Bundesanstalt für Materialforschung und -prüfung

International Symposium (NDT-CE 2003)

Non-Destructive Testing in Civil Engineering 2003
Start > Contributions >Lectures > Tendon Ducts: Print

Detection and localization of fractures in tendons by means of electromagnetic resonance measurement

H.-J. Wichmann, A. Holst, K. Hariri, H. Budelmann


The reliability of tension members is a requirement for the safety of prestressed structures. Very few methods to identify and locate fractures in prestressed tendons are currently available. Detailed informations about the prestressed tendons location in the structure are generally required.

An alternative technique is being developed in the current co-operation of two part-projects of the Collaborative Research Center SFB 477 (http://www.sfb477.tu-bs.de/) in Braunschweig. It uses the electromagnetic resonances of the tendon to detect steel fractures.

The suitability of this method was studied through application tests on experimental setups, lab-made prestressed concrete structures and on industrially produced concrete hollow slabs with simulated steel fractures. Further tests in the laboratory and on site are planned.


Due to some spectacular collapses of buildings made of prestressed steel (e.g. the former Kongresshalle in Berlin), research work started to develop new methods to detect steel fracture in the 1980s. These buildings collapsed suddenly and without forewarning. Since then, several methods to detect steel fractures have been developed and improved.

Magnetic methods are currently the foremost methods for the non-destructive detection of steel fractures [1,2]. The most prominent one uses the remanent magnetism of magnetized tendons [1]. Fractures are detected by a change of sign of the flux density.

The magnetic methods have the following limitations of usability in common:

  • the tendon has to be scanned along its full length
  • the exact position of the tendon must be known
  • if several tendons are positioned on top of each other, only the uppermost and/or the bottommost tendons can be scanned
  • only tendons up to 20-25 cm installation depth can be inspected.

A further method to detect and locate a fracture is to use an electromagnetic measurement method. The basic idea of this method is to consider the tendon itself as an unshielded resonator located in a material with electromagnetic loss (e. g. concrete). An electromagnetic wave of variable frequency is coupled into the end of the tendon. By systematically scanning the reflection coefficient |S11| over a spectrum from low to high frequencies, resonance frequencies of the tendon are recorded [3, 4].

The spacing Df between two adjacent resonance frequencies is inverse in proportion to the length of the tendon resp. to the distance to the fracture and also depends on the surrounding material, as is shown in the following equation:


where c0 is the vacuum speed of light, lthe length of the tendon and er the dielectric constant of the surrounding medium.

A main advantage of this method is that only one end of a tendon has to be accessible and that no further sensors are needed. Figure 1 shows the principle of the electromagnetic resonance measurement.

Fig 1: Principle of the electromagnetic resonance measurement.

Figure 2 shows the scheme of the experimental setup for the first lab experiments. An 8m long and 96cm wide frame was made to simulate various combinations of tendons. In order to minimize the influence on the electromagnetic waves, the frame is made of wood. The cross view shows that the frame can be equipped with up to 20 tendons (4 groups of 5 tendons each). The four groups are labeled A to D from the left to the right. The five possible tendon positions are numbered 1 to 5
(1 => upper left, 5 => lower right). A wooden cross section is placed in the frame every meter along the length. It ensures that the tendons are running parallel and do not bend along the full length. These cross sections even enable the simulation of steel fractures. A definite spacing of the simulated steel fracture is adjusted by means of a hollow plastic shaft.

Fig 2: Scheme of the experimental setup.

The cross view shows that the frame is also divided in four sections along the length. This makes it possible to fill single sections of the frame with e.g. sand and to simulate different humidities or various materials around the tendons. It allows the simulation of tendons running partly through air, partly through concrete.


Fig 3: Sensors used for electromagnetic coupling.

Two types of sensors are used for the coupling of the electromagnetic wave into the tendon (Figure 3). The first sensor (left picture of figure 3) is made of a 10cm long, standard semi-rigid cable
Æ 0.250" with a microwave connector. The inner wire sticks out of the coaxial cable with about 3 mm and is pressed to the tendons surface.

In concrete structures with an anchor plate, a different type of coupling is necessary. This sensor is shown in the right picture of figure 3. The inner wire sticking out of the supplying coax cable is connected to the tendon by a clamping piece. To assure a defined electrical ground potential the outer conductor is connected to another tendon. Due to the fact that the coax cable can dodge the anchor plate, a lead-through in the anchor plate is not necessary. This sensor has to be integrated in the construction during the course of its erection. Access to the sensor is achieved through a microwave connector.


The following figures show the reflection parameters of some experiments made using the wooden frame and various concrete structures.

Figure 4 shows the results measured on a single 59.4 m long wire in air. According to equation (1) the spacing between two resonance frequencies is Df =2.52MHz at a length of 59.4m. The calculation correlates to the measured resonance spacings Dfmeas (Figure 4, diagram to the right).

Fig 4: Reflection coefficients |S11| of a 59.4 m long wire in air.

In figure 5 to the left the reflection coefficients of single steel wires in air with lengths of 430cm, 630cm und 820cm are shown. According to equation (1), a reduction of the length results in a greater spacing Df .

Fig 5: Reflection coefficients |S11| of single steel wires in air.

Table 1 shows the averaged values of the measured spacings Dfmeas(Figure 5, diagram to the right) and the calculated spacing Dfcalcof resonance frequencies.

Length [cm] 430 630 820
Dfcalc (MHz)34,923,818,3
Dfmeas (MHz)3322,516,5
Table 1: Comparison between calculated and measured spacings.

The measured spacings Dfmeas (MHz) are about 3-5% smaller than the calculated values. The simplified equation (1) is based on a ideal open-loop characteristic at the end of the steel wire. In reality there is an additional parasitic capacity between the steel wire and the ground potential.

Steel wires of different diameters and different alloys were tested in the experiments. As expected, neither the diameter nor the alloy of the wires influence the spacing Df (MHz)of resonance frequencies.

The influence of the gap length in a steel fracture is shown in figure 6. The wire was 820cm long with a simulated fracture at the length of 330 cm. Gap lengths of 0.05cm and 1cm were tested.

Fig 6: Variation of gap length in a steel fracture.

The fracture of the 820 cm long wire at 330 cm is clearly detectable, no matter if the fracture gap is 1cm or 0,05cm long. However, if electrical contact exists at the gap, the fracture is not detectable.

The influence of wires on each other were examined by varying both the number and the distance of parallel running wires. The signal was coupled into the wire in Field b3 (430 cm length). Up to 3 wires with a length of 820 cm were applied into the Fields a3, c3 and d3 of the wooden frame. The downwards oriented arrow in figure 7 displays the change of the reflection coefficients with an increasing number of wires. The amplitude of the minima is more distinct, the more parallel wires there are. The spacing Df (MHz)of resonance frequencies is barely influenced by the number or the vicinity of parallel wires.

Fig 7: Reflection coefficients |S11| of wire b3 with different numbers of parallel wires.

Even in multi-wire tendons it is possible to determine the length of the wire resp. the distance to the fracture. Three different multi-wire tendons were examined; one without a fracture, the two others with a fracture in one wire at a length of 420cm resp. 620cm. Figure 8 shows their different reflection coefficients. The measured spacings Df (MHz)correlate with the results of the single wire experiments.

Fig 8: Fracture localization in a tendon with 5 coupled wires in air.

The electromagnetic properties of resonators like a steel wire are highly dependent on the environment (e.g. concrete, ducts). Sand has roughly the same dielectric properties as concrete. In the first experiments sand was used as fill material in the wooden frame to simulate a concrete environment. According to equation (1), the spacing of resonance frequencies shortens in sand by the factor in comparison to wires in air.

Figure 9 shows the reflection coefficients of three wires with a length of 430cm, 620cm and 820cm in sand. The dielectric constant of dry sand is about er" 3.7. The spacing Df shortens by the factor (Figure 9, right picture, comp. figure 5).

Fig 9: Reflection coefficients |S11| of steel wires of several length in sand.

During the last two years the resonance method was tested on different types of concrete structures. Figures 10 and 11 show some results measured on a 3m long concrete structure with seven wires. The positions of the seven wires are shown in both figures. The wires were labeled S1-S7. Each wire had a total length of 3m, but the wires S3 and S5 had a fracture at the length of 1.5m. From the moment the wires were set in concrete, the reflection coefficients of all wires were measured regularly during a whole year. Figure 10 shows four results of wire S3 between the third and the 52nd week and additionally one of wire S1 in the 52nd week.

Fig 10: Reflection coefficients |S11| of wire S3 and S1 in the concrete structure.

Due to the high moisture level and the ionic strength of the pore solution, the conductivity and the damping of electromagnetic waves in fresh and young concrete are very high. Resonance frequencies usually cannot be clearly determined. After 35 weeks the moisture level was still too high for a clear specification of resonance frequencies.

Fig 11: Reflection coefficients |S11| of wire S7 in the concrete structure.

In order to accelerate the drying process and to reduce the moisture content, the structure was placed in a climatic chamber. After drying for seven more weeks at a temperature of about 40°C, the resonance frequencies appeared clearly and the fracture was indicated in the results. The same can be observed in old concrete structures that have aged under normal climatic conditions. A continued drying enlarges the amplitude of the resonance frequencies.

The drying process of concrete leads to a smaller dielectric constant. Thus earlier measurements show a shorter spacing Df. For an exact calculation of the wire length the dielectric constant must be known. The dielectric constant of the tested concrete structures was measured with the coaxial-sensors developed in the project C1b of the Collaborative Research Center.

Table 2 shows the calculated and the measured resonance spacings for the wires S1 and S3 in the 52nd week with an dielectric constant er = 6 of the concrete.

Wire S1 S3
Table 2: Comparison between calculated and measured resonance spacings Df.

One possibility to minimize the influence of the high conductivity in young concrete is to cover the wire with a thin PE-coating. Figure 11 shows the corresponding reflection coefficients of the coated wire S7 during drying process. The coating isolates the wire from the ionic pore solution of concrete or grout. Thus the resonance frequencies appear earlier and the spacing Df can already be determined after a few days.


Magnetic methods are mainly used today for the determination of fracture in steel. In these methods the tendon is magnetized and scanned parallel along its full length. These methods have proved their suitability in practical use. The electromagnetic resonance method developed in the Collaborative Research Center SFB 477 in Braunschweig could be considered as a supplement to the magnetic methods. As the experiments with concrete structures have shown, it cannot be used to determine steel fracture in very young concrete due to its high conductivity. The practical use of this method is therefore limited to older structures. An advantage is that it allows the inspection of tendons in some cases, in which the magnetic methods cannot be used, e.g. ground anchors. Further investigations with more complex tendons are going on.


  1. Hillemeier, B.; K; Scheel, H.: Non-destructive location of prestressed steel fractures in post-tensioned and prestressed concrete, Transportation research board Committee A2C03-Concrete Bridges, Washington (DC), January 2002
  2. Sawade, G.: Anwendung der Methode der magnetischen Streufeldmessung zur Ortung von Spannstahlbrüchen, DGZfP-Berichtsband 66 CD, 73, 1999
  3. H. Budelmann, A.F. Jacob, H.-J. Wichmann, B. Jannsen, G. Schneider, K. Hariri: Verfahren zur Zustandserkennung von elektrisch leitfähigen länglichen Spanngliedern. Deutsches Patent- und Markenamt, DE 101 02 577 C1 (20. Juni 2002).
  4. Hariri, K., Holst, A., Wichmann, H.-J., Budelmann, H., "Assessment of the state of condition of prestressed concrete structures with innovative measurement techniques and first applications", in "Proceedings of the 1st European Workshop on Structural Health Monitoring 2002", July 10-12, 2002, Paris, 1278-1285, 2002
STARTPublisher: DGfZPPrograming: NDT.net