soft 1/2” thick
spacer.

Safety Issues
• Very large cranes should have adequate stiffness and side load
capacity >2%
• Provide tilt meter and/or alarms to monitor/warn of critical out
of level operation
• Characterize fully the soils at the site of heavy lifts
• Crane mfgs. should provide realistic track pressure distribution
which (including side load effects) will allow proper foundation
design
• Give operating crew/user better training/information regarding
crane limitations
• Planning heavy lifts requires especially close coordination by
highly skilled personnel to assure nothing of significance is
overlooked (QA principles could serve as a basis for action)

18
 The Miller Park Crane Collapse –
Analysis of the King Pin Failure

Matthew T. Kenner, P.E., John A. Wilkinson, P.E., and Charles R. Morin, P.E.

Engineering Systems, Inc.

3851 Exchange Ave.
Aurora, IL 60504
(630) 851-4566

Abstract: While lifting a one million pound section of space frame roof structure at the Miller
Park construction site, a crane failed and overturned, resulting in three fatalities and extensive
damage to the stadium. During the ensuing failure investigation, the strength of the crane’s king
pin, a 12″ diameter, 11′ long steel shaft about which the front crawler body rotates, was a key
issue along with any changes in king pin strength that may occur with various king pin connection
details. ABAQUS/Standard was used to investigate these questions with a quarter symmetry solid
element model of the king pin and its associated connection details. The model included multiple
3D deformable body-to-deformable body contact surfaces and material nonlinearities. The
analysis showed the variation of king pin strength with differing king pin connection details and
this information was used as part of the multi-disciplinary investigation to determine a probable
cause for the failure of the crane.
Keywords: Failure Analysis, Contact, Plasticity

1. Introduction

The construction of Miller Park, a retractable roof baseball stadium, required the use of a highly
specialized, heavy-lift crane. The crane was used to lift, move and position large sections of the
space frame roof structure. While lifting a one million pound section of space frame, the heavy-
lift crane failed and overturned, resulting in three fatalities and extensive damage to the stadium.
Engineering Systems, Inc. (ESI) worked as part of a multidisciplinary team to investigate the
cause of the heavy-lift crane collapse. Issues addressed during the investigation of the collapse
included material, structural, aerodynamic, geotechnical and dynamic issues, among others. This
paper focuses on the finite element analysis study of the crane’s king pin in ABAQUS/Standard,
which was used to assist in the king pin failure analysis as well as the development of a full crane
finite element model which was also used during the investigation.

2006 ABAQUS Users’ Conference 249

2. Background

The heavy-lift crane used at the Miller Park construction site had a somewhat unique
configuration, consisting of two separate crawlers 100 ft. apart and connected by a space frame
structure called a stinger (see Figure 1). The boom and mast connected to the front crawler tub,
which rested and rotated on a 14 ft. diameter slide bearing. The rear crawler carried 2.4x106 lbs.
of counterweights to resist the forward over-turning moment of the lifted load. In order to
translate a load, the front and rear crawlers, each with a separate operator, were moved in unison.
When swinging the boom, the front crawler remained stationary and the rear crawler moved about
it. The height of the load was controlled by a third operator stationed on the stinger. At the time
of the collapse, the jib tip was roughly 530 ft. above the ground.

Figure 1. Heavy-lift crane configuration.

The front crawler tub rotated on the front crawler using the king pin as a pivot. The king pin is a
12 in. diameter, 11 ft. long, solid shaft made of 4340 steel with a yield strength of 103,000 psi. In
normal service, the king pin does not see any upward loading and acts purely as a pivot. If the
crane is subjected to excessive side loading, the king pin can be exposed to upward loading due to
tipping of the front crawler tub. When this occurs, the king pin and its connections must resist an
axial, tensile load.

250 2006 ABAQUS Users’ Conference

The original design of the king pin connection had a large nut at both the top and bottom of the
king pin, both of which threaded onto the king pin as shown for the top connection in Figure 2.
The lower connection was later changed to include a roughly 2 in. thick split ring which engaged a
recess in the king pin and was locked in place by a three quarter inch thick locking ring. Both the
split ring and locking ring were made of A514 steel with a 100,000 psi yield strength. During the
collapse investigation, the as-built lower connection detail was found to vary from that specified in
the drawings (as-drawn), mainly by the addition of a half inch thick spacer made of CDA 932
bronze with a yield strength of approximately 19,000 psi. Another notable difference between the
as-built connection and the as-drawn connection was a change in thrust bearing material from the
as-drawn CDA 836 brass (17,000 psi yield strength) to as-built CDA 863 bronze (68,000 psi yield
strength).

Figure 2. King pin connection details.

On the day of the collapse, the heavy-lift crane was lifting and positioning a section of the roof
which was roughly 100 ft. by 200 ft. in size (Figure 3). The section of roof was estimated to
weigh over 900,000 lbs and, with rigging and balance weights, totaled 1x106 lbs. At the time of
the collapse, the load was approximately 250 ft. in the air. The crane, counterweights and lifted
load totaled over 6.2x106 lbs. A view of the crane and stadium after the collapse is shown in
Figure 4.

2006 ABAQUS Users’ Conference 251

 Figure 3. Heavy-lift crane with load.

Figure 4. Overview of the collapse.

252 2006 ABAQUS Users’ Conference

Examination of the lower king pin connection after the collapse showed that the end of the king
pin was sheared off below the split ring recess (Figure 5). This allowed the front crawler tub to
separate from the front crawler and was the initiating event in the collapse sequence of the heavy-
lift crane. As a result, the strength and stiffness of the lower king pin connection became one of
the main focuses of the investigation and was studied using ABAQUS/Standard.

Figure 5. Lower king pin failure.

3. Analysis
Two three-dimensional, solid element, ABAQUS/Standard finite element models were built in
order to analyze the as-drawn and as-built king pin configurations (see Figures 6 and 7). Each
model focused on the components at the lower connection of the king pin and front crawler, and
included only the lower 30″ of the king pin. The locking ring and split ring were tied together in
the area of the retaining bolts. All other surfaces in the models which may have contacted each
other were modeled with 3D-to-3D deformable body contact, resulting in 13 separate contact
surfaces in the as-built model. Because the nature of the geometry dictates that the lower king pin
connection is loaded axially, each model could take advantage of quarter symmetry to reduce the
overall degrees of freedom.

Of course, the overall strength of the lower king pin connection was of great interest to the
investigation. Equally important, though, was the stiffness of the king pin and its lower
connection. This is because once the king pin is loaded in tension, any deformation results in
additional displacement of the jib tip, which increases the load on the king pin (the so-called P-δ
effect). Because the analysis was developing a load versus deflection curve for the lower
connection, the top of the king pin was loaded via displacement control, which is much more
stable during a contact simulation.

2006 ABAQUS Users’ Conference 253

All components were assigned elastic/plastic material properties in order to account for material
yielding which occurred in each component prior to any individual failure within the connection
detail. Generic material curves were used for all components except the king pin, which was
sectioned and subjected to tensile testing as part of the investigation. The behavior of the
connection varied with friction between the components and numerous friction values were
analyzed during the investigation process. This paper focuses on the analyses with a lubricated
friction value of μ=0.1.

The models were analyzed using ABAQUS/Standard version 5.8 run on a PC (circa 1999). Run
times were as long as 24 hours, covering two steps and roughly 20 total increments. The models
were run again with ABAQUS/Standard version 6.5 on a PC (circa 2006) with run times of
roughly an hour.

3.1 As-drawn king pin model

The as-drawn king pin model is shown in Figure 6 and is comprised of the following components:

• CDA 836 leaded red brass bushing/thrust washer

• ASTM A514 steel split retainer

• ASTM A514 steel locking ring

• AISI 4340 steel king pin

• A572 steel front crawler body.

The model consists of 7983 C3D8R and C3D6 solid elements and 10435 nodes. 3D-to-3D
deformable body contact was defined between components as required and tied contact was
defined on the king pin and crawler for purposes of mesh refinement. The model has 37638
degrees of freedom.

3.2 As-built king pin model

The as-built king pin model is shown in Figure 7 and is comprised of the following components:
• CDA 863 manganese bronze bushing/thrust washer

• ASTM A514 steel split retainer

• ASTM A514 steel locking ring

• AISI 4340 steel king pin

• CDA 932 high leaded tin bronze spacer

• A572 steel front crawler body.

254 2006 ABAQUS Users’ Conference

The model consists of 8445 C3D8, C3D8R and C3D6 solid elements and 11171 nodes. 3D-to-3D
deformable body contact was defined between components as required and tied contact was
defined on the king pin and crawler for purposes of mesh refinement. The model has 40674
degrees of freedom.

Figure 6. As-drawn king pin model.

Figure 7. As-built king pin model.

2006 ABAQUS Users’ Conference 255

4. Results and Discussion

4.1 Load versus deflection curves

Load versus deflection curves for the as-drawn, as-built and nut connection are shown in Figure 8.
The curve for the nut connection is based on hand calculations, while the as-drawn and as-built
curves represent the finite element analysis results. It is clear that the nut connection is
appreciably stiffer and stronger than either connection with a split ring detail. Under normal
service this is of no consequence, as the king pin is not loaded in tension and the strength and
stiffness of the lower king pin connection is not germane. Once the king pin is loaded, an unusual
and critical condition in and of itself, the split ring design detail is significantly less robust and
will, because of its relative compliance, further amplify the load on the king pin even before any
failure occurs within the connection.

6.0E+06
Nut

5.0E+06

4.0E+06
Load [lbf]

3.0E+06

As-Built, μ=0.1

2.0E+06
As-Designed, μ=0.1

1.0E+06

0.0E+00
0.00 0.05 0.10 0.15
Deflection [in]

Figure 8. Load versus deflection curves for the king pin models.
The results shown in Figure 8 were used as the basis for a non-linear spring element in a full crane
model used to investigate the P-δ effect on the crane under side loadings large enough to load the
king pin in tension. This analysis is discussed in a companion paper presented elsewhere in the
2006 ABAQUS Users’ Conference (Wilkinson, 2006).

256 2006 ABAQUS Users’ Conference

4.2 As-drawn king pin model stress results

Figure 9 shows a von Mises contour plot of the as-drawn lower king pin connection
under a 2300 kip load. This plot clearly shows that the highest stresses in the king pin
and split retainer occur 90 degrees away from the split in the split ring. At this load,
yielding of the king pin and split ring has occurred, although yielding at the corner of the
king pin lower land is very localized.

Figure 9. As-drawn king pin at 2300 kip load [psi].

A von Mises contour of the thrust bearing in the as-drawn lower king pin connection
under a 2300 kip load is shown in Figure 10. There is significant yielding in the thrust
bearing under this load and the thrust bearing is near failure. The highest stresses in the
thrust bearing occur at the split in the split ring.

2006 ABAQUS Users’ Conference 257

 Figure 10. As-drawn thrust bearing at 2300 kip load [psi].

4.3 As-built king pin model stress results

The von Mises stress state of the as-built lower king pin connection under a 2400 kip load
is shown in Figure 11. As with the as-designed lower king pin connection, yielding of
the king pin and split ring has occurred at this load level and is most severe 90 degrees
from the split in the split ring.

Figure 11. As-built king pin at 2400 kip load [psi].

258 2006 ABAQUS Users’ Conference

Figure 12 depicts the spacer of the as-built lower king pin connection under the same
loading. The spacer exhibits significant yielding and is at the ultimate stress of the
material adjacent the split in the split ring. The analysis shows that the split ring rotates
upward under load and works to extrude the spacer from the lower king pin connection
stack-up. The split ring at its split basically cuts into the spacer as would a shear, making
the loading at this point of the spacer most severe. As the spacer is extruded, the analysis
shows the original rectangular cross-section is tapered under the severe loading. This
predicted deformation of the spacer correlates well with the physical evidence, as shown
in Figure 13.

Figure 12. As-built spacer at 2400 kip load [psi].

Figure 13. Post-accident cross section of the as-built spacer.

4.4 Failure analysis

The overall investigation showed that sufficient side loads were developed due to wind
loads, soil settlement, P-δ and dynamic effects to load the king pin significantly, but not

2006 ABAQUS Users’ Conference 259

enough to shear the end of the king pin (which requires a load of approximately 6700
kip). Figure 8 shows the strength of the as-drawn and as-built lower king pin connections
before first failure are similar, on the order of 2400 kip. At this load, which is achievable
under the conditions of the lift, the spacer in the as-built connection will fail.
Physical evidence shows that, upon failure, the spacer was ejected from the lower
connection stack-up, moving the jib tip further to the side. This dynamically increased
the load on the lower king pin connection and failed the thrust bearing (which, as can be
deduced from the as-drawn analysis, is loaded by the split ring similarly to the spacer).
The thrust bearing failure results in further movement of the jib tip and a further increase
in the king pin load. At this point, enough load can be developed to fail the end of the
king pin in shear and the front crawler tub is free to overturn resulting in collapse of the
heavy-lift crane. This step-by-step failure of the lower king pin connection, as predicted
by the finite element analysis, is supported by the physical evidence and a video (w/
sound) of the collapse.

5. Conclusions

• The lower king pin connection of a heavy-lift crane involved in a construction site accident
was modeled using ABAQUS/Standard as part of a multidisciplinary failure investigation
into the crane’s collapse.

• The finite element analysis study provided a stiffness curve for the lower king pin connection
which was used in the development of a full crane model. The strength of the as-drawn
lower king pin connection at first failure was found to be about 2300 kip. The strength of the
as-built lower king pin connection at first failure was found to be about 2400 kip.

• The failure investigation showed that loads at the time of the accident were enough to exceed
the strength of the spacer in the lower king pin connection as predicted by the finite element
analysis. The finite element analysis also provided insight into the subsequent failures of the
lower king pin connection which eventually lead to the collapse of the heavy-lift crane.

• The finite element analysis technique is a powerful and effective failure analysis tool.

6. References

1. ABAQUS/Standard User’s Manual, Version 5.8, Hibbit, Karlson & Sorenson, Inc.,
Pawtucket, Rhode Island, USA, 1998.
2. Wilkinson, J.A., Kenner, M.T. and Morin, C.R., “The Miller Park Crane Collapse – Analysis
of the Crane”, 2006 ABAQUS Users’ Conference, Boston, MA, May, 2006.

260 2006 ABAQUS Users’ Conference