Application of Stress: Wave Theory fo Pils, F B J Barends ject) © 1922 Raskema, Rotterdam, ISBN 90 54100826 Pile integrity testing and analysis ERausche Goble Rausche Likins and Associates, Inc.. Cleveland, Ohio, USA [Link] Pile Dynamics, Inc., Cleveland, Ohio, USA Shen Ren Kung Nanjing Hydraulic Research Instinne, People's Republic of China ABSTRACT: The methods of installation of drifled shafts give rise to concern among engincers regarding the structural Integrity of the shafts, ‘This hay led to a celuctance to use these foundations by some engineers. Alternatively, many engineers ate willing to accept these foundations provided that adequate inspection and integrity veritication testing are Petfovmed. Low strain integrity testing has seen substantial improvement in recent years. THe testing cost has been substan reduced and, as a result, the methods are now widely employed. However, the results of these tests must also be accurately evaluated. The evaluation may be a simple visual inspection of the data by an experienced engineer. Recent Provements in analysis capabilities now offer less intuitive interpretations at a reasonable cost. These new automated procedures ate superior to the visual techniques of the past beth from the standpoint of operator requited know-how and ‘user confidence fa results iy 1 INTRODUCTION [Link] Pile Integrity Testing (P.1°T.), a low strain method for ‘Three devices are needed to perform a low strain integrity integrity testing of driven piles and drilled shafts, uses a test: the hammer (with or without force sensor), the seasor variety of techbiques for the interpretation of force and and the processor (Figure 1). The primary difference velocity records taken under the impact of a light hammer between various interpretation methods fies in the pro- blow (Rauscie, Shen, Likins 1991). In its most basic form, gramming of the processor, The hammers and sensors ate, the input pulse signals are inspected directly in the time ‘with slight exceptions, very similar for transient dynamic domain for reflections or echoes and hence are feferred to response and sonic pulse echo methods. as the "Sonic Pulse Echo Method", ‘The interpretatea of the pile top velocity traces is further enkanced by an ampli- fication which increases exponentially in time (Paquet 1968). This concept became practical with the develop. ment of digital data processing 2.1 The hammer Depending on the pile size to be tested, the hammer mass should be between .5 and $ kg. Smaller hammers have shorter rise times and higher frequency content, while larger hammers apply greater energies to the pile top. To ‘The advent of fast spectral analyzers made possible the development of the frequency domain Non Destructive ‘Testing (NDT) methods generally applied to structaral concrete. NDT was then applied to iow strain pile testing, resulting in the “Transient Dynamic Response Method” which displays the records in the frequency domain in a so- called mobility plot. hoTION sensOn PROCESSOR WITH DISPLAY Puorren | =] FR Interpretations in frequency andlor time domain are selected either according to individual preference or locally developed practice, Generally, civil engineers feel more ‘comfortable in the time domain, particularly if they have NONANSTRUMENTED On been working with nigh strain Case Method testing. How- InsTmUMERTED HANMER, ever, the experienced P1-T. engineer should take advantage of all interpretation tools. This paper discusses various analysis methods applicable to a low strain test. It also ‘compares the results obtained with different hammers and sensors nal low-pass filter feequency. Conversion of continuous analog signals to discrete, digital representation requires & low-pass filter limit of at most one half of the sampling frequency. In [Link]. the usual digitizing frequency is ap proximately 30 kHiz and hardware low pass frequeccy response limits 10 kHiz. Additional user controlled digital low pass filtering is sometimes necessary when unwanted high frequency components mask the iinportant low fre- quency signal High-pass filtering removes low frequency components such as a constaat shift oa slow drift. Velocity records derived from acceleration records have been unjustly blamed for an erratic low frequency behavior which then yield useless dynamic stiffness (see mobility calculations below). While accelerometers are dynamic instruments whose signals slowiy drift towards zero, the average Velocity {actually the zero frequency component) is always set to zeto. This means that the displacement calculated from the velocity is also zero at the end of the record, surely a reasonable assumption considering the low impact energy. However, removing or reducing other low frequency com ponents, as suggested by Paquet (1968) and demonstrated below, may distor: records and would be tolerable only if ‘exactly the same filtering is always applied. Clearly, the caloulated stiffness values then only have relative vate. However, such a relazive effect can also be casily evaluated by merely comparing force and velocity atthe time of impact, 615 3.2 Amptiication Before the signals are converted to digital form, they are usually amplified by the analog signal conditioning to take advantage of the full digitizing range (e.g. 5 volts). This assures that smal signal changes are optimally resolved by the digital record. However, a high resolution analog-to- digital converter, such as the 16 bit A/D of the [Link]. Col lector, makes this less critical. The input wave returns at 2AJe after being reflected by the pile toc. Atcourse, sine the wave then fas traveled the longest distance, iis magi tude is usually strongly reduced by either internal (com: ‘rete) or external (soil) damping. Reflections from defects are also reduced by these damping effects: defects which are at deeper depths are therefore more ditficutt to detect ‘A effect of uniform friction om detecting defects or the toe can be minimized by using an amplification which increases ‘exponentially in tiie until time 2c. By applying this increasing amplification, exiceme weak signals from deeper locations are more easily detected. However. in gener: fiction effects are not uniformly distributed and therefore the beginning of the amplification should be delayed until the time when the major friction effects begin. An exan ple is given in Figure 5.a and Sb. In this calculated exam: ple (pile length 20 m and high side damping of 5 times pile impedance) the pile toe response was practically invisible at time 2L/¢ but was clearly apparent after 25X amplificatios The amplification started at 4 m with unit intensity and then increased exponentially as indicated below Figure 5.b. 3.3 Fourier transform and mobility Fourier transformation of pile top velocity and force versus time signals leads to the corresponding velaciey and farce spectra. Dividing the velocity spectrum by the force spec “9, VELOcHY - re pba ete Bireanennths il gether Manali, eee a oO WI Nh ch AD Ne PN Ne paps aN Fig. 5 (A) Unamplitied and (5) ampiiffed velocity time records caleulated for a 4 m pulse. Velocity spectra are respectively shown in (C) and (D).