Keywords: Concrete, Impulse-Echo-Technique, Online-lmaging Technique, Ultrasound
This paper was presented at the International Symposium Non-DestructiveTesting in Civil Engineering (NDT-CE) 26.-28.09.1995 in Berlin. NDT-CE, Full Program or the Ultrasound Part
The paper presented here reports the improvements on Ultrasound-Pulse-Echo Testing gained with a special testing-system. It will describe the testing technique in use and shows a couple of measurements done as for different testing parameters.
The testing-system used here allowes a computer-aided monitoring of the data in a B-scan. This way to display the data makes it most comfort for the tester to evaluate large numbers of single measurements ('shots') at a glance. Moving the transducer along the concrete surface he is enabled to compare directly the data from different positions, he simultaneously gets informations on intensity and phase of the signal and their relative changes. A Personal Computer included in the testing-system visualizes the B-scans in indicated colours or grayscale by running a specially developed, online-imaging software. This computed imaging turns out to be very effectiv on the localization of defects or backwall-echos little significant in the A-scan. By the online display relative deviations in the specimen are indicated and can be recognized quite fast, so after a previously coarse localization e.g. of a defect more exact views on this area can follow. The measurement may be documented afterwards by saving the scan with an additional text protocol to disc or colour printer.
A further shaping of the measured signal becomes necessary due to the heterogene structure of the concrete (aggregates, hardened cement paste, air voids). A reduction of the data-flow is appropriate to substantially ease the evaluation. For this purpose a low-pass filter cuts down the measured signal to the frequencies significant for longitudinal modes in the concrete. Former researches pointed out optimal results around 100kHz . Consequently the testing-system can be switched to perform lowpass filtering at different cut-off frequencies. The optimum frequency to adjust is most dependend on the maximum aggregate built in the specimen. With intent to suppress excessiv scattering lower cut-off frequencies must be used at larger aggregate sizes putting up with a minor resolution (s. Cpt. 3). Both this adjustment and all further functions are completely controlled via the connected PC.
Due to the high absorption of the ultrasound in the concrete and the geometrical divergence resulting from the spherical shape of the ultrasonic field  it is advantageous to implement an adjustable gain control with optional apertures over the time of flight. It enables to analyse echo signals out of different depths at equal levels. By this it becomes possible to detect both an echo of a defect and the backwall in one scan. The amplification-curve can optionally be programmed free by the tester in regard to the physical demands.
|Pic. 1 Block diagram of the testing-system|
The quality and significance of a measurement substantially depends on the system transducer / specimen. Especially the applied transducer needs to have a high, most optimal to the concrete adapted attenuation to ensure both a maximum sound-input into the concrete and a minimum relaxation time (dead time) of the transducer after excitation. Such components also used here are available since a short time, so that now Pulse-Echo-Technique at concrete comes close to practical use. Besides a further development of the transducers a further optimazation can be achieved by using different coupling-means dependend on the testing conditions and additional shaping of the excitation pulse in relation to its frequency range and energy. At the testing system this is realized by a free adjustability of the duration of the excitation pulse from 0.1 to 20µs (el.). As a rule this optimazation has to be repeated for every new, specific testing arrangement (transducer/coupling meanslconcrete), but it is executable within a few moments via the PC. Additonally slight differences between identical designed transducers can occur and should be regarded. For the measurements presented in the next chapters the testing system NFUS2300 was operated. A schematic diagram is in picture 1.
3.1 Lateral resolution
By technical and practical reasons today's transducers used for concrete testing are mostly designed with diameters about 4cm or smaller. Forcing them to operate at low frequencies from 50 to 250 kHz no ultrasonic proximity zone with directed waves occures. The ultrasound propagates in concentric sperical waves around the transducer.
Just by this geometrical divergence the energy hitting a constant area decreases with 1/r2. Additionally the soundwave is frequently scattered at the inhomogenities of the concrete along its full propagation path. Hence a reflected signal e.g. from a defect will no more be significantly distinguishable from the scattered signal components. If the depth of a defect exceeds a certain value, it will no more be resolvable.
This argumentation directly leads to the second important parameter affecting the resolving power. Intensity and frequency range of the scattered waves substantially depend on the size and shape of the scattering volumes. Dealing with concrete the maximum aggregate size and the sort of the aggregates (gravel, chipping etc.) is the second, important parameter to consider.
In the following B-scans at two identically shaped specimens both with 3 defects of different sizes are presented. The specimens (Pic. 2) only differ at the maximum aggregate size built in (German grading curves AB16 and AB32 (mm)). Each was scanned from two sides positioning the defects in depths of 10 resp. 20mm. The lateral dimensions of the defects were 2, 4 and 6cm, the dimension rectangular to the trajectories of the scans was 10cm. A consideration correlating the resolving power with the total face of a defect follows further below.
The results of the measurements are presented in the B-scans in the pictures 3 to 6. In a depth of 10cm all defects could clearly be detected at both specimens (Pic. 3, 4). In a depth of 20cm the two larger defects at specimen AB16 could be seen (Pic. 5), while at specimen AB32 only the largest one stayed detectable (Pic. 6). According to the black and white printing in this paper it may be pointed out that the B- scans become less significant compared to the displaying in grayscale or indicated colours at the testing-system.
Pic 2 - Pic 6: (78k)
Pic 2: Sketch of the Specimens
Pic 3: AB16, defect at 10cm
Pic 4: AB32, defect at 10cm
Pic 3: AB16, defect at 20cm
Pic 4: AB32, defect at 20cm
|MaxAgrs [cm ]||x||Depth [cm ] |
Rl,spec = minimal lateral size of a defect to be detected (in cm)
MaxAgrS = maximum aggregate size (rounded in cm)
Depth = single distance between transducer and defect (in cm, >5 cm)
Rl,spec may only be seen as a first, orientation giving value for this specific testing system, for the built in sort of aggregate and for the here simulated, plain shape of a defect. The empirical expression especially isn' t valid in the near surface depths below 5cm because of the transducers dead time. On the other hand further measurements also applied to other specimens proved this valuation to give good results. As a rule a defect is earliest to be detected, if its projected surface parallel to the transducers surface in at least one dimension exceeds Rl,spec Nevertheless it must be considered that the orientation and the shape of a defect do influence its representation in the B-scan. Mainly parallel to the surface orientated defects will be detected by direct echo, while diagonal orientated and more scattering defects (e.g. rock pockets) may more lead to a loss of the backwall echo than to direct echos. An interesting step between is the localization of spherical defects like a tendon duct (s. Cpt. 4).
3.2 Resolution in Depth
The resolving power in depth can only be significantly described and measured, if a direct echo of a defect or a backwall can be detected (s. Cpt. 3.1). Not included in this evaluation are furthermore the inaccuracies generated by local variation of the ultrasound speed in contrast to the callibration and difficulties with the transducers coupling to the concrete.
The measuring of the depth / time of flight of an echo is here carried out by detection on the slope of the echo pulse in the A-scan. Hence the slope cannot always be detected clearly in the computed imaging of the B-scan due to its (accidental) phasing this proceeding is more accurate. Therefore the testing-system displays both the A- and B-scan on one screen simultaneously.
In good approximation the resolution in depth is describable as a function of the maximum (cut-off) frequency used for the measurement. Experimentally the resolution is rated as the distance represented by the first echo slope's time delay between 10 to 90% of intensity (equivalent to cat 1/4 of the recieved wavelength). Hence the measuring signal is decisively shapened by the preceded low pass filter of the testing-system a low limit for the resolution in depth RD,spec can be evaluated that is determined by the used cut-off frequency:
|k||x||Vconc [cm/s] |
4 x fFilt[Hz ]
RD,spec = resolution in depth (in cm)
Vconc = actual speed of ultrasound in the concrete (in cm/s)
fFilt = used low pass filter cut-off frequency (in Hz)
k = correction faktor (without correction k = 1 )
The correction factor k was added after some measurements made to adapt the expression to values verified by experiments. Due to reading accuracy and noise values for k are found between 1.5 and 2. E.g. with the low-pass filter switched to 100kHz (k=1.5, vConc = 4000m/s) the resolution in depth can be calculated to 1.5cm. In order to ensure a maximum accuracy a callibration of the transducer's and electronic's time delaying becomes obligate if the testing arrangement (transducer / adjustments) was more than slightly changed. Hence these values are constant at the same arrangement, they can be list in a callibration chart once they are determined. Carrying out the measurements presented in the following chapter special attention was appointed to this.
At both specimens AS8 and AS32 two position were marked to measure the wall thickness of 500mm by Pulse-Echo-Method. The Ultrasound velocity at the positions was separately measured in transmission. Therefore the testing-system can be optionally operated in transmission mode. Since the purser and the system adjustments stay the same in both modes, a maximum accuracy can be reached. In Pulse-Echomode a clear backwall-response could be found at specimen AS8, the so determined thickness was 508mm with a depth resolution of 15mm (s. Ch. 3.2). At specimen AS32 the backwall-echo especially in the A-scan at the marked positions was not significantly clear detectable. Indications of a backwall-response could be found in Aand B-scans at some other positions, but were judged not to be sufficient for a defi nite statement. An additional test at a specimen with 16mm maximum aggregate size is due to follow.
4.2 Localisation of the tendon duct
The tendon duct in specimen AS8 was precisely located from the smooth surface A by 16 lateral B-scans above its estimated position. The additional A-scans were interpreted to determine its position in depth. Pic.7 shows an example of a B-scan at the not reinforced half of the specimen. At this demand the online-imaging of the Bscan is quite advantageous, because even slight differences in time of flight or intensity of the signal can directly be observed by the tester. Specially the spheric geometry of the duct produces only a very small area of minimum time of flight and maximum intensity in the B-scan, so that the tester is able to localise this (lateral) area with an accuracy of about +- 5mm. A coarse impression of the duct's geometry is also represented, though it is not appropriate to make conclusions on its exact proportions by just one single scan. Having found this area and thereby localised the duct, its position in depth can be determined by analysing the simultaneous displayed A-scan as described in Ch. 3.2 . It is yet not finally proven, if the echo-slope of a spherica/ defect exactly represents the real distance to the surface. The values found for the depth may include a systematic deviation. Because this offset should be nearly constant at similar testing conditions, it doesn't deteriorate the measurement substantially. For exact values callibration measurement could be made at the laboratory.
A second advantage of the relative displaying of an online-B-scan is that interferences produced by reinforcement bars do not unable the localisation of the duct totally. Though there are partly massive interferences to be observed, the duct continues to be detected by relative alternations in the scan as shown in Pic. 8. Naturally the interferences lower the accuracy of the localisation, but it's due to the tester to find the best line to scan e.g. in the middle between two bars.
|Pic 7 |
Duct in Specimen AS 8
- not reinforced half -
Duct in Specimen AS 8
- reinforced helf -
Pic. 9 shows all results measured at specimen AS8 at a glance. It confirms the deviation of the duct against the design also found by other groups.
|Pic. 9 Results of the Measurements at the BAM|
|Top to this page|
|| UTonline ||