THE DEVELOPMENT OF AASHTO LRFD BRIDGE

DESIGN SPECIFICATION AS N EXAMPLE OF

PROBABILISTIC-BASED SPECIFICATIONS
State University of New York at Buffalo
November 7
7, 2011
Presented By
Wagdy G. Wassef, P.E., Ph.D.
Modjeski and Masters, Inc.

A Brief History
y
1931 First printed version of AASHO Standard
Specifications for Highway Bridges and Incidental
Structures
1970s AASHO becomes AASHTO ((1990s AREA becomes
AREMA)
Early 1970s AASHTO adopts LFD
Late
L t 1970s
1970 OMTC starts
t t work
k on lilimit-states
it t t based
b
d
OHBDC
1986 AASHTO explores need to change

Design Code Objectives

Technically state-of-the-art specification.

Comprehensive as possible
possible.
Readable and easy to use.
Keep specification-type wording do not develop
a textbook.
Encourage a multi-disciplinary approach to bridge
design.
g

Major Changes
A new philosophy of safety - LRFD
The identification of four limit states
The
The relationship of the chosen reliability level, the
load and resistance factors, and load models
th
through
h the
th process off calibration
lib ti
new load factors
new resistance factors

LRFD - Basic Design Concept

Some Algebra

(R-Q) = +
2
R

R=Q+

+ = R =
2
R

2
Q

i xi
=
Q + 2R + Q2

R-Q
2
+

Q
2
R

i xi

2
Q

Load and Resistance Factor Design

i i Qi Rn = Rr

in which:
i = D R I 0.95 for loads for max
= 1/(
1/( I D R) 1.0
1 0 ffor lloads
d ffor min
i
where:
i = load factor: a statisticallyy based multiplier
p
on
force effects
= resistance factor: a statistically based
multiplier applied to nominal resistance

LRFD (Continued)
i
D
R
I
Qi
Rn
Rr

=
load modifier
=
a factor relating to ductility
=
a factor relating to
redundancy
=
a factor
f t relating
l ti to
t
importance
=
nominal force effect: a
deformation stress, or stress
resultant
=
nominal resistance
=
factored resistance: Rn

Reliability
e ab ty Ca
Calcs
cs Done
o e for
o M and
a dV
Simulated Bridges Based on Real Ones
25 non-composite
non composite steel girder bridge simulations
with spans of 30,60,90,120,and 200 ft, and
spacings of 4,6,8,10,and 12 ft.
Composite steel girder bridges having the same
parameters identified above.
beam bridges with the same parameters
P/C II-beam
identified above.
R/C T-beam bridges with spans of 30,60,90,and
120 ft,
ft with spacing as above
above.

Reliability
y of Std Spec
p vs. LRFD
175 Data Points

Major
j Changes
g
Revised calculation of load distribution
0.6

S
g = 0.075 +

2900

0.2

S

L

0.1

Kg
3
Lt s

Circa
1990

Major Changes (Continued)

Combine plain, reinforced and prestressed concrete.

Modified compression field/strut and tie
tie.
Limit state-based provisions for foundation design.
Expanded coverage on hydraulics and scour.
The introduction of the isotropic deck design.
Expanded coverage on bridge rails.
Inclusion of large
g p
portions of the AASHTO/FHWA
Specification for ship collision.

Major Changes (Continued)

Changes to the earthquake provisions to eliminate
the seismic performance category concept by
making the method of analysis a function of the
importance of the structure.
Guidance on the design of segmental concrete
bridges from Guide Spec.
The development
p
of a p
parallel commentary.
y
New Live Load Model HL93
Continuation of a long story

1923 AREA Specification

10-Ton
15-Ton
20-Ton

4k
6k
8k

16k
24k
32k
14'

VERY CLOSE!!

5.5

'

1928-1929 Conference Specification

6k 24k
14'
15-Ton

30'

6k 24k
14'
30'
15-Ton

8k 32k
14'
30'
20-Ton

6k 24k
14'
30'
15-Ton

18,000 lb for Moment

26,000 lb for Shear

6k 24k
14'
15-Ton

640 lb/ft

1944 HS 20 Design Truck Added

Live Load Continued to be Debated

Late 60s H40, HS25 and HS30 discussed
1969 SCOBS states unanimous opposition to
increasing weight of design truck wasteful
obsolescence of existing bridges
1978 HS25 proposed again
1979 HS25 again commentary
need for heavier design
g load seems unavoidable
HS25 best present solution
5% cost penalty
Motion soundly defeated

Exclusion
E
l i Loads
L d Based
B
d on TRB
Special Report 225, 1990

EXCL/HS20 Truck or Lane or 2 25

kips Axles @ 4 ft (110 kN @ 1.2 m)

Selected Notional Design Load

HL-93

EXCL/HL 93 Circa 1992

NCHRP 12-33 Project Schedule

First Draft - 1990 general coverage

Second Draft - 1991 workable
Third Draft - 1992 p
pretty
y close
Fourth Draft - 1993 ADOPTED!!
12,000 comments
Reviewed by hundreds
Printed and available - 1994

Upgrades and Changes to 1990

T h l
Technology
1996 foundation data reinserted
reinserted.
New wall provisions ongoing upgrade.
2002 upgraded to ASBI LFRD Segmental Guide
Specs.
MCF shear in concrete simplified and clarified several
times major update in 2002.
Load distribution application limits expanded several
time in 1990s due to requests to liberalize.
More commentary added
added.

Upgrades and Changes

2004 major change in steel girder design in
anticipation of
of
2005 seamless integration of curved steel bridges
ending three decade quest

Upgrades and Changes (Continued)

2005 P/C loses updated
2006 complete replacement of Section 10
Foundation Design
2006 more concrete shear options
2007 big year
Streamline MCF for concrete shear design
1,000
1 000 year EQ maps and collateral changes
Seismic Guide Spec - displacement based
Pile construction update
2008 - Coastal bridge Guide Spec

Where Do We Go From Here?

Where Do We Go From Here?

The original AASHTO LRFD live load
study
t d was b
based
d on lload
d measurements
t
made in the 1970s in Ontario. How this
relates
l t to
t todays
t d loads?
l d ?

Where Do We Go From Here?

The specifications was calibrated for the
strength
t
th limit
li it state
t t where
h
th
the d
definition
fi iti off
failure is relatively simple: if the factored
l d exceed
loads
d th
the ffactored
t d resistance,
i t
failure, i.e. severe distress or collapse, will
t k place.
take
l
What about service limit state and what is
failure under service limit states?

Where Do We Go From Here?

Two Current Projects of Special Note:
SHRP R19 B - Bridge for Service Life
Beyond 100 Years: Service Limit State
Design (SLS)
NCHRP 12-83,, Calibration of Service Limit
State for Concrete

R19B Research Team

Modjeski and Masters, Inc.:
University of Delaware:
University of Nebraska:
NCS Consultants:

John Kulicki, Ph.D., P.E.

Wagdy Wassef, Ph.D., P.E.
Dennis Mertz, Ph.D., P.E.
Andy Nowak, Ph.D.
Naresh Samtani, Ph.D., P.E.

NCHRP 12-83 Research Team

Same except that NCS Consultants are replaced with
Rutgers University:
Hani Nasif, Ph.D., P.E.

Current General SLSs

SLS s
Live load deflections
Bearings-movements and service forces
Settlement of foundations and walls

Current Steel SLS

SLSs
s
Permanent deformations in compact steel
components
t
Fatigue of structural steel, steel
reinforcement and concrete (through its
own limit state)
Slip of slip-critical bolted connections

Current Concrete SLS

SLSs
s
Load induced
Stresses in prestressed concrete under
service loads
Crack control reinforcement

Non-Load
Non Load induced
Shrinkage and temperature reinforcement
Splitting
S litti reinforcement
i f
t

Desired Attributes
IIs an SLS meaningful?
i f l? C
Can it b
be
calibrated?
Does it really relate to service---or
something else?
Can (should) aging and deterioration be
p
incorporated?
Can it reflect interventions?

General Topics

Special challenges for SLS development

Survey of owners
U off WIM data
Use
d t
Calibration process

General Topics (cont

(contd)
d)
Improvements to current SLS
Crack control in reinforced concrete
Tension in P/S beams
Load induced fatigue in steel and concrete
Use of Weigh-In-Motion
Weigh In Motion Data

Current Status
Vetted WIM data
SLS Live Load live load model
Finite Life fatigue load model
Infinite Life fatigue load model

Preliminary Betas for Service III (Tension in

P/s beams)
Work on deflections
Work on compiling info on joints and
bearings

Service and Fatigue LL has been a

challenge
Truck WIM was obtained from the FHWA
and NCHRP Project 12-76
T
Total
t l number
b off records
d about
b t 60 million
illi
about 35 million used

Initial Filtering Criteria For Non-Fatigue

SLS (FHWA Unless Noted)
Excluded Vehicles

Individual axle weight > 70kips GVW < 10

7 >Total length >200 ft
First axle spacing
p
g <5 ft
Individual axle spacing < 3.4ft
10 > Speed > 100 mph
GVW +/- the sum of the axle weights by more than 7%.
FHWA Classes 3 14

Additional Filtering
Filter #1 Questionable Records
1 - Truck length
g > 120 ft
2 sum of axle spacing > length of truck.
3 - any axle < 2 Kips
4 - GVW +/
+/- sum of the axle weights by more than 10%
5 - GVW < 12 Kips

Filter #2 Presumed Permit Trucks

6 - Total # of axles < 3 AND GVW >50 kips
7 - Steering axle > 35 k
8 individual axle weight > 45 kips

Filter #3 Traditional Fatigue Population

9 - Vehicles with GVW <20 Kips

Filtering By Limit State

Vehicles Passing Filters #1 & #2 will be
used
d ffor calibration
lib ti off allll lilimit
it states
t t
except for Fatigue, the limit state for permit
vehicles
hi l and
d possibly
ibl Strength
St
th II
II.
Vehicles filtered by Filter #2 will be
considered Permit vehicles and will be
reviewed and may be filtered further.
Vehicles passing all three filters will be
used for the fatigue
g limit state

WIM Data - FHWA

5
4
3

Standard
d Normal Va
ariable

14 sites
Representing 1 year
of traffic at most sites
The maximum
recorded GVW is 220
kips
Mean values range
from 20 to 65 kips

Arizona(SPS-1)
Arizona(SPS-2)
Arkansas(SPS-2)
Colorado(SPS-2)
Illinois(SPS-6)
Indiana(SPS-6)
Kansas(SPS-2)
Louisiana(SPS-1)
Maine(SPS-5)
Minnesota(SPS-5)
New Mexico(SPS-1)
NewMexico(SPS-5)
Tennessee(SPS-6)
Virginia(SPS-1)
Wisconsin(SPS-1)
Wisconsin(SPS
1)
Delaware(SPS-1)
Maryland(SPS-5)
Ontario

2
1
0
-1
1
-2
-3
-4
-5

50

100

150

GVW [kips]

200

250

Analysis
y
of the WIM Data
Live load effect maximum moment and
shear
Simple spans with span lengths of 30
30, 60
60,
90, 120 and 200 ft
Trucks
T k causing
i moments
t or shears
h
< 0.15 (HL93) were removed

Removal of the Heavy

y Vehicles for SLS
New York 8382 Span 90ft

Standard Norm
S
mal Variable

Filter trucks causing moments

or shears
h
more th
than 1
1.35(HL93
35(HL93
live load effect) were removed
Number of trucks before filtering
1,551,454
Number of trucks after filtering
1 550 914
1,550,914
Number of removed trucks 540
Percent of removed trucks
0.03%

-2

-4

No Trucks Removed
0.03% Trucks Removed
-6

0.5

1.5

2.5

Truck Moment / HL93 Moment

Multiple
p Presence Cases
Simultaneous
occurrence off trucks
t k
on the bridge:
Filt
Filter b
based
d on titime
of a record and a
p
of the truck
speed
Distance from the
first axle of first truck
to the first axle of the
second truck
maximum 200 ft

T1

HeadwayDistance max200ft

T2

T1

HeadwayDistance max200ft

T2

Twocasesofthesimultaneous
occurrence

Correlation Criteria
Both trucks have the same number of
axles
GVW of the trucks is within +/- 5%
All corresponding spacings between
axles are within +/- 10%

Adjacent
j
Lanes - Florida
140

Number of Trucks :
1,654,004

120

100

Frequenc
cy

Florida I10 Time

record accuracy 1
second

Number of Fully
Correlated Trucks:
2 518
2,518
Max GVW = 102 kips

80

60

40

20

20

40

60

80

100

120

Gross Vehicle Weight - Trucks in Adjacent Lanes

Adjacent Lanes Florida

2 518 off 1
2,518
1,654,000
654 000
5

Standard Normal Varia

able

-1

-2

-3
3

-4
Florida I10 - 1259 Correlated Trucks - Side by Side
Florida I10 - All Trucks
-5

50

100

150

Gross Vehicle Weight

200

250

One Lane Florida

4 190 off 1
4,190
1,654,000
654 000
5

Standard N
Normal Variable

-1

-2

-3

-4
Florida I10 - 4190 Correlated Trucks In One Lane
Florida I10 - All Trucks
-5
5

50

100

150

Gross Vehicle Weight

200

250

Conclusions for Multiple

p Presence
Vehicles representing the extreme
tails of the CDFs need not be
considered
id d tto occur simultaneously
i lt
l iin
multiple lanes.
For the SLS
SLS, only a single
single-lane
lane live
liveload model need be considered.

Statistics of Non
Non-fatigue
fatigue SLS Live Load
Based on 95% limit:
ADTT = I,000, Project Bias on HL 93 = 1.4
ADTT = 5,000, Project Bias on HL 93 = 1.45

COV = 12%
Based on 100 years:
Project Bias varies with time interval which will
be reflected in calibrated load factor
Not strongly influenced by span length

Typical Results For SLS Live Load Model

Span60ft
1.60
1.40
1 20
1.20

Bias

1.00
ADTT250
0.80

ADTT1000
ADTT2500

0.60

ADTT5000
ADTT10000

0.40
0.20
0.00
1

10

100

1000

Days

10000

100years

Conclusion For Non-fatigue

Non fatigue SLS
Not necessary to envelop all trucks SLS
expected to be exceeded occasionally
Some states with less weight
enforcement may have to have additional
considerations (site/region specific live
load)
HL-93
HL 93 adaptable
d t bl as national
ti
l notional
ti
l SLS
live load model

Non-Fatigue
Non
Fatigue SLS LL Model
Mean
Mean, Bias and project LL model at mean
plus 1.5 standard deviations tabulated with
and without DLA for parameters:
5 ADTTs = 250, 1,000, 2500, 5000 and 10,000
10 Time periods = 1 day, 2 weeks, 1 month, 2
months, 6 months, 1 year, 5 years, 50 years, 75 years
and 100 years
6S
Spans = 30 ft
ft, 60ft
60ft, 90ft
90ft,120ft,
120ft 200 ft & 300ft
With and w/o DLA

Fatigue SLS LL Model

Live
e Load
oad For
o Fatigue
at gue II ((finite
te fatigue
at gue life)
e)
NCHRP Data - Indiana

Standard N
Normal Variable

-2

Station - 9511
Station - 9512
Station - 9532
Station - 9534
Station - 9552
Ontario

-4

-6

50

100

150

200

250

300

GVW [kips]

Miners
Miner s law yields one effective moment per span
Rainflow counting yields cycles per truck
Variety of spans and locations yields Mean, bias and COV

Examples Using FHWA WIM Data

M eq

p *m
n

i 1

3
i

Meff [kip-ft] for 3 sites

30 ft (-184)* 60 ft (-360)* 90 ft (-530)*

120 ft (-762)*

200 ft (-1342)*

83
83

204
204

269
269

408
408

845
845

90

-215

-300

-452

-896

86
86

-217

-291

-439

-916

* Values in parentheses= current AASHTO fatigue moment

15 sites processed so far

Example Using FHWA WIM Data 3

sites
M eq / MFat Trk
Fatigue II Load Factors for 3 sites
30 ft

60 ft

90 ft

120 ft

200 ft

0.45

0.56

0.51

0.54

0.63

0.48

0.60

0.57

0.59

0.67

0.47

0.60

0.55

0.58

0.68

So far looks good

good, now add cycles per
Passage and compare to current

Cycles
y
Per Passage
g
4.00

C
y
c
l
e
s

Arizona(SPS1)
Arizona(SPS2)
Arkansas(SPS2)
Colorado(SPS2)
D l
Delaware(SPS1)
(SPS 1)
Illinois(SPS6)
Kansas(SPS2)
Louisiana (SPS1)
Louisiana(SPS
1)
Maine(SPS5)
Maryland(SPS5)
Virginia(SPS1)
Wisconsin(SPS1)

3.50
3.00
2.50
33% damage increase

2.00

Current

1.50
1.00

ContinuousBridges
g
MiddleSupport

0 50
0.50
0.00
30

80

Span length
Spanlength

130

180

Rainflow Cycles - nrc

Continuous Spans
ContinuousSpans
30 ft

60 ft

90 ft

120 ft

200 ft

3.13

3.03

3.38

3.02

2.36

3.09

2.85

3.00

2.76

2.38

3.30

3.30

3.52

3.04

2.44

Damage Factor Compared to Current

eq

/ MFat Trk 3

nrc
nAASHTO

Current =0.75
Current
0.75
30 ft

60 ft

90 ft

120 ft

200 ft

0.52

0.71

0.66

0.68

0.73

0.57

0.74

0.71

.73

0.78

0 55
0.55

0 78
0.78

0 73
0.73

0 73
0.73

0 80
0.80

High = 0.87 or 116% of current

MM Independent Check of UNL

UNL running all filtered trucks at a site using
the time stamps
Traffic simulation
Not individual trucks one at a time

Test axle train evaluated by UNL and MM

8 hypothetical trucks
49
9a
axles
es
963 ft
843,000 lbs

MM Independent Check of UNL

MM Cobbled together
g
existing
gp
pieces:
Variation of program MM used in early 1990s truck
study that resulted in HL93 Loading modified to
calculate moment time histories
Used rainflow counting algorithm based on ASTM E
1049 85 p
previously
y developed
p to p
process
instrumentation data for repair of in-service bridge to
calculate cycles per truck; and
Miners Law
La to calculate
calc late Meq.
Meq

MM Independent Check of UNL

Results:
O
Only
l a few
f
issues
i
negotiated

ti t d
Final results damage factors same for simple span,
veryy close for Neg
g moment at p
pier of continuous.
Sometimes intermediate results varied seemed to
depend maximum magnitude of small cycles (noise)
th t was iignored---like
that
d lik d
data
t smoothing
thi

Common sense check MM found that

equivalent
i l t single
i l cycle
l d
damage ffactor
t ffor
the 8 truck train could be used as a
comparison
i
check
h k worked
k d well.
ll

19665
19770
19775
19880
19885
19990
19991
19992
19993
19994
19995
19996
19997
19998
19999
20000
20001
20002
20003
20004
20005
20006
20007
20008

NumbeerofTruckkCombinaations

Does This Increase Make Sense?

2,500,000

2 000 000
2,000,000

1,500,000

1,000,000

500 000
500,000

Year

PercentChangee

Does This Increase Make Sense?

140.0%
120.0%
100.0%
80.0%
60.0%
40 0%
40.0%
20.0%
0.0%
20.0%

19921997
19922002

TruckWeight

Does This Increase Make Sense?

Current Status of LL Studies

Fatigue II Being calibrated now Concrete
and steel
Fatigue
g I model being
g finalized
Other SLS
Design model will be HL93 factored per calibration
LL was handed off to NCHRP 12-83 team for concrete
SLS calibration - working
SHRP team is following with deflections and foundations

Concrete-Related
Concrete
Related Limit States
LRFD
article

Description

Proposed SLS
ProposedSLS

Service IA:
Crack control of R/C
/
Service IB:
Crack control of R/C concrete deck
designed using empirical method
ServiceIIIA:Decompression
Stresses check at service III ServiceIIIB:Uncrackedsection(max
5 9 4 2 limit
5.9.4.2
li it state
t t after
ft
l
lossesfully
f ll tensilestress)
t il t
)
ServiceIIIC:Crackedsection
prestressed components
(specifiedcrackwidth)
Control of cracking by
distribution of reinforcement
Reinforcement requirements
9.7.2.5 for concrete deck designed
using empirical method
5.7.3.4

Reliability Indices of Existing P/S Conc.

Conc
Bridges
Service III Limit State

Reliability Indices of Existing P/S Conc.

B id
Bridges
5

Relialbity Inddex

Relialbity Ind
dex

4
3
2
ave=0

1
0
-1
-2
0

20

40

60

80

100

120

140

160

Span Length (ft.)

5
4
3
2
1
0
-1
-2

ave=0.2

20

40

60

80

100

120

140

160

Span Length (ft.)

Decompression

Max.AllowableTension

Relialbity Index

ave=2
2

3
2

Reliabilityindex ofexistingbridges
AssumingADTT5000

1
0
-1
-2
2
0

20

40

60
80
100
Span Length (ft.)

120

140

160

Max.AllowableCrackWidth
(0.016in.,1year returnperiod)

Reliability Indices of Existing P/S Conc.

B id
Bridges
ADTT

Reliabilityindex(returnPeriod1year)
Maximum
Maximum
Decompression AllowableTensile AllowableCrack
SStress
ess
d
Width
0.2
0.4
2.35
0.1
0.3
2.20
00
0.0
02
0.2
2 00
2.00
0.15
0.1
1.88

1000
2500
5000
10000
ProposedTarget
0
0.0
0*
02
0.2
Beta
In any one year period the limit state will be exceeded in:
500 of 1000 bridges
g for reliability
y index of 0.0
23 of 1000 bridges for reliability index of 2.0

20
2.0

Relialbity Ind
dex

4
3
2
1
0
-1
-2
-3
-4

ave=0.15

20

40

60

80

100

120

140

160

Span Length (ft.)

Decompression
4

4
3
2
1
0
-1
-2
-3
-4

ave=0.06

20

40

60
80
100
Span Length (ft.)

120

140

160

Max.AllowableTension

ave=1.9

Relialbity Index

Relialbity Ind
dex

Reliability Indices of Bridges Designed to

C
Current
tS
Specifications
ifi ti

Sameexisting bridgesexceptNo.of
strandsdeterminedusingcurrent
specifications

2
1
0
-1
-2
-3
-4
0

20

40

60
80
100
Span Length (ft.)

120

140

160

Max.AllowableCrackWidth
(0.016in.,1year returnperiod)

ReliabilityIndex
AssumingADTT5000

Reliability Indices of Bridges Designed to

C
Current
tS
Specifications
ifi ti
ADTT

1000
2500
5000
10000

Decompression
0.05
0.05
0.15
0 15
0.35

PerformanceLevel
Maximum
AllowableTensile
SStress
ess
0.26
0.11
0.06
0 06
0.21

Maximum
AllowableCrack
d
Width
2.20
2.06
1 90
1.90
1.80

In any one year period the limit state will be exceeded in:
660 of 1000 bridges for reliability index of -0.15
29 of 1000 bridges for reliability index of 1.90

Parametric Study of Reliability Index

Three cases were considered:
Bridges designed with various spacing,
span lengths
lengths, and section types
Bridges designed with different span
l
lengths
th and
d section
ti ttypes b
butt same girder
id
spacing
Bridges designed with different span
lengths and girder spacing but same
section types.

5.0

5.0

4.0

4.0

Relialbity Index

Relialbity IIndex

Parametric Study of Reliability Index

3.0
2.0

30
3.0
2.0

1.0

1.0

0.0

0.0
30.0

60.0

80.0

100

120

Span Length (ft.)

Existing
g Bridges
g

140

30.0

60.0

80.0

100

120

140

Span Length (ft.)

Redesigned
g
Bridges
g

Various girder spacing, section types, and

span lengths
lengths.
ADTT = 5000
Max
M allowed
ll
d crack
k width
idth

Conclusions Related to SLS for Concrete

St
Structures
t
Different limit states may require different
target reliability index to maintain current
performance

Bluewater Bridge
g #2
First LRFD Major Bridge
Opened 1997