comparison of five different methods for determining pile bearing capacity by jim long, univ. of...
TRANSCRIPT
Comparison of Five Different Methods for Determining
Pile Bearing Capacity
by
Jim Long, Univ. of Illinois
Wisconsin DOT
February 6, 2009
Madison, WI
Agenda
• Discuss Objectives/Tasks of Project• General Approach• Specifics
– Prediction Methods Investigated– Databases used for Assessment– Interpretation of Data
• Assessment of Predictive methods• Improved Method• Implementation into LRFD
Objective
To quantify the ability of the five methods (Wisc-EN, FHWA-Gates, PDA, corrected Gates, WS-DOT) for predicting pile bearing capacity in a way that allows Wisconsin DOT to assess when or if it is appropriate to use each of the methods and to confidently estimate the reliability/safety and economy associated with each method.
Tasks• Task 1 - Literature Review• Task 2 - Data Collection
– Collect pile information from the Marquette Interchange– Collect pile information from other past projects of WisDOT– Collect pile information from the PI’s on Collection of pile load tests– Catalog the character of the load test information
• Task 3 – Analysis– Quantify the ability of EN, Gates, and PDA to agree with capacity from static load
tests– Quantify the ability of EN, Gates, to agree with capacity from PDA, and quantify
agreement between EN and Gates– Identify limitations to the Gates method– Develop an improved modified Gates– Assess Washington State DOT method developed by Allen– Identify efficiency and impact of using promising methods compared to EN
formula• Task 4 - Report Submission
Studies Collected for DB#1
• Flaate (1964)• Olson and Flaate (1967)• Fragaszy (1988, 1989)• Paikowsky (1994)• Davidson (1996)• FHWA/Long (2001)• NCHRP 507 and Allen(2005/2007)
Results for 5 predictive methods based on DB#1
• EN-Wisc• FHWA-Gates• FHWA-Gates (corr)• PDA• Washington DOT (Allen)
Wisconsin - EN formula
• c = 0.2 for Wisconsin• Most states use built-in division by 6 to get
allowable bearing by specifying H in ft, and s in inches. Study shows that the estimate ends up to be about a FS = 3.1 wrt ultimate capacity.
)/( csWHQallowable
Methods – Gates formula
• Gates modified by FHWA
• Gates modified in this Study
100)10log(75.1 brultimate NeEQ
)(0 **** originalGatesHPSUltimate QFFFFQ
Effect of Corrected Gates
FHWA - Gates
0
250
500
750
1000
0 250 500 750 1000
Measured Capacity (kips)
Pre
dic
ted
Cap
acit
y (k
ips)
FHWA - Corrected Gates
0
250
500
750
1000
0 250 500 750 1000
Measured Capacity (kips)
Pre
dic
ted
Cap
acit
y (k
ips)
PDA
• Based on measurement of strain and velocity in the pile during driving
• Case method is applied – details in Report• There are different interpretation methods available and
different damping values that can be applied – makes the method more adaptable to local conditions, but also makes the method non-standard.
• Advantages – can determine energy going into pile• Disadvantage – does not account for setup – determines
capacity at the time of driving
Methods – Wash DOT (Allen)
)10ln(***6.6 beffultimate NEFQ
where
Feff = Hammer efficiency factor
0.55 for Air/Steam – all piles
0.47 for OED with steel piles
0.35 for CED with all piles
0.37 for OED with concrete or timber piles
Nb= Number of blows/in
E = hammer energy in ft-kips
Qult = Ultimate pile capacity (kips)
Statistical Results QP/QM
Mean COV Method
0.43 0.47 Wisc-EN
1.11 0.39 WSDOT
1.13 0.42 FHWA-Gates
0.73 0.40 PDA
1.20 0.40 FHWA-Gates for all piles <750 kips
1.02 0.36 corrected FHWA-Gates <750 kips
Observations
• In terms of scatter– corrected Gates (least scatter, limited to <750k)– WSDOT– FHWA-Gates (<750k), PDA– EN (greatest scatter)
• Trend for Gates is to underpredict at higher capacity and overpredict at lower capacity –address issue by restricting capacity < 750k
Database 2 – DB2
• Two sets of Data Collected– Wisc(JHL) 220 piles in which there are
estimates of capacity from dynamic pile behavior
– Wisc (MI) Marquette Interchange – collection of 96 piles. Estimates can be made with all dynamic methods. PDA and CAPWAP results for BOR.
– few static load tests
PDA
• EOD results with PDA determine the capacity of pile at the time of driving
• BOR results determine capacity at beginning of restrike
• BOR better accommodates effects of setup
• CAPWAP for BOR provides even better estimate of pile capacity
LRFD – Resistance Factors
• two approaches (FOSM, FORM)• Determines the resistance factor necessary for a
target reliability (index)• unknowns accounted for in both loads and
resistance• variables
– pred method (bias and cov)– loads (bias and cov)– target reliability (beta = 2.0, 2.5, 3.0)
• based on NCHRP 507
Resistance Factors - FOSM
• R= bias factor (which is the mean value of QM/QP ) for resistance• COVQD = coefficient of variation for the dead load• COVQL = coefficient of variation for the live load• COVR = coefficient of variation for the resistance• T = target reliability index• D = load factor for dead loads• L = load factor for live loads• QD/QL = ratio of dead load to live load• QD, QL = bias factors for dead load and live load
222
2
22
11lnexp
1
1
LDL
D
LD
QQRTQL
DQ
R
QQL
L
DDR
COVCOVCOVQ
Q
COV
COVCOV
Q
Q
Resistance Factor,
Using T=2.33
Predictive Method
bias,
cov
FOSM FORM
EN-Wisc 3.11 0.62 0.84 0.90
FHWA- Gates 1.09 0.50 0.39 0.42
PDA 1.67 0.50 0.60 0.64
WSDOT 1.07 0.45 0.42 0.46
“corrected” FWHA-
Gates for piles <750
kips
1.14 0.41 0.49 0.54
Since Report Submission
• We submitted report in June, 2008
• We have been continuing to work with IDOT reanalyzing and reviewing more data and methods
• If we look at the same data as we have for WSDOT, and “Fit the tail of the distribution”, we can justify higher resistance factors