EverFE Workshop
Sacramento, CA March 11, 2004 Bill Davids, PhD, PE University of Maine [email protected]JPCP Is a Complex Structure
• A 7”-12” layer of concrete on a base, sub-base, soil • Subjected to a variety of axle loads and fatigue effects • Experiences seasonal and daily temperature changes • Sawn transverse joints every 12’ – 15’ (+/-)
• Transverse joints often doweled for better load transfer • Adjacent slabs may be tied at longitudinal joints
JPCP Is a Complex Structure
I-90 in Washington State
Dowel Retrofit
Contraction Joint
Failure Modes in JPCP
Failure Modes in JPCP
Corner Break
Failure Modes in JPCP
Transverse Joint Faulting
Failure Modes in JPCP
Shrinkage CrackingMechanistic-Empirical Design
of JPCP
1. Estimate design parameters (thickness, joint spacing, etc.) 2. Predict responseunder axle loads, temperature changes, etc.
3. Assess effect of
stresses on fatigue life and durability Not OK 4. Plans and Specs, Bid, Construct OK Construction Problems?
Neg. gradient Pos. gradient
2. Linear thermal gradients through the slab thickness
Predicting Response of JPCP
Usually a Westergaard-type analysis
~slab~ wheel
1. Three critical wheel load positions are assumed
Edge Interior Corner
3. Slabs are founded directly on a dense liquid
4. Assumes an infinitely large slab, no joint load transfer
• Essential for understanding pavement behavior • Critical for developing rational design methods • Important in forensic analysis of pavement failures 2. Predictions of pavement structural response are: 1. Limitations of Westergaard-type analysis are severe
3. Clear need exists for better JPCP analysis tools
What is EverFE?
• Software for the 3D Finite Element (FE) analysis of JPCP • Incorporates specialized strategies for modeling important
response characteristics
• Utilizes problem-specific solvers for efficiency
• Integrated modeling software and graphical user interface • Intuitive model construction and result visualization
• Allows the generation of models with varying complexity
Anatomy of an EverFE Model
Basic model characteristics: • Up to nine slab/shoulder units • Up to three base/subgrade layers • Dense liquid supports model • Dowels, ties, aggregate interlock Loading:
• Multiple axle types • Thermal gradients
Extensive post-processing:
• Slab stresses and displacements • Dowel results
Workshop Objectives
1. Familiarize you with EverFE’s capabilities • Overview basic finite-element concepts • Cover details of EverFE unique capabilities 2. Give you hands-on experience with the software
• Generate and run models
• Increasing level of model complexity 3. Explain what EverFE can and can’t do
Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
Finite-Element Concepts
Mathematical definition:functional method for solving partial differential equations
Our definition:well-established numerical technique for determining stresses, strains and displacements in engineering structures
Finite-Element Concepts
Why is FEA so popular?
• Applies to wide classes of problems
• Excellent for irregular geometries
• Easily treats different boundary conditions
• Easily generalized for computer implementation • Easily handles spatially varying material properties • Well-suited to nonlinear and dynamic problems
Finite-Element Concepts
Optimization in Mechanical Design
Analysis of a Welded Connection Analysis and Design of a Floor Slab Structural Analysis of a Frame
Finite-Element Concepts
FE Procedure in a Nutshell:
• Divide a structure into discrete inter-connected “finite elements” that meet at “nodes”
• Make each finite element responsible for defining an approximate solution over its domain
• Take the original governing differential equation and re-cast it using the properties of the finite elements (the mathematically difficult part)
• Solve the resulting system of equations for unknown displacements, recover stresses, etc.
Finite-Element Concepts
Simple problem from structures/strength of materials
x
f(x)
Elastic rod of length L, elastic modulus E, area A, fixed ends
Governing differential equation: 2 ( )
2 x f dx u d EA =
Finite-Element Concepts
Finite-element discretization and solution
element nodes
stress
exact solution
Finite-Element Concepts
How does each element represent the solution?
1D linear element
nodal displ. interpolated displ.
1D quadratic element
constant stress linearly varying
stress
Finite-Element Concepts
Basic Element Types in Structures and Solid Mechanics
2D Elements Beam/Truss Element
Plate/Shell Elements
t
Finite-Element Concepts
History of FE Modeling of Concrete Pavements
• Earliest models treated slabs as plates on elastic solids • ILLISLAB, JSLAB, etc. released in late 1970s, early 1980s
Modeling of multiple slabs with 2D plate elements Methods for handling joint load transfer
• Researchers began using existing general-purpose 3D codes
Detailed models of doweled joints Treatment of slab-base interaction
• EverFE was first released in 1998
Development started in 1995, has continued until present
Finite-Element Concepts
Important Issues to Bear in Mind: FEA is an approximate method
Model must closely mimic physical reality
Accurate material properties Appropriate boundary conditions Reasonable representation of loads
The proper elements need to be used in discretization Sufficient mesh refinement is essential
Finite-Element Concepts
What is the peak tensile stress in a large slab with: 40 kN wheel load applied at the edge, r = 228 mm Slab properties: t = 254 mm, E = 27,600 MPa, v = 0.20 Subgradek = 0.027 MPa/mm
40 kN
~ Very large slab ~
Finite-Element Concepts
1948 Westergaard Solution: max= 1.43 MPa Finite-Element Solution:
Build model with quadratic solid elements Represent load with a 405mm x 405mm
square contact area (equivalent area to circle) Critical questions:
• How large a slab to model?
Finite-Element Concepts
Finite-Element Solution:
Start with a large slab (5000mm x 5000mm) Study the effect of mesh refinement on solution
2 x 2 elements 5000mm 5000 mm 24 x 24 elements increase # of elements
Examine the effect of model size on solution
Finite-Element Concepts
0 10 20 30 0.6 0.8 1 1.2 1.4 1.6 WestergaardNumber of Elements Along Edge (Only even number of elements used)
M a x im u m S tre ss (M P a )
Effect of Mesh Refinement on Results
2 elements through thickness
1 element through thickness
Discretization 12 u12 24 u24 Stress (MPa) 1.48 1.43 2 Elements through thickness
What if we change our discretization slightly?
Load is centered in element: • Element captures linear
variation in stress • Element can’t see peak
stress! 13 u13 25 u25 1.23 1.33
Finite-Element Concepts
Finite-Element Concepts
Effect of Model Size on Results
1000 3000 5000 7000 9000 0.5 0.7 0.9 1.1 1.3 1.5 M a x im u m S tre ss (M P a ) Slab Size (mm)
Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
• Obtaining EverFE and Program Architecture
Generating an EverFE Model
16-noded interface element
20-noded brick element
8-noded dense liquid element x
y z
beam elements for dowels and transverse ties
Generating an EverFE Model
Example Analysis
• Single slab, 5000mm long x 3600mm wide x 250mm thick • Founded on 125mm thick bonded CTB with E = 7000 MPa • Single 120 kN, dual wheel axle located at edge
Plan Elevation 120 kN axle 5000mm 3600m m slab bonded CTB
Workshop Topics
• Introduction• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
Slab-Base Interaction
The base layer is rarely bonded to the slab
• Slip (relative horizontal movement) between slab and base • Vertical separation of slab and base may occur
z =1.01mm, max= 2.53 MPa Unbonded Base
Consider the model we just solved
Bonded Base (as solved) z =0.91mm, max= 1.42 MPa
Slab-Base Interaction
EverFE’s treatment of slip and vertical separation
Slab-base interface may be fully bonded or tensionless Slab and base layer are meshed separately
1mm or 0.1 in slab
base
corresponding pairs of nodes
permanently tied if base is bonded (linear)
Slab-Base Interaction
EverFE’s treatment of shear stresses at interface
slip, Interface elements relate
slip to shear stress
0
δ
kSB
Defining parameters
Shear stress-slip relation:
S h e a r stre ss
Applies onlywhen slab and base
remain in contact
Slab-Base Interaction
Background on shear stress-slip relation
Shear stress is caused by several mechanisms Classical friction
Interlock (interaction of two rough surfaces) Adhesion (chemical bond)
This elastic-plastic model has seen recent use in literature Rasmussen and Rozycki (2001)
Zhang and Li (2001)
Slab-Base Interaction
What are typical values for kand 0?
Data reported by Rasmussen and Rozycki (2001): Base Type
Rough HMA Smooth HMA Rough Asphalt Stabilized Smooth Asphalt Stabilized
Cement Stabilized Granular kSB (MPa/mm) 0.270 0.068 0.200 0.065 4.100 0.027 0 (mm) 0.250 0.510 0.510 0.640 0.025 0.510
Slab-Base Interaction
Quick parametric study
• Re-run our single-slab model with an unbonded base • Let 0= 1mm, vary kSB– use say 0, 0.5, 1, 2, 5, 10, 50
• Study the effect of varying kSBon peak tensile stress
kSB 0.0 0.5 1.0 2.0 5.0 10.0 50.0 2.53 2.25 2.14 2.01 1.83 1.71 1.51 Notes
• Shear transfer has a large effect on stress • Slab and base maintained full contact • Model remained linear
Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
• Obtaining EverFE and Program Architecture
Analysis of Thermal Gradients
Corners of slab curl upward s lab th ic k nes s - T + T
Nighttime curling:top of slab cools relative to the bottom after a warm day
Weight of slab pulls downward
Analysis of Thermal Gradients
Center of slab lifts upward s lab th ic k nes s T - TDaytime curling:top of slab heats relative to the bottom during a warm day
Weight of slab pulls downward
Tension on bottomof slab
• Analytical solutions for stresses exist for simple cases • However, thermal gradients are often nonlinear
• Slab-base interaction plays a significant role in response
– Loss of contact between slab and base layer – Shear stresses develop at slab-base interface
Analysis of Thermal Gradients
s lab th ic k nes s
• EverFE idealizes gradients as linear, bilinear or tri-linear • Equal vertical spacing assumed between each T
2 E lem ents 3 E lem ents 1 E lem ent Temperature Variations Used in FE Analysis
Analysis of Thermal Gradients
Bilinear Gradient: Specified Temperature Variation s lab th ic k nes s
Analysis of Thermal Gradients
Trilinear Gradient: 1 E lem ent Temperature Variations Used in FE Analysis Specified Temperature Variation 3 E lem ents 2 E lem ents s lab th ic k nes s
Analysis of Thermal Gradients
Quick parametric study
• Re-run our single-slab model
• Consider positive (+5oC/-5oC) and negative (-5oC/+5oC) gradients
• Consider both bonded and unbonded base with no shear transfer
Results of Analyses:
Maximum Principal Stress (MPa) Bonded Unbonded Positive 1.45 0.94 Negative 1.22 0.86
Analysis of Thermal Gradients
Effect of thermal gradient nonlinearity
• Re-run our single-slab model with nonlinear gradients, unbonded base
Positive gradient Negative gradient
Results:
• 1.71 MPa for positive (82% increaseover linear gradient)
• 0.47 MPa for negative (45% decreaseover linear gradient
Analysis of Slab Shrinkage
Shrinkage can be simulated as an equivalent thermal gradient Example:
• Consider a uniform shrinkage of -0.0001 mm/mm • Coeff. of thermal expansion = 1.1x10-5/oC
•Equivalent T = -0.0001/1.1x10-5/oC = -9.09oC
-9.09oC •Re-run our single-slab model assuming:
• No slab-base shear transfer
• A rough HMA base (E = 2000 MPa, kSB= 0.27 MPa/mm, 0= 0.25mm)
Analysis of Slab Shrinkage
Results of Simulation No slab-base shear transfer
x = +/-0.25mm at x = 0mm, 5000mm No stresses are developed in slab
BOS Stresses
σmax= 0.32 MPa
Early-Age Effects
• Concrete pavements sometimes crack during curing • Primary causes are thermal and/or shrinkage gradients
that occur prior to concrete gaining full tensile strength
Shrinkage cracks in new pavement
Early-Age Effects
Simple example of how this can be studied with EverFE
• Re-run our single-slab model founded on CTB
• Consider a negative (-5oC/+5oC) thermal gradient
• Unbonded base with no shear transfer
• Examine effect of curing time on ratio of slab stress:slab MOR
Assumptions:
• MOR = E= (usual ACI equations, psi)
• Assume these relationships are valid for cure times of 1–28 days • Type I cement, published relationship between time and
• Examine effect of curing time on ratio of slab stress:slab MOR ’ c f ’ 6 fc ’ 000 , 57 fc
Early-Age Effects
Details of Analysis Parameters
Age-strength relationship 1.0 5.5 11100 1.17 2.5 10.3 15230 1.60 4.0 13.8 17580 1.85 9.0 20.7 21530 2.27 28.0 27.6 24870 2.62 Age E MOR
days MPa MPa MPa ’ c f
Early-Age Effects
Results of Analysis TOS stresses Displaced shape 0.300 5 10 15 20 25 30 0.35 0.40 0.45 0.50 M a x st re ss/ M O R Time (days)Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
• Obtaining EverFE and Program Architecture
Dowel Joint Load Transfer
The challenge: How do we model this?
separation of slab and subgrade
dowel wheel load
high stress on subgrade
Dowel Joint Load Transfer
• Early models used springs at transverse joints
• Other models used beams on elastic foundations
Primarily 2D models with plate elements
Both 2D models with plate elements and 3D models
Dowel Joint Load Transfer
Challenges for idealizing dowels in 3D FE models:
• Dowel-slab interaction and dowel looseness are difficult to treat • Conventional discretizations require slab and dowel nodes to coincide
slab mesh lines
Plan View dowels
Dowel Joint Load Transfer
Our solution is the embedded dowel element
solid
element Beam element is constrained to
displace compatibly with the embedding solid element
dowels
slab mesh lines
Immediate Benefit:
Dowel Joint Load Transfer
Specification of dowels in EverFE • Dowels can be equally spaced
• Dowels can be located in wheelpaths
• Dowels can be manually located by specifying y-coordinate • Each row of slabs can have different dowel placements Example of dowel placement
• Start a new model with say 2 rows x 3 columns of slabs • Go to dowel panel
Dowel Joint Load Transfer
Treatment of Dowel-Slab Interaction with EverFE
• Rigorous treatment
• Either bonded or unbonded • Can be severe nonlinearity
Dowel Looseness
gap length
gap
•Less rigorous treatment
• Model remains linear
• Allows intermediate bond levels
Dowel-Slab Support Modulus
Kz = modulus of
dowel support ×diameter
Dowel Looseness
Significance:
• Has been studied experimentally and numerically
• Small gaps (< 0.50mm) can greatly reduce joint load transfer
Treatment by EverFE:
• Embedded element formulation is very advantageous • Treated as a nodal contact problem
• Multiple embedded beam elements are used for each dowel multiple
elements single
914 mm 1220 m m 12 - 6.35 mm dowels 10 kN rubber pad
k
= 0.09 MPa/mm 51 m mgrease and drinking straw
Laboratory Tests of Hammons (1997)
unbonded CTB
Dowel Looseness
V e rt ic al Di spl a ce m e n t ( mm) 0.2 0.4 0.6 experimentano CTB lDistance from Joint (mm) 0 -200 -100 100 -400 model, no looseness model, gap = 0.08 mm
Dowel Looseness
Distance from Joint (mm) 0 -200 -100 100 -400 V e rt ic al Di spl a ce m e n t ( mm) 0.2 0.4 0.6
model,k= 0.09 MPa/mm, gap = 0.08mm
experimental with CTB
Dowel Looseness
model,k= 0.07 MPa/mm, gap = 0.08 mmDowel Looseness
Example for 2-slab system:
• Slabs are 4600mm long x 3600mm wide x 250mm thick • Founded directly on dense liquid, k = 0.03 MPa/mm
•E= 28000 MPa, ν= 0.20, density = 0
• Center an 80-kN axle with 2 wheels transversely, left of joint • Set linear aggregate interlock stiffness to 0
• Use 11 evenly spaced 32mm diameter dowels at the joint • Choose dowel looseness, de-select bonded, Emb = 225 mm • Set GapB to 125mm (1/2 embedded length)
Dowel Looseness
Results of Analysis Gap l u LTE (mm) (mm) (mm) (%) (MPa) 0.00 0.467 0.467 100 0.865 0.05 0.528 0.528 81 1.019 0.10 0.578 0.384 66 1.121 0.15 0.622 0.344 55 1.267 0.20 0.646 0.323 50 1.309 0.30 0.660 0.310 47 1.323 0.40 0.664 0.306 46 1.326Dowel-Slab Support Modulus
Background:
More traditional method of idealizing dowel-slab interaction Dowel-slab interface idealized with distributed springs Results in a linearly elastic model
Can specify varying degrees of bond and dowel locking
Example:
Consider the same example we just analyzed
Specify dowel-slab support modulus in lieu of dowel looseness
Dowel-Slab Support Modulus
Results of Analysis Kz l u LTE (MPa) (mm) (mm) (%) (MPa) 1e6 0.474 0.471 99 0.912 1e4 0.505 0.457 90 1.182 5000 0.517 0.447 86 1.223 500 0.612 0.357 58 1.310 100 0.771 0.200 26 1.362 1 0.969 0.004 0 1.417Dowel Misalignment/Mislocation
Inaccurately cut transverse joints mislocated dowels Improperly placed dowels dowel misalignment
Elevation View z ∆ Actual position β q s Intended position x ∆ Plan View α Intended position Actual Position q r y ∆
Dowel Misalignment/Mislocation
Treatment by EverFE
• Embedded dowel element permits implementation
• Straightforward when dowel-slab support modulus is specified • A different solver must be used when modeling looseness
Example with EverFE
• Consider the same example we just analyzed
• Vary x from 0 – 100mm with Kz= 2000 (LTE = 79% at x = 0)
• Study effect of x on response
Dowel Misalignment/Mislocation
Results of Analysis: 0 78.6 1.26 3789 135.7 20 78.6 1.26 3787 136.1 40 78.6 1.26 3783 137.2 60 78.2 1.27 3770 139.5 80 77.9 1.27 3745 143.7 100 77.3 1.27 3703 150.2x LTE Dowel Bearing
(mm) (%) (MPa) Shear Stress (N) (MPa)
Transverse Ties
• Can be independently specified for each longitudinal joint • Modeled with same embedded elements used for dowels • Can model tie-slab support and restraint moduli
• Assumed evenly spaced along each joint
• First tie is placed at ½ tie spacing from left-hand joint
Transverse Ties
4600mm (typ) 3600m m m Model Properties• 250mm slab on dense liquid • 12-32mm dowels give 80% LTE
at transverse joint • Tied shoulder
• 13mm diameter, 750mm long ties • Corner axle load and thermal
gradient considered in analyses
Example to Illustrate Tie Effectiveness
Transverse Ties
650 670 690 710 730 750 0.7 0.9 1.1 1.3 1.5 Tie Spacing (mm) Max im u m Pr inc ipa l Str e s s ( M Pa)Slab stress with NO ties:
• Axle load: 1.33 MPa • Thermal: 0.746 MPa • Axle+thermal: 1.39 MPa Axle load Axle + thermal Thermal
Transverse Ties
Observations and Conclusions:
• Ties can dramatically reduce slab stresses due to corner loads • Tie effectiveness strongly depends on its proximity to joint
700 mm spacing Max. stress = 0.719 MPa
Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
• Obtaining EverFE and Program Architecture
Aggregate Interlock
The challenge: How do we model this?
aggregate interlock wheel load
• Interaction of two rough crack surfaces
Aggregate Interlock
Usual FE Treatment of Aggregate Interlock:
Springs at Transverse Joints
• Simple, traditional approach • Model remains linear • No effect of joint opening
Coulomb Friction
• Shear depends on normal stress • Any joint opening => no shear
EverFE’s Two-Phase Model
aggregate particles crack
Relies on Walraven’s Model
concrete is two-phase medium
• aggregate particles are rigid spheres • paste is rigid-plastic
cracks follow aggregate boundaries • particles bear on paste, at point of slip
(
)
Fy =σpu Ax −µAy(
)
Fx =σpu Ay +µAx Particle Equilibrium:σ
puF
xF
yτ
pu=
µσ
pu aggregate particle deformed paste embedment crack openingEverFE’s Two-Phase Model
EverFE’s Two-Phase Model
Two-phase model parameters
1) pu= ultimate strength of cement paste
2) = paste-aggregate coefficient of friction (0.4 – 0.5)
cc pu =8.0 f
σ • Walraven suggests
• fcc= 1.25f’c (units are MPa)
3) aggregate volume fraction (usually 0.7 – 0.8) 4) Maximum aggregate size (typically 18 or 20 mm) 5) Initial joint opening (seasonally variable)
EverFE’s Two-Phase Model
Initial joint opening is a critical parameter
• Greatly affects nonlinear aggregate interlock model • Affects contact between joint faces
initial joint opening 0 4 8 12 16 20 0 0.5 1.0 Shear Str e ss
Relative Vertical Displacement inc rea sing joi nt o peni ng direct effect
EverFE’s Two-Phase Model
Tests by Colley and Humphrey (1967) Finite Element Idealization
Zero
Stiffness Two-PhaseModel
2743 mm 1219 m m Loading PL Joint Filler Pre-cracked 178 m m 229 m m
EverFE’s Two-Phase Model
178 mm Slab 1 2 3 Joint Opening (mm) 0 20 40 60 80 100 Two-phase model Experimental data 0 20 40 60 80 100 1 2 3 Joint Opening (mm) 229 mm Slab LT E ( % )EverFE’s Two-Phase Model
Example for 2-slab system:
• Slabs are 4600mm long x 3600mm wide x 250mm thick • Found directly on dense liquid, k = 0.03 MPa/mm
•E= 28000 MPa, ν= 0.20, density = 0
• Center an 80-kN axle with 2 wheels transversely, left of joint • No dowels at the joint
• Specify nonlinear aggregate interlock model
– pu= 50 MPa
– = 0.4
– volume fraction = 0.75
– Dmax= 20mm
• Examine effect of joint opening on response Same as
doweled model
0.1 0.489 0.486 99 0.863 0.5 0.519 0.456 88 1.227 1.0 0.568 0.406 71 1.271 1.5 0.632 0.343 54 1.301 2.0 0.707 0.267 38 1.330 3.0 0.860 0.114 13 1.386 4.0 0.963 0.012 1 1.414 Joint Opening l u LTE (mm) (mm) (mm) (%) (MPa)
EverFE’s Two-Phase Model
Results of analysis:
EverFE’s Two-Phase Model
Practical use of two-phase model:
Recent research has validated this type of model However, the model is not perfect:
• It assumes no fracture of coarse aggregate
(Walraven suggests scaling down puto account for this)
• EverFE does not account for smooth surface at sawcut (Will tend to overestimate joint shear transfer)
Linear Aggregate Interlock Model
Springs at transverse joints
• Simple approach • Model remains linear • Joint opening has no effect
Example for 2-slab system
0.0 0.974 0.000 0 1.418 0.1 0.647 0.328 51 1.360 0.5 0.538 0.436 81 1.290 1.0 0.518 0.456 88 1.238 10.0 0.493 0.481 97 0.997 100.0 0.488 0.486 99 0.852 Joint Stiffness l u LTE (MPa/mm) (mm) (mm) (%) (MPa)
Workshop Topics
• Introduction• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
Effect of Slab-Base Shear
Transfer and Dowel Locking
Low:kSB§0 (bond-breaker)
Intermediate:kSB= 0.035 MPa/mm, = 0.60 mm (ATB) High:kSB= 0.416 MPa/mm, = 0.25 mm (HMAC)δ0
0
δ
Prior studies have identified critical parameters: • Interface shear stiffness (kSB) and elastic limit ( ) • Dowel-slab bond (“dowel locking”)
Interface properties for given base types (Zhang and Li 2001):
0 δ •250 mm thick slab • 150 mm thick base • 32 mm dowels, no looseness • Material properties: E= 28,000 MPa = 0.20 = 2,400 kg/m3 = 1.1x10-5/oC
Parametric Study FE Model
4600 mm
3600 m
m
doweled joints
60,300 DOF
Parametric Study Loads
Uniform Shrinkage – T – 10 oC + Gradient + Shrinkage + T – T – 6 oC – 14 oC – Gradient + Shrinkage – T – T – 14 oC – 6 oCSlab Displacements and Stresses
Displacements due to – T – T Max. principal stresses due to – T – T 500 X Magnification
Selected Maximum Stresses (kPa)
Dowel Type – T + T – T – T – T Load Case – T + T – T – T – TLow Int. High
Degree of Slab/Base Shear Transfer Locked Unlocked 0 870 688 159 973 818 594 1180 991 0 872 689 118 906 669 591 1510 547 Stress Location Bottom Bottom Top Bottom Bottom Top
Discussion of Results for
– T–T
• Negative prestrain gradient produces curling, tension on top •Dowel restraint uniformly
increases tension
• Shear stresses at bottom of slab
Effect of Dowel Restraint for
+ T–T
0 400 800 1200 1600 2000 0.9 1.1 1.3 1.5Dowel Axial Restraint Modulus (MPa)
Ma x . P rin cip a l S tre ss (MP a ) High Slab/Base Shear Transfer Intermediate Slab/Base Shear Transfer
Discussion of Results for
+ T–T
• Positive prestrain gradient produces tension on bottom • Shear stresses at bottom of slab
increase tension on bottom of slab •Dowel restraint restricts relative
slip between slab and base •With high base/slab shear
transfer, restricted slip decreases tension due to base-slab shear
Workshop Topics
• Introduction
• Overview of Finite-Element Concepts • Generation and Solution of a Simple Model • Slab-Base Interaction
• Analysis of Thermal Gradients and Slab Shrinkage • Modeling Dowel Joint Load Transfer
• Modeling Aggregate Interlock Joint Load Transfer • Example of a More Complex Simulation
• Obtaining EverFE and Program Architecture
Obtaining EverFE
1. Get a cashier’s check for $5000 made out to Bill Davids
• Go to http://www.civil.umaine.edu/EverFE • Download EverFE2.23.exe
• Run EverFE2.23.exe on your computer • You can now run EverFE using the new
desktop icon, or from the Programs menu • Questions to [email protected]
Program Architecture
Basic architecture of software
User Interface (Tcl/Tk/vtk) FE meshing code (compiled C++) FE solver (compiled C++)
What you see
Nonlinear agg. interlock (compiled C++)
What does the hard work
Program Architecture
Directory structure
Top-level directory Aggregate interlock data Project definitions/results Help file and manual Finite-element solver Tcl/Tk code
Program Architecture
How Project Data is Stored
• Each project has a file with a .prj extension, and a subdirectory • The .prj file is a placeholder to allow the project to be recognized • The subdirectory contains project definition, FE input/output
Why is this important?
• These files are simple ASCII text files, but can get large • If you want to archive a project to save disk space, you simply
move the .prj file and entire subdirectory to another storage device • At any time in the future, you can copy the .prj file and
subdirectory back to EverFE2.23/data, and it will be recognized