Effective Stress Design For Floodwalls on Deep Foundations

Transcription

1 Effective Stress Design For Floodwalls on Deep Foundations Glen Bellew, PE Geotechnical Engineer USACE-Kansas City 23 April 2015 Contributors James Mehnert, PE USACE-Kansas City Paul Axtell, PE, D.GE Dan Brown and Associates US Army Corps of Engineers

2 Outline Project Background Load Cases Considered Seepage Analysis Foundation Analysis Observed Performance 1993 Flood Existing Wall Stability Alternatives Considered and Selected Design Verification Load Test Major Findings/Lessons Learned Construction Photographs

3 Project Location Fairfax Jersey Creek Levee Missouri River BPU Floodwall Fairfax-Jersey Creek Levee Unit Kansas River

4 Project and Leveed Area Details Levee/Flood Wall constructed 1940 s by USACE Highly Developed Area (~$3.3 billion) BPU Floodwall Critical Infrastructure (Power Plant, water treatment) 1400 ft Major Manufacturing (GM Plant) Kansas River

5 Existing Floodwall and Subsurface Conditions ~16 ft ~20 ft CL/ML g=119 pcf ~20 ft Sheet Pile ~80 ft Sand g=116 pcf Fluted, Tapered Steel Pipe Piles

6 Non Critical Load Case Short Term Flood Typical infrastructure analysis, buildings, bridges, etc. No time for blanket seepage Pre-flood s and stress history control S u Sand, f, g ~Horizontal Seepage

7 Critical Load Case Long Term Flood Analysis specific to water retention structures ~Vertical Seepage, reduces s Effective Stress Controls behavior, f, g flood Sand, f, g ~Horizontal Seepage

8 Effective Stress Design Process Establish seepage conditions (effective stress) Determine Ultimate Axial pile capacity Lateral response of pile group (often controls design) Calibrate analysis to observed performance

9 Seepage Analysis Criteria Dh i=dh/z z Historically criteria has focused on preventing rupture/heave of topstratum by limiting vertical gradients to less than critical gradient (i c = g /g w ). Original design (1940 s) design ensured H < z. Current requirements are FS >1.6

10 Seepage Analysis Methodology calculating h Blanket Theory (EM ) Simple geometric inputs (great for simple stratigraphy) Decades of performance to verify adequacy of method Spreadsheet solutions quick to perform

11 Seepage Analysis Methodology calculating h Finite Element Modeling (next EM ) Unlimited complexity in geometry and boundary conditions Modeling quirks can lead to unrealistic results for a novice user In situ permeabilities, boundary conditions, model extent User interface improving, but can be time consuming to set up Use when complexity warrants

12 Pile Design Methodology Axial Capacity Overall Drained Strength Parameters Effective State of stress reasonably assumed for flood conditions EM Criteria - FS min = 1.7 Side Resistance b method Nordlund for driven, tapered piles Tip resistance Bearing Capacity Factors

13 Pile Design Methodology Lateral Response Typical to use Ensoft s Lpile and/or Group Software p-y curves by soil type (drained sand, undrained clay) Unit weight Friction angle p-y modulus (k p-y ) Group effects auto p-mult. Criteria Max D = 1.5

14 Effective Stress Lateral Response - Ensoft Design Case Long Term Flood P-y curves not available for drained conditions in cohesive soil Use Sand Curves with appropriate f Cannot input U>hydrostatic directly Reduce g of blanket by g flood = g -ig w also accounts for artesian sand p-y modulus (k p-y ) estimated based on soil type/strength Loose-Medium Sand or Soft-Medium Clay Group requires an estimate of the axial load response (auto or input)

15 Performance Observations ~3 ft ~45 Day duration Seepage some reports of concentrated seeps with possible pin boils, no major boil activity Structural Performance no performance observations noted Documentation limited

16 Calibrate with Back Analysis of 1993 Flood? i avg = 0.7 RESULTS Seepage: FS~1.3 Pile Capacity: FS = 1.5 Pile Structural >failure Deflections 1.5 max g flood = 13 pcf f = 29 deg k p-y = 50 pci P-y curve API Sand g ' = 53.6 pcf f = 36 deg k p-y = 60 pci P-y curve API Sand

17 Probability of Failure (%) No failure predicted, none observed 45 Probability of Failure Brittle Response Maximum Possible Load Maximum Historical Load Loading Example fragility curve, not BPU floodwall

18 Existing Floodwall Analysis w/ TOW i avg = 0.83 FS i = 1.1 g flood = 9 pcf f = 29 deg k p-y = 50 pci P-y curve API Sand g ' = 53.6 pcf f = 36 deg k p-y = 60 pci P-y curve API Sand

19 Existing Floodwall Results w/ TOW Axial FS <1 Deflections >>1.5 Floodwall modification needed

20 Design/Site Constraints Landside The Good: Well Defined Site (<100 spaced borings) Laboratory Data (consol, R-bar, class.) Foundation Load Test during construction Riverside The Bad: Constrained ROW Maintain similar pile spacing No driven/vibrated elements Drilled Shafts Difficult Design Case (low effective stresses) Lateral Deflections a major design constraint (limit to 1.5 under extreme load)

21 Modification Alternatives 1. Cut off and Found. New Foundation $ g flood = 53.6 pcf f = 29 deg k p-y = 50 pci P-y curve API Sand Full Depth Cut-Off (~100 feet) $$$

22 Modification Alternatives 2. RW and Found. New Foundation $$ Relief Wells $ g flood = 25.4 pcf f = 29 deg k p-y = 50 pci P-y curve API Sand

23 Selected Modification Alternative RW and Found. New Foundation Cap Extension and Buttresses Structural Modification 1 st Contract 24 Steel Casing, HP 12x74 Relief Wells 2 nd contract

24 Load Test Planning and Considerations ASTM D 1143 loading procedures B Maintained Load Test and C Loading in Excess of Maintained Test (2 hr holds) Estimate drained response (need extended static holds 2 24-hr holds lateral and 1 24-hr hold axial) Production Style shafts for combined/lateral ~40 kip lateral and ~35 kip axial design loads Groundwater conditions and stress states from load test to design condition are very different (link with s )

25 Load Test Goals Variables in Axial Analysis f g Interface friction, d Reasonably Known for Design Case Nice to Validate with Load Test Variables in Lateral/Group Analysis f g Sand p-y curve K p-y Axial response curves Reasonably Known for Design Case Need to Validate with Load Test Nice to have from Axial Load Test Combined Load Test structural performance of hybrid shaft

26 Load Test Overview Axial Figures and photos courtesy Dan Brown and Associates.

27 Load Test Overview Lateral/Combined Figures and photos courtesy Dan Brown and Associates.

28 Axial Load Test Results 2 hr 130 kip 24 hr hold Axial Results Data courtesy Dan Brown and Associates.

29 head Lateral Load Test Results Lateral Results 120 kip 24 hr hold 60 kip 24 hr hold 2 hr Data courtesy Dan Brown and Associates.

30 Load Test Results Applicability to Design Case Drained conditions reasonably approximated during load test Back analyze load test responses to calibrate lateral model Need state of stress during lateral load test (including suction) effective stress model applicable to both design and load test conditions (Lpile is frictional - f, g ) K p-y will be over-estimated in back analysis of load test if suction is ignored. Design Water Surface Normal Ground Water

31 Load Test Effective Stress - Soil Suction Soil Water Characteristic Curve (SWCC) ASTM D 6836 Relates in situ volumetric water content to soil suction Suction profile with depth = effective stress profile g unsat > g

32 Shear Strength with Soil Suction Estimating shear strength with soil suction Khalili and Khabazz (1998) t s = c + s v tanf + Cytanf Where, t s = unsaturated shear strength c = drained cohesion (zero) s v = gravity stress y = matrix suction f = drained friction angle C = fitting parameter Can t input t s directly into a frictional L-Pile model

33 Considering Soil Suction in LPile Calculate a Modified Friction Angle to account for soil suction s v tanf + Cytanf = s v tanf m where f m = modified friction angle Solve for f m for blanket to get an applicable friction angle that is f(suction). Assumes f m that results in appropriate t s is reasonable to account for suction in a frictional model. Material f f m Blanket Sand Necessary because Ensoft doesn t have ability to directly account for U.

34 Axial Load Test Interpretation Soil/Casing interface friction angle Assumed f=d, measured 1.1f=d (conservatism or incomplete drainage?) Axial Response curves Develop normalized (to ultimate capacity) side resistance and tip resistance response curves for use in Group

35 Lateral Load Test Back Analysis w/ normal GWT and suction Calibrate K p-y for verification of design Assumes K p-y same for all states of stress for effective stress analysis Solve for this g gravity = 115 pcf f m = 39 deg k p-y = Variable P-y curve API Sand Verify this is appropriate g gravity = 53.6 pcf f = 36 deg k p-y = Variable P-y curve API Sand

36 Lateral Load Test Back Analysis Results Load Test Calibrated Analysis Working Load 30 kip 60 kip Original Calibrated Material K p-y K p-y Blanket Sand Conservative original estimate?

37 Major Findings and Lessons Learned Load Test Drained conditions approximated during 2 hr load steps A complete test with 24 hr minimum holds next time? Sand p-y curves approximate drained behavior of fine grained soil Modified friction angle can account for soil suction in Lpile Load and temperature variations can be problematic during extended static holds Consider direct U dissipation measurement adjacent to shaft Design Can reduce FS min if load test performed during design K p-y was reasonably estimated prior to load test Ensoft programs account for effective stress design Accounting for U directly would be an improvement FLAC or finite element could improve understanding

38 Construction Shaft Installation

39 Construction Cap Extension

40 Construction Completed Wall Modification

41 Questions?