Accurate Long-Term Deflection
Predictions for Concrete Structures
By Jonathan Hirsch, P.E.
Bentley Systems, Inc.
SE University, January, 2012 www.LearnWithSEU.com
Key Objectives
How different are the results from commonly
used deflection calculation techniques?
What are 4-5 key parameters that influence
deflection calculations?
What is a reasonable expectation of accuracy for
deflection calculations and how can they be
applied in design practice?
2
Deflection Predictions in Concrete
How accurate is accurate?
Deflection prediction methods
Comparison between calculated and measured
results
Design recommendations
3
What is Accurate?
Poll
4
Factors Affecting Deflections
Material Properties
Cracking and Tension Stiffening
Creep
Shrinkage
Externally restrained
Internally restrained
Load History
Redistribution of Forces
5
Tension stiffening
Uncracked response
Load, P
3
Cracked response (no tension in
concrete)
4
1 2
Assumed unloading response
0 Curvature,
6
Shrinkage Warping
7
Linear Elastic Deflection Theory
y f ( x) deflection
y ' f ' ( x) slope
M 5wl 4
y ' ' f ' ' ( x)
EI 384 EI
V
y ' ' ' f ' ' ' ( x)
EI
w
y ' ' ' ' f ' ' ' ' ( x)
EI
8
Deflection Prediction Methods
Deemed to Comply rules
Deflection multipliers
Spreadsheets using load distribution factors
2-D frame programs
3-D finite element programs
9
Deemed to Comply
ACI 318 9.5.2.1
Quick and easy to apply
Very crude due to failure to explicitly consider:
Concrete properties
Cracking characteristics
Load History
Creep
Shrinkage
10
Deflection Multipliers
ACI 318 9.5.2.5
Requires good calculation of initial elastic
deflections
Doesnt explicitly consider important factors:
Shrinkage
Relationship between creep curvature and creep
deflections
11
Spreadsheets Using Factors
Scanlon, A. and Suprentant, B.A., 2011, Estimating Two
Way Slab Deflections, Concrete International,V. 33, No.
7, pp 29-34
Easy and quick to use
Conservative estimate of cracking through Ieff
assumption
Good accounting of load history and material
properties
Assumed factors used for load distribution
Deflection multipliers for creep (and shrinkage)
12
Spreadsheets Using Factors
13
2-D Frames
Linear elastic analysis of frames
Ieff
for cracking/tension stiffening, difficult to apply for
column/middle strip
Long-term effects using deflection multipliers
Calculate long-term curvatures and use
geometrical methods to calculate deflections
Equivalent frame for 2-way slabs
Summing of strip results necessary
14
3-D Finite Element Analysis
Linear Elastic Analysis with Stiffness Modifiers
Using a number of tension stiffening models
Creep and shrinkage strains tracked and
superimposed
Redistribution of loads
Rigorous analysis of internally restrained shrinkage
Treatment of externally restrained shrinkage
Nonlinear Analysis
Computationally expensive
15
3-D Finite Element Analysis
Use ageing coefficient for gradually induced
strains
16
3-D Finite Element Analysis
1 d ( )
t
1 (t , ) d
Ec (t0 ) 1
(t , t0 )
(t , t0 ) t0 Ec ( ) d (t , t0 )
0.8
t0
(t , t0 )
1.8 t0
0.8
17
3-D Finite Element Analysis
Cross section calculations
Select material stress strain curves
Select tension stiffening model
Select creep and shrinkage models
18
3-D Finite Element Analysis
Perform a linear elastic global analysis and
integrate cross section forces
Perform detailed nonlinear long-term curvature
calculations on each cross section
Using the resulting axial strains and curvatures,
adjust the element stiffnesses
Iterate until the solution converges
Repeat for each load history step, breaking each
into instantaneous and sustained component
19
3-D Finite Element Analysis
20
Deflection Prediction Methods
Calculation Cracking/ Creep Shrinkage Shrinkage Load Load
Method Tension Internal External History Redistribut
Stiffening Restraint Restraint ion
Deemed to
comply
Deflection
Multipliers
Spreadsheet
with factors
2D frames -
Multipliers
2D frames
X-section
3D finite
element
21
Deflection Calculation
Comparison
Method 1 2D equivalent frames (ACI approach)
Method 2 spreadsheet using factors
Method 3 3D finite element analysis
22
Test Slabs
University of New South Wales (Guo and Gilbert
2002)
215 West Washington Street
Aqua
23
University of New South Wales
24
University of New South Wales
25
University of New South Wales
26
215 West Washington Street
27
215 West Washington Street
28
Aqua
29
Aqua
30
Slab S6
31
Slab S6
Characteristics of this test:
No restraint due to rollers at the column base
No
3-D column cracking
2-D Finite Element
Equivalent predicted
Frames deflectionsunderestimated
significantly very well no external
Spreadsheet
Loads
restraintnear
was slab
predicted
cracking
modeled deflections
loads very well
deflections due to under prediction of cracking
Slab cracks propagated over time
32
Slab S7
33
Slab S7
Characteristics of this test:
3-D Finite
Same slabElement
load as (with
S6 external shrinkage restraint
2-D
modeled)
Spreadsheet
Equivalent
Restrainedunderestimated
column
overestimated
Frames
base
significantly
deflections,
deflections,
underestimated
to to
a reasonable
a reasonable
deflections
degree
degree finaldue
Similar slab
tocrack
underpattern
prediction
to S6
of cracking
34
Slab S4
35
Slab S4
Characteristics of this test:
Slab loading far in excess of cracking load and sustained
Restrained column base
Heavy slab cracking which propagated over time
Extensive column
2-D Equivalent cracking
Frames significantly underestimated
Finitereinforcement
Heavier
deflections
3-D
SpreadsheetElement
due to underthan S6/S7
predictions
prediction
significantly wereofexcellent
cracking
overestimated deflections
36
Slab S5
37
Slab S5
Characteristics of this test:
Slab loading far in excess of cracking load and removed
3-D Finite Element
Restrained column predictions
base (without considering external
2-D
restraint)
Equivalent
Heavy slab
Spreadsheet were
cracking
Frames
overestimated
which
significantly
overestimated propagated
to a reasonable
underestimated
deflections over
to time
degree. Effects
a reasonable
deflections
of external
Same
degree reinforcement
due
restraint
to under
to
as shrinkage
S4
predictionwere
of cracking
overestimated
38
Slab S3
39
Slab S3
Characteristics of this test:
Slab loading overestimated
Spreadsheet similar to S6/S7early
sustained,
sustained
then
deflections,
reduced but
2-D
also
3-D Equivalent
Finite
overestimated
RestrainedElement
column
Frames
predictions
elastic
base
significantly
and were
creepunderestimated
reasonable,
recovery, giving
but elastic
deflections
reasonable
and
Samecreep
reinforcement
recovery
final
due to
results
under
were
as S4/S5
prediction
overestimated
of cracking
40
215 West Washington Street
41
Aqua
3D finite element (method 3) underestimated short
term
2D deflections,
Spreadsheet
Equivalent withsignificantly
(method
frames 2)significant improvements
significantly with
underestimated
overestimated
small offsets in time
deflections
42
Performance Comparison
2D equivalent frames (method 1) significantly
underestimated deflections in all cases
Spreadsheet (method 2) overestimated
deflections in all cases, dependent upon the
actual levels of cracking
3D finite elements (method 3) predicted most
of the test slabs reasonably with external
shrinkage restraint adjustment,
underestimated real world slabs
43
Design Recommendations
Maintain realistic expectations
Use spreadsheet method instead of
span/depth ratios
Never use full frame sections for
determining cracking and effective
moments of inertia
44
Design Recommendations
Be conservative with cracking
Consider load history
Consider using larger deflection multipliers
for creep and shrinkage
45
Pending ACI Publication
Practical Deflection Prediction of
Concrete Slabs
Jonathan Hirsch, Bentley Systems Inc.
Flora Calabrese, Wiss, Janney, Elstner Associates, Inc.
Eamonn Connolly, James McHugh Construction Co.
Allan Bommer, Bentley Systems, Inc.
46
Accurate Long Term Deflection
Predictions for Concrete Structures
By Jonathan Hirsch
Bentley Systems, Inc.
SE University, January, 2012 www.LearnWithSEU.com