·Home ·Table of Contents ·Materials Characterization and testing

Barkhausen effect based sensor and intellectual system for accurate stress measurements and monitoring in harsh environmental conditions V.Vengrinovich, V.Tsukerman, Y.Denkevich. Institute of Applied Physics, Akademicheskaya str. 16, 220072 Minsk, Belarus. E-mail: veng@iaph.bas-net.by Contact

1.Introduction

Many efforts have been made during last decades to develop the reliable NDE methods and instruments for stress measurements in various materials, constructions and machine parts. Vital demands of the industry for stress NDE techniques are disseminated on the manufacturing processes, in service maintenance and monitoring. Each of developed technique performs itself in terms of sensitivity, accuracy, penetration depth, spatial resolution of the sensor and of the measurement area in the material, etc. All such measurements should be calibrated with a material similar to that to be tested and in similar environmental conditions. Any deviations of material properties and/or conditions during measurements introduce an errors and corrupt the result.
Table 1, mainly taken from [1], adds together most important NDE techniques for stress measurements focusing on those their characteristics which are important for developing the advanced Barkhausen Noise Sensor (BNS).

Technique	Accuracy	Sensor spatial resolution	Penetration depth	Distinguish from microstructure and texture
Magnetic and electromagnetic: coercive force residual magnetiz. higher harmonics magnetic permeability	ca. ± 10% with calibr.	>30 mm2	< 10 mm	Partially, by combination of differ. parameters, application of differential measurem., higher harmonics
Barkhausen emission	ca. ± 6% with calibr.	> 1 mm2 [4]	< 1 mm	No. Partially by selection of appropriate set-up parameters and principle stresses computer restoration
X-ray diffraction	± 35MPa	<1 mm2	13 mm	Yes,but some errors from grain size, cold working and microstr.
Neutron diffraction	± 100MPa , Line average measurement.		3 -4 cm	Yes. The microstructure can also be measured separately
Ultrasonics (bulk)	10%. Stressrelief technique.	> 10 mm 2	> 10cm [18]	Yes. Microstructure independence for plates
Ultrasonics (surface)	10% with calibration	> 10 mm 2	One wavelength < 1mm	No. Difficulty with grain orientation compensation.
Table 1:

Magnetic Barkhausen emission (MBE) is grounded on the phenomena of discontinuous displacements of Bloch Walls. State-of-the art of the theory and applications for non destructive stress testing are given in [1,2]. Essential progress in BNS technology was done during last years [3,4].
High stress sensitivity of MBE is widely recognized. Nevertheless diverse influencing factors does not provide its using as a practical tool for applied stress on line measurements and hence to create a stress sensor alternatively to a strain gauge sensors, which in a moment has extensive industrial applications. At the same time severe constraints eliminate their use in practice, main of them being as follows:
• influenced by aging and A/C environment
• show insufficient mounting results and high maintenance effort
The reasonable obstacles on the way to develop MBE based stress sensor are temperature sensitivity, stochastic nature of Barkhausen noise (BN) excitation process , low reproducibility, critical gap and surface condition sensitivity, available calibration and verification procedure. The resultant accuracy of measurements depends on the accuracy both of calibration and measurement themselves. Repeatable and reproducible MBE based instrument and technique for applied stress measurements with the accuracy of ±0.5% in a harsh environment and under the influence of disturbing factors - is the challenge of the present development.
Starting from 1997 under the support of EC it were created and provided laboratory tests[Independent stand tests of the system were provided in Riga Technical University and Aviation Institute leaded by V.Fastritskiy , A.Shweits and V.Pavelka.] of the new BN based hard- and software for accurate stress measurements and monitoring. Innovations and new engineering ideas were developed which made it possible to exceed essentially the known MBE based sensors parameters.

2. Optimal BN frequency range and set-up parameters stability

The chart of the BN sensor used for experiments is shown in the fig.1.


Fig 1: The chart of the BN sensor used for experiments.

The experiments were provided using the specimens of steels 300M, 30HGSNA and VNS2 alloy (steel 300M in the conditions after as-ground heat treatment, shot peening and cad plated). The block-diagram of the experimental device for the investigation of BN frequency range optimization is shown in fig. 2.


Fig 2: Block diagram of the experimental unit for the investigation of BN frequency range optimization.

Main parameters for data acquisition system are represented in the table 2.

Frequency of magnetization current	125 Hz
Shape of magnetizing current	Triangular
Amplitude of magnetizing current	controlled, 1..255 mA
Band filters (two)	100 kHz and 700 kHz
Measurement time	controlled, 0.2 s, 1 s, 4 s
Table 2:

Voltage amplitude - frequency characteristics of signal in the filters' output are represented in the fig. 3.


Fig 3: Voltage -amplitude - frequency characteristics of signal in the filter's output

Fig. 3 shows distinctly that the frequency range of detected BN should be extended up to the frequency of about 1MHz in order to provide from one side the highest stability of random signal and from the other side to save more information in the signal.

3. Resolution of BN measurements

To investigate the resolution of measuring system using random BN signal the deviation of BN signal readings depending on measuring time, spectrum bandwidth and envelope shape were measured. The spectrum bandwidth was changed by band filter while envelope shape - by selecting different amplitudes of magnetizing field strength. Standard deviation was evaluated from realizations of 1000 measurements. The results are represented in the table 3 while the corresponding envelopes of BN signal - in the fig. 4.

Set-up parameters	Measurement time	Deviation
filter 100 kHz	0.2 s	1.28 %
current 170 mA	1 s	0.77 %
envelope 1 in the fig. 2.5.1	4 s	0.44 %
filter 100 kHz	0.2 s	0.70 %
current 255 mA	1 s	0.39 %
envelope 2 in the fig. 2.5.1	4 s	0.16 %
filter 700 kHz	0.2 s	0.20 %
current 255 mA	1 s	0.08 %
envelope 3 in the fig. 2.5.1	4 s	0.05 %
Table 3:

Fig 4: Envelopes of BN signal which correspond to three cases in Table 3.

The results are satisfactory agree with the theoretical value of deviation of integrated BN signal:


(1)

where s - standard deviation, T - time period of magnetizing cycle, T_c - width of envelope peak, B - spectrum bandwidth, t - time of integration (measurement).
BNS resolution was investigated experimentally and it didn't exceed 0.06%.

4. Investigation of the BNS temperature and stress sensitivity

The investigation was provided with three groups of specimens of 300M steel having the sizes 152.8x25.52x4mm in the conditions relatively: as ground (specimen 5415); quenched, released and shot peened (5408, 5413 and 5414) and also quenched, released, shot peened and finally Cad plated (5409 and 5410)[The specimens were developed and manufactured by Dowty Ltd. and British Aerospace.] . The microstructure of the first group was ferrite-perlitic, while for the second and the third group it consisted of martencite.
Stress increment s in the BNS location zone was created by cantilever loading of the sample and the temperature of the whole assembly was changed in the range -60⁰ - +70⁰.
The error of loading Ds didn't exceed 1.2 MPa. The normalized error of loading e_s was estimated of being 0.15%. The temperature variation was provided by injecting a hot air or by placing a dry ice into the thermostatic jacket. The specimen temperature was measured in immediate proximity to the pick-up coil with the help of the thermometer gauge having the absolute error ± 1⁰C and built-in BNS.
Measurement results of the BNS signal under different temperatures and loads were checked in compliance with the following model of calibration function s(S,t) of the stress on the BNS signal and temperature:


(1)

where S - BNS signal value, t - temperature value, K₀...K₃ - calibration coefficients in the function   s (S,0) of the stress via the BNS signal under the zero temperature, K_t - calibration temperature coefficients.
The corresponding diagrams which perform the functions of the stress via the BNS signal s(S,0) under the zero temperature for all three kinds of specimens treatments are shown in fig. 5. Specimens numbers are shown in the top left side of each diagram. Despite of the fact that all experimental calibration functions are close to linear one, the approximation with the polynomials of higher order was applied and corresponding approximation formulas are shown in the top right sides. One has to pay attention to the fact that the scales on the BNS signal axis are quite distinguishable due to the different heat treatment conditions of the specimens. Meanwhile all the functions are close to liner.
The series of experiments provided during investigations had definitely shown that it is necessary to provide independent calibration for the materials in different heat treatment groups, but the calibration function is stable inside the group of specimens with the similar heat treatment conditions (samples 5408 and 5413) within the specified accuracy better than 1%. Calibration tests were provided having also the goal to determine the order of polynomial p which approximates the calibration function of the stress value variation via BNS signal at zero temperature. The best fit was provided with p=4.
For any of the specimens the calculations of the coefficients K_t have been implemented. For every measurement result S_n at the temperature t_n and stress s_n the correction of the stress value Ds_n caused by temperature influence was calculated:


(2)

where K₀...K₃ - calibration coefficients estimate in the function s(S,0) of the stress via the BNS signal under the zero temperature.
For a set of pairs of values {t_n,Ds_n} by the regression method the coefficients of the following equation was calculated:


Ds(t) = Ktt,

(3)

where K_t - estimate of temperature coefficient. The corresponding diagrams which perform the functions of the temperature caused stress correction values are shown in fig.6.


Fig 6: The experimental (dots) and approximation (solid lines) calibration functions s (S,0) of the stress value determined from BNS signal values for the specimens from three groups: 5415 - as ground: 5408, 5413 - quenched, released and shot peened and 5409 - quenched, released, shot peened and finally Cad plated.

Fig. 6. The experimental (dots) and approximation (solid lines) calibration functions s(S,0) of the stress value determined from BNS signal values for the specimens from three groups: 5415 - as ground: 5408, 5413 - quenched, released and shot peened and 5409 - quenched, released, shot peened and finally Cad plated.
It is seen from fig.7. that while the temperature caused stress correction values are high enough not to neglect them in precise measurements, their absolute values are not so much influenced by the heat treatment conditions and hence by the material's microstructure. That comes out from the fact that for as ground (5415), heat treated and shot peened materials (5408, 5413 and 5414) the correction values are very close to one another. . Finally the standard deviation of the measurement resulted for the chosen model didn't exceed 0.4%.
At the same time the plating of the specimen (5409) changes and scatters the correction values considerably.

Fig 7: Functions of the temperature caused stress correction values for the corresponding specimens.

Conclusions

1. The new Barkhausen Noise Sensor, Electronic Unit, Software, Measurement, Calibration and Verification techniques for accurate applied stress measurements in harsh environment, system for safety signal transmitting are developed and tested.
2. New innovations and engineering ideas were developed which made it possible to exceed essentially the known Barkhausen noise sensors parameters particularly: resolution, time and temperature stability, accuracy, portability, insensitivity to disturbing influence and repeatability.
3. The Barkhausen noise stress sensitivity, reached with new BNS in experiments with steels 300M, 30HGSNA and VNS2 alloy (in the conditions after as-ground heat treatment, shot peening and cad plated), is shown to be sufficient to provide the resolution of 0.06% and accuracy less than 0.4% in the temperature range (-50 ¸ +70) ⁰C and bending stresses in the range (0 ¸ +800)Mpa under influence of shocks, vibrations, high humidity, electromagnetic field.

References

1. D.J.Buttle,C.B.Scruby,G.A.D.Briggs and J.P.Jacubovics, The measurement of stress in steels of varying microstructure by magnetoacoustic and Barkhausen emission, Proc. R. Soc. Lond. A 414, pp. 469-497, 1987.
2. V.Vengrinovich. Barkhausen noise. Theory and applications. Minsk, Nauka i Technika, 1991.
3. Seppo I.Tiitto. US Patent N 4634976 from Jan. 6,1987. Barkhausen noise method for stress and defect detecting in hard steel.
4. V.L.Vengrinovich,A.L.Vishnevski,V.N.Busco, On selective evaluation of stresses, residual plastic deformation and metal degradation by magnetic noise method, 14^th World conference on NDT, pp. 215-216, New Delhi, India, 1996.

© AIPnD , created by NDT.net

|Home|    |Top|