Khalid Alshibli, Ayman Okeil, Bashar Alramahi, and Zhongjie Zhang
Objectives Methodology Data Collection and Archiving Reliability Analysis Results Summary and Conclusions Implementations Acknowledgements
The main objective of this research is to: update the correlations that are currently used by LADOTD to interpret CPT data for engineering design purposes, and assess the reliability of using CPT data to predict soil shear strength in both the magnitude and spatial variations in the field with respect to the LRFD methodology.
Collect the available CPT/PCPT soundings with the corresponding boring log data from the LADOTD and other possible sources. Create an electronic archive of the CPT sites using geographic information systems (GIS). Process/ analyze the data and update the correlations of the shear strength and soil classification, which are currently used by the design section of LADOTD. Evaluate the spatial variation of soil engineering properties in the field. Develop the Louisiana CPT database for corresponding engineering properties as a general guideline for design purposes with reliability consideration in preparation for use in LRFD method. Make recommendations for future field data collections with respect to CPT and PCPT and field drilling.
CPT file header: 05-08-2000 3:18pm JD STA # 10 + 4 3 ON CL RM BENT # 1 JL BAYOU MACON LA 585 JN 332-04-0005 ID F5CKE/V539 EL 3 1. 5 GW 1 2 5 T K N This information is used to locate the project documents from LADOTD archive.
A total of 752 CPT soundings were analyzed and archived in ArcGIS Only 503 CPT were matched with adjacent borehole logs
Analyzed the CPT files using a specially developed excel template to predict the following soil properties: Bulk unit weight Undrained shear strength Soil index properties & classification Created a GIS database of all the CPT sites where all the CPT information can be retrieved by clicking on its location on the map. To compare to CPT data, the logs of the boreholes adjacent to CPT sites are located and the following information is recorded: bulk density, water content, LL, PI, S u, and classification for depth intervals of 10 ft. Archived electronic copies of the associated borehole logs in Arc GIS software.
Before performing any reliability calibrations, the compiled results in database were preprocessed by: Matching boreholes and CPT locations Averaging CPT readings over a chosen depth Limiting undrained shear strength scope 10
11 Matching boreholes and CPT locations was done by setting a limit of 150 ft maximum distance to justify the hypothesis of equal soil properties. 350 300 Distance Threshold = 150 ft 2000 1800 1600 No. of CPT Locations 250 200 150 100 50 1400 1200 1000 800 600 400 200 No. of Data Points 0 < 25' < 50' < 75' < 100' < 125' < 150' < 200' < 300' < 400' < 500' < 800' < 5,000' < 10,000' No. of CPT Locations 96 140 172 205 223 251 262 275 285 300 317 319 320 No. of Data Points 531 731 910 1068 1163 1263 1318 1419 1505 1636 1785 1804 1814 0
12 Locations where more than one borehole records were available within the distance threshold, a weighted average was used = = = = n i n j j i n j j i D D D q q 1 1 1 + + + = 2 1 1 2 2 1 2 1 D D D q D D D q q (case of 2 boreholes) 2 1 0.2414.7586 0 q q q + = (for D 1 =35 ft and D 2 =110 ft)
13 Raw CPT data was averaged over a depth equal to about 8 inches. This was necessary to eliminate sudden spikes without loosing the general trends. 0 Tip Resistance, q c (tsf) 0 50 100 150 200 250 300 350 0 Tip Resistance, q c (tsf) 0 50 100 150 200 250 300 350 spike 25 25 Depth, h (ft) 50 Depth, h (ft) 50 75 75 100 100
14 Soil properties outside of a minimum and maximum threshold were excluded, thus limiting the scope of findings from this study. Frequency (%) 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% 100 500 900 1300 1700 2100 2500 2900 3300 Undrained Shear Strength, S u (psf)
Objectives Methodology Data Collection and Archiving Reliability Analysis Results Summary and Conclusions Implementations Acknowledgements
16 Focused on the undrained shear strength, S u The calibration was performed by comparing borehole data and CPT estimates Borehole CPT Su Su By introducing the uncertainties, the limit state function is written as the difference between both quantities Z ( ) UC CPT = S S u u Random Variables
(Phoon and Kulhawy 1999)
18 First, the three main sources of uncertainty needed to be quantified γ Model from compiled database γ device from repeatability tests γ soil from literature (Phoon and Kulhawy 1999)
Conducted a repeatability test to assess uncertainty in device readings CPT Bore Hole = 3. 25 ft = 6 ft R Radius 19
20 Tip resistance showed an average COV = 19.8% Tip Resistance, q c (tsf) Tip Resistance, q c (tsf) COV(q c ) 0 0 50 100 150 200 250 0 0 50 100 150 200 250 5 0% 10% 20% 30% 40% 15 20 20 25 Depth, h (ft) 40 Depth, h (ft) 40 Depth, h (ft) 35 45 55 60 60 65 75 80 80
21 Unit weight estimates showed less COV ( ~1.5%) Unit Weight, γ T (pcf) Unit Weight, γ T (pcf) COV(γ T ) 75 90 105 120 75 90 105 120 0.0% 1.0% 2.0% 3.0% 0 0 5 15 20 20 25 Depth, h (ft) 40 Depth, h (ft) 40 Depth, h (ft) 35 45 55 60 60 65 75 80 80
Soil Classification affects COV of device readings 0 Soil Classification 2 3 4 5 6 7 40% 10 35% 30% 20 COV(qc) 25% 20% 15% Depth, h (ft) 30 40 10% 50 5% 0% 2 3 4 5 6 7 Soil Classification 60 70 80 22
23 Uncertainty in transformation model is a big factor in 2500 this study Underestimate Used the given S u expression S CPT u = q c σ N kt vo CPT overestimates S u (using N kt =15) Bore Hole Undrained Shear Strength, Su (psf) 2000 1500 1000 500 0 0 500 1000 1500 2000 2500 CPT Undrained Shear Strength, S u (psf) Overestimate
24 The bias and COV were found to be CPT S Bias λ = Mean u = UC Su CPT S Scatter = COV u = UC Su 2.01 71% Next task was to look into effects of various factors on uncertainty
25 Depth No clear correlation could be established Depth (ft) h 5 5 < h 20 20 < h 40 40 < h 60 h 60 Bias 1.473 2.203 2.191 1.826 1.940 COV(%) 85% 73% 61% 58% 73% # 123 317 189 122 111 N kt 22.1 33.0 32.9 27.4 29.1
26 Soil Classification (Zhang and Tumay) Clear correlation could be established Clay % < 25 25 50 50 75 > 75 Bias 2.112 2.273 1.954 1.822 COV(%) 67% 78% 76% 87% # 171 210 128 62 N kt 31.7 34.1 29.3 27.3
27 Soil Classification (Robertson) Clear correlation could be established Classification 2 3 4 5 Bias 1.358 1.899 2.067 2.643 COV(%) 71% 67% 63% 87% # 33 478 258 87 N kt 20.4 28.5 31.0 39.6
28 CPT Readings Clearest correlation was established q c σ vo (tsf) < 4 4 8 8 12 12 16 > 16 Bias 0.849 1.751 2.215 2.339 2.491 COV(%) 72% 62% 67% 70% 69% # 57 315 226 173 91 N kt 12.7 26.3 33.2 35.1 37.4
29 S u Magnitude Can be correlated using soil classification 2500 2000 1500 A two parameter study (classification 1000 and another parameter) is possible, however the 500 number of data points is not sufficient. 0 Bore Hole Undrained Shear Strength, Su (psf) Underestimate 0 500 1000 1500 2000 2500 CPT Undrained Shear Strength, S u (psf) Overestimate
30 Distribution type was then identified by performing a χ 2 statistical test (Normal vs. Lognormal) 0.5 0.45 Lognormal Frequency 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 Normal 0 0 1 2 3 4 5 6 7 8 Transormation Model Bias, λ
31 Based on these results, the following statistical parameters were used in the reliability study Variable Mean COV (%) Distribution Source Soil Uncertainty 1.0 33 Lognormal Phoon and Kulhawy (1999) Transformation Model varies Lognormal Current Study Tip Resistance varies Normal Current Study Overburden Pressure deterministic Current Study
Two approaches were followed in this study: 1 st Approach: A direct correlation approach to find the CPT coefficient, N kt, where a certain probability of exceedance, P e, is targeted 2 nd Approach: A detailed reliability analysis that accounts for all sources of uncertainty (device, expression, soil) explicitly to achieve a target reliability index, β T. 32
33 S u are determined using the given expression and correlated with corresponding borehole results A CPT coefficient, N kt, is determined for each datapoint and the results are statistically studied
34 PDF and CDF plots are generated and used to determine N kt for any desired P e value Frequency 100 90 80 70 60 50 40 30 20 10 0 1 0.75 Underestimated Overestimated if N kt = 15 is used 80.7% < 3 6-9 12-15 18-21 24-27 30-33 36-39 42-45 48-51 54-57 60-63 > 66 CPT Coefficient, N kt Probability 0.5 40% 0.25 19% 0 15 21 0 25 50 75 100 Correlated N kt
35 The CPT Coefficient, N kt, is calibrated for different target reliability levels, β T, using a Limit State Function (LSF) UC CPT UC η q c σ vo Z = Su Su = γ Su ξ Nkt P D F f (s ) S CPT u u f (s ) UC u S u P D F βσ Z f (z) Z Probability of Exceedance (Z>0) µ S CPT µ S u UC u UC u S, S CPT u µ Z Z
36 First, reliability is assessed for different assumed N kt values Then, an optimum N kt is identified through a minimization process 0.35 0.30 Repeat for: Soil classification q c /σ vo ratio Target reliability, β T (β-β T ) 2 0.25 0.20 0.15 0.10 0.05 Optimum N kt 0.00 0 10 20 30 40 50 60 CPT Coefficient, N kt
Objectives Research Tasks Data Collection and Archiving Reliability Analysis Results Summary and Conclusions Implementations Acknowledgements
Pe Nkt 50.0% 24.9 55.0% 26.9 66.7% 32.3 Tip Resistance - Overburden Pressure, qc - σ vo (tsf) The following Nkt values were determined 20 15 10 5 0 0 500 1000 1500 2000 2500 Borehole Undrained Shear Strength, Su (psf) 38
39 The following N kt values were determined Zhang and Tumay N P kt e ALL >75% 50-75% 25-50% <25% 50% 27.5 26.9 25.3 28.8 31.5 55% 31.1 30.0 29.3 32.5 35.3 66.7% 42.0 38.4 39.3 45.2 50.1 Robertson N P kt e ALL 2 3 4 50% 27.5 18.6 26.2 30.9 55% 31.1 21.0 28.6 34.3 66.7% 42.0 28.5 35.7 45.0
The study also showed that the CPT is capable of assessing the soil unit weight, γt, using Underestimate 125 γ T (kn / m3 ) = 8.32 log(vs ) 1.61log( z ) f s *100 qt γ ucpt Bias λ = Mean Borehole = 0.98 γu γ ucpt Scatter = COV Borehole = 12.4% γu Bore Hole Unit Weight, γ T (pcf) Vs (m / s ) = [10.1 log(qc ) 11.4] 1.67 100 0.3 75 50 25 Overestimate 0 0 25 50 75 100 125 CPT Soil Unit Weight, γ u (pcf) 40
Modern design codes specify design coefficients that vary based on site variability A procedure is established for assisting engineers in determining site variability based on CPT undrained shear strength estimates The procedure relies on comparing the COV of S u CPT and expected COVs from repeatability and soil (low, medium, high) variations combined 41
Objectives Research Tasks Data Collection and Archiving Reliability Analysis Results Summary and Conclusions Implementations Acknowledgements
The findings of this study can be summarized in the following : The database was integrated in a GIS system for ease of access and retrieval Sources of uncertainty were identified and quantified Reliability-based calibration methods were used to estimate CPT coefficient, N kt A framework for site variability assessment based on CPT results was proposed
Objectives Research Tasks Data Collection and Archiving Reliability Analysis Results Summary and Conclusions Implementations Acknowledgements
CPT S u The study demonstrated the importance of performing in depth statistical analyses of soil property estimation methods. It is therefore recommended that the LADOTD : Maintain the compiled database of CPT soundings and matching boreholes Perform periodic updates of the coefficients proposed in this study as more data become available Use recommended N kt values in conjunction with borehole results for a pilot testing period Document inconsistencies to assist in future calibrations
The authors gratefully acknowledged the financial support provided by the Louisiana Transportation Research Center (LTRC Project No. 06-6GT) and Louisiana Department of Transportation and Development (State Project No. 736-99- 1406). We thank Benjamin Fernandez from LADOTD materials laboratory and Jesse Rauser from Ardaman & Associates, Inc. for providing the CPT and borehole data. The authors acknowledge the assistance of the graduate students Chaytanya Mamidala and Ashwin Bommathanahalli who helped in data archiving and analysis. The constructive criticism and suggestions of Zhongjie "Doc" Zhang, Pavement Geotechnical Research Administrator, PRC, Report reviewers, and Mark Morvant, Associate Director of Research at LTRC are highly valued and appreciated.