Proceedings of Fifth International Symposium on Deep Foundation on Bored and Augered Piles (BAPV), Ghent, Belgium, 8-10th Sept, 2008

1 INTRODUCTION

Cross-Hole Sonic Logging has been a common method of assessing cast in place concrete founda-tions since the 1960s. It is now used extensively throughout the world and on many significant con-struction projects.

Its popularity can be attributed to two main factors firstly, there is no depth limitation to the method and secondly, the apparent ease of interpreting results, compared with low strain type integrity tests.

However, if the equipment operator is not able to view all of the raw data and select appropriate fil-ters, it can be possible to come to the wrong conclu-sion and interpret results incorrectly. Is has not been unknown for piles to be condemned and replaced, because of problems with tube installation, such as:

Tube debonding Poor tube joints Joint wrapping Bent tubes

These can all appear to alter the first arrival time of the signal, even though the true velocity of the signal in concrete between the tubes is normal.

This paper investigates the ways of differentiating between true pile shaft defects and tube defects when interpreting sonic logging results.

2 SONIC LOGGING INTERPRETATION

Figure 1. Simple CSL Schematic

The principle of cross-hole ultrasonic logging is very simple, in that it measure the time taken for a signal to travel from one transducer to another, between tubes cast into concrete. The time will depend on the distance between transducers and the material be-tween the transducers. The further apart the trans-

Interpretation and Misinterpretation of Sonic Logging Test Results

Williams, H.T. & Jones, I. Testconsult Limited, Warrington, Cheshire, United Kingdom

ABSTRACT: Cross-Hole Sonic Logging (CSL) is one of the most powerful methods of assessing the integ-rity and quality of cast in place foundations. It offers many advantages over low strain methods, in particular the ability to determine the vertical and lateral extent of anomalies at any depth. In recent years 2 and 3 di-mensional tomography is being applied to results to present a graphical visualisation of results, which are easy for engineers to understand. In addition first arrival times can be automatically picked from response signals. However, without understanding how these new developments are created, there is a real danger that results can be misinterpreted. This paper explores the causes and effects of real and apparent defects in cast in place piles on cross-hole sonic logging results.

ducers and the lower the density of material, the longer the transit time. In homogenous concrete, free of defects, the velocity of sound is constant and in the order of 4000m/sec. Concrete containing soil inclusions, bentonite, hon-eycombing etc has a lower sound propagation veloc-ity. This means that measurements of wave speed or transit time can be used as a non destructive method of assessing the quality of buried concrete founda-tions. Anomalies in the concrete are indicated by a change in signal arrival time or amplitude. A typical signal is shown in Figure 2.

Figure 2: Typical CSL signal More recently cross-hole tomography techniques have been used to produce both two and three di-mensional images of the pile shaft. Anomalies are shown as different shaded coloured area on a visual representation of the pile shaft.

A waterfall plot is usually produced, which is effec-tively a profile built up from modulated signals, taken from each test level, see Figure 3.

Figure 3: Typical CSL waterfall plot

The wave speed of ultrasonic waves in concrete is given by :

V2 = E (1-)____ (Equation 1)

(1+ ) (1 2 ) Where: V = wave speed, = density, E = dynamic modulus & = Poissons ratio

In practice, the transducers are placed in water filled tubes, cast into the concrete. So the signal actually has to pass through water/tube and tube/concrete in-terfaces twice on its journey.

First Cautionary Note: Each interface has the po-tential to alter the quality of the signal, independ-ently from the quality of concrete between the tubes the very thing we are trying to assess!

2.1 What is a significant change in first arrival time (FAT)?

Correlations between test results and exca-vated/cored defects indicate that an increase in FAT of 20% or more is significant. This corresponds to a 17% reduction in apparent signal velocity. So, con-crete with a normal velocity of 4000 m/sec would reduce to 3320m/sec.

Reductions in FAT of less than 10% are not consid-ered to be significant. This corresponds to a 9% re-duction in apparent signal velocity. So, concrete with a normal velocity of 4000 m/sec would reduce to 3640m/sec.

Reductions in FAT between 10-20% are of interme-diate significance and the total number of profiles should be taken into consideration. 2 and 3D tomo-graphy can be of assistance in visualising the lateral extent of anomalies.

Figure 4: Significant mid-shaft defect found in a retaining wall in New Zealand.

Proceedings of Fifth International Symposium on Deep Foundation on Bored and Augered Piles (BAPV), Ghent, Belgium, 8-10th Sept, 2008

Figure 4 shows 3 profiles from a test result showing a significant increase in transit time. The FAT has increased between 57 76% indicating a significant defect over a vertical zone of approx 1m. By pre-senting results alongside each other, it is visually apparent that the anomalies are connected. Signal energy has also decreased by approx 20 24 dB.

It is important to view the individual signal within defective zones, to ensure that the automatic FAT calculation is taken from the correct first arrival. Figure 5, below, shows a damped signal, however the first arrival is still visible. It may be appropriate to re-test a signal such as this with a higher signal amplification.

Figure 5: Damped signal

2.2 Base Defects Figure 6 shows a result from a pile with contamina-tion at the base. This type of defect tends to occur with tremied piles cast under bentonite, when it is difficult to clean the base.

The signal does not disappear suddenly, but gradu-ally increases in transit time. This could indicate pe-ripheral contamination; however with an increase in excess of 150% it is probably significant enough to affect the whole of the pile section. This was con-firmed by the other 5 profiles.

Figure 6. CSL profile of base defect in London

2.3 Measurement of Concrete Velocity and Bent Tubes

It is not uncommon for CSL to be used to determine concrete velocity and hence give an indication of concrete modulus. However, the test does not meas-ure velocity directly, it measures the transit time be-tween the probes. Converting this to velocity in-volves assuming a path length. The path length is only known accurately at the top of the pile, where the distance between tubes can be accurately meas-ured. In practice the tubes can bend over the length of the pile, giving rise to gradual changes in trans-mission time. If the path length is assumed to be constant, then velocity calculations will be incorrect and misleading. If the tube spacing is known, an ap-parent velocity can be calculated by dividing the tube spacing by the transit time. It must be remem-bered however, that this apparent velocity includes the water and the tubes. It should also be noted that a signal travelling around a void could yield the same velocity as one travelling through a zone of low modulus material.

Figures 7a & 7b, show the results for a pile with bent tubes. Figure 7a shows a maximum increase in transit time of 43%. Assuming that the tubes are straight this would correspond to a reduction in con-crete velocity of 30%, i.e. to 2800m/sec if normal velocity is 4000m/sec. This would be comparable to much weaker concrete. This is clearly misleading and could lead to the pile being condemned incor-rectly. The corresponding profile shown in Figure 7b shows a matching reduction in transit time. It is unlikely that concrete properties have increased so dramatically!

Figure 7a. Tubes bend out Figure 7b. Tubes bend in

As a general guide if tubes are bent, signal transit time will tend to change gradually. If the tube is se-verely bent or kinked, it is unlikely the transducer will pass anyway. The increase may also be matched by an opposing decrease by other profiles. Increases in transit time caused by voids or contamination, tend to appear more abruptly.

Second Cautionary Note: Calculation of concrete velocity from sonic logging results should be treated with caution, and clearly state they are ap-parent.

2.4 First Arrival Time versus Signal Energy Most modern CSL systems have the ability to view not just the individual signal, but also a modulated waterfall plot. A first arrival time (FAT) can then usually be determined and plotted out against depth (on some systems this is all that is displayed). The energy in the signal can also be calculated by meas-uring the area under the curve and plotted out against depth.

First Arrival time is generally considered to be the most important measurement with CSL. For this rea-son it is important to understand exactly where it is being measured. On modern digital systems, the FAT is measured automatically. It does this by set-ting two signal amplitude thresholds. The lower threshold is set to ignore background signal noise. The upper threshold is set to catch the first signifi-cant signal arrival.

A problem can occur if you are testing a large di-ameter pile or diaphragm wall unit with a large path length, especially if the system is not sensitive enough or emitter strength is insufficient. Because of the higher signal to noise ratio, the selection of a correct threshold is imperative for a correct FAT measurement. If it is set too high, it will miss the true first arrival and falsely indicate a problem.

Another reason for a low signal to background noise ratio is tube debonding. Where tube debonding oc-curs, the actual path length of the signal through concrete is unchanged. A small gap is introduced be-tween the tube and the concrete, which effectively reduces the amplitude of the signal.

Signal amplitude is therefore of secondary impor-tance to FAT and cannot be relied upon on its own, as a measure of concrete quality. It can however be used to back up FAT measurements. Figure 8 shows a sonic logging test result from a pile with tube debonding over the upper 5m of pile shaft. Signal amplitude is clearly significantly reduced, however the first arrival can be seen, albeit very faintly on the waterfall plot.

Figure 8a. CSL result for pile with de-bonded tubes over the upper 5m of pile.

Figure 8b shows the FAT and Energy plots using a correctly selected low threshold, however Figure 8c shows the same result with the threshold set too high. The signal at its most damped part has been in-correctly interpreted as having a 72% increase in signal transit time. On its own it could have lead to an incorrect interpretation. For this reason it is much more reliable to assess anomalous areas by viewing the waterfall plot and also the individual signal.

Figure 8b. Good Threshold Figure 8c. High Threshold

Plastic tubes have a tendency to de-bond from con-crete more readily than steel. They are also more prone to damage during breading out. For this reason metal tubes generally give better results. On deeper piles, plastic tubing may also suffer from heat of hy-dration or pressure and collapse.

Proceedings of Fifth International Symposium on Deep Foundation on Bored and Augered Piles (BAPV), Ghent, Belgium, 8-10th Sept, 2008

Third Cautionary Note: Do not rely purely on plots of first arrival against depth

2.5 Vertical Resolution Distance apart of readings

This varies from system to system. The smaller the vertical interval between readings, then the smaller the defect you will be able to detect, without stag-gering probes. Some systems take a reading every 20cm, whereas other take readings every 1 or 2cm. Whilst it is quite difficult to detect horizontal cracks in concrete piles with CSL, due to signal skipping, with 1cm spacing it is more likely you will detect some change in signal. With the power and memory of current computers, testing and storing results at 1cm intervals is no longer an issue and is probably best used as standard.

2.6 Poor Joints? Screwed and socketed steel tubing is the best. Welded joints can lead to transducers becoming stuck (very expensive!) or unable to pass. Another problem that can be encountered are wrapped joints. Site engineers in good faith may wrap joints with densotape type material to ensure a waterproof joint, unaware that the signal find it difficult to pass through this interface. Figure 9 shows a CSL result from such a case. This was clearly identified how-ever by the precisely spaced anomalies coinciding with the joint spacing!

Figure 9. Pile with lagged joints at 3m intervals

2.7 What is signal skipping? As you would expect, signals tend to take the short-est and easiest route wherever possible. If the trans-ducers are aligned exactly on the same level as a very thin crack, then the signal will simply travel up the tube a little and through the good concrete above or below. Even if the probes are staggered this will occur, although a slight shift in FAT and signal en-ergy may be observed.

2.8 So how can thin cracks be detected? It is recommended that low strain integrity testing is used if cracks are suspected. The signal from this type of test is unable to pass cracks and is travelling in the vertical plane rather than horizontally. This does presume however that there is good access to the top of the concrete.

2.9 When to Test? 7 days is the recommended minimum time that con-crete should be left to cure before testing. However, assuming that you are not relying on the test to measure concrete velocity (which would be inadvis-able as discussed above), CSL can be used as a comparative test and used to test concrete piles at 3 days. This would be purely to check that signal tran-sit times are constant and no changes exist. If that is the case then anomalous areas are not likely to sud-denly appear. If however, an area of increased FAT is measured, it would be advisable to re-test the pile again after at least 7 days, during which time con-crete strength may have improved.

2.10 Tube Layout what are you missing? The main drawback of CSL, is the requirement to pre-install tubes in foundations during construction (although in emergencies it is possible to core or drill holes in concrete for testing). The layout and number of tubes must therefore be chosen to suit the information that is required by the engineer. For ex-ample, if 3 tubes are used and attached equidistant to the reinforcement cage, then only 3 profiles are pos-sible and it is impossible to take a measurement across the centre of the pile. This may be critical if the pile is tremied, when core defects are more likely to occur. With 4 tubes, 6 profiles are possible, around the periphery and across the cores which is why it is the most widely used configuration.

On diaphragm walls, the tube layout will again de-pend on panel dimensions. It is recommended how-ever that tube spacing does not exceed 1.5m to en-sure good strength signals.

The more tubes, the better the lateral extent of de-fects can be determined.

2.11 Tomography how useful is it? As a quick overview of where anomalous areas are, tomography software is a useful tool. 2D tomogra-phy can also give you a clearer idea of the lateral ex-tent. However, they do not give an actual measure-ment of change in FAT. To do this, you must be able to ideally view each individual signal or if not a good waterfall plot. A simple FAT plot is no good unless you are confident it picked correctly see section 2.4 above.

First arrival time is king and should be the main ba-sis of all interpretation and used to quantify the se-verity of defects along with the number of profiles affected at the same depth.

2.12 Operator Error The most common operator errors are:

Starting tests with slack cables Carrying out tests without transducers level Not topping up tubes with water on long piles Pulling transducers up too fast

When commencing tests, particularly on longer piles, the slack should be taken out of cables and held just in tension before taking data. Figure 10 shows a result with approx 400mm of cable slack, which is evidenced by a perfectly aligned modulated signal at the base. The winch is moving but the transducers are not lifting. This will produce a test profile longer than the actual pile.

Figure 10: cable slack at base of pile

To ensure that the transducers are level, they should either be lowered together to the base, or when at the base, the signal should be viewed, and one trans-ducer raised and lowered until the transit time is at a minimum (this is best done say 2m from the base, in case of base contamination.

Many systems have a maximum speed that transduc-ers can be raised. The signal acquisition time should be such that data is processed and stored before the next signal is triggered. This is governed usually by warning lights, however with much faster computers being available, this is less of a problem. If warning lights are ignored however, signals may not to stored or processed, before the next one is acquired. This would lead to shorter profiles than expected.

On long piles, the cables will displace quite a lot of water. It is therefore necessary to top up the tubes being tested, before the transducers reach the top. The signal will not transmit through air and part of the pile profile will be lost if the water level is low.

To ensure that the full length of the pile is tested, we would recommend that the tube length is plumbed with a tape measure (if a metal weight the same size of the transducer is used, it may prevent jammed transducers). The tube top level and pile toe level should also be determined. By comparing all read-ing it is possible to confirm that the tubes go to the base of the pile, that the tubes are not blocked and that the tested length is correct.

3 CONCLUSIONS

In this paper, the interpretation of cross-hole sonic logging results has been discussed and potential pit-falls have been explored. Incorrect interpretation can be caused by many factors if care is not taken. Users should not rely too heavily on calculated val-ues such as apparent signal velocity, automatically picked first arrival time plots, signal energy plots and tomography profiles. Whilst these do give valu-able additional information, the severity of any anomaly should always be assessed mainly on the change in first arrival time, so the original signal should always be available for interpretation after testing. Where anomalies are suspected, the possibility of this being caused by the tube bonding, joints, loss of water, lagging, or bending should also be consid-ered.

REFERENCES

Stain, R.T. and Williams, H.T. (1991) Interpretation of Sonic Coring Results: a research project, Proceedings of the 4th International Conference on Piling and Deep Foundations, Stressa. Vol. 1, pp633-640.

Turner, M.J. 1997, Ciria Report 144, Integrity testing in piling practice