Centre for Marine Technology

1
Centre for Marine Technology
John D McVee Technical Manager
2
Numerical Predictions of Residual Stresses in
Welded Steel Submersible Hulls

• 1G.P. Campsie, 2A.C. Ramsay and 1J.D. McVee
• Centre for Marine Technology
• QinetiQ Rosyth
• 1QinetiQ Future Systems Technology Division,
  Centre for Marine Technology, Rosyth
• 2MSC Software Ltd., Frimley

3
Contents

• 1 Cracking in structures
• 2 Model tests
• 3 Numerical modelling
• 4 Representation of joint configurations
• 5 Heating and cooling
• 6 Predicted results
• 7 Comparison with experiment
• 8 Conclusions

4
Cracking in Structures

• Cracks in a structure are formed
• during fabrication
• during service
• Cracks in a structure are spread by mechanisms
  such as
• fatigue
• stress corrosion cracking
• fast fracture
• These mechanisms are heavily influenced by
• material and environmental variables, and/or
• applied and fabrication stresses (residual
  stresses)

5
The Military Context

• Compared to a commercial structure, a military
  structure
• may use novel materials
• will operate at high stress levels
• will operate in harsh environments
• will have additional requirements to resist
  weapon attack
• The consequences of cracks in a military
  structure are thus
• at best, loss of availability
• at worst, catastrophic failure and loss of life

6
Cracked T-Butt Weld
7
Technical Issues for Military Submersibles

• Pressure Hull
• internal ring-stiffened cylindrical structure
• high strength steel, which is difficult to
  fabricate
• tensile residual stresses locked in during
  welding ring stiffeners to hull plating
• applied stress levels are a high percentage of
  yield
• Current knowledge of Pressure Hull fatigue based
  on
• model tests
• experimental determination of residual stress
• fracture mechanics based fatigue crack growth
  predictions

8
Large Fatigue Chamber Facility
9
Schematic of Large Fatigue Chamber
10
Internal View of Large Fatigue Chamber
11
Model Tests

• Near full scale thickness Q1N steel hull plate
• Much reduced diameter compared to full scale hull
• Welded to exacting Naval Engineering Standards
• Closure domes attached
• Externally pressurised
• Soft cycled

Supporting non destructive evaluation performed
after pre-determined numbers of cycles gives a
direct indication of fatigue crack growth rates
12
Typical Fatigue Model
13
Fatigue Model being Lowered into Chamber
14
Experimental Determination of Residual Stress

• Weldable strain gauges mounted on hull
• Strains measured before and after welding
• Bending moment at toe of weld estimated by
  extrapolation of strain gauge data
• Extrapolation based on elastic shell theory
• Detailed through thickness residual stress
  distribution not known
• But, assume residual stress peaks at yield at toe
  of weld
• Postulated distribution then set to have same net
  bending moment as that deduced by extrapolation
  of elastic strain gauge data

An extensive database now exists of
experimentally determined residual stresses
15
Fracture Mechanics based fatigue crack growth
predictions

• Fatigue crack growth propagation material
  constants
• Stress intensity factor solutions
• Geometry, R/t ratio, stiffeners, weld profile
• Crack shape and crack gradient
• Applied stresses
• Residual stresses

16
Present Work

• Carried out under auspices of DERA Corporate
  Research Programme (CRP)
• Funded by UK MoD Stakeholders (primarily
  Submarine and Armoured Fighting Vehicle
  communities)
• Recognition that
• a validated numerical modelling methodology for
  prediction of residual stress levels would
  improve current fracture mechanics based fatigue
  crack growth predictions
• a validated numerical modelling methodology for
  prediction of residual stress levels would allow
  rapid assessment of proposed changes in
  fabrication practice

17
Numerical Modelling Methodology - Initial Studies

• Coupled thermo-mechanical analyses using MSC.MARC
• Issues included
• 2-D axi-symmetric representation of the ring
  stiffener to cylinder joint configuration and
  weld cross section
• an explicitly defined and idealised fusion
  profile, extending into the ring stiffener and
  cylinder plating
• weld passes defined individually within
  preprocessor, but lumped together for analysis
• nodal temperatures of 1550?C and 120?C prescribed
  as initial conditions to weld and parent plate
  respectively
• kinematic constraints applied between weld and
  parent plate

Comparison with experimental results revealed a
reasonable correlation A number of improvements
were identified for subsequent implementation to
permit 3-D analysis of thick section multi-pass
welds
18
Overview of Improvements - 1

• Acquisition of accurate material property data
• Physical
• variation of specific heat capacity with
  temperature
• variation of coefficient of thermal expansion
  with temperature
• variation of thermal conductivity with
  temperature
• variation of density with temperature
• latent heat of fusion
• Mechanical
• variation of Poissons ratio with temperature
• variation of elastic modulus with temperature
• series of stress-strain flow curves, obtained at
  different temperatures and at different strain
  rates

Temperature range from ambient up to a nominal
solidus (? 1500 ?C)
19
Overview of Improvements - 2

• Development of a tool based on MSC.Marc
  subroutines
• Functions include
• control of the heat input, i.e., proportion of
  heat directed at weld and parent plate
• moving heat flux
• control of filler material element activation
• User definable inputs include
• number of weld passes
• definition of weld pass start locations
• definition of weld paths in cartesian or
  cylindrical co-ordinates
• variation of power input per weld pass
• Variation of welding torch travel speed per weld
  pass
• Radius of weld pass and hot zone per weld pass

Permits 3-D thermal analysis by accounting for
the pre-heating effect of the welding process on
material ahead of the welding torch
20
Overview of Improvements - 3

• Introduction of Filler Material
• elements defining the filler material are present
  at commencement of analysis
• elements are inactive prior to being reached by
  welding torch
• elements are activated at the correct melt
  temperature
• heat flux is corrected to allow for specific heat
  capacity introduced
• inactive elements take up a configuration based
  on current location of nodes attached to
  activated elements and the original configuration
  of nodes attached only to further inactive
  elements

Automation of filler material element activation
permits reduced modelling and analysis time
21
Numerical Modelling Methodology - Current Studies

• Improvements implemented
• Applicability to multi-pass welding simulations
• 3-D mesh of parent plate and weld passes required
  - elements defining each weld pass grouped
  separately
• Loadcase 1
• temperature of parent plate raised automatically
  from preheat at rate dependent on user defined
  inputs
• at suitable timestep, initial filler material
  elements activated as liquid at appropriate
  temperature, mechanically attached to parent
  plate
• with continued timesteps, additional filler
  material elements activated
• Loadcase 2
• composite structure cools and shrinks, after
  all filler material elements defining weld path
  length have been activated
• Loadcases 1 and 2 sequentially repeated

22
Representation of Joint Configuration - 1

• 2-D Visualisation
• 14 weld passes laid on first side
• 13 weld passes laid on second side
• References include weld history sheets and
  macrographs of cross section
• Created on CAD as an IGES file
• Input to MSC.Marc F.E. pre-processor MSC.Mentat

23
Representation of Joint Configuration - 2

• Symmetry condition imposed at bulkhead
  mid-thickness
• 2-D planar axi-symmetric mesh created
• Rotation of 2-D planar axi-symmetric mesh
• 3-D solid mesh representing a half length 4
  segment of structure created
• 9510 8-noded, iso-parametric, arbitrary
  hexahedral lower order reduced integration
  MSC.Marc elements

24
Heating Load-cases - Weld pass 2

• Pre-heating effect of welding torch prior to
  deposition provided by user subroutine based tool
  
• rapid heating of parent plate
• diffusion of heat into parent plate
• stresses induced, as natural thermal expansion
  inhibited by cooler, stiffer surrounding material
• Activation of filler material elements provided
  by user subroutine based tool
• filler material has no strength at elevated
  temperature
• will contract under expansion of parent plate
• stress fields induced by pre-heating overtaken by
  thermal stresses due to differential cooling
• Three increments selected to move the welding
  torch through one plane of newly activated
  elements
• maintains stability of the analysis

25
Cooling Load-cases - Weld pass 13

• Film coefficient prescribed to all external
  surfaces
• Natural contraction of filler material again
  restrained by cooler, stiffer surrounding
  material
• Residual plastic strains give rise to
• external distortion
• a system of self-equilibrating locked-in residual
  stresses
• Predicted circumferential and longitudinal
  elastic strains shown opposite as output for
  comparison with experimental circumferential and
  longitudinal elastic strains

26
Experimental circumferential and longitudinal
elastic strains
27
Comparison of Results

• Predicted circumferential and longitudinal
  elastic strains extracted from Finite Element
  model internal and external surfaces
• Magnitude of predicted and experimental strains
  in reasonable agreement
• Distribution along internal and external
  surfaces of predicted and experimental strains in
  reasonable agreement

28
Predicted Results - Residual Stresses

• Circumferential residual stresses in cylindrical
  structure analogous to longitudinal residual
  stresses in flat plate - leads to tensile yield
  magnitude stresses in global circumferential
  direction and balancing compression in parent
  material, as confirmed opposite
• Longitudinal residual stresses - tensile yielded
  zone at toe of weld, with compressive zone on
  outer surface of the cylindrical structure

29
Predicted Results - Through Thickness

• Extracted through thickness longitudinal residual
  stress distribution at toe of the weld
• Bending moment calculated
• directly from distribution opposite
• using experimental method applied to predicted
  elastic strains

30
Conclusions and Recommendations

• Computations reveal that the previous
  experimental method (based on extrapolation of
  surface strains) overestimates the bending moment
  and assumed through thickness longitudinal
  residual stress distribution by 30
• Correction factor has been applied to improve the
  accuracy of fatigue crack growth predictions
• Numerical modelling methodology developed could
  be applied for parametric surveys of weld induced
  residual stresses in submersible hulls and
  surface ships

31
(No Transcript)