Views on Accuracy of Tests and Analyses

Transcription

1 Piling & Deep Foundations Asia 29 Workshop A, July 13, 29 Views on Accuracy of Tests and Analyses Bengt H. Fellenius ====================================================================================================================== 9658 First Street Tel.: (778) Sidney, BC, Canada e-address: V8L 3C6 Web site: [

2 2

3 Don't connect the reference beams to each other. A foot up on one beam and both references are disturbed! 3

4 See? 4

5 Make sure the support stakes of the reference beams are far enough away from the support of the kentledge load and the latter far enough from the pile. The distances are prescribed in the ASTM Guidelines. 5

6 Most piles are installed slightly inclined. Consider this when arranging the reaction structure and always use a swivel plate to avoid edge loading. 6

7 Fellenius

8 The error can be small or it can be large. Results from two tests at the same site using the same equipment testing two adjacent piles, one after the other Head-down O-cell Pile August 26 2, Shinho-Pile August 26 1,5 25 Error in Jack Load (KN) % Error Error (KN) 1, 5 15% Error 5 2.5% Error 2, 4, 6, 8, 1, Loadcell (KN) 5, 1, 15, Loadcell load (KN) Note, the test on the pile called "O-cell pile" is a head-down test after a preceding O-cell test. 8

9 A routine static loading test provides the load-movement of the pile head... and the pile capacity? 9

10 DECOURT 235 1

11 T e l l t a l e s A telltale measures shortening of a pile and must never be arranged to measure movement. Let toe movement be the pile head movement minus the pile shortening. For a single telltale, the shortening divided by the distance between the pile head and the telltale toe is the average strain over that length. For two telltales, the distance to use is that between the telltale tips. The strain times the cross section area of the pile times the pile material E-modulus is the average load in the pile. To plot a load distribution, where should the load value be plotted? Midway of the length or above or below? 11

12 Load distribution for constant unit shaft resistance PILE HEAD Average 1 Load A1 LOAD, Q PILE TOE 5 Midheight h 1 DEPTH, z A2 Load Distribution Q = az r s = a 12

13 Linearly increasing unit shaft resistance and its load distribution Unit Shaft Resistance az Average Load x A1 LOAD, Q A 1 = 3 ax 3 h DEPTH, z h A2 Load Distribution Q = az 2 /2 A2 = a( h 3 2 3x h + 6 A 1 = A2 2x 3 ) X is where the average load should be plotted h X = 58h 3 =. 13

14 Today, telltales are not used for determining strain (load) in a pile because using strain gages is a more assured, more accurate, and cheaper means of instrumentation. However, it is good policy to include a toe-telltale to measure toe movement. If arranged to measure shortening of the pile, it can also be used as an approximate back-up for the average load in the pile. The use of vibratory strain gages (sometimes, electrical resistance gages) is a well-established, accurate, and reliable means for determining loads imposed in the test pile. It is very unwise to cut corners by field-attaching single strain gages to the re-bar cage. Always install factory assembled sister bar gages. 14

15 Rebar Strain Meter Sister Bar Three bars?! Instrument Cables Instrument Cable Reinforcing Rebar or Strand Reinforcing Rebar or Strand Wire Tie (2 places) Rebar Strain Meter Rebar Strain Meter (3 places, 12 apart) Wire Tie Tied to Reinforcing Rebar Tied to Reinforcing Rings Hayes 22 15

16 Lunch Generator turned off break Night Night Example of Electromagnetic Interference (EMI) noise from a generator and power cable affecting vibrating wire strain gages Sources of noise are for example arc welding, machinery ignition, power generators, and power cables, etc. Continual noise will impart a general trend of overall accuracy of data by increasing its spread between data points. Sudden noise, for example, when starting up a machine, the ignition may cause a spike in the readings. Electronic noise tends to result in a reduction in strain gage values, whereas magnetic noise increases the strain gage values. Osborne, N. and Tan, G.H., 29. Factors influencing the performance of strain-gage instrumented Monitoring systems. Geotechnical News, (27)

17 We have got the strain. How to we get the load? Load is stress times area Stress is Modulus (E) times strain σ = E ε The modulus is the key 17

18 For a concrete pile or a concrete-filled bored pile, the modulus to use is the combined modulus of concrete, reinforcement, and steel casing E comb = E s A A s s + + E A c c A c E comb = combined modulus E s = modulus for steel A s = area of steel E c = modulus for concrete A c = area of concrete 18

19 The modulus of steel is 2 GPa (27 GPa for those weak at heart) The modulus of concrete is....? Hard to answer. There is a sort of relation to the cylinder strength and the modulus usually appears as a value around 3 GPa, or perhaps 2 GPa or so, perhaps more. This is not good enough answer but being vague is not necessary. The modulus can be determined from the strain measurements. Calculate first the change of strain for a change of load and plot the values against the strain. E t = Δσ Δε Values are known 19

20 Example of Tangent Modulus Plot TANGENT MODULUS (GPa) Level 1 Level 2 Level 3 Level 4 Level 5 Best Fit Line MICROSTRAIN 2

21 In the stress range of the static loading test, modulus of concrete is not constant, but a more or less linear relation to the strain dσ E t = = aε + dε b Which can be integrated to: σ = a 2 ε 2 + bε But stress is also a function of secant modulus and strain: σ = Es ε Combined, we get a useful relation: E =.5a ε + s b and Q = A E s ε 21

22 Example of Tangent Modulus Plot 1 Level 1 9 Level 2 Intercept is b TANGENT MODULUS (GPa) Slope is a Level 3 Level 4 Level 5 Best Fit Line MICROSTRAIN 22

23 Note, just because a strain-gage has registered some strain values during a test does not guarantee that the data are useful. Strains unrelated to force can develop due to variations in the pile material and temperature and amount to as much as about 5± microstrain. Therefore, the test must be designed to achieve strains due to imposed force of ideally about 5 microstrain and beyond. If the imposed strains are smaller, the relative errors and imprecision will be large, and interpretation of the test data becomes uncertain, causing the investment in instrumentation to be less than meaningful. The test should engage the pile material up to at least half the strength. Preferably, aim for reaching close to the strength. Moreover, each gage reading must be assessed as to it being true to the actual strain and free from interference. 23

24 Results of static loading tests on 4 m long, instrumented steel piles in a saprolite soil. LOAD (KN) 2, 4, 6, 8, 1, UNIT SHAFT SHEAR (KPa) DEPTH (m) DEPTH (m) ? Each gage reading provides a load value. The treatment of the values has to be with understanding of the accuracy/error of each. The unit shaft resistance is calculated as the difference in evaluated loads for adjacent gage levels divided with shaft area (= distance between gage levels times pile circumference). The load difference is about the same magnitude as the error in each load value! 24

25 A more thoughtful analysis of the data LOAD (KN) UNIT SHAFT SHEAR (KPa) 2, 4, 6, 8, 1, ß = ß = DEPTH (m) ß =.4 DEPTH (m) ß =

26 Could there be load in the pile before we start the loading test? and, if so, Could such load have any significant effect on the data evaluation?

27 Determining load from strain-gage measurements in the pile The strain-gage measurement is supposed to be the change of strain relative the no-load situation (i.e., when no external load acts at the gage location). But, is the no-load situation really the reading taken at the beginning of the test? What is the true zero-reading to use? 27

28 We often assume somewhat optimistically or naively that the reading before the start of the test represents the no-load condition. However, at the time of the start of the loading test, loads do exist in the pile and they are often large. For a grouted pipe pile or a concrete cylinder pile, these loads are to a part the effect of the temperature generated during the curing of the grout. Then, the re-consolidation (set-up) of the soil after the driving or construction of the pile will impose additional loads on the pile. 28

29 What we measure is the increase of load in the pile due to the load applied to the pile head. The external-origin load in the pile that was there before we started the test is called "Residual Load". 29

30 Normalized Applied Load Load distributions in static loading tests on four instrumented piles in clay D E P T H 3

31 Example from Gregersen et al., 1973 LOAD (KN) LOAD (KN) Residual True minus Residual DEPTH (m) Pile DA Sand DEPTH (m) True Pile BC, Tapered A. Distribution of residual load in DA and BC before start of the loading test B. Load and resistance in DA for the ultimate load applied 31

32 Clay DEPTH (m) LOAD (KN) 5 1, 1,5 2, 2,5 Static Loading Test at Pend Oreille, Sandpoint, Idaho, for the realignment of US95 46 m diameter, 45 m embedment, closed-toe pipe pile driven in soft clay 2+ m Fellenius et al

33 Distribution of Measured Loads ( False Resistance ) and True Resistance 5 LOAD (KN) 5 1, 1,5 2, 2,5 Residual Load From Test data and from Calculations 1 DEPTH (m) True Resistance From Test data and from Calculations 35 4 Fellenius et al Measured Load From Test data and from Calculations 33

34 FHWA tests on.9 m diameter bored piles One in sand and one in clay (Baker et al., 199 and Briaud et al., 2) Cone Stress and SPT N-Index (MPa and bl/.3 m) Cone Stress (MPa) Silty Sand DEPTH (m ) q c Silty Sand N DEPTH (m ) Clay Clay 1 Sand 1 Pile 4 Pile

35 ANALYSIS RESULTS: Load-transfer curves.. LOAD (KN) LOAD (KN) 1, 2, 3, 4, 5, 1, 2, 3, 4, 5,. LOAD (KN) 1, 2, 3, 4, 5, DEPTH (m) Residual Load Measured Distribution Measured Distribution True Distribution 2. Residual Load DEPTH (m) Measured Distribution True Distribution PILE PILE 4 SAND SAND PILE 7 CLAY 35

36 Results of analysis of a Monotube pile in sand (Fellenius et al., 2) LOAD (KN) 1, 2, 3, 5 Residual Load Measured Resistance DEPTH (m) True Resistance 25 36

37 Method for evaluating the residual load distribution 2 RESISTANCE (KN) 5 1, 1,5 2, Measured Load DEPTH (m) Shaft Resistance True Resistance Extrapolated True Resistance 16 Residual Load Measured Shaft Resistance Divided by 2 37

38 The approach can be applied also to dynamic tests CPT Profile A Cone Stress, q t (MPa) B LOAD (KN) 5 1, 1, CAPWAP Determined Resistance Distribution Depth (m) Depth (m) Resistance Distribution (Data from Axelsson 1998) 38

39 Analysis Procedure LOAD (KN) 5 1, 1, Residual Load Measured Load Depth (m) Shaft Resistance Distribution Divided by True Resistance Extension of Matching Fellenius

40 Results LOAD (KN) 5 1, 1,5 2, 2 4 Depth (m) CAPWAP Determined Resistance True Resistance Residual CAPWAP Shaft Divided by 2 True Shaft Resistance 2 Fellenius 22 4

41 Presence of residual load is not just of academic interest 1, 8 No Strain Softening Residual Load present 1, 8 With Strain Softening Residual Load present LOAD (KN) 6 4 No Residual Load LOAD (KN) 6 4 No Residual Load 2 OFFSET LIMIT LOAD OFFSET LIMIT LOAD MOVEMENT (mm) MOVEMENT (mm) 41

42 Toe Resistance A 1,2 1, TOE TELLTALE HEAD LOAD (KN) TOE Does not this shape of measured toe movement suggest that there is a distinct toe capacity? MOVEMENT (mm) 42

43 A 1,2 1, TOE TELLTALE HEAD B 1,2 1, HEAD LOAD (KN) TOE LOAD (KN) TOE MOVEMENT (mm) "Virgin" Toe Curve MOVEMENT (mm) No, it only appears that way when we forget to consider the residual toe load (also called the initial, or virgin toe movement) 43

44 Residual load and other influences on evaluated load distribution

45 STRAIN (µε) 9d Strain measured during the 218-day wait-period between driving (grouting) and testing d 23d 3d 39d Do these strains really represent load in the pile, as present before the start of the static loading test? DEPTH (m) d 59d 82d 99d 5 122d 55 6 At an E- modulus of 3 GPa, this strain change corresponds to a load change of 3,2 KN 218d Day of Test 45

46 Concrete hydration temperature measured in a grouted concrete cylinder pile 7 TEMPERATURE ( C) Temperature at various depths in the grout of a.4 m center hole in a 56 m long,.6 m diameter, cylinder pile HOURS AFTER GROUTING Pusan Case 46

47 7 6 Temperatures measured in a 74.5 m long, 2.6 m diameter bored pile 7 6 Temperatures measured in a 74.5 m long, 2.6 m diameter bored pile TEMPERATURE ( C) TEMPERATURE ( C) Start of Static Loading Test DAY AFTER GROUTING DAY AFTER GROUTING Temperature records during curing of grout in the Golden Ears Bridge test pile, Vancouver, BC. Data courtesy of Trow Engineering Inc. and Amec Inc. 47

48 CHANGE OF STRAIN (µε) Rebar Shortening Change of strain measured in a 74.5 m long, 2.6 m diameter bored pile Slight Recovery of Shortening DAY AFTER GROUTING Change of strain during the hydration of the grout in the Golden Ears Bridge test pile 48

49 Photo of short (2 m) pieces with plastic tube enabling them to be submerged 49

50 Temperature During Curing of a 2m Lab Specimen TEMPTATURE (ºC) S W N E April 25, Time (h) 5

51 Strain History During Curing of a 2 m Lab Specimen 5 PHC Piece CHANGE OF STRAIN (µε) A INCREASING SHORTENING OF REBAR RECOVERY OF SHORTENING (LENGTHENING) Days after Grouting Pusan Case 51

52 Submerging the short piece, letting it swell from absorption of water 1 Concrete Cylinder Piece 5 STRAIN (µε) INCREASING TENSION Pile piece submerged Time (days) S N E W 52

53 The strain gages themselves ar not are temperature sensitive, but the records may be! The vibrating wire and the rebar have almost the same temperature coefficient. However, the coefficients of steel and concrete are slightly different. This will influence the strains during the cooling of the grout. More important, the rise of temperature in the grout could affect the zero reading of the wire and its strain calibration. It is necessary to heat-cycle (anneal) the gage before calibration. (A routine measure of Geokon, US manufacturer of vibrating wire gages). 53

54 Readings should be taken immediately before (and after) every event of the piling work and not just during the actual loading test The No-Load Readings will tell what happened to the gage before the start of the test and will be helpful in assessing the possibility of a shift in the reading value representing the no-load condition If the importance of the No-Load Readings is recognized, and if those readings are reviewed and evaluated, then, we are ready to consider the actual readings during the test 54

55 Of course, we must consider also other aspects: 55

56 Interpretation of a series of tests performed at different times Change of Horizontal Stress (KPa) Cell D1 5 Days 1 Day 8 Days 4 Months 22 Months Movement (mm) Results thought due to set-up and explained as Increase in Horizontal Effective Stress Movement (mm) Fellenius 22 5 Days 1 Day 8 Days 4 Months 22 Months Results plotted According to Movement Path 56

57 Also the best field work can get messed up if the analysis and conclusion effort loses sight of the history of the data LOAD (KN) 2, 1,5 1, DYNAMIC TEST DYNAMIC TEST in a DYNAMIC series of TEST blows STATIC TEST Repeated STATIC TEST MOVEMENT (mm) The dynamic test (CAPWAP) was performed after the static test. The redriving (ten blows) forced the pile down additionally about 45 mm. 57

58 Often the past routines are nothing but jog trotting along with the humdrum past. For example, incorporating unloading/reloading cycles is useless and actually impairs a test. 58

59 Does unloading/reloading add anything of value to a test? Result on a test on a 2.5 m diameter, 85.5 m long bored pile 3 25 Repeat Acceptance test Criterion Acceptance Criterion LOAD (MN) MOVEMENT (mm) 59

60 Plotting the repeat test in proper sequence 3 25 Acceptance Criterion Repeat test test plotted in sequence in sequence of testing of i LOAD (MN) MOVEMENT (mm) 6

61 2, 1,5 LOAD (KN) 1, MOVEMENT (mm) The above series of unloading/reloading has added nothing but cost the client a lot of money. 61

62 Thank you for your attention